Nam Pyo Suh

Miguel Cavique
Joseph Timothy Foley
Editors

Design
Engineering
and
Science
Design Engineering and Science
Nam Pyo Suh • Miguel Cavique •
Joseph Timothy Foley
Editors

Design Engineering
and Science

123
Editors
Nam Pyo Suh Miguel Cavique
M.I.T. Department of Sciences and Technology
Cambridge, MA, USA Portuguese Naval Academy
Almada, Portugal
Joseph Timothy Foley
Department of Engineering
Reykjavík University
Reykjavik, Iceland

ISBN 978-3-030-49231-1 ISBN 978-3-030-49232-8 (eBook)

https://doi.org/10.1007/978-3-030-49232-8
© Springer Nature Switzerland AG 2021
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface

Many engineers chose their profession because they like to design and create
artifacts using their imagination—even the writers of fiction design their stories and
their presentation scheme early in their endeavor. People in business must design
their business plan well to succeed. Even government policy-makers must design
new policies and implementation plans to get the support of the public and benefit
their constituents. Design is the first step in creating something new. This book is
for those who are interested in learning a systematic and rational way of designing
artifacts to fulfill their goals.
An international group of professors, scholars, and practitioners of a design
prepared this book for students of all ages, especially for those college students who
want to design, innovate, and create. They have taught design in several universities
located throughout the world and practiced the design in diverse fields. This book is
a manifestation of their desire to improve the design education of the upcoming
generations and the practice of the designers and others in the field who must design
to satisfy new needs and aspirations.
The synthesis of a variety of systems is an integral part of education, especially
in engineering. This book reflects the authors’ belief that design education can and
should be done more effectively, since it teaches the basics of synthesis, i.e., how to
conceive, create, and deliver new solutions to complex problems of all kinds,
transcending specific disciplines. To be a leader in any field, it will be of immense
help to know the fundamental of the design field to be effective in delivering
solutions of all kinds. The ability to design is essential to everyone who wants to
improve societal functions, to create new imaginative products and processes, to
formulate enlightened policies, to improve organizational effectiveness, and to
deliver creative solutions to problems of all kinds.
The artifacts created through designs have changed the course of human history.
Indeed, various designed systems, combined with scientific advances in natural and
biological sciences, constitute the essence of recent human history. Newton’s
pioneering scientific work in 1686 and the Industrial Revolution that began with the
invention of the steam engine by James Watt in 1782 helped to create a modern era
with unprecedented improvements in the quality of human life as well as significant
economic advances. Overall, humanity has done well during the past four centuries
through their ingenuity. They created designed systems that have laid the

v
vi Preface

foundation for all aspects of human lives. Most of all, they initiated the movement
for knowledge-based technological, scientific, and innovative cultural era for
humanity. In the future, the design thinking that has partly led to these advances
will contribute much more to advance worthy human causes and avoid anthro-
pogenic disasters we face in the twenty-first century. We, humanity, know what
logical thinking can do in the creation of new solutions to deal with challenging
problems.
The contents of this book are primarily based on Axiomatic Design (AD). It has
been applied to a variety of different problems in many fields. The common goal in
writing this book is to improve the capability of the student and industrial practi-
tioners in designing and implementing a variety of different systems, both technical
and non-technical. Through proper teaching of design, we hope that the costly
mistakes—both financially and humans—made in the past because of poor designs
can be minimized or ultimately eliminated in the future. Mistakes made out of
ignorance or the misunderstanding of the existing knowledge that could have saved
a calamity are difficult to be excused or justified.
We are fortunate to be living in this enlightened era of the twenty-first century,
where logical and rational reasoning power of humans is guiding the thought
processes and behavior of people across many continents. As a result, the economic
development of the world is at an unprecedented level of advance; scientific dis-
coveries are taking humanity into unchartered, and often exciting, territories.
Technologies have transformed many countries to new heights of prosperity, and
healthcare has extended the longevity of people by decades. Overall, people enjoy a
higher quality of life in the twenty-first century than ever before in human history.
Many in many fields have contributed to these advances: scientists, technologists,
humanists, and societal leaders. Design thinking has been one of the final steps in
harnessing the advances in human knowledge to create practical and realizable
solutions for society.
Human history dates back several thousands of years. During most of this
period, many people lived under the dictates of a few tyrants who ruled the people
within their territory with iron fists under a class system that used most people as
power sources (with intellect) in such fields as farming, construction, transportation,
and others. Farmers—tenant farmers, hired hands, and slaves—tilted grounds that
were owned by landlords; laborers forged metal parts, transported goods, con-
structed dwellings, and even gave up their lives fighting for their masters. Most
people, except those in privileged positions, toiled without the time, freedom, and
mental capacity to think and contemplate about improving their lots and plan for a
better future for their posterity. Their daily subsistence was the critical issue that
preoccupied their minds. Poverty can be most cruel to those without independent
means. These systems lasted for centuries until relatively recently.
Starting about 350 years ago, the life of people began to change because of the
emergence of science and technology. The “new science” of the seventeenth cen-
tury rejected the Aristotelian idea of seeking the essence of four causes. The new
paradigm of science accepts that there are laws for physical phenomena, opening
the doors for modern science and technology. This birth of a new era was initiated
Preface vii

because humanity was blessed with bright people with exceptional minds (e.g.,
Isaac Newton) who changed the relationship between humanity and nature through
discoveries and inventions. New power sources that were introduced with the
invention of the steam engine by James Watt enabled people to replace human
power with machine power. It also increased their productivity, and allowed people
to use their minds to conceive, discover, and invent many artifacts that improved
the quality of life as well as productivity. Scientific theories were advanced to
explain the invisible as well as observable causality of various natural phenomena.
Also, the advances made in human-designed systems amplified the efficiency and
productivity of human efforts. In three short centuries, people with exceptional
minds have transformed the world and humanity, creating an enlightened modern
era.
The table below is a partial and arbitrary list of the significant events that
changed or might change the world. They are theories, inventions, and innovations,
some of which have transformed or had the potential of changing the future of
humanity through technological advances. These advances in science and tech-
nology began to occur only about ten generations ago, a short time relative to
preceding tens of thousands of years of human existence.
A partial and random list of science and technology that created or may change
the basis of human society in the future.
1686: Newton’s laws.
1767: Electricity by Ben Franklin and others.
1782: Steam Engine by James Watt.
1854: Thermodynamics (Lord Kelvin).
1861: Electromagnetism, Maxwell’s Equations.
1876: Telephone, A. G. Bell.
1903: Airplane (Wilbur and Orville Wright).
1905: Einstein’s Special theory of relativity.
1906: Digital Computer (Alan Turing, John Astanoff in 1939).
1911: IBM mainframe.
1917: Moving automobile assembly line by Henry Ford.
1928: Penicillin by Fleming.
1936: Turing machine by Alan Turing.
1946: Transistor.
1948: Information theory by C. Shannon.
1950: NC machine tools by John D. Parson and MIT.
1950: Jet airliner.
1953: DNA by Watson and Crick.
1956 Artificial Intelligence (AI) by Newell, Simon, et al.
1960: FDA approval of the contraceptive pill.
1960: Packet Switching (Internet) by Donald W., Davis and Paul Baran.
1960: Global Positioning Device by Roger L. Easton, Ivan A. Getting, and
Bradford Parkinson.
1964: IBM Mainframe 360.
viii Preface

1964: Medical IT software by A. N. Pappalardo.

1973: First mobile telephone.
1975: Microsoft by Bill Gates.
1976: Apple by Steve Jobs.
1976: Axiomatic Design Theory.
1980: Microcellular Plastic.
1993: 3D Printing.
1998: Google by Larry Page and Sergey Brin.
2004: Facebook by Marc Zuckerberg.
2011: On-Line Electric Vehicle (OLEV).
The modern scientific and technological era that began with the Industrial
Revolution, thanks to the invention of the steam engine by James Watt, and sci-
entific contributions by Newton and others, established the foundation for the
democracy that we enjoy today in many parts of the world. However, even today, it
will take constant vigilance to be sure that scientific and technological advances are
employed to further democracy because there is always a tendency by a few to
attempt to assume dictatorial power.
People used three different paths for the development of scientific theories and
technologies. First is the technological invention such as the steam engine by James
Watt. The second path was through scientific discoveries (such as the DNA
structure) based on observable experimental evidence. Many of them were readily
accepted as soon as they were published. The third is based on fundamental sci-
entific postulates and axioms, e.g., thermodynamics had gone through years of
debate and discussions before they were finally accepted. Even today, the second
law of thermodynamics is still being debated after more than a century. That must
be the nature of axioms that make people skeptical because of their simplicity and
yet capture profound thoughts and principles. It is remarkable that even Newton’s
laws, which were partly axiomatic at the time they were advanced, were disputed by
some of Newton’s contemporaries.
The field of Axiomatic Design is still in its early phase of replacing the old
practice of making design decisions based on experience and empiricism through
long recursive design/build/test cycles. Establishing Axiomatic Design as the basis
for decision-making in the design of systems has been challenging. Many people in
charge of development projects throughout the world depend primarily on their
experience, notwithstanding numerous failures of highly publicized projects,
attesting to the limiting nature of this practice. Universities must lead to the
transformation of design practice by generating people who are better rooted in
design science and practice. Education is the ultimate means of changing engi-
neering practice for the better.
The purpose of this book is to teach the theory and practice of Axiomatic Design
(AD) to students. AD provides a logical framework and the scientific basis for
design, which can generate better and creative solutions in many fields of design
and synthesis, including engineering, materials, information, systems, and
Preface ix

organizations. Major development projects demonstrated that Axiomatic Design

could effectively replace lengthy trial-and-error processes often used in developing
new solutions and artifacts. AD enables the designer to organize one’s thoughts
quickly and correctly in dealing with large systems. The organized methods of AD
can enhance creativity. People can improve their design practice and advance the
field of design by learning the basics of AD.
Our hope in writing this book is modest. The reader of this book acquires the
basics of how to design “systems” in a rational way.
The student who masters the subjects presented in this book should be able to
execute the following:

1. design uncoupled or decoupled systems of many different kinds;

2. manage large projects to create imaginative design solutions and satisfy cus-
tomer needs;
3. become a system architect who can lead the development of large systems
projects;
4. manage projects without incurring cost overruns and missed schedules;
5. lead significant projects by making the right decisions that do not lead to
coupled designs;
6. avoid introducing complexity to designed systems;
7. lead and manage large organizations to achieve the intended goals of the
organization.
The abilities listed above, once internalized, should give the student the in-depth
systematic knowledge on the design and operation of many different systems. The
knowledge they acquire should prepare them for diverse career paths. It should also
complement the experience-based expertise gained in the industry by practicing
engineers. Ultimately, their success in creating imaginative and rational solutions
should enhance the quality of life of humanity that began with the invention of the
Watt engine and scientific advancements made during the past three centuries. The
intellectual challenges related to the discipline of design will always be there, the
latest being the design of cybersecurity systems for the information stored in
computer systems throughout the world.

On the Role of Computers and Software in the Field of Design

In this preface, only design theories in dealing with the real problems of the world
were discussed. However, the computational power of computers and the appli-
cation of artificial intelligence (AI) in design to augment human design capability
deserve much attention. All of these developments will benefit from the theoretical
foundation laid down by design theories such as Axiomatic Design.
The attempt to use computers in design began in the mid-1950s with the effort to
replace human drafting, graphics, and sketches with computer-generated tools
under the heading of computer-aided design (CAD). Then, it quickly moved into
x Preface

the 3D rendering of complicated mechanical designs. In the 1970s, solid geometric

modeling became the fashion of the day, replacing many draftsmen and mechanical
part designers. It is soon followed by the 3D representation of complicated parts.
Now, these are the commonly used tools.
Several related developments have driven the use of computers in design: low
cost of memory, fast computational speed, ability to reduce the memory size, the
high quality of graphics, and advances in telecommunications that also depend on
the use of computers. Because of the low cost, high speed of computation, and
advances in AI, many things, even very inefficient systems, have been attempted to
advance the design field. Such an effort will continue, but ultimately, they will
increasingly be based on issues and advances discussed in this book. Probably, the
industry will drive this transformation because of their urgent need and demand for
efficiency, accuracy, and reliability. Most of all, there is the driving need to prevent
such spectacular failures as Boeing 737MAX.
The use of computers in design will continue to expand in many dimensions,
some of which we cannot even fathom today. However, one thing is clear—a firm
theoretical foundation for decision-making in design will be of immense importance
in the future development of design engineering and science, regardless of which
tools are used to improve the design efficiency.

In Conclusion

The idea of writing this book was agreed on at the 2017 International Conference
on Axiomatic Design (ICAD), which was held at Gheorghe Asachi Technical
University in Iaşi-România between September 13 and 15, 2017, kindly hosted by
Professor Laurentiu Slatineanu. It was a great deliberation. Since then, many people
worked hard to deliver what they promised.
Many outstanding people toiled and worked hard, revising their chapters many
times. In many ways, it has been a gratifying experience to work together so well,
notwithstanding the geographical distances, language barriers, and differences in
educational systems and engineering practice. It was a labor of love—a salute to all
of them.

Cambridge, USA Nam Pyo Suh

November 2019 for the Editorial Committee
Acknowledgments

Design is the third leg of the proverbial three-legged stool of engineering, science,
and nearly all other fields, including economics. The ingenuity of people created
many amazing things that have been designed over many decades and centuries to
raise the quality of life of people everywhere. Yet, the field of design has never
gained academic and disciplinary rigor because of the view of some people that
“everyone can design through the acquisition of experience,” albeit intuitively.
Humanity has paid the price for this attitude—planes that fell out of the sky, nuclear
power plant accidents with substantial human costs, cars that killed many because
of malfunctioning car keys, cost overruns in building airports, and hundreds of
other unfortunate incidents. We also have many organizations that cannot fulfill
their missions because they are so poorly designed. In 2020, the world is facing a
historical calamity due to the spread of coronavirus (COVID-19), for which we
need to design enlightened and wise policies as well as medication (i.e., vaccine and
others) and medical treatments. All of these problems in so many diverse fields
require rational, logical, and thoughtful designs. These challenges require better
design education in our schools, colleges, universities, and even in industrial firms.
At the 2017 International Conference on Axiomatic Design (ICAD), which was
held in Gheorghe Asachi Technical University in Iaşi-România, the idea of
improving design education was discussed. A solution was to write a compre-
hensive textbook on design for students throughout the world. We thus formed the
Editorial Committee. We produced a tentative Table of Contents for the book, and
sample chapters were written as a general reference and to promote consistency of
style of presentation. Professor Miguel Cavique of Portugal assumed the chair-
manship of the Editorial Committee. He performed the difficult task of coordinating
and reviewing all the chapters through a confidential review process to be sure that
the quality of writing is on par with our goal of producing a world-class textbook.
The results of his effort are proof of his energy and leadership.
This book is a result of a labor of love and the spirit of cooperation. It reflects the
shared belief that design education can be done more effectively to improve the
quality of all designed things, ranging from organizations to products and pro-
cesses. Without the dedication of the authors, we would not have this book. The
chapters they produced were thoroughly reviewed and commented by the reviewers
and by the members of the Editorial Committee. The challenge of writing this book

xi
xii Acknowledgments

was higher than usual because authors represented diverse cultures, nationalities,
educational backgrounds, various languages, religions, and ethnicity. These dif-
ferent backgrounds enriched our deliberations and increased the zeal to produce a
genuinely world-class textbook. All the authors should be complimented for their
contributions, hard work, and dedication. They deserve much credit for their effort
to improve the quality of design education worldwide.
Many enabled the completion of this book project. Oliver Jackson of Springer
promptly accepted our offer to write this book; some of the students at the New
South Wales University read the early chapters of the book and offered their candid
evaluations. Professor Evelyn Wang, Head of the MIT Department of Mechanical
Engineering, rendered her support for this project. Galia Stoyanova of MIT was an
able Editorial Assistant throughout the editing process. Emily Welsh provided her
expertise with a few figures in the book.
The members of the Editorial Committee provided valuable inputs to the review
of the contributed chapters as well as in providing editorial guidance. The members
of the Editorial Committee were (in alphabetical order) as follows:
Gabriele Arcidiacono of Italy,
Christopher A. Brown of USA,
Miguel Cavique of Portugal,
Amro Farid of USA,
Joseph Timothy Foley of Iceland and USA,
Inas Khayal of USA,
Ang Liu of Australia,
Dominik Matt of Italy,
Masayuki Nakao of Japan,
Gyung Jin Park of Korea,
Erik Puik of Netherlands,
Laurenţiu Slătineanu of Romania, and
Nam Pyo Suh of USA.
In addition to the members of the Editorial Committee, a distinguished group of
professors and scholars reviewed the manuscript. They made specific suggestions to
improve the quality of writing, including the delivery of intellectual contents of
various topics. The task of the Chairman of the Editorial Committee, Professor
Miguel Cavique, was made so much easier because of the expert support and advice
he received from many reviewers, including the Editorial Committee members. The
rigorous reviews performed by the following professors and scholars have become
an integral part of this book:
António Gonçalves-Coelho,
Hilario Oh,
António Mourão,
Petra Foith-Förster,
Taesik Lee,
Erwin Rauch,
Acknowledgments xiii

João Fradinho,
Paolo Citti,
Gheorghe Nagîţ,
Oana Dodun,
Marianna Marchesi,
Alessandro Giorgetti,
Rajesh Jugulum,
Vladimir Modrak, and
Efrén Benavides.
The Chairman of the Editorial Committee is particularly grateful to Professor
António Gonçalves-Coelho for his guidance and consultation throughout his aca-
demic career.
All the chapters were reviewed—sometimes several times—to be sure that the
book can serve the need of the readers of the book. Mr. Oliver Jackson of Springer
had the book proposal reviewed based on a table of contents and a few sample
chapters. Three reviewers positively supported this book project, while one
reviewer did not feel that there is any reason to change the design education. The
Editorial Committee is grateful for their reviews.
Education has become global—the faculty and students at most leading uni-
versities of the world are multi-racial, representing many ethnicities and religions,
and are gender-neutral. Leading universities are no longer competing and cooper-
ating only with other institutions within the same country. They operate on a global
stage to spread knowledge and produce future leaders who can deal with the needs
and problems of the future, such as global warming. The young graduates of these
universities will be operating on a worldwide stage—no longer confined to any
national boundaries or narrow confines of disciplines—just as many other human
affairs have indeed become global. In the future, all inhabitants of Earth must share
knowledge for the common good of humanity and the mutual prosperity of all
nations. The era of one group of people living better at the expense of others,
especially those downtrodden as a consequence of historical accidents, should be
replaced by a more enlightened approach for the common good of all people. It is
the hope of the authors of this book that our effort will help in elevating the level of
knowledge of all people in our global village.
Finally, it is our great personal pleasure to acknowledge many decades of
support given to our research on AD and complexity by Dr. and Mrs. Byung-Jun
(BJ) Park. With their latest gift, our community has established the Axiomatic
Design Research Foundation (ADRF) to give awards to researchers who have made
outstanding research contributions to the field of Axiomatic Design and complexity.
They have been the lifelong friends of the Suh family and supporters of our vision
for better tomorrow at MIT, KAIST, and elsewhere. Young and Nam have been
most fortunate to have them as their dear friends.
Any significant tasks such as this undertaking would be impossible without the
support and sacrifice of the spouses of the Editorial Committee members. We thank
xiv Acknowledgments

them all. It is our special privilege to acknowledge the support and encouragement
of Mrs. Young Ja Surh, Mrs. Cristina Cavique, and Mrs. Markéta Foley. Without
their support and encouragement, we would not have completed this book.

Sudbury, Massachusetts, USA Nam Pyo Suh

Almada, Portugal Miguel Cavique
Reykjavík, Iceland Joseph Timothy Foley
March 2020
Contents

1 Introduction to Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 1
Nam Pyo Suh
1.1 Human Creativity and Design . . . . . . . . . . . . . . . . . . . . . . .. 3
1.2 Design: A Basic Human Intellectual Instinct . . . . . . . . . . . . .. 5
1.3 Importance of Knowing How to Define the Problem Based
on Design Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 11
1.3.1 Invention of the Steam Engine by James Watt
Changed the History of Humanity and Created
the Science of Thermodynamics . . . . . . . . . . . . . . .. 12
1.3.2 Design as a Common Human Activity in Many
Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 13
1.3.3 Two Different Solutions for the Same Problem . . . .. 16
1.4 Designing Without Explicit Goals and Problem Identification
is Analogous to Sailing a Sailboat Without a Rudder . . . . . .. 18
1.5 A Summary of the Creative Process . . . . . . . . . . . . . . . . . . .. 18
1.6 Importance of Design and Design Thinking . . . . . . . . . . . . .. 19
1.7 What is the Most Difficult Aspect of Learning Axiomatic
Design? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.8 List of Example Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.9 Definition of Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
1.10 Fundamental Principles of Design . . . . . . . . . . . . . . . . . . . . . 28
1.11 Principle of Similitude of Systems . . . . . . . . . . . . . . . . . . . . . 28
1.12 Importance of Knowing the Basic Laws and Principles
of Science and Engineering . . . . . . . . . . . . . . . . . . . . . . . . .. 29
1.13 The Inverse Problem in Design: Extraction of Functional
Requirements from Many Design Parameters . . . . . . . . . . . .. 29
1.14 Optimization of an Existing System Versus Design
of a New System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.15 Scope of the Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.16 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

xv
xvi Contents

2 What Is Design? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 35
Nam Pyo Suh
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 36
2.1.1 Why Design Theory (DT)? . . . . . . . . . . . . . . . . . . .. 36
2.1.2 Think in Solution-Neutral Environment! . . . . . . . . .. 39
2.1.3 Design Theory: Should It Be Axiomatic
or Algorithmic? . . . . . . . . . . . . . . . . . . . . . . . . . . .. 42
2.2 Design: Transformational Process to Create Human-Inspired
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 44
2.2.1 Definition of Key Words: Design, Goals,
and Domains . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 44
2.2.2 The Concept of Four Domains . . . . . . . . . . . . . . . .. 44
2.2.3 Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 45
2.2.4 Mapping from Domain to Domain: Functional
Requirements, Design Parameters, and Process
Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.2.5 Design Axioms . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.2.6 Design Matrix [DM] . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.2.7 Design Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
2.2.8 Decomposition of Functional Requirements, Design
Parameters, and Process Variables Through
Zigzagging Between the Domains . . . . . . . . . . . . . . . 52
2.3 Design Process in Creating Various Systems . . . . . . . . . . . . . 62
2.3.1 Revisiting the Water Faucet Design . . . . . . . . . . . . . . 62
2.3.2 Design of a System for Salary Raise . . . . . . . . . . . . . 64
2.4 Role of System Architect in Managing Large System
Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
2.5 Uncertainty and the Information Axiom . . . . . . . . . . . . . . . . . 66
2.6 On Design of Machines that Can Design . . . . . . . . . . . . . . . . 67
2.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3 How Do We Design? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 71
Nam Pyo Suh
3.1 Universal Nature of Design Tasks . . . . . . . . . . . . . . . . . . . . . 72
3.2 Some Useful and Simple Corollaries and Theorems . . . . . . . . 76
3.3 Lesson on “Design It Right, i.e., No Coupled Design!” . . . . . . 78
3.4 Mathematical Optimization and Design . . . . . . . . . . . . . . . . . 82
3.5 Design of Technological Products . . . . . . . . . . . . . . . . . . . . . 85
3.5.1 Design of Easy Open Beverage Can . . . . . . . . . . . . . 86
3.5.2 Removal of Kidney Stone with a Robot . . . . . . . . . . . 87
3.5.3 Ash Tray in Automobiles . . . . . . . . . . . . . . . . . . . . . 89
Contents xvii

3.5.4 Design of a Collapsible Steering Column

for Automobiles . . . . . . . . . . . . . . . . . . . . . . . . .... 91
3.5.5 Design of a Technology that Reduces Material
Consumption . . . . . . . . . . . . . . . . . . . . . . . . . . .... 94
3.5.6 Wireless Electric Power Transmission System
to a Bus in Motion . . . . . . . . . . . . . . . . . . . . . . . . . . 95
3.5.7 Design of Ultra-Precision Lithography Machine . . . . . 104
3.6 Design of Organizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
3.6.1 Design of a Government Organization . . . . . . . . . . . . 106
3.6.2 Design of High-Technology Industrial Firms . . . . . . . 111
3.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4 Design Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 117
Paolo Citti, Alessandro Giorgetti, Filippo Ceccanti, Fernando Rolli,
Petra Foith-Förster, and Christopher A. Brown
4.1 Introduction: Design Representations and Their Purposes . . .. 119
4.2 Design Representations for Applying the Axioms . . . . . . . . .. 121
4.2.1 Axiom 1, Maintaining Independence, Function–
Physical Representations, and Module Junction
Diagram and Flow Diagram . . . . . . . . . . . . . . . . . .. 121
4.2.2 Axiom 2, Minimizing Information Content
and the Concept of Tolerances . . . . . . . . . . . . . . . .. 124
4.2.3 V-Model, Design Process Representation to Guide
a Structured Design Process . . . . . . . . . . . . . . . . . .. 127
4.3 Examples of Design Representations in Industrial
Environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 129
4.3.1 Mechanical Product Design: From the Design
Matrix to the Mechanical Drawing . . . . . . . . . . . . .. 129
4.3.2 Industrial Environment: Piping and Instrumentation
Diagram (P&ID) . . . . . . . . . . . . . . . . . . . . . . . . . . .. 140
4.4 Case Study: Design and Manufacture of Mechanical and
Chemical Polishing Machine (CMP) by Four Graduate
Students . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.5 Software Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
4.5.1 Operational Context . . . . . . . . . . . . . . . . . . . . . . . . . 147
4.5.2 Software Representation—UML . . . . . . . . . . . . . . . . 148
4.5.3 Equivalence Between Forms of Representation . . . . . . 149
4.5.4 Example of a Software Representation . . . . . . . . . . . . 150
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
xviii Contents

5 Problem Definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 167

Ang Liu
5.1 Problem Definition in Engineering Design . . . . . . . . . . . . . .. 167
5.1.1 Characteristics of Problem Definition . . . . . . . . . . . .. 167
5.1.2 Good Problem Definition Leads to Design
Innovation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 169
5.2 Definition of Customer Need . . . . . . . . . . . . . . . . . . . . . . . .. 171
5.2.1 Who Are the Target Customers and Relevant
Stakeholders? . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 171
5.2.2 How to Solicit Customer Voices? . . . . . . . . . . . . . .. 173
5.2.3 Understand Customer Voices to Capture Innovation
Opportunity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 175
5.3 Definition of Functional Requirement . . . . . . . . . . . . . . . . . .. 175
5.3.1 How to Represent Functional Requirements? . . . . . .. 176
5.3.2 How to Extract Functional Requirements from
Existing “Things” . . . . . . . . . . . . . . . . . . . . . . . . . .. 177
5.3.3 How to Classify Functional Requirements into
Different Categories? . . . . . . . . . . . . . . . . . . . . . . . . 178
5.3.4 How to Structure Functional Requirements? . . . . . . . . 179
5.4 Definition of Design Constraint . . . . . . . . . . . . . . . . . . . . . . . 181
5.4.1 What Is Design Constraint? . . . . . . . . . . . . . . . . . . . . 181
5.4.2 Examples of Design Constraint . . . . . . . . . . . . . . . . . 183
5.5 Definition of Problem Context . . . . . . . . . . . . . . . . . . . . . . . . 184
5.5.1 What Is Design Context? . . . . . . . . . . . . . . . . . . . . . 184
5.5.2 Example of Design Context . . . . . . . . . . . . . . . . . . . 186
5.6 Examples of Problem Definition . . . . . . . . . . . . . . . . . . . . . . 186
5.7 Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 188
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
6 How Should We Select Functional Requirements? . . . . . . . . . . . .. 191
Nam Pyo Suh
6.1 Importance of “Solution-Neutral Environment” in Selecting
Functional Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . .. 192
6.2 Functional Requirements and Zigzagging between Domains
to Generate Lower Level Functional Requirements through
Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 199
6.3 Role of Broad Knowledge Base/Data Base in Formulating
Functional Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . .. 202
6.4 On Selection of a Right Set of Functional Requirements—
Need to Review Functional Requirements from Time
to Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 202
6.5 On Reverse Engineering to Determine the Functional
Requirements of Existing Products . . . . . . . . . . . . . . . . . . . .. 205
Contents xix

6.6 Minimum Number of Functional Requirements . . . . . . . . . ... 206

6.7 No Relative Ranking of the Importance of Functional
Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 206
6.8 Interdisciplinary Background and Choice of Functional
Requirements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
6.9 Importance of Design Matrix and the System Architecture . . . 207
6.10 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 207
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 208
7 How Should We Select Design Parameters? . . . . . . . . . . . . . . .... 209
Nam Pyo Suh
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
7.2 Criteria for Selection of Design Parameters . . . . . . . . . . . . . . 212
7.3 Ideal Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
7.4 Generation and Selection of Design Parameters . . . . . . . . . . . 218
7.5 Integration of Design Parameters in a System . . . . . . . . . . . . . 221
7.6 Case Studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
7.7 Use of Artificial Intelligence (AI) in Axiomatic Design . . . . . . 233
7.7.1 Thinking Design Machine (TDM) . . . . . . . . . . . . . . . 235
7.7.2 Physical Laws and Thinking Design Machine . . . . . . 236
7.8 Guidelines for Choosing Design Parameters and Process
Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 236
7.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 238
Appendix 7.1 Algorithm for Changing the Order
of {FRs} and {DPs} . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 239
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 240
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 240
8 How Should We Select Process Variables? . . . . . . . . . . . . ....... 241
Nam Pyo Suh
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ....... 242
8.2 Precision Engineering . . . . . . . . . . . . . . . . . . . . . . . ....... 245
8.3 Designing a Manufacturing Process for a Difficult-to-
Machine Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
8.4 Mass Production of Microcellular Plastics . . . . . . . . . . . . . . . 248
8.5 3D Printing: Layered Manufacturing . . . . . . . . . . . . . . . . . . . 251
8.6 Process Variables in Organizational Design . . . . . . . . . . . . . . 252
8.7 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
xx Contents

9 Mapping in Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 255

Miguel Cavique, António Gabriel-Santos, and António Mourão
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
9.2 The Role of Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
9.3 Hints for Defining the Functional Requirements . . . . . . . . . . . 259
9.4 The Decomposition Process . . . . . . . . . . . . . . . . . . . . . . . . . . 263
9.5 The Zig . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 268
9.6 The Zag . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274
9.7 Decomposition and Concurrent Engineering . . . . . . . . . . . . . . 276
9.8 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
10 Redundant Designs . . . . . . . . . . . . . . . . . . . . . ................ 285
Miguel Cavique and António Gonçalves-Coelho
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285
10.2 Redundancy in Nature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
10.3 The Theorem of Redundancy . . . . . . . . . . . . . . . . . . . . . . . . . 289
10.4 Reliability-Motivated Redundancy . . . . . . . . . . . . . . . . . . . . . 290
10.5 Functionally Related Redundancy . . . . . . . . . . . . . . . . . . . . . 295
10.6 More Theorems on Redundancy . . . . . . . . . . . . . . . . . . . . . . 298
10.7 Solving Coupled Designs . . . . . . . . . . . . . . . . . . . . . . . . . . . 302
10.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 304
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
11 The Information Axiom and Robust Design . . . . . . . . . . . . . ..... 307
João Fradinho and António Gonçalves-Coelho
11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..... 308
11.2 The Information Axiom . . . . . . . . . . . . . . . . . . . . . . . . ..... 311
11.3 Independence and Information Content . . . . . . . . . . . . . ..... 312
11.4 The Computation of Information Content of Decoupled
Designs Through Graphical Methods . . . . . . . . . . . . . . . . . . . 317
11.5 Information Content and Robustness . . . . . . . . . . . . . . . . . . . 323
11.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 324
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
12 Complexity in Axiomatic Design . . . . . . . . . . . . . . . . . . . . . . . . ... 327
Erik Christian Nicolaas Puik
12.1 Motivation for the Complexity Theory in Axiomatic
Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 327
12.1.1 Applying a Complexity Theory to Real Problems . ... 328
12.1.2 What is Complexity . . . . . . . . . . . . . . . . . . . . . . ... 328
12.2 Theory of Complexity, Periodicity, and the Design
Axioms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 329
Contents xxi

12.2.1 Time-Independent Complexity in Axiomatic

Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 331
12.2.2 Time-Dependent Complexity . . . . . . . . . . . . . . . . . .. 334
12.2.3 Limiting Behavioral Options by Banning
Irregularities from the Design . . . . . . . . . . . . . . . . .. 336
12.2.4 Known and Unknown Kinds of Time-Independent
Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 337
12.2.5 Overview of All Kinds of Time-Independent
Complexity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 339
12.3 Graphical Representation of Complexity . . . . . . . . . . . . . . . .. 339
12.3.1 The Functionality Diagram . . . . . . . . . . . . . . . . . . .. 339
12.3.2 Presumed and Legitimate Position in the
Functionality Diagram . . . . . . . . . . . . . . . . . . . . . .. 341
12.3.3 Ideal Development Path for Product Design . . . . . . .. 342
12.3.4 Examples of Typical Errors . . . . . . . . . . . . . . . . . . .. 342
12.3.5 Summary of the Application of Complexity
in Functionality Diagrams . . . . . . . . . . . . . . . . . . . .. 346
12.4 Conclusions of This Chapter . . . . . . . . . . . . . . . . . . . . . . . .. 347
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 348
13 Axiomatic Design Application to Product Family Design . . . . . . .. 349
Masayuki Nakao and Kenji Iino
13.1 Axiomatic Design Application for Product Family Design . . .. 349
13.1.1 Design Concept Description as the First Step
in Axiomatic Design . . . . . . . . . . . . . . . . . . . . . . . .. 349
13.1.2 Axiomatic Design Application with Proper
Functional Requirements . . . . . . . . . . . . . . . . . . . . .. 350
13.1.3 Axiomatic Design Application with Many
Functional Requirements . . . . . . . . . . . . . . . . . . . . .. 351
13.1.4 Creating New Design Using Axiomatic Design . . . .. 352
13.2 Product and Product Family Design Cases Using Axiomatic
Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
13.2.1 Automatic Driving . . . . . . . . . . . . . . . . . . . . . . . . . . 353
13.2.2 Fan Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
13.2.3 Entrance Exam Administration . . . . . . . . . . . . . . . . . 355
13.2.4 Umbrella that Follows the Owner . . . . . . . . . . . . . . . 358
13.2.5 Stirling Engine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 358
13.2.6 CurcurPlate for Managing Peoples in a Building . . . . 360
13.2.7 Tool for Brushing the Back of Teeth . . . . . . . . . . . . . 362
13.3 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
xxii Contents

14 Design of Large Engineering Systems . . . . . . . . . . . . . . . . . . . . ... 367

Gyung-Jin Park and Amro M. Farid
14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
14.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368
14.1.2 Chapter Contribution . . . . . . . . . . . . . . . . . . . . . . . . 371
14.1.3 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 371
14.2 Axiomatic Design of Large Fixed Engineering Systems . . . . . 371
14.2.1 What Are Large Fixed Engineering Systems? . . . . . . . 371
14.2.2 Divide and Conquer: Decomposition of System
Hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 373
14.2.3 Allocation of Function to form—the Zigzagging
Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 376
14.3 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 378
14.3.1 Axiomatic Design of the Mount Type Air
Conditioning System . . . . . . . . . . . . . . . . . . . . . . . . 379
14.3.2 Automobile Cooling System . . . . . . . . . . . . . . . . . . . 385
14.3.3 Automobile Suspension System . . . . . . . . . . . . . . . . . 390
14.3.4 Mobile Harbor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394
14.3.5 The Online Electrical Vehicle . . . . . . . . . . . . . . . . . . 398
14.4 Axiomatic Design of Large Flexible Engineering Systems . . . . 402
14.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 408
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 414
15 Complexity in the Kitchen . . . . . . . . . . . . . . . . . . . . . . . . ....... 417
Joseph T. Foley, Erik Puik, Lindy Puik, Joseph Smith,
and David S. Cochran
15.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 418
15.1.1 Axiomatic Design in Other Words . . . . . . . . . . . . . . . 419
15.1.2 Complexity Theory . . . . . . . . . . . . . . . . . . . . . . . . . . 419
15.1.3 Cooking Science . . . . . . . . . . . . . . . . . . . . . . . . . . . 420
15.2 Cooking Axioms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
15.3 Cooking a Turkey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 421
15.3.1 Why Does US Thanksgiving Mean Eating
a Turkey? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 422
15.3.2 Bringing the Bird to the Table . . . . . . . . . . . . . . . . . 423
15.3.3 How Do People Really Cook a Turkey? . . . . . . . . . . 428
15.4 Baking an Apple Pie . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 429
15.5 The Complexity of Reverse Engineering a Recipe . . . . . . . . . 432
15.5.1 Reproduction of Braised Lamb Ribs . . . . . . . . . . . . . 433
15.5.2 Reproduction of Scottish Haggis . . . . . . . . . . . . . . . . 434
15.5.3 Reproduction of Swiss Hay Soup . . . . . . . . . . . . . . . 436
15.6 Achieving Functional Requirements with Less Waste . . . . . . . 438
Contents xxiii

15.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 440

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 442
16 Design of Organizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 447
Nam Pyo Suh
16.1 Why Design an Organization? . . . . . . . . . . . . . . . . . . . . . . . . 448
16.2 How Should We Design an Organization? . . . . . . . . . . . . . . . 457
16.3 Operation of Organizations . . . . . . . . . . . . . . . . . . . . . . . . . . 467
16.4 Design and Decision-Making in an Organization . . . . . . . . . . 468
16.5 Typical Organizational Design . . . . . . . . . . . . . . . . . . . . . . . . 469
16.6 When Should an Organization Be Re-designed? . . . . . . . . . . . 470
16.7 Institutional Development: The S-Curve, S-Gap, and Vector
Delta (d) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 474
16.8 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 479
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 481
17 Application of Axiomatic Design for the Design of Flexible
and Agile Manufacturing Systems . . . . . . . . . . . . . . . . . . . . . .... 483
Dominik T. Matt and Erwin Rauch
17.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 484
17.2 Use of Axiomatic Design in Manufacturing System
Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 486
17.3 Case Study 1: Design of Flexible and Changeable
Manufacturing Systems . . . . . . . . . . . . . . . . . . . . . . . . . .... 487
17.4 Case Study 2: Design of a Smart Shopfloor Management
System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 493
17.5 Case Study 3: Design of Collaborative Human–Robot
Assembly Workplaces . . . . . . . . . . . . . . . . . . . . . . . . . . .... 500
17.6 Case Study 4: Design of a Learning Factory
for Industry 4.0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 504
17.7 Case Study 5: Design of a Demonstrator for a Flexible
and Decentralized Cyber-Physical Production
System (CPPS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 510
17.8 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 518
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 518
18 Design of the Assembly Systems for Airplane Structures . . . . . ... 521
Mustafa Yurdakul, Yusuf Tansel İç, and Osman Emre Celek
18.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 522
18.2 Case Study: Airplane Fuselage Panel Assembly with
Axiomatic Design Principles . . . . . . . . . . . . . . . . . . . . . . ... 527
18.3 Assembly System Design Alternatives Simulation Results . ... 531
xxiv Contents

18.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 536

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
19 Healthcare System Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 543
Inas S. Khayal and Amro M. Farid
19.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
19.1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
19.1.2 Chapter Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . 544
19.2 Defining a Healthcare System . . . . . . . . . . . . . . . . . . . . . . . . 545
19.2.1 Healthcare System Boundary, Scope, and Scale . . . . . 546
19.2.2 Healthcare System Function and System Form . . . . . . 548
19.2.3 Variations of Healthcare System Function and
System Form Across Cultures and Regions . . . . . . . . 553
19.3 Challenges Specific to the Design of Healthcare Systems . . . . 555
19.3.1 Healthcare is a Socio-Technical System . . . . . . . . . . . 555
19.3.2 Healthcare Delivery Heterogeneity . . . . . . . . . . . . . . . 557
19.3.3 Complexity and Interoperability . . . . . . . . . . . . . . . . 558
19.3.4 Healthcare as a Legacy System . . . . . . . . . . . . . . . . . 559
19.4 Key Takeaways and Conclusion . . . . . . . . . . . . . . . . . . . . . . 560
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 562
20 Functional Periodicity, “Function Clock,” and “Solar Time
Clock” in Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 563
Nam Pyo Suh
20.1 Introduction to Functional Periodicity and Re-initialization . . . 564
20.2 “Function Clock” Versus “Time Clock” . . . . . . . . . . . . . . . . . 569
20.3 Pros and Cons of “Function Clock” Versus “Time Clock” . . . 572
20.4 Periodicity and Re-initialization of Airline Scheduling . . . . . . 572
20.5 Periodicity and Re-initialization of Biological Systems: Cell
Division and Mitosis of Cells . . . . . . . . . . . . . . . . . . . . . . .. 574
20.6 Periodicity and Re-initialization in Manufacturing System
Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 575
20.6.1 Functional Periodicity in Design of a Simple
Manufacturing Operation . . . . . . . . . . . . . . . . . . . .. 575
20.6.2 Functional Periodicity in Job Shop Scheduling . . . . .. 576
20.7 Periodicity and Re-initialization of Political Systems . . . . . . .. 579
20.8 Functional Periodicity and Re-initialization of Educational
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 582
20.8.1 Academic Calendar . . . . . . . . . . . . . . . . . . . . . . . . . . 582
20.8.2 Research Universities . . . . . . . . . . . . . . . . . . . . . . . . 583
20.9 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
Further Reading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 584
Contents xxv

21 Artificial Intelligence in Design: A Look into the Future

of Axiomatic Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 585
Erwin Rauch and Dominik T. Matt
21.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 586
21.2 Artificial Intelligence—The Next Hype? . . . . . . . . . . . . . . . .. 588
21.3 Examples of Artificial Intelligence Applications
in Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 591
21.4 Axiomatic Design Knowledge Database as Basis
for Artificial Intelligence in Axiomatic Design . . . . . . . . . . .. 592
21.5 Vision of Combining Artificial Intelligence and Axiomatic
Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 594
21.6 Artificial Intelligence-Assisted Design of Complex Systems .. 595
21.7 Artificial Intelligence-Assisted Re-design of Complex
Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 597
21.8 Impact and Advantages of Artificial Intelligence in
Axiomatic Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 599
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 601
22 Axiomatic Cloud Computing Architectural Design . . . . . . . . . . . .. 605
John Thomas and Pam Mantri
22.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 606
22.2 History of Cloud Computing . . . . . . . . . . . . . . . . . . . . . . . . . 607
22.3 Socio-Technical Stigmergy . . . . . . . . . . . . . . . . . . . . . . . . . . 608
22.4 Complex Adaptive System (CAS) . . . . . . . . . . . . . . . . . . . . . 611
22.5 Knowledge Hierarchy/Heterarchy . . . . . . . . . . . . . . . . . . . . . . 612
22.6 Power Law Versus Gaussian Distribution . . . . . . . . . . . . . . . . 614
22.7 Axiomatic Trace . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 620
22.8 Cynefin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 624
22.9 Cloud OODA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 627
22.10 Iterative Axiomatic Maturity Diagram (AMD) Ensembles . . . . 635
22.11 Non-Functional Requirements from a Complex Adaptive
System Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 642
22.12 Security of Cloud Computing . . . . . . . . . . . . . . . . . . . . . . .. 643
22.13 Econo-Complex Adaptive System Strategy in the Cloud . . . .. 646
22.14 Weickian Versus Axiomatic Adaptive Coupling in Complex
Adaptive System Architectures . . . . . . . . . . . . . . . . . . . . . .. 650
22.15 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 653
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 654
23 Future Design Challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 659
Nam Pyo Suh
23.1 Introduction: Continuing Human “Saga” . . . . . . . . . . . . . . . . . 661
23.2 A Challenging Design Problem: Creating Forest in North
Africa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 663
xxvi Contents

23.3 Design Challenges Related to Renewable Energy . . . . . . . ... 666

23.4 Design Problems Related to Transportation . . . . . . . . . . . ... 668
23.5 Design Problems Related to CO2 and Methane Gas (CH4)
Emission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... 669
23.6 Future Roles of Artificial Intelligence (AI) and Quantum
Computing in Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 670
23.7 Design and Large Databases . . . . . . . . . . . . . . . . . . . . . . . . . 672
23.8 Design Issues Related to De-Salination . . . . . . . . . . . . . . . . . 672
23.9 Design of Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 673
23.10 Control of the Weather Pattern . . . . . . . . . . . . . . . . . . . . . . . 673
23.11 Design of a Better Educational System . . . . . . . . . . . . . . . . . . 674
23.12 Design of a Democratic and Transparent Government . . . . . . . 674
23.13 Design of Improved Health Delivery Systems . . . . . . . . . . . . . 675
23.14 Design of a Peaceful World . . . . . . . . . . . . . . . . . . . . . . . . . . 676
23.15 Design of Better Drugs for Brains . . . . . . . . . . . . . . . . . . . . . 676
23.16 Design of Computer Assisted Brains (I.E., Supplementary
Brain) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 676
23.17 Recycling of Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 678
23.18 Design Problems Related to Self-Driving Cars . . . . . . . . . . . . 678
23.19 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 679

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 681
Introduction to Design
1
Nam Pyo Suh

Abstract
The human ability to design and invent has changed the world in a relatively
short time—less than ten generations since the invention of the Watt steam
engine in the eighteenth century, which led to the Industrial Revolution. Since
then, the number of inventions and design artifacts has exploded to create the
current technology-based world and the overall much higher standard of living
for a population of about 7–8 billion in the twenty-first century that is about ten
times larger than that of the pre-Industrial Revolution!
In the late nineteenth century and the twentieth century, the design of internal
combustion engines created the automotive industry and the oil industry that
dominated the world’s economy for more than a century. The jet engines
introduced in the mid-twentieth century shrunk the world—people can travel
anywhere in the world in about 24 h. In the twenty-first century, semiconductors
and advances in telecommunications have revolutionized all aspects of human
lives. The advances in biotechnology are also improving the quality of life of
many and are likely to create a new era of human history. All this progress has
been possible because of the human ability to design, inspired by challenging
goals and promising ideas.
Our ability to design creatively is likely to determine humanity’s future.
There are many challenges. We need to grow the economy, producing goods and
services, to improve the quality of life, all while improving the environment. We
also have to combat various old and new diseases through the design of better
drugs and healthcare systems. The advances that have enriched and improved the
quality of life have also created new challenges for humanity in the form of
global warming and others, which must be addressed during the next few
decades. The animal husbandry that has satisfied the human need for protein

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 1

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_1
2 N. P. Suh

intakes of people is likely to be replaced by proteins derived from vegetables,

partly related to human attempt to solve global warming problems and health
issues. Solutions to all these issues depend on our ability to design.
What is design? Design refers to a set of creative activities to satisfy human
and societal needs and goals through synthesis. Design is motivated and driven
by three basic elements of human nature: inspiration, curiosity, and necessity to
improve the quality of life through creation of something new and better.
Sometimes, these three factors reinforce each other. Regardless of the motivating
factor that initiates the design thinking and activity, the first thing the designer
must do is to “establish goals” of the design task. Then, the designer must
identify the problem(s) that must be overcome to reach the design goals. Finally,
the designer must create useful and meaningful solutions to the problem
identified.
Design and creativity have a symbiotic relationship. The ability to design
generates creative solutions. Conversely, creative minds yield unique and
appropriate designs. Axiomatic Design theory enhances the probability of
strengthening this symbiotic relationship to create solutions that are better than
random trial-and-error processes. Axiomatic Design (AD) provides a theoretical
foundation for the creation of imaginative and optimum designs that satisfy the
desired functional requirements (FRs) that the design must fulfill to solve the
identified problem. The ability to state the FRs is a foundation for modern
technologies and various non-technological systems.
The design is a series of transformational processes. It could be a routine
process or an inspirational process of creating new ideas and solutions. Briefly, it
begins with “design goals.” In the customer domain, we define the “problem” we
must address to achieve the design goals. After the problem is established, we
create a set of FRs that we must satisfy to solve the problem identified. Then, we
transform the FRs to design parameters (DPs), followed by the transformation of
DPs to process variables (PVs). Depending on the nature of the system to be
developed, DPs and PVs take on specific physical, informational, biological, and
organizational entities.
The transformational process of going from problem definition to FRs, DPs,
and PVs follows the same methods regardless of the specific nature of the design
task. All of them must satisfy the Independence Axiom and the Information
Axiom. When we complete the design, the resulting embodiment is a collection
of DPs that constitute a system or product. We must check for physical
compatibility, information flow, biological functions, and safe operations,
among others, to assemble the final system.
To find an appropriate DP for a given FR or the right PV for the chosen DP,
we must depend on the fundamental laws, principles, and known phenomena of
science, engineering, and other relevant fields. The quality of design thinking
and ideas depends on the designer’s understanding of natural laws and principles
as well as human aspirations, needs, and limitations. In scientific and
technological fields, the relationship between FRs and DPs is often governed
by natural laws, which may require mathematical modeling to predict the
1 Introduction to Design 3

outcome. In general, the uncoupled design simplifies and enables the fulfillment
of design goals. We can make right design decisions based on our understanding
of design principles, natural laws, and the human needs.
Throughout the design process, we must seek inspirations to generate designs
that are most imaginative, appropriate, and creative for the task identified or
defined. Ultimately, through creative and rational designs, we advance the lot of
humanity and even the natural world.

1.1 Human Creativity and Design

“Through design, human beings satisfy their curiosity, create solutions to satisfy the
perceived needs, and fulfill human aspirations. Inspiration, logical reasoning, perspiration,
and persistence enhances this creative process. The design axioms guide the creative
thought process.”

What sets humans apart from other creatures is “creativity” and their ability to
conceive and create solutions through design, i.e., generation of solutions that have
not existed hitherto through synthesis. It has taken many centuries to cultivate this
human creativity. Until about 6,000 to 8,000 years ago, human beings lived in the
Stone Age. Eventually, the Stone Age ended, not because humans ran out of stones,
but they had found something better, i.e., iron. Thus, the new Iron Age was born.
However, the subsequent advances in technology and science were slow. They had
to wait many tens of centuries for science and modern technologies to emerge. In
the mid-seventeenth century, people like Isaac Newton began to lay the foundation
for science, and another 100 years later, inventors like James Watt started to create
new technologies that led to the Industrial Revolution. Then, a new era for
humanity has begun, accelerating the transformation of the world through the use of
human brains. What humans have created in basic knowledge and technologies
during the past three centuries is most impressive. Furthermore, the rate of new
advances and innovation has been accelerating ever since, which has challenged
many aspects of human and societal assumptions and practices.
“Design” is an intrinsic human activity, which enables people to develop solu-
tions to the problem that they need to solve through synthesis, analysis, and dis-
covery. Curiosity plays a major role in initiating and deriving these creative
activities. Curiosity has its roots in human thought processes that culminate in
raising the question: “WHY, WHY NOT, and WHAT IF?” In some fields that do
not require manufacturing, materials, and long-term testing to answer these ques-
tions, such as the information technology industry, the transition from curiosity to
implementation has occurred relatively quickly, as attested by rapid development of
IT industry, accelerating the pace of innovation.
The design used to be done and is still practiced in many organizations, based on
experience and trial-and-error processes, including extensive prototyping, building,
and testing. Many companies depend on their expertise and accumulated database
4 N. P. Suh

to design and manufacture their products. Good experience and know-how are
invaluable in these processes, often generating a reservoir of knowledge and “trade
secrets.” People have also designed and created public organizations such as uni-
versities and governments, often through trial-and-error processes. It is simply
amazing how much humanity has improved the quality of life and advanced
technologies through these trial-and-error processes of creating new products,
solutions, and, eventually, wealth. Humanity has generated many amazing products
and processes through these empirical approaches. The human ability to design and
create has culminated through such achievement as Apollo 11 that took human
beings to the moon in 1969.
Notwithstanding the amazing human achievements of the past three centuries,
depending only on experience and adopting the “design-build-test” process of
creating new products or systems have their limitations. Such a process is slow and
expensive because it requires extensive trial-and-error procedures, costly experi-
ments, and testing of prototypes. In some cases, these purely experience-based
approaches to design have resulted in unanticipated failures of the newly created
system that resulted in the loss of human lives and extensive cost overruns. Strictly
experience-based design approaches have led to failures due to the mistakes and
inappropriate decisions made, especially when the new set of requirements deviates
from the old ones. There are many well-known examples of failed designs due to
these “design-build-test” practices in product and system development. Famous
examples are airplanes that fell out of the sky, nuclear power plants that exploded,
newly constructed airports that have incurred significant cost overruns and delays,
and many costly failures of consumer products such as automobiles due to faulty
ignition switches.
Design based on fundamental principles of design should augment
experienced-based know-hows, which should assure the creation of successfully
designed systems and products that work the first time around without repetitive
redesigns and incremental changes to overcome design errors. Rationally designed
products and systems based on design principles also simplify analysis and testing,
reducing the time and cost associated with innovations of new technologies,
products, and systems of all kinds, including organizations, software, and hardware.
Some people believe that design is an “experiential subject,” which cannot be
taught well at universities. They claimed that design could be learned only through
experience. This “experiential school of design” has dominated the design field
until fairly recently. As a result, many people engaged in the creation of new design
solutions, including engineers, depended on repeated trial-and-error processes and
their experience in creating new systems and products. They use the repetitious
cycle of “design-build-test” and “redesign-build-test” in developing new products
and systems, including organizations. This experience-based design practice is
equivalent to attempting to improve the efficiency of jet engines without knowing
the fundamental laws of thermodynamics. Similarly, design cannot be done
rationally, minimizing mistakes, in the absence of fundamental design principles.
1 Introduction to Design 5

Axiomatic Design (AD) was advanced about 40 years ago to overcome and
eliminate the costly trial-and-error processes of design. The goal was to enable us to
make the right design decisions and avoid making mistakes and wrong solutions in
the development of new products and systems. Since the advent of AD, many
innovative products, processes, and organizations have been created quickly and
reliably. Many people have become much more creative after learning AD. These
results are not surprising, because the design axioms are distillations of common
features found in rational designs. The axioms were discovered through the
examination of past design decisions that had generated good designs and often,
highly creative products. The purpose of this book is to teach AD to students in all
fields of intellectual endeavor, who are interested in the synthesis of innovative
systems and products.
To learn AD, students must acquire the “language of AD,” i.e., definitions of
keywords, axioms, mapping, domains, the Independence Axiom, and the Infor-
mation Axiom. They are not difficult concepts to understand but must be able to
follow the logical reasoning used in AD.

1.2 Design: A Basic Human Intellectual Instinct

The need to design exists in most fields of human endeavor. We encounter “design”
in many different contexts and situations. Sometimes it is driven by human curiosity
and by problems that must be solved. Design principles discussed in this book help
in both of these situations. Curiosity arises when “something” violates the design
principles. These principles also guide the process of creating solutions to the
perceived problem.
The design is done to achieve a set of goals. In other words, without a clear set of
goals, we cannot commence design. Once we define the goal, we can identify the
problem that must be solved to achieve the goal. Based on the identified problem,
the designer can establish FRs that must be satisfied to solve the problem. Then, the
designer must look for design ideas and the corresponding DPs that can satisfy the
FRs of the design. It is a creative process that could be lots of fun because the
process of conceiving something new that no one else thought about in the past is a
challenging and exciting endeavor. There can be many equally acceptable design
solutions, but often, there is a superior design. The joy that follows when we come
up with creative ideas can be intoxicating. Furthermore, the same design thinking
that leads to innovative design applies to many different problems in many diverse
fields!
This book presents the process and the principles that will lead to the creation of
designs after we identify the problem—creatively, effectively, and efficiently—
regardless of the specific field of application. AD leads to rational and useful design
solutions in all areas of synthesis. By being able to identify weak or bad designs
early, it prevents the creation of a design that is not acceptable, thus improving the
6 N. P. Suh

efficiency of the design process. In practical terms, it means that a person who is
good at dealing with design issues can be useful in many other fields where syn-
thesis is essential.
The following are real design stories that illustrate how important it is to
establish the goals and identify the problem that must be overcome to achieve the
goal, leading to new design solutions:

Design Story 1.1:

An aspiring engineering student had to earn enough money to pay for his living
expenses while attending a university in the United States because his parents,
being recent immigrants to the U.S., could not support him financially. To support
himself, he had to hold a series of odd jobs at the university. His first job was to
work as a janitor, later moved up to be a telephone operator, lab assistant, and
library assistant, working roughly 25 h a week during the academic year to pay for
his room and food. He received the legal minimum wage for his work at the
university. After his third year at the college, he was most fortunate to get a job in a
small industrial firm near his university that manufactured disposal plastic products
such as cups and dishes. The best part of the job was the pay! The hourly wage at
the company was 120% more than the compensation for student assistants at the
university. He felt rich!
Moreover, he was in charge of his project with flexible working hours so that he
could attend his classes at school on Mondays, Wednesdays, and Fridays, while
working at the company the rest of the week, including Saturdays. He was the only
“engineer” in the small company, although he had not yet completed his under-
graduate studies. He worked with a dozen or so skilled machinists and technicians,
who taught him a lot of practical skills and know-how.
One day, the president of the company asked him to design a new product that
could replace a competitor’s product that had been used in vending machines for
dispensing hot coffee. The young engineer then learned, for the first time, about the
problem his boss identified, that had to be solved. The problem was the following:
The company decided to replace paper cups used in vending machines with a
plastic cup. However, the cup made of thin polystyrene sheet by vacuum-forming
could not replace the paper cup, because the cup made of the plastic sheet was too
hot to hold with a bare hand when hot coffee is poured into it. Furthermore, the cup
did not have enough rigidity to be held by a hand when it contained hot coffee. The
alternative was to use a thick foamed plastic sheet (sometimes called Styrofoam that
had high thermal insulation) to make the cup, but it was not acceptable, because the
wall thickness of the cup made of foamed polystyrene was too thick, limiting the
number of cups that can be stacked in the vending machine. The president of the
company assigned him to develop a new solution. It was an exciting challenge for
the young engineer, who had not yet graduated from college.
The young engineer’s solution was to satisfy two FRs, i.e., stiffness and thermal
insulation, by laminating foamed plastic sheet with un-foamed straight polystyrene
sheet to provide both the required thermal insulation and rigidity. In addition to
1 Introduction to Design 7

creating the product, he also designed and built a continuous manufacturing system
for mass production of the laminated plastic products. It took about 2 years to finish
the project and go into mass production. The product was a major commercial
success. The company did very well with the new invention, making many prod-
ucts. The U.S. Patent Office granted a patent for the product and processes. He
received a bonus that was equal to his weekly pay, which was appreciated but did
not make him rich!
Many years later after he became a professor at the same university where he
was a student, he and his family visited one of the manufacturing plants of the
company and found that the same machine he designed and built as an under-
graduate student, a la a development engineer, was still being used to make the
product, except that there were many more of them humming at the same time! He
was happy to show what he did many years ago to his wife and daughters.

Design Story 1.2:

As the above design story unfolded, across the town, a recent graduate of the same
engineering school got a job at a major teaching hospital. While working there, he
had this inspiration that the hospital could use a computerized information man-
agement system. He realized that the hospital, which was well known for its ad-
vanced medical care system throughout the world, was inefficient in its operation
because the hospital lacked an integrated information system because all the
information was written by hand and stored in file cabinets by people. Such
information could not be shared among different departments of the hospital
without human intervention.
He designed and implemented a central software system for electronic record
keeping as well as providing diagnostic assistance to medical doctors based on the
data collected from various tests done on a patient. He established a new company
in 1968 to make a software system for automation and management of hospitals and
healthcare. His company might have been one of oldest software systems company
in the world. (Note: Bill Gates started Microsoft in 1975.)
Initially, the company struggled but eventually received funding from a local
venture capitalist. Now the company is the leading software company in the
healthcare industry in the United States. The company did superbly well, making
him and his wife one of the wealthiest couples in the country. He and his wife have
become generous philanthropists, supporting many worthwhile causes at univer-
sities; hospitals; and various educational, civic, and cultural organizations in the
United States and other countries. Many of the recipients of their largess were glad
that he had that inspiration and started a new industry.

Design Story 1.3:

Another young engineer established his company that tested imported merchandise
for big department stores in the United States for quality assurance. He had learned
from one of his classmates that large department stores had a problem controlling
the quality of merchandise they were importing from overseas. They needed the
8 N. P. Suh

confidence that the products they were about to buy in large quantities from a
vendor in other countries were of good quality. His company tested various mer-
chandise (ranging from clothing, furniture to even medicine) at the request of
department stores.
The manufacturer of the merchandise wishing to sell its product to a department
store in the United States paid this testing company to test and certify the quality of
their products. When his company approved the quality of the product, the
department store purchased the merchandise directly from the manufacturer. It was
a perfect arrangement for this testing company, i.e., testing done at the request of
large department stores but paid for by the manufacturers of various merchandise
that is trying to sell their products to the department stores.
His company designed and invented many new testing methods and machines.
They also established the standards for an acceptable quality of various products,
creating an extensive database. The company became a reputable and highly suc-
cessful merchandise testing company in the world. One of the reasons for success of
this company was the reputation and personality of the founder. He always had
impeccable reputation for hard work, honesty, and the highest ethical standards.
Later, he and his wife sold the company to a large company at a high premium.
With their immense wealth, they became philanthropists, supporting many worth-
while causes at universities, hospitals, and needy students in many countries.

Design Story 1.4:

There are other interesting stories related to design. Recently, students at a

well-known university organized a team to create a solar-energy-powered electric
car to solve the problem of global warming caused by CO2 emission. They came up
with their unique design for the solar-powered electric car, learning about many
issues related to developing such new products. The students would have done a
better job if they had known more about fundamental aspects of the design of such
vehicles and solar power. If they had defined the FRs of such a car first (i.e., in a
solution-neutral environment) without letting the pre-conceived notion of what such
a vehicle should be like, they would have created a better vehicle. They could have
learned the lesson from those who designed the “Sunraycer” that won the
1,867-mile Pentax World Solar Challenge (Wilson et al. 1989). The Sunraycer team
defined FRs first without any specific design in mind, whereas other competitors
had decided on the physical configuration first without clearly defining the FRs of
such a vehicle and then tried to optimize the design afterward. Doing so, they
encountered too many problems due to the coupling of FRs.

Design Story 1.5:

A young professor at a leading university presented a seminar on how to manu-

facture thin, single-crystal III–V semiconductors without defects. The idea he
offered for making thin semiconductors was creative and smart. It had the potential
of manufacturing thin-film semiconductors that can be used to make devices.
1 Introduction to Design 9

We have known for a long time that if we deposit a new layer of a known
crystalline material (e.g., III–V semiconductor compound) on a crystalline substrate
by vapor deposition, the newly deposited material assumes the crystal structure of
the substrate. After the deposition of several layers of the crystalline material, if we
could separate the newly deposited crystal from the substrate, it could be used to
make a memory or logic device. The problem was that it was difficult to peel off the
newly deposited semiconductor layer from the substrate.
The young professor’s idea, which he developed while he was working at IBM,
was to satisfy two FRs independently from each other, i.e., the FR of growing
thin-film semiconductors and the FR of separating the thin-film semiconductor from
the substrate. To satisfy the second FR of separating a newly deposited semicon-
ductor layer, an intermediate atomic layer of graphene was deposited first on the
solid semiconductor substrate. The thin graphene sheet is a two-dimensional
material and does not bond to any material perpendicular to its surface because the
interatomic force of graphene is planar. Then, if the semiconductor material is
deposited on top of the graphene by vapor deposition, the newly deposited semi-
conductor would then assume the structure of the substrate crystal below the gra-
phene sheet, but not bond to the graphene. Hence, when the semiconductor material
is deposited on top of the graphene layer, the crystalline structure of the newly
deposited semiconductor material on the top of the graphene would be the same as
the original substrate below the graphene layer. Then, the freshly deposited crystal
on top of the graphene sheet can be peeled off to make thin semiconductor devices.
If the design can produce atomic scale-thin semiconductors on a mass production
basis, it may open up a new chapter in mass production of semiconductor devices.

Design Story 1.6:

Famous researchers and professors got together at a research institute, which was
established by a generous gift given by a leading industrialist and an alumnus of the
university, to discuss the progress made in the field of brain science and technology
related to autism. They were bright people with impressive credentials. Young
researchers and their professors made presentations in front of these experts
assembled from many different regions of the world. They presented the results of
various measurements, correlations, hypothesis, and models. They got into heated
discussions to clarify multiple concepts presented. What was clear from the pre-
sentations and subsequent discussions was that they could not define the cause or
the problem that lead to autism. Until they can identify the problem, their research
is likely to take longer to find a cure for autism. Their research, which is in their
early stages, needs to be conducted to determine the problem. They should perhaps
do more research based on hypothesis to narrow down the probable cause of autism.
To do this type of hypothesis-based research, they will probably need to adopt
“design thinking” to make more rapid progress.
10 N. P. Suh

Design Story 1.7:

In 2019, one of the most challenging issues in the world is the “cyberattack” by
hostile governments to disable computer networks or steal the information stored in
the computer systems of other nations. Many countries are engaged in such hostile
attacks, hoping to tilt the public opinion. They either try to steal or compromise the
information stored in the computer system of the targeted institutions. Certainly, the
attempt to influence the U.S. election has dislocated the U.S. political system in
2016. Corporations and institutions are spending a vast sum of money to protect their
information system from those who are attempting to steal the secret information and
strategic plans stored in their computer systems. In some cases, the goal of the
attacker is to destroy or disable specific computers to render the entire information
system malfunction and thus destroy the infrastructure of their competitors.
This kind of cyberattack is of primary concern for governments, corporations,
and even individuals. The problem is how to deal with cyberattacks and safeguard
the information systems from these intruders. The attackers may use email systems
and the Internet system to penetrate the information system of the organizations
they wish to compromise. One of the major design issues in cybersecurity is the
continuing escalation of attacking strategy to avoid newly installed defense sys-
tems, which require a response system that periodically updates and adjust the
defensive strategy. One design approach might be to attack the source of the
cyberintruders at their base.
This problem can be solved only through the superior design of the software
system that can identify the attacking system, protect its information system,
mislead the attacking system to self-destruct, and, at the same time, disable the
attacking computer system.

Design Story 1.8:

One of the most critical issues of the twenty-first century is global warming. If it is
unchecked, the temperature of the earth atmosphere will rise, creating many
unacceptable calamities such as flooding, the rise of sea level, creation of desert,
and many others. One of the major causes is the anthropogenic emission of carbon
dioxide (CO2) by automobiles and electric power plants. A widely accepted goal is
to limit the temperature rise to below 2 degrees Celsius relative to the temperature
of the atmosphere before industrialization. Professor William D. Nordhaus of Yale
University received the Nobel prize in economics for his work on how to reduce
carbon emission. His idea was to introduce a carbon tax to limit CO2 emission. In
2016, 196 nations signed the United Nations Framework Convention on Climate
Change (UNFCCC). Its implementation has not been easy, some countries refusing
to join in, citing that its negative impact on the economy and the high cost of
replacing fossil fuels with solar and wind energy.
It is a classic design problem. There are many FRs we have to satisfy to solve
this global warming problem. We need to identify all the FRs we have to satisfy and
the corresponding DPs. The carbon tax, although a creative idea, cannot solve the
1 Introduction to Design 11

global warming problem by itself. As we will demonstrate in later chapters, when

there are more FRs to be satisfied than the number of DPs (in this case, the carbon
tax), the design is not acceptable, i.e., the problem cannot be solved. We will show
that the number of DPs must be equal to the number of FRs that must be satisfied to
solve the global warming problem. Thus, the carbon tax alone will not bring about
an acceptable solution to global warming! We have to identify all the FRs that must
be satisfied and then develop an equal number of DPs, which will satisfy the FRs
without coupling them to each other.
Question 1.1:
“What is the common element in all these stories? How are they related to the main subject
of this chapter, i.e., why design?”.

1.3 Importance of Knowing How to Define the Problem

Based on Design Goals

Every design task, regardless of the specific field of application, has a set of goals.
In the case of the environment, the goal may be the reduction of greenhouse gases,
whereas, for the head of a university, the goal may be making the university one of
the best in the world. In the case of product development, the goal may be to make
the most efficient electric vehicle. The goal of the government may be to provide
universal healthcare to all citizens.
Once the goal is determined, the designer must ask, “What is the PROBLEM(s)
that has to be solved to achieve the goal?” Once the problem is defined, we can
proceed with the design task by going through the following four-step transfor-
mational processes:

(a) understanding the problem that needs to be solved in the customer domain;
(b) defining functional requirements {FRs} that must be satisfied to solve the
problem identified;
(c) identifying design parameters {DPs} to satisfy the {FRs};
(d) selecting process variables {PVs} that can satisfy the {DPs}.

This transformational process occurs over the four domains of the design world: the
customer domain, the functional domain, the physical domain, and the process
domain. The relationship between the domain on the left and the domain on the
right is “What” versus “How.” In other words, {FRs} with respect to {DPs} rep-
resent “what we want to achieve,” whereas {DPs} are “how we are going to satisfy
{FRs}.” Figure 1.1 illustrates this transformational process.
12 N. P. Suh

Fig. 1.1 Design is done to achieve the goals (i.e., the customer attributes, CA) stated in the
customer domain. Typically, the PROBLEM that is preventing the achievement of the goal must
be identified and solved. We state the goal in the form of functional requirements {FRs} in the
functional domain that will solve the problem. Then we go through a series of transformational
processes, i.e., from {FRs} to design parameters {DPs} in the physical domain, from the design
parameters, {DPs}, in the physical domain to process variables {PVs} in the process domain. The
terms DP and PV can be interpreted in different ways, depending on what we are designing, e.g.,
an organization rather than a machine

The importance of identification of the problem cannot be over-emphasized as

the following story about the invention of the Watt steam engine illustrates: James
Watt, who transformed the history of humankind, began his quest for mechanical
power generation with the use of thermal energy by discovering the problem with
then-existing Newcomen engine!

1.3.1 Invention of the Steam Engine by James Watt Changed

the History of Humanity and Created the Science
of Thermodynamics

From a historical point of view, the most critical invention made by human beings
is the steam engine by James Watt. His invention was not only responsible for the
Industrial Revolution in the eighteenth century but also chartered a new path for
human history by awakening the human brainpower for scientific and technological
thinking. The Watt engine freed humankind from hard physical labor and changed
the trajectory of civilization through a series of scientific and technological
advances that followed. Now 300 years after the invention of the Watt engine,
science and technology are taking humanity to a hitherto unimaginable path that
depends on science and technology, the outcome of which we cannot even fathom
with any degree of certainty. The only thing we know for sure is that in the future,
humankind will create new problems and new solutions that will accelerate the pace
of change. The design will be central in this transformation of human history, as
people will continue to use their creativity to design artifacts that will take humans
1 Introduction to Design 13

to a new plateau of not only technological progress but also better understanding of
humanity itself and societal fabric.
History attributed the beginning of the Industrial Revolution of the world to the
invention of Steam Engine by James Watt in 1736. In some ways, James Watt was
fortunate to be at the right place at the right time, in addition to being gifted with
brainpower and persistence to solve a problem that led to his invention of the Watt
steam engine. His genius was his ability to identify the problem (i.e., shortcomings)
associated with the Newcomen engine (invented in 1712), which led to the
invention of the Watt steam engine (1763–1775). At the time, the Newcomen
engine was primarily used to pump water out of mine shafts.
History states that James Watt was asked to repair the Newcomen engine that
belonged to the University of Glasgow. While repairing the machine, he found the
major shortcoming with the Newcomen engine (see Fig. 1.2), i.e., its intermittent
motion. As shown in Fig. 1.2, steam is injected into the cylinder of the Newcomen
engine. Then, the valve from the boiler to the cylinder is closed, and cold water was
injected into the cylinder to lower the temperature and condense the steam in the
cylinder. The vacuum created in the cylinder due to condensation of the steam
pulled the piston down. This downward motion of the piston, which was connected
to the water pump of the mine, sucked the water from mine shafts. James Watt
noticed that since one cylinder was used for both expansion and condensation of the
steam, the Newcomen engine was slow and not efficient. This discovery of the
problem by James Watt led to the invention of the Watt engine. His solution was to
separate the two functions, i.e., expansion and condensation by adding a separate
cylinder for condensation. This solution is consistent with the Independence Axiom
of AD, which is a formalization and generalization of what James Watt did,
although the origin of AD was not based on this observation of the Watt engine.

1.3.2 Design as a Common Human Activity in Many Fields

“Design is a universal human activity to satisfy human aspirations.”

In 2006, researchers at KAIST (The Korea Advanced Institute of Science and

Technology) identified the elimination of CO2 emission from internal combustion
(IC) engines as the central problem they must solve in order to alleviate the global
warming problem. Electrification of ground transportation systems should eliminate
about 30% of all anthropogenic CO2 emission of the world by using electricity
generated at more efficient electric power plants as well as using renewable energy
sources such as wind, solar, and hydropower.
This goal of removing of anthropogenic emission of CO2 has resulted in the
invention of new kinds of electric buses, cars, and trains that receive electric power
wirelessly from the underground power supply system to propel the vehicle. This
technology is called the “On-line Electric Vehicle (OLEV).” The OLEV bus carries
a small battery on board for operations of the vehicle on roads without the
underground power supply system. The driver of the OLEV bus drives the vehicle
14 N. P. Suh

Fig. 1.2 Schematic drawing of the Newcomen engine. The sequence of operation: Hot steam is
injected into the cylinder, followed by the closing of the valve. The piston of the cylinder then
moves upward due to the internal pressure created by the steam. This piston is connected to the
water pump of the mineshaft. When the steam in the cylinder condenses by the cold water injected
into the cylinder, a vacuum was created in the cylinder, pulling down the piston. The downward
motion of the piston pulled the piston of the water pump upward, pumping water out from the
mineshaft. (Reproduced from Black and Davis 1913)

without ever worrying about recharging the battery because it is done automatically
when the bus is on the top of the road with underground power supply system.
Figure 1.3 shows a bus commercially operating in Gumi City in Korea.
In addition to the design issues related to technology, similar design needs exist
in other fields such as organizations, economics, finance, public policy, and liter-
ature. For instance, a senior professor at a leading university got a telephone call
from overseas, asking him to lead a university to the next level of academic
excellence. He accepted the presidency of the university and redesigned the uni-
versity through strategic planning with the support of some of the faculty members,
eventually making it one of the best universities in the world. What the new
president did first was to identify the problems the university had to solve through
institutional transformation. After reaffirming the new goal of the university, he and
1 Introduction to Design 15

Fig. 1.3 OLEV bus in Gumi City in Korea

his team redesigned the university, including organizational structure, operational

policies, financial structure, and personnel policies. They used the principles of AD.
The university has indeed emerged one of the best innovative universities of the
world.
The design is performed in many fields, although in some fields they may not
call it “design” and use other synonymous phrases. It is interesting to note that
recently, graduates of liberal arts colleges in many countries are eager to become
writers of fictions, inspired by the work of Kazuo Ishiguro, the 2017 Nobel laureate
who is the author of “The Remains of the Day.” To write such a novel, the author
had to design the book first before actually commencing the writing of the book.
Although the specific nature of their tasks appears to be vastly different, all of the
people mentioned in this chapter have performed similar tasks! The first thing all of
them had to do to achieve their goal was to identify the problem to be solved and
then develop design solutions for the problem identified through a transformational
process illustrated in Fig. 1.1. The engineers and scientists design their products,
processes, and systems. The writers of fictions also design their book, i.e., identify
the “problem (i.e., the theme of their story),” design the structure of the story to be
written, and then write the narratives of the book. University administrators must
also do similar things: design the goals, strategy, and policies, and organize the
university for the execution of the plan.
Question 1.2:
Through DESIGN, humans have created modern society; fulfilled their aspirations,
curiosity, technological and cultural advances; and extended human life. Unfortunately,
people have also conceived the means of harming nature and human civilization through
design. How do we design all of these different things? Are there common elements in all
these designs?
16 N. P. Suh

In this introductory section of this chapter, we emphasized, repeatedly, the need to

discover or identify the “problem,” either before or after the goal of the design is
established. Then the designer translates or transforms the problem into a set of
specific objectives or goals the design must satisfy (or achieve). The goals are then
transformed into specific FRs that must be satisfied through design in order to
achieve the stated goals. Then, we have to search for specific means of satisfying
the FRs, which is defined as DPs in AD. All designs must go through these steps.
Fortunately, all designs, regardless of the field or the subject matter, involve similar
thought processes and the same basic principles a good design must satisfy. In
summary, to be useful in design, we must go through the actual process of
designing and understanding the design process, similar to learning how to ride a
bicycle, i.e., one has to try it, even at the risk of falling!
The process of the design described above applies in all fields that require design
solutions. It is not a difficult task. Everyone can do it! In design, the experience can
either be helpful or become a hindrance if the experience consists of purely random
trial-and-error processes.

1.3.3 Two Different Solutions for the Same Problem

Space travel fascinates many people of all ages, both young and old. The idea of
people going to the Moon was a big challenge in the 1960s. When Apollo 11 of the
United States landed on the Moon for the first time, on July 20, 1969, it was a
momentous and aspiring moment for all humankind. It gave confidence that human
beings can claim space as part of the human habitat. It has undoubtedly inspired
many young people all over the world to be interested in space travel and science.
In the early 1980s, the U.S. National Aeronautics and Space Administration
(NASA) established the Space Shuttle Program in order to build a space trans-
portation system primarily between Earth and the International Space Station.
Figure 1.4 shows a space shuttle taking off the launch pad. It shows the orbiter
vehicle (OV) that looks like an airplane attached to the external tank (ET), which
carries oxygen and hydrogen in two spherical tanks inside the external tank. Two
solid-fuel booster rockets are attached to the external tank on its two sides, which
assist during the liftoff. The external tank, which is made of aluminum, has about
half-inch thick polyurethane foam layer on its surface as thermal insulation in order
to prevent the formation of ice on the cold surface of the external tank while waiting
for launch. If ice forms, it can fracture during the ascent, which may hit the OV,
damaging the ceramic tile on the OV that protects the vehicle from burning up
during its re-entry into the earth’s atmosphere. The surface area of the external tank
is about the same as two football fields. The students, who visited the
1 Introduction to Design 17

Fig. 1.4 The space shuttle

discovery and its
seven-member STS-120 crew
head toward earth’s orbit and
a scheduled linkup with the
International Space Station.
(Reproduced with permission
from Petrushenka 2019)

manufacturing facility in Louisiana, U.S.A., where the external tank was being
built, were duly impressed by the size of the external tank.1
The problem with the Space Shuttle Program was its enormous cost. In com-
parison to the cost of launching communication satellites using a rocket, the cost of
launching it using the space shuttle was much more expensive. NASA justified the
cost because the space shuttle was needed to service the International Space Station.
In order to reduce the cost, it was decided to reuse the solid booster rockets. They
jettisoned the rockets mounted on the external tank fall into the ocean upon com-
pletion of their mission during ascent. Then they have recovered the jettisoned
rocket for future use.
Another solution to the reuse of the solid rocket is to let it descent along a
pre-programmed trajectory back to the launch pad, which has been demonstrated in
2017 by Space X, a company founded by Elon Mask. Conceptually, this technology
appears to be more elegant because it eliminates the need to search the ocean for the

1
As part of the contract between a NASA contractor and MIT, one of the students under the
supervision of a faculty member developed a method of putting on thermal insulation on the
external tank, which significantly reduced the cost of manufacturing.
18 N. P. Suh

recovery of used rockets. We have two different designs to solve similar highest
level FRs.
The FRs of the Space Shuttle Program involved many additional functions than
the Space X because it had to serve the International Space Station and launching of
satellites. However, the problem of recovering used rockets is similar at the highest
level of the design hierarchy. The lower level of FRs and DPs was different,
yielding two different solutions.

1.4 Designing Without Explicit Goals and Problem

Identification is Analogous to Sailing a Sailboat
Without a Rudder

In a major global company that makes steering systems for automobiles, the
management assembled their technical and marketing leaders to review their
strategic direction for the development of their next generation of new products.
Their goal was to be more competitive in the market place, especially in light of the
competition coming from Chinese companies that were offering similar products at
a lower price. The people gathered in the conference room of the company were all
experienced and bright people. Various ideas and solutions were proposed and
discussed.
Their products may be classified into the following four different kinds of sys-
tems: purely mechanical systems, electro-mechanical systems, all electrical sys-
tems, and hydraulic systems. The price ranges from a few hundred dollars per unit
to a few thousand dollars. After 3 h of intense discussion, the meeting was con-
cluded without developing any new major ideas and decisions. They decided to
have another meeting in about a month. Unfortunately, these unsuccessful meetings
are often a common occurrence in many companies. It is highly probable that the
meeting failed to produce any concrete ideas because they concentrated on lowering
the manufacturing cost of their products rather than reviewing their design after
clearly defining the PROBLEM.
The situation discussed above is similar to trying to steer a sailboat without a
rudder. The sailor will not go very far, or even worse, may not turn to the home
base!

1.5 A Summary of the Creative Process

The first step in developing innovative design typically requires the establishment
of the need (or the goal) and the identification of the problem that must be over-
come to achieve the goal.
1 Introduction to Design 19

Based on the identification of the goal and the problem, we establish specific
FRs that our design must satisfy to meet the goal. The next step in design is to come
up with DPs that will enable us to satisfy the FRs. Then, we configure the designed
system through the integration of various DPs as an integrated system. The inte-
grated system may consist of hardware, software, information systems, natural
elements, and others such as sensors. In some ways, it is an obvious way of coming
up with good designs.
Most people confront many problems, large or small throughout their life. They
have to solve them through design and perhaps by other means as well. The
difficulty is that often they are trying to find solutions without first defining the
problem. They may come up with various and contradictory claims and counter-
claims, aggravating each other. If we can define “what the problem is,” it may be
easier to develop solutions to the problem identified. The ability to define the
problem can be acquired by accumulating broad knowledge base, experience, and
design thinking. One of the goals of this design book is to teach those who are not
yet initiated into the field of design the ability to identify and define the problem
and FRs.
Once the FRs are defined, most people can synthesize their unique solutions,
unless they can find a quick solution from their “library” of past solutions that are
similar to the current problem. When there are no obvious past examples that can be
adapted, people should seek a new solution through “design” with confidence
without being intimidated. The decision to either adopt an old existing solution or
create a new “solution” depends on one’s knowledge base or experience or the
confidence that one can create something new from scratch that will be superior to
the existing system. Many people instinctively want to create their unique solution.
That may be a good aspect of human nature.

1.6 Importance of Design and Design Thinking

Design and design thinking are essential in dealing with all aspects of any system.
Design thinking implies that a system must be designed first before analyzed.
Sometimes based on the analysis, we may have to go back to the original design of
the system to improve the design. Unfortunately, sometimes, engineers are mired in
an analysis of wrong designs rather than changing the design. Engineering edu-
cation sometimes biases students’ thinking by giving well-defined problems at the
end of each chapter, without explaining the design that led to the problem. To
repeat, design thinking emphasizes the need to design first, followed by analysis.
This chapter outlined how one should begin the design process.
We design two kinds of systems: technical and non-technical. Much of this book
deals with technical and scientific design. However, the design of non-technical
systems such as government and universities has equally significant consequences
on society and humanity as much as technical systems. Institutions and govern-
ments must be designed well for them to serve their constituents as well as
20 N. P. Suh

intended. Unfortunately, this is not always the case. In short, “Design Thinking” is
equally applicable to all designs.
When we deal with systems, technical as well as non-technical, we must deal
with synthesis and analysis. In designing new systems or modifying old systems, it
is often instructive to consider the synthesis (or design) issues first before delving
into an analysis of a subset of the overall system. One can spend a lifetime ana-
lyzing a poorly designed system, because coupled systems, i.e., poorly designed
systems, are often mathematically intractable. Unfortunately, more people in all
intellectual persuasions tend to delve into an analysis of lower level issues and then
become bewildered by conflicting details of a poorly defined or designed system.
To reiterate the significant point of this chapter: identify the PROBLEM(s) of a
system that must be improved (in the case of an existing system) and then establish
goals in the form of FRs that must be satisfied. Then find DPs that can satisfy the
FRs, before undertaking a detailed analysis of lousy design, thinking that opti-
mization will improve the system. An optimized, poorly designed system is still a
bad design!
In many professions, the precedents provide guiding lights, and therefore they
look for similarities of a case with an old example for decision-making. For
instance, not too many political leaders in many countries have a technical or
scientific background. In the United States, many of them have legal or business
backgrounds. Occasionally, there were exceptions. Two of the past presidents of the
United States, Herbert Hoover (1929–1933) and Jimmy Carter (1977–1981), had
technical backgrounds. Hoover was a mining engineer and Carter was a nuclear
engineer who served on nuclear submarines in the United States Navy. Even today,
they are regarded as honest and ethical people with deep convictions. However,
historians might not see their presidencies as successful ones. During their presi-
dencies, the United States encountered economic difficulties. The Hoover admin-
istration had to deal with the world’s worst depression of 1929. During the Carter
period, inflation was out of control. Either they were unlucky or mismanaged the
economy. Did they concentrate too much on individual detailed issues rather than
the systems issues to develop suitable designs for economic policy? The simple
conclusion may be that they focused too much on details before they really iden-
tified and understood the problems and then design their policies accordingly. We
must identify the overall problem first and then design sound policies. Analysis of
poorly designed systems typically is not enlightening. Accurate analysis of a poorly
designed system cannot improve the poorly defined system. If the system is cou-
pled, i.e., FRs are not independent, improving one FR may result in deterioration of
other FRs, negating the intended effect.
It should be emphasized again here: The first step in developing innovative
design typically requires the identification of the problem.
Once we know the problem, we establish the goals that our design must satisfy
to solve it and other associated issues. (Sometimes, we set the goal first based on the
“customer need” and then identify the problems that must be solved to achieve the
goal.) These goals must then be stated as specific FRs that our design must satisfy.
The final step in design is to come up with DPs that will enable us to fulfill the FRs.
1 Introduction to Design 21

The last step is to configure DPs through the integration of various DPs as an
integrated system that may consist of hardware, software, information systems,
natural elements, and others, such as sensors. In some ways, it is an obvious way of
coming up with good designs. However, many companies have repeated the error
made by some manufacturers, i.e., trying to be more competitive, without first
discovering the specific problem faced by their company other than lowering the
cost of manufacturing. This situation may not be only confined to manufacturing
companies but also individuals as well.
Most people confront many problems, large or small, throughout their lives.
They have to solve them through design and perhaps by other means as well. The
difficulty is that often they are trying to find solutions without first defining the
specific problem. They may come with various and contradictory claims and
counterclaims, aggravating each other. If we can determine “what the problem is,”
it may be easier to develop solutions. The ability to define the problem can be
acquired by accumulating broad knowledge base, experience, and design thinking.
One of the goals of this design book is to teach the ability to identify and define the
problem.
Once the problem is recognized and defined, most people can synthesize their
unique solutions, unless they can find an immediate solution from their “library” of
experience that is similar to the current problem. When there are no prominent past
examples that can be adapted, people should seek a new solution through “design”
with confidence without being intimidated. The decision to either adopt an old
existing solution or create a new “solution” depends on one’s knowledge base or
experience or the confidence that one can create something new from scratch that
will be superior to the existing system. Many people instinctively want to create
unique solutions. That may be the meritorious aspect of human nature.

1.7 What is the Most Difficult Aspect of Learning

Axiomatic Design?

Every era has certain businesses that dominate the economy. In the first half of the
twentieth century, it was the automotive companies (such as Ford, General Motors,
Toyota, and Daimler Benz). In the late twentieth century, manufacturers of com-
puters (such as IBM and Microsoft) and semiconductor chips and integrated circuit
devices (such as Intel) have dominated the economy and technology. In the first half
of the twenty-first century, it was the era of high-speed Internet and social net-
working and telecommunication businesses (dominated by companies such as
Amazon, Facebook, Google, Uber, and others) that have built their business using
digital technologies and the Internet. In all cases, they were successful in creating
new industries, because they could identify problems that need to be solved and
came up with FRs and DPs for their businesses, using newly emerging
technologies.
22 N. P. Suh

One of the most stumbling mental blocks in learning AD could be the lack of
experience in defining the FRs for the problem identified in the customer domain.
To some, defining FR is a trivial task, and to some others, it is a difficult task. To
many, it is a new experience to think of design as the transformational process of
going from problem definition to FR, going from FR to DPs, and from DPs to PVs.
There could be many reasons for this difficulty—different depending on one’s
experience and educational background. Some people might have worked on
problems someone else defined for them throughout their life. Students are often
taught to solve the problem defined by others, especially in textbooks. In some
countries, the college entrance examinations ask only analysis-oriented questions.
Not surprisingly, students prepare for the exam accordingly. Also, in some cases,
one might have spent most of their career, modifying, through trial-and-error
processes, designs someone else has made. In many companies, high-level exec-
utives define the problem, and engineers/designers execute them. In universities,
professors assign problems to be solved, and students are taught to solve them
following well-established methods and processes. Also, throughout their educa-
tion, students are taught that there is ONE correct solution, which is true in most
cases of analysis-oriented subjects. However, in design, there can be many equally
good solutions!
The best way of learning how to define the problem to be solved and establish a
specific set of FRs is to go through the experience doing them a few times to
internalize the process in one’s brain. Pretty soon, it can become second nature to
young students.
Design is ecumenical in the sense that the design methodology is not a
field-specific subject. The same method and approach can be used in all fields that
require synthesis and design, although the specific design task will depend on the
nature of the problem and the FRs that we must satisfy. Some students claim that
the most challenging part of learning and using AD is the process of coming up
with FRs, which are derived from the problems that were identified in the customer
domain. Some claim that they could delve into the design without specifying FRs,
which may indicate that the person either modified or copied an existing design.
When someone designs without specifying FRs, one may come up with a coupled
design that does not withstand the test of times or fails to perform. One must invest
time and effort to state FRs correctly and creatively. It is not difficult to learn how to
state FRs, but it may take longer to state FRs than the designer has been accustomed
to.
The basic rule in defining FRs is the following: Define FRs in a solution-neutral
environment! That is, “DO NOT THINK OF THE SOLUTION FIRST
BEFORE DEFINING THE PROBLEM AND THE FRs.” When this simple rule is
violated, the proposed design is simply an old reconstituted design.
Some designers and organizations spend “five minutes” in defining the design
task and spend months or years to make the designed system work by correcting all
the problems that continue to pop up because of the poor and unsystematic design
practice. This practice leads to long development times and high costs. There are
many well-publicized failures where a critical project failed to work, airplanes
1 Introduction to Design 23

plunge into ground, cars stop all of a sudden leading to fatal accidents, and R&D
projects had to be redone, all because the designers used their gut feelings to design
some new products, purely based on their years of experience in designing systems
and what they have seen before somewhere rather than defining FRs a priori.
Designing Orbital Space Plane (OSP)
A principal defense contractor in the United States received a government contract to
design and manufacture the orbital space plane (OSP), which can replace the space shuttle
that transported people and goods to the International Space Station. One of the primary
goals of OSP is to reduce the cost and improve the versatility of space transportation. The
manager in charge of the project decided to produce a better product at a lower cost. To
achieve this goal, the visionary the leader of the OSP program chose to replace the past
practice of repeating the “design-build-test” cycle of system development and production,
because the company leadership found that such a practice is costly and invite major
failures after they deploy the system. They hired a consultant to teach AD to their “lead
engineers,” about 250 engineers and scientists.
These highly experienced and skilled engineers and scientists learned AD rather quickly.
However, it was much more challenging to teach experienced engineers and designers than
undergraduate students because they tended to jump right into the physical domain, i.e.,
DPs and PVs, without ever defining what they want to achieve, i.e., FRs, explicitly. In many
cases, many experienced designers and engineers often jumped right into solutions, i.e.,
DPs and PVs, without first establishing FRs and constraints, even when FRs are simple to
state and in some cases, almost trivial. Sometimes, people try to state FRs after they come
up with DPs, which is counterproductive in practicing AD. When FRs are not defined in a
solution-neutral environment, the design may turn out to be a replication of a product that
already exists. For innovation, it is of utmost importance to define FRs in a solution-neutral
environment.
The experience of teaching AD has been that it is easier to teach undergraduate students
than graduate students or experienced industrial engineers. The underlying reason might
be that experienced people do not define FRs in a solution-neutral environment. They may
instead think of a solution first based on their experience and justify their decision
afterward.

In the corporate world, people who become top executives need the ability to define
the problem and FRs from a systems point of view. They often have many com-
petent people working for them who are well trained to solve specific problems if
they are presented with well-defined problems. It is harder to find people who can
define the problem the company needs to address. Even in academia, people who
can lead an organization need the ability to identify the problem and create specific
goals in the form of FRs. Sometimes, universities select students based on their
ability to solve problems someone else defined for them, and then teach more of the
same skill for another 3 or 4 years. Sometimes, the student rarely has the oppor-
tunity to identify the problem themselves and establish goals in the form of FRs.
24 N. P. Suh

1.8 List of Example Problems

Consider the following problems and think about how one might approach them to
develop a design solution for each example. Some of these examples will require
much thinking and work to develop the desired design solutions. In some cases, one
may have to acquire some fundamental knowledge of the field by reading reference
books or by getting information from the Internet. Most of all, one may have to
THINK with an open mind in a solution-neutral environment! We should not forget
is that all these design tasks can be achieved, given enough time and resources.
Example 1.1 Improvement of the Protocol System for Social Networks
Social networks (SN) such as Facebook and Twitter facilitate interaction between and
among people as well as between people and business for fast social information distri-
bution or disinformation. In the 2016 presidential election in the United States, certain
groups or nations used these SN services to tilt the election by spreading distorted or
fabricated information to American electorates. One of the issues we have to deal with is
the privacy issue, because Facebook, for example, has accumulated so much data on
individuals that some may use them for illegal purposes. How would we improve the
protocol of SN to prevent such misuse of the SN systems?
Example 1.2 Water Faucet to Control Water Temperature and Flow Rate
We want to have a water faucet that will enable a person to control the water flow rate and
the water temperature independently. How should we achieve these goals through the
design of a new faucet?
Example 1.3 Operation of Emergency Room
In a typical emergency room (ER) of a hospital, as many patients as possible must be
treated as quickly as possible. One problem faced by the hospital is that patients with many
different kinds of illnesses come into ER without any prior notice, e.g., some because of the
injury sustained in a car accident, another due to a flare-up of the chronic disease. Often, the
doctors are so busy that patients must wait after checking in with the admitting nurse before
the patient can see the doctor, especially if the illness appears to be chronic. Many hospitals
use the FIFO (first in–first out) system to control patient flow in the ER unless someone is
about to die, who should be given the highest priority. If one is to design a better triage
system for the ER operation, what should the new system? Would an AD designer be able
to create a better triage system to maximize the throughput rate of the ER?
Example 1.4 Global Warming
One of the societal goals is to solve the global warming problem by reducing anthropogenic
CO2 emission. Roughly 27% of CO2 is generated by ground transportation systems and
about 33% from electric power generating plants that burn coal. Suppose that our goal is the
reduction of CO2 as the primary design goal to deal with global warming. To achieve the
goal, what specific actions would we undertake?
Example 1.5 Ski Bindings
Ski bindings enable a skier to transmit control loads from the boot to the ski to maneuver
the ski while sliding on snow at varying speeds. One of the main problems is to avoid
transferring loads from the ski to the boot that leads to injuries. They need to transmit
control loads from the skier to the ski reliably and not transmit injurious loads from the ski
1 Introduction to Design 25

to the skier. The solution is to have the binding release the ski from the boot under
certain situations that lead to injury. Because the ski bindings do not perform these func-
tions well, many skiers end in hospitals especially with knee injuries most to the ACL and
from collisions after an inadvertent release of the ski by the binding. How would we solve
this problem?
Example 1.6 3D Printing
One of the final outputs of design could be the manufacture of reliable goods. Some of
these products have complicated three-dimensional shapes such as inside holes not con-
nected to the outside surface. We have used, and often still use, various casting and
machining processes to make these parts. The disadvantage of these traditional processes is
the need to prepare expensive tooling, which necessitates that we make a large number of
the same pieces to distribute the fixed manufacturing cost over a large number of parts to
lower the unit manufacturing cost. How can we make these 3D parts using a
general-purpose machine that can produce several products with minimal tooling and in
small volumes on short notice?
Example 1.7 Money Circulation and Economy
To an engineer, it appears that one way of strengthening the economy is to increase the
velocity of money circulation because the faster the money circulates, the higher will be the
economic activity and increased economic growth. Thus, everyone within the economic
system benefits as the velocity of money circulation increases. However, real rich people
will only spend a small fraction of their wealth, whereas poor people do not have the money
to spend. This situation limits the velocity of money circulation. Assuming that the wealth
distribution is Gaussian, determine the ideal wealth distribution that will maximize the
money circulation. As a designer, our task is to design several policies for increasing money
circulation. State your FRs.
Example 1.8 Healthcare Delivery
A medical doctor has spent many years to deliver healthcare services to developing
countries. One of the problems he has encountered is not the medical problem, but the
logistics of delivery of medical care and medicine to remote areas. He is looking for
“system engineers” who can solve the logistics problem working with MDs. What FRs
should we try to satisfy in our design? How should we create a solution to this problem
through the design of a system for delivery of medical care to remote regions?
Example 1.9 Global Warming
Many people in many countries are concerned about global warming. The Intergovern-
mental Panel on Climate Change (IPCC) of the United Nations determined that unless we
can keep the temperature rise of Earth to within 2 °C relative to the temperature at the time
of Industrial Revolution, many calamities will make the Earth much less habitable with
more floods, hurricanes, forest fires, drought, and rising ocean submerging low lying lands.
The consensus developed based on scientific data is that one of the major causes of these
problems is the anthropogenic generation of CO2. As a designer, what would you do to deal
with this problem? What should be our goals to deal with the challenge created by global
warming?
Example 1.10 Reduction of Plastic Consumption
A high-level executive of a major industrial firm that manufactures many products out of
polymers (i.e., plastics) is searching for means of reducing the consumption of plastics to
lower the cost of the company’s products. Since the materials cost constitutes about 50% of
the manufacturing cost, the executive reasoned that the company could lower the cost of
26 N. P. Suh

their products by 5% if they can reduce the materials consumption by about 10%. Their
products must look and perform the same as their current products. As a designer of
manufacturing processes, develop a design for achieving the above-stated goal—first state
the FRs you have selected.
Example 1.11 Mobile Harbor
Containerships carry as many as 16,000 containers (TEUs2) and cross the Pacific Ocean to
deliver goods in containers to the United States and elsewhere. Many of these ships unload
their containers in Long Beach, or Los Angeles harbors in California, which is then
transported to the rest of the United States by freight trains because the Panama Canal was
too narrow to accommodate these giant containerships. Now the Panama Canal has been
widened, but the problem is not entirely solved, because the harbors in the eastern seaboard
of the United States are not broad enough and deep enough to accommodate these large
containerships. As a solution to this problem, mobile harbor (MH) was invented while
visiting Singapore and developed at KAIST. The central idea for the MH is as follows:
“Why should ships come into the harbor? Why not have the harbor go out to the ship?”
Under this scheme, big container ships moor in deep waters rather than come into a harbor.
The MH (see Fig. 1.5) that can handle 600 containers goes out to the containership to
unload the containers and deliver them to their final destinations. MH has a relatively flat
bottom rather than streamlined V-shape, which enables it to turn quickly in harbors,
maneuver easily in shallow waters, and go to any place to unload the containers.

Fig. 1.5 Mobile harbor concept invented at KAIST (four mobile harbors are unloading containers
from a large ocean-going containership in open sea to transport the containers from the ship to the
shore)

To develop MH, we had to deal with two problems. The MH had to be firmly tied to the big
containership during the loading and unloading of containers, so they move in unison in the
rough and windy sea. A new design issue was how to tie the MH to the container-
ship. Another problem is related to unloading the containers from the big ship to MH. The

2
A TEU (20-foot equivalent unit) is a measure of volume in units of 20-foot long containers. For
example, large container ships are able to transport more than 18,000 TEU (a few can even carry
more than 21,000 TEU). One 20-foot container equals one TEU.
1 Introduction to Design 27

containers on the big containerships are lifted using a crane and steel rope. During the
transfer of the container from the ship to MH, the container may oscillate due to the motion
of the ship and MH. How should the containers be unloaded from the containership to MH
on a windy day in a rough sea? What FRs should we satisfy?

In all of the above examples, better solutions may involve “design” of a new
solution rather than adapting the past or existing designs. Some of these examples
are not simple tasks and may require in-depth thinking and extensive work to
provide the answer. They are given here to let the reader think about various issues
involved in design. Many of the solutions will be systems that consist of many
subsystems.
The purpose of this book is to teach the fundamentals of design to those
interested in “developing superior design solutions.” To achieve this goal, in the
subsequent chapters, the design principles based on AD will be introduced with
many examples and case studies. Many case studies show that projects executed
based on the teachings of AD cost less and deliver a superior design.

1.9 Definition of Systems

The final output of design is a “system” that solves the problem and satisfies the perceived
original need. A system is defined as an entity that generates a set of pre-determined
outputs when a set of pre-defined inputs is supplied.

The operation of all systems consumes energy, i.e., the energy input to the system is
higher than the energy output of the system. Some also are net consumers of
materials and human resources. Some systems require financial resources to
operate, and some are net generators of financial returns. The performance of a
system is measured in terms of economic measures, or efficacy and fidelity in
satisfying FRs, or social and human benefits.
Some systems are massive both in terms of the number of FRs they satisfy and
their physical size. However, there is no relationship between physical size and the
number of FRs they must fulfill. Some systems, such as semiconductor devices, are
tiny physically, but they fulfill a large number of FRs. Some systems are measured
in terms of people involved in operating the system.
Some systems are relatively simple, whereas some systems are complicated
partly because of the number of FRs involved in a system and the nature of DPs
chosen. Some systems are complex because they may not satisfy the FRs at all
times. Sometimes a complicated system is also complex, but not always. For
instance, the design of Boeing 787 is complicated because of the number of
functions it must perform is very large as well as the number of parts that make up
the airplane. However, they are not complex, because they satisfy their FRs with
100% certainty. On the other hand, the ignition key of a car manufactured by one of
the largest automobile companies in the world could be extremely complex, if the
28 N. P. Suh

probability of the key performing its functions is much less than 100%. Complexity
increases when the system cannot satisfy FRs with 100% certainty.
The constituents of systems vary depending on FRs, DPs, and PVs. Exemplary
system elements are physical elements, natural elements, software programs, bio-
logical units, humans, ideas (e.g., books), and some combinations of all of the
above. Many systems operate within a set of constraints. There are two kinds of
constraints: pre-existing external constraints and some constraints created during
the design process or operation of the system.

1.10 Fundamental Principles of Design

The following two axioms constitute the basis of AD. All designs must satisfy
them. We can separate good designs from unacceptable designs by checking
whether or not they are consistent with these two axioms.
The Independence Axiom
Maintain the independence of FRs.
The Information Axiom
Minimize the information content.

Subsequent chapters explain and apply these axioms to various systems and
problems.

1.11 Principle of Similitude of Systems

This book deals with many different systems, including mechanical, electrical,
chemical, software, organizational, healthcare, and others. Once we understand the
basic design principles and methodologies in one field, we should be able to deal
with other design problems in many different fields. The reason we can treat many
design problems in many diverse fields is that they are all systems with similar
structures and characteristics. That is, the concept of design domain, the design
axioms, and the design process apply to all design problems regardless of the
specific field of application.
The Principle of Similitude of Systems may be stated as follows:
All systems follow the same design principles and processes, and therefore the
same concept of design applies to all systems, although their specific functions,
components, and usage may be domain-specific.
After students learn the materials presented in this book, they should be able to
solve the 18 examples given in this chapter and be able to solve many other original
design problems.
1 Introduction to Design 29

1.12 Importance of Knowing the Basic Laws

and Principles of Science and Engineering

To be a creative designer of engineering artifacts, it helps if the designer has a

strong background in basic disciplines of relevant engineering fields. Similar
comments are equally valid for other disciplines. Without a strong knowledge of the
related subjects, it is difficult to identify the problems that need to be solved and
follow through the steps involved in the transformative process for the design of
creative solutions outlined in this book.
The output of design is the functional requirements, {FRs}, which is what the
designer wishes to satisfy through design. The inputs are the design parameters,
{DPs}. If the design is related to physical things, the designer should have a
fundamental understanding of natural laws and principles, constitutive relation-
ships, and conservation principles. If the design involves information technology,
the outputs are codes and background in algorithms should be invaluable. If the
system is biological or medical, the designer should have background life science
subjects such as biology. If the design is related to organizations, {DPs} may be
organizational entities. After the design is completed, a more detailed mathematical
modeling may be necessary to choose the correct values for DPs and PVs.
The results of the design we see are the assemblage of DPs. How DPs should be
physically arranged is an issue that needs to be addressed, sometimes by examining
the physical proximity in the case of mechanical design.

1.13 The Inverse Problem in Design: Extraction

of Functional Requirements from Many Design
Parameters

In the preceding sections of this chapter, the emphasis was on identifying the
problem to be solved through design. In subsequent chapters, we will go through
how the design can be done to solve the problem identified through the transfor-
mation of the problem to FRs, which are then transformed to DPs. Similarly, DPs
will be transformed into PVs. We will show how FRs can be stated to represent the
design task. We transform FRs to select appropriate DPs. This approach may be
called the “direct” approach to design.
There is another approach to design, an “inverse” approach, i.e., going from
existing designs to uncover the DP that can satisfy the desired FR. Usually, it is a
difficult task to go from DP to FR, because a DP may satisfy many FRs. For
example, the FR of a “coffee mug” can be many. It could be “hold hot coffee,” “act
as a paperweight,” “commemorate a special event,” and others. The fact that a DP
can be related to many different FRs is one of the difficulties of “reverse engi-
neering” because determining the FR from geometric shapes is challenging.
However, with the enormous computational power of modern computers, we can
manipulate a vast database and try many different combinations of existing DPs to
30 N. P. Suh

satisfy FRs. In other words, the inverse method consists of going through a vast
database of DPs to identify the design of a system that is close to the problem
identified. This “inverse method” is possible because of the extensive data that can
be processed by supercomputers and cloud computing. This approach is, in essence,
the use of artificial intelligence (AI) to synthesize design solutions. With the
increasing use of AI techniques, the inverse approach to design may be employed
more in the future.
IBM has developed a machine named “Watson.” Watson is an intelligent
question answering (QA) computing system. IBM built it to apply natural language
processing, information retrieval, knowledge representation, automated reasoning,
and machine learning technologies to answers questions posed by people using
natural language. It has access to 200 million pages of structured and unstructured
content consuming four terabytes of disk storage. These machines can store an
immense amount of data, which are used to answer questions posed by the user.
IBM has been attempting to make Watson be the next generation of products that
can generate new revenues for the company. It won chess games over the best
human master chess player. Watson has been used in medical applications to guide
medical practitioners as well. Google has its version similar to Watson that can
provide answers to queries made by people. These technologies are possible
because of large computers and easy access to the database produced by large
networks and cloud computing. The difficulty lies in satisfying many FRs at the
same time, which many design tasks require.

1.14 Optimization of an Existing System Versus Design

of a New System
“Optimization of a poorly designed system yields yet another poorly designed system.”

Many engineers, economists, and others in many different fields are engaged in the
analysis and optimization of an existing system. They devoted significant effort to
get the most out of existing systems, which were designed and have been used,
sometimes, for decades.
There are many optimization techniques, mostly, mathematical, that have been
developed. Typically, the mathematical approach is to express the problem in terms
of an objective function with constraints. Even when there are many objectives, the
problem is formulated for one objective function with many constraints for math-
ematical convenience and treatment. Such a brute force approach would not be
applicable when an entirely new system must be designed to satisfy many FRs.
James Watt would not have invented his steam engine if he tried to optimize the
Newcomen engine.
1 Introduction to Design 31

1.15 Scope of the Book

The purpose of this book is to enable the reader to “design” on her/his own, with
imagination and creativity, to satisfy human needs and societal aspirations. Human
intellect is a powerful tool that enables human beings to achieve many things
through design—it only requires the imagination and willingness to learn how the
design should be done. This book outlines the steps involved in design based on
AD to develop a rational design solution. The basic idea is to approach design
systematically so as not to make wrong designs. There are well-known design
mistakes that cost a great deal to re-do them to correct the errors, a la the design
problem associated with control of Boeing 737 MAX. Many of these mistakes are
often due to the coupling of FRs during the decomposition process. These coupled
designs lead to accidents of nuclear power plants, crashing of aircraft, delayed
construction of airports, and unreliable products.
This book presents the basic concepts involved in AD: the idea of four domains
in the design world, the mapping between the domains, and the transformations
involved during the design process. After the problem that needs to be solved is
identified, the designer sets the ultimate goals of the design task in terms of
functional requirements {FRs} and constraints (Cs). Then perform specific design
tasks of identifying design parameters {DPs}. The idea of the design matrix is also
introduced in this chapter. The design process, including the decomposition through
zigzagging, is illustrated in this chapter. Many examples are given to clarify the
new concepts presented.
The subject of “DESIGN” treated in this book is not limited to engineering,
although many examples are derived from engineering and technology. Design is
equally important in many other fields that involve synthesis to achieve a set of
goals, such as in the design of software, organizations, and even in cooking
gourmet foods. The same thinking and methodologies apply to all these subjects.
That is, although the specific topics and subject matters are field-specific, all fields
share the standard design process and the same design axioms.

1.16 Conclusions

The design is one of the most critical subjects in engineering. Synthesis of inno-
vative products and solutions is the essential foundation for solving societal
problems and advancing commerce, engineering, science, and social science fields.
The ability to design well determines the quality of most things: products, pro-
cesses, manufacturing, organizations, governments, technologies, the quality of life,
and others.
Good designs depend, the foremost, on the quality of problem identification and
definition. Once the problem is defined, the designer can proceed to the subsequent
steps of design. The steps consist of the transformation of customer needs of the
customer domain into functional requirements {FRs} of the functional domain,
32 N. P. Suh

followed by transformation of {FRs} to design parameters {DPs} of the physical

domain, and finally the transformation of {DPs} to process variables {PVs} of the
process domain. The relationship between the domains is “What” and “How.” The
{FRs} in the functional domain is “what we want to achieve,” whereas the {DPs} in
the physical domain represent “how we are going to satisfy the FRs.”
Similar design processes govern the design of diverse systems. A person who
can design technical systems well can also apply the same skill to other design
problems such as organizations, although specific issues are domain-specific.
Problems
1. In the twenty-first century, telecommunications and social network systems
(SNS) have become the dominating information dissemination mechanism,
displacing the printed media. As a result, the information distributed in SNS
throughout the world can be corrupted by those with ill-intentions and nations as
a warfare tool. Our job is to solve this problem to safeguard the system and
make the information in the digital communication system from being cor-
rupted. What FRs, would you satisfy this problem?
2. Faulty designs or poor designs cause many failures in various systems. The
recent crash of Boeing 737 airplanes, the delayed opening of the new Berlin
airport, and the Fukushima Daiichi nuclear power plant disasters are some of the
well-publicized failures. These accidents occur despite numerous tests. What is
the best way of preventing these failures?
3. If you are the president of your university, how would you improve the
admissions process? What is the problem you are trying to solve? What func-
tional requirements {FRs} should your university try to satisfy through their
admission process?
4. Many companies are working on driverless automobiles. What functional
requirements {FRs} should the designer of the automobile satisfy?
5. What do you think makes Starbucks so successful in selling coffee even though
so many other companies had already been selling coffee before their
emergence?
6. Define the FRs we must satisfy to teach AD well.

References
Black NH, Davis HN (1913) Practical physics for secondary schools. Fundamental principles and
applications to daily life. Macmillan and Company, public domain
Petrushenka A (2019) The launch of the space shuttle. With fire and smoke. Against the
background of the starry sky. Elements of this image were furnished by NASA. Photo licensed
from Shutterstock, ID: 1480426817
1 Introduction to Design 33

Further Reading

Hatamura Y (2008) Learning from design failures. Springer

Senor D, Singer S (2009) Start-up nation: the story of Israel’s economic miracle. Twelve Hachette
Book Group
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Suh NP (2010a) On innovation strategies: an Asian perspective. The Glion Colloquium. https://
glion.org/on-innovation-strategies-an-asian-perspective/
Suh NP (2010b) Theory of innovation. Int J Innov Manag (IJIM) 14:893–913. https://doi.org/
10.1142/S1363919610002921
Suh NP, Cho DH (eds) (2017) The on-line electric vehicle: wireless electric ground transportation
systems. Springer International Publishing
Wilson HG, McCready PB, Kyle CR (1989) Lessons of Sunraycer. Sci Am 260(3):90–97
What Is Design?
2
Nam Pyo Suh

Abstract
The design is a commonly used English word that describes a variety of different
human activities, depending on the context and the field of interest. In this book,
the word “design” refers to a series of creative activities that are related to
solving a problem identified in the customer domain to achieve a goal through
synthesis. The identified problem is solved through synthesis. A set of functional
requirements (FRs) is established in the functional domain through the mapping
between the customer domain and the functional domain. The FRs are then
satisfied by selecting design parameters (DPs) in the physical domain through
the mapping between the functional and the physical domains. The selected DPs
must satisfy the Independence Axiom, i.e., FRs at a given level of decomposition
must not affect other FRs by the selected DP or be affected by other FRs. DPs
are, in turn, satisfied by selecting process variables (PVs) in the process domain
through the mapping between the physical domain and the process domain.
Thus, the design consists of a series of transformation processes, beginning from
the goal and the problems identified in the customer domain to FRs of the
functional domain, FRs to DPs in the physical domain, and DPs to PVs in the
process domain. In this design process, analysis often follows synthesis for the
quantification of design decisions made.

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 35

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_2
36 N. P. Suh

2.1 Introduction

2.1.1 Why Design Theory (DT)?

Many engineers and scientists often wonder why we need design theory. Many of
them are proud of their past accomplishments because they have designed many
things based on their intuition and experience. However, consider the following
simple example:
Disaster Faced by a Major Airplane Manufacturer in 2019
The CEO of an airplane manufacturer was a man of full confidence for having come up
through the ranks of the company to become the leader of one of the largest airplane
manufacturers in the world. More importantly, he was confident of his engineers, who have
racked up years of experience. Overall their airplanes performed well in the field. Their
new airplane, which was an extended version of their best-selling aircraft but with larger
more fuel-efficient engines, was selling well.
Then, a disaster struck the company, bringing down their stock price and losing potential
new customers to their rival company. Two of their planes fell from the sky, killing all the
passengers and the flight crews! The pilots of the airplane could not control the aircraft
during its ascent right after takeoff and thus plunging into the ground. Many organizations,
in addition to the manufacturer of the airplane, began investigations. They will eventually
come up with explanations and solutions after much testing and information gathering.
We cannot draw any firm conclusions without knowing the actual design of the control
system developed by the airplane company and the data in the black box of the airplane. In
the following example, we will illustrate how Axiomatic Design (AD) might be used to
provide answers to the failure of an aircraft as a means of explaining the use of AD. To
determine the real cause of failure, the airplane manufacturer usually works with
authorities, analyzing all available data, and simulating with the real similar airplanes.
They will also test the actual system operation using simulators. Based on these results, the
airplane manufacturer will make appropriate changes in software and hardware to prevent
future failures. However, all the persons perished in those accidents would never know
what happened to them.
In this simulation of what might have happened, we will make up a fictitious case study to
show the power of AD in the proper designing of complicated systems. The irony is that it is
not difficult to develop the right system, but if one makes design decisions based on “gut
feeling,” one can make serious mistakes that are very costly both financially and also in
terms of human lives.

Design of the airplane control system:

Assume that the pilots must control four functions of the airplane for its safe operation,
which we will designate as FRs, i.e., FR1, FR2, FR3, and FR4. We will assume that the
outputs are controlled using four input control variables, which we will designate as DP1,
DP2, DP3, and DP4, respectively. Furthermore, we will assume that DP3 is automatically
controlled by the system installed as part of the airplane control system, and thus the pilot
cannot manually control DP3. However, the pilot can control DP1, DP2, and DP4. The
relationship between FRs and DPs may be represented as
2 What Is Design? 37

FR1 ¼ a DP1 þ b DP2 þ c DP3 þ d DP4

FR2 ¼ e DP1 þ f DP2 þ g DP3 þ h DP4
ð2:1Þ
FR3 ¼ i DP1 þ j DP2 þ k DP3 þ m DP4
FR4 ¼ n DP1 þ q DP2 þ s DP3 þ w DP4

The above relationship may be rewritten in a matrix format as follows:

ð2:2Þ

Suppose that they designed their control system such that the coefficients a, b, c, … w are
fixed and may not be varied during the flight. Furthermore, the airplane company sold
these features (i.e., a, b, c, … w) for an extra price as options. Therefore, some of the
airplanes did not have all the features provided by a, b, c, … w.

Questions the reader should try to answer:

Case #1: Suppose the design is such that all coefficients are zeroes except a, f, k, and w. The
desired values of FR1, FR2, FR3, and FR4 are given as 5, 9, 3, and 8. The control task is to
set the values of DPs (i.e., DP1, … DP4) to satisfy FRs. What should be the values of DPs?
The design matrix for this case is a diagonal matrix as shown below:

ð2:3Þ

This diagonal matrix represents an ideal design, where one DP affects only one FR. The
control of this kind of systems is the simplest. We should attempt to develop systems that are
of this kind.
Case #2: Suppose the design is such that all coefficients are non-zeroes (i.e., a full matrix).
The desired values of FR1, FR2, FR3, and FR4 are still 5, 9, 3, and 8. The control task is to
change one of the FRs without affecting any other FRs by varying appropriate DPs. What
should be the values of DPs? What should we do if the coefficients vary from time to time?
If we want to change only FR1 without changing other FRs, which DPs should we change
to vary the FR1?
Case 2 represents a coupled design. When one of the FRs is to be changed, it will be
necessary to vary all the DPs simultaneously. Such an airplane will be hard to control in
actual flight. The control of the aircraft by the pilot will be extremely challenging if the pilot
38 N. P. Suh

does not know which DPs are changing because the computer onboard directly controls
the DPs, i.e., if the pilot is out of the control loop. Even if he knows that the computer is in
the loop, the pilot would not know how to control the airplane because the computer will
change DPs.
Was this coupled design the reason the two Boeing airplanes plunged into the ground?
Case #3: Anything in human-made systems can malfunction. That is, one of the elements of
the design matrix can malfunction, i.e., change arbitrarily on its own without any input
from the pilot. In that case, an unexpected coupling may occur, or the plane may lose some
of its functional capabilities. In that case, the pilot must be able to override the automated
systems and operate the airplane manually before they lose control over the airplane.
For simplicity, we discussed three extreme cases in this example. Other cases will be
covered in later chapters.
Think of the panic faced by the pilot! The pilot cannot see the design matrix and has to try
to control the plane with his feeling, control lever, and engine speed. In this case, the
situation was even worse. The automated control system also tried to control the plane,
probably over-riding the inputs of the pilot—a classic example of a coupled design!
This example shows that the design with a diagonal matrix is the best design in terms of
satisfying FRs by varying the input variables, i.e., DPs. When the matrix is a full matrix, it
is complicated to vary only one FR with one DP, i.e., all DPs may have to be changed
simultaneously, which increases the complexity of the design and operation of the system.
Typically, it cannot be done. This kind of design is a coupled design. When many groups of
engineers start changing the design without informing the “design architect”—by
changing one or two elements of the design matrix—and without the group consensus, it
will be extremely difficult to make the airplane controllable and safe when unexpected
events occur. This coupling of FRs adds to the complexity of the airplane operation.
Without knowing the details of what happened to those two airplanes, one can only
speculate that the failure of the aircraft might have been due to the coupling of FRs. This
failure may be due to the wrong design of the control-system software. Pilots cannot avert
disasters in such a case.
The vice president in charge of engineering or the president of the company should have
asked a simple question: “Does the aircraft control system, which includes both hardware
and software, satisfy the Independence Axiom? That question could have saved hundreds of
lives!”

We devoted most of Chap. 1 to one issue: the importance of identifying the goal to
be achieved and defining the PROBLEM in the customer domain to accomplish the
goal. To repeat: identify and define the problem well, before embarking on the
design activity! That is, the ability to identify the most critical problem that needs to
be solved is the first and the most defining step for successful design and inno-
vation. The preceding statement is true, regardless of the specific nature of the
problem. Even if it takes several trials, it is less expensive and fruitful to spend time
identifying the right problem at the early stages of design. Without James Watt’s
identification of the shortcomings of the Newcomen engine, the Industrial Revo-
lution might not have begun in the latter part of the eighteenth century! Although
many other people had used the Newcomen engine before James Watt did, they
merely used it without recognizing and questioning its shortcomings. The power of
2 What Is Design? 39

observation, intelligent questioning of well-accepted wisdom, and penetrating

logical reasoning are some of the essential requisites in defining the problem (or
shortcomings) and creating innovative designs to achieve the goal identified in the
customer domain.
In this book, we present a specific theoretical and practical approach to design
based on AD theory. According to the theory, one should not jump into analysis or
experimentation right away before a design solution is conceived or completed.
When the problem is identified in the customer domain, the designer should think
about the overall design issues, i.e., goals of the design and desired final output. The
first step in AD should be the identification of THE PROBLEM, as discussed in
Chap. 1. The subsequent steps of AD involve the transformation of the problem
identified in the customer domain into FRs in the functional domain, followed by
two additional transformations to generate DPs and PVs.
To be proficient in design, the designer should initially think broadly about many
related issues to the design task, before settling on a design solution. It is highly
advisable to consider all the peripheral matters before defining the problem to solve.
It is also good to think deeply about the implications of the problem to be chosen.
Any changes made to the problem definition or FRs and DPs have consequences,
and therefore one should consider all alternate possibilities before making the final
decision on design goals and problems. It is much cheaper and better to make
changes in the early stages of design activity as soon as one realizes that there is a
flaw or shortcomings in decisions made in the early design cycle rather than persist
with a flawed design. This initial thinking process does not take much time.

2.1.2 Think in Solution-Neutral Environment!

Designers should select the problem in the customer domain and the FRs in the
functional domain in a “solution-neutral environment,” i.e., do not think about a
solution before defining the problem! Many people come up with “a solution” first
before determining the problem. In other words, one should not come up with a
solution and then think about the problem that fits the solution. Similarly, one
should not start to develop their design by copying their competitor’s product or
thinking about a solution first even before they commence the task of defining the
problem. Many industrial firms have difficulties in making their people think in a
solution-neutral environment. The “marketing groups,” which are charged to
develop broad guidelines for their desired new products, often “cheat” by studying
their competitors’ products and develop “specifications” for their new proposed
products based on what they have already seen in their competitors’ products. In
many cases, such an approach yields systems or solutions that are nearly identical to
its competitors’ or their old product. It is hard to be competitive or creative this
way!
40 N. P. Suh

A Story of Design Failure: The RIM Machine designed by an outstanding graduate

student and his professor
Identifying the real problem and creating a new design that addresses the identified
problem is not a trivial exercise, as one young professor learned the hard way! The
following story is a real story that happened at a well-known technological university.
Many professors enjoy working with bright students in research because students often
contribute to their research projects while also furthering their knowledge as well as
learning through their participation in research. In an ideal arrangement, professors and
students should become “partners-in-learning,” which implies that both the professor and
the student must be willing to learn new things while working on research. The professor
teaches the student how to define the problem and how to think about it. When teaching is
done through research, both the student and the professor learn. The professor has more
experience, and thus can teach the student how to think and how to approach the unknown
question. It also strengthens the university, since the involvement of students in research
makes in the university a live theater for education and research. Through research,
students learn to exercise independent thinking. Therefore, the young professor was
delighted when a new graduate student signed up to work with him! The student was not
only bright but also mature, having served for 5 years as an officer in the U.S. Army, after
graduating from West Point, the military academy of the United States. For his doctoral
research, the professor and the student jointly decided to develop a new machine for
producing polyurethane parts. A large industrial firm sponsored the project. Polyurethane
is one of the favorite materials in the automotive industry since it is easy to make com-
plicated parts such as bumpers and fenders with polyurethane as well as reducing the
weight of the vehicle—this reduction in weight results in the improved fleet gas mileage of
automobiles.
To make polyurethane parts, two liquids, i.e., diisocyanate and polyol, a viscous resin, are
mixed in a precise ratio to make solid polyurethane parts when these two components react
in a mold. Turbulent mixing of two liquid components occurs when two liquid streams
collide with each other at high speeds in a small chamber. The higher the impingement
speed, the smaller is the turbulent eddy size, thus better mixing. Commercial machines used
two precision gear pumps to deliver polyol and diisocyanate in a pre-set ratio at high flow
rates for impingement mixing in the mixing chamber. The mixture then flows into a mold
where the reaction is completed, forming a solid part. The fast-reacting mixture is then
injected quickly into a large mold to produce polyurethane parts when the chemical
reaction is completed. For his doctoral research, the student and the professor jointly
decided to develop a new machine for producing polyurethane parts, because the industrial
machines used precision gear pumps, which had to be replaced frequently due to wear.
The student worked very hard and was productive, thanks to his intelligence, personality,
and the experience he gained in the U.S. Army. He quickly designed and built the machine
and demonstrated its capability. The central idea behind the new design was to replace the
high-speed precision gear pumps with accumulators and inexpensive gear pumps. In this
newly designed machine, the resin and diisocyanate were pumped into two separate
accumulators with a rubber pouch (shown in Fig. 2.1). The resin was injected into the
accumulator slowly, using an inexpensive gear pump. A similar arrangement was also
made for diisocyanate. As the pouch expanded with the injection of resin taking up the free
2 What Is Design? 41

Fig. 2.1 Bag-type and piston-type accumulator. (Reproduced with permission from Casey (2009))

space of the accumulator, the gas pressure in the accumulator increased. When the gas in
the accumulator was compressed to a pre-set high pressure, the valve was opened to
discharge the resin at high speeds. A similar arrangement was made for diisocyanate. They
mixed in the impingement mix chamber and then discharged into the mold. These accu-
mulators eliminated the need for large, expensive gear pumps. The newly designed machine
seemed to work as intended. After a couple of years of research, the student, a technician,
and the professor jointly applied for a patent. The student received his Ph.D. after adding a
detailed analysis of his machine. All the inventors were very proud and hopeful about the
potential of their RIM machine. However, the industrial firm that sponsored the research
never adopted the device for their production! After all that hard work and confidence in
their design, it was very disappointing that the industrial people did not share their
enthusiasm!
The industrial sponsor found that the new machine did not perform as well as the old
industrial machine. It could not deliver the liquids in the set ratio, precisely and consis-
tently. The student and the professor tried to figure out why the device they designed with
such confidence did not perform as well as the commercial machine! It turns out that their
new design was a wrong design! The original problem was well defined, but a wrong
solution was proposed. All the hard work of the student and all the money the sponsor spent
did not yield the desired result. Now we know the shortcoming of this newly developed
machine: it violated the Independence Axiom. If they had known the design theory covered
in this book, they might not have made the mistake of proceeding with the wrong design!
Unfortunately, when this project was undertaken, the AD theory had not been developed.
Unfortunate timing?
42 N. P. Suh

Question 2.1:
Based on the preceding saga of the RIM machine development, what do you think were the
fundamental shortcomings of the design? Could you have known a priori that the design
had flaws? Would you have designed a better machine? Did the design violate any fun-
damental principles?

2.1.3 Design Theory: Should It Be Axiomatic or Algorithmic?

A few years after the retired U.S. Army captain graduated with his doctorate and
became a professor at another leading university, his former advisor began to think
about a more fundamental and systematic way of dealing with the design and
synthesis of systems. The basic theory-based design would be better than the
intuition- and experienced-based design practice. He concluded that
experienced-based design is a highly limited form of education and that university
may not be the best place to teach experience-based design. Better places for
learning experience-based design would be leading industrial firms with years of
design experience. However, there is a role for university-based design education if
universities can teach design based on fundamental design principles that can
generalize the characteristics of excellent designs.
He reasoned that there are two different ways of teaching design: axiomatic and
algorithmic.
In the algorithmic approach, we provide step-by-step instructions, which, if the
student follows it, the student will arrive at an answer, a la software algorithm.
Under an algorithmic design approach, one would design a system following a
step-by-step algorithm. An example of the algorithmic approach to teaching can be
found, for instance, in training “electricians.” In this case, a young person (who
might have graduated from a technical high school) works for a “licensed master
electrician” for a few years to learn the “trade” by following the footsteps of the
master electrician for a few years as an assistant. A similar approach is also used in
large companies to train their technicians, draftsmen/designers, and machinists
through apprenticeships. In these companies that have made similar machines for
many years, there is a system of designing a specific class of machines (that might
be called the algorithm), which was based on the knowledge gained while making
similar machines, with minor variations, for many years. The shortcoming of an
algorithmic approach is that it is challenging to develop so many algorithms for so
many different cases and products, especially in rapidly changing industries that
must introduce new systems to satisfy many different sets of FRs.
In the axiomatic approach, general design principles for design are created for all
designs that satisfy different sets of needs and requirements. The axiomatic
approach to design should apply to all designs regardless of the specific nature of
the system to be designed. Such general principles should be valid in all fields and
for all applications. The shortcoming of the axiomatic approach is that there are no
2 What Is Design? 43

algorithms that would yield the right answer if one were to follow it. In the axio-
matic approach, broad general principles are given without a product-specific
algorithm. If we had the design algorithms for the RIM machine, we might have
produced a machine similar to the then-existing commercial RIM machine. How-
ever, that algorithm would not have created a better RIM machine than the
then-existing industrial machines.
On the other hand, had we had the design axioms when we worked on the RIM
machine, we would have been cautious not to create the accumulator-based RIM
machine because it would have violated the design axioms! The RIM machine
developed without the benefit of design axioms “coupled” FRs. The three FRs—
function of maintaining a fixed ratio of two fluids, the function of providing
high-quality mixing, and the function of controlling flow rates—all depended on
pressure in the accumulators (which decayed as the fluid is being discharged) and
material-specific properties such as viscosity and temperature of each liquid.
Had the AD theory existed at the time the Ph.D. student worked on the RIM
project, they would not have developed the RIM machine that is based on the use of
accumulators. Design axioms eliminate the need for expensive and extensive testing
to find out the fundamental shortcomings of a system because the flawed design can
be identified at the conceptual stage based on generalizable axioms! Unfortunately,
even today, many companies make mistakes in system development by depending
on their experience and intuition! There are some well-known examples of failure
that have cost millions of dollars. These costly mistakes are a result of experien-
tially based or intuitive designs. Students who learn and practice rational design
presented in this book should not make these expensive mistakes!
AD is created based on the realization that all good designs always satisfy two
fundamental axioms regardless of the specific nature of the design or the field of
application. One has to do with the independence of FRs, and the other has to do
with simplicity (or minimum) of information contained in the design. AD is based
on the identification of common features that all good designs possess. The axioms
are stated as follows:
The Independence Axiom:
Maintains the independence of FRs.

The Information Axiom:

Minimizes the information content.

The meaning of these two axioms will be made more apparent throughout this
book. Many examples will be given to clarify the statements of these two axioms.
These two axioms were extracted by identifying the common elements that exist in
all good designs. It should be noted here that “cost” is not a FRs in nearly all
designs, but the design that violates the Independence Axiom can become extre-
mely costly, and many times, the system must be abandoned or redesigned and
built.
44 N. P. Suh

2.2 Design: Transformational Process to Create

Human-Inspired Systems

2.2.1 Definition of Key Words: Design, Goals, and Domains

The design consists of a series of the following actions and processes:

(a) identifying a problem that needs new or improved solutions to achieve a set of
goals;
(b) transforming the problem into a set of goals in the form of FRs to be achieved;
(c) fulfilling the goals with specific DPs within a given set of constraints;
(d) synthesizing the final embodiment in the form of hardware, software, systems,
organizations, and policies through the integration of DPs, preserving the
simplicity of the ultimate system; and
(e) maintaining the independence of FRs throughout the transformation process.

The above definition of “design” is comprehensive and applicable to many different

kinds of designs, ranging from systems, hardware, software, and organizations.
Many examples given in this book should clarify the significance of these
statements.

2.2.2 The Concept of Four Domains

The description of design given in Sect. 2.2.1 may be represented graphically as

shown in Fig. 2.2.
Figure 2.2 shows the four domains of the design world, i.e., the customer
domain, the functional domain, the physical domain, and the process domain1. We
define the “PROBLEM” in the customer domain (e.g., James Watt’s recognition of
the shortcomings of the Newcomen engine).

1
These four domains are applicable to all kinds of designs, including hardware, software, and
manufacturing processes. The particular names assigned to the domains are due to the fact that
when AD was first developed, most examples came from hardware- and manufacturing-related
activities. We are using the same names, although this general concept is applicable to many other
design tasks.
2 What Is Design? 45

mapping mapping mapping

{CAs} {FRs} {DPs} {PVs}

Customer Functional Physical Process

domain domain domain domain

Fig. 2.2 Four domains of the design world

2.2.3 Goals

One of the prominent words used in the definition of design is the word “goal or
goals.” We establish the goals in the customer domain after we identify the per-
ceived customer needs. After establishing the goals, we can identify the problem we
must solve to achieve the goals. Based on the problem identified, we can create FRs
in the functional domain that are consistent with the specific design goals. One of
the tasks of a designer is to come up with “means” of satisfying the FRs, which will
be called DPs in the physical domain.

2.2.4 Mapping from Domain to Domain: Functional

Requirements, Design Parameters, and Process
Variables

The problem defined in the customer domain is transformed into a set of FRs in the
functional domain, which represent the “GOALS” of the design (e.g., in both the
Watt engine and the Newcomen engine, one FR was for expansion of steam and
another FR was for condensation of steam). Once FRs are identified in the func-
tional domain, we identify and conceptualize DPs in the physical domain that can
satisfy the FRs of the functional domain. Then, in the process domain, we choose
PVs that can satisfy DPs or make DPs possible. We call this process of trans-
forming the problem into FRs, FRs into DPs, and DPs into PVs the “MAPPING”
between domains.
46 N. P. Suh

{FRs} in the functional domain may be considered as a vector with n compo-

nents that defines the problem. Similarly, {DPs} in the physical domain constitute a
vector, whereas {PVs} is a vector that defines the process domain2. Transforming
the FR vector to the DP vector represents the design process. Similarly, we are
designing the process when we transform {DPs} to {PVs}3.

Example 2.1 Illustration of the Mapping Concept (Shoe Soles for Children’s Shoes)
Background story: A recent immigrant to the United States just graduated from an engi-
neering school with a bachelor’s and master’s degree. He joined the largest shoe-machinery
company in the world, a successful and great company to work for, which was founded by
two families many decades ago in Massachusetts, U.S.A. One of the good things about this
company was that they did not do much defense-related work, and therefore even a recent
immigrant could get a job there since they did not require its employees to have a defense-
security clearance. They made nearly all the machines that are used in making shoes
worldwide.
Until a few years before this young engineer joining the company, they had a monopoly in
the shoe-machinery business by only leasing their machines rather than selling them. They
even built factories for shoe companies with their machines in them and collected a royalty
each time their machine was used to make shoes. They were active in research and obtained
many patents to protect their business. Other companies could not break into this
shoe-machinery business because of their business model. However, sometimes, good
things do not last forever! The United States government sued the company for practicing
monopoly and won. The company had to sell all their leased machines at a discount, which
gave the company lots of cash. This loss of monopoly ultimately resulted in the company
exploring ways of expanding its business under the new rules.
The problem (or the need) in the customer domain: The first assignment given to the young
engineer was to develop the process for manufacturing children’s shoe soles at a low cost.
The shoe soles had to meet several requirements. They had to be flexible, lightweight, and
wear-resistant. In the functional domain, the FR was to manufacture inexpensive shoe soles
for children’s shoes. In the process domain, he decided to use a variation of the commonly
used injection molding process. The idea was to directly injection mold the foamed PVC
(polyvinyl chloride) shoe soles to the upper (i.e., a stitched leather, which is stretched and
mounted on a last to create the three-dimensional shoe shape) to satisfy the FRs of the shoe
soles. In the process domain, he designed the manufacturing process that can yield the
product with specific DPs.
The goal in the functional domain was quite clear:

FR = manufacture foamed PVC soles for children’s shoes.

Mapping this FR in the physical domain (i.e., DP domain), DP may be stated as

2
The bracket {x} is used to denote a vector in this book.
3
Note that FR, DP, and PV are also written as FR, DP, and PV interchangeably throughout this
book.
2 What Is Design? 47

DP = Injection molded shoe soles.

We map this DP in the physical domain into a PV, which may be stated as

PV = Injection molding process.

The above set of FR, DP, and PV represent design decisions made at the highest level. This
young engineer’s task was to decompose these highest level FR, DP, and PV to
mass-produce the injection-molded shoes with PVC soles with desired properties at low
cost. To implement this design, we need more detailed designs. To produce detailed
designs, we will decompose them to generate lower level FRs, DPs, and PVs, which is
discussed in Sect. 2.3.4.

2.2.5 Design Axioms

We presented design examples from different fields in the preceding sections to

show how the design should be conceived and executed. In the examples presented,
we judged design based on the independence of FRs without much elaboration. We
will try to formalize what we have done in previous sections by presenting the
design axioms more formally.
Axiom is defined as
“The truths for which there are no counter-examples or exceptions.”

We use two design axioms to be sure that the design decisions we make are the
right ones. These two axioms were created based on the observation of common
features or elements that all good designs possess.4 Axioms, by definition, cannot
be derived. All students are familiar with some axioms; for example, in Euclid’s
geometry, they learned that the shortest distance between two points is the straight
line that connects these two points, which is an axiom because we cannot derive it,
but we cannot find counterexamples.
In the examples given in the preceding section, we defined FRs to solve the
problem stated in the customer domain and then selected DPs to satisfy the FRs.
How would we know whether or not we chose the most suitable set of DPs? In the
case of the shoe sole manufacturing, the example discussed in the preceding sec-
tion, we were fortunate that the process designed to make the composite shoe soles
with a foamed core and a skin layer worked as intended, although we did not invoke
design axioms. In this section, we present design axioms that we must satisfy in
creating rational designs more formally.
There are many questions that we must answer to be sure that our thought
processes on design are progressing correctly. We have to know if we selected the

4
For historical details, see Nam P. Suh, The Principles of Design, Oxford University Press, 1991.
48 N. P. Suh

right set of DPs for the given set of FRs. To be sure that our design can indeed
achieve the intended design goals, we must be able to answer the following
questions:
• Have we selected the right problem?
• How do we know we selected the right FRs and DPs?
• What are the fundamental requirements we must satisfy to be sure that we have
made the right design decisions?
• What constitutes a good design?
• How do we know that our design is excellent?
The design rationale presented in this book is based on two design axioms: the
Independence Axiom and the Information Axiom. These axioms were postulated by
asking the following question: “What were the common elements (or features)
that were present in all successful designs?”
These two axioms have been applied to a variety of different problems, the
design of hardware, software, organizations, manufacturing processes, complicated
systems, and others. Historically, axioms have provided the basis for scientific and
technological advances in many fields such as geometry and thermodynamics.
Newton’s laws at the time they were advanced might be considered to be also
axioms in that they were not based on scientific measurements of force and
acceleration, i.e., they were conceptual. Perhaps the following two questions will
remind the reader what axiom is and how axioms have advanced many fields:
Question 2.2:
Why is the straight line that joins two points the shortest distance?
The answer to the question is: In Euclidean geometry, an axiom states that the straight line
is the shortest distance between two points. We believe in the axiom because we cannot find
exceptions or counterexamples.
Question 2.3:
List three axioms taught in other subjects such as geometry and thermodynamics.

A design task may involve both synthesis and analysis. To do analysis, we invoke
certain natural principles such as Newton’s laws and thermodynamic laws. They
may be regarded as axioms. Axioms are defined as the evident truth for which there
are no counterexamples and exceptions. Of course, we apply all known natural
laws and principles to develop design solutions, in addition to the Independence
Axiom and the Information Axiom.
In AD, it was stated that there are two axioms: the Independence Axiom and the
Information Axiom. They are stated again here in declarative form as follows:
2 What Is Design? 49

The Independence Axiom

Maintain the independence of FRs.
The information Axiom
Minimize the information content.

These axioms were extracted by identifying the common elements that were always
present in good designs, but not present in bad designs. This entire book is devoted
to the use and application of these two axioms. They are applied to a variety of
different tasks in many different fields.

Example 2.2 Water Faucet Design

Given the two faucets shown below that control the flow rate of water and the temperature
of water, which is a better design?
We cannot answer the question until we know what the FRs of the design are. If we define
the FRs to be the following:
FR1 = Control temperature;
FR2 = Control flow rate.
The faucet shown in (a) has two valves, one for cold water, and one for hot water, i.e.,

DP1 = Hot water valve;

DP2 = Cold water valve.
In this design, when one of the valves is turned to change the temperature of water flowing
out the faucet, the flow rate changes as well. Thus, it violates the Independence Axiom. In
the faucet shown in (b), the up-and-down position of the lever only changes the flow rate
and turning the lever left and right only changes the temperature. The design shown in
(b) maintains the independence of FRs, and thus is a better design.

Fig. 2.3 Two water faucets: a two valves, one for cold water and the other for hot water, b One
valve with two independent controls for flow rate and for water temperature. (Reproduced with
permission from Kurguzova (2019a, b))
50 N. P. Suh

Example 2.3 Design of Cutting Tool

A variety of cutting tools are used for machine steel and other metals. Cutting speeds are
affected by the quality of cutting tools. Industrial productivity is directly related to cutting
speed, which is often limited by the wear of cutting tools. However, besides the wear of
cutting tools, they must be able to withstand the cutting force without deflection and
fracture. If the task is to improve the quality of cutting tools, what are the appropriate FRs
that must be satisfied to improve the performance of the cutting tools? What are the DPs
that can satisfy the FRs?
The FRs of a cutting tool may be stated as follows:

FR1 = Provide wear resistance;

FR2 = Prevent fracture;
FR3 = Have sufficient stiffness.
Many cutting tools were made of a single material such as high-speed steel, cemented
carbides, and ceramics. Therefore, FR1, FR2, and FR3 could not be independently satisfied.
These FRs must be then mapped in the physical domain by finding DPs, i.e., DP1, DP2,
and DP3, which may be stated as

DP1 = Tool material;

DP2 = Tool material;
DP3 = Thickness of the cutting tool.
Since the same DP controls FR1 and FR3 because DP1 = DP2, we cannot satisfy the FRs
independently, thus violating the Independence Axiom. As a consequence, a material that
has an excellent wear resistance may lack toughness. Conversely, if we select a material
with toughness, it may wear too fast. This is not an ideal tool design.
How should we improve the performance of cutting tools by satisfying the three FRs
independently?

Examples 2.1 through 2.3 are given earlier to illustrate the concept of mapping from
the functional domain to the physical domain. The question now is: are those
designs good as per the Independence Axiom? A convenient way of deciding
whether or not the proposed design solution is consistent with the independence
Axiom is to examine the design matrix.
Note that design involves both synthesis and analysis, which reinforce each
other. In the early stages of design, the detailed analysis may not be feasible or
necessary until the design details emerge from conceptual design. The analysis
becomes more critical after the overall initial conceptual design has been com-
pleted, although analytical thinking is always essential throughout the design
process. However, a coupled design cannot be improved through analysis!
2 What Is Design? 51

2.2.6 Design Matrix [DM]

The relationship between FR1 and FR2 of the water faucet may be written as

FR1 ¼ A11 DP1 þ A12 DP2

ð2:4Þ
FR2 ¼ A21 DP1 þ A22 DP2

One way of representing the relationship between FRs and DPs is to construct a
matrix between FRs and DPs. The water faucet shown in Fig. 2.3a is a coupled
design, which has the following design matrix:

Valve 1 Valve 2
FR1 (control the temperature of water) X X
FR2 (control the flow rate of water) X X

Since valve 1 and 2 affect both FR1 and FR2, it is a coupled design, violating the
Independence Axiom.
We can represent Eq. (2.4) and the above table as follows:

fFRsg ¼ ½DMfDPsg ð2:5Þ

{FRs} is a vector, consisting of two components: FR1 and FR2. Similarly {DPs} is
a vector, consisting of two components: DP1 and DP2. [DM] is called the “Design
Matrix.”
To satisfy the Independence Axiom, the design matrix [DM] of a system must be
a diagonal or a triangular matrix. From a mathematical point of view, when the
matrix is diagonal, the governing equation is a set of the one-input one-output
problem regardless of the number of FRs and DPs involved, which can be solved
without much difficulty either analytically or numerically. When [DM] is a trian-
gular matrix, we can get solutions to the design problem, if we vary DP in the
sequence given by the triangular matrix. When [DM] is a full matrix, we have to
find a unique (or compromised) solution, which is difficult to solve mathematically
or evaluate and calibrate. As a result, even when a unique sweet spot for the design
is found, the design requires tight tolerances to work correctly. Furthermore, as the
system wears out, it can fail more readily than an uncoupled system.

2.2.7 Design Equation

Equation (2.5) relates the output, i.e., {FRs}, to input, i.e., {DPs}. A similar
equation that relates vector {DP} to input vector {PVs} can be written. These will
be called the design equations. The solution to these design equations is simple to
52 N. P. Suh

solve if the design is uncoupled since there is only one output and one input
variable. When the design is coupled, we have to solve many equations simulta-
neously, which may not be solvable. If the design equation is non-linear but
uncoupled or decoupled, it can be solved either analytically or numerically. When
the design is coupled, an analytical solution would be difficult to obtain since many
equations must be solved simultaneously.
In practical terms, the following should be noted as a reminder to all designers.
When there are n FRs to be satisfied, one may have n equations with n DPs. When
the design is coupled, the design matrix is going to be a full matrix, and one may
have to solve n simultaneous equations for a unique solution, which is a difficult
task, especially when the equations are non-linear. Even if one can solve it, a slight
change in one of the DPs will throw the entire system into chaos. On the other hand,
if the design is uncoupled, each FR can be solved by itself, regardless of whether
the equation is linear or non-linear.
Many systems we design have many FRs. If the design is coupled, the imple-
mentation of the system is likely to bog down, leading to system failures. The
reason an international airport cannot be made to operate even after spending
billions of more dollars and years of delay is a symptom of this problem, i.e., a
coupled design.

2.2.8 Decomposition of Functional Requirements, Design

Parameters, and Process Variables Through Zigzagging
Between the Domains

In Sect. 2.2.1, we discussed the idea that in the design world, we have four
domains: customer, functional, physical, and process domains. For the shoe sole
example given in Example 2.1, we defined the FR in the functional domain, DP in
the physical domain, and PV in the process domain. These were the highest level
characteristic vectors that define each respective domain. When further details are
needed for implementation, we must decompose FRs, DPs, and PVs further to
generate detailed designs that can be implemented.
To decompose, we have to zigzag between the domains. For instance, after we
define FR1 in the functional domain, we move to the physical domain to select the
corresponding DP1. If the selected DP1 does not have sufficient details that can be
implemented, we must return to the functional domain, and select children-level
FRs, i.e., FR1.1, FR1.2, etc., by decomposing the highest level FR1. These
children-level FRs may be thought of as the FRs of DP1. Figure 2.5 illustrates the
zigzagging between the domains (Fig. 2.4).
After FR1.1 and FR1.2 are chosen, we need to go back to the physical domain
and select corresponding DPs, i.e., DP1.1 and DP1.2, which are the children of the
highest level DP. This zigzagging process must continue until sufficient details of
the design that can be implemented are generated. This process is illustrated in
Example 2.5.
2 What Is Design? 53

Fig. 2.4 Zigzagging

between domains to FR1 DP1
decompose FRs and DPs
FR1.1 FR1.2 DP1.1 DP1.2

FR1.2.1 FR1.2.2 DP1.2.1 DP1.2.2

Functional Physical
Domain Domain

Example 2.1 (cont’d)

Continuation of the Decomposition of the Children’s Shoe Sole Design
(To illustrate decomposition of FRs, DPs, and PVs through zigzagging.)
Going back to the example related to children’s shoes, we completed the design at the
highest level as follows:

FR = Manufacture foamed PVC soles for children’s shoes;

DP = Injection molded shoe soles;
PV = Injection molding process.
These highest level FR, DP, and PV are rather conceptual without details, and therefore
cannot be implemented without further details. In order to generate the details, we have to
decompose FR, DP, and PV. The decomposition of FR, DP, and PV creates in each domain
a tree-like structure with lower level FRs, DPs, and PVs.
The highest level FR may be decomposed as follows:

FR1 = Mold the entire shoe sole onto the upper;

FR2 = Make the sole flexible, i.e., pliable;
FR3 = Make the sole wear-resistant;
FR4 = Control the weight (i.e., density) of the sole.
Every designer might have decomposed the FR differently, resulting in different products
that may perform the same highest level FRs.
The corresponding DP in the physical domain may be chosen to be the following:

DP1 = Injection molding against the upper mounted on last;

DP2 = Density of foamed PVC;
DP3 = Solid skin layer for wear resistance;
DP4 = Total weight of the PVC injected into the mold.
In Fig. 2.5, the cross section of shoe soles is shown. DP1, the injection molding, controls
the shape of the molded shoe soles. To make the sole flexible (FR2), foamed PVC (DP2) is
used by mixing PVC with a chemical blowing agent (which decomposes in the plasticating
extruder section of the injection molding machine) to generate gas when it reaches the
decomposition temperature of the blowing agent. To make the shoe sole wear-resistant,
PVC in contact with the movable mold surface was prevented foaming by keeping the mold
54 N. P. Suh

Fig. 2.5 A shoe sole (Boot reproduced with permission from Alexlukin (2019))

surface in contact with PVC cold. The flow front of PVC tends to foam and expand because
of the low pressure at the flow front. Therefore, to prevent the expansion of the molten PVC
while the molten plastic flows into the mold, the cross-sectional area of the mold was made
very narrow by making the bottom of the mold (which is in contact with the bottom of the
shoe sole) moveable. To increase the pressure on the plastic during injection and thus
prevent it from expanding when the blowing agent decomposes, the cross-sectional area of
the flow channel was narrowed by moving in the movable part of the mold. After the mold
cavity is filled, the movable part of mold (which is in contact with the bottom of shoe soles)
was moved out to expand the mold volume, and thus allowing the plastic to expand forming
tiny bubbles everywhere, except the bottom of the shoe sole. During this expansion process,
the bottom of the PVC sole in contact with the cold mold surface cannot expand because of
the low temperature of PVC, thus forming a solid skin layer. To bond the molded shoe sole
directly to the shoe upper mounted on the last, an adhesive is applied to the shoe upper
mounted on the last, before the injection of PVC (Fig. 2.5).
Now that DPs are selected to satisfy FRs, we can check the relationship between FRs and
DPs. One way of checking it would be to use a matrix relationship. A means of representing
the relationship is to ask: “Does DP1 affect FR1? FR2? And FR3?” If it does, we put down
X, and if not, 0.

DP1 DP2 DP3 DP4

FR1 X 0 0 0
FR2 0 X 0 X
FR3 0 x X 0
FR4 0 0 0 X

By rearranging the above sequence of FRs and DPs, we obtain

2 What Is Design? 55

DP4 DP2 DP3 DP1

FR4 X 0 0 0
FR2 X X 0 0
FR3 0 x X 0
FR1 0 0 0 X

It is a triangular matrix. According to the Independence Axiom, we need to maintain the

FRs independent from each other, but since it is a triangular matrix, FR3 is affected by DP2.
It states that the density of the foam core can affect the wear rate after the skin layer wears
off. Therefore, we need to change the DPs in the right order to maintain the independence of
FRs. In this case, we first control the foam core thickness for flexibility and then determine
the skin layer thickness for wear resistance. We call such a design that requires a sequential
change of DPs a “decoupled design.” In this design, the sequence of change of DPs should
be in the order of DP4, DP2, DP1, and DP3. Then the design matrix becomes triangular.
Note that we purposely chose the same number of DPs and the number of FRs for a good
reason. It can be shown that in an “ideal design,” the number of FRs and DPs must be the
same (Theorem 4).
As the foregoing paragraph stated, the PVs were as follows5:

PV1 = Injection of PVC with chemical blowing agent;

PV2 = Expansion of the mold volume by moving the bottom plate;
PV3 = Temperature of the mold that comes into contact with the bottom of the shoe sole;
PV4 = Expansion of the mold volume by moving the bottom mold plate.
In Fig. 2.5, the cross section of shoe soles is shown. DP1, the injection mold, controls the
shape of the molded shoe soles. To make the sole flexible (FR2), foamed PVC (DP2) is
used by mixing PVC with a chemical blowing agent (which decomposes in the plasticating
extruder section of the injection molding machine) to generate gas when it reaches the
decomposition temperature of the blowing agent. To make the shoe sole wear-resistant, a
solid layer of PVC skin at the bottom of the shoe sole is created by making the mold surface
that is in contact with PVC cold to prevent the expansion of PVC by the blowing agent. To
prevent the expansion of the PVC during the injection of plastic into the mold, the cross
section of the mold was made very narrow to build up the pressure on the plastic with
blowing. After the mold is filled, the bottom part of the mold that is in contact with the
bottom of shoe soles was moved to expand the mold volume, thus allowing the expansion
of the plastic with the decomposed blowing agent. This process creates a foamed shoe sole
with a solid skin layer. The adhesive is applied to the shoe upper mounted on the last,
which bonds the molded shoe sole directly to the shoe upper mounted on the last. It should
be noted that this project was finished in 3 months with field tests after the young engineer
took over the project that had been going on for a couple of years, all thanks to the rational
design of the process.

5
The PVs can be selected if one has some knowledge on injection molding of foamed PVC. This is
a rather specialized knowledge that must be acquired either by reading reference books or through
actual practice. A special molding technique was invented to achieve PV1, PV2, and PV3, which
was named the “high-pressure injection molding.” It will be described in Chap. 7
56 N. P. Suh

Example 2.4 Illustration of the Mapping Concept (Global Warming)

In this example, the goal is to keep the temperature rise of Earth to within 2°C above its
temperature at the time of the Industrial Revolution6. The problem we identified to solve in
the customer domain is global warming to slow down the temperature rise of Earth. Global
warming is primarily due to the anthropogenic emission of CO2. Now we need to “map”
this need to reduce CO2 into the functional domain and we need to define a set of FRs in the
functional domain to reduce CO2 emission.
In the functional domain, we may establish the following FRs as a means of solving the
problem identified in the customer domain:

FR1 = Replace internal combustion (IC) engines with electric drives;

FR2 = Generate electricity with solar energy;
FR3 = Generate electricity from wind power;
FR4 = Plant trees.
Note that the designer chooses the FRs. Other designers may choose a different set, which
may be larger or small than the four FRs listed above. The quality of the ultimate design
will depend on the choice of FRs. Each designer will attempt to solve the problem identified
in the customer domain in different ways. Therefore, in design, there is no unique solution,
i.e., uniqueness theorem does not hold. There may be many equally acceptable solutions as
long as they satisfy the same set of FRs. We will later cover the Information Axiom, which
may be used to evaluate two equally good designs from the viewpoint of the Independence
Axiom, but one particular design may be superior to the other from the simplicity point of
view, which is what is measured by the Information Axiom.
To satisfy the four FRs identified above, we have to come up with DPs in the physical
domain (i.e., transform FRs into DPs). How many DPs should we select to satisfy the four
FRs chosen? What would you pick as DP1, DP2, DP3, and DP4? Should we have more (or
less) than four DPs? What should be the relationship between the FRs and DPs? How
should we represent the relationship between FRs and DPs? These questions may be
difficult to answer at this stage of learning AD, but we will answer all these questions later
in this book.
Different designers may select a different set of DPs to satisfy the FRs. Suppose that a
designer selected the following set of DPs to satisfy the FRs chosen:

DP1 = Electrical ground transportation system;

DP2 = Solar cell “farm” for the generation of electricity;
DP3 = Windmills along the shore to generate electricity;
DP4 = Designated national tree planting day.
Sometimes it is hard to distinguish FRs and DPs. It is suggested that always state FRs
starting with a verb and DPs with a noun.
Once we select DPs, FRs and DPs are related as

fFR1; FR2; FR3; FR4g ¼ DM fDP1; DP2; DP3; DP4g ð2:6Þ

The FR vector, i.e., {FRi} is related to the DP vector, i.e., {DPi} by the design matrix DM.

6
On December 12, 2015, 195 nations at the United Nations Conference on Climate Change in
Paris (Conference of Parties COP21) agreed to reduce their CO2 emission to limit the rise of the
earth’s temperature to below 2 C above the pre-industrial level.
2 What Is Design? 57

Equation (2.6) may also be expressed as

FR1 ¼ A11 DP1 þ A12 DP2 þ A13 DP3 þ A14 DP4

FR2 ¼ A21 DP1 þ A22 DP2 þ A23 DP3 þ A24 DP4
ð2:7Þ
FR3 ¼ A31 DP1 þ A32 DP2 þ A33 DP3 þ A34 DP4
FR4 ¼ A41 DP1 þ A42 DP2 þ A43 DP3 þ A44 DP4

In indices notation, Eq. (2.7), may be written as

X
FRi ¼ j
Aij DPj ; where i and j are 1; 2; 3; 4 ð2:8Þ

{FRs} and {DPs} are vectors, and DM is a matrix that relates these two vectors. The nature
of DM determines whether or not the proposed design at this highest level is acceptable or
not. If the design violates the Independence Axiom, it is not acceptable. The Independence
Axiom states that the FRs must be independent of each other.
A straightforward way of checking the independence is to construct the design matrix and
check if a DP affects FRs. For example, does DP1 affect FR4? If the answer is affirmative,
we put down X and if not, 0.

DP1 DP2 DP3 DP4

FR1 X 0 0 0
FR2 0 X 0 0 ð2:9Þ
FR3 0 0 X 0
FR4 0 0 0 X

The designer decides whether there is a relationship between a DP and an FR, making
certain assumptions. In the above example, it was decided that DP2 (solar power) does not
affect FR1, because, in the early days of installing the electric car system, they will rely on
the power sources currently used. Eventually, if the entire electrical power generation is
done using solar, the power generation may affect the transportation system.
The above design represented by Eq. (2.9) has a diagonal matrix, and therefore each FR can
be changed by varying one DP without affecting any other FRs. Such a design is much
easier to manage and optimize than when some of the off-diagonal elements are non-zero.
This kind of design is called “uncoupled design,” which is the most desirable design.
Although there are many FRs, each one of them can be treated as a one-input/one-output
problem, which makes it much easier to optimize, control, and operate.
Once we define DPs, we have to map them in the process domain and identify means of
satisfying the DPs. We call them PVs. For instance, PV1 should provide a means of
enabling DP1. Possible PV1 could be “rechargeable batteries” or “wireless electric power
transmission.”
Assembly of DPs into a physical system
DPs must be assembled into a whole system. In the case of solid components, this process
may be aided by determining the interaction between DPs to put them into geometric
proximity during assembly of the parts, if the design involves hardware. For this, we can
determine the interaction between DPs, creating a matrix similar to that given by Eq. (2.10).
Such a matrix is called the geometric proximity matrix (GPM) or design structure matrix
58 N. P. Suh

(DSM). The table states that there are physical interactions when the element of the matrix
is non-zero. For example, DP2 and DP4 may be close together physically. It does not imply
that there is a functional relationship between FR2 and FR4. Sometimes, it takes a great
deal of effort to package all the components that go into the automobile engine compart-
ment, which is limited in space.

DP1 DP2 DP3 DP4

DP1 a c e g
DP2 c d 0 h ð2:10Þ
DP3 e 0 n f
DP4 g h f m

Given the GPM, DP1 should be surrounded by DP2, DP3, and DP4, if there is appropriate
space. Similarly, DP4 can be placed in close proximity of other DPs. DP2 and DP3 do not
need to be placed in close proximity.

Example 2.5 Illustration of the Mapping Concept (Mass production of glass bottles)
Suppose you are in charge of developing a mass manufacturing system for the production
of glass bottles by blow molding molten glass gobs. Molten glass runner at about 1100 C
(note: runner is sometimes called parison in plastics blow molding industry) is extruded as
the round rod, which must be cut to gobs using a special cutting tool which shears the glass
runner (the shearing tool is similar to scissors with two blades). The gobs are then inserted
into a mold where it is molded into a bottle by blowing air into the gob, i.e., similar to blow
molding of plastic bottles. After the glass bottle solidifies in the mold, it is transported away
from the machine. The desired production rate is 100 bottles per hour. The cutter must cut
the glass parison within a tight tolerance without leaving any residues of glass particles on
the parison surface, which, if left on the surface, would then show up as defects on the
bottle surface. The problem we are concerned with is that the shearing tool must be replaced
frequently due to the wear of the cutting tool, which slows down the production rate and
create defective parts. Our job is to design the entire system for shearing off the glass
runner.
Determining the highest level FRs:
Our job is to improve the glass/gob cutting operation to minimize the wear of the shearing
tool to cut the parison. We begin by asking: “What are the FRs of the cutting operation?”
Each designer may come up with a different set of FRs based on the designer’s under-
standing of the problem in the customer domain.
Suppose that we chose the following FRs:

FR1 = Control the diameter of the runner;

FR2 = Measure the length of the runner where it must be cut;
FR3 = Shear off the runner when the runner reaches the pre-set position;
FR4 = Lubricate the cutting operation;
FR5 = Control the temperatures of the glass, lubricant, and the cutting tool.
We will know whether or not this set of FRs is complete after we complete the design by
choosing the right set of DPs. In some cases, the final verdict will come when the actual
shearing operation is done to test the design. Even if the above five FRs are not complete,
2 What Is Design? 59

this systematic approach will enable us to spot any mistakes made. Our experience shows
that in most cases, design done this way is complete and performs the intended functions
well.
Mapping of FRs into DPs at the Highest Level:
Having defined the FRs of our design, we need to map the FRs into the physical domain by
identifying the corresponding set of DPs. As we try to select the corresponding DPs, we
should note that the control of temperature is critical, because the glass has a narrow glass
transition temperature where the glass behavior changes drastically from a viscous–elastic
state into a glassy state. As an initial attempt to designing a right shearing tool for the
molten glass, we may select the following DPs to satisfy the FRs chosen:

DP1 = Diameter of the extrusion die opening;

DP2 = Optical gage;
DP3 = Mechanical shear with sharp blades;
DP4 = Water-based lubricant with surfactants, corrosion inhibitors, and water-soluble oil;
DP5 = Temperature controllers for glass, machines, and shearing blades.
Note that each one of these FRs and DPs must be further decomposed to develop further
details since DPs cannot be implemented for lack of detailed designs.
Also, note again that FRs are stated starting with verbs and DPs begin with a noun in order
to easily identify FRs and DPs.
Design Matrix
For the chosen FRs and DPs, we must construct the design matrix and investigate if the
proposed design at this highest level satisfies the Independence Axiom, i.e., is it an
uncoupled or decoupled design. Design matrix links the {FR} vector to the {DP} vector.
The design matrix is shown below, which indicates that the design at this level is an
uncoupled design:

DP1 DP2 DP3 DP4 DP5

FR1 X 0 0 0 0
FR2 0 X 0 0 0
FR3 0 0 X 0 0
FR4 0 0 0 X 0
FR5 0 0 0 0 X

Decomposition
This design at this level does not have sufficient details, except DP1 and DP4. Therefore,
other FRs and DPs need to be further decomposed. Ultimately, it should show all the
detailed designs through continuous decomposition until the implementable details are
designed. This subject is discussed further in a later chapter.
60 N. P. Suh

Example 2.6 Money Circulation for Economic Growth

In the customer domain, the problem we identified is the slow economic growth of a small
country in Asia. Our goal is to solve the problem through a new design of its taxation and
economic policies.
Some economists seem to believe that one way of promoting the economic growth of a
nation is to increase the circulation of money (sometimes called the velocity of money
circulation). The rationale seems to be the following: When the money circulates, people
with money either buy or sell goods and services. Thus, the faster the circulation money,
the greater is the economic activity, and hence faster economic growth. The question is how
a taxation system should be designed to increase the speed of money circulation. We will
assume this assumption to be accurate and determine how we may develop economic
policy based on AD.
Poor people spend all their income but still cannot buy all the things required for a decent
living because they make less than what is needed to maintain “middle-class life.” People in
the top 1% spend as much money as they please, but they cannot use all of their income,
because their income is far higher than the amount of the money they need to maintain their
luxurious lifestyle, and therefore they circulate only a small fraction of their annual income.
Rich people may directly deposit their money to collect interest on the money, which may
not affect money circulation in a significant way. Your task is to transform the economy of
this nation. We need to transform the problem identified in the customer domain into a set
of FRs in the functional domain and a set of DPs in the physical domain in order to design
an economic system that will maximize the economic growth rate of the country by
increasing the velocity of money circulation, assuming that this is sound economic policy.
Your task as a renowned economist is to transform the problem we identified in the
customer domain into a set of FRs in the functional domain and a set of DPs in the physical
domain. The goal is to transform the economic system and accelerate the economic growth
rate of this country by increasing the velocity of money circulation.
The following facts about this country are known: There are five income groups: A, B, C,
D, and E. Group A is the lowest income Group; B the second lowest income group, which
makes twice as much as those in Group A; Group C makes three times that of B; Group D
makes five times that of Group C; and Group E makes 100 times that of Group D. Income
of Group C is just enough to support a typical family. The population distribution in each
one of the economic income groups is as follows: A = 100,000; B = 5,000,000;
C = 1,000,000; D = 50,000; and E = 300. People in Group C spend all the money they
make to take care of their minimum expenses. Groups A and B need more money to
achieve middle-class living quality but do not have the necessary income. The goal is to
design fiscal and monetary policies through design to maximize the velocity of money
circulation.
The new tax system will be designed to achieve the following three FRs:

FR1 = Maximize the velocity of money circulation;

FR2 = Invest in public infrastructure, including national defense;
FR3 = Provide incentives to all groups to work efficiently and effectively.
2 What Is Design? 61

The DPs may be stated as follows:

DP1 = Monetary policies;

DP2 = Investment policies for a public good;
DP3 = Fiscal policies.
These highest level FRs and DPs must be decomposed to develop detailed designs that can
be implemented, because FR1, FR2, FR3 and DP1, DP2, DP3 are still conceptual and do
not provide sufficient details for us to implement.
To achieve the goal of having a strong economy, we need more details than the design done
so far. We will decompose FR1, FR2, and FR3 as well as DP1, DP2, and DP3. Different
designers may come up with different sets of lower level FRs, depending on how they
decompose them based on their understanding of the problem and issues involved.
We decided to choose the following children-level FRs to address the problem of maxi-
mizing money circulation, i.e., FR1 (maximize the velocity of money circulation):

FR11 = Give subsidies and tax credits to Groups A and B;

FR12 = Establish a progressive income tax rate for Groups C, D, and E;
FR13 = Reduce the tax rate on the net income that exceeds the preceding year’s income;
FR14 = Levy fixed % tax on all income groups for healthcare;
FR15 = Levy fixed % tax on all income for social security.
For FR2 (invest in public infrastructure, including national defense), we selected the fol-
lowing as the next-level FRs:

FR21 = Invest in infrastructure building;

FR22 = Fund national defense;
FR23 = Provide free education from primary to tertiary education;
FR24 = Invest in 5% of GDP in extensive R&D;
FR25 = Strengthen patent policies;
FR26 = Provide free public transportation.
For FR3 (provide incentives to all groups to work efficiently and effectively), we decided to
choose the following as the next-level FRs:

FR31 = Provide free education for primary, secondary, and tertiary education;
FR32 = Legislate a minimum pay policy that is tied to the inflation rate;
FR33 = Promote merit-based reward system;
FR34 = Implement a generous overtime pay policy.
To satisfy these FRs, we need to select the means of satisfying them by choosing the right
set of DPs in the physical domain. Again note that DP1 is chosen to satisfy FR1, and so on.
One possible set of DPs that are chosen to satisfy the children-level FRs listed above are as
follows:

DP11 = Special tax incentives for Groups A and B;

DP12 = Progressive tax rate for Groups C, D, and E;
DP13 = Tax incentive system for exceptional achievers;
DP14 = Healthcare tax;
DP15 = Social security tax;
DP21 = Infrastructure fund;
62 N. P. Suh

DP22 = Defense budget;

DP23 = Strong public educational institutions;
DP24 = Competitive R&D funding;
DP25 = Patent policy for public good;
DP26 = Public infrastructure;
DP31 = Public education system;
DP32 = Mandatory minimum pay system;
DP33 = Civil servant examination systems;
DP34 = Overtime pay legislation.
Questions 2.4:
How do we know that we have selected good DPs? Do the selected DPs constitute an
acceptable set of DPs? What is the role of the Information Axiom in designing these
systems?

2.3 Design Process in Creating Various Systems

To summarize the design process:

The first step is to have a clear understanding of the problem that requires a
design solution (in the customer domain). From this problem statement, we
establish design goals in the form of FRs. After the designer chooses FRs, the DPs
must be selected that can satisfy the FRs. Then, to be sure that the DPs are
correctly chosen, we create the design matrix (DM) to check for the independence
of FRs. Then, we assemble DPs into a system for ease of use, physical compati-
bility, physical contiguity, geometric compatibility, transfer of forces, electrical
compatibility, etc. In other words, in the case of physical products, DPs must be
“packaged” into a physical assembly, e.g., the engine compartment of cars is
packaged to house most mechanical components such as engine, electricity gen-
erator, air-conditioning compressor, etc., to fit under the hood.

2.3.1 Revisiting the Water Faucet Design

Example 2.2, two water faucet designs were shown to ask which design is superior
in satisfying two FRs, one controlling the water flow rate and the other for con-
trolling temperature. We stated that the design that satisfies the temperature and the
flow rate independently is a superior design as per the Independence Axiom. We
will now go back to the water faucet design and discuss the actual design process.
The FRs for the water faucet may be stated as follows:

FR1 = Control water flow rate = Q;

FR2 = Control temperature = T.

The desired design is an uncoupled design that will enable us to write the design
equation as
2 What Is Design? 63

      
FR1 Q A11 0 DP1
¼ ¼ ð2:11Þ
FR2 T 0 A22 DP2

We should try to choose DP1 and DP2 that will give us a diagonal design matrix.
Then, we can proceed to determine what the coefficients A11 and A22 should be in
order to satisfy FR1 and FR2.
One way of achieving the independence of these two FRs might be to control the
cross-sectional areas of the valve through which hot water and cold water flow. If
Ah is the cross-sectional area of the valve through which hot water flows through
and Ac is that for cold water, then Q and T are functions of Ac and Ah as

Q ¼ f ðAc þ AhÞ
T ¼ gðAc=AhÞ

Thus, Eq. (2.11) may be written as

         
FR1 Q X 0 DP1 X 0 f ðAcþ AhÞ
¼ ¼ ¼ ð2:12Þ
FR2 T 0 X DP2 0 X Ac
g Ah

Equation (2.12) states that if FR is to control water flow rate, we vary f(Ac + Ah),
and if we want to vary the water temperature we vary g(Ac/Ah). A valve design
must be such that it will control the total flow rate of water without affecting the
temperature of the water and vice versa. To achieve the design goal, we need to
develop a mechanism that will let the flow rate change without affecting temper-
ature by keeping the ratio of cross-sectional areas remain the same, while the sum of
the cross-sectional areas changes. Conversely, it must be able to change the ratio of
the cross-sectional areas without affecting the total flow rate.
Now we must develop the design that will define DP1 and DP2 more explicitly.
The design question now becomes what kind of mechanism will provide us DP1
that will let us vary the (Ac + Ah) without affecting (Ac/Ah) and DP2 that will
affect (AC/Ah) without affecting (Ac + Ah).
One of the mechanisms that will let us achieve these design goals is shown in
Fig. 2.6
Then, the design equation, Eq. (2.11), may be written as
         
FR1 Q X 0 DP1 X 0 /1
¼ ¼ ¼ ð2:13Þ
FR2 T 0 X DP2 0 X /2

Equation (2.13) verifies that the water faucet is an acceptable design because the
DM is a diagonal matrix.
64 N. P. Suh

Fig. 2.6 Uncoupled water

faucet design (©Oxford
Publishing Limited.
Reproduced with permission
of the Licensor through
PLSclear (Suh 2001))

2.3.2 Design of a System for Salary Raise

Many organizations, including universities, give annual salary raise to their

employees. It is in recognition of the contributions they have made to the organi-
zation during the previous 12 months and also to adjust for the inflation of living
costs. The central administration typically allocates a lump sum for each unit of the
organization with the approval of the board of directors (or trustees). Each group
then distributes to its employees by the head of the unit based on an established
rationale. The salary administration must be fair to the employees as well as for the
organization to achieve the goal of improving performance and rewarding expected
contributions. Our job is to design a salary raise system that performs these tasks.
Design Task:
PROBLEM to be solved through design (in the customer domain)
Usually, the available pool of funds for the salary raise is not sufficient to give a
significant increase to everyone. Furthermore, the organization must recognize
those who have made significant contributions to achieving the goals of the orga-
nization. Therefore, those who have made more significant contributions to the
organization should receive a higher raise. Such a system may be called a
merit-based system. However, those who think they received smaller raises than
they deserve can criticize the system. One temptation for the administrator to avoid
2 What Is Design? 65

criticism is to give a uniform raise based on their current salary, which violates the
goal of advancing the organizational goals through a merit-based reward system.

GOALS of the salary distribution (in the functional domain)

Having defined the problem, we need to define the goals we have to achieve
through design.

FR1 = Assess the contributions made by each employee;

FR2 = Reward based on specific contributions.

Implementation of the goals (DPs of the salary system)

DP1 = Merit-based salary system consisting of six groups of merit;
DP2 = A Gaussian distribution, giving the two times the average raise to the best
performing and 20% of the average raise to the least effective
employees.
Constraint: To determine the average raise, the total sum of the raises to be given must be
the same as the available fund to be distributed.

2.4 Role of System Architect in Managing Large System

Development

When the system to be designed is vast and complicated (for example, the on-line
electric vehicle), many people participate in the execution of the project. To manage
such a large system development with many FRs, it is necessary that a system
architect is appointed. The primary job of the system architect is to make sure that
any member of the design and management team introduces no coupled design.
Since there may be a large number of FRs and DPs with many layers of
decomposition, different groups of people are in charge of different subsets of the
project. When they make design decisions, they tend to make them by considering
their tasks alone. However, their judgment may affect the design decisions made in
some other branches of the design team. These mistakes usually increase the cost of
the project substantially and lengthen the development period. Therefore, someone
has to monitor whether or not the decisions made by everyone working on the
project are proper from the systems point of view, i.e., prevent the inadvertent
introduction of DPs that couple FRs of the system. Such a person is the system
architect.
The role of the system architect is to construct the master design matrix for the
entire project and check if any design decisions have inadvertently introduced the
coupling of FRs by any participant in the design project. To achieve this goal, the
design architect must construct a master design matrix for the entire system and
must follow all the branches of design and the rationality of the design decisions
66 N. P. Suh

made by checking for the inadvertent coupling of FRs. When KAIST7 developed
the on-line electric vehicle (OLEV) and mobile harbor, system architect teams
monitored the decisions made by various team members, thus preventing the
introduction of coupled designs.

2.5 Uncertainty and the Information Axiom

In this chapter, we discussed the implication of the Independence Axiom in

designing various systems, including mechanical, electrical, economic, and orga-
nizational systems. No in-depth discussion was presented on the Information
Axiom. Information Axiom states that a better design is the one with less infor-
mation content. According to the second axiom of AD, a robust system that satisfies
the Independence Axiom has required less information than coupled systems that
violate the Independence Axiom. When two designs that satisfy the Independence
Axiom are compared, the design with less information content is the best one.
The Information Axiom deals with the “robustness” of a design, reliability of
designed systems, optimization of the system, and complexity of a system.
Robustness implies that the system is easy to make and operate and reliable. We
will show that a system that satisfies the Independence Axiom inherently requires
less information to construct and to operate. Therefore, when we design a new
system, we must make sure that the Independence Axiom is not violated before we
concern ourselves with the Information Axiom. In Chap. 8, we will discuss the
Information Axiom and its implications in great depths. We will show how the
Information Axiom can be used to choose the design range and the system range to
develop an optimum design for faster economic growth.
In Example 2.6 (money circulation for economic growth), we designed a ficti-
tious taxation system as a way of illustrating how AD may be used in non-technical
fields. We considered the effect of increasing the velocity of money circulation as a
way of increasing the economic growth of a nation, assuming that indeed the
velocity of money circulations will affect economic activity. (In economics, there
seems to be a diversity of opinions among economists on many economic matters.)
The simple example illustrated how the taxation system could be designed to
increase the circulation rate by changing the tax rates, tax credit, and tax brackets.
In Chap. 11, we will show how the Information Axiom can be used to choose the
design range and the system range to develop an optimum design for faster eco-
nomic growth.

7
KAIST is an acronym for the Korea Advanced Institute of Science and Technology.
2 What Is Design? 67

2.6 On Design of Machines that Can Design

In recent years, many companies have introduced AI-based technologies and

products (e.g., IBM’s Watson, self-driving cars, machines that employ facial
recognition, and many others), but so far most design decisions are made by people
other than routine graphics that use robust modeling tools.
Question 2.5:
Do you think someday intelligent machines can develop complicated systems, taking over
the functions human beings are performing today? Following the sequence of the design
described in Chaps. 1 and 2, (i.e., identification of the problem, distillation of FRs,
selection of DPs without violating the Independence Axiom, and choosing PVs to satisfy
DPs), which part or parts are likely taken over by an intelligent design machine?

Already routine design functions (e.g., graphics that are part of known products
such as airplanes and automobiles) are performed by computer-assisted machines.
However, the original highly creative problem identification is still in the realm of
human ingenuity. No one can predict the future of technological and scientific
innovations when we realize how much human ingenuity has advanced these fields
during the past 350 years.
One thing is clear! As we systematize the logic and procedure of design through
the development of theories such as AD, it can be the basis for creating a “Thinking
Design Machine” that performs many routine functions of design such as playing
the role of system architect. It may create a design matrix (DM), suggesting better
FRs and DPs, identifying coupled designs and decoupled designs, coming up with a
proper sequence of operations when decoupled designs are involved, and assem-
bling DPs into a unified physical system. Computers are powerful tools, especially
when we have to deal with massive data that can facilitate the design process.
Human creativity has done amazing things. It is difficult to imagine the day
when machines can replicate (or replace) the incredible intellectual advances people
like Einstein, Watts and Crick, Shannon, and others made during the past century. If
one may be permitted to speculate on what the future holds, human ingenuity will
continue to lead high-level thought processes that are essential in genuinely creative
designs.

2.7 Conclusions

Synthesis builds on the human ability to quantify and understand broader social and
natural issues. In this chapter, we strived to show the role of design, not only in
engineering but also in many other fields. Leaders in all fields must possess the
ability to design since their responsibility often requires the ability to integrate and
synthesize based on the analytical database provided by various groups within the
organization. Head of industrial firms, university presidents, and national political
leaders must understand the complementary role of synthesis and analysis in
68 N. P. Suh

creating systems. Some leaders possess both synthetic and analytic capabilities.
However, if one can have only one, i.e., either analytic or synthetic capability, it
may be better to choose a leader with a strong ability to synthesize available
information to generate a new design or solutions.
Educational goals of a university must be diverse, educating people who are
analytic thinkers, also people who are more talented in synthesis, and even people
who are comfortable in both analysis and synthesis. Creativity is needed both in
analysis and synthesis. The ability to design well must be nurtured through edu-
cation, practice, and deep thinking.
It is much harder to teach synthesis for many different reasons. Therefore, even
in engineering schools, design or synthesis education has received marginal
emphasis. Many students are expected to learn design and synthesis in the industry.
However, the industry cannot provide a formal pedagogy in design. As a result,
many engineers learn design through experience. More formal teaching of design is
needed in many countries.
There are a few important messages that we hope to convey in the first three
chapters of this book. First, the design is a process of creating solutions. Second, we
show how the design should proceed. In Chap. 7, the Independence Axiom is
discussed in depth to explore the significance of the independence of FRs. In
Chap. 8, the Information Axiom is examined to show how design can be made
more robust. In later chapters, these two design axioms are applied to a variety of
engineering problems to illustrate the AD process of creating solutions.
Problems
1. The 2014 report of the Intergovernmental Panel on Climate Change (IPCC) of
the United Nations stated that the emission of anthropogenic greenhouse gases
must be reduced by 40–70% in 2050 in comparison to the emission level of
2010 to limit the earth’s temperature rise below 2 °C relative to the
pre-industrial level. Define FRs that your country must satisfy to reduce CO2
emission by 50%.
2. Identify potential problems of driverless cars and then define the FRs and
constraints of such a system for an ideal driverless car.
3. In many countries, unemployment or under-employment of young people is a
significant social issue. The goal established by the government is a 50%
reduction in the unemployment rate of college graduates. Define FRs and DPs to
alleviate the problem.
4. The supply of clean potable water is a significant issue in many parts of the
world. Traditional desalination processes such as evaporation technology and
reverse osmosis (RO) processes are too energy-intensive. For the country, you
are a resident of, develop FRs and DPs to deal with the portable water problem.
5. University education is important. There are many problems associated with
providing higher education: cost, inadequate preparation of incoming students,
and lack of jobs for college graduates. You just became the Minister of Edu-
cation of your country. Define the problems faced by your country in improving
2 What Is Design? 69

university education. Develop a set of FRs that can solve this problem. Identify
corresponding DPs.
6. Social networks (SN) play a vital role in forming public opinions. Unfortu-
nately, some manipulate the public opinion for personal gains using SN. Design
an SN system that can filter out erroneous information.
7. When you drive a car and try to steer it to either right or left, the car tilts,
indicating that the steering function is coupled to the tilting of the vehicle. Stiffer
suspensions are used in sporty cars for tight steering, while the softer suspension
is in limousines for comfort. The suspension and steering systems are coupled
with designs. Can we develop a design that uncouples steering and tilting?
8. Construct the design matrix for the problem discussed in Example 2.6: money
circulation for economic growth.

Bibliography
Alexlukin (2019) Yellow leather stylish shoes isolated on white background. Photo licensed from
Shutterstock, ID: 96707662
Casey B (2009) Advice for maintaining hydraulic accumulators. Machinery Lubrication. https://
www.machinerylubrication.com/Read/2305/hydraulic-accumulators
Kurguzova O (2019a) Mixer cold hot water. Modern faucet bathroom. Kitchen tap. Isolated white
background. Chrome-plated metal. Photo licensed from Shutterstock, ID: 529620055
Kurguzova O (2019b) The water tap, faucet for the bathroom and kitchen mixer, isolated on a
white background. Chrome-plated metal. Side view. Photo licensed from Shutterstock, ID:
524192626

Further Reading

Farid A, Suh NP (eds) (2016) Axiomatic design in large systems. Springer, Cham, Switzerland
Fiege R (2009) Axiomatic design: Eine Methode zur serviceorientierten Modellierung. Gabler
Verlag
Park GJ (2007) Analytic methods for design practice. Springer
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
How Do We Design?
3
Nam Pyo Suh

Abstract
The design of systems foreshadows the outcome of human efforts in many fields.
Many successful designs (e.g., the U.S. Constitution, automobile, airplane,
telephone, Internet, electric power, electric grid system, a banking system,
university, refrigeration, atomic bomb, computer, microelectronics, new mate-
rials) have altered the course of human history. This chapter describes how a
system can be designed, following the design process outlined in Chaps. 1 and 2.
The designer envisioned in this book is a person with a large, clean canvas on
which to design a system freely and creatively, unhindered, and unencumbered by
any prior bias and preconceptions. The role of the designer is analogous to that of a
composer of a symphony, who uses creativity, inspiration, and knowledge to
orchestrate the entire process of creation and performance of music. When they
achieve their goals successfully, their work can mesmerize thousands of music
lovers for centuries. The result of the designer’s work must be a complete, practical,
and realistic system that embodies the most rational, creative, and imaginative
thoughts. The design should yield the solution(s) for the problem identified within
a set of bounds (i.e., constraints such as cost, safety, reliability, rules, and
regulations). History shows that the work of great designers has changed the course
of nations, technology, and humanity, perhaps as much as the composer and the
scientist who are still affecting humanity decades after they had done their work.
This chapter shows how Axiomatic Design (AD) has been used to create
solutions to various problems in diverse fields, technical and non-technical.
Some are simple examples, such as the design of ashtrays to complicated
subjects such as the design of wireless electric power transfer for on-line electric
vehicle (OLEV). Regardless of the problem, the same thought process is used in
designing these items. The use of AD theory has shortened the time and cost of

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 71

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_3
72 N. P. Suh

developing new products, systems, processes, and organizations. A series of

different examples of designs illustrate the design process. They demonstrate
how the designer begins the design process with a clear goal in mind and solves
the problem through the use of ingenuity, creativity, and practicality within a
given set of constraints. Perhaps most importantly, the AD shows the designer
how to avoid making wrong decisions.
After studying this chapter, the reader should know the following:
(1) What we must do to achieve the design goal(s). (2) The first step in design is
defining the problem we must solve to accomplish the design goal. (3) A clear
understanding of the characteristics that distinguish each domain, i.e., FR, DP,
PV. (4) The transformational process of mapping between the domains.
(5) Design matrix. (6) Importance of satisfying the Independence Axiom.
(7) The similitude of all design tasks, i.e., the same processes used for both the
design of technologies and organizations. Perhaps equally important: how to
imagine and even dream of what appears to impossible things!
One of the Most Important Questions to Think About throughout This
Chapter:
This chapter presents some of the essential theorems, including the one on an
ideal design. These theorems state that in an ideal design, the number of FRs and
the number of DPs must be equal to each other with a diagonal matrix between
the FR vector and DP vector. Then, mathematically, the optimization of design
should be simple and straight forward, i.e., transform any matrix into a diagonal
matrix, which is possible mathematically. However, in design, such mathemat-
ical operations cannot and should not be done. Why?

3.1 Universal Nature of Design Tasks

Chapter 2 presented the four domains of the design world. All designs can be regarded
as a cascade of activities starting from the customer domain, where the goal of design
is established. Also, in the customer domain, the problems that must be overcome to
achieve the goal are explicitly stated. The identified problem is then mapped into the
functional domain, which is characterized by the functional requirements (FRs). The
set of FRs constitutes the characteristic vector {FRs}, which represents the design
goals. Design is the activity of choosing FRs and satisfying the chosen FRs.
After the FRs are chosen, then they are transformed into a set of design
parameters (DPs). Each DP is chosen to satisfy a given FR. A set of DPs, repre-
sented as {DPs}, constitutes the characteristic vector {x} that satisfies the FR
vector, {FRs}, in the physical domain. Finally, DPs are mapped into the process
domain by choosing process variables (PVs). {PVs} is also the characteristic vector
in the process domain that satisfies {DPs}. This transformation process from the
problem identification to the establishment of FRs, similarly from FRs to DPs, and
3 How Do We Design? 73

DPs to PVs embodies the design process discussed in this book. This process of
design, which may be called the GPFDP (Goal-Problem-FR-DP-PV) transforma-
tion process, is universally applicable in all design activities. Therefore, as stated in
Chap. 1, the Principle of Similitude of Systems applies to all systems. This chapter
illustrates how these ideas for design have created new systems such as:

• new products;
• new technologies;
• new systems;
• ultra-precision machines such as a lithography machine;
• new organizations, including a government agency and venture firms, and
• strategies.

Design Story 3.1: Battleship Design and the Design of a Winning Battle
Strategy
Throughout history, there were many wars between and among nations. Preparing
for battles or trying to defend oneself from invaders, many countries invested
enormous financial resources and devoted precious human capital to the military
and armaments. As a by-product of the R&D investment in developing new
weapons, many innovative technologies have been created throughout human
history, especially since the 1940s.
The story described in this sub-section is based on a real story that happened in
the sixteenth century. The goal of this storytelling is to illustrate rational thinking
that led to innovative ship design, albeit for military purposes, as well as a winning
military strategy.
Historically, neighboring nations often fought wars or had various conflicts. This
story is about two neighboring nations geographically close to each other but
separated by sea. One of these nations that had a more extensive population base
invaded the other country several times during the past 500 years. One of these
occurred in the mid-sixteenth century. The invading nation assembled a large fleet
of naval ships and invaded its neighboring country. Notwithstanding a large fleet
they deployed for the battle, they lost the war. Why?
The credit must be given to the commanding admiral of the smaller country, who
led their smaller navy to victory, defeating a much more powerful navy with many
more ships. The admiral had done two things right to win the battle: innovative ship
design and the design of winning war strategy. Surprisingly, what he did was
consistent with AD.
The commanding admiral of the smaller nation might have known several
important things: (1) the invading navy would have a larger number of warships
than he had; (2) the ships of the opposing navy would have a conventional
streamlined hull shape designed for faster speed with the least amount of resistance;
(3) these ships could not make sharp turns quickly because of the hull shape, and
(4) it would take a long time to reload their guns after each shot.
74 N. P. Suh

Fig. 3.1 Model ship with a flat bottom for ease of maneuver (i.e., turnaround) known as the turtle
ship, the design of which is credited to Admiral Yi Sun-Sin. It was built around 1590. (Reproduced
with permission from Papilionem 2019)

Based on these observations, the admiral designed a warship that had a flat
bottom and iron deck, as shown in Fig. 3.1. These ships with a flat bottom, rather
than a streamlined ship hull, could turn around quickly.
The admiral engaged his enemy's naval armada in the open sea. He then pre-
tended as if his ships were retreating from the fight. He led his ships into a narrow
strait where the water flowed fast due to the retreating tide. The invading navy
chasing after these flat-bottomed ships had to line up in the strait because the
channel width was too narrow to accommodate many ships.
When the ships reached a narrow section of the strait, the admiral engaged the
invading navy in a fight by firing guns broadside of the invading ships, hitting
vessels one at a time. As soon as the flat-bottomed ship fired its weapon, the ship
was able to quickly turn around and then fire again, using the guns that had been
pre-loaded. By keep rotating his naval ship and reloading the guns, they could sink
the enemy vessels one by one. The ships of the opposing navy did not have quick
maneuverability and the capability to rapidly fire.
Since the ships with flat bottom could turn around quickly and fire almost
continuously, which the invading ships could not, they could demolish the other
side. Thus, the invading naval ships suffered a major defeat.
The FRs for the admiral’s fighting strategy might have been the following:

FR1 = Engage in fight with one enemy ship at a time;

FR2 = Quickly rotate the ship;
FR3 = Uninterrupted firing of guns (from both sides of the ship).
3 How Do We Design? 75

The DPs were:

DP1 = Narrow sections of the strait (where the current flow of water is fast due
to large difference in high and low tides);
DP2 = Flat bottom ship (rather than the streamlined ship);
DP3 = Fast rotation of the ship.

The design of the ship is illustrated in Fig. 3.1.

It should be noted that the relationship between FRs and DPs can be represented
by the following matrix (note: X indicates that there is a relationship between the
FR and the DP):

DP1 DP2 DP3

FR1 X 0 0
FR2 0 X 0
FR3 0 0 X

The design of the war strategy, including the ship design, was uncoupled. The
warships with the flat bottom and the war strategy were the creation of the admiral.
The admiral defeated the invading navy and won the war, becoming a national
hero to this day. However, soon after his triumph, his government put him in jail
because some politicians spread unfavorable innuendos and rumors to protect their
political positions and interest in the kingdom. Unfortunately, such injustice has
occurred in too many countries.

Design Story 3.2: Student Enrollment in an Academic Department of a

Well-Known University
Many land-grant colleges in engineering and technology were established at the end
of the American Civil War around 1865. One of these universities became a leading
university in the world. Initially, it had two engineering departments: civil engi-
neering and mechanical engineering. Both departments became one of the highest
ranked engineering departments in the world. In the late twentieth century, the
mechanical engineering department had about 65 professors. Many of these pro-
fessors were the leaders of their respective fields. However, many undergraduate
students would not choose this department as their major field of study. They would
go into other engineering fields such as electrical engineering and computer science.
This low enrollment created a massive headache for the department chair. The
professors’ opinions had three different kinds of views, each with a different
diagnosis of the problem. The faculty members of the department argued over this
issue and tried many different things for three decades without a solution.
Then a new department head was appointed to lead the department. One of the
first things he did was to identify the problems that were the barriers to attracting
undergraduate students to the department. One of the issues identified was that the
76 N. P. Suh

curriculum was not suited to deal with contemporary issues of the engineering field.
It was the curriculum the professors went through when they were students 20 to
30 years ago. Students wanted to study the topics related to their era, but the
curriculum was not flexible enough to accommodate their interest because there
were too many required subjects that the professors thought were important based
on their experience.
A new curriculum had to be developed to enable students to construct their
personalized curriculum with a minimum set of required common subjects by
identifying the FRs of the curriculum. One of the FRs was: “Allow flexibility in
establishing students’ academic programs.” Having identified this FR, a new DP
was introduced. The new DP was: allow students to construct an individualized
curriculum, selecting 50% of their academic courses from any one of the six aca-
demic fields outside of their department, substantially reducing the number of
required subjects of the old curriculum. To administer this new curriculum, a
professor who had a multidisciplinary academic background became the professor
in charge of this unique undergraduate educational program, i.e., (PVs). This
change of the educational program increased the number of undergraduate students
enrolled in the department by a factor three. The department became one of the
most popular academic departments in the university.

3.2 Some Useful and Simple Corollaries and Theorems

As a result of the Independence Axiom and the Information Axiom,1 we can state
corollaries, and derive theorems, some of which are simple and yet useful to re-
member. Corollaries and theorems can be derived from the axioms. We will state
them here. The readers of this chapter should strive to derive them based on the
axioms. The definitions of corollary and theorems are as follows2:
Corollary:

1. a proposition inferred immediately from a proved proposition with little or no additional

proof;
2. something that naturally follows from axioms or other propositions that have been
proven;
3. something that incidentally or naturally accompanies or parallels;
4. some corollaries and axioms may be interchangeable.

1
For those interested in knowing how the design axioms were created in 1976, see (Suh 1990,
pp 17–22).
2
“… it is useful to distinguish between theorems and corollaries in mathematics vs in design
science. Mathematics, as natural science, states axioms, theorems, and corollaries in a descriptive
manner. This is because the underlying field itself simply seeks to true statements on the natural
world and its underlying world. In contrast, design science has a prescriptive nature. Its goal is to
guide the designer with the guidance of what good design is. For that reason, in AD the axioms,
theorems, and corollaries are stated in a prescriptive manner rather than a descriptive manner.”
(Farid and Suh 2016).
3 How Do We Design? 77

Theorem:

1. a formula, proposition, or statement in mathematics or logic deduced or to be deduced

from other formulas or proposition;
2. an idea accepted or proposed a demonstrable truth often as a part of a general theory;
3. a theorem is valid if its referent axioms are valid.

Some corollaries of the Independence Axiom (i.e., maintain the independence of

FRs) and the Information Axiom (i.e., minimize the information content) of AD are
the following:
Corollary 1: (Decoupling of Coupled Design)
Decouple or separate parts or aspects of a design solution if FRs are coupled
or become interdependent in the design proposed.
Corollary 2: (Minimization of FRs)
Minimize the number of FRs and constraints.
Corollary 3: (Integration of Physical Parts)
Integrate design features in a single physical part if FRs can be independently
satisfied in a proposed solution to reduce the information content.
Corollary 4: (Use of Standardization)
Use standardized or interchangeable parts if the use of these parts is con-
sistent with FRs and constraints.
Corollary 5: (Use of symmetry)
Use symmetrical shapes and/or components if they are consistent with the
FRs and constraints.
Corollary 6: (Largest Tolerances)
Specify the largest allowable design range in stating FRs.
Corollary 7: (Uncoupled Design with Less Information)
Seek an uncoupled design that requires less information than coupled designs
in satisfying a set of FRs.

There are many theorems, but only a limited number will be listed here.
Theorem 1: (Coupling Due to Insufficient Number of DPs)
When the number of DPs is less than the number of FR, either a coupled
design results, or the FRs cannot be satisfied.
Theorem 2: (Decoupling of Coupled Design)
When a design is coupled due to the greater number of FRs than DPs, it may
be decoupled by the addition of new DPs to make the number of FRs and DPs
equal to each other, if a subset of the design matrix containing n x n elements
constitutes a triangular matrix.
Theorem 3: (Redundant Design)
When there are more DPs than FRs, the design is either a redundant design or
a coupled design.
Theorem 4: (Ideal Design)
In an Ideal Design, the number of DPs is equal to the number of FRs.
78 N. P. Suh

Theorem 5: (Need for New Design)

When a given set of FRs is changed by the addition of a new FR, or sub-
stitution of one of the FRs with a new one, or by selection of an entirely
different set of FRs, the design solution given by the original DPs cannot
satisfy the new set of FRs. Consequently, a new design solution must be
sought.
Theorem 6: (Path Independence of Uncoupled Design)
The information content of an uncoupled design is independent of the
sequence by which the DPs are changed to satisfy the given set of FRs.
Theorem 7: (Path Dependence of Coupled and Uncoupled Designs)
The information contents of coupled and decoupled designs depend on the
sequence by which the DPs are changed and on the specific paths of the
changes of these DPs.

There are many additional theorems, some of which are presented in the appendix
of this chapter.
These theorems and corollaries are important, because they guide the designer’s
thoughts as they search for best or acceptable design. For instance, Theorem 4 states
that in an ideal design the number of FRs and DPs must be the same. This theorem
is simple and important to remember.

3.3 Lesson on “Design It Right, i.e., No Coupled Design!”

More frequently than we realize, many designers create coupled designs, thinking
that they can optimize it through modeling and optimization. Then, they try to make
the flawed design work through extensive analysis, modeling, “optimization,”
testing, and other means—wasting time, effort, and resources as well as damaging
their professional reputation. Their coupled design may even end up leading to
fatalities (e.g., airplanes plunging into the ground!). Those who do not understand
the design axioms may work hard, spend lots of money (often someone else’s), and
end up creating a poorly functioning and unreliable product. Then, they produce a
lengthy and complicated instructional manual, hoping to minimize complaints from
those who end up buying the “broken” system. Advice: create an uncoupled design
from the beginning. If a coupled design is introduced inadvertently, identify the
coupling early and change the design to an uncoupled design! The following
example illustrates the importance of creating uncoupled or decoupled design:
Engineers and scientists at the Diamond Light Source of the United Kingdom
published an excellent paper on the importance of the Independence Axiom. They
decoupled a coupled design. In short, they first came out with a coupled design, and
then they analyzed it to optimize the system. However, the system optimized based
on their analysis failed! Later, they learned about AD, redesigned it to eliminate the
coupling of their FRs. It solved their problem! Their paper describes what they have
3 How Do We Design? 79

gone through for the benefit of others who make a system complicated by creating a
coupled system, wasting time, money, and resources!
It should be noted that their problem is similar to designing a water faucet that
will control temperature and the flow rate of water independently from each other!
Often when a coupled design feature is found, the best thing to do is to develop a
new design that is uncoupled rather than tweaking the system to make the flawed
design work, which will not work well as long as the design is coupled. Their
abstract reads as follows (Reproduced from Drakopoulos et al. (2015), originally
published open access under a CC BY license: https://journals.iucr.org/s/issues/
2015/03/00/ie5138/):
“I12 Joint Engineering, Environmental, and Processing (JEEP) is a high-energy imaging,
diffraction and scattering beamline at Diamond. Its source is a superconducting wiggler
with a power of approximately 9 kW at 500 mA after the fixed front-end aperture; two
permanent filters aim at reducing the power in photons below the operating range of the
beamline of 50–150 keV, which accounts for about two-thirds of the total. This paper
focuses on the design and simulation process of the permanent secondary filter, a 4 mm
thick SiC disk. The first version of the filter was vulnerable to cracking due to thermally
induced stress. Therefore, a new filter based on an innovative concept was proposed: a
water-cooled shaft rotates, via a ceramic interface, the SiC disk; the disk operates up to
900 ◦C, and a copper absorber surrounding the filter dissipates the heat through radiation.
We utilized analysis data following the failure of an initial prototype to model the heat flow
using FEA successfully. This model informed different iterations of the re-design of the
assembly, addressing the issues identified. The operational temperature of the final product
matches within a few degrees Celsius the one predicted by the simulation.”
The schematic of their optical and functional layout of the I12 JEEP beamline is shown in
Fig. 3.2.

Fig. 3.2 Schematic optical and functional layout of the I12 JEEP beam line. (Reproduced from
Drakopoulos et al. 2015, originally published open access under a CC BY license: https://journals.
iucr.org/s/issues/2015/03/00/ie5138/)
80 N. P. Suh

Fig. 3.3 A round SiC disk is

attached to a disk holder made
of copper with circulating
water. The round disk is
positioned and attached to the
copper holder with
molybdenum interface
material. Three FRs and two
DPs is a coupled design since
there are more FRs and DPs
(a theorem). (Reproduced
with permission from Tizzano
et al. 2018)

Design Issue: The purpose of the second filter is to reduce the power coming in from the
primary filter from 6.2 to 2.6 kW. The filter made of a 4-mm-thick SiC disk (shown in
Fig. 3.3) is attached to the Cu carrier with a molybdenum interface binder. There are three
FRs for the secondary filter:

FR1 = Lower the power level from 62 to 2.6 kW;

FR2 = Transmit heat to cool the filter;
FR3 = Hold the SiC disk in place.

The original design made by engineers and scientists at the Diamond Light Source of the
United Kingdom is shown in Fig. 3.3. The SiC disk is bonded with a molybdenum binder
to the copper fixture, which is cooled by circulating water. The energy absorbed by the SiC
disk (to reduce the power transmitted by 3.8 kW) is transmitted to a copper fixture by
conduction to maintain its operating temperature low.
The DPs are as follows:

DP1 = Thickness of SiC disk;

DP2 = Heat conduction from SiC to the copper fixture through the molybdenum binder;
DP3 = Molybdenum Binder = DP2.
Since there are three FRs and only two DPs, we know that it is a coupled design, because
the number of DP is less than the number of FRs (A theorem). The design matrix may be
represented as:
3 How Do We Design? 81

DP1 DP2
FR1 X 0
FR2 0 X
FR3 0 X

It is a coupled design because there is only one DP, i.e., the molybdenum binder, for two
FRs, i.e., FR2 and FR3. As a result, both FR2 and FR3 were not satisfied, and the SiC disk
cracked due to the temperature rise and thermal stress induced due to the differential
expansion of the SiC disk and the copper holder. Before constructing the design shown in
Fig. 3.2, the scientists and engineers performed extensive analysis and optimization to be
sure that the design was going to work as intended. However, after the construction, they
found that the designed system failed to satisfy FR2 and FR3.
Because they had cracking and other problems with the coupled design, they applied AD
and came up with a new uncoupled design, which is shown in Fig. 3.4.
In this uncoupled design, SiC disk (shown in blue) is sandwiched between two copper disks
(shown in brown, with copper tubing circulating water) to remove heat. Heat transfer
occurs through the large surface area in contact between the SiC sandwiched between the
two copper plates. The DPs are

DP1 = Thickness of SiC disk;

DP2 = Heat conduction from the SiC surface to copper plates on both surfaces of the
SiC;
DP3 = Two copper plates bolted together.

Fig. 3.4 New uncoupled design. SiC disk (shown in blue) is held in place by sandwiching the SiC
disk between two copper plates (shown in brown), which are cooled by circulating water.
(Reproduced with permission from Tizzano et al. 2018)
82 N. P. Suh

This is an uncoupled design, i.e., the design matrix is a diagonal matrix.

This design performed well. The lesson of this example is: if the design violates the
Independence Axiom, the design is a flawed design that cannot be solved through
optimization.
This design was later modified for ease of mounting the disk assembly and also to promote
additional heat transfer through the rotating shaft (Papilionem 2019).

3.4 Mathematical Optimization and Design

It was stated that FRs is a vector, {FR}, and DPs is a vector, {DP}. What relates
these two vectors is the design matrix [DM], which may be expressed as

fFRg ¼ ½DM fDPg ð3:1Þ

When there are two FRs, there should be two DPs, as per Theorem 4. The best
design, i.e., an ideal design, then is the one where the FRs are independent from
each other, i.e., the design with a diagonal matrix, i.e.,

FR1 ¼ ðA11ÞDP1
FR2 ¼ ðA22ÞDP2


A11 0
or fFRg ¼ fDPg ð3:2Þ
0 A22

If we design without the Independence Axiom in mind, the relationship between

{FR} and {DP} may be stated as follows:
 
 
FR1 A11 A12 DP1
¼ ð3:3Þ
FR2 A21 A22 DP2

Equation (3.3) can be mathematically manipulated to make it a diagonal matrix,

like Eq. (3.2), by determining characteristic values (in mathematics, it is referred to
as “eigenvalues”). However, the resulting changes in DPs would be such that they
could not be implemented physically. What a pity that the mathematical transfor-
mation cannot be applied in design! Otherwise, the designer’s task would be
straightforward. In design, we have to choose the right set of DPs to prevent the
coupling of FRs.
Example 3.1
Suppose that FR1 and FR2 in Eq. (3.3) are given as 5 and 10, respectively. Furthermore,
the proposed design yielded the following diagonal design matrix:
3 How Do We Design? 83
 
 
FR1 1 0 DP1
¼ ðNote : This is an uncoupled design:Þ
FR2 0 5 DP2

What should DP1 and DP2 be? The correct answers are DP1 = 5 and DP2 = 2. In this case,
if FR1 is to be changed, the only thing we have to do is change DP1. FR1 and FR2 are
independent from each other.
Now the design matrix for another design is a triangular matrix given as follows:
 
 
FR1 1 4 DP1
¼
FR2 0 5 DP2

(Note: This design is called the decoupled design, because the design matrix is a triangular
matrix.)
What should DP1 and DP2 be?
In this case of “decoupled design,”

FR1 ¼ 5 ¼ 1  DP1 þ 4  DP2

FR2 ¼ 10 ¼ 5  DP2

DP2 must be determined first, which should be 2. And then we can determine DP1 as

DP1 ¼ 5  8 ¼ 3

Therefore, in this case of a decoupled design, we have to vary DPs in a proper sequence,
i.e., we have to vary DP2 first and then determine DP1. As long as we follow this sequence,
FR1 and FR2 are independent from each other.
Suppose now, the design matrix is a full matrix as given below:
 
 
FR1 1 4 DP1
¼
FR2 3 5 DP2

In the above design, FR1 and FR2 are coupled to each other. If we change one of the DPs, it
will affect both FRs, i.e., a coupled design. In this coupled design, there is only one set of
DP1 and DP2 that will uniquely satisfy FR1 and FR2. Similarly, if one of the FRs is
changed, both DPs must find a new design spot, i.e., change their values, to satisfy the new
value of the FR. When DPs represent physical things, it becomes an enormous task to
change them.

Example 3.2
The design that is easiest to implement is the one where all elements of the design matrix
are zero, except A1, B2, C3, and D4, as shown below. In this case, we change D11 and
84 N. P. Suh

DP12 to satisfy FR11 and FR22, which are the children FRs of FR1, which enable us to
satisfy FR1. This is an uncoupled design, which satisfies the Independence Axiom.

DP1 DP2
DP11 DP12 DP21 DP22
FR1 FR11 A1 0 0 0
FR12 0 B2 0 0
FR2 FR21 0 0 C3 0
FR22 0 0 0 D4

Now, suppose that the design has two additional design elements, A4 and D1, as shown
below. In this case, if we vary DP22 to satisfy FR2, FR1 also changes. Similarly, if DP11 is
changed to satisfy FR1, FR2 also changes. This is a coupled design. This design should not
be implemented.

DP1 DP2
DP11 DP12 DP21 DP22
FR1 FR11 A1 0 0 A4
FR12 0 B2 0 0
FR2 FR21 0 0 C3 0
FR22 D1 0 0 D4

The above example is a straightforward design in that we had to decompose FRs and DPs
only once. However, if Boeing 737 Max had many layers of decomposition and many of
the off-diagonal design elements were non-zero elements. The pilot would have had dif-
ficult times to control the airplane if the pilot was not aware of the fact that a computer was
controlling one of these elements independently of the pilot’s command! That is why
satisfying the Independence Axiom is essential. Unless we construct a master matrix of the
entire design to trace all the design decisions made, it will be impossible to determine if the
design is satisfactory. Many companies, instead of constructing the design matrix for the
entire system, resort to the practice of developing products following the repetitious cycle
of “design-build-test-redesign to correct mistakes-test” to develop their products. Such an
approach to design and development is costly and unreliable.
In executing a sizeable complicated system design with many FRs, it will be necessary to
appoint a “system architect,” whose team constructs the entire design matrix and checks all
the design decisions made by any member of the project team. When the system architect
determines that a particular design decision made by one of the project teams has intro-
duced the coupling of FRs, the architect should instruct the team that introduced coupling to
develop another design that does not create such a coupled design.

Example 3.3
In the mathematical world, we can take any square matrix and determine its eigenvalues
(i.e., characteristic values). However, in design, we cannot make a coupled design and
manipulate it mathematically to find “characteristic values (i.e., eigenvalues)” to uncouple a
coupled design. If we could discover eigenvalues in design through mathematical manip-
ulation, any bad design can be made into a good design that satisfies the Independence
Axiom, but unfortunately, in the design world, that cannot be done.
3 How Do We Design? 85

Suppose we have two FRs for a water faucet design: FR1 = control temperature,
FR2 = control flow rate. If the design of the water faucet has a cold water valve and a hot
water valve, we cannot independently satisfy FR1 and FR2 without changing two valves
continuously to get the right temperature and the flow rate. This design is a coupled design
with a full design matrix, i.e., DP1 affects both FR1 and FR2. Likewise, DP2 also affects
both FR1 and FR2. Then, the design equation may be written as
 
 
FR1 A11 A12 DP1
¼
FR2 A21 A22 DP2

The elements of the design matrix are non-zero elements.

In a pure mathematical world, the above equation may be transformed into
 
 
 
FR1 A11 A12 DP1 1 0 DP1
¼ ¼
FR2 A21 A22 DP2 0 1 DP2

where DP1* and DP2*, which are sometimes called the “characteristic values” or
“eigenvalues.”
For instance, we encounter this kind of transformation in applied mechanics. Stress rij and
strain eij are second-order tensors. In the two-dimensional case where i and j vary from 1 to
2, rij represents an ellipse with a major axis and a minor axis. Along with these principle
directions, only the normal stress and strain are present with no shear components. How-
ever, a mathematical transformation cannot be done in the case of design to determine the
characteristic values, because DP1* and DP2* represent something that does not have any
physical significance. This difficulty of finding DPs that is physically meaningful is the
reason why design is challenging, i.e., we have to choose the right DPs a priori based on
our understanding of the “physical or the real” world to satisfy the Independence Axiom.
Similarly, the same comments apply to the software world. Many software companies have
had problems that stem from their misunderstanding of this issue. Software developers
know that coupling FRs create a spaghetti code, but they find that it is complicated to
uncouple legacy codes because it is difficult to determine the logic behind the existing
legacy codes.
The lesson of this example is simple and straightforward but important: satisfy the Inde-
pendence Axiom by selecting a correct set of DPs to satisfy FRs from the beginning. In
most cases, uncoupled design cannot be created through mathematical manipulations.

3.5 Design of Technological Products

“The purpose of this section is to illustrate that even a system that appears to be com-
plicated or complex can be designed by following the process used to design a simple
system such as the water faucet. We have to use a systematic process to check for the
coupling of FRs because as the number of FRs and the number of the layer of decompo-
sition increase, the probability of creating coupled design increases.”
86 N. P. Suh

Many companies, university researchers, individual inventors, and government

laboratories design and develop new technologies. Anyone with some science and
engineering background can create new technologies. The person(s) should identify
the problem(s) in the customer domain, be able to transform the problem into a set
of FRs, and then select appropriate DPs to create a system that fulfills the FRs.
Then, they have to deal with PVs such as investment and personnel. In this section,
two examples will be presented to show how new technologies can be designed and
created.

3.5.1 Design of Easy Open Beverage Can

Bauxite is the raw material that is used to produce aluminum through an electrolytic
process of separating aluminum atoms from oxygen atoms, which is an
energy-intensive process. The cost of electricity is the most significant contributor
to the manufacturing cost of aluminum. The largest fraction of aluminum produced
worldwide, i.e., about 60% of the aluminum produced worldwide, is used to make
beverage cans. The rest is used to manufacture airplanes, buildings, and the like.
Although they tried, it has not yet penetrated the automobile market. 70% of an
aluminum can is recycled.
A manufacturer of aluminum beverage can has continued to reduce the amount
of aluminum used to make a can to be able to compete with cheaper materials such
as steel and plastics. Aluminum is the most expensive material among the three.
The ability to make a thin-walled can cheaply enable the aluminum industry to
dominate the beverage can business. The aluminum ingot is made to a thin sheet,
from which a disk is punched out, which is then deep drawn into a thin-walled
cylindrical tube with a bottom. Then, the top of the cylinder is rolled to reduce its
diameter. The top of the can is made separately, which is then attached to the main
body.
An aluminum beverage can with an easy-open tab is shown in Fig. 3.5. What are
the FRs of this can? What are the constraints? Why is the top of the can narrower
than the rest of the body? Why is the bottom of most aluminum cans not flat but
rather concaved-in?
The major constraints are relatively easy to identify:
Constraint 1 = Minimize the overall cost of manufacturing;
Constraint 2 = Use a minimum amount of aluminum;
Constraint 3 = Allow recycling of used aluminum.
The FRs of an aluminum beverage can are more challenging to identify because the
geometry does not reveal all the FRs. The FRs for the easy opening can appear to be the
following:
3 How Do We Design? 87

Fig. 3.5 Aluminum beverage

can. (Reproduced with
permission from Gossip 2019)

FR1 ¼ Contain the radial internal pressure of 30psi;

FR2 ¼ Allow easy opening of the can;
FR3 ¼ Allow easy stacking of the can;
FR4 ¼ Coat the inside of the can to prevent the beverage
acquire the taste of aluminum;
FR5 ¼ Allow printing on the outside of the can;
FR6 ¼ Make the diameter of the top of the can as small
ð3:4Þ
as practical to minimize the consumption of aluminum;
FR7 ¼ Make the bottom of the can to be concave in to contain the pressure;
FR8 ¼ Enable stacking of cans on top of each other;
FR9 ¼ Contain the pressure at the bottom of the can;
FR10 ¼ Contain the pressure acting on the lid;
FR11 ¼ Attach the tab for ease of opening the can;
FR12 ¼ Join the lid to the body of the can:

Exercise 3.1:
An engineer of a major aluminum producer attended a special program for engineers in the
industry on AD at a university. During one of the discussions, he stated that the beverage
can satisfies 12 FRs. However, he realized that the company was struggling with its
product because he could only identify 11 DPs. While attending the class, he figured out
what the missing DP was and thus was able to develop an uncoupled design. Assuming that
the above 12 FRs are what he had identified as FRs, what are the 12 DPs? Does the can
satisfy the Independence Axiom?

3.5.2 Removal of Kidney Stone with a Robot

Some people suffer severe flank and lower back pain because of kidney stones,
which are hard deposits of minerals and salt that stick together in the kidney, or the
urinary tract. Some stones are small being in the order of 5 mm or less, which
88 N. P. Suh

Kidney

Ureter

Bladder
Urethra

(a) (b)

Fig. 3.6 a A schematic drawing of kidneys, bladder, and ureter in the human body. (Reproduced
from NIH: National Cancer Institute 2009.) b Kidney stones (Reproduced with permission from
Lorne 2019)

generally pass through the urinary system by themselves, but they can be much
larger, several stones filling the entire kidney (see Fig. 3.6). The common symp-
toms are severe back and flank pain, stomachache, blood in the urine, and nausea or
vomiting. A well-known technique of removing the stone is to use the ultrasound to
crush the stone and then let the urine to carry the stone out of the kidney or the
urinary tract.
A well-known professor in the field of robotics at a leading science and tech-
nology university in the world decided to develop a robotics technology to solve the
kidney stone problem. His technique worked so well that he decided to establish a
new venture firm with his former students to commercialize his technology. He
raised the initial capital in his country to start the business. He and his team visited
major hospitals, venture capitalists, and major universities in many countries to
show-tell his technology. His patented technology combines vision to identify the
stone in the kidney or the urinary tract with a mechanical means of transporting the
stone out of the patient. It also can crush the stone in the kidney. In some cases, the
robot can grab the stone in the kidney and transport it out of the patient. Before
inserting the robot through the ureter (the duct that connects the kidney to the
bladder and finally urethra, a soft lining tube may be inserted through the ureter.
Without knowing how the professor initially designed his surgical robot for
kidney stone removal, we may attempt to create it based on the teachings of AD and
using the information he verbally explained what his robot does over a lunch in
Boston, Massachusetts, U.S.A., during his visit to the Massachusetts General
Hospital (MGH) and a high-tech company. The benefit we have is that we know
what he designed and made, although we have not seen the actual robot and do not
know the detailed design of the robot.
3 How Do We Design? 89

The goal of the design is clear: remove kidney stones from human beings. The
overall constraint is that it has to be safe, i.e., extremely safe. The FRs, for this robot
may be stated as follows:
FR1 = Flexible robot that can be inserted into the kidney through ureter and urethra;
FR2 = Protect the lining of ureter and urethra;
FR3 = Provide camera at the end of the robot;
FR4 = Attach a clamp mechanism at the end of the robot to grab the stones;
FR5 = Transmit mechanical means of manipulating the clamp mechanism;
FR6 = Provide light to capture the image.

Now we have to come up with DPs to satisfy the above FRs. Conceptually, we can
visualize many different physical configurations. For example, a robot made of a
soft-flexible and corrugated “long tube,” in which electric wiring is placed for LED
lighting in the kidney and to control the end effector (e.g., a mechanical suction
device that sucks in the stone into the tube and transport to outside), the illumi-
nating light at the head of the robot, a camera that can see the inside of kidney, the
mechanism that sucks the stones into the tube, and the means of
pressurizing/de-pressurizing the corrugated tube to insert the tube into kidney. The
robot can “crawl” toward the kidney through cyclic pressurization of the soft,
flexible, corrugated tube. The tube can stretch and move forward slowly when
pressurized. When the tube is de-pressurized, the tube will stay stationary. By
repeating the pressurization cycle, it may reach the kidney. Another possible
insertion method may be to put a flexible but stiff wire in the tube and push the front
end of the tube.
Exercise 3.2:
Based on the above description of the robot, define DPs that can satisfy the six FRs stated
above.

3.5.3 Ash Tray in Automobiles

About four or five decades ago, smoking of cigarettes was prevalent among people,
especially young people. Therefore, passenger cars had to have an ashtray in the
dashboard of a vehicle. What would be a useful design feature? What FRs should
the ashtray satisfy?
Suppose that we want to make it convenient for the driver to use the ashtray by
letting it pop out of its stored position when the front of the ashtray is pushed in
slightly to unlock it. To lock it in, the driver pushes it about a halfway then the
ashtray is pulled into the locked position. Then, when the driver wants to use it
again, the driver pushes it in slightly to have the ashtray pop out of the closed
position to its open position. For an ashtray that can perform these two functions,
what should be the FRs? We may state them as follows:
90 N. P. Suh

FR1 = Push in the front of the ashtray (that is in a locked position) to have it unlocked
and pop out to the fully open position;
FR2 = Push in the front of the ashtray in its open position about halfway to move it into
move into its closed and locked position.

The corresponding DPs are

DP1 = Unlocking mechanisms;
DP2 = Locking-in mechanisms.

The design matrix is diagonal, indicating that it is an uncoupled design. FR1 and
FR2 should be decomposed since we need to have more detailed designs for DP1
and DP2 to be able to implement it.
FR1 and FR2 may be decomposed as
FR11 = Unlock the ashtray from its locked position;
FR12 = Push the ashtray out when activated;
FR21 = Pull the ashtray in and lock it when it is pushed in toward the closed position;
FR22 = Lock the ashtray in the closed position.

The decomposed DPs, the corresponding FR11 and FR11, FR21, and FR22, may
be chosen as follows:
DP11 = Unlocking mechanism (release cam);
DP12 = Stiff spring (which releases the stored energy);
DP21 = Locking mechanism (locking cam);
DP22 = Compression of a soft spring.

The actual mechanism is sketched in Fig. 3.7.

The design matrix for this design is given below. It has a diagonal matrix,
indicating that the design is an uncoupled design. In this design, one DP satisfies a
specific FR and no other FRs, thus satisfying the Independence Axiom. In an
uncoupled design, each FR is independent of other FRs. This is an essential
characteristic of the uncoupled design.

ð3:5Þ

In the above design matrix, X indicates that there is a relationship between FR and
DP. For instance, DP11 affects FR11 only. This design matrix demonstrates that we
have an uncoupled design, which is acceptable as per the Independence Axiom.
3 How Do We Design? 91

Fig. 3.7 a Sketch of an ashtray design; b Photograph of an automobile ashtray with coil spring
rather than a linear spring sketched in (a). In this commercial product, the linear spring is replaced
by coil springs to make the product more compact

3.5.4 Design of a Collapsible Steering Column

for Automobiles

Automobiles use steering systems that turn the front wheels to change the direction
of the vehicle motion. There are many commercial steering systems with different
designs, ranging from simple mechanical systems to power-assisted steering sys-
tems. Power-assisted systems use either electric or hydraulic power. The price of
steering systems varies over a wide range, as much as by order of magnitude. We
will illustrate the design process by considering the design of a simple steering
column of an automobile.
The steering column must transmit the rotational motion of the steering wheel at
the command of the driver. It must also protect the driver in the event of a frontal
collision of the vehicle. When the car comes to a sudden stop due to the frontal
impact, the driver’s chest may hit the steering wheel of the vehicle due to the
forward inertia of the body. The goal of the design is to protect the driver by
making the steering column collapse when the impact occurs. At the same time, the
column must be strong and stiff enough to allow the regular operation of the
vehicle.
The {FRs} of a simple steering column are
92 N. P. Suh

FR1 ¼ Transmit steering motion by the driver to the front wheel;

FR2 ¼ Collapse the steering column to protect the driver in a frontal collision;
FR3 ¼ Absorb energy during the collapse of the steering column;
FR4 ¼ Allow the telescoping motion of the steering column adjust to the drivers position:
ð3:6Þ

A rough sketch of the design is shown in Fig. 3.8, which shows a steering
column that consists of two steel tubes. A portion of a smaller metallic tube is
inserted tightly inside a larger metallic tube. The serrated interface transmits the
rotational motion of the inner tube to the outer tube. The larger tube is attached to
the car body through a bracket that holds a bearing in place to allow the rotational
motion of the outer tube. The two-tube assembly rotates together under normal
steering operations. When a major frontal collision occurs, the smaller tube inside
the larger tube slides down inside the large tube, plastically deforming the narrowed
down section of the larger tube, and thus dissipating energy to protect the driver as
the chest of the driver slams down the steering wheel. The steering wheel position is
fixed by attaching the inner tube to the inside hole of a bearing. A steel bracket that
holds the inner tube is attached to the car body. To telescope the steering wheel, the
bolt that holds the steel bracket to the car body is loosened to allow the inner tube to
move in and out of the larger tube.
The DPs corresponding to FRs are as follows:

DP1 ¼ Serrated interface between the inner tube and outer tube ðto transmit the rotational motion of the
inner tube to the outer tube; allowing relative axial movementÞ;
DP2 ¼ Pin attached to the outer tube that breaks when the inner tube is pushed down under high force
ðwhen the drivers body hits the steering wheel and pushing the inner tube downwardÞ;
DP3 ¼ The narrow section of the outer tube ðwhich absorbs energy as the inner tube
attached to the steering wheel is pushed down duringfrontal impact;
which deforms the outer tube plastically deforming it and dissipating energyÞ;
DP4 ¼ A journal bearing ðthe inner hole of which is attached to
the inner tube of the steering columnÞand the outside is attached to another tube
that is attached to the car body through a bolt and an elongated hole;
which allows telescoping of the steering wheel to adjust its
position to match the drivers height:
ð3:7Þ

The design matrix for the design is as follows:

ð3:8Þ
3 How Do We Design? 93

Steering wheel

Journal bearings
(Attached to car body)
(DP4)

Serrated interface
with axial V groove
(DP1)

Pin
Narrowed down section
(DP2)
of the outer tube for
energy absorbtion
(DP3)
Ball bearings
attached to car
frame

Fig. 3.8 A schematic drawing of the collapsible steering column. It consists of two tubes: the
inner tube is connected to the steering wheel at one end and the other end of the inner tube is inside
an outer tube. A shearable pin connects the inner tube and the outer tube. The inner tube and the
inside surface of the outer tube are serrated, which prevents the slippage while transmitting the
rotational motion. The other end of the outer tube is connected to the front wheels through a
steering mechanism (not shown). The inside diameter of the outer tube narrows down. During the
collision, the smaller tube is pushed down, breaking the pin, and, as it is pushed down inside the
outer tube, the plastic deformation of the outer tube absorbs the impact energy

The design matrix is triangular, indicating that it is a decoupled design. The pin
that holds the outer tube to the inner tube absorbs a small amount of energy when it
breaks due when the chest of the driver hits the steering wheel upon impact. It is an
uncoupled design.
94 N. P. Suh

3.5.5 Design of a Technology that Reduces Material

Consumption

Problem
One of the well-known manufacturers in the United States uses several freight-car
loads of plastics to make a variety of their products every day. They surmised that if
they can reduce their plastic consumption by 10% in making their products, it
would have a significant impact on the profitability of their products. Many of their
products are injection-molded or extruded thermoplastics. The constraints are the
following: the product shape cannot change from those currently manufactured, the
toughness of their product must remain the same to prevent cracking of the parts
when they are dropped, and the production rate must be competitive. In other
words, the product shape cannot change, and yet it must use less material with
similar mechanical properties. How should we proceed?

Design Solution:
Based on the description of the problem given above, we have to be able to define
FRs. They may be stated as follows:

FR1 ¼ Reduce material consumption;

FR2 ¼ Maintain mechanical properties within an acceptable range; ð3:9Þ
FR3 ¼ Keep the geometric shape the same as the currently produced parts:

Now we have to come up with DPs that can satisfy the FRs. What DPs would
you choose to satisfy the above three FRs?
To select good DPs that will create innovative products, it is useful to know
about other related disciplines. Sometimes, the designer may read reference books;
go into Google for information; ask friends, co-workers, and colleagues for sug-
gestions; and most important of all, “think” of fundamental principles, one has
learned in science and engineering.
One possible set of DPs may be selected as shown below:

DP1 ¼ Plastic parts with a plethora of tiny bubbles;

DP2 ¼ Bubble size; ð3:10Þ
DP3 ¼ Injection molding process:

The rationale for choosing the above set of DPs is as follows: (1) By introducing
a large number of tiny bubbles, voids take up space, thus using less material.
(2) When the bubble diameter is less than a critical size, its toughness is not
negatively affected, because they may arrest cracks, and therefore the material may
even be tougher than the solid plastic parts. (3) By using a conventional injection
molding process, we can keep the geometric shape the same. This kind of material
has been invented at a university and currently used in industry worldwide to make
3 How Do We Design? 95

automotive parts. The tiny bubbles of about 20 microns in diameter are introduced
dissolving CO2 in molten polymer and suddenly releasing the pressure to nucleate
tiny bubbles.3
The design matrix for this material that is commercially known as microcellular
plastics is shown in the following:

ð3:11Þ

The above matrix given by Eq. (3.11) indicates that it is a decoupled design, and
thus this design does not violate the Independence Axiom if we follow the sequence
given above to FRs, i.e., change DP1 first; then DP2; and finally DP3 to change
FR1, FR2, and FR3 independently from each other.
Figure 3.9 shows the cross section of a microcellular plastic. It has about 109
bubbles per cubic centimeter. The bubble size is about 10 to 30 microns. It is
currently produced by extrusion of plastics or by injection molding.
Exercise 3.3:
Develop ideas as to how we can put in about one billion bubbles of 10 microns per cc of
plastics on a mass production basis using either by extrusion or by injection molding
processes.

3.5.6 Wireless Electric Power Transmission System to a Bus

in Motion

Problem
Suppose that your university has decided to replace the diesel buses from your
campus with electric buses to reduce CO2 emission and help in solving the global
warming problem.4

3
See the review article: “Microcelluar Plastics” at Wong et al. (2016).
4
One of the significant issues of the twenty-first century is global warming. According to IPCC, we
must keep the temperature rise of the earth to below 2C relative to its temperature at the time of the
Industrial Revolution. One of the major causes of weather change is the accumulation of
anthropogenic generation of CO2. We must reduce the CO2 emission to prevent the weather
change, which is the primary cause for the rise of sea level, forest fire, preservation of agriculture,
and many other calamities. Roughly, 27% of CO2 is generated by internal combustion
(IC) engines, and about 33% by coal-burning electric power plants. To reduce the CO2 emission
by automobiles that are powered by IC engines, we need to replace it with electric vehicles, which
will reduce anthropogenic emission of CO2 by nearly a half. After considering the vast
consumption of lithium needed if lithium batteries power all the cars, we decided to employ
electric vehicles that receive electrical power wirelessly from external electrical power sources
embedded under the ground.
96 N. P. Suh

Fig. 3.9 Cross section of microcellular plastics. (Reproduced with permission from the Trexel
Corporation)

Design Solution:
After considering all the factors involved in the electrification of the transportation
system, it was decided that the best option is to use the wireless transmission of
electric power from an underground power supply system to the bus either in
motion or stationary. A proposed design idea was to install the power supply system
underground just below the road pavement and transmit the electric power to
moving buses while the vehicle is on the top of the power supply system beneath
the road. We need to transmit about 100 kW of power to the bus over a distance of
over 20 cm. Since such a wireless electric bus system is not commercially available,
it was decided to design the system as a university research and development
(R&D) project. The bus will have a small lithium–ion rechargeable battery onboard
for limited mobility even when the bus is on the road without the underground
power supply system. Our job is to design such a system that is robust, reliable, and
low cost.
There are many FRs that such a transportation system must satisfy to work well
as shown in Eq. 3.9.
3 How Do We Design? 97

FR1 ¼ Propel the vehicle with electric power;

FR2 ¼ Transfer electricity from an underground electric cable to the vehicle;
FR3 ¼ Steer the vehicle;
FR4 ¼ Brake the vehicle;
FR5 ¼ Reverse the direction of motion;
FR6 ¼ Change the vehicle speed;
FR7 ¼ Provide electric power when there is no external electric power supply;
FR8 ¼ Supply electric power to the underground cable:
ð3:12Þ

The constraints the final design should not violate are

C1 ¼ Safety regulations governing electric systems;

C2 ¼ Price of OLEVðshould be competitive with cars with IC enginesÞ;
C3 ¼ No emission of greenhouse gases;
C4 ¼ Longterm durability and reliability of the system;
C5 ¼ Vehicle regulations for space clearance between the road and the vehicle:
ð3:13Þ

The DPs chosen to satisfy the highest level FRs given by Eq. (3.12) are as follows:

DP1 ¼ Electric motor;

DP2 ¼ Wireless power transfer system;
DP3 ¼ Mechanical steering system;
DP4 ¼ Hydraulic braking system;
ð3:14Þ
DP5 ¼ Electric polarity;
DP6 ¼ Motor drive;
DP7 ¼ Rechargeable battery;
DP8 ¼ Electric power supply system:

The relationship between {FRs} and {DPs} is given by

fFRsg ¼ ½DM fDPsg ð3:15Þ

98 N. P. Suh

For the eight FRs and DPs chosen for OLEV, the design matrix [DM] is given in
Eq. (3.16) by checking whether or not a given DP affects the specific FR:

ð3:16Þ

Design Matrix [DM] for the OLEV.

As mentioned before, X indicates that an FR is affected by a DP. For example, DP2

affects FR1 in the design represented by Eq. (3.16).
Our part of the project is to design the wireless power transmission system from
an underground power supply system to moving or stationary buses, i.e., FR2.
Other FRs for moving buses can be satisfied using existing technologies.
Decomposition of FR2 and DP2:
At the time this project was undertaken, there was no existing technology for FR2
(transfer electricity from an underground electric cable to the vehicle) and DP2
(wireless power transfer system). Therefore, we have to design a new system that
can transmit heavy electric power (>100 kW) over a considerable distance
(*25 cm) by identifying children of FR2 and DP2 through the decomposition of
FR2 and DP2.
In physics, we learned that when electric current flows through conducting wire, a
magnetic field is created around the wire. Conversely, if we wrap copper wire
around a rod made of ferromagnetic material (e.g., a ferrite) with high permeability
and send an electric current through the electric coil surrounding the ferrite core, a
magnetic field is created in the rod. One end of the rod becomes the “north
(N) pole” and the other end becomes the “south (S) pole” just like the North pole
and South pole of the earth’s magnetic field. Between the N pole and the S pole, we
create a magnetic field, emanating from one pole and terminating at the other pole
as shown below for a U-shaped ferrite core as shown in Fig. 3.10. By using
alternating current in the coil, we generate an oscillating magnetic field above the
ground. If we then catch the alternating magnetic field at the bottom of the bus, we
can transfer the power wirelessly if we capture the magnetic field energy and can
generate electric power on the bus.
3 How Do We Design? 99

Fig. 3.10 Design of the wireless electric power transfer system based on “Shaped Magnetic Field
in Resonance” (SMFIR) principle

We may decompose FR2 into second-level FR2s as follows:

FR21 ¼ Create an alternating magnetic field above the ground;

FR22 ¼ Control the shape of the magnetic field;
FR23 ¼ Control the power level of the magnetic field;
FR24 ¼ Pick up the energy of the magnetic field by the vehicle; ð3:17Þ
FR25 ¼ Confine the electromagnetic waves between the vehicle and the
underground power supply systemðshielding of EMFÞ;
FR26 ¼ Deliver the electricpower to the vehicle while it is in motion:

Figure 3.10 shows the magnetic field outside of the electromagnet with a ferrite
core with high permeability. It illustrates the concept of sending power over a
distance H wirelessly. First, a two-dimensional oscillating magnetic field is created
above ground by sending oscillating electric current around a ferrite core of the
underground power supply system. The distance between the magnetic poles W
determines the magnetic field shape above the ground. The greater the distance, the
higher is the reach of the magnetic field. Once the field shape is determined, the
field strength can be adjusted by controlling the electric power that generates the
100 N. P. Suh

magnetic field. When we place a ferrite core above the ground (attached to the
vehicle) opposite to the ferrite core below the surface, the shape of the magnetic
field above the ground changes due to the high permeability of the ferrite core, i.e.,
more magnetic field permeates through the ferrite core above the ground. The
oscillating magnetic field energy in the ferrite core is then captured by a set of
electric conductors “wrapped around” the ferrite core of the power pickup system
attached to the vehicle. Because the power pickup system is tuned to have the same
natural frequency as the power supply system below the ground, the magnetic field
forms a continuous loop from the power supply system to the power pickup system
because they are in resonance. Note that the power transfer efficiency is maximum
only when the power pickup system is in resonance with the underground power
supply system. Only at resonance, a continuous magnetic loop is created from
Magnetic Pole A to Magnetic Pole B through the top power pickup system without
interference and a phase lag. We tune the system by varying the capacitance and
inductance of the power pickup unit on the bus.
For the magnetic field to reach a certain height, H (FR22 in Fig. 3.10), we can
adjust the width between the magnetic poles W (DP22). If we use an alternating
electric field to generate an alternating magnetic field above the ground, we can pick
it up by using a conducting coil attached to the bus. The frequency is chosen to
maximize the electric power transmission at a minimum energy loss during the
transmission. The chosen frequency is between 20 and 60 kHertz, because, at a
higher frequency, the loss of power may be too significant. This technology, which
was developed at KAIST, is named SMFIR (shaped magnetic field in resonance).
DP2 was decomposed into second-level DPs as follows:
DP21 = Electromagnet design—ferrite core inside electric field;
DP22 = Distance between magnetic poles (W);
DP23 = Amplitude of the electric current that generates the magnetic field around the
underground ferrite core;
DP24 = Resonating magnetic energy pickup unit on the vehicle;
DP25 = (Passive or active) shield for stray electromagnetic field;
DP26 = Two-dimensional magnetic field that does not vary along the direction of
vehicle motion.

DP2 was decomposed into second-level DPs as follows:

3 How Do We Design? 101

DP21 ¼ Electromagnet design  ferrite core inside electric field;

DP22 ¼ Distance between magnetic polesðWÞ;
DP23 ¼ Amplitude of the electric current that generates the magnetic field
around the underground ferrite core;
ð3:18Þ
DP24 ¼ Resonating magnetic energy pickup unit on the vehicle;
DP25 ¼ ðPassive or activeÞshield for stray electromagnetic field;
DP26 ¼ Two-dimensional magnetic field that does not vary along
the direction of vehicle motion:

The constraints that the design could not violate were

C1 ¼ Maximum allowable EMF level : 62:5 mGauss;

C2 ¼ Maximum weight of the pickup unit;
C3 ¼ Electric shock resistance of the system;
ð3:19Þ
C4 ¼ Temperature rise should not exceed 20 degrees Celsius;
C5 ¼ High magnetic permeability of the core material m;
C6 ¼ Minimize power loss:

The design matrix for FR2s and DP2s is given in Eq. (3.20).

ð3:20Þ

The design at the second level of FR2 and DP2 is a decoupled design. Therefore,
DPs must be varied in the specific sequence specified in Eq. (3.20), i.e., DP23
should be set first to satisfy FR23 before setting the value of DP21 to satisfy FR21.
Figure 3.11 shows the schematic arrangement of the SMFIR design that is given by
the design matrix.
To satisfy FR21 (create an alternating magnetic field above the ground), DP21
(electromagnet design—ferrite core inside the electric field) is the ratio W/H, which
must be much larger than 1, i.e., W/H > > 1. FR25, the shielding of EMF, can be
satisfied by either reactive or passive shielding. Passive shielding would consist of
placing a barrier in the ground and grounding the EMF picked up around the
receiving unit; reactive shielding would include generating a signal that is opposite
to the EMF emanating from the receiver unit.
102 N. P. Suh

Fig. 3.11 Schematic design of SMFIR. To have the strongest field extend from the underground
magnetic pole to the pole of the pickup unit on the vehicle (FR21), L should be much larger than
H. Having the electric current flow perpendicularly to the ferrite cores (parallel to the direction of
motion of the vehicle) satisfies FR26

The actual physical arrangement of the electric coil is shown through which the
current flows to generate the magnetic field in the ferrite core by aligning magnetic
domains. Because the ferrite core has low magnetic permeability, the magnetic field
is concentrated in the core. One end of the core becomes the S pole and the other
becomes the N pole. Since the electric current that flows in the cable is alternating
current (AC), the magnetic field above the ground also oscillates. The magnetic
poles alternate between N and S. To create a magnetic field that is strongest above
the ground, pointing toward the vehicle rather than directed toward the magnetic
poles, the distance H from the magnetic pole in the ground to the poles of the
pickup unit on the vehicle should be smaller than the distance W between the
magnetic poles of the underground power supply system (see Figs. 3.10 and 3.11).
The frequency of the field must be carefully chosen. It should be high enough to
reach the necessary height and low enough to minimize loss of energy in the
conductors and ferrite core. The initial frequency chosen was 20 kHz. The wire for
electric current in the underground cable has to be Litz wire, which is a type of
cable used in electronics to carry alternating current at frequencies up to about
1 MHz, to reduce the skin effect and proximity effect losses in conductors. For
high-speed trains, the chosen frequency was 60 kHz. The advantages of using
60 kHz are as follows: if the current of the embedded power line is maintained the
same as for the 20 kHz case, the cost and weight of the power pickup system can be
reduced by about 1/2 of the 20 kHz design, because the induced voltage will
3 How Do We Design? 103

increase by three times. On the other hand, if the induced voltage of the power
pickup system is the same as the 20 kHz case, the cost and weight of inverter and
embedded power cable can be decreased by about 1/2 because the current of the
inverter and embedded power cable decreases by 1/3.
SMFIR is unique in its transmission of electric power to moving vehicles, i.e.,
FR26. This feature is achieved by creating a two-dimensional magnetic field
(DP26) that is not a function of the direction of vehicle motion, i.e., the magnetic
field generated is a two-dimensional planar field perpendicular to the vehicle’s
direction. Therefore, the pickup unit mounted on the vehicle sees the same magnetic
field while the vehicle is moving, independent of the vehicle’s position along the
direction of motion, and the vehicle’s motion does not affect power transfer and its
efficiency. Thus, OLEV receives electric power while in motion or stationary.
Furthermore, the height of the magnetic field is controlled by placing magnetic
poles at a pre-determined distance apart.
The power transfer efficiency of SMFIR decreases when the magnetic poles of
the receiver unit mounted on the bottom of the vehicle are not aligned with the
poles of the underground power supply system. However, because these poles are
far apart (*25 cm), the transmission efficiency is not too sensitive to slight
misalignment, unlike the magnetic induction between two circular rings.
Many wireless power transfer systems use circular coils for both the transmitter
and the receiver, which requires that the centers of two coils are well aligned, and
the coils close together for maximum power transfer. Such a device can be used
only when a vehicle is stationary. FR24 (pick up the energy of the magnetic field by
the vehicle) and DP24 (resonating magnetic energy pickup unit on the vehicle) must
be decomposed to the next level to define the lower level FR24s and DP24s. One
way to decompose the FR24s is
FR241 = Control the impedance of the pickup coil to that of the incoming magnetic
field;
FR242 = Control the natural frequency of the pickup coil to match that of the magnetic
field;
FR243 = Control the temperature of the pickup coil within an acceptable range;
FR244 = Control the flow path of the magnetic field.

DPs must be selected to satisfy these FRs. Since the pickup unit must be
designed using a coil and ferrite core, it can be characterized as an RLC circuit built
inside a ferrite core. We can choose capacitance to satisfy FR241, inductance to
satisfy FR242, a convective air channel for FR243, and a shaped ferrite core for
FR244. Many physical configurations are possible, all variations on this basic idea.
There are also many constraints at this level of design that cannot be violated. The
details of these design issues are discussed in detail in subsequent chapters.
104 N. P. Suh

3.5.7 Design of Ultra-Precision Lithography Machine

One of the most advanced technologiesis the lithography machine that prints
electric circuits (e.g., transistors) on semiconductor chips, either silicon-based or
gallium arsenide III–V semiconductors. Commercially, many of the current
lithography equipment use a 193 nm UV light source. Lithography machines that
use UV lights can print billions of transistors on postage-stamp-size silicon. The
density of these transistors is so high that the storage density of memory space and
the speed of computation are no longer the limiting factor in many applications.
Also, transistors with dimensions as small as 5 to 7 nm can be printed on semi-
conductors, using EUV (extreme ultraviolet) light source with a wavelength of
13.5 nm. A commercial machine is shown in Fig. 3.12. These machines are so
precise that everything used in these lithography machines must be extremely
accurate. It is difficult to measure the curvature of lenses and alignment of all
components, etc., used in lithography because of the ultra-precision required. The
light source must have a short wavelength to be able to make the critical dimensions
of the line width of the circuit as narrow as possible. Now the preferred technique is
to use EUV (extreme ultraviolet) as the light source, which eliminates the need for
multiple deposition/etch processes and multiple masks.5 The ambient air in these
fabrication facilities in which these lithography machines are operated must be
almost particle-free. Also, to increase productivity, the power of light source
matters. The industry has been trying to increase the brightness of EUV to 250
watts.
To increase productivity, the components in these lithography machines must
move at very high speeds. For instance, the silicon wafer must be loaded on a platen
by a robot, and then the platen moves the wafer at high speeds to the location where
the wafer will be subjected to UV light to make the circuit. The platen must attain a
top speed from a stationary position. As given by Newton’s second law, when the
platen and associated components are accelerated very fast, it involves large forces.
These forces make the machine vibrate, which may affect these precision lithog-
raphy machines.
Exercise 3.4:
Consider a platen attached to a linear electric motor, on which a silicon wafer is mounted.
Suppose that the platen moves with an acceleration of 30 g. We cannot allow any vibration
of the machine, because we are printing electric circuit patterns of 10 nm in width. Design
a mechanism that eliminates the mechanical vibration of the device under these operating
conditions.

5
There used to be several companies that used to make these lithography machines. A couple of
Japanese companies (Canon and Nikon), ASM Lithography (ASML) of the Netherlands, and a few
others were significant suppliers of lithography machines. Now ASML dominates the lithography
business globally because of its advanced technologies. In 2016, ASML sold about 140
lithography machines, making its market share to be around 60%. Each lithography machine can
cost approximately $110 million. It takes an entire Boeing 747 cargo plane to transport the
machine.
3 How Do We Design? 105

Fig. 3.12 A lithography machine for semiconductor wafer manufacturing (Permission granted by
ASML)

3.6 Design of Organizations

Public and private organizations perform diverse functions in society and a nation.
They constitute systems that deal with a variety of different problems and issues
they confront during their operation. The design of their organization determines its
effectiveness. Some of these organizations are government agencies and private
entities, either for-profit (e.g., industrial companies) or non-profit organizations
(e.g., universities). These organizations function within the laws governing orga-
nizational entities of a country, which are different from country to country and
from region to region. In the United States of America, the constitution, which is
the ultimate guarantor of the rights, goals, and limits of these organizations, is an
amazing design of governing principles.
In this section, the design of two different kinds of organizations will be illus-
trated. In Chap. 21, the design of other organizations will be more thoroughly
examined, including the design of universities. One of the significant differences in
designing organizations is the primary role of people in an organization. In many
cases of organizational design, often the DPs and PVs involve people with free will.
Therefore, sometimes, organizations do not perform as designed initially because
people in the organization may not function or behave as assumed initially.
106 N. P. Suh

3.6.1 Design of a Government Organization

Suppose someone is suddenly appointed to head up a government agency that

supports engineering research at universities throughout the country, although she
has not served in the government before. What should she do?
The simple answer is that she must go through the four domains of the design
world, which were presented in Chap. 2 and design the policies, procedures, and the
organization within the broad bounds and constraints that exist within the overall
government policy and within the bounds established by the law. Once she
establishes the policies and programs for her organization, she must administer
them well to serve the public to the best of her ability. One significant constraint, at
least in the case of the United States government, is that she does not have much
time to do all these things unless she works hard and prepares her actions even
before she assumes the job. As the common saying goes, “you have to run as soon
as you hit the ground.”
Before she assumes the job, she may have time to prepare her plan of action
while waiting for her nomination by the president of the country after the back-
ground check, which may take about 6 months (in the case of the presidential
appointment in the United States, clearance by security agencies and presidential
personnel office is needed), and confirmation by the U.S. Senate. During this time,
it will be highly desirable to acquire the current and historical information about the
agency she is about to join and think about what she wishes to achieve in the job. If
she has no idea as to what she wants to do to contribute to the nation, it will be wise
not to take on such an assignment. Then she must plan! She may wish to identify
the problem(s) the agency must address in the customer domain, and then formulate
specific goals by establishing the FRs of the organization. Talking to important
constituents of her agency may be a good idea to get a sense of how they would
react to her designs as well as learn about their thoughts and concerns. This process
is needed to find out the problem she must deal with in the customer domain.
After FRs are established in the functional domain, she must create an organi-
zational structure (i.e., DPs) in the physical domain that will enable the execution of
the chosen FRs. Then, in the process domain, she has to establish PVs (resources,
budget, and personnel), which will be needed to support the DPs. This way we are
mapping from the customer domain to the functional domain, from functional
domain to physical domain, and finally from the physical domain to process domain
(where resource issues such as funding and personnel are addressed). As she
transforms from domain to domain, she must check to be sure that there are no
coupled designs (i.e., the coupling of FRs as per the Independence Axiom). Since
she will need financial and personnel resources, she should secure the necessary
resources and support of other government agencies of the U.S. government and the
U.S. Congress. She then should execute the policies and programs logically, fairly,
and legally for the public good, as she swore to do at the time of her appointment.
3 How Do We Design? 107

Case Study 3.1: The Story about the Design of the NSF Engineering
Directorate:
The purpose of this case study on organizational design is to illustrate the
following:

1. importance of knowing the PROBLEM(s) as discussed in Chaps. 1 and 2;

2. importance of defining FRs to address the problem;
3. developing an organizational structure in terms of DPs to fulfill FRs;
4. importance of addressing resource issues: personnel and budget;
5. importance of addressing potential criticisms that usually follow when decisions
are made that change the direction of an organization.

These five issues should be addressed in the design of organizations, because in

many situations of organizational design, most attention is given to DPs, i.e., the
creation of organizational structure, without articulating the goals of the organi-
zation in the form of FRs. Once FRs and DPs of the organization are appropriately
addressed, one of the most critical tasks of the person heading up the organization is
related to resource issues, i.e., budget, staffing, and facilities.
Item 5 is unavoidable when changes are introduced to any organization. Since
some have benefited by the existing organization, any change made may affect them
negatively or in some cases, adversely. Sometimes, some of those affected negatively
by the changes introduced may even take personal attacks. It can consume a great deal
of mental and physical energy to deal with negative attacks, either overtly or behind
the scene. One way of avoiding Item 5 is to maintain the status quo and maintain
existing organizations and try to make the influential stakeholders “happy” until the
end of the tenure. Perhaps such a person should not take on the job since organizations
must move forward to serve the people for whom the organization exists.

A Real Story:
The National Science Foundation (NSF) is a United States (U.S.) government
agency that provides research funding primarily to universities in the United States.
It was established in 1950 to enable universities to continue to make the critical
scientific and technological contributions to the country as they had done during
the Second World War. NSF supports research in science, engineering, and edu-
cation. Its budget is relatively modest in comparison to other R&D agencies such
as the National Institute of Health (NIH), Defense Advanced Projects Agency, and
U.S. National Aeronautics and Space Agency (NASA). However, it is the largest
agency that supports only extramural research at universities, whereas NIH and
NASA have important intramural programs. It promotes research in physics,
mathematics, chemistry, biology, materials, engineering, and education. The
Engineering Directorate was established in the late 1970s. Before that, it was part
of the Mathematics, Physics, and Engineering Directorate.
In the early days of its establishment, basic scientists dominated NSF, perhaps
because, during the Second World War, scientists played critical roles in
108 N. P. Suh

developing new weapons. Beginning in the 1970s, NSF attempted to strengthen the
support for engineering research. However, NSF had difficulties in defining its role
in engineering. By 1984, the United States had formidable competition from
abroad, especially in manufacturing-related industries and NSF was under pres-
sure to strengthen its engineering programs. To chart a new course for engi-
neering, President Ronald Reagan appointed an engineering professor as the first
presidential appointee in charge of the Engineering Directorate of NSF as Assis-
tant Director (AD) of NSF for Engineering. What should he do to strengthen the
engineering community of the United States?
Introduction to the NSF Engineering Directorate:
NSF was established to achieve the following three primary missions: (1) to pro-
mote progress in science and engineering; (2) to provide health, welfare, and
prosperity to the people; and (3) to secure the national defense.
In the early days of NSF, engineering was part of the Directorate for Mathe-
matics, Physics, and Biology. Then, it became a separate directorate in the late
1970s. Even then, engineers at NSF defined its role as “applied science,” and thus
fields such as design and manufacturing were not supported. Furthermore, it
dwelled on engineering issues of the first half of the twentieth century, even though
microelectronics, semiconductors, and biology were revolutionizing the world of
science and technology. The then-existing policies were in part established because
well-known engineering professors had undue influence on the NSF Engineering
Directorate, emphasizing those topics that were important in the first half of the
twentieth century. This practice forced young professors to work on old engi-
neering problems to get funding. The people in the White House, who were
responsible for science and engineering policies in President Reagan’s adminis-
tration, thought that the NSF should strengthen its support for engineering R&D to
make the United States more competitive in the world.
Problems Identified in the Customer Domain:
The problems the new AD for Engineering identified through extensive consultation
with professionals in industry, academia, and the government were the following:

1. The NSF mission was, as defined in the NSF Act of 1950, (1) to promote
progress in science and engineering; (2) to provide health, welfare, and pros-
perity; and (3) to secure the national defense. The NSF Engineering Directorate
was primarily serving the interest of existing engineering educational programs.
It was organized like a mirror image of a typical engineering college and forgot
about their primary mission related to the future well-being of the nation.
2. Funding was given primarily to well-known senior professors in engineering
who were solving (or resolving slightly different approaches) the problems of
the past to improve the accuracy rather than dealing with the future progress of
engineering and technology in a rapidly changing world. Young professors who
were yet to establish their reputation had difficulty in getting NSF grants. To
write credible research proposals with many references, they tended to extend
their doctoral thesis work rather than venturing into new challenging fields.
3 How Do We Design? 109

3. Research in the design and innovation of technologies was neglected entirely.

4. Emerging and critical technologies were neglected, although these areas will
become important in the future. Many grants supported engineering problems of
the first half of the twentieth century.
5. The budget was allocated in proportion to the number of proposals received in
a given area. Consequently, new fields and creative ideas did not get much
funding, because the number of proposals in these emerging and critical areas
was small, and therefore there was no budget allocated for newly emerging and
innovative fields. For example, there was no funding for research in design or
emerging technologies (e.g., biotechnology). As a result, many universities did
not hire professors in these emerging engineering fields.
6. Cross-disciplinary research did not get any funding, including the fields in which
the collaboration between academia and industry is of paramount importance.
7. Although NSF is a government agency to deal with national issues, NSF
Engineering was not providing funding research areas that are important for
the U. S. competitiveness in the twenty-first century. Funding went to those who
submitted thick proposals that received high scores in peer review rather than
proposals with creative and risky ideas.
8. The number of women in engineering was inadequate because they shied away
from engineering because of its traditional image, i.e., engineers wearing hard
hats.

These were the problems that had to be dealt with to be sure that NSF fulfills its
mission, as outlined in the NSF Act of 1950.
Design Solution:
Defining FRs in the Functional Domain:
Based on the problems identified, the FRs for the NSF Engineering Directorate
were defined as follows:

FR1 ¼ Advance engineering science base;

FR2 ¼ Support engineering fields where the science base needs to be developed;
FR3 ¼ Support emerging technologies;
FR4 ¼ Advance critical technology areas for U:S:competitiveness;
FR5 ¼ Promote collaboration between universities and industry;
FR6 ¼ Strengthen the support for young researchers;
minority researchers; and women researchers:
ð3:21Þ

Constraints: There are many constraints when one attempts to reorganize any
organization, especially government agencies. It is especially difficult when the FRs
deal with newly emerging subjects for which there are no large groups of
researchers and professional societies (e.g., at the time of reorganization, nan-
otechnology). Furthermore, if the budget has to be changed by more than a fixed
amount, it has to be approved by the director of NSF, the Office of Management and
110 N. P. Suh

Budget of the White House, and the Congressional Committee that oversees NSF,
which may take a long time for approval. An equally important constraint was
personal. He promised his family that he would work at NSF for only 1 year and
return home, mostly for financial reasons. Therefore, the reorganization had to be
done during the first 3 months of his tenure at NSF. No one believed that such an
ambitious redirection of a government agency could be achieved in less than 2 or
3 years. It was reorganized, received the approval of OMB and Congress for
budgetary changes, and implemented in 3 months, a record that defied the pre-
vailing culture of Washington! The new structure and policies of the NSF Engi-
neering fundamentally changed the NSF engineering support structure.
When a new direction for an organization is implemented, what people often
search for are the DPs and PVs rather than FRs. DPs, which are created to achieve
FRs, are represented by an organizational chart. PVs are resources such as per-
sonnel and budget that are needed to support DPs. However, to design an effective
and efficient organization, we must establish FRs first and then develop DPs. That
is, in the design of organizations, DPs are the organizational entities that exist to
achieve (or satisfy) a given set of FRs. For example, DP3 that must satisfy FR3 (i.e.,
“support emerging technologies”) could be called the “Division for Emerging
Technologies.”
Exercise 3.5:
(a) Define DPs for the above set of six FRs chosen for the new NSF Engineering Direc-
torate. DPs will be entities called “divisions” within which we will have “programs” for
various sub-topics of engineering. For example, DP1 could be “Division for Engineering
Sciences.” Remember that the DPs must satisfy the Independence Axiom, i.e., FRs must
remain independent from each other by creating proper DPs that do not couple FRs.
(b) Construct the design matrix. Is the design coupled?
Exercise 3.6:
Based on the DPs chosen, design an organizational structure. Remember that the Division
Directors must report to AD for Engineering. In each division, there may be many
“Programs,” each with a Program Director. The total budget for the Engineering
Directorate is $800 million a year. How and on what basis should the budget be allocated
to various FRs?
Essence of the Design Process for Organizational Design
We demonstrated through this organizational design was to show the process of applying
the design methodology using the design of the organization as an example. The steps
involved were in the GPFDP sequence: (1) establish the goal, (2) define the problem,
(3) establish FRs to solve the problem identified, (4) develop DPs to satisfy the FRs, and
(5) identify PVs (i.e., resources in the form of personnel and financial resources). We have
to make sure that the FRs are not coupled to each other during this design process. We will
now repeat this process over and over again for all sorts of different designs involving
machines, processes, and systems, in addition to other organizations.
It is good to remember that unlike the design of machines and manufacturing processes, the
design of organizations that affects people can become controversial, especially if the
budget has to be cut in some favored traditional areas. Money matters even in academia,
often over-ruling their rational judgment! Often these well-established fields have
3 How Do We Design? 111

champions—often better known scholars, who believe that their field is the most important
among all academic disciplines. Such a conviction is often necessary, but we need a
balanced perspective in all matters we deal with, but usually, our behavior follows instincts
rather than a fair rationale. This NSF AD, who did not fully appreciate the sensitivity and
politics of the academic community, was severely criticized for his actions by some pro-
fessors. Sometimes, this is the price one has to pay to achieve what appears to be the right
thing to do to achieve the institutional goals.
Micro-Management of an Organization
If by mistake, one creates an organization that is a coupled design (i.e., an organizational
structure that couples FRs to each other), the person who is in charge of any one of the FRs
cannot decide without affecting other FRs that are coupled to it. In this case, the decisions
made by one of the division heads may affect those made by others, creating conflicts and
tensions among them. Then the person who is in charge of all FRs ends up managing all the
details of the entire organization. Such an organization cannot function effectively. Such a
manager is sometimes called a “micro-manager,” which should not be taken as a com-
pliment! An organization that must be micromanaged tends to be inefficient as well as
ineffective, often with unhappy and frustrated staffers.
Exercise 3.7:
Suppose that you have been asked to be the president of your university to make it to make
it one of the best research universities in the world. What are the problems the university
has you must deal with to achieve the goal of making it one of the leading universities in the
world? What are the FRs you will try to satisfy? What DPs would you choose to satisfy the
FRs? Where would you get the necessary resources (PVs) you will need to achieve your
goals? Show how you would perform the PFDP transformation to make your university a
great university.

3.6.2 Design of High-Technology Industrial Firms

Most people would agree that during the past 40 years, there have been explosive
growths of new high-technology industries, especially in the United States. Many
well-known companies, such as Microsoft, Google, Intel, Apple, Broadcom,
Samsung, Boeing, and Amazon, have changed the world. The combined revenue of
these companies is far larger than the total income of more traditional companies
such as General Motors, General Electric, IBM, Hitachi, Ford, and BMW. Some of
the new companies have demonstrated that new industrial firms based on tech-
nology innovation can make significant contributions to society and make their
investors wealthy. Therefore, many young students hope to start their own com-
panies during their careers.
Only a few new venture firms, out of thousands, become successful. Many factors
determine the outcome of new ventures. However, the most successful companies
are those that have designed their business goals and strategy well. They were also
“lucky” in terms of timing, financing, and market receptivity for their products. The
most important elements behind the successful companies are the correct identifi-
cation of the “problem,” selection of a right set of FRs, selection of appropriate DPs
in the physical domain, and securing (or generation) of financial resources (PVs) for
proper execution.
112 N. P. Suh

Case Study 3.2: Creation of Successful High-Technology Company

Problems identified in the Customer Domain:
A well-known high-technology firm grew rapidly because of its superior propri-
etary products, dominating some of the high-technology product areas. They made
computers, semiconductors, digital printers, measurement devices, and others. They
competed with other high-technology firms, which had also diversified into many
product areas. Then, new firms that only specialized in a limited number of product
areas entered the market place, competing with this large company with many
different products in highly competitive markets. To be more competitive, this old
firm with an extensive portfolio had to invest more and more in research and
development (R&D) to stay competitive. Eventually, this large firm realized that it
must divest some of its divisions that were no longer competitive because of the
severe competition. They sold their semiconductor device division to a private
equity firm, which generates funds from institutions such as universities to invest in
profitable areas and return handsome profits to the investors.
A private equity firm, which bought a division of this large company, launched a
new semiconductor manufacturing company. They hired a brilliant and capable
CEO for this new company. The new CEO organized the new company and made it
profitable in a couple of years. Then, he took the company public selling its shares
to the public. The private equity firm sold its shares at a considerable profit and left
the company.
The CEO of this new company developed a set of FRs and corresponding DPs to
make the new company successful. If you were the CEO, what would you choose
as your FRs and DPs?
Design Solution:
The CEO of this newly formed company developed a new set of FRs for the
company. The highest level FR was the following:
FR = Become a dominant manufacturing company in semiconductor chip
products for the telecommunications and enterprise systems (i.e., storage
equipment).
The corresponding DP was
DP = High-technology semiconductor chip manufacturing company in
telecommunications and storage.
The following were the next-level FRs:

FR1 = Dominate a few selected product areas in the market place;

FR2 = Consolidate R&D and other activities to increase efficiency and reduce
cost;
FR3 = Show a substantial profit and raise the stock price;
FR4 = Acquire other semiconductor companies that are in a related business;
FR6 = Increase profit margin by offering high-performance products.

He then selected the DPs to satisfy the above set of FRs.

3 How Do We Design? 113

Concluding Note:

This new company became hugely successful. The stock market rewarded this
company by increasing its stock price. It soon acquired a larger semiconductor
company that had been around for about three decades and became one of the
dominant semiconductor device companies. This newly merged company became
one of the dominant semiconductor companies in the world. The company CEO
donated large gifts to his alma mater.

3.7 Conclusions

In this introductory chapter, many examples of design—both technical and

non-technical—are used to illustrate the design process involved in creating various
systems. It is shown that the design of systems, regardless of the specificity of the
field, is a transformational process of going from problem identification to FRs, FRs
to DPs, and DPs to PVs, in short, the “PFDP transformation.” We have illustrated
this typical transformational process for various designs such as organizations,
transportation systems, and others. During this transformational process, we must
maintain the independence of FRs as per the Independence Axiom to create designs
that are robust and easily implementable. Construction of design matrix can help in
identifying coupled designs, especially when the design involves many FRs and
many layers of decomposition. It was emphasized that the design process is com-
mon to all designs, regardless of the field of application.
Problems
1. Design an alternate steering column that may be better than the one shown in
Sect. 3.5.4.
2. In design, it was stated that we cannot make a non-diagonal design matrix into a
diagonal matrix through mathematical manipulation in most cases. Is this
statement valid?
3. In many countries, political problems arise because educational opportunities for
children depend on family income levels. Usually, better educational opportu-
nities are available to children from financially better-off families than those
from financially struggling families. Better educated people end up getting better
paying jobs and get more opportunities to advance in society. Design a system
where all children will get equal opportunities in education. State your FRs and
DPs clearly. Construct the design matrix for your design. Is your design a good
one that provides equal opportunities?
4. If you were the CEO of the above semiconductor company, what would you
pick as your DPs to satisfy the FRs? For the DPs chosen construct the design
matrix. Is your design uncoupled? Decoupled design? Coupled design?
114 N. P. Suh

5. Design a new venture firm that specializes in artificial intelligence (AI) in the
field of financial auditing of industrial firms. Identify the problem (in the cus-
tomer domain). List the FRs of your new venture firm in the functional domain
that can solve the problem. Then, list DPs in the physical domain that will
satisfy the FRs. Finally, list the investment you will need to finance your new
venture firm (PVs in the process domain). Construct the design matrix for your
design. Is your design acceptable?
6. At the Diamond Light Source of the United Kingdom, they have been devel-
oping high-energy imaging, diffraction, and scattering beamline. Its source is a
superconducting wiggler with a power of approximately 9 kW at 500 mA after
the fixed front-end aperture; two permanent filters aim at reducing the energy in
photons below the operating range of the beamline of 50–150 keV, which
accounts for about two-thirds of the total. How would you remove the heat
generated in the permanent filters? Propose a design solution.
7. One of the modern technologies that have changed the world is the airplane with
jet engines. Now we can reach almost anywhere within 24 h because these
planes fly at nearly the speed of sound, i.e., sub-sonic. They are quite efficient.
However, they produce more CO2 than other transports per passenger per mile
flown than any other transports. Therefore, we need to reduce fuel consumption
and thus CO2 emission.
The jet engines mounted typically on the aircraft wings are quite big and heavy.
The engine size is determined by the power the airplane needs to reach the liftoff
speed within a fixed length of the runway. However, once it reaches the cruising
altitude, it uses only a tiny fraction of its power, since the pressure drop across
the plane and air drag on the aircraft body is rather small. The massive engine
needed for takeoff increases the drag on the airplane. Propose designs that can
reduce the drag or the engine size.
8. The engineers at the Diamond Light Source published a paper on the design and
simulation process of the permanent secondary filter, a 4-mm-thick SiC disk
used in the system. The first version of the filter was vulnerable to cracking due
to thermally induced stress, so a new filter was designed based on an AD. The
new design solved their problem. Review the following paper and comment on
their design. Propose another design that can accomplish the same goal.

Reference: W. Tizzano, T. Connolley, S. Davies, M. Drakopoulos, Design and FEA of an

innovative rotating SiC filter for high-energy X-ray beam,, Mechanical Eng. Design of
Synchrotron Radiation Equipment and Instrumentation MEDSI2018, Paris, France JACoW
Publishing ISBN: 978-3-95,450-207-3. https://doi.org/10.18429/JACoW-MEDSI2018-
THOAMA04 [G. E. Howell Diamond Light Source, OX11 0DE Didcot, United Kingdom].
3 How Do We Design? 115

Appendix

Additional Theorems (Note: some of the theorems are related to information,

which is presented in Chaps. 8 and 11).
Theorem 8 (Independence and Tolerance)
Design is an uncoupled design when the design range is greater than
Pn @FRi
i¼1;i6¼j ð@DPjÞDDPj, in which case, the non-diagonal elements of the design matrix
can be neglected from design consideration.
Theorem 12 (Sum of Information)
The sum of information for a set of events is also information, provided that
proper conditional probabilities are used when the events are not statistically
independent.
Theorem 13 (Information Content of the Total System)
If each DP is probabilistically independent of other DPs and affects only its
corresponding FR, the information content of the total system is the sum of the
information of all individual events associated with the set of FRs that must be
satisfied.
Theorem 14 (Information Content of Coupled versus Uncoupled Design)
When the state of FRs is changed from one state to another in the functional
domain, the information required for the change is greater for a coupled process
than for an uncoupled process.
Theorem 16 (Equality of Information Content)
All information content that are relevant to the design task are equally important
regardless of their physical origin and no weighting factor should be applied to them.
Theorem 17 (Design in the Absence of Complete Information)
Design can proceed even in the absence of complete information only in the case
of decoupled design if the missing information is related to the off-diagonal
elements.
Theorem 18 (Existence of an Uncoupled Design)
There always exists an uncoupled design that has less information than a
coupled design.
Theorem 19 (Robustness of Design)
An uncoupled design and a decoupled design are more robust than a coupled
design in the sense that it is easier to reduce the information content of designs that
satisfy the Independence Axiom.
Theorem 20 (Design Range and Coupling)
If the design ranges of uncoupled or decoupled designs are tightened, the
designs may become coupled designs. Conversely, if the design range of some
coupled design are relaxed, the designs may become either uncoupled or decoupled
designs.
Theorem 22 (Comparative Robustness of a Decoupled Design)
Given the maximum design ranges for a given set of FRs, decoupled designs
cannot be as robust as uncoupled designs in that the allowable tolerances for DPs
of decoupled design are less than those of uncoupled design.
116 N. P. Suh

Bibliography
Drakopoulos M, Connolley T, Reinhard C, Atwood R, Magdysyuk O, Vo N, Hart M, Connor L,
Humphreys B, Howell G, Davies S, Tim Hill and GW, Pedersen U, Foster A, Maio ND,
Basham M, Yuan F, Wanelik K (2015) I12: the Joint Engineering, Environment and
Processing (JEEP) beamline at Diamond Light Source. J Synchrotron Radiat 22:828–838
Farid A, Suh NP (eds) (2016) Axiomatic Design in large systems. Springer, Cham, Switzerland
Gossip (2019) Template aluminum soda can isolated on white background, mockup, can for beer
and carbonated drinks. 3D rendering. Photo licensed from Shutterstock, ID: 1300349536
Lorne E (2019) Kidney stones after ESWL intervention. Lithotripsy. Scale in centimeters. Photo
licensed from Shutterstock, ID: 179862602
NIH: National Cancer Institute (2009) SEER training modules: Anatomy & physiology. https://
training.seer.cancer.gov/anatomy/urinary/components/,publicdomain
Papilionem K (2019) Geobukseon (Korean Turtle Ship) replica sculpture in Gwanghwamun agora.
Photo licensed from Shutterstock, ID: 1425185789
Suh NP (1990) The Principles of Design. Oxford University Press
Tizzano W, Connolley T, Davies S, Drakopoulos M (2018) Design and FEA of an innovative
rotating SiC filter for high-energy X-ray beam. In: Proceedings of Mechanical Engineering
Design of Synchrotron Radiation Equipment and Instrumentation (MEDSI2018) in Paris,
France, JACoW Publishing, pp 306–312
Wong A, Guo H, Kumar V, Park CB, Suh NP (2016) Microcellular Plastics. Wiley. pp 1–57.
https://onlinelibrary.wiley.com/doi/pdf/, https://doi.org/10.1002/0471440264.pst468.pub2

Further Reading

Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press

Suh NP (2005) Complexity. Oxford University Press
Suh NP, Cho DH (eds) (2017) The on-line electric vehicle: wireless electric ground transportation
systems. Springer International Publishing
Design Representations
4
Paolo Citti, Alessandro Giorgetti, Filippo Ceccanti, Fernando Rolli,
Petra Foith-Förster, and Christopher A. Brown

Abstract
The results of design activities must be transmitted to people who need them for
their tasks, e.g., manufacturing, construction, software development, etc. The
objectives of this chapter are to understand how design information should be
represented and conveyed using standards, geometric drawings, design matrices
for the complete system, DPi/DPj matrices, and industry-specific functional
diagrams. The goal of this chapter is to introduce how the design information is
typically conveyed to its ultimate user.
Proper descriptions of design must address the needs of the users of the
design results. For example, the manufacturing group may need the information
on the geometry of each part, acceptable tolerances for each dimension,
materials, the hardness of each piece, the complete assembly of the system, etc.
On the other hand, those charged with the task of evaluating and implementing
the design may need information on the entire assembly of parts, operating
procedure, power requirements, etc. To facilitate these processes, different
professional groups have established commonly used methods, conventions, and
practices.
The “design information” is typically represented using representation
methods that are used in a given profession, sometimes adapted by each
company to deal with their specific needs. This chapter reviews some of the

P. Citti
A. Giorgetti (&)
F. Ceccanti
F. Rolli

Department of Engineering Science, Guglielmo Marconi University, Rome, Italy
e-mail: [email protected]
P. Foith-Förster
Fraunhofer Institute for Manufacturing Engineering and Automation IPA, Stuttgart, Germany

C. A. Brown
Department of Mechanical Engineering, Worcester Polytechnic Institute, Worcester, MA,
USA

© Springer Nature Switzerland AG 2021 117

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_4
118 P. Citti et al.

fundamental representation methods of design that have been developed by

various professional groups, typically non-government entities. For instance,
there are national professional organizations such as the American Society of
Mechanical Engineers (ASME) that have established the standards for certain
products such as pressure vessels and boilers to assure the safety of certain
products. Globally, there is the International Organizations for Standardization
(ISO), an international non-governmental organization that has established
voluntary international standards, which facilitates world trade by providing
common standards worldwide.
In this book that emphasizes Axiomatic Design (AD), the relationship
between functional requirements (FRs) and design parameters (DPs) is the basis
for product design. In AD, the design process begins with the identification of
FRs first, followed by the development of DPs, which are specifically chosen to
satisfy the FRs. Therefore, in AD, the relationship between FRs and DPs forms
the core of design representation, in addition to the representation of geometric
shapes in the case of the design that involves solid objects. A design matrix is a
form of design representation that describes the relationship between the
functions and physical entities. The design matrix between FRs and DPs is the
most effective means of identifying the coupled designs that are to be avoided in
AD.
To highlight the powerfulness of the design matrix representation and the
wide applicability of AD, several families of representations, as stated above,
have been considered. In particular, the chapter is structured in such a way to
explain, in a first instance, what should be the connections between designing
with AD and representing the results. The concept of module and tolerance will
be introduced. Therefore, representation families will be presented: standard
mechanical drawing, piping and instrumentation diagram (P&ID), and software.
A case study is presented as well, to bring a real example of a complete
application of AD. The choice to illustrate both mechanical drawing and
software representation comes to the authors’ will to emphasize that the design
process should follow a structured approach, in particular, the AD one,
regardless the nature of what is designed.
Proper descriptions of a design must address the needs of a variety of users of
the design information. Some may only be interested in knowing the functional
and physical relationships in terms of FR and DP hierarchy. Some may need to
exact geometric details of the designed parts in terms of DPs, their tolerances,
the geometric shape, and their relationships. Some may need the information on
the assembly of DPs, i.e., information on DPi/DPj relationships. The objectives
of this chapter are to describe how design information is typically represented
and conveyed using standards, geometric drawings, design matrices, DPi/DPj
matrices, and industry-specific functional diagrams.
4 Design Representations 119

4.1 Introduction: Design Representations and Their

Purposes
A Real Story
In the late 1950s, a financially struggling engineering student got a job in a small local
machine shop near his university. It was a god-sent job because he was struggling
financially to pay for his dormitory room and meals. The job paid much better than what he
was getting from other odd jobs. It was also unusual. His job was to draw up the machines
the head machinist built without any drawings. He would look at a machine or a pho-
tograph of a machine, and proceed to make a similar machine by machining the parts with
steel or aluminum without any drawings. This practice was contrary to what his school
taught him. He was taught to draw the assembly drawing as well as part drawings for
someone to make the part based on the drawings that had nominal dimensions and tol-
erances. However, his boss at this company proceeded to produce parts without drawings.
He took a photograph of a machine that was similar to what he had in mind. Then, without
any drawings, he would proceed to make various machines such as vacuum-forming
machines, punching press, molds, and so on. For a few years, the company liked him,
because without hiring any designer or engineer, he produced to manufacturing equipment
for vacuum-formed plastic parts and punching press quickly. The company wanted to have
a record of what he had made. The company decided to hire someone to draw up what he
just made without any drawings! That is the reason the company hired this young aspiring
engineer.
This young “engineer” went around the machines, examining the parts the “chief engi-
neer” had already made to produce paper documents for what he had already made. It was
a very difficult job! Measuring parts that had been already machined and assembled, and
then drawing them on a piece of paper was not only dull but could easily be misleading,
because the geometric information did not reveal functions of the part, i.e., FRs. He could
only measure nominal dimensions and had no information on the tolerances . He just
assumed that the tolerance was around ±0.005 inches in the case of linear dimensions. As
the company’s business grew, the management realized that they could no longer depend
on this archaic method of making machines and machined parts without designing the
machine or the part first on paper. The management changed their policy: design machine
parts first on paper and then make the machine based on the drawing that had all the
information on materials, dimensions, and tolerances. This new approach was much more
reliable and less expensive than the previous method, i.e., start cutting a piece of metal
without the benefit of having drawings that specified tolerances and nominal dimensions.
In some ways, history repeats itself. It turns out that a similar situation existed in the late
eighteenth century. Then, muskets were made without standardization and interchangeable
parts. Artisans uniquely created each gun, and none of the parts could be exchanged with
similar parts in other guns made by the same manufacturer. Every musket was “custom
made.” Therefore, they could not supply the number of muskets needed for wars. Then, the
idea of interchangeable parts was finally introduced, which required drawings of the parts
with nominal dimensions, tolerances, and material specifications, which were needed to
standardize the components and increase manufacturing productivity. Each piece was
made separately within a tight tolerance and assembled with other mass-produced parts
because they were interchangeable.
Mechanical technologies have come a long way since then. Now in semiconductor man-
ufacturing, we are dealing with nanometer-scale dimensions with fractions of nanometer
accuracy for mass production of integrated circuits (IC) with a very high yield of
acceptable products. The required accuracy of these chips is a fraction of nanometers.
Indeed, these amazing technologies would not be possible without the capability to
120 P. Citti et al.

represent the design graphically for a single part as well as a system of parts. The
drawings must carry all the information needed to manufacture the part. Sometimes,
extensive documentation is required to specify all the conditions that must be satisfied in
manufacturing these parts that produce integrated semiconductor parts.

Almost everything humans do or produce is designed first, either formally or

informally. Representations of design embodiments are required elements of
design. Final design representations communicate design solutions to others. They
represent the outcome of syntheses processes. When the design is done for a client,
the design representation is an integral part of the contract. When the design of a
product is done for manufacturing within a company, it becomes the basis for
designing and planning the production system or the factory.
Sometimes, especially during the early phase of design, the design is sketched as
an abstract and idealized model of the design concept for future implementation.
They describe reality in a simplified and pragmatic way to allow for scientific or
technical analysis or description or exploration. Final design representations should
communicate the details of the design intent, results of analyses, calculations,
alternatives, and decisions carried out during the design phase. There are three main
uses of representations during the life cycle of a product:

1. support for design processes and documenting progress;

2. final documentation of completed design solutions;
3. inputs for manufacturing process design and quality assurance.

First, design solutions are not created for the first time after design problems are
completely solved. Representations evolve with developments of solutions during
zigzagging decomposition processes between FRs and DPs, as described in the
previous chapters. Second, representations guide design processes. They represent
progress in solution developments. Also, they can provide a record of how solutions
were reached. Usually, more than one candidate for a design solution should be
considered. Reasons for selections and rejections of candidate solutions should be
recorded. Good representations of evolving designs help designers follow effective,
structured design processes that can build on experience. Third, representations of
designed products are instruments for transferring manufacturing-relevant infor-
mation from designers to manufacturing and industrial engineers, who are
responsible for designing production systems and processes, including quality
assurance.
Good design representations do not, per se, mean that design solutions are good.
Design representations merely document the design embodiment. They do not alter
it. A poor design could be well documented and still be insufficient to accomplish
the CNs. Good design solutions start with good problem statements, followed by
appropriately structured, functionally oriented design processes. Based on the CNs,
different FRs and, consequently, different DPs, might be considered for good design
solutions.
4 Design Representations 121

Design representations use special semantics for explaining things that have
been designed, like process, objects, and functions. These special semantics can be
combined in different ways for different purposes. Different semantic systems are
used, based on what is designed. Typically, representation systems are chosen for
their effectiveness in communicating the features of each design process
step. Conventional graphical representations of product design solutions, such as
2D and 3D drawings, e.g., as produced by CAD software, are well-established
notation systems used to represent an object’s physical dimensions, i.e., DPs. Note
that while CAD often refers to “computer-aided design,” this is not “design” in the
sense used here. CAD is rather computer-aided drafting or drawing.
Drawings fail to record design intents, i.e., FRs. Drawings show only physical
aspects of components and assemblies. Drawings also fail to explain how DPs
influence FRs and which ones they influence. These deficiencies in graphical-only
design representations make it impossible to modify DPs without risking unin-
tended consequences, a result of violating axiom one, albeit inadvertently because
the FR-DP relation modules are not communicated. It is essential to make extra
efforts, integrating design intent, FRs, and FR-DP modules into design represen-
tations, to represent solutions more fully.
Axiomatic Design (AD) offers more complete and useful representations of
design solutions, through FR-DP modules and hierarchies. These are the semantics
for functional–physical representations of design solutions. AD’s module-junction
diagram and flow diagram further deliver object and process representations of
design solutions.
The following section elaborates on AD semantics for representations that
facilitate robust, adjustable, controllable design solutions that avoid unintended
consequences and have the highest probabilities of success.

4.2 Design Representations for Applying the Axioms

4.2.1 Axiom 1, Maintaining Independence, Function–

Physical Representations, and Module Junction
Diagram and Flow Diagram

Complete design solutions need to indicate the design intent for each physical
feature, and they need to show the interactions between physical solutions and
design intents. In AD processes, this starts by writing text to record FR–DP pairs.
Relations between FRs and DPs are then described mathematically in the design
equation within a design matrix.
Design matrices progress with design processes. Design matrices encapsulate
design decisions, and the progress made toward complete design solutions. Full
design matrices contain all the individual design equations between each FR and all
the DPs that influence it.
122 P. Citti et al.

Design matrices can be used to check quickly for the satisfaction of the Inde-
pendence Axiom. FR–DP interactions can be summarized together in design
matrices with Xs and 0s, representing significant interactions or lack of them, as
described in the previous chapters. If a particular DP is influencing a certain FR, it
is indicated by an X in the appropriate position in the matrix or, not, shown by an 0.
An uncoupled design has a matrix with Xs only on the diagonal. DPs can be
adjusted in any order to fulfill FRs without iterating to correct unwanted influences
of DPs on FRs. Off-diagonal Xs indicate unwanted influences of DPs on FRs.
Triangular matrices show decoupled designs, which must be adjusted in a specific
order to avoid iterating. If a matrix indicating unwanted influences cannot be
arranged by linear algebra to be triangular, then the design is coupled.
Knowing, quantitatively, how each FR is influenced by each of the DPs inde-
pendence can be maintained as prescribed by axiom one, adjustability and con-
trollability are provided and unintended consequences can be avoided.
AD provides a unified representation, including the various levels of decom-
position of FRs, based on modules. Modules are defined in terms of FR–DP or DP–
PV relationships. A module is the row of the design matrix that calculates the value
of an FR when it is provided with inputs of corresponding DPs. Hence, the design
modules show how each FR is satisfied by its corresponding DP and may be
affected by the other DPs.
An FR is satisfied if there is a function M such that FR = M(DP). Therefore,
interactions among use cases, DPs, can be interpreted as system inputs. For sim-
plicity in notation, the operator * can be introduced. This operator represents, in
equivalent modular form, mapping relationships (FR–DP or DP–PV), even for
non-linear systems at high abstraction levels. Then the following simplification can
be used: FR = M(DP) = M * DP.
In complex systems, consisting of many modules, this simplification is useful for
identifying the types of relationships existing between the various modules. Three
types of relationships can be considered. The relations between the modules of a
system are defined as uncoupled (S—summation junction) when FRs are mutually
exclusive, i.e., independent. This situation is equivalent to having a diagonal matrix
design (Eq. 4.1):

   

FR1 a DP1 FR1 ¼ M 1  DP1
¼ ! ð4:1Þ
FR2 b DP2 FR2 ¼ M 2  DP2

Figure 4.1 illustrates the graphical representation in terms of the flowchart

diagram for the system of Eq. (4.1).
In practice, relations between modules of systems correspond to connections
parallel to these modules.
Similarly, relationships among system modules are defined as decoupled (C—
control junction), if the system of equations corresponds to a triangular design
matrix. In this case, connections between modules of a system are a bit more
complex.
4 Design Representations 123

M1
S

M2

Fig. 4.1 Flowchart diagram for an uncoupled system where M1 and M2 modify DP1 and DP2 to
satisfy FRs 1 and 2, respectively, and S is a summation junction

   
FR1 a DP1
¼
FR2
b c DP2
FR1 ¼ aDP1 ¼ M 1  DP1 ¼ f ðDP1 Þ
! ð4:2Þ
FR2 ¼ bDP1 þ cDP2 ¼ M 2  DP2 ¼ f ðDP1 ; DP2 Þ

where
 
DP1
M2 ¼ b þc ð4:3Þ
DP2

This path starts from module M1 and develops to module M2.

The modules corresponding to the same DPs can be connected in series. They
are sub-assemblies of modules arranged with respect to the execution of main
information paths of systems. Therefore, overall graphical representations of sys-
tems are equivalent to flowcharts, constituted by subsets of modules in series,
connected in parallel. Modules M1 and M2 are connected in series. This means that
the output of M1 constitutes the input for M2. Figure 4.2 shows the graphical
representation as a flowchart diagram of the system in Eqs. 4.2 and 4.3.
Connections among system modules are coupled (F—feedback junction) when
FRs are not independent. In this case, system behavior can no longer be represented
by simple connections of modules in series or in parallel. In order to define the
system status, some modules need to have a return. In these situations, there is a
feedback relationship between the various modules. In terms of matrices, connec-
tions of this type can be represented by scattered matrices.

   

FR1 a d DP1 FR1 ¼ aDP1 þ dDP2 ¼ M 1  DP1
¼ ! ð4:4Þ
FR2 b c DP2 FR2 ¼ bDP1 þ dDP2 ¼ M 2  DP2

where
124 P. Citti et al.

M1 C M2

Fig. 4.2 Flowchart diagram for a decoupled system

M1 F M2

Fig. 4.3 Flowchart diagram for a coupled system where F is a feedback junction

8 
< M 1 ¼ a þ d DP1 ¼ f ðDP1 ; DP2 Þ
 DP2 ð4:5Þ
: M 2 ¼ b DP1 þ c ¼ f ðDP1 ; DP2 Þ
DP2

In graphical form, Eq. 4.4 is represented by the flowchart in Fig. 4.3.

This module representation with AD is particularly interesting. It can be rep-
resented graphically as a module-junction structure diagram or as a flowchart. In
this treatment, there is recourse to representations of system modules, like flow-
charts. This typology of representation has extensive use. It can also be used in
industrial and mechanical systems design. It is a representational equivalent of a
design matrix. Moreover, modular representations can be particularly effective in
describing complex systems, consisting of components of different natures.
Flowcharts of system modules facilitate integrated designs of non-homogeneous
parts. Each module constitutes an abstract entity. However, these diagrams give
considerable relevance to connections among various components of systems,
regardless of their nature. In this way, it is possible to pursue integrated designs,
which allow the contextual and coordinated development of mechanical, electronic,
and software components. In fact, the decomposition of FRs can be carried out to
define designs of individual components and interfaces between components.
Design matrices and modules for each FR that compose them need to connect
with other representations can be applied in each engineering field. Mechanical
drawings, process maps, etc. are included with design matrices for more complete
representations of design solutions.

4.2.2 Axiom 2, Minimizing Information Content

and the Concept of Tolerances

DPs are physical solutions to design problems. They can be almost any sort of thing
because the science of design is applicable in practically all fields. DPs can include
physical and electrical components, industrial processes, healthcare delivery, and
4 Design Representations 125

monitoring of operations. Regardless of the application, in almost all designs, there

will be parameters more critical than others for success. There can be uncertainties
in determining independence and in estimating the probability of success for cal-
culating information contents for the candidate solutions. These uncertainties can be
enough, such that, even starting from common CNs, it is possible to get different
design solutions of indistinguishable quality.
The success of a complete design is measured by its ability to satisfy the original
CNs. Success is measured in decompositions by abilities of DPs to fulfill FRs and
of PVs to produce DPs. In most cases, CNs are represented by things like satisfying
needs, component life, sustainability, and interfaces with other components. Based
on these, and considering potential constraints, such as cost, maximum weight and
maximum dimensions, many different design solutions that satisfy all points and
axiom one might be developed within specified tolerances. Probabilities of
achieving these tolerances can be used for calculating information contents and
ranking solutions to determine the best design through Axiom 2. Metrics are
essential for this.
Good metrics reduce uncertainties in determining information content. With
metrics, all potential design solutions can be ranked by their information contents,
i.e., probability of success. Uncertainty in knowing probabilities of success limits
the ability to distinguish information contents for candidate solutions, which limits
the application of information content, i.e., Axiom 2 for ranking candidate
solutions.
Short Metrics Examples
To understand how metrics can work in design, imagine designing an electrified transport
system to replace traditional internal combustion buses and taxis in a downtown envi-
ronment. Regardless of other DPs to be chosen, the power line voltage delivered to the
transport unit must match the electric motor specifications. Power drops or spikes that
could damage the motors need to be addressed. The power supply should be as constant as
possible, within limits that the motors can withstand. The tolerance for the power supply
variance depends on what the motors can tolerate while maintaining functionality. These
considerations establish tolerances on DPs.
Consider, as another example, mechanical components subjected to cyclic bending loads
that can cause fatigue failures, like a shaft. Fatigue cracks generally initiate on the surface
and cycle by cycle; they grow until the component fails. Maximum stress on shafts sub-
jected to bending loads is initially on the surface, until a crack starts, then they are at the
crack tip. Surface integrity is critical because surfaces are where cracks initiate. Toler-
ances on surface topographies should be specified by designers to ensure the required
component life. Typically, this is done by specifying maximum values for roughness
parameters. Because surface characterizations are vital for design success, the roughness
characterization parameters to be used in each situation should be specified by designers.
The selection of topographic characterization parameters to be specified for a given sit-
uation is based on several considerations, including materials, processes used to create the
surface, and surface accessibility, which should all be specified in representations of design
solutions.
An approach to reducing uncertainty is stress estimation. The best solution can be selected
by evaluating the performance of each candidate design with finite element method
(FEM) analyses. In which case, metrics, such as maximum stress, maximum strain, and
126 P. Citti et al.

maximum displacement, can be used to determine information contents of each design

solution.
Another approach could be to build prototypes to test design solutions physically in real
working conditions, i.e., pilot tests. Different kinds of metrics, such as the number of cycles
to the break, maximum load withstood, and hours to failure, can be adopted to score
candidate design solutions.

Note that different detailed metrics, such as maximum stress and the number of
cycles to failure, address standard high-level performance metrics, such as com-
ponent life. From this perspective, simulations and tests can be used either to enrich
designers’ knowledge for improving design solutions or to confirm and freeze a
design solution concluding the design phase and starting the production phase.
Iterative steps can be design phases, where the axioms are applied to design rep-
resentations consistently through all attempts, which is why design representations
are essential during design development phases.
A brief history of the concept of tolerance
The concept of tolerance was born during the first industrial revolution to supply guns for
the military more efficiently. Before that moment, “quality control” was carried out directly
by artisans during fabrication. Therefore, there was no need to check dimensions and
shapes, because, in the case of incongruencies, artisans would correct them immediately, to
get the desired components and systems functioning properly.
When demands for products increased, production paradigms evolved. From a one-person
artisanal work, goods started to be produced in sub-assemblies. The creators of this
production method are usually identified in Honoré Blanc and Eli Whitney. Sub-assemblies
were pre-assembled modules of components designed to be put together when the final
product was needed. Also, modules could be mixed to make different kinds of guns. This
production paradigm modification, even though it seems natural, introduced several
aspects completely ignored until then. One of them was tolerancing. Because modules were
supposed to be assembled, their dimensions and all their interfaces needed to satisfy
interface and clearance requirements.
This need was satisfied by thinking of dimensions of modules such that, with all the
stack-ups of dimensional variations, the resulting modules could be assembled into a
properly functioning gun. This apparent smart dimensioning is the first massive application
of the modern tolerance concept.

The concept of tolerance is fundamental in design because it connects the design

world, which is always an abstraction with semantics modeling reality, and the real
world, with uncertainties and constraints. Tolerancing is fundamental to design
because calculations, considerations, and selections are usually carried out only in
ideal and simplified environments. The real world is, on the contrary, an imperfect
and complex environment in which designed products will live their entire lives.
Therefore, design solutions should consider intrinsic, real-world imperfections, and
provide robust choices for robust design solutions. Like design performance, there
are metrics for design robustness. Axiom two is all about probabilities of success in
achieving tolerances and robustness to the selection of design solutions.
Whenever there are critical DPs, regardless of what is designed, their values
require tolerance intervals. Tolerance intervals are the ranges in which values can
4 Design Representations 127

vary, while satisfying the CNs. Tolerance interval definitions comprise manufac-
turability, performance, safety, and lifetime.
Regardless of why tolerance is used, designers should identify DPs, i.e., features,
that need special tolerances, which require special PVs, because these features can
increase information content. Specification of tolerances that challenge capabilities
of conventional PVs can introduce additional costs in prototyping and manufac-
turing. Therefore, the number of tolerances specified should always be defined
considering the criticalities of specific applications. They should be defined
according to the target performance. Target performance defines tolerances, and
tolerances define cost. Therefore, if tolerances are defined, based on CNs and FRs,
to be as large as possible and still supply required values, the costs of products
come down.

4.2.3 V-Model, Design Process Representation to Guide

a Structured Design Process

In conventional industrial practice, designs often evolve iteratively from a first

prototype without a dedicated analysis of FRs and feasibility testing of design
solutions, as can be achieved by applying the axioms. The success of such a design
process relies purely on the designers, who might find satisfactory solutions by
chance or experience. Systems-engineering methods promote representations of
actual design processes to support designers with a structured approach. Instead of
luck or designers’ experience, design processes should be driven by decomposition
of design process functions and concerted design process steps.
A popular design process representation in industry is the V-model. Originally
defined for software development, it breaks the design process into three phases
(Fig. 4.4). The alignment of the phases in the form of a V gives the procedure its
name.
During the first phase, a specification and decomposition of FRs lead down to
the level of a detailed design of system modules. This becomes the basis of the
implementation phase, located in the tip of the V, where design representations (i.e.,
system models) are created and realization takes place. Different product domains
may be realized independently (e.g., mechanical, electronical, and software real-
ization). The third phase is dedicated to integration and testing. It is located on the
right flank of the V-model. Consequently, each of the first phase’s specifications
and decomposition steps on the left flank of V has a testing and integration
counterpart.
The V-model can be run through several times during a design project.
Depending on the design progresses, outputs of the V-model could be laboratory
prototypes, functional prototypes, pilot-run products, and so on, until the final
design stage is reached.
The benefit of the V-model is its representation of a structured,
requirements-oriented, design process. Product architectures are defined as real-
ization begins to integrate components from functional decompositions into
128 P. Citti et al.

Requirements Product

Requirement assurance of properties Acceptance

analysis Testing

Concept & System

Architecture Testing

Detailed Testing and

Design Integration

Modelling
I and III
Specification Realization Integration
& Decomposition & Testing

II Implementation

Fig. 4.4 V-model design process representation (as introduced by Gausemeier and Moehringer
(2002))

physically realizable units. Another benefit of the V-model is that it promotes

testing for multiple levels of integration of physical subsystems. However, the
V-model misses iterations between phases. However, in AD, needs for iterations are
largely eliminated by application of the axioms during functional decompositions
and physical integrations. Its underlying concept is that of a sequential design
process. As a rule of thumb, early testing and user involvement in the evaluation of
the design are discouraged until the late design phase of the right flank of V.
AD also promotes a structured design process, focusing on customer needs,
functional–physical decompositions, and physical integrations. The design process
is represented by customer, functional–physical, and process domains, populated by
zigzagging between them (Fig. 4.5). Through application of the information axiom,
every single mapping of an FR to a DP solution into the physical domain is
associated with verification of its probability of success in satisfying its respective
FR. To find out about system ranges, testing of design solutions can be indis-
pensable. Consequently, testing can be included in design steps in AD, and
zigzagging between domains allows for iterations when necessary.
AD processes continue through the domains as required, to reach a final design
stage via prototypes of increasing maturity and detail, until the solution is obvious.
4 Design Representations 129

CNs FRs DPs PVs

Customer FuncƟonal Physical Process

Domain Domain Domain Domain

Fig. 4.5 Axiomatic Design domains

4.3 Examples of Design Representations in Industrial

Environment

This section introduces example design representations: The first example shows
how a technical drawing is created from a design decomposition and matrix. The
second explores piping and instrumentation diagrams.

4.3.1 Mechanical Product Design: From the Design Matrix

to the Mechanical Drawing

Consider a vehicle braking disk design solution starting from an FR “brake the
vehicle.” The example shows how, starting from a high-level FR “Brake the
Vehicle,” it is possible to complete the detailed design of the component brake disk.
A brake disk (Fig. 4.6) is part of a sub-system, the braking systems, that allows
vehicles to stop. Braking is an FR to be developed for vehicles. A portion of the
design matrix is shown in Table 4.1.
This high-level representation is not detailed enough for drafting a technical
drawing, or for generating a solid model. Therefore, only by considering lower
level FRs and DPs it is possible to draft a technical drawing. The line related to FRn
is expanded in more detail in Table 4.2.
This example shows how technical drawing drafting processes should proceed.
The design brake disk sub-assembly is represented by its technical drawing
(Fig. 4.7), annotated with numbers corresponding to FR–DP pairs. Ideally,
decompositions are taken far enough so that specific dimensions and tolerances can
be calculated, before completing component drawings.
130 P. Citti et al.

Fig. 4.6 Brake disk

Table 4.1 A section of the design matrix for braking

DPn−1 DPn DPn+1
… Brake the system …
FRn−1 …
FRn Brake the vehicle X
FRn+1 …

Table 4.2 A portion of the decomposition of the brake system

… DPn DPn.1 DPn.1.1 DPn.1.1.1 …
… Brake Disk Self-cooling Determine …
system brake disks roughness
system level
… … … … … … … …
FRn Brake the vehicle … X …
FRn.1 Provide adequate … X …
breaking power density
FRn.1.1 Dissipate friction heat … X …
FRn.1.1.1 Maintain stable contact … X …
between sliding surfaces
… … … … … … … …

In Fig. 4.7 DPs are highlighted in red with arrows to the features. Physical
integrated in of DPs are depicted. DP numbers link to the FRs, which indicate
design intents. Knowing design intents, features on designed artifacts can be
changed, without risking unintended consequences. Definitions of actual
4 Design Representations 131

Fig. 4.7 Technical drawing of the brake disk and identification of DPs (in red)

dimensions, such as the disk thickness, disk diameters, and the number of holes for
connecting disks to vehicles, are defined with consideration of other FRs and their
modules, showing how they are influenced by the DPs.

4.3.1.1 Mechanical Drawing General Rules

Semantics for representations of design solutions, i.e., rules for interpreting sym-
bols, are discussed here for mechanical drawings. There are many rules to be
followed, including standardized communication methods, for drawings. These
rules make the many kinds of technical representations of design solution lan-
guages, like musical scores. These representations transcend ordinary human lan-
guages. Just as it is not necessary to speak Italian to play music written by Vivaldi,
it is not necessary to speak Italian to understand technical drawings and manu-
facture components for Fiat. Technical drawings are special methods of commu-
nicating, and similar rules are followed, regardless of what is designed.
Consider mechanical drawings, like the braking disk example. In the following
example, we examine the most important aspects of this kind of design represen-
tation. The first consideration about mechanical drawing is how 3D components are
represented on paper in 2D. The answer is with multiple views.
Consider parallel projections. These look at a component from an exact point
and remove perspective effects, i.e., vanishing points, which consider perceived
diminishing of sizes with distance from the observer. Components are represented
by looking at them from as many views as needed to understand its entire shape.
The next question is about how views are arranged on drawings. There are two
standardized conventions shown in Figs. 4.8 and 4.9, the first angle projection
according to ISO, the International Standards Organization, and the third angle
projection according to ASME, and the American Society of Mechanical Engineers.
132 P. Citti et al.

Fig. 4.8 Projection angles

Fig. 4.9 ISO and ASME representations

Knowing which convention, ASME or ISO, is used in the drawing is fundamental

to understanding it properly.
Usually, notes are inserted in title blocks as shown in Fig. 4.10. These notes are
a crucial part of any mechanical drawing. Title blocks are usually placed on the
bottom right side of a drawing. They contain all the notes necessary for under-
standing the drawing content, such as the projection angle, component name, and
scale.
Component dimension can be either explicit or implicit, determined by adding or
subtracting explicit dimensions. There are only two rules to follow when dealing

Fig. 4.10 Title block

4 Design Representations 133

with component dimensioning: include all dimensions, and only define dimensions
once, either explicitly or implicitly, never both.
Consider details of all the different dimensions reported in the brake disk in
Fig. 4.7, particularly:

• Numbers express linear dimensions. The measurement unit can be either mil-
limeters or inches. Inches all have the symbol, ″, after the number, e.g., 3″.
• Indications, such as 36°, express an angular dimension, as in the disk example,
this is the angular difference between two adjacent holes connecting the disk to
the boss.
• Numbers proceeded by the symbol, ø, e.g., ø219, indicate a diameter. Diameters
can also be represented with a number without the ø symbol, but in that case the
measures should be referred to a diameter of a drawn circumference, e.g., 206 in
the brake disk example, which means that all the hole centers are placed on a
circumference of 206 mm in diameter.
• When dimensions are followed by smaller number values with a sign or signs,
such as 8.15±0.05, it means that a dimensional tolerance has been specified, on
that specific measure, which must be satisfied when making that component.
This means that the nominal dimension 8.15 mm can vary within the range
specified by the tolerance, in the example ±0.05 mm. The specification of
dimensional tolerances usually indicates that those dimensions are important for
the function of that component. Dimensional tolerances are given only where
they are important for fulfilling FRs.
• When a dimension is followed by a multiplication symbol and a number, such as
8.15±0.05(10), it means that the dimension is valid for all the ten similar
features in the view. In the brake disk example, the holes that fix the disk to the
boss are specified on only one feature and apply to the rest of them, although
written only once.
• Surface topographies are defined through specific symbols indicating surface
roughness at fine scales, waviness at larger scales and lay, or directionality,
according to ASME standard B46.1 on surface texture. There are many topo-
graphic characterization parameters that represent different geometric compo-
nents of roughness, waviness, and lay (see also ISO 25178, ISO 21920, ASME
B46.1, and Y14.36M).
• Usually, surface properties, mainly textures, are indicated by placing symbols on
drawings of surfaces that are specified. The number above the triangle tradi-
tionally is the arithmetic average roughness, Ra, the most used topographic
characterization parameter. In Fig. 4.11, the symbol means Ra should not exceed
12.5 µm.

Another fundamental attribute found on mechanical drawings is geometric tol-

erances on form. Unlike dimensional tolerances, these are applied on surfaces, and
not on dimensions, like lengths, widths, and heights. Geometric tolerance on form,
as the name suggests, specify how much the actual shape of a surface can differ
from nominal. As with dimensional tolerances, surface tolerances are only specified
134 P. Citti et al.

Fig. 4.11 Traditional indication of surface roughness

Fig. 4.12 Example of a geometric tolerance on component form for a surface

when shapes are critical for function. An example of a geometric tolerance is shown
in Fig. 4.12. The face indicated by an arrow should have a flatness of 0.05 mm.
This means that all points on that surface should be between two ideal planes
0.05 mm apart.
Geometrical tolerances are divided into two groups: self-referenced and
cross-referenced. Flatness is a self-referenced tolerance. An example of a
cross-referenced tolerance is parallelism, shown in Fig. 4.13.
In cross-referenced tolerances, there is always a reference surface, called the
datum, which is labeled A in Fig. 4.13. Tolerances are specified with respect to the
datum. In Fig. 4.13, the face indicated by the arrow should be parallel with respect
to datum A with a tolerance of 0.1 mm. This means that all the points on the surface
subjected to the tolerance should be between ideal planes 0.1 mm apart, parallel to
the datum A.
4 Design Representations 135

Fig. 4.13 Cross-referenced geometric tolerancing

Artifacts may have more than a datum, like a plane, a face, and a center. If
possible, the datum is independently defined. On other applications, each datum
needs a certain sequence of definitions. Therefore, the design will be decoupled and
a sequence on drawing and manufacturing needs to be followed. For example, a
datum center may depend on a face datum. If specifications do not follow the
correct sequence for the datum, then dimensions and tolerances will form coupled
designs. Coupling of dimensions and tolerances incur extra manufacturing costs.
Costs can arise because drawings define lower values for tolerances than necessary,
and because specified designs are impossible to manufacture. Impossible designs
cause parts to be rejected without the manufacturing enterprise understanding why.
It is because a design problem caused a manufacturing problem.
Moreover, in assemblies without movements between components, geometric
tolerances should be defined for worst-case scenarios, which allow tolerance
bonuses. The worst-case scenario in GD&T is the maximum material condition
(MMC), which is identified in the tolerance by a circle with an M inside. For
example, a hole could be as short as allowed and a stud that should fit in it is as
large as allowed. Maximum material conditions, in many cases, provide the worst
case. This condition allows manufacturing to save non-conforming parts that can be
reworked, to ensure that when tolerances are at their worst condition, the compo-
nent still functions properly. These are just two examples of the most common
geometrical tolerance that could be necessary for design solution representations.
Note that not all the kinds of dimensioning listed above have the same importance.
Figure 4.14 shows some symbols commonly found in technical drawings. They
are potentially required to successfully represent design solutions, i.e., DPs. For a
detailed description of each symbol refer to mechanical drawing books and
manuals.
Tolerance independence is another fundamental principle to be considered.
Dimensional tolerances and geometric tolerances should be considered separately.
This influences design processes. When dealing with real components, both
dimensional and geometric tolerance specifications might be required at the same
location. In this case, careful definitions of tolerance intervals can avoid conflicts
136 P. Citti et al.

Fig. 4.14 List of geometric tolerances

with axiom one. Complete dimensional information is vital for representing DPs.
Surface properties and tolerances are specified only when they are required to
assure function, because they incur manufacturing costs.
4 Design Representations 137

Fig. 4.15 Example of a section representation along A-A left

In principle, physical tolerances on DPs should be derivable from functional

tolerances on the FRs, and traceable through design equations. These equations are
design modules that show how an FR is influenced by DPs. FRs should have
functional tolerances that indicate the acceptable range for satisfying the CNs.
Tolerances can be mapped to DPs using the appropriate module.
When an FR is influenced by more than one DP, and DPs influence more than
one FR, then tolerances on the DPs might need to be considered with two or more
FRs. The partitioning of the tolerances should follow axiom two so that they
provide the least information content for the system.
In practice, tolerances are often based on experience with similar designs.
Guidance can also be found in the literature. For example, ANSI/ASME B4.1 is a
standard guide for tolerancing cylindrical features. It gives textual descriptions
corresponding to functions, FRs, and then it gives physical tolerances for DPs, so
that components function as intended.
Representations of designed components have been considered, i.e., how to
describe sizes and shapes of DPs with dimensions and tolerances. These concepts
are valid for any possible view used in a drawing. Components with internal
features are considered below.
Internal features are not visible from outside. To view inside a component
section is used. A section is a view, or portion of a view, in which a component is
opened virtually to see inside. All the above concepts can be applied to exposed
interiors. There are two ways to show interiors: traditional section views and
break-out section views.
Traditional section views consist of virtual cutting of whole components fol-
lowing the plane specified through a section line. Section lines can be either simple
or dashed. Section lines are defined on a view, and section drawings consist of
additional views of cut components.
138 P. Citti et al.

The left side of Fig. 4.15 has a section line. Section lines are typically long
dash-dot lines extending beyond components where two arrows, one per extension,
indicate the section orientation. Arrows at section line extremities in Fig. 4.15
indicate that the section shows the right side of the component. On the right side of
the drawing, the section itself is shown. Usually section views are indicated by a
name like “Section X-X,” where X is the letter near the arrows indicating the
section orientation, e.g., Section A-A.
In section drawings, all intersections between cutting planes and solid material
are represented by filling with a crosshatch pattern. Different types of crosshatch
patterns are used to indicate different parts and materials, and therefore a discon-
tinuity in filling indicates an interface between two parts. In this way, section views
are characterized by section lines in one view, and actual sections in separate views.
Break-out section views help to reduce the number of views required for a
complete representation. Break-out sections are defined directly on views on which
sections are made. Therefore, there is no need for an additional view in which the
section line is defined. Figure 4.16 illustrates the advantages of break-out sections
in drawing simplification.
The choice of one or the other kind of representation is completely up to
designers.
The most important aspects are drawing readability and understandability.
Drawings represent design solutions, which are products of design phases, they
should be clear and free from misunderstanding. Representations complying with

Fig. 4.16 Break-out view

4 Design Representations 139

Fig. 4.17 Example of a system made with multiple components: Gearbox

axiom two should minimize the information content and in so doing, maximize the
probability of success in correctly communicating intentions of design solutions.
Intentions are clearer when corresponding decompositions are supplied with the
drawings. This is an essential requirement for any good AD solution representation.
The last topic regarding the design representation of mechanical components is
assembly drawings. These show how elements of decompositions are physically
integrated, with relationships between multiple components, in one drawing. An
example of a gearbox is shown in Fig. 4.17.
Assembly drawings follow the same rules already discussed for component
drawings, such as view orientations and section views. Assembly drawings, how-
ever, have different scopes than component drawings. They satisfy different CNs for
the representation of design solutions. Component drawings are intended to com-
pletely represent single components, with all the details of its dimensions and
shapes.
Typically, components are physically integrated, i.e., assembled, into systems
that fulfill FRs at a higher level. Satisfying CNs and fulfilling FRs typically require
an assembly of components, and design representations should include assembly
drawings.
As mentioned above, assembly drawings and component drawings have different
purposes. Therefore, they contain different kinds of information. Assembly draw-
ings should identify components that comprise a system, and show how these
components are arranged in a complete assembly. An example of an assembly
drawing is shown in Fig. 4.18.
Individual components are identified with numbers, in balloons. Code numbers
in the balloons should indicate numbers of individual DPs as they appear in the
decomposition.
140 P. Citti et al.

Fig. 4.18 Example of an assembly drawing of a gearbox

Lists of components, also called a bill of materials (BOM), are important

because they reference what is needed to deliver functioning products.

4.3.2 Industrial Environment: Piping and Instrumentation

Diagram (P&ID)

Piping and Instrumentation Diagrams (P&ID) come with well-defined semantics in

the language of process industries. They represent details of piping, process
equipment, instrumentation, control devices, and all their interconnections. P&IDs
are chosen as an example here because their applicability is wide in industry, and
because they represent connections between design processes and representations.
P&IDs show how a designer wants to arrange equipment, instruments, control
devices, etc., and how to connect them to get a system to provide its FRs. Many
systems are represented through P&IDs, e.g., water cycles inside a power plant that
uses steam turbines, oil inside hydraulic/oleodynamic actuators, and an automatic
fire-extinguishing system.
Every P&ID representation is detailed and characterized by specific symbology.
Because of their high level of precision in representation, P&IDs are suitable and
effective in many applications, such as
4 Design Representations 141

• planning and construction of power and manufacturing plants;

• operations processes;
• maintenance and modification of processes;
• reference documentation for mechanical technicians and safety personnel;
• reference documentation for HAZOP (hazard and operability study) studies; and
• control of documentation formally issued at various stages of the project.

It is interesting to note that this kind of design representation has been thought to
be exhaustive from many perspectives at the same time. Looking at P&IDs, it is
possible to understand simultaneously how systems are arranged and controlled,
how to plan maintenance, and what the critical parts are.
P&ID representation of an air supply system is shown, as an example, in
Fig. 4.19.
Air supply systems are used in industries to obtain a clean air supply for the
operations, e.g., pneumatic circuit and conditioning systems. An array of instru-
ments is used to condition air supply, and each component has a specific task. The
system functions as follows:

• air passes through a filter (used to clean the air) and arrives to a compressor;
• a compressor (actuated with a motor) takes filtered air and compresses it;
• compression makes the air hot, so a cooling system is needed (aftercooler);
• a cooling process condensates moisture from the air. So, moisture needs to be
removed with a separator;
• then the air is sent to a receiver, which removes the dirt and oil from it;
• the air is sent to the dryers to remove the remaining moisture.

Fig. 4.19 Example of a P&ID representation of an air supply system

142 P. Citti et al.

• Finally, the air pressure is regulated to target values for operational needs.

Each component is represented in a specific way usually referred to normative

(pneumatic system, hydraulic/oleodynamic, electric, instrumentation devices, etc.).
As briefly observed in the example, reading a P&ID diagram can be complicated
when readers are not so confident with this kind of representation. However,
symbols are standardized, and their meaning is listed in ANSI/ISA S5.1 and ISO
14617-6. Furthermore, each component symbol can be enriched with detailed
information, such as dimensions, lengths, and temperatures.

4.4 Case Study: Design and Manufacture of Mechanical

and Chemical Polishing Machine (CMP) by Four
Graduate Students

Background Story: A major industrial firm in the semiconductor manufacturing

equipment industry was interested in getting into a related but new business in the
semiconductor processing equipment industry. After considering several possibil-
ities, they decided to buy a small company that manufactured equipment that
polished semiconductor surfaces between deposition processes to make the surface
“atomically” smooth for the next manufacturing step. This process is called CMP
(chemical–mechanical polishing), which polishes the semiconductor surface after
each deposition of a new layer of materials to make a semiconductor device. After
each round of deposition of materials on the semiconductor surface, it was polished
flat before the next deposition can commence.
This business was thought to be growing. In order to increase the density of
devices and shorten computational times, deposition layers were increasing, and
dimensions of devices were getting smaller. They wanted to buy a small company in
Southern California, but they were asking for an unreasonable price. Therefore, a
decision was made to let MIT design and build a prototype machine to learn the
intricacies of CMP processes and consider the possibility of going into the business
with a better tool. For MIT, it was an opportunity to teach students how to design
such a complicated large commercial scale machine based on the AD theory. It was
determined that such a machine could be built in about 2 years while teaching
graduate students how to design and manufacture such a precision machine. CNs
were to demonstrate the CMP process using a machine that is better than existing
commercial machines.
Five graduate students were assigned to the project. The students had finished
undergraduate studies at several different universities and had just entered grad-
uate school at MIT. None of them had any prior industrial experience. To provide
proper faculty supervision, an adjunct professor with years of industrial experience
was invited to supervise the students. The professor in charge of the project
believed that the students could do a successful job and learn what they need to
know by executing the project, because he had developed complicated industrial
4 Design Representations 143

processes and mass production systems when he was a senior at an engineering

school. He believed in his students.
One student was assigned the development of the control system for the complete
system, including building circuits and devices, although he majored in mechanical
engineering. Another student was in charge of designing the entire mechanical
system in cooperation with two other graduate students. One of the students, Jason
Melvin, did such a great and extensive work that his doctoral thesis committee
accepted his thesis as a doctoral thesis rather than letting him go through the
typical routine of getting a master’s degree first and then writing another thesis to
receive a Ph.D. degree. Some of his PowerPoint slides are added below for
illustrative purposes.
The students started by defining the highest level FRs based on what they
learned by studying industrial machines on the market. They decomposed to the
lowest level FRs and DPs, developing the design matrix during the decomposition
process. In Fig. 4.20, the design of one of these components and the FRs and the
DPs are shown. The according flow diagram for the spindle design is shown in
Fig. 4.20.
The finished disk had to be perfectly flat without dished-in parts for subsequent
manufacturing operations, using optical imaging to create transistors and other
devices. One critical variable in getting an optically flat surface is controlling the
pressure exerted on the disk. The decomposition of FR3.2.1.2 is done to control the
pressure distribution, as shown in Fig. 4.21.
Figure 4.22 shows the detailed assembly design for wafer career. It enables the
transport of the wafer and applies the right pressure distribution during the pol-
ishing operation.
Note that only a few parts are shown to illustrate the design process based on
AD.

• Theorem 3: Upon decomposition, M2 expands

to include modules M2.x
• Corollary 1: DP2 = DP2.1

DP2 M2 FR2
speed input spindle speed

DP2 M2.1 M2

M2.3 M2.4 M2.5 M2.6 M2.7

FR2
M2.2

Fig. 4.20 Spindle flow diagram. (Reproduced with permission from Melvin (2003))
144 P. Citti et al.

Rigid Plate Rubber Membrane Wafer

P/Pnom
DP 3.2.1.2.2: Pressure
1
distribution

DP 3.2.1.2.1: Membrane
compartment areas

DP 3.2.1.2.4: Compartment
divider vent length & I.D.

DP 3.2.1.2.3: Front membrane

thickness, hmem
Polishing Pad

Fig. 4.21 Design of the system that controlled the radial polish rate. (Reproduced with
permission from Melvin (2003))

Isolation bellows
DP 1.1.1.5.5

Compartmentalized membrane
DP 1.1.1.5.3, DP 3.2.1.2.1,
DP 3.2.1.2.3, DP 3.2.1.2.4
Retaining ring flexure
DP 3.2.1.1.1, DP 3.2.1.1.2, DP 1.1.1.3.1.2
Retaining ring
DP 1.1.1.3.1.1

Fig. 4.22 Design of wafer carrier components. (Reproduced with permission from Melvin
(2003))
4 Design Representations 145

Fig. 4.23 The design matrix for the CMP machine. (Reproduced with permission from Melvin
(2003))

The design matrix is shown in Fig. 4.23, indicating if there is any coupling,
which requires re-design. Based on the design matrix, Melvin made the necessary
and appropriate changes to be sure that the machine would function as intended. In
many firms, they make the part and try it out to see if it works appropriately, which
is a costly and time-consuming process, rather than checking it during the design
stage to find any design flaws done by checking the design matrix.
The final assembled machine (right) and solid modeling (left) are shown in
Fig. 4.24. The control system was designed and built by Douglas Lee as part of his
thesis for his Science of Master’s degree in mechanical engineering at MIT. The
machine was designed, built, and tested in 2 years. CNs to develop a
better-than-market CMP process machine could be met. Industry sponsors were
surprised that students with no prior industrial experience could achieve such a
feat in 2 years. The secret is:
THROUGHOUT THE DESIGN PROCESS, DO NOT CREATE A COUPLED
DESIGN! CONSTRUCT DESIGN MATRICES TO BE CERTAIN THAT THERE IS
NO COUPLING IN THE DESIGN DECISIONS MADE!
146 P. Citti et al.

Fig. 4.24 The completed machine. The CAD model on the left and the final fabricated machine
on the right. (Reproduced with permission from Melvin (2003))

4.5 Software Design

In previous sections, traditional design representation techniques have been intro-

duced and analyzed. Traditional, in that they are referred to representations of
classic components and processes. Because these have been studied for a long time,
the basic ways in which they are represented are largely frozen.
In the last decades, however, less traditional products have experienced expo-
nential growth in terms of numbers, relevance, and fields of application, ranging
from industry to everyday use. In fact, one of the most important products, maybe
the most important product, developed, produced, and sold in the last 50 years, is
software. Today, almost everything is managed by, or must interact with, software.
There is software for every application, and, more recently, with the growth of
artificial intelligence applications, software for software management has been
released as well.
Software is so important that it is impossible to overlook how software designs
can be represented. Explanations follow on how software is usually represented,
relationships of design structure matrices (DSM), and applications of modulus
concepts in design processes. Reference is made to direct correspondence between
unified modeling language (UML) diagrams and design matrixes resulting from
decompositions of FRs. Relations between FRs and DPs or DPs and PVs decom-
positions are shown graphically in module-junction structure diagrams. Through
these, a class diagram of systems can be built, on the basis of the existing
4 Design Representations 147

correspondences between module concepts in AD and concepts of class in

object-oriented programming. These make the axiomatic approach compatible with
object-oriented programming techniques so that design solutions can be optimized
in all phases of software life cycles.

4.5.1 Operational Context

Software is a result of human creativity, which is made up of intangible procedures,

and yet pervasive across many human activities. The development of software
systems is a complex process involving several actors, according to production
cycles, that can follow different models, with multiple intermediate phases. Each
development phase is characterized by interactions of skilled professionals who
often have completely different roles, experiences, and cultural backgrounds.
The success of development projects depends on stakeholders engaging in
dialog. Thus, just the availability of skilled developers does not guarantee success,
if contractors fail in communicating clients’ needs (CNs) to project analysts.
Analysts must formalize CNs into FRs for their developers, who, in turn, must
correctly translate FRs into DPs and PVs.
Software engineering emphasizes modeling languages and semantics that
facilitate design processes. Representative forms of output design defined by the
unified modeling language (UML) are particularly noteworthy. This modeling
language was introduced in the late 1990s by the Object Management Group
(OMG), in order to unify and standardize various software modeling techniques in
use at that time. Object-oriented programming defines abstract software objects,
which interact by exchanging messages. Systems can be designed with modular
architectures. At each level of abstraction, these systems can be defined by simple
diagrams. In this context, UML allows designers to analyze, describe, specify, and
document software systems, even complex ones, using visual models, called dia-
grams. UML is in continuous evolution, and has different forms of graphical rep-
resentation, depending on the intent. It simplifies dialog among stakeholders
involved in software development processes. Project customers and contractors can
understand one another through well-defined, clear rules. However, simple repre-
sentations of cognitive artifacts do not provide all the tools needed to improve
projects. In some cases, excessively specific representations of projects could also
inhibit creative charges of stakeholders, who might not have an incentive to con-
sider alternative solutions.
AD goes beyond simple, formal representations of projects. AD allows the
selection of representations of products that are not simply resulting from the
author’s creativity or experience. Axiom one guarantees logical coherence of rep-
resentations themselves, while axiom two selects the solution with the least com-
plexity. However, not all types of UML diagrams are compatible with AD. After
all, AD requires representations that are not simply descriptive of the project.
For AD representations, unequivocal relationships can be inferred among various
components. Therefore, on a case-by-case basis, it is necessary to adapt UML
148 P. Citti et al.

modeling to AD. Not to distort representative techniques, only to redefine them

within the scope of design intents to be pursued. Each form of design output
representation is specific to adopted software development processes and phases
being represented. In this regard, an examination of axiomatic forms of represen-
tation of software design cannot ignore contextualization of adopted production
processes.

4.5.2 Software Representation—UML

AD is well suited to optimize the software system designs based on object-oriented

programming techniques. This is because the AD approach is based on decom-
posing FRs. For simplification, we can identify at least three levels of FR
decomposition for software development. These levels can correspond to three
specific classes of representations of software systems (Fig. 4.25).
The first class of representations should describe the conceptual designs of
systems. It can be formalized in several ways. It starts by introducing system use
cases, i.e., formalizations of specific operation modes to be designed. Each case
represents a high-level FR, also from an AD perspective. This definition introduces
two types of UML diagrams, use case diagrams and collaboration diagrams. In
UML language, it is a matter of two distinct forms of formal representations of the
same system. Diagrams of use cases represent general operations of systems and
users. This type of representation of a software system has prevailing descriptive
nature.

First Level of Decomposition

Equivalent UML representations

Use case
Collaboration diagram

Functional requirements
related to customer needs

Sequence and Activity Second Level of Decomposition

Diagrams

Activity Matrix representation

of the activities and events flow
Functional
Class diagram
Third Level of Decomposition requirements
related to the
Elementary function technical solution
Matrix representation
of the technical solution

Fig. 4.25 Axiomatic decomposition levels and related software representations

4 Design Representations 149

A collaboration diagram has more quantitative elements than a matrix. It con-

stitutes a graphical representation of collaborative interactions among various use
cases. Every collaboration between use cases is activated by sending a message,
which allows the execution of elaborations. Both types of diagrams provide
high-level general representations of systems to be designed. They are used to make
structures of systems understandable to non-technical users, such as project clients.
They represent standards in the UML 2.0 modeling language, approved by the
international OMG committee. The second class of representations defines logical
system designs. It corresponds well to the second level of axiomatic decomposi-
tions of FRs. In this case, the second-level design matrix can be represented
graphically by a sequence diagram. This diagram describes sequential flows of
activities inherent to the system to be implemented. This type of representation of a
software system constitutes a standard in the modeling language UML 2.0,
approved by the international committee OMG. It is always a high-level description
intended for analysts, particularly. These professionals act as a bridge between the
customer and the developers who are responsible for implementing the procedure.
The third class of representations starts to describe, instead, physical designs of
systems. In this case, the third-level decomposition matrix of FRs begins detailed
representations of objects of the system, consisting of methods, elementary func-
tions, and datasets applied to them. This representation is addressed to the devel-
opers of the system, which already describes a design solution. In this case, the
third-level design matrix can be graphically represented by a class diagram in UML.
The class diagrams allow for formulating descriptions of operations of software
systems based on abstract entities called classes.
A class can be constituted by two or more objects. Each object is defined by data
structures, called attributes, and by active procedures on such data, called methods,
according to what is defined or declared by the respective classes. However, as a
simplification in the course of the discussion, we consider the terms “classes” and
“objects” as synonyms, even if an object is an instantiation of a class. Also, class
diagrams constitute standards in UML 2.0, approved by the international com-
mittee, OMG. The entire functional decomposition process is top-down. It starts
with high-level requirements and goes on to define detailed technical specifications.
The writing phase of the source code is bottom-up. In this case, from the
implementation of the elementary object, processes move on to developing inter-
faces, and up to overall system implementation. The whole process is iterative. The
redefinition of an FR, here a use case, or its incorrect interpretation, involves the
overall recycling of the process. The redefinition of use cases corresponds to
changing the behavior of a system. This means reviewing the entire system.

4.5.3 Equivalence Between Forms of Representation

Software systems designed with object-oriented programming techniques can also

have modular representations compatible with AD. To put this into practice, it is
necessary to introduce a useful simplification to UML.
150 P. Citti et al.

Fig. 4.26 Module M representation

One key phrase “Objects with indices” is used to represent all levels of FRs, i.e.,
class, object, and behavior. For example, classes or objects may be called Object i,
which is equivalent to FRi. The behavior will be denoted as Object ij to represent a
second-level FRij. Therefore, each module (M) is an object that holds data structures
(DPs) and functions that refer to that structure (Aij), as seen in Sect. 4.2.3.
Therefore, a generic module (M) is defined as the row of the design matrix
(Eq. 4.1), which yields the FR of the row when it is multiplied by the corresponding
DP, i.e., data. In practice, a module is an object, which contains the data and the
methods that implement FRs (Fig. 4.26). Data structures correspond to data files,
while (Aij) are transactional functions, i.e., methods.
Starting from these definitions a series of graphical correspondences can be
introduced between modular representations through flowchart diagrams and class
diagrams in UM. These correspondences can translate representations of projects in
terms of AD, both in the matrix and modular form, in an equivalent form of class
diagrams. Regarding Eqs. 4.1 and 4.2, we can consider, respectively, Figs. 4.27 and
4.28 to analyze correspondences with class diagrams.

4.5.4 Example of a Software Representation

Consider the implementation of web services that allow family doctors to monitor
the progress of therapies of patients at home with chronic hypertension. Consid-
ering clinical cases of hypertension control, in the first phase, consider only two
people: a patient and a general practitioner. The stories of both people can be
collected in semi-structured interviews. By putting these two interviews together,
processes that involve them both can be schematized. The general scenario of the
system (Fig. 4.29) can be defined. Each case of identified use must have at least one
activation element, i.e., participant. Symbols for participants are little person icons.
The two people of the system are also participants in the general scenario because,
with their actions, they activate home management processes for blood pressure
control. For the example shown, five main use cases are identified, which describe
the main functioning of the system. They are represented in Fig. 4.29 by ovals.
In summary, the process is activated by periodic measurements of blood pressure
by the patient at home. Every 10 days, a report is sent to a platform, which is
4 Design Representations 151

Class M1
-DP1
+FR1()

M1
Class M
S

M2

Class M2
-DP2
+FR2()

Fig. 4.27 An uncoupled software system like flowchart diagram and class diagram

Class M

M1 C M2
Class M2
Class M1
-DP1
-DP1
-DP2
+FR1()
+FR2()

Fig. 4.28 A decoupled software system like flowchart diagram and class diagram

Fig. 4.29 Use case diagram describing the general treatment of hypertension
152 P. Citti et al.

responsible for sending it to the general practitioner, GP, i.e., family doctor. The GP
analyzes the report. If the pressures are considered abnormal then medication
dosage can be changed, another medication can be prescribed, or a visit with a
specialist can be scheduled. These are recorded in the patient’s health record
(EHR) and a message is sent via email and text message to the patient. Patient signs
in to their EHR accounts and picks up the prescriptions. This identifies primary and
recurrent activities. These activities can be summarized in the following synthetic
semiformal description (high-level case story):

• measure BP;
• data collector;
• GP evaluation;
• EHR registration; and
• follow therapy.

This general system scenario describes relationships between two people using
the system. This representation is simplified and easy to understand. Relationships
between components on the diagram are descriptive. However, they are not suitable
for directly feeding AD decomposition processes. Therefore, UML diagrams should
be identified that can have graphic representations directly referable to matrices. At
this level of detail, representations of systems as collaboration diagrams can be
adequate.
Collaboration diagrams are graphical representations in UML of software sys-
tems. They are used to represent interactions, or collaborations, between various use
cases of systems in main modes of operation. Such modality is shown in the
diagram of the general scenario of the system of Fig. 4.29. However, to make it
entirely compatible with AD, conditions, consisting of considering only interactions
that produce elaborations in interaction use cases, must be imposed. Collaborations
between use cases X and Y can be defined as the use case X enabling processing in
use case Y. In other words, use case X interacts or cooperates with use case Y only
when it activates a method causing use case Y to change its state. This forcing
allows design processes to focus on events that involve processing. From the
general scenario of Fig. 4.29, consider relative use cases of systems and place them
in square matrices. In the same way, collaborations can be placed in columns of the
same matrix. Then the design matrix of Table 4.3 can be obtained.
Black Xs in Table 4.3 are internal elaborations specific to the use case. Red Xs
represent methods that, starting from a specific use case, activate a process in
another use case.
From Table 4.3, the equivalent collaboration diagram can be obtained easily
(Fig. 4.30). Here, use cases are indicated by rectangles, while arrows represent
collaborations.
AD can provide modular representations of use cases related to general treat-
ments of hypertension. As shown in Sect. 4.2.1, the matrix design related to
Table 4.3 can also be represented as a system of equations.
4 Design Representations 153

Table 4.3 First-level design matrix describing the general treatment of hypertension
DP1 DP2 DP3 DP4 DP5
collaboration collaboration collaboration collaboration collaboration
measure BP data collector GP HER follow
evaluation registration therapy
FR1 measure BP X
FR2 data X
collector
FR3 GP X
evaluation
FR4 HER X
registration
FR5 follow X
therapy

Fig. 4.30 Collaboration diagram describing the general treatment of hypertension

154 P. Citti et al.

Fig. 4.31 Flowchart diagram

of Eq. 4.6. system M1

C M2

C M3

C M4

C M5

8 9 2 38 9
>
> FR1 >
> >
> DP1 >
>
>
> > 6X 7> >
< FR2 >
= 6 7 < DP2 >
> =
X X
FR3 ¼ 6
6
7 DP3
7 ð4:6Þ
>
> > 4 XX > >
> FR4 >
> >
> X 5> >
> DP4 >
>
>
: ; X : ;
FR5 XX DP5

Starting from previous equations, a modular representation in terms of a flow-

chart diagram can be provided as shown in Fig. 4.31.
A new decomposition of FRs can be carried out from the use cases in Table 4.3.
Use cases are broken down into actions. A higher level of detail is added
(Table 4.4). The proposed decomposition is not the only one possible. Professional
experience guides designers through analyses and decompositions of FRs. Thus,
AD allows the use of quantitative comparison tools of alternative solutions, based
on matrix algebra, such as reangularity and semiangularity from Principles of
Design (Suh 1990).
In this case, the second-level design matrix is equivalent to the UML sequence
diagram in Fig. 4.32. Its construction is simple. The sequence diagram describes the
exchange of actions and messages related to the use case diagram of Fig. 4.29. In
this diagram, actions are distributed between participants of the system, patients,
medical devices, platforms, EHRs, and GPs. In practice, actions are attributed to
specific participants in the process that activates them. Going through Fig. 4.32,
each participant refers both to a main action and other actions, which are initialized
following interactions with other participants in the system. Actions referring to the
same participant are the same color. Messages addressed to other participants are
straight lines, dashed lines instead represent those that constitute a return.
Representations of output design can be re-elaborated in terms of flowchart
diagrams (Fig. 4.33). This form of representation holds different conceptual levels
together. Taking the functional decomposition of Table 4.4 as a reference, the
following systems of equations relative to the blocks M1, M2, M3, M4, and M5 can
be written as follows:
 4

Table 4.4 Second-level design matrix describing the general treatment of hypertension
DP1 collaboration measure BP DP2 collaboration data collector DP3 collaboration GP evaluation DP4 collaboration EHR registration DP5
collaboration
follow therapy
DP1.1 DP1.2 periodic DP1.3 E5 BP DP2.1 DP2.2 DP3.1 DP3.2 DP3.3 DP4.1 DP4.2 collaboration DP5.1
daily data pressure report to collaboration collaboration collaboration collaboration collaboration collaboration send message collaboration
BP report platform registor BP send message to get report analyse report prescription register (prescription) get
report GP prescription prescription
Design Representations

FR1 FR1.1 read BP X

measure FR1.2 periodic X X
BP data pressure
report
FR1.3 send BP X X
report to
platform
FR2 data FR2.1 registor X
collector BP report
FR2.2 send X X
message to GP
FR3 GP FR3.1 get X
evaluation report
FR3.2 analyse X X
report
FR3.3 X X
prescription
FR4 EHR FR4.1 register X
registration prescription
FR4.2 send X X
message
(prescription)
FR5 follow FR5.1 get X
therapy prescription
155
156 P. Citti et al.

Fig. 4.32 Sequence diagram describing the general treatment of hypertension

8 9 " #8 9
< FR1:1 = < DP1:1 =
X X
FR ¼ DP1:2 ð4:7Þ
: 1:2 ; X X X : ;
FR1:3 DP1:3

  

FR2:1 X DP2:1
¼ ð4:8Þ
FR2:2 X X DP2:2
8 9 " #8 9
< FR3:1 = < DP3:1 =
X X
FR ¼ DP3:2 ð4:9Þ
: 3:2 ; X X X : ;
FR3:3 DP3:3

  

FR4:1 X DP4:1
¼ ð4:10Þ
FR4:2 X X DP4:2
4 Design Representations 157

Fig. 4.33 Flowchart diagram describing the general treatment of hypertension

FR5:1 ¼ XDP5:1 ð4:11Þ

The flowchart in Fig. 4.33 can be represented in terms of class diagram. In this
case, the modules Mi are replaced by classes (Fig. 4.34).

Class M1

Class M2

Class M1.3
Class M3
Class M1.1 -DP1.2
-DP1.3
-DP1.1
+FR1.3() Class M4
+FR1.1()

Class M1.2 Class M2.1 Class M3.1

-DP1.1 -DP2.1 -DP3.1 Class M5
-DP1.2 +FR3.1()
+FR1.1()
+FR1.2()
Class M2.2
-DP2.1 Class M3.3
-DP2.2 Class M3.2 -DP3.2 Class M4.1 Class M4.2
+FR2.2() -DP3.1 -DP3.3
-DP4.1 -DP4.1
-DP3.2 +FR1.3()
+FR4.1() -DP4.2 Class M5.1
+FR3.2()
+FR4.2() -DP5.1
+FR5.1()

Fig. 4.34 Class diagram describing the general treatment of hypertension

158 P. Citti et al.

In summary, software system representations in terms of UML diagrams are

equivalent to AD forms of design representations. This is possible because
object-oriented programming techniques allow systems with modular architectures
to be implemented. Throughout software life cycles, representations of design
solutions in terms of project matrices or flowcharts of modules can easily be
translated into an equivalent UML diagram. These equivalences allow software
design activities to be framed within broader scope of engineering studies.

Problems
1. Nowadays, tablets have become one of the most important devices for enter-
tainment, since they allow us to watch films and series, as well as net surfing and
social media. The most important problem with tablets is that they need, usually,
to be held in place. Thus, how should we design a tablet holder? On the market
there are lots of solutions, some of them patented and some others not. How-
ever, are they good designs? What about to try to design a tablet holder using
AD?
Consider a customer as a person who uses the tablet mainly to watch movies
or other media while lying down. Consider the following customer needs:

• CN1: Fits on as many different beds and couches as possible;

• CN2: Holds tablets/other smart devices of various sizes and weights;
• CN3: Pushes easily out of the way when the user is in need of getting up;
• CN4: Adjusts easily while remaining stable.

To facilitate the work, we also propose some FRs defined basing on the CNs:

• FR1: Attaches securely between any hard surface and a cushion;

• FR2: Holds tablets/smart devices up to 14 inches and 1 kg. One of the largest
and heaviest tablets on the market today is the Toshiba Excite 13;
• FR3: Pushes out of the way with minimal force, when the users are in need of
getting up;
• FR4: Adjusts with minimal force while remaining stable.

A student can also try to figure out different FRs.

Try to represent a design solution basing on this information.
For further detail, you can find an example in: Helgason H, Þórarinsson T,
Ingvason S, Foley JT (2018) “Design of a tablet holder with the help of
Axiomatic Design” International Conference of Axiomatic Design 2017,
MATEC Web of Conferences.
2. Adjustable desks are a valid solution to reduce prolongated sitting time. This
fact, basing on recent studies, has been demonstrated to be helpful for people’s
health. As well as in the previous exercises, we would like to try to design an
adjustable height desk using AD.
4 Design Representations 159

Even in this case, the market offers lots of solutions, but in this exercise, we
would like to compare, at the end of the exercise, commercial solutions with the
obtained design representation.
Even in this case, CNs, as well as FRs, will be provided.

• CN1: Adjustable desk that suits people in common sizes;

• CN2: The desk has to be steady so you can lean on it while working;
• CN3: It should be affordable;
• CN4: It should have electrical sockets;
• CN5: The top panel should endure hammer beatings and fluids without
deteriorating.

In this exercise, we shall also consider some constraints as follows:

• C1: Total cost cannot exceed 400 USD;

• C2: It can be manufactured in a workshop.

FRs proposed are the following:

• FR1: Positions a work surface vertically;

• FR2: Stays stable when worked on;
• FR3: Locks in position;
• FR4: Powers appliances;
• FR5: Endures hammer strikes and fluid spills on the surface.

The student is supposed to define a design representation basing on these FRs

(or on other FRs defined basing on the CNs).
For further detail, you can find an example in: Foley, J.T., Símonarson, A.F.,
Símonarson, H.T., Ægisson, L.F., and Goethe, A.T. 2017, “Adjustadesk—An
Adjustable Height Desk,” MATEC Web of Conferences.
3. Consider the following representation of a simple plate with two holes
(Figs. 4.35 and 4.36):
What is the difference between the two design representations? What is
supposed to specify dimensioning in the second design representation? What
would this representation tell about the difference in FRs that these fulfill?
4. Consider the following representation of a cylinder (Fig. 4.37):
Think about a reason why the internal part of the cylinder has more speci-
fication about tolerancing than external? State this in terms of the FRs that might
need this tolerance to be fulfilled satisfactorily by this specification on a DP.
5. Imagine needing to realize the assembly represented in section in Fig. 4.38.
The element 1 and element 2 are represented by the following drawings
(Figs. 4.39 and 4.40):
160 P. Citti et al.

Fig. 4.35 Representation A of a plate with two holes

Fig. 4.36 Representation B of a plate with two holes

4 Design Representations 161

Fig. 4.37 Representation of a cylinder

Fig. 4.38 Representation of a system composed of two elements (a single-hole plate—element 1

and a cylindrical pin—element 2)
162 P. Citti et al.

Fig. 4.39 Representation of element 1

Fig. 4.40 Representation of element 2

4 Design Representations 163

Is it possible to say something about the connection between 1 and 2? What

could be the DP represented by the hole and pin tolerancing? Do you have any
idea of how to assemble the two parts?
6. Explain why AD may be an optimization tool in software system design
processes.
7. Construct the design matrix for an FR–DP relationship of a case study related to
home administration of drug infusion in pediatric oncology which is described
below.

Problem:

The interviews with the actors (healthcare workers, parents, young patients) of
the process allowed to identify the primary and recurrent activities for this
process. These activities can be summarized in the following use cases:

• checking for drug availability (locker state, non-automated);

• change of therapy;
• change of modality;
• change of drug;
• prescription of drug by the hospital doctor;
• withdrawal of drug at the hospital pharmacy; and
• administration of drug by parents or healthcare professional (nurse, doctor).

Practically, the administration of a drug is activated by the management

device. At this point, the system detects that the drug is about to end and sends
the alert to the control center. The control center contacts the hospital doctor
who prescribes the medicine administered by the hospital pharmacy where the
patient’s parent goes to collect it. At the end of the process, the availability of
the drug is updated.
8. Decompose the FRs related to the design matrix of the abovementioned prob-
lem, in order to construct a modular representation of the process in terms of
flowchart.
9. Draw for the abovementioned problem the representations in terms of class,
collaboration, and sequence diagrams, starting from the previously defined
design matrices.

References
ANSI/ASMEB4.1-1967 (2009) ANSI/ASME B4.1-1967 (R2009) Preferred limits and fits for
cylindrical parts
ASME Y14.36-2018 (2018) ASME Y14.36-2018 Surface texture symbols
Gausemeier J, Moehringer S (2002) Vdi 2206—a new guideline for the design of mechatronic
systems. IFAC Proc 35(2):785–790. https://doi.org/10.1016/S1474-6670(17)34035-1
ISO1302 (2002) ISO 1302:2002 Geometrical product specifications (GPS)—indication of surface
texture in technical product documentation
164 P. Citti et al.

ISO21920 (2020) ISO/DIS 21920 Geometrical product specifications (GPS)—Surface texture:

profile
ISO25178-2 (2012) ISO 25178-2:2012 Geometrical product specifications (GPS)—Surface
tex-ture: areal
Melvin JW (2003) Axiomatic system design: chemical mechanical polishing machine case study.
PhD Defense Presentation Slides
Suh NP (1990) The principles of design. Oxford University Press

Further Reading

Bell E, Davison J (2012) Visual management studies: Empirical and theoretical approaches. Int J
Manag Rev 15:167–184. https://doi.org/10.1111/j.1468-2370.2012.00342.x
Booch G, Rumbaugh J, Jacobson I (1999) The unified modeling language user guide.
Addison-Wesley Professional
Brown C (2011) Axiomatic design applied to a practical example of the integrity of shaft surfaces
for rotating lip seals. Procedia Eng 19:53–59. https://doi.org/10.1016/j.proeng.2011.11.079
Brown C, Hansen H, Jiang X, Blateyron F, Berglund J, Senin N, Bartkowiak T, Dixon B, Le Goïc
G, Quinsat Y, Stemp W, Thompson M, Ungar P, Zahouani H (2018) Multiscale analyses and
characterizations of surface topographies. CIRP Ann. https://doi.org/10.1016/j.cirp.2018.06.
001
Do SH, Suh NP (2000) Object-oriented software design with axiomatic design. In: Thomp-son
MK (ed) First international conference on axiomatic design, Institute for Axiomatic Design,
Axiomatic Design Solutions, Inc., Cambridge, MA, pp 278–284. https://axiomaticdesign.com/
technology/icad/icad2000/icad2000_027.pdf
Foley JT, Símonarson AF, Símonarson HT, Ægisson LF, Goethe AT (2017) ADjustadesk—an
adjustable height desk. In: Slǎtineanu L (ed) 11th international conference on axiomatic design
(ICAD), MATEC Web of Conferences, Iasi, Romania, 01002, p 7
Girgenti A, Pacifici B, Ciappi A, Giorgetti A (2016) An axiomatic design approach for customer
satisfaction through a lean start-up framework. In: Liu A (ed) 10th international conference on
axiomatic design (ICAD), Procedia CIRP, Elsevier ScienceDirect, Xi’an, Shaanxi, China, vol
53, pp 151–157. https://doi.org/10.1016/j.procir.2016.06.101
Helgason H, Þórarinsson T, Ingvason S, Foley JT (2018) Design of a tablet holder with the help of
axiomatic design. In: Puik E, Foley JT, Cochran D, Betasolo M (eds) 12th international
conference on axiomatic design (ICAD), MATEC web of conferences, Reykjavík, Iceland, p. 7
Melvin JW (2003) Axiomatic system design: chemical mechanical polishing machine case study.
Doctor of Philosophy in Mechanical Engineering, Massachusetts Institute of Technology, 77
Massachusetts Ave., Cambridge MA 02139, USA
Oberg E, Jones FD (1916) Machinery’s handbook volume 1916. Industrial Press
Object Management Group (2017) Unified modeling language specification (2.5.1). https://www.
omg.org/spec/UML/About-UML/
Parretti C, Pourabbas E, Rolli F, Pecoraro F, Citti P, Giorgetti A (2019) Robust design of web
services supporting the home administration of drug infusion in pediatric oncology. In: Liu A,
Puik E, Foley JT (eds) 12th international conference on axiomatic design (ICAD), MATEC
web of conferences, Sydney, Australia, vol 301, p 00013. https://doi.org/10.1051/matecconf/
201930100013
Pecoraro F, Luzi D, Pourabbas E, Ricci FL (2017) A methodology to identify health and social
care web services on the basis of case stories. In: Proceedings of IEEE E-health and
bioengineering conference (EHB), Sinaia, Romania
Pimentel AR, Stadzisz PC (2006) A use case based object-oriented software design approach using
the axiomatic design theory. In: Thompson MK (ed) 4th international conference on axiomatic
4 Design Representations 165

design, ICAD2006, Axiomatic Design Solutions, Inc., Firenze, Italy, p 8. https://

axiomaticdesign.com/technology/icad/icad2006/icad2006_29.pdf
Reed D, Bohemia E (2012) Representations of design outputs in cross-functional teams.
International conference on engineering and product design education. Artesis University
College, Antwerp, Belgium, pp 425–430
Simmons C, Maguire D (2012) Manual of engineering drawing, 4th edn. Butterworth-Heinermann
Suh NP (1998) Axiomatic design theory for systems. Res Eng Design 10:189–209
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Vulliez M, Gleason M, Souto-Lebel A, Quinsat Y, Lartigue C, Kordell S, Lemoine A, Brown C
(2014) Multi-scale curvature analysis and correlations with the fatigue limit on steel surfaces
after milling. Procedia CIRP 13:308–313. https://doi.org/10.1016/j.procir.2014.04.052
Problem Definition
5
Ang Liu

Abstract
The first step in developing an innovative design is to know the problem we are
trying to solve. Once we know the problem, we need to establish the goals our
design must satisfy to solve the problem. Sometimes, we establish the goals first
and then determine the problems we must solve to achieve the goals. These goals
must then be stated as functional requirements (FRs) in the functional domain
that our design must satisfy. In some ways, it is an obvious way of coming up
with good designs. However, many companies have made errors by trying to be
more competitive, without first discovering the real problem faced by their
products. The information on their existing products may come with diverse and
contradictory claims and counterclaims, aggravating each other. If we can define
“what the problem is,” the problem can often be solved through design. The
ability to define the problem can be acquired by accumulating broad knowledge
base, experience, and repeating the steps outlined in Axiomatic Design (AD).
One of the goals of this chapter is to teach those who are not yet initiated into the
field of design the ability to identify and define the problem.

5.1 Problem Definition in Engineering Design

5.1.1 Characteristics of Problem Definition

Problem definition is one of the most critical steps in design. It is an open-ended

process. Given the same set of customer needs and design constraints, a variety of
different problems can be formulated by different designers. It depends on their

A. Liu (&)
Sydney, Australia
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 167

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_5
168 A. Liu

experience, background, and knowledge. For example, in order to treat cancer, one
person may define it as a biological problem that should be solved by “designing a
biological experiment to understand the genesis and propagation of cancer.”
Alternatively, another person may define the problem as a pharmaceutical problem
of “designing a new drug to retard cancer development.” Another person may
characterize it as a public healthcare problem and formulated it as “design a public
campaign to discourage smoking.” For engineers, multiple engineering problems
can be formulated, such as “designing a scanning machine to identify and visualize
stage-1 cancer,” “designing a big-data platform to analyze a person’s potential risk
of developing cancer,” “designing a medical device to remove cancer completely,”
and so forth. In another example, the same goal of “addressing climate change” can
be converted into a scientific problem of “designing an assessment framework to
evaluate the influence of global warming,” a policy-making problem of “designing
an environmental policy to reduce the greenhouse gas emissions,” a political
problem of “designing a political discussion to develop a consensus among global
political leaders,” or an engineering problem of “reversing the trend of global
warming through climate engineering.”
Problem definition is an iterative process. No problem definition will remain
valid forever. As new data, information, and knowledge are acquired, people should
be prepared to revisit and revise your problem definition. For example, the success
of century-old IBM (International Business Machines) can be attributed to its
strategic shift of core business by continuously focusing on the emerging,
high-value, profitable new markets. In other words, IBM repeatedly redefined its
essential problem from typewriter to personal computer, to IT service, to artificial
intelligence (AI), and so on. The development of new problem definition can be
triggered by new technological development. Every time a breakthrough technol-
ogy is introduced, not only it solves existing problems, but also it introduces new
problems. For instance, the advancement of additive manufacturing (3D printing) is
promoting the development of material science.
Problem definition serves to bridge ideality and reality. The former refers to the
ideal state you desire to achieve without any limitations in the distant future,
whereas the later means the possible state you can afford to achieve against various
constraints at present. A problem should be defined in light of the ideality, while the
problem should be solved following the reality. Those successful company, orga-
nization, and person, with no exception, are all keenly driven by an ambitious
vision of ideality. For example, the ideality of BMW is to produce the “ultimate
driving machine,” whereas the ideality of Toyota is to “lead the way to the future of
mobility.” The ideality of Elon Musk is “immigration in Mars,” and his projects
(e.g., Space X, electric vehicle, and Hyperloop) were all intended to serve such an
ideality in the reality.
The design problem is a multi-faceted notion. Generally speaking, a complete
problem definition involves customer need (CN), FRs, design constraint (DC), and
problem context. Firstly, new artifacts are developed to satisfy unmet customer
needs. In order to generate new values, impacts, and profits, it is imperative to fully
understand your target customers and relevant stakeholders, concerning their voices
5 Problem Definition 169

of what is desirable and undesirable. Secondly, the unstructured customer voices

should be translated into a set of well-defined FRs that can be assigned to engineered
systems. This is where engineers come into play to leverage their engineering
knowledge and design thinking to connect the customer domain and the physical
domain. Thirdly, different from art design, engineering design is always performed
against various limitations. Therefore, a variety of design constraints should be
identified to limit the choice of FRs and design parameters (DPs). Lastly, a problem
definition is unique in a particular context. The same problem should be defined
differently in different contexts. When the context changes over time, the problem
should be redefined correspondingly. Among the four critical components of problem
definition (i.e., customer need, FRs, design constraint, and problem context), FR is
arguably the most important but difficult decision to make in problem definition.
Different from CNs and DCs that are mostly imposed by the third parties (e.g., client,
management, supplier, government, and union), FRs are entirely decided on your own
based on your engineering knowledge, experience, and creativity.

5.1.2 Good Problem Definition Leads to Design Innovation

Design Story 5.1:

As illustrated in Fig. 5.1, iPod was one of the most successful product offerings in
the development history of Apple Inc. It is widely agreed that the subsequent
successes of iPhone and iPad were more or less built upon the foundation of iPod.
The original problem definition for iPod, proposed by Steve Jobs, was to design a
portable device that could “carry 1,000 songs in the pocket.” As a result of such a
problem definition, the engineers at Apple were inspired to solve multiple “un-
conventional” problems, such as “where to download 1,000 songs in a legal way,”
“how to pay for so many songs,” “how to deal with the copyright issue,” “how to
index and search so many songs,” etc. Through solving these problems, not only
Apple engineers developed sophisticated hardware, but also it created the media
library of “iTunes.” Thereafter, the definition of iPod continuously evolved from a
portable music player to a multimedia digital player, and eventually to today’s
multi-purpose pocket computer. It should be noted that, in the year of 2011, MP3
player was by no means an entirely new concept. Many competitors had their own
product offerings of MP3 player. Assume the problem definition given to the
engineers had been to “develop an MP3 player with an Apple logo,” it should not
be too surprising that a set of very different problems would have been addressed,
such as “storage capacity,” “battery life,” “sound quality,” etc.

Design Story 5.2:

RSI (repeated stress injury) is a common problem caused by using the computer
keyword. As the name suggests, RSI occurs when a routine motion that stresses the
body is repeated. Over time, stress accumulation will lead to injury. Ergonomic
170 A. Liu

Fig. 5.1 Case study of iPod

design. (Reproduced with
permission from Delyk 2019)

keyboard, as a traditional solution, was designed to address the problem of “how to

provide users with the most comfortable (or least stressful) typing position.” In
other words, the focus of problem definition lies in the keyword of “stress.” A few
years ago, this design task was given to a group of KAIST students, and another
innovative solution was created. By focusing on another keyword of “repetition,”
they redefined the problem as “how to interrupt the repetition of the same gesture.”
As a result, they developed a “floating keyboard” that is characterized by a subtly
moving plane to help unlock stressful muscles and engage other muscles. This
concept has been applied to other industries. Take the office chair for example, by
replacing the static piece of furniture with a yoga ball, the user subtly engages other
muscles to keep an unstable balance, and hence reducing the RSI. This is called a
yoga ball chair, as illustrated in Fig. 5.2. In addition, the footwear industry has seen
this trend as well with “rocking shoes” that promote engaging ankle muscles in a
way traditional shoe cannot.
A good problem definition enables you to discover those invisible innovation
opportunities. As illustrated in Fig. 5.3, although the three products (i.e., smart-
watch, jogging stroller, and yoga ball chair) belong to entirely different product
categories (i.e., wearable device, children product, and office furniture), they can
actually be recommended to the same target customer, because they are all intended
to fulfill the same purpose (i.e., “perform exercise while multitasking”) by solving a
variety of different problems.
5 Problem Definition 171

Fig. 5.2 Case study of yoga ball chair. (Reproduced with permission from Popov 2019a, b)

Fig. 5.3 Comparison of smartwatch, jogging stroller, and yoga ball chair. (Reproduced with
permission from Lyjak 2019; Popov 2019a, b)

5.2 Definition of Customer Need

5.2.1 Who Are the Target Customers and Relevant

Stakeholders?

Problem definition should be collaboratively developed between you with a group

of target customers and relevant stakeholders. On the one hand, customer voices
should be solicited, analyzed, and interpreted to consolidate the basis of problem
172 A. Liu

definition, as Steve Jobs once proposed, “you‘ve got to start with the customer
experience and work backward to the technology.” On the other hand, problem
definition cannot be dictated only by customer voices, as Steve Jobs once proposed,
“you can’t just ask customers what they want and then try to give that to them. By
the time you get it built, they’ll want something new.” Therefore, the rational
attitude toward customer inputs should be “always listening to your customers,
avoid being uttered by their voices, and make your own decisions.”
Problem definition begins with involving target customers and relevant stake-
holders. Generally speaking, there are three models of customer involvement
(Kaulio 1998): design for customer (i.e., the process of creating new product based
on customer’s preferences), design with customer (i.e., the process of engaging
customers in evaluating new products), and design by customer (i.e., the process of
empowering customers to build individualized products by themselves).
Given an ill-defined problem statement, the first question to ask is “who are the
target customers.” Apparently, depending on your answer to this question, problem
definition will be driven toward different pathways, because different customers
more or less demand different things. In practice, brainstorming is a common
method for identifying target customers. As illustrated in Fig. 5.4, a simple 4-P
method can also be followed to search for target customers. Specifically, a set of
four interrelated questions are asked in sequence: “what is the name of a product,”
“what problem is the product designed to solve,” “what purpose is the problem
intended to serve,” and “who are the people behind the purpose.” By repeating this
process, you will practice your divergent thinking to think outside the box. It should
be noted that, in industry, Big Data analytics is growingly employed to recognize
target customers, analyze customer preference, and build customer personas.
Take the toy design for example, who are the real target customers? Is it parent,
children, toy store, or all of them? An obvious answer is children, who however has
no money to pay for the toys he/she really prefers. An alternative answer is the
parent, who has more purchase power than the child. In reality, nevertheless, most

Fig. 5.4 4P Model to identify target customers and relevant stakeholders

5 Problem Definition 173

of the children simply won’t play the toys bought by the parent, which are primarily
intended to “stimulate intelligence and creativity” (i.e., a typical customer need of
most parents). Interestingly enough, in certain markets, grandparents are the most
generous customers who are willing to pay for those very expensive toys. A smart
designer like yourself should never underestimate the decision of target customers,
because your choice will directly/indirectly affect the subsequent decisions of
production, packaging, distribution, retailing, repairing, and recycling. For exam-
ple, if you chose grandparent as your primary target customer, it may not be a good
idea to sell the product online (e.g., Amazon), since most of the grandparents still
prefer the conventional shell shopping experience.
In addition to target customers, it is equally important to involve the relevant
stakeholders. Although the relevant stakeholders will not directly buy or use your
product/service, their decisions will influence the design with respect to where, when,
how, in what ways, and under what conditions the product can and cannot be sold,
used, and maintained. For example, the relevant stakeholders of stroller design
include government, supplier, car manufacturer, school, airline, grocery store, and so
forth. A square design structure matrix, where the domain elements are the relevant
stakeholders, can be created to indicate the conflicting or mutual interests between any
two stakeholders. For example, passenger and airline share conflicting interests for the
space airplane seat, while airline and airport share mutual interests for airplane safety.
Exercise 5.1:
If you were one of the chief designers of Boeing 787 Dreamliner, who are your real target
customers? Is it passenger, pilot, airline, airport, Federal Aviation Administration (FAA),
Transportation Security Administration (TSA), Department of Homeland Security, or
simply all of them? Be careful, your answer to this question will drive the direction of a
billion-dollar project.
Exercise 5.2:
If you were the dean of a leading engineering school, in the event of redesigning the
existing curriculum for the accreditation purpose, who are your real target customers? Is it
college student, faculty member, professional staff, industry partner, university president,
government official, or simply all of them? If your answer is “all of them,” is there a way to
make everyone happy? What will you do in front of conflicting customer needs?
Exercise 5.3:
If you were the CEO of a global car ride-sharing service provider, can you name a list of
key stakeholders who are relevant to the development, test, commercialization, and reg-
ulation of self-driving taxi technologies? Do they have any conflicting or mutual interests?

5.2.2 How to Solicit Customer Voices?

After you have involved a group of target customers and relevant stakeholders, some
design methods can be followed to solicit their voices. Firstly, the lead user theory
can be used to involve those lead users who possess in-depth knowledge of both
“know-what” and “know-how” (Franke et al. 2006). Secondly, ethnographic studies
174 A. Liu

can be conducted to observe and discover how and in what ways customers utilize,
interact, and alter a product, being situated in a particular context. Ethnography is
commonly used in social science to investigate how a particular culture is developed
based on social interactions. It can also be employed in product design to understand
user habits. Thirdly, you can interview customers to directly solicit their voices. In
comparison with the traditional interview approach that occurs in a formal setting
(e.g., conference room), contextual inquiry is a more user-centered method, which
enables you to discover some contextual information. Focus group is a group
interview approach, through which, dynamic interactions between peer users are
leveraged to trigger common opinions and implicit findings that cannot be revealed
by individual interviews. However, the success of a focus group session hinges on an
experienced moderator. Lastly, surveys can be conducted to solicit customer opin-
ions and aggregate their preferences. On the one hand, surveys are the most fre-
quently used method in practice. On the other hand, surveys are by no means the
most effective method. Not only you should have in-depth knowledge to design a set
of meaningful survey questions, but also you should find a way to escape from the
Arrow’s Paradox when aggregating the survey responses.
The abovementioned methods all require direct interactions between you and the
customer. Due to the rapid development of e-commerce and social networking,
there emerged a huge volume of crowd-sourced online customer voices in the
format of product reviews. Take a popular product “Kindle Paperwhite eReader” for
example, over 58,000 product reviews are published on Amazon. Product reviews
can be collected from a variety of online channels such as e-commerce platform,
search engine, APP store, social media/networking, video sharing website, product
forum, and so forth. Rich information can be abstracted from the product reviews.
Take a typical review on Amazon for example, the data that is potentially useful for
problem definition includes product rating, reviewer ranking, review content, photo
and video, comment and Q&A, and peer evaluation of the review’s helpfulness.
Since the vast majority of product reviews are contributed by end users, the contents
are oftentimes unstructured and inconsistent. Therefore, massively crowd-sourced
reviews must be carefully analyzed through qualitative data analysis. A complete
qualitative data analysis process consists of data collection, transcription, seg-
mentation, categorization, and coding.
Customer voice includes customer need, want, expectation, complaint, reflection,
emotion, etc. We are most interested in customer needs, especially the unmet customer
needs that cannot be fulfilled well by any existing solution. Maslow’s Hierarchy of
Human Needs, which is one of the most fundamental theories in psychology, can be
followed to classify CNs into different categories (i.e., “self-actualization,” “esteem,”
“social needs,” “safety and security,” as well as “physiological needs”). Through the
classification of CNs, you will be guided to filter those distracting noises that are
unnecessarily customer needs (you are “forced” to double-check the vitality of those
CNs that cannot be classified into any of the categories).
Exercise 5.4:
Given the task to design a coffee machine, how do you propose to solicit customer voices?
Which method(s) will you select? How do you combine different methods?
5 Problem Definition 175

5.2.3 Understand Customer Voices to Capture Innovation

Opportunity

The above-solicited customer needs should be extrapolated in consideration of

various macro-environmental factors based on the PEST framework (i.e., the
political, economic, socio-cultural, and technological factors). By doing so, the
purpose is to rank-order the CNs in terms of their probabilities of triggering
breakthrough innovations. Since most of the political factors (e.g., government
regulation, environmental law, tax policy, etc.) and economic factors (e.g., eco-
nomic growth, interest rate, inflation, etc.) are not directly related to a typical
engineering design project, as much as possible, you should focus on the
socio-cultural and technological factors.
One of the most important socio-cultural factors is the lifestyle meaning behind a
product. For example, the success of iPhone can be attributed to not only its
functions but also the lifestyle meaning it represents in the customer’s mind. Since
the intangible lifestyle meaning is socially constructed via back-and-forth interac-
tions, it must be examined in a particular context. For example, the coffee culture is
notably different in different contexts (e.g., in the USA, UK, and Australia). Hence,
a coffee machine should be designed differently to reflect such variations of lifestyle
meaning. In addition, you should make sense the CNs in light of the ongoing social
trends such as the sharing economy, human multitasking, product–service inte-
gration, etc.
With respect to the technological factors, in the interest of promoting innova-
tions, the CNs should be unfolded in consideration of the ongoing technological
trends. For example, multiple smart products can be interconnected, through the
Internet of Things (IoT), to perform functions collectively and collaboratively that
cannot be performed by any individual product. The Big Data analytics and
machine learning algorithms can be leveraged to recognize, record, and analyze the
complex human behaviors. A variety of sensors can be employed to make a product
more aware of the surrounding contexts such as the physical context (e.g., time,
location, weather, etc.), social context (e.g., peer product, complementary service,
resource supply, etc.), and user context (e.g., user demographics, habit, mood, etc.).
The additive manufacturing technologies make it possible to produce some com-
plex structures that cannot be manufactured before. The digital twin enables the
designer to compare a product’s expected behaviors (i.e., that are derived from
functions) and actual behaviors (i.e., that are derived from structures) in real time.

5.3 Definition of Functional Requirement

Problem definition is essentially a translation process from customer voices to

engineer voices. The vast majority of customer voices are poorly defined, ill-
structured, highly inconsistent, mostly biased, and full of noises. By contrast,
176 A. Liu

engineer voices should be clearly defined, well organized, and consistently phrased.
Engineers communicate through the language of FRs.

5.3.1 How to Represent Functional Requirements?

Function represents the purpose of design. By using FRs to describe an artifact, you
are guided to diverge your thinking to explore a broader solution space for more
possibilities. For example, if a watch is described as an artifact to “tell the time,”
then you are naturally inspired to think “are there any other means to tell the time”?
As a result, more alternative solutions will automatically emerge, such as computer,
smartphone, clock, etc. It should be noted that, in comparison with novice
designers, expert designers are characterized by their awareness and ability to
describe artifacts in terms of FRs in place of the physical name, appearance,
material, and structure.
Functions are assigned to an artifact as its requirements, and hence, namely, FRs.
Not only the same FR can be assigned to different products, but also the existing
artifact can be assigned with new functions through iterative problem definition. Take
the evolvement of mobile phone for example, initially, it was only assigned a few
basic FRs, such as <make phone call>, <send message>, <manage contacts>, etc.
Gradually, new and higher FRs were assigned, such as <take phone>, <respond
email>, <manage calendar>, etc. Finally, mobile phone became smartphone that
carries a variety of “smart” functions, such as <navigate direction>, <play game>,
<browse Internet>, etc. As a matter of fact, it can be argued that today’s popularity of
smartphone is largely attributed to the great availability of smartphone apps, which
enabled the users to customize the functions of their smartphones.
The formulation of FRs should follow some basic principles:

(a) FRs should be represented in the format of <verb + object>.

(b) FRs should be described in a solution-neutral fashion.
(c) FRs should be specified by a range of target value (i.e., design range).

Firstly, FRs should be represented in the format of <verb + object> or its variations
as follows:

• Function = <Verb + Object>.

• Function = <Verb + Object1> to/from/with/through <Object2>.
• Function = <Verb + Object> in <context>.
• Function = <Verb + Object> against <constraint>.

Secondly, in order to inspire novel ideas, FRs should be formulated in

solution-neutral fashion. By doing so, it enables you to explore a broader solution
space in the concept generation phase. In practice, it is helpful to show your FRs to
5 Problem Definition 177

some fellow engineers and ask whether they can immediately associate the FRs
with an existing product/service. If their answer is “yes,” it is an indicator that you
should step back and reformulate your FRs.
Thirdly, every FR should be associated with a quantifiable design range. In
practice, specifying a design range requires a wealth of design knowledge, expe-
rience, and even in-depth research. It is important that the design range is carefully
determined because it will be compared with the system range to compare multiple
alternative DP options.

• FR1 = Adjust the seat angle between 10° and 80°;

• FR2 = Support 15–20 kg (weight of a baby);
• FR3 = Stop motion within 1–2 s;
• FR4 = Convert human force (value range) to torque;
• FR5 = Prevent direct sunshine by 50–70%;
• FR6 = Carry 10–15 kg grocery;
• FR7 = Hold a water bottle of 2–3 kg;
• FR8 = Unfold the stroller (to what extent?);
• FR9 = Make seat waterproof (volume of liquid);
• FR10 = Limit degree of freedom (4°–5°);
• FR11 = Dissemble seat within 15 s;
• FR12 = Install seat as car seat within 20–30 s;
• FR13 = Absorb shock impulses by 80%.

It should be noted that function formulation is an iterative process. It is common

that you will need to revisit and revise your FRs once new market is cultivated, new
knowledge is acquired, new constraints are imposed, and new decisions are made.

5.3.2 How to Extract Functional Requirements from Existing

“Things”

FRs are intangible and solution-neutral. Therefore, it is cognitively difficult to

formulate FRs without any tangible thing in mind. This is especially true for novice
designers who have inadequate design knowledge and experience. In practice, it is
common that young designers tend to extract FRs based on the observation,
abstraction, and interpretation of existing things.
FRs can be extracted from not only engineered systems but also biological
systems. Biologically inspired design is one of the most widely adopted design
approaches. For example, “FR” can be extracted based on the human body to
describe the roles played by different organs. It should be noted that the
<verb + object> format is strictly followed to formulate the FRs of these organs.

• Nose: Breathes fresh air and senses smell;

• Mouth: Breathes fresh air, tastes food, speech, facial expression, etc.;
178 A. Liu

• Larynx: Generates sound, swallows food;

• Heart: Pumps blood;
• Lungs: Pull oxygen, push CO2 gas out, add air to blood, etc.;
• Stomach: Digests food, absorbs small molecules, controls secretion, senses
nutrition, etc.;
• Liver: Synthesizes protein, breaks down insulin, detoxifies chemicals, secretes
bile, etc.;
• Kidneys: Clean blood, regulate blood pressure balance water, activate Vita-
min D, filter water;
• Small Intestine: Digests food, absorbs nutrition.

It is interesting to point out that these internal organs are all functionally inde-
pendent of each other, even though they are physically integrated, in a highly
sophisticated manner, by the blood vessels. This can be considered as a vivid
example of how the Independence Axiom applies for biological systems as well.
From this perspective, cancer can be viewed as a redundant organ (or DP) that only
consumes resources (or adding process variables (PVs)) without contributing to any
useful function. As a matter of fact, one of the major challenges of treating cancer is
how to identify cancer at a very early stage. This challenge is more or less attributed
to cancer’s functional invisibility.
The FRs extracted from an existing artifact (or biological system) can be
reformulated, reorganized, and then recommended to the new artifact. For example,
many functions of smartphone were extracted from its peer products such as
computer, camera, PDA, etc. The FRs carried by the coffee grinder, milk frother,
and mug warmer can be transferred to a coffee machine.
Exercise 5.5:
Can you specify a range of acceptable values for the above-listed FRs of different organs of
human body?
Exercise 5.6:
Can you extract multiple candidate functions from other home appliances and recommend
them to the robotic vacuum cleaner?

5.3.3 How to Classify Functional Requirements into Different

Categories?

FRs should be classified into different categories. By doing so, it facilitates you to
evaluate the weights and priorities of different FRs. According to the Kano customer
satisfaction model, product features can be classified into three kinds: excitement
function, performance function, and basic function. The Kano model is a qualitative
measure of human psychological reactions (excitement) to a product based on
Maslow’s hierarchy of needs. Firstly, basic features are features that are required in a
product or it won’t sell. On the other hand, even the basic feature is available,
customer reaction is neutral. For example, a customer will never purchase a car
without any brakes. Basic features are often a result of government regulations. The
5 Problem Definition 179

second feature is the performance feature. The more performance features a product
has, the more customer is willing to pay. Based on the car example, the performance
features include gas mileage, air-conditioning, horsepower, airbags, etc. Thirdly,
excitement features are a mirror image of basic features—if the product lacks them,
the customer will not be disappointed because they didn’t expect them in the first
place. If a product has excitement features, the customer has to have it! An example
would be a self-parking car. As time progresses, today’s excitement feature will
become tomorrows’ performance feature, which eventually becomes basic features.
With social media, this timeframe has accelerated tremendously! In certain cases,
yesterday’s basic feature may come back and become tomorrow’s excitement fea-
ture. For example, consider a cigarette lighter in a car, while the technology hasn’t
changed, consumers can now use the cigarette lighter to charge electronic devices. In
practice, the Kano model can be integrated with customer surveys to collect relevant
information for accurate classification.
Secondly, according to the Long Tail model, FRs can be classified into two
categories: popular functions and unpopular functions. Function popularity can be
measured in terms of how many products are carrying the same function and how
frequently the function is used by the customer. According to the traditional design
thinking, the designers should focus on the 20% most popular functions, based on
the principle of Pareto efficiency. The Long Tail model, however, suggests that you
should focus on the 80% unpopular functions. It is increasingly difficult to maintain
user excitement simply by providing popular functions because popular functions
tend to be the focus of social media. As a result, most of the popular functions will
soon lose their surprising effect, and therefore become performance features.
Accordingly, customer excitement is increasingly stimulated by the “unpopular”
functions that are tailored precisely to the unique demographic, need, and prefer-
ence. Therefore, heavier weight should be assigned to the unpopular functions that
are anticipated to trigger customer excitement.
During the classification process, the redundant FRs should be eliminated. In
practice, redundant FRs refer to those functions that are undesirable to customers
and the functions that can be better fulfilled by another artifact. It should be noted
that redundant FR is not equivalent to unpopular FR. The former is measured by the
customer desirability, whereas the latter is measured by the usage frequency.
According to the previous studies of the Long Tail model, in many cases, customers
tend to gain a much higher satisfaction when an unpopular function is met.
Exercise 5.7:
What are the 20% popular functions and 80% unpopular functions of a smartphone? What
are the exciting function, performance function, and basic function of a smartphone?

5.3.4 How to Structure Functional Requirements?

The classified FRs should be organized into a design structure. Hierarchy is a

commonly used structure to organize FRs. Hierarchy is a structured arrangement of
180 A. Liu

entities based on their abstraction levels and dependency relationships. AD pre-

scribed three basic principles for organizing FRs into a hierarchy, namely, “com-
plete” (i.e., no necessary FRs are missing), “minimum” (i.e., no redundant FRs are
included), and “independent” (i.e., no functional couplings exist). The functional
hierarchy can be created by alternating the analysis and synthesis operations
through a zigzagging process.
Generally speaking, FRs are organized according to their dependency relation-
ships and abstraction levels. Specifically, the general FRs should be accommodated
in the upper layers of the hierarchy, whereas the specific FRs should be placed in
the lower layers of the hierarchy. The parent FRs should be placed above the
children FRs. The relationships between FRs at adjacent layers of the hierarchy are
the “part-of” relationship. By organizing FRs into different layers, the problem
definition is decomposed into different levels, respectively. Figures 5.5 and 5.6
illustrate the functional hierarchies of stroller and baby incubator, respectively.
Exercise 5.8:
Can you formulate a set of FRs for a robotic vacuum cleaner, organize the FRs into a
functional hierarchy, and check to what extent these FRs are independent of each other?
Exercise 5.9:
Can you formulate a set of FRs for a 3D printer, organize the FRs into a functional
hierarchy, and check to what extent these FRs are independent of each other?
Exercise 5.10:
Can you formulate a set of FRs for a self-driving car, organize the FRs into a functional
hierarchy, and check to what extent the FRs are independent of each other?

Fig. 5.5 Functional hierarchy of stroller design

5 Problem Definition 181

Fig. 5.6 Functional hierarchy of baby incubator

5.4 Definition of Design Constraint

Design constraint serves to define the boundary conditions of a design problem.

Not only it limits the formulation of irrational FRs, but also it prevents infeasible
solutions from being generated or selected. Design constraints should be clearly
differentiated from FRs.

5.4.1 What Is Design Constraint?

Problem definition involves defining not only FRs but also design constraints. The
former serves as objectives to illuminate the direction of ideation, whereas the latter
serves to define the boundaries of solution space. Different design theories and
methodologies model constraints in different ways. For example, AD treats
“weight” as a constraint that limits the choice of DPs and PVs (Suh 2001), the
function–behavior–structure ontology treats “weight” as a behavior derived from
physical structures (Gero and Kannengiesser 2004), and the analytic hierarchy
process treats “weight” as a criterion to select the best design alternative (Saaty
2008). It is most challenging to identify and manage constraints during the problem
definition stage when everything remains intangible, subjective, and dynamic.
182 A. Liu

In problem definition, most of the design constraints are intangible and difficult
to determine. Constraints are factors that would limit an artifact from achieving a
more ideal state (i.e., performance, reliability, robustness, etc.). Accordingly, the
instantiation, specification, and management of design constraints hinge on two
necessary conditions: a specific goal and a tangible system. In the context of
engineering design, the former means an initial intent or thought (i.e., customer
need), whereas the latter means a completed artifact (i.e., DPs). In the problem
definition phase, however, the intent is yet to be solicited, clarified, and analyzed,
whereas the artifact is yet to be created, evaluated, and optimized. As a result,
design constraints cannot be stated in an explicit, quantitative, and precise manner.
For example, in the problem definition phase, it is difficult to accurately estimate the
delivery time of a working prototype, though deadline is an important constraint for
any product development.
In problem definition, design constraints could be easily confused with FRs. FRs
are the real target of engineering design, whereas constraints are bounds to viable
solutions, which are intended to fulfill the FRs. Different from FRs that should
always be formulated and maintained independent of each other, it is difficult to
make various design constraints entirely independent of each other. For example,
the cost of an artifact is contributed by all the components, while it is mostly
unnecessary to decompose cost into “sub-cost” and allocate them for different
components. In terms of the relationship between FR and DC, according to AD, it
becomes more efficient to select FRs when design ideation is appropriately con-
strained. Whatever the case may be, innovative design must be driven by FRs and
limited by design constraints.
According to AD, design constraints are classified into “input constraints“ and
“system constraints.” Input constraints are designer-independent and must be
complied by all design solutions. On the other hand, system constraints are intro-
duced by your own decisions, and therefore they are designer-dependent and
context-dependent. The major distinction between input constraint and system
constraint hinges on the original source. If a constraint comes from the explicit
requirements imposed by relevant stakeholders other than yourself, it is classified as
an input constraint. In contrast, if a constraint comes from your own decisions, it is
classified as a system constraint. The difference between input constraint and
system constraint can be illustrated by the comparison between “budget” and
“cost.” At first glance, “budget” and “cost” seem to mean the same thing. However,
they belong to different categories of constraints. Budget means a financial plan of
expenses for designing an artifact, whereas cost means the number of resources for
producing the artifact. In practice, a budget is mostly decided by the management
and imposed on you. Therefore, it should be attributed as an input constraint. In
contrast, cost is primarily determined by your own decisions in terms of your
unique choice and synthesis of components, structures, materials, manufacturing
process, etc. As a result, cost should be attributed as system constraint.
Constraints can be either internal or external of an artifact. Design begins with an
initial though that keeps growing toward a sophisticated artifact. The artifact’s
evolution is constrained by both internal and external forces. Internal constraint is a
5 Problem Definition 183

part of the artifact; hence, it limits the evolution of the artifact from the inside. For
instance, the choice of combustion engine will limit a car’s emission performance
and fuel efficiency. This internal constraint can only be altered if the combustion
engine is replaced by, for example, electric motor. In contrast, external constraint is
not a part of the artifact, and as a result, it bounds the evolution of the artifact from
the outside. For instance, if an artifact’s structure cannot be manufactured by the
readily available machine tools, then the machine tools will become an external
constraint. In particular, government regulation is a typical external constraint,
which can prevent illegal products from being developed.
Various design constraints are classified into four categories: internal input
constraint, external input constraint, internal system constraint, and external system
constraint (Liu et al. 2019).

• Internal input constraint: constraint that is a part of the artifact, though it is not
chosen by you but imposed to you by the third-party stakeholders. For example,
as per the government regulation, seat belt is a mandatory part of a car seat
independent of the designer’s decision.
• External input constraint: constraint that is not a part of the artifact, while it is
included in the design task or problem statement. For example, if a product is
developed for a particular country, the country’s local culture and regulation will
impose constraints to product design.
• Internal system constraint: constraint that is chosen by you to be a part of the
artifact, which will intentionally or unintentionally limit behaviors of the artifact.
For example, the seat belt alarm is purposefully added to constrain the driver–car
interactions.
• External system constraint: constraint that is not a part of the artifact, which is
introduced by your previous decisions. For example, the oil grade will limit a
car’s fuel efficiency, though oil is not a part of car. In other words, bad decisions
will introduce unnecessary constraint.

5.4.2 Examples of Design Constraint

Uber is is a popular smartphone app that provides peer-to-peer ride-sharing ser-

vices. As a smartphone app, the performance of Uber is inherently limited by the
internal constraints of smartphone. For example, the GPS will limit Uber’s ability to
locate the user in areas where the GPS signal is weak, the battery life will limit the
long-time usage of Uber, and the Big Data analytics will limit Uber from finding the
best match between driver and rider among numerous possibilities and calculating
the best route against the dynamically changing traffic. In the meantime, the
development of Uber has been limited by some external constraints. An apparent
external constraint is government regulation that varied in different contexts (i.e.,
different countries and cities). For example, some countries have explicit laws
against any kind of phone usage during driving; some countries forbid UBER cars
184 A. Liu

from picking up passengers in the international airport; and some counties require
that the Uber drivers should be hired as employees in place of contractors. In the
early days, the expansion of Uber was constrained by the resistance of taxi drivers.
Take the toy design for example, since the target customers are mostly children,
toy design is bounded by a number of hard constraints that cannot be violated in
any circumstance. Material is an external constraint imposed by government reg-
ulation. Most of the governments have mandatory standards in terms of which
materials cannot be used in toy design. Size is a critical internal constraint. For
example, a minimum size is imposed on the LEGO bricks to prevent child choking.
The size constraint varies in accordance to the child age.
Exercise 5.11:
Can you identify a list of design constraints for the child car seat, and specify a maximum
or minimum value for each constraint?
Exercise 5.12:
Acceleration is an important measure of a car’s performance. Can you identify a list of
design constraints that will limit a car’s acceleration performance?
Exercise 5.13:
Can you identify a list of design constraints that will limit the performance of a delivery
drone to transport packages? For each constraint, can you specify a maximum or minimum
value?

5.5 Definition of Problem Context

Problem definition is situated in a particular context. A variety of contextual

information should be collected, analyzed, and integrated in order to characterize
the uniqueness of problem definition. A unique problem is only valid in a particular
context. As the context changes over time, the problem should be redefined
accordingly.

5.5.1 What Is Design Context?

The notion of “context” has been investigated from different disciplinary per-
spectives. By definition, context refers to the situation in which something occurs.
For problem definition in engineering design, context can be regarded as a col-
lection of information that characterizes a particular situation or scenario, within
which a product/service interacts with the users in appropriate ways to satisfy their
demands against certain constraints. Contextual information can be acquired in
different ways such as the explicit way (e.g., direct communication among product,
user, and environment), the implicit way (e.g., user survey, product review, and
usage report), and the statistical means (e.g., data analytics to discover meaningful
patterns shared by many products).
5 Problem Definition 185

The commonly applicable design contexts include the following:

• Physical context: Information about the surrounding environment, in which the

product is being used, such as temperature, humidity, weather, etc.;
• Social context: Information about how users interact with each other, such as
with whom, when, where, in what occasion, and under what condition the
product is being used;
• User context: Information about end users, such as user demographics,
knowledge, mood, habit, preference, cognition, health, etc.;
• Interaction context: Information about the user–product interactions, such as
browsing history, personalized setting, graphical user interface, etc.;
• Operational context: Information about a product’s operational state, such as
battery life, computing power, intelligence level, maintenance, etc.

Contextual information can affect problem definition in various ways. Firstly, user
context can affect the selection of target customers, and social context can affect the
solicitation of customer voices. For example, contextual inquiry is proven to be a
more effective method than formal interview, because it occurs in the field when the
product is being used by customers. Therefore, some invisible contextual infor-
mation can be revealed. Secondly, the physical context can affect the formulation of
FRs. Even the same FR should be defined differently against different contexts. For
example, watch is generally used to fulfill the FR of “tracking time,” while the
accuracy requirement of tracking time varies significantly in different contexts, such
as Olympic Games, stock trading, business meeting, lecture in the classroom, etc.
Lastly, context can affect the definition of design constraints. The most apparent
example is government regulation, which varied significantly in different countries.
Contextual information is especially useful for designing context-aware smart
products. The advancement of information communication technology paves the way for
the increasing popularity of smart products (e.g., smart car, smart home, smartphone,
etc.). A critical facet of product intelligence is context-awareness. Context-awareness
means a product’s ability to accurately interpret a particular situation, in which, to
intentionally perform appropriate actions and avoid inappropriate behaviors. So far,
many context-aware information systems have already been developed to make per-
sonalized recommendations against different contexts. In the future, it is expectable that
more and more context-aware smart product/service will be developed. For example, a
user-context-aware robot vacuum should purposefully avoid interfering user activities
within the home environment, and a social-context-aware robot vacuum should recog-
nize nearby peer products and available resources.
New artifacts are created not only for the real contexts but also for the imaginary
contexts, such as the movie themes and holiday themes. For example, from time to
time, LEGO offers special sets designed for certain movie themes such as the
Disney theme and the Star War theme. During the Christmas season, customers
tend to purchase various holiday products such as Christmas tree. In most of the
American universities, bookstore is one of the most profitable departments, where a
186 A. Liu

Fig. 5.7 Square watermelon.

(Reproduced with permission
from Volkov 2019)

variety of souvenirs (e.g., T-shirt, coffee mug, etc.) with the university logo are sold
at high prices. Such souvenirs are mostly designed according to the context of
university culture.

5.5.2 Example of Design Context

As illustrated in Fig. 5.7, square watermelon is a popular product that was “in-
vented” in Japan. It is produced by letting the baby watermelon grow inside a
transparent square cube. The original intention of “designing” the square water-
melon was to enhance space efficiency in small refrigerators and facilitate trans-
portation. Compared to the regular watermelon, square watermelon occupies a
much smaller space. However, because of the high cost of growing square water-
melons, in practice, they can only be afforded by the rich families and hence go
beyond the original design context of “small refrigerator.”
Exercise 5.14:
Can you create a list of contextual information that is relevant to the design of child car
seat? In other words, under what contexts the child car seat will be used?
Exercise 5.15:
Can you reflect the history and define the design context of iPhone by Apple?
Exercise 5.16:
Can you define the design context of the kimchi refrigerator, which is a very popular
product in South Korea?

5.6 Examples of Problem Definition

Consider the following problems and think about how you might approach them to
develop a design solution for each example. Some of these examples will require
considerable thinking and work to develop the desired design solutions. In some
5 Problem Definition 187

cases, you may have to acquire some fundamental knowledge of the field by
reading reference books or by getting information from the Internet. Most of all,
you may have to think with an open mind!
Example 5.1 Mars Exploration Rover
In light of the successful launch of Falcon Heavy by Space X, NASA is interested in
initiating another ambitious mission to explore Mars. Key equipment is an autonomous
exploration rover. Assume you were the Director of NASA, can you prepare a one-page
proposal to lobby the Congress for support of this project? What are the FRs and design
constraints?
Example 5.2 Urban Farming System
Farming is one of the earliest human endeavors to alter the environment to satisfy basic
human needs. Before the Industrial Revolution, farming plays a critical role in driving the
historical development of human society. In modern society, however, farming is no longer
accessible to ordinary people who live in the megacity. Against this background, can you
develop a problem definition for designing a small-scale farming system that can be sus-
tained in urban apartments? Who are the target customers? What are the design constraints?
How is the new context different from traditional farming in the filed?
Example 5.3 Smart Factory
Making factories more intelligent is one of the key value propositions of Industry 4.0. The
advancement of new technologies such as sensor, actuator, Big Data analytics, and artificial
intelligence paves the way for the development of smart factory, which is characterized by
information transparency, autonomous operation, decentralized decision-making, etc.
Assume you were the CEO of a global auto manufacturer, can you frame a problem
definition for designing a brand new smart factory?
Example 5.4 Mobile Learning Management System
Learning management system (LMS) is widely used in higher education for both teacher
and student to manage learning activities in the cyberspace. Traditionally, the LMSs are
mostly designed for the context of computer access. In light of the increasing popularity of
smartphone and hence the transition of student attention from computer to smartphone, can
you frame a problem definition for designing a mobile LMS? What is the new design
context, and how is it different from the old context?
Example 5.5 Samsung Store
The Apple Store successfully revolutionized customer’s impression of retailing. Assume
you were the CEO of Samsung Electronics, in order to compete with Apple, can you frame
a problem definition for designing a Samsung Store, which will be given to your senior
managers and chief designers? Who are the target customers? Who are the relevant
stakeholders? What are the customer needs? What are the design constraints? What are the
FRs? What is the problem context?
Example 5.6 Access to Clean Water
Access to clean water is one of the 14 grand challenges for engineering in the twenty-first
century, which were identified by the National Academy of Engineering in the USA. Every
year, more deaths are caused by lack of clean water than wars. Assume you were the
director of a global NGO foundation, can you design a 5-min presentation to highlight the
significance of this grand challenge, and call for innovative, affordable, and reliable
engineering solutions to address it?
188 A. Liu

Example 5.7 Kitchen Product

Of all the rooms in your home, kitchen is the center of energy, creativity, and comfort. In
the coming decade as our habits change, the kitchen as we know it will evolve drastically.
Today’s youth, especially college students, tend to avoid cooking by themselves in the
kitchen because it is time-consuming, inconvenient, and intuitive. This is evidenced by the
fact that the food delivery service is becoming increasingly popular worldwide. Can you
frame a problem definition for designing a new smart appliance that can facilitate cooking
in the kitchen?
Example 5.8 Pet Entertainment System
Dog/cats have been a fundamental part of many households and are known to be quite
active and responsive to their owners. However, oftentimes, owners must have periods of
absences due to other commitments such as jobs and study, leaving their dog/cats at home
alone. Can you develop a problem definition for designing a new system that can entertain
or accompany dog/cat when their owners are away from home? Who are the target cus-
tomers and relevant stakeholders? What are the FRs? What is the problem context?
Example 5.9 Bicycle Security on Campus
Bicycles are one of the most popular transportation means on the university campus.
However, bicycle security has been a long-standing problem on the university campus. Can
you create a problem definition for designing a new system that can prevent bicycles from
being stolen or damaged on campus?
Example 5.10 Research Submarine
Some researchers need to carry out expeditions to explore the deep ocean. Can you develop
a problem definition for designing a research submarine for scientific purposes?
Example 5.11 NGO for Climate Change
Climate change is one of the greatest challenges for humanity. Can you create a problem
definition for developing a not-for-profit organization that is intended to promote social
awareness in climate change? How do you propose to tailor the problem definition dif-
ferently for different contexts (i.e., developing country and developed country)?

In the above examples, the success of design largely depends on how, in what ways,
and to what extent a good problem definition is formulated. As much as possible,
your problem definition should cover all aspects concerning customer need, FRs,
design constraint, and problem context. For design instructors, these examples, as
well as their variations, can be assigned as the topic of project-based learning.

5.7 Summary and Conclusion

Chapter 5 illustrates the critical role of problem definition in driving the design
process toward more innovative solutions. It has been repeatedly proven by
numerous product development successes as well as failures that a good problem
definition is a necessary condition of design innovations. The quality of problem
5 Problem Definition 189

definition determines the direction of subsequence design activities such as concept

generation, evaluation, and selection. Problem definition involves the iterative
formulation, specification, reformulation, and mapping of customer needs, FRs,
design constraints, and design contexts. It requires both divergent thinking and
convergent thinking to develop a good and unique problem definition. Divergent
thinking is needed to expand the searching for innovating opportunities, customer
voices, boundary conditions, and contextual information. Convergent thinking is
equally needed to convert the unstructured, separate, and inconsistent information
into a set of well-defined FRs.

References
Delyk O (2019) MP3 player audio musical portable headphones Walkman iPod iPhone, Apple
isolated 3D. Photo licensed from Shutterstock, ID: 272031950
Lyjak B (2019) Mother with child in baby stroller enjoying summer sunset and mountains
landscape. Jogging or power walking woman with pram. Beautiful inspirational mountains
landscape. Photo licensed from Shutterstock, ID: 1045287292
Popov A (2019a) Side view of a relaxed businesswoman stretching her arms. Photo licensed from
Shutterstock, ID: 1083974525
Popov A (2019b) Young businesswoman sitting on Pilates ball working in office. Photo licensed
from Shutterstock, ID: 396426541
Volkov V (2019) Cubic watermelon with slice on a white background. Photo licensed from
Shutterstock, ID: 106115846

Further Reading

Franke N, Von Hippel E, Schreier M (2006) Finding commercially attractive user innovations: a
test of lead-user theory. J Prod Innov Manag 23(4):301–315
Gero JS, Kannengiesser U (2004) The situated function-behaviour-structure framework. Des Stud
25(4):373–391
Goldratt E (1999) Theory of constraints. North River Press
Kaulio MA (1998) Customer, consumer and user involvement in product development: a
framework and a review of selected methods. Total Qual Manag 9(1):141–149
Liu A, Wang Y, Teo I, Lu S (2019) Constraint management for concept ideation in conceptual
design. CIRP J Manuf Sci Technol 24:35–48
Saaty TL (2008) Decision making with the analytic hierarchy process. Int J Serv Sci 1(1):83–98
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Suh NP (2010) Theory of innovation. Int J Innov Manag (IJIM) 14:893–913. https://doi.org/
10.1142/S1363919610002921
How Should We Select Functional
Requirements? 6
Nam Pyo Suh

Abstract
The preceding chapters provided broad outlines of the design of various systems
based on Axiomatic Design (AD). Two axioms were the basis of making design
decisions: the Independence Axiom and the Information Axiom. The process of
AD consists of the transformation of the problem identified in the “customer
domain” into a set of functional requirements (FRs) in the “functional domain,”
which was in turn transformed as design parameters (DPs) in the “physical
domain” that are chosen to satisfy FRs. DPs were, in turn, transformed into
process variables (PVs) in the “process domain” to fulfill DPs. The mapping
process was illustrated.
Once the highest FRs and DPs are finalized at the highest level of the system
design hierarchy, we may have to decompose FRs and DPs, if the selected FRs
and DPs lack sufficient details to complete the design. This decomposition
process must continue until the design has enough details that can be
implemented. To decompose, we should zigzag between FR and DP domains.
Similarly, DP versus PV can be decomposed through zigzagging.
The resulting designs were classified as uncoupled, decoupled, and coupled
designs. Coupled designs violate the Independence Axiom and, thus, should not
be implemented. They are unreliable and require monitoring, resulting in the
waste of resources, cost overruns, and delays in implementation. When a system
design is uncoupled, it can readily be satisfied because each FR is a function of
only one DP, irrespective of the total number of FRs and DPs that the system has
to satisfy. The design matrix is constructed to identify coupling. Throughout
these chapters, “real-life” examples were presented.
The quality of the design is determined by how well the problem is identified
and how FRs are selected. Although every step of the transformation is essential,

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 191

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_6
192 N. P. Suh

the selection of FRs ultimately determines the functional quality of the design
output. It is helpful to have an in-depth understanding of basic sciences,
engineering, and other relevant fields to be proficient in system design. It may
also take different kinds of information, depending on the field. In designing
commercial products, market information on the current state and prospect is
critical in selecting the right set of FRs. In other fields, such as the economy, the
set of information required to choose the right set of FRs would be entirely
different. However, the basic system structure is similar. The domain-specific
knowledge is field-specific.
In this chapter, the process of selecting FRs is elaborated further. One of the
essential concepts to remember in selecting FRs is the idea of choosing them in a
“solution-neutral environment,” that is, the designer should not think of a design
solution first and then define FRs for the assumed solution. Given the critical
importance of selecting the right set of FRs for the problem identified, the
designer should have a broad knowledge base and also access to extensive
“database for scientific facts and various technologies,” including the commer-
cial database for existing products.

6.1 Importance of “Solution-Neutral Environment”

in Selecting Functional Requirements

The preceding chapters provided broad outlines of the design of various systems
based on AD. Two axioms were the basis of making design decisions: the Inde-
pendence Axiom and the Information Axiom. The process of AD consists of the
transformation of the problem identified in the “customer domain” into a set of FRs
in the “functional domain,” which was in turn transformed as DPs in the “physical
domain” that are chosen to satisfy FRs. DPs were, in turn, transformed into PVs in
the “process domain” to satisfy DPs. The mapping process was illustrated.
Once the highest FRs and DPs are finalized at the highest level of the system design
hierarchy, we may have to decompose FRs and DPs, if the selected FRs and DPs
lack sufficient details to complete the design. This decomposition process must
continue until the design has enough information for implementation. To decom-
pose, we should zigzag between FR and DP domains. Similarly, DP versus PV can
be decomposed through zigzagging.
The resulting designs were classified as uncoupled, decoupled, and coupled
designs. Coupled designs violate the Independence Axiom and, thus, should not be
implemented. They are unreliable and require monitoring, resulting in the waste of
resources, cost overruns, and delays in implementation. When a system design is an
uncoupled design, it can readily be satisfied because each FR is a function of only
one DP, irrespective of the total number of FRs and DPs that the system has to
6 How Should We Select Functional Requirements? 193

satisfy. The design matrix is constructed to identify coupling. Throughout these

chapters, “real-life” examples were presented.
The quality of the design is determined by how well the problem is identified and
how FRs are selected. Although every step of the transformation is essential, the
selection of FRs ultimately determines the functional quality of the design output. It
is helpful to have an in-depth understanding of basic sciences, engineering, and
other relevant fields to be proficient in system design. It may also take different
kinds of information, depending on the field. In designing commercial products,
market information on the current state and prospect is critical in selecting the right
set of FRs. In other fields, such as the economy, the set of information required to
choose the right set of FRs would be entirely different. However, the basic system
structure is similar. The domain-specific knowledge is field-specific.
In this chapter, the process of selecting FRs is elaborated further. One of the
essential concepts to remember in selecting FRs is the idea of choosing them in a
“solution-neutral environment,” that is, the designer should not think of a design
solution first and then define FRs for the assumed solution. Given the critical
importance of selecting the right set of FRs for the problem identified, the designer
should have a broad knowledge base and also access to extensive “database for
scientific facts and various technologies,” including the commercial database for
existing products.
Example 6.1 Lesson Learned: Role of Solution-Neutral Environment in Developing
On-Line Electric Vehicle (OLEV)1
In Chap. 3, the basic design of OLEV (on-line electric vehicle) was presented by giving
details of the design of the shaped magnetic field in resonance (SMFIR). In this chapter, the
background story will be presented to illustrate the difficulties one may encounter in
undertaking innovative R&D projects to show the importance of thinking in a
“solution-neutral environment” in creating innovative products.
As mentioned in a previous chapter, in 2006, KAIST (Korea Advanced Institute of Science
and Technology) designed a new plan to make it one of the best universities in the world in
science and technology. To achieve this goal, one of the FRs chosen for the new KAIST
strategic plan was to solve some of the most critical problems of the world in the
twenty-first century. One of the issues identified was “global warming.” To solve this
problem, we decided to replace all internal combustion (IC) engines used in automobiles
with electric drives (EV). Such a shift from IC to EV would reduce the anthropogenic
emission of CO2 by about 27%. Also, to avoid many problems associated with using a large
number of batteries to propel all-electric vehicles, the decision was made to transmit electric
power wirelessly to the vehicle from underground electrical power supply systems. We
named it the on-line electric vehicle (OLEV). KAIST approached the government of the
Republic of Korea for significant funding (*about $25 million during the first year, a large
sum of money for research and development, by any measure).

1
OLEV was partly introduced in Chap. 3 to illustrate the design process.
194 N. P. Suh

When the Korean government tentatively agreed to fund this OLEV project pending the
congressional budget approval, two unexpected things happened. First, none of the KAIST
professors who were specialists in electric power would be willing to lead this project,
stating that the idea of transmitting 100 kW of electric power wirelessly over a distance of
20 * 25 cm could not be done technologically. They claimed that the idea was meritless,
although they were too polite to use these words! Therefore, the KAIST administration
organized a project team led by professors who were not specialists in heavy electric power
engineering. They did a great job because they were able to design a new system based on
the first principles of electromagnetism and AD.
The second event was much more serious. Some of the professors at other universities, who
were working on battery-powered electric automobiles, mounted a significant opposition to
the KAIST project. They mobilized a member of the National Congress (equivalent to the
U.S. Congress) with a Ph.D. in physics from a university in the U.S. and former professor
of physics in Korea to mount a significant opposition campaign to the KAIST OLEV
project. They argued that such massive electric power (*100 kW) could not be transmitted
wirelessly over a distance of over 20 cm, similar to the argument presented by some
specialists in KAIST. (For proprietary reasons, we could not reveal the basic idea behind
OLEV.) They unduly interfered with the project proposal review process to stop the
funding of the KAIST project. KAIST president was called in to answer questions at a
special meeting of the National Assembly (i.e., counterpart the U.S. Congress) of Korea.
The opposition to OLEV was based on their claim that the proposed idea was not tech-
nically feasible. This incident is similar to the opposition mounted in the United States
against the NSF Engineering Research Centers (ERC) program when it was launched in
1985, albeit based on a different argument.
The professors who opposed the OLEV project were not thinking in a “solution-neutral
environment.” Instead of thinking about the goal of transmitting vast electric power over a
considerable distance in a solution-neutral environment, they were thinking of the
well-established designs that have been used in electric power transformers, where the
primary coil is adjacent to the secondary coil. OLEV was based on the idea of creating an
oscillating magnetic field of required shape and strength by controlling the distance
between the two magnetic poles. The idea was generated based on the principles taught in
undergraduate physics courses by people who were not specialists in heavy electric power
transmission.
The KAIST project ultimately received its funding after a great deal of effort to overcome
the opposition of influential professors at two leading universities and some members of the
National Congress. It took 2 years—a short time after receiving the grant to develop OLEV
and install it in Seoul Grand Park and entertain people around the park. Throughout the
execution of the project, we made sure that there was no coupling of FRs by creating
“system architects,” who monitored design decisions made by various groups to be sure
that no one in the design team introduced coupled designs. Now OLEV buses are running
in five cities in Korea. AD played a significant role in achieving this feat.
Thinking in a solution-neutral environment is essential for creative design, which is a
difficult thing to do for people with extensive design experience. The human instinct is to
go directly to a solution without defining FRs.
6 How Should We Select Functional Requirements? 195

Example 6.2 Space Life Support Systems2

The closed system human life support architecture now implemented in the International
Space Station has been virtually unchanged for 50 years. In contrast, brief missions such as
Apollo and Shuttle have used open-loop life support. As mission length increases, greater
system closure and increased recycling become more cost-effective. Closure can be grad-
ually increased, first recycling humidity condensate, then hygiene waste water, urine,
carbon dioxide, and water recovery brine. A long-term space station or planetary base could
implement nearly full closure, including food production. Dynamic systems theory supports
the axioms by showing that fewer requirements, fewer subsystems, and fewer intercon-
nections all increase system stability. If systems are too complex and interconnected,
reliability is reduced, making operations and maintenance more difficult.
Using AD, we need to show how the mission duration and other requirements determine the
best life support system design, including the degree of closure in a solution-neutral en-
vironment. The highest level FR and DP are as follows:

FR = Support human life in space;

DP = Life support system.
The next-level FRs and DPs may be stated in a solution-neutral environment by decom-
posing FR1 and DP1 as follows:

FR1 = Provide atmosphere (transmit oxygen to humans3);

FR2 = Provide water (maintain healthy level of water in humans);
FR3 = Handle waste (remove gas waste products);
FR4 = Suppress fire (restrict combustion to designated areas);
FR5 = Provide food (maintain healthy level of nutrition in humans).

DP1 = Atmosphere system;

DP2 = Water system;
DP3 = Waste system;
DP4 = Fire system;
DP5 = Food system.

H.W. Jones states the following:

“A key method of the AD approach is to match each requirement with its design imple-
mentation at each stage of the top-down elaboration of the requirements. The lower level
requirements are more specific and detailed. This approach is a deliberate direct contrast to
the usual method of creating a detailed multilevel requirements tree, freezing it, and then
developing the hardware design to meet that set of detailed requirements. It is supposed, but
rarely happens, that the requirements are developed without assuming some system
design.”

“While it helps in clarifying requirements, the main purpose of going back and forth
between requirements and systems in the AD approach is to ensure maximum decoupling
of each requirement from the systems implementing other requirements. The extent of
decoupling obtainable can be limited by the environment and the available hardware
systems.”

2
From Jones (2017).
3
FRs in italics were suggested alternate statements by a reviewer of this chapter.
196 N. P. Suh

The design matrix [DM] for the above FRs and FPs is given as follows by Jones (2017):

DP5: Food DP4: Fire DP3: DP1: DP2:

system system Waste Atmosphere Water
system system
FR5: Provide X
food
FR4: X
Suppress fire
FR3: Handle X (food and X
waste packaging
waste)
FR1: Provide X (De- and X
atmosphere re-pressurize)
FR2: Provide X (extract X
water water)

The above design given by H. W. Jones is a decoupled design given by the above design
matrix. This design matrix satisfies the Independence Axiom.

Example 6.3 Consequences of Operating with Wrong FRs: Eastman Kodak Company
The importance of selecting the right set of FRs cannot be over-emphasized. Consider the
case of Eastman Kodak. It was one of the most successful companies until about 1980.
They dominated the camera and film business. They employed thousands of people and
created the City of Rochester, New York. (The founder of the company, George Eastman,
even gave a sizable gift to MIT to build the current campus on the bank of Charles River in
Boston, changing the name from Boston Polytechnic Institute to the Massachusetts Institute
of Technology. Without the money, MIT might have become a part of Harvard!) There
were outstanding engineers, scientists, managers, and executives who made Eastman
Kodak a leading industrial firm in the world. They pioneered silver-halide-based photog-
raphy technologies well into the twentieth century.
Today, the company does not exist anymore. It is not appropriate to blame anyone or a
group of people. However, the board of the company should bear the most blame because it
is their function to be sure that the company is headed in the right direction by hiring an
able leader as the Chief Executive Officer of the company. Apparently, “they dropped the
ball.” In many of these situations, however, it is the CEO who should bear the primary
responsibility, because CEO’s job is to steer that company toward success by taking
appropriate actions to make the company successful. CEOs of companies are highly
compensated because CEO’s actions indeed determine the success of corporate goals. One
of the last CEOs at Kodak, the board of directors hired, had made his reputation in
semiconductor and electronics business, and yet he persisted, thinking that the silver-halide
system could compete with the electronic photography business. He chose a wrong set of
the highest level FRs!
When the electronic camera was being introduced, the upper management of Eastman
Kodak continued to believe in the silver-halide system of making images on thin polymeric
films and printing them on photographic paper. The irony is that Eastman Kodak was one
6 How Should We Select Functional Requirements? 197

of the first companies to develop electronic photography.4 In hindsight, it is clear that the
top management of Kodak chose a wrong set of FRs. They believed in the superiority of the
silver-halide system because the company continued to improve the silver-halide system.
Also financially, they had so much invested in old photography business based on the
silver-halide system that they could not believe that digital technology could ever overtake
their traditional business and obsolete their massive capital investment. When the digital
camera was becoming popular, they even imported a top manager from electronics and
semiconductor business, but he also became a convert to the old silver-halide system of
photography. They did not fathom how fast semiconductor technology would miniaturize
images (i.e., pixels per inch) and how digital technology would transform the communi-
cation technology involving images. Therefore, they did not fully support their digital
technology group, although they got into digital photography first, because they could not
foresee such rapid progress of electronic photography and demise of the old photography
business in the marketplace. Because of the mindset of the top executives of Eastman
Kodak, the lower level managers set the goal of saving cost and increasing productivity as
their goals.5 Many of them have done outstanding jobs, but they were not the captains of
the ship.

The cited example above illustrates the importance of “knowledge” and the ability
to make the right decisions in selecting the right set of FRs based on the available
knowledge on a periodic basis. It also illustrates the conflict in decision-making
between the short-term profits and creating new business by gutting an existing
business that is bringing in cash, profits, and the current prosperity. The top
management of Eastman Kodak failed to protect its employees and investors by
having made a wrong decision and ultimately led to the demise of such a great
technology firm. It is clear that when the CEO had a chance to choose a new set of
FRs upon becoming a new CEO of Kodak, he stuck with the FRs his predecessors
had chosen, which eventually bankrupted the company. The mystery is why a
person who had a strong career in the semiconductor and electronics industry chose
to stay with the silver-halide system. Is this because the system rewards the CEOs
based on their short-term performance? This story highlights the importance of
selecting the right person as the CEO of a company, who has the intellect, expe-
rience, and the ability to design a long-term strategy for the company.
Exercise 6.1: Design of Driverless Car
Many companies are developing driverless cars, such as the one shown below. Is the
driverless car a good design? State FRs and constraints in a solution-neutral environment
(Fig. 6.1).

4
An engineer at Eastman Kodak, Steven Sasson, invented and built the first
self-contained electronic camera. It used a charge-coupled device image sensor in 1975, which
was initially used in mainly military and scientific application. Later, it was used in medical and
news applications.
5
This emphasis on cost-saving led to the creation of the microcellular plastics by MIT under the
sponsorship of Eastman Kodak and other companies in the MIT-Industry Polymer Processing
Program. Today, MuCell (tradename of Trexel, Inc., for microcellular plastics) is used in
automotive industry to reduce the weight of cars so as to increase the fuel efficiency of
automobiles. Reference: Wong et al. (2016).
198 N. P. Suh

Fig. 6.1 Driverless car.

Reproduced with permission
from Tim (2020)

Exercise 6.2: Unbeatable Combination: Human and Machine Intelligence

Many things people do every day depend on a combination of human and machine intel-
ligence. People decide what to do, and machines take over the task and deliver the results to
the person. For instance, people order goods through the Internet, and automated systems
take over the order and deliver the good to the door step of the person who ordered it.
One of the least developed areas in human society is “politics.” Politicians often lie to
people and argue with other politicians with lots of heat but with limited enlightenment to
resolve issues and establish policies. We wish to develop an intelligent machine system that
can settle political arguments and make decisions. Develop FRs and constraints that can
improve the democratic decision-making process.

Exercise 6.3: Admission to Leading Colleges

Colleges and universities accept students for their freshman class each year. Many students
apply for admission to the colleges they would like to attend. Colleges use several different
criteria for admission. Along one axis, they measure the scholastic aptitude of the students
by several different means such as entrance examination, high school grades, and letters of
recommendation. Along the orthogonal axis, they attempt to measure personal charac-
teristics for future success, such as student activities, letters of recommendation, particular
unique talents, and extracurricular activities. The problem that colleges and universities
are facing is that the number of qualified students exceeds the number of students they can
accept. Design an admissions system for screening students by stating FRs and DPs in a
solution-neutral environment.

Exercise 6.4: Mission to Moon

The space agency of the United Nations decided to colonize the Moon by establishing a
base where people can stay for an extended period. To achieve this goal of colonization of
the Moon, heavy equipment and supplies must be transported to the Moon from Earth.
There are two options. The first option is to use a large enough booster rocket to carry the
cargo from Earth to Moon in one shot by being able to escape the gravitational force of the
6 How Should We Select Functional Requirements? 199

Earth. Then, once it reaches a point where the gravitational pool of the Moon is greater
than that of the Earth, it will travel to and ultimately go into Moon’s orbit for final landing
on the Moon by slowly descending to the lunar surface. The second option is to use smaller
rockets to take the payload to an orbit around Earth in several launches. Then, assemble
the parts into a larger container in space while circling the Earth. When the entire space
vehicle is assembled in the earth orbit, it can fire a rocket to propel it to Moon.
If you are the director of this Moon mission, which option would you choose? Why? List
FRs in solution-neutral environment and develop DPs that can satisfy the FRs. Discuss the
pros and cons of these two different approaches.

Exercise 6.5: Stabilizing the Weather

It is desirable to change the weather pattern of Earth to prevent global warming. Weather
is controlled by a series of instabilities in the atmosphere. When instability is initiated in a
local region (e.g., due to the abnormally hot zone in the equator), it will create fast
circulating vortices. These become tornados that will gather more and more energy by
entrapping vapor from the ocean surface and then unleash the condensate when it
encounters a cold atmosphere, creating a large pressure gradient and additional instability
in the atmosphere. If these unstable natural phenomena can be created by first initiating
local instabilities at a preselected region (e.g., west coast of Northern Africa), we may be
able to change the weather pattern of Earth. How would you change the desert in North
Africa next to the Atlantic Ocean into a green forest with many trees, making use of
human-made instabilities in the atmosphere? List FRs and DPs at the highest level.

Exercise 6.6: Energy Storage

One of the issues in using electric power generated by solar and wind is the storage of
excess electrical energy when the electric power generated is higher than the demand.
Ideally, we should be able to store excess energy and use it when the demand is greater
than supply. How would you store electric energy that may exceed the demand by tens of
million watt-hours? List your FRs and DPs.

6.2 Functional Requirements and Zigzagging

between Domains to Generate Lower Level Functional
Requirements through Decomposition

At the highest level of design, the FRs may be stated based on the problem iden-
tified in the customer domain. However, we cannot identify the lower level FRs,
without first identifying the corresponding DPs that can satisfy the FRs identified at
the highest level. The lower level FRs can be stated after we first select and define
the DPs at a higher level. These FRs are children-level FRs that represent the
functions of the DP at one higher level. Therefore, mapping and zigzagging are
essential concepts that the designer should be familiar with since they are the
fundamental tools of AD.
200 N. P. Suh

Example 6.4 Designing Chemical–Mechanical Polishing (CMP) Machine for Silicon

Wafers6

In manufacturing semiconductors, there are many steps involved. One of the processes is
the chemical-mechanical polishing (CMP) process. Flat silicon wafers (diameter of up to 12
inches) is first coated with a thin layer of photoresist by spinning the wafer at high speeds
after the photoresist is put on the surface. Then, electric circuits are printed on the pho-
toresist with a lithography machine, using a short wavelength light (e.g., typically ultra-
violet light (UV)) that goes through a mask with open slots in the form of electric circuits.
Then, the resin is cured through a cross-linking chemical reaction, except the area that was
exposed to light. When the uncured resin is removed, it exposes the original silicon surface
area where the surface was exposed to light. Then it is etched, creating thin channels. Then
copper is deposited, which covers the entire wafer. When the wafer is polished using
abrasive slurry with polishing pads by the CMP machine, it removes the copper except
where the etched channel exists, making a conducting path for electricity (see Fig. 6.2).
Then, another layer of semiconducting material is deposited by vapor deposition, which is
subjected to similar processes until a semiconductor device is created after the repetition of
the above process. The task of the CMP machine is to remove unwanted materials to make
the electric circuit layer by layer. Throughout this process, the CMP process must remove
materials uniformly to prevent “dishing” of the surface where the center of the wafer is
removed more than the edges or removing more materials at the edges of the wafer.
There are many FRs that must be satisfied. To simplify for illustration, we will choose a
second-level FR and show how that FR is decomposed to create the next level of FRs and
DPs (Melvin 2003a).

FR1.1 = Process wafer;

DP1.1 = Front layer removal.
The next-level FRs will be stated by decomposing FR1.1, which are really the FRs of
DP1.1. They may be stated as follows:

FR 1.1.1 = Remove surface material;

FR 1.1.2 = Enable multi-step processes;
FR 1.1.3 = Control remaining thickness;
FR 1.1.4 = Exchange wafers.
The corresponding DPs are as follows:

DP 1.1.1 = Abrasive removal processing;

DP 1.1.2 = Multiple removal station design;
DP 1.1.3 = End point signal;
DP 1.1.4 = Wafer exchange sequence.
It should be noted that another designer might have decomposed it differently, which is
equally acceptable if it is done consistently. In the design, there is no uniqueness theorem,
but the decomposition must be consistent within the design framework chosen.

6
The MIT Laboratory for Manufacturing and Productivity developed a commercial CMP machine
in about 2 years. This example is based on the extensive design work done by Jason Melvin who
received his Ph.D. partly based on this work and the work of other students. As discussed in
Chap. 4, this project was sponsored by SVG Corporation, which was acquired by ASML. We are
grateful to Papken Der Torossian, CEO of SVG, a man with a vision.
6 How Should We Select Functional Requirements? 201

Cu
m SiO2

Fig. 6.2 Etched wafer after copper deposition (left) and after removing the copper layer by the
CMP process (right). Reproduced with permission from Melvin (2003b)

The design matrix for the above design is given below:

8 9 2 38 9
>
>
FR1:1:1 >
>
X O O O > DP1:1:1 >
> >
>
< >
= >
6 X X O O 7 DP1:1:2 >
< =
FR1:1:2 6 7
¼6 7
> > 4 X X X O 5>
> FR1:1:3 > > DP1:1:3 >
>
>
: >
; >
: >
;
FR1:1:4 X O O X DP1:1:4

The design is a decoupled design, which satisfies the Independence Axiom.

The above example is a small part of the overall design of the MIT CMP machine, which
was presented in Chap. 4 as part of the discussion on the representation of design results,
which is shown again in Fig. 6.3.

Fig. 6.3 The figure on the left is “Solidworks model” and the picture on the right is the actual
fabricated machine. Reproduced with permission from Melvin (2003b)
202 N. P. Suh

6.3 Role of Broad Knowledge Base/Data Base

in Formulating Functional Requirements

Problem identification is a critical step in the successful execution of the design. In

previous chapters, the design was defined as the transformation process of going
from domain to domain, e.g., to generate FRs from the problem identified in the
customer domain, which is facilitated by “knowledge” and “data.” At the beginning
of the transformational process, we sometimes do not even know “the question” or
“the problem.” Initially, we depend on the “knowledge stored in our brain” or the
“data stored in some accessible database” to generate questions and ideas to cor-
rectly identify the problem we need to solve in the customer domain, which pre-
cedes the selection of FRs or DPs or PVs. The generation of DPs and PVs is much
easier than the generation of FRs, because we may not have adequately understood
the “problem” that resides in the customer domain. Although the computer is
powerful with extensive database and increasing intelligence with the use of arti-
ficial intelligence (AI), human imagination and the human brain are still the most
critical factor in being able to define appropriate FRs.

6.4 On Selection of a Right Set of Functional

Requirements—Need to Review Functional
Requirements from Time to Time

In 1950, General Motors Company was the largest publically held company in the
United States. In 2018, it was no longer. What happened?
General Motors dominated the U.S. economy in the 1950s. Since it was founded, it
absorbed several companies and under the chairmanship of Alfred P. Sloan, and grew to
became the largest automobile company in the world. When President Eisenhower nomi-
nated Charles Erwin Wilson, GM Chairman, for defense secretary, he faced the Senate
Armed Services Committee in 1953. What he said at the hearing has been quoted widely:
“because for years, I thought what was good for our country was good for General Motors,
and vice versa.”

Some corporations go through a crisis when they cannot renew themselves in time as
the world around them changes. Furthermore, some of the large companies become
so bureaucratic that they are no longer creative and competitive. They do not attempt
to identify new problems they should solve, establish new FRs, develop new DPs,
and deal with resource issues (PVs) to continue to revitalize the business. Fortu-
nately, many companies continue to do well by renewing themselves periodically.
Some universities have problems, too. One of the potential crises facing some of the
leading universities is that the cost of operating their universities is exponentially
escalating when the number of students and faculty has not changed. They often
blame new government regulations, but that may not be the only reason or even the
primary reason for the inefficient operation. When governments no longer expand
6 How Should We Select Functional Requirements? 203

the budget for research and do not guarantee student loans, these universities might
have significant problems.
These issues boil down to the following question: how should these institutions
renew themselves? The answer is that they should periodically review and
re-establish FRs. The right set of FRs is the one that solves the problem identified in
the customer domain. Once we select a new set of FRs, we must review the design
and attempt to implement new DPs when the changes are required. These institu-
tions are systems that should reinitialize themselves periodically, which is a subject
matter discussed in the chapter on complexity.
Selecting the right set of FRs is the most important and the most challenging task in
any decision-making process. Anybody can make decisions, but the question is: is
the decision the right one? Fortunately, there is not only “one right decision” in that
there are equally good equivalent decisions and FRs. Once the FRs are chosen, DPs
and PVs must be created or adjusted. Often there are oppositions to institutional
changes when they are introduced, especially during the early stages of a new
administration.
Example 6.5 “You Should Quit for the Sake of Your Great Department!”
When the head of a distinguished academic department got promoted to a deanship, young
faculty members of the department asked the university administration to name a specific
senior faculty member of the department to take on the job. The concern was that the
department has been living off its past laurels. They thought that it was time to change its
direction to be ready for the twenty-first century. Notwithstanding the objections of some
senior faculty in one major group of the department, the administration made the
appointment as recommended by the search committee. He accepted the appointment
because he agreed that the department was now ripe for renewal.
He introduced significant changes, including the renovation of old physical infrastructure,
faculty personnel policy, research emphasis, faculty hiring policy, and raised a substantial
amount of gifts. The fundamental problem he identified was that the department was
dwelling on issues and problems of the late nineteenth-century and early twentieth-century
engineering. The department was famous because the many well-known professors had
made a fundamental contribution to automotive engineering, macro-scale heat engineering,
experience-based instruction of design, and others. These were essential topics to teach, but
in terms of research for the future, the department was not in sync with new emerging
technologies, e.g., semiconductors, nanotechnology, software systems, biomedical engi-
neering, modern materials, and design theory. The department needed different kinds of
professors who can deal with the topics of the twenty-first century and forge a new
mechanical engineering department. The department had to change from a discipline that is
primarily based on physics into a discipline that is based not only on physics, but also
computer science and engineering, biology, information, and modern materials. The
department embarked on the task of hiring professors whose doctorate degrees were in
disciplines outside of the traditional mechanical engineering to lead the department into a
direction that is relevant in the twenty-first century. Also, the old defunct laboratory was
demolished to create new updated physical facilities. He worked day and night to bring
about these changes because he wanted to move onto other tasks after 3 years of this job.
All these changes were conceived in a “solution-neutral environment,” which can have
unpleasant consequences.
204 N. P. Suh

About 5 months into this new task, the chairman of the department’s outside advisory
committee, who was also a member of the trustees of the university, knocked on the door of
the new department head unannounced. They met each other 4 months earlier when the
new department head went to see the chairman of the visiting committee at his home in the
Midwest of the United States, hoping to raise money for the renovation project underway in
the department. Unlike the last time the department head met him, he was not smiling and
looked very serious. For a good reason!
He said that he met with the senior faculty members of the department at their request. He
said that they had many complaints about the new direction the department was headed.
Some of the senior faculty stated to the chairman of the visiting committee: “We must have
been doing something right to be so highly ranked all these years. Then, why change?” He
stated that 50% of those present at the meeting voted to have the department head removed.
That was not a pleasant message. The top administration of the university met and asked the
department head to appear and explain. The initial reaction of the department head was to
step down for the sake of the department. However, others advised him against the idea,
stating that “if you step down, no one else will be able to make unpopular changes, because
the idea that faculty opposition can stop any changes can become a part of department
legend.” The fact that at least 50% of the senior faculty did not join the revolting group
gave enough courage to continue the reform process.
The lessons of this story are the following: (1). Changes are hard to make, especially when
they affect people. (2). Before making changes, one should clearly define FRs and DPs in a
solution-neutral environment. (3). If one wants to be popular and have friends, do not make
significant changes that will tilt the apple cart! (4) Be willing to sacrifice personally for the
sake of the institution. It should be noted that if the new FRs were not conceived in a
“solution-neutral environment,” it would have been more challenging to go through a
difficult period.
He stayed on the job for a long time to be sure that some of the younger professors brought
in from other disciplines into mechanical engineering get tenure and sustain the changing
culture of the department. It took about 10 years for this process! Indeed these “young”
professors have become the leaders of the department and the new mechanical engineering
field worldwide during the period. His contributions to the department and the university
were later widely recognized because of the contributions of these “young” professors.

Factors that affect our ability to choose a right set of FRs:

There may be many factors that may affect decision-making ability, such as the
following:

1. board and relevant knowledge that applies to the issue in hand;

2. in the case of innovation of technologies, a strong base in engineering and
science disciplines;
3. relevant experience in the industry, government, universities, technology
development, and other related fields;
4. for organizational design, experience in establishing and administering systems;
5. actual design experience for any systems;
6. innate creativity;
7. deep perspective in design thinking;
8. ability to think out of the box;
6 How Should We Select Functional Requirements? 205

9. intense curiosity;
10. questioning mind;
11. asking “why” five consecutive times until the root cause is clearly explained;
12. ability to learn new subjects on one’s own;
13. willingness to collaborate with others;
14. an open mind to listen to opposing viewpoints;
15. ability to tolerate criticisms;
16. confidence in one’s knowledge and belief;
17. honesty with high ethical standards;
18. ability to admit mistakes made;
19. ability to listen to viewpoints of others;
20. abundant imagination.

The above list is long. No one person may have all those qualities, but they are
personal characteristics that should be cultivated.

6.5 On Reverse Engineering to Determine the Functional

Requirements of Existing Products

Some companies also try to determine the FRs of the competitor’s products through
reverse engineering, which is difficult or impossible to do. For example, if someone
gives you a hammer and ask you to determine its FRs, you, like most people, would
describe one of its FRs as “drive a nail into wood.” However, your answer might be
wrong if it is used primarily as a paperweight. In other words, it is possible to
reverse engineer DPs because they can be measured, tested, and evaluated, but FRs
can only be guessed.
Every manufacturer, especially those in consumer products, tries to learn about their
competitors’ products. They break down or tear down their competitors’ products to
learn about the merits and demerits of the competitor’s product. However, reverse
engineering has significant limitations. Although it is possible to measure the
geometric shape of the product and determine specific properties through mea-
surements and testing, it is difficult to decide on the FRs of these products through
reverse engineering.
Equally challenging is the determination of tolerances associated with the product,
although the nominal dimensions can readily be measured, their tolerances cannot
be measured. Although we have not covered the Information Axiom in detail yet,
the information contents of a product are also difficult to determine through reverse
engineering, since information content is a function of the tolerance and the
nominal dimensions.
206 N. P. Suh

6.6 Minimum Number of Functional Requirements

The Information Axiom states that information content should be minimized.

Consistent with this axiom, experienced designers try to satisfy a minimum number
of FRs at any given level of design. It may be challenging to deal with many FRs at
any level of decision-making, since as the number of FRs increases, it may become
more challenging to satisfy the Independence Axiom.
Through the decomposition process, we can develop detailed designs. It is easier to
come up with good design ideas and robust designs when we try to limit the number
of FRs we have to satisfy at any given level of design hierarchy. If there is only one
FR, it is always independent.
When we design large systems with many FRs and many decomposition branches,
at any given node of FR and DP, it is better to minimize the number of FRs.

6.7 No Relative Ranking of the Importance of Functional

Requirements

When we specify FRs, they are all equally important. However, the Information
Axiom determines their robustness and the relative importance in terms of the
information content, which will be further discussed in Chap. 7.

6.8 Interdisciplinary Background and Choice

of Functional Requirements

Every designer is likely to choose a different set of FRs and DPs. People with
broader disciplinary or multidisciplinary backgrounds may select better sets of FRs,
because they may be able to access their information base for more appropriate FRs.
For instance, the knowledge acquired on the mixing of liquids through the
impingement of two streams of liquids enabled the creation of mixalloy. Similarly,
if Crick and Watson did not get the information that DNA structure has a helical
shape by visiting the laboratory of Dr. Rosalind Franklin, they might not have
designed and discovered the structure of DNA molecule.
In the future, we will depend more on the database stored in computers to acquire
knowledge on FRs and DPs. In some cases, access to the information will be
facilitated by the use of AI.
6 How Should We Select Functional Requirements? 207

6.9 Importance of Design Matrix and the System

Architecture

In the early stage of the design process, it is easy to change FRs without incurring
much cost and time. However, if we discover that we have chosen a wrong FR, we
may end up redesigning the system from the beginning, incurring a higher cost and
delayed implementation. Then, the designer should change it until the design
adequately addresses the original problem. It is much easier to make changes before
the design is committed to hardware or software development.
Like the example of the CMP machine design indicated, we have to construct the
design matrix to be sure that we do not have coupled design. When a large project is
undertaken, someone or a group of outstanding designers should be designated as
the system architect to make sure that during the decomposition and design process,
a coupled design is not introduced. The responsibility of the system architect is to
construct the design matrix for the entire project to be sure that some of the
decisions made by various participants have not created coupling of FRs.

6.10 Conclusions

Defining FRs is an essential step in design. If a wrong set of FRs is identified, one
has to discard the whole design process and embark on a new process.
This chapter presented different means of defining FRs based on the problem
statement. Sometimes FRs must be defined as part of the decomposition process,
which was briefly reviewed.
The best way of learning how to define FRs is to practice it. By asking the right
questions and by practicing the decomposition process correctly, designers learn
how to decompose an FR and identify children-level FRs.
Once FRs are defined, we need to identify DPs. This process is where the designer’s
creativity may make a difference.

Problem
Six (6) exercise problems are given in the text of this chapter. Choose
even-numbered exercise problems if your last name begins with an alphabet letter
between “a” and “m.” If your last name begins with the letter “n” to “z,” answer the
odd-numbered exercise problems.
208 N. P. Suh

Bibliography
Jones HW (2017) Axiomatic design of space life support systems. In: Proceedings of the 47th
International Conference on Environmental Systems, South Carolina, USA, p 12. https://ttu-ir.
tdl.org/handle/2346/72908
Melvin JW (2003a) Axiomatic system design: chemical mechanical polishing machine case study.
Doctor of Philosophy in Mechanical Engineering, Massachusetts Institute of Technology,
Massachusetts Ave., Cambridge, MA, 02139, USA
Melvin JW (2003b) Axiomatic system design: chemical mechanical polishing machine case study,
Ph.D. Defense Presentation Slides
Tim J (2020) Self-driving intelligent driverless car goes through the city with happy passenger
relaxing. Photo licensed from Shutterstock, ID: 383436070
Wong A, Guo H, Kumar V, Park CB, Suh NP (2016) Microcellular plastics. Wiley, pp 1–57.
https://onlinelibrary.wiley.com/doi/pdf/10.1002/0471440264

Further Reading

Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press

How Should We Select Design
Parameters? 7
Nam Pyo Suh

Abstract
Previous chapters emphasized that the first step in the design process is the
identification of the problem(s) after establishing the goal of the design. Then,
we define the functional requirements (FRs) the design must satisfy to deal with
the problem. Once FRs are determined, we conceptualize a design by selecting
DPs that can satisfy the FRs.
Unlike FRs, design parameters {DPs} are problem-specific, i.e., some deal
with material things, while others may deal with software or a combination of
hardware and software. The DPs dealing with organizations and economic
activities are substantially different from those related to the design of
technology. For DPs that are physical things, their physical integration or
packaging should consider geometric proximity. For those DPs that depend on
information and communication exchange, system integration is done through
software systems.
Each DP should affect only one FR to generate an uncoupled design. The next
acceptable design is a decoupled design, where the relationship between FRs and
DPs results in a triangular design matrix. All other designs are not acceptable
because they violate the Independence Axiom. The selected DP should not
compromise the independence of FRs.
The theorems that govern the design process are given in this chapter. For
example, Theorem 4 states that in an ideal design, the number of FRs and the
number of DPs must be equal to each other. The number of DPs should never be
less than the number of FRs because the design will be a coupled design, which
is not acceptable. When the number of DPs is greater than the number of FRs,
the design is a redundant design.

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 209

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_7
210 N. P. Suh

In developing a large system with many layers of decomposition, there can be

many layers of FRs and DPs, i.e., children-level FRs and DPs, and
grandchildren-level FRs and DPs. In executing a large project, there may be
many layers of decomposition. Thus, to monitor the selection of DPs by all
participants, a Master Design Matrix should be created and monitored by a
system architect to be sure that no coupling is introduced by some of the
participants in the design project.
The task of selecting the right set of DPs to create an uncoupled or decoupled
design is not a trivial task for a complicated system. One way of dealing with
this issue is to mine an extensive database available to come up with a design
matrix that is either diagonal or triangular. Now with the availability of powerful
computers, we may be able to keep searching the database until we find a set of
DPs that yield uncoupled or decoupled designs. In the future, this process may
be assisted through the use of artificial intelligence (AI) and powerful computers.
The coupling of FRs by chosen DPs in any system is the source of many
failures in the field.

7.1 Introduction

The selection of FRs was the topic covered in Chap. 6. In this chapter, we discuss
the selection of DPs to satisfy FRs.
The history of modern science and technology is only about 350 years old
(about ten generations), although humans existed millions of years before that.
Since the Industrial Revolution, people began to use mechanized power. They used
their creativity and ingenuity to make things, mostly through trial-and-error pro-
cesses. Then, the idea that they have to define the problem in terms of FRs and
conceive DPs to satisfy FRs would have been a foreign concept. They solved
problems by jumping into making things (i.e., DPs) through trial-and-error pro-
cesses without stating the FRs formally. This empirical approach has been highly
successful. Even today, many people create innovative technologies successfully
through this brute force approach. Similar things happened in many other fields, but
eventually, logical, rational, and science-based methods and theories have gradually
replaced empirical approaches, because these trial-and-error processes cost too
much, both in financial and human terms, with unpredictable results.
A Unique Lesson for a Young Engineer! Experience has its limitations.
A young student who just finished his third year at an S&T university in the United States
was very fortunate. He got a summer job in the industry as an engineer in a small company
that manufactured disposable plastic products such as cups, dishes, and the like! When he
joined the company, he thought that he would be designing new automatic machines,
because the job was advertised as such. The company assigned him to work for the “chief
engineer,” a creative, experienced, and hardworking head of the machine shop, among
other things. The “chief engineer” did not have formal engineering education, but learned
to engineer through many years of practical experience and have made many things
intuitively even without any drawings. The “chief engineer” and his crew made many new
7 How Should We Select Design Parameters? 211

things: vacuum forming molds, the press for punching out vacuum-formed cups, and also
other related machines needed for the production of plastic products. The young engineer
also worked with the foreman of the production line when a new manufacturing system had
to be installed to manufacture various new products. He learned a lot from these experi-
enced people. None of them had gone to colleges for formal engineering education, but
they were smart people who had acquired technical skills in the jobs they had over their
careers. The young engineer learned a great deal about the “real world” from these
craftsmen that he would not have learned in classrooms.
The fantastic thing about the chief engineer was that he did everything without any
drawings or written plans! Everything came out of his head! He also conveyed his ideas
verbally, and his design was done intuitively. He did not verbalize his thoughts in terms of
FRs. The chief engineer made things similar to what had been made before and, therefore,
did not worry about FRs. He mostly replicated DPs he saw earlier. He built a small steam
locomotive at home from photographs of a locomotive had taken! The young engineer was
impressed to see the pictures of the train he constructed without any drawings!
The primary job of the young “engineer” was to go around and measure the dimensions of
machines, parts, and molds retroactively that the chief engineer had already built. He then
drew the part on paper, mostly for record-keeping. It was not an easy task, because,
although the nominal dimensions could be measured, it was impossible to discern the
tolerances of those nominal dimensions. Furthermore, it was impossible to state FRs based
on DPs.1
Then, things got worse. One day, the “chief engineer” bought an old press frame made of
solid cast iron in the form of “yoke” to convert it into an automated punching press. His
rationale was that it was thick and, therefore, would not deform and damp out vibration
during the punching process. Then, he asked the young engineer to use the old machine
frame and design a stamping press! This idea of making a new punching press out of the
old cast iron part was a costly way of making a press without any FRs and DPs for anyone
to see or read because only the chief engineer knew what he had in mind! After a great deal
of discussion, the top management decided to abandon this project. This invaluable
experience taught this young engineer a lesson of his life! Experience is essential but has its
limitations.
A couple of months later, the young engineer got promoted (i.e., became independent of the
chief engineer). He ended up inventing and patenting a method of making laminated
foam/straight plastic cups and dishes for the company, which was a hugely successful
product sold worldwide.
Many years later, back at his university, he encountered a similar but entirely different
issue in the field of design. A famous design professor in his department articulated his
view on the design that reinforced what the “chief engineer” in the small company
practiced. He advocated that design is an “experientially based subject,” and thus, it can
only be learned through experience. Based on that view, he opposed offering graduate
courses in design. However, this engineer, now also a professor in the same department,
had learned in the industry that an “intuitive and clever” approach to design might lead to
failures that could have been avoided and certainly not the best way to teach design to
aspiring young students. Experience is useful and valuable, but the experience-based

1
Sometimes, we use the term “reverse engineering” when someone tries to reproduce a
well-designed product, typically that of a competitor by taking it apart and make a similar product
by coping its components. The reverse engineering can reproduce the physical shape, but it is
difficult to know the FRs of the original product. Some companies tried to catch up in technology
development through reverse engineering of their competitors.
212 N. P. Suh

intuitive design is not the most effective way of designing products, machines, and systems.
Such a design practice without a theoretical foundation would likely lead to wrong ideas
and inferior designs, wasting time and money. The haphazard way of doing engineering
design based purely on experience without a theoretical basis is highly risky, leading to
high expenses and frequent failures. We read about these failures often in newspapers.
The story of the young engineer in the small company was told to emphasize the importance
of starting the design with a clear definition of FRs, followed by DPs that satisfy the FRs.
There are still many designers who jump right into DPs before defining the FRs, costing his
employer both in time and money.

In this chapter, we will discuss many issues related to DPs. Consider the questions
listed below:

1. Given a set of FRs, how do we develop a right set of DPs?

2. In an ideal design, how many DPs should there be when there are n FRs to be
satisfied?
3. What happens if there are more DPs than FRs?
4. Can the number of DPs be less than the number of FRs?
5. In designing a large system, there may be many layers of decomposition. How
do we make sure that the system is correctly designed?
6. How do constraints affect the choice of DPs?

7.2 Criteria for Selection of Design Parameters

As articulated many times in previous chapters, the design process begins with the
identification of the problem in the customer domain that we need to solve through
design. From the problem identified, we establish a set of the highest level FRs that
the design must satisfy. FRs must be selected in a solution-neutral environment.
Then we search for DPs that can satisfy each FR of the highest level FR set without
violating the Independence Axiom. Choosing the highest level DPs is most
important because all subsequent design decisions are affected by the highest level
choices.
The selection of DPs depends very much on the knowledge of the designer(s).
The more the designer knows about the problem and FRs, it is the more likely that
the designer can come up with a good DP. There is no criterion for selecting the
best DP at the highest level of the design process unless a similar design had been
done before. However, there are some general criteria for not selecting particular
DPs. Some of those criteria are as follows:

1. violation of law;
2. hazardous to humans;
3. harmful to the environment;
4. morally and ethically unacceptable;
5. violation of social norms;
7 How Should We Select Design Parameters? 213

6. an affront to commonly accepted religious and social standards;

7. government regulations;
8. unacceptable cost of the final product;
9. extensive use of rare materials that are difficult to acquire, e.g., rare earth
elements.

The criteria for selecting acceptable DPs are as follows:

1. each DP selected must satisfy a corresponding FR;

2. DP must not violate the laws of nature;
3. DP must be compatible with the constitutive relationships of matter;
4. DP should not be expensive;
5. DP should be easily manufacture-able;
6. well-known DP is preferable;
7. when DP is a person, the person must be ethical, trustworthy, and knowledgeable;
8. when DP is an organization, it is preferred if the past track record is outstanding;
9. DP should be conceptually simple.

Case Study 7.1: Plastic Cup

A company that has been manufacturing paper cups decided to manufacture plastic
cups for the rapidly expanding market for hot coffee. The product has to satisfy the
following three FRs:

FR1 = Contain hot beverage such as coffee;

FR2 = Must be stiff enough at 90 C to be held by hand;
FR3 = Have low thermal conductivity to be held by bare hands with hot coffee
inside.
We may choose the following as the DPs that can satisfy the FRs:

DP1 = Container in the shape of a cup;

DP2 = Wall thickness of the cup;
DP3 = Foamed plastic.

The design matrix that relates the FRs and DPs is shown below:

DP1 DP2 DP3

FR1 X 0 0
FR2 0 X X
FR3 0 X X

The above design matrix indicates that the proposed design is a coupled design.
It violates the Independence Axiom. In this design, if we try to make the cup stiffer,
it also affects the heat transfer rate through the wall. This design is a coupled design,
214 N. P. Suh

i.e., the functions of providing stiffness and thermal insulation are coupled. That is,
if we increase the wall thickness to make the cup more rigid when held in hand, it
also decreases the heat transfer rate through the wall.
Is there any other way we can uncouple FR2 and FR3 of this design?
One possible way is to vary the density of the foamed plastic to satisfy FR3 and
the wall thickness to satisfy FR2. It may not be the best solution because the density
variation may also affect both the stiffness and the thermal conductivity.
Is there another way? Perhaps a more effective way is to laminate the foamed
plastic sheet with a straight unfoamed plastic sheet to control the stiffness, i.e., FR2.
This second solution was patented through the U.S. Patent Office. The company did
very well with these laminated products. The inventor got a small bonus for the
laminated “composite” product and also for the manufacturing processes and the
machine that automatically made the product. The president of the company
identified the need for such a product, i.e., thin-walled cup with high thermal
insulation.
Whichever DP the designer selects, it has to be executed to the lower level
detailed designs through decomposition until the design is completed. Even before
the design is fully implemented, the design may encounter insurmountable and
unanticipated obstacles in terms of satisfying the FRs within the constraints given
such as cost, government regulations, hazardous to human health, performance, and
others. Otherwise, one proceeds with the design to completion and then compares
the final design with the primary criteria in terms of initial customer expectation,
cost, reliability, and competitiveness. Sometimes, the project must be halted in the
middle of the project if one discovers that it has been pursuing the wrong DPs.
At each level of decomposition, the FRs and DPs at that level are the
children-level FRs and DPs of the higher level FRs and DPs that were chosen
earlier. When we select children-level FR and DP, the process of selecting rational
and sound FRs and DPs is repeated. When the project is relatively small in scale,
the designer can make many decisions without involving others. However, as the
project is large involving many people and many FRs, the project management
team must make sure that all design decisions are made well by selecting the right
sets of DPs for a given set of higher level FRs.
Ultimately, a large design project depends on the decision of many designers.
The decisions made by so many designers at all levels may not be acceptable
because they violate the Independence Axiom. In that case, there must be a “system
architect,” who monitors all the decisions made by constructing the design matrix at
the system level. When the coupling is introduced inadvertently, the design decision
has to be replaced by another that is acceptable.
In conducting a large project involving a sizable number of staff members and
engineers, it is good to hold group meetings at least once a week. Everyone working
on the project should share information to identify any wrong decisions made by
anyone in the project group. The following two case studies illustrate the process:
7 How Should We Select Design Parameters? 215

Case Study 7.2: CO2 Level in Atmosphere

Suppose the FR is: “Reduce the CO2 level in the atmosphere?” One or more DPs
could be chosen from among the following possibilities as well as many others:

1. electric cars in place of automobiles with internal combustion engines;

2. reduction of the number of automobiles with IC engines in cities;
3. replacement of coal-burning electric power plants with solar panels;
4. using human-powered automobiles;
5. elimination of the use of automobiles in cities;
6. storage of CO2 underwater;
7. conversion of CO2 into CH4 and useful compounds.

One or a combination of the above DPs may satisfy the FR of reducing CO2
concentration in the atmosphere. Depending on which one of the above potential
DPs is chosen, we may end up having different designs to achieve the same FR.
Once we choose a DP for a given FR, we will end up with an entirely different
design, because we will follow different decomposition processes depending on
which first-level DP solution we choose. As resources must be expended to
implement the design, we may end up having very different design embodiments,
depending on how much resources are available for the task. The first decision we
have to make as a designer is: “Which DP shall I choose?”.
The above example deals mostly with DPs in technology-related fields. How
about DPs in non-technology-related fields? For example, suppose the FR is:
“Improve high school education in science and mathematics.” The possible DPs
might include:

1. allocation of more time for instruction of science and mathematics;

2. better teachers of mathematics and science;
3. development of computer games that teach mathematics and science;
4. more homework related to mathematics;
5. summer “math” camp for primary school children;
6. interactive computer software for teaching mathematics and science.

Although these DPs are not physical, we can treat them like the
technology-related case to design the system. The design process is the same,
although the contents are different.
Exercise 7.1: The P–V Diagram
You are assigned to design a compressor with a cylinder and reciprocating piston con-
nected to a crank mechanism. In order to determine its performance, you wish to measure
the pressure–volume relationship of the CO2 gas that will be compressed in the cylinder.
One way of determining the P–V relationship is to attach a sensor to the cylinder to
monitor the pressure as a function of the piston position. Your design assignment is to
develop the sensor that can automatically generate the P–V trace as the machine operates.
What are your FRs? What are your corresponding DPs?
216 N. P. Suh

7.3 Ideal Design

The minimum requirement for an ideal design is that it is consistent with the
Independence Axiom and the Information Axiom. The Independence Axiom states
that FRs must be independent of each other. The information states that the design
must be simple to design and operate by minimizing the information content of the
design.
Question Raised at a Seminar at a Leading University:
A young professor who was working on Axiomatic Design (AD) was invited to give a
seminar on design at a leading university in upstate New York. He went there despite the
bad weather. It was windy and cold. After learning about the Independence Axiom, a
distinguished professor in the audience, raised his hand to ask the following question:
“Isn’t it a good idea to have a functional coupling. For example, if I turn on the electric
switch to turn the light on for illumination and at the same time, heat the room with the
thermal energy emitted by the light bulbs, isn’t that a good idea? What is wrong with that
design?” The lecturer, in turn, asked the questioner, “What would you do during the
summertime when the temperature is high? Teach in a room without light?”.
Exercise 7.2:
Consider a design that must satisfy three FRs, i.e., FR1, FR2, and FR3. To satisfy these
three FRs, how many DPs do we need?

Suppose we have a design that has three FRs and two DPs. We may express the
relationship between FRs and DPs as follows:

FR1 ¼ f1ðDP1; DP2Þ

FR2 ¼ f2ðDP1; DP2Þ ð7:1Þ
FR3 ¼ f3ðDP1; DP2Þ

In this case, it is easy to see that, all three FRs cannot be independent of each
other.
Now suppose that we have four DPs to satisfy three FRs as follows:

FR1 ¼ f1ðDP1; DP2; DP3; DP4Þ

FR2 ¼ f2ðDP1; DP2; DP3; DP4Þ ð7:2Þ
FR3 ¼ f3ðDP1; DP2; DP3; DP4Þ

In this case, we have a redundancy of having an extra DP for three FRs we need
to satisfy. This kind of design where the number of DPs is larger than FRs is called
the redundant design. (It is a theorem.)
In an ideal design, the number of DPs must be equal to the number of FRs. (It is
a theorem.)
7 How Should We Select Design Parameters? 217

The best design would be the following:

FR1 ¼ f1ðDP1Þ
FR2 ¼ f2ðDP2Þ ð7:3Þ
FR3 ¼ f3ðDP3Þ

In this case, each FR is controlled and satisfied by one DP. It is an uncoupled

design. It satisfies the Independence Axiom. The design matrix is diagonal as
shown below2:

DP1 DP2 DP3

FR1 A11 0 0
FR2 0 A22 0
FR3 0 0 A33

If we have three DPs for three FRs that have the following three special rela-
tionships exist, we can satisfy the Independence Axiom:

FR1 ¼ f1ðDP1Þ
FR2 ¼ f2ðDP1; DP2Þ ð7:4Þ
FR3 ¼ f3ðDP1; DP2; DP3Þ

In this special case, we must vary DPs in a particular order to satisfy the
Independence Axiom. In the design given by Eq. (7.4), we need to vary DP1 first to
satisfy FR1, followed by DP2 and DP3, in that order. Then, the independence of the
FRs can be satisfied. This kind of design is called a decoupled design.
It should be noted that the beauty of satisfying the Independence Axiom
becomes apparent by examining Eq. (7.4). Although the system represented by
Eq. (7.4) involves three inputs and three outputs, the system behaves as a one-input
and one-output system, making it easy to control and satisfy. A coupled system
with n outputs and n inputs is complicated to operate or control.

2
In termsP of indices notation, the relationship between FRi and DPj may be written as.
FRi ¼ Aij DPi ;
where i and j are 1, 2, or 3. FRi and DPj are vectors, and Aij is a second-order tensor, like the
stress tensor ij and the strain ij tensor in solid mechanics. In mathematics, we learned to transform
the second-order tensor to find the eigenvalues or characteristic values through coordinate
transformation. That is how we determine the maximum principle stress and the minimum
principle stress. That is, there are only normal stresses and no shear stresses along a certain
coordinate axes in a solid body under load. Unfortunately, we cannot do the coordinate
transformation in the case of design, because the resulting DP’s become something that do not
have any physical significance. In design, we choose DPs so as to make the design to “have
eigenvalues” through proper choice of DPs. That is, we cannot do mathematical transformation of
a design to search for DPs that will give a diagonal matrix (i.e., equivalent to eigenvalues) through
proper choice of DPs that will yield a diagonal or triangular matrix.
218 N. P. Suh

The following theorems are useful to remember:

Theorem on Rational Ideal Design:
In an ideal design, the number of FRs and DPs is the same.
Theorem on Coupled Design:
When a design cannot satisfy each one of the n FRs independently from each other, the
design is a coupled design.
Theorem on Decoupled Design:
When the design can satisfy the independence of FRs by varying DPs in a specific
sequence, the design is a decoupled design.
Theorem on Related FRs:
If FR1 is related to FR2 as FR1 = f(FR2), the design has only one independent FR.

Question 7.1:
Can we manufacture PVC parts faster with plastisol?

One way of manufacturing thin PVC plastic parts such as children’s boots and PVC-coated
tools is to use plastisol. Plastisol is a slurry of PVC (polyvinyl chloride) powder suspended
in a viscous liquid called polyol (an organic compound containing multiple hydroxyl
groups). Plastisol is a highly viscous liquid. When plastisol is poured into a mold and heated
in the mold, polyol diffuses into solid PVC particles at an elevated temperature. PVC swells
up and softens, finally fusing, forming a substantial flexible part in the shape of the mold.
Many PVC parts are made from plastisol. The final property of the solidified part is a
function of the ratio of the resin to the polyol. The productivity of this plastisol processing
is low because it takes a long time to raise the temperature of the slurry in the mold due to
the low thermal diffusivity of plastisol. Therefore, the problem with plastisol processing is
its slow processing speed.
A young engineer had a bright idea for curing plastisol parts quickly. His brilliant idea:
eliminating the need to heat polyol in the mold. The idea was to pre-heat about 75% of the
polyol separately to a temperature higher than the average processing temperature of
plastisol and then quickly mix it with the remainder of unheated PVC-rich plastisol, which
has the PVC resin powder and remaining 25% of the polyol that has not yet been heated.
Then the mixture would quickly reach the fusion temperature of plastisol due to the
physical mixing of hot polyol and plastisol, shortening the processing time. However, his
“great” idea for rapid plastisol processing was not accepted for commercialization.
Exercise 7.3:
What is wrong with this new idea? Is it a coupled design? Why is it a coupled design?

7.4 Generation and Selection of Design Parameters

Given an FR, we must choose an appropriate DP that does not violate the Inde-
pendence Axiom. Are there fundamental theories that can guide us in this process
of transforming FR into DP? Are there algorithms that can lead us to the right DP?
7 How Should We Select Design Parameters? 219

In the absence of a simple way of sorting all possible DPs and choosing the very
best DP for a given FR and constraints, we choose DPs based on our understanding
of natural laws, materials, and other databases, which are typically stored in
designer’s head, i.e., “knowledge of the designer matters!” As of now, there is no
ideal way of developing ideas that can lead to the selection of the best DP for a
given FR. However, when there are many FRs we have to satisfy at a given level of
the design hierarchy, it is possible to choose the best set of DPs among a given set
of acceptable DPs by constructing the design matrix to check for coupling. This
observation can be stated as theorems:
Theorem 1 on Selection of DPs for a given Set of FRs:
In system design, when there are n FRs at a given level of the design hierarchy, an
acceptable set of n DPs is those that yield either diagonal or triangular design matrix.
Theorem 2 on Selection of DPs for a given set of FRs:
In many cases, when a specific DP that satisfies a given FR cannot be identified, select a
conceptual DP that only satisfies the FR and decompose the FRs and DPs.

Theorem 1 above states that when human designers are given the task of
selecting DPs for a given FR, the designer should use the existing knowledge
accumulated in the database to create an uncoupled or decoupled design. However,
when such a DP cannot be readily identified, Theorem 2 given above states that we
should select a conceptual DP that satisfies the FR intending to decompose the FR
and the DP further until all the details of the design emerge.
Given the rapid advances made in AI, in the foreseeable future, we should be
able to create generalized search tools for finding the best DPs and PVs for a given
FR. However, it may be still best for human designers to decompose FRs and DPs
to limit the search space of AI.
Notwithstanding the limitations of the current knowledge and database on the
FR, DP, and P–V relationships, the following guidelines may aid in developing
ideas for DPs and selecting the most appropriate DPs:
For technology-related fields (in random order):

1. think of the implication of all basic concepts used in science and technology
such as conservation laws, Fick’s laws, Newton’s laws, electromagnetism,
thermodynamic laws, and others;
2. try to ignore the current technologies to avoid becoming mental slaves of
current technologies and the thinking that goes with them;
3. review or acquire the fundamental knowledge in subjects related to FR by
reviewing the basics of the related fields, e.g., physics, chemistry, and biology;
4. keep reviewing the original problem identified: what are the shortcomings;
5. the best place to start may be introductory textbooks!
6. attend and listen to lectures in fields that may be closely or remotely related to
the topic;
7. remember, good ideas are simple!
220 N. P. Suh

8. great ideas may be generated by similitude based on developments in other

fields;
9. establish the upper and lower bounds of the FR, i.e., the allowable variation of
the chosen FR;
10. do something else, such as walking on a treadmill, in order to divert the thought
process from time-to-time;
11. think out aloud by talking to other colleagues and friends about the problem. It
may force you to organize your thoughts;
12. think of analogies;
13. do not let others belittle your ideas;
14. remember the principle of similitude;
15. if no specific DP can be identified, just put down a conceptual DP, based on
which the next level FRs can be selected and subsequently, the corresponding
DPs;
16. think!
For non-technology fields:
Some of the ideas listed above for technical fields apply here as well. In addition, the
following may be useful:

1. understand the history of the organization, including past efforts made to deal
with the FR;
2. review the primary objectives of the organizational goals;
3. listen to people who have been working in the field or the organization;
4. understand the fundamental theories or laws or rules that govern the field or the
organization;
5. be able to state the goals of the organization succinctly;
6. gather existing opinions of the leaders of the organization or the field;
7. identify who is going to benefit and who is going get hurt if the current
organization is changed;
8. have a detailed understanding of the subject matter, à la Warren Buffett;
9. develop a competitive strategy for the operation of companies;
10. do not believe everything you hear;
11. think!

The designer should remember that it may take a while to come up with good DPs,
but it is better to take time and think to select appropriate DPs than rushing in with
half-baked ideas, ending up with a wrong set of DPs. Otherwise, one might not
develop a design that solves the original problem identified in the customer domain.
Role of Constraints in Choosing Design Parameters
In design, it is essential that the constraints imposed on design be stated. The cost of
finding out the existence of constraints after the design is completed can be sub-
stantial. Some of the constraints on design are generated in the customer domain.
Also, during the design process, some constraints are generated due to the design
7 How Should We Select Design Parameters? 221

decisions made during decomposition. One of the typical constraints is the cost of
the product.
There may be many constraints that eliminate specific DPs from consideration.
For example, in choosing DPs that can reduce the CO2 emission, suppose that the
following constraints are imposed:

1. the cost of reducing CO2 should be equal to economic gains generated by CO2
reduction in 10 years;
2. the measures undertaken to reduce CO2 in the atmosphere should not affect the
sulfur content of the atmosphere;
3. the measures used to reduce CO2 should not adversely affect the safety of
people;
4. the technology adopted to reduce anthropogenic CO2 should not increase the
average temperature of the ocean.

These constraints may eliminate some of the DPs from further consideration.
Exercise 7.4:
We are charged with developing national health policies. The highest level FR for
healthcare is stated as “Provide healthcare for all citizens.” Design the next level FRs and
DPs. List possible DPs for this FR for healthcare. Also, list Cs (constraints).
Exercise 7.5:
Driverless autonomous electric vehicles (DAEV) are being introduced in urban areas. If
the vehicle malfunctions, injuring pedestrians, and damaging properties, we will need to
develop insurance policies to pay for the loss of lives and properties. The FR may be stated
as: “Develop insurance policies for damages done by DAEV.” Design the insurance policy
in terms of FRs and DPs by decomposing the highest level FR and DP. Also list constraints.
Exercise 7.6:
The use of social networks (SN) has spread worldwide. One of the significant problems with
SN is “fake news” that some people spread for political and other illegitimate purposes.
Define FRs and DPs, as well as constraints that can eliminate the intentional misuse of SN.

7.5 Integration of Design Parameters in a System

The DPs generated through the design process discussed in the preceding sections
must be integrated into a system. The methodology used in the integration of DPs
into a system depends on what the DPs are, i.e., physical things, personnel, soft-
ware, organization, a combination of these “things.”
When DPs are organizational entities that depend on communications between
DPs, system integration is done through software and communications systems. In
the pre-modern communications era, co-workers were placed in close physical
proximity, but with the availability of modern communications tools, many
222 N. P. Suh

companies are depending on interconnections through the Internet, telephones, and

software systems. Even in the case of organizations, it is more efficient and, thus,
cheaper to operate even substantial entities without having to have everyone in
close geographic proximity. Many large companies operate globally and, therefore,
and it is not realistic to depend on close physical proximity. Many large companies
allow their employees to work at home, which reduces the need for physical space
at one location for everyone.
When the DPs are tangible things such as parts in automobiles, physical location,
and physical integration of DPs can be critical. It is better to have certain DPs close
together through physical integration if they have to interact in the system. In that
case, it is useful to create DP to DP map, using a DP versus DP matrix to identify
which DPs interact frequently and thus, need to be in close geometric proximity.

7.6 Case Studies

Case Study 7.3: Identification of DPs Based on Physics and Engineering

Science
The goal of this case study is to illustrate how DPs and PVs were conceived to
satisfy FRs based on physics.

a. Design Problem:
In Example 2.1, the design of the molding process for the manufacture of shoe soles was
discussed.
We will review the steps involved in conceiving the process variables (PVs) to create the
DPs that satisfy the three FRs in order to illustrate how the design was conceived. In
Chap. 2, the FRs, DPs, and PVs for the shoe sole were listed, which is repeated below:
The highest level FR may be decomposed as:

FR1 = Mold the entire shoe sole onto the upper;

FR2 = Make the sole flexible, i.e., pliable;
FR3 = Make the sole wear-resistant;
FR4 = Control the weight (i.e., density) of the sole.
The corresponding DPs in the physical domain may be chosen as:

DP1 = Injection molding of PVC against the upper mounted on last;

DP2 = Density of foamed PVC;
DP3 = Solid skin layer for wear resistance;
DP4 = Total weight of the PVC injected into the mold.
7 How Should We Select Design Parameters? 223

The PVs that enabled the achievement of DPs are given as:

PV1 = Injection of PVC with a chemical blowing agent in pre-determined space;

PV2 = Expansion of the mold volume by moving the bottom plate;
PV3 = Temperature of the mold that comes into contact with the bottom of the shoe
sole;
PV4 = Injection volume.
The company tried various approaches without success before this young engineer joined
the company. One of the methods tried in the past was to melt PVC (polyvinyl chloride)
resins with a chemical blowing agent (that decomposes at the decomposition temperature)
in the plasticating screw pump of the injection molding machine. The mixture of the molten
gas/polymer is then injected into the mold. The mold cavity was shaped like a shoe sole
enclosed on the top by the stitched upper mounted on a last. When the mixture is injected
into the mold, the plastic/gas mixture forms gas bubbles and expands as it flows into the
mold. This process creates a polymer/bubble structure that is inhomogeneous because the
gas bubbles at the flow front expand more than the plastic/gas mixture near the gate. This
poor density distribution, a result of the flow front of the molten plastic is subject to low
pressure, while the pressure at the gate is high. Therefore, the material distribution is
uneven throughout the mold. The density of the polymer/bubble mixture is highest near the
gate with small bubbles, while the plastic at the leading edge of the flow has the lowest
density. Thus, the material distribution is not uniform throughout the part. Furthermore, this
process does not create the wear-resistant skin layer at the bottom of the shoe sole.

b. Conceptualization of the Process Based on Basic Physics:

If we look at this problem, the physics tell us the following:

1. We must distribute the material evenly in order to have the right amount of
material in the right place throughout the mold.
2. To achieve (1), we cannot allow much expansion of plastic through foaming
while it flows into the mold.
3. To achieve (2), we need high pressure while the plastic is being injected into the
mold to minimize foaming.
4. To achieve (1) and (3), we need to raise the pressure by narrowing the flow
channel in the mold between the last (where the upper is mounted) and the
bottom mold plate.
5. (4) can be achieved if we make the bottom of the mold movable to make the
flow channel very narrow during the injection.
6. After the injection is completed, if we move back the bottom plate of the mold
outward, the plastic with the foaming agent (now gaseous state) will expand
with bubbles, filling the mold.
7. To make the wear-resistant skin layer at the bottom of the shoe sole, we keep the
bottom mold plate, which is in contact with the hot plastic cold. The plastic in
contact with the cold plate cannot foam due to its low temperature, thus creating
an unfoamed skin layer.
224 N. P. Suh

Case Study 7.4: Identification of DP Based on Tribo-Physics for Electric

Connector

a. Design Problem:

The use of electrical connectors is ubiquitous. They are used wherever an electrical
connection must be made. They consist of male and female parts, which can slide
against each other under normal pressure for good electrical contact with a contact
resistance of less than 20 milliohms. One type of commercial connector is shown in
Fig. 7.1. There are many different kinds of connectors made for many applications
ranging from printed circuit boards and simple circuit breakers. In order to have
acceptable electric contacts, high normal pressure is applied to the contact area
through tight tolerances and sometimes applying elastic “spring” loading. The
contact area is plated with gold to prevent oxidation of the contact area, which
increases electric contact resistance. Depending on applications, the current density
at the contact area can be high, which leads to heating of the connector.
After the extended use of these connectors, the gold plating gradually wears off,
exposing substrate metal. When the substrate metal wears off and form a loose wear
particle, it can oxidize. These oxide particles may be lodged between two rigid
sliding surfaces, separating the two electrodes. Under further sliding, they may
cause abrasive wear creating more particles and disrupting the electrical contact.
The question here is what are the DPs and how we create them.

Fig. 7.1 Triple row spring-pin connectors and mating target connectors. (Reproduced with
permission from Mill-Max Mfg Corp. 2020)
7 How Should We Select Design Parameters? 225

b. Design Solution:

To prevent the wear and disruption of electrical contacts between the pin and the
receptor, we need to satisfy the following FRs:

FR1 = Provide good electrical contacts between the pin and the receptor;
FR2 = Allow sliding action between the electrodes.

How shall we now determine two DPs for these two FRs?
AD does not provide domain-specific expert knowledge needed to determine
DPs, other than providing the general framework, i.e., the fact that we need two
DPs to come up with an ideal design.
Unfortunately, there is no magic in determining DPs. One has to know the basics
of tribology3; in this case, mostly tribo-physics. One can read introductory reference
books on tribology or ask someone who knows the basics of tribology. It is not too
difficult to learn the subject and come up with right solutions to this problem.
Before we present DPs for these FRs, the following story may shed light on the
challenges associated with changing pre-conceived notions.
The DPs for the two FRs for electric connectors may be stated as follows:

FR1 = Provide good electrical contacts between the pin and the receptor;
FR2 = Allow sliding action between the electrodes.

DP1 = Good contact areas for continuous flow of electric current;

DP2 = Sliding interface

FR1 and DP1 may be decomposed as follows:

FR11 = Prevent oxidation of the contact area;

FR12 = Large number of contact areas at the interface for maximum current
flow;
FR13 = Provide constant normal load;

FR21 = Modulate the contact forces for ease of sliding;

FR22 = Remove any entrapped wear particles from the sliding interface.

The corresponding DPs are chosen as follows:

DP11 = Gold plating of contact areas;

DP12 = Woven conductor structure;
DP13 = Spring force to limit the maximum load.

Tribology refers to the field of science and engineering of friction, wear, and lubrication.
3
226 N. P. Suh

DP21 = Tensioning fiber;

DP22 = Open space between fibers for wear particles to escape.

There can be many different designs that can accommodate the above three DPs.
Two of the physical embodiments that satisfy the above DPs is shown in Figs. 7.2
and 7.3.

Fig. 7.2 One embodiment of the design. Pin is gold-plated copper. The conducting fibers are
wrapped around the tensioning fiber

Fig. 7.3 The pin is made of gold-plated metal rods. The female part of the connector is made up
of a flexible, bristle-like material held together by a substrate, similar to a velvet-like fabric
structure. The fibers are plated with a highly conductive metal such as gold
7 How Should We Select Design Parameters? 227

The insertion force should be extremely low in both cases. The normal load
between the pin and the bristle during the insertion is controlled by either the
tensioning fiber (Fig. 7.2) or stiffness of the bristle (Fig. 7.3). Therefore, the wear of
the pins and the strips should be negligible, allowing an “infinite” number of
insertions.
Too Simple to Be True! (Stories About the Visit of a Professor of Practice,4 AD, and
Korean Alphabet).
One of the engineering professors with extensive industrial experience could not believe
that some of the technological innovations he heard about were done based on AD! He
paid a special visit to a colleague to discuss his misgivings about design axioms. His point
was that it is not the AD theory but rather the ingenious inventor who came up with his
various inventions mentioned in the literature. It was apparent that he was not prepared to
accept the Independence Axiom. His thinking might have been: “Can such a simple
statement as the Independence Axiom be the basis for all these creative designs?”.
In some ways, this story is not so unique when so many theories, including Newton’s, had
so many skeptics. About 600 years ago, King Sejong, the third king of the Lee Dynasty of
Korea, invented the Korean alphabet, which consists of 24 letters. Most people can read
and write Korean words within a day. The Korean alphabet, HanGeul, is so simple that one
can learn to write them in a few hours. That was the reason Korean scholars refused to use
it, writing poems, poetry, and prose in Chinese for more than 500 years after King Sejong
invented HanGeul.
Until about 100 years ago, educated Korean people would not write poems and books in
Korean, because the Korean alphabet was so easy to learn that scholars could not dis-
tinguish themselves from the peasant, (so they might have thought). Korean scholars spent
their lifetime memorizing Chinese characters, which was used as a measure of scholarly
achievements. During the Lee dynasty, government civil service examinations consisted of
composing poems and poetry in Chinese rather than in Korean! The one who could
compose a better poem in Chinese became government officials. Even Korean family books
were written in Chinese! The tragedy is that now most Koreans cannot read their family
book, because they are written in Chinese characters.
We may have a similar problem in the field of design. Some still say: “How can it be so
simple when people spent decades to learn to design by acquiring years of experience.” It
is time to change because the old way of designing is costly and not reliable.
Exercise 7.7:
Construct the design matrix and write down the design equation, i.e., FRs as a function of
DPs for the connectors shown in Fig. 7.3.

Case Study 7.5: Identification of DP in Organizational Design

a. Design Problem:

In designing organizations, we have to decide the FRs of an organization first

before we do anything else. Since there are many people involved in large orga-
nizations, it is essential that one develop a consensus on the selection of FRs, if

4
This title of Professor of Practice is typically given to people who made important contributions in
industry and joined universities to teach students based on their industrial experience.
228 N. P. Suh

possible. FRs are the goals of the organization, and DPs are simply the organiza-
tional units to fulfill the FRs. In some organizations, FRs are stated under the
heading of a “mission statement.” However, in many organizations, FRs are not
explicitly stated and go directly into selecting DPs, which are often named as
“department” or “divisions.” There can be more than one DP for a given FR if the
task is immense.
The design of an organization is only as good as the people who are in charge of
various FRs. When a new person inherits an existing organization, it is necessary to
work with the people who have been in the organization. It is more effective to
work with these people rather than replacing them with newly hired people unless it
is absolutely necessary.
An intricate part of organizational design is the selection of “people in charge” of
a given task. One needs to talk to many people to identify the best people already in
the organization. However, one should remember that the performance of a person
depends on the clarity of the mission given by FRs of the organization, the freedom
to execute the task within broad guidelines, and budgetary support.
In many organizations, the interaction with the outside world is essential,
because their support can be critical in securing the financial support, in coalescing
the support of inside people and implementing the internal programs. In a sense, no
organization exists in isolation. It is a part of a continuum of society or a larger
organization.
In selecting people who can run different parts of the organization (i.e., PVs), it
takes a finite time to assess the performance of a person in a given position.
Initially, one has to wait and see what the person does, but it can take a year or more
to be able to assess the performance. Some people may occupy a position not doing
much, only looking for the next promotion or position. By the time one realizes the
situation, one’s tenure in that position may be up!
In many organizations, the new leader just appointed to the position must work
with available financial resources. The budget was already set 2 or so years
beforehand because the budgetary process in public institutions goes through a
lengthy decision-making process. Typically, it may take up to 2 years to get the
new budget through the system.
In the private sector, things can move faster if the right people back the effort.
Often the private sector operates based on personal trust and past performance
record. If someone at the top of an organization selects a person for a “fast track”
career path, one can become an executive of a large company in a relatively short
time. It is partly “luck” for this to happen. An equally able person may never get the
promotion if a person is at the wrong place, at the wrong time, working for the
wrong people. In this situation, it may be best for the person to leave the organi-
zation and seek new opportunities. However, one has to remember that there is no
guarantee of success.
7 How Should We Select Design Parameters? 229

Note on Importance of the Reputation of a Person

A new high-tech company established based on a technology developed at a university was
struggling. Things were not going well. The young people he hired to run the company and
commercialize the invention were using up the initial fund raised from the friends of the
founder. They could not raise new money to bring technology to a commercial-stage and
find customers. In desperation, the founder of the company asked his acquaintance, who
was the CEO of a successful high-tech company, if he would be able to help this struggling
company by becoming a member of the board of directors. His answer was immediate:
“No.” He said that he was too busy with so many activities. The answer was not
unexpected.
Surprisingly, the following day, early in the morning, the founder of the company received
a telephone call from the CEO, who turned down the request for help less than 24 h ago.
He said that he thought about the proposal overnight and changed his mind. He said that
he would become a member of the board if he can also invest money into this struggling
company. The founder was more than grateful! He then asked the CEO to be the chairman
of the board of the company and oversee the management of the company.
Once he assumed the chairmanship, several things happened. Many people wanted to
invest money in this struggling company! He could also recruit an experienced business
person to be the CEO of the company. Under his leadership, the technology was com-
mercialized successfully in the market place.
The lesson learned is that the impeccable reputation of the industry leader was the key to
his success. He was extremely ethical, creative in his thinking, willing to take risks, and
dedicated to the public good. He treats people fairly and equally, including the same size
cubicle for office for everyone, including the CEO himself, the same class airplane ticket,
and the like. He is an exceptional person. Fortunately, people like him make the world a
better place for everyone.

b. Design Solution:

Again, the design of organizations follows the same sequence as the design of
technologies, i.e., the Problem–FR–DP–PV transformation. First, one has to
determine the problem the organization has vis-à-vis its mission and goals. Then the
FRs must be established, followed by DPs and PVs. In this process, one must
realize that there will be criticisms coming from all directions since any change will
hurt those who have been the beneficiaries of the old system. Such a controversy
should be avoided, but it may be difficult to avoid them if one wants to solve the
problem identified through the design of a new organization. However, the com-
pensating feature of such an effort is the advancements made by these transformed
organizations. Sometimes, one has to be patient and nurture the newly changed
organizations.
Case Study 7.6: Identification of DP Based on Material and Mechanics

a. Design Problem:

In materials science, it is assumed that one cannot have both high strength and high
toughness at the same time. That is, if one chooses a DP to make the metal stronger,
then one has to accept the decrease in the toughness (i.e., the energy the material
230 N. P. Suh

can absorb before fracturing) of the material. Therefore, it has been a challenge in
the materials field, to be able to gain in both toughness and strength at the same
time.
One way of increasing strength is through dispersion hardening. It is a simple
idea. Metals deform plastically because crystal defects called “dislocations” move
along the crystal plane when stress is applied. One way of making the metal
stronger is to block the motion of these dislocations or make it harder to slide on the
slip plane, which can be done by various metallurgical means. In the case of steel,
we add carbon in iron to form iron carbides “particles,” which are harder than the
iron matrix. When these particles are present, the dislocations cannot go through
them and must be “extruded” between the hard particles, which require higher
stress. However, these particles must be microscopic, around 100 nm, to increase
the strength. When these particles are large, they fracture and crack, lowering the
toughness. In the case of copper, it is done by diffusing oxygen slowly into the
copper matrix that has a small amount of aluminum dissolved in it. The oxygen
then oxidizes aluminum, forming oxide particles in the copper matrix. The problem
with this oxidation process is that it is slow and expensive.
Hard and tough copper alloy is needed in many applications. One of the
applications is in making electrodes for spot welding guns used in the automotive
industry. In this application, a copper rod of about 1–2 cm in length and diameter is
forged into a conical electrode shape. Two of these “welding tips” are attached to
robot arms. They are then pressed against each other through the two steel sheets to
be welded. When a high electric current is sent through the electrodes, the tem-
perature of the compressed junction of the steel sheets increases to the melting
temperature, forming spot-welds which join the sheets. Through this welding
process, we assemble car bodies. If pure copper is used, the tips deform right away
because they do not have the strength to withstand the compressive load.

b. Design Solution:

The new design solution of getting both the toughness and high strength in copper
was proposed. It consists of the following steps:
1. Prepare two copper solutions in two separate crucibles. In one of the crucibles,
dissolve about 5% of titanium (Ti) in molten copper and the other dissolve boron(B)
in molten copper. When these two hot streams of copper are quickly mixed, they
form tiny titanium boride (TiB2) particles in molten copper to lower the free energy
of the two solutions. Then, if the molten copper is quickly solidified, the particles
do not have time to grow, forming sub-micron-scale hard particles in the copper
solution. Such dispersion-strengthened copper is both strong and tough. The matrix
phase must be pure copper without any solutes to maintain the high thermal and
electrical conductivity.
7 How Should We Select Design Parameters? 231

The FRs of this dispersion-strengthened copper are the following:

FR1 = Increase the strength of copper;

FR2 = Increase the toughness of copper;
FR3 = Maintain the thermal conductivity of copper.

The DPs are the following:

DP1 = Number of titanium boride particles;

DP2 = Diameter of titanium boride particle;
DP3 = Purity of the copper matrix.

The manufacturing process for making this dispersion-strengthened copper was

also designed and developed from scratch. Three brand new Ph.D.s built a factory
to manufacture the alloy. In 3 years or so, the alloy was manufactured using newly
developed manufacturing processes and sold to car manufacturers. This alloy is
given the name Mixalloy.
The remarkable feat of this project was that the manufacturing process could not
be tried out in small-scale tests in the laboratory because of the scale of molten
copper required to reach a steady-state flow. Therefore, a full-scale production
machine was built from scratch without the benefit of small-scale experiments. The
electrodes produced by this Mixalloy process, using the production machine built
from scratch, were shipped to automotive companies in Detroit, Japan, and Korea in
3 years from the date the company was established. This actual case shows the
power of AD and is a testimony for hard-working able and smart engineers and
technicians.
Exercise 7.8:
Construct the design matrix for the proposed design of the Mixalloy.

Case Study 7.7: Identification of FRs and DPs Based on Fluid Mechanics.5

a. Design Problem:

Automotive parts such as bumpers and fenders are made of a polymer called
polyurethane because of its lightweight, resistance to damage, and low manufac-
turing cost. Polyurethane parts are made by mixing two chemical components,
polyol (viscous resin) and the other is diisocyanate. Trial-and-error processes
developed the industrial process of mixing these two chemicals. The problem is
how to design the impingement mixing process.
The obvious way of mixing these two liquids is to stir them using a mechanical
impala. However, for these fast-reacting liquid systems, it is too slow to mix them
before they partially react. Another way of mixing them is by “impingement
mixing,” which collides two liquid components by shooting them against each other
5
This design problem was partially discussed in Sect. 2.1.1.
232 N. P. Suh

at high speeds in a small mixing chamber. This process, which is known as

Reaction Injection Molding (RIM), was developed in industry empirically and
widely used.

b. Design Solution:

The FRs can be stated as follows:

FR1 = Inject polyol and diisocyanate into a mixing chamber;

FR2 = Keep the ratio of the two-fluid components the same;
FR3 = Create turbulent mixing in the chamber;
FR4 = Make the mixing chamber to be free of the dead area;
FR5 = Discharge the mixture of polyol and diisocyanate into a mold;
FR6 = Control temperature of the monomers.

To achieve the above stated FRs, we have to pump the two fluid components at
the same rate at any instant. The mixing chamber should be small to be sure that all
the fluid in the chamber at any instant undergoes turbulent mixing, i.e., no dead
pocket of stagnant fluids. We have to make sure that the turbulent eddy size must be
of the order of magnitude to promote the instant reaction of the two-fluid com-
ponents. Thens the mixture must be discharged continuously. The mixing chamber
and the overall system are sketched (Fig. 7.4).
The DPs may be stated as follows:

DP1 = Injection nozzle;

DP2 = Positive displacement pumps in synchronous motion.

Fig. 7.4 A Typical RIM machine and the mixing chamber

7 How Should We Select Design Parameters? 233

DP3 = Turbulent eddy size;

DP4 = Small mixing chamber;
DP5 = Discharge port;
DP6 = Heat exchanger.

A brilliant student6 showed, through dimensional analysis, that the turbulent

eddy size, d, is proportional to the Reynolds’ number (Re) as follows:

d / ðReÞ3=4 ¼ ðqVD=lÞ3=4;

where V is the velocity of the fluid, q is the mass density of the fluid, µ is the
viscosity of the fluid. Since the polyol has a higher viscosity, its velocity must be
high enough in order to be sure that the turbulent eddy size is sufficiently small to
provide a complete chemical reaction of polyol and diisocyanate.
The design matrix for the proposed design is as follows:

DP1 DP2 DP3 DP4 DP5

FR1 X X 0 0 0
FR2 0 X 0 0 0
FR3 X X X 0 0
FR4 0 0 0 X 0
FR5 0 0 0 0 X

The design matrix indicates that the design is a decoupled design. We need to set
DP2 first and then DP1, followed by DP3, DP5, and DP4 in any order.
Although this turbulent mixing analysis was done for highly viscous fluids, this
technology has been applied to make Mixalloys with molten metals to create
nano-scale ceramic particles to produce dispersion-strengthened copper with tita-
nium di-boride as the dispersoids.

7.7 Use of Artificial Intelligence (AI) in Axiomatic Design

Suppose a design task has five FRs. Then, we know that we need to find five DPs.
The design matrix that relates FRs to DPs is as shown below:

6
Charles L. Tucker, III, who later became a professor and Associate Provost at the University of
Illinois.
234 N. P. Suh

DP1 DP2 DP3 DP4 DP5

FR1 A11 A12 A13 A14 A15
FR2 A21 A22 A23 A24 A25
FR3 A31 A32 A33 A34 A35
FR4 A41 A42 A43 A44 A45
FR5 A51 A52 A53 A54 A55

The design that satisfies the Independence Axiom is the one with a diagonal
matrix, i.e., all none diagonal elements in the design matrix such as A12, A34, etc.,
are zero and all diagonal elements such as A22, etc., are non-zero quantities. That
is, find a DP1 that does not affect any other FRs than FR1. Similarly, DP2 does not
affect all FRs but FR2. And so on. Sometimes we may be so fortunate that we can
readily choose the five DPs that will yield either a diagonal matrix or a triangular
matrix based on the knowledge we have in our brain. If that is not possible, we may
have to resort to the computational power of a computer. One of the techniques of
searching the database may be the use of AI techniques in order to search the
database and look for the right DPs to come up with the following design:

DP1 DP2 DP3 DP4 DP5

FR1 A11 0 0 0 0
FR2 0 A22 0 0 0
FR3 0 0 A33 0 0
FR4 0 0 0 A44 0
FR5 0 0 0 0 A55

The search routine may be stated as

a. search for DP1 that affects FR1 but does not affect FR2, FR3, FR4, and FR5;
b. if (a) is not possible, search for DP1 that affects only FR1 and then search for
DP2 that affects FR2 and FR1 but no other FRs;
c. then search for DP3 that affects FR3, FR1, and FR2 but no other FRS;
d. continue this process until a triangular matrix is obtained as follows:

DP1 DP2 DP3 DP4 DP5

FR1 A11 A12 A13 A14 A15
FR2 0 A22 A23 A24 A25
FR3 0 0 A33 A34 A35
FR4 0 0 0 A44 A45
FR5 0 0 0 0 A55
7 How Should We Select Design Parameters? 235

e. often, a neat triangular matrix as shown may not result, although DPs chosen
satisfy the conditions outlined in (a)–(d), requiring the re-arrangement of FRs
and DPs until a triangular matrix is obtained.

In some design tasks, there may be a large number of FRs at the highest level. It
may also be a result of many layers of decomposition of FRs and DPs. As the
number of layers of decomposition increases and as the number of FRs increases,
the task of coming up with a set of DPs that can assure either a diagonal design
matrix of an uncoupled system design or a triangular design matrix for a decoupled
design may become an onerous task and sometimes perhaps beyond human
capability without the help of computational and information storage capability of
computers. We need to use the capability of large computers that can deal with
extensive database and computational issues in comparing various combinations of
FRs and DPs.
One way of searching for DPs is to use the computational power to try many
combinations and seek a set of DPs for a given set of FRs that yield acceptable
designs. Some people have done this using AI, which has created a new business
boom in AI-based technologies. This new approach to AI—rather than the old
paradigm of “if…, then…” logic—is based on neural network and by using fast
learning algorithms.7 This new AI approach has become feasible, in part, because of
the computational power of computers that can accumulate a vast database and
compute at a rapid computational speed. There are many industrial firms, IBM,
Amazon, Google, and others, that have developed computational machines for
dealing with a vast database and means of extracting the desired information. Such
a technique is used in self-driving cars, drug discoveries, stock market analysis, and
many others. In many of these fields, the highest-level FRs are often known a priori
under a prescribed set of conditions and the task is to come up with lower-level FRs
and DPs, using the available database, the computational power of computers, and
inputs from sensors. Often the challenge is to create an extensive database from the
available information and manage it in a cost-effective and timely manner. Such a
machine that can generate is named the Thinking Design Machine (TDM).

7.7.1 Thinking Design Machine (TDM)

A TDM is defined as an intelligent machine that can generate creative designs. The
concept for TDM is based on the Independence Axiom and the Information Axiom.
As presented in preceding chapters, the first thing we need to do in design is to
identify the PROBLEM that we need to solve or address. Based on this problem
definition in the customer domain, we identify the FRs the designed system should
satisfy. Once the highest level FRs are defined, we find a set of DPs that can satisfy

7
Recently, Yann LeCun, Geoffrey Hinton and Yoshuua Bengio received the 2018 Turing award
for their contributions to AI.
236 N. P. Suh

the FRs, satisfying the Independence Axiom, as discussed in preceding paragraphs

in (7.7). This search process can be taken over by the TDM, which can search the
database stored in TDM or any other machines, perhaps using the cloud computing
system. This task can be complicated if the number of FRs is large at any level of
decomposition hierarchy. If we cannot satisfy the highest-level FRs because the
selected DPs cannot be implemented for lack of specific details, we must decom-
pose the highest FRs and DPs further, which then creates the next level FR-DP
hierarchies. This decomposition process must continue until the DPs have been
decomposed to the lowest level that can be implemented. As the layer of FR–DP
decomposition hierarchy increases, the database we have to search for the right set
of DPs that will not create coupling at the highest level can be vast.
When the search is done without the benefit of AD, the search process for the
right design that satisfies the FRs can be quite complicated, because one has to
search for a right set without the benefit of the Independence Axiom and the
Information Axiom where the universe of the search space can be vast. One may
use AI techniques to facilitate this search process. Now that the computational
power of computers is immense, and the cost of computation has significantly come
down, it is possible to search for the right FR–DP sets in a well-defined universe
such as in the synthesis of drugs for various illness such as cancer, self-steering
automobiles with many ©sensors, pilot-less air transport planes, and many others.
The power of TDM is that it can deal with many design tasks more effectively
than purely AI-based machines since TDM starts with a well-defined FRs and deal
with clear FR–DP structure, and follow the logic dictated by the Independence
Axiom and the Information Axiom, following the clearly defined decomposition
process. TDM can also use a search engine when it searches the FR–DP database to
find the right DP for a given FR.

7.7.2 Physical Laws and Thinking Design Machine

In searching for the right DPs for a given FR, we should always make sure that our
decision is consistent with the laws of nature and proven theories. TDM must check
if the selection of a DP for a given FR satisfies or consistent with the laws of nature.
Fortunately, there are a limited number of natural laws we must consider. In
engineering, we often deal with Newton’s laws, thermodynamic laws, thermal and
mass diffusion laws, Ohmic laws, mass and charge conservation laws, thermal
radiation laws, and Boltzmann’s relationship.

7.8 Guidelines for Choosing Design Parameters

and Process Variables

Once we define the FRs to be satisfied, we have to select DPs. This is a challenging
task. There are some guidelines that can be helpful in identifying DPs.
7 How Should We Select Design Parameters? 237

Guideline #1
Out of all these potential DPs, which one should I pick as the DP?
Create the Design Matrix to identify and eliminate DPs that can potentially
cause coupling!
In searching for the right DPs for a given FR, we should always make sure that our
decision is consistent with the laws of nature and proven theories. TDM must check
if the selection of a DP for a given FR satisfies or consistent with the laws of nature.
Fortunately, there are a limited number of natural laws we must consider. In
engineering, we often deal with Newton’s laws, thermodynamic laws, thermal and
mass diffusion laws, Ohmic laws, mass and charge conservation laws, thermal
radiation laws, and Boltzmann’s relationship.
Guideline #2 Depend on Knowledge base.
Ultimately, the quality of design depends on the basic knowledge of the
designer.
An aspiring designer should learn the laws of Nature, i.e., physics, chemistry, mathe-
matics, biology, etc. If one is more interested in the non-technical aspect, one should
learn the common laws (that govern organizations, national laws, local laws, etc.).
In fields where the government regulations (environment, safety, health, etc.) are
relevant in decision-making, the designer should work with people who are familiar
with these regulatory issues. Sometimes, the regulations may appear to be irrational,
but one cannot avoid the reality (or fragility) of human society.
Guideline #3 Create a Database.
Database of important ideas should be maintained.
Data provides a quick path to the final decision. One should compile a database
by writing down useful ideas and systems.
Guideline #4 Past experience.
Experience—good or bad—can be valuable. It will be good to remember
them.
Many things people do are based on their experience. Good experience and the
context that led to an excellent experience should be remembered. However, one
should not become a mental slave of the experience.
Guideline #5 Design cannot violate constraints.
It may be good to remove as many constraints as possible to give more room
for creativity.
There are many different kinds of constraints. Some are created during the
design process because of the decisions made during the design. It is essential to
have a complete list of constraints during the design process.
Guideline #6 Do Not Choose DP that Violates Ethics.
The importance of personal ethics cannot be over-emphasized in any field,
especially the design field, which deals with the creation of new things that have
not been done in the past.
238 N. P. Suh

7.9 Conclusions

The quality of design depends on the selection of appropriate DPs, which is the
third step in the AD process after the problem is defined, and FRs are selected.
There are a few simple rules that govern the DP selection process. One of the
essential rules is that in an ideal design, the number of FRs and DPs is the same. If
the number of DPs is less than that of FRs, we always have a coupled design,
violating the Independence Axiom. When there are more DPs and FRs, we have a
redundant design.
We can create a database on DP for a given FR, but there are no general
governing principles that guide the designer to choose the best DPs. We have to
invoke the Information Axiom to select the best DPs for a given set of FRs.
We can use the computational power of computers in searching for the best of
the set of FRs and DPs, although nothing can replace human capability to select and
manage FRs and DPs.
Examples from diverse fields are given to illustrate how DPs should be chosen.
Problems
1. Students should choose three of the exercise problems given in the main text of
this chapter and work on them.
2. Engineers at the Technical Center of Nissan Motor Company of Japan published
a paper with the following abstract (Reference: “Technical Papers of the
Automobile Technology Association”, Vol. 50, No. 6, November 2019):
“The automotive industry continues to develop new power plant technologies
such as downsizing, rapid combustion for fuel efficiency. But engine excitation
force will be increased by those, it’s very important to improve the isolation of
engine mount. This paper proposes a new active torque rod by using AD which can
be a simple design methodology against conflicting requirements. Our technology
principle consists of big improvement of the isolation level by lowering the reso-
nance of torque rod and applying the damping force by inertial mass actuator. We
successfully achieved a good NVH level on L4 downsizing engine as same as V6
engine.”
If you were the designer of this engine mount, what would you choose as the
FRs of the engine mount. What would you pick as the DPs that satisfy the FRs?
7 How Should We Select Design Parameters? 239

Appendix 7.1 Algorithm for Changing the Order of {FRs}

and {DPs}8

The uncoupled design is obvious, because each DP affects by only one FR,
regardless of how many FRs and DPs we have in the design. However, when we
have many FRs and DPs, and some DPs affect more than one FRs, we need to
re-arrange the order of FRs and DPs to determine if the design matrix is a triangular
matrix and thus, satisfy the Independence Axiom. Then, it is helpful to have an
algorithm for re-arranging the order of FRs and DPs to determine if the design is a
decoupled design with a triangular matrix.
The algorithm is as follows:
Find the row which contains one non-zero element. Rearrange the order of {FRs}
and {DPs} by putting the row and the column which contains the non-zero element
first (i.e., if i-th row contains one non-zero element at j-th column, then put the i-th
component of {FR} first and put j-th column of {DP} first. For example, the design
matrix for a new design is given below:

DP1 DP2 DP3 DP4

FR1 A11 A12 0 A14
FR2 0 A22 A32 0
FR3 0 0 A33 0
FR4 0 A442 A43 A44

The above matrix may be changed to

DP3 DP1 DP2 DP4

FR3 A33 0 0 0
FR1 0 A11 A12 A14
FR2 A23 0 A22 0
FR4 A43 0 A42 A44

Then, excluding the first row and column, find a row which has one non-zero
element, and put the row and the column second.

8
From Suh, N.P.: The Principles of Design. ©Oxford Publishing Limited. Reproduced
with the permission of Licensor through PLSclear.[p 383]
240 N. P. Suh

DP3 DP2 DP1 DP4

FR3 A33 0 0 0
FR2 A23 A22 0 0
FR1 0 A12 A11 A14
FR4 A43 A42 0 A44

Finally, excluding the first and second rows and columns, find a row which has
one non-zero element, and put the row and column third.

DP3 DP2 DP4 DP1

FR3 A33 0 0 0
FR2 A23 A22 0 0
FR4 A43 A42 A44 0
FR1 0 A12 A14 A11

This process can be done automatically by creating a computer algorithm.

References
Mill-Max Mfg Corp. (2020) Triple row spring-pin connectors and mating target connectors.
https://www.mill-max.com/products/new/triple-row-spring-pin-connectors-and-mating-target-
connectors
Suh NP (1990) The principles of design. Oxford University Press

Further Reading

Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press

How Should We Select Process
Variables? 8
Nam Pyo Suh

Abstract
Preceding chapters presented the transformation of customer needs or problems
identified in the customer domain into functional requirements (FRs), and then
FRs to design parameters (DPs). In this chapter, we consider the issues involved
in selecting the process variables (PVs) of the process domain to satisfy DPs.
When we are designing a hardware product, PVs are typically manufacturing
processes, human resources, and financial support. In organizational design, PVs
are primarily human resources. PVs are equally crucial as FRs and DPs in
completing a design project; PVs constitute an element in the continuum going
from “need” (in the customer domain) to delivery of solutions via FRs and DPs.
In dealing with PVs, domain-specific knowledge is required, such as expert
knowledge on manufacturing processes and software systems to create complete
systems solutions. In organizational design, PVs are typically people who will
determine the achievement of the organizational goals. Ultimately, PVs
determine the productivity in the manufacturing and operations of organizations.
The steps involved in selecting PVs to satisfy DPs are similar to the
transformation of FRs to DPs. A great deal of imagination and creativity is
always the necessary and required elements in sound design. Sometimes, it pays
to think out of the “mental” box we sometimes create for ourselves! It is often
more productive if designers should start with a clean sheet of paper and create a
design rather than being constrained by what was done in the past. There is no
assurance that what was done in the past is appropriate to the current problem or
was a good design!

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 241

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_8
242 N. P. Suh

8.1 Introduction

Process variables (PVs) are the final embodiment of customer needs or solving the
problem identified in the customer domain. In the field of product development, the
selection of PVs determines manufacturing cost. In the case of organizational
design, PVs represent people and other resources of the organization. As the
number of FRs increases, the number of DPs and PVs increases as well, which may
result in increased complexity and higher cost of manufacturing depending on the
quality of design. PVs, the characteristic vectors in the process domain (i.e., the last
of the four domains), are the final outputs of the design process, and they determine
the manufacturing process, the cost and the quality of manufactured goods when the
design involves manufactured goods.
Steering Column of Automobiles: How Much Should It Cost?
Every car must have a steering column to steer the vehicle. A professor who was invited to
teach “Axiomatic Design” to engineers at an industrial firm that manufactures steering
columns for automobiles was surprised to learn that there were considerable differences in
cost, complexity, and weight among the steering columns. (Fig. 8.1) A manual steering
column that performs simple steering functions for small cars may cost around a hundred
dollars, but the steering column that performs more functions in larger vehicles may cost
an order of magnitude more. He also learned that the most expensive part of these products
is the component cost rather than the labor cost. Many of the steering columns that had
problems were coupled designs. These classes held in industrial firms are beneficial to
everyone, including the instructor, because they deal with real issues.
It should be noted that the production cost is a function of the production volume of the
product. If each one of the components of an automobile is manufactured one at a time,
rather than using mass production technologies, cars would cost an order of magnitude
more expensive than mass-produced products. The power of mass production in lowering
the cost of various manufactured products has led to the consolidation of the industry.

Fig. 8.1 Different steering columns of automobiles (a and b reproduced with permission from
Nikiforov (2019a, b); c reproduced with permission from Dmitrii (2019))
8 How Should We Select Process Variables? 243

Manufacturing is a relevant field for the economic well-being of a nation.

Industrial firms engaged in manufacturing strive to improve their productivity to
maintain their competitiveness, which depends on how well they deal with the
entire transformation process from the customer need to manufacture products.
Moreover, the quality of design determines how much they charge for their prod-
ucts. Many companies have developed and adopted specific techniques to be pro-
ficient in their manufacturing business, starting from their design operations. Some
of these advances made in the industry have influenced academic research as well
as teaching. Among the better-known techniques, some universities teach are the
“House of Quality” and “Six Sigma.” Some of these techniques came out of Toyota
Motor Company of Japan. Some companies subscribe to the idea that their products
should have coupled FRs so that other companies would not easily be able to
replicate the quality of their parts and compete with them. This philosophy, which
might have served them well in the past when their traditional business depended
on many decades of refinement, may not be the best approach to innovation of new
products or developing large, complicated systems that are made in small quanti-
ties. Successful companies in highly competitive businesses often owe their success
to innovation rather than the mass production of well-developed products. A better
approach is to satisfy the Independence Axiom to make it easy to manufacture as
well as producing products that perform better.
Replication of PVs versus Innovation
Many developing nations often replicate products that had been made in advanced
countries for many decades. That is, without going through the entire transformation
process starting from the Customer Domain to PVs, they merely learn to replicate DPs and
PVs and sell the product at a lower price. This replication of existing products appears to
be a relatively simple process.
Around 1972, an American professor was invited to visit a manufacturing plant in an Asian
country that had just decided to industrialize the nation by manufacturing “heavy”
products such as machines and automobiles. The new owner of the company, who just had
taken over a defunct factory, was to manufacture diesel engines under a license from a
German company. The professor was surprised to learn that the executives of this company
did not know much about manufacturing (they did not even know the basics of milling
machine)! However, in a matter of a year, they were manufacturing diesel engines! This
incidence demonstrated that the replication of an existing product is relatively easy if one
has an educated workforce and the will to learn and succeed! During the subsequent
20 years, the Asian country has become a prosperous industrialized country, manufac-
turing sophisticated industrial products for the global market! Such rapid progress was
possible because of their emphasis on education. Now their goal is to become innovators of
new technologies, which will be much more challenging than becoming proficient in
manufacturing existing products. To become innovators, they must execute the transition
from the customer domain to the functional domain, to the physical domain, and finally to
the process domain most creatively.

In Example 2.1 and Case Study 7.1, the transformation of DPs to PVs in
designing and manufacturing shoe soles with three DPs was described. This project
illustrated the transformation of customer needs, to FRs, to DPs, and finally,
manufacturing processes (PVs). The lesson learned in that example was that once
244 N. P. Suh

we define FRs based on customer needs, we need specific expert knowledge on the
manufacture of such products. We need to develop the right set of PVs to satisfy
DPs. Since then, this technology was further developed to manufacture automotive
parts, furniture, and many others.
In the design of organizations, the same mapping idea applies between the four
domains, although the specific nature of FRs, DPs, and PVs are different from the
case of product development. In organizational design, PVs represent the resources,
(e.g., people, funds) needed to satisfy DPs. In previous chapters, the design of a
government organization was presented to discuss the application of AD to
organizations.
“Ban” Axiomatic Design (AD)!?
There once was one of the largest consumer electronics companies in the world. They made
good innovative products. Many children from all over the world wanted to have one of their
favorite products, a portable radio. The company was, at one point, one of the leading tech-
nology innovators in the world. It is no longer. It was speculated that one of the leading causes
of their “demise” was the company’s insistence on sticking with their “coupled designs”!
The speculative reasoning given to explain what happened to the company is as follows:
Before semiconductors were invented, all electrical circuits were made of analog circuits,
which makes electric circuits with resistors, capacitors, and inductors. In general, it is
difficult to satisfy the Independence Axiom with analog circuits with individual resistors,
capacitors, and inductors, especially complicated large circuits! They are mostly coupled
designs. Digital circuits with semiconductors do not have this problem! This famous
consumer electronics company made most of its products with analog circuits. Even when
the entire world is switching to digital systems, this company insisted on analog circuits.
This company reasoned that since it is so challenging to develop analog circuits, other
companies would not be able to compete with them! So even with the emergence of
semiconductors, they refined and improved their products using analog circuits. However,
other companies, all relatively new, switched from analog circuits to the digital circuit,
which enabled them to create better products that effortlessly satisfy the independence of
FRs. Today, these new companies that adopted digital circuits have now become the
dominant consumer electronics companies worldwide.
In some cases, it is difficult to satisfy the Independence Axiom with an analog circuit.
Digital circuits give the flexibility to meet the independence of FRs.
The lesson to remember: What appears to be a good business strategy, if it violates a good
design practice, cannot win!

Exercise 8.1: Battery-Powered Electric Car

Conceptually, the design of battery-powered electric cars (i.e., EVs) is quite simple. It has
to deliver the power stored in a set of batteries to the wheel. The most challenging part of
EVs is the management of electric energy stored in a large number of cells of lithium
batteries to the electric motors that deliver power to the wheels. Unlike cars powered by
internal combustion (IC) engines, which have low torque at low speed, EVs do not need a
transmission gearbox. The torque and speed of electric motors are proportional to electric
power delivered, i.e., current at a constant voltage, to the wheel. Design the electrical
power management system for lithium battery-powered cars.
8 How Should We Select Process Variables? 245

8.2 Precision Engineering

Many mechanical and electronic parts require ultra-precision. These parts are
manufactured in a “clean room” with no particles in the atmosphere and an
isothermal environment. People who enter the room, including workers, must wear
clean garments to be sure that particles from the human body do not contaminate
the working environment. All the machines are maintained at the isothermal con-
ditions, and those machines that generate thermal energy due to electrical motors
and others must be cooled using cutting fluids and other means. Most of the
semiconductor fabrication factories (commonly called the “FAB”) and
ultra-precision machining facilities belong to this group of “precision manufac-
turing.” Some of the components used in military weapons are made in such a clean
environment.
Design Problem faced by those in LLNL of California:
For some applications related to defense and space, we need parts that are atomically
smooth and accurate. Can we manufacture such parts by machining?

Design Solution:
Precision engineering was born because of the extremely accurate (to atomic scale) parts
needed for U.S. defense applications (e.g., nuclear weapons), X-ray telescope, and laser
fusion research. The pioneering research and development for precision engineering were
initiated at the Lawrence Livermore National Laboratory (LLNL) of the United States,
which is known as the place where the creation of the hydrogen bomb was initiated. The
Large Optics Diamond Turning Machine (LODTM) developed at LLNL is the world’s most
accurate machine tool as of 2019. LODTM can machine metal work-pieces as large as
1.5 m (5 feet) in diameter and 46 cm (18 inches) in height to an accuracy of greater than

Fig. 8.2 Surface generated

by the Large Optics Diamond
Turning Machine (LODTM).
(Courtesy of Lawrence
Livermore National
Laboratory (Klingmann
2001))
246 N. P. Suh

30 nm. It has been used to make optical mirrors for observatory as well as special
weapons. One of the surfaces cut by that machine with a diamond-cutting tool is shown in
Fig. 8.2. To measure such surfaces, they also developed measurement techniques that
measure optical errors to the atomic level while machining.
In cutting atomic-scale precision surfaces and contours, the machine, the work-piece,
cutting tools, and cutting fluids must be maintained at absolutely isothermal conditions.
Also, the feedback of the surface contour is essential as the surface is cut. In this machine,
the contours are monitored continuously, as well as the tool positions during cutting.
Although this technology was developed after decades of research in optics and precision
engineering without using the AD theory, this technology can be understood by applying
the reasoning process of AD.
The problem identified in the customer domain is the manufacture of atomic-scale surfaces
and contours on various non-ferrous metal surfaces. The highest FR is the creation of
atomically smooth precision surfaces with the diamond-cutting tool (i.e., DP). This highest
FR may be decomposed as.

FR1 = Cut surface contours to atomic-scale precision;

FR2 = Monitor the contour of the surface being cut during the cutting operation;
FR3 = Maintain the temperature of the work-piece, cutting tool, machine, and cutting
fluids at a constant temperature throughout the manufacturing operation;
FR4 = Provide vibration-free machining of the work-piece;
FR5 = Isolate the building from external vibration.
The DPs may be stated as the following:

DP1 = Precision diamond cutting machine;

DP2 = Laser-based metrology equipment;
DP3 = Temperature controller device;
DP4 = Vibration free machine;
DP5 = Isolation mount for the building.
The PVs in the fourth domain may be stated as follows:

PV1 = Cutting with a diamond tool;

PV2 = Measurement of the surface generated with laser surface profilometer;
PV3 = Constant system temperature, including cutting fluid;
PV4 = Monitoring of machine vibration;
PV5 = Monitoring of building vibration.
It should be noted that in this kind of precision machining, the maintenance of temperature
at isothermal conditions is necessary. To achieve this goal, sometimes they pour the cutting
fluid over the work-piece to guarantee that it is at the same temperature as the rest of the
machining system.
We have to make sure that the FR-DP relationship is diagonal or triangular. Similarly, the
DP–PV relationship should be consistent with the Independence Axiom.1 This was done
because they have generated atomically smooth surfaces reliably and repeatedly.

1
They developed the idea of the error budget to delineates how much uncertainty or
non-repeatability can be tolerated at each step in the production process. Predictability and
repeatability were maximized in these machines with tolerances of fractions of a micrometer.
8 How Should We Select Process Variables? 247

Exercise 8.2:
Construct the design matrix for FR–DP and DP–PV relationship of the above precision
machining process.

8.3 Designing a Manufacturing Process

for a Difficult-to-Machine Material

Many parts are manufactured using conventional manufacturing processes. How-

ever, sometimes new manufacturing processes must be designed because of either
unique features or unusual materials that cannot be machined. The design of
manufacturing processes and machines follows the same reasoning process as all
other design tasks.
Design Problem:
A major materials company in the United States invented a new composite material
consisting of aluminum matrix loaded with silicon carbide (SiC) particles. Such a
composite material is expected to be tough, abrasion-resistant, and strong. How-
ever, this material is difficult to machine to the desired shape. When ordinary
cutting tools are used to cut this composite material, the tools wore out rapidly
because of the presence of SiC grits in the composite. Therefore, they chose to grind
as the means of removing materials from this composite. However, when they tried
to grind it using SiC grinding wheels, the aluminum matrix phase of the composite
filled up porous space between the SiC grits of the grinding wheel, stopping
abrasive action of the grinding wheel.
How should we process this composite material? This research project was given
to a university as a research project. The professor assigned an undergraduate
student to work on this project under the supervision of a graduate student.
Design Solution:
The function of removing materials from this composite may be divided into two
components: remove carbide particles in the composite with grinding wheel and
then removing the aluminum matrix phase electrochemically. This process requires
using an electrically conducting grinding wheel.
The FRs may formally be written as

FR1 = Remove SiC particles in the composite with an abrasive grinding wheel;
FR2 = Remove aluminum matrix phase electrochemically.

Then, the DPs may be written as

DP1 = Carbide abrasive grits in the grinding wheel;

DP2 = Electrolysis.
248 N. P. Suh

The PVs may be stated as

PV1 = Abrasive action by the grits in the grinding wheel;

PV2 = Electric current flowing from the grinding wheel to the composite.

It was decided to make an electrically conducting grinding wheel by plating a

regular commercial SiC grinding wheel with copper. The copper plating was done
by flowing electro-less copper plating solution through the grinding wheel at room
temperature after the wheel is first catalyzed by passing through a catalytic solution
such as palladium chloride solution and hydrochloric acid.
During grinding, an electric potential was applied between the grinding wheel
and the workpiece to remove the aluminum phase of the composite electrolytically.
Abrasives of the grinding wheel removed silicon carbide particles in the composite.
It worked great!

8.4 Mass Production of Microcellular Plastics

Unlike the previous case of removing materials to manufacture discrete mechanical

parts by machining, polymeric components are typically processed by melting the
polymer for extrusion or injection molding. These mass production processes
reduce the cost of manufacturing discrete parts by orders of magnitude when the
production volume is significant that we can amortize the cost of tooling over a
large number of parts. These things are designed and manufactured using
well-established routine processes. However, some cases require innovation of new
manufacturing processes and materials. Such is the case with microcellular plastics,
which is now used to manufacture automotive parts, among others, worldwide
(Wong et al. 2016).
Design Problem:
In Sect. 3.5.5, the creation of microcellular plastics was presented. The idea was to
put in tiny bubbles (*10–30 microns in diameter) in large numbers (*billion
bubbles per cm3) into plastics to reduce material consumption, enhance the fracture
toughness of the plastic, and provide dimensional stability by reducing the residual
stress developed in the plastic part. The idea for and design of microcellular plastics
were done over a luncheon meeting, but the subsequent design of the manufacturing
process took more time.
Initially, to demonstrate the feasibility of introducing bubbles into solid plastics,
a technique similar to that used in making popcorns was used, which utilizes the
thermodynamic instability phenomenon. The plastic sheet was put into a
high-pressure chamber with CO2 to diffuse and dissolve CO2 in a solid polymer
matrix, creating a polymer/gas solution. At a suitable temperature, if the pressure is
suddenly lowered, the dissolved CO2 in the polymer matrix tries to form a new gas
phase by diffusing out of the polymer. However, since they do not have time to
8 How Should We Select Process Variables? 249

diffuse a long distance, they nucleate a large number of tiny bubbles, typically a
billion bubbles per cubic cm, i.e., 109/cm3. The bubbles are about 10 microns in
diameter.
This batch process demonstrated that microcellular plastic could be made, but
the batch process could not be used to manufacture commercial products since it
was too slow to be economical. What was needed was a continuous process of
making microcellular plastics at a rate that is suitable for industrial production.
Design Solution:
What is needed is the mass production of microcellular plastics. One of the most
commonly used processes for thermoplastics is the extrusion using a plasticating
screw extruder, which is shown in Fig. 8.3. It consists of a barrel and a screw inside
the barrel. Plastic pallets are fed into the extruder through a hopper at one end. The
plastic melts in the extruder due to the mechanical work done when it is continu-
ously deformed by the rotational motion of the screw inside the extruder barrel.
When the plastic is completely molten, it is extruded through a die to make plastic
sheets, tubes, or other profiles.
In extruding microcellular plastics, a conventional extruder such as the one in
Fig. 8.2 cannot be used, because to make the microcellular plastic, the following
four different functions must be performed independently from each other but in a
specific sequence, which conventional extruders cannot satisfy:

FR1 = Melt and mix the polymer to be foamed;

FR2 = Inject CO2 in a critical state of molten polymer;
FR3 = Dissolve CO2 in the molten polymer, forming a solution of polymer and
CO2;
FR4 = Extrude the solution of polymer and CO2 through a die under high
pressure;
FR5 = Cause sudden pressure drop to change the thermodynamic state of the
solution.

To achieve the FRs stated above, the extruder shown in Fig. 8.3 was modified.
The extruder was made longer, and CO2 at the super-critical state was pumped into
the plastic/barrel interface in small quantities continuously at the midsection of the
extruder where the polymer is completely molten. After CO2 dissolves in the
molten plastic, the solution of plastic/gas was extruded through a specially designed
die. Near the exit of the die, the pressure of the polymer/CO2 mixture is dropped
quickly, which initiates the nucleation of bubbles in large numbers. The nucleation
of a large number of bubbles occurs when the pressure is suddenly reduced because
the gas in the polymer/gas solution is no longer in thermodynamic equilibrium at
the new low pressure. This sudden thermodynamic instability of the polymer/gas
solution nucleates bubbles. The die was so designed that the pressure drop occurs
over a predetermined distance of the die lip. Then, the extrudate, i.e., plastic sheets,
has a plethora of nucleated gas bubbles (about a billion bubbles per cm3), which
250 N. P. Suh

Fig. 8.3 Plasticating extruder for polymers by (Extract from Sukoptfe:ptfe-machinery.com

SUKO PTFE Machinery (2017))

Fig. 8.4 Cross section of

extruded microcellular plastic
(Average cell size *20
microns) (Courtesy of Trexel,
Inc.)

grows until the polymer sheet is cooled. The density reduction could be controlled
from a few percent to 90%.
The DPs of this process are as follows:

DP1 = Plasticating section of the extruder;

DP2 = CO2 injection port;
DP3 = Mixing section of the extruder;
DP4 = Extrusion section of the extruder;
DP5 = Die lip design.

The PVs are the following:

PV1 = Rotation of the screw;

PV2 = Control the pumping rate of the critical CO2;
PV3 = Length of the extruder after injection of CO2;
8 How Should We Select Process Variables? 251

PV4 = Die design;

PV5 = Control of time rate of change of pressure in the die.

Cross section of microcellular plastic is shown in Fig. 8.4.

8.5 3D Printing: Layered Manufacturing2

3D Printing has become an essential industrial technology of quickly producing

parts from computer graphics into a solid part. There are many different approaches
to 3D Printing, but all of them are based on the idea that complicated 3D parts can
be manufactured by laying 2D layers of thin materials layer-by-layer. Initially, it
was called the “rapid prototyping” technology to meet the needs of the United
States defense industry, funded by the U.S. Defense Advanced Research Projects
Agency (DARPA), but it has expanded into many sectors of industrial production
technology. Some large companies in the United States have established a separate
business unit to commercialize the 3D Printing for mass production.
The highest FR of rapid prototyping is

FR = Produce 3D parts rapidly.

The corresponding DP is

DP = Layered manufacturing process.

The PV in the manufacturing domain is

PV = Layered manufacturing process.

Through zigzagging among three domains, (i.e., the functional domain to the
physical domain, and finally the process domain), we develop the lower level FRs,
DPs, and PVs as follows:

FR1 = Provide 2D information for each layer of a 3D body;

FR2 = Deposit a thin layer of materials;
FR3 = Bond the materials into a continuum;
FR4 = Remove unwanted materials;
FR5 = Build up the body layer by layer.

2
In this section, the materials presented earlier (Suh 2001) are more or less re-produced here.
252 N. P. Suh

The DPs are

DP1 = Digitalized 2D information;

DP2 = Thin layer of photoresist (resin);
DP3 = Light beam;
DP4 = Vacuum suction;
DP5 = Table motion.

The product design matrix, i.e., FR/DP matrix, is a diagonal matrix, which
indicates that design of the product is acceptable.
The PVs are

PV1 = Computer memory for matrix table of n x m pixels;

PV2 = Resin dispenser and horizontal table;
PV3 = On–off intense light beam that scans in two dimensions;
PV4 = Suction tube that scans in two dimensions;
PV5 = Step motor with ball screw for motion in the vertical direction.

The product and the process conceived by the above FRs, DPs, PVs constitute a
layered manufacturing process in which thin photoreactive resin (i.e., photoresist) is
deposited on a table that is scanned by a light source to cross-link the resin based on
the information supplied by the computer. A vacuum suction tube that also rasters
the resin surface after the reaction is completed removes the unreacted resin.
The process design matrix for the DP/PV relationship is

PV1 PV2 PV3 PV4 PV5

DP1 X 0 0 0 0
DP2 0 X 0 0 0
DP3 X 0 X 0 0
DP4 0 0 0 X 0
DP5 0 0 0 0 X

The process is a decoupled design. Each one of the first-level FRs, DPs, and PVs
must now be decomposed.

8.6 Process Variables in Organizational Design

In preceding chapters, we stated that in organizational design, PVs are either people
or financial resources. When many people are involved, we need to be concerned
about the issues that arise because people are the key elements of the design. For
example, the following could be important issues: the accuracy of transmission of
information among people, fidelity in the execution of the defined tasks, the ability
8 How Should We Select Process Variables? 253

of human operators to be consistent for an extended period, and assessment of the

effectiveness of people in a given job. We need to design the financial and other
reward systems for the people in the organization. To deal with these issues,
organizations have used compensation and promotion systems as a means of
encouraging people to perform the task of PVs as designed.
Perhaps the most challenging systems to design and operate are those that
involve humans as PVs. When people are the PVs, we cannot design the person to
fit PVs. We can only choose the best people among those available for a given PV.
Since human operators have their independent thoughts, different cultural norms,
ethical standards, and personal bias, decisions involving humans must be sensitive
to these variable factors when a human being is chosen as a PV. All of these
characteristics of human operators are strengths, but unlike machines, the
input/output relationship of human beings as PVs is less predictable. There can be
significant misunderstandings among people because the communication channels
among people are imperfect. The situation is worse if the language of communi-
cation involves diverse ethnic backgrounds and cultures. Unlike mechanical or
electronic components, when people are the PVs, consistency can be a problem;
people can change their views, values, and performance. The cultural norms that
prescribe how people should behave can help the operation of organizations, but it
may take away independent thinking that is the major strength of having people as
PVs.

8.7 Conclusions

PVs are a set of vectors that define the process domain. In organizations, PVs are
typically performed by humans to satisfy DPs. PVs can also be the financial
resources needed to enable the manufacture of products or run an organization. The
wise selection of PVs has a significant effect on the quality of products and the cost
of manufacturing. It also determines the efficiency of the system designed. In
manufacturing firms, PVs are manufacturing processes that determine the cost of
their products and their competitiveness.
Problems
1. “3D Printing” is finding many new applications in manufacturing solid parts
made of metals, polymers, ceramics, and composite of various materials. The
process consists of making 3D parts by depositing 2D layers to build up
three-dimensional shapes of all kinds. In “the old days,” engineers used to think
that solid metallic parts must be worked by forging or deformation to get high
strength parts. 3D Printing showed that a large number of parts could be
manufactured cheaply when the volume of the manufactured parts is small. The
mechanical properties of many 3D parts are acceptable for the intended
applications.
254 N. P. Suh

Design a manufacturing process for the human knee joints, assuming that you
can get the exact shape that has to be manufactured through X-ray scanning of
the human joint that has to be replaced. Define FRs, DPs, and PVs. Determine
the materials to be used and the sequence of the 3D printing process.
2. You are just appointed to be the new CEO of General Electric Company. Your
job is to restore the company to its old eminence. Define FRs. Design your new
organization. Clearly, state your DPs and PVs.
3. You just got elected to be the President of the United States. You have only four
years to turn the country around. Develop FRs, DPs, and PVs.
4. It appears that BREXIT is going to face many problems. If you just became the
Prime Minister of the United Kingdom, what would you do? State your FRs,
DPs, and PVs.
5. A country in the Middle East just decided to make all their public universities
“semi-private.” The government will provide the necessary funding, but each
university will have their board of trustees and operate as a private university. If
you have the responsibility of running one of these universities, what would you
do?
6. You are responsible for designing a software system for the management of
hospitals. Describe how you are going to develop the appropriate FRs, DPs, and
PVs. Develop the flow diagram to show how the FRs will be administered.

References
Dmitrii B (2019) New car steering column on white background. Photo licensed from
Shutterstock, ID: 672174733
Klingmann J (2001) The world’s most accurate lathe. Science and Technology Review, Lawrence
Livermore National Laboratory, p 13
Nikiforov V (2019a) 3d model of steering column and car suspension on white background. Photo
licensed from Shutterstock, ID: 1208610586
Nikiforov V (2019b) 3d model of the steering column on a black background. Drawing. Photo
licensed from Shutterstock, ID: 1208610007
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
SUKO PTFE Machinery (2017) Polymer screw extruder. https://ptfe-machinery.com/wp-content/
uploads/2017/02/Polymer-Screw-Extrusion.gif
Wong A, Guo H, Kumar V, Park CB, Suh NP (2016) Microcellular plastics. Wiley, pp 1–57.
https://onlinelibrary.wiley.com/doi/pdf/10.1002/0471440264.pst468.pub2
Mapping in Design
9
Miguel Cavique, António Gabriel-Santos, and António Mourão

Abstract
The preceding chapters presented the basic concept of Axiomatic Design (AD).
This chapter takes “MAPPING,” one of the important concepts of AD, to deeper
understanding. Mapping between the functional and physical domains, and
between the physical and process domains, accomplishes a number of important
design tasks. First, it enables the designer to decompose functional requirements
(FRs), design parameters (DPs), and process variables (PVs) to many branches
and levels, in order to advance design solutions while maintaining independence
and minimizing information content. Through proper mapping, designers can
track the development of large projects. As the number of FRs and DPs increases
with many levels and branches of decomposition, and as various participants do
mapping, we need to capture the results of the mapping process by creating a
master design and process matrices. This chapter also presents some additional
hints regarding the best way to define the FRs.

M. Cavique (&)
UNIDEMI & Escola Naval, Base Naval de Lisboa - Alfeite, 2810-001 Almada, Portugal
e-mail: [email protected]
A. Gabriel-Santos
A. Mourão
UNIDEMI & DEMI, NOVA SST, Campus de Caparica, 2829-516 Caparica, Portugal
e-mail: [email protected]
A. Mourão
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 255

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_9
256 M. Cavique et al.

9.1 Introduction

The four domains of the design—customer, functional, physical, and process—

have already been presented in Chapter 2. They are again shown in Fig. 9.1. The
arrows between the domains show the mapping process between the domains. This
chapter will focus on the mapping process, explaining the zigzagging process
indicated by the arrows between the domains. The design method in AD explicitly
asks what the customer wants and how to satisfy the customers’ needs. The method
treats the design as a continuum, from customer needs (CNs) to PVs.
Many engineering approaches to design erroneously jump directly to the
physical domain, based on physical concepts learned in universities or colleges,
while not clearly defining the goals or needs the system must achieve. Designers
often work from their own previous experience and knowledge, using creativity in a
non-structured environment. This working model has a reduced benefit to the
customer. Design solutions for systems or artifacts must compete based on price or
important minor features. Furthermore, such a design practice, jumping to physical
solutions, is more likely to lead to a coupled design that violates the Independence
Axiom. Unnecessary iterations for adjustment and unintended consequences lead to
long development times, continuing modification of the design, higher cost, and
poor performance.
To innovate, the FRs need to be formulated in a solution neutral environment
based on problems identified in the customer domain. This chapter provides an
in-depth understanding of the mapping process involved in defining the FRs.
We begin by discussing the use of optimization by experienced designers,
attempting to manage coupled designs, which fail to maintain independence, vio-
lating Suh’s first axiom. Subsequent sections show the role of the decomposition
process, i.e., zigzagging process, in maintaining independence and minimizing
information.

Fig. 9.1 Four Domains of the Design World

9 Mapping in Design 257

9.2 The Role of Optimization

A consequence of not satisfying the Independence Axiom is that when one of the
DPs is changed to adjust one of the FRs, other FRs also vary. Therefore, it is
difficult to satisfy each specific FR and all the FRs as specified and desired. When
such a design is created, many firms and engineers try to “optimize.” A typical
optimization technique creates an objective function that incorporates most of the
DPs and tries to find a set of values that will be as close to the desired performance
as possible. Often some FRs must be compromised by establishing relative
importance among the FRs. Such an approach involves the creation of an objective
function that incorporates the effect of each DP and finds the set of values that
maximize the value of an objective function rather than satisfying each FR within
an acceptable tolerance. Some call such an approach an “optimization algorithm.”
For example, in the world of finance, the yield curve is typically “convex,” which
changes as the interest rate changes. Then, they search for the peak value of the
objective function. One of the fundamental goals of AD is to avoid “optimization”
by eliminating the coupling of FRs and compromising the satisfaction of each FR.
In many engineering and management schools, optimization is taught as a
typical way of solving problems. Optimization implies sacrificing some FRs for the
best compromise operation of the system. Many engineers and managers attempt to
improve products and processes by optimizing what they have in the time and the
budget available for the project. They might improve the performance of a system
by five or ten percent in terms of cost, energy, weight, or any other selection
criterion through optimization. Regrettably, many times the system might be per-
forming at a fraction of what it could have achieved if each FR were satisfied
exactly as specified. The deficient performance is usually the result of choosing a
coupled design that does not satisfy axiom one.
Under the heading of “optimization,” which is so prevalent in the “real world,” they
attempt to improve features or cost, rather than satisfying all specified FRs. Therefore,
we explore three questions: “What made the optimization so common?” “When might
we use optimization?” and “What is the role of optimization in design?”.
The optimization in animal feed
George Dantzig developed the simplex method for optimization just after World War II,
inspired by the methods he used previously for the US Army. The method can find the best
solution in a convex region of hyper-planes studied in operation research (OR). In the
1960s, the simplex produced a revolution in the animal feed industry. The animal feed used
dozens of foods, including corn, wheat, oat, vegetables, forage, barley, fruit, and vitamins.
Animal food might have certain amounts of calories, vitamins, proteins, fat, the quantity of
food to eat, and many other requirements. Creating a mixture that can fulfill the animal
needs was a difficult task, which was only achieved by a laborious trial and error method.
In those days, it was virtually impossible to create the right mix at the minimum cost when
grain prices change constantly. The simplex method made it possible, not only providing a
solution but the best solution.
The question is how we should deal with this problem if we are to design the system based
on AD. First, we define FRs, which may include the following:
258 M. Cavique et al.

FR1 = Provide 3,000 cal per day;

FR2 = Supply the necessary Vitamin A, B, and C;
FR3 = Provide water;
FR4 = Provide 300 cal of protein.
Constraint or optimization criterion (OC, Thompson 2013): Minimum cost.
Then for DPs, we can select the following:

DP1 = grains with the highest calorie per Euro;

DP2 = Supplementary vitamins;
DP3 = Water supply;
DP4 = 100 g of soybeans.
The design matrix for this design is as follows:

DP1 DP2 DP3 DP4

FR1 X 0 0 x
FR2 0 X 0 x
FR3 0 0 X 0
FR4 x 0 0 X

In the above design, FR4 and DP4 create a coupled design. One easy solution, if it is
acceptable, is to eliminate FR4 and DP4. Then, it becomes an uncoupled design. Another
possible solution is to give a large tolerance on protein intake, then the relationship
between FR4 and DP1 becomes zero.

This last example shows two approaches to the same problem. Some animal
food requires a difficult balance between the amounts of products for the animal
feed. In such a case, the design is coupled, and the inclusion of an optimization
criterion f for cost allows for an optimization process to define the best solution.
However, if protein is not a requirement, or has a large tolerance, the design
problem is uncoupled.
Writing the design equation (DE) and defining the minimum number of FRs that
characterize the solution are keys to any design.
Exercise 9.1: Define FRs for Animal Feed
Create the DE for a horse feed using the products given in the example. Discuss the type of
design you get and the way to solve the quantity of food a horse needs to eat every day in
order not to feel hunger.

Many problems in engineering and management use analysis approaches. In

many cases, the problems have couplings, and universities tend to like to handle
these types of problems. After graduation, the engineer or manager wants to use the
same tools of analysis to address design problems. Many tools apply optimization
algorithms. These types of algorithms usually need a starting point, benefiting from
9 Mapping in Design 259

being a first attempt for the solution. If an optimization algorithm finds a better
solution than the previous attempts, then the user feels like it is a good solution.
However, the user just accepted a trade-off between requirements and got a poor
design.
The optimization approach forgets how difficult it will be to maintain the system
working in slightly different circumstances, how narrow the tolerances will be for
the DPs, and what will happen if tolerances change overtime. Optimization is a way
to find a trade-off point of operation, therefore achieving a balance between the
influences of different DPs on the desired FRs. In other words, optimization applies
to coupled designs.
According to AD theory, good design solutions are decoupled or uncoupled.
Usually, these ask for tuning the DPs so that the system ranges of all FRs are within
the design ranges. Tuning is the process of setting the values of the leaf, or low
level, DPs. Tuning applies to decoupled or uncoupled designs to define the DPs
within their tolerances. In audio equipment, you may want to enhance bass or
treble, or change the balance between the right and left speakers. You can use
independent knobs to set the volume, the pitch, and the balance. Tuning a design is
like using different knobs to set each of the DPs and fulfill the FRs.
A design achieves its target if the FRs are fulfilled, or in other words, the design
provides values of FRs in their acceptable ranges. An acceptable range for an FR in
a certain design might be different than the range of the same FR in another design.
Students should understand how difficult it can be to make a coupled design work
in acceptable ranges for all FRs, and how easy it would be when the design is
uncoupled.
Optimization and tuning are part of the design process, but not the design
solution itself. Therefore, designers first might find a good design solution and tune
it afterward. If the design solution is coupled, it could need to be optimized.
However, it is best to seek uncoupled or decoupled designs rather than optimize a
coupled design.
This discussion brings us back to explain in further detail the role of the FRs,
what an FR is, and how to express it. This consideration is important, as designers
are keen on defining the physical parts of the design, i.e., the DPs, but many times,
they do not explicitly express what they want to achieve, i.e., FRs.

9.3 Hints for Defining the Functional Requirements

FRs define good designs that solve problems articulated in the customer domain. In
other words, the design solution cannot be better than its FRs. This section provides
some hints on how to develop FRs in a solution neutral environment. Many
designers pass directly from “what they want” to “how they can” do it, based on
their experience or market standards in the area of the design. This way, designers
copy what already exists, i.e., the physical parts, supposing the DPs expresses
reality.
260 M. Cavique et al.

The allegory of the cave

The allegory of the cave is a metaphor of the Greek philosopher Plato who places Socrates
telling a story of a group of prisoners that lives in a cave. The prisoners see shadows
moving in the walls of the cave and suppose it is the reality. One of them leaves the cave
and learns that those shadows come from what passes in front of the cave, interfering with
the entrance of the sunlight. He got enthusiastic about the discovery and came back to the
cave to explain the new reality. Entering the cave, he cannot see because he had gotten
used to the outside sunlight. It made the others suppose that going out would make them
blind. The story ends by creating a group rule to kill anyone who wants to drag them out of
the cave.

Applying this interesting metaphor to the field of design, make us ask what the
reality in the design is. The reality is what we see in the physical parts or something
equivalent. Are physical parts the shadows of reality? We live in a world of
functions. We seek for functionalities, and the physical parts are the way to achieve
the functionalities.
Lawrence D. Miles introduced the idea of “functions” in design when he worked
at the General Electric Company. During World War II, GE felt shortages of certain
parts, forcing Miles to seek for replacement parts that performed the same func-
tions. The parts used were physically different from the originals but attained the
same requirements. It made Miles develop a new methodology of design, the
so-called “Value Analysis.” Sometimes, it was possible to use new parts that were
cheaper and could improve the performance of the system.
The following exercise is an example of what can happen if a design team copies
what they see, rather than understanding what the FRS are:
Exercise 9.2: Define the FRs of a Tram
A team of engineers from an electrical tram company is developing a new tram for a hilly
city. They decide to visit another city that had installed a new tram system recently. In that
city, they saw the tram equipment, and then submitted a technical report on the design.
They reported that each tram had two wagons, which was able to transport about 200
passengers. The report also described the number of trams in operation, the number of
persons transported per tram, average and maximum velocity of the tram, power and
voltage, accessibility, etc. The tram company agreed to create a similar concept for their
city, realizing immediately that the new tram restricts its use in the flat part of the city.
Furthermore, the new trams had excess capacity. Therefore, the company started to use
fewer trams and increased the time between operations, which made the passengers wait
longer to catch a tram. As a result, the customers used the new trams less and less,
reducing the number of trams in operation.
Question 9.1:
State what the customers want. Establish the main FR of this project that can address the
customers’ needs.

This section provides some hints to help the students define the FRs. We first go
into the ontology of what a function is in design. Assuming that students are
familiar with physical and economic functions that describe natural and sociologic
phenomena, the question proposed in this section is, “what is the difference between
9 Mapping in Design 261

a function of a natural phenomenon and a function in design”. The following is an

example of a physical function described by Pascal:
The Pascal experiment
Blaise Pascal allegedly made the barrel experiment in 1646 to prove that the pressure a
fluid exerts depends on the hydraulic head, not on the mass of the fluid. Opponents said the
larger the mass, the higher the pressure would be. To prove his ideas, Pascal inserted a
long thin tube into a barrel and started pouring water into the barrel. At a particular
moment, the barrel started bursting. Was Pascal concerned about the material of the tube,
or the number of wood staves? No! His concern was about the function: what made the
pressure increase? For a fluid, water in this experiment, pressure depends on the head.

Students can repeat the barrel experiment many times in different ways to verify
the validity of the Pascal equation. Experimenting intends to check the function, not
the physical parts of the experiment. Therefore, functions in nature are abstractions
of the physical domain.
Science claims to be objective because anyone who uses the correct apparatus
can verify the natural laws under observation. Objectivity in Design has a slightly
different meaning. Different projects can perform the same functions, so repeata-
bility in design occurs in the functional domain. In this sense, functional repeata-
bility in Design Science is not so different from the objectivity in science.
So, what is the main difference between Natural Sciences and Design Science?
In natural sciences, the function is checked by the experiment, or the experiment
is given, and the function is verified. In design, the functions are given, and the
physical concretization allows for checking if the DPs can fulfill the functions. In
natural science, a theory, or a function can be rejected if it fails the verification of an
experiment; in design science, one rejects the artifact if it fails to perform the
functions.
These opposite checking directions between natural science and design science
have an expression in the use of DEs. DEs relate the FRs to the DPs, the same way
a function relates independent variables to dependent variables. However, in a
function, the independent variables are known, and we want to obtain the value of
the function; in design, we know the function, the FRs, and want to discover what
the DPs are! First and Second axioms help to analyze solutions and can deny a
solution, but the application of the axioms does not lead to the solution.
The reader already saw in Chapter 5 that a FR is a verbalized action. Examples
of FRs are, regulate the water flow, maintain the indoor temperature, cut the beam,
or any other phrase with a verb that involves action. The FRs and from DPs submit
to the bounds defined by the constraints (Cs). The Cs are derived from CNs, laws,
and regulations, arise from physics, or are derived from parent FRs or DPs.
FRs are developed by considering “what we want to achieve?”. The “what?” or
“what do we want to achieve?” are how to develop the FRs. On the other hand,
“how to attain it” locates elements in the physical domain that fulfill FRs. There-
fore, the dialogue, or mapping, between the “what” (FRs) and the “how” (DPs) is
the basis of the design solution.
262 M. Cavique et al.

The linguistic structure of an FR is a verb and noun implying an action.

Therefore, “adjust temperature,” or “control airflow” are examples of FRs. More-
over, FRs need to have a target value and a range of acceptance, i.e., a tolerance that
can be used for computing the information content of the design. Instead of “adjust
the temperature,” an FR should be “adjust the temperature to 24 ± 1 °C.”
Mistakes of novices and ontological incongruences that can be encountered
when developing FRs are discussed below. The following three types of mistakes
are common from novice applicants:

– non-designs, “negative FRs”;

– mixing FRs with DPs;
– mixing FRs of the artifact with FRs of the design process.

Stakeholders sometimes warn designers that they do not want things that pre-
viously had a negative impact on the system. Novice AD users might respond to
these warnings with negative FRs. These are formally incorrect. Because the
complement of any element is everything else, not creating anything is a
non-design. Examples are: “should not be dangerous”; “must not have lead”; “must
not create legionella.” Design teams still must take these warnings into account in
formulating design solutions because they can express important needs of the
stakeholders.
Another common novice mistake is to define FRs by a DP the designer intends
to use. Instead of “adjust the temperature in a room,” the novice designer may
describe the FR as “use a fan-coil system.” In this design, the DP is the “fan-coil
system.” Anytime the FR has a syntax “to use + noun” or “to have + noun,” and
the DP is the noun, the FR is in a non-neutral environment, and should be avoided.
Another class of mistakes is to mix the FRs of an artifact with the FRs or Cs of
the design process. The designer knows that “deliver the design solution in two
months,” or the Cs “budget available,” are important constraints on developing a
new artifact. Therefore, FRs and Cs of the design process can apply just to the
process, and not to the artifact.
Ontological incongruences, inconsistencies in the nature of the AD process, are
more difficult to handle. AD users should know these incongruences, avoid them, if
possible, and make use of them, if necessary, to solve design problems. Incon-
gruences can be of the following types (Thompson 2013 ICAD):

– mixing FRs with Cs;

– the b-FRs;
– mix FRs with selection criteria (SC) or optimization criteria (OC).

Cs might sound like FRs, although Cs have no tolerance and need not be
independent of other Cs or FRs. Cs create boundaries for the design. Usual
examples of Cs are limits on cost, or weight, like “maximum cost,” or “maximum
weight.” Nevertheless, an FR can have a range of acceptance with a lower bound
but no upper limit, which is a common concept in engineering and management.
9 Mapping in Design 263

In the design of an air conditioning system, the CN “delivery of outdoor airflow

greater than…” is often used. However, no one cares about having a huge amount
of air, except for fulfilling another requirement regarding energy use. Similarly, in
the enterprise field, financial ratios have lower or upper bounds. A high-level
requirement for an enterprise design may be “maintaining a financial autonomy
greater than 0.5,” meaning the enterprise will have financial strength.
Additionally, it is possible to transform some common Cs into FRs, as long as
the FR has a target and a range of acceptance. In the example of the cost, changing
“maximum cost” to “control the budget of the artifact to a certain amount in a range
of” allows one to define cost as an FR and computing its information content.
The b-FRs usually express an emotion about the product with a semantic
“be + adjective” or similar. Examples are “be aesthetic,” “smell good,” or “look
like a.” Strictly speaking, the b-FRs are not FRs, as there is not a defined target
value and a range of acceptance. Moreover, “be something” or “like something” can
mix an FR with a DP. Anyway, AD experts should not forget that the b-FRs
describe decision criteria for many customers. The designer should then create
empirical or emotive scales for those FRs.
Selection criteria and optimization criteria are often used as FRs, a common use
being in the energy design processes. “Minimize energy use” may be an FR
expressed in different ways along the decomposition process. Ontologically the
minimum has no defined target, nor a range, so the information content is not
possible to compute. However, an expert in AD might realize that an FR relating to
energy use exists. This FR needs to be reworded and specified appropriately, not
eliminated.

9.4 The Decomposition Process

The decomposition process runs over all four domains of the design representation,
from the highest level CNs to the lowest level PVs. The decomposition works in all
the domains, at each level of decomposition, finding DPs and PVs to formulate the
best solution by following Suh’s two design axioms.
As explained in Chapter 2, CNs are the reverse side of the declaration of a
problem, so that CNs usually relate to high-level FRs. It is wise to check all along
the decomposition process to see if the FRs are consistent with the CNs. Many
times, there is no need to decompose the CNs to lower levels.
Figure 9.2 shows the domains and the zigzagging between adjacent domains.
The mapping between CNs and FRs is conceptual design. Product design is
mapping between FRs and DPs. Most applications of AD are conceptual and
product design. Production, or process, design maps between the physical domains
and the process domain, defining the technology and the processes to create the
DPs. At the end of this chapter, we introduce “concurrent engineering,” where
product and production designs occur at the same time.
264 M. Cavique et al.

Fig. 9.2 Zigzagging between domains

Figure 9.2 does not show Cs arriving from the CNs, can be legal, and ethical, nor Cs
from the FRs and DPs. A common approach for conceptual design is mapping CNs to
high-level elements of the functional domain. Concerning product design and pro-
duction design, Fig. 9.2 shows the zigzagging between FRs and DPs, and between
DPs and PVs, along with the various levels of decomposition. However, it may not be
necessary to define a PV for high-level DPs when the designer accepts an existing
technology and process to manufacture the DPs. In an enterprise of manufacturing
home appliances, if the DP is a “washing machine” the designer might not need to
describe the PV as being “processes that produce a washing machine.” Otherwise,
low-level DPs might need to be defined too, including, how to manufacture, install,
and dismantle, making it necessary to define the corresponding PVs.
However, the common approach described in the last paragraph may have
exceptions. CNs might need to be mapped onto FRs at low levels of decomposition,
and high-level PVs might need to be defined. The following example shows the
need to define low-level CNs that map onto FRs and DPs at a low level, and the
need to consider the high levels PVs:
Play and stop button
On audio recorders, the symbols for play, stop, record, forward, and backward have been
used since the sixties. They are low-level DPs concerning the manufacturing of an audio
recorder, but as everybody recognizes the symbols, they became a standard in videos,
television, and media. The use of symbols became a CN that needs to have a definition on
the leaves DPs.

A new fan-coil factory

A European air conditioning enterprise with good local reputation was facing in 2000
competition from the Far East. They wanted to make a new type of fan-coil and decided to
consider delocalizing the production. This decision would mean that about a hundred
production line workers would lose their jobs, which would cause heavy economic pain in
the region. The manager of the enterprise asked for two new equipment designs. One would
be manufactured in the Far East by the traditional methods, and another designed for a
robotized line. The cost of production plus investment of the second option was 10% higher
9 Mapping in Design 265

than the off-shore option, but the manager decided to invest in the robotized line. To off-set
the investment in the new production line, he decided to improve sales of those products.
This type of decision benefits from mapping between the four domains and includes an
ethics Cs for the manager’s decision.

According to Fig. 9.2, design processes are active dialogs between FRs and DPs,
the choice of one influencing the other. Zig is the mapping from the functional
domain to the physical domain, to find DPs best able to fulfill FRs. Zag operates in
the opposite direction coming back from the physical domain to the functional and
finding FRs at a lower level. Zigzagging between domains starts at the highest level
FRs and develops all the leaf level DPs to complete the solution.
The next example shows the strong interdependency between the FRs and the
DPs at higher levels of the abstraction hierarchy.
Thermal comfort design
You are designing a solution for FR0:“provide thermal comfort to your team”. Choosing
DP0 will affect all remaining FRs.

• The probable DP0 might be an “air conditioning system”. Thus, next FRs will be about
adjusting temperature and providing a certain amount of outdoor airflow for indoor air
quality purposes.
• However, if people are not constrained to work indoors they could lie outside and work
with a computer in a hammock, as long as the weather outside is nice.
• Another solution might be to take a “metabolism pill,” reducing the metabolic rate so
that no one will feel hot any longer.
• Another solution is to develop a new “thermal suit” that would be able to heat or cool
the body according to the metabolism and the indoor temperature.
Therefore, depending on the choice of DP0, child FRs will be different. Nevertheless,
medical Cs from FR0 forbid a “metabolism pill”, and depending on the context of working,
hammocks could be forbidden. Both situations are examples of Cs that arise from FR0. On
the other hand, choosing DP0, the “air conditioning system”, many regulations impose Cs
that need to be taken into account.

Design theories can be applied to help achieve a good solution in less time,
avoiding trial and error experiments. AD should provide the best solution, at the
first attempt, eliminating non-productive iterations. Students might suppose that
there are no iterations in the zig process when looking for a DP, or in the zag to
attain the FRs, although this is not always the case. AD can avoid iterations
between levels of decomposition, and it can avoid ending with solutions that need
to be checked again from the beginning. Iterations might happen at each zig and
each zag several times until the design team feels comfortable with their solutions.
Arrows in Fig. 9.2 are final expressions of design process components.
The following example clarifies what zigzags are while providing some rationale
for this design process. The example goes to the second level of decomposition,
showing the possible solutions and the reasons for selecting one solution over
another. In this example, the experience of the designer helps to choose the solution
that has the best probability of success. In seeking a design situation, students might
266 M. Cavique et al.

try to understand the probability of success by getting data from examples, talking
to experts in the area, making simple models of how the system works, and finally
calculating information contents.
Rotate a reel
A manufacturing company uses reels of metal sheet transported in the factory over roller
conveyers with the axis vertical. A crane picks up each reel at the end of a conveyor. It
needs the roll axes to be horizontal, making it necessary to turn the reel. Figure 9.3 shows
the CN that will solve the problem of the factory and, in a box, the new product to design.

The designer translates the CN into FR0 and then to DP0 according to.
FR0 = Rotate steel reel.
And the new product as
DP0 = Mechanical turning device.

Notice that the decision on DP0 is crucial for the development of the design. Another DP0
could create a completely different design, by choosing instead “manpower” or “elec-
tromagnetic field” as DP0. With this DP0 accepted as a good solution, the zig at the
top-level ends, and the process zags back to the functional domain at the next level of the
decomposition. Figure 9.4 shows the zag and the corresponding FRs that should be
according to DP1 and at the same time with FR1.
The figure expresses the FRs in a neutral environment on the second level of
decomposition:
FR1 = Receive the steel reel;
FR2 = Rotate the reel axis from vertical to horizontal;
FR3 = Keep the reel axis in a horizontal alignment.
Many DPs are possible for each FR. Figure 9.5 depicts a rationale for the zig, showing
possible solutions. Because the transportation system in the factory uses roller conveyors,
the receiver can be a roller conveyor (DP1-A). A belt conveyer (DP1-B) is also possible,

Fig. 9.3 The CN to turn the reel. (Reproduced from Gabriel-Santos et al. (2017), originally
published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201711209010)
9 Mapping in Design 267

Fig. 9.4 The zag from the first level of decomposition. (Reproduced from Gabriel-Santos et al.
(2017), originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/
matecconf/201711209010)

Fig. 9.5 The possible solutions at a zig for the “mechanical turning device”. (Reproduced from
Gabriel-Santos et al. (2017), originally published open access under a CC BY 4.0 license: https://
doi.org/10.1051/matecconf/201711209010)

although less suitable because the surface of the conveyer belt can tear during operation.
Similarly, conveyor rollers could get damaged during service, although repairing this
implies just changing one or two rolls instead of the entire conveyer belt.
DP2 can be a handling structure (DP2-A), or a cradle support (DP2-B), or any other
solution the student can try to imagine. The handling structure was the choice because it is
cheaper than the cradle, and any deformation does not affect the sheet in the reel. Finally,
to “keep the reel axis in a horizontal alignment,” the options are to use two rolls (DP3-A) or
two hinged convex metal supports (DP3-B), the former having the disadvantage of creasing
the metal reel.

According to the above rationale, the chosen DPs are

DP1 = Roller conveyor (DP1-A);
DP2 = Handling structure (DP2-A);
DP3 = Convex metal supports (DP3-B).
268 M. Cavique et al.

Fig. 9.6 The decomposition process with a solution (Adapted from Gabriel-Santos et al. 2017,
originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201711209010)

Figure 9.6 shows the decomposition process, integrating into the image of DP1 the “me-
chanical turning device.”

The previous example appeals to experience and good sense and shows how to
perform the zigzagging, the decisions needed, and the rationale for it. Moreover, it
shows on DP1 the integration of the child DPs.
Next sections formalize each path, the zig, and the zag in deeper detail, using a
structured flow chart.

9.5 The Zig

The zig is the mapping between two adjacent domains at the same level of detail,
usually expressed by an arrow from left to right. This section focuses on the zig
between the functional domain and the physical domain, but it is possible to zig
between any two adjacent domains.
The zig arrow can give the student a first impression that making a zig is easy,
although it represents the end of a decision process that may have had many
iterations, decisions, and counterdecisions. The inputs for the zig are the Cs, FRs,
ranges of acceptance of FRs, variation of FRs with time, and tolerances of the DPs.
Mapping involves two main tasks:

– “Defining and selecting DPs” and

– “Setting a parameter value for the selected DP.”
9 Mapping in Design 269

“Defining and selecting the DPs” is the most important part of the zig because it
should be maintain independence and minimize information content for that
component of the design solution. If the design is coupled, the design team might
try decoupling it or going back to check alternative DPs and FRs. After selecting a
physical solution, the design moves to “setting a parameter value for the selected
DP.” To set the value for a DP may involve an optimization or tuning process that
may end with the definitions of the DPs, or in turn, by coming back and starting the
zig process again. Optimization and tuning are well-known processes of engi-
neering and management, as explained in the introduction of this chapter. There-
fore, we will give special attention to the “Defining and selecting the DPs.”
At high levels in designs, an FR may be “machine wash” using as a DP
“washing machine.” This type of solution for the DP is not completely neutral but
defines that the design uses a machine and allows the design to proceed.
For each FR, a DP might be sought, and the design matrix (DM) evaluated by
checking the relationships between each DP and the FRs. One way of starting
processes at each level is by defining the system at a nominal condition (NC) and
obtain a first draft for the DM.
NCs of operation are common in engineering and management. However, final
designs must consider all conditions of operation. Otherwise, depending on the
condition of operation, the system could fail.
The NC error
A hotel group owning a four stars hotels asked a design team to propose an HVAC
(heating, ventilating and air condition) system for a new hotel. The hotel faced West, with a
large glass façade on this side. Therefore, during the afternoon, the solar heating load
would increase. The solution proposed by the design team can heat and cool at any time.
This solution is the so-called “four-pipe system” with two pipes for heating and two other
for cooling. However, the owner of the hotel had good experiences in other ones with
cheaper two-pipe systems that work in only one mode at a time, cooling or heating. This
two-pipe system is less expensive than the four-pipe system. Overriding the advice of the
design team, the owner selected two-pipes. He rationalized his argument by saying that the
two-pipe solution could fulfill nominal winter and summer conditions.
The hotel opened by November, and the HVAC system was able to fulfill the heating
requirements during the winter season. However, in the middle season, there were needs
for heating in the morning and cooling in the afternoon in rooms adjacent to the west
façade, while just heating all day in the other rooms. Moreover, as the water in the pipes
was hot in the morning, the change to cooling requires changing all the water, and it needs
more than an hour to start cooling. The hotel had many complaints about uncomfortable
temperatures. After a year of operation, they had to change to a four-pipe HVAC system.

The above example shows that the nominal working condition can be a good
starting point. However, it is not enough to define a system. The system might work
at any actual working conditions. However, all possible working conditions may be
hard to define.
In structural engineering, codes define design working conditions. The codes set
minimum loads for structures as well as combinations of loads, e.g., wind and
snow.
270 M. Cavique et al.

In enterprise management, nominal working conditions can be the target for the
core business. However, demands change, and it can be necessary to find temporary
ways to increase working capacity, e.g., research for special development needs and
lawyers when legal problems arise.
Verifying system performance in all predictable conditions can reveal previously
unanticipated coupling between DPs and FRs, missing DPs or FRs. Evaluating all
the DPs and FRs lets design teams use the DM to maintain independence in actual
working conditions, to direct designers to decoupled or uncoupled solutions. The
design team could decide to use DMs that express the behavior of the system as a
whole or use different design matrixes for each working state of the system.
Figure 9.7 shows a flowchart of a possible process for “defining and selecting
DPs.” The top of the figure is about solving the design at NCs, the middle at actual
conditions (AC), and the bottom of the figure, checking the information content
considering the tolerances of the DPs.

Fig. 9.7 Defining and

selecting DP. (Reproduced
from Cavique et al. 2017,
originally published open
access under a CC BY 4.0
license: https://doi.org/10.
1051/matecconf/
201712701007)
9 Mapping in Design 271

The designer might find one or more DPs that can fulfill an FR. According to AD,
good solutions should be uncoupled or decoupled, maintaining the independence of
functional elements. To select between DPs that all maintain the independence,
choose the one with the least information content, i.e., the greatest chance of success.
Students already know that a 100% probability of success corresponds to zero
information. Design teams should compute the information content in ACs. In other
words, check if the system performs throughout the intended functional ranges.
Actual working conditions for an artifact are usually easy to define, an example
being the earlier case of tilting the steel reel. In engineering systems that depend on
nature, ACs are probabilistic, as with weather, wind, earthquakes, and fires. Similar
probabilistic approaches happen on security systems, hospital emergency, or
enterprise manufacturing, that depend on customer behavior.
Coming back to Fig. 9.7, to define FRs at NC, the inputs IN_1 are the DPs and
FRs at the immediately higher level of decomposition and the Cs. Thus, the
designer might synthesize the DPs, obtain the DE, and compute or estimate the
information content of the design. At high levels in decompositions, the compu-
tation of the information content may follow a non-formal way, as expressed in the
example of tilting the steel reel. Chapter 11 shows more detail of computing
information content.
Besides, it is necessary to check if the child FRs and DPs combine to equal the
parent and if the children are independent of each other (collectively exhaustive and
mutually exclusive, CEME, Fig. 9.7). Section 9.6 describes the CEME concept in
more detail. At the end of a zig, it may be possible to specify values, or preliminary
values, and tolerances for the DPs.
Further, we need to understand how systems work under ACs, using DEs,
computing the information, and checking again if FRs are CEME. This phase needs
additional inputs (IN_2) regarding the DPs, FRs, and Cs. Some designers may
proceed directly to the AC phase, using IN_1 and IN_2 to define the DE.
In many engineering or management environments, it is possible to finalize the
zig at this stage and proceed to specify DP values and tolerances.
Finally, design teams might want to evaluate ranges of variation of FRs, and
corresponding tolerances of the DP, dDP (IN_3). Codes seldom enforce this
evaluation in structural engineering, air conditioning design, fire safety, and many
building engineering activities, although they are common in machine manufac-
turing. Once more, if designs fulfill requirements, design teams can proceed to DP
specifications at the working level of decomposition and tune them. If design is
coupled, an optimization process could be required. Optimization and tuning might
not be necessary at the higher levels of decomposition.
Air conditioning design
Variable air volume (VAV) is a common air conditioning system that supplies a variable
amount of air at a cold temperature to each room. Figure 9.8 shows spaces 1 and n of n
spaces of a building cooled by the VAV system. The figure depicts the air handling unit
(AHU), the supply duct (1), and a VAV box (2) controlled by a thermostat (T). The system in
the figure is the simplest VAV able to work just in the cooling mode.
272 M. Cavique et al.

Fig. 9.8 A VAV system. (Reproduced from Cavique et al. (2017), originally published open
access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/201712701007)

When the temperature goes up in any room, the room thermostat asks the regulator in the
VAV box (2) to open the damper step-by-step to increase the amount of cooled air entering
the room. Thus, the pressure in the duct (1) goes down, and the sensor (P) gives the signal
to speed up the fan of the AHU.
The AHU supplies a mixture of outside air (OA) and return-air (RA) throughout duct (1),
making the amount of OA delivered at each room to depend on the damper position.
Therefore, the amount of OA delivered to each room is a function of the heat removed from
the room and all other rooms.

At the higher level of decomposition, FR0 and DP0 are

FR0 = Give comfort to room occupants.
DP0 = VAV air conditioning system.
The intent of the VAV systems of the middle of the twentieth century is to “achieve indoor
thermal comfort,” therefore controlling room temperature. In the seventies, the oil crisis
forced a reduction in energy use in buildings, and new designs then reduced the usage of
outdoor air. Unfortunately, scarce fresh air brings problems of indoor air quality (IAQ),
and the latter forced air conditioning systems to include a new function: “provide IAQ.”
Table 9.1 displays the FRs and DPs for the VAV system, showing that the system has more
FRs than DPs.

At NCs, VAV boxes of each room are defined knowing the temperature supply (Ts) and the
needs of cooling of each room. The definition of the total airflow of the system and the total
cooling requirements can be used for specifying the AHU flow system and cooling systems.
Finally, the OA set at the AHU provides the total amount of air needed according to the
IAQ of the building.
The OA of each space FR21 and FR22 depends only on the VAV boxes and OA system of the
AHU. Equation 9.1 shows the DE of the VAV system at NCs, which allows specifying the
AHU, the VAV boxes, the duct, and the OA system, but not expressing the behavior of the
entire system.
9 Mapping in Design 273

Table 9.1 FRs and DPs for the VAV system

FRs DPs
FR1- Achieve indoor thermal comfort DP1- Temperature control system
FR2- Provide IAQ DP2- Outdoor air control system
FR11- Remove heat from space 1 DP11- 1′s VAV box airflow
FR12- Remove heat from space n DP12- n’s VAV box airflow
FR13- Provide total airflow supply DP13- AHU flow system
FR14- Adjust Ts of the AHU DP14- AHU cooling coil system
FR21- Provide space 1 OA
FR22- Provide space n OA
FR23- Provide building total OA flow DP23- AHU OA system

8 9 2 3
>
> FR1 > > X 8 9
>
> FR2 > > 6X 7 > DP1 >
>
> >
> 6 X 7 > >
>
> FR11 >> 6 7 >> DP2 > >
>
> >
> X X 7 >> >
>
>
> > 6 > >
< FR12 >= 6
6 X X 7 >
7 < DP >
11 =
FR13 ¼ 66 X X 7
DP12
7 > ð9:1Þ
>
> > >
>
> FR14 >>
>
6
6 X 7 >
7 >> DP13 >>
>
>
> >
> 6 7 >
> >
>
>
> FR 23 >
> 6 X7 > > DP 14 >
>
>
> >
> 4 : ;
>
> > X5
: 21 >
FR X DP23
;
FR22 X X

Exercise 9.3: Define the Design Equation of a VAV

In a VAV system, anytime the damper of a VAV box opens, all the system adjusts, as the
amount of air delivered in all other boxes changes.

• Spaces with higher cooling loads receive more supply air, and therefore they receive
more OA. Therefore, FR21 depends not only on DP11 but also on the behavior of all
other VAV boxes, creating a coupling between the needs for OA and the cooling load of
any space. Try to show this issue by changing the above Eq. 9.1;
• The OA airflow is a requirement of the type “the more, the better.” Therefore, if the OA
airflow delivered in each space is high enough, it can fulfill the needs of any space at
any load. Therefore, neglecting the problem of having more FRs than DPs, it might be
possible to meet the minimum requirements for OA at each space.
Explain how this type of coupling can be solved.

After “defining and selecting DPs,” design teams need to specify DP values,
which can be computed using DEs. DMs, however, usually indicate only non-zero
derivative relationships of the DEs with Xs, without showing equations. Normally,
it is possible to specify or pre-specify DPs with DEs.
Having the set of DPs, it may help to perform the following tasks to estimate the
values of the DPs:
274 M. Cavique et al.

– layout child DPs;

– write parent–child, or decomposition, equations to show how child DPs fulfill
and are CEME concerning the parent DP;
– verify DP tolerances and check against the range of the FRs (IN_3 of Fig. 9.7).

Afterward, systems can be studied and tuned or adjusted using specific values of
each DP. In case an irreconcilably coupled design results, an optimization process
could be applied to get the best possible behavior. On the other hand, it might be
possible to tune a coupled design so that it works acceptably through some useful range.
If a design solution fails to comply with the axioms at any phase, go back and
redefine the DPs and the DE, or redefine the tolerances of the DPs. If necessary,
design teams could need to start the zig again and again. The work already done
will help a lot in the following attempt to reach a solution.
The zig arrow is the final representation of many attempts to reach a good
solution at a level of decomposition, i.e., selecting a DP for an FR. The experience
and knowledge of the design team, as well as their creativity and method, play an
important role in attaining a good solution.
The first step in the critical assessment of a zig is to apply the first axiom of AD.
If several candidate DPs have equal independence, then the second step is the use of
the information axiom to check the probability of success of the design. Finally, the
values and tolerances of the DPs need to be specified.

9.6 The Zag

The zag is the process represented by the arrow coming back to the left from a
parent level of a domain to a child level of decomposition of the previous domain
(Fig. 9.2). The usual task is to come from a DP in the physical domain back to the
functional domain. In the functional domain, the new FRs are developed in the
context of the upper level DPs. The new FRs must fulfill, i.e., be collectively
exhaustive with respect to the parent FR and mutually exclusive to each other.
The next exercise aims to show the role of selecting a DP when zagging back
into the functional domain.
Exercise 9.4: Design a Thermal Suit
When a person is in thermal equilibrium, the heat produced by the metabolism is removed
from the human body by radiation, convection, conduction, and evaporation. The evapo-
ration accounts for vapor transfer due to respiration and transpiration. To feel comfort-
able, people need to be in thermal equilibrium while meeting constraints for skin
temperature and transpiration.
Therefore, suppose the design starts with the following FR0 and DP0:
FR0 = Achieve indoor thermal comfort.
DP0 = Thermal suit.
What would be the next level FRs to fulfill FR0?
9 Mapping in Design 275

The zig synthesizes the DPs that fulfill the FRs by answering “how we achieve
it?”. The zag comes back to the functional domain and answers, “what we want to
achieve?”. The new FRs need to be in accordance with the parent FRs and the
previous DPs. However, there is no special tool in AD to check if the new FRs are
in accordance with the desired design. FRs are derived from the parent DPs and
collectively express the parent FR. At the end of a zig, we find a point of consis-
tency in the design.
To help to make the zag, Thompson (2014) introduced the question of “Why did
we choose this DP?” the answer is the selection of the child FRs. In addition, child
FRs need to be within the Cs. Therefore, the design process using AD theory is a
synthesis process to discover the DPs, and it includes a subsequence process of
analysis to achieve the child FRs.
Once more, when designers perform the zag, they should not be afraid of coming
back to the parent DP at any time, if it seems the design failed. Moreover, if necessary,
redo the previous zig. The following steps summarize a way of doing the zag:

– Select an FR–DP pair for decomposition. The DP could have specification or

pre-specification values from the previous zig.
– Consider the Cs for FRs and DPs.
– Define child FRs answering to “why did we choose the parent DP?”. It can be
seen as reverse engineering from the parent DP. Notice that a parent FR needs to
have two or more child FRs. Otherwise, the child FR is the parent FR, and no
decomposition happens, or there is a missing sibling.
– Check if the proposed FRs align with CNs. If you realize there is a need for a
new CN, start the design again from the beginning, from the FR0.
– Check the FRs for consistency with the parent FR. The set of child FRs needs to
be collective exhaustive (CE) regarding the parent FR. Moreover, the FRs need
to be independent of each other, so their intersection is nil, and therefore, they
are mutual exclusive (ME). Hence, the consistency with the parent FR is CE plus
ME, or CEME.
– Make a double check to evaluate if you need all the FRs. The number of FRs
should be as low as possible to enable achieving a good design.
Exercise 9.5: Design of a Purchasing Department
The purchasing department of an enterprise needs a redefinition from scratch. The design
should develop in these four domains: customer, functional, physical, and process. Services
of the department are in the physical domain. Data treat belongs in the process domain.
Most material is purchased locally. Some technical equipment has international orders.
There are three major groups of purchased materials:

– repair and maintenance material, usually purchased in small quantities at low prices;
– material for maintenance and manufacture acquired in sets of parts;
– technical equipment with detailed specifications.
276 M. Cavique et al.

Do the mapping between the four domains until you can define the following outcomes for
the design solution:

– define the internal services of the department;

– what are the roles of each service?
– procedures to implement at each service (define it descriptively);
– outline the flow of data between services and external companies;
– what are the control and audit operations to implement?
Students might need to define the operation mode of the department by contacting a real
enterprise.

9.7 Decomposition and Concurrent Engineering

The ontological classification of design entities in the four domains helps to define
many problems. In designing artifacts, enterprises, management organization, and
many other things, identifying the CNs, FRs, DPs, and PVs fosters clear reasoning.
This section presents the Concurrent Engineering approach, which is an
important industrial application of the mapping between domains. Concurrent
Engineering is intended to integrate all product and process development tasks
simultaneously, in the design phase, to develop a product quickly.
The Society of Concurrent Product Development (SCPD) defines (1993): Inte-
grated product development is a philosophy that systematically employs a teaming
of functional disciplines to integrate and concurrently apply all the necessary
processes to produce an effective and efficient product that satisfies the CNs.
Moreover, SCPD states Concurrent Engineering involves the correct interplay of
the functional department, including customers and suppliers, and the supporting
infrastructure technologies.
Traditional engineering approaches classify design processes in three sequential
stages: conceptual design, product design, and process design. The three stages
define the most common way to design. However, this approach depends on the
technological or field knowledge of the design team. To overcome the possible lack
of field knowledge, the design teams usually have persons with different skills and
experiences. The target is always to have a good design that can be constructed,
processed, and installed in a short time, so enterprises save money in development
and catch the market opportunity.
About 80% of the processing cost can be due to decisions made during the
product design phase. Therefore, important product design decisions made without
the knowledge of manufacturing, assembly, and inspection, may have extra costs
involved. Concurrent Engineering provides rules and techniques for “design for
manufacturing” and “design for assembly.” In this context, the design process
should involve the process domain at the upper levels of the decomposition, unlike
the situation when the manufacturing technology is already defined.
9 Mapping in Design 277

Concurrent Engineering provides rules and techniques for “design for manu-
facturing,” and for assembly, hence reducing the total time to market of new
products. However, product design by itself might take longer when it interacts with
manufacturing design in the full realization of new products. Therefore, Concurrent
Engineering can be more difficult to implement in “traditional” enterprises that tend
to have new products “designed” in a short time. These shorter “design” times
result in iterations at the prototype, production, and test phases and result in a long
time to reach the market.
Traditional enterprises accept CAD (computer-aided drafting, or “design”) files
as representations of completed designs. Typically, these CAD files are grossly
inadequate representations of design solutions from an AD perspective. They jump
directly to the physical integration phase and include no indication of the design
intent (FRs). They lack compliance with Suh’s design axioms. Enterprises need to
learn to value time spent in rigorous decompositions and physical integrations that
produce design solutions that comply with Suh’s axioms. Representations of
completed design solutions should include decompositions, metrics, design and
decomposition equations, design matrices, and physical integration. For Concur-
rent Engineering solutions, the process domain and process matrix and physical
integration of the manufacturing process should be included as well.
Figure 9.9 shows the traditional and Concurrent Engineering approaches. The
traditional approach is sequential, knowledge flows from the product design to the
process design, and feedback from the process design to the product design is
informal or via the marketing information. In contrast, the Concurrent Engineering
approach mixes knowledge from product and process designs, and it involves
people with different knowledge bases and skills. In Concurrent Engineering
environments, the availability of processing knowledge can cause redesigns of the
artifact that are problematic from a manufacturing perspective at any phase of the
design process.

Fig. 9.9 Engineering Time to market

approaches to the Product and
Process Design
“TradiƟonal” Engineering

Product Design Process Design

“Concurrent” Engineering

Product Design

Process Design
278 M. Cavique et al.

In this context, when the manufacturing technology is not yet known, e.g.,
large-scale integrated circuits in the early 1960s, then, at high levels of the
decomposition, mapping should go from the CNs to the FRs, from the FRs to the
DPs, and from the DPs to the PVs. If manufacturing technology is already known,
e.g., conventional cars in 2000, the DPs to the PVs can be mapped just at lower
levels of decomposition.
Figure 9.10 shows the Concurrent Engineering approach to the zigzag, where
the zig to the product design flows to the process design, and the zag to the FRs
contains inputs from the process domain.
This approach reduces the subjectivity of decision-making at the early phase of
the product development, because it generates viable and cost-effective design
solutions, by including the knowledge about the PVs.
Mathematically speaking, the DE for the product design is

fFRg ¼ ½ AfDPg ð9:2Þ

For the production process design:

fDPg ¼ ½BfPV g ð9:3Þ

The combination of Eqs. 9.2 and 9.3 yields

fFRg ¼ ½C fPV g ð9:4Þ

where [C] = [A]  [B] is the design matrix for the concurrent design process.
According to the Independence Axiom, in an ideal design, all the three design
matrices, [A], [B], and [C], must be square and diagonal or, failing that, triangular.

Fig. 9.10 The decomposition process as performed in a Concurrent Engineering environment

9 Mapping in Design 279

Geometric Dimensioning and Tolerancing (GD&T) design to guarantee the assembly

of a component
A component (Fig. 9.11) shall be assembled to a machine through a set of four holes and
shall be used to attach another component through a set of three holes. The surfaces of the
component are previously prepared, and the two sets of holes are analyzed here.
These points are necessary to assure the desired assembly:

(a) ensure the position of the component in the machine has no ambiguity;
(b) the set of four holes must be positioned relative to the structure of the machine;
(c) the set of four holes defines the position of the set of three holes;
(d) the subsequent assembly, by the set of three holes, must be done with this component
laid on the machine;
(e) through the set of three holes, a new component will be assembled and must be
perpendicular to the plane of the first component.
The first-level mapping of this case based on the concurrent engineering approach is

FR1 = Ensure the position of the component;

FR2 = Ensure holes position.

DP1 = Isostatic support system;

DP2 = Dimensions and geometric tolerances related to holes.

PV1 = Pins in a three-plane datum system;

PV2 = Work-holding solution.
This mapping between functional, physical, and process domains is represented by these DEs:
    
FR1  0 DP1
¼ ð9:5Þ
FR2   DP2

Fig. 9.11 Component used in the exercise: a plate with two sets of holes
280 M. Cavique et al.
    
DP1  0 PV1
¼ ð9:6Þ
DP2   PV2

In the second level, the decomposition of FR1 and FR2 is a consequence of the previous zig,
from functional to process domains.

FR11 = Seat one face of the component on the machine (by three seating points on one
reference plane);
FR12 = Guide one edge of the component (by two guiding points another reference
plane perpendicular to the first);
FR13 = locate one end of the component (by a single locating point on a third reference
plane perpendicular to the first two);
FR21 = Allocate at each one of the four holes a position tolerance at the worst condition
to assembly operations, related to the three reference planes;
FR22 = Adopt the set of four holes referenced to the position of the set of 3 holes;
FR23 = Allocate to each of the three holes a position tolerance related to the plane seated
on the machine and to the set of four holes, at the worst condition to assembly
operations, and at the same condition, this tolerance must be tighter when related
to the mentioned plane.
The worst condition to assembly operations occurs when a feature or part has the maxi-
mum amount of material (volume/size) within its dimensional tolerance. This situation is
the maximum material condition (MMC). The MMC of the shaft would be the maximum
diameter; The MMC of the hole would be its minimum diameter. (Georg Henzold 2006).

The set of DPs that satisfy the specified FR are:

DP11 = The face which plane is D;

DP12 = Plane A defines the edge that guides the component;
DP13 = Plane B position the component;
DP21 = The position tolerance at maximum material condition, in any direction, of each
hole of the set of four holes, follows the sequence of relationship to faces D–A–B;
DP22 = The geometrical center of the set of four holes is a reference (C) to the position
of holes of the set of three;
DP23 = The position tolerance in any direction at maximum material condition, of each
hole of the set of three holes, is primarily related to face D, and thus to the four
holes as a whole (C) for its worst situation (maximum material condition), and
simultaneously has a tight tolerance related to face D at the same material
condition.

DPs are depicted in a technical drawing (Fig. 9.12). Letters A, B, C, and D represent
datum, and small letters (r, s, and t) represent the value of each tolerance since numerical
values are irrelevant for this study.
Equation (9.7) shows the relation between the functional domain and the physical domain
and represents the product design:
8 9 2 38 9
>
> FR11 >
>  0 0 0 0 0 >> DP11 >
>
>
> FR12 >
> 60  >
> DP12 >
>
>
> >
> 0 0 0 077>
> >
>
< = 660 0 < =
FR13  0 0 077 DP13
¼6
6 ð9:7Þ
>
> FR > 6 
>   0 077>
> DP >
>
>
21
> > 21 > >
>
> FR22 >
>
>
4   0  0 5>
>
> DP > >
: ; : 22 > ;
FR23  0 0 0   DP23
9 Mapping in Design 281

Fig. 9.12 Component with datum system and geometrical tolerances

The set of PVs that characterizes the process that is used to implement the design (achieve
or create the DPs), are:

PV11 = Define plane D by three seating points;

PV12 = Define plane A by two seating (guiding) points;
PV13 = Define plane B by one seating (locating) point;
PV21 = Clamp the component after positioning according to the sequence D–A–B;
PV22 = Determine the geometrical center of the four drilled holes set, which is datum C;
PV23 = Support the component on the three seating points (plane D), and use
dimensions referred to the center of the four holes set to their production and
inspection.
Equation (9.8) shows the relation between the physical domain and the process domain
and represents the process design.
8 9 2 38 9
>
> DP11 >>  0 0 0 0 0 >> PV11 >
>
>
> DP12 >> 60  >
> PV12 >
>
>
> >
> 0 0 0 077>
> >
>
< = 660 0 < =
DP13  0 0 077 PV13
¼6
6  ð9:8Þ
>
> DP >
>   0 077>
> PV >
>
>
21
> 6 > 21 > >
>
> DP22 >>
>
4   0  0 5>
>
> PV > >
: ; : 22 > ;
DP23  0 0 0   PV23
282 M. Cavique et al.

The ease of manufacturing, assembly, and inspection operations is a consequence of the

GD&T specifications (FR) making the DM show the independence between FRs and DPs
and between DPs and PVs.
Concurrent Engineering leads to the specification of the functionalities in agreement with
the manufacturing, assembly, and inspection operations. This example shows that GD&T is
an engineering language that helps the creation of independent solutions, as long as, at
each mapping, the DMs are triangular (Eqs. 9.7 and 9.8). Moreover, this example shows
that AD domains and mapping give good theoretical support for the foundation of Con-
current Engineering.

Exercise 9.6: A Small Change in the Design of the Component

After designing the component, a factory gets to manufacture it. The manufacturing
engineer decides to use the datum A, B, and D on both sets of holes. He argues the
dimensions are the same using one datum system or the other.
Discuss the manufacturing engineer’s idea by redefining the design matrixes from FRs to
DPs and from DPs to PVs at the second level of decomposition. Moreover, discuss the
problems the manufacturing process will face.

9.8 Conclusion

AD theory (AD) avoids trial and error, helps to classify a solution as good or poor
but does not give the solution by itself. A good design process needs to develop and
define good FRs. A design solution can be no better than its FRs. This statement
encouraged the authors of this chapter to give extra hints on how to define FRs.
The main message of this chapter is that it can take many attempts to perform
good zigs and zags. The design team might not be depressed about starting the
mapping process repeatedly. AD is about doing a good design at the first attempt,
avoiding trial and errors. However, at each level of decomposition, many attempts
might need to finalize a zig. Finalizing a zig with success provides consistency to
the design because it makes it possible to apply Suh’s two axioms.
The zag requires child FRs that are collectively exhaustive with respect to the
parent and mutually exclusive with respect to each other.
Finally, this chapter presents the mapping to help to reason in different domains
of design and shows Concurrent Engineering as an example that needs decompo-
sition in all four domains at the same level.
Problems
1. Deploy up to the third level a possible FR and DP zigzag decomposition for a
two-stroke internal combustion engine.
2. Comment on the statement: “Decisions at each level have important conse-
quences at lower levels.”
3. A manufacturing company works in a traditional industry of shoemaking.
A consultant said to the enterprise chair that “there is no such thing as a poor
technological market”. Discuss what the consultant said:
9 Mapping in Design 283

– the industry needs to map between the physical and process domains;
– the industry needs to map at the same time between the functional, physical,
and process domains.

4. An HVAC system has at the first level of decomposition the same FRs and DPs
as the ones shown in the air-conditioning example:

FR1 = Achieve indoor thermal comfort;

FR2 = Provide IAQ;

and

DP1 = Temperature control system;

DP2 = Outdoor air control system.

A system uses an AHU to treat the outdoor-air and a convection-type system to

cool the indoor ambiance.

– define the DM for the first level of decomposition;

– decompose the design until the next level.

5. The traditional salt industry harvests a mixture of salt crystals and clay.
A possible way to separate salt crystals from clay is by dissolving the clay. The
process uses a solvent, called brine, which is a saturated aqueous salt solution.
This solution can no longer solve salt but dilutes the clay and maintain the salt
crystals. The main FRs of this design are as follows:

FR1 = Separate clay from salt;

FR2 = Dilute the clay;
FR3 = Remove the clay;
FR4 = Remove the solvent;
FR5 = Collect the washed salt.

– Define the DPs and PVs while meeting the first Axiom of AD.

References
Cavique M, Fradinho J, Gabriel-Santos A, Gonçalves-Coelho A, Mourão A (2017) The Iterative
Nature of the “Zig” and How to Define the “Hows”. In: Slǎtineanu L (ed) 11th International
Conference on Axiomatic Design (ICAD), MATEC Web of Conferences, Iasi, Romania.
https://doi.org/10.1051/matecconf/201712701007
Gabriel-Santos A, Martinho A, Fradinho J, Cavique M, Gonçalves-Coelho A, , Mourão A (2017)
How Axiomatic Design can promote creativity in the design of new products. In: Slǎtineanu L,
Nagit G, Dodun O, Merticaru V, Coteata M, Ripanu M, Mihalache A, Boca M, Ibanescu R,
Panait C, Oancea G, Kyratsis P (eds) 21st Innovative Manufacturing Engineering & Energy
International Conference—IManE&E 2017, MATEC Web of Conferences, vol 112, p 8.
https://doi.org/10.1051/matecconf/201711209010
284 M. Cavique et al.

Further Reading

Cavique M, Gonçalves-Coelho A (2011) Repeatability in Design Science. In: Thompson MK

(ed) 6th International Conference on Axiomatic Design (ICAD 2013). KAIST, Axiomatic
Design Solutions Inc., Daejeon, Korea, pp 31–34
Dickinson A, Brown CA (2009) Design and Deployment of Axiomatic Design. In: ao de Azevedo
Gonçalves Coelho AMFR (ed) Proceedings of the 5th International Conference on Axiomatic
Design, Universidade Nova de Lisboa, accessed at Axiomatic Design Solutions, Inc., Lisboa,
Portugal
Gonçalves-Coelho A (2004) Axiomatic Design and the concurrent engineering paradigm. Acad J
Manufact Eng 2(2):6–15
Henzold G (2006) Geometrical Dimensioning and Tolerancing for Design, Manufacturing and
Inspection: A Handbook for Geometrical Product Specification using ISO and ASME
standards, 2nd edn. Butterworth-Heinemann
Suh NP (1990) The Principles of Design. Oxford University Press
Tate DE (1999) A roadmap for decomposition : activities, theories, and tools for system design.
Doctor of philosophy in mechanical engineering, Massachusetts Institute of Technology, 77
Massachusetts Ave., Cambridge MA 02139, USA
Thompson MK (2013) A classification of procedural errors in the definition of functional
requirements in Axiomatic Design theory. In: Thompson MK (ed) 7th International Conference
on Axiomatic Design (ICAD 2013), CIRP, accessed at Axiomatic Design Solutions, Inc.,
Worchester, MA, vol 32, pp 1–6
Thompson MK (2014) Where is the ‘Why’ in Axiomatic Design? In: Thompson MK (ed) 8th
International Conference on Axiomatic Design (ICAD 2014), CIRP, accessed at Axiomatic
Design Solutions, Inc., Lisboa, Portugal, vol 33
Redundant Designs
10
Miguel Cavique and António Gonçalves-Coelho

Abstract
The concept of redundancy is common in day-to-day life. In engineering,
redundancy allows a system to work safely, often by creating a backup that
enters on service if the primary system fails. Redundancy in management may
have the meaning of plan-B if the initial plan fails. In nature, redundancy appears
in many living organisms, man included. This chapter starts by presenting
examples of redundancy in nature, then showing how redundancy applies to
many engineered systems, and how redundancy may have different meanings in
engineering. According to Axiomatic Design (AD), a redundant design has more
design parameters (DPs) than functional requirements (FRs). There are two types
of approaches for redundancy: reliability motivated and the functionally
motivated approach. These thoughts give room for discussing the ontology of
redundant design, allowing the derivation of new theorems on redundancy. One
of these theorems helps to decouple coupled designs.

10.1 Introduction

The standard definition of redundancy relates superfluousness or to have surplus

resources to perform a specific task. The surplus resources may be required in case
part of the system fails. This definition has many applications in engineering as failure
happens in computer, energy, and structural engineering. In reliability, redundancy

M. Cavique (&)
UNIDEMI & Escola Naval, Base Naval de Lisboa – Alfeite, 2810-001 Almada, Portugal
e-mail: [email protected]
A. Gonçalves-Coelho
UNIDEMI & DEMI, NOVA SST, Campus de Caparica, 2829-516 Caparica, Portugal

© Springer Nature Switzerland AG 2021 285

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_10
286 M. Cavique and A. Gonçalves-Coelho

helps to maintain the entire system working in case something fails. As an example of
the last case, the air-conditioning of a data center may use more indoor units than
necessary. A standard solution is to use two more indoor units. Each additional unit
switches on in case of failure occurs on any of the working units. In the communi-
cations field, redundancy may mean that a message will flow through two different
channels that can reach the receptor. Therefore, the system still operates in case one of
the channels fails. The system may use two channels of the same kind to deliver the
message or two completely different ways to fulfill the same requirement.
Redundancy is common in the day-to-day use of computers, by using back-ups,
cloud systems shared by distinct computers, or data storage in separate disks.
RAID
The Redundant Array of Inexpensive Drives (RAID) is a way to store data in separate hard
disks. There are many types of architectures for RAID systems, the simplest one consisting
of replication of data in two separate disks. Therefore, if a failure occurs in one disk, the
system can keep operating without any visible disturbance to the user.

Exercise 10.1: Redundant Design Equation

Wellington sent a message to London regarding the victory at the Waterloo battle, using a
semaphore chain and a carrier pigeon. Due to the fog, the pigeon was the first to reach the
destination. Define the design equation of this design.

In “education,” a message sent by the teacher may not reach the student in the first
attempt. Therefore, teachers use redundant techniques to make the message reach
the student. One way is to repeat the same subject; another is to use many exercises
on the same topic or to observe the occurrence in a lab or videos; or a blend of all
these processes that produces an interesting course.
Exercise 10.2: Defining an AD Course
The student read the last nine chapters of an AD course and may have an opinion about the
key concepts of the subject. Define the FRs of an introductory course on AD using the
action verb “Understand …”. Define the DPs for the FRs identified and check if the design
is a redundant design.

This section presents examples of dissimilar concepts of redundancy. Next sections

give a closer look at this subject in what concerns reliability and functionally
motivated redundancy. At the end of this chapter, the theorems of redundancy are
presented.

10.2 Redundancy in Nature

Natural systems depend on many variables that usually make the design redundant.
In a lake, the water temperature depends on wind velocity, air temperature, and
humidity, solar radiation and absorption of radiation by the plants and soil, the soil
temperature, etc. The soil temperature depends on depth, thermal amplitude,
10 Redundant Designs 287

radiation, and average outdoor temperature. The number of fishes in the lake, or the
photosynthesis of plants, depends in turn on some of these variables. Therefore,
Eq. (10.1) may represent a natural phenomenon, where y’s represent the outputs
and x’s are input variables, with m larger than n:

y1 ¼ f1 ðx1 ; x2 ; :::; xm Þ
y2 ¼ f2 ðx1 ; x2 ; :::; xm Þ
ð10:1Þ
:::
yn ¼ fm ðx1 ; x2 ; :::; xm Þ

In biology, equations are harder to define, but biologists know the way variables
interfere in the phenomenon under analysis. In medicine, medical doctors may not
be able to quantify all variables in a medical episode, but they know what
cross-dependencies exist between the variables and symptoms. To make a decision,
medical doctors need to assume some variables as fixed and work with all the
others. A similar situation may happen in management practice, on a marketing
campaign or at a military theater, making the person in charge of the decision to fix
some variables to allow a solution for the set of equations.
The Laplace’s demon
Laplace lived in the eighteen century and died at the beginning of the nineteenth century. It
that period, Newton’s laws were the fundamentals of mechanistic theories that tried to
explain all phenomena based on mechanic causalities. Laplace supposed that if it would be
possible to write all the equations of all atoms, then the Universe would have a deter-
ministic behavior. Therefore, knowing their present location and momentum, the equations
would reveal the future and the past of the universe. This reasoning is known as the
Laplace’s demon, which would be a mind that can know the past and future.

Calculation algorithms in engineering frequently address redundant designs. Sup-

pose we need to define the diameter of a duct that delivers water in some spots of
the line. Topologically, the duct is a sequence of edges in series. The calculation of
the diameters of the edges starts by defining the available head loss. Thus, knowing
the flow at each edge makes it possible to use many combinations of edge diameters
that satisfy the available head loss. Therefore, the design is a redundant design.
Duct algorithms use a common parameter or a dimensionless number to define the
ducts diameters. It is usual to define as a dimensioning criterion a constant velocity,
a constant head loss per unit of length, a constant energy loss, or a minimum
investment over a period. In any case, using these parameters, the diameters turn to
depend on flow at each edge. This method changes the design from redundant to
non-redundant and uncoupled. The following equations help to explain this subject.
Equation (10.2) shows that a set of n branches with diameters Di and flows Qi
can fulfill the available pressure drop ΔP. In this equation, what we want to achieve
(FR) is the pressure drop ΔP, and the DPs are the diameters Di, for the known flow
rates Qi. Therefore, the equation to solve is:
288 M. Cavique and A. Gonçalves-Coelho

DP ¼ f ðD1 ; D2 ; . . .; Dn ; Q1 ; Q2 ; . . .; Qn Þ ð10:2Þ

The equation may change to an ideal design by using a dimensioning criterion: if

the diameters Di are functions of the flows Qi. Knowing the flows at each edge
allows defining the diameters. According to AD, “what we want to achieve” are the
diameters, and the DPs are the known flows Qi at each branch. Equation (10.3)
shows this idea:
2 3 2 3 2Q 3
D1 X 1
6 D2 7 6 76 Q 7
6 7 6 X
7:6
27
6 .. 7 ¼ 4 .. 54 . 7
6 . ð10:3Þ
4 . 5 . . 5
Dn X Qn

Exercise 10.3: Selection of a Beam

A student wants to select a beam with a uniform rectangular section to support a maximum
bending moment Mmax. For a section with dimensions b and h, the maximum axial stress
Mmax h=
rmax is rmax ¼ I 2, where h is the thickness of the beam, and I is the moment of inertia
3
of the cross section, I ¼ bh
12 . Therefore, the student might be able to:

write the design equation for the FR = “be able to sustain a maximum bending moment
Mmax”; and
define a variable so that the design turns from a redundant to a non-redundant design.

The last examples show that some redundant designs may turn into non-redundant
designs using quantified dimensions. In many engineering applications, it is pos-
sible to aggregate variables in dimensionless variables and relate the FR, “what we
want to achieve,” with some dimensionless quantities. In fluid mechanics, the
friction coefficient of a pipe, f, is a function of the Reynolds number, Re, and the
dimensionless roughness, e. Moody, after Rouse developments, depicted the dia-
gram that shows the equation f ¼ f ðRe; eÞ, where f is the friction coefficient. The
friction coefficient defined by the Moody diagram applies to most of the engi-
neering fluids.
The advantage of using dimensionless numbers is the possibility of using a
smaller number of experiments to find the function between the dimensionless
numbers than the number of experiments that would be necessary if all the variables
in the dimensionless number were used. Thus the findings can apply to other
applications.
Usually, the design team knows what the main variables are, but does not know
the dimensionless numbers to use. The Buckingham theorem of dimensionless
variables, the so-called p theorem, is very common in engineering but less used in
other fields. The authors encourage the reader to study and apply the p theorem as a
means to turn redundant designs into uncoupled or decoupled designs.
10 Redundant Designs 289

The student may find more comprehensive approaches to this subject in the
theories regarding the design of experiments.
From the aforesaid, redundancy in nature exists. The following section intro-
duces, in a formal way, the concept of redundancy according to the view of AD.

10.3 The Theorem of Redundancy

According to AD, the ideal design is uncoupled (Theorem 4). If there are fewer DPs
than FRs, either the design is coupled, or the FRs cannot be simultaneously satisfied
(Theorem 1).
Equation (10.4) shows a redundant design with three FRs and four DPs. It is a
redundant design because DP2 and DP3 are used to fulfill FR2. This design equation
may express two different designs: a design that uses DP2 and DP3 at the same time to
fulfill FR2, and a design that uses DP2 on specific states and DP3 on other situations.
2 3
2 3 2 3 DP1
FR1 X 6 7
4 FR2 5 ¼ 4 X X 5  6 DP2 7 ð10:4Þ
4 DP3 5
FR3 X
DP4

On the second interpretation, the design equation is the amalgamation of the

states the system performs. This concept is similar to the one used on the
common-sense definition of redundancy. The hyperstatic structures in civil engi-
neering are examples where all DPs work at the same time. These examples show
redundant design of the aforementioned first type. On the other hand, a UPS (un-
interruptible power station) enters on service when the main fails, belonging to the
second kind of redundant designs.
We may now introduce the 3rd Theorem of AD. This theorem regarding
redundant designs states:

– “When there are more DPs than FRs, the design is either redundant or coupled.”

In a better assertion, Theorem 3 states:

– “When there are more DPs than FRs, the design is a redundant design, which can
be reduced to an uncoupled design or a decoupled design, or a coupled design.”
Exercise 10.4: Hybrid or Electric Car
The automotive industry has introduced innovative hybrid and electric cars to solve the
environmental problem created by CO2 emissions. Carbon dioxide (CO2) has been a
pollutant of concern due to the greenhouse effect it causes in the atmosphere. Electric cars
do not emit CO2 directly but increase the needs for electricity, the generation of which
releases carbon depending on the energy mix. To define the design equation, we start by
defining the following two high-level FRs:
290 M. Cavique and A. Gonçalves-Coelho

FR1 = Move the car;

FR2 = Control emissions.
Define the design equation for an electric car and a hybrid car and discuss how these
solutions fulfill these FRs. Discuss what might be a better solution with a higher probability
of success.

The theorem of redundancy may include two different realities: the reliability
motivated and the functionally motivated approaches. Section 10.4 shows this
reliability approach, classifying the redundancy in two groups: active and passive
redundancy. On the other hand, Sect. 10.5 discusses the functionally motivated
approach, which covers designs with different DPs to satisfy a range of FR that may
change over time.

10.4 Reliability-Motivated Redundancy

Reliability-motivated redundant designs are classified into two types: active and
passive redundant. If the design is an active one, then it has two or more states of
operation, and the redundant design equation can be split in so many equations as
the number of states of operation. Equation (10.410.4) may express a passive or an
active redundant design. Case Eq. (10.4) is an active redundant design it expresses
two states of operation, defined by two different design equations, as per Eq. (10.5).
2 3 2 3
2 3 2 3 DP1 2 3 2 3 DP1
FR1 X 6 DP2 7 FR1 X 6 7
4 FR2 5 ¼ 4 X 56 7 or 4 FR2 5 ¼ 4 X 5  6 DP2 7
4 DP3 5 4 DP3 5
FR3 X FR3 X
DP4 DP4
ð10:5Þ

Heating and cooling system

Figure 10.1 shows a system that heats a house and produces domestic hot water (DHW).
A fan-coil (FC) delivers the heat into the house, and the faucet represents the delivery of
DHW. In many countries, it is more cost-effective to use the heat pump rather than the gas
boiler. If the heat pump fails, or if it cannot work due to a very low outdoor temperature,
then the gas boiler starts, and the heat pump turns off.
This type of redundant designs can fulfill the FRs regardless of the working mode. Active
redundant designs are quite common in the energy field.

The designer intends to fulfill the range of acceptance for each FR at any mode of
operation.
10 Redundant Designs 291

Fig. 10.1 System for heating a room and for delivering domestic hot water

According to the example mentioned above, the FRs of the design are:
FR1 = Heat the house;
FR2 = Produce domestic hot water.
The system has a gas boiler (GB), a heat pump (HP), a solar collector (SC), and a storage
tank (ST). The hot water supply of the fan-coil comes from the GB or the HP, according to
the position of the three-way valve V1. Regarding the hot water production, it comes from
the ST, where two heat exchangers can heat the water coming from the mains supply.
The SC feeds the lower heat exchanger of the ST. Case the solar heat is not enough to heat
the water, then the HP or the GB feed the upper heat exchanger depending on the position
of the three-way valve V2.
Exercise 10.5: Design Equation of Fig. 10.1
Regarding Fig. 10.1 and using the above-defined FRs, find a design equation and all states
of the functioning of the system. Further, write the design equation of each state of
operation. Notice that, concerning the production of domestic hot water, the design may
have passive redundancy.

Any security system needs a redundant design. Usually, the security systems are
active redundant designs. Case a system fails, another system comes into operation.
The following example shows the case of electrical supply to a network operation
center:
292 M. Cavique and A. Gonçalves-Coelho

Electric supply to a data center

The electric power that feeds some data centers comes from an uninterruptible power
supply (UPS) that isolates the computers from the grid, the so-called galvanic insulation.
Batteries are the heart of the UPS and produce direct current (DC). DC is transformed into
alternating current (AC) that supplies the computers in the data center. The batteries have
autonomy for a certain period without external energy supply. The external supply comes
from the electrical grid (EG) or a diesel generator (DG). Therefore, the design equation of
the electrical supply of the data center is according to Eq. (10.6):

2 3
EG
½ES ¼ ½ X X X   4 UPS 5 ð10:6Þ
DG

The system may operate in three modes. The most common mode uses the supply of the EG to
the UPS and thus transforms the DC into AC; in case of a grid failure, the DG starts working
providing energy to the UPS; during the start-up of the DG the electricity supply comes only
from the UPS. In the case of total failure of the EG and DG, the UPS can supply the
data-center during a defined period. Equation 10.7 shows the three modes of operation:

2 3
EG
6 7
½ES ¼ ½ X X 0   4 UPS 5or
DG
2 3
EG
6 7
½ES ¼ ½ 0 X X   4 UPS 5 or ð10:7Þ
DG
2 3
EG
6 7
½ES ¼ ½ 0 X 0   4 UPS 5
DG

Exercise 10.6: Data-Center without Galvanic Isolation

Write the design equation of a data-center that receives electricity directly from the grid or
the diesel generator. This system uses UPS as a side supply during transition or total
failure. Discuss the fulfillment of energy supply in the three operating modes.

If the design is a passive redundant design, then all DPs are always in operation,
and no action changes the operating mode of the system.
For a steel structure to hold specific loads, it is possible to use a topology with a
minimum number of bars. However, many structures use more bars than the
minimum and distribute the loads through all of them. In such cases, it is necessary
to add the equations of displacement to the equations of the sum of forces and sum
of bending moments to allow defining the forces at each bar. These structures are
hyperstatic and very well known in civil and mechanical engineering. In case a bar
fails, the structure does not fail, at least immediately, as it happens in a
non-hyperstatic structure, the isostatic ones.
10 Redundant Designs 293

Fig. 10.2 A hyperstatic structure

Hyperstatic structure
Figure 10.2 shows a 2D hyperstatic structure with five loads, F, twelve nodes, an
anchoring node on the left, and a sliding node on the right. The loads are applied to the
nodes according to the picture so that each bar is subjected to compression or traction
loads, but not to bending.

Exercise 10.7: Hyperstatic Structure

Using the example of Fig. 10.2, define the minimum number of bars that allows supporting
the loads F. Define the design equation for the structure presented in the figure.

Fluid networks are other examples of redundant passive systems. In many water
networks, the ducts, which are topologically edges, form different paths from the
injection node to the consumption nodes. Therefore, it makes possible the flow to
vary in each edge depending on the consumption flows. Figure 10.3 depicts a
topology of a water network, with four nodes, two of them consumption nodes, and
one injection node. The network has five edges and forms two loops. The pressure
equilibrium in the loops makes the pressure to vary on the nodes depending on their
consumption. The pressure equilibrium causes the flow to vary in each duct along
the time. The network would need just two edges to supply the two nodes, for
example, edge 1 and 2. All other edges make the network to be a redundant design.
The student is invited to discuss why this type of networks is so prevalent in real
life.
In conclusion, regarding the reliability point of view, the design may be passive
or active. If the design has passive redundancy, then:

– all redundant parts are always on duty;

– no special action needed.

Regarding active redundancy, the design meets these characteristics:

– an action puts the redundant parts working;

– the modes of the system change over time.
294 M. Cavique and A. Gonçalves-Coelho

Fig. 10.3 A water network topology

Fig. 10.4 Deployment of forces during the Battle of Austerlitz. Figure sketched from public
domain maps of the Department of History Atlas (courtesy of the United States Military Academy
Department of History)
10 Redundant Designs 295

In both types of designs, the system aims to fulfill the FRs in the ranges defined
in the project. Therefore, the above examples assume that at any time, the range of
acceptance of the FRs remain constant.
However, the designer may want to change the FRs range over time. In such a
case, the FRs may have more than a target value and more than a range of
acceptance. These type of problems belongs to the classification of functionally
redundant design.

10.5 Functionally Related Redundancy

According to the AD theory, the acceptance of a solution depends on the fulfillment

of the FRs. Therefore, the acceptance of the design occurs in the functional domain.
Intrinsic redundancy, discussed at the beginning of this chapter, is inherently
related to the functions and therefore belongs to the functionally related type of
redundancy. In such a case, the functions to perform, their target value, and range of
acceptance remain.
Other designs need to change their target value and their range of acceptance of
the FRs over time, called “alternative redundancy.” In this way, an alternative
redundant design is always an active redundant design.
Therefore, regarding functional redundancy, the classification of designs is:

– intrinsic redundancy;
– alternative redundancy;
– adaptive, or augmentative, redundancy.
PTO/PTI Marine Engines
When a vessel docks, the main engine usually stops to reduce energy consumption, as well
as to reduce pollution in the port. At this working state, the energy supply to the vessel
comes from the dock infrastructure. However, for safety reasons, the vessel might need to
set sail urgently. As the main engine takes some time to start and cannot drive the propeller
in time according to the urgency of the maneuver, a quicker engine needs to start.
Many vessels use the concept of PTO/PTI to maneuver on time. During regular operation,
the main engine drives the propeller and supplies energy to a generator by engaging a
clutch to move its shaft. In this case, the generator charges a set of electric batteries. This
situation is called the power take-off, or PTO mode. The PTI mode, power take in, occurs
during an emergency start. Some port regulations enforce the use of PTI mode for safety
and pollution reasons. In this case, the electrical batteries power an electric motor that
connects to the propeller shaft by another clutch.
The generator and the motor are usually the same device that works as a generator when it
receives mechanical energy and works as a motor when it receives electrical energy.

The last example is an alternative redundant design. Based on the classification of

Sect. 10.4, it is an active redundant design.
296 M. Cavique and A. Gonçalves-Coelho

The FR is “produce shaft power,” and the DPs are the main engine, usually a
Diesel motor (DP1), the electric motor system (DP2), and the alternator system
(DP3). The electrical motor system includes the engine, clutch, shaft, electric
supply, and all other equipment and control systems needed to make it work.
Similarly, the alternator system contains the alternator, clutch, batteries, electrical
system, and control.
The design equation is according to the following Eq. (10.8):
2 3
DP1
½FR ¼ ½ X X X   4 DP2 5 ð10:8Þ
DP3

This equation has two states of operation defined in the following equation,
where FRPTO regards the propulsion by the diesel engine at PTO mode and FRPTI,
the propulsion through the electrical motor. In Eq. (10.9), we suppose FRPTO and
FRPTI may have different target values and different ranges of acceptance:
2 3 2 3
DP1 DP1
½FRPTO  ¼ ½ X X   4 DP2 5 or ½FRPTI  ¼ ½ X   4 DP2 5 ð10:9Þ
DP3 DP3

The design is a redundant design, classified as alternative regarding functional

redundancy, and active regarding the reliability point of view. As FRPTO and FRPTI
have different ranges and target values, then the design is adaptive.
The system may also work with the main engine only, FRD, without delivering
energy to the alternator.
In some arrangements, the main engine, as well as the electrical motor, may
contribute to the shaft power. In this case, the shaft power may add the Diesel
engine and electric power, called combined diesel and electric (CODAE), the
design is adaptive. Equation (10.10) shows the four possible states of this
arrangement: diesel only, PTO, PTI, and both diesel and electrical motor:
2 3
DP1
6 7
½FRD  ¼ ½ X   4 DP2 5 or
DP3
2 3
DP1
6 7
½FRPTO  ¼ ½ X X   4 DP2 5 or
DP3
ð10:10Þ
2 3
DP1
6 7
½FRPTI  ¼ ½ X   4 DP2 5 or
DP3
2 3
DP1
6 7
½FRDE  ¼ ½ X X   4 DP2 5
DP3
10 Redundant Designs 297

Exercise 10.8: New Definition of the Design Equation for the PTO and PTI Arrangements
Last example shows a redundant design for the FR “produce shaft power.” Redo this
example by considering an adaptive redundant design, using the following two FRs:
FR1 = Produce shaft power;
FR2 = Produce electricity to the service grid.

Using the design equation, discuss the possible working modes and identify the
design matrix for each mode of operation.
Blue Angels
The Blue Angels squadron is a flying aerobatic squadron of the United States Marines
formed in 1946. One of the planes of the squadron is a Lockheed Hercules C-130 with
rock-assisted take-off (RATO) capability. During take-off, the plane uses the four
turbo-propellers plus rockets to increase thrust. Therefore, this arrangement is an adaptive
redundant design that increases power during take-off.

The last example addresses an adaptive redundant design. To fulfill part of the
range of the FR: “produce thrust power” the Blue Angels Lockheed C-130 uses four
turboprops, and the extra power needed at take-off comes from rockets. In this
situation, the FR definition remains, and more DPs allows the system to be able to
fulfill the whole working range.
The student might be aware when facing a redundant design if the design lacks
the definition of some FR. Known examples are very important to discuss the
application of a theory, but knowing the DPs may cause a misunderstanding of the
targets of the system. Next exercise asks the student to discuss this problem.
Exercise 10.9: The Battle of Austerlitz
The Battle of Austerlitz opposed Napoleon Bonaparte to the Austro-Russian coalition
commanded by the Russian czar Alexander. This battle is a masterpiece of strategy that
offered one of the greatest victories of Napoleon. He could defeat a larger army and affirm
the new French empire. On the 1st December 1805, Napoleon had his forces near the
Pratzen Heights, which he abandoned some days before to make the enemy feel he was in a
weak condition. Moreover, he purposely weakened his right flank at Telinitz.
The Russian commander decided to attack the right flank of Napoleon to roll up the French
line and cut the French contact with Vienna. However, Napoleon has ordered Marshal
Davout to do a risky and crucial maneuver: to march during 48 h from Vienna to help his
right flank. Davout was one of the best French marshals and was able to accomplish this
movement, just in time to support General Legrand at the right flank. Meanwhile, the
Austro-Russian army has left their forces at the Pratzen Heights, which allowed the French
Marshal Soult to make a frontal attack and cut the Austro-Russian forces into two pieces.
Napoleon said: “If the Russian force leaves the Pratzen Heights to go to the right side, it
will certainly be defeated.”
298 M. Cavique and A. Gonçalves-Coelho

Napoleon may have had in mind the following two high-level FRs:

– persuade the Austro-Russian forces to attack the French flank;

– attack the center of the Austro-Russian army.
The DPs were the forces of Legrand, Davout, and Soult.
Is it a redundant design? Define the design equation of Napoleon’s strategy.

10.6 More Theorems on Redundancy

AD’s Theorem 3 says a redundant design “can be reduced to an uncoupled design

or a decoupled design, or a coupled design”. This section shows how to classify
redundant designs and define new theorems.
Equation (10.11) shows a redundant design with three FRs and five DPs. The
subset of the design matrix regarding DP1, DP2, and DP3 shows a coupled design.
Anyway, the set of DP1, DP2, and DP3 allows fulfilling FR1, and then DP4 can tune
FR2 and DP5 tunes FR3. This equation shows the zero elements of the matrix to
improve matrix readability:
2 3
2 3 2 3 DP1
FR1 X X X 0 0 6 7
6 DP2 7
4 FR2 5 ¼ 4 X X X X 0 56 DP 7 ð10:11Þ
6 37
FR3 X X X X X 4 DP4 5
DP5

The sequence for tuning the FRs is DP1 + DP2 + DP3, DP4, and finally, DP5,
showing it is a decoupled design.
Theorem R1 states:

– All redundant designs with right-trapezoid design matrix are decoupled.

Next section uses this theorem to decouple coupled designs, showing some
examples and applications.
If the design matrix has diagonal blocks, then the designs matrixes are similar to
Eq. (10.12) and
Theorem R2 applies:

– Redundant designs with design matrices composed by contiguous diagonal

blocks are uncoupled.

In this case, DP1 + DP4 fulfill FR1; DP2 + DP5 fulfill FR2, and DP3 fulfills FR3.
The sets of DPs can fulfill an FR independently, making the design redundant and
uncoupled. Notice that on any column of the matrix appears just one “X”.
10 Redundant Designs 299

2 3
2 3 2 3 DP1
FR1 X 0 0 X 0 6 7
6 DP2 7
4 FR2 5 ¼ 4 0 X 0 0 X 6 DP3 7
5 6 ð10:12Þ
7
FR3 0 0 X 0 0 4 DP4 5
DP5

If the matrix has triangular blocks, then Theorem R3 applies:

– Redundant designs with design matrices composed by contiguous triangular

blocks are decoupled.

Equation (10.13) shows an example of theorem R3, with a redundant design

formed by two decoupled designs. In this example, the sequence of tuning is
DP1 + DP4, thus DP2 + DP5, and lastly, DP3.
2 3
2 3 2 3 DP1
FR1 X 0 0 X 0 6 7
6 DP2 7
4 FR2 5 ¼ 4 X X 0 X X 6 DP3 7
5 6 ð10:13Þ
7
FR3 X X X 0 0 4 DP4 5
DP5

Using both triangular and diagonal matrices, Theorem R4 states:

– Redundant designs with design matrices composed by contiguous diagonal and

triangular blocks are decoupled.

Equation (10.14) shows an application of Theorem R4. Notice the sequence of

tuning of Eqs. (10.13) and (10.14) is the same, but FR2 does not depend on DP4,
which allows a wider tolerance for DP4 in the design expressed by Eq. (10.14).
2 3
2 3 2 3 DP1
FR1 X 0 0 X 0 6 7
6 DP2 7
4 FR2 5 ¼ 4 X X 0 0 X 56 DP 7 ð10:14Þ
6 37
FR3 X X X 0 0 4 DP4 5
DP5

If it is necessary to create a redundant design, it is easy to use already available

systems. On Exercise 10.5, the heat pump, boiler, and solar panel helped to define a
redundant new system. We apply this idea in the next theorems that help to syn-
thesize new systems. Therefore, it is possible to raise new in the following
paragraphs.
Theorem R5 states:

– The combination of an arbitrary number of uncoupled designs with common FRs

and unshared DPs is again an uncoupled design.
300 M. Cavique and A. Gonçalves-Coelho

As an example, having the following three designs defined by their design equa-
tions (10.15), it is possible to create a redundant design according to Eq. (10.16):
2 3 2 3 2 3
FR1 X 0 0 DP1
6 7 6 7 6 7
4 FR2 5 ¼ 4 0 X 0 5  4 DP2 5 and
FR3 0 0 X DP3
     
FR1 X 0 DP4 ð10:15Þ
¼  and
FR2 0 X DP5
     
FR2 X 0 DP6
¼ 
FR3 0 X DP7

The three designs create a new redundant design:

2 3
DP1
6 7
2 3 2 3 6 DP2 7
FR1 X 0 0 X 0 0 0 6 DP3 7
6 7
4 FR2 5 ¼ 4 0 X 0 0 X X 056 7 ð10:16Þ
6 DP4 7
FR3 0 0 X 0 0 0 X 6 DP5 7
6 7
4 DP6 5
DP7

To fulfill each FR, the new design expressed by Eq. (10.16) has two or more
DPs. The designer might be aware of the interconnections of the DPs not to create a
coupled design at lower levels of decomposition. In many circumstances, the
interconnection couples the systems and may create a coupled design.
Theorem R6 states:

– The combination of an arbitrary number of decoupled designs with common FRs

and unshared DPs is again a decoupled design.

Equation (10.17) and (10.18) show an example of matching three existing

decoupled designs into a redundant decoupled design:
2 3 2 3 2 3
FR1 X 0 0 DP1
6 7 6 7 6 7
4 FR2 5 ¼ 4 X X 0 5  4 DP2 5 and
FR3 X X X DP3
     
FR1 X 0 DP4 ð10:17Þ
¼  and
FR2 X X DP5
     
FR2 X 0 DP6
¼ 
FR3 X X DP7
10 Redundant Designs 301

Create:
2 3
DP1
6 7
2 3 2 3 6 DP2 7
FR1 X 0 0 X 0 0 0 6 7
6 DP3 7
4 FR2 5 ¼ 4 X X 0 X X X 0  6 DP4 7
5 6 ð10:18Þ
7
FR3 X X X 0 0 X X 6 DP5 7
6 7
4 DP6 5
DP7

Finally, Eqs. (10.19) and (10.20) show the existing designs and the redundant
design created by Theorem R7, which states:

– The combination of an arbitrary number of uncoupled and decoupled designs

with common FRs and unshared DPs is again a decoupled design.
2 3 2 3 2 3
FR1 X 0 0 DP1
6 7 6 7 6 7
4 FR2 5 ¼ 4 X X 0 5  4 DP2 5 and
FR3 X X X DP3
     
FR1 X 0 DP4 ð10:19Þ
¼  and
FR2 0 X DP5
     
FR2 X 0 DP6
¼ 
FR3 0 X DP7
2 3
DP1
6 7
2 3 2 3 6 DP2 7
FR1 X 0 0 X 0 0 0 6 DP3 7
6 7
4 FR2 5 ¼ 4 X X 0 0 X X 056 7 ð10:20Þ
6 DP4 7
FR3 X X X 0 0 0 X 6 DP5 7
6 7
4 DP6 5
DP7

This section helps the student to classify a redundant design using Theorems R1
to R4. Moreover, it helps synthetizing redundant designs using existing systems by
using Theorems R5 to R7. A corollary of Theorem R1 can help to solve coupled
systems. Due to the importance of solving coupled design, a complete section
concerns this matter. Following section addresses this problem, giving some
examples to illustrate the application.
302 M. Cavique and A. Gonçalves-Coelho

10.7 Solving Coupled Designs

Theorem R1 states that “all redundant designs with right-trapezoid design matrix
are decoupled” no matter the classification of the square part of the design matrix.
Therefore, it is possible to join a right diagonal or right triangle matrix to any design
matrix to create a right-trapezoid matrix.
Thus, Theorem R8 states:

– Any coupled design can be reduced to a redundant decoupled design by joining

an upper triangular matrix with unshared DPs.

Therefore, if a design is a coupled design, as shown in the example of

Eq. (10.21), then it is possible to reduce it to a decoupled design.
2 3 2 3 2 3
FR1 X X X DP1
4 FR2 5 ¼ 4 X X X 5  4 DP2 5 ð10:21Þ
FR3 X X X DP3

Knowing “what we want to achieve,” the designer may identify new unshared
DPs than can fulfill a certain FR. In Eq. (10.22), DP4 and DP5, new unshared DPs,
fulfill FR2 and FR3, respectively. Therefore, knowing a working mode of DP1 +
DP2 + DP3 that fulfills FR1 makes it possible to tune the other FRs. Therefore, the
sequence of tuning is freezing DP1, DP2, and DP3, and then set the unshared DPs.
2 3
2 3 2 3 DP1
FR1 X X X 0 0 6 7
6 DP2 7
4 FR2 5 ¼ 4 X X X X 0 6 DP3 7
5 6 ð10:22Þ
7
FR3 X X X 0 X 4 DP4 5
DP5

The de-carbonization of the world economy

Most of the World countries agreed in the 2015 Conference of Paris to reduce their carbon
emissions. Moreover, they decided to apply the best practices to keep global warming”well
below 2 °C” of the pre-industrial era. According to the International Panel on Climate
Change (IPCC) of the United Nations, the World needs to perform a hard path to reduce
carbon in the atmosphere. Many evolutions of future carbon emissions have been studied
and cataloged according to the average radiative forcing. The representative concentration
path (RCP2.6) allows the temperature of the earth's atmosphere to increase in the range of
0.9 to 2.3 °C until 2100. This RCP requests to reduce human emissions to values well
below the ones experienced in 1980. This path needs a substantial increase in the use of
renewable energy as well as nuclear energy. To ensure the needs of energy for 2100 while
reducing the carbon emissions, the technologies using coal, natural gas, and bio-energy,
must be helped by technology for carbon capture and storage (CCS).
10 Redundant Designs 303

In the context of the last example, the thermoelectric generators (TEG) are facing a
difficult challenge. There has been a huge effort to increase the efficiency of the
TEGs and to substitute them with cleaner technologies. However, according to the
RCP2.6, coal TEGs will be necessary for the future, due to the increase in energy
demand.
The FRs for a coal thermoelectric power plant are:

– FR1 = Adjust power production;

– FR2 = Reduce carbon emissions.

A central power station starts with a certain number of thermoelectric generators

of specific power and adjusts the power production at any time by changing the
number of generators in operation. The DPs are:

– DP1 = Power of a thermoelectric generator (PTEG);

– DP2 = Number of thermoelectric generators (n).

Equation (10.23) shows the design equation. The second row harms the
reduction of emission, making the “Xs” of this row to be negative.
     
FR1 X X PTEG
¼  ð10:23Þ
FR2 X X n

The above FRs and DPs solve the demand for energy, but the reduction of
carbon emission makes to reduce the number of TEG in a power plant. Therefore,
on a given power plant, carbon emission depends strongly on power production.
How can the World solve this problem? According to the IPCC, CCS is a solution
by changing the design into a redundant design, described by Eq. (10.24). CCS
gives an extra possibility to lower the carbon emissions to the atmosphere despite
the production of carbon on the power plants.
2 3
    PTEG
FR1 X X
¼ 4 n 5 ð10:24Þ
FR2 X X X
CCS

Exercise 10.10: The Energy-Mix Equation

The RCP2.6 asks for using by 2020 an energy-mix of carbon resources (coal, oil, natural
gas), bio-energy, renewable, and nuclear sources. Also, there is a need to use a technology
for carbon capture and storage. Write the energy-mix equation using the FRs of the
previous example:
- FR1 = Adjust power production;
- FR2 = Reduce carbon emissions.
304 M. Cavique and A. Gonçalves-Coelho

Redundant designs are more common than we may realize. In this chapter, most of
the examples considered energy supply or safety, but many situations in the
day-to-day work use redundant designs.
The two-visa problem
A person goes to a country where it is necessary to travel with a visa in the passport.
However, that person was urgently asked to travel to a second country by the end of the
month, requiring a new visa added to the passport. It is necessary to leave the passport to
apply for a new visa, but during that month, the person will need it for traveling. However,
in such cases, the foreigner affairs services allow a person to have two passports. However,
it is necessary to present the booking of the hotel and the flight tickets to receive a second
passport. Nevertheless, it is no sense to pay for the hotel and plane without knowing if the
passport visa would arrive on time. So, how would you solve this problem?
Exercise 10.11: The Two-Visa Problem
Write down what the requirements of the passenger are and define the design equation.
Then, look for a solution using a redundant design.

Managers often use redundant designs to solve organizational problems. With time,
organizations tend to slow down the fulfillment of their original stated require-
ments. High-level decision-makers or managers introduce new requirements in the
organization to fulfill new needs or hidden personal agenda. Redirecting the
organization to perform the original requirements is often a traumatic experience for
many persons. A way to solve this problem is by using a redundant design.

10.8 Conclusions

When there are more DPs than FRs, the design is either a redundant design or a
coupled design (as per Theorem 3). Redundant designs are common on energy
grids applications, on safety systems, and many applications of real life.
This chapter identifies three kinds of redundancy: adaptive, alternative, and
intrinsic.
Moreover, this chapter presents theorems on redundancy: theorems R1 to R8.
All redundant designs with a right-trapezoid design matrix are decoupled (as per
Theorem R2). Redundant designs with design matrices composed by contiguous
diagonal blocks are uncoupled (as per Theorem R2).
Redundant designs with design matrices composed by contiguous triangular
blocks are decoupled (as per Theorem R3). Four base theorems (R1 to R4) help to
classify uncoupled and decoupled redundant designs. Redundant designs with
design matrices composed by contiguous diagonal and triangular blocks are
decoupled (as per Theorem R4).
10 Redundant Designs 305

Moreover, Theorems R5 to R8 give a framework for creating redundant designs.

Linking two uncoupled or decoupled designs with unshared DPs create new
uncoupled or decoupled designs. Finally, theorem R8 helps to solve coupled
designs. Therefore, it is a powerful tool for AD applications. It states that case the
design is coupled, then it is possible to turn it into a decoupled design by adding
DPs so that the design matrix becomes right-trapezoidal.

References
United States Military Academy, Department of History (2020a) The Battle of Austerlitz, 1805—
Situation, 0900 h, 2 December. https://www.westpoint.edu/sites/default/files/inline-images/
academics/academic_departments/history/Napoleonic%20wars/Nap23.pdf
United States Military Academy, Department of History (2020b) The Battle of Austerlitz, 1805—
Situation, 1800 h, 1 December. https://www.westpoint.edu/sites/default/files/inline-images/
academics/academic_departments/history/Napoleonic%20wars/Nap22.pdf

Further Reading

Gonçalves-Coelho A, Neştian G, Cavique M, Mourão A (2012) Tackling with redundant designs

through axiomatic design. Int J Precis Eng Manuf 13(10):1837–1843
Suh NP (1990) The principles of design. Oxford University Press
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
The Information Axiom and Robust
Design 11
João Fradinho and António Gonçalves-Coelho

Abstract
In preceding chapters, it was emphasized that the first task in innovative design
is the problem definition. The problem is then transformed into functional
requirements (FRs), then into design parameters (DPs) and process variables
(PVs), the vectors that define the four domains of the design world. Design
matrices give the relationship between these vectors. The Independence Axiom
states that the FRs must always be independent of other FRs as DPs and PVs are
chosen. In this chapter, the role of the Information axiom in creating the best
design is presented, including how to measure the information content, how to
make a design robust by satisfying the Independence Axiom and the Information
Axiom. A coupled design has large information content, in some cases,
approaching infinite information content, which implies that the design will
never work as intended. An uncoupled design can be made to have zero
information content, making the design easy to implement and the resulting
design most reliable. The determination of information content is presented in
this chapter with many examples.

List of abbreviations

AD Axiomatic Design
FR Functional Requirement (FRs)
DP Design Parameter
PV Process Variable

J. Fradinho (&)
A. Gonçalves-Coelho

UNIDEMI & DEMI, NOVA SST, Campus de Caparica, 829-516 Caparica, Portugal
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 307

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_11
308 J. Fradinho and A. Gonçalves-Coelho

11.1 Introduction

Axiomatic Design consists of a simple and logical general design framework to deal
with the design of products, services, or organizations. It provides designers with
criteria to support decision-making along with all the steps of the designing process.
Figure 11.1 depicts a simple workflow for the AD framework.

Generated solutions

Yes

Conforms to Redesigned
No No Discarded
Independence? solutions? solutions
(Axiom I)

Yes

There are more No

than one design Accepted solution
solution?

Yes

Is it clearly
Is the information No
the simplest solution? No Discarded
content computation (qualitative solutions
possible? evaluation)

Yes Yes

It has the smallest No Discarded

information content? solutions
(Axiom II)

Yes

Elected solution Elected solution

Fig. 11.1 An Axiomatic Design workflow

11 The Information Axiom and Robust Design 309

True Story: “Poor” Ph.D. Candidate!

Sometimes academic research and academic career can be challenging in many different
ways, especially to be a graduate student working on what appears to be a simple idea!
Sometimes, it can be more challenging to advance “simple ideas” than solve some
“complex problems.” Appearance can be deceiving in many fields, but especially in
academia!

When a young professor was invited to join the faculty of a renowned technological
university, he was overjoyed. The only problem was that the salary offer was much less
than what he had been receiving at the current university he had been teaching for a few
years. Furthermore, the housing cost and the tax on the house in the region where the
technical university was located was at least double the cost of where he was. With three
small children to educate, the offer was not as attractive as it first appeared. However, his
wife thought that it would be a good thing for children’s education to go to the more
expensive area and suffer financially! She had always been right on these matters, so they
left a comfortable place and went to a more challenging environment.

When he got to the new university, there were many prominent professors in the depart-
ment. Many of them were his professor when he was a student there a few years back.
Furthermore, to earn his tenure, he could not offend too many of these senior professors.
However, that was the first thing this young professor did! He came up with a view that the
design education that had been taught at the university, which had been well recognized
throughout the world, needs to be improved! The eminent professor in design and his
younger professors in design believed in the notion that “design can be learned only
through experience and no lectures and research are needed.” Therefore, when this new
young professor proposed a contrary view, his new view was not well received, not only at
this university but also throughout the national and international community of traditional
design professors! Many openly criticized him. Since he had not yet had a tenured
appointment, he told his wife that they should be mentally prepared to seek another job.
Fortunately, the primary national funding agency for research liked his new idea on design
and gave substantial funding to his design research project. Fortunately, he also got his
tenure soon after he joined this new university—before the mandatory tenure date.
Therefore, he could pursue his idea for a new design approach without worrying about his
job!

Four graduate students were hired to work on AD. They all suffered since there was no
precedent for the research they were conducting. One Ph.D. student was dealing with the
question: “what is INFORMATION?” The second Axiom stated that the information
content should be minimized.” He and his professor struggled with this question of in-
formation content and other related questions. They could not even state clearly the
meaning of “information content.” They struggled with the question: “Is there more
information in a square than in a circle?”

Somehow, this first Ph.D. student put together a doctoral thesis, which his advisor thought
was good enough for his doctorate and should get his degree. The fact that he and his wife
wanted to raise a family and that he had received an excellent job offer also weighed in on
the decision of his advisor.

On the day of his thesis examination, the room was full of design professors. After the
presentation, the senior professor in the design section of the department refused to pass
him based on what the student presented, although his advisor explained the difficulty of his
thesis topic! The objection was more on the research the young professor was conducting
than what the graduate student had done! There was a shouting match between the Ph.D.
candidate and the traditional design professor.
310 J. Fradinho and A. Gonçalves-Coelho

The faculty vote was against accepting the doctoral thesis by this student. The student
stayed another six more months and finally passed the thesis examination. The contribution
of his thesis was that it established the problematic questions the group had to answer
related to the second axiom, i.e., “Minimize the information content.” The notion that the
information in design is related to tolerances versus the nominal dimension was advanced,
which was a promising beginning in formalizing the second axiom. Also, the idea that the
information is related to the “design range” and the “system range” was advanced. Once
these ideas were crystallized, much progress has been made, but the first Ph.D. student had
to take the brunt of criticisms. Some people compared what this first Ph.D. student had gone
through to what happens to the first-born piglet. A pig gives birth to many piglets at the
same time, but the first piglet suffers the most to come out of the womb! The professor still
thinks that he did a great job! Later many other researchers joined the AD research group
and made many important contributions, including visiting professors, Professors H.
Nakazawa and Professor G. Solenius.

The length of the Shore Line of Australia!

Sometimes, it is complicated to answer simple questions related to information, especially

those related to geometry. For example, what is the length of the shoreline of Australia?
How much information do we need to specify the length of the shoreline? Is it difficult to
answer it? Why?

The difficulty is that although we draw shorelines around Australia, it is only approximate
because what is presented as a straight line is made up of jagged lines. If we look at the
jagged lines in higher magnification, they are made of more jagged lines. Mathematicians
call it a “fractal,” a subset of Euclidean geometry. Fractal appears to be nearly the same at
different levels of magnification. However, the area does not change much although the
actual length can be much longer. Likewise, when we determine the information content
related to FRs and DPs, we have to define the tolerance we need to satisfy within which we
define these quantities. That is the functional independence is defined by the tolerance of
interference we are willing to tolerate. Similar comments apply to DPs and PVs.

Similarly, when we specify FRs, DPs, and PVs, we have to define the tolerance within
which we must satisfy them.

As it was shown in the previous chapter, the functional independence clearly dis-
tinguishes the preferred design alternatives. This is the case of the uncoupled and
the decoupled design solutions, while coupled solutions shall be discarded or
reformulated to attain functional independence. The Independence Axiom is
therefore an early acceptance criterion for alternative solutions.
When more than one suitable design alternative exists, then the question that
arises is: which solution is the best one? The answer can be found through the
Information Axiom.
The process terminates after a number of iterations, according to the designer’s
criterion.
11 The Information Axiom and Robust Design 311

11.2 The Information Axiom

The Information Axiom provides the last criterion in the choice of the best design
solution. It can be stated as (Suh 2005, p. 23):
Minimize the information content of the design;

or, alternatively, as Park (2007, p. 18):

The best design is a functionally uncoupled design that has minimum information content;

or yet as Gonçalves-Coelho et al. (2005):

In a set of designs that satisfy the same FRs and conform to the independence axiom, the
best is the one with the minimum information content.

The Information Axiom lays on the information content, similar to that defined by
Shannon (1948), which is based on the probability of success of each solution. For a
single FR design, the mathematical expression of the information content, I, is:


1
I ¼  log2 p ¼ log2 ð11:1Þ
p

where p is the probability of success of achieving a given FR.

Therefore, the information content of a design is null when its success is always
guaranteed. For any other value of the probability between 0 and 1, the information
content will be a positive number.
If the probability density function (pdf) of the existing FR is known, then the
probability of accomplishing the FR can be computed by a quotient of areas, as
shown in Fig. 11.2.

Fig. 11.2 The system range, the design range, and the common range
312 J. Fradinho and A. Gonçalves-Coelho

In Fig. 11.2, the system range represents the whole ability of the system, while
the design range represents the working range that the designer is looking for. The
common range is the intersection of the system and design ranges.
Most of the designs have a set of FRs that are satisfied by a set of DPs. The
information content of those designs is a measure of the probability of the simul-
taneous success of all the FRs. Then, the information content of a design with
m FRs is given by:

X
m
Itotal ¼ Ii ð11:2Þ
i¼1

The probability of success, pi, of satisfying FRi is expressed by:

Z
pi ¼ pðFRi ÞdFRi ð11:3Þ
design range

or, graphically, by:

Area Common Range

pi ¼ ð11:4Þ
Area System Range

In some cases, a uniform distribution is assumed as an approximation.

11.3 Independence and Information Content

There are two types of designs where independence is fulfilled: uncoupled designs
and decoupled designs. Uncoupled designs are always preferable to decoupled
designs. Theorems 6 and 7 (Suh 2005, p. 46) relative to path dependency apply.
Theorem 6—The information content of an uncoupled design is independent of the
sequence by which the DPs are changed to satisfy the given set of FRs.

Theorem 7—The information contents of coupled and decoupled designs depend on the
sequence by which the DPs are changed to satisfy the given set of FRs.

In uncoupled designs, each DP only affects one FR. As a consequence, the order in
which they are attained is irrelevant and the information content of the design Itotal,
is the algebraic sum of the individual information content, Ii, of the m FRs:
!
X
m X
m Y
m
Itotal ¼ Ii ¼  log2 pi ¼  log2 pi ¼  log2 ptotal ð11:5Þ
i¼1 i¼1 i¼1
11 The Information Axiom and Robust Design 313

Example 11.1
Consider a game that consists of throwing a dice once and withdrawing a card from a
52-card deck. A prize is awarded to anyone who has a “six” in the dice and a “king” in the
cards. What is the probability of success? What is the information content of the designed
game (in the player viewpoint)?

Solution: The result of the game is made of two independent events which order
does not matter. Therefore:

1 4 1
ptotal ¼ p1  p2 ¼  ¼
6 52 78

and


1
I ¼  log2 ’ 6; 29
78

Example 11.2
The owner of a small coffee shop intends to provide customers with a new game consisting
of the sequential throwing of two darts on a single target. Players earn a prize when they hit
both darts in the center of the target. There are two alternatives (A and B) for the shape of
the target, as shown in Fig. 11.3. Considering that the players always hit the target, what is
the best game in the owner’s viewpoint?

Solution: In both cases, the two events (throwings) are independent. So, their
probabilities of success are:

Areacenter A ¼ 802 ¼ 6400 Areacenter B ¼ p  502 ¼ p  2500

Areatotal A ¼ 3002 ¼ 90000 Areatotal B ¼ p  2002 ¼ p  40000

Alternative A Alternative B

100 mm
80 mm

300 mm 400 mm

Fig. 11.3 Two alternatives for an arrow launching game

314 J. Fradinho and A. Gonçalves-Coelho

6400 p  2500
pA ¼ ’ 0; 0711 pB ¼ ¼ 0; 0625
90000 p  40000

The coffee shop would have better profitability with alternative B because
alternative A is easier for the players.
Decoupled designs are designs where independence is also assured, but there is
one only order to fulfil the FRs without the need of reiteration. Consider the 2-FR,
2-DP decoupled design equation presented below:
    
FR1  0 DP1
¼ ð11:6Þ
FR2   DP2

To satisfy both FRs at the first attempt, one should begin by choosing the value
of DP1 as to satisfy FR1. The chosen value of DP1 also affects FR2, which must be
then fulfilled by choosing the value of DP2. The information content of this design
depends on the sequence by which DP1 and DP2 are chosen to satisfy FR1 and FR2.
The satisfaction of FR1 through DP1 has a certain probability, p1, but the satis-
faction of FR2 through DP2 only occurs after satisfaction of FR1 trough DP1. This is
a case of conditional probability. The probability of satisfying FR2 is
pð1 \ 2Þ
pð2n1Þ ¼ ð11:7Þ
p2

and the probability of success of the design is

ps ¼ p1  pð2n1Þ ð11:8Þ

The probability of success of a decoupled design with m FRs and the same
number of DPs, is

ps ¼ p1  pð2n1Þ  pð3n2Þ  . . .  pðmnm1Þ ð11:9Þ

It should be noted that in practice, the conditional probability is difficult to

determine during the design phase.
Example 11.3
Consider that a given design requires that a body with mass, m, to slide on an inclined plane
that makes a h angle relative to the horizontal. The body should reach certain acceleration,
a, and should exert a normal force on the plane, FN, within the following limits:

3 ms2  a  5 ms2 ð11:10Þ

0:5 N  FN  4 N ð11:11Þ

The designer has two alternative devices, A and B, embodied by a mass, m, and by a plan
with the slope angle h, within the following ranges:
11 The Information Axiom and Robust Design 315

Device A : 0:1 kg  mA  0:5 kg ð11:12Þ

10  hA  30 ð11:13Þ

Device B : 0:5 kg  mB  2 kg ð11:14Þ

20 
 hB  60 
ð11:15Þ

Which device is the preferable one? Which one has the smaller information content?

Solution: In the context of this example, Eqs. (11.10) and (11.11) describe the
design range and Eqs. (11.12), (11.13), (11.14), (11.15) denote the system ranges of
the two devices. Figure 11.4 depicts the vector diagram of the devices.
The relations between the FRs a and FN, and the DPs m and h, are given by

a ¼ g sin h ð11:16Þ

FN ¼ mg cos h ð11:17Þ

where g is the gravitational acceleration. For the sake of simplicity, its value will be
considered as 10 ms−2.
If there is no friction, the design equation is
    
a  0 h
¼ ð11:18Þ
FN   m

Equation (11.18) depicts a decoupled design, where a definite order in the ful-
fillment of the FRs must be followed by virtue of the independence axiom: First,
selecting the value of h determines a, then selecting the value of m, together with
the previously chosen value of h, determines FN. Thus, the total probability of
success, ps, is given by

Fig. 11.4 Vector diagram of

the devices

θ
316 J. Fradinho and A. Gonçalves-Coelho

ps ¼ p1 p2n1 ð11:19Þ

For the device A,

aAmin ¼ 10  sin 10 ’ 1:74 ms2

aAmax ¼ 10  sin 30 ¼ 5 ms2

53
p1A ¼ ’ 0:613
5  1:74

The common range varies between 3 ms−2 and 5 ms−2. So, these are the values
for a to be considered for the normal force, FN. An acceleration of 3 ms−2 corre-
sponds to an angle of 17.4º.

FNAmin ¼ mmin g cos hmax ¼ 0:1  10  cos 30 ’ 0:866 N

FNAmax ¼ mmax g cos hmin ¼ 0:5  10  cos 17:4 ’ 0:470 N

0:866  0:5
p2n1A ¼ ’ 0:924
0:866  0:470

pA ¼ p1A p2n1A ’ 0:613  0:924 ’ 0:566 IA ’ 0:82

For the device B,

aBmin ¼ 10  sin 20 ’ 3:42 ms2

aBmax ¼ 10  sin 60 ’ 8:66 ms2

5  3:42
p1B ¼ ’ 0:302
8:66  3:42
The common range varies between 3.42 ms−2 and 5 ms−2. These are the values
for a to be considered for the normal force, FN. An acceleration of 5 ms−2 corre-
sponds to an angle of 30º.

FNBmin ¼ mmin g cos hmax ¼ 0:5  10  cos 60 ’ 2:5 N

FNBmax ¼ mmax g cos hmin ¼ 2  10  cos 30 ¼ 10 N

4  2:5
p2n1B ¼ ¼ 0:2
10  2:5

pB ¼ p1B p2n1B ’ 0:302  0:2 ’ 0:060 IB ’ 4:06

11 The Information Axiom and Robust Design 317

Device A has a higher probability of success and consequently the smaller

information content. Accordingly, device A is the best alternative solution.

11.4 The Computation of Information Content

of Decoupled Designs Through Graphical Methods

The computation of the information content of decoupled designs is a hard task.

A graphical method to deal with 2-FR, 2-DP decoupled designs with uniform
probability density FRs was developed by Suh (2001, p. 528) and Park (2007,
p. 38).
For this case, the random variations of the FRs due to the random variations of
the DPs are given by
     
dFR1 A11 0 dDP1 dFR1 ¼ A11 dDP1
¼ , ð11:20Þ
dFR2 A21 A22 dDP2 dFR2 ¼ A21 dDP1 þ A22 dDP2

Equation (11.20) shows that dFR1 is statistically independent, while dFR2 is

statistically dependent with respect to dFR1.
Let us consider Aij  0, as well as the tolerance ranges:

DFRi  dFRi  DFRi ð11:21Þ

DDPi  dDPi  DDPi ð11:22Þ

Combining Eqs. (11.20), (11.21), and (11.22), we have the limit lines of the
system range and of the design range in the physical domain:

DFR1  A11 dDP1  DFR1 ð11:23Þ

DFR2  A21 dDP1 þ A22 dDP2  DFR2 ð11:24Þ

DDP1  dDP1  DDP1 ð11:25Þ

DDP2  dDP2  DDP2 ð11:26Þ

In a similar way, for the functional domain, we have:

DFR1  dFR1  DFR1 ð11:27Þ

DFR2  dFR2  DFR2 ð11:28Þ

A11 DDP1  dFR1  A11 DDP1 ð11:29Þ

318 J. Fradinho and A. Gonçalves-Coelho

A21
A22 DDP2  dFR2  dFR1  A22 DDP2 ð11:30Þ
A11

Figure 11.5 depicts Eqs. (11.20)–(11.22) in the physical domain.

The parallelogram [ABCD] of Fig. 11.5 is the system range of the design, while
the quadrilateral [EFGH] is the design range and the hexagon [IFJKHL] is the
common range. Thus, for the case of uniform pdfs, the information content is given
by



1 Area of quadrilateral EFGH
I ¼ log2 ¼ log2 ð11:31Þ
p Area of hexagon IFJKHL

The areas of the quadrilateral and of the hexagon can be analytically computed.
Those areas can be much more easily evaluated using a 2D CAD system capable of
measuring areas of closed polygons.
By the same token, the functional domain of the 2-FR, 2-DP decoupled design of
Fig. 11.5 can be depicted in Fig. 11.6, by using Eqs. (11.27)–(11.30).
Notice that this graphical method is valid only for uniform pdfs. The method can
be extended to 3-FR, 3-DP decoupled designs with uniform pdfs (Fradinho et al.
2017). The random variation of the generic FRs, dFRi , due to the random variation
of the DPs, dDPi , is expressed by
8 9 2 38 9
< dFR1 = A11 0 0 < dDP1 =
dFR2 ¼ 4 A21 A22 0 5 dDP2
: ; : ;
dFR3 8A31 A32 A33 dDP3
< dFR1 ¼ A11 dDP1
, dFR2 ¼ A21 dDP1 þ A22 dDP2 ð11:32Þ
:
dFR1 ¼ A31 dDP1 þ A32 dDP2 þ A33 dDP3

Fig. 11.5 The isogram of a

2-FR, 2-DP decoupled design
in the physical domain.
(Reproduced from Fradinho
et al. (2017), originally
published open access under a
CC BY 4.0 license: https://
doi.org/10.1051/matecconf/
201712701004)
11 The Information Axiom and Robust Design 319

Fig. 11.6 The isogram of a

2-FR, 2-DP decoupled design
in the functional domain

Considering unbiased tolerance ranges, and that

DFRi  0; DDPj  0; Aij  0;

dDPj are statistically independent;
dFR2 is statistically dependent to dFR1 ;
dFR3 is statistically dependent to dFR1 and dFR2 :

From Eq. (11.25), one can obtain the shapes of the system range and of the
design range in the physical domain:

DFR1  A11 dDP1  DFR1 ð11:33Þ

DFR2  A21 dDP1 þ A22 dDP2  DFR2 ð11:34Þ

DFR3  A31 dDP1 þ A32 dDP2 þ A33 dDP3  DFR3 ð11:35Þ

DDP1  dDP1  DDP1 ð11:36Þ

DDP2  dDP2  DDP2 ð11:37Þ

DDP3  dDP3  DDP3 ð11:38Þ

The first three conditions characterize the system range and the last three denote
the design range.
The system range is the intersection of three pairs of semi-spaces. The first
condition represents two profile planes that are parallel to the lateral projection
plane and orthogonal to the dDP1 axis; the second one represents two vertical
planes that are parallel to dDP3 axis and orthogonal to the horizontal projection
320 J. Fradinho and A. Gonçalves-Coelho

plane; and the third condition represents two planes that are oblique to the three
projection planes. The design range is a rectangular parallelepiped that is centered
in the origin.
Similarly, the equations of the design range and of the system range in the
functional domain are given by

DFR1  dFR1  DFR1 ð11:39Þ

DFR2  dFR2  DFR2 ð11:40Þ

DFR3  dFR3  DFR3 ð11:41Þ

A11 DDP1  dFR1  A11 DDP1 ð11:42Þ

A21
A22 DDP2  dFR2  dFR1  A22 DDP2 ð11:43Þ
A11


A32 A31 A21 A32
A33 DDP3  dFR3  dFR2   dFR1  A33 DDP3 ð11:44Þ
A22 A11 A11 A22

In the physical domain, the system range is determined by six planes.

Figure 11.7 shows the intersections of the six planes with the orthogonal projection
planes. The values of Aij , DFRi and DDPj were considered as positive in the
graphical representation. Thus, the vertical and the oblique planes are opened to the
right.
Figure 11.8 depicts the resulting geometric solid, which is a six-face polyhedron
centered in the origin and composed of three pairs of parallel faces. Four of those
faces are defined by the profile planes (p and p0 ) and the vertical planes (h and h0 ).
All these four planes are orthogonal to the horizontal projection plane. The upper
and the lower limits of the polyhedron are the oblique planes (a and a0 ).
The design range is a quadrangular prism that is also centered in the origin.
The system range is the six-face polyhedron [ABCDA′B′C′D′] with two profile
faces ([ADA′D′] and [BCB′C′]), two vertical faces ([ABA′B′] and [CDC′D′]), and
two oblique faces ([ABCD] and [A′B′C′D′]).
The design range is the quadrangular prism [EFGHE′F′G′H′] with faces that are
parallel to the projection planes.
The common range is the ten-face polyhedron [AA‴II′JJ′KL′C″C′MM′NN′PQ′],
which is the intersection of those two solids and is made of five pairs of parallel
faces.
The coordinates of all the vertices are easily computed because each one results
from the intersection of three known planes.
The probability of success of this design, ps , is given by
11 The Information Axiom and Robust Design 321

δDP3≡δDP2

fπ≡hπ fπ’≡hπ’
fθ fθ’ fα
fα’

δDP1

hα’

hα hθ’
hθ

Fig. 11.7 The six planes that determine the system range in the physical domain of a 3-FR, 3-DP
decoupled design. (Reproduced from Fradinho et al. (2017), originally published open access
under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/201712701004)

Fig. 11.8 The system, the

design, and the common
ranges in the physical domain
of a 3-FR, 3-DP decoupled
design. (Reproduced from
Fradinho et al. (2017),
originally published open
access under a CC BY 4.0
license: https://doi.org/10.
1051/matecconf/
201712701004)
322 J. Fradinho and A. Gonçalves-Coelho

Vc
ps ¼ ð11:45Þ
Vdp

where Vc is the volume of the common range and Vdp is the tolerance volume for
the DPs, which is the design range.
Example 11.4
The design depicted in Fig. 11.7 will be used as an example, where the following data is
assumed:
Profile planes intersect the dDP1 axis at points 2 and (−2);
Vertical planes intersect the dDP1 axis at points 3 and (−3) and open 45° to the right;
Oblique planes intersect the dDP1 axis at points 4 and (−4) and open 45° to the right;

DFR1 ¼ 2 length units;

DFR2 ¼ 3 length units;
DFR3 ¼ 4 tan 35 length units.

A11 ¼ A22 ¼ A33 ¼ 1

DDP1 ¼ DDP2 ¼ DP3 ¼ 3 length units.

As a result, from Eq. (11.32), the design equation of this decoupled design is
8 9 2 38 9
< dFR1 = 1 0 0 < dDP1 =
dFR2 ¼ 4 1 1 0 5 dDP2 ð11:46Þ
: ; : ;
dFR3 tan 35 1 1 dDP3

Thus, the volume of the common range, Vc , is the volume of the feasible range
represented by the polyhedron [AA‴II′JJ′KL′C″C′MM′NN′PQ′]. This volume can
be computed, although very hard to attain. Therefore, a 3D solid modeler software
was used to evaluate Vc .

Vc ¼ 90; 31 volume units ð11:47Þ

The volume of the tolerance region for the DPs, Vdp , is

Vdp ¼ 8  DDP1  DDP2  DP3 ¼ 216; 000 volume units ð11:48Þ

Thus, the probability of success and the information content are given by

90; 31
ps ¼ ’ 0; 42 ð11:49Þ
216; 00

I ¼  log2 ps ’ 1; 26 ð11:50Þ
11 The Information Axiom and Robust Design 323

If DDP1 , DDP2 , and DDP3 were sufficiently small, then the whole design range
would be contained in the system range and the information content would be null.

11.5 Information Content and Robustness

The Information Axiom states that, among designs that conform to independence,
the best is the one with the minimum information content. In Sect. 11.2, it was
shown that the information content decreases (or the probability of success
increases) when the area of the common range increases. The minimum information
content can be achieved by moving the median value of the design range to the
median value of the system range. This distance is called bias. Thus, a small or null
bias should always be preferred. This may be easy for designs that conform to
independence because each FR can be tuned through a DP only.
Another way to increase the area of the common range is to extend the design
range. Designs that allow large random variations in the DPs with small response
deviation are known as robust designs. One of the possible strategies to increase the
robustness of design is to identify the limit values of the largest range of the DP that
achieves the looked-for functional tolerance range.
The relationship between a FR and the corresponding DP is either already
known or may be experimentally determined. Graphically, it can be represented by
a line, as shown in Fig. 11.9.
For a certain ΔFR, (ΔDP)1 represents the DP limits for a design, between points
A and B, while (ΔDP)2 represents the DP limits between points C and D. Since
(ΔDP)2 is larger than (ΔDP)1, then (ΔDP)2 corresponds to a more robust design.
The most robust design is the one with the smallest absolute value of the average
variation rate between the DP limits. In other words, the preferable situation is the
one with the smallest slope modulus of the straight line linking the two points of the
curve.
Robust designs have a small information content and large capacity to allow DP
random variations with small disturbance of their response.

Fig. 11.9 Different design

ranges for the same functional (∆FR)
FR
tolerance
B C
D
A

(∆DP)1 (∆DP)2

DP
324 J. Fradinho and A. Gonçalves-Coelho

11.6 Summary

This chapter presents the Information Axiom and the meaning of information
content. For simple designs it is presented how to make the computation of the
information content, and it is shown that it measures the probability of success of
the design.
The computation of information content is only possible for designs that con-
form to independence (uncoupled and decoupled designs). Three simple examples
with two FRs were presented.
The computation of the information content of decoupled designs is hard to
preform because it involves the use of conditional probability. A graphical method
is presented to compute the information content of 2-FR, 2-DP and 3-FR, 3-DP
decoupled designs considering uniform probability density for FRs.
Finally, the concept of robust design is discussed, and desirable design ranges
withdrawn from the analytical relationships between FRs and DPs are indicated.
Problems
1. Explain the following expression according to the Information Axiom of AD:
P
n Q
n P
n
Itotal ¼  log2 pi ¼  log2 pi ¼ Ii
i¼1 i¼1 i¼1
2. Comment the following statement: “The calculation of the information content
of a decoupled design implies the use of conditional probability.”
3. The following figure shows the probability density function (pdf) of the system
range of a certain FR with a uniform distribution. The figure also shows the
limits of the design range and of the common range. Explain how the infor-
mation content for this FR could be computed (Fig. 11.10).
4. Consider the following equations of two designs. Explain how the information
contentof each
  design can be calculated.
     
FR1 A11 0 DP1 FR1 B11 B12 DP1
¼ ¼
FR2 0 A22 DP2 FR2 0 B22 DP2
5. The following figure shows the pdf of a design with one only FR. The figure
also shows the design range of three possible solutions (A, B, and C). According
to AD theory, which solution is the best one? And the worst? Why? (Fig. 11.11)
11 The Information Axiom and Robust Design 325

Fig. 11.10 Pdf of a uniform

distribution

Fig. 11.11 Non-uniform pdf Design

of a design range C
pdf

Design
Design
range B
range A System
range

FR

References
Fradinho J, Cavique M, Gabriel-Santos A, Mourão, A, Gonçalves-Coelho A (2017) How to
compute the information content of 3-FR, 3-DP decoupled designs with uniform probability
density functions for their FRs. In: MATEC web of conferences, vol 127, no. 01004, The 11th
International Conference on Axiomatic Design (ICAD 2017). https://doi.org/10.1051/
matecconf/201712701004
Gonçalves-Coelho A, Mourão A, Pereira ZL (2005) Improving the use of QFD with axiomatic
design. Concur Eng Res Appl 13(3):233–239. https://doi.org/10.1177/1063293X05056787
Park GJ (2007) Analytic methods for design practice. Springer-Verlag, London
Shannon CE (1948) A mathematical theory of communication. Bell Syst Techn J 27(3):279–423,
623–656
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press, NY
Suh NP (2005) Complexity. Oxford University Press, NY
Complexity in Axiomatic Design
12
Erik Christian Nicolaas Puik

Abstract
In the previous chapter, the concept of information in design was introduced. It
was shown how the Information Axiom could be applied to increase the
robustness of processes. It was also shown that the axioms in Axiomatic Design
(AD) should be addressed in a distinct order.
In this chapter, four different kinds of complexity in AD are explained that
can be applied for typical situations. Also, a way to visualize complexity in
design is introduced; the “Functionality Diagram.”

After studying this chapter, the reader should know the following:
The reader will understand the particular but powerful definition complexity of
in AD, which kinds of complexity in AD have been defined, and how they can
be applied. The reader will also learn how to apply complexity in functionality
diagrams, that offers a powerful way to visualize the design process as it evolves
over time.

12.1 Motivation for the Complexity Theory in Axiomatic

Design

So far, this book has been focusing on two axioms: (i) the Independence Axiom and
(ii) the Information Axiom. In this chapter, a third axiom will be introduced and
explained: The “Complexity Axiom.”

E. C. N. Puik (&)
Knowledge Centre for Healthy and Sustainable Living, HU University op Applied Sciences
Utrecht, Postbus 182 Padualaan 99, Utrecht 3584 CH, The Netherlands
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 327

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_12
328 E. C. N. Puik

12.1.1 Applying a Complexity Theory to Real Problems

After the AD methodology was refined for over 20 years, it appeared still difficult to
explain simple things such as “How do we control the semiconductor processing
machines such as the track machine to maximize productivity?”, “Why is the control
of job shops (i.e., machine shops) so difficult?”, “How does mitosis work so well?”,
“Why did the GM ignition key fail, killing several people?”, and, “Why did it take so
long to build the new Berlin airport?”, etc. These systems appear to have high
complexity. Trying to understand why some commercial and industrial systems have
so many problems, Professor Nam Suh started to study Shannon’s information theory
that was also invoked by many others to investigate complexity in scientific literature.
After spending quite some time trying to understand what they had done, he con-
cluded that there wasn’t a practical and accurate definition of the concept of “com-
plexity.” All variations in literature defined complexity as a mathematical issue but
that approach does not seem suitable for AD that is focusing to solve real problems.
The conclusion was that complexity is a design issue.
If a task cannot be achieved, it looks complex. For instance, even a short
computer software program appears to be complex when it is a coupled design. It is
not the length of the program that creates the complexity of a software system, but
the coupling of FRs. Some computer scientists unsuccessfully tried to explain
complexity in terms of the length of the software code.
If we cannot solve real problems, the theory is not useful in the field of design
and engineering. The complexity theory on AD tries to change this and apply the
concept of complexity in a practical manner, enabling the capability to address real
problems in practice.

12.1.2 What is Complexity

There is not a single explanation for the term “Complexity”, though everyone will
have a certain interpretation of the general concept. Also, from a scientific per-
spective, there are many interpretations of complexity or complex systems. Many of
these concepts were inventoried and explained by Suh in the context of AD (Suh
1999). Some interpretations of complexity as were mentioned by him are the fol-
lowing ones:

• Gallagher and Appenzeller define complex systems as systems whose properties

are not fully explained by an understanding of its component parts (Gallagher
and Appenzeller 1999);
• Goldenfield and Kadanoff state that complexity means that we have a structure
with variations. Thus, a living organism is complex because it has many different
working parts, each formed by variation in the working out of the same genetic
coding (Goldenfeld and Kadanoff 1999);
• Whiteside and Ismagilov state that a complex system is one whose evolution is
(i) very sensitive to initial conditions or to small perturbations, one in which the
12 Complexity in Axiomatic Design 329

number of independent interacting components is large, or one in which there are

multiple pathways by which the system can evolve and (ii) the system is
complicated by some subjective judgment and is not amenable to exact
description, analytical, or otherwise (Whitesides and Ismagilov 1999);
• Weng, Bhalla, and Iyengar state that complexity arises from a large number of
components, many with isoforms that have partially overlapping functions, from
the connections among components; and from the spatial relationship between
components (Weng et al. 1999);
• Shannon and Weaver, as well as Gellmann and Lloyd equate complexity with
the number of bits of information it takes to describe an object or a message
(Shannon and Weaver 1949). Gellmann and Lloyd introduce the concept of
“Irregularities” that describes that elements are not organized well and therefore
should be considered complex.

In order to be able to imagine what complexity is in daily life, here are some
examples of complex systems. It concerns systems that are really difficult to
comprehend:

• weather forecasting;
• understanding the interactions in a rainforest;
• understanding the dynamics of social media;
• governing a country;
• understanding and prediction of global warming.

12.2 Theory of Complexity, Periodicity, and the Design

Axioms

Among the definitions of complexity, AD has a generic though particular approach

of complexity. Where many definitions define complex systems in the physical
domain, AD defines the complexity in the functional domain. This means that
complexity in AD is not determined by what a system is or what it looks like, but it
is determined by the functional behavior of the system. A complex system does not
automatically have the capability to perform as it should according to its functional
requirements (FRs) and as such its complexity should be reduced. Therefore, the
definition of complexity in AD is

• the Complexity Axiom: “Reduce the complexity of a system”;

• in which complexity is defined as “A measure of uncertainty in satisfying the
FRs”.

The Complexity Axiom is also referred to as the “Third Axiom,” “Axiom 3,” or the
“Main Axiom.” The Complexity Axiom is directly related to the Independence Axiom
and the Information Axiom since their goal is to make sure that the FRs are satisfied
accordingly. To be more precise: the Independence and the Information Axioms
330 E. C. N. Puik

appear to be a subset of the Complexity Axiom (Puik and Ceglarek 2014a). However,
this relation is still being investigated and as such the Independence and the Infor-
mation Axioms are maintained as unbound axioms for the benefit of the designer.
If we want to deal with complexity, understand and eventually eliminate it, it is
essential to perceive what the fundamental cause is of complexity in a system. To
understand this cause, it helps to have a closer look at some complex systems:

(1) Let’s say the challenge is to replace the front wheel of a car. Even someone
who never has done this may understand that the car needs to be lifted by use
of the jack. When this task is successfully completed, the next step is to
remove the bolts from the wheel. When trying to exceed the torque needed to
unscrew the bolts from the wheel hub, it appears that the wheel freely rotates in
the air when a torque is applied to the bolts. The bolts needed to be loosened
first when the car was still standing on the ground.

This first example shows a decoupled system. With the right knowledge of the
design matrix, it would have been possible to determine the right order to
address the DPs on forehand in order to satisfy the FRs. However, without this
knowledge, the person resorts to trial and error methods of evaluation trying
the possible sequences of addressing the DPs to satisfy the FRs. As long as he
has not found the right sequence, the design shows an unwanted response to
his actions. This fairly simple problem is uncertain to succeed for people that
miss the knowledge and therefore, for them it is a complex task.
(2) A second example is that of a combination lock (Suh 2005a; Foley et al. 2016).
A combination lock is hard to open without the right knowledge of the
sequence of numbers we have to activate. Using trial and error it is extremely
laborious and time-consuming to try all combinations of the lock since the lock
was designed to be complex. However, with the knowledge of the right
numbers and the instruction manual at hand, the lock is easy to open. Without
the right knowledge, opening the lock is quite complex. When the right
knowledge is available, opening the lock is not complex at all.
(3) A space shuttle consists of over 2 million parts. A vast majority of these parts
should perform within the tolerance window of their FRs to enable safe launch
and return. This means that the error margin per part is incredibly low. To
satisfy over 2 million FRs with their respective DPs is very complex.
(4) If a designer is sloppy when decomposing the FRs of a system, it can happen
that one of FRs with lower hierarchy is forgotten. As the design process
continues, there will not be a matching DP appointed for the forgotten FR. As
a result, the main FRs (at higher hierarchical levels) will not be satisfied under
all conditions. Since it is not known that the FR and DP are missing, the
system is misunderstood and seems complex.

All these examples have a common feature and that is that the designer lacks
knowledge about the system he is working on. Because of that, it is not possible to
fully understand all constituent elements of the system. This lack of knowledge
12 Complexity in Axiomatic Design 331

leads to uncertainty in satisfying the FRs, leads to failing systems, and makes a
system complex.
If we zoom in further to Complexity in AD, it appears that we have to make a
distinction regarding how complexity develops over time. Some kinds of com-
plexity do not change with time, others do strongly depend on time as will be
shown in the next sections.

12.2.1 Time-Independent Complexity in Axiomatic Design

This section deals with “Time-Independent” Complexity. This means that the
complexity basically does not change with time. It changes due to the involvement
of the designer and his environment.
Time-independent complexity consists of two components: “Real” and “Imag-
inary” time-independent complexity, further to be referred to as real complexity and
imaginary complexity (CR and CIm).
Real Complexity
Real complexity is inversely related to the probability of success that the associated
FRs are satisfied according to one of the following relations:

X
m
CR ¼  logb Pi ð12:1Þ
i¼1

X
m
CR ¼  logb Pijf jg for f jg ¼ f1; 2; . . .; i  1g ð12:2Þ
i¼1

depending if the system is uncoupled (12.1) or decoupled (12.2). Relation (12.1) is

under the reservation that the total probability Pi is the “joint probability of pro-
cesses that are statistically independent” so the different processes that are sum-
marized do not influence each other. Relation (12.2), for decoupled systems, is
modified to correct for dependencies in the probabilistic function (Suh 2005a). “b”
Is in both cases the base of the logarithm, usually in bits (b = 2) or nats (b = 2.718)
depending on the preferred definition. Given (12.1) and (12.2), CR can be related to
the information content in AD, which was defined in terms of the probability of
success of achieving the desired set of FRs (Suh 1990), as

CReal ¼ I ð12:3Þ

in which CR is the real complexity and I being information as defined by the

Information Axiom.
CR is by definition addressed by the Information Axiom. It is addressed by
matching the design and system ranges of the product design.
332 E. C. N. Puik

Examples of systems with CR are

• tolerances of a manufacturing process;

• a paint job of a car with color difference;
• sloppy work of a shoemaker;
• people continuously crossing road markings while driving a car;
• being too late for a job interview.

These examples describe systems that need (constant) attention to be executed

well, but when procedures are closely followed they will deliver good results.

Imaginary Complexity
“Imaginary Complexity” is a little harder to understand. Suh defines it as “uncer-
tainty that is no real uncertainty” and “it arises because of the designer’s lack of
knowledge and understanding” (Suh 2005b). When a design is uncoupled or
decoupled, the imaginary component of complexity is equal to zero (Suh 2005b).
Note that for a decoupled design this is only guaranteed if the optimization order of
the design relations is known. This makes CIm inversely related to the satisfaction of
the Independence Axiom; an uncoupled system is a system with no CIm. Vice versa,
a coupled system is a complex system.
To illustrate CIm, Suh describes a designer trying to optimize a system without
writing down the design equations (Suh 2005a). Even if this design is a decoupled
design, the designer will have to resort to trial-and-error methods of evaluation. The
design will tend to behave erratically and not conforming his (limited) under-
standing. He will have to try many different sequences of DP adjustments to see if
the FRs get satisfied. There are n! distinct sequences of DPs, of which only one is
correct. The probability of finding the right sequence n DPs to find the entire set of
m FRs is given by

1
P¼ ð12:4Þ
n!
The probability of finding the right sequence through a random trial-and-error
process goes down rapidly with an increase in the number of DPs, as shown in the
table below:

n n! P = 1/n!
1 1 1.000
2 2 0.500
3 6 0.1667
4 24 0.04167
5 120 0.8333  10–2
6 720 0.1389  10–2
7 5,040 0.1984  10–3
12 Complexity in Axiomatic Design 333

When n = 5, the probability of finding the right sequence is 0.008 per try which
is very low. Therefore, this design appears to be quite complex because the
uncertainty that a solution is found and the FRs will be satisfied is large. This kind
of complexity is typical for CIm and it resides in the designer’s head due to a lack of
knowledge of the system. CIm is a kind of complexity that is caused by a lack of
fundamental understanding of the AD theory and its application. To the designer, a
system may look complex, although it may not be when the right knowledge is
present.
Examples of systems with CIm are

• you get an assignment on the first day of a new job that requires an under-
standing of the organization;
• finding your way without a map or GPS in a foreign city;
• cooking without a recipe;
• opening a combination lock without the code;
• you offer a friend to help moving to a new house without knowing how much
stuff he owns.

Complex Versus Complicated Systems in AD

We have been elaborating about complex systems. However, in many other com-
plexity theories distinction is made between complex and complicated systems
(Snowden 2000; Kurtz and Snowden 2003; Boone and Snowden 2007). These are
not the same and for those readers that have experience with other definitions the
differences are explained here:

• For complex systems, cause and effect relations are not (yet) understood. This
the traditional domain of complexity theories, which study how patterns emerge
through the interaction of many elements. It is not that everything in this context
is unknown or unsure. However, emerging cause and effect relations that are still
fuzzy and the number of elements are too mind-boggling to oversee the situation
and analyze it. Emergent patterns can be perceived but not predicted. In the work
of Snowden and Boone, this phenomenon is called “retrospective coherence“
(Boone and Snowden 2007); things become clear in retrospect, but could not be
predicted on forehand. As such, the designer can expect to be surprised as his
design is not fully understood yet;
• Complicated systems however, deal with knowable cause and effect relations.
Complicated systems, unlike simple ones, may contain multiple right answers,
and although there is a clear relationship between cause and effect, not everyone
can see it. The fact is that problems can be solved by bringing in the right team
of experts, however, the issue is whether we can afford the time and resources to
reduce complicated systems to simple systems with clear and documented cause
and effect relations. In general, this is not possible and, instead, it relies on expert
opinion, which in turn creates a key dependency on trust between expert advisor
and decision-maker. Complicated systems need systems thinkers, learning
334 E. C. N. Puik

organizations, and the adaptive enterprise, all of which are too often confused
with true complex systems.

By this definition complex systems are systems without a cause and effect
relation. Complicated systems are systems that do have a causality but not everyone
can see it.
There is an important difference how AD approaches complexity. Due to its
definition “A measure of uncertainty in satisfying the FRs,” it defines complexity
as an issue that compromises functionality. This means that any system that lacks
the capability to meet the FRs is a complex system. Here emerges a discrepancy in
comparison to most complexity theories as described above. A system that is
considered “complicated” in other complexity theories may appear to be “complex”
in the complexity definition of AD. This is because complexity is considered from
the perspective of functionality. If a designer is not able to gather the required
knowledge to understand and solve the problem, a system is considered complex
according to the definition in AD (note that “the designer” may represent a group of
designers with various expertise). Even if the designer would have some under-
standing of the problem but he is not given enough time to adequately address all
issues of the system, it is considered a complex system. Complexity in AD focuses
on functional performance, not on implementation, nor on a solution that could be
realized by engineers that cannot be mobilized, nor what Einstein could do if he
were here.
In summary, the distinction of complex and complicated systems is not made in
AD. Both are referred to as CIm and they exist because the designer does not have
the knowledge to fully understand the design. Complexity in AD is inversely
related to the capability of the designer to address problems well. The result is
assessed in the functional domain and depends if the system actually does what it
should do.

12.2.2 Time-Dependent Complexity

Sometimes, complexity changes with time. Even if no involvement with a system

takes place, it still tends to become more complex. Time-dependent complexity
occurs because future events occur in unpredictable ways and therefore the nature
of these events also is unpredictable (Suh 2005b). This is caused by a system range
that moves away from the design range. The overlap of system and design range
will become smaller and the certainty that the FRs will be satisfied by the DPs
decreases. This is shown in Fig. 12.1.
There are two kinds of time-dependent complexity:

• combinatorial Complexity;
• periodic Complexity.
12 Complexity in Axiomatic Design 335

Probability Density Function (pdf)

System Range changes
Design Range
over the course of time

Common
Range

Functional Requirement (FR)

Fig. 12.1 If the system range moves away from the design range in time, the overlap between
design range and system range will decrease. Chances that the FRs are successfully satisfied by
their DPs is negatively affected

Time-Dependent Combinatorial Complexity

Time-dependent Combinatorial Complexity is defined as the complexity that
increases as a function of time due to a continued expansion of the number of
combinations with time, which may eventually lead to a chaotic state (Suh 2005a).
An example is the airline scheduling problem. When there is a major storm in
Chicago, the airplanes that were to arrive in and leave from Chicago cannot operate
according to their schedules. In fact, some flights may be canceled. Eventually, this
weather in Chicago affects the flight schedules at all the airports that are connected
to Chicago either directly or indirectly. As a result, there will be increasing
uncertainty in dispatching aircraft as a function of time. This is an example of
time-dependent Combinatorial Complexity. When a system is affected by
time-dependent Combinatorial Complexity, it eventually breaks down because it
will go into a chaotic state.
Examples of systems of time-dependent Combinatorial Complexity are

• a computer that operates slower when active for a longer time;

• a developing traffic jam during traffic hours;
• an increasing failure rate of a production process due to tool wear;
• search for exact and exhausting information on the internet;
• brexit discussions in the UK House of Commons.
Time-Dependent Periodic Complexity
Time-dependent Periodic Complexity can be understood if we go back to the airline
scheduling problem caused by a major snowstorm in Chicago. Typically, the airline
scheduling problem continues to worsen throughout the day because of the cas-
cading effect of airlines having to make decisions on the deployment of available
airplanes. However, since the airline schedule is periodic each day, all of the
uncertainties introduced during the course of the day terminate at the end of a 24-h
cycle, and hence this Combinatorial Complexity does not extend to the following
336 E. C. N. Puik

day. Each day, the schedule starts over again (Suh 2005a). The periodic nature of
the airline schedule resets between periods and the uncertainties created during the
prior period are irrelevant. Within the same period, complexity may be combina-
torial, but over many periods we speak of Periodic Complexity.
Examples of systems of time-dependent Periodic Complexity are

• rebooting your computer every morning;

• a traffic jam that disappears during evening hours;
• a decreasing failure rate of a production process when tools are regularly
replaced;
• daily life and its worries;
• even in the UK House of Commons, brexit discussions have ended at some
point.

One important conclusion of the complexity theory is that complexity is greatly

reduced when a system with time-dependent Combinatorial Complexity is trans-
formed to a system with time-dependent Periodical Complexity. The period is the
functional period, not necessarily temporal period, where the same set of functions
repeats, i.e., the periodicity does not necessarily have a constant time period. Where
Combinatorial Complexity tends to keep emerging further and further till something
goes horribly wrong, Periodic Complexity is reset from time to time and because of
that it is far more manageable.
Combinatorial- and Periodic Complexity are mostly time-dependent forms
of CR due to the fact that they deal with a changing overlap between system range
and design range. In theory, it is possible to distinguish “time-dependent Imaginary
Complexity,” where even the FRs may change over time, and though they are
referred to in literature (Lee 2003) they are extremely hard to deal with since a fix
would require conceptual changes of the system.

12.2.3 Limiting Behavioral Options by Banning Irregularities

from the Design

A practical view on dealing with Complexity in a design and the uncertainties that
arise from it was described by Gell-Mann and Lloyd (1996). As they state, the
cause of uncertainties is found in a “Lack of Regularities“ of the considered system
(in our case a product or system design). The lack of regularities, which may be
seen as a missing structure in the design, will be reduced as the knowledge of the
designer increases. While the designer is working on his design, the lack of reg-
ularities will be reduced as much as the knowledge of the designer permits. It works
two ways, (i) a designer that has no knowledge of the design cannot be expected to
deliver a well-structured design, and (ii) the irregularities in this ill-structured
design will not be addressed (Puik and Ceglarek 2015). As a result, the design will
not be regulated well and, at some point, it will show unexpected behavior.
12 Complexity in Axiomatic Design 337

This makes Complexity inversely related to the knowledge of the designer. In a

situation where the designer has unlimited time and resources to acquire and
implement the right knowledge, the product design will be fully regulated and
therefore it will have no freedom of choice in its operation other than the envisioned
way by the designer; as a result, all complexity is eliminated from the system and it
will behave exactly the way it should; all FRs of the product design satisfied by
their DPs.
Summarized, it can be said that the way of looking to a design as proposed by
Gell-Mann and Lloyd, leads to an accessible way to deal with complexity in design.
Complexity in AD is a measure for the lack of regularities in a product design and it
leads to uncontrolled behavior of the system.

12.2.4 Known and Unknown Kinds of Time-Independent

Complexity

A problem with complexity in general is that lacking knowledge does not auto-
matically come to the surface. Complexity may show that it is present but may also
remain hidden. These are called “Known” and “Unknown” kinds of complexity.
Known and Unknown Kinds of Imaginary Complexity
CIm also has known and unknown variants. In a lucky situation, CIm in the design
plays up at some point and warns the designer that he does not yet fully understand his
design. A perceptive designer will be alarmed and try to replicate the situation till he
learns what irregularities cause unexpected behavior. In the less lucky situation, the
designer is not confronted with the system’s misbehavior, or he does not acknowl-
edge it. In this case, the CIm stays hidden in the system for some time longer. Note that
this is virtually always of a temporary nature; in a later stage, the irregularities in
design will play up and need to be restricted by an appropriate design solution.
Based on these examples, two specific kinds of CIm may be recognized:
The first one is “Unrecognized Imaginary Complexity,” the subset of CIm that is
not recognized by the designer and therefore remains hidden in the system. It may
be addressed by the determination of the right FRs, DPs, and PVs. However, this
will typically not happen because the designer is not aware of its existence.
Examples of Unrecognized CIm are

• a cook trying to copy a recipe, however, without exactly knowing what the right
ingredients are;
• partial modelling of a system (when elemental parts were forgotten);
• a DP that was not recognized that changes an FR.

Second, “Recognized, Imaginary Complexity” is the subset on CIm that indeed is

recognized by the designer, however, the knowledge to address the problem is
lacking. Because of this, it cannot yet be eliminated from the design. It is addressed
by definition of the right design matrix and successively decoupling it.
338 E. C. N. Puik

Examples of Recognized CIm are

• during the design process, it appears that a conceptual error was made and the
design matrix needs repair but this has far-reaching consequences;
• insufficient means or time to address a problem, or the problem needs to be
solved by an external party;
• the back-log in a Scrum session.
Known and Unknown Kinds of Real Complexity
Analog to CIm, it is also possible to decompose CR one step further.
Comparable to the definition above, “Unrecognized, Real Complexity” is the
subset of CR that is not recognized by the designer and therefore remains hidden in
the system. It is addressed by the knowledge how the DPs satisfy the FRs.
Examples of Unrecognized CR are

• A German car manufacturer was suddenly confronted with the problem that the
windshields showed large cracks within hours after completion of the manu-
facturing process. It appeared to be a problem with the material properties of the
glass material. It took 6 weeks to find the cause and understand the problem and
another four weeks to fix the problem;
• A manufacturing process that runs out of its operating window without the cause
being known;
• A redundant situation where drifting DP’s compensate for each other’s devia-
tions. The related FR will remain satisfied but the system might be operating out
of their design ranges.

And second “Recognized, Real Complexity” is the subset of CR that indeed is

recognized by the designer but cannot yet be eliminated from the design. It is
addressed by understanding how the system ranges and their tolerances satisfy the
design ranges (and their tolerances).
Examples of Recognized CR are

• when the German car manufacturer understood the cause of cracking windows, it
was largely dependent on a supplier to fix the problem;
• basically, all situations where well-engineered systems fail and a diagnosis has
not yet be executed;
• a traffic situation where traffic is slowing down and a car is driving at too high
velocity while keeping insufficient distance to the car ahead.
12 Complexity in Axiomatic Design 339

12.2.5 Overview of All Kinds of Time-Independent

Complexity

To complete this explanation of complexity in AD, now follows an overview of the

all definitions that were described. Figure 12.2 shows a graphical overview:

• Complexity Axiom; complexity related to the Complexity Axiom based on

information in design according to the definition of Shannon and that affects the
FRs.
• CIm; a specific kind of useful information due to a discrepancy in design ranges
and system ranges according to the Information Axiom.
• CR; Information in design as defined by the Information Axiom. It is caused by
insufficient overlap between the design and system ranges.
• Unrecognized CIm; the part of CIm that is not recognized by the designer and
therefore remains unaddressed. It should, when eventually discovered, be
addressed by all relevant FRs and DPs of the design matrix. Once this kind of
complexity is recognized, it instantly changes to Recognized CIm.
• Recognized CIm; the part of CIm that is recognized by the designer but the
knowledge to address the problem in an appropriate manner is still lacking.
Recognized information is addressed by decoupling of the design matrix.
• Unrecognized CR is eliminated by understanding how the DPs satisfy the FRs,
knowing all relevant parameters (decomposition of the DPs).
• Recognized CR is eliminated by optimizing the system and design ranges.

12.3 Graphical Representation of Complexity

In this section, complexity in AD from the former section will be applied to

visualize the design process as it develops during the execution of the design
process.

12.3.1 The Functionality Diagram

The “Functionality Diagram,” also referred to in literature as “Axiomatic Maturity

Diagram” or “AMD” visualizes complexity of a product design (Puik and Ceglarek
2014b). As complexity in AD is related to the satisfaction of the FRs, it plots the
extent to which functionality of a system is met. The diagram, shown in Fig. 12.3,
uses two axes, one for each axiom, plotting the degree in which the axioms are
satisfied.
The horizontal axis is the “axis of conceptual organization” starting at “No
Organisation” and ending with “Proof of Concept.” Proof of concept indicates that
the product design is a viable design; the design matrix is decoupled and therefore
340 E. C. N. Puik

Complexity Axiom
(Useful Complexity)

Are the design matrices organized?

(FR-DP-PV relations known and decoupled)

No Yes

Imaginary Complexity Real Complexity

Are FR-DP-PV relations Is it understood how the

recognized? DPs satisfy the FRs

No Yes No Yes

Unrecognized Recognized Unrecognized Recognized

Imaginary Imaginary Real Real
Complexity Complexity Complexity Complexity

Address it by Decouple Understand stochastic Make robust by

defining all FRs the design FR-DP relation and matching Design-
and DPs matrices all parameters and System-Range

Fig. 12.2 Breakdown of complexity starting with the complexity axiom in AD

CIm has become equal to zero. This implies that both Unrecognized CIm and
Recognized CIm have been eliminated. The vertical axis represents robustness of the
design from “Not Robust“ to “Fully Robust.”
Note that both axes have a continuous measure. Irregularities in design can have
all sizes and therefore the progression on the axes can have any intermediate value.
For the horizontal axis, that plots CIm (related to the Independence Axiom) this
contains; (i) the definitions of the right FRs, (ii) the according DPs, including all
DPs that need to be fixed during the process of conceptual validation, and (iii) the
process of decoupling the FRs and DPs one by one. Removing irregularities to
guarantee independence is much more than just decoupling the design matrix.
12 Complexity in Axiomatic Design 341

Fig. 12.3 The functionality

diagram. The horizontal axis
plots the independence axiom,
the vertical axis plots the
information axiom. Together
they plot complexity. The
development path is arbitrary.
(Reproduced with permission
from Puik and Ceglarek
(2014b))

The lower left-hand corner in the functionality diagram indicates a high level of
ignorance and high complexity accordingly; the designer has little knowledge of
how to satisfy FRs with his DPs and therefore the functionality is low. The upper
right-hand corner shows low complexity and maximum probability of FRs being
satisfied. Therefore, this is the area of high functionality. Development of products
starts in the lower left-hand side and moves to the upper right-hand side. Products
are fully mature when they reach the upper right-hand corner of the functionality
diagram, as marked with a dot (Fig. 12.3).

12.3.2 Presumed and Legitimate Position

in the Functionality Diagram

At any moment of development, the designer may presume an actual position in the
diagram according to the current status of the design, but this position may differ
from the real and legitimate position of the design; the presumed and legitimate
positions may have discrepancies. A discrepancy is caused by a lack of knowledge
of the designer because he has missed some essential design artefacts. As a result,
the designer rates the level of engineering of the current product design higher than
it actually is good for. When he finds the design error that causes the discrepancy,
the problem can be addressed. However, if it is not discovered, the discrepancy will
present itself at some point in the remaining part of the development process or after
market introduction as a surprise to the engineers. The presumed position in the
diagram needs to be corrected and that may lead to a project delay. Discrepancies
between the presumed and legitimate position in the functionality diagram are the
342 E. C. N. Puik

result of Unrecognized CIm and due to its disruptive character, it may have a large
impact on the remaining product development process. Therefore, the goal is to
discover discrepancies between presumed and legitimate positions as early as
possible.

12.3.3 Ideal Development Path for Product Design

Product development, as indicated above, will start somewhere at the lower

left-hand side and will move diagonally upward. The exact starting point will
depend on the difficulty of the project. A high-tech project that is new to the world
might start with a high amount of ignorance in the deep lower left corner. A project
that aims to develop according to the Right-First-Time philosophy should start
without Unrecognized CIm and starts further to the lower right-hand side of the
diagram.
The chosen path may be dependent on the amount of risk that is acceptable to the
company, e.g., the most efficient development path in terms of investment (SME), a
path that reduces the lead time (semiconductor industry), or a path that minimizes
development errors (medical or avionics). As explained in Chap. 11, it is preferred
to start with the Independence Axiom followed by the Information Axiom due to
the disruptive character of Unrecognized CIm, thus:

• define FRs and find all relevant DPs to address Unrecognized CIm;
• decouple the design matrix to address Recognized CIm;
• learn about all parameters that influence the FRS to address Unrecognized CR;
• tune the design ranges and system ranges to guarantee an adequate common
range to address Recognized CR.

This leads to a preferred path that first moves to the right and then angles
upward. It is plotted in the left-hand graph of Fig. 12.4.
Depending on the preferred project strategy, a more or less risky path could be
followed. In case of the rather conservative and slow but safe path of the Waterfall
Model, (Royce 1970) the procedure of following Independence and Information
Axioms in that order would be persistent (Fig. 12.4, right-hand graph). A slightly
more risky path that in practice enhances the development speed of projects is the
path of “Simultaneous Engineering” (Bullinger and Warschat 2012). This gives the
designer more room to start early work on robustness, process technology, and
other life cycle elements. This merges the work on Independence and Information
Axioms and possibly shortens project lead time.

12.3.4 Examples of Typical Errors

Unexpected errors in the development process are mostly related to the discovery of
Unrecognized CIm. This reveals the discrepancy between the presumed and
12 Complexity in Axiomatic Design 343

Fully Robust Fully Robust

Information Axiom

Information Axiom
Operational Room for

right
Project Management style

Proof of Concept

Proof of Concept
No Organisation

No Organisation
Speed vs Risk Mitigation

things
Doing
Simultaneous
Engineering

Doing the right things

Waterfall
Not Robust Not Robust
Independence Axiom Independence Axiom

Fig. 12.4 Left; preferred development path through the functionality diagram, as indicated in
literature, first moves to the right to satisfy Axiom 1. After this, Axiom 2 is satisfied in an upward
direction. (Reproduced with permission from Puik and Ceglarek (2014b)). Right; depending on the
nature of the project, a different strategy may be followed. The right lower curve would represent a
waterfall management approach, while the upper would represent the path in case of a
simultaneous engineering strategy

legitimate position. It will divert the development path in the functionality diagram.
Depending on the kind of error, a discontinuity will appear. This discontinuity is the
result of the conversion of Unrecognized CIm to Recognized CIm. It may show as a
kink in the development path or a jump to a different position in the diagram,
depending on

• availability of a solution to address the problem;

• robustness of the current design being affected or not.

The following typical design errors could occur:

Example 12.1 Dealing with Coupling of the Design Matrices
The first example that will be analyzed with the functionality diagram is that of coupling in
a system. As many examples in this book already have explained the problem of coupling, a
different kind of example is chosen; the problem of repairing a defective system.
Let’s say we want to repair an old radio that produces bad quality sound. The problem
could be that it just needs to be adjusted by trimming the various potentiometers, capacitors,
and inductors. However, it is also possible that one of the electronic components is
defective and needs to be replaced. A repairman will typically start asking questions, e.g.:
When did the sound degrade, did it occur all of a sudden or did it occur gradually, what was
the temperature and maybe humidity when the problem occurred for the first time. All these
questions will help the repairman to understand the problem. Now let’s say the radio still
produces some sound and repairman does not understand that a problem is caused by a
defective component. He chooses to repair the radio by readjusting some of the components
of the radio. Since this is an old radio, and system ranges may have drifted away from their
344 E. C. N. Puik

design ranges, he might succeed in improving the functionality of the radio. However, at
some point, this process stops because the defective component structurally prohibits
further improvement. As shown in Fig. 12.5 (left diagram, left blue line), it is possible to
readjust the radio to a higher level. The process does unfortunately not reach the black dot
of the diagram since the defective component causes a structural barrier for further opti-
mization. In this situation, CR is eliminated, however, Unrecognized CIm still resides in the
system as the repairman is not aware of the defective component. At some point, the
repairman will notice that his efforts do not sort the effect he was aiming for. He will start
digging deeper into the design till his knowledge increases and enables him to locate and
repair the defective component. From this point, there are two possibilities; (i) Replacement
of the component and the tolerances it introduces in the system require him to repeat the
adjustment procedure, or (ii) the tolerances of the component do not affect his earlier work
and the radio is repaired. The former of these two possibilities is shown with the dotted line
in the left diagram of Fig. 12.5, the latter is shown in the right diagram.
This last observation is comparable with a designer developing a new system of which he
starts optimizing the DPs before he is sure that the conceptual design is stable. Further
optimizations of the conceptual design may eliminate the optimized DP causing earlier
optimization efforts to be spent in vain.
Example 12.2 Wrongly Chosen DP
A wrongly chosen DP leads to the situation that this DP does not actually satisfy the related
FR. It will seem to the designer that the design matrix is understood and decoupled, but in
fact, this is not the case. In our radio repair example, this would mean that the repairman
closely has located the circuit that is malfunctioning, however, he is not 100% sure which
part to replace and from a limited number of options he guesses wrong. The efforts to repair
the radio do not reduce Unrecognized CIm and the position in the functionality diagram
does not advance to the right. There may even be a fallback in CR due to the added
tolerances of the new part. The right diagram of Fig. 12.6 plots the possible discontinuities
when the right part in the radio is replaced.
Example 12.3 Non-Matching System and Design Ranges
In Chap. 11, the example was given how to determine the length of the shoreline of
Australia. The difficulty was that although shorelines around Australia are drawn, they are
only approximate because what is presented as a straight line is made up of jagged lines. If
we look at the jagged lines in higher magnification, they are made of more jagged lines. It
was explained that the shoreline may be described by a fractal and that it has a mathe-
matical basis. When zoomed in on the fractal, it appears that at a smaller scale the shape of
the shoreline remains the same, and when zoomed further this happens again. The deter-
mination is a complex problem because every time the level of accuracy is increased, there
is the challenge to go a level deeper and increase the accuracy; it is impossible to com-
pletely match the system and the design range for this problem.
A non-matching system and design range for one or more of the design relations between
FRs and DPs leads to the situation that the Information Axiom cannot be fully satisfied.
Note that the definition of CR is based on joint probability or the sum of all information in
the design relations. Therefore, the mature state is only reached if all system- and
design-ranges are matched (Fig. 12.7). In this case, there is no discrepancy between pre-
sumed and legitimate positions. Problems with non-matching systems and design ranges
can be the result of Unrecognized CR. However, when this is discovered Unrecognized CR
is changed in Recognized CR but an infinitely accurate determination cannot be made.
12 Complexity in Axiomatic Design 345

Fully Robust Fully Robust

Information Axiom

Information Axiom
Proof of Concept

Proof of Concept
No Organisation

No Organisation
Not Robust Not Robust
Independence Axiom Independence Axiom

Fig. 12.5 Irregularities in the design matrix do not necessarily conflict with satisfaction of the
information axiom. However, if decoupling of the matrix needs replacement of DPs, the
information axiom is not automatically satisfied for the new DPs and efforts may be lost (left). The
second option shows a luckier situation that the DPs can be maintained. In this case the impact on
the design is minimal (right)

Fully Robust Fully Robust

Information Axiom

Information Axiom

Correction
Proof of Concept

Proof of Concept
No Organisation

No Organisation

Correction

Not Robust Not Robust

Independence Axiom Independence Axiom

Fig. 12.6 Discovery of a wrong DP leads to a discontinuity in the development process. In the
unlucky situation that an obsolete DP was already optimized, efforts are lost and the new DP again
needs optimization and a correction takes place (left). In a lucky situation, the problem can be
solved with minor efforts. In this case, the related unrecognized as well the Unrecognized
Imaginary Complexity disappears (right)
346 E. C. N. Puik

Fig. 12.7 Non-matching

Fully Robust

Information Axiom
system- and design ranges

Proof of Concept
prevent the mature state from
being reached. The design
will not become robust.
(Reproduced with permission
from Puik and Ceglarek
(2014b))

No Organisation

Not Robust
Independence Axiom

12.3.5 Summary of the Application of Complexity

in Functionality Diagrams

The complexity theory of AD, plotted in functionality diagrams can be applied to

monitor the development process of products and systems. It has the capability to
monitor if complexity and the axioms are addressed in the right order. Typical
patterns that were defined so far:

• If the designer starts optimizing the design according to the Information Axiom,
before decoupling the design, the horizontal axis will start rising early. It may to
discrepancies in presumed and legitimate positions in the functionality diagram
since the designer is not always aware of this problem.
• The different development paths for SMEs, semiconductor industry, safety
systems, or medical industry, follow different trajectories as indicated in
Fig. 12.4. It is up to the designer or his organization which path is the most
optimal path for each situation.
• It was shown what errors can occur, how to recover from it, and what the
consequence is in terms of work that needs to be redone in Figs. 12.5 and 12.6.
• If the designer does not succeed in making the design robust it will not reach the
fully robust status marked with the dot Fig. 12.7.

In learning organizations like universities but as well as companies, the func-

tionality diagram can serve as a tool to explain the origin of errors made in projects
to students and novice designers. Causes and consequences become clear lessons
for future design projects and it will contribute to the learning experience of the
12 Complexity in Axiomatic Design 347

designer (design team). Communication, supported by these visual means, could

function as a universal language to widen the scope of personnel, increasingly being
capable of understanding what went wrong, for students, engineers, but also
managers, and executives.

12.4 Conclusions of This Chapter

This chapter has explained how complexity in AD is defined and how it can be
applied. Complexity in AD is defined as “a measure of uncertainty in achieving the
specified FRs.” Four main kinds of complexity in AD are defined:

• real Complexity;
• imaginary Complexity;
• combinatorial Complexity;
• periodic Complexity.

The first two kinds of complexity are Time-Independent: these kinds do not
change with time. The last two are Time-Dependent and they indeed do change
with time. There are two kinds of CR and CIm; complexity that is recognized or is
not unrecognized by the designer.
Second, it is shown that complexity in AD being defined in the functional
domain, is very applicable for engineers. As Gell-Mann and Lloyd state it, com-
plexity is the result of a “lack of regularities in the design.” If these regularities are
either; (i) not recognized, or (ii) ignored they will reside in the system and, as a
result, the system will behave in a stochastic manner outside its specifications. On
the other hand, if all irregularities are removed from the system, it will thereafter
behave fully according to expectations and the designer has made a good design.
Third, when complexity in AD is visualized in a functionality diagram, it
becomes a valuable tool to monitor the design as it evolves over time. This diagram
reveals the design consequences of errors in the development process when the
Independence Axiom and Information Axiom are not fully satisfied or if they are
not applied in the right order. A number of typical situations have been analyzed
and the causes of their risks are explained. The acceptable amount of risk differs per
project as lead times, budget, and available resources are not the same. The diagram
can be used to estimate remaining development risks and create awareness of how
to best deal with them.
348 E. C. N. Puik

References
Boone ME, Snowden DJ (2007) A leader’s framework for decision making. Harvard Bus Rev 85
(11):68–76
Bullinger HJ, Warschat J (2012) Concurrent simultaneous engineering systems. Springer Science
& Business Media, Berlin/Heidelberg, Germany
Foley JT, Puik ECN, Cochran DS (2016) Desirable complexity. In: Presented at the 10th
international conference on axiomatic design ICAD2014, Xi'an, 2016, p 6
Gallagher R, Appenzeller T (1999) Beyond reductionism. Science 284(5411):79–80
Gell-Mann M, Lloyd S (1996) Information measures, effective complexity, and total information.
Complexity 2(1):44–52
Goldenfeld N, Kadanoff LP (1999) Simple lessons from complexity. Science 284(5411):87–89
Kurtz CF, Snowden D (2003) The new dynamics of strategy: sense-making in a complex and
complicated world. IBM Syst J 42(3):462–483
Lee T (2003) Complexity theory in axiomatic design. Massachusetts Institute of Technology,
Cambridge, U.S
Puik ECN, Ceglarek D (2014a) A review on information in design. In: Presented at the 10th
international conference on axiomatic design ICAD2014, Lisbon, 2014, 1st ed, pp 59–64
Puik ECN, Ceglarek D (2014b) A theory of maturity. In: Presented at the 10th international
conference on axiomatic design ICAD2014, Lisbon, 2014, 1st ed, pp 115–120
Puik ECN, Ceglarek D (2015) The quality of a design will not exceed the knowledge of its
designer; an analysis based on axiomatic information and the cynefin framework. Proce-
dia CIRP 34:19–24
Royce WW (1970) Managing the development of large software systems. In: Proceedings of
IEEE WESCON, pp 328–338
Shannon CE, Weaver W (1949) The mathematical theory of communication. The University of
Illinois Press, Chicago
Snowden D (2000) Cynefin: a sense of time and space, the social ecology of knowledge
management
Suh NP (1990) The principles of design (Oxford series on advanced manufacturing). Oxford
University Press on Demand, Oxford, England, U.K.
Suh NP (1999) A theory of complexity, periodicity and the design axioms. Res Eng Design 11
(2):116–132
Suh NP (2005a) Complexity: theory and applications. Oxford University Press, Oxford, England,
U.K.
Suh NP (2005b) Complexity in engineering. CIRP Ann 54(2):46–63
Weng G, Bhalla US, Iyengar R (1999) Complexity in biological signaling systems. Science 284
(5411):92–96
Whitesides GM, Ismagilov RF (1999) Complexity in chemistry. Science 284(5411):89–92
Axiomatic Design Application
to Product Family Design 13
Masayuki Nakao and Kenji Iino

Abstract
The basic concepts and the framework of Axiomatic Design (AD) provide
powerful tools in the design of products and product families, especially for
visualizing the design goals and improving the design process. When learning
how to apply AD, however, nearly a half of the uninitiated designers like
students may need to devote much effort to advance a sufficient number of
different design concepts in terms of functional requirements (FRs) and/or design
parameters (DPs), which is often done in abstract phrases like the first step in
AD. The instructors must encourage them to think freely and squeeze out all the
FRs and DPs they have in their minds and must guide them to integrate FRs
functionally and DPs physically to obtain the desired design matrix.

13.1 Axiomatic Design Application for Product Family

Design

13.1.1 Design Concept Description as the First Step

in Axiomatic Design

The idea of AD shows powerful effects in designing product families.

By “product family,” we mean a group of interacting or interrelated entities that
form a unified entirety. In other words, a product family is a unified whole whose
structural elements have effects on one another, thus, applying Independence

M. Nakao (&)
K. Iino

The University of Tokyo, Tokyo, Japan
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 349

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_13
350 M. Nakao and K. Iino

Axiom for decoupling such interference of FRs increases the possibility of satis-
fying all FRs; that is good design.
AD is a universal conceptual enabler without any constraints (Cs) of the design
objects, thus, we can apply it to subjects including engineering designs of
mechanical, electrical, buildings, or chemical processes, as well as such social
demands as planning of structuring organizations, proposing policies, developing
new products, or improving lifestyles; we can apply the method in any creation.
AD, in other words, is a set of general axioms that can effectively support design no
matter what the subject is.
In learning how to apply AD, one can read the methods described in detail in the
earlier chapters. Just reading AD methods, however, may lead only half of the
readers, to quickly acquire the skills in applying them, because they need to
describe the design concept in a natural language as the first step in AD.
When learning AD with Suh’s textbook (Suh 2001), half of the uninitiated
designers like students may not even reach the stage of an axiom application. The
reason is the difficulty for such designers in describing design concepts of FRs or
DPs in abstract phrases instead of actual shapes laid out in drawings. Suh intro-
duced a teaching method to list the FRs and DPs of a beverage can. The can has
only three parts of the body, top lid, and the bottom, but it has more than 12 FRs.
These three parts are physically integrated from more than 12 DPs. For example,
about half of the students missed the FRs of the cylindrical body, the size of the pull
tab, or the beautiful body print showing images about the beverage.
Another useful design example is an ongoing research topic for a bachelor’s or
master’s degree (see problem 2). The students have to list up, at least, ten FRs and
DPs related to arcs to construct the FR–DP charts, and identify the critical DP that
signifies the novelty of the research. Although the students are always concerned
about this requirement, only about half of the students may complete this task
(Nakao and Iino 2018). A half dropped primary FRs or DPs that the instructor could
recognize or made the mistakes of “mixing the FRs of the designer (the project
budget, the project deadline, or so)” as pointed out by Thompson (2013). Without
establishing the FRs and DPs that construct the design, the students cannot proceed
to the next step of applying AD.
To guide the students, as we will describe with case studies in Sect. 13.2, we
have them squeeze out all the FRs and DPs from their brains as a preparatory step
for AD. General methods of mind mapping or work breakdown structure
(Fig. 13.1) will work just fine.

13.1.2 Axiomatic Design Application with Proper Functional

Requirements

For the next step, the students group the design concepts they generated into FRs
and DPs, then connect related FRs and DPs with arcs to produce FR–DP charts.
Next is a key technique in applying AD of listing up an equal number of FRs and
DPs (Nakao and Iino 2018). This step leads to a regular design matrix (regular:
invertible square matrix with a non-zero determinant) that can be decoupled into a
diagonal or triangular one with proper row operations.
13 Axiomatic Design Application to Product Family Design 351

Failing to list up proper FRs and DPs blocks the students from reaching the
entrance to AD application. Many students can describe DPs that are visible,
however, they often cannot spell out the FRs that are hard to visualize. We have to
loudly emphasize “FR first!” or “set FRs under a solution-neutral condition,”
otherwise, they will end up with smaller numbers of FRs compared to those of DPs.
The AD textbook (Suh 2001) teaches that zigzag thinking is effective in setting FRs
and DPs. The zigzagging starts from an abstract high-level concept toward low-level
ones that are easier to picture and alternates between the functional and physical spaces
going FR, DP, FR, DP, and so on. The method leads to an equal number of FRs and
DPs, and at the same time, avoids describing multiple FRs with combinations of the
same DPs. Therefore, the design matrix becomes regular, and its determinant is
non-zero. The situation with students in the early stage, however, lacks efficient
numbers of FRs and DPs in their minds. Thus, even with zigzag thinking, they
overlook important aspects. It is just like an excellent recipe without the right mate-
rials, failing to produce a good dinner. Proper FRs are necessary for AD application.
The instructors twist the students’ arms to list up FRs and DPs, and they tend to
list FRs chronologically and DPs spatially (Nakao and Iino 2018). The way they
work comes from imagining how they would use the product, i.e., the sequence of
work, to list up the FRs, and next referencing the bill of material (BOM) of similar
existing products to list up the DPs. Listing up FRs and DPs in separate mindsets
against the zigzag thinking process naturally leads to discrepancies in their num-
bers. Once the students produce their imperfect FR and DP lists, the instructors and
teaching assistants (TAs) guide the students to integrate FRs functionally and DPs
physically. This step is a grouping of low-level concepts of FRs and DPs, and it can
rearrange the FR and DP vectors to have the same dimension. After this integration,
retrying the above mentioned zigzag thinking can operate perfectly. The students
have enough numbers of the right materials now.
Management professors teach the need for skills in setting the problem in a
“mutually exclusive and collectively exhaustive (MECE)” manner if one wants to
be a business consultant. The first part of mutual exclusiveness is the same as the
Independence Axiom in AD, and engineering students can manage to set mutually
independent FRs by avoiding trade-offs. The second part, collective exhaustiveness,
is more challenging for the students who always have some mind slips. After a
one-semester-long design class or seminar, for any objective designs, half of the
young participants can directly build the design equations by tacitly setting
FRs/DPs and integrating them in the brain.

13.1.3 Axiomatic Design Application with Many Functional

Requirements

Generally, real product family design in industries have so many FRs that the
designers cannot easily check the trade-offs or interferences with their brains or
hands. In later chapters, however, AD shows positive and effective work for large or
complex systems.
352 M. Nakao and K. Iino

A software system design starts by listing up the FRs. This process, with a
business consultant, describes what the customer wants to happen in natural lan-
guage. Then a computer scientist translates the FRs into detailed specification, and
programmers map the specifications to programs to realize the functions. A typical
number of FRs, for these cases, easily exceeds 1,000 and the number of steps of
checking for interferences among them turns out huge to reach 1 million cases, i.e.,
the square of the number of FRs. Testing in software design is said to take about the
same number of days as designing takes, e.g., if the design took one year, its testing
will take another 1 year because checking the interferences takes huge manpower.
AD can split the FRs into explicit ones FRe and implicit ones FRi as shown later
in Fig. 13.8a. The former are those that the customer wants with the design, i.e.,
FRs that AD explained up to the previous chapters. The latter, on the other hand,
are those without customer voices. If the design cannot meet the voice of the
customer, the customer may file claims, and so the makers prepare those FRs to
prevent some risks of future claims. For example, large-scale programs like one for
an automatic teller machine (ATM) in banking is said to have 70% of its program
lines to realize implicit FRs. Examples of these implicit FRs include operation
schedules, future development plans, recovery plans upon problems, transition
plans for new systems in the future, prevention of unauthorized access, aseismic
reinforcement, installation weight, electrical power consumption, and so on. These
problems will arise in situations like; a 24 h a day, 365 days a year operation
without not even a minute of margin for update to program modifications;
expanding the capacity to eight times after a successful operation caused congestion
due to narrow data bus; or loss of electrical power following an earthquake caused
loss of live data, and thus, the makers have to prepare against such emergency
states.
AD often teaches to set these requirements into Cs, such as cost, safety, physical
proximity, durability, and so on, as shown later in Fig. 13.8b. It is adequate if
narrowing the tolerable ranges for DPs alone can satisfy Cs, however, if the nar-
rowing lowers the probability of realizing FRs, the solution is not desirable.
A different method, frequent in practice, is to prepare a separate DP for satisfying an
implicit FR. For example, in preparation against the above problems; halt the
operation for 10 min every day starting at 2 o’clock in the middle of the night;
design the system in advance with a high data transfer frequency to allow 16 times
the expected information transfer volume, or place a mirror server in a city located
1,000 km away. In general, describing the FRi gives better chances of finding
interferences with other FRs, as shown in Fig. 13.8a. For example, security and
electric power consumption relate to all programs, and they result in rows with all
Xs meaning interference with everything.

13.1.4 Creating New Design Using Axiomatic Design

Design assignments of creating new designs, instead of improving existing ones are
now globally common, especially in the information business. This type of new
13 Axiomatic Design Application to Product Family Design 353

assignment, however, gives further hurdles in listing up all the FRs and DPs,
especially the FRs. The problem is not in the lack of linguistic ability to express
concepts but in the overlooking of FRs that will surface later. The main cause is not
recognizing the values of customer attributes (CAs). Some examples are; a change
in a rival organization disturbing the designer, a competitor filing suit on a patent
issue, a customer applying the product in ways the designer did not expect, a
sudden change in regulation that prohibits using the product, or a workers’ strike
unrelated to the designer’s responsibility. In such cases, the designer has to set a
new set of CAs and FRs, and creation is always faced with such changes in
reaching a successful design.
To find what element is missing from the formation, relying on imagination
while sitting in the office will never lead to discovery. One will have to quickly go
through the cycle of the first prototype, testing, improvement, the second prototype,
testing, improvement, and so on, to find what concepts are missing from the for-
mation. Mark Zuckerberg said, “Done is better than perfect.”
The design solution is not necessarily unique. The FR itself, changes with the
customer and situations that surround the society, forcing changes in the optimum
DP. This transformation makes the design different from mathematics that has a
single unique and eternal solution, and that is what gives compelling attraction to
the act of designing. One of the most effective design methods is AD when we want
to teach the philosophy of design to young designers visually.

13.2 Product and Product Family Design Cases Using

Axiomatic Design

13.2.1 Automatic Driving

Figure 13.1 shows the method for exhausting design concepts with the example of
designing an automatic driving system. Figure 13.1a is the result of applying mind
mapping, and (b), work breakdown structure. Both methods start from a single
concept and reach multiple concepts following the association game method. They
also allow grouping of concepts so the player can exhaust all concepts without
leaving out any. In the end, the designer separates the FRs and DPs, for example, by
collecting verbs for FRs and nouns for DPs, and in step (c), they are aligned in the
FR–DP chart with arcs connecting related FRs and DPs. As noted with gray bal-
loons in (c), the discrepancy in the counts of FRs and DPs is evident, as well as
design interference indicated with intersecting arcs.
When a designer is at the stage in (c), the design matrix is irregular and coupled,
and advancing to the decoupling phase is quite discouraging. These problems look
complicated in design. We thus tried concept integration. As the dotted boxes of
stage (d) shows, for example, “shoulder, pedestrians, and vehicles” are all obstacles
and can form a single group FR, and “GPS + map, steering, brake, and gas pedal”
are all in constant use to form a single group DP. The resulting design equation for
354 M. Nakao and K. Iino

Fig. 13.1 Design of automatic driving. (Reproduced from Nakao and Iino (2018), originally
published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201822301011)
13 Axiomatic Design Application to Product Family Design 355

(d) is 4D and is easy to understand the design definition. An interference is seen in

the column in the dotted box for DP1, “GPS + map + actuators.” Without them, no
matter how sophisticated the sensors may be, there is no way to accomplish
automatic driving. In other words, they are the key technologies. The design
equation in (d) solved these problems, and now we can use the axioms of AD for
decoupling.

13.2.2 Fan Design

Automatic driving we saw in Fig. 13.1, with a great deal of attention from the
society and a large number of articles about it in a variety of journals and magazines,
allows the students to search the internet and easily collect articles and pick up
concepts of FRs and DPs from them. A fan design in Fig. 13.2, on the other hand, is
a mature product, and there are no articles that discuss it. The students have to think
for themselves. Figure 13.2a is a typical FR–DP chart by a student who visualizes a
fan in the air and sets the FRs following the process of activating one, while on the
other hand, the student virtually disassembles one and sets the DPs following the
BOM. Naturally, the two methods force different mental processes. Thus, the
numbers of FRs and DPs do not match with intersecting arcs for related FRs and
DPs. In this unstructured situation, zigzag thinking does not work well, either.
To escape the situation, we rearrange the breakdown by integrating multiple
lower level FRs into a single FR at a higher level like “set airflow power” or “stop
upon falling asleep,” or combine related DPs into a higher level module DP like
“motor + fan” or “motor + knob.” For the DP “cover” without a corresponding FR
in (a), we add the hidden FR of “injury-free finger poking” in (b). A hidden FR is
one unnoticed during the early stage of design. The DP with influences on all FRs,
shown with a corresponding column with all Xs is “motor + fan.” This interference
shows that this DP is the key technology for the product fan.
Figure (c) shows the design equation for the bladeless fan that was a recent hit
product. Its shape is clearly different from a conventional model, but it only has an
additional attractive FR of “hide blades.” All the remaining FRs are carried over
from a conventional model. The attractive FR, however, was so effective. The FR
brought the large value of being “bladeless,” and led consumers to purchase them at
$300 even though a conventional model would only cost $50. Within the set of
DPs, the novel technologies are “ring-shaped blower” and “place blades inside the
base.” The key technology remained with “motor + fan,” but a new small syn-
chronized motor with rare-earth magnets hid the motor in the base.

13.2.3 Entrance Exam Administration

Figure 13.3 shows the case of “Planning administration of admission exam.”

Shown in (a) is the first FR–DP chart. A student listed up the FRs following the
schedule and wrote down the DPs looking at the list of stakeholders. Naturally, the
356 M. Nakao and K. Iino

Fig. 13.2 Fan design. (Reproduced from Nakao and Iino (2018), originally published open
access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/201822301011)

FR and DP counts did not match, and intersections were there among their relations.
Applying the integration techniques, respectively, to the FRs and DPs led to a 4D
design equation in (b). In the end, a column with all Xs in the design matrix that
influenced all FRs was the DP of “exam committee.” The committee takes the
leading role in all aspects with the responsibility to all the FRs. As shown in (c), a
flaw in the exam questions one year was found after the exam was over, and the
university received social blame. For the following year’s exam, a hidden FR of
“eliminate errors in questions” was added with a corresponding DP of “exam
review committee” consisting of young teaching staff tackling the exam questions
13 Axiomatic Design Application to Product Family Design 357

in advance of the real exam. The DP of “exam committee” appears to also have
influence on this FR, however, such an influence would discourage the young staff
to point out errors by tenured professors. Thus, this exam review committee alone
was kept independent on purpose.

Fig. 13.3 Planning administration of admission exam. (Reproduced from Nakao and Iino (2018),
originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201822301011)
358 M. Nakao and K. Iino

13.2.4 Umbrella that Follows the Owner

Figure 13.4 is the design of an “Umbrella that follows the owner.” It was a student
creation in a design exercise class. The first idea was to mount an umbrella on a
drone. However, that resulted in a noisy follower like a mosquito above the head.
The next design iteration was a helium-filled balloon to counter the weight of the
umbrella and a pair of propellers mounted on the two sides to control
forward/backward, and left/right turns. A camera mounted on the umbrella balloon
recognized a red hat and controlled the propellers to follow its motion. The test
session resulted in the balloon flying away after 20 s or so following the hat, and
the testers had to pull the balloon back with the “emergency string.” The designers
had failed to recognize the FRs of controlling rolling and pitching. Only two
propellers were insufficient to control rolling and pitching additionally. Looking
into an airplane design led the team to find the need for a tail wing. Also, the camera
had a narrow view angle and would easily lose sight of the red hat. The students
placed a fish-eye lens on the camera to counter this problem.
What improvements to make are easy to find through quick prototyping and
testing. Many large-sized corporations like to “start with a perfect solution” and
extend the development period. However, they often lose their business chances.
Startups like to quickly place products still under evaluation into the market and
have the market tell them what improvements to make. The latter attitude is needed
for creative design to find hidden FRs.

13.2.5 Stirling Engine

Figure 13.5 shows two sets of FRs of a Stirling engine, one when they are set
following the chronological operation, and the other following functional evalua-
tion of laws of thermodynamics. The former referenced the case of setting FRs for a
steam engine in Suh’s textbook (Suh 2001). Four FRs of producing hot air, raising
the piston, producing cold air, and lowering the piston form a lower triangular
matrix. The latter FR set, on the other hand, from the point, that the difference in
injection and extraction of heat produces work, sets four FRs of injecting heat,
extracting heat, doing work, and repeating the cycle. The two designs are physically
different from different sets of FRs and DPs. Both design matrices, however, are
also lower triangular ones. In other words, both approaches lead to correct answers
for decoupling. The difference in their descriptions comes from matrix multipli-
cation, just like performing a coordinate transformation to FR and DP. The burner
also heated the air cooler; the difference between hot and cold temperatures became
zero; the engine eventually stopped.
This discussion showed that there are cases of describing FRs and DPs of the
same machine in design equations with different concepts, but both descriptions are
correct.
13 Axiomatic Design Application to Product Family Design 359

Fig. 13.4 Design of umbrella that follows the owner. (Reproduced from Nakao and Iino (2018),
originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201822301011)
360 M. Nakao and K. Iino

Fig. 13.5 FRs of Stirling engine that are listed chronologically or functionally. (Reproduced from
Nakao and Iino (2018), originally published open access under a CC BY 4.0 license: https://doi.
org/10.1051/matecconf/201822301011)

13.2.6 CurcurPlate for Managing Peoples in a Building

This section introduces “curcurPlate,” a software system designed for monitoring

people’s whereabouts. Implicit FRs, mentioned in Sect. 13.1.3, are introduced and
Fig. 13.6a illustrates the implicit FRs (FRis), compared with Cs. Although AD
allows both methods, preparing new solutions (DPis) for a set of new FRis usually
is more feasible than narrowing the DP ranges against new Cs. Both FRis and Cs
may have many couplings with other DPs as shown in FRi1 or C1 in Fig. 13.6.
Figure 13.7a shows the presence display panel, a hardware system placed at our
office entrance that lab members can flip their nameplates to show their presence
and absence (FR1). If one is running an experiment in a lab other than the office, a
little magnetic sticker with the name of the lab placed on the steel nameplate shows
13 Axiomatic Design Application to Product Family Design 361

Fig. 13.6 Implicit FRs versus constraints for preventing the future trouble

the whereabouts (FR2). Further, in case of an emergency like an earthquake or fire,

any lab member can take the entire frame to the evacuation site and if someone
present is not around at the site, others can head out for rescue (FR3).
Figure 13.7b is “curcurPlate” the tablet version of this tool. The phrase “ku-
rukuru (curcur)” is the onomatopoeic word for flipping a nameplate. DPe1 and DPe2
are input by tapping, and DPe3 is to store the data in a remote server so one can
output the data in case of evacuation. The structure is simple and free of interfer-
ence even with 200 laboratories using it.
When put in practice, however, everyone trying to update their whereabouts
information caused a delay in the server response, and an increased number of labs
further pushed back the response and the system needed to counter this problem
(FRi1). Moreover, if an earthquake or fire shuts down the server, data immediately
before the evacuation are unavailable as well as the state of evacuation (FRi2).
Another requirement rose to register the whereabouts information from off-campus
362 M. Nakao and K. Iino

Fig. 13.7 “CurcurPlate” design with implicit FRs

locations (FRi3). These implicit FRs greatly affected the success of the tool, espe-
cially “FRi1: Respond within 3 s with 1000 simultaneous updates” led to its
acceptance.

13.2.7 Tool for Brushing the Back of Teeth

This section discusses a tool for brushing the back of teeth with some constraints.
The target is to design a tool that allows brushing the back of teeth for elderlies that
13 Axiomatic Design Application to Product Family Design 363

cannot open their mouths widely and avoid aspiration pneumonia at the same time.
The conventional solution has a caregiver insert a thin toothbrush into a gap of only
about 1 cm and blindly brush the back of teeth as Fig. 13.8a shows. FR1 is “brush
the back of teeth after meal” and FR2 “insert the brushing tool through a 1 cm gap
between the upper and lower teeth.” The conventional method used a toothbrush
(DP1) and a small brush head (DP2) to insert it. However, the small brush head
failed to give a good thorough brushing and interfered with FR1. We set the third

Fig. 13.8 A tool for brushing the back of teeth

364 M. Nakao and K. Iino

requirement FR3, “check if the brushing was successful,” however, there was no
way of looking at the back of the teeth so the third parameter DP3 “gargle with
fluorine mixed water” did not satisfy FR3. In the end, sniffing the mouth was the
method for judging how good the brushing was.
We then replaced the toothbrush with water jet (DP’1), as Fig. 13.8b shows.
After the nozzle was inserted past the teeth, waterjet squirted out through a bent
nozzle (DP’2), and the tool successfully cleaned the entire back of teeth without
interference with FR1. We even attached an LED mounted miniature camera (DP’3)
by the nozzle to directly check if there is still food debris left especially between
teeth. So far, so good.
When we, however, wanted to apply the solution to visitors that cannot open
their mouths wide, safety (C1) turned into a large obstacle. We claimed that “it is
just brushing of the teeth,” however, if a dentist or hygienist uses the tool, it is a
medical procedure. We had to demonstrate through experiments that the user will
not gag with water, and the nozzle will not come off to choke the user, or the
electricity to the camera with illumination will not electrocute the user. Next, we
had to run the experiments against a variety of people, mandatorily thinking, ethics
(C2). We had to repeat explaining that waterjet is a device available to the general
public and anyone can use one, the cleansing tool will not stay within the body, and
that we will not keep the private information of teeth data and will erase them, but
the ethics committee gave us a hard time to reach approval to use it.

13.3 Conclusions

The idea of AD shows powerful effects, especially for visualizing the design def-
inition and improving design problems in product family design no matter what the
subject is as shown in Figs. 13.1, 13.2, 13.3, 13.4, 13.5, 13.6, 13.7 and 13.8. When
learning how to apply AD, half of the beginner designers, however, fail to describe
enough numbers of design concepts of FRs or DPs in abstract phrases. The
instructors have them squeeze out all the FRs and DPs in their minds as a
preparatory step for AD with general methods of mind mapping or work breakdown
structure as Fig. 13.1 shows. For the next step, the instructors guide them to
integrate FRs functionally and DPs physically for getting a regular design matrix, as
shown in Figs. 13.1, 13.2, 13.3, 13.4. We also discussed further applications.
Finally, some tacit requests which customers do not claim should be set as implicit
FRs or Cs, as shown in Figs. 13.6, 13.7 and 13.8.
Problems
1. Design your future life. Here, you need to set money as FR, DP, or C: the dream
to become a millionaire (FR), the inevitable tool to eat enough meals or enjoy
the hobbies (DP), or one of the minimum necessary resources like health or
academic background to realize your FRs (C).
13 Axiomatic Design Application to Product Family Design 365

2. Make the design equation on your current research or your job. You should
clarify the purpose (FR) and the method (DP), at least. Do not mix the FR of the
designers, that is, the project budget, the project deadline, the promotion, the
thesis, and so on.

References
Nakao M, Iino K (2018) Students list FRs chronologically and DPs spatially, and need to integrate
FRs functionally and DPs physically. In: Puik E, Foley JT, Cochran D, Betasolo M (eds) 12th
international conference on axiomatic design (ICAD). MATEC web of conferences, Reykjavík,
Iceland
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Thompson MK (2013) Improving the requirements process in axiomatic design theory. In: Annals
of the CIRP, 1, vol 62, pp 115–118 (2013)
Design of Large Engineering Systems
14
Gyung-Jin Park and Amro M. Farid

Abstract
One defining characteristic of twenty-first-century engineering challenges is the
breadth of their scope. The National Academy of Engineering (NAE) has
identified 14 “game-changing goals.” At first glance, each of these aspirational
engineering goals is so large and complex in its own right that it might seem
entirely intractable. Fortunately, design science provides a continually advancing
perspective built upon a meta-problem-solving skill set. This chapter introduces
the design engineer to the world of large complex systems from an Axiomatic
Design (AD) perspective. In particular, the chapter focuses on two critical
“systems-thinking” design skills that help the designer manage the inherent and
abstract complexity of large systems. They are (1) system decomposition and
(2) the allocation of function to form. The chapter also practically demonstrates
these design skills in several design stories and case studies. The chapter
discusses why and how these design skills are used differently when the system
has a fixed versus flexible structure. Finally, the chapter concludes with several
avenues for further investigation.

G.-J. Park (&)

Hanyang University, Ansan, South Korea
e-mail: [email protected]
A. M. Farid
Dartmouth College, Hanover, NH 03755, USA
e-mail: [email protected]; [email protected]

A. M. Farid
Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, USA

© Springer Nature Switzerland AG 2021 367

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_14
368 G.-J. Park and A. M. Farid

14.1 Introduction

There are two kinds of large systems. The first kind is a tree-like large system. It
starts with a limited number of functional requirements (FRs) and design parameters
(DPs) at the highest level but requires many layers of decomposition for actual
implementation, involving a large number of lower level FRs to satisfy the limited
number of the highest level FRs. For example, the original idea for making a
dispersion-strengthened copper alloy had only a limited number of FRs at the
highest level. The central idea was to manufacture a dispersion-strengthened copper
alloy that consists of a pure copper matrix phase with a plethora of nanoscale
titanium di-boride particles dispersed throughout the copper matrix as dispersoids to
strengthen the alloy. The process designed involved high-speed impingement
mixing of Cu/Ti solution with Cu/B solution. The mixed liquid was then rapidly
quenched to create an alloy with copper matrix phase with nanoscale titanium
di-boride particles dispersed throughout the matrix.
To satisfy the high-level FRs, they had to be decomposed, creating many lower
level FRs and DPs. For example, the mixture of Cu/TiB2 solution had to be quickly
solidified on a cold spinning copper disk to prevent the coagulation of the ceramic
TiB2 particles. Then, the liquid mixture of Cu and TiB2 alloy was rapidly quenched
on the rotating copper disk. Then, the ribbon of the Cu/TiB2 alloy was shredded.
The shredded chips of Cu/TiB2 alloy were then compacted into a solid rod using a
high-pressure hydrostatic compaction/extrusion process.
In this case, the number of FRs at the highest level was relatively small.
However, when the design of the process was finalized, the entire manufacturing
system became quite large. The final system designed had many FR, DPs, and PVs
with many layers of decomposition, making the system large with many lower level
FRs and DPs.
The second kind of a large system has a large number of FRs at the highest level
from the beginning. For example, if we are designing an airport for a major hub,
there are many highest level FRs that must be satisfied. These FRs, in turn, generate
many lower level FRs, making the system very large. The materials presented in
this chapter are equally applicable to any large systems design.

14.1.1 Motivation

One defining characteristic of twenty-first-century engineering challenges is the

breadth of their scope. The National Academy of Engineering (NAE) has identified
14 “game-changing goals”:

1. advance personalized learning;

2. make solar energy economical;
3. enhance virtual reality;
4. reverse engineer the brain;
5. engineer better medicines;
14 Design of Large Engineering Systems 369

6. advance health informatics;

7. restore and improve urban infrastructure;
8. secure cyberspace;
9. provide access to clean water;
10. provide energy from fusion;
11. prevent nuclear terror;
12. manage the nitrogen cycle;
13. develop carbon sequestration methods;
14. engineer the tools of scientific discovery.

At first glance, each of these aspirational engineering goals is so large and complex
in its own right that each might seem entirely intractable. Furthermore, each goal
might appear so different from the next that an aspiring engineer might easily
conclude that the skills needed to solve one challenge are entirely distinct from
those of another. Consequently, our engineering education system would have to
turn “on a dime,” orient itself toward each of these 14 challenges, and ask our
first-year engineering students to commit themselves to one of these challenges;
never to change direction again. And in the unlikely event that we are successful on
such a course, the engineering education system would have to pivot again years
later to address the newly cropped up grand challenges.
Fortunately, design science provides an alternative and continually advancing
perspective. While each of these aspirational NAE goals might seem entirely dif-
ferent, in reality, they exhibit many common characteristics which can be integrated
into a consistent design framework. In time, this design framework increasingly
spans individual engineering disciplines and real-life problem domains. It also
increasingly sets aside limiting paradigms and assumptions in its quest toward a
refined meta-problem-solving skill set.

Design Story 14.1:

The design engineer reconsiders the 14 NAE challenges in the context of the four
AD engineering domains shown in Fig. 14.1.

• First, they recognize that the “design solution” to each of the grand challenges
can be viewed as a newly designed large complex system that is very much a
refined version of the large complex system that exists in its place today.
• Second, they realize that unlike “traditional products,” these systems have not
just one “customer” in the stakeholder requirements (SRs) domain but rather a
diversity of internal and external stakeholders. These stakeholders impose a
wide variety of hard and soft requirements which must ultimately be resolved
into the FRs and constraints (Cs) of the functional architecture domain.
• There, the aspiring design engineer finds that large complex systems are char-
acterized by a large number of FRs. Managing such a large number is a psy-
chological and organizational challenge in its own right.
• Beyond just number, the design solution is complicated by the heterogeneity (or
diversity) of the FRs. Functions are closely tied to siloed engineering disciplines.
370 G.-J. Park and A. M. Farid

synthesis synthesis synthesis

SR FR DP PV

analysis analysis analysis

Stakeholder Functional Physical Process

Requirements Architecture Architecture Architecture
Domain Domain Domain Domain

Fig. 14.1 Four domains in the engineering design of systems—an Axiomatic Design perspective

At their simplest, an electrical engineer knows electricity, a chemical engineer

knows chemical reactions, and a mechanical engineer knows classical
mechanics. A diversity of function means that the design engineer must either
absorb the functions associated with other disciplines themselves or work
effectively with other design engineers who have.
• It is at this point, when the design engineer crosses the synthesis path from the
functional architecture domain to the physical architecture domain and its
associated DPs that they face their greatest challenge: creativity. The word
strikes fear in the hearts of many young designers. As young designers our-
selves, we recall being perpetually in awe of how our design professors seemed
to magically come up with design solutions from “thin air.” And yet, creativity is
not a binary characteristic encoded in DNA. Rather, it is cultivated by a will-
ingness to loosen and expand one’s established mental constructs, whether by
picking up new books or immersing oneself amongst and appreciating the
perspectives of a broad diversity of people. Each person, their field, and personal
set of experiences can be characterized by a set of “mental constructs” that serve
as “meta-design-parameters” that the design engineer deploys creatively in the
moment of synthesis. The design engineer soon recognizes that the 14 NAE
goals are indeed challenges for the specific reason that they require a broad
number and diversity of such “meta-design-parameters.” No single design
engineer will have them all, but design teams can produce maximally effective
large complex systems by fostering an environment where new ideas are
encouraged and easily communicated.
• As the design engineer implements their solution and crosses over to the process
architecture domain, they find that the large complex system has already been
implemented—at a cost of millions, billions, or even trillions of dollars—poorly
no less—in the form of the legacy system! The likelihood of implementing their
14 Design of Large Engineering Systems 371

“forward-engineering design” is low. Despair can set in. But with newfound
energy, the design engineer sets off in the reverse direction; now along the
“analytical path,” “reverse-engineering” the legacy system. They task them-
selves with modeling the PVs, DPs, FRs, and SRs of the legacy system; knowing
full well that they will have to make a sequence of meaningful piecewise
transformations toward an “ideal” design solution that reconciles their forward
design with the existing system all while it remains operational.

14.1.2 Chapter Contribution

This chapter introduces the design engineer to the world of large complex systems
from an AD perspective. It would be entirely intractable to try to address all of the
specific challenges mentioned in the large complex system design story relayed
above. Rather, this chapter focuses on two critical “systems-thinking” design skills
that help the designer manage the inherent and abstract complexity of large systems.
They are (1) system decomposition and (2) the allocation of function to form. The
chapter also practically demonstrates these design skills in several tractable design
case studies. The chapter also discusses why and how these design skills are used
differently when the system has a fixed versus a flexible structure. The chapter
concludes with a discussion of several open challenges in the design of large
complex systems.

14.1.3 Chapter Outline

The remainder of the chapter is organized as follows: Section 14.2 treats the AD of
large fixed systems. Section 14.3 then provides several design stories of large fixed
systems. Finally, Sect. 14.4 discusses large flexible systems and contrasts them
with the large fixed systems discussed earlier in the chapter.

14.2 Axiomatic Design of Large Fixed Engineering

Systems

14.2.1 What Are Large Fixed Engineering Systems?

The previous section used the examples of the 14 NAE challenges to motivate the
topic of large complex systems. This section addresses an important subset of these
called “large fixed engineering systems (Farid and Suh 2016).” To gain insight, the
term must be deconstructed into its constituent words.
There is no shortage of definitions in the literature for the term “system.” This
chapter adopts the definition below:
372 G.-J. Park and A. M. Farid

Definition 14.1. System: A set of components (subsystems, segments) acting

together to achieve a set of common objectives via the accomplishment of a set of
tasks.
Note that the mere definition of the term requires the definition of three more
abstract systems thinking concepts:

Definition 14.2. System Boundary: The delineation between the system and its
environment or context.

Definition 14.3. System Form: What a system “is.” It includes a description of

1. all the system’s components (or DPs);

2. how the components are interconnected;
3. what portion of the total system behavior/function is carried out by each
component.

Definition 14.4. System Function/Behavior: What a system “does.” It is its

reason for existence. A set of subfunctions (or FRs) that must be performed to
achieve a specific objective.

Returning to the AD framework shown in Fig. 14.1, the system form constitutes
the physical architecture domain and the system function constitutes the functional
architecture domain.
“Large systems” are referred to as such because they have a “large” number of
system elements; be they FRs or DPs. How large is a “large number”? For all
practical purposes, the answer is driven by the psychological limitations of the
human mind as it attempts to design the system. In 1956, Miller recognized that
human beings can typically recall 7 ± 2 numerical digits in short term memory.
Consequently, as a rule of thumb, the literature refers to systems with approxi-
mately 7 elements as “small-sized,” 72
50 elements as “medium-sized,” and 73

300 elements as “large-sized.” Interestingly, at 74
2500 elements or greater, the
system can no longer be designed practically by a single designer (or system
architect). Instead, multiple designers or design teams with their respective
responsibilities must cooperate to produce a well-functioning large system.
The mere existence of multiple designers implies that they must manage the
interdependencies between the systems’ elements. The term “complex system” is
often used to refer to systems with a large number of interdependencies between its
constituent elements. For clarity, it is useful to distinguish between two types of
such interdependencies:
Definition 14.5. Interactions: An interdependency is caused by the sequence of
one FR followed by another.
14 Design of Large Engineering Systems 373

Definition 14.6. Interfaces: An interdependency is caused by the relationship

between one DP and another. For example, there may exist a flow of matter
between two components.

One can imagine that a system with N elements can have up to N2 interdepen-
dencies. Consequently, a large system by virtue of its number of elements is likely to
be complex as well by virtue of the interdependencies between these elements.
As the large complex system grows in size, potentially over many years, it is
likely that its elements and their interdependencies will change. Consequently, AD
distinguishes between “large fixed engineering systems” and “large flexible engi-
neering systems.”
Definition 14.7. Large Fixed Engineering System: An engineering system with a
large set of FRs which do not evolve over time and whose components (DPs) also
do not change over time.

Definition 14.8. Large Flexible Engineering System: An engineering system

with many FRs that not only evolve over time but also can be fulfilled by one or
more DPs.

Finally, this book is devoted to “engineering systems” rather than “systems”

broadly; in that the latter includes (natural) systems (e.g., the solar system, the
human body) that are not engineered but still adhere to the three minimal
requirements of being a system stated above.

14.2.2 Divide and Conquer: Decomposition of System

Hierarchy

The first critical “systems-thinking” design skill is decomposition of the system

hierarchy. Here, the engineer must use a “divide and conquer” mentality to manage
the design of the large system. In brief, the system’s elements; be they FRs or DPs
must be decomposed into their constituent parts (Park 2007).
For the moment, let’s assume a large fixed engineering system. By Definition
14.7, it has many FRs. Following Fig. 14.1, these were identified in a requirement
engineering process where the requirements of the SRs domain were transformed
into the high-level FRs of the functional architecture domain.
To recall, the FRs describe what the system must do, and sometimes they are
simply referred to as the systems’ functions to describe what the system does. Said
differently, they are a mutually exclusive and collectively exhaustive set that
describes the system functionally. Each FR must be defined in a solution-neutral way
that doesn’t presuppose the technologies of the design solution. By convention, each
function is defined as a transitive verb stated in the third person singular followed by
its associated object/operand. For example, in a home design, one FR may be:

FR1 = Protect Internal Climate.

374 G.-J. Park and A. M. Farid

There is a lot packed into such a “high-level” FR. For this reason, it is often
necessary to decompose each function into a number of lower level of functions that
aggregate to achieve the same end. In decomposing a given FR, the designer must
be careful to keep the lower level FRs mutually exclusive and collectively
exhaustive. For example, FR1 can be decomposed into:

FR1.1 = Keep out Moisture;

FR1.2 = Damp out Hot/Cold Fluctuations in the External Environment;
FR1.3 = Heat and Cool Interior Area to desired temperature;
FR1.4 = Redirect falling rain away from the house;
FR1.5 = Admit natural light;
FR1.6 = Protect from Insects;
FR1.7 = Allow entry/exit of inhabitants;
FR1.8 = Protect from Intruders.

The designer can further decompose each of these FRs into even lower level FRs
until the large system has sufficient functional detail to be fully realized. In an
abstract and general sense, this iterative process generates a functional hierarchy as
shown in Fig. 14.2. Furthermore, in many cases, it is useful to explicitly identify the
functional interactions between each system function at each level of functional
decomposition. At such a point, the functional hierarchy becomes a complete
functional architecture.
In an analogous fashion, a large system also has a physical hierarchy composed
of multiple decompositions of DPs; be they systems, subsystems, components, or
single numerical parameters. At the highest level of hierarchy, the large system is a
single DP. For example, in-home design:

People People
Money
Energy
System Money
Energy
Material
Signals A0 Function Material
Signals

System System System

A1 Function Function A3 Function
A2
Decomposition

System
Function
A4
Aggregation

System System
Function
A2.1 A2.2 Function

Fig. 14.2 A functional architecture that has been decomposed two levels. Parallel and serial
interactions are shown between each function
14 Design of Large Engineering Systems 375

DP1: House’s External Barrier.

This high-level DP can be decomposed into several lower level DPs (as
subsystems):

DP1.1 = Waterproof Shell;

DP1.2 = Insulation layer;
DP1.3 = Air-source heat pump;
DP1.4 = Gutter system;
DP1.5 = Window;
DP1.6 = Window screens;
DP1.7 = Doors;
DP1.8 = Door locks.

Finally, when the physical hierarchy is depicted with the physical interactions
between each DP, it becomes the physical architecture. At such a point, it resembles
the functional architecture in Fig. 14.2; but with the functions replaced with
components (or DPs).
Functional and physical decomposition create the AD dual hierarchy shown in
Fig. 14.3. It serves as the primary means by which a designer tackles the inherently
large number of elements (i.e., FRs and DPs) in the system. Rather than view the
large system as a loose collection of functions and components, these sets are
organized methodically in a tree-like structure. The primary advantage is to reduce
the “mental-load” of the designer so that they are only thinking of the elements that
are directly related within the associated hierarchy.

Fig. 14.3 The Axiomatic Design dual hierarchy: Functional decomposition, physical decompo-
sition, and the allocation of function to form in a “Zigzagging” process
376 G.-J. Park and A. M. Farid

One question that often arises is how to organize the functional and physical
hierarchies. One may imagine that a given FR or DP can be decomposed all at once
into many (potentially hundreds) of elements. Alternatively, one may imagine that
the designer organizes these elements into many decomposition layers. In theory,
the choice of the number of decomposition steps is arbitrary. In practice, it matters
tremendously. For example, the physical hierarchy may be organized into assem-
blies and subassemblies that have real manufacturing significance. Similarly, design
teams may have certain functional expertise that drives how we may think of the
FRs. Finally, the 7 ± 2 rule mentioned above provides a practical “design rule of
thumb” from which designers should not depart too far.

14.2.3 Allocation of Function to form—the Zigzagging

Process

The second critical “systems-thinking” design skill is the allocation of function to

form in what is called the “zigzagging process.” In the previous section, functional
and physical decomposition were presented independently. In reality, a given FR
can rarely be decomposed without first assuming some technological solution that
fulfills it. In the meantime, decomposing a given DP has little meaning without
understanding the FRs that the decomposed DPs fulfill. Consequently, in
forward-design, the designer considers a given FR, conceives a DP to fulfill it, and
then allocates the function to this new element of form. At that point, the designer
decomposes the FR while assuming the newly conceived DP. In such a way, the
two critical design skills of decomposition and function-to-form allocation are used
in an alternating fashion. A single designer can handle several zigzagging processes
to accommodate several hundred DPs. Beyond that, a design team can work
together to accommodate a system of potentially arbitrary size.
As part of the zigzagging process, AD keeps track of the allocation of function to
form using a design equation:

FR $ f ðDPÞ

where FR is the set of functional requirements on a given level of decomposition,

DP is the set of design parameters on the same level of decomposition, and f() is a
function representing the laws of physics that govern the design and the relatively
new symbol $ means “satisfies” when read from right to left. When the $ symbol is
replaced with an = and the first derivative is taken it yields a linear equation:

DFR ¼ BDDP

where B is the design matrix. The Independence Axiom instructs designers to

ensure that this design matrix is square and diagonal to reach an uncoupled design,
and if not then square and lower triangular to reach a decoupled design. When the
14 Design of Large Engineering Systems 377

1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8

1.1 X
1.2 X X
1.3 X X X
1.4 X X
1.5 X X
1.6 X X
1.7 X X
1.8 X X

Fig. 14.4 Axiomatic Design matrix of a house’s external barrier

design matrix is anything else, the design is said to be coupled and it violates the
Independence Axiom. For example, the decomposed FRs and DPs mentioned in the
previous section yield the design matrix below (Fig. 14.4).
One important consequence of the Independence Axiom is that it affects the
project management of the design or the design workflow. A designer’s task is to
complete the selection of all the DPs that fulfill the FRs. The design matrix shows
the selection of DPs must be taken in groups. The group of DPs that fulfill a given
FR is called a module. The design matrix also shows that there is a specific order in
which to design each module. In the case of an uncoupled design, the modules can
be designed entirely independently in parallel. As shown in Fig. 14.5, in a design
flow diagram, such a case is defined by a summation (Ⓢ) junction of two or
modules. In the case of decoupled design, Fig. 14.5 shows a control ( ) junction to
indicate the design of one module must follow another sequentially. Finally, in the
case of a coupled design, Fig. 14.5 shows a feedback ( ) junction to indicate that
the design of one module is in an iterative feedback loop with another module. The
problem with such feedback loops is that it is not always clear how many design
iterations will be required to escape the design feedback! In the design of large
systems, such design feedback loops can be entirely debilitating and ultimately
bring design progress to a screeching halt! In short, the Independence Axiom does
not just facilitate good designs, it also facilitates efficient design workflows.
To summarize, the forward-design zigzagging process follows several rules. The
designer must:

1. Conceive the DPs associated with the FRs in the same level.
2. Respect the Independence Axiom when allocating function to form. The design
is ideally uncoupled and otherwise decoupled. If necessary, draw a design
workflow diagram to operationalize the sequence of design task amongst the
design team.
3. Decompose the FRs of the subsequent level based upon the choice of DPs in the
level immediately above.
378 G.-J. Park and A. M. Farid

Fig. 14.5 Elements and

junctions in a design
workflow diagram. Summing,
control, and feedback
junctions correspond to
uncoupled, decoupled, and
coupled designs, respectively

4. Ensure that the FRs and DPs in a lower level are a mutually exclusive and
collectively representation of the FRs and DPs in the level immediately above.
5. Continue the zigzagging process until the AD dual hierarchy includes sufficient
detail to implement the design solution.

Zigzagging in reverse-engineering simply runs in reverse. The designer must

start from the individual components (or DPs) and then deduce the associated FRs.
From there, the designer proceeds to a higher level of aggregation in the DPs and
then deduces the associated FRs on that level. Such a reverse-engineering approach
is necessary when a part or a whole of the system has already been built and the
development of the functional architecture is required to determine how to “evolve”
the system forwards to a more advanced stage of development. In some cases, the
deduction of the associated FR is straight forward. Many tried-and-true physical
solutions have well-known functions. For example, I-beams in buildings support
weight and railways transport trains.

14.3 Examples

The two critical “systems-thinking” design skills of system decomposition and the
allocation of function to form are surprisingly powerful tools in the engineer’s
armament toward the design of large complex systems. This section presents case
studies where these two skills are applied practically. The provided examples have
been chosen for their manageable size so as to facilitate learning.
14 Design of Large Engineering Systems 379

14.3.1 Axiomatic Design of the Mount Type Air Conditioning

System

The mount type air conditioning system is a type of heating, ventilating, and air
conditioning (HVAC) control system that is installed between a ceiling and the
ceiling boards of a room to control room temperature. Such a product is investi-
gated from AD point of view (Lee et al. 2009). The overall decomposition of the
system hierarchy is established down to the “atomic” level components. Some
coupled aspects are found, but ultimately kept at the request of the design sponsor.
The sponsor wants to keep some of the existing component-level parts, and so the
coupled aspects of the design are reported as warnings in the design process.
The basic operating principle of the mount type air conditioning system is to
absorb heat in a room and emit it to the exterior. Figure 14.6 illustrates the com-
ponents of the mount type air conditioning system which consists of indoor and
outdoor machines. Most of the air conditioning systems employ the refrigeration
cycle using a refrigerant to generate cool air in a room. Figure 14.7 presents the
refrigeration cycle using the refrigerant. In the outdoor part, a compressor generates
the flow of refrigerant and a condenser releases heat. The indoor part consists of a
capillary tube to control the flow of the refrigerant and an evaporator to absorb the
heat in the room. The refrigerant periodically circulates between the two sides so
that the indoor heat is absorbed and released outside. Currently, the design is carried
out based on the conventional process illustrated in Fig. 14.8. Since the machine is
not a new one, most of the design components are already known. Therefore, the
designer selects appropriate components and sets the values of the associated DPs.
Performance tests are then conducted to meet the requirements. The designer
iteratively redesigns the components until the performance requirements are satis-
fied. The associated workflow is illustrated in Fig. 14.9.
Each customer presents a new set of performance requirements. In such a case,
the product should be re-designed and re-manufactured each time to satisfy the new
conditions. The conventional design method relies on a trial-and-error approach to

Fig. 14.6 Indoor and outdoor machines. (Reproduced with permission from Lee et al. 2009)
380 G.-J. Park and A. M. Farid

Fig. 14.7 Composition of an Outdoor

air conditioning system.
(Reproduced with permission
from Lee et al. 2009) Condenser
Capillary Compressor

Evaporator
Indoor

Fig. 14.8 The conventional

design process for an overall Decision of positioning for installation
air conditioning system.
(Reproduced with permission
from Lee et al. 2009) Design of components for the system

No
Conϐirmation of the
performance tests

Yes

End the design process

Fig. 14.9 The conventional < Design of HVAC system >

design process for the design
of a part of an air conditioning
system. (Reproduced with Design of Evaporator/Condenser
permission from Lee et al.
2009) Design of air ϐlow

Design of drainage

Design of control system

Design of exterior/ assembly parts

meet the new set of requirements because it is not easy to identify the relationships
between the components and their characteristic DPs. Furthermore, the designer
may find it difficult to follow a rational design process that converges toward
14 Design of Large Engineering Systems 381

Table 14.1 Customer needs for a mount type HVAC control system
Customer needs from general users Customer needs from design engineers
People should feel cool Dust or bad smell should be eliminated
The room should be cooled as soon as The maintenance and management of the system
possible should be easy
The cool air must sufficiently The system should need a small space and have an
circulate aesthetic shape
Uniform temperature is required Regulations must be satisfied
Each vane should be controlled Production costs should be minimized
separately
Temperature control should be easy
Control from a distance is required
The machine should operate quietly
Energy consumption should be
minimized

satisfied FRs. Along the way, the designer may exercise engineering judgment that
relies on tacit knowledge. Such knowledge is neither written nor systematic and
consequently is difficult to transfer to junior designers. All of these challenges make
the conventional design process costly and slow.
In contrast, a new AD approach is demonstrated. Customer needs (CNs)—a
subset of the SRs described in Sect. 14.1—are gathered first based on customer
surveys and customer interviews with design engineers. The CNs are shown in
Table 14.1. FRs are then identified based on the CNs, and DPs are selected to
satisfy the independence of the FRs. The FR–DP relationship is made by the design
matrix and the dual hierarchy of the design process is established. Again, the
decomposition of the functional and physical hierarchies is made by the zigzagging
approach. CNs are defined from the customer survey and interviews with design
engineers.
Based on the CNs in Table 14.1, the FRs and Cs are defined at the top level.

FR1 = Minimize a possessed space of the mount type air conditioning system;
FR2 = Generate appropriate air current in the room;
FR3 = Make enough cold air;
FR4 = Minimize the vibration/noise of the mount type air conditioning system;
FR5 = Maintain purity of the air quality in the room;
FR6 = Control the temperature under user’s directions.

Note that the FRs are stated in a solution-neutral language. All of the nouns in
the FRs either (1) refer to the higher level DP of the “mount-type air conditioning
system” or (2) refer to nouns in the design context (e.g., air space, air current, cold
air, noise, air purity, temperature). Cs are defined as well. They provide bounds on
acceptable design solutions and differ from the FRs in that they do not have to be
independent.
382 G.-J. Park and A. M. Farid

C1 = Satisfy the first-grade for energy efficiency;

C2 = Satisfy related standards;
C3 = Satisfy the product size to sufficiently insert the air conditioning system
between the ceiling and ceiling board;
C4 = Minimize production cost;
C5 = Make maintenance and repair of the system easy.

To meet the FRs, an appropriate set of DPs are defined.

DP1 = Ceiling type structure;

DP2 = Air current formation system;
DP3 = Mutual assistance system;
DP4 = Vibration/noise reduction system;
DP5 = Air cleaner system;
DP6 = Temperature control system.

Then the allocation of function to form is captured in the design matrix:

DP1 DP2 DP3 DP4 DP5 DP6

FR1 X 0 0 0 0 0
FR2 X X 0 0 0 0
FR3 0 X X 0 0 0
FR4 0 X X X 0 0
FR5 0 0 0 0 X 0
FR6 0 X X 0 X X

where X represents where the FR is being fulfilled by one or more DPs and O
represents otherwise. At the top level of decomposition, the design is decoupled,
and the design matrix is lower triangular. Therefore, the independence of FRs (in
the Independence Axiom) is guaranteed if the DPs are determined in increasing
numerical order from FR1 to FR6.
The FRs and DPs at the top level are then iteratively decomposed using the
zigzagging process described in the previous section until the bottom or “atomic”
level components are reached. For the sake of brevity, a full explanation of the
system decomposition is omitted here. The complete design matrix, however, is
presented in Fig. 14.10. The left column of the table represents the FRs and the
upper row includes the corresponding DPs. The mount type air conditioning system
has a decoupled design at the top level as shown in Eq. (14.1). The entire design
matrix at its lowest level of decomposition is, however, non-square and the
resulting product design is classified as redundant and coupled. This situation is the
direct result of the sponsor’s request to keep the existing component-level parts.
14 Design of Large Engineering Systems 383

Fig. 14.10 Complete design matrix for mount type air conditioning system. (Reproduced with
permission from Lee et al. 2009)

Fig. 14.11 Design flow of a mount type air conditioning system. (Reproduced with permission
from Lee et al. 2009)

The coupled aspects are explained. The entire matrix in Fig. 14.11 is a
non-square and consequently, some FR–DP relationships are coupled at some
points of the hierarchy. For example, consider FR2.2.2.

FR2.2.2 = Generate the air flow in the perpendicular direction using a rotary
motion.
384 G.-J. Park and A. M. Farid

The existing system has five corresponding DPs that make up the components of
the turbofan.

DP2.2.6 = The number of blades;

DP2.2.7 = Ratio of the inside/external diameters;
DP2.2.8 = The angle among blades;
DP2.2.9 = Shape of the blade;
DP2.2.10 = Ratio of pitch/chord.

Consequently, the relationship between FR2.2.2 and its DPs is:

DP2.2.6 DP2.2.7 DP2.2.8 DP2.2.9 DP2.2.10

FR2.2.2 X X X X X

Obviously, it is a redundant design and a lot of feedback is required. The

redundancy is also found in other components such as the motor, the orifice, the fin
of the evaporator, and the condenser.
Several subsystems within this detailed design warrant further discussion. The
FRs and DPs for the compressor system are defined as:

FR3.1 = Generate the pressure to change the state of the refrigerant;

FR3.2 = Generate the air-flow for the ability of air conditioning to sufficiently
cool the room;
DP3.1 = Compressor system.

The associate portion of the design matrix is:

DP3.1
FR3.1 X
FR3.2 X

In this case, the compressor system has an insufficient number of DPs, and its
design matrix is full and therefore the compressor subsystem constitutes a coupled
design. The capillary tube subsystem suffers from the same problem.
The heat exchanger system has an equal number of FRs and DPs but they are
fully coupled.

FR3.3.2 = Change refrigerant from gas to liquid;

FR3.3.3 = Change refrigerant from liquid to gas;
DP3.2.2 = Evaporator system;
DP3.2.3 = Condenser system.
14 Design of Large Engineering Systems 385

DP3.2.2 DP3.2.3
FR3.3.2 X X
FR3.3.3 X X

Because the refrigerator cycle consists of the compressor system, the capillary
tube, and the heat exchanger system, the refrigerator cycle as a whole also con-
stitutes a coupled design. Note that only one coupled subsystem is required for the
system as a whole to be classified as coupled. In an ideal situation, the DPs of the
coupled design should be modified or replaced with a set that yields either an
uncoupled or decoupled design. In this case, the firm decided to flag the coupled
characteristics to alert designers to the potential for iterative feedback loops in the
design process.
Returning back up to the top level of design decomposition, the associated
design flow diagram is illustrated in Fig. 14.11. Because it is a decoupled design at
the top level of hierarchy, the design flow links the modules sequentially. That said,
if the design workflow were detailed further, feedback loops would appear within
each of the system’s modules.
In summary, this example has served to apply AD for large fixed systems to a
mount type air conditioning system. AD demonstrates a rational and effective
design process that decomposes the FRs and DPs at each level of hierarchy and then
allocates function to form using the design matrix. These two systems thinking
techniques allow the design team to manage the complexity of the design. Finally,
the design flow diagram organizes the design’s project management and is estab-
lished from the design matrix.
The AD approach highlighted some differences between the existing conven-
tional approach. In the conventional approach, each module was designed in par-
allel and then later integrated into a larger system; resolving inconsistencies in the
design all at once. In contrast, AD highlighted the need for a sequential approach to
the design of the high-level modules; thus avoiding the need to resolve downstream
design inconsistencies. The improved design process is more effective for designers
to not only understand the overall design process but also to meet the diversity of
customer demands. Finally, the AD approach highlights the presence of coupled
design feedback in lower levels of the dual hierarchy. These are fairly inefficient
and time-consuming and have the potential to derail the design of the system as a
whole. These coupled aspects are identified and flagged for management attention
without eliminating them from the design as a whole.

14.3.2 Automobile Cooling System

We consider the design of a cooling system for a hybrid vehicle as illustrated in

Fig. 14.12 (Park 2018). Coolant material is circulated in a cooling system to expel
the generated heat from the engine. The circulation is illustrated in Fig. 14.13.
Some of the heat is reused for heating the inside of the car.
386 G.-J. Park and A. M. Farid

Fig. 14.12 An automobile cooling system

Fig. 14.13 Circuit of coolant and air in the cooling system

The CNs of the system are defined as:

CN1: The inside temperature should be kept within a certain range;

CN2: Environmental pollution is undesired;
CN3: The vehicle fuel economy should be high.

The top-level FR is defined based on the CNs: Increase the energy efficiency of a
hybrid car while maintaining the inside temperature.
14 Design of Large Engineering Systems 387

From the top-level FR, the next level FRs, DPs, and design matrix are

FR1 = Maintain the temperature of the air inside the vehicle;

FR2 = Reduce the emission of hazardous substances during operation;
FR3 = Control fuel consumption within acceptable bounds;
DP1 = Air conditioning system;
DP2 = Hazardous material reduction system;
DP3 = Energy management system.

DP1 DP2 DP3

FR1 X x x
FR2 0 X 0
FR3 x 0 X

Note that FR1 and FR3 have a coupled design with respect to DP1 and DP3.
This coupling is discussed in greater detail at lower levels of decomposition. DP1
and FR1 are decomposed to:

FR1.1 = Decrease room temperature;

FR1.2 = Increase room temperature;
DP1.1 = Air cooling system;
DP1.2 = Air heating system.

DP1.1 DP1.2
FR1.1 X 0
FR1.2 0 X

The air cooling system is not considered further in this example. FR1.2 and
DP1.2 are then decomposed to:

FR1.2.1 = Generate a heat source;

FR1.2.2 = Generate additional heat;
FR1.2.3 = Absorb the generated heat;
FR1.2.4 = Deliver the heat;
FR1.2.5 = Heat the air;
FR1.2.6 = Move the heated air;
FR1.2.7 = Adjust the air direction;
DP1.2.1 = Engine operation request logic;
DP1.2.2 = Positive temperature coefficient (PTC) system;
DP1.2.3 = Coolant;
DP1.2.4 = Coolant circuit;
DP1.2.5 = Heat core;
DP1.2.6 = Blower;
DP1.2.7 = Flap.
388 G.-J. Park and A. M. Farid

The associated design matrix is:

DP1.2.1 DP1.2.2 DP1.2.3 DP1.2.4 DP1.2.5 DP1.2.6 DP1.2.7

FR1.2.1 X 0 0 0 0 0 0
FR1.2.2 0 X 0 0 0 0 0
FR1.2.3 0 0 X 0 0 0 0
FR1.2.4 0 0 0 X 0 0 0
FR1.2.5 0 0 x 0 X 0 0
FR1.2.6 0 0 0 0 x X x
FR1.2.7 0 0 0 0 0 x X

FR2 and DP2 can be decomposed to:

FR21 = Reduce the hazardous material before the combustion;

FR22 = Reduce the hazardous material after the combustion;
FR23 = Reduce the hazardous material during the combustion;
DP21 = Evaporative gas reduction system;
DP22 = Engine emission reduction system;
DP23 = Catalyst system.

The associated design matrix is:

DP21 DP22 DP23

FR21 X 0 0
FR22 0 X 0
FR23 0 0 X

FR3 and DP3 are decomposed as well. Figure 14.14 presents the entire design
matrix after several more decomposition and zigzagging steps.
At first glance, the design of the automobile cooling system appears to be highly
coupled. There are filled X elements in both the upper and lower triangles of the
design matrix. Recall that FR1 and FR3 have a coupled design by virtue of DP1 and
DP3 at the highest level of decomposition. These coupled elements manifest
themselves in lower levels of decomposition as well. The situation, however, is not
as bleak as one might think. The high-level coupling between FR1, FR3, DP1, and
DP3 does not mean that all of the low-level FRs and DPs are coupled. Quite a bit of
sparsity is introduced into the design matrix with each subsequent decomposition.
After careful inspection, the designer finds that many of the filled elements in the
upper triangle of the design matrix do not have their associated filled elements in
the lower triangle. Indeed, a minimum condition of a truly coupled design is that
there exist two DPs that are coupled to two FRs. Or mathematically, there exists at
14 Design of Large Engineering Systems 389

Fig. 14.14 Complete design matrix of an automobile cooling system prior to resorting FRs and
DPs

least one pair of elements in the design matrix DM such that DM(i, j) = DM(j, i),
where i 6¼ j. Such a condition occurs in exactly one place above: FR1.2.6 and
FR1.2.7 are coupled with DP1.2.6 and DP1.2.7. All of the other filled elements in
the design matrix are asymmetric and indicate a decoupled design for the remainder
of the automobile cooling system. To emphasize this point, the rows and the
columns of the design matrix can be sorted in such a way as to bring all of the
off-diagonal terms to either the matrix’s upper or lower triangle. Figure 14.15
shows the complete design matrix of the same automobile cooling system after the
FRs and DPs have been sorted. It reveals clearly the original conclusion that only
FR1.2.6 and FR1.2.7 are coupled. This example illustrates that a high-level coupled
390 G.-J. Park and A. M. Farid

1.1 1.2.2 1.2.4 1.2.7 1.2.6 1.2.5 1.2.3 2.1.1 2.1.2 2.2.1 2.2.2.1 2.2.2.2 2.3.1 2.3.2 2.3.3 3.1.1 3.1.2 3.2.1.1 3.2.1.2 1.2.1 3.2.2.1 3.2.2.2
1.1 X
1.2.2 X X
1.2.4 X X X
1.2.7 X X
1.2.6 X X X
1.2.5 X X
1.2.3 X
2.1.1 X
2.1.2 X
2.2.1 X
2.2.2.1 X X
2.2.2.2 X
2.3.1 X
2.3.2 X
2.3.3 X
3.1.1 X X
3.1.2 X X
3.2.1.1 X X
3.2.1.2 X X
1.2.1 X
3.2.2.1 X
3.2.2.2 X

Fig. 14.15 Complete design matrix of automobile cooling system after resorting FRs and DPs

design can be “rescued” with respect to the Independence Axiom to become a

decoupled design in lower levels of decomposition if the nature of the coupling is
reflected in an asymmetric design matrix! Fortunately, a creative designer can often
find clever ways to introduce such asymmetry into the design.

14.3.3 Automobile Suspension System

The Independence Axiom is used to design the automobile suspension system

shown in Fig. 14.16 (Park 2018). It connects the wheels and the body to make the
automobile move. There are two suspension subsystems; one for the front wheels
and the other for the rear wheels.
The CNs are:

CN1 = The vehicle must have a driving capability;

CN2 = Safety must be provided to the vehicle when driving.

From the CNs, the top-level FR is:

FR = Suspend the vehicle with stability while driving (CN1 + CN2).

which is in turn decomposed into six FRs:

FR1 = Generate forward thrust from transmission system torque;

FR2 = Permit wheel rotation;
FR3 = Secure the car body;
FR4 = Suspend the vehicle with stability while driving on rugged roads;
14 Design of Large Engineering Systems 391

Fig. 14.16 An automobile suspension system. (Reproduced with permission from Nosorog 2019)

FR5 = Suspend the vehicle with stability while turning;

FR6 = Provide turning ability.

To meet FRs, an appropriate set of DPs are defined as follows:

DP1 = Wheel system;

DP2 = Knuckle and bearing;
DP3 = Mounting system;
DP4 = Independent link system;
DP5 = Stabilizer bar and shock absorber;
DP6 = Tie rod and ball joint.

The associated design matrix shows a decoupled design at the highest level of
decomposition.

DP1 DP2 DP3 DP4 DP5 DP6

FR1 X 0 0 0 0 0
FR2 x X 0 0 0 0
FR3 0 0 X X 0 0
FR4 0 x x X x 0
FR5 x 0 x x X 0
FR6 0 x 0 0 x X
392 G.-J. Park and A. M. Farid

FR1 and DP1 are then decomposed.

FR11 = Generate friction force from transmission system torque;

FR12 = Deliver forward thrust from friction force;
DP11 = Tire;
DP12 = Wheel.

The associated design matrix shows a decoupled subsystem.

DP11 DP12
FR11 X 0
FR12 x X

FR2 and DP2 are then decomposed and the corresponding design matrix is
defined.

FR21 = Impede the motion of the wheel system through five coordinates (3
translation and 2 rotation);
FR22 = Permit the rotation of the wheel system about its axis;
DP21 = Knuckle;
DP22 = Bearing.

DP21 DP22
FR21 X x
FR22 x X

FR3 and DP3 are then decomposed and the corresponding design matrix is
defined.

FR31 = Absorb vibration in mounting;

FR32 = Connect the car body;
DP31 = Mounting bush;
DP32 = Mounting bolt.

DP31 DP32
FR31 X 0
FR32 x X

FR4 and DP4 are then decomposed and the corresponding design matrix is
defined.

FR41 = Absorb vibration and noise at the junctions;

FR42 = Connect the mounting and wheel system;
14 Design of Large Engineering Systems 393

FR43 = Allow the wheel system to move independently;

DP41 = A,G bush;
DP42 = Sub-frame;
DP43 = Lower arm.

DP41 DP42 DP43

FR41 X 0 0
FR42 x X x
FR43 x x X

FR5 and DP5 are then decomposed and the corresponding design matrix is
defined.

FR51 = Provide rigidity for roll mode;

FR52 = Provide rigidity for the ride mode;
FR53 = Decrease in vertical energy;
FR54 = Limit the vertical distance of motion;
DP51 = Stabilizer bar;
DP52 = Spring;
DP53 = Damper;
DP54 = Bump stopper.

DP51 DP52 DP53 DP54

FR51 X x 0 0
FR52 x X 0 0
FR53 0 x X 0
FR54 0 x x X

FR6 and DP6 are then decomposed and the corresponding design matrix is
defined.

FR61 = Restrain the wheel system to turn;

FR62 = Provide a turning input;
DP61 = Ball joint;
DP62 = Tie rod.

DP61 DP62
FR61 X 0
FR62 x X
394 G.-J. Park and A. M. Farid

1.1 1.2 2.1 2.2 3.1 3.2 4.1 4.2 4.3 5.1 5.2 5.3 5.4 6.1 6.2
1.1 X
1.2 X X
2.1 X X X
2.2 X X
3.1 X
3.2 X X X X
4.1 X X X
4.2 X X X
4.3 X X X X
5.1 X X X
5.2 X X X X X
5.3 X X
5.4 X X X
6.1 X X X
6.2 X X

Fig. 14.17 Complete design matrix of an automobile suspension system

The entire design matrix is illustrated in Fig. 14.17. Although the suspension
system has received many design iterations over a long period of time, there are
many and more importantly non-negligible points of coupling between the FRs and
the DPs. In the early stage of product development, the suspension system had
relatively few and simple FRs. Over time, more FRs were added, and the design
became quite complex with many non-zero off-diagonal terms in the design matrix.
Despite this large number, only three pairs of DPs (shown in red above) have
coupled designs. They are DP2.1 and DP2.2, DP4.2 and 4.3, and DP5.1 and 5.2.
Although these pairs of DPs formally violate the Independence Axiom, the situation
is quite manageable. These three pairs of DPs can be treated as three DPs, with each
pair being treated as a single entity. Furthermore, the DPs appear at the bottom of
the AD dual hierarchy, and so their coupled nature does not “ripple” across the
design of the rest of the suspension system. Finally, product design companies can
find it beneficial to have highly coupled components at the bottom of the dual
hierarchy because they effectively embody proprietary and potentially secret design
know-how that is not easily reproduced. That said, this special condition should not
be interpreted as a license to ignore the Independence Axiom. The product’s overall
dual-hierarchy should remain uncoupled or decoupled (as above) and only if nec-
essary introduce coupled DPs at the very bottom of the hierarchy.

14.3.4 Mobile Harbor

As the worldwide volume container shipments increases and very large container
ships emerge as a dominant player in the maritime cargo transport market, the
14 Design of Large Engineering Systems 395

functional capabilities of container ports need to be greatly enhanced. For example,

large container ships with streamline hulls lack maneuverability in tight and highly
suggested maritime ports and often require tug boats which may in and of them-
selves be otherwise occupied. To address this problem, the Korea Advanced
Institute of Science and Technology (KAIST) is undertaking a project to design a
novel container transport system called Mobile Harbor (MH). A conceptual illus-
tration of the MH is presented in Figs. 14.18 and 14.19 (Lee and Park 2010). MH
refers to a system that can go out to a large container ship, anchor in the open sea,
load and unload containers between itself and the container ship, and transport them
to their destination. It has a flat bottom design so that it is maneuverable in narrow
shipping lanes and stable enough to handle a wide variety of ocean-going cargo.
Like many other large-scale engineering projects, the design of MH presents a
number of challenges at the beginning stages of the project. In the conceptual

Fig. 14.18 Illustration of a revised mobile harbor concept (approved for funding by a national
R&D program)

Fig. 14.19 The mobile harbor as conceptualized by Axiomatic Design (republished with
permission of SAE international, from managing system design processusing Axiomatic Design: a
case on KAIST mobile harbor Project, Lee and Park, 3, 1, 2010; permission conveyed through
Copyright Clearance Center, Inc.)
396 G.-J. Park and A. M. Farid

design phase of the project, the design team must properly define and disseminate
FRs, clarify interface requirements between its subsystems, identify and reconcile
potential design conflicts like functional coupling.
To begin the AD, the top-level FRs, DPs, and Cs are defined.

FR = Transfer containers from ships anchoring in the sea to a harbor;

FR1 = Load/unload the containers;
FR2 = Keep the containers in MH (Mobile Harbor);
FR3 = Let MH float on the sea;
FR4 = Let MH navigate on the sea;
FR5 = Dock MH against the container ship in the open sea.
DP = Mobile harbor (MH);
DP1 = Crane system;
DP2 = Deck system for loading the containers;
DP3 = Floating body structure;
DP4 = Driving system <propulsion + navigation>;
DP5 = Mooring system + Docking system.

Constraints

C1 = The capacity of the power generator should be large enough to cover the
needed energy (Generate a certain power for operating.);
C2 = The total weight should be less than a specified value;
C3 = The MH carries the containers up to 250 TEU;
C4 = Safety, reliability, efficiency, cost-effective, environmentally acceptable,
automation, the satisfaction of standard and shipping registration or other
rules.

According to the relationship between FRs and DPs, the design matrix is defined
as follows:

DP1 DP2 DP3 DP4 DP5

FR1 X x 0 0 0
FR2 x X X 0 0
FR3 x x X 0 0
FR4 x 0 x X 0
FR5 x 0 x x X

For illustrative purposes, the AD of DP1 is presented in detail and the completed
design matrix of the MH is shown in Fig. 14.19.
Decomposition of FR1 (DP1: Crane system)
FR11 = Translocate the containers between MH and a container ship in the open
sea or a harbor;
14 Design of Large Engineering Systems 397

FR12 = Load and unload the containers, while MH is on the sea;

DP11 = Container crane;
DP12 = Dynamic control system of MH movement.
DP11 DP12
FR11 X 0
FR12 X X

Decomposition of FR11 (DP11: Container crane)

FR111 = Let the container crane of MH approach to the containers in a container
ship or open sea;
FR112 = Hold the container;
FR113 = Move the container;
DP111 = The size and shape of the container crane and moving system of the
container;
DP112 = An appropriate configuration of the crane structure to have sufficient
strength;
DP113 = The hoist device including a trolley.
Decomposition of FR12 (DP12: Dynamic control system of MH movement)
FR121 = Minimize the movement of the container due to the vibration of the
boom while working on the sea;
FR122 = Minimize the movement of the container due to the vibration of the
hoist while working on the sea;
DP121 = Zero moment point (ZMP) system for stability control;
DP122 = Dynamic control system of the trolley;
Cs = The condition of sea state 3 (a wave level) should be satisfied.
DP121 DP122
FR121 X 0
FR122 x X

The MH constitutes an entirely novel large complex product. It is able to go out

to a ship anchored in the open sea to load and unload cargo from it without
occupying a pier of a land-based harbor. Such a functionality relieves the cargo ship
from having to come into a stationary harbor to unload and load its cargo. Fur-
thermore, the increasing size of container ships means that fewer and fewer harbors
have a sufficiently large pier and are sufficiently deep. The MH is also capable of
loading and unloading cargo to and from a large ship to the stationary land-based
harbor without a container quay by pre-loading the cargo in freight cars placed on
the MH which themselves can be moved onto and removed from the MH. The MH
was designed so that it can approach both sides of a cargo ship and attach itself
securely; further reducing the time to load and unload cargo. In addition to these us
398 G.-J. Park and A. M. Farid

cases, the MH can be used inland within a canal to allow the crossing of transport
containers and goods. One integral part of the design is positioning. Not only must
the MH control horizontal translation on the sea and vertical height alignment but it
also uses a large gyroscopic apparatus to stabilize rotation even in the presence of
large waves.
The final design matrix is illustrated in Fig. 14.20. It is almost entirely decou-
pled. FR2 is weakly coupled with FR1 and FR3. Keeping the containers in the MH
has the potential to impact their loading/unloading as well as the overall floating
functionality of the MH. Over most conditions, these coupling are quite weak. They
only represent a few DPs and can be given sufficient design attention upfront so as
to allow the remainder of the design to proceed. Nevertheless, one must recognize
that as the designers pursue ever-more “aggressive” MH designs with an ever-larger
number of increasingly heavy containers, the coupling will strengthen and even-
tually lead to design infeasibility. Indeed, the presence of such coupling is entirely
inherent to the mobile application and is predicted by AD theory. The “Design
Range and Coupling” theorem in AD theory states:
Theorem 20. Design Range and Coupling: If the design ranges of uncoupled or decoupled
designs are tightened, they may become coupled designs. Conversely, if the design ranges
of some coupled designs are relaxed, the designs may become either uncoupled or
decoupled.

Many mobile applications ultimately suffer from such coupling because the
mobility of the application effectively places hard Cs on the design range for space,
energy, and functionality. The energy and power capabilities of many robots and
drones are limited by their size and weight. Sea-faring vessels must also ensure
flotation, and aerospace applications must ensure flight. Finally, the most recent
smart city research demonstrates that as an urban population expands, and the
demands for water, energy, mobility, and other infrastructure services grow, the
design and planning of the city’s infrastructure systems become increasingly
interdependent within the city’s confined geography. Whereas such couplings can
provide the integrated delivery of infrastructure services, they are also susceptible to
failures propagating from one subsystem to another. Alternatively, the city’s con-
fined geography can be loosened to avoid such a situation; at which point urban
sprawl becomes the concern.

14.3.5 The Online Electrical Vehicle

The On-Line Electrical Vehicle (OLEV) is a type of electric vehicle that was
developed at KAIST to overcome many of the problems posed by conventional
electric vehicles: limited driving range on a single charge, the battery’s heavy
weight, and its high cost. The conceptual configuration is illustrated in Fig. 14.21
(Hong and Park 2010). The OLEV design team is composed of many designers
from many engineering disciplines including automobile engineering and
electromagnetics.
14 Design of Large Engineering Systems 399

Fig. 14.20 Complete design matrix of the mobile harbor

The integration of these multiple fields presents a significant design challenge.

Consequently, the design was conducted by a “design team” and an “integration
team.” Their roles and workflows are illustrated in Fig. 14.22. The integration team
defines the FRs, at which point, the design team defines the DPs, at which point the
integration team defines the design matrix and defines the new set of decomposed
FRs. This alternation between the two teams realizes the zigzagging process on the
fly and prevents the inadvertent introduction of design coupling.
The FRs and DPs at the top level are

FR1 = Import electric power;

FR2 = Transmit electric power to the vehicle;
400 G.-J. Park and A. M. Farid

Fig. 14.21 Conceptual organization of the OLEV (approved for funding by a national R&D
program)

Fig. 14.22 Design process of the OLEV. (Republished with permission of SAE International,
from Design Information Management of an On-Line Electric Vehicle Using Axiomatic Design,
Hong and Park, 3, 1, 2010; permission conveyed through Copyright Clearance Center, Inc.)
14 Design of Large Engineering Systems 401

FR3 = Operate the vehicle using electric power;

FR4 = Protect systems from external loads;
DP1 = Three phase AC 370-440 V;
DP2 = Induction Coupling System;
DP3 = Electric Vehicle;
DP4 = System protection devices.

From the FR-DP relationship, the design matrix is defined.

DP1 DP2 DP3 DP4

FR1 X 0 0 0
FR2 X X 0 X
FR3 0 X X 0
FR4 0 X 0 X

Throughout the design process, much effort was exerted to maintain a decoupled
design. Nevertheless, the design matrix above shows a high-level coupling by
virtue of the FR2–DP4 coupling. Investigating further, this coupling arises because
the system protection devices are coupled to the transmission of electric power to
the vehicle. This coupling is ultimately inevitable because the safety considerations
of transferring many kilowatts of electricity inductively through an air gap require a
highly integrated design between the induction coupling and the protection devices.
From this high-level design matrix, the zigzagging process is carried out until the
bottom level is reached. The final design matrix is presented in Fig. 14.23. The
FR2–DP4 coupling is highlighted to alert the design and integration teams to the
potential for design iterations.
This design example, much like the MH example before it, serve to demonstrate
the utility of AD in large systems of unprecedented function. The engineers on the
project cannot rely on other products or large systems for inspiration. Rather, they
must deeply reflect on what the system must achieve and consequently synthesize
the associated FRs and DPs. From there, the two critical systems thinking skills
detailed at the beginning of the chapter become the guiding principles of
forward-design: decomposition of the functional and physical hierarchy and the
allocation of function to form through the zigzagging process. When used in
concert, they serve as the primary means by which a designer or design team can
tackle the design of large fixed engineering systems.
402 G.-J. Park and A. M. Farid

Fig. 14.23 Complete design matrix of the OLEV (Republished with permission of SAE
International, from design information management of an On-Line Electric Vehicle using Axiomatic
Design, Hong and Park, 3, 1, 2010; permission conveyed through Copyright Clearance Center, Inc.)

14.4 Axiomatic Design of Large Flexible Engineering

Systems

Thus far, the chapter has motivated the role of design science as a means by which
engineers can tackle the 14 grand challenges identified by the National Academy of
Engineers. It recognized that design science provides engineers with a
meta-problem-solving skill set composed of several systems-thinking design skills.
The chapter focused on two of these: decomposition of the system function and form
as a means to “divide-and-conquer” and the allocation of function to form in a manner
consistent with the Independence Axiom. These two design skills were demonstrated
on a half-dozen design examples and in so doing demonstrated their salience to a wide
variety of large fixed engineering systems in many application domains.
A curious engineering design student may naturally ask: “What if we wished to
design a large flexible engineering system?” (see Definition 14.8 at the beginning of
the chapter.). Indeed, many of the NAE grand challenges identified at the beginning
of the chapter will require enduring engineering solutions that respond to the
changing needs of society. Furthermore, they are likely to require such large
investments that it would be entirely cost prohibitive to design systems from
scratch. These design solutions will need to build upon existing legacy systems and
will need to continually evolve their functionality over and over again. Unfortu-
nately, the answer to the curious student’s question cannot be answered
14 Design of Large Engineering Systems 403

comprehensively in a single book chapter, or perhaps even book, because in many

ways it presents a relatively open frontier for the development of design science as
an engineering discipline. The literature on the AD of large flexible engineering
systems has expanded over the years and has developed to now be called
hetero-functional graph theory; given its heavy reliance on graph theoretic con-
cepts. Nevertheless, this chapter can serve to open the door to this fascinating realm
of design science and systems engineering.
The AD of large flexible engineering systems was first mentioned by Suh in his
2001 text. He presented the following abstract example of a large flexible engi-
neering system knowledge base:

FR1 $ ðDPa1 ; DPb1;...; DPr1 Þ

FR2 $ ðDPa2 ; DPb2;...; DPq2 Þ
FR3 $ ðDPa3 ; DPb3;...; DPw
3Þ
...


FRm $ DPam ; DPbm;...; DPsm

where the first line means that FR1 (as indicated by the $) can be satisfied by DPa1
or DPb1 or DPr1, etc. These few lines are remarkably profound. First, note that the
knowledge base does not say that DPa1 and DPb1 and … and DPr1 satisfy FR1. If it
did, it would become the design matrix used throughout the earlier parts of this
chapter, and DPa1 and DPb1 and … and DPr1 would form either a decoupled or
coupled engineering system. Instead, the focus is on the word “or.” In other words,
at a given moment in time, DPa1 alone satisfies FR1. Such a statement is simply
another way of stating adherence to the Independence Axiom. Consequently, the
knowledge base above (as presented in the Suh 2001 text) did not allow for
decoupled and uncoupled designs. In so doing, it implicitly reached a profound
conclusion: large flexible engineering systems must adhere to the Independence
Axiom if they are to maintain their ability to evolve their FRs and DPs over time.
We offer a logical proof by contradiction. In the case of an ideal uncoupled
design (with an identity design matrix), the engineer would remove a DP from a
functioning system, and an entire FR would be removed at the same time. The
system would then continue to function with reduced functionality. In contrast, if a
given FR demonstrated a decoupled design and was coupled to multiple DPs, or if
the FR demonstrated a couple design and was coupled to other FRs, then when one
of the associated DPs were removed, the system would demonstrate broken func-
tionality! Furthermore, removing additional DPs to remove an entire module and its
associated FR would lead to downstream effects on other FRs. By Definition 14.8,
such a condition contradicts a functioning large flexible engineering system.
Indeed, large flexible engineering systems demonstrate a “plug-and-play” func-
tionality similar to that found in modern computers where functional and physical
elements can be added or removed at will.
404 G.-J. Park and A. M. Farid

The knowledge base above also serves the design of large flexible engineering
systems for two other reasons. First, by Definition 14.8, large flexible engineering
systems have FRs that can be fulfilled by potentially many DPs. An identity design
matrix does not show this. Therefore, in order to reveal this functional redundancy,1
the set of FRs instances FR must be distinguished from the set of functional
requirement classes FR.2 Second, the knowledge base allows a single DP to fulfill
more than functional requirement class FR. The importance of doing so is dis-
cussed later in the context of hierarchy in large flexible engineering systems. In
brief, the knowledge base becomes the single most important concept in the design
of large flexible engineering systems, in much the same way that the design matrix
is important to the design of large fixed engineering systems.
As the AD of large flexible engineering systems developed into
hetero-functional graph theory, the knowledge base was given an explicitly quan-
titative definition.
Definition 14.9 System Knowledge Base : A binary matrix J of size r(FR)x
r(DP) whose element J(w,v) 2 {0,1} is equal to one when an action ewv (in the
SysML sense) exists as a FR class FRw being executed by a design parameter DPv.
These actions represent the “capabilities” in the engineering system.

Consequently, the design equation of the large flexible engineering system can
be written in terms of the system knowledge base.

FR ¼ J  DP

where  represents matrix Boolean multiplication.

Definition 14.10 Matrix Boolean Multiplication ⊙. Given sets or Boolean
matrices B and C and Boolean matrix A, C = A⊙B is equivalent to
_
C ði; kÞ ¼ Aði; jÞ ^ Bðj; kÞ
j

where ^ refers to the scalar “AND” operation and _ is an “OR” operation over
j
j elements much like the well-known sigma sum R.

1
Note that functional redundancy refers to having the same functional requirement repeated. For
example, FR1 = generate electric power and FR2 = generate (backup) electric power. Functional
redundancy should not be confused with “redundant designs” in the AD of large fixed systems
were the number of DPs exceeds the number of FRs.
2
The terms class and instance are drawn from the fields of software/systems engineering. Note that
many works on AD do not make this distinction between functional requirement instances and
functional requirement classes because it is rarely needed within a single design work. Here, the
distinction is made in order to maintain the conceptual link between large fixed and large flexible
engineering systems and the universality of the Independence Axiom in both cases.
14 Design of Large Engineering Systems 405

The design equation of the large flexible engineering system stated above in no
way replaces the design matrix B. In fact, the design equation stated at the
beginning of the chapter can be written in a set-theoretic expression as well.

FR = B ⊛ DP

where the aggregation operation ⊛ is defined as:

Definition 14.11 Aggregation Operator ⊛ : Given Boolean matrix A and sets
B and C, C = A ⊛ B is equivalent to
[
C ðiÞ ¼ aði; jÞ ^ bð jÞ
j
S
where the operation is a union operation of j elements much like the well-known
j
sigma sum R.
The engineering design student must recognize that these two statements of the
design equation are equivalent in meaning, but must absolutely be distinguished
from each other. They simply express the allocation of function to form in terms of
the system knowledge base J or the design matrix B. The fine mathematical dis-
tinctions between the aggregation operator ⊛ and matrix Boolean multiplication ⊙
and the set of functional requirement instances FR and the set of functional
requirement classes FR facilitates the analysis of large flexible engineering systems
in complementary ways.
Example 14.1
To solidify these large flexible engineering systems, consider the example of the simple
manufacturing system depicted in Fig. 14.24. It consists of a drill press and milling
machine. The former is able to drill a hole and the latter is able to do the same and mill
surfaces. Each contains its respective fixture. The manufacturing system also has two
one-way conveyors between them.

A quick analysis of the system yields:

FR = drill hole, drill hole, mill surface, store the part at point A, transport part
from point A to point B, transport part from point B to point A, store the
part at B.
DP = {drill press, milling machine drill, milling machine end mill, drill press
fixture, conveyor 1, conveyor 2, milling machine fixture}.

Consequently, the design matrix B = I7x7. As expected, the system is uncoupled,

and the Independence Axiom is satisfied. To continue the analysis, the FR classes
are viewed instead of their instances.
FR = drill hole, mill surface, store the part at A, transport part from point A to
point B, transport part from point B to point A, store the part at B}. Rather than
viewing the DPs at the very lowest level of aggregation, it is often useful to
406 G.-J. Park and A. M. Farid

Conveyor 1

Drill Press Milling Machine

Conveyor 2

Fig. 14.24 A simple manufacturing system that consists of one drill press, one milling machine
and two conveyors

aggregate the DPs to a higher level of abstraction. In this case, a logical aggregation
yields:
DP ={drill press, milling machine, conveyor system}
At this level of physical aggregation, the system knowledge base J is defined.
2 3
1 1 0
60 1 07
6 7
61 0 07
J¼6
60
7
6 0 177
40 0 15
0 1 0

Note that the first column of the knowledge denotes that the drill press is cable of
both drilling holes as well as storing the part. The second column shows that the
milling machine is cable of doing both of those functions as well as also milling
surfaces. The third column shows that the conveyor system is capable of trans-
porting parts back and forth between the two machines. In all three cases, the
columns assigned to each of these aggregated DPs, (often called resources in
hetero-functional graph theory), was greater than one because each was capable of
more than one function. Conversely, the first row has a sum greater than one to
reflect that the function “drill hole” has two instances in the system. In short, the
design matrix and the system knowledge base give complementary insights into
how function is allocated to form in large flexible engineering systems.
Returning to the original exposition of large flexible engineering systems in the
2001 Suh text, the flexible nature was emphasized by a set of FRs that changed in
time.

@t ¼ 0 fFRg0 ¼ fFR1 ; FR5 ; FR7 ; FRm g

@t ¼ T1 fFRg1 ¼ fFR3 ; FR5 ; FR8 ; FRz g
@t ¼ T2 fFRg2 ¼ fFR3 ; FR9 ; FR10 ; FRm g

As the AD of large flexible engineering systems developed into

hetero-functional graph theory, this time-dependent system functionality was
quantified using a system Cs matrix.
14 Design of Large Engineering Systems 407

Definition 14.12 System Cs Matrix: A binary matrix K of size FR  DP whose

element B(w,v) 2 {0,1} is equal to one when a constraint eliminates action ewv
from the action set.

A reconfiguration process is said to change the value of the system Cs matrix.

Therefore, the system knowledge base contains information on the existence of
capabilities in the engineering system. Meanwhile, the Cs matrix contains infor-
mation on their availability. Quantitatively keeping track of these capabilities is
done via the system's structural degrees of freedom as a quantitative measure.
Definition 14.13 Structural Degrees of Freedom3: The set of independent actions
ES that completely defines the available capabilities in a large flexible engineering
system. Their number is given by

ðFRÞ rX
rX ðDPÞ
DOF ¼ rðeÞ ¼ ½J  Kðw; vÞ
w v

Perhaps the last necessary topic in this introductory discussion of large flexible
engineering systems is that of the AD dual hierarchy. When a functional or physical
element is added or removed, it has the potential to disrupt their respective hier-
archies as well. Simply speaking, a plane ceases to be one if it were to lose a wing.
And its high-level function of flight would be impossible if it loses propulsion.
Nevertheless, the designer can proceed cautiously.
Developing the AD dual hierarchy for large flexible engineering systems ,
downward in the direction of design synthesis, proceeds in the same way as for
large fixed engineering systems. The system is viewed in terms of FR instances
rather than classes. Because the Independence Axiom has been strictly maintained,
each structural degree of freedom can be designed as previously described as if it
were its own system. The engineering design problem is separable. Therefore, the
addition or removal of a structural degree of freedom adds or removes all of the
associated lower branches in the dual hierarchy .
It is also useful to consider the dual hierarchy of a large flexible engineering
system upward in the direction of design analysis. Here, it is no longer required to
aggregate the physical and functional hierarchies simultaneously. It is particularly
common in bottom-up design to aggregate only the physical hierarchy into higher
level DPs (or resources). A corresponding functional aggregation does not need to
occur. Such was the case of the manufacturing system example above. This is
because, in bottom-up design synthesis and analysis, physical aggregation and
functional aggregation do not have the same meaning and do not necessarily imply

3
The term structural degrees of freedom is appropriately named. Previous works on
hetero-functional graph theory have shown that it is a generalization of the well-known concept
of degrees of freedom in mechanical systems.
408 G.-J. Park and A. M. Farid

each other. Consider, for example, five tasks as FRs and five individuals as DPs;
each of whom completes one task. This a large flexible engineering system that
fulfills the Independence Axiom. The five individuals may be aggregated into a
resource called a team without making any statement about the five tasks. They may
not be related in any way (i.e., share any functional interaction). Similarly, the five
tasks may be aggregated into a project without making any statement about the five
individuals who complete them. They may have never met (i.e., share any physical
interface). Physical aggregation is particularly interesting because it yields
resources with many capabilities. An addition or removal of a DP yields the cor-
responding change in a resource's capabilities. In contrast, the functional aggre-
gation of a large flexible engineering system may result in a rigid top-down
structure. Any time the set of FRs changes, the functional hierarchy would need to
change as well. In a project, the elimination of a single task causes the elimination
of the project as a whole.

14.5 Conclusion

This chapter has introduced the design engineer to the world of large complex
systems from an AD perspective and motivated the pressing need for such design
skills in terms of the 14 grand challenges of the NAEs. The chapter focused on two
critical systems-thinking design skills that help the designer manage the inherent
and abstract complexity of large systems. They are (1) system decomposition as a
means to “divide-and-conquer,” and the allocation of function to form in what is
often called the “zigzagging” process. These two skills were demonstrated in
several tractable design case studies of large fixed engineering systems. The chapter
also discusses why and how these design skills are used differently when the system
has a fixed versus a flexible structure.
In distinguishing between large fixed and flexible engineering systems, this
chapter has also opened the door to a fascinating realm of design science and
systems engineering; one that remains a very open frontier for methodological
development. Three broad directions of enquiry are worthy of note here:

• a quantitative understanding of life cycle properties;

• the treatment of cyber-physical systems;
• hetero-functional networks in large flexible engineering systems.

First, the field of systems engineering is increasingly concerned with developing

a quantitative understanding of life cycle properties which can be deployed within
engineering design. Many life cycle properties are called “ilities” because they end
with that suffix. Flexibility, sustainability, reconfigurability, maintainability, and
interoperability are but a few that follow this grammatical pattern. In the meantime,
research into safety, quality, and even stability is well established in many engi-
neering curriculums as independent design methods. Still, other life cycle properties
14 Design of Large Engineering Systems 409

like resilience are relatively new and garner active research interest. These “ilities”
are usually classified as non-FRs; which are further classified as Cs in AD theory.
Nevertheless, most of these life-cycle properties have a highly integrative nature
and consequently, their integration into formal engineering design approaches like
AD theory presents a significant intellectual challenge.
Second, most large complex systems have a cyber-physical nature. In addition to
the underlying engineering physics, the system as a whole deploys control,
automation, and decision-making components that make the system operate effi-
ciently, reliably, and stably. These components can be entirely automated as in the
case of PID controllers, or entirely manual as in the case of aircraft pilots. Fur-
thermore, the control, automation, or decision-making system can have centralized,
distributed, or decentralized algorithms. Figure 14.25 contrasts four types of
cyber-physical systems: (a) open-loop physical systems, (b) closed-loop
cyber-physical systems, (c) closed-loop cyber-physical systems a centralized con-
troller, and (d) closed-loop cyber-physical systems with a distributed control
architecture. Each of these can be modeled as a SysML block diagram, analyzed for
its system behavior, or inspected in terms of its underlying system structure. The
challenge, here, is that many closed-loop control systems end up creating
cyber-physical systems with coupled designs. Consequently, the design workflows
develop feedback loops that are not easily managed and prone to design error;
particularly for large complex systems with demanding engineering physics. One
can perhaps speculate whether the recent news about the Boeing 737 MAX air-
planes is the result of the feedback design loops caused by the plane’s underlying
cyber-physical nature. In any case, we must recognize that as the modern world
continues to automate its technologies, it designs cyber-physical systems that are
increasingly safety–critical. Developing engineering design approaches that target
the closed-loop nature of cyber-physical systems is imperative.
Finally, the study of large flexible engineering systems must continue to develop
through hetero-functional graph theory. Three broad communities are actively
working to develop the methods to tackle the types of systems that sit at the heart of
the NAE’s grand challenges. The network science community has deployed graph
theory as a means to analyze the form of large flexible engineering systems. These
works, however, are relatively divorced from the established understanding of
function within the engineering design field. Furthermore, many of the large-scale
architectural transformations of the twenty-first century appear within the functional
architecture or in the system knowledge base, but not in the physical architecture
alone. In the meantime, model-based system engineering has established graphical
modeling techniques like SysML to approach systems of arbitrary size and com-
plexity. However, the quantitative understanding of the underlying system structure
has remained elusive. In the meantime, much of the AD community focuses on the
quantification of such a system structure, but most investigations have been limited
to large fixed engineering systems. To the network science community,
hetero-functional networks present a new view as to how to construct graphs with
fundamentally different meanings and insighst. To the systems engineering com-
munity, hetero-functional graph theory may come to be viewed as quantification of
410 G.-J. Park and A. M. Farid

Fig. 14.25 Cyber-physical systems from the perspectives of SysML, transfer functions, and
Axiomatic Design

many of the structural concepts in model-based systems engineering. Finally, to the

engineering design student interested in large flexible engineering systems,
hetero-functional graph theory presents itself as a natural extension of AD theory
where the Independence Axiom is applied to study systems of flexible structure.

Problems
1. Former mayor of Atlanta in the state of Georgia and the U.S. Ambassador to the
United Nations learned about the MH project at KAIST. He thought the MH
offers an ideal solution to the problem he has been concerned about for years.
His idea was to make the state of Georgia a hub for cargo transportation, just as
he made the city of Atlanta the hub of air transportation by enlarging the Atlanta
airport, which had made a significant impact on Georgia's economy.

His new idea was as follows: When large containerships come through the
Panama Canal from Asia, there are no large harbors that can accommodate these
large ships that require ports with deep waters. One solution is to moor these
ships off the Georgia coast and then let MHs unload containers from the large
container ships in the open sea about 35 miles offshore from the Georgia coast.
Then, these MHs can transport the containers from the containership in the open
sea to Savannah, Georgia, or Jacksonville, Florida, as well as other harbors in
14 Design of Large Engineering Systems 411

the Florida Pan-Handle. The containers transported to the Savannah harbor from
the containerships by the MH can then be transported to the East Coast cities as
well as the Mid-West cities of the United States by the railroad system that is
already there.

This proposed system can solve the current problems. There are no harbors with
deep enough waters along the Eastern seaboard of the United States to
accommodate these large containerships. Therefore, most of the containers from
Asia are unloaded at harbors at Long Beach, or Los Angeles, California. They
are then transported by rail to the East Coast or the Mid-West region of the
United States.

Design a platform that can be anchored in the open sea with the capability of
supplying fuel and water to the large containership moored in the open sea.
2. One of the goals of automobile companies is to manufacture driverless auto-
mobiles that will take passengers to new destinations without the intervention of
drivers. The idea is to use sensors of many different kinds, computational power,
information storage systems, GPS (ground positioning system), AI (artificial
intelligence), automated steering system, telecommunications system, and
intelligent engines.
Propose your approach in developing such vehicles, identify potential problems,
define FRs, and select DPs and PVs.
Based on the FRs, DPs, and PVs selected, develop the design matrix for your
vehicle.
3. Many universities are large systems. Typical research universities have teaching
staff (i.e., professors, research staff, technical staff), administrative staff,
undergraduate students, and graduate students. They must have the right and
adequate physical facilities for teaching, research, housing for students, and
sports facilities. Your job as its president is to make this university one of the
best research universities in the world. The university has 4,000 undergraduate
students and 6,000 graduate students. It receives an annual budget from the
government, which can meet roughly 50% of its budget. All of its students do
not pay any tuition, which is covered by the budget allocated by the govern-
ment. Define the highest level FRs, DPs, and PVs after identifying the potential
problems that must be overcome to make it into one of the leading universities
in the world. Develop a strategic plan for the next 5 years.
4. Boeing 737 MAX, the new airplane introduced by Boeing Airplane Company,
developed flight control problems during take-off. Two of these planes crashed
soon after take-off during ascending. The speculation is that the computer
control system, which was installed to prevent the potential stalling problem
during ascending, failed to do its function. When the airplane stalls, it loses its
lifting force and can fall. The control system was so designed that the pilot did
412 G.-J. Park and A. M. Farid

Fig. 14.26 Schematic view of the front part of a train

not have any control over the computer system. Define the first three-levels of
FRs and DPs for this airplane, if you are in charge of the development of this
airplane. What caused this plane to crash?
5. Analyze the design of your cell phone using AD. The functions of the cell
phone do not have to be considered. Define the hierarchies of FRs and DPs only
for physical parts of the phone.
6. A schematic view of the front part of a train is illustrated in Fig. 14.26. The
figure indicates (1) train structure, (2) coupler, (3) honeycomb structure,
(4) headstock. The train is going leftward. The front part is composed of (2), (3),
and (4), and it absorbs the impact energy in a crash event. Generally, the coupler
absorbs the impact energy first and the headstock and honeycomb structure

Fig. 14.27 The airbag in the

steering system
14 Design of Large Engineering Systems 413

absorb the rest of the energy. The patent KR20090059520A can be referred for a
detailed explanation. Design the front part of the train using AD.
7. An automobile airbag system is illustrated in Fig. 14.27. The airbag is for a
driver and is installed inside of the steering system. Design the airbag using AD.

Various drones are being sold in the market due to an increase in demand.
Design a drone according to the following steps:

(1) define the purpose of your own drone (FRs);

(2) define the design specifications based on your environment and conditions
(Cs);
(3) design a product (DPs).

Using a hierarchy of FRs and DPs is recommended. You can pick a drone in the
market and modify it according to your definition.

14.8 Recently, pedestrian protection is a hot issue in the automotive industry.

Some engineers are studying the installation of a pedestrian protection airbag
system as illustrated in Fig. 14.28. When the car impacts a pedestrian, the airbag is
deployed according to the following process:

Fig. 14.28 A pedestrian protection airbag system (Hyun-Ik Yang, Yong-Won Yun, and
Gyung-Jin Park, Design of a pedestrian protection airbag system using experiments, Proceedings
of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering (230.9)
pp. 1182–1195. Copyright © 2016 (Sage Publishing). https://doi.org/10.1177/0954407015603854)
414 G.-J. Park and A. M. Farid

(1) a sensor in the bumper detects the collision;

(2) an actuator opens the hood a little. The actuator is equipped at the hinge that
connects the main body and the hood;
(3) the inflator generates gas to inflate the airbag cushion;
(4) the head of the pedestrian is impacted by the airbag.

The pedestrian airbag is installed in the housing at the back of the hood. The
airbag has a disadvantage that it can block the driver’s sight. Design the pedestrian
protection airbag using AD. Refer to Yang et al. (2016) for further information.

References
Farid A, Suh NP (eds) (2016) Axiomatic design in large systems. Springer, Cham, Switzerland
Hong EP, Park GJ (2010) Design information management of an on-line electric vehicle using
axiomatic design. SAE Int J Mater Manufactur 3(1):133–141. https://saemobilus.sae.org/
content/2010-01-0279/
Lee S, Hong E, Park GJ (2009) Analysis of a mount type HVAC control system using axiomatic
design. In: Coelho AG (ed) Proceedings of the 5th international conference on axiomatic
design, Universidade Nova de Lisboa, accessed at Axiomatic Design Solutions, Inc., Lisboa,
Portugal
Lee T, Park GJ (2010) Managing system design process using axiomatic design: a case on KAIST
mobile harbor project. SAE Int J Mater Manufact (1):125–132. https://saemobilus.sae.org/
content/2010-01-0278/
Nosorog UA (2019) Car suspension separately from the car isolated on white 3D illustration.
Photo licensed from Shutterstock, ID: 415090513
Park GJ (2007) Analytic methods for design practice. Springer
Park GJ (2018) Application of axiomatic design to the design of automobile parts (keynote
speech). In: Puik E, Foley JT, Cochran D, Betasolo M (eds) 12th international conference on
axiomatic design (ICAD), MATEC web of conferences, Reykjavík, Iceland, contained
proprietary data, contact author for details
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Yang HI, Yun YW, Park GJ (2016) Design of a pedestrian protection airbag system using
experiments. In: Proceedings of the institution of mechanical engineers, Part D: journal of
automobile engineering, vol 230.9, pp 1182–1195

Further Reading

Buede DM (2009) The Engineering Design of Systems: Models and Methods. Hoboken, N.J.:
John Wiley & Sons, 2nd Ed.
Crawley E, Cameron B, Selva D (2015) System Architecture: Strategy and Product Development
for Complex Systems. Upper Saddle River, N.J.: Prentice Hall Press
Farid AM, Suh NP (2016) Axiomatic Design in Large Systems: Complex Products, Buildings and
Manufacturing Systems. Berlin, Heidelberg: Springer
14 Design of Large Engineering Systems 415

National Academy of Engineering (2019) “NAE Grand Challenges for Engineering,” https://www.
engineeringchallenges.org/challenges.aspx
Schoonenberg WC, Khayal IS, Farid AM (2018) A Hetero-functional Graph Theory for Modeling
Interdependent Smart City Infrastructure. Berlin, Heidelberg: Springer
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Complexity in the Kitchen
15
Joseph T. Foley, Erik Puik, Lindy Puik, Joseph Smith,
and David S. Cochran

Abstract
Axiomatic Design (AD) and Complexity Theory are often applied to highly com-
plex and technological systems that provide educators with many engineering
examples and case studies. The use of AD is applicable outside of these areas.
However, there are not many examples outside of these areas. As a result, stu-
dents often have trouble understanding the breadth and impact of Axiomatic
Design’s application to problem-solving. One large complex system that is often

Author’s Note: This chapter is an adaptation of the paper “Complexity in the Kitchen” presented by
the same authors at ICAD2019 in Sydney Australia (Foley et al. 2019) . This paper was published
open access at https://doi.org/10.1051/matecconf/201930100007 under a CC BY 4.0 license.

J. T. Foley (B)
Reykjavík University, Menntavegur 1, Reykjavík, Iceland
e-mail: [email protected]; [email protected]
Massachusetts Institute of Technology, 77 Massachusetts Ave, Cambridge, MA, USA
E. Puik
HU Utrecht, Hudsondreef 32, 3565 AV Utrecht, The Netherlands
e-mail: [email protected]
L. Puik
Eindhoven, The Netherlands
e-mail: [email protected]
J. Smith
Purdue University—Fort Wayne, 2101 E. Coliseum Blvd, ETCS233, Fort Wayne, IN, USA
e-mail: [email protected]
D. S. Cochran
Purdue University—Fort Wayne, 2101 E. Coliseum Blvd, ETCS229B, Fort Wayne, IN, USA
e-mail: [email protected]
© Springer Nature Switzerland AG 2021 417
N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_15
418 J. T. Foley et al.

overlooked is that of the kitchen. In this chapter, we present different food-related

preparation tasks that are inherently complex: cooking a turkey, baking an apple
pie, reverse engineering a recipe, and designing ecologically minded food packaging
while also discussing the impact of prepared food’s packaging approaches on the
environment. The authors believe such examples demonstrate Axiomatic Design’s
applicability in a new aspect that is approachable to a wide audience.

15.1 Introduction

Axiomatic Design (AD) has been already shown to be effective in highly technical
realms and creative areas. Technical applications include Shape Memory Allow actu-
ator testing (Pétursson et al. 2017), Industry 4.0 Human–Robot interfacing (Gualtieri
et al. 2018), industrial safety practices (Iino and Nakao 2018), and reconfigurable
manufacturing system design at Mercedes-Benz (Kujawa et al. 2018). Nontechnical
interest comes in the areas of football1 (Rolli et al. 2018), diabetes treatment (Smith
et al. 2018), university department organization (Suh 2001, 2015), special education
curricula (Wettasinghe and Koh 2008) and interactive art (Foley and Harðardóttir
2016).
These articles provide case studies and analogies that are relevant to engineering
students and the technically literate. Unfortunately, for those who wish to use AD
outside of those fields, educators are left with examples that resonate with engineering
students, but not with the design and student community at large. The nontechnical
analogy most commonly used by educators is that of designing a water faucet that is
easy to use. This example was initially published by Suh (2001) and further expanded
and revised by Foley et al. (2017).
One area that contains highly technical challenges that are approachable by a
large audience is that of food preparation. The ubiquity and success of cooking show
clearly indicate large interest from the viewing public. In the latest edition of “On
Food and Cooking,” Harold McGee celebrates that the worlds of science and culinary
arts have become tightly integrated and collaborative since he first wrote the work
that has brought him fame (McGee 2007). The authors agree that this indicates that
it is an excellent area to exploit for interdisciplinary research while also providing
rich examples to use in teaching design. The idea of using food preparation as a
design exercise is not unique: Slocum (2008) uses cooking dinner as one of his
common examples in his modified AD method called FRDPARRC. Cooking itself
is a transformative process, which fits well within the functional requirement (FRs)
mapping to the physical requirement aspect of AD.

1 Also known in the United States as Soccer, which is distrinct from what some might call “hand-
egg.”
15 Complexity in the Kitchen 419

15.1.1 Axiomatic Design in Other Words

The reader should be now familiar with the two axioms: (Suh, 2001).

Independence Axiom (1): “Maintain the independence of the FRs”

Information Axiom (2): “Minimize the information content of the design”

The authors have found that the axioms are often confusing to beginning prac-
titioners of the design discipline and non-technical parties of interest. Joe Foley
presented these at the ICAD2017 tutorials in a zen koan riddle (Foley 2017):

Independence Axiom (1): Find harmony in conflict

Information Axiom (2): Prepare for the unexpected

One might clarify this explanation further:

Independence Axiom (1): Modularize

Information Axiom (2): Choose robust elements and their combinations

To go into the creative realms, (Foley and Harðardóttir 2016) changed “Functional
Requirement” into a “Feeling Requirement”. The axioms also changed focus

Independence Axiom (1): Ensure the chosen emotions or interactions can be sep-
arately adjusted
Information Axiom (2): Choose implementations that reach the largest audience
with the highest impact

We have considered AD through multiple lenses to apply the same core con-
cepts into different disciplines. Now, we will quickly review the concept of Suh’s
complexity.

•? AD in other disciplines

Think of other disciplines, technical or not, and augment the two Axioms to be
appropriate for that area. Are two axioms sufficient? Do you need more or less?

15.1.2 Complexity Theory

To simplify the understanding of how many reliability (primarily Axiom 2) issues can
be addressed, Suh’s later focus combines the AD concepts into a singular complexity
Theory (Suh 2005).
420 J. T. Foley et al.

He explores the meaning of complexity, finally settling on “Complexity is defined

as a measure of uncertainty in achieving the specified FRs” (Suh 2005, p.58). Vosse-
beld et al. (2018) summarizes the four different types succinctly:

• Time-independent real complexity which is simply the information content of

a design: C R = I , where Ii = − log2 Pi and Pi is the probability of meeting
satisfying FRi . For uncoupled designs, the total information of the system is simply
the sum I = i Ii . For other cases, we refer the reader to Suh’s deeper discussion
in Suh (2005).
• Time-independent imaginary complexity arises in coupled or path-dependent
solutions where the order in which DPs should be addressed is unclear or improp-
erly ordered.
• Time-dependent combinatorial complexity develops in systems in which opera-
tion has a higher probability of going out of specification due to “continued expan-
sion in the number of possible combinations with time.” In short, time-dependent
combinatorial complexity describes systems that progress toward chaotic states
over some time.
• Time-dependent periodic complexity is similar to combinatorial complexity
except that a functional period has been identified over which the system can
be reset before it enters an unpredictable state.

15.1.3 Cooking Science

Harold McGee wrote one of the seminal works in trying to bridge the gap between
cooking and science in his work “On Food and Cooking” in 1984 later revised in
2004 (Donovan 2006; McGee 2007). This book is considered one of the required texts
for those interested in the science behind food and the cooking processes considered
commonplace today. When the book was first introduced, technical terms such as
“emulsified” and “denatured” were foreign to chefs except perhaps those previously
schooled in chemistry (Donovan 2006). McGee’s book is highly relevant to this
article due to its focus on a deeper understanding of how food and cooking processes
interact. To AD practitioners, the word “interact” should instantly remind us of the
concept of “coupling” and for good reason.

•? Definition

What is the definition of cooking? Must it involve heat? Does your definition fit
dishes such as kimchee,2 lutefisk,3 salami, and ceviche?4

2 Spicy Korean-style fermented cabbage often started with a raw oyster to begin fermentation.
3 Norwegian whitefish preserved in lye.
4 South American raw fish dish where the fish is cured in fresh citrus juices.
15 Complexity in the Kitchen 421

15.2 Cooking Axioms

We believe the AD axioms can be easily translated to cooking:

Independence Axiom (1): Reduce interaction between ingredients and processes

to only the ones desired
Information Axiom (2): Produce the desired food aspect reliably

Of note, there is one fast food chain that is famous for mastering Axiom 2 on a
global basis: the McDonald’s corporation. It is well known to the point of an adage
that “A McDonald’s hamburger is the same no matter where you are.” The unspo-
ken secondary part is that they are consistent, but not considered high quality, i.e.,
“consistent mediocrity” (Bramlett 2018); they could be considered high precision,
but low accuracy. In Sect. 15.5, we examine a structured way to evaluate the quality
of a cooking result, particularly for “complex” recipes. Perhaps we need to consider
this in terms of a quality definition heuristic (Cochran et al. 2016) shown in Eq. 15.1:
Satisfaction of Needs
quality = (15.1)
Resources consumed

•? Evaluation of food

When you last ate food prepared by another person, how would you rate its quality?
Can you state a set of FRs and measure how many resources were consumed? How
would a McDonald’s hamburger rate in the quality heuristic you developed?

To demonstrate how these axioms can direct cooking to a more desirable result
reliably, we will now present challenging recipes in the next sections.

15.3 Cooking a Turkey

In the United States, the holiday known as Thanksgiving is associated with roasting
a turkey in modern times (Fig. 15.1). How this came to be is a fascinating story which
we will explain in brief.5

5 Anexcellent resource on Thanksgiving’s turkey tradition is Davis’s “More than a Meal” (Davis
2001) which provided much of our historical resources and may convince some readers to go vegan.
422 J. T. Foley et al.

Fig. 15.1 The annual US

Thanksgiving tradition of
roasting a turkey.
(Reproduced with
permission
from Weinersmith (2011))

•? Holiday Gobble Gobble

If you eat turkey during the holidays, what was your favorite part and why? If you
don’t, what is your favorite alternative and why? If you were the one to prepare the
turkey, how happy were you with the result last time?

15.3.1 Why Does US Thanksgiving Mean Eating a Turkey?

The word turkey for the bird we know came into use during the Middle Ages.
Its origins are best explained as being associated with the Turkish Empire being
the main European trade route from which exotic birds such as the peafowl were
arriving (Davis 2001, p.28).

Three centuries before any actual turkeys appeared in sixteenth-century Europe, the word
turkey was being used to describe exotic birds from Asia (Schorger 1966, p. 16).

The choice of it being a roasted bird (Fig. 15.1) in the United States are clear due
to it being plentiful, especially in the New England area where the Puritan settlers
arrived (Davis 2001, p. 33). That said, records of the feast from the first Thanksgiving
mention a large variety of birds but no specific mention of turkey (Bradford 1981,
p. 100). Thanksgiving was not a significant national holiday in the United States
until 1863 when President Abraham Lincoln used it as a mechanism to promote
15 Complexity in the Kitchen 423

unity (Davis 2001, p. 53). Alexander Hamilton, the first secretary of the Treasury is
perhaps the earliest proponent of turkey being critical to Thanksgiving, saying “[n]o
citizen of the United States should refrain from turkey on Thanksgiving Day” in
1805 (Schorger 1966, p. 369; Davis 2001, p. 53). By 1857, it had become a traditional
part of Thanksgiving in New England. The English had an even earlier introduction
to Turkey in 1573, being referred to as “Christmas husbandlie fare” (Wright 1914, p.
338; Davis 2001, p. 54). The bird had been shipped there from Mexico by Spanish
explorers in the sixteenth century and became commonly bread during Renaissance
England (Davis 2001, p. 54). Strangely enough, the breeds that made their way there
were then brought back to the United States to become the forerunner of modern
domesticated turkey breeds (Davis 2001, p. 54). In short, the patriotic United States
holiday Thanksgiving, strangely enough, is all about devouring a Latin American
repatriated bird species!

15.3.2 Bringing the Bird to the Table

The challenge of cooking a turkey consistently was deeply considered in November

1993 by the cooking journal Cooks Illustrated.6
Anyone who has tried to cook a frozen turkey in any reasonable amount of time
has run into the problem of having the meat cooked thoroughly while also remaining
juicy. The problem stems from a few factors:

• a frozen turkey requires a huge amount of energy to heat the center;

• a large turkey has a large amount of thermal mass. (16 kg or greater);
• the turkey is not homogeneous, so the various parts cook at different rates;
• the US Department of Agriculture recommends an internal temperature of at least
165 ◦ F (73.9 ◦ C) to “destroy bacteria and prevent food-borne illness” (United
Stated Department of Agriculture 2015).

Rather than focus on the details of meat thermal models which have already been
explored in Papasidero et al. (2015), McGee et al. (1999), and Chang et al. (1998), we
place our attention on the overall design aspect. McGee7 discusses the challenge of
roasting whole birds including chickens and turkeys (McGee 2007). He agrees that
the challenge is that the meats are “best cooked differently.” Breast meat becomes
an unpleasantly chewy and tough texture to the palate if cooked much over 68 ◦ C.
Conversely, the leg and other dark meat have significant connective tissue that is
chewy at temperatures below 72 ◦ C. We have a contradiction, similar to what the
Russian design methodology TRIZ (Atshuller 1994) delights in; what is a cook to
do?

6 The first reference to analyzing turkey storage and cooking the authors were able to find was from

1962 in Goodwin et al. (1962).

7 Who is a co-author on one of the meat thermal models.
424 J. T. Foley et al.

•? TRIZ Turkey

Read about TRIZ:“Theory of the resolution of invention-related tasks”. What kind

of approaches would TRIZ suggest for cooking a frozen turkey?

Rather than be further distracted by TRIZ’s enthusiasm, let us consider the problem
in an axiomatic framework:

• FR1 = Heat breast meat to a maximum 68 ◦ C;

• FR2 = Heat thigh meat to a minimum 72 ◦ C.

Traditionally, cooks have tried to solve this with a single DP, which results in the
design matrix shown in Eq. 15.2.

• DP1 = Heated enclosure at 180 ◦ C for X minutes per kg.

   
FR1 X  
= DP1 (15.2)
FR2 X

•? TV Dinners

Frozen pre-compiled meals on included trays i.e. “TV dinners”8 must include instruc-
tions for preparation. Previously, these instructions were simply to place them into an
oven of the correct temperature for a set amount of time. In the modern era, instruc-
tions for a microwave oven of a “given wattage” are also included. What mechanisms
have the manufacturers done to ensure that the food is consistently cooked according
to these instructions? What are the failure cases for under-cooking and over-cooking?

This will clearly have issues: There is a single DP for 2 FRs, so it is inherently
coupled. Besides, the chances of meeting both FRs are very low since breasts and
thighs are both at the surface and have varying thicknesses. Perhaps we could change
the thickness of the meat, choosing turkeys with thicker breasts and following the
thermal transfer model derived by Chang et al. (1998). Unfortunately, measuring
the exact size of the turkey breast without removing it is rather challenging, so this
approach is abandoned.
One other thought is to compress our FRs:

• FR1 = Heat breast and thigh meat to a individual optimal temperatures.

8 Probably because people would often prepare them quickly to eat while watching a television
program.
15 Complexity in the Kitchen 425

Which does not show unreasonable coupling due to it being a 1FR1DP design
(Eq. 15.3), but has a large information content because the two temperatures are
different.  
 
FR1 = X DP1 (15.3)
An obvious improvement is to add feedback:

• DP1 = Heated enclosure at 180 ◦ C;

• DP2 = Thermometer.

Unfortunately, this still won’t work because of the contradiction. Clearly, we need
to uncouple the two FRs somehow. One answer is to physically uncouple them with
a kitchen knife as shown in Fig. 15.2:

• FR1 = Separate meat pieces according to dark vs. light meat;

• FR2 = Heat each type of meat to the optimal temperature:

– FR2.1 = Heat breast meat to a maximum of 68 ◦ C;

– FR2.2 = Heat thigh meat to a minimum 72 ◦ C.

• DP1 = Knife and knowledge of turkey anatomy;

• DP2 = Heated enclosure at 180 ◦ C and thermometer probe in meat.

The resulting design matrix in Eq. 15.4 is de-coupled or “path-dependent” indi-

cating that we must do things in the correct order. Clearly, we have to cut the turkey
before we can put it into the oven.
    
FR1 X 0 DP1
= (15.4)
FR2 X X DP2

Fig. 15.2 Joseph T. Foley’s

interpretation of un-coupling
a turkey’s anatomy to reduce
the cooking complexity.
(Reproduced from Foley
et al. (2019), originally
published open access under
a CC BY 4.0 license: https://
doi.org/10.1051/matecconf/
201930100007)
426 J. T. Foley et al.

Fig. 15.3 The Spidurky: a

novel method of increasing
drumstick availability.
(Reproduced with
permission
from Weinersmith (2018))

This new configuration greatly reduces the information content (Axiom 2) because
the temperature of each type of meat can be carefully controlled. Seeing that people
seem to prefer some parts of the turkey more than others, this seems like a very
simple way to uncouple. Unfortunately, such desirability of a particular type or part
of the meat such as the often desired drumstick can result in decisions that lead to
the terrifying implications supposed in Fig. 15.3.
Joseph T. Foley’s grandfather often told the story of having to do KP9 while on a
large naval vessel during the Second World War:
Design Story 15.1:
The ship had an impressive kitchen, which made sense since it had to feed a few
hundred seamen. It had a steam-heated cooking pot that took up much of the room
and was as deep as a man. To make it easy to get ingredients in, the top was accessible
from an upper floor. You had to be careful with that thing; any leaking steam would
cook you instantly and that’s how you cleaned it.
The head of the kitchen once came to me and said “Foley, today you’re making
chicken soup. The crates of frozen chicken are over in the freezer. Get 10 of them
out and put ’em in the pot. Get to it!”
Well, I dragged a couple of those crates from the freezer (which was also something
to see) and brought them into the room. I was pretty tired at that point because they
were heavy, but food needed to be made. I got a crowbar and started levering off the
sides of the wooden crate, pulling out the nails to make it easier to pry. I’d finished
one crate and discovered that the chickens were frozen in a solid block. I was prying
one out with the crowbar, covered in sweat, when the head of the kitchen came by
and gave me quite the look.
He shouted, “What are you doing, Foley?10 ”
I answered, “Preparing the chicken like you said, Sir!”
He retorted, “No! You’re wasting time!”
Can you guess what he did?

9 Kitchen Patrol, i.e., cooking.

10 Slightly paraphrased from the original naval method of speaking.
15 Complexity in the Kitchen 427

He grabbed a fire ax from the wall and began swinging with great speed and force
on one of the wooden crates. He kept at it until the crate was broken into large chunks.
Then he used a shovel to dump the chunks into the giant pot and handed the ax to
me. I asked, “But Sir, how will that make soup?” His reply was simple and to the
point: “Wood floats and nails sink. We’ll scoop ’em out later”
And that is how you made chicken soup in the Navy.

•? Scaling up food

Imagine you have been hired by a company to break the Guinness Book of World
Records for a food item such as pizza. How would you modify the traditional recipe
for that item so that it can break the current world record? Consider that the finished
product should be both food-safe and edible.

One must not forget the heat transfer aspect of cooking in an oven: the outside
temperature of a cut of meat will be much higher than the target temperature unless
special care is taken to remove the meat from the oven before it reaches the target
temperature for “carry-over” cooking.
To further reduce the chances of failure, we can take a lesson from the current food
movement of Sous-vide water baths similar to the Paté cooking described by McGee
(2007), p. 171:

• FR1 = Heat breast meat to a maximum 68 ◦ C;

• FR2 = Heat thigh meat to a minimum 72 ◦ C.

• DP1 = Immerse plastic-sealed breast meat chunks in 68 ◦ C water until equilibri-

ated;
• DP2 = Immerse plastic-sealed breast meat chunks in 72 ◦ C water until equilibri-
ated.

It would seem that we have found the perfect turkey cooking method, and in fact,
this is similar to what is done to cook turkey meat in restaurants when it is used
as a component or covered in sauces. Unfortunately, we have not considered our
customer’s needs carefully enough as in Vossebeld et al. (2018), and Girgenti et al.
(2016). The customers want the food to be presented in a particular way, not just be
at the correct temperature.

• CN1 = Put a cooked whole bird on the table so it looks pretty;

• CN2 = The skin must be crispy.

We need some way to cook certain parts of the bird selectively to address these
needs without going back into our coupled state and without mechanically separating
them.
428 J. T. Foley et al.

•? Super crispy

How do you make food crispy? Is it about the temperature? Is it about the cooking
process? How would you keep food crispy if it needs to be stored for at least a week?

15.3.3 How Do People Really Cook a Turkey?

Several innovative solutions have arisen to address these challenges to varying

degrees. With minimal effort, one can find videos and equipment for deep fat-frying
a turkey to speed up the process, obvious even in 1962 (McGee 2007; Goodwin et al.
1962). While faster, this greatly increases the chances of injury as the ice crystals
(or just moisture) may cause the hot oil to splatter and atomize, occasionally turning
the cooker into a fireball of epic proportions (Butler 2015; Osias 2006). The issue is
a large enough concern that there is a patent on a deep fryer that claims to make this
process safer (Osias 2006).

•? Fire in the hole!

Working with hot oil near an open flame is dangerous. How would you design an
oil-based cooking process that is less likely to destroy your kitchen? Consider the
axioms for guidance.

Defrosting the turkey ahead of time also can assist, if planning permits. This does
not address the problem of the differing composition of dark and white meat resulting
in certain parts cooked while the others are not. Again, necessity has bred innovation
in the form of placing aluminum foil as a radiant heat shield over the white meat
areas. One suggestion was to place ice packs over the white meat to cook them
selectively. As well as some of these approaches work, they still require significant
effort and quite a bit of skill to apply for repeatable success. This makes it clear that
cooking the turkey is a “complex system” in the Suh sense.
The concept of brining a turkey was the initial issue of Cooks Illustrated in 1993
since updated in 2004 (Hays 2004). Very recently, Lan Lam, Senior Editor of “Cook’s
Illustrated” re-examined the turkey process to see if the process could be further
streamlined in Lam (2016). Brining a turkey allows us to remove the contradiction
inherent in our two incompatible target temperatures by changing the meat chemistry
in a way that it can accept a larger variety of temperatures and maintain a high
level of moisture. This wider range of acceptable temperatures could be considered
“softening the spring” in Suh’s terminology. Lam’s improved recipe replaces the
short brining with a long refrigerated salting step, which accomplishes the same goal
and reduces the moisture to improve the skin’s crispiness. Doing this with a small
amount of sugar also caused the skin to brown nicely, effectively caramelizing the
skin.
15 Complexity in the Kitchen 429

The previous recipe suggested starting the turkey upside down so that the majority
of the heat focused on the dark meat (on the bottom of the bird). Lam wanted to find
an easier way to apply more heat to the bottom without the extra effort and found
it in the current method of making pizza at home: a pizza stone. This large stone,
when preheated correctly, stored enough heat that the turkey could simply be placed
in the roasting V-rack and not have to be manipulated. This innovation caused the
dark and light meat to finish cooking at the same time.

• FR1 = Keep white meat tender at 72 ◦ C and higher;

• FR2 = Crisp skin of turkey;
• FR3 = Heat meat to 72 ◦ C.

• DP1 = Salt to change the protein structure of meat to retain water;

• DP2 = Sugar solution;
• DP3 = Preheated pizza-stone, heated enclosure at 180 ◦ C, and thermometer probe
in meat.

In effect, the new procedure is effectively uncoupled as shown in Eq. 15.5. Each
of the FRs is only affected by its DP.
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎨ F R1 ⎬ X 0 0 ⎨DP1 ⎬
F R2 = ⎣ 0 X 0 ⎦ DP2 (15.5)
⎩ ⎭ ⎩ ⎭
F R2 0 0 X DP3
As can be easily seen, the continuing innovation in cooking a challenging roast
such as a turkey involves understanding the coupling and information content inher-
ent in each of the process choices made for having it arrive on the Thanksgiving
table.

15.4 Baking an Apple Pie

A famous simile often used in the United States is “As American as Apple Pie.” This
references the idolized atomic family period of the 1950s where the working spouse
would arrive home to the smell of a freshly baked apple pie (Fig. 15.4).

•? Think Global, Act Local

In the North East of the United States in 1980–1990s, it was popular to go to a “pick
your own” apple orchard to get apples for making apple pies (see Sect. 15.4) and other
dishes. These orchards had significant financial troubles due to an interesting change:
China has flooded the market with cheap applesauce. Why does this matter? It turns
out that after the apple picking season is done, most orchards take the remaining
apples, turn them into applesauce, and sell them to packaged food producers as a
healthy way to add flavor and sweetness to food. China and cheap shipping have
430 J. T. Foley et al.

Fig. 15.4 A classic apple

pie. (Reproduced with
permisssion from America’s
Test Kitchen (1997),
©America’s Test Kitchen)

suddenly caused this to often not be worth the effort to compete: the US producers
cannot make up the costs in labor for producing the sauce.
It turns out that the successful orchards have noticed that this is an issue and
have examined their CNs to realize that people come to an orchard not only for the
apples but for the experience. This has caused smart orchard owners to expand their
operations into baked products and amusement rides to bring people in to make back
that lost profit. What happens to the leftover apples now?
Consider a local business in your area that is having trouble competing with a
global competitor. Are there CNs and FRs that the local business have missed to
maintain their business or even grow? Propose a design of how this concept might
turn into a business venture.

The concept of a pie and apples themselves did not originate in the US (Eschner
2017). In fact, the association seems to have come about due to WW2 involvement
by the US (Kohatsu 2017). One of the author’s own experiences in baking pies
allowed him to immediately identify what makes them challenging when following
the traditional approach:

• CN1 = Top and bottom crusts should be flaky;

• CN2 = Crust should hold the fruit in but not be thick like pizza.

• FR1 = Separate many small blobs of fat and wheat flour;

• FR2 = Roll crust to consistent thickness of 3 mm±1 mm.

• DP1 = Butter cut into flour at a cold temperature until pea-sized lumps form, then
made into a ball and rolled out;
• DP2 = Water added until it can be manipulated with a rolling pin.
15 Complexity in the Kitchen 431

The problem becomes clear when examining the design matrix in Eq. 15.6:
    
FR1 X X DP1
= (15.6)
FR2 X X DP2
This concept is coupled because the water affects the layers of the starch and fat.
If you do not add enough water, the dough is hard to work with: it does not stay
together, crumbles, and is very fragile. If you add too much water, the dough is no
longer flaky when baked. The more you manipulate the mixture, the more gluten
forms and it gets leathery in texture.
In addition, unless the cook can work in a refrigerator, the butter will begin to
melt as it is worked. This creates a time dependence and reduces the chances of
success. Depending upon your dietary preferences or desired tastes, you can use
hydrogenated vegetable shortening, coconut oil, or even traditional lard (Lopez-alt
2007). Many of these fats fix the melting problem but does not have much taste, so
some butter needs to be added back in.
What we need is a different fluid that does not interact with starch and fat to
eliminate coupling. The choices of liquid in the standard kitchen are limited to
mostly water-based hydrolysates and oils. Pastry already has a fat, so additional oil
will affect the butter-starch interaction. Julia Childs comes to the rescue in the form
of the only other commonly available liquid: alcohol (Lopez-alt 2007). Alcohol acts
similar to water at room temperature but disappears at oven baking temperatures.
Also, alcohol inhibits the gluten-forming process, ensuring that flakiness is preserved
as you work the dough into shape as shown in Fig. 15.5.
These changes affect our DPs:

• DP1 = High-melting temperature fat with low moisture content cut into flour at a
cold temperature until pea-sized lumps form then made into a ball and rolled out;
• DP2 = Vodka added until it can be manipulated with a rolling pin.

Fig. 15.5 Effect of alcohol on pie dough. (Reproduced with permisssion from Lopez-alt (2007),
©America’s Test Kitchen)
432 J. T. Foley et al.
    
FR1 X 0 DP1
= (15.7)
FR2 0 X DP2
As shown in the Design Matrix in Eq. 15.7, these two small changes have reduced
complexity by removing coupling between the various elements, increasing the
chances that we will meet our requirements.

15.5 The Complexity of Reverse Engineering a Recipe

The complexity of cooking becomes painfully clear when a recipe needs to be

reversed engineered. Typical examples of this are found in a cooking show pro-
duced by Endemol, “Herman Against the Others” (inspired by the German show
“Kitchen Impossible”) in which top chef Herman den Blijker challenges known and
award-winning chefs to reproduce a foreign regional signature dish. A third for-
eign award-winning chef prepares the pretty complicated signature dish, typically
a regional specialty, and challenges Herman and his opponent to reproduce it. The
chefs have no or little experience in preparing it, as it is a regional dish that does
not meet their expertise. By tasting the food and observing its structure and char-
acteristics, they have to find out what the exact ingredients are and determine the
right procedure for preparation. They can taste the dish, take pictures, ask employees
in stores, and/or ask people in the streets. Obviously, time is limited so they have
to apply readily available knowledge for reproduction. The competing chefs both
present their dish to a qualified local jury consisting of 6–8 local experts that will
taste the dish, assess, and rate it. As the cooking show presents a battle between top
chefs, complexity is desired to make the show interesting for its viewers. This is
not unique as there are situations where complexity is desirable (Foley et al. 2016).
However, complexity should be reduced by the acquisition of knowledge at the end
of the show (Puik and Ceglarek 2014a, 2015). A few typical situations are described
and analyzed from a perspective of complexity in AD.

•? Taste Testing

How well could you identify the ingredients the last time you ate out or ate food that
was created by someone else? Could you identify the process used to cook it? Can
you estimate measures to tell when the food is finished being prepared?

The Axiomatic Maturity Diagram (AMD) (Puik and Ceglarek 2014b) will be
applied to visualize the processes, analyze the scores, and explain the complexity
of the reverse engineering process. The AMD, as applied in Figs. 15.7, 15.8, 15.9,
15.10 and 15.11 plots the status of the Axioms through the development process
(in time). The actual position of design activity in the AMD is determined by the
extent to which the Axioms are satisfied. The Independence Axiom is plotted on the
horizontal axis of the AMD and it describes the organization of a product design
15 Complexity in the Kitchen 433

or in this case the understanding of the cooking process. The vertical axis plots the
Information Axiom which is a measure for the robustness of a product’s design and
in this case, it plots how well the cooking process is executed. The goal of the design
process is the top-right dot in the AMD; both axioms are fully satisfied and the design
may be considered a “Good Design” (Suh 1990).
In this chapter, the process of cooking is considered to be a design process and the
process of reconstructing the dish after reverse engineering is plotted in the AMD.
In the cooking show, the chefs’ performances are rated with a grade on a scale of
1–10. To visualize this, the AMD is expanded with scores around the dot at the top
right side. The grades of the jury members (6–8 people) are averaged into a single
score. The open dot at the end of the development path represents the average and
final score per dish of the jury (one extra decimal for accuracy on a scale of 0–100).
The lines represent the development path that the chefs followed during their quest
to reconstruct the dish.

15.5.1 Reproduction of Braised Lamb Ribs

Michelin Guide “Bib Gourmand” awarded Icelandic chef Gísli Matt prepares Lamb
Ribs in a sweet and sour sauce (Fig. 15.6). While reverse engineering, the competing
chefs are both quite successful in determining the ingredients of this dish. Even the
way of preparing the dish is chosen well, both chefs infer that the meat is braised.
Unfortunately, the also Michelin Guide “Bib Gourmand” awarded chef Alain Caron,
who was born in France and moved to the Netherlands at the age of 26, is mis-
taken about the exact cut of meat to choose. He chooses a lamb loin instead of the
lower ribs. Because of this, his cooking result is not satisfactory. The lamb loin does
not have enough fat to make the meat soft during the braising process. His perfect
reconstruction of the sweet and sour sauce cannot prevent that his dish is assessed
disappointingly by the jury. Alain has made a structural error by selecting a wrong
DP (loin instead of ribs). Figure 15.7 shows the AMD.
The process of reconstructing the dish starts at the lower left of the AMD, where
ingredients and preparation methods are still unknown. The analysis of the Icelandic
dish leads to the organization of the recipe, coupling FRs and DPs, and as such

Fig. 15.6 Example of

braised lamb shanks.
(Reproduced with
permission from Gough
(2019))
434 J. T. Foley et al.

Fig. 15.7 Analysis of chef

Fully Robust 40 50 60 70 80 90
Alain’s mistake. There is a
structural error in the design

Information Axiom
of his recipe. Though the
cooking process is executed
well, he cannot recover from
the mistake, but he

Proof of Concept
No Organization
minimalizes damage as
much as possible.
(Reproduced from Foley
et al. (2019), originally
published open access under
a CC BY 4.0 license: https://
doi.org/10.1051/matecconf/
Alain
201930100007)
Herman

Not Robust
Independence Axiom

satisfying the Independence Axiom. Unfortunately for chef Alain, the FR “delivering
soft but rich mouth feeling” cannot be addressed by the DP “braised lamb loin”
because the muscles in the loin are too rigid and will not become soft enough within
the available time. In his recipe, the Independence Axiom is not fully satisfied and the
curve in the AMD does not reach the right side of the AMD. His perfectly executed
sweet and sour sauce brings him quite high in the AMD (correct ingredients that are
perfectly treated). His final score is a mere 71 points, while Herman’s dish scores 80
points while not being executed to perfection.

•? Morbitity and Mortality

Have you ever had a sense that your recipe was going to turn out well or not? What
was instinct telling you at the time? How did you feel at that moment? Did you do a
post-mortem of a failed recipe to determine what was wrong?

15.5.2 Reproduction of Scottish Haggis

The second example is about replication of the Scottish Haggis (Fig. 15.8). Haggis
is a savory pudding consisting of sheep’s pluck (liver, lungs, heart). It is prepared
by Haggis champion and butcher Fraser MacGregor. Herman’s opponent is Maaike
Dogan, a specialist in merging western and middle eastern food. In this battle, Herman
has a substantial advantage; he is familiar with Haggis, recognizes the dish, knows
about the ingredients, and instantly has ideas to reproduce it. Maaike is not familiar
15 Complexity in the Kitchen 435

Fig. 15.8 Example of

Scottish Haggis.
(Reproduced with
permission
from Stockcreations (2019))

Fig. 15.9 Herman starts off

Fully Robust 40 50 60 70 80 90
perfectly well, however,
though all his ingredients
Information Axiom

and procedures are perfectly

chosen, he is not capable of
matching the right flavors
and loses the battle.

Proof of Concept
No Organization

(Reproduced from Foley

et al. (2019), originally
published open access under
a CC BY 4.0 license: https://
doi.org/10.1051/matecconf/
201930100007)
Maaike

Herman

Not Robust
Independence Axiom

with the dish at all, and she is not able to determine the right ingredients. In fact,
she is far off. Instead of using lamb’s pluck, she is using minced beef and beef liver.
Halfway through the cooking process, it seems that her last chances of a good result
are gone when she almost burns her ingredients. This, however, is where the odds
turn. Though her ingredients are not the right ones, she shows her excellent cooking
skills and qualities in tasting. Due to her middle eastern origin, her skills of seasoning
with herbs and spices appear decisive in replicating the right flavor. Where Herman
had a big lead, he loses due to Maaike’s excellence in reproducing the right taste.
The AMD is shown in Fig. 15.9.
Though Herman’s potential is considerably better than Maaike’s, Herman gets
lower grades due to the incorrect flavor of his dish (68 vs 47 points).
436 J. T. Foley et al.

Fig. 15.10 Example of

Bavarian Hay Soup (Swiss
Hay Soup image not
available). (Reproduced with
permission from DM-Media
(2019))

•? Death by Chocolate

Imagine you are trying to make a chocolate cake with these ingredients: eggs, water,
butter, wheat flour, cocoa, baker’s chocolate. Without looking at an existing recipe,
develop the FR–DP matrix and fill in the transfer coefficients to determine what you
think are the correct ratios. Consider each ingredient and write down what you think
will happen if you add too much or too little of the ingredient. Try baking your recipe
and evaluate its result. How close were you in your first attempt? What would you
change in the next attempt?

15.5.3 Reproduction of Swiss Hay Soup

The third and last example is the reconstruction of a Swiss “Hay Soup.” It is a
rare Alpine dish usually only served above 1000–2000 m altitude and made of dried
alpine grass, flowers, and herbs. The dish is prepared by top chef Lukas Pfaff by
infusing the local hay with water (like preparing hot tea). The hay extract is mixed
with beef/veal and vegetable stock. It is served with a milk-foam and ashes from
burnt hay. Herman’s opponent is Jermain de Rozario, a young talented and fanatic
chef who recently received his first Michelin star.
This is a very unusual recipe because hay is an unaccustomed ingredient for
soup and the ashes of burnt hay on top of the milk-foam is a challenging attribute
to recognize. Both competing chefs are completely unfamiliar with the concept and
ingredients of hay soup and have knowledge making it difficult to properly reproduce
it. At this stage, Jermain takes a lead; being a young and eager chef he is willing to
learn. He has tasted the ashes of hay in the foam because he suspects that hay is used
in that part of the soup. Further, he shows two valuable behavioral characteristics:

• He visits a farm to acquire hay and gathers knowledge by asking the farmer about
the preparation of the soup. He does the same in the stores he visits. This is how
he learns about the right ingredients and the preparation of the soup;
15 Complexity in the Kitchen 437

Fig. 15.11 Jermain applies

Fully Robust 40 50 60 70 80 90
an iterative approach in

Information Axiom
which he optimizes his
recipe several times before
he finalizes it. He reflects
after every attempt and
gathers an understanding of

Proof of Concept
No Organization
the dish leading to an
excellent score. (Reproduced
from Foley et al. (2019),
originally published open
access under a CC BY 4.0
license: https://doi.org/10.
Jermain
1051/matecconf/
201930100007)
Herman

Not Robust
Independence Axiom

• While preparing the dish, he iterates the recipe of the soup. He tries something,
evaluates the result very explicitly, and adjusts the recipe to approach the flavors
of the soup as close as possible. This is how he improves his recipe several times
and executes that step again and again. Herman, on the other hand, compiles the
soup in a single procedure. Obviously, he tastes the result several times but he does
not change his recipe, he continues with what he has cooked in the last step.

Again the AMD of these processes is derived (Fig. 15.11).

Jermain’s performance is praised by the jury and they can hardly believe that this
soup was made by a different chef since he is so close to the initial dish. His score is
a total of 95 points in the AMD. Though Herman has used the right ingredients, his
hay is not infused in the right way. The process of infusing took too long and made
the soup bitter. His average grade does not go beyond 60 points.

•? Delicacies

Iceland has several foods that are “delicacies” eaten at certain times of the year:
fermented Greenland shark, fermented whale fat, and blowtorch-seared split sheep’s
head.11 To many people, this food seems very strange (similar to the previously
described hay soup). Determine why these foods became part of the Icelandic culture.
Do you have similarly highly-regional foods that are not found outside of your region
of the world? How did this happen?

11 The eyeball is a particular delight to many.

438 J. T. Foley et al.

15.6 Achieving Functional Requirements with Less Waste

With the move from family farm-centered food distribution to supermarkets, the need
for food packaging has increased. In today’s busy world, convenience is the name of
the game. Disposable, single-serve food items have become a staple at the grocery,
and even if you are preparing your meal at home, many of the ingredients come
packaged in a variety of different materials. With an AD viewpoint, convenience is
an FR that is customer need-driven, and the common DP is usually centered around
easily disposable packaging.

•? The Dangers of Too Much Coffee

Due to the high price and significant processing of coffee, it tends to be a crop that
uses a lot of pesticides. Read about how coffee is produced and look for the coupling
elements. Propose changes to eliminate that coupling such that the need for pesticides
is lower while maintaining coffee quality.

More recently, sustainable/earth-friendly packaging is becoming an FR born out

of the moral and ethical responsibilities felt by the actual consumer of the product.
Although this customer need is being driven by the actual consumer of the product,
could “The Environment” be the actual customer representing this need? The AD
framework provides a way to express the environment as a customer and derive FRs
based on the needs of the environment.
For choosing DPs to minimize the impact that a product or process has on the
environment, we can use the Design For Environment (DFE) methodology. The DFE
methodology provides the means for an organization to minimize the impact of its
product and processes on the environment (Ulrich and Eppinger 2015) and could aid
in choosing more earth-friendly DPs. To understand how adding FRs derived from
environmental needs can affect the product design, the first axiom can be used to
understand the resulting path dependency of the design.
One FR could be, “Ensure product packaging is sustainable.” For example, there
are many possible DPs that could achieve this requirement such as recyclable mate-
rials, biodegradable materials, or even returnable/reusable packaging. The question
is what effect does the addition of an FR related to sustainable packaging have on
the ability to achieve the other FRs?
Consider the “K-Cup” for a moment shown in Fig. 15.12. This invention has
brought the ultimate convenience to brewing a cup of coffee in the morning, but
at the expense of millions of single-use plastic cups hitting the landfill every day.
A few companies have recognized this incredible amount of waste and have added
an FR to “reduce packaging waste” to their product design to remedy the problem.
Biodegradable K-Cups (Fig. 15.13) are now manufactured by those companies, and
the customer experiences the same level of convenience as a normal K-Cup, but
without the pollution waste from the plastic. The coffee may even taste better since
it is not being brewed in plastic.
15 Complexity in the Kitchen 439

Fig. 15.12 Traditional

K-Cup. (Reproduced
from Foley et al. (2019),
originally published open
access under a CC BY 4.0
license: https://doi.org/10.
1051/matecconf/
201930100007)

Fig. 15.13 Compostable

K-Cup. (Reproduced
from Foley et al. (2019),
originally published open
access under a CC BY 4.0
license: https://doi.org/10.
1051/matecconf/
201930100007)

The effect of adding a DP for sustainable packaging can be seen in the comparison
of the design matrices for Design A (generic plastic K-Cup, Eq. 15.8) and Design B
(Biodegradable K-Cups, Eq. 15.9). Both designs are coupled. How can the market-
place accept and use product designs that are seemingly coupled?

Design A:     
FRConvenient X X DPConvenient
= (15.8)
FRLow Cost X X DPLow Cost
Design B:
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎨FREarth−Friendly ⎬ X X X ⎨DPEarth−Friendly ⎬
FRConvenient = ⎣X X X ⎦ DPConvenient (15.9)
⎩ ⎭ ⎩ ⎭
FRLow Cost X X X DPLow Cost
With AD, the designer can achieve the design intention by choosing DPs that
decouple a design (Suh 2001), but this decoupling can also result from customer
perception and acceptance of a DP. For example, packaging coffee into single-serve
plastic cups results in a higher cost per cup of coffee for the consumer, thus the choice
of DP for convenience will affect the achievement of a low-cost cup of coffee, but the
choice of DP for the low cost may also affect the achievement of convenience. There-
fore, the design is coupled, but with the sales and widespread success of single-serve
440 J. T. Foley et al.

coffee makers, clearly, the consumer is willing to pay more for convenience. For the
consumer, there may be no significance of the increased cost for added convenience.
Therefore, there is little to no relationship between low cost and convenience as
expressed by Eq. 15.10.

Customer-Perceived Design A:
    
FRLow Cost X 0 DPLow Cost
= (15.10)
FRConvenient X X DPConvenient
Also, customer perception may affect the design matrix for Design B which con-
siders the earth-friendly FR. For instance, a consumer looking to purchase a more
earth-friendly K-cup may not perceive additional cost or less convenience to be a
barrier to purchasing the product. Therefore, Design B becomes a partially coupled
design as expressed by Eq. 15.10. Both customer-perceived designs become lower
triangular.

Customer-Perceived Design B:
⎧ ⎫ ⎡ ⎤⎧ ⎫
⎨FREarth−Friendly ⎬ X 0 0 ⎨DPBiodegradable ⎬
FRLow Cost = ⎣X X 0 ⎦ DPLow Cost (15.11)
⎩ ⎭ ⎩ ⎭
FRConvenient X X X DPConvenient
The thinking is that customer perception and the acceptance of innovations can
decouple designs that were previously considered to be coupled. Design intention
allows the designer the opportunity to recognize that if the customer accepts the
addition of an FR, then a seemingly coupled design can become decoupled. In this
way, a designer can rely on marketing to create the customer need that drives this
additional FR.

15.7 Conclusion

In this chapter, we have considered several complex kitchen-related tasks and oper-
ations in the lens of AD. When cooking a turkey, it advises the cook to decouple the
different incompatible cooking materials: dark meat and white meat. If dis-assembly
was not acceptable, the cook needs to increase the design range by brining and
changing the heat-transfer between the different compositions of meat.
For apple pie, the coupling appears in the choice of fat, moisture, and its interaction
with gluten in making a pie crust. The traditional recipe is high-information content
because it requires tightly controlled conditions to have a crust that is flaky but can
be manipulated into the proper shape. AD again asks us to find ways to uncouple
these three elements by finding a different working fluid, alcohol, which separates
the workability from gluten formation.
Reverse engineering of food has many similarities with traditional reverse engi-
neering of product and system designs and AD’s complexity definition may be well
15 Complexity in the Kitchen 441

used. As an analog to traditional engineering, knowledge acquisition is the cen-

tral theme here. Knowledge and complexity have an inverse relationship; as relevant
knowledge increases, complexity is reduced. Just like a product designer that is learn-
ing while designing, a chef reconstructing a dish needs knowledge and understanding
of the recipe: structured analysis in (re)evaluation supports this process.
In addition to the recipe, food preparation requires the equipment and the food as
the raw materials. Innovation in both equipment and packaging design was discussed
from the customer viewpoint. AD was used to describe that customer acceptance of
equipment and packaging innovations can decouple designs that are coupled. The
K-Cup was used to illustrate these points and further illustrates the significance and
consequence of design intention.
We hope these examples of complexity in the kitchen have inspired you to look for
your own opportunities to create food the way you like it more easily and robustly.
Problems
1. Kolmogorov complexity measures complexity by considering how difficult it is to
describe a sequence using a programming language. Is this an effective measure
of complexity in a cooking recipe? If not, how would you adjust Kolmogorov’s
complexity to give an effective measure of complexity in a cooking recipe?
2. One of the big innovations in beer packaging was the incorporation of a small
plastic ball in Guinness cans to ensure that the foamy “head” that was produced
when pouring was similar to a draught from a tap. Consider another packaged
food item that you found inferior to the product in another venue. Based upon
these differences, build up a CN to FR to DP mapping and look for where the
complexity is coming into your perception of the product. Propose a solution.
3. With the popularity of cooking game shows, it is challenging to come up with
a new concept. The Desirable Complexity approach in Foley et al. (2016) states
that for a “fun” game, you must use invert AD best practices to create appropriate
challenges. Based upon violating Axioms 1 and 2, develop a cooking game show
that would be at an appropriate level for amateur cooks.
4. You are developing new Meals Ready to Eat (MRE) for extreme adventurers
going into remote locations. What are the CNs, FRs, and DPs you would choose
to keep the adventurers fed when you are unsure of where they will be going?
How would you deal with the constraints of someone with a nut allergy or a
religious food restriction?
5. In US homes, a staple for sending food for children to school is the peanut-
and-jelly sandwich.12 Such food is not appropriate for feeding in homes for the
elderly due to the high risk of inhalation and suffocation. Design a better portable
and convenient food item for distribution to the elderly during outings.

12 The lead author is not a fan of these due to his severe nut allergy.
442 J. T. Foley et al.

6. Consider the task of feeding yourself (or your family). Apply the Complexity
Theory method to this problem. Start with CNs, FRs, and develop DPs. Create a
Design matrix, design decomposition, or Design Record Graph (Iino and Nakao
2016) to show the dependencies between the FRs. Finally, consider the com-
plexity for the design of the meal: time-independent real complexity for meeting
each of the goals outlined in the FRs (such as having food ready at a given time)
and time-dependent combinatorial or periodic complexity as appropriate.
7. How to consider the level of recyclability in a product (such as a K-cup) is
hard to incorporate into the design methodologies as described. Is it possible to
use the “battery symbol” approach described in Iino and Nakao (2016) to tag
elements that can be recycled easily? Discuss how you would modify AD to
better incorporate sustainable elements such as this.
8. Consider a single-use food preparation item such as the K-cup. Develop the
Design Matrix for it from the user’s standpoint. Develop the Design Matrix for
it from the product designer’s standpoint. Why are they different? How would
you adjust the design to include sustainability in the design intent?

References
America’s Test Kitchen (1997) Recipe: Classic apple pie. https://www.cooksillustrated.com/
recipes/1377-classic-apple-pie
Atshuller G (1994) And suddenly the inventor appeared. Technical Innovation Center, translated
from Russan by Leev Shulyak
Bradford W (1981) Of Plymouth Plantation 1620–1647. Modern Library, New York, first printed
in entirety 1856
Bramlett J (2018) The unconventional thinking of dominant companies. CreateSpace Independent
Publishing Platform
Butler G (2015) 15 greatest deep-fried turkey disasters caught on video. https://www.oregonlive.
com/cooking/2015/11/12_greatest_deep-fried_turkey.html
Chang HC, Carpenter JA, Toledo RT (1998) Modeling heat transfer during oven roasting of unstuffed
turkeys. J Food Sci 63(2):257–261. https://doi.org/10.1111/j.1365-2621.1998.tb15721.x
Cochran DS, Foley JT, Bi Z (2016) Use of the Manufacturing System Design Decomposition for
comparative analysis and effective design of production systems. Int J Prod Res 55:870–890
Davis K (2001) More than a meal. Lantern Books, New York, NY, USA
DM-Media (2019) Bavarian specialty, special soup, hay soup with bread and spices, vegetarian.
Photo licensed from Shutterstock, ID: 667673017
Donovan M (2006) On food and cooking: the science and lore of the kitchen, 2nd edn. Gastronomica:
The J Crit Food Stud 6(4):117–118. https://doi.org/10.1525/gfc.2006.6.4.117, http://gcfs.ucpress.
edu/content/6/4/117, book Review
15 Complexity in the Kitchen 443

Eschner K (2017) Apple pie is not all that American. Smithsonian. https://www.smithsonianmag.
com/smart-news/why-apple-pie-linked-america-180963157/
Foley JT (2017) Tutorial: Axiomatic Design for everyone. In: Slǎtineanu (2017), p 68
Foley JT, Harðardóttir S, (2016) Creative Axiomatic Design. 26th CIRP Design Conference. Pro-
cedia CIRP, Elsevier ScienceDirect, Stockholm, Sweden, pp 688–694
Foley JT, Puik E, Cochran DS (2016) Desirable Complexity. In: Liu (2016), pp 101–106. https://
doi.org/10.1016/j.procir.2016.05.023
Foley JT, Puik E, Cochran DS (2017) The faucet reloaded: improving Axiomatic Design by example.
In: Slǎtineanu (2017), p 7. https://doi.org/10.1051/matecconf/201712701009
Foley JT, Puik E, Puik L, Smith J, Cochran DS (2019) Complexity in the kitchen. In: Liu A, Puik
E, Foley JT (eds) 12th International Conference on Axiomatic Design (ICAD), MATEC web of
conferences. Sydney, Australia, p 12. https://doi.org/10.1051/matecconf/201930100007
Girgenti A, Pacifici B, Ciappi A, Giorgetti A (2016) An Axiomatic Design approach for customer
satisfaction through a lean start-up framework. In: Liu (2016), pp 151–157. https://doi.org/10.
1016/j.procir.2016.06.101
Goodwin TL, Mickelberry WC, Stadelman WJ (1962) The effect of freezing, method of cooking, and
storage time on the tenderness of pre-cooked and raw turkey meat. Poultry Sci 41(4):1268–1271.
https://doi.org/10.3382/ps.0411268, https://academic.oup.com/ps/article/41/4/1268/1614461
Gough J (2019) Roasted lamb shanks. Photo licensed from Shutterstock, ID: 55000708
Gualtieri L, Rauch E, Rojas R, Vidoni R, Matt DT (2018) In: Puik et al (2018) Application of
Axiomatic Design for the design of a safe collaborative human-robot assembly workplace. p
01003. https://doi.org/10.1051/matecconf/201822301003
Hays R (2004) How to brine a turkey. Cook’s Illustrated. https://www.cooksillustrated.com/articles/
36-how-to-brine-a-turkey#print
Iino K, Nakao M (2016) Design Record Graph and Axiomatic Design for creative design education.
In: Liu https://www.sciencedirect.com/science/article/pii/S2212827116307764
Iino K, Nakao M (2018) Human design parameters for safety of products and systems. In: Puik
et al, p 01002. https://doi.org/10.1051/matecconf/201822301002
Kohatsu K (2017) Why are we ’as American as apple pie’? Huffington Post. https://www.huffpost.
com/entry/why-are-we-as-american-as_n_6227462, originally published 2014
Kujawa K, Weber J, Puik E, Paetzold K (2018) Exploring and ADAPT!—extending the ADAPT!
method to develop reconfigurable manufacturing systems. In: Puik et al (2018), p 01006. https://
doi.org/10.1051/matecconf/201822301006
Lam L (2016) Easier roast turkey and gravy. Cook’s Illustrated. https://www.cooksillustrated.com/
articles/331-easier-roast-turkey-and-gravy
Liu A (ed) (2016) 10th International Conference on Axiomatic Design (ICAD), vol 53, Procedia
CIRP, Elsevier ScienceDirect, Xi’an, Shaanxi, China
Lopez-alt JK (2007) Foolproof pie dough. Cook’s Illustrated. https://www.cooksillustrated.com/
articles/37-foolproof-pie-dough
McGee H (2007) On food and cooking. Scribner, revised from 1984 version
McGee H, Mcinerney J, Harrus A (1999) The virtual cook: modeling heat transfer in the kitchen.
Phys Today—PHYS TODAY 52:30–36. https://doi.org/10.1063/1.882728
Osias J (2006) Safe turkey deep fryer. US Patent 20060272633A1
Papasidero D, Pierucci S, Manenti F, Piazza L (2015) Heat and mass transfer in roast beef cooking.
temperature and weight loss prediction. Chem Eng Trans 43:151–156. https://doi.org/10.3303/
CET1543026
444 J. T. Foley et al.

Pétursson E, Karlsson IN, Garðarsson OG, Pálsson P, Saulius Genutis VO, Foley JT (2017)
Axiomatic design of equipment for analysis of SMA spring degradation during electronic actua-
tion. In: Complex systems engineering and development proceedings of the 27th CIRP design con-
ference, Procedia CIRP, Elsevier ScienceDirect, Cranfield University, UK, pp 261–266. https://
doi.org/10.1016/j.procir.2017.01.055
Puik E, Ceglarek D (2014a) A review on information in design. In: Thompson. https://
axiomaticdesign.com/technology/icad/icad2014/9-Puik-et-al-paper.pdf
Puik E, Ceglarek D (2014b) A theory of maturity. In: Thompson. https://axiomaticdesign.com/
technology/icad/icad2014/17-Puik-et-al-paper.pdf
Puik E, Ceglarek D (2015) The quality of a design will not exceed the knowledge of its designer; an
analysis based on axiomatic information and the cynefin framework. In: Thompson et al. https://
doi.org/10.1016/j.procir.2015.07.040
Puik E, Foley JT, Cochran D, Betasolo M (eds) (2018) 12th International Conference on Axiomatic
Design (ICAD), MATEC web of conferences, Reykjavík, Iceland
Rolli F, Fradinho J, Giorgetti A, Citti P, Arcidiacono G (2018) Axiomatic decomposition of a zero-
sum game: the penalty shoot-out case. In: Puik et al, p 01005. https://doi.org/10.1051/matecconf/
201822301005
Schorger AW (1966) The wild turkey: its history and domestication. University of Oklahoma Press,
Norman
Slocum AH (2008) FUNdaMENTALS of design. MIT precision engineering research group. http://
pergatory.mit.edu/resources/FUNdaMENTALS.html
Slǎtineanu L (ed) (2017) 11th International Conference on Axiomatic Design (ICAD), MATEC
web of conferences, Iasi, Romania
Smith JJ, Shah SA, Cochran DS (2018) Prevention, early detection, and reversal of type-2 diabetes
using Collective System Design. In: Puik et al, p 01018. https://doi.org/10.1051/matecconf/
201822301018
Stockcreations (2019) Cooked sliced open haggis and vegetables with mashed turnip, potato and
fried onions on a rustic wood table with copy space. Photo licensed from Shutterstock, ID:
611174096
Suh NP (1990) The principles of design. Oxford University Press
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Suh NP (2005) Complexity. Oxford University Press
Suh NP (2015) Challenges in dealing with large systems. In: Thompson et al, pp 1–15, keynote
Thompson MK (ed) (2014) 8th International Conference on Axiomatic Design (ICAD 2014), vol
33. CIRP, Axiomatic Design Solutions Inc, Lisboa, Portugal
Thompson MK, Giorgetti A, Citti P, Matt D, Suh NP (eds) (2015) 9th International Conference on
Axiomatic Design (ICAD), vol 34. Procedia CIRP, Elsevier ScienceDirect, Florence, Italy
Ulrich K, Eppinger S (2015) Product design and development, 6th edn. McGraw-Hill Education,
New York, NY
United Stated Department of Agriculture (2015) Turkey basics: safe cooking. https://www.fsis.
usda.gov/wps/portal/fsis/topics/food-safety-education/get-answers/food-safety-fact-sheets/
poultry-preparation/turkey-basics-safe-cooking/CT_Index
Vossebeld DM, Foley JT, Puik E (2018) The complexity of mapping customer needs ... (and the myth
of a unanimous customer). In: Puik et al, p 7. https://doi.org/10.1051/matecconf/201822301024
Weinersmith Z (2011) Saturday morning breakfast cereal comic: dinosaur to turkey. https://www.
smbc-comics.com/comic/2011-11-08
Weinersmith Z (2018) Saturday morning breakfast cereal comic: GMO. https://www.smbc-comics.
com/comic/gmo
15 Complexity in the Kitchen 445

Wettasinghe CM, Koh TH (2008) Axiomatic Design Theory for the analysis, comparison, and
redesign of curriculum for special education. Disabil Rehabil Assist Technol 3(6):309–314.
https://doi.org/10.1080/17483100802281012
Wright AH (1914) Early records of the wild turkey. The Auk: A Q J Ornithol 31(3):334–358
Design of Organizations
16
Nam Pyo Suh

Abstract
Sound well designed organizations are most important organizations should be
design well for many different reasons, such as efficiency, reliability, rational
decisions, and productivity, and most of all, in achieving the mission of the
organization. Some of the significant failures of high-technology products and
major corporations, as well as successful enterprises and products, may be
attributed to organizational and personnel issues.
People design a variety of organizations that are fundamental units of human
society to achieve specific goals such as those related to educational,
manufacturing, financial, legal, and political purpose. An organizational unit
could be a nation, industrial firm, university, army, hospital, and a non-profit
organization. An organization may consist of several lower level organizations,
forming a hierarchical structure among higher level and lower level units to
fulfill the missions and objectives of the organization.
Organizations have missions or goals, i.e., raison d’ être. They must operate
within a set of external and internal constraints such as governing rules and legal
regulations that may constrain the decisions and activities of the organization.
Organizations need resources, i.e., operating funds, people, and infrastructure.
Most of all, the organization must have a leader or a group of leaders to oversee
the entire design, operation, execution, and administration of the organization to
achieve its stated goals.
The first step in the design of an organization is the establishment of its goal
and mission. The process of design begins with the identification of problems the
organizations must overcome to achieve the established goal. The design of an
organization follows the same transformational process as any other design. The
goals and problems must be translated into functional requirements (FRs), which

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 447

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_16
448 N. P. Suh

is then followed by the transformation of FRs into design parameters (DPs) and
finally into process variables (PVs). One significant difference between
organizational design and other designs is that in organizational design, DPs
are people, whose future behavior and performance are not always predictable.
The design of the organization should be either uncoupled or decoupled for
the organization to be effective. When the structure of the organization is a
coupled design, the organization cannot function effectively. Then, the decisions
made in one part of the organization may conflict with decisions made in another
part of the organization, thus compromising the attainment of the overall goals.
Organizations with coupled FRs are resource-intensive with uncertain outcomes.
Organizations—their missions and goals as well as personnel—may evolve or
change over time. Many organizations are complex time-dependent systems with
functional periodicity. Thus, a periodic re-initialization of the system should be
required.
The role of the chief executive officer (CEO) is most important in any
organization. Many CEOs play two roles: both the role of a composer of music
and the conductor of a symphony that plays the music. A wise CEO creates an
uncoupled organization. The CEO then can delegate each one of the highest
level FRs to each one of the executives who report directly to the CEO. The role
of the CEO is to achieve corporate goals by modulating and combining the
outputs of all FRs. When the CEO designs a coupled system, the organization
cannot effectively achieve its goals. The organization that couples FRs will be
chaotic, undoing the actions of other parts of the organization.
In large organizations, many lower level FRs are the responsibility of lower
level DPs. The organization must be sure that all the decisions made throughout
the organization are consistent with the highest level goals. Before any decision
is made, one should ask “Is this decision good for our organization?” If the
answer is not affirmative, one should not implement the decision.
Complexity theory applies to organizational design. Time-dependent combi-
natorial complexity should be avoided, whereas time-dependent periodic
complexity should be built into renew the goals of the organization periodically.
Finally, this chapter presents the idea of S-factor, S-Gap, and Delta (d) to
illustrate why it is difficult to change the relative standing of organizations and
what the leader of the organization must do to put the organization on a different
trajectory to make the organization to be outstanding and distinguished.

16.1 Why Design an Organization?

Beginning in prehistoric days, humans have instinctively designed and imple-

mented organizations at all levels, probably to improve collective survivability1.
They probably used trial-and-error processes to act in unison and survive in a

1
This instinct for survivability seems to be possessed by all animals. The schools of deer that come
around our yard also have a hierarchy; the mother deer always protects her young breeds.
However, their organizational hierarchy seems to be shallow, being limited to one generation.
16 Design of Organizations 449

hostile environment and address everyday needs and common problems collec-
tively. Today, we have a variety of organizations to deal with diverse human needs
and societal goals, promote economic and technological advances, and ultimately
achieve national and global aspirations for a better quality of life.
Organizations are the basic units of society, nations, and humanity. Each one of
them has a purpose and objectives. They perform a variety of functions and achieve
specific missions and goals. The size of the organization may vary, from a few
people to millions of people. Within an ideally designed organization, people share
common values and depend on each other for collective well-being. Without
organizations, individuals may not be able to function, survive, and secure basic
needs.
The goal of organizations is to achieve a set of missions that must be addressed
to meet human, societal, national, and international needs. Different common laws
and regulations of a nation may govern each one of these organizations, which were
established for governance. Ultimately, the overarching legal basis for these
organizations is the constitution of a nation or a royal decree, which may impose the
ultimate limits of acceptable designs.
Many different kinds of organizations exist to serve a variety of different societal
goals. Well-known organizations are schools, industrial firms, banks, governments,
manufacturing organizations, religious organizations, military, charitable organi-
zations, the World Bank, and many others. As human knowledge expands and the
population increases, “super” organizations (e.g., the United Nations) have been
created to deal with a common interest and to assure the collective survival and
collaboration.
Many of the modern non-governmental organizations are chartered legal entities
(e.g., schools, companies, law firms, schools, hospitals, churches), each with a
mission or objectives. In the case of a university, the mission is to generate highly
educated people for various fields of human endeavor, who can advance organi-
zational or societal goals. For an industrial firm, the mission could be the manu-
facture of specific products by hiring people and gathering human and financial
resources and generate profit for those who have invested in the firm.
Both the design and operation of all of these organizations must be done well to
achieve their stated goals. From time to time, these organizations need to be
re-designed (or re-initialized), often within the same legal framework, to be more
productive or compatible with the changed external conditions and needs. To
achieve this goal, many organizations change their internal structure from time to
time to respond to the external changes through the re-initialization of the
organizations.
In many countries, regionalism, race, sex, and religion still affect or influence the
organizational structure as a means of protecting their diverse interests. In some
countries, the tribes used to be the basic unit of societal organizations, i.e., people
belong to tribes for security, jobs, marriage, the survival of posterity, and family
life. In some countries, hiring and promotion of people in an organization are not
done based on merits and potential contributions, but rather on some other criteria
450 N. P. Suh

such as “whom one knows” and “what region one is from.” In such a society, a real
democracy based on individual merits cannot survive for long.
Organizations need resources—human and financial—to sustain the organiza-
tions. Therefore, securing these resources is the most critical responsibility of the
chief executive and the management.
Industrial firms exist to generate wealth through their activities. There are two
kinds of industrial firms: those privately owned by individuals or a group of
individuals, and those publicly traded companies that derive their finance through
the issuance of company stocks in the public market. Public companies are con-
trolled by regulations and rules of the national and local governments. The board of
directors (BoD) governs the public companies, whose primary function is looking
after the interest of the shareholders of the company. The BoD appoints the chief
executive officer (CEO) of the company, who holds the executive power for
operation of the company. The CEO or the board nominates the board members,
and the shareholders typically elect the board members at their annual meeting. For
the CEO, the membership of the board is vital for two reasons, i.e., to get sound
advice and support for the CEO’s actions. Many CEO attempts to create a friendly
board that backs the CEO’s position and policies.
Relevant units of an organization typically include the following functions: chief
executive, finance, information, marketing, sales, operations, human
resource/personnel, physical facilities, and the board of directors. For non-profit
organizations, similar functions are performed but often under some other names.
The board of directors of a public company protects the interest of the public (i.e.,
shareholders) and also appoints the chief executive. The chief executive officer
(CEO) is responsible for all operational aspects of the company. Having an out-
standing CEO is a prerequisite for a successful public company. The CEO appoints
all key personnel of the organization. The fortune of a company often depends on
the ability of its CEO. There are many examples of renowned companies facing
bankruptcy because the board chose an inept CEO. Unfortunately, it takes a few
years to find out the actual ability of the CEO, sometimes too late to save the
company.
The CEO must identify the problem(s) that the organization must address or
solve. Depending on the selection of a specific nature of the problem(s) the orga-
nization must address, different FRs may be selected, and thus, the outcome can be
substantially different. It is the responsibility of the CEO to determine FRs.
Therefore, the selection of the CEO is the most important decision the board of
directors makes. The CEO is also responsible for the operation of the organization.
She/he must design the organization to achieve the goals of the company. The CEO
then must select the right, able, and ethical people (i.e., design parameters (DPs))
who can perform their functions (i.e., FRs) effectively.
University presidents are also CEOs, although their role differs from
profit-oriented organizations, and the issues and goals are different. However, many
issues that must be handled by university presidents are similar to corporate CEOs.
16 Design of Organizations 451

Design Story 16.1: Important Roles of CEO in a Major Corporation

In many major public industrial corporations, the chief executive officer (CEO) is
also either the president or the chairman of the company. The COE performs the
most critical role in determining the future of a company. The CEO designs the
organizational structure in consultation with the CEO’s staff, including COO (Chief
Operating Officer), vice presidents (VPs), CFO (chief financial officer), and legal
counsel. CEOs develop policies and strategies, as well as implement them. CEOs
can make or break the company.
The board of directors of a company determines the compensation package (i.e.,
salary, stock options, etc.) of CEOs and senior executives of companies based on
their performance. Sometimes, they gather data on the compensation of other
similar companies to justify their decisions to shareholders. Relative to the average
salary of employees in the company, many CEOs of American companies are well
compensated. In many high-tech companies in the United States, CEOs hold about
2% of the company stock as incentives for outstanding performance. If the com-
pany does well, the CEO of profitable companies can become quite wealthy.
Outstanding CEOs of companies deserve every penny they earn for their per-
formance. Under a strong and capable CEO, the shareholders of these companies
have seen their investment grow by leaps and bounds. In 2018, Amazon, Apple,
Boeing, Facebook, Google, Broadcom, and Microsoft were such companies.
Amazon is the largest and fastest growing e-commerce company in the United
States. Apple, Inc. is known as the dominant American company in portable
telephones worldwide and is one of the largest multi-technology, multi-national
companies in the world. Broadcom is the leading manufacturer of critical semi-
conductor devices that are used in making telecommunication, computation, and
other applications. The gross revenue of these high-tech companies exceeds that of
all the oil companies of the world combined, although they use less natural
resources than many others for their business. One of the remarkable things about
many of these high-technology companies is that the founder of the company is still
a dominant business leader of their industries.
Inept CEOs can ruin a reputable company. One CEO has ruled a major and
oldest public company in the United States with iron hands for 15 years. The
company used to be one of the original companies listed on the New York Stock
Exchange. Until early 2000, the company was one of the fast-growing
multi-technology companies in the world. Today, the company is a
much-diminished company with many questions about its future. Many of its
shareholders lost their investment.
Notwithstanding the poor performance, the high-level executives of the com-
pany were well-compensated relative to their performance and the compensation of
its employees. There may be many reasons for the demise of the company that
outsiders can only speculate. However, some of the contributing factors might have
been due to the poor design of the company strategy. From the AD perspective, the
lingering questions are did the CEO understand the problem(s) the company had to
452 N. P. Suh

address? Did the CEO define a correct set of FRs and select the right set of DPs?
Did the design of the company couple FRs?
Why did this company fail to perform under the leadership of this new CEO? It
is difficult to judge without access to the inner workings and the business strategy of
the company. However, it appears that he was a manager, not a leader with a vision.
He enjoyed the trappings of being the CEO of a large conglomerate but did not have
a clear understanding of his FRs and DPs. He made promises that he could not
deliver or would not deliver. The board of directors (BOD) is partially responsible
for letting the CEO manage without a clear strategy for a highly diversified global
company. It may be safe to assume that the FRs and DPs were not carefully
designed, and the company was an organization with coupled design. The BOD
should have replaced the CEO within five years rather than waiting for 15 years.
Oriental folklore says that “even mountains move in 10 years!”
Functional Periodicity, discussed in Chap. 12 on complexity, is highly
relevant in corporate management. Corporations should review their strategy
and execution periodically, and re-initialize and execute the plan well!
Design Story 16.2: Organizing a Start-Up Industrial Firm

Researchers at a university developed a new technology to measure the moisture

content in hygroscopic polymers such as nylon and PVC (polyvinyl chloride)
in situ. These plastics tend to adsorb moisture because these polymers are made of
polar molecules. When the moisture in the plastic pellets exceeds a certain mini-
mum level, the long-chained molecule of the polymer breaks down (i.e., molecular
cession) during processing in high temperatures in extruders and injection molding
machines, degrading the physical properties of the polymer. Therefore, these
polymers are typically dried before processing. It is desirable to measure the
moisture content in situ before processing the polymer pellets. Under the spon-
sorship of industrial firms, students developed a means of measuring the moisture
content of such polymers based on the measurement of the dielectric properties of
the polymers.
The industrial firms that sponsored the research wanted to use the technology.
The professor, who conceived the idea for the measurement technology, asked two
of his graduate students who just finished their masters’ degrees if they would be
interested in starting a new venture firm to commercialize the monitoring tech-
nology. They were enthusiastic about the idea of starting a new high-tech company.
The professor raised funds from his friends and relatives to commercialize the
technology to dispense with the time and effort required to raise funds from venture
capitalists.
Some of the investors and the professor hired a lawyer to incorporate a company,
which is relatively simple in the United States. They formed a board of directors
and appointed one of the newly graduated engineers as the president and CEO of
the company. The CEO hired several of his friends to develop the technology
further and commercialize it. Creating a new business was relatively easy. How-
ever, making the company profitable was not easy!
16 Design of Organizations 453

They thought that the most crucial part of the new company is technology. That
was naiveté or a significant mistake. They did not fully appreciate the fundamentals
of what it would take to commercialize a high-tech product successfully. Many
things, such as the importance of the reliability of the instrument in the production
environment, sales, marketing, manufacturing, and financing of a company, were
not fully appreciated. Because they did not have sufficient background in the
commercialization of new technology such as marketing and economical manu-
facture of the product, the company was not successful in becoming a viable
commercial venture after they exhausted the capital raised. They did not appreciate
how difficult it is to make reliable industrial equipment. They learned an expensive
lesson, and the investors lost money.
They learned a few lessons from the failure. The market size for the new product
was much smaller than they had assumed. Furthermore, the instrument had to be
calibrated for every material at the operating temperature. The organizational design
was also flawed because they did not correctly define the “problem.” They also
learned that it is not a good idea to have inexperienced people to take on the critical
task of making business decisions. They should seek out highly qualified people to
join the board of directors to benefit from their knowledge and experience.
The young engineers who operated the company learned a lot, perhaps more
than they would have learned in a business school, at a great expense to the
investors. The professor learned a lot, too, about the commercialization of new
technology and how not to start a new venture firm. The key lesson learned was that
a new venture firm must be designed correctly with the clear identification of FRs,
DPs, and PVs. Perhaps an equally important lesson was that experience in starting a
new business can eliminate apparent mistakes made by novices.
Design Story 16.3: Organizing an Industrial Firm—Story of Another Venture
Firm

A bright young graduate, who studied business management at a leading technical

university and inherited some money from his family, joined a successful industrial
firm. He was soon fired because he raised too many questions about the way the
company was operating as a young engineer/manager. Then he joined another
company and was fired there also. After he was fired the third time, he joined a new
start-up company established by a brilliant technologist who graduated from the
same university a few years earlier. The only product made by the company was the
testing equipment that measured the quality and functionality of semiconductor
chips that were just manufactured through complicated fabrication processes. He
became the second in command in charge of marketing the new product. The
company did very well, especially after the CEO, the technologist who founded the
company, left the company, leaving the helm of the company to him as the new
CEO. The company grew rapidly as the semiconductor industry expanded at an
unprecedented rate for the next three decades. They defined the problem well in the
early phase of the semiconductor industry by identifying the need to measure the
quality of each one of the semiconductor devices.
454 N. P. Suh

This new CEO knew the importance of the underlying technology and had
expertise in and appreciation of the importance of finance, marketing, and human
resources in newly launched high-technology companies. He was a brilliant
designer of organizations and a superb leader of his employees. He hired out-
standing technologists and treated them with respect. They continued to improve
and innovate their products to be able to deal with ever more complicated semi-
conductor chips being produced by companies such as IBM and Intel. His CEO
office was a small cubicle just like those of low-level engineers because he knew
that the most valuable asset of his high-tech company, i.e., the creative and pro-
ductive technologists, must be treated with respect and fairly. He used to say that
the most valuable assets of his company went home every day at 5 p.m. Everyone
in his company, regardless of their job title, traveled in the same class of the
airplane when they were on business trips.
In these high-technology companies, the chief financial officer (CFO) is the
second most important executive in the company, who oversees nearly all aspects
of managing high-technology companies, including sales, manufacturing, market-
ing, engineering, and legal. Successful companies always need more capital to
expand their business. They raised the capital by selling the stocks of their company
in the public market. The CEO and CFO of this company worked together well for
a few decades. They also had an active board made up of experienced business and
technology leaders.
He became super-rich and funded many other new start-up companies. He
became successful in his business because he was ethical and moral, in addition to
his expertise in marketing, financing, and high-technology. He commanded the
respect of those who worked with and for him in many different capacities. He
became a philanthropist who funded many programs at universities, hospitals, and
other worthy causes. People like him render hope and faith in continuing the
advancement of humanity for the better. His name was Alex d’Arbeloff.
Government organizations are designed to achieve the goals established by the laws
of the nation. For example, the U.S. Congress established the U.S. National Science
Foundation (NSF) by enacting the NSF Act of 1950. It specified its mission of the
agency (i.e., to promote the progress of science and engineering; to provide welfare,
health, and prosperity to people; and to secure national defense), a broad outline of
its organizational structure, including the appointment of the leaders of the orga-
nization, i.e., presidential appointees, (the director, deputy director, and a few
assistant directors). It also specified how the agency is to be financed and the chain
of command under the presidency of the United States (e.g., the NSF director
reports directly to the President of the United States). Its budget plan is approved by
the White House through the Office of Management and Budget (OMB) and
submitted to the U.S. Congress for final approval. Unlike other government
agencies, the NSF Act mandate that a board (i.e., the National Science Board) be
appointed by the President to oversee the activities of NSF.
16 Design of Organizations 455

Design Story 16.4: Design of Government Organizations2

A professor, who was just sworn in as a new presidential appointee of a govern-

ment agency of the United States, came into his office after waiting for 6 months for
confirmation of his appointment by the U.S. Senate after receiving the clearances of
the Federal Bureau of Investigation (FBI) and the Presidential Appointment Office
of the White House. It was an unexpected turn of events in his career since going
into government service was not part of his career plan! He took the job of serving
in the government in a critical position, although his family had to suffer financially.
When he entered his large office, he was impressed by its cleanliness with a new
carpet on the floor and a fresh coat of paint on the wall. It was apparent that the staff
members, who were extremely dedicated and competent, worked hard to make the
office as pleasant as possible for the new appointee. The bookshelves were empty,
and the desktop was polished. Then, he noticed a piece of paper on his desk with a
drawing similar to that shown below. The caption said, paraphrasing it “A political
appointee comes to Washington, thinking that he/she can make a big difference in
government. Then after two years, they return home not having achieved anything,
mired in Washington bureaucracy with is tail between the hind legs like a scared
dog.” The cartoon was amusing but did not deter this new political appointee from
his goal of changing how engineering research and education were funded to
strengthen U.S. engineering education and research in the twenty-first century
(Fig. 16.1).
He decided to re-design his directorate to make it more effective in fulfilling the
duties and responsibilities outlined in the NSF act of 1950, as amended. He began
this process of designing the new NSF Engineering Directorate by defining the
problems that must be overcome, translating the issues identified into FRs. A new
structure was created by specifying DPs (e.g., divisions). Division Directors
designed the programs within the divisions in consultation with the Engineer-
ing AD. The appointment of the AD by the President of the United States gave him
the authority to make these decisions.
NSF was created to achieve three tasks: to promote the progress of science and
engineering; to provide welfare, prosperity, and health to people; and to secure the
national defense. The newly introduced changes were to deal with these three tasks
more effectively and make the United States more competitive in engineering in the
twenty-first century. One result of this change was more support for younger
researchers and new fields of science and technology. Also, it was to promote
multidisciplinary research, in some cases with the support of and in collaboration
with industry. The new design of the NSF Engineering was well received by many,
including those in the U.S. Congress, the White House, universities (especially
young professors), and industry, but there was strong opposition, too, from those
who benefited under the old system. The new system was to support new and
innovative research rather than refining the well-established ideas and technologies.

2
This story is partially told in Sect. 3.6.1.
456 N. P. Suh

Fig. 16.1 A cartoon to make fun of presidential appointees running away from Washington after
having failed to achieve anything meaningful during their tenure. (Copyright 2013, Center for
Shelter Dogs, Cummings School of Veterinary Medicine at Tufts University. Illustrated by Lili
Chin. Reproduced with permission)

The hostility of those who opposed the changes made at the NSF Engineering
Directorate lingered. Their feeling was reasonable and understandable in the sense
that they had been champions of their fields, many of whom had made significant
contributions. However, the new direction and policy of the Engineering Direc-
torate, which was to support more emerging technological fields and younger
researchers who are entering the profession for the first time, resulted in the loss of
easy access to research grants to established professors. This loss of easy access to
government support changed the quality of life of these well-established professors,
who used to get the NSF support easily working on more or less similar problems
for decades. These changes instituted by this presidential appointed have had a
lasting positive impact on the nation’s research and education infrastructure,
although some senior professors criticized him at the time.
In the new design of the NSF Engineering Directorate, there were seven FRs the
new organization was designed to satisfy with seven DPs. Seven divisions were
proposed, increasing the number of divisions from four to seven to assure the
independence of FRs. Then, the chief financial officer (CFO) of NSF warned that
Congress would reject the new organizational structure, because there would be too
many divisions that had to be headed up by additional senior executives, increasing
the top ranks of civil servants. As a result of the CFO’s advice, he compromised by
combining two proposed divisions into one. Therefore, two of the seven FRs had to
be administered by one division director. The result was not satisfactory because
one of the FRs of the two FRs in this division was not given the same priority as the
other FR by the division director. He tended to favor the FR he was most familiar
with and neglect the other FR. Therefore, after waiting for a year, this division was
split into two, so each division can handle only one FR and its lower level FRs.
The Congressional Committee that had the oversight on NSF did not have any
concerns about the number of divisions in the new organization. The new FRs of
NSF Engineering changed the landscape of engineering research and education of
many engineering schools in the United States. For the first time, NSF Engineering
supported research in micro-technology, design, and newly emerging fields of
technology. Also, NSF established the Engineering Research Centers (ERC) to
16 Design of Organizations 457

promote multidisciplinary research in collaboration with industry. Many out-

standing scholars from universities and industry joined the NSF Engineering
Directorate to make their share of contributions toward newly emerging fields. The
overall consensus was that the transformation resulted in the stronger engineering
research community in the country.
When he decided to go back to his university after almost four years of gov-
ernment service, the U.S. Congressional Committee with oversight on NSF com-
manded the NSF Engineering AD for the job well done. NSF awarded him the
Distinguished Service Award and a Gold Medal, notwithstanding the cartoon that
was on his desk on the first day of his arrival in his office at NSF.
Finally, it should be noted that such significant changes in a government
organization, programs, and policies are challenging. There are too many forces that
try to maintain the status quo at any given instant in the institution’s history. In
institutions in any country, there are always those who benefit from the existing
system, although it may no longer serve the interest of the people. One must be
devoid of personal agenda if one wishes to make such a significant change. In fact,
in the case of NSF Engineering Directorate, about 1,600 engineers of “Concerned
Engineers of America” sent a petition into the White House asking the President to
fire the NSF AD, although the appeal was not heeded. Some eminent professors
even “blackballed” him to prevent his election to a particular academy. Unless one
is willing to swallow some of these abuses, one should not assume the position. Also
holding the government job was extremely expensive for the professor with a large
family.
Notwithstanding the cartoon of the dog, many leave their government positions
after two years of service for financial reasons. However, the most important reason
to accept such an assignment is because of the unique opportunity to serve one’s
nation with a critical mission. For him, it was an honor to get a telephone call from
the White House.

16.2 How Should We Design an Organization?

The CEO of an organization has a great deal of latitude in designing his/her

organization. CEO typically decides on a set of FRs the organization must achieve
during her/his tenure as the CEO. The CEO gets the approval of the board of
directors for the plan. In the case of universities, the governing body is often called
the Board of Trustees, which plays a similar role as the board of directors of
companies.
To achieve the highest level FRs, the CEO must decide on a strategy of satis-
fying the FRs by coming up with a plan (i.e., DPs). The CEO has to find the right
person to be in charge of each set of FR/DP/PV. In general, it is not a good idea to
have one person to be in charge of more than one FR, since the person may make
“coupled” decisions, which forces the upper management to check double all the
decisions made by his/her subordinates, leading to micro-management. As we
458 N. P. Suh

decompose the design of the organization, the person in charge should design the
organization by decomposing the highest level FR and DP and appointing the right
people to be in charge of the next-level FRs and DPs. One of the most successful
CEOs, who have received the highest compensation based on his performance
among all CEOs in the United States, gives his key division managers only one
specific FR to achieve after he designs his company’s strategy.
Many firms and organizations have tried to improve their operations by trying
many different forms of organizational structure, including “matrix organization,”
“flat organization,” “hierarchical organization,” and others. However, the common
mistake many CEOs have made was creating the organizational structure first
before deciding on the goal of the organization and FRs. That is, the CEO must
determine the overall goals and select specific FRs. After the FRs are chosen, the
CEO can create an organizational structure and choose the right person to lead each
pair of FR/DP. Then, the CEO can monitor the progress made toward achieving the
overall goal of the company. The CEO should construct a master design matrix to
quickly identify decisions that can lead to the coupling of the highest FRs of the
company.
As the following example shows, a large organization designed based on
Axiomatic Design (AD) is much easier to operate and achieve its goals. It also
illustrates how difficult it is to change the existing design of an organization.

Design Story 16.5: Transformation of a Science and Technology University

Research universities are distinguished from other universities because they con-
duct significant research and have active graduate programs that emphasize research
and original contributions to their fields. There are many excellent research uni-
versities throughout the world, which have contributed to the development of
modern society and laid the foundation for human advancement. Some of the
leading research universities have generated societal leaders, scholars, and
researchers who have contributed to humanity through innovative ideas, scientific
discoveries, and technological advances. For example, their research has played a
significant role in the agricultural revolution, technological innovation, scientific
discoveries, and medical advances. In the twentieth and twenty-first centuries alone,
they played important roles in creating an electric power distribution system,
automobile revolution, electronics, digital technologies, computers, software rev-
olution, fission and fusion of atoms, scanning electron microscope, printing tech-
nologies, mobile telephones, and many others. Many of the older research
universities, e.g., Harvard, Cambridge, MIT, and Oxford, have evolved and
advanced over a long period. However, there are some outstanding young research
universities, which are less than 50 years old. Initially, many of these younger
institutions tried to emulate older research universities. They were partly successful,
but many of them are still not competitive relative to the best research universities
in the world. The main reason is that universities with more resources and repu-
tation advance faster than those without them and tend to receive more financial
support from both public and private. For younger universities to become a great
16 Design of Organizations 459

university, they must make their unique contributions to humanity with visionary
goals and programs, in addition to emulating the excellent features of leading older
universities.

Background:

An Science & Technology university was established in an Asian country about

50 years ago in 1972 when the nation was in the early stages of industrialization. It
was founded as a government-funded, but independent graduate school in science
and technology. It had its board of trustees as the governing entity, unlike national
universities that are run by the government. The goal was to generate human
resources in science and technology to support the ambitious industrialization plan
of the country.
To attract the best talents to this university, the government gave particular
support to this then-new university, its students, and faculty. For example, once
admitted, students were waived from military service. To attract best faculty
members from abroad in its early days, the university offered the professors salaries
comparable to those paid in the United States, about three times higher than the
current salaries of national universities. They built a new campus in 1986 away
from the highly congested capital city. At the same time, it also absorbed a nearby
undergraduate school of science and technology, making it a university with about
3,200 undergraduates and 6,000 graduate students. In 2006, there were about 400
professors. It became one of the most selective universities in the country. Many of
its students were top graduates of “science high schools,” an exclusive high school
for gifted students. Students were adequately supported, i.e., no tuition, free room,
and board. Professors were as well compensated as those at comparable leading
universities in the United States. It was a good university but was not rated as being
one of the best in the world. Its ranking was about 200th in the world in 2006.

Problem Statement in the Customer Domain:

The university elected a new president from overseas because the board of trustees
and the government were interested in improving its quality, global status, and
competitiveness. The customers, i.e., the government and the public, were not
satisfied with the rate of progress being made at the university. Their aspiration for
this university was to see it become one of the leading universities in the world.
What should this new president do?
His first task upon assuming the job was to define the “problem” the university is
facing before he could design a solution. Before assuming the position, he collected
information about the university by talking with young alumni/ae of the university,
asking for written inputs of the department heads, and talking to the board of
trustees. He also spoke to outside people such as journalists, government officials,
and politicians—all to identify and understand the problems faced by the university.
He accepted the invitation to be the president of this university because he realized
that this university had the potential of becoming a world-class university under the
right circumstance. The university had built-in advantages such as generous
460 N. P. Suh

government support, a special admissions policy that enabled it to attract the best
students, and public support for education.
As explained in Chaps. 1 and 2, the first thing we have to do in designing a
university is to define the problem that must be solved to transform this university
to achieve its goal of becoming one of the best universities in the world. The new
president, in consultation with members of its faculty and staff, surmised that the
following are the problems that must be addressed to move forward:

• The University has not progressed in recent decades because of the use of a
seniority system rather than having a merit-based system.
• Professors were not pursuing new intellectual challenges in their research and
education, because the number of publications rather than the impact made by
the researchers measured their performance. Many professors published papers
by letting bright graduate students research a topic the professor is already
familiar with rather than pursuing intellectually challenging new issues that are
important in the future development of science and technology.
• The senior professors who joined the faculty when they were young in the early
days of the University continued to dominate the university culture for decades,
depriving young faculty members do their challenging work and creating a crony
system.
• The University depended on the government for funding, personnel, and oper-
ation of the University. The government bureaucrats controlled the University
through its budgetary process. As a result, the median age of the faculty was
55 years, with almost no faculty members in the thirties and some in the forties.
• Because the students were fully supported and were waived from serving in the
military, many of the students had delayed their graduation to improve their
grade point average by retaking the same subject to improve the grades and
replace the poor grades. Thus, the number of undergraduate students was higher
than what the budget allowed by about 800, which strained the housing and
other operating expenses.
• The University did not raise any funds by soliciting private gifts, and therefore,
the University did not have any endowment.
• The infrastructure was not maintained since they were built in the 1980s. The
buildings did not have adequate heating and cooling system. Therefore, they
closed all research buildings at night during the winter season.
• Faculty members were appointed to key administrative positions for two years
with extra compensation. Under this rotational system, inexperienced faculty
members without managerial expertise often made significant mistakes.
• The number of female faculty, as well as non-native faculty members, was only
about 10%.
• Because teaching and all the administrative functions were performed in the
local language, its graduates were not prepared for jobs in major industrial firms,
which sell their products overseas and thus, conduct their business in the English
language.
16 Design of Organizations 461

The problem he identified was the “stagnation” of the university as if the forward
motion of the university was frozen in place. The cause for the stagnation was the
“intellectually lazy” culture that permeated throughout the university, i.e., many of
the faculty members and students were happy with the then prevailing status quo,
although they could achieve a lot more if they tried to distinguish themselves. They
also depended too much on government bureaucrats’ directives and did not attempt
to be the leaders in their professional fields. They were satisfied with the status quo.
The idea of becoming one of the leading universities in the world was beyond their
imagination or dream, i.e., they thought that it could not be done! The university
had gradually deteriorated over the previous 30 years since they moved into a brand
new campus.
The most challenging part of becoming a professor or student at this university
was “getting into the university.” Once one becomes a professor at this university,
the job was secure for life. Similarly, a student admitted to the university graduates
from the university with a degree. Students also get free room and board for at least
four years. Once one gets into this university, there were no incentives and reasons
to perform. The prestige of the university guaranteed a good life for a lifetime for
most graduates and faculty of the university. Professors’ salaries increased each
year regardless of the performance, too.
When the university was established some 50 years ago, it started with full of
enthusiasm, because the government conferred special status to the university to
nurture highly educated engineers and scientists. They worked hard to make the
university to be one of the best in the country, but the notion that they could make
their university to be the world’s leading university was dismissed as being unre-
alistic. The university administration and the faculty blamed the government for all
their shortcomings.
Having identified the “problem,” i.e., the “stagnation,” the following FRs were
established to solve the problem and achieve the goal of making it one of the best
universities in the world:
FR1 = Increase the faculty size from 400–700 to be competitive with the best universities
in the world;
FR2 = Hire only the best scholars regardless of their field of specialty;
FR3 = Work on important problems of the twenty-first century;
FR4 = Increase government support for research;
FR5 = Change the language of instruction to English;
FR6 = Increase diversity;
FR7 = Increase the overhead rate to recover the indirect cost of sponsored research;
FR8 = Increase externally funded research;
FR9 = Improve the decision-making process;
FR10 = Raise gifts from donors;
FR11 = Require students to complete their degree requirements on time;
FR12 = Recruit and admit the best students;
FR13 = Strengthen interdisciplinary research;
FR14 = Improve and expand physical facilities;
FR15 = Replace the seniority-based compensation policy to a merit-based system.
462 N. P. Suh

Many of these FRs are what a typical research university would aim to achieve.
However, there are a few that deserve further explanation regarding their issues:
FR1: A few universities would dare to commit to increasing the faculty size by
75% without a financial guarantee either from the state resources or a significant
increase in the endowment. However, without such an increase in the faculty size, it
was clear that this university will not be able to be as good as some of the world’s
leading universities. The government bureaucrats were extraordinarily skeptical and
in some ways annoyed by such a daring plan when they had no intention of
increasing its support of this university even by 10%.
FR2: Most universities have traditionally allocated a fixed number of faculty
positions to different departments; for example, 50 faculty for physics, 20 for
biology, 50 to mechanical engineering, and ten in civil engineering, etc. In many
cases, the number of faculty members was determined based on student enrollment
and tradition. However, in some ways, they are entirely arbitrary. Students do not
prefer specific departments because there are more professors in the department.
The number of students in a given department tends to fluctuate over the years.
Sometimes, the professors from other departments may have to help out if the
enrollment suddenly jumps, but it rarely happens. At this university, it was decided
that every department head should hire as many outstanding professors as they can
find, provided that they have exceptional intellectual accomplishments and are
incredibly gifted. Initially, there were many skeptics of the idea that each depart-
ment could hire as many new professors as they could find, provided that they are
truly outstanding people.
FR3: Many professors are under pressure to publish. Some end up publishing
papers that are neither good nor important to increase the number of publications.
FR3 states that professors should work on significant problems rather than on
frivolous or trivial issues.
FR4: The government allocates about 5% of the national budget for research
support. Some of this fund is available to researchers on a competitive basis.
Winning proposals can receive this grant under the competitive government
program.
FR5: In science and technology, English is used globally. For example, seminal
papers are written in English for a broader readership. Many international journals
are in English. Many of the new scientific and technological keywords are in
English. Also, this country depends on foreign trade for the economy. Young
people are learning English at a very young age. For these reasons, the language of
instruction should be in English.
FR6: Male professors with limited international representation dominated this
university. The number of women students in science and technology is too small,
being about 25%. The brainpower of both genders should be utilized to be active in
the twenty-first-century economy.
FR7: The prevailing overhead rate was too low to recover the overhead cost of
sponsored research. Thus it should be raised.
16 Design of Organizations 463

FR8: Competing for publically available research funds has many excellent
benefits. It improves the quality of research and also focuses the attention of
researchers on critical research issues.
FR9: Departments could not make any difficult decisions because the senior
members of the faculty controlled the department by developing consensus, and the
central administration controlled the real power by over-riding department
decisions.
FR10: The university depended primarily on government largess. The new
administration undertook significant fundraising activities.
FR11: The old system allowed students to stay at the university indefinitely,
receiving free room and board without paying any tuition. This system enabled
students to stay longer taking the same subject over and over again until they got a
better grade, replacing the lower grade received in the past. Professors also liked
this old system, since they were not paying for the student support, and therefore,
they tended to keep their Ph.D. students longer, since they were more useful to the
professor.
FR12: This university attracted the best students in the country because of its
prestige, free education, military service deferment, better educational environment,
and future career development. Many students came from “science high schools”
that were extremely selective. Some bright and able students from regular high
schools could not get admission to this university, except a few, for a variety of
different reasons, often financial. Therefore, 15% of the admitted students to the
university were selected from these high schools based on the recommendation of
the school principal, who was asked to nominate one student. These students did as
well as those who graduated from science high schools.
To satisfy the FRs, we may choose the following DPs:
DP1 = Increased faculty size to be competitive with MIT. Hiring criterion: Department
heads to search for as many high-quality faculty members as they can find irrespective of
their field of interest and department size;
DP2 = Quality as the sole selection criteria for new faculty;
DP3 = Emphasis on energy, environment, water, and sustainability;
DP4 = Active solicitation of government support for bold new research projects;
DP5 = Faculty support for the change of the language of instruction;
DP6 = Active recruitment of woman students and faculty;
DP7 = New rate with faculty and government;
DP8 = Active proposal submission;
DP9 = Department head system (rather than chairman system);
DP10 = Active fundraising;
DP11 = Time-Limit on financial support (i.e., 4 years only for undergraduate students, and
6 years only for Ph.D. students);
DP12 = Special admissions policy to supplement the regular admissions policy.

These DPs were chosen to satisfy the FRs. Some of these FRs and DPs had to be
decomposed, which will not be presented here.
There was strong resistance to some of the DPs chosen, especially on DP5 (the
language of instruction), DP9 (the department head system), and DP11 (new tuition
system). There were many reasons for the resistance, some quite understandable.
464 N. P. Suh

Rationale for Making Design Decisions (i.e., Experience and Knowledge

Matters!)

The person in charge of an organization cannot choose specific DPs without a plan
of implementation in mind. For example, how would anyone dare to increase the
faculty size from 400 to 700 without any guarantee of funding from the government
funding agencies for the contemplated increase in the faculty size? Since there were
no available funds and no outstanding commitment for future funding, the leader of
the organization must have had a plan for implementation when one chooses a
particular DP. Otherwise, one may quickly become a laughing stock by creating
false hope and talking about a project without the ability for actual implementation.
Based on his experience in his previous jobs, he had contingency plans in mind for
implementing the plan. His rough thinking and strategy were as follows:
At a typical research-intensive university, the fraction of the salary fund that
goes toward paying professors is approximately 16% of the total salary pool of the
university. After a rough calculation, he affirmed that this ratio was also valid at this
university, which he just joined as the new president. Although he let every
department hire as many professors as they can find, it would be difficult for them to
meet the highest criterion established for the quality of faculty. He also knew that it
would be difficult to find more than 40 (i.e., *10% of the existing faculty) highly
qualified faculty candidates. If the government does not provide the additional
funding needed to increase the faculty by 10%, he identified how he could generate
1.6% of the budget required to pay for the newly hired professors from the existing
resources by eliminating unproductive activities.
He also realized that the university could raise a lot more research funds if the
professors were more active in generating creative ideas for their research. Many
professors had been attempting to increase the number of publications by refining
what they had done before. Furthermore, at this university, professors did not have
to seek external research support because the graduate students were paid from the
central university budget. Therefore, the professors did not have to raise research
funds to support their research students. This system created a culture where pro-
fessors relied mostly on the university to support their research rather than seeking
external funding. Thus, their competitiveness in research had not been tested. Also,
a few professors, who brought in outside research funds, complained about the
overhead charge, although it was only 10%, which would be the envy of all pro-
fessors at research universities in the United States, where the overhead rate would
be more than 50%. Such a low overhead rate meant that the university was, de
facto, subsidizing the externally funded research from internal resources, because
the real cost of supporting research was about 50%! Therefore, professors were
encouraged to seek research funds from external sources to pay part of their
research cost. The overhead rate was increased from 10% to 25% to reflect the
actual research cost!
The president of the university also initiated activities to raise new external
support from two external sources: (1) government funding for special education
and research projects, (2) the generation of large gifts from private donors. Through
16 Design of Organizations 465

active solicitation, both of these financial plans succeeded in a country that did not
have the philanthropic culture. The university increased its faculty size to 650 from
400 by hiring 350 new faculty members in five years and built 17 new buildings to
accommodate the increased faculty size and expanded research activities as well as
new student dormitories to retire old dilapidated facilities.
The faculty hiring was done solely based on the qualification of the individual
faculty candidate regardless of their field of research interest. By eliminating the
departmental quotas for the faculty size, each department had to concentrate on the
intellectual quality of the faculty candidate under the new faculty hiring policy.

Result:

The hiring of the best faculty, regardless of their disciplinary interest, strengthened its
research and teaching activities, promoting interdisciplinary research. The central
administration approved hiring most professors who appeared to be outstanding
irrespective of their professional specialization. As a consequence, some departments
grew faster than those that could not recruit exceptional people. The new policy
accelerated the initiation of new research programs, especially interdisciplinary
research, which has resulted in significant advances in science and technology. The
faculty members were encouraged to conduct bold research or technology innova-
tion, i.e., two ends of the research spectrum. Many pursued innovative research at
both ends of the research spectrum. The quality of the work replaced the number of
publications as the primary criterion for promotion and salary raise.
These highly capable professors brought in significant research funding, which
also generated additional overhead income, strengthening the financial resources of
the university. Teaching in English made both the students and faculty not only more
proficient in collaborating with their colleagues in many countries but also made
them more active and competitive, enriching both education and scholarly activities
of the university. Physical facilities were expanded drastically to accommodate the
increased number of professors and enhanced teaching and student activities by
constructing 14 new buildings. A new teaching format was introduced, which
eliminated formal lectures and introduced a discussion format for learning, aided by
making formal lectures available through the Internet. In five years, the sponsored
research volume and the overhead income increased by a factor of 2.7.
The number of publications and patents obtained increased, although qualitative
measures replaced the quantitative measures of performance. Professors became
active globally than ever before. For instance, the robotics team of this university
won in an international robotic contest that carried $2 million prizes, competing
against the leading universities in the world. The university also introduced the
world’s first wireless transportation technology (i.e., On-Line Electric Vehicle) and
“mobile harbor” technology. The electric vehicle is commercially operating in five
cities.
The international standing of this university rapidly advanced, becoming #6
among the top ten most innovative universities in the world, the only non-U.S.
based university to be in the top ten.
466 N. P. Suh

It should be noted that such a rapid transformation of a university could not be

done without the personal sacrifice of the leadership of the university and strong
outside support of the government and the citizen of the country. Significant
financial donations came in response to the new direction established for the uni-
versity, which helped in building 16 new buildings and increasing the faculty size
from about 400–630 by hiring 300 new faculty members based on their quality
regardless of their specialization. The “faculty union” opposed many of the changes
from the beginning, mainly because the old seniority-based system was replaced by
the merit-based system, including compensation, i.e., younger professors could be
paid more than senior professors, depending on their performance as a researcher,
teacher, and scholar. They also opposed the change of the language of instruction to
English from their native language. However, the majority of the faculty, especially
the younger professors, supported the move. The support of the newspapers and the
average citizen was instrumental in implementing the difficult changes. They helped
the changes because they were well aware of the abuse of power by the senior
faculty, which prevented the advancement of the university despite the generous
resource allocation by the government.

Design Story 16.6: Designing a Competitive Industrial Firm through Merger

and Consolidation

In the early twentieth Century, Henry Ford revolutionized the automobile industry.
He did not invent the automobile but knew the “problem,” i.e., the high cost of
manufacturing automobiles due to the practice of assembling the entire car by a
team of craftsmen. He introduced the concept of the moving assembly line to
reduce the manufacturing cost of automobiles. In this new Ford production system,
the car was assembled on a moving production line by a large team of workers, each
of whom performed only a limited but highly specialized assembly task as the
vehicle came to his/her station. When the car reached the end of the assembly line,
the fully assembled vehicle was driven off the production line. Today, the pro-
duction rate is about one car a minute.
Thanks to the Henry Ford idea for mass production, a large number of small
industrial firms were created to manufacture machines and automotive components
in Detroit, Michigan, U.S.A. The Detroit area was the center of technology inno-
vation and wealth generation of the United States during the first half of the
twentieth century. Many automobile companies were sprung up around the state of
Michigan and nearby states. Besides Ford Motor Company, many automotive
companies were established and competed. Ford dominated the automotive busi-
ness for a long time. Many smaller companies did similar things to produce
automotive components and automobiles. Many of them were not competitive.
Finally, many of the automobile companies merged to form the General Motors
Corporation (GM), which dominated the automobile business in the United States
for decades until the mid-twentieth century. Alfred P. Sloan was one of the founders
of General Motors Corporation.
16 Design of Organizations 467

A monopoly is harmful to the overall economy because it may lower the overall
efficiency of capital utilization and innovation. At the same time, the scattering of
resources by a large number of sub-critical enterprises that cannot withstand sig-
nificant risks involved in technology innovation cannot be sustained. The optimum
consolidation of a given industry yields substantial benefits through the better
service of customers and enhanced innovation and innovation activities. GM
became the wealthiest and most profitable automotive company by the
mid-twentieth century. It became more prominent than the Ford Motor Company.
In the twenty-first century, one of the dominant industries is the semiconductor
industry. They manufacture semiconductor devices that are used in computers,
cameras, and all other instruments and equipment that used information technology.
There were many competing companies in semiconductor fabrication and systems
that use semiconductors. A brilliant and capable CEO of a semiconductor device
manufacturing company designed a strategy for creating a dominant semiconductor
company by acquiring or merging with companies in similar businesses (a la the old
GM model) and established a dominant high-technology company in the semi-
conductor device industry. He eliminated redundant R&D and consolidated product
lines. He also concentrated on critical technology products to become the most
competitive firm in that space. His company has grown by leaps and bounds by
being the most productive and efficient firm in the semiconductor industry.
His organization tends to be simple, although it has become a hundred
billion-dollar company. Each product line is managed separately and indepen-
dently, all reporting to the CEO directly without going through a COO (Chief
Operating Officer). This organizational structure enabled the CEO to monitor each
highest FR, which may be possible only when the CEO knows the technology
associated with the FRs. In this type of organization, R&D groups are closely tied
to product divisions. His goal is to produce the best products in each business area.
One of the criticisms has been that he does not invest in long-term R&D, which
cannot be substantiated unless one is more specific about the goals of long-term
research that have been neglected. In a matter of about ten years, the company has
become one of the leading semiconductor device manufacturers in the world, a
major success story of the twenty-first century.

16.3 Operation of Organizations

The raison d’être of any organization is to achieve its missions by solving a set of
challenging problems. For example, universities exist to educate the future leaders
of society and to pursue scholarships for enlightening and improving the quality of
life of people through the generation of knowledge. Nearly all universities do not
have financial gains as their goal. Industrial firms exist to produce goods and
services to improve the quality of life of people and to generate wealth for a higher
standard of living, especially those who take risks investing their wealth for a better
future. Most industrial firms must make financial gains to attract investment in their
468 N. P. Suh

firms. Organizations must be administered well and efficiently to achieve these

societal and institutional goals,
To accomplish the mission of the organization, they must be organized as it was
discussed in the preceding section. Many organizations have a tree-like structure
with the CEO at the top, leading and orchestrating the operation of the organization.
Most of the legal entities have a board of directors or trustees to be sure that the
organization does indeed abide by laws and protect the investors or stakeholders.
CEOs of industrial organizations tend to have nearly absolute power since they
drive the organization toward achieving their corporate goals. At most universities,
the power is distributed between the administration and the faculty. Individual
members of the faculty contribute to the financial and operational well-being of
universities.
As a consequence, some universities deliberate even simple issues endlessly
because of the broad participation in the decision-making process by both faculty,
students, and the administration, whereas timely and efficient decision-making
ability is essential for the well-being of an institution. There are pros and cons to
these power structures. For a quick response to changing the external environment,
the concentration of power at the top can be justified, since consensus-based
decision-making can take much longer, and some decisions must be made on a
timely basis. Furthermore, there is no guarantee that consensus-based choices are
superior to the decisions made by professional administrators with appropriate prior
experience and specific knowledge that applies to a particular situation.
One of the interesting things about the operation of an organization is that one’s
perspective changes drastically once one assumes the role of CEO. There may be
many reasons for it. First, being a CEO is like being both the composer and the
conductor of a symphony. The CEO must know where the organization should be
headed and gather all the resources to generate the orchestrated music. CEO must
make sure that the entire organization is moving in the direction to achieve its
mission, which requires knowing the overall make-up of the organization and
placing the right person in the right job. CEOs may not be the best violinist of the
symphony orchestra but must know what kinds of musical sound should come out
of the violin section of the symphony. The CEO has to identify and assign the best
person for the critical position of the company. However, when a wrong person has
the power of the CEO, he/she can run the company into the ground. There are
well-known cases providing examples of how a CEO has ruined a great company
because of the inept top management.

16.4 Design and Decision-Making in an Organization

Like other designs, once an organization is designed and finalized, the organization
must perform as intended to achieve the FRs of the organization. To achieve this
organizational goal, people at all levels of the organization must strive to make the
right decisions consistent with the overall organizational goals. However, it is
16 Design of Organizations 469

impossible to guarantee that everyone in the organization would make the right
decisions. Therefore, one of the challenges in designing and implementing orga-
nization goals is how to ensure that people in the organization make the right
decisions.
Thus, the critical question is “How should decisions be made at all levels?” If the
CEO has to make all the decisions within an organization, the organization cannot
function and move forward. Everyone in the organization has to make decisions
that belong to the person’s organizational purview. Furthermore, every decision
cannot be monitored by someone else because there is not sufficient bandwidth and
resources to do so. At the same time, we must make sure that everyone in the
decision-making chain of command makes the right decisions.
One criterion for sound decision-making by everyone in the organization is “Is
this decision right for my organization?” In the case of the university: “Is this
decision good for the university?” In the case of people in a company, “is this
decision right for my company?” In the case of a citizen, the question to ask is “Is
this decision good for our country?” The decisions made cannot be alike, but we
know that everyone has done their best to make the right decision. Having asked
this question, they can later evaluate the quality of their decision and attempt to
make improvements if their decision was a misguided one.
Finally, everyone at all levels must ask this simple question. “Is the decision
made good for my organization and humanity?”

16.5 Typical Organizational Design

Question 16.1:
Which industrial/commercial firm between Amazon and Walmart has a better business
model and a stronger corporate structure?

The main point of this book is that systems, including organizations, must be
designed based on FRs that were created based on the problem(s) the organization
must solve. If an organization is designed based on DPs (rather than FRs), it may
become a personality-cult based organization rather than a goal-oriented company.
By choosing FRs first and then identifying the right DPs (e.g., the most qualified
persons) who can satisfy the FRs best, the organization can achieve the goals of the
organization. If DPs are chosen first, one may end up having to teach FRs to
stubborn old-timers. Sometimes, the conventional wisdom states that it is difficult to
teach new tricks to old dogs—“old” meaning in terms of attitude rather than
physical. In this transformational process of defining FRs to solve a problem
identified, and then selecting DPs, followed by resource allocation (i.e., PVs), may
create an organization that is efficient in terms of making the right decisions and
implementing them. For example, Walmart may not be able to become Amazon,
whereas Amazon can easily compete with companies like Walmart, because to an
470 N. P. Suh

outsider, it appears that Amazon is a very functional organization, whereas Walmart

is a mix of FR-based and DP-based organizational structure.
Re-construction of the FRs of an existing business is challenging. It is most
likely that the reconstructed FRs may not be what they had in mind. However, to an
outsider, the highest FRs of Amazon appears to be the following:
FR1 = Sell to customers electronically without the physical display of merchandise;
FR2 = Quick delivery of goods;
FR3 = Guarantee return policy;
FR4 = Centralize warehousing of some goods;
FR5 = Centralize the financial and administrative operation using computers;
FR6 = Integrate merchandise-delivery services with manufacturers;
FR7 = Electronic display of merchandise;
FR8 = Contract with manufacturers for guaranteed return policy;
FR9 = Collect payment as soon as a customer orders the merchandise.

Assuming that these nine FRs are indeed the correct ones, what should be the
corresponding DPs? This question is given as a homework problem #4 presented at
the end of this chapter.

16.6 When Should an Organization Be Re-designed?

Changing any organization involving a large number of people is not a trivial task.
Typically, the rationale for not changing, or resisting changes, at some leading
departments and universities are “we must be doing well to be so highly ranked.
Why change?” They often forget that their reputation was based on what they did in
the past. However, organizations are like living beings that must do well in the
future in a continuously changing environment. The FRs change because society
and business continually evolve, requiring re-examination or transformation of
missions, goals, internal policies, the organizational structure, allocation of
resources, and make-up of the personnel.
Many different symptoms or indicators suggest the need for reorganization.
When it is not addressing the problem(s) identified or when it does not have an
organizational structure that can implement FRs, sometimes organizations may not
be performing their intended functions or have low productivity. In some cases, the
organizations are merely repeating its traditional role that worked well in the past,
although the FRs had changed and are no longer valid because the society sur-
rounding the university is no longer the same as when the policies were enacted.
Implementing a new organizational design to replace the one that has existed for
many years or a few decades is not a trivial matter. Depending on the organization,
one may create committees to study the proposed new organization to make sure
that enough people buy in the new organizational structure. It is particularly difficult
16 Design of Organizations 471

at universities where the faculty and students are significant stakeholders as the
following example shows.

Design Story 16.7: Re-organizing the Departmental Structure at a Leading

Research University

Most engineering schools have chemical engineering departments. In the U.S. they
were established in the 1920s in response to the need for petrochemical engineers as
the automotive industry expanded, requiring more petroleum for automobiles. Later
as the nuclear power industry developed beginning in the 1950s, some of the
chemical engineering departments merged with nuclear engineering. Later some
universities made nuclear engineering as an independent department. The latest
trend has been to transform the chemical engineering department into the bio-
chemical engineering department. Now some schools made bio-engineering into a
separate department. All these permutations of the chemical engineering department
are somewhat meaningless unless the FRs are identified and stated.
At a leading university, there once was a department of food technology. The
department had significant problems of not attracting undergraduate students to
major in their discipline. Even their graduate program had a problem in enticing
outstanding graduate students to their programs. People consume vast amounts of
food, but the discipline of food technology was not the most challenging or
attractive to prospective students. Finally, they changed their department name to
“Department of Applied Biology.” However, the mission of the department was
ambiguous because the university had a separate department of biology. In other
words, they could not state their FRs succinctly. Having had years of unsatisfactory
performance of the department, the administration of the university decided to
eliminate the Department of Applied Biology. Then, some of the faculty members
of the universities start condemning the university administration, arguing that the
administration took administrative action without due consultation with the faculty.
The university delayed making this tough decision for several decades, probably
because some of the administrators did not deal with the vocal group of faculty.
Indeed the provost of the university who made the decision has been hounded for
decades. When another university was considering him for the presidency of their
university, his detractors campaigned to deny him the opportunity. Sometimes,
administrators at universities do not make the decision they need to make, for this
reason, leading to bloating administrative structure and the number of committees,
increasing the cost of running a university. At many universities, the number of
students and faculty has not changed, but their cost and administrators increased the
significantly untenable situation.
Since they did not wish to dismiss tenured faculty, the university provided three
different options: join the Department of Biology (if invited), or the Department of
Chemical Engineering (if invited). Professors who were not welcomed by either
biology or chemical engineering departments belonged to the Provost office and
continue their research work without any teaching duties. This reorganization was a
suitable arrangement because it strengthened both the biology department and the
472 N. P. Suh

chemical engineering department. A few who could not join these two departments
continued to research the umbrella of the Provost Office.
Under an ideal scenario, each academic department should transform itself
periodically based on the changes that have occurred in their field or to respond to
socio-economic changes that have occurred. It takes a leader who is willing to face
fair criticisms and still do what is right. The design is the required first step in
improving an organization, but the subsequent difficulties that accompany changes
are an integral part of the transformation of institutions.

Design Story 16.8: Periodic Re-initialization of Organization at a Medium-

Sized Producer of Consumer Goods

Starting from the previously outlined AD based complexity theory, a long-term

study performed in a medium-sized industrial company investigated the effects of
economic periodicity as a trigger for a regular organizational reset on the agility and
performance of the corporate system.
The company was founded in the early 80 s and literally started in a garage. For
almost two decades, the founder himself managed the entire value chain with only a
few employees: from product design to procurement, materials management and
production to sales and marketing. However, the person-centered approach of the
company structure set a strong growth limit. Finally, due to growing market
demand and the associated need to overcome organizational constraints, a new
organization was designed and implemented in 2000 that allowed the company to
take a major step forward in growth in just a few years. However, a market-driven
explosion of product variants quickly pushed the organizational system to its limits.
In the first 5-year interval, a gradual deterioration in organizational efficiency was
observed, which finally led to quality problems and efficiency losses until 2005.
The decision was made to carry out a further reorganization, this time with a focus
on assembly, materials management, and product development. Assembly was
restructured and a new hierarchical level of assembly team leaders was introduced.
Purchasing was separated from Materials Management and the Materials Man-
agement position was assigned to a new young manager. In product development, a
new profile of an innovation project manager was introduced. Continuous perfor-
mance monitoring showed that the organization had again lost performance
between 2006 and 2011. This time, the focus of the organizational changes had to
be on sales and marketing. Due to further growth, the organization of the company
was completely revised again in 2017. This time it was done to consider the
development of a new strategy and its impact on management processes.
The long-term study showed that the company had “intuitively” established an
organizational “rhythm” (i.e., periodicity) to overcome time-dependent complexity.
In order to continue with the good experiences made in the past, now a regular
5-year reorganization rhythm was introduced in order to prevent future organiza-
tional performance losses.
16 Design of Organizations 473

The previous example in Design Story 16.8 shows that an organization needs to
change periodically: a maximum performance can be achieved only if an organi-
zational structure matches the rate of change in its environment. The organizational
structure and the underlying design principles are, therefore, key factors for a
company’s successful and sustainable development within a turbulent environment.
The logic behind can be explained with the life cycle model of the Systems
Engineering concept (Matt 2011), which roughly divides the life span of a distinct
artificial system into design, realization, utilization, and disposal stages. For orga-
nizational design, this model might be adapted.
The first stage is the system’s design initialization. It follows the general prin-
ciples of AD: identify customer attributes, derive FRs, and finally assign suitable
DPs so that the final result of the organizational design ideally satisfies the two
design axioms. T0 represents the start of the very first design of a completely new
system. Within an ideally very short time frame (t0-t01), the system is set up for the
first time, and the introduction of the new organization starts at t1. However, as we
know from experience, any organizational model “deteriorates” after a certain
period due to internal (e.g., fluctuation of personnel, new strategy, new products)
and/or external changes (e.g., market changes, environmental requirements, new
competitors) and thus needs to be updated. It is necessary to find a viable way to
make necessary adjustments to internal and external events on the one hand, but on
the other hand not to have to re-design the organization with every single change
event fundamentally. For this time-dependent need for change in an organizational
model the AD-based complexity theory delivers a helpful approach. The com-
plexity of any dynamic system is determined by the uncertainty in achieving the
system’s FRs and is caused by two factors: by a time-independent poor design that
causes a system-inherent low efficiency (system design), and by a time-dependent
reduction in system performance owing to system deterioration or to market or
technology changes (system dynamics).
There are two types of time-dependent complexity: the so-called periodic
complexity exists only in a finite time period, resulting from a limited number of
probable combinations. These probable combinations may be partially predicted
based on existing experiences with the system or with a very systematic research of
possible failure sources. The second type of time-dependent complexity is called
combinatorial complexity. It increases as a function of time proportionally to the
time-dependent increasing number of possible combinations of the system’s FRs.
Leaving the pre-defined flexibility tolerance of a system design, it may lead to a
chaotic state or even to a system failure. The critical issue with combinatorial
complexity is that it is completely unpredictable.
To control combinatorial complexity, a functional periodicity must be intro-
duced. First, a set of FRs that repeats cyclically must be identified. Among these,
those FRs and their related DPs must be identified that may be the subject of a
combinatorial process. To introduce functional periodicity, the selected set of FRs
must be re-initialized at a defined (periodically turning) point in time t2.
474 N. P. Suh

16.7 Institutional Development: The S-Curve, S-Gap,

and Vector Delta (d)3

Many organizations aspire to be one of the best in their fields. However, history
shows that only a few surpass the leading organizations in terms of the institutional
reputation, accomplishment, and development. Most stay where they are over many
decades relative to other institutions in the same field. Sometimes, it appears that the
leading organizations do not have to try very hard to remain as the leading insti-
tutions. There may be many different reasons for this comparative standing. First,
human and financial resources tend to go to the leading institutions rather than to
those that are in dire need. There may be many reasons for this situation. Is it
possible to change this situation? It is the question every leader is striving to answer
with varied conclusions. In this section, one way that has successfully transformed
three major institutions is discussed.
The strategy for transforming an institution for the better was designed based on
AD. It was implemented at three major institutions—a government agency and two
top universities. These institutions were significantly improved through transfor-
mational changes, thanks to the re-design. The strategy used in these three insti-
tutions is outlined in this section. The central idea is encapsulated in the S-Curve,
S-Gap, and Vector Delta (D).
If we define the “S-Factor” as the integrated total measure of the quality of a
university, the development of the institution over as a function of time may be
depicted as shown in Fig. 16.2. This figure assumes a linear upward change of both
A and B, which would be the case when each institution works hard to improve
productivity, quality of education, and increase its financial resources. These curves
are a simplified depiction of the progress made, although the improvement of an
institution may be highly non-linear.
The development of an institution appears to be similar to the change in the
wealth of two individuals. As shown in Fig. 16.3, one with much more wealth than
the other. T. Piketty, a French economist, demonstrated that the rate of capital return
in developed countries is persistently higher than the rate of economic growth of
poorer nations and that this will cause increasing wealth inequality in the future.
Based on the historical data in France, he showed that wealth produces more wealth
faster than the wealth that can be generated through productivity increases and hard
work when the rate of return on capital is higher than the rate of economic growth
over the long term. Consequently, rich people (or rich nations) tend to get richer
faster because of the wealth that they already possess, i.e., wealth breeds more
wealth than just harder work or more productivity increase. This finding is
schematically shown in Fig. 16.3. Similar phenomena may be present in the
development of research universities (Picketty 2014).

3
Based on the paper: Nam P. Suh, Presented at the Univer-Cities conference, Newcastle, Australia,
November 14, 2016
16 Design of Organizations 475

Fig. 16.2 The development of two universities is shown as a function of time. It shows that
university B is developing faster than university A, but both universities are improving linearly as
a function of time. The gap between these two universities is denoted as the S-Gap. This change of
S-Factor may not be a realistic depiction of what happens

Fig. 16.3 The Picketty curve of wealth growth. The curve on the top shows that the wealthier
person gets richer at a much faster rate than the person who tries to accumulate wealth by working
harder

A similar phenomenon seems to occur with institutions such as universities. The

leading universities in all of their chosen fields have a much higher probability of
growing even stronger, because of the intellectual, human, and financial resources
that they already have. If we designate the aggregation of all the elements that make
a university strong, the “S-Factor,” the growth may be depicted as shown in
Fig. 16.4. The figure is a composite strength of a university, represented as the
S-Factor, as a function of time. The top curve is for a stronger university, and the
lower curve represents a less well-known university. The stronger university tends
to attract better faculty, students, and significant financial gifts. Thus, the gap
between the two, the “S-Gap,” grows as a function of time. The S-Factor is a
composite measure made of many elements that make a university great, such as the
476 N. P. Suh

Fig. 16.4 The S-Factor of a university is a composite index of what constitutes the reputation of a
university. The lower curve represents a university with a lower S-Factor, which develops more or
less linearly with time through hard work, etc. The S-Gap between the wealthier institution with a
higher reputation grows by a combination of linear growth and the compound rate due to the
leading status. The S-Gap between these two universities continues to increase unless some drastic
actions are taken

quality of faculty, students, and staff; the size of the endowment and financial
resources; its current reputation; past academic and scholarly achievements; and
prospects. Stronger universities with more “intellectual and financial assets” will
grow faster than universities with a low S- factor. Therefore, the gap between them,
the “S-Gap” shown in Fig. 16.4 will grow larger with time. For this reason, stronger
universities attract more resources and people, which accelerate their growth.
The much faster growth (or advance) of the university with a higher S-Factor is
due to many factors. The principle ones may be the following:

• more competitive students and faculty go to universities with better reputations

and resources;
• resources (financial gifts, funds, etc.) tend to concentrate at richer universities or
better known universities;
• faster growth of the S-Factor (e.g., outstanding faculty, students, facilities,
reputation, etc.) at stronger universities may be due to the existing advantages;
• safety factor—for students and faculty members considering their choice of a
university, joining a successful enterprise may be deemed safer;
• quality of life may be better at wealthier institutions;
• more opportunities may exist at a university with a higher S-Factor.

Universities that have a low S-Factor at a given instant in time (indicated by the
lower curve) should not stay on their current trajectory if their long-term goal is to
become one of the best universities in the world. They must transition to a higher
16 Design of Organizations 477

Fig. 16.5 For the university

represented by the lower
curve to be better than the
university represented by the
upper curve, it has to adopt
non-linear changes indicated
by vector d

trajectory as indicated by taking actions to go from the lower curve to the upper
curve employing Vector d as shown in Fig. 16.5, by adopting significant changes.
The changes may take many different forms depending on specific institutional
conditions. For example, the necessary actions may include the following: adding a
large number of outstanding faculty members, attracting the most competitive
students, generating significant financial resources, making major scientific dis-
coveries, and innovating new technologies.
Case Studies at Two Different Universities
University A: This university, University A, had been rated as one of the top three uni-
versities in the United States for decades. Among the more than 20 (twenty) academic
departments, this department had been rated to be the best in the country ever since various
organizations ranked universities. Because the department was so far ahead of other
similar departments in the United States, the department personnel, especially the pro-
fessors, were very proud of their achievements. In some ways, they became complacent.
However, young professors in the department were concerned about the stagnant nature of
their department. It had not changed much during their days in the department as students
and later as faculty members. They became restless.
When an opportunity came along to look for a new department head, they advocated the
idea of finding who can take the department to new heights. They recommended one of the
senior professors, who has transformed some other institutions as well as creating new
research paradigms in the past, which had become a national model throughout the
country. After thinking about two options he was presented with at the same time, he
agreed to stay on, although compensation, etc., were less attractive than other offers. One
big reason was that his family loved where they had been living for many years, and he
loved the institution where he also attended as a student.
Among all its peer departments in the country and the world, the department was on a
steeper slope, making other universities to catch up, similar to the wealth gap between the
rich and poor. The well-known professors in the department were so satisfied and proud of
478 N. P. Suh

their achievements, and they did not see any reason to change the department or their
research and teaching activities. However, the new department head did not agree with
their views on the future direction of their department.
The department had become the leading department of its kind in the world by excelling in
more traditional engineering science and technology. The professors in the department
were leading scholars in fields that were fundamental in developing the industrial base in
the first half of the twentieth century. Because they were so good, they wanted to replicate
what they had built by hiring their former students who were also good in the same
disciplines. Most other leading departments of its kind in the world followed their footsteps
in programmatic undertakings and their disciplinary research. This happy situation has
persisted for decades. Yet young professors knew that the world outside was changing
rapidly, passing their department by, although their senior professors could not be happier
with the status quo.
The new department head decided to transform the department to be ready for the twenty-
first century. One of the significant issues was the disciplinary mix of the faculty. Then, the
existing faculty members were the best in classical disciplines that were important in the
first half of the twentieth century. The best strategy was to bring in new faculty members
who had backgrounds in fields that could change the direction of the academic field of the
department. The department had been active in disciplines of classical physics, but now it
needed to build expertise in the fields that would be important in the twenty-first century,
such as communications, biology, optics, nanotechnology, quantum mechanics, and the
like. The department started hiring new professors whose doctoral degrees were outside of
the traditional disciplines of the department. These changes constituted the big delta shown
in Fig. 16.5.
These significant changes in the direction of the department did not sit well with the famous
senior professors of the department. They said “we must have done something right to
create this department of international repute. Why change?” They try to disrupt the new
direction of the department by forcing the department head to step down. But the
department head prevailed. Now two decades of the transformation, the department has
become a model for other departments of the same university as well as departments at
other universities.
University B: Another well-known university hired a new president because the internal
candidates chosen from their current faculty have not satisfied the expectation of the public.
The taxpayers had supported this university by granting them special privileges and
generous financial support since its founding about 40 years ago. Right after its estab-
lishment, the university had developed rapidly. However, its development soon reached a
plateau, because privileges given to the university by taxpayers had the opposite conse-
quences. The faculty members were researching familiar topics they been doing research in
for the sake of publishing more papers. The life was too comfortable to search for chal-
lenging issues, with full funding of their graduate students and selection of highly qualified
undergraduate students under special provisions given to the university by the government
as a means of generating human resources with advanced education.
Because of the several privileges given to this unique institution, it gradually lost its
competitiveness. The professors had repeated similar research over and over to publish
more papers that were used as a metric for academic achievements, and teaching was
neglected. They hired faculty members they were comfortable with rather than searching
for the best. All of this nurtured the seniority-based culture rather than merit-based
competitive culture. They were also content with the budget the government provided
yearly with inflation-adjusted increases. No attempt was made to undertake innovative
research as an institution.
16 Design of Organizations 479

This university needed the vector d to become a leading research university by coming out
of the comfortable nest they had built for themselves. The new president of the university
had to institute several significant changes, which were opposed by the senior faculty. He
recruited and hired new faculty members in massive numbers purely based on their
intellectual and scholarly quality. The specific expertise and contributions of their research
were given the highest priority in selecting new faculty members rather than their field of
study. He abolished the headcounts for each department; a merit-based system replaced the
seniority-based system for compensation and promotion; and innovative research was
emphasized by eliminating the policy that emphasized the number of papers published, etc.
New faculty members were hired to double the faculty size by raising external financial
support. New faculty members were given two years of research support to allow them to
explore new research topics rather than merely continuing what they had done before.
Facilities and equipment were modernized and expanded by building new physical
buildings. In short order, the global ranking of this university. At one point, this university
was recognized as the most improved university in the world.

16.8 Concluding Remarks

Organizations constitute basic units of society, a nation, and a group of nations.

They are designed to achieve specific goals and missions. Unlike other design tasks,
organizational design often deals with or affects the immediate well-being of people
or organizations or nations and, ultimately, the world. Therefore, it must be dealt
with thoughtfully and carefully. Many societal and organizational problems are
caused by and the people in and around them. Consequently, we must design
organizations right, following a rational and wise implementation of the goals of the
organization.
The design of organizations must be done rationally by identifying the problem
that must be solved through the creation of a viable set of FRs that are consistent
with the goals. The designers should go through the design process objectively and
rationally to come up with the best sets of DPs and PVs. DPs are typical organi-
zational entities (such as universities, divisions of a company, government agen-
cies), and PVs are human and financial resources needed to enable DPs to fulfill the
FRs.
Some designers, engineers, and technologists are less interested in non-technical
societal issues. However, it should not be the case. Ultimately, most engineers and
technologists function in organizational settings, and as they assume more
responsibility in their organization, people and organizational issues become
increasingly more important. They must face the people-related problems to be
good technologists as well as outstanding leaders. Everyone ultimately influences
the decisions being made by various groups in companies, universities, and gov-
ernments. More often than commonly acknowledged, engineers and scientists
neglect personal and humane aspects of organizational life at their peril.
Organizations perform certain functions relevant to their fields of specialization.
Almost all organizations are subject to functional periodicity, which requires
re-initialization from time to time to make them viable in fulfilling their missions.
480 N. P. Suh

This re-initialization is hard to do in organizations because organizations are made

up of people since, invariably, some of them would oppose any changes. The most
challenging organizations to change are the organizations that provide guarantees to
those in the organizations, because changes may disturb a set comfortable life of
those in the organization. However, periodic re-initialization will benefit everyone
associated with organizations.
Problems
1. Is a matrix organization better than a straight hierarchical organization? From
the AD point of view, is this matrix organization better than a vertical line
organization, which is responsible for certain products, including design,
development, and manufacture? How would you design the organization if you
were the VP in charge of the R&D group? Is the matrix organization a coupled
design?
2. A major automotive company just appointed a new VP for research and
development. He received his Master of Business Administration (MBA) from
one of the leading business schools. Before enrolling for his MBA, he worked as
an engineer in another large corporation after receiving his bachelor’s degree in
engineering. As the first order of business, he decided to re-organize his R&D
group, which consisted of 450 people in a matrix structure. The vertical line of
the matrix organization was to be the line organization that consisted of several
divisions headed by a Project Director (PD). The horizontal line of the orga-
nization was for professional expertise led by an Engineering Director (ED), i.e.,
ED for hydraulics, air-conditioning, materials, etc., who oversees the continuing
development and advancement in these fields. The Project Director could draw
in experts from various engineering groups as needed with the concurrence of
the EDs. Engineers would typically be in the group headed by an ED, but the
PD may hire them for a specific project. In this type of matrix organization,
people with similar skills are pooled for work assignments, reporting to two
bosses, i.e., the vertical line PD and the horizontal ED. Is this matrix organi-
zation a coupled design? How would you make this organization function as an
uncoupled or decoupled system?
3. Is the design of the university presented in the main text (i.e., Design Story 16.7:
Re-organizing the Departmental Structure at a Leading Research University) an
uncoupled or decoupled design? Construct a design matrix for the department
strategic plan.
4. The following nine FRs were presented as a possible set of FRs that Amazon
might be satisfying:

FR1 = Sell to customers electronically without the physical display of

merchandise;
FR2 = Quick delivery of goods;
FR3 = Guarantee return policy;
FR4 = Centralize warehousing of some goods;
FR5 = Centralize the financial and administrative operation using computers;
16 Design of Organizations 481

FR6 = Integrate merchandise-delivery operations with manufacturers;

FR7 = Electronic display of merchandise;
FR8 = Contract with manufacturers for guaranteed return policy;
FR9 = Collect payment as soon as a customer orders the merchandise.
Develop nine DPs for these FRs that will satisfy the Independence Axiom.
Construct the design matrix for your design. Then, for your design, establish the
design range and the system range for your design. Show how the information
content can be minimized.
5. Managing a hospital is a complicated task with so many patients with a variety
of illnesses and managing outpatients as well as those admitted to hospital for
extended in-house care. It has to manage its finances to be sure that it can sustain
its operations without having to incur financial losses. Suppose that you are just
appointed as the president of a mid-size general hospital with 300 beds and daily
visits to the outpatient flow of 500 patients. What would be the FRs that you
have to consider and design a management system for the hospital?
6. One of the leading engineering departments of the world has many giants (i.e.,
well-known professors) who have contributed to the development of
twentieth-century technology. Consequently, the department has enjoyed a
stellar reputation worldwide because of these leading scholars. They continued
to hire their former students and new professors from other universities, who
have done similar work. The problem faced by the new department head is that
these well-known professors want to continue to expand their kind of research
that was important some 30–50 years ago. However, the world is moving into
new technologies based on telecommunications, digital technologies, biotech-
nologies, quantum computing, and quantum computers. Help the new depart-
ment head in choosing a new set of FRs. How would you overcome the
resistance of the influential professors in the department?

Reference
Piketty T (2014) Capital in the twenty-first century. Belknap Press of Harvard University Press,
Cambridge, Massachusetts, USA, translated by Arthur Goldhammer from French

Further Reading

Belanger DO (1997) Enabling American innovation: engineering and the National Science
Foundation (History of Technology). Purdue University Press
Matt DT (2007) Reducing the structural complexity of growing organizational systems by means
of axiomatic designed networks of core competence cells. J Manuf Syst 26(3–4):178–187
Matt DT (2011) Application of axiomatic design principles to control complexity dynamics in a
mixed-model assembly system: a case analysis. Int J Prod Res 50(7):1850–1861
Suh NP (1990) The principles of design. Oxford University Press
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
482 N. P. Suh

Suh NP (2005) Complexity. Oxford University Press

Suh NP (2016) On strategy for developing an innovative university: S-factor, s-gap, and vector
delta. In: Univer-Cities Conference, University of New Castle, Australia. https://www.
newcastle.edu.au/community-and-alumni/-alumni-archived/univer-cities-conference-2016/pro-
gram, session 3: strategic and evolving implications of the single key node Eco-system of
globalizing Universities
Züst R, Schregenberger JW (2003) System Engineering: A Methodology for Designing
Sustainable Solutions in the Field of Engineering and Management. Verlag Eco Performance,
Zurich. https://www.swissinstitute.ch/publikationen/buecherjournal.html, distributed by Swiss
Institute for Systems Engineering
Application of Axiomatic Design
for the Design of Flexible and Agile 17
Manufacturing Systems

Dominik T. Matt and Erwin Rauch

Abstract
In the previous chapters, we learned the basics of Axiomatic Design (AD) as
Design theory. We learned that users very often express their wishes as customer
needs (CNs). These wishes must then be examined by the designer in order to
define functional requirements (FRs). Based on the FRs and the two axioms in
AD, design solutions are finally derived, so-called Design Parameters (DPs).
AD can be applied in many different areas such as product development,
healthcare, software development but also manufacturing system design.
Manufacturing systems are complex entities which can be broken down into
their functional elements or requirements by the application of AD and for which
suitable DPs can be found.
AD has been used for many years for the design of manufacturing systems.
The chapter also shows that the number of uses of AD in manufacturing has
increased continuously. While in the past, there were many topics related to
specific manufacturing processes and systems as well as lean manufacturing, the
focus has changed somewhat in recent decades. Today AD is often used in
manufacturing to derive design characteristics for the intelligent and sustainable
manufacturing of tomorrow.
In this chapter, several case studies from the field of production management
and manufacturing system design are shown. By means of these practical case
studies, the students will understand in which cases AD can be applied in
industrial practice.

D. T. Matt (&)
E. Rauch

Faculty of Science and Technology, Free University of Bozen-Bolzano, Universotätsplatz 1,
Bolzano, Italy
e-mail: [email protected]
D. T. Matt
Fraunhofer Research Italia s.c.a.r.l., Via A.-Volta 13a, 39100 Bolzano, Italy

© Springer Nature Switzerland AG 2021 483

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_17
484 D. T. Matt and E. Rauch

The first case study deals with the derivation of design guidelines for flexible
as well as changeable manufacturing systems. The second case study describes
the digitization in shopfloor management by developing a tool for a systematic
acquisition, analysis, and evaluation of production data. In the third case study,
AD is used to design an assembly station in which robots and humans work
together and collaborate safely. The fourth case study shows how AD can also
be used to derive design guidelines for Industry 4.0 learning factories. Finally,
the fifth and final case study deals with the re-design of a demonstrator for
cyber-physical production systems.
At the end of all case studies there are short exercises, which students should
work on. The exercises encourage students to reflect on what they have learnt, to
think on it independently and to apply AD in practical case studies.

17.1 Introduction

Axiomatic Design (AD) is applied not only in mechanical engineering, but also in
many other applications. Through the systematic approach and the consideration of
Independence Axiom and Information Axiom, even highly complex projects can be
mastered reducing the complexity in the design task. In addition to product design,
system design, software design and many other fields, AD is, therefore, also used in
the design of manufacturing systems.
The design of manufacturing systems has a major influence on the sustainability
of manufacturing companies. On the one hand, an efficiently designed manufac-
turing system generates a higher profit for the company, which in terms of eco-
nomic sustainability contributes to securing the existence and growth of a company
and thus also to the prosperity of its employees. With regard to ecological sus-
tainability, the reduction of energy consumption or the reduction of emissions
during production can make a contribution to the preservation of our environment
and our earth. Last but not least, a manufacturing system should also focus on the
employee, and therefore contribute to social sustainability. This is reflected in the
design of the work systems and in the organization of the working environment.
Manufacturing systems can also be subdivided fundamentally into three different
design areas: (1) system, (2) organization, and (3) people. A functioning manufac-
turing system can only be achieved through a symbiotic design of all these three areas.
Machines must be correctly selected and arranged and an organizational set of rules
must ensure that people work in a structured manner in this production environment.
Engineers in the industrial sector have been working on the design of manu-
facturing systems for many decades. While in the past 20 to 30 years mainly
organizational innovations with the concept of Lean Management have been
introduced, the trend today, with the introduction of Industry 4.0 and Internet of
Things, rather goes in the direction of technological innovations. This inevitably
leads to the fact that the design of manufacturing systems has become a very
17 Application of Axiomatic Design for the Design of Flexible … 485

complex task. Designers of manufacturing systems should, therefore, make use of

methods and instruments to design manufacturing systems as efficiently as possible.
AD is one such tool that allows manufacturing system designers to solve problems
in a structured way, break them down into their functions and derive the right
solutions and implementation measures.
Therefore, this chapter first gives an overview of the extent to which AD has
been applied in manufacturing system design to date. In the following sections,
various case studies are presented on how AD has been applied in practice.

Design Story 17.1: Changing Needs in the Design of Manufacturing Systems

A young graduate engineer started his career in a production company as a project

manager in the department of process optimization. He was very much looking
forward to his new job in a large company and was already excited about his area of
responsibility. One of the first projects he was working on was the planning and
design of a new production line for a new product line.
Although the task was very exciting, it was also very challenging. He knew
neither the current processes nor the previous products, which made it difficult for
him to get into the subject. He, therefore, tried to learn from proven methods and
first looked at the old production facilities. They produced standardized product
components for the automotive industry in large quantities. The young graduate
thus began to use an old plan of a production line as a basis and adapted it to the
new product line. The head of the department observed him in his work and let him
work out an initial concept. The young engineer proudly presented his concept draft
to the round and showed the advantages resulting from the application of
best-practice design guidelines of the old production lines. At a certain point, the
head of the department stopped the young engineer and asked him if he had
considered what requirements the new product line would have in comparison to
the previous products? The young engineer was not prepared to answer the question
and was only able to make inadequate explanations. The manager replied that his
approach would be very good in itself, but that the new product line contains a large
number of variants that would continue to increase over the next few years. In
addition, the product life cycle of the new product would probably be much shorter
than that of the old products. The production system would, therefore, have to be
much more flexible and changeable than the previous rigid mono-product lines.
So, the young engineer was tempted to reapply what had been tried and tested
without pursuing new and creative approaches, even though he did not have the
“operational blindness” that many long-standing engineers have. Besides, he has
not dealt enough with the special needs arising from the new product design and the
resulting FRs for the manufacturing system. The young engineer learned this and
revised his concept, developing a much more flexible and adaptable design for the
production line, which helped the company to save much money for otherwise
necessary investments in dedicated machinery for new variants.
The story of the young engineer was told to emphasize the importance of ini-
tiating a design with a clear and unbiased view keeping the focus on the needs and
FRs and using this information then for deriving the right design solutions for the
design of the manufacturing system.
486 D. T. Matt and E. Rauch

17.2 Use of Axiomatic Design in Manufacturing System

Design

AD is a systematic approach for design by the top-down decomposition of “What

we want to achieve,” to “How we can satisfy the requirements.” The theory of AD
is applicable to many different kinds of complex systems. A manufacturing system
can be defined as a dynamic and complex system, because it is subject to temporal
variation and must be reconfigurable and adaptable. In such cases AD shows a
suitable and helpful method to reduce complexity in the manufacturing systems
design (Matt and Rauch 2011). As learned in the previous chapters, AD is based on
four domains to transform the so-called customer needs or customer attributes (CA)
into FRs, DPs, and process variables (PVs) (Suh 2001).
Through its top-down approach, AD is a very systematic and structured design
methodology. Starting from a main goal, a hierarchically structured catalog of FRs
with proposed design solutions is developed. By breaking down (decomposition) of
the top goals and design proposals specific DPs can be identified at a lower
operational level. This is helpful for manufacturing system designers, which start
very often with an overall objective and need to develop the manufacturing system
accordingly in all its details on an operational Shopfloor level.
The number of studies using AD principles is gradually increasing as AD’s
superiorities create important advantages for decision-makers in solving
multi-criteria decision-making problems (Kulak et al. 2010). Investigating the
scientific literature regarding works on AD in manufacturing using the keywords
“Axiomatic Design” and “manufacturing system” there can be identified a trend
toward an increasing use of AD in manufacturing. In the following a brief overview
is given to summarize the most important use of AD in manufacturing system
design over the last 20 years and to give students the opportunity to deepen their
knowledge by looking up relevant literature:
Cochran developed an approach for the design of manufacturing systems, which is based
on the principles of AD (Cochran et al. 2001). Cochran's methodology “Manufacturing
System Design De-composition” (MSDD) visualizes the derivative FR–DP tree in a very
clear manner and is easy to understand. AlGeddawy and ElMaraghy (2009) describe AD
as a very suitable and frequently used method to derive the target system as well as the
requirements and evaluate the interactions of the identified requirements in a systematic
way. Bergmann applies the MSDD-methodology and thus the AD approach for the
derivation of requirements for a sustainability-oriented holistic manufacturing system
(Bergmann 2010). The work of Bergman proves once again, that the application of the AD
methodology is suitable for a systematic and structured derivation of requirements and
DPs. Authors like Vinodh and Aravindraj (2012) apply AD for the development of lean
manufacturing systems. Durmusoglu and Satoglu (2010), Matt et al. (2016) and Rauch
et al. (2019b) use AD for the development of design guidelines to make manufacturing
systems more flexible and agile. Cochran et al. (2016) extended later his Manufacturing
System Design Decomposition approach to implement manufacturing systems that are
sustainable. Puik et al. (2017) developed an assessment of reconfiguration schemes for
reconfigurable manufacturing systems based on resources and lead time. The trend toward
Industry 4.0 and Internet of Things had also an effect on AD. Farid (2017) used AD to
design intelligent manufacturing systems and Cochran et al. (2017) to model human–
17 Application of Axiomatic Design for the Design of Flexible … 487

machine interaction in manufacturing cells. Also Delaram and Fatahi Valilai (2018) used
AD to develop an architecture for an intelligent use of computer-integrated manufacturing
in modern production systems.

In most of this research works the Independence Axiom is mainly applied and discussed by
application of the decomposition of FRs and DPs. Contrary to this there are not so many
research works dealing and investigating the Information Axiom in manufacturing systems
design. This could encourage practitioners and scientists to discuss in the future also the
quality of alternative solutions by the use of the Information Axiom. At the domain level, the
decomposition between FR and DP are treated in most of the works. The discussion about
CAs and PVs played up to now only a subordinate role. Due to the fact that CAs are
important to define the right first-level FRs and DPs they should be considered more in
future works. In the literature, AD is mainly used as a method for practical applications
and case studies of manufacturing system design. The analysis of previous and actual
specific topics of AD applications shows further, that it is increasingly used also for new
challenges in manufacturing like sustainability in manufacturing, smart and cloud manu-
facturing or agile/changeable manufacturing systems (Rauch et al. 2016).

In the following, 5 case studies show how AD can be applied in different design
tasks in manufacturing.

17.3 Case Study 1: Design of Flexible and Changeable

Manufacturing Systems

In the first case study (see also Holzner, 2015), we want to design a flexible and
changeable manufacturing system for a small industrial manufacturer. The firm in
our case study is a small-sized company in the North of Italy with 25 persons
employed, which started in 1995 as a small crafts enterprise processing solid surface
material for bathroom furnishings, kitchens, and modern interior design. Over time,
the firm focused its activity on the production of exclusive bathroom furnishings in
solid surfaces. The company produces different types of furnishings for bathrooms:
(a) washbasins, (b) shower trays, and c) bathtubs. The production of bathtubs is a
highly specialized production process, while the production of shower trays and
washbasins—even if they are entirely different products—need similar manufac-
turing and assembly steps. Today shower trays and washbasins are produced on
different assembly workstations and go from milling/gluing to a next workstation for
grinding. The aim of the application of AD in this case study was to develop a new
concept for a more flexible and changeable manufacturing and assembly system for
different types and dimensions of shower trays and washbasins.
The main difference between flexibility and changeability is that flexibility in
manufacturing only permits a system change in a specific corridor. Changeability,
however, describes the responsiveness over the existing flexibility corridor and
usually requires a longer time for reaction (Zäh et al. 2005). In this sense, flexibility
describes the ability of production to change a manufacturing system very quickly,
with little effort, and therefore with low costs. By flexibility, it is possible, within a
defined flexibility corridor, to adjust the manufacturing system. Further flexible
488 D. T. Matt and E. Rauch

manufacturing systems allow the adaptation of the manufacturing systems for the
production also of new products, but only if those are very similar and of the same
product family. We distinguish in literature different types of flexibility:

1. Variant flexibility: Ability of manufacturing / assembling multiple versions of a

product.
2. Volume flexibility: Ability of adaptation of production systems to fluctuating
sales volumes.
3. Internal flexibility: Ability to change the system without modifications (example
change of internal numerical code program).
4. Staff flexibility: Ability to work with a variable number of employees and
different worker skills.

Changeability, however, is the ability to switch from one product family to

another and making the appropriate changes in the production capacity of a com-
pany or manufacturing system. A change can have a significant impact on the
production and logistics systems. It also impacts on the equipment structure as well
as on the organizational or operational structure. Such a change requires a longer
lead time for planning and takes place relatively quickly. Wiendahl et al. (2007)
define changeability as characteristics to accomplish early and foresighted adjust-
ments of the factory's structures and processes on all levels to change impulses
economically. To reach changeability in companies and manufacturing systems,
five enablers (see Fig. 17.1) can be found in the literature.
Especially for small- and medium-sized enterprises (SMEs), costs and flexibility
in a production system are very important issues because the products are generally
produced in small batches. In addition to a high percentage of manual production,
these kinds of enterprises use mainly universal machines to guarantee certain
flexibility. SMEs that have a high degree of automation and manufacture in large
batches have to address their production more flexible and changeable to react
quickly to market changes and consumers’ preferences. SMEs with a robust and
highly flexible manufacturing system have usually a greater market share, a better
financial condition, and a better sustainable technology.
As learned in the previous chapters, the AD-based approach starts with the
identification of customer needs. In our case study, the company was asked about
future challenges and changes in their business environment. Principal CAs for the
future manufacturing system were defined as follows:

CA1 = Handle an increased variety of individual products (Increasing variety and

Individualization);
CA2 = Being competitive in price and costs (price competition in the market);
CA3 = Handle increasing quality requirements (increasing quality requirements);
CA4 = Deliver products in shortest time (increasing demand on delivery).
17 Application of Axiomatic Design for the Design of Flexible … 489

Fig. 17.1 Enablers of changeability in manufacturing systems. (Reproduced with permission

from (Spena et al. 2016) (Reprinted from Procedia CIRP, 41, Pasquale Russo Spena, Philipp
Holzner, Erwin Rauch, Renato Vidoni, Dominik T. Matt, “Requirements for the Design of
Flexibleand Changeable Manufacturing and Assembly Systems: A SME-survey”, 207–212,
Copyright (2016), with permission from Elsevier.). Figure based on (Wiendahl et al. 2007)
(Reprinted from CIRP Annals — Manufacturing Technology, 56, H.-P. Wien-dahl, H.A.
ElMaraghy, P. Nyhuis, M.F. Zäh, H.-H. Wiendahl, N. Duffie, and M.Brieke, “Changeable
manufacturing-classification, design and operation”, 783–809, Copyright (2007), with permission
from Elsevier).)

In the next step, these CAs were translated into FRs and DPs for manufacturing
system design. The identified CAs were translated into further first-level FRs
showing the technical and practical requests for manufacturing system design.

FR1 = Increase flexibility and changeability;

FR2 = Produce at lowest costs;
FR3 = Improve quality;
FR4 = Reduce lead time.

Corresponding DPs to meet these FRs were defined as follows:

DP1 = Flexible and changeable manufacturing/assembly system;

DP2 = Low-cost manufacturing systems;
DP3 = Zero defects and TQM in production;
DP4 = Pull principle and “0” WIP.
490 D. T. Matt and E. Rauch

The design matrix on the first hierarchical level shows the relationship of the
identified solutions (DPs) on the derived FRs:
8 9 2 38 9
>
> FR1 >
> X 0 0 0 > > DP1 >
>
< = 6 < =
FR2 0 X 0 077 DP2
¼6
40 5 ð17:1Þ
>
> FR3 >
> 0 X 0 > > DP3 >
>
: ; : ;
FR4 0 0 0 X DP4

The design matrix shows an uncoupled design. This means that FRs are dis-
tinguishable from each other. This case study focuses on the design of flexible and
changeable SME manufacturing systems. Therefore, FR1 and DP1 are now further
decomposed in additional AD levels.
Then, the decomposition process on the next hierarchy level continues with
mapping and “Zig-Zagging”. FR1 can be deduced into further two general FRs:

FR1.1 = Increase flexibility of manufacturing system;

FR1.2 = Increase changeability of manufacturing system;
DP1.1 = Flexible manufacturing systems guidelines;
DP1.2 = Changeable manufacturing systems guidelines.

The design matrix shows a decoupled matrix. Changeable manufacturing sys-

tems are usually flexible at the same time, while a flexible manufacturing system
does not have to be changeable.
    
FR1:1 X X DP1:1
¼ ð17:2Þ
FR1:2 0 X DP1:2

DP1.1 and DP1.2 are very general and abstract design solutions. Therefore, they
need to be further decomposed in a next level to break down DP1.1 and DP1.2 into
more tangible proposals for solutions according to the different types of flexibility in
production:

FR1.1.1 = Produce different variants;

FR1.1.2 = Employ worker on different workstations;
FR1.1.3 = Increase or reduce quantity based on demand;
FR1.1.4 = Minimize time for changeover between different variants;
DP1.1.1 = Fixation positions for shower tray variants (variant flexibility);
DP1.1.2 = Qualified personnel who know all variants and can work at flexible
times (staff flexibility);
DP1.1.3 = (De)Activation of assembly tables depending on demand (volume
flexibility);
DP1.1.4 = Memorized fixation positions changing with barcode scanning (in-
ternal flexibility).
17 Application of Axiomatic Design for the Design of Flexible … 491

The revised design matrix of FR–DP shows a decoupled matrix.

8 9 2 38 9
>
> FR1:1:1 >
> X 0 0 0 > > DP1:1:5 >
>
< = 6 < =
FR1:1:2 0 X 0 077 DP1:1:2
¼6
40 5 ð17:3Þ
>
> FR1:1:3 >
> 0 X 0 > > DP1:1:3 >
>
: ; : ;
FR1:1:4 X X 0 X DP1:1:4

After the decomposition of FR1.1 also FR1.2 needs to be further decomposed:

FR1.2.1 = Produce different products and product families;

FR1.2.2 = Extend the manufacturing system stepwise;
FR1.2.3 = Provide modular functional plug and produce units for working;
FR1.2.4 = Move machinery easily when changing product;
FR1.2.5 = Standardize connection of the manufacturing system;
DP1.2.1 = Universal assembly tables for shower trays and washbasins
(universality);
DP1.2.2 = Scalable manufacturing layout (scalability);
DP1.2.3 = Selection of necessary tools on the table (modularity);
DP1.2.4 = Mobile arrangement of tools (mobility);
DP1.2.5 = Quick-connector for vacuum and power supply for quick start-up
(compatibility).

8 9 2 38 9
>
> FR1:2:1 >
> X 0 0 0 0 >> DP1:2:1 >
>
> FR1:2:2 >
> > 6X >
> >
>
< = 6 X 0 0 077< DP1:2:2 =
FR1:2:3 ¼ 6
6X 0 X X X77> DP1:2:3 ð17:4Þ
>
> >
> >
>
> FR1:2:4 >
>
40 0 X X X 5>
>
> DP1:2:4 >
>
>
: ; : ;
FR1:2:5 0 0 X X X DP1:2:5

Figure 17.2 illustrates the result of the conceptual design, based on the AD
decomposition. The figure shows a proposal for a new universal assembly table.
Instead of a process-oriented assembly it was developed as an object-oriented
assembly bringing together the processes and technologies for gluing the different
components, for milling and for surface finishing by grinding. The system allows a
flexible positioning and fixation of different products (shower trays and washbasins)
—therefore every assembly table is able to produce variants from both product
families.
Exercise 17.1: Development of a Smart Assembly System for the Case Study Company
You are engaged by the case study company to develop a more intelligent assembly system
as the one shown in Fig. 17.2. For this, you should use the concepts of Industry 4.0 and
Internet of Things. Main objective of the re-design is to increase reactivity of the assembly
system when a changeover to another product becomes necessary. What are your FRs?
What are your corresponding DPs?
492 D. T. Matt and E. Rauch

Fig. 17.2 Case study results—design of a universal assembly table. (Reproduced with permission
from Holzner (2015) (Reprinted from Procedia CIRP, 34, Philipp Holzner, Erwin Rauch,
PasqualeRusso Spena, and Dominik T. Matt, “Systematic Design of SME Manufacturingand
Assembly Systems Based on Axiomatic Design”, 81–86, Copyright (2015),with permission from
Elsevier.).)
17 Application of Axiomatic Design for the Design of Flexible … 493

17.4 Case Study 2: Design of a Smart Shopfloor

Management System

Traditional shopfloor management concepts are changing to new and digitally

supported approaches for the coordination and management of production at the
shop floor level. With Industry 4.0 and the Internet of Things production data will
be provided in a completely new quality and with real-time information about
production processes. This will be possible by the comprehensive equipping of
production with sensors for data acquisition and the consistent integration of
intelligent machines (so-called cyber-physical systems) and products. The future
representation of the data models in real time makes production transparent and
thus easier to control. In the future, such a production control system can also cope
with short-term changes in demand and capacity utilization. The digitization and
collection of production data promises also a better use of the data for shopfloor
management to increase efficiency and sustainability. In this second case study, we
report about the design of digital ad smart shopfloor management systems (see also
Rauch et al. 2018).
In traditional lean management, shopfloor management has often taken place in
the so-called lean war room (Japanese obeya) with the support of analog visual-
ization and communication tools. The digitization of shopfloor management
requires the introduction of specific software systems for production monitoring and
management. Such mobile- and web-based software applications are suitable
instruments for combining human decision capacities, methods, relevant production
data, and innovative technologies in a smart shopfloor management.
The software company Solunio GmbH recognized the trend of the time and dealt
relatively early with the development of a software to collect data from the pro-
duction, to structure this data and to visualize it as relevant information for man-
agers, shift leaders, and workers in production. Based on existing expertise in
industrial companies the company developed the commercial software “Visual
Shop Floor” for smart shopfloor management with three layers (see Fig. 17.3).
At layer 1 (Data Collection), data are collected by connecting the application
with various systems and equipment in the company. Such systems can be the
internal ERP (enterprise resource planning), a MES (manufacturing execution
system) or MDA/PDA (machine or production data acquisition), as well as specific
systems such as CRM (customer relationship management), APS (advanced plan-
ning system), QMS (quality management system), or intelligent sensors installed on
machines.
At layer 2 (Data Management), all data collected is then aggregated and
structured by filtering relevant data and transferring it into structured data models.
The initially large data flood (“Big Data”) is thus transformed into smart and
relevant data (“smart data”). The data are then grouped into (i) order data,
(ii) machine data, and (iii) worker data.
494 D. T. Matt and E. Rauch

Fig. 17.3 Existing functionalities of the commercial software Visual Shop Floor. (Reproduced
with permission from Solunio GmbH)

Layer 3 (Shopfloor Management) uses the data for shopfloor management by

means of various functions. The “Performance Monitoring” module makes it
possible to display key performance indicators (KPIs) or statistics in real time
avoiding complex updating of reports. The “Smart Info Board” module enables
general information to be displayed to all or specific monitors in production at the
17 Application of Axiomatic Design for the Design of Flexible … 495

touch of a button, eliminating the need for time-consuming manual update of paper.
The “Smart Layout” module enables the user to visualize the current situation in
production by means of a dynamic graphical layout representation. If the user
zooms in or out, more details such as machine status or job data appear.
However, the company, which is mainly run by software developers and not
production experts, found it difficult to understand the comprehensive needs of the
users for developing a new release of the software with additional
functionalities/modules. In order to analyze the FRs in a systematic way and to
define appropriate modules in the software, the company decided to use AD
together with experts. As a first step, several workshops were held with production
managers and staff to better understand the customer needs (see Fig. 17.4).
The highest level of FRs and DPs can be expressed as follows:
FR0 Manage the production shopfloor in a smart way.
DP0 Smart shopfloor management software.

Fig. 17.4 Customer needs collected in workshops with production leaders and workers.
(Reproduced from Rauch et al. (2018), originally published open access under a CC BY 4.0
license: https://doi.org/10.1051/matecconf/20182230101)
496 D. T. Matt and E. Rauch

Based on the results of the initial workshops the following high-level FRs and
related DPs were defined:

FR1 = Collect real time data of machines, processes, resources, and energy;
FR2 = Avoid inconsistency of data;
FR3 = Visualize needed and related data everywhere for individual roles in
production without using paper;
FR4 = Monitoring of performance (input–output) in production;
FR5 = Allow individual and production related data analysis;
FR6 = Avoid problems and increase reactivity in decision-making if problems
arise;
FR7 = Facilitate acceptance of employees;
DP1 = Data collection module with interfaces and smart sensors;
DP2 = Data structuring module;
DP3 = Digital visualization technologies;
DP4 = Monitoring tools;
DP5 = Tools for production data analysis;
DP6 = Problem prevention and problem-solving tools;
DP7 = Dedicated training of employees and tutorials.

The design matrix on the first level is decoupled and shows the dependencies
between the solutions (DPs) and the FRs:
8 9 2 38 9
>
> FR1 >
> X O O O O O O >> >
DP1 >
>
> >
> 6O O7 >
> >
>
>
> FR2 >
> X O O O O >
> DP2 >
>
>
> > 6 7> >
< FR3 >
= 6
6O O X O O O O7 > >
7< DP3 =
FR4 ¼ 6 O O O X O O 7
O 7 DP4 ð17:5Þ
>
> > 6 > >
> FR5 >
> > 6
> 6 O O O O X O O7 >
> DP5 >
7> >
>
>
> > 4 > >
>
> FR6 >
>
> O O O X X X O 5>
>
> DP6 >
>
>
: ; : ;
FR7 O O X X X X X DP7

DP1 (data collection) does not need a further decomposition as this module is
already realized in the existing commercial software and works well for the planned
software re-design. Same as above occurs also for DP2 (data structuring). The data
structuring method has already been realized and is not part of the re-design.
DP3 (digital visualization technologies) is supporting to visualize specific con-
tents to operators or other users in production. Mobile apps help to visualize data
everywhere, while an own function is visualizing only data relevant for the own
position (shift-leader, operator, etc.) while an Info Board contains general infor-
mation. Problems are visualized graphically in the layout as problem map.

FR3.1 = Visualize data everywhere;

FR3.2 = Visualize relevant data for the own role;
FR3.3 = Visualize general information;
FR3.4 = Visualize problems transparently;
17 Application of Axiomatic Design for the Design of Flexible … 497

DP3.1 = Mobile App module;

DP3.2 = MyRole View module;
DP3.3 = Info Board module;
DP3.4 = Problem Map module.

The design matrix shows a decoupled design. The Mobile App module consists
of an app containing a mobile version of the software. Thus, it has a dependency to
other FRs. The Info Board module is also able to visualize problems transparently.
8 9 2 38 9
>
> FR3:1 >
> X O O O >> DP3:1 >
>
< = 6 < =
FR3:2 X X O O77 DP3:2
¼6
4X ð17:6Þ
>
> FR3:3 >
> O X O 5>
> DP3:3 >
>
: ; : ;
FR3:4 X O X X DP3:4

DP4 (monitoring tools) needs to be further decomposed. Production perfor-

mance indicators are monitored in a performance monitoring function, while energy
consumption is monitored in an energy cockpit. The smart layout function shows
changes in the layout.

FR4.1 = Monitoring of production performance;

FR4.1 = Monitoring of energy performance;
FR4.1 = Monitoring of changes in the layout;
DP4.1 = Performance Monitoring;
DP4.2 = Energy Cockpit;
DP4.3 = Smart Layout.

The design matrix shows an uncoupled design:

8 9 2 38 9
< FR4:1 = X O O < DP4:1 =
FR4:2 ¼ 4 O X O 5 DP4:2 ð17:7Þ
: ; : ;
FR4:3 O O X DP4:3

Also DP5 (tools for production data analysis) needs a decomposition. The
software should not only allow to see general monitoring dashboards but also to do
more detailed data analysis using the data stored in the software. Thus, the software
should provide a digital value stream mapping function in order to compare dif-
ferent situations and to identify potentials for optimization. Further individual data
analytics should also be possible.

FR5.1 = Allow lean production data analysis;

FR5.2 = Enable individual data analysis;
DP5.1 = Digital Value Stream Map;
DP5.2 = Data Analytics.
The design matrix shows an uncoupled design:
498 D. T. Matt and E. Rauch

    
FR5:1 X O DP5:1
¼ ð17:8Þ
FR5:2 O X DP5:2

DP6 (problem prevention and problem-solving tools) needs a further decom-

position. Problems due to machine stops should be avoided in future through early
warning systems or predictive maintenance. A digital problem issue tracking
function should also reduce the duration of the procedure to solve a problem. The
introduction of a location-based decision-making can further improve also the
reactivity for problem-solving. This means that, e.g., the nearest shift-leader will be
informed if the production system identifies a problem where a certain decision is
needed.

FR6.1 = Prevent stops of production machines;

FR6.2 = Reduce duration of problem-solving;
FR6.3 = Increase reactivity in problem-solving;
DP6.1 = Predictive maintenance module;
DP6.2 = Issue tracking;
DP6.3 = Location-based decision.

The design matrix shows a decoupled design. Predictive maintenance as well as

issue tracking increases the reactivity for problem-solving showed in the matrix by
the dependencies of the DPs with the last FR.
8 9 2 38 9
< FR6:1 = X O O < DP6:1 =
FR6:2 ¼ 4 O X O 5 DP6:2 ð17:9Þ
: ; : ;
FR6:3 X X X DP6:3

Also DP7 (dedicated training of employees and tutorials) needs a decomposition.

It is crucial for the implementation of such a shopfloor management system that all
user are accepting the software. Thus the design foresees two further lower level
FR–DP pairs.

FR7.1 = Support users if problems arise;

FR7.2 = Increase skills of users in data management;
DP7.1 = Online Tutorial;
DP7.2 = Dedicated trainings in the shopfloor management software.

The design matrix shows a decoupled design. There is a dependency as the

regular use of online tutorials is also helping to train people working with the
software.
    
FR7:1 X O DP7:1
¼ ð17:10Þ
FR7:2 X X DP7:2
17 Application of Axiomatic Design for the Design of Flexible … 499

Figure 17.5 shows the final concept and structure of modules/functionalities of

the software. The module “Data Analytics” enables the user to create individual
data analyses in an intuitive way without any specific knowledge, thus using
existing data effectively and efficiently. The “Mobile App” module enables the user
to access the software system from mobile devices. In order to keep distances short
and reaction speed fast, the module “Location-based Decision” supports calling the
closest decision authority to the place of the immediate decision. The “myRole
View” module enables the user to switch between a standard view and a
role-specific view in order to display only that information that is important and
useful for their role (e.g., worker, shift-leader). The “Smart Layout” module enables
the user to visualize the current situation in production by means of a dynamic
graphical layout representation. If the user zooms in or out, more details such as
machine status or job data appear. The “Performance Monitoring” module makes it
possible to display KPIs or statistics in real time avoiding complex updating of
reports. The “Smart Info Board” module enables general information to be dis-
played to all or specific monitors in production at the touch of a button, eliminating
the need for time-consuming manual update of paper. Based on internal data
algorithms the “Predictive Maintenance” module determines automatically, when
the next maintenance intervals are to be carried out. The “Problem Map” module
graphically shows where quality problems arise in the company, and therefore
enables intuitive and quick detection of problem areas or machines. The “Issue

Fig. 17.5 Future modules of the Shopfloor management software. (Reproduced from Rauch et al.
(2018), originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/
matecconf/20182230101)
500 D. T. Matt and E. Rauch

Tracking” module digitally tracks the systematic processing of occurring problems

in production by allowing the status of open problems to be continuously and in
real time updated by the parties involved. The “Energy Cockpit” module shows the
current energy consumption in the company and warns the user as soon as the
consumption exceeds a defined warning limit or follows a negative or positive
trend. The “Digital Value Stream Map” module is the digital image of a traditional
value stream map, which is updated continuously and in real time and thus avoids
time-consuming manual data collection. The “Dedicated Trainings” module offers
access to training documents or video material directly at the workplace in order to
facilitate the training phase of new employees at the workplace or during the
start-up of new products. The last module “Online Tutorial” enables decentralized
access to tutorials for problem-solving and for frequently asked questions (Rauch
et al. 2018).
Currently, the company has already implemented the modules DP3.1, DP3.3,
DP4.1, DP4.3, DP5.2, DP6.2 in a new software release. The other modules form the
basis for further developments and will be implemented in subsequent releases.
Exercise 17.2: Detailed Design of a Shopfloor Management Software Module
Imagine you are in charge of implementing the remaining modules and planning their
implementation. Pick a module that has not yet been implemented and try to design it with
AD using a similar approach to the one in this case study.

17.5 Case Study 3: Design of Collaborative Human–Robot

Assembly Workplaces

With new technologies and more intelligent control systems for machines and
robots, the collaboration between human and machines has become easier. In smart
manufacturing systems, robots are working hand in hand with people and support
them, when their assistance is needed.
However, the implementation of such collaborative human–robot workplaces is
not so easy in practice. The design of collaborative workplaces presents completely
new challenges in terms of safety of the worker. Commercial collaborative robots
are safe as such, but as soon as they are used in a specific application situation, this
often changes. For example, the robot can enter or stop in a safety mode when in
contact with the operator. However, if the robot is equipped with a dangerous
gripper (sharp, pointed), the potential danger can change or increase. This means
that the use of collaborative robotics no longer makes sense or that the robot can in
turn be used only with a safe enclosure. Therefore, possible sources of danger must
be identified and eliminated or minimized by appropriate design solutions of col-
laborative workspaces.
There are several international standards for the safety of workstations with
robots. The various standards regulate different cases and situations and are often
difficult to understand due to their complexity and scope. In particular, there is a
17 Application of Axiomatic Design for the Design of Flexible … 501

lack of an overview of which standards can be applied and for which situations.
Manufacturing system designers, therefore, often have difficulties in applying these
standards in the right way (Gualtieri et al. 2018).
Such a complex problem requires a systematic and structured approach for
concept design, in order to avoid loops in the design stage or even worse during
implementation and to identify the relevant international standards for safety.
In this third case study, AD is used to examine how FRs can describe sources of
danger and classify them according to the applicable standards (see also Gualtieri
et al. 2018). This provides practitioners with a clearer overview of the most relevant
safety standards to keep in mind when designing a collaborative workplace. In
Fig. 17.6, we can see the main standards available for the design of the safety
system in collaborative robotic cells.

Fig. 17.6 Collaborative robotic cell: main standards for the safety systems design. (Reproduced
from Gualtieri et al. (2018), originally published open access under a CC BY 4.0 license: https://
doi.org/10.1051/matecconf/201822301003)
502 D. T. Matt and E. Rauch

The highest level of FRs and DPs are

FR0 = Reduce the mechanical risks that could arise from not intentional human–
robot physical interaction;
DP0 = ISO 12100:2010 and ISO 10218-2:2011 (safe collaborative workstation
by reducing as far as possible the risk probability and gravity).

The top-level FRs and relative DPs were defined as follows:

FR1 = Prevent unexpected human–robot contacts;

FR2 = Reduce the intensity of unexpected human–robot contacts;
FR3 = Avoid the access to the dangerous zone physically;

DP1 = Contact prevention measures according to ISO 13849-1:2015;

DP2 = Power and force limitation according to ISO 10218-1:2011 and ISO TS
15066:2016 (Sect. 5.5.5);
DP3 = Physical barriers/limitations according to ISO 14120:2015.
8 9 2 38 9
< FR1 = X O O < DP1 =
FR2 ¼ 4 X X O 5 DP2 ð17:11Þ
: ; : ;
FR3 0 0 X DP3

FR1 (Prevent unexpected human–robot contacts) and DP1 (contact prevention

measures) need a further decomposition on a lower level. The design of the safety
systems that aims to actively prevent human–robot unexpected contacts can be
satisfied using different complementary approaches. Depending on the final appli-
cation, it is possible to apply one or more combined solutions. Starting from FR1,
further FRs and DPs of the successive hierarchical level can be defined as follows:

FR1.1 = Determine the minimum distances to a hazard zone with respect to

approach speeds of parts of the human body;
FR1.2 = Monitor when a person enters the safeguarded space;
FR1.3 = Timely sensing of a possible collision;
FR1.4 = Reduce speed if a robot comes nearby people;
FR1.5 = Prevent unexpected machine start-ups;
DP1.1 = Position safeguards with respect to the human body part speed according
to ISO 13855:2010;
DP1.2 = Safety rated monitored stops according to ISO TS 15066:2016
(Sect. 5.5.2);
DP1.3 = Sensitive protection according to IEC 62046:2018;
DP1.4 = Monitoring of speed and separation according to ISO TS 15066:2016
(Sect. 5.5.4);
DP1.5 = Measures for prevention of unexpected machine start-up according to
ISO 14,118:2017.
17 Application of Axiomatic Design for the Design of Flexible … 503

The design matrix shows a decoupled design:

8 9 2 38 9
>
> FR1:1 >
> X 0 0 0 0 > > DP1:1 >
>
>
> > 6 > >
< FR1:2 >
= 6X X 0 0 07 >
7< DP1:2 =
>
FR1:3 ¼ 6
6X 0 X 0 7
0 7 DP1:3 ð17:12Þ
>
> >
> > >
>
> FR1:4 >
>
4X 0 X X 0 5>> >
>
> DP1:4 ;
>
: ; :
FR1:5 0 0 0 0 X DP1:5

For FR2 (reduce the intensity of unexpected human–robot contacts) and DP2
(power and force limitation) there is no need for a further decomposition. The
design of the safety systems that safeguard the operator during unexpected but
allowed human–robot contacts can be achieved by reducing the energy exchange
according to the “power and force limiting” approach. For the implementation of
this DP, the guidelines explained in ISO 10218-1:2011 and ISO TS 15066:2016
Sect. 5.5.5 can be applied.
FR3 (avoid the access to the dangerous zone physically) and DP3 (physical
barriers/limitations) need a further decomposition on a lower level. The design of
the safety systems that aims safeguard the operator by avoiding human–robot
contacts using physical limitations can be satisfied through the preventive design of
the components of the robotic cell as well as the design of safeguards.
Starting from FR3, further FRs and DPs of the successive hierarchical level can
be defined as follows:

FR3.1 = Prevent hazard zones being reached by operator during manual work
activities;
FR3.2 = Avoid crushing of parts of the human body;
DP3.1 = Protective structures according to ISO 13857:2008;
DP3.2 = Minimum gaps relative to parts of the human body according to ISO
13854:2017.

The design matrix shows a decoupled design:

    
FR3:1 X O DP3:1
¼ ð17:13Þ
FR3:2 X X DP3:2

Figure 17.7 explains a preliminary concept layout of the new collaborative

workstation, including the main safety systems and their related standards. As
shown, there will be a collaborative space, where human and robot will share the
workspace in order to perform a common production task, and a not-collaborative
space, where operators are not allowed and the robot can work, performing more.
Of course, different limited spaces involves different safety requirements and
systems, which are regulated by different standards.
504 D. T. Matt and E. Rauch

Fig. 17.7 Exemplary design of a collaborative assembly workplace. (Reproduced from Gualtieri
et al. (2018), originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/
matecconf/201822301003)

Exercise 17.3: Design a Learning Factory Lab at Your University

Imagine you have to design a collaborative workplace for a previously purely manual
assembly. Discuss in small groups, which possibilities exist in detail for the implementation
of DP1.1 to DP1.5. If possible/available, use the relevant standards as an aid and then
present your proposals to the group.

17.6 Case Study 4: Design of a Learning Factory

for Industry 4.0

Learning factories are an idealized representation of real production environments.

In the last decades numerous learning factories have been built in industry and
academia. The first example of a learning factory was established in the United
States at Penn State University in 1994. Later in the 1980s, one of the first learning
factories in Europe was established with the “Lernfabrik” for computer-integrated
manufacturing (CIM) at the research center Fraunhofer IAO in Stuttgart. Since the
17 Application of Axiomatic Design for the Design of Flexible … 505

last decade more and more learning factories appeared like the Process Learning
Factory at TU Darmstadt, Pilotfabrik at TU Vienna, ifactory at the University of
Windsor or the Laboratory for Manufacturing Systems and Automation at
University of Patras. The specific objectives of learning factory concepts in an
academic environment are mainly to offer a practice-based engineering curriculum,
balancing analytical and theoretical knowledge with manufacturing skills as well as
hands-on experience in the design of manufacturing systems and product realiza-
tion. While the majority of learning factories have mainly worked on concepts of
lean management, today the trend is going toward learning factories for Industry 4.0
(I4.0). Modern learning factories are, therefore, acting as application centers for
Industry 4.0 in order to familiarize with new technologies, to test them in practice
and to learn how to use them in the industrial environment.
When setting up such I4.0 learning factories, it is important to define the extent
to which the learning factory is to act. Here, the requirements of students and
stakeholders, i.e., companies, must be taken into account. This case study attempts
to explain the use of AD to derive design guidelines for I4.0 learning factories. The
case study is based on the development of an I4.0 learning factory at the Free
University of Bolzano named Smart Mini Factory showing the single steps how the
design guidelines were implemented in practice (Rauch et al. 2019a).
According to the collective system design approach of Cochran and Kim (2010),
it is important to include all members of a design (user and stakeholder) in the
design process in order to achieve a collective agreement on the requirements of the
system. Based on a literature review and workshops with user and stakeholders,
their needs in engineering education were identified (see Table 17.1). The “X” in
Table 17.1 means that the identified need can be assigned to one or more stake-
holders. University students, high school students, researchers, teachers, and
experienced people from industry were identified as main users, while universities,
high schools, and industrial companies are the main stakeholders.
After the collection of CNs in Table 17.1, the next step in the design approach
was to define top-level FRs to start with the development of an FR–DP decom-
position. For the purpose of this work, the following top-level FRs–DPs were
defined:
FR0 = Increase the qualification level of engineers and technicians in Industry
4.0 technologies;
DP0 = I4.0 Learning Factory for practically oriented engineering education.
According to AD, this vague and very abstract design solution in DP0 needs to
be further decomposed until DPs are detailed enough to work with them. Based on
FR0–DP0, the following first-level FRs and DPs as well as design fields (DF) were
defined. For this first-level DPs, so-called design fields (DF) were defined, where a
DF is a main field of responsibility. In case of the identified DFs, the approach
suggests one responsible for each.
FR1 = Transfer newest insights from research in Industry 4.0;
FR2 = Teach industry 4.0 technologies and methods in a practically oriented
way;
506 D. T. Matt and E. Rauch

Table 17.1 Collectively identified customer needs (CNs) of users and stakeholders. (Reproduced
with permission from Rauch et al. (2019a).)
CN Customer Need (CN) University High Industry
School
CN1 Real-world practical experience (Cochran and Smith 2018) X X X
CN2 Safe and comfortable environment to ask questions (Cochran X X X
and Smith 2018)
CN3 Interesting contents (Cochran and Smith 2018) X X X
CN4 One-to-one interaction with instructor (Cochran and Smith X X X
2018)
CN5 Application of theories (Cochran and Smith 2018) X X X
CN6 Chance for trial and error without negative consequences X X X
(Cochran and Smith 2018)
CN7 Getting in contact with new and emerging (I4.0) X X X
technologies
CN8 Understanding if an engineering study could be the right X
choice
CN9 Training the own teachers and instructors in new X
technologies
CN10 Training the own skilled workers or engineers already hired X
CN11 Test new technologies before investing and implementing X X
them
CN12 Use equipment to perform experiments or for research X X
purpose
Reprinted from Procedia Manufacturing, 31, Erwin Rauch, Florian Morandell, and Dominik T. Matt,
“AD Design Guidelines for Implementing I4.0 Learning Factories”, 239–244, Copyright(2019), with
permission from Elsevier

FR3 = Exploit the resources and competences developed in the lab;

FR4 = Provide the resources to sustainably establish and run a learning factory;
DP1 = Research projects and collaborations (DF1: Research);
DP2 = Education program in the learning factory (DF2: Teaching);
DP3 = Collaboration models with local industry (DF3: Industry);
DP4 = Funding for setting up a learning factory and business model (DF4:
Management).

Based on the top-level FR–DP pair and the first-level decomposition, the fol-
lowing lower level FR–DP pairs could get derived (see Fig. 17.8). On the left side,
we can see “what” should be achieved (FR) and on the right side “how” it could be
satisfied (DP). The sum of the lowest level DPs of every branch (highlighted in
blue) builds the final list of a total of 20 design guidelines (DG) for I4.0 learning
factories for practically oriented engineering education in universities.
The Smart Mini Factory Laboratory at the Free University of Bozen-Bolzano has
been successfully set up using the previously shown design guidelines to convert
the learning factory from a lean oriented lab to an I4.0 learning factory. Table 17.2
summarizes the DG and the activities implemented in the Smart Mini Factory:
17 Application of Axiomatic Design for the Design of Flexible … 507

Fig. 17.8 Axiomatic Design mapping and decomposition of FR0–DP0. FRs on the left side and
corresponding DPs on the right side, the in blue highlighted DPs represent the derived design
guidelines for learning factories (LF) (screenshot created with AD software Acclaro DFSS).
(Reproduced with permission from Rauch et al. (2019) (Reprinted from Procedia Manufacturing,
31, Erwin Rauch, Florian Morandell, and Dominik T. Matt, “AD Design Guidelines for
Implementing I4.0 Learning Factories”, 239–244, Copyright (2019), with permission from
Elsevier.)

Exercise 17.4: Design a Learning Factory Lab at Your University

As a student, you are one of the main users of learning factories alongside other possible
target groups (such as companies). As a student, consider the requirements you place on a
possible learning factory laboratory at your university. Then define the FRs, derive pos-
sible design proposals (DPs) and check your design in each step by using a design matrix.
Discuss your design with your professor. He will be grateful for your valuable input from a
student's perspective.
508 D. T. Matt and E. Rauch

Table 17.2 Collectively identified customer needs (CNs) of users and stakeholders. (Reproduced
with permission from Rauch et al. (2019a).)
DGi DPi Exemplary application in the Smart Mini Factory
(SMF) lab
DG1 – Vision and mission DP1.1.1 A new vision and mission statement as well as
I4.0 oriented research topics were defined for the
SMF (see also the following link: www.
smartminifactory.it)
DG2 – Fundraising DP1.1.2 To strengthen existing and to develop new I4.0
competences, fundraising was started acquiring
1.25 mio Euro from 2017–2018 for research
projects on Industry 4.0 related topics
DG3 – Regional DP1.2.1 SMF is member of a trans-regional network of
collaboration network partners from Italy and Austria working on
innovative education and training models
(ongoing research project “Engineering Education
4.0”)
DG4 – International partner DP1.2.2 An international network of I4.0 and learning
network factory experts could be developed through an
international researcher exchange project on
Industry 4.0 funded by the EC H2020 MSCA
RISE program as well as through the participation
at the Conference on Learning Factories
DG5 – Study projects DP1.3.1 Students from university were involved in
research projects through thesis projects for
undergraduate and graduate students as well as
mandatory study projects for graduate students
DG6 – Student assistant DP1.3.2 Students were involved financing “120 h student
positions jobs” where students assist research teams in the
conduction and preparation of research activities
on Industry 4.0
DG7 – Projects with high DP1.3.3 High school students were involved in research
schools projects serving as test persons for lab
experiments or by outsourcing parts and tasks of
research projects
DG8 – Exercises and DP2.1 The learning factory is used in ten lectures in the
seminars for university undergraduate and graduate program for practical
students lab exercises and case study trainings. Further,
students can participate to Industry 4.0 seminars
for industry
DG9 – Seminars for high DP2.2 The learning factory offers two Industry 4.0
school teacher seminars to high school teacher from technical or
scientific high schools to enforce the knowledge
transfer to the high school level
DG10 – Exercises and DP2.3 The learning factory is used for practical lab
seminars for high school exercises and case study trainings with high
students school classes. The goal is first to promote
engineering study courses and secondly to
(continued)
17 Application of Axiomatic Design for the Design of Flexible … 509

Table 17.2 (continued)

DGi DPi Exemplary application in the Smart Mini Factory
(SMF) lab
increase the qualification level of technicians with
high school degree
DG11 – Specific I4.0 DP2.4.1 The SMF offers nine specific Industry 4.0
seminars for industry seminars for professionals in industry in order to
increase the qualification level of existing
workforce in industrial companies
DG12 – I4.0 exercises in DP2.4.2 The SMF offers practical training sessions
executive training courses integrated in training programs organized by other
associations and the local Chamber of Commerce
DG13 – Funding of PhD DP3.1 Through the funding of PhD positions with a
positions duration of 3–5 years, industrial companies can
outsource long-term research projects to SMF and
often take over the doctoral candidate afterwards
DG14 - Commissioned DP3.2 For medium-term research projects, industrial
industry projects companies engage single researchers or individual
teams in the SMF to work on a specific research
problem
DG15 – Rent of equipment or DP3.3 In case of short-term necessity of equipment of the
lab tests SMF as well as the conduction of tests, industry
companies can engage the SMF (equipment is,
therefore, well described on the website of the
SMF)
DG16 – Start-up funding for DP4.1.1 The SMF received public funding of 2.3 mio Euro
core team for financing a core team consisting of a lab
manager, post-doc researcher, and lab technicians
DG17 – Start-up funding for DP4.1.2 0.5 mio Euro of the total funding of 2.3 mio Euro
I4.0 equipment from public is addressed to finance up to date
equipment and technologies related to industry 4.0
DG18 – Service portfolio DP4.2.1 According to the results of the conducted AD
decomposition, the SMF offers services in the
following fields: (i) Research, (ii) Teaching, and
(iii) Industry
DG19 – Pricing and payment DP4.2.2 In case of the SMF, the definition of the pricing
structure and the payment model was supported
by a market analysis of common prices for
training courses as well as by discussions with
local providers of training courses for
professionals. This was necessary to avoid that
SMF owned by the university is in competition
and in conflict with private provider of training
courses
DG20 – Marketing strategy DP4.2.3 Creation of a professional and responsive website
for the SMF lab. Further marketing instruments
were posting of regular news on the website,
brochures for high schools and brochures for
industrial companies, the organization of regular
(continued)
510 D. T. Matt and E. Rauch

Table 17.2 (continued)

DGi DPi Exemplary application in the Smart Mini Factory
(SMF) lab
events or workshops with companies and the
organization of lab visits for target groups (high
school students and industrial companies). In a
next step also the activation of social media
channels will be started in order to reach
companies (LinkedIn), students (facebook) and
scholars (ResearchGate)
Reprinted from Procedia Manufacturing, 31, Erwin Rauch, Florian Morandell, and Dominik T.
Matt, “AD Design Guidelines for Implementing I4.0 Learning Factories”, 239–244, Copyright
(2019), with permission from Elsevier

17.7 Case Study 5: Design of a Demonstrator

for a Flexible and Decentralized Cyber-Physical
Production System (CPPS)

Cyber-physical production systems (CPPS) are equipped with intelligent units,

so-called cyber-physical systems (CPS), and enable new possibilities in the factory
of the future through the connectivity between the digital world and the physical
production system. Many enterprises are still quite skeptical regarding the vision of
Industry 4.0. The term CPS stands for the continuously increasing presence of
computing and communication capabilities in physical objects in the real world.
The decision-making process of such systems in production may be supported or
handled autonomously by computational intelligence.
It has been so far difficult to transfer such advantages and concepts in a clear and
practical-oriented manner. It is not always possible to train and show these
emerging concepts directly in a real factory environment. Therefore, demonstration
models are a popular alternative and complementary solution, where the concepts
and technologies can be demonstrated and explained in a miniaturized way to
employees as well as to students. Therefore, Fraunhofer Italia Research built a
demonstration model for a flexible and decentralized CPPS system for showcase
purposes (Egger et al. 2017). The demonstration model created by Fraunhofer Italia
is intended to facilitate the knowledge transfer of Industry 4.0 concepts and CPPS
to project partners, industrial firms as well as students from schools and universities.
The case study shows an AD based re-design of the realized demonstration
model of a flexible and decentralized CPPS. The previous demonstration model was
built by a team of Fraunhofer Italia researchers and students to demonstrate the
potentials of a CPPS in the factory of the future. Mainly the model aims to
demonstrate the following concepts of typical and modern factory of the future
concepts to students and companies (see also Fig. 17.9):
17 Application of Axiomatic Design for the Design of Flexible … 511

Fig. 17.9 Previous architecture of the CPPS demonstrator at Fraunhofer Italia Research.
(Reproduced from Egger et al. (2017), originally published open access under a CC BY 4.0
license: https://doi.org/10.1051/matecconf/201712701016)

• flexible Transport System;

• intelligent Work-piece Carrier;
• decentralized Control;
• digital Interconnection;
• efficient Human–Machine Interface.

Figures 17.9 and 17.10 illustrate the system architecture and the realized demon-
stration model consisting of:

• laser engraving head;

• safety housing (custom design);
• fume extraction unit;
• air compressor;
• NFC (near field communication) pads;
• vehicles for product transport.
512 D. T. Matt and E. Rauch

Fig. 17.10 Realized CPPS demonstrator with vehicles for product transport. (Reproduced from
Egger et al. (2017), originally published open access under a CC BY 4.0 license: https://doi.org/
10.1051/matecconf/201712701016)

The vehicles presented in Fig. 17.10 consist of a commercial robotic platform

with 4 DC motors, an Arduino-compatible controller board, an internally developed
apple-spin mechanism (apple symmetry axis aligns to spin axis), a custom built line
sensor with eight analog sensors, an IR distance sensor for collision avoidance, an
apple-eject mechanism, and a control algorithm (line follower) with routing capa-
bility and the ability to detect and take crossings. In addition, a user interface was
developed to get orders from a computer terminal. Visitors may enter their name
and a personal message to be engraved on the apple. When ready, a screen informs
the visitor to pick up his apple. Visitors need to present their RFID card in order to
start the delivery of the apple.
The research team collected the main requirements and needs for the re-design of
the CPPS demonstrator and categorized them as shown in Table 17.3. The table
shows also if the needs were fully or only partly fulfilled (improvable) in the
previous demonstration model or if they are new.
The highest level of FRs and DPs are

FR0 = Demonstrate Industry 4.0 concepts for CPPS in a practical way;

DP0 = Demonstration Model for a flexible and decentralized CPPS.

Based on the outcome of the initial workshop, the following Cs for the (re)
design were defined:

C1 = Maximum total budget of 20,000 Euro for new investments;

C2 = Maximum space of 6 square meters;
C3 = Modular structure with standard components;
C4 = Mobility of the demo model for flexible use.
17 Application of Axiomatic Design for the Design of Flexible … 513

Table 17.3 Collectively identified customer needs (CNs) of users and stakeholders. (Reproduced
from Egger et al. (2017), originally published open access under a CC BY 4.0 license: https://doi.
org/10.1051/matecconf/201712701016)
CN Customer Needs C, FR, Existing
n-FR
CN1 Demonstrate intelligent and decentralized control of FR4 Yes
individualized production (lot size 1)
CN2 Permit rapid scaling of the capacity up and down FR2 In part
automatically
CN3 The product should have a local link n-FR1 Yes
CN4 The model should be realized with students for training n-FR2 Yes
purpose
CN5 The demonstration model should show a complete n-FR3 In part
production process
CN6 The maximal budget for realization are 20,000 Euro C1 Yes
CN7 Modularity in order to reuse the single components also in C3 Yes
other educational or research settings
CN8 Maximum space for the demonstration model is 6 m2 C2 Yes
CN9 The demonstration model needs to be movable to use it in C4 In part
different events and facilities
CN10 Safety for user and visitors has to be guaranteed FR3 Yes
CN11 Allow interaction between visitor and the CPPS demo FR3 Yes
model
CN12 Traceability of orders FR4 In part
CN13 Quality check of the order before delivery FR3 In part
CN14 Delocalization of order entry (e.g., on mobile devices) FR3 Yes
CN15 Demonstration of advanced industrial robotics in the CPPS FR1 No
CN16 Experience the production from the point of view of the FR5 No
product

Further, the following FRs were defined:

n-FR1 = Products should have a local link;

n-FR2 = The model should be realized and run with the support of students for
training purpose;
n-FR3 = The demonstration model should show a complete production process.

Finally the remaining CAs were associated to top-level FRs deriving DPs:

FR1 = Apply advanced industrial robotics in the CPPS;

FR2 = Allow automatic scaling of capacity up and down;
FR3 = Ensure safety during operation of personnel, visitors, and equipment;
FR4 = Event-based dynamic control and monitoring of a production line for
mass customized products;
514 D. T. Matt and E. Rauch

FR5 = Visualize the production from the point of view of the product;
DP1 = Lightweight robot and vision system for bin picking at the loading
station;
DP2 = Buffer for waiting vehicles and automatic call;
DP3 = Safe user interface between user/visitor and the cyber-physical system;
DP4 = Intelligent and autonomous vehicles driven by a decentralized control
architecture;
DP5 = Camera system on the work-piece carrier.

The design matrix on the first level is decoupled and shows the dependencies
between the solutions (DPs) and the FRs:
8 9 2 38 9
>
> FR1:1 >
> X 0 0 0 0 > > >
DP1:1 >
>
> > 6 > >
< FR1:2 >
= 6X X 0 0 077
>
< DP1:2 >
=
FR1:3 ¼ 6
60 0 X 0 077> DP1:3 > ð17:14Þ
>
> >
>
>
> FR1:4 >
>
4X 0 X X 0 5>> >
> DP1:4 >
>
: ; : ;
FR1:5 0 0 0 X X DP1:5

Regarding FR1–DP1, industrial robotics can be integrated using a lightweight

robot for picking apples from the container and loading them on the vehicles.
Starting from FR1, further FRs and DPs of the successive hierarchical level can be
defined as follows:

FR1.1 = Localize and identify apple for flexible feeding;

FR1.2 = Grasp sensitive products;
DP1.1 = Lightweight robot combined with vision system for bin picking of the
apple;
DP1.2 = Flexible gripper for complex and sensitive products.

The design matrix shows an uncoupled design:

    
FR1:1 X O DP1:1
¼ ð17:15Þ
FR1:2 0 X DP1:2

For flexible feeding, the research team used existing equipment. A mobile sta-
tion with a mounted UR3 lightweight robot combined with a camera system allows
flexible feeding without additional investments. For grasping the apples, a flexible
and sensitive gripper is required to avoid damages to the product.
Scalability (FR2-DP2) is a major requirement of modern production systems.
While in the current demonstration model the number of vehicles is fixed, a buffer
should be created in the re-designed demonstration model. All other stations (laser
engraving and robotized loading station) are sufficiently rapidly scalable in their
performance. Vehicles not needed in periods with low demand can be parked in the
buffer to reduce energy consumption in the system, while they are called auto-
matically, when demand is rising. FR2 can be decomposed further as follows:
17 Application of Axiomatic Design for the Design of Flexible … 515

FR2.1 = Provide function for rapid scaling down;

FR2.2 = Provide function for rapid scaling up to guarantee capacity for higher
demand;
DP2.1 = Crossing points with NFC pads to detect vehicle direction for ejection
of the vehicles in a buffer line;
DP2.2 = Buffer line with sufficient length to guarantee the needed capacity.

The design matrix shows an uncoupled design:

    
FR2:1 X O DP2:1
¼ ð17:16Þ
FR2:2 0 X DP2:2

To realize the buffer line some more NFC pads had been integrated in the
demonstration model. In addition, the dimensions (length) of the buffer line were
defined according to the maximum number of vehicles, in order to guarantee the
expected performance of the model during visitor presentations.
For FR3–DP3 (safe user interface) the model should allow interaction between
the visitor/user and the CPPS. In the previous demonstration model, visitors may
create an individual order for writing an individual text on an apple by using a
desktop station or their smartphone. Any order requires approval by a supervisor.
To avoid injuries and malfunctions of the production equipment, the visitor cannot
touch the vehicles or any other stations, and the finished apple ejects to a with-
drawal tray, where the visitor may grasp it without interfering with the vehicle
itself. In the previous system, a quality check was missing for the simulation of a
complete production process. Further, in addition to the visualization on a screen,
the result of the quality check as well as the availability of the apple at the delivery
station should be sent to the visitor via app. Thus, this new functions had to be
integrated in the re-designed demonstration model. FR3 could be decomposed in
the following lower level FRs and DPs:

FR3.1 = Prevent direct intervention by the user in unsafe areas;

FR3.2 = Create individual order in situ or remotely;
FR3.3 = Check compliance of incoming orders;
FR3.4 = Check the quality after processing;
FR3.5 = Inform user/visitor about the order progress;
DP3.1 = Separation of unsafe areas (e.g., through acrylic glass screen);
DP3.2 = Order creation (individual text on the apple) at the order terminal or via
smartphone (app);
DP3.3 = Approval by supervisor on a monitor screen;
DP3.4 = Camera system at laser engraving station to compare the result with the
text in the order;
DP3.5 = Notification to the visitor after laser engraving station and at delivery
station.
516 D. T. Matt and E. Rauch

The design matrix shows a decoupled design:

8 9 2 38 9
>
> FR3:1 >
> X 0 0 0 0 > > DP3:1 >
>
>
> > 6 > >
< FR3:2 >
= 60 X 0 0 07 >
7< DP3:2 =
>
FR3:3 ¼ 6
60 X X 0 7
0 7 DP3:3 ð17:17Þ
>
> >
> > >
>
> FR3:4 >
>
40 0 0 X 0 5> >
> DP3:4 >
> >
: ; : ;
FR3:5 0 0 0 X X DP3:5

While DP3.1 to DP3.3 were already part of the original model, the other two
DPs were implemented in the re-designed demonstration model.
The decentralized control (FR4–DP4) as well as the traceability of mass cus-
tomized products in the demonstration model was previously solved by the use of
intelligent (NFC technology) and autonomous vehicles. FR4 can be decomposed in
the following lower level FRs and DPs:

FR4.1 = Vehicles shall be aware of their position and communicate with the CPPS;
FR4.2 = Bring mass customized products decentralized to their next processing
station;
DP4.1 = NFC technology for both communication and location awareness;
DP4.2 = Autonomously navigating vehicles with their own drive, routing
capability, and controller for every work-piece carrier.

The design matrix shows an uncoupled design:

    
FR4:1 X O DP4:1
¼ ð17:18Þ
FR4:2 0 X DP4:2

The choice for NFC is motivated by the fact that it provides superior robustness
against electromagnetic disturbances typical for an industrial production environ-
ment and the well-defined range of operation, which allows its use not only for
communication purposes but also for unambiguous position detection. Same also
for the modular vehicles built with standard components, and hence there were no
design changes in this FR–DP pair.
Regarding FR5–DP5 (user experience), the demonstration model should allow
the visitor to follow the steps in the production process from the point of view of the
product. In the previous demonstration model this function was not integrated.
There is no need to decompose this FR–DP pair any further, as DP5 can be
implemented through a standard camera mounted on the vehicle. The livestream of
the camera was realized by transmission to the information screen and allows now
the visitors following the production process from loading the apple, rotating the
apple in the right position, laser engraving, and ejection in the delivery station.
Most of the improvements are reflected in the layout of the demonstrator
illustrated in Fig. 17.11. In the re-designed demonstration model there is a clear
distinction between active waiting for jobs (green), active waiting for delivery
(purple), and standby in the buffer line (blue) vehicles. The loading mechanism not
17 Application of Axiomatic Design for the Design of Flexible … 517

Fig. 17.11 System architecture of the re-designed CPPS demonstrator. (Reproduced from Egger
et al. (2017), originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/
matecconf/201712701016)

only grips the apples and places them on the vehicles, but also aligns them so that
the rotation mechanism of the vehicles becomes obsolete. Hence, the vison process
is now part of the loading. The operator is relieved from repetitive loading and may
assist visitors. In addition, visitor may place orders remotely.
Exercise 17.5: Re-design of the Demonstrator Considering Artificial Intelligence
Artificial intelligence (AI) is the next trend that will strongly influence production in the
coming years. The task is to demonstrate the possibilities of AI based on the current
demonstration model. Discuss this in a small group and define the CNs resulting from this
brainstorming. Derive then the top-level FRs from them and perform a decomposition to
obtain possible DPs for a combination of an AI and CPPS demonstrator.
518 D. T. Matt and E. Rauch

17.8 Conclusions

As this chapter has shown, AD is a highly suitable design theory that can be applied
to the design of complex tasks in manufacturing systems design. The various
examples and case studies also show the many possible uses of AD. Thus, both
very technical problems regarding the design of individual machines or production
cells can be addressed as well as more extensive projects for the design of the entire
production system with additional organizational or strategic aspects. With the
increasing introduction of topics such as Industry 4.0, Digitization and AI, the
design of manufacturing systems will very often be expanded to include topics
relating to software instruments for the acquisition, analysis, and evaluation of
production data. The case studies have also shown that AD provides very valuable
input here. Students should get an insight through the different case studies how AD
can be applied in the most different questions in production and how an AD project
can be handled in its approach. The different exercises at the end of each case study
should also encourage the students to think along.

Problems
Students should choose two of the exercise problems given in the main text of this
chapter and work on them.

References
AlGeddawy TN, ElMaraghy HA (2009) Changeability effect on manufacturing systems design,
Springer, Chap 15. Series in Advanced Manufacturing
Bergmann L (2010) Nachhaltigkeit in Ganzheitlichen Produktionssystemen. Vlukan, Essen (in
German)
Cochran DS, Kim YS (2010) Collective system design in systems engineering education.
American Society for Engineering Education
Cochran DS, Arinez JF, Duda JW, Linck J (2001) A decomposition approach for manufacturing
system design. J Manufact Syst 20(6):371–389
Cochran DS, Arinez JF, Collins MT, Bi Z (2017) Modelling of humanmachine interaction in
equipment design of manufacturing cells. Enterp Inf Sys 11(7):969–987
Cochran JJ, Smith (2018) A Systematic Design Approach to Manufacturing Education. Procedia
Manufacturing, 26:1369–1377
Delaram J, Valilai OF (2018) An architectural view to computer integrated manufacturing systems
based on axiomatic design theory. Comput Ind 100:96–114
Durmusoglu MB, Satoglu SI (2011) Axiomatic design of hybrid manufacturing systems in erratic
demand conditions. Int J Prod Res 49(17):5231–5261
Egger G, Rauch E, Matt DT, Brown CA (2017) (re-)design of a demonstration model for a flexible
and decentralized cyber-physical production system (cpps). In: Slǎtineanu L (ed) 11th
International Conference on Axiomatic Design (ICAD), MATEC Web of Conferences, Iasi,
Romania, p 01016
Farid AM (2017) Measures of reconfigurability and its key characteristics in intelligent
manufac-turing systems. J Intell Manuf 28(2):353–369
Gualtieri L, Rauch E, Rojas R, Vidoni R, Matt DT (2018) Application of axiomatic design for the
design of a safe collaborative human-robot assembly workplace. In: Puik E, Foley JT,
Cochran D, Betasolo M (eds) 12th International Conference on Axiomatic Design (ICAD),
17 Application of Axiomatic Design for the Design of Flexible … 519

MATEC Web of Conferences, Reykjavík, Iceland, vol 223, p 01003, doi: https://doi.org/10.
1051/matecconf/201822301003
Holzner P, Rauch E, Spena PR, Matt DT (2015) Systematic design of sme manufacturing and
assembly systems based on axiomatic design. Procedia CIRP, Elsevier 34:81–86
Kulak O, Cebi S, Kahraman C (2010) Applications of axiomatic design principles: a literature
review. Expert Syst Appl 37(9):6705–6717
Matt DT, Rauch E (2011) Continuous improvement of manufacturing systems with the concept of
functional periodicity. Key Eng Mater 473:783–790
Matt DT, Rauch E (2016) Design and implementation approach for distributed manufacturing
networks using axiomatic design. In: Farid AM, Suh NP (eds) Axiomatic design in large
systems, Springer, Cham, chap 9, pp 225–250
Puik E, Telgen D, van Moergestel L, Ceglarek D (2017) Assessment of reconfiguration schemes
for reconfigurable manufacturing systems based on resources and lead time. Robot
Comput-Integ Manufact 43:30–38
Rauch E, Matt DT, Dallasega P (2016) Application of axiomatic design in manufacturing system
design: a literature review. In: Liu A (ed) 10th International Conference on Axiomatic Design
(ICAD), Procedia CIRP, Elsevier ScienceDirect, Xi’an, Shaanxi, China, vol 53, pp 1–7
Rauch E, Morandell F, Matt DT (2019) AD design guidelines for implementing I4.0 learning
factories. Proc Manufact 31:239–244
Rauch E, Spena PR, Matt DT (2019) Axiomatic design guidelines for the design of flexible and
agile manufacturing and assembly systems for smes. Int J Interact Des Manufact (IJIDeM) 13
(1):1–22
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Vinodh S, Aravindraj S (2012) Axiomatic modeling of lean manufacturing system. J Eng Des
Technol 10(2):199–216
Wiendahl HP, ElMaraghy HA, Nyhuis P, Zäh MF, Wiendahl HH, Duffie N, Brieke M (2007)
Changeable manufacturing-classification, design and operation. CIRP Ann 56(2):783–809
Zäh MF, Mäller N, Vogl W (2005) Symbiosis of changeable and virtual production the emperor’s
new clothes or key factor for future success. In: Proceedings (CD) of the international
conference on changeable, agile, reconfigurable and virtual production, Munich, Germany
Design of the Assembly Systems
for Airplane Structures 18
Mustafa Yurdakul, Yusuf Tansel İç, and Osman Emre Celek

Abstract
This chapter deals with the design of automated assembly systems of airplanes.
Axiomatic Design (AD) methodology is used to develop a general design
hierarchical tree for the assembly of an airplane fuselage panel. Introduction of
new and advanced machinery and assembly equipment in material handling,
measurement, robotics, plant modeling and simulation, manufacturing processes
require the usage of systematic methodologies for the design of assembly
systems. In this chapter, multiple alternative design solutions are demonstrated
from the general hierarchical tree, which are compared for their acceptability in
satisfying various factory objectives. In two case studies presented in this
chapter, design solutions are obtained and described in detail. The first assembly
system alternative consists of the following: (1) an automated laser tracking
measurement system, (2) reconfigurable fixtures for fixing the panels, (3) frame
clip riveters in frame assembly cells, (4) 3D projection devices to control the
position of the fastener on the panels, (5) autonomous mobile robots for material
handling, and (6) automatic riveting machines for the final fuselage panel
assembly. The second assembly system alternative includes the following:

M. Yurdakul
Faculty of Engineering, Department of Mechanical Engineering, Gazi University, Eti Mah.
Yükselis Sokak, No: 5, 06570 Maltepe/Ankara, Turkey
e-mail: [email protected]
Y. T. İç (&)
Faculty of Engineering, Department of Industrial Engineering, Baskent University, Bağlıca
Kampüsü, Fatih Sultan Mahallesi, Eskisehir Yolu 18.km, 06790 Etimesgut/Ankara, Turkey
e-mail: [email protected]

O. E. Celek
Turkish Aerospace Headquarter, Havacılık Bulvarı No: 17, 06980 Fethiye Mahallesi,
Kahramankazan Ankara, Turkey
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 521

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_18
522 M. Yurdakul et al.

(1) robotic measurement systems, (2) robotic stringer placement robots for
stringer positioning, (3) reconfigurable fixtures for holding the panels, (4) frame
clip robots to assemble clips to frames, (5) 3D projection devices for fastener
position controls on the panels, (6) modular crane for material handling, and
(7) mobile automated drilling and fastening robots for final fuselage panel
assembly. These two alternative designs are compared, and results are made
available for system designers to assess the capabilities of each alternative. The
application of the proposed design methodology provides a reference guide for
system designers to apply in designing assembly systems in an aerospace
assembly factory.
After studying this chapter, the reader should know the following:
(1) How we can design airplane fuselage panel assembly systems. (2) How to
develop a general assembly system design hierarchy tree. (2) The clear
understanding of functional requirements (FRs) and design parameters (DPs) at
each level of the assembly system design hierarchy tree. (4) How to obtain
various alternative solutions from design hierarchy tree of the assembly system.
(5) Using simulation to compare alternative design solutions.
One of the Most Important Questions to Think About throughout This
Chapter:
This chapter presents multiple alternative solutions instead of a single solution.
Although with multiple solutions the general decision tree is larger and more
widespread, it lets the system designers to see all solutions. The system designer
can make a more thorough comparison among the solutions and then make
recommendations for final selection.

18.1 Introduction

Aerospace manufacturers are bursting at the seams with a backlog of orders. To

remain competitive, aerospace manufacturers must reduce time-to-market on new
airplane variants, improve capability, increase flexibility and shorten lead time to
reduce backlog on current orders (Groover 2007). One viable solution is to auto-
mate their factories. The industry is investing heavily in systems that reduce cost,
improve quality, and boost productivity. Systems include lightweight fixtures,
reconfigurable tools, automated part positioning, automated scanning, countersink
control, automatic riveting, and robotic measurement. Furthermore, investments are
made in state-of-the-art robots for drilling, riveting, sealing, coating, and painting
applications, in addition to material handling, carbon fiber layup, and advanced
machining operations.
Airlines will need 38,050 new airplanes valued at more than $5.6 trillion in the
near future. Asia Pacific region (China, Northern Asia, South Asia, Southeast Asia,
and Oceania) is expected to have the largest expected deliveries in total as shown in
Fig. 18.1.
18 Design of the Assembly Systems for Airplane Structures 523

Fig. 18.1 New airplane

deliveries by region (2015–
2034) 1150, 3% Asia
1170, 3%
3020, 8%
North
America
3180, 8% Europe
World Total 14330, 38%
38.050
Middle East

7310, 19% Latin

America
Africa
7890, 21%

Based on the growth trends in air travel demand, airplane production volumes
have been increasing over the years, as shown in Fig. 18.2 especially for
single-aisle airplanes such as A320 or B737. Boeing increased the production of its
popular narrow-body single-aisle 737 jets from 42 airplanes to 52 in 2018. The
company plans to ramp up production volume on the more technically complex
B787 to 14 airplanes per month by 2020. The aerospace giant currently has more
than 5,700 airplane on order, valued at more than $450 billion. Boeing’s archrival,
Airbus, also plans to boost output of its most popular airplane, the A320. Airbus
expects to ramp up production rates from 42 planes a month to 50 units. Airbus
plans to crank up A320 production to 60 planes a month at its factories in China,
France, Germany, and the United States. That kind of production volume would be
unprecedented in the commercial aerospace industry (Weber 2015).
Assembly lines at major suppliers, such as GE Aviation, Honeywell, Rockwell
Collins, Spirit Aero Systems, and UTC Aerospace Systems increased production
rate. Pratt & Whitney invested $1 billion worldwide to prepare its factories for jet
engine production increases. The facility featured an overhead horizontal moving
assembly line in order to produce the Pure Power PW1100G-JM engine for the
Airbus A320neo (Weber 2015).
Automation in assembly lines of aerospace companies is necessary for mini-
mizing hand assembly operations in order to increase the efficiency. Benefits of
automation include lower manufacturing costs, increased quality levels, consistent
and higher throughput rates, and fewer repetitive motion injuries. When all the
benefits are added, automation is continually replacing human workers not only in
the final assembly line but also in the sub assembly lines.
The assembly centers within airplane factories are the focal points for collecting
all the varied pieces of an airplane and bringing them together to be joined into an
airframe. The airframe is approximately 75% of the cost of the airplane; assembly is
approximately 65% of the cost of the airframe; and drilling and fastening is
approximately 65% of the labor cost of assembly. The greatest opportunity to
reduce costs is based on unchanged processes and activities related to the assembly
line of the airframe (Bullen 2013).
524 M. Yurdakul et al.

30000
26730, 70%

25000

20000

15000

10000

4770, 13%
5000 3520, 9%
2490 ,7%
540, 1%
0
Regional jets Single-aisle Small wide-body Medium wide-body Large wide-body

Fig. 18.2 Airplane Deliveries: 38.050 (2015–2034)

The fuselage is the main structure or body of the airframe of an airplane. It

provides space for cargo, controls, accessories, passengers, and other equipment. In
single-engine airplane, the fuselage houses the power plant. In a multiengine air-
plane, the engines may be either in the fuselage, attached to the fuselage, or sus-
pended from the wing structure. Wings, tail cone, vertical fin, and engine are
attached to the fuselage in an airplane. In some airplane designs, the engines and
landing gear are also attached directly to the fuselage structure. In modern airplanes,
as shown in Fig. 18.3, the fuselage takes the form of a tube, which houses the flight
deck, passenger cabin, freight holds, and the majority of the equipment required to
operate the airplane. For passenger airplanes that fly above 10,000 ft. the fuselage
also forms a pressure hull so that a cabin altitude of 8,000 ft. can be maintained
throughout a normal flight (Federal Aviation Administration 2018). The airplane
fuselage consists of six different panels, namely, lower, lower left, lower right,
upper, upper left, and upper right.
A typical airplane fuselage panel is mainly composed of skin, stringer, clips,
frame, and other support parts. Main components of a fuselage panel assembly are
presented in Fig. 18.4. Skins carry the major stresses and functions as the main part
of the airframe. Stringers are the main lengthwise members of the fuselage panel
structure. Stringers work in combination with the skin to create a strong beam.
Stringers prevent tension and compression from bending the fuselage. They have
some rigidity but are chiefly used to give the fuselage shape and attach the skin.
18 Design of the Assembly Systems for Airplane Structures 525

Fig. 18.3 Airplane structural components

Fig. 18.4 Main components of a fuselage panel assembly

526 M. Yurdakul et al.

Frames on the other hand serve a dual purpose, they give cross-sectional shape
to the fuselages, and also they add rigidity and strength to the structure. The shape
and size of these members vary considerably depending on their function and
position in the fuselage. They have multiple functions and break up skin panels so
that they behave as panels supported on all four edges. They distribute applied loads
created by doors and fairings and maintain the cylindrical shape of the skin. Frame
assemblies are the most numerous and important members of the fuselage. Clips are
structural elements that connect stringers to skins. Also, frames are attached to the
stringers with stringer clips. Sometimes frames attach directly to the skin by means
of shear ties (Jeppesen Sanderson, Inc. and Atlantic Flight Training, Ltd. 2007).
All these members are delivered to fuselage panel assembly center and are joined
together to form fuselage panel assemblies. In this study, AD principles are applied to
develop alternative assembly systems for a fuselage panel assembly.
The most relevant studies to the content of this chapter are briefly discussed as
follows:
An AD methodology that relates production system design objectives to opera-
tional DPs was developed at Massachusetts Institute of Technology by Reynal and
Cochran. The methodology also focuses on the design of production operations by
eliminating non-value-added operations. Internal and external customer requirements
are also taken into consideration with this methodology. It is used to increase the
efficiency of manufacturing cells in the assembly areas of two different manufacturing
companies. AD axioms and decomposed processes are analyzed in existing and new
manufacturing systems. As a result of the study, the part stocks in the first manufac-
turing company are decreased, the processing time of the detail parts is reduced by
50%, the assembly process time is reduced by 60%, and production area reduced by
40%. Production of a higher quality final product is provided. In the second manu-
facturing company, bottlenecks in manufacturing area are solved. Some assembly
fixtures are re-designed and loading process time is reduced by 30 s per piece. The
single-piece flow system is integrated and U shape arrangement is applied in the
assembly cells. Work in process time reduced by 80%. The objective for the production
of 400 pieces is achieved (Reynal and Cochran 1996).
Kulak et al. (2005) developed an effective methodology of converting a
process-based manufacturing system into a cellular one in a conventional manu-
facturing firm. Their study reduces average material handling distance from 67 to
31 m and delivery time from 18 to 7 days. The biggest improvement is observed in
80% reduction of overtime.
Jefferson et al. (2016) addressed the design complexity issue by creating design
methodology for a novel reconfigurable assembly system. Their methodology is a
holistic, hierarchical approach to system design which integrates reconfigurability
principles, AD, and design structure matrices. A wing assembly case study is used
to illustrate how the methodology translates reconfigurability requirements into a
system that is scalable and flexible from the outset. The resultant reconfigurable cell
design assembles the wing's spars and ribs with ramp up capability from 40 to 100
airplane per month.
18 Design of the Assembly Systems for Airplane Structures 527

18.2 Case Study: Airplane Fuselage Panel Assembly

with Axiomatic Design Principles

This section presents a case study of the assembly of the fuselage panel to develop a
cost-effective assembly system to ramp up to future production rates based on the
AD methodology.
Figure 18.5 shows typical airplane fuselage sections. Airbus and Turkish
Aerospace have signed the contract related with the manufacturing of Single-Aisle
(A319/320/321) Airplane Section 19. Turkish Aerospace started manufacturing of
Section 19 shells under the first phase of the program and started deliveries of these
shells to Premium Aerotech Augsburg facilities. At the second phase, in addition to
the shells, Turkish Aerospace will also manufacture Section 19 barrels and deliver
these barrels to RUAG Oberpfaffenhofen facilities (Turkish Aerospace 2020).
Based on the contracts signed by the Turkish Aerospace, the customer needs and
input constraints for the assembly of airplane fuselage structures are as follows:

• rate 5–20 airplane fuselages per month for the first ramp up;
• use certified processes and materials;
• meet all Airbus regulations and certifications;
• limited assembly area.

Fig. 18.5 Airplane fuselage section

528 M. Yurdakul et al.

Customer needs and input constraints convert FRs into (DPs) and process
variables (PVs) to determine design properties of the assembly system. The
decomposition process is designed to create a discrete system in order to make the
system parameters independent of each other. They are consistent with constraints
and needs. The first-level FRs and DPs for fuselage panel assembly is specified as
follows:

FR1 = Assemble Airplane Fuselage Panels at Required Rate;

DP1 = Flexible and Scalable Assembly System.

At this stage, DPs and FRs are decomposed to their next levels, which are
presented in Table 18.1. Focusing on the DPs, the methodology provides a solution
for reconfigurability characteristics. Customization and scalability are embedded in
the DPs. For second-level FRs and DPs to be consistent with the first-level and
requirements are specified. The second-level FRs and DPs are specified as follows:
As an example, Consider FR13 and DP13 shown in Table 18.1, FR13 is needed
to be satisfied by DP13 which is an automated drilling and fastening system. The
systems are generally capable of carrying out drilling, countersinking, and fastening
operations automatically by using computer-controlled numerical codes. They are
originally designed to drill and to fasten airplane parts like panels, stringers, frames,
and sections.

Table 18.1 Second-level FRs and DPs for fuselage panel assembly system
FRs-1 DPs-1
FR11: Position and locate beam component to DP11: Fixture system
fixture
FR12: Position and locate skin-stringer-clip DP12: Fixture system
components to the beam with tack rivets
FR13: Drill all components DP13: Automated drilling and fastening
systems
FR14: Remove components from the fixture DP14: Scalable manual deburr and
deburr and clean clean
FR15: Apply promoter, sealant relocate all DP15: Scalable manual promoter,
components to the fixture sealant application and relocation
FR16: Complete manual operations before final DP16: Scalable semi-automatic riveting
fastening system
FR17: Final fasten and assemble all components DP17: Automatic riveting machine
systems
FR18: Complete manual operations after final DP18: Scalable manual after machining
fastening operations
FR19: Apply paint, sealant, and varnish DP19: Scalable robotic painting, sealant
and varnish application
18 Design of the Assembly Systems for Airplane Structures 529

Also at this stage, the design equation is written to check if the system is
coupled, decoupled, or uncoupled. Since the design matrix is diagonal, the system
is an uncoupled design. The design matrix is constructed with the DPs where the
FRs are met for this level as follows:
    
 FR11  1 0 0 0 0 0 0 0 0  DP11 
    
 FR12  0 1 0 0 0 0 0 0 0  DP12 
     
 FR13  0 0 1 0 0 0 0 0 0  DP13 
     
 FR14  0 0 0 1 0 0 0 0 0  DP14 
    
 FR15  ¼ 0 0 0 0 1 0 0 0 0  DP15  ð18:1Þ
    
 FR16  0 0 0 0 0 1 0 0 0  DP16 
    
 FR17  0 0 0 0 0 0 1 0 0  DP17 
    
 FR18  0 0 0 0 0 0 0 1 0  DP18 
    
FR19 0 0 0 0 0 0 0 0 1 DP19

Design decomposition is divided into subgroups by the zigzagging method, in

which the design is expressed in more detail. It is aimed to obtain a physical and
reasonable solution. The decomposition of the parameters continues until an
existing solution becomes concrete and further decomposition is not physically
possible. For the third level, the FRs and DPs are created as follows in Table 18.2.
The Following FRs and DPs in Table 18.2, FR132 is needed to be satisfied by
DP132 which is frame assembly cell. In order to transfer holes to frame and fasten
clips to frame, a more specific frame assembly cell is required. Frame assembly cell
is capable of carrying out drilling, countersinking, and fastening clips to frame in
airplane fuselage components either by using robots or advanced drilling and fas-
tening machines.
Design decomposition proceeds with the same method and FRs and DPs are
created for the fourth level as in Table 18.3.
DPs and FRs are specified at the next level, they are decomposed to be detailed
more. The Following FRs and DPs in Table 18.3, FR1322 is needed to be satisfied
by DP1322 which is frame clip robot. Frame clip robots that can be used as an
alternative to frame clip riveters are designed for assembly of frames to clips in
different types of airplane fuselage panels. The robot moves on rail systems which
is robot’s seventh axis movement. The accuracy of the robot is ±0.3 mm,
repeatability is ±0.05 mm. The precision is achieved by calibration, pressure, and
temperature compensation. Clamping force is 3000 daN, drill speed is 80 daN and

Table 18.2 Third-level FRs and DPs for fuselage panel assembly system
FRs-13 DPs-13
FR131: Constrain and positions clips DP131: Manual operator
FR132: Transfer holes to frame and fasten clips to frame DP132: Frame assembly cell
FR133: Increase drill rate for ramp up DP133: Scalable architecture
FR134: Fasten components temporarily with tack rivets DP134: Tack riveting module
530 M. Yurdakul et al.

Table 18.3 Fourth-level FRs and DPs for fuselage panel assembly system
FRs-132 DPs-132
FR1321: Fasten with riveting machine DP1321: Frame riveter assembly cell
FR1322: Fasten with riveting robot DP1322: Frame clip robot

drill spindle speed 500–6000 rpm with a peck-drilling functions. The robots are
capable of clamping, drilling, countersinking, 5/32″, 6/32″, 7/32″, 8/32″, fastener
insertion and squeezing processes up to 7 mm thickness of components. Fig-
ure 18.6. shows drilling and fastening robot similar to the frame clip robot.
FRs and DPs are created for the fifth level as in Table 18.4.
The Following FRs and DPs in Table 18.4, FR13222 is needed to be satisfied by
DP13222 which is an all-electric riveting head. The all-electric riveting head
includes drill feed unit, rivet inserter, rivet injector, tool changer, drill spindle,
resynch camera, distance and normality sensors, rivet length measurement sensor,
drill lubricator, pressure foot, and sealant applicator.
FRs and DPs at the lowest level correspond to a specific part feature, mecha-
nism, or system. Finally, design hierarchy decomposition is illustrated as an
example up to level 6. FRs and DPs are created for the sixth and last level as shown
in Table 18.5.

Fig. 18.6 Drilling and fastening robots

18 Design of the Assembly Systems for Airplane Structures 531

Table 18.4 Fifth-level FRs and DPs for fuselage panel assembly system
FRs-1322 DPs-1322
FR13221: Move the robot for appropriate position DP13221: Rail platform
FR13222: Fasten clips to frame DP13222: All electric riveting head

Table 18.5 Sixth-level FRs and DPs for fuselage panel assembly system
FRs-13212 DPs-13212
FR132121: Position the fuselage panel DP132121: Position sensors
FR132122: Select appropriate drill bit and drill the DP132122: Drilling unit
component
FR132123: Insert rivet DP132123: Rivet inserter
FR132124: Squeeze rivet DP132124: Upper pressure foot and
lower anvil

General assembly system design hierarchy tree (Fig. 18.7) is obtained using FR
and DP information which is explained in detail above. Different alternative sys-
tems can be obtained by using the general assembly system design hierarchy tree.
Figures 18.8 and 18.9 present an alternative assembly system. This system includes
primarily an automatic riveting machine and autonomous mobile robots. On the
other hand, a second alternative which includes riveting robots and modular crane
for panel handling purposes is presented in Figs. 18.10 and 18.11.

18.3 Assembly System Design Alternatives Simulation

Results

The first assembly system design alternative includes fully automated laser tracker
for measurement systems, reconfigurable fixtures for fixing the panels, frame clip
robots in frame assembly cells for the clip to frame assembly, 3D projection devices
for fasteners position controls on the panels, autonomous mobile robots for material
handling, and automatic riveting machines for final fuselage panel assembly.
Second assembly system design alternative includes robotic measurement sys-
tems for measurement, robotic stringer placement robots for stringer positioning,
reconfigurable fixtures for fixing the panels, frame clip robots for clip to frame
assembly, 3D projection devices for fasteners position controls on the panels,
modular crane for material handling and mobile automated drilling and fastening
robots for final fuselage panel assembly.
After completion of simulations, results show that the second alternative is better
than the first alternative by production rate.
 532
M. Yurdakul et al.

Fig. 18.7 General assembly system design hierarchy tree

18 Design of the Assembly Systems for Airplane Structures 533

Fig. 18.7 (continued)

 534
M. Yurdakul et al.

Fig. 18.8 Assembly system design alternative-1

18 Design of the Assembly Systems for Airplane Structures 535

Fig. 18.8 (continued)

536 M. Yurdakul et al.

Fig. 18.9 Simulation of assembly system design alternative-1

• First alternative: 138 fuselage panels per month.

• Second alternative: 153 fuselage panels per month.

Koren and Shpitalni (2010) developed a comparison table for combine features
of reconfigurable/flexible manufacturing systems (dedicated or random order).
A new table is created for comparison results of assembly system design alterna-
tives for airplane fuselage panels (Table 18.6).

18.4 Conclusion

A novel assembly system design methodology is developed by using AD principles

in order to propose a solution to design airplane fuselage panel assembly. The
framework of design methodology is developed based on system design methods,
academic research, industry requirements, and industrial case studies. An Airbus
A320 airplane Section 19 fuselage panel assembly case study is carried out for a
better understanding of how the methodology is applied. Customer requirements are
transformed into FRs, and FRs are transformed into DPs. The DPs are detailed in
the design structure matrix and related sub-functions are created. Different alter-
native systems can be created by using the design hierarchy tree. Accordingly, two
system alternatives for airplane fuselage panel assembly are presented to illustrate
18 Design of the Assembly Systems for Airplane Structures 537

Fig. 18.10 Assembly System Design Alternative-2

 538
M. Yurdakul et al.

Fig. 18.10 (continued)

18 Design of the Assembly Systems for Airplane Structures 539

Fig. 18.11 Simulation of assembly system design alternative-2

Table 18.6 Comparison of Assembly system design for

the assembly system design airplane structures
alternatives
Alternative-1 Alternative-2
System structure Fixed Adjustable
Machine structure Fixed Adjustable
Customizationa Yes Yes
Convertibilityb No Yes
Scalabilityc Yes Yes
Modularityd No Yes
Integrabilitye No Yes
Diagnosabilityf No Yes
Reconfigurability Customized Customized
Throughput High Very high
Productivity High Very high
Lifetime cost High High
Configuration time High Medium
Time loss Low Very low
Customizationa: flexibility limited to part family
Convertibilityb: design for functionality changes
Scalabilityc: design for a capacity change
Modularityd: components are modular
Integrabilitye: interfaces for rapid integration
Diagnosabilityf: design for easy diagnostics
540 M. Yurdakul et al.

the application of AD methodology. This case study is aimed to be used as a

reference guide for the design of similar assembly systems in the aerospace
industry.
Problems
1. What can be the system-wide FRs and DPs that can be put at the first level of a
design hierarchy for an assembly system?
2. Return back to the case study. Develop a hierarchy tree that provides a single
system solution instead of multiple ones. Compare your result with the solutions
obtained in the case study. Discuss whether it is better to limit your solution
space in developing your decision hierarchy.
3. What advantages did the usage of AD methodology bring to the assembly
system design?
4. Does the proposed approach always encourage usage of automated and
advanced but more expensive technologies? Under which conditions more
conventional technologies can be recommended for usage in assembly systems?
5. One of the authors stated that “Application of the AD Approach required a
detailed and complete review of all technologies that are used/have potential to
be used in a fuselage panel assembly system.” The authors spent more time and
effort in this review process than the development of the hierarchy tree. Test the
correctness of the above statement by applying AD approach to a simple
assembly system?
6. Are there any other design approaches available in the literature that can be used
in development of alternative assembly system solutions? Try to apply those
approaches to the same case study.

References
Bullen GN (2013) Automated/mechanized drilling and countersinking of airframes. SAE
International, Warrendale, Pennsylvania, USA. https://www.sae.org/publications/books/
content/r-416/, product code: R-416
Federal Aviation Administration (2018) Aviation Maintenance Technician Handbook—Air-frame,
vol 1. US Department of Transportation. https://www.faa.gov/regulations_policies/handbooks_
manuals/aircraft/media/amt_airframe_hb_vol_1.pdf, document FAA-H-8083–31
Groover MP (2007) Automation, production systems, and computer-integrated manufacturing, 3rd
edn. Prentice Hall, Upper Saddle River, NJ, USA
Jefferson TG, Benardos P, Ratchev S (2016) Reconfigurable assembly system design method-
ology: a wing assembly case study. SAE Int J Mater Manufact 9(1):31–48. https://doi.org/10.
4271/2015-01-2594
Jeppesen Sanderson, Inc, Atlantic Flight Training, Ltd (2007) Airframes and systems: JAA ATPL
training, 2nd edn. No. 4 in ATPL series, Jeppesen Sanderson, Inc., Neu-Isenburg, Germany
Koren Y, Shpitalni M (2010) Design of reconfigurable manufacturing systems. J Manufact Syst 29
(4):130–141. https://doi.org/10.1016/j.jmsy.2011.01.001
Kulak O, Durmuşoğlu B, Tufekci S (2005) A complete cellular manufacturing system design
methodology based on axiomatic design principles. Comput Indus Eng 48(4):765–787. https://
www.sciencedirect.com/science/article/pii/S0360835204001901
18 Design of the Assembly Systems for Airplane Structures 541

Reynal VA, Cochran DS (1996) Understanding lean manufacturing according to axiomatic design
principles. The Lean Aircraft Intiative Report Series, URL https://axiomaticdesign.com/
technology/papers/RP960728Cochran_Reynal.pdf, document RP96–07–28
Turkish Aerospace (2020) AIRBUS—PAG SA Section 19 Shells&Barrel Program. https://www.
tai.com.tr/en/product/airbus-pag-sa-section-19-shellsbarrel-program
Weber A (2015) Assembly automation takes off in aerospace industry. Assembly Magazine.
https://www.assemblymag.com/articles/92790-assembly-automation-takes-off-in-aerospace-
industry
Healthcare System Design
19
Inas S. Khayal and Amro M. Farid

Abstract
This book includes several design applications in several fields: energy, large
systems, manufacturing, and many more. In this chapter, we turn our focus to the
design of healthcare systems. But first, we ask what exactly do we mean by
healthcare systems? This chapter will describe the large scale and scope of what
is meant by the phrase “healthcare systems.” A broad definition of Healthcare
System Design consequently leads to the recognition of the large number of
stakeholders affected by such designs. The need for appropriate, effective,
timely, sustainable, and cost-effective healthcare services has been recognized
globally, and has led to an increasingly growing healthcare sector. As the
healthcare sector continues to expand, there are several unique challenges that
have risen due to the many siloed designers acting on different parts of the
healthcare system. The goal of this chapter is to orient the reader to the various
scale and scope of healthcare systems and to present a structure that allows the
reader to thoughtfully consider the design process relative to stakeholders,
system boundary, system functions for functional requirement (FR) selection,
system form for design parameter selection (DP), and consequently the process
variables (PV).

I. S. Khayal (&)
The Dartmouth Institue for Health Policy and Clinical Medicine and Biomedical Data Science,
Geisel School of Medicine at Dartmouth, Lebanon, NH, USA
e-mail: [email protected]
I. S. Khayal
A. M. Farid
Department of Computer Science, Dartmouth College, Hanover, NH, USA

A. M. Farid
Thayer School of Engineering at Dartmouth, Hanover, NH, USA

A. M. Farid
MIT Mechanical Engineering, Cambridge, MA 02139, USA

© Springer Nature Switzerland AG 2021 543

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_19
544 I. S. Khayal and A. M. Farid

19.1 Introduction

19.1.1 Motivation

As a reader of this book, your career may include designing across multiple sectors
or may focus on a single sector. This chapter will focus on healthcare. While it may
seem like a niche service, healthcare is currently the largest employer in the United
States, accounting for about 18% of the gross domestic product (GDP) or an
expenditure of $11,000 per person in 2018. Why is that?
People are living longer today than in any time in human history. The aging
population along with a continuous growth in medical spending has led to an
increased rate of healthcare jobs in the United States and globally. Interestingly,
although science and technology have allowed for significant advancements in
health treatments, humanity is facing an unprecedented chronic disease burden.
While we may live longer, most will struggle with many more long-duration
chronic diseases—such as diabetes, heart disease, autoimmunity, and cancer. Such
a significant disease burden has caused the healthcare sector to expand to address
the many challenges that arise when treating or managing health conditions today.
This sector is one of the most labor-intensive fields, requiring expensive medication
and individualized services, and highly educated service personnel.

19.1.2 Chapter Outline

Section 19.2 starts with how to define the healthcare system. Healthcare design has
many varying levels of scale and scope. Therefore, understanding and identifying
the scale and scope is critical to a successful design [Sect. 19.2.1]. Then,
Sect. 19.2.2 presents a framework for thinking about what the healthcare system
does (functions) and who or what performs these functions (resources). While
many engineered designs are for stock products, healthcare designs are in many
cases customized and will vary by region, country, culture, or even individual
patients. Understanding the variation in FRs and/or DPs in this broad context is
discussed in Sect. 19.2.3.
Section 19.3 discusses the characteristic and often unique challenges found
within healthcare systems. The discussion addresses the significant heterogeneity of
function and form even within a specific country, region, or hospital [Sect. 19.3.1].
While many engineered systems are typically defined as technical systems,
healthcare is fundamentally socio-technical. Unique implementation and outcome
challenges emerge as a result [Sect. 19.3.2]. Despite the range in scale, scope,
heterogeneity, complexity, and regional impact of healthcare services, there is a
need to interconnect healthcare systems. Consequently, designing for interoper-
ability is a critical consideration for success [Sect. 19.3.3]. Finally, healthcare is a
legacy system. It is hard to change what has been already designed—be it technical
19 Healthcare System Design 545

in nature or social in nature. Therefore, understanding the existing system, which of

its aspects are flexible, and why is a critical consideration to healthcare design
[Sect. 19.3.4]. Finally, Sect. 19.4 summarizes the key takeaways and conclusions.

19.2 Defining a Healthcare System

Defining a healthcare system includes identifying the system boundary, the scale
and scope of the problem, and a framework to organize the system. As described in
previous chapters, with a dedicated chapter to the topic (Chap. 5: Problem Defi-
nition), accurately identifying the problem is absolutely critical. This section will
stress the importance of accurately identifying the problem and consequently help
the design reader to distinguish between “what is designed” and the remainder of
the healthcare system.

Design Story 19.1: Pacemaker

In 1951, a Boston cardiologist is given credit for ushering in the modern era of
clinical cardiac pacing, the fitting of a device—a pacemaker—to maintain a regular
heart rhythm. Since then, cardiac pacemakers have advanced tremendously. One of
the primary customer requirements for a permanent (internal to the body) pace-
maker is its ability to function for a long period of time, to avoid further invasive
surgeries to the patient. This customer requirement appears as a design constraint on
two DPs: the size of the battery and the power consumption of the pulse generator.1
Traditional biomedical engineering design devotes much effort to continuously
improve on this customer requirement. Despite these efforts, the traditional pace-
maker design neglects an important FRs, namely, to turn it off!
A classic “device-engineer” may wonder, “why would you ever need to do
that?” In reality, the complex needs of today’s diverse cardiac patients have made
this FRs evermore critical to not just the patient, but also their entire healthcare
team. Unfortunately, such a requirement did not arise from a sterile stakeholder
requirement identification process but rather from design failure and emotional
heart break.
Although people are living longer today, they are often doing so with chronic
as-yet incurable diseases. For example, dementia is a progressive condition where
the brain deteriorates, and the individual loses their ability to remember. Such a
condition is emotionally, socially, and physically taxing on both the individual and
their family. Many patients with dementia, typically in their 60s, 70s, or 80s, also
tend to have deteriorating heart conditions. Many of them received a pacemaker
prior to dementia or during its early onset. Some patients believe that as dementia

1
Interestingly, this design constraint creates an inevitable coupled design between the battery and
the pulse generator. Such a coupled design appears in many applications where “on-board” energy
storage and the need for miniaturization create an energy budget and a constraining volume for the
device as a whole.
546 I. S. Khayal and A. M. Farid

comes to a late stage, it can cause so much suffering that they would prefer not to
prolong their lives with artificial measures like cardiac devices. In the United States,
patients at least have the right to refuse or discontinue treatment, including a
pacemaker that keeps them alive. Physicians also have a right to refuse to turn it off.
These two seemingly contradictory legal rights have raised a significant ethical
question. Is it ethical to deactivate a pacemaker in patients with a chronic debili-
tating condition given that its use in such patients significantly extends their life, but
at a questionable quality of life for both the patient and their family? Many engi-
neers would consider this a “messy” ethical or legal issue. However, in reality, the
identification of the “turn-off” FRs is a fundamental engineering design question
that would have alleviated many stakeholder ramifications. From a design per-
spective, we can include an FR to “turn-off” the device with a corresponding
DP. Where the issue becomes difficult is setting up a corresponding DP that
addresses the various types of individuals that may turn-off the device (e.g., clin-
icians, family members, etc.) and a thorough protocol to do so that considers the
ethical, moral, and religious questions of who makes the decision to “turn-off” the
device. Therefore, the design of healthcare protocols and procedures that resolve
these ethical questions is an integral part of healthcare organizational design.

19.2.1 Healthcare System Boundary, Scope, and Scale

The concept of the system implies and requires a boundary that distinguishes it
from everything else. Such a boundary can be defined with varying scope. For
example, one may be interested in reducing knee pain, which may be stated as an
FR. A designer can create one or more systems defined with increasing scope: “a
knee implant,” “a knee replacement surgical suite,” “a knee replacement service
including the surgical suite and surgical team,” “an orthopedic clinic with integrated
physical therapy services,” or “a hospital”. Each of these scopes provide additional
functions that ultimately serve to reduce knee pain. The logic here is that while a
knee implant can reduce knee pain, it has limited efficacy without the technical
facilities and human resources to conduct the knee replacement, the post-surgery
follow-up care, and the integration with other healthcare services that may be
required in the event of complications. A designer may be responsible for designing
a healthcare system with any of these degrees of scope. Consequently, the choice of
a widening system boundary may result in a design that could be described as a
“product,” “software,” “both product and software,” and “organizations” (see
Chap. 16).
Unfortunately, the designer may not necessarily have control over the delin-
eation or the clarity of the system boundary. Its specification may be set externally,
by the design-client organization, by politics, or even healthcare regulation laws.
While external specification of the system boundary alleviates this design choice
from the designer’s responsibility, it does not alleviate the burden of ensuring that
the to-be-designed-system integrates well into the larger healthcare system and
produces the expected health-outcome results. In short, and like many other
19 Healthcare System Design 547

large-scale systems, the designer is only responsible for the “system” but is actually
accountable for its integration with a larger “whole-system.”
The choice of the system boundary and scope also implies the scale of the
system design. As the designer broadens the scope of their healthcare system, they
are inevitably forced to view the system’s function and form at a fairly high level of
abstraction. Rather than focusing on the individual features of a given device, they
must now consider entire healthcare system resources be they operating rooms,
departments, or entire healthcare facilities. Once the designer has chosen such a
scale, the system can then be decomposed iteratively into its component parts,
starting from the highest FRs and DPs to a lower level FRs and DPs. Once the
decomposition is done up to a point without introducing any coupling of FRs, the
lower level FRs can be parceled out and given to different groups for further
decomposition and implementation. This is how complicated systems such as air-
planes (see Chap. 18) and wireless electric buses (i.e., On-Line Electric Vehicle
(OLEV) (Suh and Cho 2017)) have been created. The important thing to implement
in this process is the careful monitoring of the design tree and the creation of the
Master Design Matrix discussed in Chap. 3 to quickly identify any coupling in the
design that may be introduced by one of the participants in the design group.
It is generally accepted that after 1000 components have been identified, the
system can no longer be practically designed by a single designer. In such a case,
the designer must form a design team to address different part of the system which
can be later integrated. At a scale of 10,000 components, the system can be referred
to as a system-of-systems that inevitably requires multiple design teams. In such a
situation, the coordination of the design effort becomes as significant as the design
itself. The possibility of introducing coupling due to the decisions made by a
designer who is working on other branches of decomposition exists, which can be
prevented if the team under the system architect follows the entire process and
constructs the master design matrix to identify any inadvertent coupling of FRs.

Design Story 19.2: Endovascular Aortic Aneurysm Repair Device

A company in California manufactures an internal implant called an endograft used

in a procedure called an endovascular aortic aneurysm repair (EVR). EVR was
developed almost three decades ago as a minimally invasive alternative to a surgical
procedure to address abdominal aortic aneurysms (AAA)—a bulge in the wall of
the main blood vessel leading away from the heart to the abdomen. They inno-
vatively designed their endograft to have the main body stabilized by resting
directly on the aortic bifurcation rather than using hooks for fixation with the fabric
of the endograft affixed to the stent skeleton only at the top and bottom of the
device. This design created a novel sealing concept that reduces the possibility of
device migration and leakage (i.e., reducing the risk of what is considered a type 1
endoleak—flow around the proximal or distal seal zones of the endograft).
At the time of initial manufacturing, the graft was a stratra fabric. Upon intro-
duction into the healthcare delivery system, it yielded increased reports of Type III
endoleaks (i.e., those caused by defects in the endograft device) for patients with
548 I. S. Khayal and A. M. Farid

significant aortic remodeling. Endoleaks are a type of uncommon but nevertheless

serious late stage complication that has a high risk of rupture and death. A study
found that there was a 24% chance of Type III endoleaks due to fabric breakdown.
The company consequently provided a safety notice to providers. The United States
Food and Drug Administration (FDA) classified this notice as a Class I recall (i.e.,
the most serious). Since then, the company has updated their graft material to a new
formulation.
This design story demonstrates how a new and unique biomedical product can
be completely successful upon first use, but later have unintended life-threatening
consequences as the human biology (in the form of aortic remodeling) adapts to the
integration. Furthermore, the device’s novelty created an endoskeleton design that
may complicate guidewire and device entry on redo procedures. Such a design
considers the biological integrity of an initial procedure, but neglects the very
realistic potential for complications in this relatively minimally invasive EVR
procedure.
The above discussion emphasizes the importance of the system boundary in
healthcare system design. Not only does it affect the scope and scale of the design,
but it also requires a deep appreciation of how the system will affect and be affected
by its introduction into the healthcare delivery system. The chapter now provides a
formal systems framework for the description of a healthcare system.

19.2.2 Healthcare System Function and System Form

In addition to the system boundary, a system is defined by what it does—its system

function, and how these functions are performed—its system form. The system
function is made up of a set of functions that fulfill the system’s FRs. The system
form (in large-scale systems) is made up of a set of resources that are described by
DPS.
An understanding of system function and form is now particularly important as
the designer begins to understand the state of the healthcare system today. His-
torically, the healthcare system has been organized based on form and not function.
A typical hospital has a departmental structure that addresses specific body parts
(e.g., Rheumatology [Joints], Neurology [Brain], Cardiology [Heart], Endocrinol-
ogy [Hormones], etc.). Similarly, at a higher level of aggregation, healthcare sys-
tems are also defined by location. For example, a typical hospital has inpatient and
outpatient wings to distinguish between patients that will or will not stay overnight.
Furthermore, within these wings and departments, healthcare is typically
described by who provides the function, rather than the function itself. Simply
speaking, when someone needs care, they go to see a general physician, surgeon, or
psychiatrist. In other words, the hospital is organized by system resources rather
than the system functions that they provide. While it seems intuitive to describe
surgery-care by the resource performing the surgery (i.e., the surgeon), from a
system design perspective, such a description omits the inclusion of the
19 Healthcare System Design 549

care-function itself! For example, surgery can be defined as “a procedure that

involves cutting of a patient’s tissues or closure of a previously sustained wound.”
Furthermore, the Merriam-Webster Medical Dictionary defines surgery as a noun
and not a verb. To a lay patient audience, this system resource-based description of
healthcare may be appropriate. However, it is pervasive and appears as an integral
part of clinical models used in professional practice and academic research (e.g.,
behavioral health models, serious illness care models). Consequently, the very
models used as a system-level description of clinical practice do not define the
needed functions but instead describe the type of provider that should be per-
forming these non-explicitly stated functions.
While healthcare is not perfect, it does seem to be functional. At first appearance,
patients receive the care that they need. So, what is the problem with how the
healthcare field describes and defines its system? The answer lies in the nature of
the care itself. Conventional clinical practice utilizes a reductionist approach
directed primarily toward acute conditions. In recent years, however, healthcare
needs have significantly shifted from treating acute conditions to treating chronic
conditions. 78% of total healthcare costs in the United States are now due to chronic
disease. Consequently, clinical medicine needs to shift toward addressing the
growing chronic disease epidemic. Unlike acute conditions, chronic conditions are
particularly complex in that they tend to involve multiple factors with multiple
interactions between them. The medical science literature has now established that
combating chronic disease requires treating the patient holistically. Doing so will
require systems thinking that entirely redesigns existing healthcare approaches
rather simply “reworking” or “reapplying” of acute care models.
Despite the growing need posed by chronic conditions, the relative absence of
systems-level engineering design descriptions of healthcare inhibit designers and
non-medically trained individuals from understanding the field and its terminology.
This linguistic and conceptual gap has isolated healthcare, leaving many engi-
neering design efforts to contribute in a siloed fashion. For example, biomedical
engineering focuses on developing devices that measure and deliver treatment. The
growing efforts in health informatics develop software solutions that support clin-
ical decision-making. Finally, the field of healthcare logistics optimizes the flow of
the patients through the healthcare system. Nevertheless, it is less than clear how
these disparate efforts actually improve patient outcomes. In contrast, the recent
efforts in healthcare system design focus on redesigning the clinical patterns of
practice with patient outcomes firmly in mind, drawing upon not just medical
science but also the many types of engineering solutions mentioned above. Here,
the designer(s) must intersect engineering, systems thinking, and formal design
methods from a diversity of practitioners in different fields. The first step in doing
so is getting everyone “to speak the same language.” That is, the designer must gain
an awareness of and reconcile the linguistic and conceptual discrepancies between
these many fields.
In the past, designers remained within the technical realm; thus, avoiding much
of these discrepancies. They could become experts in knees or hearts; addressing
the acute injuries to these parts of the body. However, the chronic diseases of today
550 I. S. Khayal and A. M. Farid

are multifaceted and consequently require services from several healthcare

departments in potentially different locations. In effect, the patient experience of a
multifaceted chronic condition creates a significant need for integration. Therefore,
designers must simultaneously understand how their technology, devices, and
software affect the patient experience and integrate within the larger healthcare
system. For example, at the outset, diabetes care involves an endocrinologist to
regulate the insulin hormone. However, advanced diabetes patients can experience
complications. Foot damage requires podiatry, eye damage requires ophthalmology,
and kidney damage requires nephrology. These many types of medical specialists
are not typically trained to coordinate care between each other. Consequently,
Diabetes Centers have emerged as places where all of a diabetes patient’s needs can
be provided. Despite the presence of such facilities, and patients’ preference for
them, most patients still receive care in a typical, medically siloed, healthcare
environment.
Healthcare systems are complex. Introducing technical artifacts that were
designed using Axiomatic Design (AD) is one level of complexity. Using AD to
design the healthcare system incorporating many technical and human resources is
more complex. The following emergency department design story describes the
importance of the need to understand the many aspects of a healthcare system.

Design Story 19.3: Emergency Department

Emergency departments (EDs) are considered the safety net within the United
States healthcare system. Unlike typical healthcare departments, emergency
departments are open 24 hours a day, 7 days a week. Emergency departments
provide care to individuals with illnesses or conditions requiring immediate
attention. They are also a safety net for uninsured individuals that only have
emergency departments as care options.
Growing demand for emergency departments have not led to increases in ED
capacity. On the contrary, ED numbers are decreasing. Consequently, challenges of
ED overcrowding have led many medical professionals to turn to systems scientists
and complex system designers in order to find robust solutions.
With the understanding that the ED is a complex system that is working toward
specific goals, it is necessary to understand those goals clearly on all levels in order
to properly analyze and re-design the system (Peck 2008). The design team
developed a four-level emergency department design decomposition (ED3). Initial
analysis showed very high levels of coupling between the DPS and FRs, thus a
clearly coupled system. This design example focuses on one area of the ED—triage
to address and manage patient flow. The classic system of triage sorts patients based
on urgency called the emergency severity index (ESI). When using the ESI system
to facilitate patient flow, its original purpose may suffer due to a coupled design.
The use of AD led to the creation of a new index based on how long a patient is
likely to remain in the ED. By introducing a new index for patient flow, the design
problem was then uncoupled. By doing so, patient wait times were significantly
decreased (  50%) compared to typical operations.
19 Healthcare System Design 551

Fig. 19.1 Diagram showing system form, system function, and the allocation of the latter to the
former called the system concept. (Reproduced from Khayal (2019), originally published open
access under a CC BY 4.0 license: https://doi.org/10.3390/systems7010018).

The remainder of this section describes a system model architecture for per-
sonalized healthcare delivery (Khayal and Farid 2018). It formally defines the
system function and system form for healthcare. A visual representation is shown in
Fig. 19.1.

System Functions

The primary functions in healthcare are identified based on how clinical medicine is
practiced: the diagnostic model. First, it examines the individual’s concern or
complaint. Second, it attempts to determine its cause. Third, it applies a treatment
regimen to that cause. Using solution-neutral language, these functions sequentially
are.
(1) measurement: understand, quantify, or classify individual health state;
(2) decision: determine what to do for the individual and when;
(3) transformation: perform treatment service(s) for the individual;
(4) transportation: move the individual between any of these processes, if
needed.
Here, it is important to recognize the term “individual” as the primary operand of
the healthcare system. It is used instead of the more common term “patient” to
reflect that the current “healthcare” system is actually a reactive “sick-care” system
that treats individuals only after they have become ill. In order to design a proactive
“healthcare” system then the primary operand must be a healthy individual who
uses the system to maintain or improve a healthy state, prevent its degradation, and
only recover the health state if illness prevails.
Measurement is a cyber-physical process that converts a physical property of the
operand (i.e., the individual) into a cyber, informatic property to ascertain their
health state. Typical healthcare measurement processes include clinical evaluation,
diagnostic tests (e.g., blood, urine, and stool) and diagnostic procedures (e.g.,
medical imaging, endoscopy, and electrocardiography).
Decision is a cyber-physical process occurring between a healthcare system
resource and the operand-individual. In cooperation with the individual, the
healthcare system resource generates a decision on how to proceed next with the
552 I. S. Khayal and A. M. Farid

healthcare system. Several types of decision processes exist. Planning is defined as

the decision of which healthcare system functions need to occur for the individual
in the form of a treatment plan or strategy. Scheduling includes two integrated
decisions; a choice of who/what is going to perform the functions in the treatment
plan, and a choice of when to do so in the form of an individual’s booking. The
distinction is important because, in many cases, exclusivity relationships may
appear between the individual and the healthcare system resource, be it imposed
externally as in the case of many healthcare insurance providers or freely chosen as
rapport is developed between a medical professional and the individual. This
chapter focuses on the atemporal aspects of planning and scheduling decisions;
those that directly affect the choice of treatment and who/what is going to execute it.
While many engineers may gravitate quickly toward the development of mea-
surement and treatment technologies, human-centered design encourages us to
step-back and understand how the clinician and the individual will together decide
on using the measurement or treatment technology as part of a larger treatment plan.
For example, the best way to see and address gastro-intestinal and colon cancers is
through preventative colonoscopy. The procedure uses a device called a colono-
scope. It is a long, flexible, tubular instrument used to image the lining of the colon
and take biopsy samples. During the procedure, it is inserted through the rectum
and advanced to the other end of the large intestine. Many individuals refuse this
uncomfortable procedure and select a less-effective imaging-based procedure
instead. Such a planning decision naturally draws upon further information such as
cancer risk, family history, and reaction to drugs.
Transformation is a physical process that transforms the operand, specifically,
the internal health state of the individual. Such processes include surgical proce-
dures such as amputation, ablation, and laparoscopic surgery, and therapeutic
procedures such as pharmacotherapy, chemotherapy, and physical therapy.
Transportation is a physical process that moves the individual between
healthcare resources. It is included here for the sake of completeness because only
incapacitated individuals need to be moved. For example, a nurse can move an
individual from the emergency department’s operating theatre to the recovery room.

System Form

As mentioned previously, the system form (in large-scale systems) is made up of a

set of resources that are further described by DPs. Each of the resources defined
below includes a set union of human and technical resources. In the healthcare field,
a designer should expect that each newly designed technical resource requires a
human resource to operate it. For example, the operation of a push wheel chair can
be performed by any adult whereas the operation of a surgical robot system requires
years of specialist clinical training. Again, the design of a technical resource must
explicitly consider the system boundary and more specifically the often
non-technical nature of the human operator. The consequent implementation and
outcome challenges are discussed in Sect. 19.3.
19 Healthcare System Design 553

A Measurement resource is a resource capable of measuring the operand; here,

the health state of an individual. Human measurement resources include MRI
technicians, sonographers, and phlebotomists. Technical measurement resources
include magnetic resonance imaging scanners, ultrasound machines, and syringes.
A Decision resource is a resource capable of advising the operand-individual on
how to proceed next with the healthcare delivery system. Human decision resources
include oncologists and general practitioners. Technical decision resources include
decision support systems and electronic medical record decision tools.
A Treatment resource is a resource capable of a transformative effect on its
operand: the health state of an individual. Human transformation resources include
surgeons, oncology infusion nurses, and physical therapists. Technical transfor-
mation resources include operating theatres, chemotherapy infusion rooms, and
physical therapy rooms.
A Transportation resource is a resource capable of transporting the
operand-individual. Human transportation resources include emergency medical
technicians, clinical care coordinators, and transporters. Technical transportation
resources include ambulances, gurneys, and wheelchairs.

19.2.3 Variations of Healthcare System Function and System

Form Across Cultures and Regions

Engineering design is often viewed as the utilization of universal scientific

first-principles to produce designs that can be used universally across cultures and
regions. More recently, the field of manufacturing has recognized that many
products must be “mass-customized” in order to serve the preferences of a wide
variety of customers. In much the same way, and probably to an even greater extent,
the healthcare system has many levels of design that are impacted by the biology
and culture of people in different regions.
National-Financial Level
While the term healthcare elicits a focus on medicine, the financing of healthcare is a
significant driver of how, where, when, and what healthcare services are available to a
population. For example, Japan simultaneously faces a growing aging population and a
shrinking young population. Consequently, the healthcare industry has focused on devel-
oping technical resources like social robots rather than human resources like home-nurses.
Such a design decision delivers a healthcare system delivery model within the financial
constraints of the country.
Social Level
The social level affects many aspects of healthcare services; most importantly its
community-wide acceptability.
Healthcare systems are increasingly utilizing new community structures to deliver their
services. For example, many healthcare systems are utilizing social media platforms to
provide peer-patient-to-peer-patient support networks. Alcoholics Anonymous is one such
554 I. S. Khayal and A. M. Farid

successful network in the United States. Such peer networks create a strong emotional bond
across individuals with a similar condition. In certain cultures, however, there remains
significant social stigma towards certain diseases. For example, in some cultures women
have significant difficulty marrying if they are identified as having certain diseases. Fur-
thermore, some societies, cultures, and/or religions place restrictions on certain types of
healthcare interventions like birth control and abortions.
Biological Level
In addition to national and social factors, healthcare system design must also consider
variations in biological factors. The testing of healthcare treatments today requires that
clinical trials specifically note the ethnicity of the studies’ participants. Medical science has
generally accepted that such variations as they stem from ethnicity and culture affect study
efficacy.
Medical research today has also recognized that although new treatments are tested for a
wide variety of ethnicities, the design of new treatments still follows a one-size-fits-all
approach. The latest healthcare trend advocates for “personalized medicine” where thera-
pies are tailored to specific individuals. Such an approach takes into account an individual’s
unique molecular and genetic profile to predict whether a given medical treatment will be
safe and effective.

Design Story 19.4: Healthcare Labor Market Volatility Assessment

The quality and reliability of a nation’s healthcare system is often driven by the
number and diversity of its healthcare professionals. Unfortunately, many devel-
oping nations often suffer from very constrained segments in their highly skilled
labor market and hence must “import” this human capital. Volatility in key
healthcare professions can threaten reliable and sustainable healthcare delivery. In
this design story, AD was used to develop a system to assess volatility of healthcare
human resource in a country in the middle east. The FRs for the system were
defined and aggregated as a set of healthcare professions: FR = {Healthcare Pro-
fessions}. The functional domain is only addressed at this level of hierarchy as it is
important to be able to distinguish between these individual healthcare functions
throughout the analysis. In contrast, the physical domain can be analyzed at several
levels of hierarchy. But one of the hierarchies would describe a decoupled system.
Each individual healthcare professional is a distinct healthcare resource. Healthcare
professionals can be aggregated by their professional licensing degree:
DP = {Healthcare License Degree}. When designing based on resources, each
individual can be aggregated based on where they practice: DP={Healthcare
Facility} and in what region: DP = {Geographic Region}. For this example, the DP
level is chosen such that this system is decoupled.
19 Healthcare System Design 555

Example FRs include: and Example DPs include:

19.3 Challenges Specific to the Design of Healthcare

Systems

The previous section served to describe several systems thinking abstractions for a
healthcare system: system boundary, scale, scope, function, and form. All of these
are key aspects that need to be considered when designing a healthcare system. In
doing so, Sect. 19.2 touched on several challenges specific to the healthcare field.
This section enumerates four key challenges to consider when designing healthcare
systems. They include healthcare as a socio-technical system (in Sect. 19.3.1),
heterogeneity (in Sect. 19.3.2), complexity and interoperability (in Sect. 19.3.3),
and understanding legacy systems (in Sect. 19.3.4).

19.3.1 Healthcare is a Socio-Technical System

Section 19.2 described that every designed technical resource requires a human
resource to operate it. For many engineering designers, the defined system
boundary may be a completely technical artifact (e.g., a knee implant). However,
556 I. S. Khayal and A. M. Farid

the success of this technical system is based on the success of the human user as a
human resource. For example, the orthopedic surgeon must place the knee
implant, and thereafter the physical therapist must rehabilitate the knee in order to
achieve the highest likelihood for normal knee function with minimal pain. In many
cases, it is not easy to distinguish whether an unsuccessful outcome is due to the
human or technical resource. Years of clinical studies and reporting are often
required to distinguish between these two factors. Returning to formal AD theory
(and its Independence Axiom), the socio-technical nature of the healthcare solution
means that the FRs of normal knee motion are achieved by two DPs in an inevitable
coupled design.
The socio-technical nature of healthcare systems presents further design chal-
lenges. One may naively believe that an implantable device, such as a knee implant,
will have hundreds or thousands of identical copies. In reality, the device itself is
produced using materials and manufacturing methods that have quality tolerances
that leave a level of variability in the final product. Similarly, the surgical team
installing the implant provides several types of parallel and sequential services that
also introduce a level of variability. The variability in the healthcare outcome can be
further exacerbated by the orthopedic rehab center that provides inpatient and
outpatient rehabilitation services, physical therapy services, and skilled nursing
services. Only after all of these technical and human resources are combined can a
health outcome be truly evaluated. Furthermore, in a clinical study, a sufficiently
large sample of implants need to be placed by many different surgical teams and
rehabilitated by many different orthopedic centers in order to truly assess the
technical design. Consequently, the variability in the social aspects of the healthcare
system must be controlled in order to maximize the likelihood of success of the
designed technical resource. That said, the variability introduced by both the human
and technical resource may need to accommodate the natural variations of the
individual-patient. In short, the Information Axiom instructs the designer to achieve
a successful healthcare outcome by minimizing the variability of both the human
and the technical resources.
As medical advancements continue, especially in artificial intelligence (AI) and
data analytics, some believe that we may, someday, minimize or eliminate human
resources from healthcare solutions. Such a perspective, nevertheless, neglects the
importance of the human-patient to human-doctor connection, be it conscious or
subconscious, in providing care. Furthermore, their social connection, in terms of
rapport and trust, shapes the patient and clinician’s decision-making process, affects
how patients proceed with healthcare options, and determines their degree of
cooperation as they proceed with treatment. These many human aspects of
healthcare are even more pronounced in services that require a high level of social
connection like psychology, psychiatry, cancer, palliative care, and pediatric care.
19 Healthcare System Design 557

19.3.2 Healthcare Delivery Heterogeneity

While the Information Axiom instructs the designer to minimize the variability in
the human and technical resources of the healthcare system, there are practical
limitations to doing so. Even as the technical resource is designed within very tight
quality tolerances, it must also be designed to fit in a wide variety of heterogeneous
healthcare delivery systems. This heterogeneity stems from its different human
resources, in the implementation of the integrated design, and in the financial
models under which the healthcare delivery system operates. Different doctors have
different healthcare patterns and outcomes, different healthcare systems integrate
technology in different ways, and the different financial models present hard con-
straints and soft influence on how healthcare services are chosen and delivered.
Heterogeneity in human resources
Beyond the heterogeneity in the different types of doctors (e.g., cardiologists,
surgeons, psychiatrists), there exists variations in their number, training, expertise,
and patterns of practice across a given region. These human factors affect the design
of the technical resource. Even if the design is purely technical (e.g., social robots),
the device’s market uptake will depend on the number and nature of the doctors in
that region. For example, rural areas are less likely to benefit from advanced
healthcare technologies because they also suffer fewer doctors and specialists. Still
other countries have shortages in certain types of doctors. Japan has a shrinking
ratio of working individuals to aging individuals. In the United Arab Emirates,
expatriates make up 85% of the healthcare workforce; and consequently, half the
workforce is turned over every two years.
Heterogeneity in implementation into a healthcare delivery system
In addition to the heterogeneity in the DP of the healthcare system’s human and
technical resources, the healthcare system also introduces heterogeneity through the
implementation of its PVs. One common problem is that the PV hierarchy is not
decomposed to the same degree as FR and DP hierarchies. In other words, the PV are
described at a high level with no specifics as to how the implementation is carried out
in detail. For this reason, many medical advancements are never implemented and
the typical implementation lag time is 17 years. To address this problem, a new
medical field called “implementation science” was developed to study how to im-
plement medical discoveries and interventions once they are known. It recognizes
that the lack of detail of how a scientific discovery or intervention was tested in a
clinical trial can lead to widely different outcomes. For example, in some cases the
implementation of the electronic medical record (EMR) into the healthcare system
has improved patient health outcomes and doctor productivity, while in other cases it
degraded them. Similarly, some implementations of behavioral health and opioid
recovery programs have improved patient health outcomes and lowered costs, while
in others there has been little or no improvemen.
558 I. S. Khayal and A. M. Farid

Heterogeneity in healthcare delivery system financial models

The operation of a healthcare system depends significantly on how it is financed.
Many different payment strategies and stakeholders have been implemented, and
each has its own set of stakeholders. For example, individuals can pay out of
pocket, employers or individuals can purchase private healthcare insurance from
private companies, or a country or region can provide public healthcare through the
government. Furthermore, a healthcare system’s financial model can and usually
does vary over time.
Many countries are currently revising the financial models of their healthcare
systems. The conventional model today is “fee-for-service”; a specific fee payment
for every time a service is rendered. Much research suggests that such a payment
model incentivizes the provision of large quantities of healthcare service with less
attention to quality in terms of patient health outcomes. In contrast, a new concept
called “value-based care” provides reimbursements for demonstrated value, and
improved health outcomes at reasonable cost. Many countries are iteratively
experimenting with reimbursement schemes for value-based care. As the associated
financial incentives change, the efficacy of healthcare interventions has also been
found to change as well.

19.3.3 Complexity and Interoperability

As mentioned earlier in the chapter, the current healthcare system developed

organically to meet “one-off” acute illnesses and injuries. This same system is
facing an unprecedented chronic disease burden. These conditions, unlike acute
conditions, are particularly complex in that they are ongoing and tend to involve
multiple factors with multiple interactions between them. Consequently, many
patients require many different services and resources in their treatment of the
chronic condition. A great deal of healthcare system complexity arises when a
single patient utilizes many different types of services offered by different human
and technical resources. For example, a cancer patient may meet with an oncologist
in a clinic, receive chemotherapy from an infusion nurse at an oncology
infusion-center, and receive imaging at an imaging center.
From the perspective of the patient, the complexity appears as the burden of
moving between many human and technical resources. Meanwhile, from the per-
spective of the healthcare system, the complexity creates the need for coordination
and communication of care between these resources. Such resources are accus-
tomed to providing services without coordination despite the potential for adverse
outcomes of varying severity: missed appointments, drug reactions, ineffective,
delayed or inaccurate care plans, and other safety issues.
Greater healthcare system interoperability may alleviate some complexity. Here,
it is equally important to improve the interface both within and between different
healthcare systems. For example, an electronic medical record (EMR) system from
a single vendor may enhance interoperability across healthcare systems. However,
19 Healthcare System Design 559

given the relative absence of EMR standards, data sharing across EMR systems
from different vendors remains elusive. That said, it is not uncommon to find
different EMR systems in different departments within the same healthcare system.
The lack of interoperability appears at multiple scales and scope. While inter-
operability concerns can occur between different healthcare systems, they can also
be found within a single healthcare system (e.g., hospital), a single department (e.g.,
surgery), or a single resource location (e.g., operating room). For example, a given
operating room utilizes a vast array of medical devices. They should ideally gather
data from all available source sensors so that they can be processed (automatically)
into actionable data for the surgical team. Instead, this device data is often uploaded
into the EMR system and monitored manually. In this regard, interoperability
presents a viable opportunity to improve patient safety.
Efforts to improve healthcare interoperability can have several drivers. From a
regulatory perspective, the connection between patient safety and interoperability is
likely to cause regulatory reform proactively or via the judicial system. Alterna-
tively, when there is a pressing industry-wide need, industrial consortia will often
form to write relevant standards. Finally, in some cases, EMR system vendors may
recognize that proprietary but transparent interoperability standards is conducive to
a “platform-economics” business model. For example, the DOC, XLS, PPT, and
PDF file formats are all proprietary de-facto interoperability standards. There also
exists the DICOM imaging file format in healthcare, a standard that revolutionized
the ability to share images across boundaries.

19.3.4 Healthcare as a Legacy System

Finally, it is important to recognize that the healthcare system functions as a large

flexible engineering system. Not only is it a “soft infrastructure system” composed
of social elements like rules, regulations, policies, and institutions, but it is a
veritable “hard infrastructure system” in terms of the billions of dollars worth
of capital assets that constitute its technical resources. The sheer scale of the ex-
isting healthcare system means that a designer is obliged to recognize its legacy
nature. Neither can its existing capital stock of technical resources be immediately
replaced with a new set of new technologies, nor can its stock of human resources
be fired and replaced with newly trained medical professionals. Doing so is not only
cost-prohibitive but also entirely unsafe given that new technical and human
resources often require significant time before they are properly integrated in the
remainder of the system.
This distinction between large flexible engineering systems and new product
design requires a fundamental change in the engineer’s design methodology. As
described earlier in this book, the forward-engineering “zig-zagging” method is no
longer entirely appropriate. The designer cannot simply determine a high-level FRs,
allocate it to a DPs, and then decompose the FRs into a set of lower level FRs, etc.
Such an approach maximizes creativity, free of design constraints, and is most
appropriate for new products. It is not appropriate for a large flexible engineering
560 I. S. Khayal and A. M. Farid

system design because it does not take into account legacy nature of existing
technical and human resources.
Instead, in an engineering systems context, the functional and physical hierar-
chies must first be “reverse engineered.” The designer must observe the physical
hierarchy—particularly at a relatively detailed level, infer the associated FRs, then
aggregate the DPs into their associated human and technical resources, and then
repeat. Once the existing legacy structure of the healthcare system has been
modeled in terms of its functional and physical hierarchies it can serve as the
designer’s “initial condition.” From there, the designer can “evolve” the system to a
more advance stage of development recognizing the design priorities that appear in
a modified set of stakeholder needs and FRs. Furthermore, in developing the new
design, it is often unnecessary to “reinvent the wheel.” Well-known functions have
tried-and-true physical solutions which may be re-implemented successfully. More
specifically, the design of technical resources in a healthcare system often relies on
an existing and extensive body of validated medical science and biomedical
engineering research which can be leveraged as a rich repository from which to
draw. Such a design approach recognizes and emphasizes the need for validation by
clinical trial prior to system-level design. Similarly, the design of human resources
in a healthcare system requires the designer to select appropriately qualified per-
sonnel given that the time and money required for training is practically limited. In
short, the legacy nature of the healthcare system places practical constraints on
methodology and outcome of any design effort.

19.4 Key Takeaways and Conclusion

In summary, this chapter has described some of the key considerations that a
designer must make as they approach a healthcare system design project. It has
discussed several systems abstractions in the context of healthcare systems: system
boundary, scale, scope, function, and form. These abstractions were described in
terms of practical limitations and requirements that are often imposed from outside
of the system boundary. This chapter also described a system framework for
designing healthcare systems that is consonant with Axiomatic Design’s FRs, DPs,
and process variable domains. The chapter concluded with four key healthcare
design challenges that the designers must consider as they design healthcare sys-
tems at various scales and scope.
Problems
1. From Design Story 19.1, design a “turn-off” system for a pacemaker that
addresses the ethical and policy issues of two types of turn-off functions. One of
the turn-off functions is to be performed by a physician and the other is a more
advanced turn-off system that can be performed by the user or family member,
but that ensures it is performed with authorization and not by mistake or malice.
19 Healthcare System Design 561

2. From Design Story 19.2, it was identified that the designed system makes for a
complicated re-entry. What FRs would you add to the system to ensure that
future surgical routing to fix any issues would be uncomplicated? How would
you setup your FRs and DPs for this analysis?
3. You are hired to lead the design for a pacemaker. If the following high-level FRs
are designated, design a set of DPs for an uncoupled system.
FRs:

1. produces rhythmic electrical signals;

2. does not negatively affect the host body;
3. produces safe signals regardless of external environment;
4. eliminate surgical maintenance or expiration (does not need to be replaced).

4. You consult for a baby company that wants you to develop a baby monitoring
system. Below are comments from parents. Convert each comment into a FRs.

a. “I live in a rural area in a high-energy performance home (i.e., 12-inch thick

walls) on a large lot.”
b. I work outside and I need a baby monitoring system that can provide a secure
private connection that guarantees zero interference, clear sound, with range
up to 1,000 feet.
c. I need to know when my unit is out of range of the baby unit.
d. I need to be able to respond to my baby so I can comfort them with my voice
and face.
e. I need the parent unit to function outside in wet or dry conditions, and under
extreme hot and cold temperatures.

5. You are a design engineer traveling abroad to Africa for vacation. You notice a
tired pregnant passenger on a local bus trying to get to a pre-natal visit at a local
hospital (65 miles away). You converse with the passenger who tells you the
following FRs for her ideal pre-natal care. Identify the DPs to develop a
decoupled system solution.

a. send me a reminder when I need to connect for a pre-natal visit;

b. perform most visits virtually;
c. communicate efficiently and with speed (within 30 min) between patient and
clinician;
d. allow for anonymous patient-to-patient communication that is monitored for
clinical accuracy (so I can ask private questions that I don’t want my doctor
to know about me).
562 I. S. Khayal and A. M. Farid

References
Khayal IS, Farid AM (2018) Architecting a system model for personalized healthcare delivery and
managed individual health outcomes. Complexity 2018:24. https://doi.org/10.1155/2018/
8457231
Peck JS (2008) Securing the safety net: applying manufacturing systems methods towards
understanding and redesigning a hospital emergency department. S.m. thesis, Massachusetts
Institute of Technology, Engineering Systems Division, Technology and Policy Program, 77
Massachusetts Ave., Cambridge MA 02,139, USA. https://dspace.mit.edu/handle/1721.1/
42934
Suh NP, Cho DH (eds) (2017) The on-line electric vehicle: wireless electric ground transportation
systems. Springer International Publishing

Further Reading

Crawley E, Cameron B, Selva D (2015) System architecture: strategy and product development for
complex systems, 1st edn. Pearson
Farid AM (2016) An engineering systems introduction to axiomatic design. In: Farid AM, Suh NP
(eds) Axiomatic design in large systems: complex products, buildings and manufacturing
systems. Springer, Cham, Switzerland, Chap 1, pp 3–47
Khayal IS (2019) A systems thinking approach to designing clinical models and healthcare
services. Systems 7(18):26. https://doi.org/10.3390/systems7010018
Khayal IS, Farid AM (2016) The need for systems tools in the practice of clinical medicine. Syst
Eng 20(1):3–20. https://doi.org/10.1002/sys.21374
Functional Periodicity, “Function
Clock,” and “Solar Time Clock” 20
in Design

Nam Pyo Suh

Abstract
Some systems are designed to perform the same set of functions cyclically and
repeatedly. These systems have “functional periodicity.” The best-known system
with “functional periodicity” is a biological cell. A cell divides into two identical
cells through mitosis, which involves five sequential phase transitions:
interphase, prophase, metaphase, anaphase, and telophase. These transitions
occur when the cell is ready for phase change rather than after a fixed time delay.
Similarly, in engineered systems such as the “track machine” used in
manufacturing semiconductor chips, functional periodicity can be introduced
to the system to improve the reliability of the manufacturing system and
overcome inevitable variations of processing times. The time duration of the
functional period may vary from cycle to cycle for a variety of reasons.
There are other engineered systems that can benefit from functional
periodicity, e.g., the assembly line of automobiles, the flight schedule of
airlines, and operation of universities. We may incorporate “functional
periodicity” in these design systems to maximize productivity or to improve
system reliability. In each functional period, these systems perform an identical
set of functions. Before initiating a new functional period, the system should be
re-initialized for optimum operation of the system for next cycle. A designed
system can extend its useful life when its functions at the beginning of each
functional period are re-initialized.
In engineered systems, we use two different kinds of periodicity: time-based
and function-based. The time-based systems use the conventional solar time
clock to control the system. The function-based system uses the completion of
the same set of “functions” before it initiates the next cycle. These systems with
functional periodicity can be re-initialized at the beginning of each new cycle to

N. P. Suh (&)
M.I.T., Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 563

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_20
564 N. P. Suh

get the best performance. The inclusion of the functional periodicity in designed
systems generally improves their operations. An example of functional
periodicity that is both time-based and function-based is the election of political
leaders.
Application of functional periodicity in designed systems should lengthen the
life of products, improve the efficiency of systems, reduce the downtime,
increase the effectiveness of transportation systems, and minimize deterioration
of political and socio-economic systems. For best results, the “re-initialization”
of the system should be done at the beginning of each functional period.
In this chapter, two ideas are introduced and discussed: the application of
“functional periodicity” in system design and the utilization of the concept of
“function clock.”

20.1 Introduction to Functional Periodicity

and Re-initialization

A system must fulfill its functional requirements (FRs), at all times without failure,
to be useful throughout its life. If the system is a production machine, its reliability,
performance, and productivity are of utmost importance. A large, complicated
system must run more or less continuously throughout its life, producing quality
goods reliably, with occasional maintenance. On the other hand, if the system is a
research university, it must perform its functions well. It must continuously improve
and enhance its programs, contents, quality of human resources, and finances to
provide the best education to students and create knowledge for humanity. If the
system is a government agency, it must renew itself to be sure that it is fulfilling the
original legislative mandate for establishing the agency. Many of these complicated
systems must satisfy a set of FRs over a pre-determined period, as discussed in
previous chapters.
The primary goal in design is to create a system that satisfies the FRs. An even
more ambitious goal is to design a system that will renew itself periodically for an
indefinite period to fulfill its original FRs. To achieve this goal, we introduced two
concepts: “functional periodicity” and “re-initialization” to show how a system that
renews its FRs and DPs periodically can be designed. One of the most well known
among such systems is the democratic political system. Political leaders (e.g.,
president, members of Congress, etc.) serve a fixed term in office, and a new
election is held. Academic programs re-initialize at least once a year to achieve this
goal. These systems have time-fixed periods for renewal, i.e., each cycle ends at a
fixed time.
20 Functional Periodicity, “Function Clock,” and … 565

Educational Institutions and Functional Periodicity and Re-initialization

Nearly all educational entities from primary school to colleges and universities have used
functional periodicity and re-initialization, at least once or twice a year. Students change
each year, but the educational programs are typically organized on a semester or yearly
basis. In the United States, schools begin a new academic year and semester in the fall of
each year. After four months of intensive study, students and teachers have breaks. Then
the schools repeat the same process for different students.

Some systems have periods that end when their functions are finished rather than
stopping at the pre-determined time. A well-known example is the Supreme Court
of the United States. The Supreme Court justices do not have a term limit that
occurs at the end of a fixed time. Similarly, some manufacturing operations do not
end at the end of a set time but instead ends when its manufacturing task (i.e., a
function) is finished. A biological cell undergoes five sequential phase transfor-
mations during mitosis to split into two identical cells when it is functionally ready
rather than at a fixed time interval.

Actual Story: Is it a monopoly or an effective use of functional periodicity and

re-initialization for an industrial firm?

New England of the United States consists of the six northeastern states of the
United States, i.e., Maine, Massachusetts, Rhode Island, New Hampshire, Vermont,
and Connecticut. The region is not only rich in history but also the center of higher
education and the birthplace of many advanced technologies of the Americas. It is
also one of the most scenic sections of the United States. Many of these states are
culturally and politically liberal states vis-a-vis the conservative southeastern states
of the United States. Economically, technically, and politically, these states have
continued to evolve and transform. Many well-known universities have their roots
in New England.
Until about 1950, New England was the home for many shoe-related businesses.
There were many shoe factories and leather tanneries in Massachusetts and New
Hampshire. The world’s largest shoe machinery manufacturer, USM Corporation,
was located in Beverly, Massachusetts, about 35 miles northeast of Boston. They
dominated the machinery business worldwide. It was incredibly generous to its
employees. It created wealth and many high-paying jobs in the Commonwealth of
Massachusetts and the United States!
USM’s business model was unique. Instead of selling their machines, they only
leased them to shoe manufacturers. If someone wanted to go into the shoe manu-
facturing business, the company would set up a factory and install its machines on
lease. Then, they would collect lease-fees each time one of their machines was used
to manufacture shoes. They conducted active research and development to invent
new machines, new materials, and processes. The main goal of their research and
development was to innovate new technologies and secure patents to protect its
market share. They had a warehouse full of modern machines they invented, but
they did not sell them. They only leased some of their patented machines. The
patents and their leasing policies protected them from their potential competitors.
566 N. P. Suh

Therefore, they had a de facto monopoly in the shoe machinery business world-
wide. Once in a while, they would introduce new machines, but their business was
based on leasing the same machines for many years, servicing them regularly,
replacing bearings, etc., periodically. Anyone who wanted to get into shoe business
could do so if they could meet the lease terms of USM Corporation.
They practiced functional periodicity to make their customers happy. After their
machines had been used to manufacture a fixed number of shoes, they would
periodically service them in the shoe factory. They replaced worn parts and bear-
ings, etc., to make the machines function as if they were new. Since their machines
were rugged to last for a long time, once the worn components were replaced, they
worked like new machines again. Their business practiced “functional periodicity”
and “re-initialization”—through the re-initialization of their leased machines,
although they did not use those terms. The company had a gold mine because it was
difficult for other companies to penetrate the shoe machinery business. They took
care of their employees generously. Many stayed with the company throughout
their lifetime until they retired.
Sometimes, good things cannot last forever! The U.S. government sued USM for
monopoly and anti-trust practices, and the government won. The U.S. Federal
Court ordered USM to sell their machines rather than only leasing them. When all
those leased machines had to be sold to the shoe manufacturers all at once, USM
became temporarily cash-rich. Unfortunately, USM invested their cash in the wrong
businesses and could not compete in highly competitive companies; they did not
have expertise and deep roots. USM, a great company, is no longer in business. In
hindsight, the top management of USM chose a wrong set of FRs when they
re-initialized the corporate strategy under the government mandate.
The lesson we should learn from the sad saga of USM Corporation is that the
functional periodicity and re-initialization in business are effective strategies.
However, the business model should quickly adapt to changing external conditions,
such as changing Federal laws and regulations. That is one of the reasons why so
many companies hire “lobbyists” to protect their back by influencing the process of
enacting government regulations and laws.
Actual Story: Functional Periodicity and Re-initialization as a Design Tool
A couple of decades later after the demise of USM Corporation, the idea for
incorporating functional periodicity and re-initializing FRs into a designed system
was advanced to solve a problem encountered by a semiconductor equipment
company.
Functional Periodicity and Re-initialization: Background Story
An executive of a company and a professor were waiting at the Armonk airport, a small
airport, outside of New York City, near the headquarters of IBM (International Business
Machines) Corporation, for their flight to Rochester, New York. They were on their way to
visit the optics research laboratory at the University in Rochester. The executive and the
professor knew each other well, having collaborated on many projects in the past. The
professor was a board member of the company and recommended hiring the executive a
few years earlier. While waiting for their flight, the executive described a problem the
20 Functional Periodicity, “Function Clock,” and … 567

company had had with their “Track Machine.” He stated that he had tried to hire a
specialist in the field of operations research (OR) as a consultant to solve the problem.
However, he said that he had decided against it. The OR specialist did not know the
solution to their problem. Instead, he proposed a major research contract to the company.
However, the problem needed an immediate answer because the track machines were being
used in many semiconductor fabrication facilities (commonly known as FAB), causing
production problems.
In FABs, silicon wafers are made into semiconductor devices (sometimes, called “chips”).
One of the machines used is the so-called “track machine” (note: the old name “track”
stuck, although modern tools do not use tracks anymore). “Track machine” performs many
functions in conjunction with the lithography machine. It takes in wafers at a regular
interval and clean the surface of semiconductors, etch the surface, spin-coat the surface of
wafers with photoresist, and transport the photoresist-coated wafer to the lithography
machine for exposure to UV light through photomasks to create electric circuits (line-width
as small as 20 nm) on the surface coated with photoresist. After exposure in the lithog-
raphy machine, the wafers are brought back to the track machine for removal of unexposed
photoresist and cross-linking of the photoresist.
The process in the track machine begins when a silicon wafer is loaded at its loading
station by a wafer-handling robot. There are many robots and many stations within the
track machine for simultaneous transport and processing of many wafers. These robots
carry the wafers from station to station for various operations done in the track machine.
The operations include cleaning of the wafer surface, spreading photoresist, curing pho-
toresist after exposure to UV light in lithography machine, etching, etc. After each oper-
ation, the wafer is moved to the next station in the track machine to complete another
manufacturing operation. This process continues until the wafer has been subjected to all
manufacturing processes, which follow a pre-determined sequence of operations. Since a
large number of wafers must be processed, there are multiple stations for each one of the
manufacturing operations. The processing time for each operation is different. Therefore,
the robot must be instructed to transport a given wafer from its current station to the next
station to complete the required manufacturing operations. The company used an artificial
intelligence (AI) technique to program the transportation sequence of the robot based on
“If …, then do …” type of AI algorithm. After the completion of each assigned manu-
facturing operation, the robot transports the wafer to the next station for the subsequent
fabrication process.
The track machine made by this company normally functioned well, but then once in a
while, the device would stop working, because the old AI-based algorithm—(if …, then do)
—would not know what to do next, i.e., the machine got “confused” after many hours of
operation due to accumulated errors. When the machine stops, many wafers in the track
machine at various stages of processing might become unusable. To continue the manu-
facturing operation, they have to re-boot machine, which is very disruptive and costly.
The executive thought that a better software program should be developed based on a
better process model, perhaps using an optimization method developed in operations
research (OR). They both agreed that the problem might have stemmed from the fact that
there are slight variations at each station from the assumed processing times, temperature,
etc. When the errors accumulate, the algorithm based on the “if, then” logic might fail to
function correctly.
The professor made a different suggestion. “Why not introduce functional periodicity and
automatically “re-initialize the machine” after one (or more) complete cycle measured
from when a finished wafer leaves the track machine. The re-initialization is done by
assessing the status of every wafer currently in the machine. For example, it determined
568 N. P. Suh

how many more minutes a specific wafer has to be kept in Module A, which modules are
empty and waiting for a new wafer to process. Then, decide on where each wafer has to go
for the next operation, i.e., re-initialize the system for the next cycle. This idea became the
basis for “functional periodicity” and “re-initialization.” It turns out that Nature has been
practicing “functional periodicity” and re-initialization to maintain life on Earth, which
may be responsible for the incredibly healthy life of living beings!

The idea for functional periodicity and re-initialization was initially developed to
solve a problem associated with the track machine used for semiconductor pro-
cessing. However, the principles are equally applicable to other technical and
non-technical systems. For example, companies that sell expensive automobiles
provide regular “free” maintenance. During the maintenance, they replace parts that
may imminently fail to enhance the image of the quality of their cars.
Nature, including all living beings, uses functional periodicity every day, which
is known, in biology, as the circadian cycle. We sleep at night and study or work
during the daytime, more or less, regularly. Our biological cells go through different
phases in our body and divide into two “identical” cells once approximately every
24 h. When our biological systems lose functional periodicity and the ability to
re-initialize, the system becomes chaotic and does not survive for too long.
Many human-designed systems have incorporated periodic changes to harmo-
nize several actions taking place in engineered systems. Transportation systems are
good examples. For instance, airlines schedule thousands of flights every day,
covering vast geographic areas. The flight schedules repeat regularly, typically
daily. When airlines must reschedule their flights because of the unexpected
weather change, they can recover quickly because of the periodic nature of airline
schedules. For instance, suppose that a hurricane hit the Atlanta airport in the State
of Georgia. Therefore, they had to close one of the busiest airports for a day—this
closing of the Atlanta airport forces airlines to divert the aircraft headed to Atlanta
to other airports. The effect of closing the Atlanta airport will propagate to other
airports. Some of the flights from Boston to other cities must be canceled because
the aircraft from Atlanta to Boston has been canceled. A chaotic situation will arise
not only in Atlanta but also in Boston and elsewhere, which may last hours.
Passengers stranded in the Boston Logan Airport may have to sleep on chairs and
floors at the airport. This weather in Atlanta may create chaos in scheduling
problems that propagate throughout the country and even to other countries.
However, if the weather finally clears up after so many hours or days, the airlines
can re-initialize their system. They can dispatch the airplanes to various airports and
gradually resume their flights on schedule. Airlines can resume their regular flights
rather quickly because airline flights typically repeat on a 24-h cyclic basis. Most
airline flight schedules have functional periodicity.
Many human-made systems have functional periodicity. Just like the airline
schedule, we can build in functional periodicity in engineered systems. These
systems with functional periodicity can be re-initialized to continue the operation
without disruption. For instance, every morning, the availability and locations of
airplanes should be determined to determine if the aircraft should be relocated or
dispatched to different airports to meet flight schedules. When functional periodicity
20 Functional Periodicity, “Function Clock,” and … 569

and re-initialization are designed into a system at the design stage, the system will
be more robust and easy to maintain as well as lasting a long time.
The purpose of this chapter is to introduce the concept of “Functional Period-
icity” and “Re-Initialization” of systems. The goal is to design a system with
functional periodicity and re-initialize it periodically. A system that re-initializes
periodically can continue its operation for an extended period without a chaotic
breakdown of the system operation. Many large systems designed in the past have
encountered problems both during the development stage and also in operation,
because they had made design mistakes, introducing coupling of FRs. Had they
introduced the functional periodicity and the means of re-initializing itself, they
might have operated more robustly and survived longer.

20.2 “Function Clock” Versus “Time Clock”

Most people perform various tasks based on the time clock, i.e., 24 hours a day,
60 minutes per hour, and 60 seconds per minute. The time clock is used as the
standard reference for most events that occur in our daily life. In the operation of
design systems, we set almost everything based on the time elapsed instead the
function accomplished. The time referred to is the “time clock,” which is set to the
rotation of Earth about its axis. However, in some cases, it is more appropriate to
use the completion of “functions” as the reference frame, because some functions
may take longer or shorter to complete. For example, suppose we create a school
system based on the “function clock”1 rather the “time clock” that is prevalent
today. How would that school be different from the school that operates based on
“time clock”?
Definition of “Function Clock”
A function clock is defined as the clock that tracks when a specific function of a system is
completed. When there is a set of functions that must be satisfied to complete the task
repeatedly, the function clock tracks the completion of each function. When a new cycle of
functions begins, the function clock begins to tick again and monitors when all the func-
tions are completed. There is no correlation between the time clock and the function clock.

Many things in Nature operate based on “function clock.” For instance, in

biological systems, the cell division, i.e., “mitosis,” goes through four sequential
phases before they split into two identical cells. However, the duration of each
phase change does not occur at precisely the same time. The phase transition occurs
after the functions of each phase are completed, which take different durations in
terms of time. Similar variations occur in some manufacturing operations. For
instance, consider the task of maximizing the productivity of the “track machine,”
which process semiconductor wafers by subjecting them to various manufacturing
processes using many modules. In this machine, the wafers are moved from module
1
The term “Function-Clock” is not a commonly used term. It is introduced in this book for the first
time, because in some design applications, the concept of “Functional-Clock” may be useful.
570 N. P. Suh

to module using a robot after a given operation is completed. The wafer can be
transferred to the next station of the track machine only after the current procedure
is completed. The completion of each activity cannot be timed precisely because of
variations in processes. The wafers can be removed from each processing station of
the track machine only after the specific processing function is completed. The
traditional way is to pre-program the robot motion to transport the wafer from a
given station to the next module.
The difference between “function clock” and “time clock” may be illustrated
using how the education of students can change under these two different systems.
Example 20.1 Difference between “Function Clock” and “Time Clock”
In nearly all countries, schools operate based on a fixed schedule. Schools in the United
States typically starts new school year right after labor day, around September 4 and ends in
the middle of June. They have a couple of weeks of a short break between December and
January as winter vacation. The school year is divided into two semesters, each lasting
about four months. Every student is taught various subjects during the semester. Most of
these schools operate based on a “time clock.” All students are evaluated for their under-
standing of the subjects taught by giving them “grades” after each semester. We can
rationalize this system, but the system is unfair to slow learners who could do well given
more time, even better than those who get good grades. It is also unjust to quick learners,
because they may be wasting their time once they complete what they are programmed to
learn. The school that operates based on the “time clock” justify their system by giving the
students “grades” such as “A, B, C, D, and F” at the end of each semester, which is called
the semester.
If schools operated based on a “function clock,” each student will be asked to learn a given
set of materials without any fixed time limit. Whenever the student masters the subject
matter, she/he can move onto the next level of learning after demonstrating that they
learned the subject matter. Some students may learn the subject matter in 45 days, whereas
another student may take 120 days to learn the same subject matter. The students move up
to the next level as soon as they master the subject matter. Under this system, the student
follows the “function clock,” which measures the progress made by each student in the
understanding of the subject matter. Under this system of education, every student may
achieve a complete understanding and comprehension of a given subject matter. Some
students may “graduate” in one calendar year, whereas some others in ten years. Some
countries in Asia had this system of individualized education before the twentieth century.

Example 20.2 Difference between Teaching and Learning in Terms of “Function

Clock” and “Time Clock”
All the teachers were once students, too. As a student, she/he needed to satisfy the
requirements and get passing grades. When someone becomes a teacher and teaches other
people for the first time, the new teacher realizes that as a teacher, he/she must fully
comprehend the subject matter. In other words, students operate based on “time clock,”
whereas the teacher must perform based on “function clock.” The intensity of effort needed
to operate under the “function clock” can be either higher or less than that required to
operate under the “time clock,” depending on one’s knowledge base.

For some of the machines used in semiconductor manufacturing such as the

“track machine” (shown in Fig. 20.1), it is essential that the machine continuously
runs. The consequence of the machine stopping in the middle of its operation can be
quite costly. To achieve the goal, some companies used the “if … then …” logic to
20 Functional Periodicity, “Function Clock,” and … 571

Fig. 20.1 SVG Track Machine, that cleans, coats with photoresist, and cures silicon wafer.
A track machine is typically linked to a lithography machine. The number of stations of some track
machines can be much larger than the one shown

schedule the wafer-handling robot in a track machine. Usually, such a system works
well in production, but once in a while, the system would stop because the accu-
mulated error is so significant that the system that uses “if … then …” logic to
operate the machine breaks down. Then, the machine has to be re-booted at a great
expense.
One way of overcoming this problem is to assess the state of the track machine
when a finished wafer leaves the machine. That is, determine where the remaining
wafers are located, what additional processing steps must be undertaken for each
wafer. Based on the information, transfer and process the wafers until the next
functional period, when the next wafer is ready to be removed from the track
machine. Then, again re-assess the state of each wafer in the machine and decide for
the next processing step. This “re-initialization” process should continue after each
“functional period.” This re-initialization should be done utilizing the information
stored in the computer and quick sensing of the state of each wafer. This
re-initialization process each time a finished wafer leaves the machine avoids the
need to pre-determine the entire operation of the track machine by creating the “if
… then …” rules a priori and eliminates the problem associated with accumulated
errors related to “if … then …” rules.
Unlike the “time clock,” the “function clock” starts ticking at the beginning of
each new functional period of the designed system. For example, in the track
machine used in semiconductor wafer processing, a new functional cycle of the
system begins when a finished wafer leaves the track machine. In manufacturing
airplanes, a new period starts when a finished aircraft leaves the assembly plant.
Schools re-initialize twice a year. In all of these systems, the FRs of the system are
“re-initialized” when a finished product leaves the system. Re-initialization is done
using the latest information available on the status of each unfinished work-piece in
572 N. P. Suh

the machine at the time of re-initialization. For example, when a finished wafer is
taken off the track machine, another wafer in the track machine must be moved into
the final module. The track machine is re-initialized after determining the status of
each wafer in each module of the machine. The data are used to determine the best
placement of the wafers for the next cycle. Based on the data of the initial state of
each piece at the beginning of a new period, the best operational sequence for the
new cycle can be determined. The “function clock” begins to “tick” at the begin-
ning of each functional period. The function clock is not synchronized with the
circadian clock. This process repeats as long as the system has to produce or
operate.

20.3 Pros and Cons of “Function Clock” Versus “Time

Clock”

Most of the societal functions are based on the “Time Clock” rather than the
“Function Clock.” In fact, the term “Function Clock” has not been used before and
is not a commonly used term. However, there are many reasons for using Time
Clock in many societal roles. The most important reason is that humans behave
based on the circadian cycle. It is the least resource-intensive to plan group
activities based on the Time Clock. Unlike controlling machine functions, the cost
of operating a system that involves people based on the Function Clock can be high.
Some people may not finish their tasks within an allocated time, slowing down the
operation of the entire system. On the other hand, a highly efficient person may
suffer in the system that uses “Time Clock” waiting for the less efficient person to
finish their task. Although there is no definite proof or evidence, the overall effi-
ciency of society and any organization with diverse activities and people with
different levels of competency would be much higher under a system that use the
“Time Clock.”
“Function clock” is expected to be more efficient in a system that repeats a set of
similar operations but with slight variations from cycle to cycle.

20.4 Periodicity and Re-initialization of Airline Scheduling

A major goal of airlines is to maximize their profit, just like most commercial firms.
To achieve this goal, they attempt to transport as many passengers as possible in the
shortest possible time. In the U.S., after airlines were de-regulated, airlines lowered
the airfare of the economy class, eliminated many in-flight services (i.e., meals),
and reduced the number of flight attendants. They also use smaller airplanes such as
Boeing 737 and fill these planes to its full capacity with almost no empty seats.
They often overbook, assuming that some customers will not show up. They also
increased the number of seats by reducing the legroom between the rows of seats.
20 Functional Periodicity, “Function Clock,” and … 573

Some of the airlines adopted these and other extreme measures to reduce the cost of
operation. Airline profits are affected by the cost of fuel, personnel, and equipment
(i.e., airplanes). When the oil price is low, airlines make healthy profits and suffer
when the fuel price is high. They tend to hedge against high oil prices.
Airlines provide direct, non-stop flight services between major hubs with suf-
ficient passenger loads such as the New York–San Francisco leg. However, they
eliminated many direct, non-stop flights between smaller cities because of the
limited passenger loads. Therefore, they use regional hubs to collect passengers
from diverse cities and then to transport them to their final destinations, using
smaller aircraft. To achieve this goal, the airline synchronizes their flights by letting
airplanes land and take off from a given airport nearly at the same time. Therefore,
airports are busy at certain times of the day. Many flights of a given airline arrive
and take off within a couple of hours of each other to allow passengers to transfer
and reach their destination without long layovers. For example, airplanes from
Europe typically arrive in Boston in the afternoon and flights from Boston to
European destinations leave early in the evening, which is around mid-night in
Europe due to the time difference. These airplanes then arrive in European cities
early in the morning, after the airports officially open. (Airplanes are not allowed to
land at the Frankfurt International Airport, Germany, before 6 a.m., probably for
noise reduction.) Therefore, airlines have introduced certain functional periodicity
to harmonize flights between various cities.
When severe weather forces airports to close, flights are either canceled or
diverted to other airports. If the bad weather persists for an extended period, the
airlines cancel their flights. However, when the weather clears, the airlines
“re-initialize” their systems by relocating their aircraft to the right airports. The fact
that at night the number of flights is small helps to relocate airplanes. Therefore,
they can dispatch aircraft to the correct airports during the night to resume regular
flights starting in the morning.
The important message of this example is that when a system is in operation for
a long time, it is inevitable that errors accumulate, eventually resulting in system
failures. These errors can be removed by re-initialing the system at the beginning of
each functional period. Nature has evolved through the solar system that has yearly
and daily periodicity. The periodic nature of biological systems has created func-
tional periodicity and re-initialization through the long history of natural evolution.
Exercise 20.1: Travel to “Solar System II” in Distant Universe in Year 2520
Assume that by the thirtieth century, human beings finally developed spacecraft that can
take people to another universe and found a solar system similar to ours. However, when
they got there, they found two significant differences from our solar system of Earth in the
twenty-first century. The planet revolves around its “sun” once every three months (rather
than once a year) and rotates about its axis every 12 h, i.e., twice the rotational speed of
Earth. The “people” on this newly discovered planet are very much like the people living
on Earth but with some differences. What would be the difference? How would the dif-
ference in functional periodicity affect people? Perhaps the information given in the next
section may help in thinking about this question.
574 N. P. Suh

20.5 Periodicity and Re-initialization of Biological

Systems: Cell Division and Mitosis of Cells

Cells are the basic building blocks of all living beings. Cells in our body divide into
two daughter cells, roughly once every 24 h. The phenomenon of cell division has
been known since 1835 when Hugo von Mohl, a botanist, discovered the cell
division in green algae in 1835.
Biological cells have functional periodicity, re-initializing after each cycle, i.e.,
cells replicate themselves periodically, following a sequence shown in Fig. 20.2.
After the cells split into two identical cells, each cell goes through three different
phases (in biology, it is called the “interphase”): the G1 phase, S-phase, and the G2
phase, before creating two identical cells through mitosis. The regular interval for
the cell division is equivalent to the functional periodicity of biological systems.
Another interesting fact is that the cell goes through re-initialization. The cell
division process may halt for a while by going into the G0 phase until it is ready to
resume the cell division process. This is equivalent to the “re-initialization” process.
Biology textbooks provide the details of this process.

Fig. 20.2 Cell cycle. A cell goes through the interphase consisting of G1, S, and G2 phases.
Through mitosis, one cell generates two daughter cells. Some cells go into G0 phase, where the cell
cycle is stopped until it is ready to repeat the cell division process by going through the cell
interphase and mitosis. (Reproduced with permission from DAntes Design (2019))
20 Functional Periodicity, “Function Clock,” and … 575

20.6 Periodicity and Re-initialization in Manufacturing

System Design

There are many different kinds of manufacturing systems, depending on the nature
of the product, the production rate, and the total number of products to be manu-
factured. In “old” days, a large number of identical parts were made by machining
and stacked up on a cart until the cart is fully loaded. Then, the cart was then taken
to the next machining station for additional machining or to an assembly station to
be joined with other parts. This type of manufacturing system is, in general, inef-
ficient, because the inventory of parts is extensive. They may also need a large
warehouse to keep the machined parts until they are required. Modern manufac-
turing factories may use a manufacturing cell, where a group of machines does
various operations to a part until it is finished. In this type of manufacturing cell, a
part is subjected to a series of manufacturing processes (e.g., cutting, milling, etc.).
Similarly, if the manufacturing operation is an assembly operation, a number of
assembly stations are placed in series to assemble the product in a continuously
moving platform.
Henry Ford’s Transformed Manufacturing System
One of the biggest revolutions in manufacturing is the mass-production system. The first
such system was the manufacturing assembly line introduced by Henry Ford in the early
twentieth century. He manufactured automobiles by assembling them on a continuous
moving platform. Workers assembled automobile parts on a moving “conveyer” system,
which substantially increased productivity and reduced the manufacturing cost of auto-
mobiles. Parts were brought to the assembly stations, which were assembled in a specific
sequence as the platform moved along carrying the partially assembled cars. The assembly
process was synchronized with the motion of the assembly line. When there is a problem at
any station, the entire assembly line is stopped until the problem is fixed. Modern assembly
plants roll out a new car one a minute.

20.6.1 Functional Periodicity in Design of a Simple

Manufacturing Operation

Consider a simple manufacturing system with four machines, a, b, c, and d. The

geometric shape of the part is so designed that it must be machined in all four
stations in the sequence given, i.e., a, b, c, and d. The problem to be solved is the
following: we want to maximize productivity and minimize the manufacturing cost.
The operational time for machining and loading/unloading for each machine is
different, the operational time at station b being the longest. The machining time at
stations c and d are the same. There are several different ways of setting up the
manufacturing system. One scheme is to run each machine as fast as the machine
can be operated. Since it takes the longest at station b, parts that have gone through
machining at the station a will pile up in front of station b. Station c will starve for
lack of parts since Station b is slower than Station c.
576 N. P. Suh

Table 20.1 Design matrix PVa PVb PVc PVd

for machining parts at four
machining stations DPa X 0 0 0
DPb X X 0 0
DPc X X X 0
DPd X X X X

The best manufacturing sequence in terms of manufacturing cost and produc-

tivity is to adopt a functional periodicity based on the speed of machine b. When
machine b is to begin a new machining cycle, other machines, i.e., a, c, and d,
should be started at the same time. Machines a, c, and d should be slowed down so
that they would finish their operation at about the same time as machine b. Slowing
down the cutting speed may prolong the tool life, reducing the manufacturing cost.
The relationship between DPs (i.e., the shape of the part) and PVs (i.e.,
machining operation) are given by the design matrix given in Table 20.1.
The design matrix shows that the manufacturing system is a decoupled design.
Machining operation on “a” affects the functions of machines “b”, “c”, and “d”.
Even if each machine is operated at its fastest speed, the productivity will not
increase. The production rate of machine b controls the number of parts that can be
assembled. The slowest machine controls the overall production rate. Machine b
controls the functional periodicity of this manufacturing system. Therefore, the
productivity is highest, and the cost is the least if all stations produce parts at the
rate machine b process the parts.

20.6.2 Functional Periodicity in Job Shop Scheduling

One of the classical problems in manufacturing research has been the scheduling of
job shops. A job shop is a manufacturing system with various machines to produce
machined parts for multiple customers on demand. The machining orders coming
into the job shop are unpredictable in terms of timing, specific machining needed to
make the part, the number of pieces, etc. Typically, in a job shop, a part is subjected
to various machining operations, and when the part is done with the final machining
operation, it is shipped out to the customer.
Consider a job shop with 20 machines of various kinds. Some of the commonly
used machines may be identical. However, the job shop must have a sufficient
variety of different tools to service whatever parts the customer orders for
machining. Typically, when an order comes in for machining, the foreman of the
shop attempts to process it as soon as possible based on a rule such as
first-in/first-out (FIFO). Each part may be subjected to several different machining
operations. Since several different components are machined in the job shop at the
same time, this FIFO process may not be the most efficient way of utilizing the
machine shop. Some parts will take up too much time on some machines, thus
20 Functional Periodicity, “Function Clock,” and … 577

denying access to these machines for other jobs. A classical question related to the
job shop that has not been answered thus far is “How should the job shop schedule
their operation to maximize the productivity of the job shop?”.
A Possible Solution for the Job Shop Scheduling Problem Based on Functional
Periodicity and Re-initialization:
When the first machining order comes into an empty machine shop, the issue is
simple. i.e., use the available machines to manufacture the part. However, once the
machine shop is ‘fully loaded’ with lots of orders, scheduling the sequence of
machining different parts becomes a critical issue for maximizing the productivity
of the job shop. The scheduling task is complicated because the job shop does not
know the future incoming orders. How should we schedule the operation of the job
shop? The answer to this question has eluded many researchers and scholars in
production engineering.
Consider the manufacture of a part that requires the following machining
sequence: milling on a milling machine, followed by machining on a lathe, grinding
on a surface grinder, and electric-discharge machining (EDM). The last operation
before the part leaves the job shop is the EDM process. We can set the functional
period to be whenever the finished piece leaves the job shop. That is, for this
particular part, when the EDM is done with the part, it will be the end of the
functional period. We then re-initialize the parts still being processed to determine
the next operation. The next part that is finished and leaves the job shop will be the
end of another functional period, and the beginning of a new functional period and
the time to re-initialize.
In job shops, the incoming orders are random and unpredictable in terms of
specific machining tasks and the number of parts to be machined. The problem is to
enable the job shop to schedule which part should be machined by what machine
and in what sequence. It has been a difficult problem to solve, although many
attempts have been made in the past. We will apply the functional periodicity and
re-initialization to solve this job shop scheduling problem.
For this job shop problem, we will assume the following:

1. A new functional period to begin at that instant the latest finished part leaves the
job shop, creating an empty machine, i.e., machine not occupied by a machining
operation.
2. At that instant, a machine becomes available for another new workpiece.
3. At the beginning of this new cycle, re-initialize the entire system. The data
include the status of each part, the status of each machine, and how long it
would take to finish the piece.
4. Move partially finished parts to machines where subsequent machining should
be performed.
5. Load a new piece to an unoccupied machine after moving the parts that have
been machined at another station for following machining operations. This
process can continue as long as there is a new stock to be machined.
6. There will be a buffer station to accommodate parts that cannot be moved to the
next machining station right away, although it is done with a given operation.
578 N. P. Suh

After one functional period, i.e., when a completed part leaves the job shop, we
assess the status of all the parts that are currently being processed. Then decide the
best way of handling the remaining operations, i.e., which machining operation
should be done next for part A, etc. To achieve this goal, we should make sure that
functional independence is always maintained. Because we determine the quickest
sequence of processing the parts in the job shop at the end of each functional
period, this solution should maximize the productivity of the job shop. The pre-
diction of which part will be the next that will be leaving the job can also be
determined at the time of re-initialization. However, the uncertainty increases,
because the next part that will be coming into the job shop is not known a priori. If
we develop an algorithm, then we can do this more or less instantly in a computer to
determine the next sequence of machining operations for the parts in the job shop at
that instant.
Considering the forgoing discussion on the job shop scheduling problem, the
FRs of a job shop may be stated as follows:

FR1 = Fulfill machining orders as quickly as possible.

FR1 may be decomposed as follows:

FR11 = Maximize the flow rate of completed job orders through the job shop;
FR12 = Minimize the number of idle machines;
DP11 = Scheduling algorithm;
DP12 = Load a part when a machine becomes available.

FR11 and DP11 may be decomposed as follows:

FR111 = Introduce functional periodicity to the job shop operation;

FR112 = Re-initialize the system when a completed part leaves the job shop,
determining the remaining time on the current machine;
DP111 = Completion of a part for delivery;
DP112 = Assess the status of each part in the job shop, i.e., remaining machining
time and subsequent machining operation to complete the part.

FR12 and DP12 may be decomposed as follows:

FR121 = Calculate the remaining machining time;

FR122 = Load empty machines at the time of re-initialization;
FR123 = Provide buffers for partially finished parts awaiting the next operation;
DP121 = Algorithm for calculating remaining machining time for all parts
currently being worked on by various machines;
DP122 = Bring in a new part to the job shop;
DP123 = Buffer table.
20 Functional Periodicity, “Function Clock,” and … 579

Exercise 20.2: Job Scheduling

A job shop has two horizontal milling machines, two lathes, three drill presses, and a
grinding machine. The job shop has received orders for the two parts: Part a and Part b.
They have to make 50 pieces of each. For Part a, it must be milled for 15 min; turned on a
lathe for 23 min; drilled to make two holes for 5 min each, and ground in the grinding
machine for 35 min. For Part b, it must be machined in the machining center for 10 min,
drilled for five holes which will take 7 min each, and ground by the grinding machine for
12 min. The loading time and unloading time of the parts on each machine are 6 min and
3 min, respectively. Develop the best schedule for the job shop.

Exercise 20.3: Scheduling for Construction Site

The engineer-to-order industry is under constant pressure to optimize production and
deliver the right components at the right time. For instance, in the building industry, they
have to install their parts at the construction site. Synchronization between fabrication and
on-site installation is difficult to realize. Outline a plan of real-time-capable production
planning and control in engineer-to-order companies by minimizing time-dependent
combinatorial complexity in the value chain.

Exercise 20.4: Throughput Rate in Emergency Room (ER) of Hospitals

Emergency rooms (ER) of hospitals are one of the most demanding departments of hos-
pitals to operate. Patients with many kinds of different illnesses come in for treatments
without any prior appointments. Some with the urgent need for care, such as those who
were in automobile accidents and gunshot wounds, need quick attention when they show
up. Some patients with chronic illness come in because they feel worse than usual. The goal
is to take care of ER patients well as efficiently as possible. The FRs of ER may be assumed
to be the following three:

FR1 = Prioritize ER patients based on the urgency of the illness of the patients;
FR2 = Separate patients who need to be admitted to the inpatient unit;
FR3 = Estimate the duration of treatment;
FR4 = Fast track (FT) patients who can be discharged after quick treatment.
Design a system that will maximize the patient flow rate through the emergency room of a
city hospital.

20.7 Periodicity and Re-initialization of Political Systems

Conceptually, it may be easier to understand the significance of periodicity and

re-initialization in the design of systems by considering the political system. Many
different political systems exist. (For example, the dictatorship a la North Korea;
presidential system, a la the United States; parliamentary system, a la the United
Kingdom, and a political system that is an amalgam of these systems.) Here, we
will compare the three: the presidential system, the parliamentary system, and
dictatorship. The key characteristics of these two systems are as follows:
580 N. P. Suh

1. Key features of a Presidential System:

• president is elected by direct vote of voters or by electors elected by popular

vote;
• fixed length of the term in power (periodicity and re-initialization applies);
• checks and balance: Power sharing with legislative body, and judiciary;
• annual budgetary approval.

2. Key feature of a Parliamentary System:

• the prime minister is elected by majority party in legislature;

• no fixed length of the term in power, election has to be called once in a fixed
period;
• checks and balance: Power sharing with legislative body, and judiciary;
• annual budgetary approval.

3. Key feature of a dictatorship:

• dictator controls power, often, until the dictator’s death or forceful removal;
• no fixed length of the term in power;
• no checks and balance;
• personal enrichment.

Under the presidential system, major new policies are advocated at the beginning
of the president’s term, which generally lasts four to six years. Then, the president
has the rest of the term to implement the principal legislation, even if it is
unpopular. Congress must pass new legislation that supports the new primary
policy of the president if they are needed. The central power of Congress is the
control of the budget. During the term of office, a presidential system allows the
implementation of even unpopular changes because, during the term of office, the
president cannot be removed except through impeachment. Functional periodicity
exists in this kind of presidency. Re-initialization takes place at once every four
years in the United States and every six years in France. This presidential appears to
be superior to the parliamentary system because of a clear initiation point for
re-initialization.
Under the parliamentary system, there has to be a general election at least every
four or five years. As long as the party in power has a majority, they can re-elect
prime minister as often as they like. If the majority party does not support the prime
minister, the cabinet is dissolved, and a new election is held. The prime minister can
set a new set of policies when the PM has the backing of the majority party when
they take power. It is difficult to design new legislation on a long-term basis unless
the party in control has a supermajority in the parliament and unified among
themselves. The end of the term is not clear, although there is a period during which
20 Functional Periodicity, “Function Clock,” and … 581

a new election does not have to be called. Some times the prime minister can call a
snap election to prolong the control of power. In a parliamentary system, it is more
challenging to create functional periodicity and re-initialize the system. Govern-
ments can be volatile, changing hands whenever difficult issues must be resolved.
In general, the worst political system is a dictatorship. In most cases, it does not
have functional periodicity and re-initialization process, since there are no set terms
for tenure. Often the consequence of losing their power is a significant upheaval.
Some dictators seize control with good intentions, but they eventually lead the
country into turmoil because of the absence of functional periodicity. The transition
from dictatorship to a political system with functional periodicity is challenging.
Many dictatorships fail, but the dictator tends to amass a significant fortune or leave
the power with an unfortunate end. Ultimately, the country loses out when a dictator
takes over a country.
When corruption in government is an over-riding issue, functional periodicity is
even more critical. Unfortunately, in many of the countries where corruption of
public officials is prevalent, it may be difficult to institute a functional periodicity.
Under such a system, re-initialization may not be implemented peacefully.
The Miracle of South Korea and Dictatorship:
In 2018, most people in the world who visit Korea found it to be a modern and prosperous
democratic nation built on an educated workforce and technology. Korea had gone
through tumultuous periods since the beginning of the twentieth century to be where she is
today. Korea is one of the few nations that transformed themselves through several
re-initializations and the adoption of the political functional period that was part of the
constitution.
Because of the political ineptitude and corruption of the ruling class, Korean people
suffered for almost seven decades. Japan colonized Korea from 1909 until 1945. In 1945,
Korea was split into two by the U.S. and Soviet Union, and then in 1950, communist North
Korea invaded South Korea, which destroyed the country. Since 1953, the Koreas main-
tained armistice. In 1961, General Park Chung Hi took over South Korea through coup de
ta. Then, Korea was an impoverished country with one of the lowest GDP in the world.
Many people give him credit for having created modern Korea. Modern Korea can trace
back the beginning of many things that has made Korea a modern nation to him. He was
not corrupt and was a visionary of what a modern country should be like. Most people
would agree that he had built a modern nation in 50 years. During his presidency, Korea
had a façade of democracy but was really under one-person control. He built from
highway, world-class industrial firms, educational infrastructure that is highly competitive
worldwide, but he was a dictator. In 1960, ‘Korea’s per capita income was than that of the
Philippines and Egypt. Today ‘Korea’s economy is in the top ten in the world and has one
of the most modern infrastructures. President Park had a vision, and he realized his vision.
His economic plan set his government’s functional periodicity. The ten-year period of the
1960s may be considered one period, where the emphasis was building on national
infrastructure using low-cost labor. His second period of functional periodicity was in the
1970s, during which he emphasized the heavy industries, shipbuilding, automobiles, con-
sumer electronics, and high-speed rails. Unlike the 1960s, the 1970s were a difficult period
because the return on investment was not immediate in these heavy industries. After his
assassination in 1979, Korea finally reached its initial goal of creating the base for heavy
industries, unlike labor-intensive apparel business.
582 N. P. Suh

Throughout the presidency of President Park, Korea maintained the appearance of the
democratic form of government, although the president had ruled the country with absolute
power. He extended his tenure for more than two terms thorough questionable political
maneuvers. He was an absolute dictator. The question remains: would Korea be where it is
today economically if President Park Chung Hi served only two four-year terms? Two of
the presidents who succeeded President Park, who used to be army generals, ended up in
jails after their presidencies were over because of their financial corruption while in power.
President Park did not enrich himself while in office. Fortunately for Korea, the industries
that were nurtured by President Park became highly successful industrial firms, several
becoming world leaders in their respective fields.
After the demise of two generals who became presidents after the assassination of Presi-
dent Park, Korea has become a genuinely democratic nation. The civilian presidents, who
succeeded the generals, brought in real democratic reform without harming industrial
infrastructure and economic development. Today, Korea is a truly free democratic country.
Such an open political environment nurtured its economic growth, the foundation of which
were laid during the Park presidency has been no less than spectacular and stellar. Today,
Korea is one of the top ten economies of the world. One of the significant factors for this
economic and technological growth is the zeal of Korea people for education in general
and higher education in particular. 70% of high school graduates go to college. And
equally important is the willingness to make massive investments in new technologies by
major companies in Korea.
With economic prosperity and democratization of the political system, Korea has become a
genuinely democratic nation. It has strengthened the independence of the three branches of
the government. The president of the country heads the executive branch. Its legislative
body is active with the usual chaos. The Supreme Court heads the judiciary branch. The
system is similar to those of the United States of America. When the Korea War ended in
1953, Korea was one of the poorest countries in the world without any industry, natural
resources, electric power, and infrastructure.

Exercise 20.5:
Many leading industrial firms of the 1960s are no longer leading industrial firms in their
country or the world in the early part of the twenty-first century. Identify one of the large
industrial firms that fit this description and show how you might revive the firm through
re-initialization.

Exercise 20.6:
Some of the countries in the world have not improved their economies during the past
several decades. How would you re-design their economic and industrial policies to
improve their economy and quality of life?

20.8 Functional Periodicity and Re-initialization

of Educational Systems

20.8.1 Academic Calendar

The school calendar is periodic and cyclic. The school year is punctuated by winter
vacation and summer vacation throughout the world. Time off from intensive
20 Functional Periodicity, “Function Clock,” and … 583

learning in schools is necessary to give young people to grow intellectually, psy-

chologically, and physically. During the summer vacation, some students worked
on family farms and other jobs, which helped their families and their finances.
During the long summer holidays, students also rest and re-charge to meet another
year’s challenges. After each academic year, students are promoted to the next
grade or level. In some ways, it is similar to the cell cycle of physical systems.

20.8.2 Research Universities

Before the Second World War, most universities in the world did not do extensive
research. They were primarily teaching institutions. However, during World War II,
some of these universities in the United States, especially those with strong science
and engineering schools, actively researched to help the war effort. As the war was
coming to an end, the United States government, under the directives of President
Frank Roosevelt, decided to continue the involvement of universities in national
efforts. The goal was to strengthen economic competitiveness and national security
by enhancing universities. In 1950, the U.S. National Science Foundation (the NSF
Act of 1950 of the U.S.A.) was established to support university research to pro-
mote science and engineering. It has provided funding for basic research in science
and engineering.
During the Cold War following the Second World War, some universities in the
United States received substantial research support through various government
agencies and minor assistance from industrial firms. Today, many of the top 20
universities in the world are the universities in the United States. They have con-
ducted extensive research mostly with public funds, a significant portion of which
was provided by defense-related departments and agencies of the United States
government. It should be noted that in 2019, the United States spent more on
defense than the next 19 nations combined, including Russia, Germany, Japan, the
United Kingdom, France, Korea, and China.
In the United States, the universities researched the significant investment made
by the U.S. government through its many agencies and departments. Major uni-
versities established specialized laboratories devoted to defense and space-related
research such as the Jet Propulsion Laboratory of California Institute of Technol-
ogy, the Lincoln Laboratory of MIT, and the Applied Research Laboratory of Johns
Hopkins University. In addition to NSF, the U.S. Government invested in defense-
related research through the Defense Advanced Research Projects Agency
(DARPA), the Office of Naval Research (ONR), the Air Force Office of Scientific
Research (AFOSR), and as part of the development of various defense technologies
as part of the advanced weapon development. The U.S. government also funded
five National Laboratories such as the Lawrence Livermore National Laboratory,
Sandia National Laboratory, Oak Ridge National Laboratory, Argonne National
Laboratory, and Los Alamos National Laboratory. The U. S. government also
supported basic research in the field of biology and health sciences, which has
created some of the most prominent universities in these fields.
584 N. P. Suh

The research conducted by universities has created a new culture in universities

as well as in the business world of generating new business ventures funded by
venture capital (VC). The well-known firms that were created as a result of this new
culture entrepreneurship are such firms as Microsoft, Apple, Google, Facebook,
Amazon, and many others. These firms were created based on “innovative design
using software.” It is noteworthy that the combined revenue of these firms is much
larger than the income of all oil companies and automotive companies of the world
combined.

20.9 Conclusions

This chapter examined the design of systems that have cyclic behaviors. Two kinds
of systems were considered: (a) a system that satisfies the same set of FRs cycli-
cally; and (b) a system that satisfies the same set of FRs, but requires
re-initialization of the system at the beginning of each new cycle due to variations
of DPs in the previous period. An example of the second kind of a system that
requires re-initialization is a biological cell that undergoes mitosis. The latter sys-
tem follows a “function clock.” That is, at the beginning of each new functional
period, the FRs of the system must be re-initialized to assure the optimum operation
in the subsequent cycle.
Problems
Take any two of the Exercise problems given in the main text and provide
solutions.

Reference
DAntes Design (2019) Division cycle of eukaryotic cell divided into four phases: G1, s, g2 and
mitosis. Photo licensed from Shutterstock, ID: 1251867079

Further Reading

Alberts B, Bray D, Lewis J, Raff M, Roberts K, Watson JD (1994) Molecular biology of the cell,
3rd edn. Garland Science
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Suh NP (2005) Complexity. Oxford University Press
Artificial Intelligence in Design: A Look
into the Future of Axiomatic Design 21
Erwin Rauch and Dominik T. Matt

Abstract
In the previous chapters, we learned all about Axiomatic Design (AD), where
AD comes from, how it was developed, the axioms, complexity theory, and
in-depth content for learning AD. This chapter looks at how AD will evolve in
the future and introduces an approach based on Artificial Intelligence (AI). The
approach presented is to be understood as a hypothesis to venture into the future
together with the students analyzing how AI will change not only their daily
lives, but also the work of a design engineer.
Many of the technological innovations related to Industry 4.0 pave the way
for achieving a next level in engineering and especially in manufacturing. For
the next decisive steps, we need the combination of disciplines such as
engineering design and AI to create assistance systems for designers of highly
complex systems, which support them step by step in the design process as well
as for the re-design of systems affected by time-dependent complexity.
In the proposed AI-assisted design approach, we propose to combine AD as
an established and proven theory for the design of complex systems with the
latest methods of AI. For the design phase, AI offers an enormous potential to
transform customer needs into functional requirements (FRs) and to support the
designer in identifying and selecting the best design solution for a design
problem based on existing data sets of previous successful or not successful
designs.

E. Rauch (&)
D. T. Matt

Faculty of Science and Technology, Free University of Bolzano-Bozen, Universitätsplatz 1,
Bolzano, Italy
e-mail: [email protected]
D. T. Matt
Fraunhofer Research Italia s.c.a.r.l., Via A.-Volta 13a, 39100 Bolzano, Italy

© Springer Nature Switzerland AG 2021 585

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_21
586 E. Rauch and D. T. Matt

Similarly, the combination of modern technologies for data collection and AI

also holds great potential to monitor complex systems determined by
time-dependent complexity, like manufacturing systems. Based on improved
technologies for data collection and processing as well as AI we can generate
data-based predictive suggestions for system adaptation or even optimization,
either with or without the interaction of the designer.
The presented AI-assisted design approach has potential to usher in the next
era of engineering design, which will take us a huge step toward the vision of
intelligent and self-optimizing systems.

21.1 Introduction

For many years, designers have dreamt about having an intelligent design machine
that automatically generates designs or design concepts superior to those currently
available. Already in 1990, Suh and Sekimoto (1990) presented the idea of the
“Thinking Design Machine” (TDM) with the aim to provide designers a powerful
and computer-aided design tool to improve the quality of the design and to reduce
the time needed for creating a high-level design concept.
Since 1990, the world has changed rapidly and significantly. Computer-aided
tools have been improved and further developed in order to take later stages of the
design process on a new level. Especially recent developments in computer-aided
design (CAD) and computer-aided engineering (CAE) facilitated the work of
designers in geometrical modeling and drawing as well as in structural and dynamic
engineering of parts and products. In the meantime all larger and most of the
smaller enterprises are equipped with latest computer software for 3D CAD
drawing, which allows a completely new level of visualization and a better platform
for discussing designs together with other specialists in the product development
and realization process (manufacturing specialists, material specialists, quality
experts, …). Technologies like Virtual Reality (VR) give designers now the pos-
sibility to immerse in the virtual world and to inspect their design in a virtual but
realistic and three-dimensional environment by using VR headsets for some hun-
dreds of dollars of investment. Such powerful 3D CAD tools allow to creation of a
Virtual Mock-Up (VMU). Instead of testing a product design via crash tests or
fatigue tests in the laboratory using Physical Mock-Ups (PMU), most of the testing
can be done nowadays on a virtual level. As a result, costs for engineering and time
to market could be reduced significantly over the last 30 years. Examples of such
tools in the engineering phase of product development are tools for Finite Element
Analysis (FEA) or multibody dynamics simulation. The ongoing trend toward
open-source software especially allows also smaller enterprises to get access to
simulation software and modern CAE technologies. Further developments in
computational engineering will lead also in the near future to a further increase of
tools that assist designers in their daily work.
21 Artificial Intelligence in Design: A Look into the Future … 587

While the last 30 years show many innovations in the later stages of the design
and engineering process, there is little to no computational aid system available
today in the early and conceptual design stage. As the most critical decisions are
usually taken in the conceptual design phase this implies in many cases that a
design with a lot of potential for optimization is handed over to the colleagues in
CAD and CAE and thus creating problems and costs in later stages of the design or
even worse during manufacturing or assembly of the product. Thirty years ago
designers using AD in the conceptual design phase had only Microsoft Excel
spreadsheets with very limited space and no automation to document and visualize
their design decompositions and matrixes in a digital way. Nearly 20 years ago,
Acclaro DFSS has been launched providing the designer with a computational tool
to support the AD process in the early design stage. Acclaro DFSS allows the
designer to document the steps from one domain to another by encoding CNs, FRs,
and DPs. During the mapping and decomposition process, Acclaro DFSS helps the
designer significantly to analyze the design according to Axiom 1 (Independence
Axiom) providing an automatic check if a design is coupled, decoupled, or
uncoupled. Further, the software tool includes a function to rearrange a decoupled
design matrix in such a way that the designer automatically gets the ideal sequence
for implementation. In addition to the before-mentioned functionalities, Acclaro
DFSS provides different forms for visualization like the design matrix, the FR–DP
tree, or process flowcharts.
After this pioneering step forward in the automation of the design process and
computational support of designer using AD, there has not been any further in-
novation over the last nearly 20 years. Compared to the developments in CAD and
CAE as well as in manufacturing and assembly (e.g., rapid increase in flexible
automation, robotics, advanced manufacturing technologies as well as smart and
connected factories) the early stage of design is still working with “stone-age”
design tools.
The actual design solution depends mostly on the individual experience and
knowledge of the designer itself. Up to now, there is no tool available for archiving
past (successful and not successful) designs and to support the designer in
decision-making or in finding better design solutions outside of the individual
experience and knowledge of the designer. At the same time, we can observe that
since some years AI has become a trend and will become much more important in
the next years. Due to the introduction of the concept of Industry 4.0 (mainly used in
Europe and Asia) and Smart Manufacturing (used in the US) data has become a new
status in engineering. Based on new sensor technologies, industrial Internet, and
smart and connected factories the amount of available data increased exponentially.
This leads currently to a stronger focus on computational methods for taking
advantage of this new quantity and quality of data. According to many experts, “data
will become the new gold” and those that are able to use advanced AI methods for
analyzing data in an intelligent way will benefit from a competitive advantage on the
global market. The race to become leader in AI has already started. United States,
China, and Europe are already developing and implementing important initiatives to
develop their competences in AI.
588 E. Rauch and D. T. Matt

Therefore, the hypothesis for the future development of AD is to take advantage

of recent developments in AI. The time has come to make use of computational
power, Big Data technology, and new AI methods to fundamentally renew and
automate the design process. After many years of insufficient computational aid, AI
has the potential to introduce a new era of AD and to realize the dream of the
Thinking Design Machine.

21.2 Artificial Intelligence—The Next Hype?

AI is currently on everyone’s lips. There are many examples of AI in our lives.

Apple’s Siri is one such example. Another is Amazon’s Alexa. Natural language
processing technology, a form of AI, is used to translate languages in Google
Translate. Indeed, up to $30B has been invested in AI in the past five years and
90% of it on research and development by companies such as Google and Microsoft
(Bini 2018).
In the annual Gartner Hype Cycle Curve, new promising innovations are
examined again and again and their status is presented on the so-called “hype cycle
curve.” The main aim is to show to what extent a certain technology is still in an
initial “hype phase” or whether it has already reached the “plateau of productivity”
and is thus ready for practical application. AI is currently at the zenith of the hype
curve, which means that a great deal of future potential is currently seen in this
technology, but the extent to which AI will really change our lives and work is still
outstanding. Besides the general Gartner Hype Cycle Curve there is also an own
hype cycle curve for AI technologies, which allows an even deeper view into the
current development and the future technology leaps of AI (Goasduff 2019). The
following can be read from this curve, for example:

• Speech Recognition has already reached the “plateau of productivity” and is

already used in many daily applications.
• Text- or voice-based Chatbots are still on the top of the hype curve offering a
high potential for increasing efficiency in customer service. For example, the car
manufacturer KIA talks to 115,000 users per week, or Lidl’s Winebot Margot
provides guidance on which wine to buy and tips on food pairings. Chatbots are
changing customer service from “the user learns the interface” to “the chatbot is
learning what the user wants.”
• Machine Learning already passed the top of the hype curve moving toward a
more realistic use of this technology. ML uses mathematical models to extract
knowledge and patterns from data. Adoption of ML is increasing as organiza-
tions encounter exponential growth of data volumes and advancements in
computer infrastructure. For example, Volvo uses data to help predict when parts
might fail or when vehicles need servicing, improving its vehicle safety.
21 Artificial Intelligence in Design: A Look into the Future … 589

• Augmented Intelligence is at the beginning of the hype curve with an expected

time of 2–5 years to reach a practical use in daily business. Augmented intel-
ligence is a human-centered partnership of people and AI working together to
enhance cognitive performance. It focuses on AI’s assistive role in advancing
human capabilities. AI interacting with people and improving what they already
know reduces mistakes and routine work. The goal of augmented intelligence is
to be more efficient with automation, while complementing it with a human
touch and common sense to manage the risks of decision automation.

In other words, AI will have a huge impact on both our daily lives and the world
we work in, perhaps more than any other technological innovation in recent years.
This also means that engineers and engineering students should be more involved
with AI to be prepared for these technological changes. For this reason, this chapter
pays special attention to this topic. However, before we get into the application of
AI in engineering and then in engineering design, we first want to better understand
what we mean with terms such as AI, machine learning (ML), or deep learning
(DL). Basically, we can say that these three terms are different concepts of different
levels. In general, deep learning is a subset of machine learning, and machine
learning is a subset of AI (Garbade 2018; Nicholson 2019) (Fig. 21.1).
Artificial Intelligence can be seen as the science and engineering of making
intelligent machines. AI is a branch of computer science dealing with the simulation
of intelligent behaviors in computers. A goal of AI is to further develop the
capability of a machine to imitate intelligent human behaviors. In AI, a computer
system is able to perform tasks that normally require human intelligence, such as
visual perception, speech recognition, decision-making, and translation between
languages (Skymind 2019). The main objective of AI is to teach the machines to
respond like humans do to flows of data. Although AI is a branch of computer

ArƟficial Intelligence (AI)

Engineering science of making intelligent machines using logic, rules, decision trees and machine
learning

Machine Learning (ML)

Learning algorithms based on staƟsƟcal techniques and structured data
/experience without the need of being programmed

Deep Learning (DL)

Learning based on
mulƟ-layer deep
neural networks able
to process also
unstructured data

Fig. 21.1 Artificial intelligence, machine learning, and deep learning

590 E. Rauch and D. T. Matt

science, there is hardly any field which is unaffected by this technology (Sadiku
et al. 2019).
Machine learning is a subset of AI. That is, all ML counts as AI, but not all AI
counts as ML. For example, rules engines, expert systems, and knowledge graphs—
could all be described as AI, and none of them are machine learning. To give an
example for rules engines: rules engines could be like an accountant system with
knowledge of the tax code, which takes information you feed it, runs the infor-
mation through a set of static rules, and gives you the amount of taxes you owe as a
result. One aspect that separates machine learning from the knowledge graphs and
expert systems is its ability to modify itself when exposed to more data; i.e.,
machine learning is dynamic and does not require human intervention to make
certain changes. That makes it less brittle, and less reliant on human experts. As the
name suggests, machine learning can be loosely interpreted to mean empowering
computer systems with the ability to “learn.” The intention of ML is to enable
machines to learn by themselves using the provided data and make accurate pre-
dictions. In 1959, Arthur Samuel, one of the pioneers of machine learning, defined
machine learning as a “field of study that gives computers the ability to learn
without being explicitly programmed.” The “learning” part of machine learning
means that ML algorithms attempt to optimize along a certain dimension; i.e., they
usually try to minimize error or maximize the likelihood of their predictions being
true. This has three names: an error function, a loss function, or an objective
function, because the algorithm has an objective. Here we need also “neural net-
works.” They keep on measuring the error and modifying the parameters until they
cannot achieve any less error (Garbade 2018; Skymind 2019).
Deep Learning is a subset of machine learning and also the next evolution of
machine learning. DL algorithms are roughly inspired by the information pro-
cessing patterns found in the human brain. Whenever we receive a new information,
the brain tries to compare it to a known item before making sense of it, which is the
same concept deep learning algorithms employ. Usually, when people use the term
deep learning, they are referring to deep artificial neural networks, and somewhat
less frequently to deep reinforcement learning. Deep artificial neural networks are a
set of algorithms that have set new records in accuracy for many important prob-
lems, such as image recognition, sound recognition, recommender systems, natural
language processing, etc. “Deep” is a technical term. It refers to the number of
layers in a neural network making it able to process also unstructured data com-
pared to machine learning techniques where features for classification need to be
provided manually. Multiple hidden layers allow deep neural networks to learn
features of the data in a so-called feature hierarchy, because simple features (e.g.,
two pixels) recombine from one layer to the next, to form more complex features
(e.g., a line). Nets with many layers pass input data (features) through more
mathematical operations than nets with few layers, and are therefore more com-
putationally intensive to train. Requirements of DL are high-end computing
machines and considerably big amounts of training data to deliver accurate results
(Garbade 2018; Skymind 2019).
21 Artificial Intelligence in Design: A Look into the Future … 591

21.3 Examples of Artificial Intelligence Applications

in Engineering

The literature contains various applications of Industrial AI in the fields of engi-

neering and manufacturing. In the following, we want to give an overview of
examples and applications of Industrial AI in engineering.
AI Technologies are already used also for engineering design. As an example,
the generative design platform DesIA uses an object-oriented and open-source
language to describe the specifications and the desired product/system. Afterwards,
a rule based system combined with decision trees generates a number of admissible
design concepts. In the next step, machine learning algorithms are used to choose
the best design concept. In the last step, modern and advanced simulation and CAE
tools help the designer to optimize the concept in its details (Masfaraud and
Dumouchel 2019).
Kumar (2017) in his literature review describes a number of applications in
production planning. AI technology has already been used in many
computer-aided process planning (CAPP) applications in the past. Furthermore, AI
is used in knowledge-based expert systems where AI technologies access the
experience of experts (collected in databases) and give a designer or user sugges-
tions for design solutions.
A further field of application of AI is domestic or industrial robotics. Today’s
AI-powered robots, or at least those machines deemed as such, possess no natural
general intelligence, but they are capable of solving problems and “thinking” in a
limited capacity. From working on assembly lines at Tesla to teaching Japanese
students English, examples of AI in the field of robotics are plentiful. Home robots
use AI to scan room size, identify obstacles and remember the most efficient routes
for cleaning (Daley 2018). Fanuc, the robot manufacturer, uses AI-based tools to
simplify to teach industrial robots to do their work. AI simplifies the training
process, so the human operator just needs to look at a photo of parts jumbled in a
bin on a screen and taps a few examples of what needs to be picked up, like
showing a small child how to sort toys. This is significantly a less training than
what typical vision-based sensors need and can also be used to train several robots
at once (Shu 2019).
In maintenance, AI technologies are used for realizing predictive maintenance
in the form of machine learning and artificial neural networks to formulate pre-
dictions regarding asset malfunction. Knowing that a certain component of a
machine will fail with a defined probability on a certain day and at a certain time
has an enormous influence on the way we organize maintenance work in the
company in the future. AI in predictive maintenance allows for drastic reductions in
costly unplanned downtime, as well as for extending the remaining useful life
(RUL) of production machines and equipment. In cases where maintenance is
unavoidable, technicians are briefed ahead of time on which components need
inspection and which tools and methods to use, resulting in very focused repairs
that are scheduled in advance (Kushmaro 2018).
592 E. Rauch and D. T. Matt

Another potential field of application of AI in industry is quality management

and quality inspection. Manufacturing units that make complicated items like
microchips and circuit sheets are already making use of machine vision, which
furnishes AI with amazingly high-goal cameras. Advanced vision systems com-
bined with AI algorithms can pick even minute defects unmistakably, more reliably
than the human eyes. Defects are identified immediately and a response is auto-
matically configured, sent, and managed in order to reduce inefficiencies and waste
of material due to non-quality. This helps to increase productivity and at the same
time to improve, also, the ecological sustainability of modern manufacturing pro-
cesses (Admin 2019).
The use of Industrial AI is not only increasing in manufacturing but also in
logistics and supply chain management (SCM). Supply chain planning is among
the most important activities included in SCM strategy. Therefore, it is crucial to
have reliable tools for developing efficient plans. Implementing AI or machine
learning, the supply chain decision-making processes can be optimized signifi-
cantly. The advantage in logistics and SCM is that we usually have a lot of data,
which is a prerequisite for using AI or ML techniques. Analyzing huge data sets and
applying intelligent algorithms, we can balance demand and supply, and at the same
time optimize the delivery processes. In addition, human intervention is minimal.
AI algorithms will do everything autonomously and save companies from making
mistakes.
In addition to the above-mentioned examples of AI in engineering, a number of
other examples for the application of AI could be listed (warehouse management,
manufacturing system design, self-optimized machining, etc.). In summary, it can
be said that AI is still in its infancy in the industrial environment and that not all
potentials have been exhausted. In the future, the interdisciplinary combination of
engineering competences and AI methods from computer science will enable us to
take full advantage of AI in industry and engineering.

21.4 Axiomatic Design Knowledge Database as Basis

for Artificial Intelligence in Axiomatic Design

As already mentioned in Suh and Sekimoto (1990), a computational aid system for
AD requires a database. According to them, such a database needs to have at least
enough knowledge to give the designer plausible design solutions. Such a database
should also have the knowledge of many designers by providing a function to store
past designs and to retrieve possible design solutions for a common set and subset
of FRs. A FR–DP database may also be convenient to evaluate various functional
aspects of a candidate DP, such as side effects not considered by the designer.
According to Suh and Sekimoto (1990), DPs may be constructed and stored in
different ways, such as (i) parts/components, (ii) systems and subsystems,
(iii) materials, and (iv) physical phenomena/status. Since 1990, such a database has
never been realized as the technical possibilities for storage of Big Data has been
21 Artificial Intelligence in Design: A Look into the Future … 593

limited over the last years. Also Floss and Talavage (1990) already proposed in
1990 such a knowledge-based design assistant. Khan and Day (2002) developed a
similar concept of a knowledge-based database some years later.
Due to poor scalability and low performance, many traditional computing
technologies were inadequate for handling Big Data, which are characterized by the
volume, velocity, variety, and veracity of the data (Cheng et al. 2018). The latest
developments in the area of Big Data today also allow large amounts of data to be
handled and processed. Big Data concern large-volume, complex, growing data sets
with multiple, autonomous sources. With the fast development of networking, data
storage, and the data collection capacity, Big Data are now rapidly expanding in all
science and engineering domains (Wu et al. 2014). Research in the areas of com-
puter graphics, database management systems, and AI along with the development
of faster and more powerful hardware platforms accelerated and widened the use of
computers for engineering problem-solving. Knowledge-based expert systems
(KBESs) are one of the first realizations of research in the field of AI, in the form of
a software technology. KBESs are computer programs designed to act as an expert
to solve a problem in particular domain. The program uses the knowledge of the
domain coded in it and a specified control strategy to arrive at solutions. Such
systems consist of a knowledge base and an inference engine subdivided in one or
more inference mechanisms (Krishnamoorthy and Rajeev 1996). The research
findings of Quintana-Amate et al. (2015) provided by literature survey confirm the
existence of a gap on knowledge sourcing in engineering, and more precisely they
underlined the need for an extended knowledge-based engineering (KBE) devel-
opment process which integrates AI tools and expert intervention to systematically
manage the knowledge efficiently captured and modeled (employing AI algorithms
and expert involvement). Therefore, there is a need for further research on the
integration of KBE systems and AI implementations as a potential solution to
achieve a next level of engineering design.
As seen before, with the recent progress in capability for data storage, handling,
analysis as well as AI methods, such a database becomes much more realistic and
opens completely new opportunities to take advantage for an optimized design. Such
a database would be a complete novelty in the AD community as well as in the
community for engineering design. Such a database means an immense concentra-
tion of knowledge, which would mature over years. As a side effect, this database
would also play an important role for the practical application and teaching of AD, as
examples can be taken from it for learning purposes. A big difficulty is surely the
correct structuring of the database and the form in which the earlier designs are
documented and archived, in order to be able to extract and use the data afterward in
the best possible way. In the phase of building up such a database, first the ontology,
architecture, and data representation of the database need to be defined as this is a
fundamental basis that needs to be considered. As there is a great interest from the
AD community to support the creation of such a database, there would be a high
willingness of designers to fill the database with life. Another way of filling the
databases is to use a reverse engineering approach (see also Girgenti et al. 2016;
Vickery et al. 2018; Rauch and Vickery 2020). Such a reverse-mapping allows to
594 E. Rauch and D. T. Matt

decompose past designs easily into hierarchical FR–DP pairs. In a first step, such a
database could be filled only with freely available designs or designs from Open
Access papers. This means that people uploading their past designs give free access
to this information. At a later stage, a model might be developed how designs can
also be commercially exploited, e.g., by allowing the designer to access more than
just the freely accessible data by paying a fee.

21.5 Vision of Combining Artificial Intelligence

and Axiomatic Design

According to the sections before, we introduce the vision of an AI-assisted design

and re-design of complex systems in engineering design. While a lot of research is
currently focused on how to design products and systems according to the princi-
ples of the digital transformation in industry and Industry 4.0, the following
hypothesis aims to innovate and automate the design process itself and thus to
initiate a revolution in the field of engineering design.
Especially in the design of complex systems that are changing over time, e.g.,
manufacturing systems, the design process can be divided into a “new design” and
adaptations (“re-design”). Adapting an existing system to changing environment or
new requirements is a difficult mission. The world is becoming ever more
short-lived, and product life cycles as well as system life cycles are becoming ever
shorter. This also means that many products or systems have to be adapted at ever
shorter intervals. Therefore, in the future, the design of products or systems must be
as fast and agile as possible and be based on the principles of self-optimizing
systems by using intelligent and smart technologies like AI.
Figure 21.2 shows the development of AD over the years focusing on the most
important developments in AD theory, its field of applications, and tools to support
the designer.
From the early phase of AD to the beginning of the twenty-first century,
researchers focused more on the first axiom. In the following years, advances in
complexity theory concentrated more on the second axiom and in the detailed
analysis and procedure to define CNs in the customer domain and FRs in the
functional domain. Regarding the application of AD in practice, it has been used in
its beginnings mainly for the design of products and later for the design of complex
systems (e.g., manufacturing systems, software design, and organizational design).
In recent years, designers started also to use AD for the design of intelligent
products and systems according to Industry 4.0 and Smart Manufacturing (Vickery
et al. 2019). With the work of Kim et al. (2019), the first time researcher started also
to use AI technologies for making a next important step in the development of AD
in engineering design. As already mentioned in the introduction of this chapter
there are only limited number of tools supporting the designer in using AD during
the design process. From the 90s, the designer used Microsoft Excel spreadsheets
for the application of AD with all its limitations in space as well as automation.
21 Artificial Intelligence in Design: A Look into the Future … 595

Fig. 21.2 Past, present, and future of complex system design using Axiomatic Design

Roughly ten years later, Acclaro DFSS has been launched facilitating larger design
projects and offering functions for a semi-automated FR–DP decomposition.
However, over the last nearly 20 years there were no more updates or launches of
new aid tools for conceptual design based on AD, which is now (in an increasingly
digital world) seriously limiting and affecting the use and dissemination of AD in
design projects.
Therefore, the hypothesis is to realize in the future the vision of a Thinking
Design Machine, as mentioned by Suh and Sekimoto (1990) by using recent AI
methods or AI technologies to be developed in the near future creating the basis for
a new platform of a computer-aided tool for conceptual design. In the following two
sections, this hypothesis and vision of an AI-assisted design and re-design of
complex systems will be described more in detail, thus giving an outlook on
possible future developments in AD for engineering design.

21.6 Artificial Intelligence-Assisted Design of Complex

Systems

Figure 21.3 shows the new proposed AI-assisted design approach (see Fig. 21.3). It
starts with an automated clustering of customer feedbacks into meaningful CNs
with the help of data mining and knowledge discovery methods. In previous
approaches, this was usually done manually by collecting and looking over the
customer feedback. The big challenge for users of AD in this phase is that large
quantities of information in the customer domain can only be reduced with great
effort to meaningful CNs in a manual way, whereby also a guarantee to consider all
596 E. Rauch and D. T. Matt

Fig. 21.3 Concept of AI-assisted design of complex systems

information as completely as possible is usually not given. Therefore, for the future,
the development of assisted and intelligent methods is needed to simplify this and
improve the quality of the data. Kim et al. (2019) are taking a first step in this
direction showing in their work a first attempt/experiment based on available data
of Airbnb (customer feedback) using Pytorch, an open-source deep learning plat-
form, and MATLAB. By using a hierarchical clustering algorithm, customer
feedback can be translated and clustered into key features, and therefore relevant
customer needs. According to them a challenge will be to deal with different types
and abstraction levels of customer feedback, which requires in future to find
appropriate state-of-the-art algorithms for the automated identification, extraction,
and clustering of CNs. Further research is still necessary, in particular, due to the
need to transform unstructured and, sometimes, ill-defined user specifications into
meaningful key CNs, which requires the research and application of state-of-the-art
natural language processing techniques (Kulkarni et al. 2019; Kang et al. 2019a, b).
In the second step, these CNs will be encoded to FRs and to constraints. This
describes the link from the customer domain to the functional domain, which
should occur as automated as possible and without expensive involvement of the
designer (except for logical checks). As also described in the work of Kim et al.
(2019), one challenge lies in the automatic encoding and transition from natural
language expression of customer needs to FRs that can be then further used in the
AD design approach. First tested AI abstraction tools for natural language by Kim
et al. (2019) have resulted to be not effective in extracting FRs. Word embedding
tools are currently not directly capable of translating key CNs into functions of a
21 Artificial Intelligence in Design: A Look into the Future … 597

system. It is also necessary to research, test, and validate possible AI solutions for
this step in the AD design approach, or to determine whether fully automated
solutions can be implemented or to what extent humans should intervene in a
semi-automated solution. Using the proposed AD knowledge database, also a
knowledge-based encoding of FRs could also be done, where based on earlier
transitions from the customer domain to the functional domain, the designer is
provided with suggestions for possible FRs for recurring or similar CNs.
In the third step, a semi-automated and AI-assisted FR–DP mapping and de-
composition process takes place. During the mapping and decomposition, the
proposed approach includes automated proof checking (Mantravadi et al. 2019) of
the Independence Axiom (Axiom 1) to avoid a so-called coupled, and therefore
complex design suggesting possibilities for a re-design or rearrangement of FR–DP
pairs. The FR–DP mapping (finding a design solution for a certain FR) is
semi-automated in the assisted design approach as the system will propose the
designer automatically several alternative DPs for a certain FR. The selection of the
best DP will be assisted by an automated check of the Information Axiom (Axiom
2), evaluating the complexity of a design solution. Similarly to this, such an
AI-based approach is used also to propose possible decompositions of a higher
level FR–DP pair into lower level ones based on similar decompositions found in
the AD knowledge database. Prerequisite and crucial for this is a suitable ontology
for the formal representation of such knowledge (Akmal and Batres 2013) and a
central database in the background, which is fed with labeled successful and
unsuccessful designs in order to be able to learn from past successes and failures
through supervised AI techniques.

21.7 Artificial Intelligence-Assisted Re-design of Complex

Systems

In Fig. 21.4, we can see the behavior of a system, e.g., a manufacturing system,
over the time. Future events are very often unpredictable and might shift the system
range away from the defined design range, and thus create time-dependent com-
plexity. According to AD, the information content of a system with defined FRs is
described by the joint probability that all FRs are fulfilled by the respective set of
DPs and measured by the ratio of the common range between the design and the
system range. As shown in Fig. 21.4, a system might deteriorate during its service
life and its design range will move outside the required system range. According to
AD, the first type of time-dependent complexity is periodic complexity, which can
be managed through the analysis of previous typical time periods. Simple examples
of periodic complexity are tools that wear out (Matt and Rauch 2011). If we are able
to define the right periodic intervals for their change and re-design, this information
helps to reduce complexity. The second type of time-dependent complexity,
598 E. Rauch and D. T. Matt

Fig. 21.4 Overlapping and shifting of design and system range

combinatorial complexity (Suh 2005), can be caused in case of a manufacturing

system by the dynamics of unpredictable technological, socio-economic, or political
influences. According to Suh (2005), this unpredictable combinatorial complexity
can be managed by, transforming it into a periodic complexity. When the system
does not renew itself by resetting and reinitializing, it becomes a reason for wasting
resources. Manufacturing systems are especially characterized by such a combi-
natorial complexity and in order to be responsive to unforeseen changes. They must
be reset in periodic intervals to avoid or minimize the effect of shifting outside of
the design range. The time-dependent combinatorial complexity must be changed
into a time-dependent periodic complexity by introducing functional periodicity.
This allows trigger of the re-design of the system and to re-adjust to changing
conditions. If this could be achieved, the system will be more agile and resilient
than traditional ones.
In the proposed AD, approach for AI-assisted re-design the analysis of the
system over time might show degrees of periodicity, which can be used to actively
trigger the re-design of a company’s manufacturing system or parts of it, before a
too strong deviation of system range and design range force it to (see also
Fig. 21.5). The analysis would imply searching for patterns in collected temporal

Fig. 21.5 Triggered re-design of manufacturing systems

21 Artificial Intelligence in Design: A Look into the Future … 599

series, like, for example, sensor measurements, evolution of performance measures,

maintenance interventions (Pertselakis et al. 2019), and logging data for functional
fault diagnosis (Ye et al. 2014).
Modern technologies of Industry 4.0, such as sensors and human–machine in-
terfaces, enable the collection of large amount of data and thus the current status of
systems like a manufacturing system to be called up in real time. A re-design should
not be triggered only after an emergency alarm based on the self-monitoring, as this
would mean a reactive re-design. Re-design must be based on the digest of the daily
production measured over time (a trend over a longer period), not only in terms of
the real-time data generated during the daily manufacturing process. Based on this
fact also predictive suggestions need to be generated “it is worth to re-design before
the end of the week/month/year” and only trigger alarms of the form “urgent
re-design required” when the suggestions had been ignored and the “trigger point”
had been reached without any measurable trend improvement. For self-monitoring,
the tolerance bands of FRs and DPs are determined by metrics during the initial
configuration or design of the system. Depending on whether the adaptation of the
system can take place automatically (e.g., an adaptation of parameters of a man-
ufacturing process) or whether human support is required, we speak of
self-adaptation or assisted adaptation of the system. Due to technological or
organizational changes, it may be that for certain FR–DP pairs, an earlier rejected
DP alternative would achieve now a better result. To achieve this the “AI” must be
able to doubt about its own beliefs. It is required that an “entity” able to simulate the
use of “alternative” DPs to re-weight their influence on the overall performance.
This topic is related to Reinforcement Learning (Lison 2012), a ML technique to
find the best possible behavior or path in a specific situation. Similarly, new DPs
may join the AD knowledge database (e.g., a new technology), which may replace
an old DP due to better performance. In these cases, not only the previous status
should be restored by self-adaptation, but the aim is to achieve a self-optimization
of the manufacturing system. During this process, some DPs may save time and
money by automatic adaptation, while other DPs may be of strategic relevance
requiring a human-driven decision-making process.

21.8 Impact and Advantages of Artificial Intelligence

in Axiomatic Design

In the current application of AD for the design of complex systems, there are and
there have been certain limitations. Many experienced designers agree that AD
helps them to cultivate insightful thinking. But many still find it difficult to apply
AD principles to design practice since using AD effectively also requires designer’s
insights and experience in AD (Kim et al. 2019). Based on own experiences in
research and industrial practice the following are currently detectable weaknesses of
AD:
600 E. Rauch and D. T. Matt

• need of experience in applying AD;

• difficulty in the holistic consideration of customer needs;
• designer very often struggle to define solution-neutral FRs;
• the identification of design solutions/parameters is based on their knowledge and
experience;
• existing designs from other sources cannot be taken into account because there is
no central data source;
• the process is very manual, and there is a very limited computer-aided support
with current systems available;
• current approaches do not allow a real-time monitoring of whether the design
parameters (DPs) still meet the FRs over time;
• not possible to automatically determine the time for a re-design of a system
(trigger point);
• does not allow an AD-based self-optimization of the system.

However, the time may be now just right to take AD to a completely new and
revolutionary level by taking advantage of modern technologies from Industry 4.0
(such as sensor technology, real-time data gathering, AI, deep learning, machine
learning, or cloud computing) to eliminate the before-mentioned limitations.
Planning can be carried out much faster, with less effort, with more accuracy, and
an enormous planning quality through the presented AI-assisted and automated
design concept. Table 21.1 shows the most important innovations due to the use of
the proposed new AD design approach.

Table 21.1 Assisted (re)design of complex systems—traditional versus new proposed

AI-assisted approach
Traditional AD design approach New proposed AI-based AD design approach
Manual and subjective analysis of key Automated clustering of customer feedback
customer needs through the designer or into meaningful CNs
design team
Manual and subjective translation of the CNs Semi-automated translation of CNs into FRs
into FRs through the designer or design team
Designer very often struggle to define System supports the correct syntax and
solution-neutral FRs formulation of solution-neutral FRs
Experience-based approach depending on the Knowledge-based approach depending on the
experience of the designer or design team quality and quantity of past designs in an AD
knowledge database
No guarantee of a comprehensive Due to an increasing filling level of an AD
consideration of all potential design solutions knowledge database over time, an immense
(DPs) compendium of potential DPs can be created.
Very manual process with only limited Development of a highly automated
computer-aided support computer-aided conceptual design (CACD)
tool
(continued)
21 Artificial Intelligence in Design: A Look into the Future … 601

Table 21.1 (continued)

Traditional AD design approach New proposed AI-based AD design approach
Difficult to apply for novices or people not Many of the decisions and reviews (e.g.,
experienced with AD check of AD Axioms) are done automatically
in the background, making it easy to use even
for novice designers
No possibility for real-time monitoring of Integration of a self-monitoring with
whether the DPs still meet the FRs over time adjustable sensitivity limits and alarm if
defined setting limits are exceeded
No possibility to automatically determine the Autonomous determination of the trigger
time for a re-design of a system (trigger point for a re-design of the system (or its
point) parts) based on the self-monitoring
Does not allow a self-optimization of the Enables self-adaptation and self-optimization
system of a system

As mentioned above, such an AI-assisted design approach will have an

important impact in the field of the design of complex systems. However going in
this direction, there will be many related fields of action, where such an approach
may help. A possible additional goal for the future is sustainability. The proposed
approach could also provide information on the extent to which certain adjustments
could also have positive or negative effects on the environment or sustainability in
general in all their facets (economic, ecological, or social).
Problem
Students should concentrate on one domain (customer, functional, design domain)
and discuss the potential of AI in the respective domain.

References
Admin (2019) 10 use cases of AI in Industry 4.0 revolution. GPUonCloud Website. https://
gpuoncloud.com/10-use-cases-of-ai-in-industry-4-0-revolution/
Akmal S, Batres R (2013) A methodology for developing manufacturing process ontologies. J Jpn
Ind Manag Assoc 64:303–316. https://doi.org/10.11221/jima.64.303
Bini SA (2018) Artificial intelligence, machine learning, deep learning, and cognitive computing:
what do these terms mean and how will they impact health care? J Arthroplasty 33(8):2358–
2361. https://doi.org/10.1016/j.arth.2018.02.067
Cheng B, Zhang J, Hancke GP, Karnouskos S, Colombo AW (2018) Industrial cyberphysical
systems: realizing cloud-based big data infrastructures. IEEE Ind Electron Mag 12(1):25–35.
https://ieeexplore.ieee.org/document/8322328
Daley S (2018) 19 examples of artificial intelligence shaking up business as usual. Builtin Website.
Accessed 29 Oct 2019
Farid A, Suh NP (eds) (2016) Axiomatic design in large systems. Springer, Cham, Switzerland
Floss P, Talavage J (1990) A knowledge-based design assistant for intelligent manufacturing
systems. J Manuf Syst 9(2):87–102. https://doi.org/10.1016/0278-6125(90)90024-C. http://
www.sciencedirect.com/science/article/pii/027861259090024C
602 E. Rauch and D. T. Matt

Garbade MJ (2018) Clearing the confusion: AI vs machine learning vs deep learning differences.
Medium Magazine Website. https://towardsdatascience.com/clearing-the-confusion-ai-vs-
machine-learning-vs-deep-learning-differences-fce69b21d5eb
Girgenti A, Pacifici B, Ciappi A, Giorgetti A (2016) An axiomatic design approach for customer
satisfaction through a lean start-up framework. In: Liu A (ed) 10th international conference on
axiomatic design (ICAD), Procedia CIRP, vol 53. Elsevier ScienceDirect, Xi’an, Shaanxi,
China, pp 151–157. https://doi.org/10.1016/j.procir.2016.06.101
Goasduff L (2019) Top trends on the gartner hype cycle for artificial intelligence, 2019. Gartner
Website. https://www.gartner.com/smarterwithgartner/top-trends-on-the-gartner-hype-cycle-
for-artificial-intelligence-2019/
Kang S, Patil L, Rangarajan A, Moitra A, Jia T, Robinson D, Dutta D (2019a) Automated
feedback generation for formal manufacturing rule extraction. Artif Intell Eng Des Anal Manuf
33(3):289–301. https://doi.org/10.1017/S0890060419000027
Kang S, Patil L, Rangarajan A, Moitra A, Robinson D, Jia T, Dutta D (2019b) Ontology-based
ambiguity resolution of manufacturing text for formal rule extraction. J Comput Inf Sci Eng 19
(2):9. https://doi.org/10.1115/1.4042104
Khan A, Day AJ (2002) A knowledge based design methodology for manufacturing assembly
lines. Comput Ind Eng 41(4):441–467. https://dl.acm.org/doi/10.1016/S0360-8352\%2801\%
2900067-5#sec-ref
Kim SG, Yoon SM, Yang MC, Choi J, Akay H, Burnell E (2019) AI for design: virtual design
assistant. CIRP Ann 68:141–144. https://doi.org/10.1016/j.cirp.2019.03.024
Krishnamoorthy CS, Rajeev S (1996) Artificial intelligence and expert systems for engineers, 1st
edn. CRC Press, Boca Raton
Kulkarni S, Verma P, Mukundan R (2019) Assessing manufacturing strategy definitions utilising
text-mining. Int J Prod Res 57(14):4519–4546. https://doi.org/10.1080/00207543.2018.
1512764
Kumar SPL (2017) State of the art-intense review on artificial intelligence systems application in
process planning and manufacturing. Eng Appl Artif Intell 65:294–329. https://doi.org/10.
1016/j.engappai.2017.08.005. http://www.sciencedirect.com/science/article/pii/
S0952197617301896
Kushmaro P (2018) 5 ways industrial AI is revolutionizing manufacturing. CIO Magazine
Website. https://www.cio.com/article/3309058/5-ways-industrial-ai-is-revolutionizing-
manufacturing.html
Lison P (2012) An introduction to machine learning. University of Oslo Website. https://heim.ifi.
uio.no/plison/pdfs/talks/machinelearning.pdf, hiOA
Mantravadi S, Li C, Møller C (2019) Multi-agent manufacturing execution system (MES):
concept, architecture and ML algorithm for a smart factory case. In: Proceedings of the 21st
international conference on enterprise information systems (ICEIS), vol 1, pp 477–482. https://
doi.org/10.5220/0007768904770482
Masfaraud S, Dumouchel PE (2019) DessIA—an engineering design software with artificial
intelligence. Telecom Paris Website. https://www.telecom-paris.fr/wp-content-EvDsK19/
uploads/2019/07/2-2b-JPE-TP-2019-KeynoteE_Dessia_Commercial_V6_short.pdf, keynote
Paper Presentation
Matt DT, Rauch E (2011) Continuous improvement of manufacturing systems with the concept of
functional periodicity. Key engineering materials, vol 473. Trans Tech Publications,
Switzerland, pp 783–790
Nicholson C (2019) Artificial intelligence (AI) vs. machine learning vs. deep learning. A.I. Wiki
on Pathmind Website. https://skymind.ai/wiki/ai-vs-machine-learning-vs-deep-learning.
Accessed 28 Oct 2019
Pertselakis M, Lampathaki F, Petrali P (2019) Predictive maintenance in a digital factory
shop-floor: data mining on historical and operational data coming from manufacturers’
information systems. In: Proper HA, Stirna J (eds) Advanced information systems engineering
21 Artificial Intelligence in Design: A Look into the Future … 603

workshops, CAiSE 2019. Lecture notes in business information processing, Rome, Italy, vol
349. Springer, Cham, pp 120–131, 3–7 June 2019
Quintana-Amate S, Bermell-Garcia P, Tiwari A (2015) Transforming expertise into
Knowledge-Based Engineering tools: a survey of knowledge sourcing inthe context of
engineering design. Knowl-Based Syst 84:89–97
Rauch E, Vickery AR (2020) Systematic analysis of needs and requirements for the design of
smart manufacturing systems in SMEs. J Comput Des Eng. https://doi.org/10.1093/jcde/
qwaa012
Sadiku MNO, Musa SM, Musa OM (2019) Artificial intelligence in the manufacturing industry.
Int J Adv Sci Res Eng (IJASRE) 5(6):108–110
Shu C (2019) Industrial robotics giant Fanuc is using AI to make automation even more
automated. Tech Crunch Magazine Website. https://techcrunch.com/2019/04/18/industrial-
robotics-giant-fanuc-is-using-ai-to-make-automation-even-more-automated/?guccounter=1
Skymind (2019) Artificial Intelligence (AI) vs. Machine Learning vs. Deep Learning. https://
skymind.ai/wiki/ai-vs-machine-learning-vs-deep-learning. Accessed 28 Oct 2019
Suh NP (1990) The principles of design. Oxford University Press
Suh NP (2001) Axiomatic design—advances and applications. Oxford University Press
Suh NP (2005) Complexity. Oxford University Press
Suh NP, Sekimoto S (1990) Design of thinking design machine. CIRP Ann 39(1):145–148. https://
doi.org/10.1016/S0007-8506(07)61022-1
Suh NP, Cavique M, Foley JT (eds) (2021) Design engineering and science. Springer Nature
Vickery A, Rauch E, Brown CA (2018) Deriving functional requirements for industry 4.0 from
industry’s assessment of needs. In: Peruzzini M, Pellicciari M, Bil C, Stjepandić J, Wognum N
(eds) Transdisciplinary engineering methods for social innovation of Industry 4.0: proceedings
of the 25th ISPE Inc. international conference on transdisciplinary engineering. Advances in
transdisciplinary engineering, vol 7. IOS Press, pp 23–32. https://doi.org/10.3233/978-1-
61499-898-3-23
Vickery AR, Rauch E, Rojas RA, Brown CA (2019) Smart data analytics in SME manufacturing
an axiomatic design based conceptual framework. In: Liu A, Puik E, Foley JT (eds) 12th
international conference on axiomatic design (ICAD), MATEC web of conferences, Sydney,
Australia
Wu X, Zhu X, Wu GQ, Ding W (2014) Data mining with big data. IEEE Trans Knowl Data Eng
26(1):97–107
Ye F, Zhang Z, Chakrabarty K, Gu X (2014) Knowledge discovery and knowledge transfer in
board-level functional fault diagnosis. In: Purtell M, Mitra S (eds) 2014 international test
conference (ITC), Washington, DC, USA, pp 1–10
Axiomatic Cloud Computing
Architectural Design 22
John Thomas and Pam Mantri

Abstract
The design of modern cloud computing makes available a plethora of scalable
cloud computing offerings. The cloud is increasingly becoming the backbone of
the highly complex modern knowledge economy that includes social, mobile,
IoT, Big Data, and AI. Knowledge-based products and services follow fat-tail
distributions such as the power law that poses major opportunities and
challenges for the designer. The Axiomatic Designer is uniquely positioned in
designing for the de novo situations that the fat-tailed distributions expose. Also,
the cloud frees up the architectural decision-making away from the legacy
compatibility-burden, and toward various cloud-native (i.e., de novo/solution-
neutral) as well as hybrid (on-prem/cloud and cloud/cloud) architectures.
Furthermore, the competitive landscape around the cloud is not static; it is
adaptive and evolving rapidly. Here again, Axiomatic Design (AD) is uniquely
positioned in rising up to the various de novo challenges. This, however,
requires contributions from frameworks such as knowledge as heterarchically
hierarchical (KA|h|H), stigmergy, complex adaptive systems (CAS), Cynefin,
Boyd’s OODA loop theory of asymmetric fast transients, axiomatic maturity
diagram (AMD), as well as Weick’s loose-coupling approach to help unify and
strengthen the axiomatic approach. This chapter unifies the above approaches in
order to tackle the architectural challenges of cloud computing.

This chapter is an adaptation of the paper “Axiomatic Cloud Computing Architectural Design”
presented by the same authors at ICAD 2019 in Sydney Australia (Thomas and Mantri 2019). This
paper was published open access at https://doi.org/10.1051/matecconf/ 201930100024 under a
CC BY 4.0 license.
J. Thomas (&)
P. Mantri
New Rochelle, NY, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 605

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_22
606 J. Thomas and P. Mantri

22.1 Introduction

The cloud is fundamentally disruptive. If its strategic import is properly understood,

it has immense potential to help scale businesses in space (geographical space,
product/service space, governance/regulation space, etc.) and time (start-up to IPO,
industry business cycle, market seasonality, etc.).
With hindsight, it is understandable that the web would bankrupt a successful
legacy business model such as the Borders bookstore that in the mid-90’s mis-
takenly invested heavily on brick-and-mortar stores across the globe, failed to
develop an e-reader, and outsourced its online sales (in 2001) over to Amazon. By
the time (i.e., in 2008) Borders realized its online outsourcing error and retracted, it
was too late. It filed for bankruptcy on February 16, 2011 (Wikipedia 2019a).
In hindsight, it is easy to see the strategic mistakes that the management at
Borders bookstore had made. But from a mathematical perspective, what killed
Borders was an inadequate appreciation of the power law (Wikipedia 2019b) that is
operative in the modern knowledge-based, network economies. Human knowledge
is a network of concepts which has been aggregating across millenniums. It has a
certain shape, structure, and overall dynamic. But most significantly, it has a certain
directionality in its growth patterns, as dictated by the power law which results from
the phenomenon of preferential attachment (colloquially known as the Matthew
effect or the Rich-Get-Richer effect). In other words, what is popular becomes even
more popular by virtue of the fact that it is better known. Thus, a brick-and-mortar
bookstore that tries to give equitable shelf space to the mega-successful Harry
Potter books as well as the also-rans, simply cannot compete with an online store
that only incurs remote-location storage cost that is comparatively cheap. Note that
shipping costs are borne by the buyer who pays for it for the convenience of
shopping from his/her home. Along with lower inventory costs, an astute vendor
such as Amazon also gets to harvest critical insights about the knowledge growth
patterns of the buyer (Siegel 2013). When this is aggregated across the population
dimension that spans cities, states, and nations—it provides deep strategic insights
about the meristematic growth patterns in the overall economy, and which may be
gainfully tapped into. Note that it is not just the purchase of books that provide
insights about the struggles and aspirations of a nation that is logging into browse
and purchase. Each item that is being browsed and purchased has aspects of
knowledge that went into its design and manufacture. The five-star rating that
captures the wisdom of the crowd (Surowiecki 2005) is the engine that propels the
preferential attachment and the resultant power law. These strategic insights help
the company outsmart its competition.
The power-law distribution of knowledge-based products is fundamentally dif-
ferent from the familiar Gaussian-normal distribution. Power-law-based products
exhibit fat tails that are often underestimated when mistakenly framed using the
Gaussian distribution. Fat tails force the designer into the uncharted de novo space.
And it is here that the AD approach has a unique role to play. Here, we trace the
22 Axiomatic Cloud Computing Architectural Design 607

potential of AD in the context of de novo situations that fat-tailed distributions

expose in the cloud.
Cloud computing provides various benefits, including agility, scalability, cost
reduction, mobility, disaster recovery, etc. Also, the cloud ecosystem is fairly
complex. Designing an adaptive architecture that can withstand the onslaught of
change is fairly challenging. This chapter holistically unifies a smorgasbord of
architectural concepts and frameworks that help unify and strengthen the AD
approach for tackling cloud computing design.
Without the benefit of a holistic design framework, cloud architectures remain
fragile. This is especially true in the context of cloud security. Unaddressed gaps in
design become salient points of architectural weakness. Cloud offerings are
increasingly vulnerable in this context, given the joint ownership model that the
cloud operates within. The AD framework is rare in upholding the holism of design.

22.2 History of Cloud Computing

Cloud computing was born in the shadow of the Cuban missile crisis (1962) that
almost precipitated a nuclear holocaust (Vleck 2019). During the crisis, the Pen-
tagon discovered that its existing information infrastructure was practically unus-
able in a coordinated, large-scale, human–machine endeavor. Simultaneously, early
prototypes of time-sharing systems were being developed in the early 1960s at MIT
and BBN (Bolt, Beranek, and Newman) under the leadership of Prof. John
McCarthy and Joseph C. R. Licklider (Lick). The missile crisis gave the necessary
impetus to launch the time-sharing Project MAC (Multiple Access Computing) at
MIT that spanned 1963–74. By 1969, Lick had expanded and democratized the
time-sharing Project MAC vision to what he colorfully called the “Intergalactic
Computer Network (Wikipedia 2019c),” which seeded the ARPANET—the historic
precursor to the modern Internet and gateway to the cloud.
To help stabilize the infrastructure that was constantly being upgraded under
Moore’s law, the early 1960s also saw market-initiated (primarily IBM, GE, and
Bell Labs) development of server virtualization (Wikipedia 2019d).
Cloud computing was born when these two seminal concepts (i.e., time-sharing
and server virtualization) fused in the 1990s. The term itself was coined by Prof.
Chellappa in a 1997 talk (Chellapp 1997):
Cloud computing can be defined as a set of frameworks that provides on demand, scalable,
customized, quality services in Software, platform and also provides sharable infras-
tructure through internet that are accessible and available everywhere. (Emphasis
added.)

In 2011, NIST standardized the official definition of cloud computing (Mell and
Grance 2011):
Cloud computing is a model for enabling ubiquitous, convenient, on-demand network
access to a shared pool of configurable computing resources (e.g., networks, servers,
608 J. Thomas and P. Mantri

storage, applications, and services) that can be rapidly provisioned and released with
minimal management effort or service provider interaction. …Cloud systems auto-
matically control and optimize resource use by leveraging a metering capability at some
level of abstraction appropriate to the type of service (e.g., storage, processing, bandwidth,
and active user accounts). (Emphasis added.)

Based on the NIST standard, cloud computing may be formally captured in the
lower triangle, FR–DP configuration as shown in Fig. 22.1. The design leads off
with two seminal concepts that historically define cloud architecture: virtualization
and time-sharing. The virtualization design parameter (DP11) answers the
Economies-of-Scope functional requirement (FR11) for abstracting away from the
underlying hardware resources (compute, storage, networking) in order to set up
highly customizable resources. Virtualization allows the user to break away from
tight-coupling with any given hardware vendor (i.e., lock-in). It also allows the
architect to mix-and-match technological offerings from various vendors that would
have been impossible otherwise. Time-sharing (DP12) allows load-sharing-driven
Economies-of-Scale of software/hardware resources for substantial cost reductions
(FR12) across multiple users. The service is accessible (FR2) anywhere, anytime, for
anyone, for any business, and from any of the client devices that can access the
Internet (DP2). Given the virtualization functionality across time-shared resources, it
is now possible to programmatically provide rapid elasticity (DP3) on an as-needed
basis. Such rapid built-up and tear-down of an arbitrary set of resources from the
resource pool would have been unthinkable in previous on-prem architectures. From
an FR perspective, what this accomplishes is the agility to rapidly reconfigure and
repurpose the strategic business thrust, and move on a dime (FR3). It is this agility
that aligns closely with the Boydian strategy framework of asymmetric fast tran-
sients (see Sect. 22.9). FR4 demands that all of the above FR/DPs need to be
managed and achieved on an automated (on the vendor side) and self-service (on the
customer side) basis with minimal managerial overhead. Again, given the
virtualization/time-sharing flexibilities, this requirement is also programmatically
feasible (DP4). FR5 requires usage and billing transparency as provided by cloud
telemetry (DP5) in order to honor the service contract as well as for providing usage
feedback that the customer may use for adapting to changes in the demand curve.
With the above brief historical review of cloud computing along with the
framing of the basic architecture using the AD, we may now review the rest of the
13 architectural concepts (Siegel 2013; Surowiecki 2005; Mell and Grance 2011;
Vleck 2019; Wikipedia 2019c, d; Chellapp 1997; Grassé 1959; Parunak 2006;
Fehling et al. 2014; Erl et al. 2017; Akhtar 2018).

22.3 Socio-Technical Stigmergy

Stigmergy (Grassé 1959) denotes call to work based on local signs or markings left
by collaborating agents (a) at some time in the past and during the course of their
work (either as a side effect of the said work or as something in addition to the
22 Axiomatic Cloud Computing Architectural Design 609

Fig. 22.1 Cloud computing functional requirement–design parameter (FR–DP) map (using NIST
definition (Mell and Grance 2011)). (Reproduced from Thomas and Mantri (2019), originally
published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201930100024)

work). These markings aggregate to provide organizational directives (b-logic)

available at various levels, both within the environment and within and between
agents. Thus, even though there is no one controlling the set of agents in a
top-down sense, there is nevertheless system-wide control being established in a
bottom-up sense. Case in point is the ant trail that emerges (see Fig. 22.2) from
pheromone droppings by ant agents. The trail then helps organize the ant swarm in
its various activities. Across multiple iterations, the stigmergic trail smoothens and
tightens, thus indicating minimization of the embedded information content.
The concept of stigmergy was discovered while searching for governing orga-
nizational motifs among eusocial insects such as ants and termites. Research
indicates that these same organizational motifs may be observed in various human
activities. Parunak reports on a variety of such human-level socio-technical stig-
mergic processes, including forest trail formation, highway traffic flows, democratic
elections, document editing, social media groupings, viral marketing, Google page
ranks, peer-to-peer computing, Amazon-style recommender-systems, etc. (Parunak
2006) Stigmergic problem-solving occurs wherever the problem context is beyond
the ken of any one agent. In other words, one should expect stigmergic solutions to
dominate in regimes that would be considered as complex. Architecting solutions
within the cloud computing/Big Data space clearly fall within this context. Indeed,
610 J. Thomas and P. Mantri

Fig. 22.2 Stigmergic trails by marching ants (created using NetLogo Wilensky (1997)). Note the
tightening of the stigmergic trail. (Reproduced from Thomas and Mantri (2019), originally
published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201930100024)

the very essence of Big Data is in capturing and tracking stigmergic patterns of
economically interesting behaviors across large population sets. This could be in the
realm of finance, healthcare, threat modeling and cybersecurity, gaming, education,
entertainment, etc. In architecting such systems, it is therefore important to track the
stigmergic patterns that are forming across the vast cloud/Big Data ecosystem
(Fehling et al. 2014; Erl et al. 2017; Akhtar 2018; Erl et al. 2016). Prior to the
cloud, socio-technical stigmergic processes and patterns were either trapped in the
confines of the enterprise or merely left as shallow traces across the lightweight
Internet weblinks. The modern cloud has the potential to allow data flow patterns to
aggregate without necessarily compromising privacy.
Stigmergic signals from the cloud include DP5: Telemetry-based transparency
for usage-based billing (see Fig. 22.1). Organizations are perhaps unaware of the
stigmergic significance of such accumulating data. Contractual agreements (Tollen
2016) with vendors need to be carefully negotiated, especially in the context of
shape-shifting complex adaptive systems (see Sect. 22.4).
Example 22.1
Discuss stigmergic tightening from an AD framework.

Solution: Nature often starts with a coupled design. For example, to obtain the shortest path
from point A to B, an ant trial starts with too many random detours away from a smooth
ideal path. Each such detour that deviates from the ideal path may be accounted as a
needless DP that over-complicates and adds needless information complexity to the travel
path. Nevertheless, across multiple iterations, as the ants march back and forth across the
jagged terrain, a number of these spurious, control-point DPs are eliminated, with the final
result being a minimally complex information trail. Here, the design matrix begins with too
many DPs, but over time, the design settles into a lightly decoupled (with necessary
coupling between adjacent control points), lower triangular design.
22 Axiomatic Cloud Computing Architectural Design 611

22.4 Complex Adaptive System (CAS)

To fully understand the strategic import that design plays in the cloud, one has to
appreciate the underlying adaptive dynamics propelling the cloud. As Urquhart
suggests in Urquhart (2012):
Cloud as an adaptive system: The thing is, however, a certain class of complex systems,
complex adaptive systems, have the additional trait that they can change their behavior in
response to the success or failure of previous behaviors when a given event occurs—or
when a certain series of events occurs. This ability to “learn” and adapt to the surrounding
system environment creates amazing outcomes, including many of the most rich, enduring,
and powerful systems in our universe.

Shape-shifting adaptive dynamics makes the cloud ecosystem a complex

adaptive system (CAS). As Prof. Holland describes it (Holland 2006), CASs “are
systems that have a large number of components, often called agents that interact
and adapt or learn.” Holland then proposed a two-tiered system as shown in
Fig. 22.3a.
The lower a-tier follows a fast dynamic and is engaged in the flow of resources
between diverse agents, while the upper b-tier follows a slow dynamic that captures
knowledge artifacts and aggregates from these which are then emitted system-wide
as stigmergic signals that help the agents organize and scale.
Figure 22.3b is an iterative variation on the basic CAS. At each follow-on
feedback-loop/iteration, the CAS trace bifurcates the target population into higher
levels of organizational complexity. As evidence of the cloud bifurcation process,
the Azure Resource Manager (Karthikeyan 2017) enables the organization and
management of the cloud resource sprawl that used to exist in the classic model.
Example 22.2
Map the bipartite axiomatic FR/DP framework alongside the CAS a/b framework using
Francis Bacon’s dictum: “Nature to be Commanded, must be Obeyed [Novum Organum].”

Fig. 22.3 Complex adaptive system. a Basic; b Iterative. (Reproduced from Thomas and Mantri
(2019), originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/
matecconf/201930100024)
612 J. Thomas and P. Mantri

Solution: All designs involve working closely with base-level materials available in nature
along with the laws that govern their behavior. It is in this sense that Bacon’s dictum that
nature to be commanded must be obeyed is relevant in design. To command nature (i.e., to
design and compose with elements found in nature), it is wise to obey the laws that govern
the basal elements as well as the adaptive, self-organizing structures and behaviors that they
exhibit. Many complex natural as well as man-made systems (including biological, euso-
cial, large-scale physical, as well as socio-technical systems such as the cloud) follow the
iterative CAS processes that exhibit hierarchical as well as heterarchical patterns (see
Sect. 22.5). In this context, the built-up of the bipartite b/a hierarchy (as shown in
Fig. 22.3b) maps well with the bipartite FR/DP hierarchy. The fundamental difference
between nature’s CAS-driven hierarchy versus human-driven design hierarchy is that the
former is often a long drawn-out, bottom-up zigzag process, while the latter is a relatively
rapid, top-down zigzag process. But the elements of FRs and DPs may be discerned from
carefully studying the elements available in the b and the a layers, respectively. In this
sense, at all levels, design involves three distinct and strategic selections: the right FR
elements from the b set, the right DP elements from the a set, and the proper mapping
between the two.

22.5 Knowledge Hierarchy/Heterarchy

The following is a short review of the knowledge hierarchy framework (Thomas

and Zaytseva 2016) which could aid in the mapping of various frameworks such as
AD, Cynefin, OODA, agile, etc., into an integrated whole. Human knowledge is the
engine that drives the knowledge economy. Given the abstract nature of human
knowledge, it may be observed that domain knowledge is conical in shape; i.e.,
there are many more concretes than abstractions. The knowledge corpus captures
the sum total of truths/facts gleaned from nature and painstakingly accumulated
across time. Induction (depicted in Fig. 22.4a as the upward flowing arch) involves
creating higher level generalizations. Note the similarity between induction and the
upward flowing, regularity creating CAS emergence (Fig. 22.3a). In contrast, a
deduction is the downward flowing arch involved in the application of the induced
generalizations. The abductive cascade combines the inductive as well as the
deductive flows into a long sequence of step-by-step problem-solving trace.
When multiple domains are mapped side by side along with their shared con-
ceptual linkages, the various hierarchies map onto a heterarchical span that share
and cross-pollinate across the domain barriers (Fig. 22.4b). Hierarchies are denoted
as |H, heterarchies as |h, knowledge hierarchy as K|H, and knowledge as heterar-
chically hierarchical as KA|h|H.
Reverse salients are gaps that appear between knowledge hierarchies when
mapped heterarchically, side by side. These create knowledge asymmetries between
individuals and groups of individuals (including corporations and nations). Such
units may hold opposing mental models based on the underlying knowledge
asymmetries. For example, the management of Borders and Amazon was basing
their corporate decisions on mental models that were at variance with each other.
By the time gap closure had occurred, Borders as a corporation was bankrupted.
22 Axiomatic Cloud Computing Architectural Design 613

Fig. 22.4 Knowledge hierarchy/heterarchy. (Reproduced from Thomas and Mantri (2019),
originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201930100024)

Systems that display strong heterarchical engagements have very different

growth patterns and architectural opportunities/vulnerabilities. For example, bac-
terial colonies exchange genetic material not just vertically (in a parent-to-child
sense), but also laterally (Wikipedia 2019e) between cells that come into contact
with each other. Vertical exchange is hierarchic and horizontal exchange is heter-
archic. Evolutionary adaptation that uses the vertical exchange is slow as it has to
work across organismal life spans. In contrast, an adaptation that uses lateral
exchange is rapid as it is able to quickly share successful mutations across large
populations. But when both mechanisms work in tandem, organisms can rapidly
navigate large search spaces in order to solve species-wide existential threats. For
example, this allows bacterial colonies to rapidly evolve and attain antibiotic
resistance (Wikipedia 2019e).
The advent of social media, IoT (Internet of Things), Big Data, AI/machine
learning, and mobile and cloud computing has accelerating such heterarchic link-
ages. Differences in heterarchic strengths and weaknesses could be complementary
or noncomplementary. When pursued for competitive advantage, they result in
asymmetric warfare (which is explored more fully in Sect. 22.9 under cloud
OODA).
614 J. Thomas and P. Mantri

Heterarchic problems exist in the context of cloud data and cloud security. For
example, the data modeling complexity involved in a Big Data/Data Lake context
has to do with ironing out the inevitable ontological inconsistencies across multiple
domains, each competing for abstraction dominance as well as in establishing a
wider inductive base spanning structured, semi-structured, and unstructured data
(Lee et al. 2018). Likewise, in the context of cloud cybersecurity, the current
dominance of the hierarchical defense-in-depth (Igbe 2017) approach is vulnerable
to heterarchic attacks. Defense-in-depth assumes that data/applications are secure in
the innermost layer of its hierarchical and concentrically structured onion rings. But
such an approach is vulnerable, especially in the increasingly IoTized technology
landscape where data and applications are placed at the edge with both hierarchical
and heterarchical vulnerabilities. Alongside defense-in-depth, what is therefore also
needed is an integrated defense-in-breadth focus (Igbe 2017).
Example 22.3
Provide an example where large-scale/real-time heterarchic linkages between the customer
experience base and the minute-by-minute operational business logs that are captured in the
cloud could lead to rapid closure of the reverse-salient gaps between the two.

Solution: Stigmergic processes operates whenever and wherever problem contexts over-
whelm the problem-solving capacity of the agents involved. The highly fractalized and
heterarchic sprawl of modern business problems that clients and employees of a major
cloud-based corporation faces often leads to reverse-salient gaps between the problem and
the available solution base. By amassing and analyzing minute-by-minute operational logs
being accumulated in the cloud, a business could quickly narrow down and pinpoint
opportunities and business trends, their overall market potential as well as strategically
generate valuable lead-customer (i.e., influencer)-driven design insights. In this sense, both
problem abstraction and its rapid resolution are being done in a democratic, “wisdom-of-
the-crowd” sense. Given the large-scale fractal sprawl of the cloud footprint, it would be
valuable to incorporate principles of design into what is currently and predominantly an
otherwise ad hoc process.

22.6 Power Law Versus Gaussian Distribution

Consider a variable which tracks a phenomenon that has multiple contributing

factors, each of which obeys its own unique probability distribution. If these factors
aggregate in an additive fashion, the resultant summing distribution that charac-
terizes the phenomenon would be a bell curve. As per the central limit theorem, it
would result in the Gaussian-normal distribution (Fig. 22.5a) if the contributing
distributions were independent and identically distributed. The issue with the
normal distribution shows up in the tail regions where the probabilities attenuate
drastically. This gives rise to the problem of unreasonably thin tails. Seldom do
natural phenomenon follow the Gaussian in the tail regions as it is attenuating too
sharply by following the exponential of the negative square of the variable in
question (Fig. 22.5a). Even so, many human-centric and natural phenomena
approximately follow the Gaussian distribution. Examples include temperature
22 Axiomatic Cloud Computing Architectural Design 615

distribution in a city at a given day of any year, height/weight distribution of a

population, delays in the arrival of public transportation, size and weight of fruits
and vegetables, experimental observational errors, etc.
In contrast, if these factors aggregate in a multiplicative fashion, the resultant
multiplicative distribution that characterizes the phenomenon would be a power-law
distribution which when plotted on a log–log plot shows up as a linear plot (see
Fig. 22.5b). Note that from an empirical perspective, there could be many varia-
tions of the power law (Wikipedia 2019b), including the broken power law which
consists of piecewise combinations of multiple power laws as well as smoothing of
the power law with the exponential.
To illustrate the generative multiplicative process behind the power-law phe-
nomenon, consider the following example of programmer productivity (Louridas
et al. 2008; Ward 2016). To begin with, suppose that just a few software pro-
grammers are proficient in the software-creation tools at their disposal. Even small
differences in better tool usage quickly aggregate to the advantage of the slightly
exceptional individual. Each successful project completion builds confidence in the
individual as well as in the eyes of the management that oversees the project. In
time, with repeat deliveries and accrual of choice experiences, the end result is that
these individuals are orders of magnitude more productive than the rest. Seeded by
minor differences in the initial conditions (for example, here the slight difference in
tool usage proficiency), the generative multiplicative process behind the power law
has the potential to bifurcate target populations (in a CAS sense) into distinct
groups and sub-groups. In time this could lead to deep-set social hierarchies
between those who lead versus those who are led. Indeed, wherever human intel-
lectual work is involved and made available to large populations in a free-market
economy, the rich-get-richer style power-law distributions are also likely to show
up. This is one of the reasons why the power law is increasingly relevant when
considering a knowledge economy.
Other examples of the power law include earthquake intensities, city popula-
tions, best sellers sold, the number of citations received, etc. In (Barabási and Albert
1999), Barabási and Albert highlight the generative process behind the power-law
distribution of the vertex degree in a network of webpage links:
Because of the preferential attachment, a vertex that acquires more connections than
another one will increase its connectivity at a higher rate; thus, an initial difference in the
connectivity between two vertices will increase further as the network grows.

From a symmetry perspective, the above argument could also be extended to

include preferential detachment for nodes that fall out of comparative favor. In time,
this creates significant bifurcations between the haves and the have-nots. Given the
foundational level wherein the generative multiplicative process behind the power
law is operating from, no egalitarian legislative action could effectively overcome
these biases without concomitantly also damaging the knowledge economy.
Instead, the proper solution to the above politico-economic problem (in a knowl-
edge economy) is to leverage the neglected heterarchy generation potential of the
very same hierarchy generating power-law-driven, iterative CAS. Over and above
616 J. Thomas and P. Mantri

Fig. 22.5 Normal vs log-normal vs power-law distribution. (Reproduced from Thomas and
Mantri (2019), originally published open access under a CC BY 4.0 license: https://doi.org/10.
1051/matecconf/201930100024)
22 Axiomatic Cloud Computing Architectural Design 617

the hierarchy generation potential, an iterative CAS module (if and when assisted
by favorable factors such as the cloud) also has the ability to generate heterarchic
hierarchies which (as explained below) is the fundamental solution to the above
egalitarian conundrum. While knowledge hierarchies fragment and bifurcate a
given domain (and therefore the social groups that cultivate it) into smaller and finer
grains, knowledge heterarchies create bridging artifacts that allow concepts and
propositions to disperse across the overall knowledge fabric. This is similar to the
lateral transfer of genes in a bacterial population.
Mindless specialization (into ever-deepening hierarchies) is the bane of modern
life. Fortunately, countervailing forces are at play that promises to link the hier-
archies via heterarchic bridge artifacts. For example, AD (Suh 1990; Suh 2001) is
fundamentally a hierarchy-bridging heterarchic artifact that considers all creative
designerly activities under a common breadth-seeking rubric. It is breadth-seeking
because the very same two principles of design (i.e., Axiom I and II) apply
regardless of the domain of interest. In other words, regardless of whether the
design pertains to engineering, software, education, organizations, medicine, etc.,
the same two design principles apply (Suh 1990).
By balancing the depth-seeking hierarchic drivers against the breadth-seeking
heterarchic drivers, iterative CAS has the potential to coordinate and disperse
knowledge agents across the totality of the knowledge economy, instead of
crowding around just a few clusters. Thus, instead of egalitarially spreading the
wealth (i.e., the product of human creativity) around, the heterarchy-generating
iterative CAS is capable of spreading the wealth-generating engine of human cre-
ativity around. In colloquial terms, it is about “teaching how to fish instead of giving
fish” (Wiktionary 2019).
Iterative CAS has the potential to flatten deep hierarchies in favor of a network
economy that bridges isolated hierarchies using a multitude of lateral linkages.
Furthermore, these heterarchic linkages allow the search space along all stages of
the design process to be vastly improved. We may, therefore, be more confident that
the search imperative embedded in Axiom II (i.e., minimize information content) is
indeed delivering a true minimum. This is similar to how bacterial populations are
able to rapidly navigate large search spaces in order to solve species-wide exis-
tential threats by utilizing both horizontal and vertical gene transfer mechanisms.
In a knowledge-driven political economy, the proper nurturing of the link
between human creativity embedded in the act of design and finding meaning and
fulfillment in life is of paramount importance. Note that in such an economy, the
neglected art and science of design ought to be center stage. Also, in such an
economy, the flattening of deep-set knowledge hierarchies in favor of heterarchic
hierarchies allows for greater freedoms in human actualization. As an analogy, this
is similar to the phenomenon of seed dispersal far removed from the parent, which
allowed plants to colonize and spread itself across the globe. Dispersal of seeds
opened up ecological pathways for the evolution of plants into rich, diverse forms
that could successfully exploit the local micro-ecologies. However, the difference
between seed dispersal and knowledge growth patterns is that while the available
global surface area with sufficient sunlight for plant growth is limited, such is not
618 J. Thomas and P. Mantri

the case for the highly fractalized heterarchically hierarchical knowledge archi-
tectures (Thomas and Zaytseva 2016). Indeed, such an abstract space is only limited
by our imagination. In other words, there is sufficient “surface area” across the
richly fractalized knowledge fabric for humans to flourish without having to
dominate and extinguish the creative entelechy in each other in the narrow hier-
archies of disjointed knowledge fragments. Such ought to be the proper solution to
the Rich-Get-Richer power-law conundrum. And in this context, cloud computing
is like the wind that is dispersing the seeds of human imagination into ever wider
mash-ups of creative, uncharted territories.
While power laws go hand in hand with CAS-generated hierarchies, they do not
necessarily vanish with CAS-generated heterarchic hierarchies. Both depth-driving
hierarchies and breadth-seeking heterarchies are necessary for balanced growth, but
need to exist in symbiosis. For example, the ancient redwood trees (symbolizing
deep hierarchies) did not go extinct when grasslands (symbolizing wide heterar-
chies) started appearing 55 MYA. Instead, each continued to thrive separately
within their respective niches while collaborating on the wider photosynthetic gas
exchange cycles. Also, within its respective niche, each exhibits various power laws
in its relevant allometric measurements (Anfodillo et al. 2013; Niklas 1994). In
other words, it is not a single power law across the overall span. Instead, it is broken
into separate segments, each governing a separate niche. This is what was referred
to earlier as the broken power law (Fig. 22.5a).
In cloud computing, Loboz finds significant backing for the power law when he
analyzed Microsoft’s Azure Resource usage patterns (Loboz 2010):
Analysis of daily resource usage by customer accounts on two Azure storage clusters had
shown that the distribution of the resource usage on any given day is very heavy-tailed. We
have found, for five different resource types that distributions are far from normal, expo-
nential, or even log-normal—in fact they either are power-law or closer to power-law than
any of the aforementioned distributions.

When jointly plotted against each other, the power-law distribution highlights
key regions of over/underestimation of probabilities under the guidance of the
bell-curve logic. As shown in Fig. 22.5a, the tails at the extremes are often
underestimated, while the middle is overestimated. Underestimated tail regions are
denoted as fat tails. For example, suppose that the T-shirt industry was to be
designed under the guidance of the normal distribution but in fact exhibited
power-law distribution. If that were to be the case, it would be as if ready-made
clothes were being designed mostly for the middle region of the normal curve (i.e.,
small, medium, and large); but sizable populations continue to show up at the retail
stores who are extremely large or extremely small. This, of course, does not happen
in human body proportions which generally follows the Gaussian-normal curve.
But if that were to be the case, huge sections of the market that pertained to the
massive fat tail at the lower end (i.e., extra small) of the market would effectively be
invisible to the designers. Likewise, some extremely large-bodied individuals
would also not be serviced and therefore forced to obtain custom-tailored clothing.
Also, in the middle regions (i.e., small, medium, and large), there would be a huge
22 Axiomatic Cloud Computing Architectural Design 619

surplus and wastage. Clearly, misjudging the market would be a major problem for
the designer.
Note that fat tails may also form under non-power-law situations such as shown
in Fig. 22.5c which overlays the normal against the log-normal. Like the power
law, log-normal is also multiplicative, but places restrictions such as stock prices
never being allowed to fall below zero. Examples of log-normal distributions
include the following:

• file size distribution on the Internet;

• the Internet traffic rate;
• how long users stay and peruse an online article.

As shown in Fig. 22.5b, the defining property of the power-law distribution is

that it is scale-free. In other words, in transitioning from x to 2x (or any multiple),
the ratio of the distribution is constant (i.e., invariant) no matter what that x is. In
other words, it exhibits the fractal property of self-similarity (i.e., the whole has the
same shape as the part). Under purely hierarchic dominance, the power-law
structure that results would be the trivial and featureless straight line (Fig. 22.5b),
but when mixed with heterarchic influences, the resultant broken power-law
structure is far more interesting and realistic.
From an AD perspective, what is critically relevant about the power-law dis-
tribution is that the fat tails are often underestimated and fall outside the main-
stream. This means that the design requirements need to be freshly induced, and the
design itself needs to be attempted in a solution-neutral, de novo fashion. This is
also probably the reason why the agile movement has resonated so well in our
modern knowledge economy. If the requirements need to be freshly induced in
order to cater for the fat tails, any top-down waterfall-type approach that caters to
the middle regions of the bell-curve logic would incur major costs of wasted effort
and misdirected resources. Power-law distribution is central to the modern
knowledge economy. And fundamentally, given its structuring ability when dealing
with solution-neutral/de novo problem contexts, AD is uniquely situated in rising
up to the modern challenge of designerly misguidance under the normal curve.
Note that the two fat tails as shown in Fig. 22.5a are neither symmetric nor
perhaps of equal significance. Given the relative probabilities between the two,
occurrences of the fat tail at the short end at left are far more likely than that of the
long end at right. In other words, there is far more statistics available for the short
end as compared to the long end (for example, far more small earthquakes versus
just a few really large ones). But from an impact perspective, events in the long end
are probably far more consequential than that at the short end (for example, the
energy released as well as damage from a very large earthquake). Also, tracing the
causal linkages at the short end is far more hierarchical and Markovian (Wikipedia
2019f) (i.e., lacking in memory) than that at the long end. In contrast, the tracing of
causal linkages in phenomenon that exhibit memory requires far more heterarchical
thinking (which currently is more demanding and therefore in short supply).
620 J. Thomas and P. Mantri

Example 22.4
Compare and contrast the normal versus the power-law distribution from an AD
perspective.

Solution: If the knowledge economy was to be erroneously considered as normally dis-

tributed, then the idea that the mass market is homogeneous and lumped in the central regions
of the bell curve could have deleterious consequences in understanding the demand curve,
including keeping major sections of the fractalized and non-homogeneous fat tails invisible to
the designer who may be expecting the normal curve. It is effectively as if the designer is only
aware of a single, homogeneous mass market FR from the central zone of the bell curve when,
in fact, the power-law distribution exhibits at least three distinct FR regions:

i. a mass market that is highly fractalized and located at the short end of the distribution;
ii. a middle region that is thinner than the short end of the power-law distribution as well
as the central zone of the bell curve; and
iii. a high-impact fat tail at the far end that is thicker than the corresponding section of the
bell curve.
From an axiomatic framework, such a problem-statement mismatch would be
akin to designing for a single FR when the actual problem had at least three, if
not more (especially in a broken power-law context). Given the cloud-driven
acceleration of our knowledge economy, such broken power-law mismatches
may be expected to increase if this problem is not properly understood and
corrected.

22.7 Axiomatic Trace

Given the knowledge hierarchy framework, the AD trace may be depicted as shown
in Fig. 22.6. Given that human knowledge is hierarchical, the design trace (Tho-
mas 1995) that leverages this knowledge is likewise hierarchical. As explained in
Sect. 22.6, fat tails expose designerly blind spots that need to be problem-abstracted
afresh and designed in a solution-neutral, de novo fashion. De novo designs also
occur in uncharted contexts (such as cloud-native architectures) that require the
designer to freshly induce the highest level problem context that will govern the
overall design. In contrast, on-prem designs that compete with the cloud have legacy
commitments that need to be carefully re-engineered in the context of the cloud as
they have substantial amounts of the designtrace locked up in long-shelf-life, cap-ex
obligations. In all of these de novo, cloud-native/hybrid contexts, the axiomatic
approach could provide critical insights as to how best to proceed.
As Prof. Suh indicates in (Suh 1990), design “involves four distinct aspects of
engineering and scientific endeavor” as listed below:

• problem definition;
• creative leap;
• analytical process; and
• overall testing and validation.
22 Axiomatic Cloud Computing Architectural Design 621

Fig. 22.6 Axiomatic Design

trace along K|H. (Reproduced
from Thomas and Mantri
(2019), originally published
open access under a CC BY
4.0 license: https://doi.org/10.
1051/matecconf/
201930100024)

Problem definition involves problem perception (PP), problem abstraction-

synthetic (PAs) as well as problem abstraction-analytic (PAa). PP is how we per-
ceive the problem (e.g., patient perceiving pain in the chest and showing up at the
doctors). PAs is the diagnosis of the problem at the right level of abstraction in an
essentialized sense (e.g., the medical doctor having done sufficient tests on the
patient decides that the patient has a life-threatening blockage in the coronary
artery). When the patient takes the diagnosis and asks for a second opinion about
the diagnosis that would be PAa: problem abstraction-analysis. If the presiding
doctors were Sigwart and Puel (Wikipedia 2019g), the creative design leap
622 J. Thomas and P. Mantri

(DCL) could be the world’s very first heart stent. Prior to the advent of the coronary
stent, the medical profession has had close to two centuries of experience in stenting
of vessels in other minor organs. All of this knowledge would be relevant as prior
art and therefore inform the de novo design of the coronary heart stent. Ideally, it
should be explicitly captured as part of the growing K|H. Along with the axiomatic
tools and corollaries from AD, the prior art (whether implicitly or explicitly cap-
tured in the K|H) helps in the proper analysis of the design (i.e., in the Da: design
analysis step). Despite the prior art, every de novo design probably has unique
elements (that falls outside the current prior art) which requires rigorous testing.
Commenting on the hierarchical nature of design, Prof. Suh indicates in (Suh
1990) that

• Everything we do in design has a hierarchical nature to it. That is, decisions

must be made in order of importance by decomposing the problem into a
hierarchy… When such a hierarchical nature of decision making is not utilized,
the process of decision making becomes very complex.
• The designer must recognize and take advantage of the existence of the functional
and physical hierarchies. A good designer can identify the most important FRs at
each level of the functional tree by eliminating secondary factors from consider-
ation. Less-able designers often try to consider all the FRs of every level simul-
taneously, rather than making use of the hierarchical nature of FRs and DP’s.

The above sentiment is the strongest indication that AD is closely aligned with
K|H. The only distinction is that hierarchies are not just relevant in the top-down
decompositional phase, but it is also of equal (if not more) relevance in the original
problem abstraction phase too (i.e., PAs and PAa). The familiar set of design
matrices (as shown in the red-dot-outlined offset in Fig. 22.6) also captures the
hierarchical trace.
The step-by-step decomposition of the abstract design is aided by four auxiliary
design processes:

• DD = Design decomposition;
• DDa = Design decomposition analysis;
• PAns = Problem abstraction (nested)-synthesis;
• PAna = Problem abstraction (nested)-analysis.

These steps are templated along the abductive cascade as shown in Fig. 22.4a.
The fourth major step in the design process, namely, overall system testing and
validation (OSTV) makes sure that the original problem (in our above example, the
pain in the chest) has been adequately addressed.
Tracing designs across the knowledge repository could be of value in at least six
different ways:

• By tracing the design across a well-explicated heterarchic knowledge hierarchy,

the design is also well documented. Documentation is apparently burdensome in
22 Axiomatic Cloud Computing Architectural Design 623

modern agile practices on account of the effort involved. The design-trace

approach could help to overcome this burden by leveraging and reusing that
which is common.
• When the design trace is used for capturing evolving families of designs that are
related by a common problem context, it creates a phylogeny which could be
mined for stigmergic patterns which otherwise would be missed.
• The design trace would add a valuable pedagogic tool for teaching design.
• In the hierarchical composition/decomposition of the design, the trace could help
assure that the conceptual order is being maintained. In other words, it would be
out of place to witness higher level abstractions showing up at lower level
designs. And vice versa, it would also violate the knowledge trace if lower level
abstractions show up in the higher rungs.
• As per the ironic process theory (Wikipedia 2019h), attempts to suppress certain
thoughts, unfortunately, make it all the more likely to happen. Colloquially this
is called the “don’t think about the white bear” problem which results in the
subject trapped in the very same thought process that is taboo. Likewise, the
requirement to think out of the box in a solution-neutral, de novo sense is much
harder when a solution already exists. Such cognitive traps may be avoided if the
mind could free-range and view the overall conceptual landscape with the cur-
rent de vetus design being included rather than excluded.
• Given the nature of the intense specialization in modern knowledge economies
(i.e., the problem of the dearth of generalists), problems and solutions are posed
within the limited domain expertise of the designer. By tracing the design across
the heterarchic knowledge hierarchy, such self-limiting parochialisms may be
avoided.

These are some of the myriad ways that the tracing of design across the heter-
archic knowledge hierarchy could benefit AD.
Example 22.5
Discuss the critical role that PAns: problem abstraction (nested) synthesis plays in the
proper decomposition of a conceptual design.

Solution: Creative induction plays a significant role throughout the design process. This is
especially true during the upward arching problem abstraction phase PAs: problem
abstraction-synthesis. The inductive base for such an upward leap would involve amassing
sufficient evidence from the forensic engineering of multiple cases that relate to the
underlying problem. While it is clear that inductive synthesis plays a major role in the root
problem-statement phase, it is less well known that induction is also involved in each of the
decompositional phases of a conceptual design (i.e., PAns). Furthermore, higher up the
conceptual hierarchy that the decompositional phase is situated, greater is the need for the
nested PAns. Each of these PAns steps is creatively inductive, and therefore favoring the
human touch. There is a fundamental difference between PAs and PAns: while PAs is
open-ended and problem-focused, PAns is circumscribed by the conceptual design from
above and is, therefore, solution-focused. Also, having a database of well-documented past
designs could help rapidly triangulate and structure the otherwise error-prone inductive arch
of PAns.
624 J. Thomas and P. Mantri

22.8 Cynefin

While earlier technologies had recognizable life-cycle trajectories that could be

analyzed along simple, clear, well-structured, top-down frameworks such as SWOT
matrices (Wikipedia 2019i), Porter’s five competitive forces (Wikipedia 2019j),
etc., the strategic- and business-oriented framing of cloud computing has been
addressed in mostly a piecemeal fashion (as for example, frameworks for cloud
security (Jansen and Grance 2011), governance (The Open Group 2019), migration
(Passmore 2016), vendor selection (Cloud Industry Forum 2019), etc.). Snowden’s
Cynefin framework (Snowden and Boone 2007) is an integrated, inductive,
bottom-up sensemaking framework that is complexity-aware and therefore of
substantial relevance in the cloud context. It has, however, yet to be adapted for the
cloud computing context (Wong 2011). Along with casting the Cynefin approach in
the knowledge hierarchy framework, the following discussion highlights the cloud
computing potential for Cynefin.
The Cynefin framework (Fig. 22.7) highlights both the opportunities and the
challenges faced by architects embracing the complexity challenge (Snowden and
Boone 2007):
In a complex context, however, right answers can’t be ferreted out. It’s like the difference
between, say, a Ferrari and the Brazilian rainforest… Ferraris are complicated machines,
but an expert mechanic can take one apart and reassemble it without changing a thing. The
car is static, and the whole is the sum of its parts. The rainforest, on the other hand, is in
constant flux—a species becomes extinct, weather patterns change, an agricultural project
reroutes a water source—and the whole is far more than the sum of its parts. This is the
realm of “unknown unknowns,” and it is the domain to which much of contemporary
business has shifted.

Cynefin is unique in emphasizing distinct and discernable managerial realms and

heuristics (Fig. 22.7) where either reductionism (i.e., the whole is the sum of its
parts) or holism (i.e., the whole is more than the sum of its parts) is the dominant
operative. Holism and reductionism have ancient Greek heritage. As shown in
Fig. 22.8, cloud architects may present three distinct temperaments when consid-
ering holistic versus reductionistic tendencies. The ideal cloud architect (i.e., the
architect engineer) is both a generalist (in a big-picture sense) and a specialist (from
a fast-moving technology perspective). Such an architect is able to engage the a-b
CAS structures seamlessly.
Such a skill-set profile could be characterized as the T-profile (i.e., broad as the
head of the T as well as deep as the leg the T). The ideal cloud architect ought to be
able to anticipate and envision technological shifts that might trigger business
emergences as well as business shifts that might trigger technological emergences
(in a CAS sense). Furthermore, the ideal architect ought to have the necessary
people skills/soft skills to communicate, persuade, motivate, and navigate across
complex corporate terrains in order to help bring about the requisite corporate
realignments that the envisioned emergences entail. As mentioned in Sect. 22.6,
given the fact that the AD framework is capable of both hierarchical and
hierarchy-bridging heterarchic designs, it is uniquely positioned in training the
22 Axiomatic Cloud Computing Architectural Design 625

Fig. 22.7 Cynefin and knowledge hierarchy. (Reproduced from Thomas and Mantri (2019),
originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201930100024)

Fig. 22.8 Holism, reductionism, and cloud architects (Images from Wikipedia (Zapf 1801;
Jastrow 2006)). (Reproduced from Thomas and Mantri (2019), originally published open access
under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/201930100024)
626 J. Thomas and P. Mantri

aforementioned architect engineer. Also, if the design trace is mapped against the
knowledge hierarchy phylogeny, it could help anticipate potential emergences.
Furthermore, given the fact that it is able to smoothly range across the abstraction
spectrum, it is also able to assimilate both managerial and technical expertise into
an integrated approach. Even the development of soft skills as a well-organized
collection of building-block skill sets could be reduced to a problem of design
(Sonmez 2014). Nevertheless, AD is not a panacea; it requires collaborative con-
tributions from auxiliary frameworks (such as Cynefin) to help it in solving modern
complex problems (such as cloud computing).
The vertical axis in Cynefin divides the ordered (on the right) versus the
unordered (on the left). Of vital strategic essence is properly locating the problem in
the proper regime. Here, the onus is to lead with bottom-up data in finding the right
regime rather than applying any given framework in a top-down sense (i.e., let
bottom-up induction have dominance).
By casting the Cynefin framework alongside the knowledge hierarchy frame-
work (Fig. 22.7), it becomes clear that both the simple and the complicated are
operating along well-structured knowledge hierarchies. In contrast, the necessary
inductive base has yet to be established in both the complex and the chaotic
regimes. The difference between the complex and the chaotic is that the former has
at least partial conceptual order that overlaps with the conceptually known world.
Furthermore, in the chaotic realm, higher values are under imminent threat (as in a
medical emergency room situation) and require quick heuristic-based thinking and
safety-enhancing actions. In the cloud context, chaos is when there is a major data
breach with the host organization facing an existential crisis.
The knowledge hierarchy framework shows the continuum between the various
regimes, i.e., reductionism and holism are not set against each other. Instead, each
requires the other in order for knowledge to progress. Given that many of modern
business problems manifest first in the complex realm, one should, therefore, expect
that after analysis, some parts of the problem would be treated in a reductionist
sense, while others in a holistic sense. Cynefin, however, warns about the danger of
treating a complex problem as if it were simple or complicated. The warning is that
this could lead to failover of the project into disorder (shown centrally as well as
with a bottom swoosh in Fig. 22.7) where the managerial governance is itself lost.
Problem-solving in the Cynefin world is to be contained within the separate
regimes. Such a heightened sense of alarm would be short-sighted. As indicated
above, the right approach would have been to partition and transition the reducible
parts of the problem over to the simple/complicated regimes while dealing with the
non-reducible parts in its own rights.
Consider, for example, some of the professional practices in software engi-
neering. The top-down, deduction-biased, process-heavy, waterfall framework was
perhaps adequately suited for an earlier era of simpler software development; it,
however, fails in any of the other regimes where induction dominates, and the
rigidity of the process becomes a bureaucratic ball-and-chain against agility and
innovation. In contrast, the agile framework is better suited for the complicated
regime. Also (with adequate care), it could iteratively move the complex into a
22 Axiomatic Cloud Computing Architectural Design 627

more manageable “complicated” regime. This is similar to the transformation of

time-dependent complexity in the Axiomatic Design/complexity theory (AD/CT)
framework (Suh 2005) with the caveat that order and disorder are not merely
temporal—it could also be geometric, chemical, informational, biological, etc.
The last line in the aforementioned quote (i.e., …this is the realm of “unknown
unknowns,” and it is the domain to which much of contemporary business has
shifted) is relevant in the cloud context. The failure of design here is that while the
majority of FR-DPs in a given problem context is situated in the well-established
industry practices of the simple/complicated realms, a few top-level components of
the problem are frequently situated in the complex realm. Failure also happens when
engineering and management locate the totality of the problem in one or the other
realms and therefore miss the combinatorial. These same blind spots probably existed
even prior to the advent of cloud computing. But what has changed with the cloud is
the rapidity with which novel, top-level problems with the potential for major impact
shows up. The fundamental problem of cloud computing, therefore, involves expe-
ditiously coming up with de novo designs for a few of the top-level, rapidly evolving
FR–DP problem components even while much of the adjacent/lower level FR–DP
components remain static and therefore re-targetable (with minimal change) from the
existing legacy/de vetus play. The challenge, therefore, is in facing the rapidity with
which the top layers need to be continuously shaped, reshaped, and reformulated. It is
in these de novo, de vetus, and mixed cases that the sensemaking framework of
Cynefin, along with the structuring that the AD framework provides, helps. In the
presence of |h/|H, a judicious mixture of the various components (i.e., complex,
complicated, and simple) would be more realistic.
Example 22.6
Cast the three types of cloud architects (CAs) in the axiomatic as well as the CAS a/b
framework.

Solution: This is a case of two FRs and either one or two DPs. CAs need both breadth and
depth. Satisfying just one of the job requirements would yield a coupled design.
Architect-Engineers with both breadth and depth could satisfy both these requirements, thus
yielding a decoupled design. This would have to be a lower triangle decoupled design as
the breadth stricture sets the necessary context for the build-up of depth. Framing it now in
a CAS framework, the scope and context for the breadth requirement form at the b1 level
while that for depth forms at the b2 level. As a general rule, each new a $ b iteration
increases the fractalization of the domain under purview.

22.9 Cloud OODA

We now turn to Boyd’s OODA framework (Coram 2002) which is well placed in
coming to terms with two of the most fundamental concepts in modern strategy,
namely, asymmetric warfare and fast transients. Asymmetric warfare is related to
heterarchies, while fast transients is related to hierarchies. Military
628 J. Thomas and P. Mantri

strategist/systems architect (of the LWF: lightweight fighter program, which gave
rise to the legendary F16 fighter plane) Colonel John Boyd highlighted the
importance of fast transients in his OODA (observe–orient–decide–act) loop
framework (see Fig. 22.9 as well as (Wikipedia 2019k)):
Idea of fast transients suggests that, in order to win, we should operate at a faster tempo or
rhythm than our adversaries—or, better yet, get inside our adversary’s
Observation-Orientation-Decision-Action time cycle or loop.

The four-stage OODA loop is depicted in Fig. 22.9a. In Fig. 22.9b, the four
stages of the OODA loop (as well as that for OODA feedback) are traced across the
knowledge hierarchy along with the trace for AD (i.e., key elements from Fig. 22.6)
we had discussed in Sect. 22.7.
In (Richards 2004), Colonel Richards reviews the strategic value of the OODA
framework:
What Boyd discovered was that the side with the quicker OODA loops began to exert a
strange and terrifying effect on its opponent. Quicker OODA execution caused the slower
side to begin falling farther and farther behind events, to begin to lose touch with the
situation. Acting like the “asymmetric fast transients” experienced by fighter pilots, these
mismatches with reality caused the more agile side to start becoming ambiguous in the
mind of the less agile.

The key phrase “asymmetric fast transient” needs to be carefully dissected and
elaborated upon.
In both OODA and AD, the critical step is that of problem abstraction, which in
OODA is denoted as the orientation stage. As Boyd writes (Wikipedia 2019k)
The second O, orientation—as the repository of our genetic heritage, cultural tradition,
and previous experiences—is the most important part of the O-O-D-A loop since it shapes
the way we observe, the way we decide, the way we act.

Likewise, as Prof. Suh indicates in (Suh 1990):

It may be useful to state once more the importance of proper problem definition: the
perceived needs must be reduced to an imaginative set of FRs as the first and most critical
stage of the design process.

Both approaches highlight the seminal value of problem abstraction. When

placed in the K|H context, it becomes clear why the problem abstraction phase (i.e.,
orientation) has such strategic import. In comparison to every other step, it has the
longest arc (i.e., PAs/PAa in Figs. 22.6 and 22.9b) along the K|H. Furthermore,
being upward oriented, it is fundamentally inductive, which makes it more
error-prone. The manner in which Boyd came up with the E-M
(energy-maneuverability) theory (Coram 2002) that informs fighter-aircraft design
illustrates the inductive challenge. The E-M theory resulted from synthesizing
various contributing insights, including firsthand practical experience battling
Mig-15 fighters in the Korean War, studies in strategic warfare,
thermodynamics/aerodynamics as well as voluminous computer simulations
designed to help create flight performance envelopes. From a design theory
22 Axiomatic Cloud Computing Architectural Design 629

Fig. 22.9 a The OODA

loop; b OODA, K|H, and AD;
c OODA strategies.
(Reproduced from Thomas
and Mantri (2019), originally
published open access under a
CC BY 4.0 license: https://
doi.org/10.1051/matecconf/
201930100024)
630 J. Thomas and P. Mantri

perspective, what was being induced was not the FR specifications for any specific
fighter plane; instead, it was establishing the overall theoretical design envelope for
all possible fighter planes. With the benefit of this framework, a fighter-aircraft
designer could reasonably articulate a feasible set of FRs based on the perceived
needs of the customer (i.e., the defense department).
But the E-M theory helps structure-only part of the OODA loop (i.e., PAa/PAs).
Furthermore, it could not explain the 10:1 kill-ratio between the F-86 (predecessor
to F-16) and the technically superior Mig-15 (higher ceiling, tighter turn radius,
higher maximum speed). In fact, the Mig-15 was better positioned on the E-M
profile. As Colonel Richards observed in (Richards 2004), the controversy was that
the “MiG’s theoretically higher EM performance rarely led to wins in actual or
even in practice air-to-air combat.” Puzzling over this ambiguity, Boyd noticed the
following countervailing facts in favor of the F-86 (Richards 2004):

• The F-86’s bubble canopy provided a simple, direct 360-degree field of vision
that helped the pilot become better situationally aware (in visually detecting the
enemy aircraft) as compared to the constraining view (i.e., the rear view was
blocked) from the Mig-15 canopy. Engaging in a fast-transient dog fight requires
better tools for situation awareness and problem perception (PP in Figs. 22.6 and
22.9b). This pertains to the observe phase of the OODA loop. Likewise, in the
context of the cloud (with divided responsibility between the client and the cloud
vendor), it is worthwhile creating automated monitoring algorithms based on the
telemetric signals (i.e., DP5 in Fig. 22.1) along with simple dashboards to help
the DevOps team be continuously situation-aware. Clutter, complexity worship
and confusion need to be removed in favor of simplicity. This is in contrast to
the Cynefin framework which cautions against moving from the complex realm
into the simple for fear of falling into disorder. In other words, while Cynefin is
valuable in becoming situationally aware as to where one begins with, it may not
be the right choice in transitioning the problem (either whole or in part) from the
complex into the more manageable complicated and simple regimes. Note that in
pursuing the simple versus the complex/complicated, OODA agrees with AD’s
Axiom II which recommends the minimization of information content.
• The F-86 had fully hydraulic controls which allowed the pilot to command the
aircraft with a single finger. In contrast, the Mig-15 pilot had to strenuously exert
physical labor in controlling the aircraft that relied on mechanical linkages. The
ability of the Mig-15 pilot to act in a consistent, coordinated fashion degraded
under physical exhaustion. Note that this pertains to the act phase of the OODA
loop. Poor actions in one loop feed every phase in the follow-on OODA loop,
thereby creating vicious cycles. Here again, we see the importance of simplicity,
but now in the realm of human action. The tactical end result of the hydraulic
controls was that in comparison to the Mig-15 pilot, the F86 pilot could engage
in faster transitions from one OODA maneuver to another.
22 Axiomatic Cloud Computing Architectural Design 631

Along with comparable EM credentials, the addition of a bubble canopy as well

as hydraulic controls made the F-86 fighter plane strategically superior to the
Mig-15 in an “asymmetric fast-transient sense.” The asymmetry here is from the
OODA-related strategic value of the bubble canopy as well as the hydraulic con-
trols. And it is this asymmetry that is feeding the step-by-step degradation of the
adversary, the process whereby the slower side begins “falling farther and farther
behind events” and “to lose touch with the situation (Richards 2004).”
It is important to realize that cloud computing is strategic and needs to be
considered in warfare terms. Critical assets deployed on the cloud are no longer
business-as-usual; it, in fact, is being positioned on a winner-take-all basis. In this
context (and as was illustrated in the fighter-plane example), every aspect of the
OODA loop needs to be examined for its asymmetry potential. These insights,
when abstracted and generalized using the KA|h|H framework along with AD, have
the potential to scale.
It is true that as the top-level design is decomposed, there are many more nested
abstractions (i.e., the steps denoted as PAns in Figs. 22.6 and 22.9b). Each of these
is a baby step (compared to PAs) and closely triangulated with the aid of the
downward arching deductive decompositions (i.e., the steps denoted as DD in
Figs. 22.6 and 22.9b). These nested PAns are the orientation phases of the nested
OODA loops which when viewed from the top create a fractal pattern of spirals (see
top-left in Fig. 22.9b).
Consider once again the problem of problem abstraction (i.e., PAa/PAs). In a
negative sense, if the problem is poorly stated, or worse if the wrong problem is
being addressed, all downstream effort is wasted. In the medical context, this is
called misdiagnosis. To put a human face on the cost of misdiagnosis, consider the
following healthcare summary from a John Hopkins (Newman-Toker 2013) report:

• significant cases of permanent injury or death from misdiagnosis estimated to be

80 K to 160 K/year;
• diagnostic-error-related medical claims dominate in the total count (28.6%) and
amount (35.2%);
• for new diagnosis, error may range up to 15%.

As we saw in Sect. 22.6 (where we discussed the problem of fat tails in the
context of the power law), if we perceive the problem to be normally distributed
when in fact it is operating under the power-law or log-normal distributions, it
would force the designers to effectively “bark up the wrong tree.” Referring back to
the Borders bookstore case discussed in Sect. 22.1, the management failed to
perceive the threat of the Internet as well as the demise of the brick-and-mortar,
resulting in the bankruptcy of the firm in a short 10-year period.
In contrast, if designers could rapidly diagnose and orient (i.e., problem abstract
in essentials) around real and pressing issues, it would make all the difference in
orienting the design and execution teams in the right direction and solving the real
problem.
632 J. Thomas and P. Mantri

Note that there are subtleties and nuances in the concept of asymmetric fast
transients associated with the OODA loop. For example, in the above illustration of
how hydraulic controls contributed to fast transitions from one OODA maneuver to
another, speed is of the essence. But note that such an OODA loop is situated at the
tactical level. In cloud computing, this would be similar to the roles and respon-
sibilities of the DevOps team. In Fig. 22.9b, the tactical OODA has been demar-
cated within a red-dotted boundary in the middle. In contrast, in the case of the
overall OODA loop which included the E-M theory, the issue is more strategic as it
is not so much about physical speed; it is about swiftly tapping into accurate mental
models (Coram 2002). In other words, the issue is not so much about how quickly
the fighter pilot can tactically transition from one maneuver to another in order to
get around and get behind the enemy aircraft for establishing air dominance in
three-dimensional physical space; the issue is more about how quickly and effi-
ciently the aircraft system architect could design, test, and transition from one
configuration to another in the abstract conceptual space, as and when the
requirements change.
Given the broad centrality and reach of cloud computing, the architectural design
of the cloud is fundamentally strategic in nature. But given the speed at which cloud
systems may be assembled and torn-down, everyone is operating under tactical time
pressures. Change is relentless in the cloud, and time is of the essence. In other
words, architects do not have the luxury of time in formulating their architectural
designs. They have to deliver strategic designs under tactical time constraints.
In Cloudonomics (Weinman 2008), Weinman highlighted a similar theme in his
7/10 laws of Cloudonomics:
A real-time enterprise derives competitive advantage from responding to changing business
conditions and opportunities faster than the competition.

It is in the fast-transient challenge that the cloud poses an “existential threat” to

the legacy (e.g., banking) as well as an “irresistible opportunity” to the upstarts
(e.g., FinTech).
As was the case with Boyd (who had hands-on experience flying sorties in the
Korean war), the cloud architect needs to have hands-on experience in all facets of
the cloud. And as indicated in Fig. 22.8, the cloud architect also needs to be able to
span the strategic business/technological context in the widest possible abstract
terms. In the context of cloud architecture, this means that in contrast to the tac-
tically oriented DevOps role, the above strategically significant architectural role
ought to be rightly designated as BizArch. Such combinations occur because the
looping mechanism in OODA is a guided search that is looping across organiza-
tional levels and responsibilities in trying to solve the problem posed by the
top-level FR–DP. Here OODA is genuinely heterarchical in creating and encour-
aging information flows across organizational/disciplinary spaces. In this process,
OODA brings about creative mash-ups such as the aforementioned DevOps and
BizArch. For example, ArchOps is a role being popularized by the Amazon Web
Services (DevOps@Logicworks 2019; Hohpe 2019). In a similar vein, other such
22 Axiomatic Cloud Computing Architectural Design 633

mash-ups may include GovArch and BizVend (with Gov for governance and Vend
for vendor relations), etc.
While OODA is about decision-making in rapidly evolving strategic and tactical
terrains, Boyd also highlighted the importance of exploiting strategic asymmetries
between allies and combatants. Between two or more allies and/or adversaries,
asymmetries may exist along various dimensions, including asymmetries in wealth,
culture, manpower, mental models, technological prowess, physical skill sets, team
cohesion, group dynamics, etc. For example, the bubble canopy as well as hydraulic
controls pertain to technological asymmetries. But among all the above asymme-
tries, mental model asymmetries are unique in that they fall out of heterarchical
knowledge asymmetries. And as indicated in Sect. 22.5, when corporations (such as
Borders versus Amazon) lock horns across these asymmetries, they have the power
to systematically and inexorably lay waste (in classic OODA style) the
knowledge-gapped corporation or nation.
Figure 22.9c captures a few of the OODA patterns whereby asymmetric fast
transients could make or break a corporation. The context is that of a far-flung,
multi-national corporation that is hierarchically administered. The view is the
top-down view as was described in the context of a similar view shown in the
top-left of Fig. 22.9b. The various OODA strategies may be characterized as listed
below:

i. Base OODA in sync with apex and executing harmoniously: This is the
benchmark case where the apex and basal layers of the organization are
working in close coordination (both in time-synchrony and in policy). In the
cloud context, it means that the organization is well aware of the heightened
cloud cadence and is fundamentally organized top to bottom with this in
view.
ii. Heterarchical connect and working in sync: This is the more demanding
benchmark that requires heterarchically hierarchical units of a far-flung
multi-national corporation being able to work in close coordination. In the
cloud context, it means that the corporation is deft and experienced in
navigating the international regulatory strictures regarding data location, data
privacy, and security.
iii. Base OODA faster than apex and lacking in coordination: This is where the
apex and basal layers are asynchronous, with the basal layers evolving at a
much faster rate. This situation is not uncommon, given that many corpo-
rations treat technology merely as an enabler, and not sufficiently strategic. In
extreme cases, it could lead to counterproductive corporate pathologies such
as insubordination and toxic workplace cultures. Such a situation might arise
in the cloud context if the top management bought into the cloud-as-hype
without properly understanding its strategic implications. In other words,
there was no serious rethinking of the current architectural strategies in view
of the cloud.
634 J. Thomas and P. Mantri

iv. Base OODA slower than apex and lagging in execution: This could happen
when the founders of the firm who are technically and managerially astute
are leading the firm, but under growth pressures they went on a hiring spree
that failed to do due-diligence and quality check.
v. Heterarchical disconnect unknowingly at cross-purpose: This is the classic
case of “the left hand not knowing what the right hand does (Wikipedia
2019l).” Given that there are no linking mechanisms between far-flung
multi-national corporate units, the disconnect continues unabated for long
durations. In the cloud context, such multi-national corporate dysfunctions
could be disastrous given the fact that miscommunications and cultural
insensitivities could escalate rapidly out of control (both within the firm and
in the larger marketplace sense).
vi. Base OODA intimate with apex OODA and executing faster and at
cross-purpose: This is clearly corporate sabotage. In a legacy corporation,
there usually exists sufficient checks and balances to make sure that such
intentional and highly coordinated actions and their actors do not find refuge.
But in the cloud context, given the speed at which policies and personnel can
change, it is not unlikely that highly coordinated sabotage teams could take
residence, and no one is the wiser.
vii. Adversary intimate with your OODA and executing cross-purpose deep
within and at a faster rate: This is the case of an external agent that is
somehow privy to the internal corporate strategies and technological initia-
tives. Being intimate with the corporate agenda and capabilities, such an
agent could competitively outsmart the corporation. This is the case of
corporate espionage. In the cloud context, a single breach (at the firm,
partner, or vendor boundaries) could drop unwanted listening assets within
the vast sprawl of the firm’s cloud belongings. Stigmergic listening (see
Sect. 22.3) would fall in this category.
viii. OODA swarm attack—some internal, some external, and some that change
shape: This is akin to the distributed denial-of-service (DDoS) attack. The
classic DDoS swarm is synchronous and distributed but coordinated (in an
algorithmic sense). It, however, lacks a clear center which could be targeted.
Also, what it may lack in sophistication, it makes it up in the sheer number of
resource exhausting attacks that are launched. A CAS swarm is similar, but it
is shape-shifting and does not have to be synchronous. It could, therefore,
play out in time, giving it more ambiguity and cover. An OODA-CAS swarm
could be asynchronous, shape-shifting, and executing along asymmetric fast
transients. If such sophistication exists on the cloud, it would probably be at
the behest of a state sponsor.

Given the close alignment between OODA and AD, the question is what added
value does OODA provide AD? Likewise, what added value does AD provide
OODA? OODA was formulated for the purpose of establishing dominance in
asymmetric warfare, i.e., how to exploit subtle differences (in mental models,
technological prowess’s, physical skill sets, and group dynamics between adverse
22 Axiomatic Cloud Computing Architectural Design 635

and aligned participants) in order to obtain strategic (and often changing) objectives
in fluid, fast-changing environments. In contrast, AD was formulated for the
explicit purpose of establishing how design may be “made into a science (Suh
1990).” It is true that both OODA and AD track closely when mapped along K|H,
but they are not dealing with the same issues. AD is more generic than OODA and
could be used to enhance the design aspects of OODA. In other words, OODA
could benefit from the logical tripartite mappings between FRs, DPs, and PVs. In a
similar vein, AD could benefit from OODA in recognizing the strategic significance
and asymmetric competitive value of certain key elements of a proposed design.
Example 22.7
How would you correct an organizational disconnect within a heterarchic multi-national
such as shown in Fig. 22.9.c.v using the axiomatic approach?

Solution: The fundamental problem that organizations face is the rapidly increasing pace of
business decisions that need to be synchronized across the global sprawl of far-flung
multi-national units. In this context, the cloud computing initiative has only accelerated this
pace. The differential rates in the OODA loop between two or more such organizational
units often lead to a lack of coherence and synergy, both between individual units and in
an overall sense. In the hierarchical case (i.e., Figures 22.9.c.iii, c.iv), the addition of an
OODA-sensitive (vertical) synchronization control unit as an organizational DP would help
coordinate the handling of the increasing pace of modern business decisions. The heter-
archical disconnect as exhibited in Fig. 22.9.c.v would likewise require the addition of
similar OODA-sensitive synchronization control units as organizational DPs to help smooth
and coordinate the lateral disconnect. While organizational control units (i.e., the gover-
nance bodies) that synchronize business decisions (both vertically and horizontally) are not
new, what is currently lacking is the awareness of the need for synchronizing the rate of
change across the vast organizational sprawl.

22.10 Iterative Axiomatic Maturity

Diagram (AMD) Ensembles

If failure could be de-stigmatized, it has potent stigmergic value in systematically

learning and becoming familiar with the design landscape. Unfortunately, the
various “Fail-X” phrases in use today have created needless confusion.
Failure as a worthwhile end goal does not make sense; it only has value in an
interim sense when it is being harnessed for learning the topography of a complex
design surface or improving a given design that is flawed. It is never the end goal.
A company that prides itself on delivering nothing but failures will cease to exist
before long.
Following are two Fail-X listings, the former which could lead to stigmergic
learning, and the later which could very well thwart it. From a strategic OODA
perspective, we would want our own teams (and those of our allies) to embrace the
former while encouraging our adversaries to embrace the later:
636 J. Thomas and P. Mantri

Fail-X’s that encourage stigmergic learning:

i. Safe-Fail (Snowden 2006): Through deliberately engineered failures,

designers learn the complex terrain being navigated.
ii. Fail-Fast: To find if a system is ill-designed, best to quickly test it early on,
rather than prolonging the discovery of the flaw.
iii. Fail-Often: Setting up a sequence of small bite-size goalposts, which creates
opportunities for many successes and failures.
iv. Fail-Early: Provide greater latitude for failure, but only during the early
phases of a project which could create the right attitude of seriousness toward
successes and failures.
v. Fail-Forward: In failure, take advantage of the lessons learned for the next
iteration.
vi. Fail-Small: Set up small, bite-size goalposts, which creates opportunities for
small successes and failures.
vii. Fail-Well: Compartmentalize and contain the failure from spreading. This
agrees well with the uncoupled/decoupled design in AD.
viii. Fail-Safe: In production, if and when you fail, fail safely by not endangering
life and property.

Fail-X’s that discourage stigmergic learning:

i. Fail-Backward: Lack of team resilience/ability to recover from failure, hence

no learning.
ii. Fail-Big: Fail colossally at something big. This is high risk for high rewards. It
is not driven by incremental, iterative stigmergic learning.
iii. Fail-Badly: When the system fails, it is catastrophic, and therefore no learning.
iv. Fail-Silent: When failure happens, it is suppressed from public view with no
indication of failure, and therefore no learning.
v. Fail-Deadly: Mutually assured destruction (MAD) such as is the case for
nuclear deterrence. Cloud development has yet to reach the stage of cloud wars
where MAD may be relevant.

These are some of the colloquial ways to characterize candidate designs. In

general, designs (as well as design processes) may be characterized and critiqued in
at least six different ways:

i. Viability of the select design in regard to the CRs/FRs. This includes the
Independence Axiom.
ii. Performance of the select design in functional comparison to a family of other
valid designs. This includes the Information Axiom as a selection criterion.
iii. Performance of the select design in cost comparison to a family of other valid
designs.
iv. Performance of the select design in view of the change dynamic that is evident
in the phylogeny of the problem domain. In other words, what is the rate of
22 Axiomatic Cloud Computing Architectural Design 637

change in the FRs? And how does the candidate design cater to such a rate of
change in the FRs?
v. Performance of the design process that created the design with respect to
time-to-market? How quickly can such designs be designed and implemented?
vi. Performance of the design as part of a family of other designs that exist in the
same enclave (for example, being hosted by the same cloud vendor).
vii. Performance of the design as well as the design process in catering to emergent
FRs that may not be known a priori until the design/design process has
matured sufficiently.

While the first two criteria from the above list do have axiomatic representation,
the last five items do not have representation (except perhaps as rigid, a priori
constraints). The agile methodology has made valuable contributions in addressing
the last of the above cases, namely, how to go about designing in the complex realm
(see Fig. 22.7) where FRs are emergent and unknown a priori, where statistics is
rare, and where fat tails and power laws are common. AD could learn from the agile
gambit and address this lacuna in a principled fashion without compromising its
holistic strengths. Note that in a similar vein, items iii–vi above also remain
unaddressed in AD.
As discussed earlier in Sect. 22.9, the problem abstraction phase (i.e., PAs in
Fig. 22.9b) arches upward on the inductive design trace. But induction takes its
own time; it does not have the same rapid cadence of deductive logic. In other
words, it takes time to marshal the necessary holistic view that the axiomatic
approach prefers.
Also, the inductive component is more error-prone. Misdiagnosis of the problem
can be costly. Agile adopts various Fail-X approaches in order to mitigate this risk.
For example, the safe-fail approach (Snowden 2006) accommodates FRs being
emergent in a CAS sense. In other words, the FRs do not exist a priori. Referring
back to the manner in which the b-layer forms in a CAS setup (Fig. 22.3a), agile
seems to suggest that FRs are stigmergically emergent (using agile-style Post-It
notes, etc.). They cannot be discerned a priori except through repeated trials and
errors. They emerge along the design pathway that is engaged in solving a larger
problem. In such a situation, a top-down linear approach such as the waterfall
model that does not iterate back to the root FRs will misdiagnose the problem and
therefore fail to address the emergent issues.
As would be the case in other approaches, true rapidity/agility in agile occurs in
problem contexts that have been well plowed. And in problem contexts that are
more bottom-up, inductive, and tentative in nature (such as Cynefin’s complex
regime), agile adopts an iterative approach of sprints and retrospectives which
would necessarily take longer. If that is the case, where exactly is the agility in
agile? In order to understand the agility aspect of agile, let us compare the waterfall
approach to agile. If in waterfall-type approaches, spurious FRs are being addressed
while relevant FRs (which happen to be emergent) are left unaddressed, it is
obvious that such designs are never timely. As an analogy, if the train arrives at the
wrong destination, it is indefinitely late for arrival at the right destination. The
638 J. Thomas and P. Mantri

agility of agile is in solving the right problems, and not because it is inherently
agile. In other words, it is fundamentally on the basis of the emergent FR problem
(which is real and salient, especially in a knowledge economy) that agile has staked
its claim on the totality of design.
However, such a broad claim on the totality of design needs to be challenged.
For example, the differences between agile and a more formal/structured approach
have been couched in the agile manifesto as “X over Y (Kern 2018).” Such a
conflicted approach is unnecessary. For example, in the AD context (as shown in
Fig. 22.10), it is more than likely that it is a case of “X because of Y.” The industry
is increasingly becoming aware of many of the agilist blind spots (Meyer 2014;
Winters 2016; Brizard 2015).
Agile fails to master the problem of design in at least two significant ways:

• Design is holistic. Piecemeal designs seldom scale, especially in the complex

regime. Given the significance of the emergent FR problem (especially in a
knowledge economy), agile downplays the very concept of system-wide/holistic
design. This can be a problem when considering cybersecurity which tends to
expose unaddressed gaps in non-systemic, ad hoc designs. However, it would
not be too difficult to bring in holism (especially during the final-stage sprints
and retrospectives).

Fig. 22.10 X over Y (Agile) versus X because of Y (AD). (Reproduced from Thomas and Mantri
(2019), originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/
matecconf/201930100024)
22 Axiomatic Cloud Computing Architectural Design 639

• Without formal documentation, the stigmergic pattern-making process rarely

takes root. In the current technological context where anything and everything is
being dutifully noted and recorded (thanks to myriad IoTs and other instruments
of constant vigil), it is unfortunate that seamless and effortless documentation is
not de-jure in the design realm. The agile manifesto on documentation (i.e.,
“working software over comprehensive documentation”) is therefore misguided.
In the modern age, design documentation should be automated and effortless.
The bias that agile has against documentation is quite anachronistic and
self-defeating especially given the role that documentation plays in the formation
of stigmergic patterns. Once again, it would not be too difficult to correct this
problem.

From an AD perspective, there are valuable insights to be learned from agile. As

mentioned earlier, the key distinction between agile and AD is in the context of
emergent FRs. This is an area where AD could learn from agile. While keeping the
holistic view, the axiomatic approach can strategically borrow agile’s iterative
stance (which incidentally agrees with the OODA loop). This is precisely what has
been proposed in (Puik and Ceglarek 2018) wherein Puik and Ceglarek have
advocated using the axiomatic maturity diagram (AMD) in an ingenious way for
bringing synergy between the axiomatic and agile approaches. The following dis-
cussion extends the AMD approach by explicitly adding the stigmergic tightening
of the AMD patterns along the time dimension (see Fig. 22.11).
As suggested in (Puik and Ceglarek 2018), the key is in understanding the
competing thrusts of the main three drivers of the design process (Fig. 22.11a):

• Do the right thing: Find the right set of FRs, DPs, PVs that has total, holistic
capture of the problem at hand and a design for it that satisfies the Independence
Axiom (Fig. 22.11/X-axis).
• Do things optimally: Minimize the holistic, system-wide information content
among candidate designs in order to find the right solution (Fig. 22.11/Z-axis)
• Do things fast: Navigate the design space in order to reach the target solution at
a rapid pace, i.e., minimize h, the temporal splay along Fig. 22.11/Y-axis.

Emphasizing any one of these drivers stand-alone or even two-by-two will only
succeed up to a certain point. As suggested in (Puik and Ceglarek 2018), the
axiomatic maturity diagram (AMD) could be modified ever so slightly to incor-
porate agile’s iterative insight (Fig. 22.11b).
Normally, the AMD is a 2D plot that captures the path dominance between
Independence and Information Axioms. For de novo designs, usually the Inde-
pendence Axiom dominates the initial stages; it is only after the design trace has
reached a certain level of maturity that the information axion is triggered. For de
vetus cases, the Information Axiom may gainfully be put to use early on given the
level of experience and history that is readily available. For the problematic case of
emergent FRs (which by its nature is de novo), an iterative approach from the very
beginning (i.e., as in Probe ! Sense ! Respond from the complex realm of
640 J. Thomas and P. Mantri

Fig. 22.11 Iterative axiomatic maturity diagram (AMD) ensembles (adapted from (Puik and
Ceglarek 2018)). (Reproduced from Thomas and Mantri (2019), originally published open access
under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/201930100024)
22 Axiomatic Cloud Computing Architectural Design 641

Cynefin, Fig. 22.7) is probably the right way to proceed. Providing such early and
repeated reality checks is one of the hallmarks of the agile approach. As suggested
in (Puik and Ceglarek 2018), the dotted curves in Figs. 22.11b3 to b4 capture such
agilest iterations that have been abstracted as the underlying wedge. The distinction
between Fig. 22.11b3 and b4 is that the former (i.e., Fig. 22.11b3) has a slight bias
toward the Information Axiom which should show up in colloquialisms such as
Fail-Fast, Fail-Often, Fail-Early, and Fail-Small. In contrast, the latter (i.e.,
Fig. 22.11b4) is more conservative in its approach and might include colloqui-
alisms such as Safe-Fail, Fail-Well, and Fail-Forward. These loose colloquialisms
are merely suggestive and they are not categorical distinctions. The tan-colored
wedge captures a variety of iterative/combinatorial possibilities that span between
the boundaries of Figs. 22.11b3 and b4.
In each sprint (as Fig. 22.11b1 indicates), only a partial list of the required
functionality is being delivered. There are four sprints that have been outlined, of
which only the third is colored tan and bounded by Figs. 22.11b3 to b4. Fig-
ure 22.11b2 indicates the end of all sprints, with the full working functionality
being delivered. Each of the sprints could have fresh additions (of emergent FRs).
FRs may also be deleted from a previous partially working solution.
The slight temporal splay (h) of the very first wedge captures the time taken in
the underlying iteration (Fig. 22.11c). The agilest argument is that without an
iterative approach, the splay would be much wider; the FRs often hastily and
improperly induced; and the resultant design rigid, fragile, and even abandoned.
These agilest arguments are completely valid. But there is nothing stopping the
axiomatic approach from adopting an iterative stance as shown. And the compar-
ative advantage of the axiomatic approach is that it never loses sight of the fact that
the problem of design is holistic. With that ideal in mind, it is therefore motivated to
reach for the holistic view.
The contour plot of the four sprints is depicted in Fig. 22.11d. It captures all four
partial/fully functional iterative AMD patterns along with an overall containing
envelope (in light green) that captures it in abstract. Just the abstract outline is
repeated in the lower AMD figure in order to help capture the larger stigmergic
patterns if proper documentation was to be enforced.
Figure 22.11e is the stigmergic capture of AMD patterns of similar projects (i.e.,
the phylogeny) which may exist both within the firm and across the industry.
Figure 22.11f captures the contour plot of the stigmergic tightening of the various
iterative AMD patterns as displayed in Fig. 22.11e. Such stigmergic tightening is
feasible, given the fact that the axiomatics encourages holistic information mini-
mization along with succinct, documentation-friendly, design matrix design cap-
ture. Thus, across multiple AMD wedge iterations (Fig. 22.11e, f), the axiomatic
approach is capable of not just capturing the holistic demand, but it is also capable
of leveraging the stigmergic tightening of the AMD patterns.
In time, the AMD trace is well established and settles into a thin slice (displayed
as a dark-orange slice at the center). Once the iterative uncertainty has been
removed and the stigmergic wedge well established, a waterfall approach would
work just as well. In other words, the complex has now been tamed (at least in part)
642 J. Thomas and P. Mantri

into the complicated/simple regimes, i.e., good practice and best practice available
in the industry (see Cynefin, Fig. 22.7). It is indeed a waste of scarce resources to
indiscriminately treat every problem as if it were always attached to the complex
regime in each of its decompositional details.
Example 22.8
Discuss practical ways to aid the stigmergic tightening of an ensemble of iterative AMD
patterns (as shown in Fig. 22.11f).

Solution: Any form of stigmergic tightening requires the existence and accumulation of a
wide variety of stigmergic markings that have been carefully curated across multiple
iterations. In the case of software/hardware assets (such as in the case of the cloud), this
would imply the capture of succinct documentation of the design as well as the operational
logs of the application in use. As shown in Example 22.3, stigmergic tightening occurs (in
the local context) between control-point DPs that need to be discerned and explicated. With
each such iterative/tightening run, the local context becomes more and more global in scope
until the overall contour plot is in play. The key step in the above stigmergic tightening run
is the discernment and explicit establishment of successive control-point DPs.

22.11 Non-Functional Requirements from a Complex

Adaptive System Perspective

Non-functional requirements (non-FRs) dominate the design of cloud architectures.

These include system-wide -ilities such as

• scalability;
• adaptability;
• reliability;
• security;
• maintainability;
• availability;
• customizability;
• testability, etc.

But as a term of common usage, the non-FRs are indeed a misnomer; there is
nothing non-functional about the concern at hand. Thus, the ability to rapidly
scale-up or scale-down a certain website based on the seasonal load at hand is most
definitely a FRs—except that instead of it being at a final user level, it is now at a
system-wide/population-wide level. It is, therefore, a failure in the design com-
munity to understand the functional domain when it asserts that the above list of
requirements is somehow non-functional (Thompson 2014; Adams 2016).
The deeper question that, however, needs to be probed is where do these
non-FRs come from? To solve the puzzle of the origin of functionally relevant
non-FRs, one has to study the requirement formation as an iterative CAS process
(Fig. 22.3b). If the regular FRs are to be found in the b1 ensemble, the non-FRs are
22 Axiomatic Cloud Computing Architectural Design 643

to be found in the b2 ensemble. Thus, the so-called non-FRs are indeed FRs and
therefore subject to the standard design approaches. Indeed, it is easy to project a
future point where there will be a b3 ensemble one day. It is, therefore, best to
acknowledge levels of design in preference to ad hoc terminology (such as non-FR)
when dealing with systems that are fundamentally CAS in nature. In other words,
the various -ilities requirements are likely to adapt and evolve away from the
current strictures. Or to put it differently, of all the ilities mentioned above, the
adaptability requirement is dominant and overarching over asset/agent/artifact space
as well as time. The cloud is fundamentally operating and evolving at a much faster
cadence than the systems that are residing in the traditional on-prem ecosystem.
Architecting such rapidly evolving systems requires the architect to understand the
CAS a $ b pattern forming mechanism and consider the problem of design from
the highest b-level reached thus far, as well as the projected CAS trajectory.
Restating the above in Boydian OODA terms, asymmetric fast transients in the
cloud are operating with latency in milliseconds instead of weeks or months.
Missteps can be fatal. This is especially true when considering cloud cybersecurity.
Example 22.9
Discuss the possible consequences of treating system-wide FRs as non-FRs.

Solution: When legitimate, system-wide FRs are treated as if they were non-functional
stipulations that exist outside the purview of design principles, it effectively results in ad -
hoc designs that are fragile, non-holistic, and seldom scale. As they are considered to be
non-functional, valuable compositional insights from the AD framework are likely to be
ignored. Larger the number of such competing non-FRs, greater is the risk of lack of
coherence. In some cases (such as security as a non-FR), such dereliction of the designerly
mandate could be fatal.

22.12 Security of Cloud Computing

Once the corporate assets have been migrated (in part or whole) over to the cloud, the
legacy threat surface is significantly altered. Depending on the cloud footprint
(Sroczkowski 2019) (i.e., IaaS, PaaS, SaaS), the onus of securing the assets is now a
joint responsibility. What was previously an in-house responsibility is now a shared
undertaking that juxtaposes the evolving footprint and complexity of the vendors’
cloud infrastructure and operations (with its global reach and geophysical asset spread)
against the cloud maturity of the in-house architects, developers, users, and operators.
Some recent quotes regarding cloud-related security breaches in the news
include the following:

• A group of angry customers filed a lawsuit against Capital One…following the

hack that affected more than 106 million people…the group also named Amazon
Web Services, Capital One’s cloud provider, alleging the tech giant is also
culpable for the breach. (August 9, 2019, (Levy 2019))
644 J. Thomas and P. Mantri

• Unprotected Database Puts 65% of American Households at Risk…data

included on the 24 GB database [hosted on Microsoft Azure] is people’s full
names, full street addresses, marital status, date of birth, income bracket, home
ownership status and more. (April 29, 2019. (Morris 2019))
• Verizon Partner Exposed Millions of Customer Accounts…a misconfigured
cloud-based file repository exposed the names, addresses, account details, and
account personal identification numbers (PINs) of as many as 14 million US
customers of telecommunications carrier Verizon. (December 12, 2018.
(O’Sullivan 2018))
• In December [2018], Google revealed the details of…data breach that hap-
pened…leaving the data of close to 52.5 million Google + users vulnerable to
hackers…Google + is shutting down in 2019. (Roussey 2019)
• Between July and September 2018, hackers leveraged the “view as” feature on
Facebook to steal tokens for access profiles. This breach compromised the
personal details of close to 29 million users across the globe. It divulged per-
sonal information including names, phone numbers, email addresses, and other
personal details Facebook collected over time. The breach was disclosed to the
general users on September 28. (Roussey 2019)

According to a recent multi-factor industry survey of the leading concerns of

400,000 cloud cybersecurity professionals (Cloud Security Report 2018):

• Biggest threats to cloud security—misconfiguration of cloud platforms (62%).

• Legacy on-prem security tools are ill-designed for the virtual, dynamic, and
distributed cloud. Traditional security solutions either don’t work at all in cloud
environments or have only limited functionality (84%).
• Operationally, the leading security control challenges (SOCs) include (a) poor
visibility into infrastructure security (43%); (b) compliance (38%); (c) Setting
consistent security policies across cloud and on-premises environments (35%);
and (d) security not keeping up with the pace of change in applications (35%).
• Compared to an on-prem deployment, 49% of the respondents believe that the
public cloud poses a greater security risk, 30% responded about the same, and
only 17% responded in favor of the cloud.

Many cloud-related cyberinsecurities are related to poor design. These design

issues could show up as misconfiguration, poor consistency, lack of transparency,
inadequate security tools, inability to keep pace with application changes, etc. With
the rapid evolution of the threat surface (Thomas and Mantri 2015a, b), it is no
wonder that even well-configured systems sprout leaks.
But the greatest cybersecurity vulnerability in cloud computing has to do with a
flawed defense-in-depth (DiD) mindset that currently dominates the industry (see
Sect. 22.5 for a prior discussion on this topic in connection to knowledge
hierarchy/heterarchy). There are two major problems with the DiD posture (Igbe
2017):
22 Axiomatic Cloud Computing Architectural Design 645

• Technological Exposure: Rapid evolution of technology exposes interface

mismatches across the various onion layers. Ubiquitous and remotely located
edge nodes and IoT devices can be compromised. Each system is designed for a
certain system range. Swarm attacks can be designed to overwhelm the elasticity
of these system ranges. While traditional attacks used to be sequentially directed
against any single layer, modern swarm attacks target multiple layers simulta-
neously. Traditional attacks were directed at the network layer that is easier to
detect, while modern attacks target the application layer that is harder to detect.
• Socio-technical Exposure: Human cognitive biases and vulnerabilities are
constantly being probed via social engineering techniques and strategies such as
weak passwords, robocalls, and phishing attacks. The performance overhead
from any of the security defenses can reach a point of intolerance for the human
agents to become overwhelmed and then switch it off.

In contrast to defense-in-depth, the emerging defense-in-breadth (DiB) approach

also considers the wider angle of vulnerabilities. NIST has helped in defining and
contrasting the two (Ross et al. 2016):

• Defense-in-depth: Information security strategy integrating people, technology,

and operations capabilities to establish variable barriers across multiple layers
and missions of the organization.
• Defense-in-breadth: A planned, systematic set of multidisciplinary activities
that seek to identify, manage, and reduce risk of exploitable vulnerabilities at
every stage of the system, network, or subcomponent life cycle (system, network,
or product design and development; manufacturing; packaging; assembly;
system integration; distribution; operations; maintenance; and retirement).

It is the multidisciplinary emphasis in DiB that provides the hint that the
underlying system and the concerns thereof are heterarchical in nature (see
Fig. 22.12). In contrast to traditional hierarchical systems, the combinatorial space
that the architect has to master when dealing with heterarchical systems is vastly
more complex and expanded. As shown in Sect. 22.10 (i.e., iterative AMD), such
systems may best be designed in a systematic and principled way by leveraging the
stigmergic patterns that accrue over time.
Example 22.10
Why is it easier to penetrate heterarchic vulnerabilities of a system as compared to the
hierarchical?

Solution: All designs require the basic awareness that a problem exists in the first place.
Given the highly specialized training that professionals (such as engineers and scientists)
receive, it is far more likely that they are aware of the hierarchical linkages within their
areas of specialization as compared to heterarchical linkages between specialties. In other
words, heterarchical vulnerabilities are located in the collective blind spot of a hierarchi-
cally trained workforce. This is the fundamental reason why it is easier to penetrate
heterarchic vulnerabilities of a system as compared to the hierarchical.
646 J. Thomas and P. Mantri

Fig. 22.12 Defense-in-depth/width (hierarchy vs. heterarchy). (Reproduced from Thomas and
Mantri (2019), originally published open access under a CC BY 4.0 license: https://doi.org/10.
1051/matecconf/201930100024)

22.13 Econo-Complex Adaptive System Strategy

in the Cloud

None of the top four companies by Market Cap from the year 2000 (i.e., General
Electric, ExxonMobil, Pfizer, and Citigroup) were able to retain their leadership
positions. Instead, it is now Microsoft, Amazon, Apple, and Alphabet (Google).
Among these four, three have strong vendor presence in the cloud; only Apple is
weak and is dependent on Amazon. As reported in (Statt 2019):
Apple is deeply reliant on AWS to operate core parts of its business, even though doing so
means working with a soon-to-be-rival in online video and a current competitor in areas
like artificial intelligence, streaming music, and smart home products.

Apple is facing a similar problem as was the case with Borders. Many other
major firms (including Netflix) face the same conundrum that Borders faced when
dealing with a power-law-driven knowledge economy.
As was mentioned earlier, cloud computing is strategic and needs to be con-
sidered in warfare terms. Critical assets deployed on the cloud are no longer
business-as-usual, it, in fact, is being positioned on an asymmetric, winner-take-all,
22 Axiomatic Cloud Computing Architectural Design 647

fast-transient OODA basis. In such a context, every aspect of the OODA loop needs
to be examined for its asymmetry potential.
In order to scale a higher peak, one often has to climb down from the current
summit. Traditionally, it is economics that has provided navigational guidance in
the summit-to-summit route-finding endeavors. But what if the economic heuristics
we have here-to-fore depended on themselves change? What if the governing
economic rules are being rewritten to accommodate fat tails even as the corporate
econo-strategist tries to navigate across the shifting landscape? It is in this sense that
cloud computing is “both an existential threat and an irresistible opportunity
(Weinman 2012).”
It is indeed an existential threat for those wedded to the status quo, but it is also
an opportunity for those willing “to climb down” from their current summits and
look at the de novo econo-design landscape that is opening up. These include social
network economies (Facebook, LinkedIn, Twitter etc.), Big Data plays (Green-
Plum, Cloudera/Hortonworks, Palantir, etc.), streaming economies (Netflix, Hulu,
Amazon Prime, YouTube TV, etc.), gaming plays (PlayStation Now, Shadow,
GeForce NOW, etc.), and others.
As one steps back from the micro-view in order to then take in the big
picture/macro-view, it is becoming increasingly clear that the architectural design of
the cloud computing play is anything but simple and straightforward. For example,
in the context of cybersecurity, the attack surface is the map of all ports of entry/exit
whereby an attacker may launch an attack and/or spirit off corporate assets. And as
the business grows, the dangers of cyberinsecurity increase as the attack surface
proliferates and mutates across pathways and resources that the corporation does
not fully command.
Likewise, the economic attack surface for cloud-based corporate ventures is
orders of magnitude more complex than the traditional on-prem ventures. Once the
corporation establishes key assets in the cloud, it is operating in a shared envi-
ronment where its business activities leave open and visible stigmergic traces.
These include what is openly known about the strengths/weaknesses of the cloud
vendor.
Once the business model proves viable in the cloud, the competitive attack
surface can bring in a swarm of traditional/non-traditional challengers uncon-
strained by erstwhile barriers to entry that have either been leveled or rendered
irrelevant. The cloud fundamentally lowers many of the traditional barriers to entry.
For example, it is true that the Chinese firm Ant-Financial was rebuffed in estab-
lishing a FinTech foothold in the U.S. via the purchase of MoneyGram (Roume-
liotis 2018). But there are no such barriers for the rapid migration/replication by
native agents of a successful business model such as the Ant-Financial. And it is in
such a rapidly evolving cloud ecosystem that a well-thought-out adaptive archi-
tectural design could take advantage of the new economies-of-scale and elasticity
that the cloud makes available. Creating a viable cloud enterprise increasingly
involves architecting of a complex adaptive system (Urquhart 2012). The following
648 J. Thomas and P. Mantri

Fig. 22.13 CAS-based extension of Porters 5 a-level competitive cloud forces (Wikipedia 2019j)
to include 3 b-level forces. (Reproduced from Thomas and Mantri (2019), originally published
open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/201930100024)

discussion is a CAS-based extension of Porters original five a-level competitive

cloud forces to include three b-level forces for a total of eight (see Fig. 22.13).
The fundamental difference to notice is that all eight competitive forces are at the
periphery and jostling with each other for dominance, whereas Porter’s framework
had centralized on current rivalry as the core node. Also, the forces are charac-
terized using the CAS a-b notation to help identify the level at which agents are
forming their strategic intent. Porter’s formulation had the original five forces
occurring at the inter-agent a-level. The three new entries (i.e., cybersecurity,
technological shifts, and global regulations) exist at the b ! a level. These addi-
tions are just a sampling of the missing entries in the original formulation; there
could, of course, be many more than just these three additions.
The set of eight competitive forces is split into three groups: hierarchical,
heterarchical, and mixed. Hierarchical forces are incremental, slow-moving, and
works within the confines of the current competitive landscape. For example, if
there is a PaaS Lock-In (as is at node 6 in Fig. 22.13), it is not easy to shift out of
this. Heterarchical forces could also be slow-moving. But a few are strategic, rapid,
22 Axiomatic Cloud Computing Architectural Design 649

and works orthogonal to the current competitive landscape (e.g., 5G at node 8,

Fig. 22.13). Mixed items have both heterarchical and hierarchical elements within
them.
To illustrate the framework, consider the IaaS cloud offering. Here, there is a low
threat from new entrants (see node 1, Fig. 22.13) given the sizable upfront Capex
outlays that the hosting of a full-stack cloud infrastructure requires. Nevertheless,
given the state-level strategic significance of the cloud for any given nation, the
above deep-pocket Capex stricture, therefore, does not preclude state-level agents
from entering the competitive landscape. Yet given the speed at which the under-
lying technology shifts (for example, the impending shift from 4G to 5G), any state
sponsor faces steep odds in keeping up with the fast-moving heterarchic front. State
sponsors tend to be hierarchical in nature; they have yet to master the unwieldy
socio-technical heterarchic hierarchies of today. But when cybersecurity issues are
raised (see node-2, Fig. 22.13), the state may have a strategic incentive to take on
the infrastructural challenge head-on. All three competitive forces as shown in
brown (new entrants, technological shifts, and cybersecurity) fall in the heterar-
chical space as each of these driving forces has the power to shift the competitive
landscape from a point of view that is orthogonal to the a $ a focus (in Porter’s
framework), and that too, potentially overnight.
Now consider the hierarchical tranche. From a consumer’s perspective, the shift
from Capex to Opex has fundamentally lowered the bargaining power of the IaaS
supplier as there is very little lock-in (see node 6, Fig. 22.13). If one of the com-
petitors offers a similar or enhanced set of infrastructural offerings at a sufficiently
competitive price (competitive enough to overcome the switching costs), the con-
sumer has every incentive to shift. The real bargaining power of the supplier exists
not at the level of any given a-agent-level supplier–consumer contract; instead, it is
in successfully leveraging the b-level -ilities which the industry mistakenly iden-
tifies as the non-FRs (see Sect. 22.11).
The mixed set includes the realm of substitutes that rise up either hierarchically
within the current competitive landscape or heterarchically orthogonal to it. The
mixed set also includes the shifting regulatory fractal landscape that has a global
reach. It is “mixed” in the sense that much of the regulatory landscape is hierar-
chically constrained by precedent, but occasionally, the regulatory bodies do reach
across and claim jurisdiction (especially in uncharted areas that new technological
mash-ups have recently opened up). From a regulatory perspective, any of the
major state-level agents have the power to overnight shift the competitive landscape
with the understanding that when dealing with a CAS, the overall system will react
back. In other words, in a CAS system, any unilateral action would face resistance
at various levels. This is the reason why all the eight forces (and there may be many
more b-level forces) are depicted at the periphery and engaging each other in a
complete network. Porter’s framework only considered the a-level agents and the
direct interplay between them; it is silent about the b-level play wherein much of
the competitive landscape has shifted.
650 J. Thomas and P. Mantri

Example 22.11
What is the strategic impact of 5G in the context of cloud computing?

Solution: 5G is supposed to be 100 times faster than the prevalent 4G, thus bringing the
latency between devices to near-zero levels. This will flatten the current hierarchies such as
cloud$smartphones$wearables to be just between cloud and wearables. In other words,
compute, storage, and networking in the middle layers will face increasing obsolescence.
The overall architecture between the cloud and the edge devices will become more and
more bipartite with large-scale consolidation in the centralized cloud offerings on account
of the aforementioned (see Sect. 22.2) economies of scale and scope. Edge devices will
increasingly become highly sensitive and action-oriented agents alongside the begin and
end of the OODA loop (observe/act), while the middle two roles (orient/decide) are retained
at the centralized cloud.

22.14 Weickian Versus Axiomatic Adaptive Coupling

in Complex Adaptive System Architectures

Organizational Psychologist, Karl Weick originated the modern concept of loose-

versus tight-coupling in the mid-1970s. Coupling plays a central role in AD. The
following discussion frames both the Weickian and the axiomatic approaches on
coupling in the CAS framework.
As Prof. Suh noted in (Suh 1990, 2001):
When the design matrix [A] is diagonal, each of the FRs can be satisfied independently by
means of one DP. Such a design is called an uncoupled design. When the matrix is
triangular, the independence of FRs can be guaranteed if and only if the DPs are deter-
mined in a proper sequence. Such a design is called decoupled design. Any other form of
the design matrix is called a full matrix and results in a coupled design.

Likewise, as Prof. Weick noted in (Weick and Orton 1990):

…loose coupling is evident when elements affect each other “suddenly (rather than con-
tinuously), occasionally (rather than constantly), negligibly (rather than significantly),
indirectly (rather than directly), and eventually (rather than immediately).

The dynamics embedded in the Weickian concept of loose-coupling may be best

understood using CAS framework that emphasizes levels, temporal spans as well as
impact:

• Levels: Rejects direct intervention (a ! a) in favor of the indirect (a ! b ! a)

(i.e., “indirectly (rather than directly)”).
• Time Delay: Rejects the chronic and continuous in favor of episodic in the short
term or long term (i.e., “suddenly (rather than continuously), occasionally
(rather than constantly), …eventually (rather than immediately)”).
• Impact: Favors small versus the large impact (i.e., “negligibly (rather than
significantly)”).
22 Axiomatic Cloud Computing Architectural Design 651

In (Weick 1976), Prof. Weick further notes that:

By loose coupling, the author intends to convey the image that coupled events
are responsive, but that each event also preserves its own identity and some evi-
dence of its physical or logical separateness.
Based on the above characteristics, Prof. Weick defines the following types
(Weick and Orton 1990):
If there is neither responsiveness nor distinctiveness, the system is not really a
system, and it can be defined as a noncoupled system. If there is responsiveness
without distinctiveness, the system is tightly coupled. If there is distinctiveness
without responsiveness, the system is decoupled. If there is both distinctiveness and
responsiveness, the system is loosely coupled.
A system of CAS agents is responsive when sufficient b-level orchestrating
patterns have accumulated (within a given iteration) in order to help a-level agents
respond in a coordinated fashion to the events at hand. Furthermore, the system is
distinctive (within a given iteration) when the a-level agents have retained sufficient
degrees of freedom (DOFs) to move orthogonally to the restrictions placed by the
previous b ! a coordination.
Based on the above CAS reframing, the four Weickian categories (see
Fig. 22.14 above) may be restated as follows:

• Noncoupled: No a $ a, a ! b, No b ! a.
• Tightly coupled: a $ a & b ! a exists, but no orthogonal a ! b exists in the
next iteration (i.e., no degrees of freedom left).

Fig. 22.14 Axiomatic versus Weickian coupling. (Reproduced from Thomas and Mantri (2019),
originally published open access under a CC BY 4.0 license: https://doi.org/10.1051/matecconf/
201930100024)
652 J. Thomas and P. Mantri

• Decoupled: a ! b exists, but the b-level patterns are either non-existent or not
mature enough to provide b ! a coordination and control.
• Loosely coupled: Minimum a $ a; both b ! a coordination and control as
well as next orthogonal/iterative a ! b exists (i.e., remainder degrees of free-
dom exist even after b ! a coordination and control).

Loosely coupled systems are an organizational ideal for social, technical as well
as socio-technical systems. They are neither brittle (i.e., they are resilient in the face
of change dynamics) nor anarchic (i.e., not lacking in coordination and concerted
action). The resilience is because (when faced with novel situations) the agents
retain sufficient degrees of freedom to coordinate and self-organize new structures
outside the current strictures. But for the most part, the agents are operating within
the current strictures. Or as Weick asserts in (Weick 2001), “the real trick in highly
reliable systems is somehow to achieve simultaneous centralization and
decentralization.”
Here, the centralization mandate is achieved via b ! a coordination (Thomas
and Mantri 2018). The a-level decentralization captures the remaining orthogonal
degrees of available freedom given the current state of centralization. As the overall
system gears across multiple a ! b iterations (see Fig. 22.3b), and the agents get
more organized, the loose-coupling frontier shifts to the highest iterate. Everything
lower down is stable, highly orchestrated, coordinated, waterfallish, and tightly
coupled. Since there are no degrees of freedom available at the lower levels, con-
sequently the problem of design itself does not exist. An example of this is
assembly programming, i.e., for most higher level programming language cases,
there are no degrees of freedom left at this base layer. All the remaining degrees of
freedom exist in one of the higher level programming languages. Thus, the problem
of design exists only in the outermost iterate where degrees of freedom exist for the
respective agents. All the lower levels exist as constraints and context for the
problem of design.
The fundamental problem of design only exists where there are degrees of
freedom available, which typically exists in the outermost layers of the CAS sys-
tem. The problem of coupling in AD, therefore, exists at the outermost iterate layer
(see bottom-right in Fig. 22.14). Given the symbiotic reach of both of these sys-
tems, the AD framework (as well as that for agile) could, therefore, work well with
the Weickian loose-coupling framework. Across each iteration (as the system gears
up), the axiomatic approach of preferring uncoupled/decoupled in preference to the
coupled is, therefore, sound advice.
The question now is how does the above discussion relate to cloud architectures?
As we indicated in Sect. 22.9, the cloud is anything but static. Instead, it is a
socio-technical CAS system that is rapidly morphing, adapting, and evolving. And
it is in this context that the Weickian approach could work symbiotically alongside
the axiomatic approaches, i.e., via Weickian loose-coupling at the basal layers
alongside AD uncoupling/decoupling at the growing meristem of the CAS edifice.
For example, as in (Fehling et al. 2014), by using an intermediary broker mecha-
nism, direct a $ a interactions are streamlined and minimized. Instead, new broker
22 Axiomatic Cloud Computing Architectural Design 653

agents are used (under the guidance of b ! a patterns) for indirect information
exchange between agents. AD could assist in the establishment of such a CAS
scaffold. This would provide for framing the loose-coupling between a $ a. Once
the a-b architecture of the loosely coupled bipartite CAS framework is established,
AD could further be used in designing each of the layers as per the AD
coupling/decoupling logic.
Example 22.12
Compare the axiomatic versus Weickian structures to that found in plant biology between
meristematic and non-meristematic tissue.

Solution: In plant biology, tissues (such as the shoot and root systems) that engage in cell
division and growth are considered as meristematic. After division, a certain number of
these daughter cells differentiate and lose their stem-cell-like flexibility to become per-
manent tissues such as the dermal tissue that functions as the protective skin for the plant.
Likewise, the axiomatic versus the Weickian structures have a very similar meristematic
versus non-meristematic property in that the growing front affords maximum design flex-
ibility in the form of designerly degrees of freedom, while the Weickian under-structures
are well defined and minimally invested in designerly degrees of freedom. But there is a
fundamental difference between the two in that the loose-coupling aspect of the underlying
Weickian structures helps to retain and exercise all the degrees of freedom that were left
unassigned. In other words, there could be valuable stem-cell-like designerly degrees of
freedom in the Weickian sub-structures that when resolved may avoid downstream
couplings.

22.15 Conclusions

A number of theoretical/practical issues related to an adaptive architectural design

surrounding cloud computing were covered in this chapter. Nine mutually sup-
portive decision-making frameworks were formally integrated for the first time in
providing a comprehensive, overarching approach to help tackle design problems at
large. These included axiomatics, knowledge hierarchy/heterarchy, Cynefin,
OODA, stigmergy, CAS, AMD, Non-FRs, and adaptive loose-coupling.
Some of the key findings include the following:

• The cloud is a CAS system. The architectural design of the cloud requires
coming to terms with the underlying CAS dynamics and patterns.
• The framework of knowledge hierarchy/heterarchy provides a simple but robust
approach to help integrate many orthogonal, but mutually supportive frame-
works such as axiomatics, OODA, and Cynefin.
• Time-axis extension of the iterative AMD approach as reported in (Puik and
Ceglarek 2018) provides a pragmatic way to bring agility, axiomatics, and the
stigmergic-tightening pattern logic together in one place to help bring about
principled design, but in an agile fashion.
• Using CAS, the non-FRs have been folded into the standard FR–DP mappings.
654 J. Thomas and P. Mantri

• Security of the cloud has been critiqued, both from a design perspective and
from a defense-in-depth/defense-in-breadth perspective.
• Using CAS, the fundamental concept of coupling has been broadened to include
loose-coupling which powers much of modern technology, including the cloud.

The global economy is facing unprecedented challenges. While cloud computing

is not an all-round panacea for alleviating the human condition, it has much to offer
in bringing us together in solving the myriad large-scale problems spanning the
globe. And as shown, design plays a central role across all levels in this
conversation.
Problems
1. Compare and contrast the historical development of space exploration versus
cloud computing using the axiomatic framework. Based on your study, what
future trends could be projected in each of the above endeavors?
2. Eusocial entities (including humans) fall back on stigmergic techniques when
the scale and scope of the problem reach beyond the problem-solving capacity
of the individual or the group. Make a case for how stigmergy may have played
a role in the creation of each of the seven wonders of the ancient world.
3. Find three examples where complex adaptive systems are operative at the micro,
meso, and macro-scales in human biology, sociology, and politics.
4. Create a conceptual trace of the knowledge hierarchy in any two domains of
your choice. Then link them up to illustrate the heterarchical connections
between the two. Study the historical chronology in the development of the
above two conceptual networks.
5. Forensically illustrate the importance of power laws in the modern knowledge
economy by studying three cases of the decline and fall of legacy corporations
in the face of upstarts.
6. Develop the econo-CAS framework (as shown in Fig. 22.13) to include three
more competitive cloud forces from Sect. 22.11 on non-FRs.
7. Investigate the role of design in cloud security.
8. Compare and contrast Weickian loose versus tight-coupling in engineering
versus socio-technical systems. Compare and contrast the role of axiomatic
decoupling in each of the above cases.

References
Adams KM (2016) Non-functional requirements in systems analysis and design. Springer
Akhtar SMF (2018) Big Data Architect’s Handbook: a guide to building proficiency in tools and
systems used by leading big data experts. Packt Publishing
Anfodillo T, Carrer M, Simini F, Popa I, Banavar JR, Maritan A (2013) An allometry-based
approach for understanding forest structure, predicting tree-size distribution and assessing the
degree of disturbance. Royal Society Publishing. https://doi.org/10.1098/rspb.2012.2375
22 Axiomatic Cloud Computing Architectural Design 655

Barabási AL, Albert R (1999) Emergence of scaling in random networks. Science 286:509. https://
doi.org/10.1126/science.286.5439.509
Brizard TJ (2015) Broken Agile. Apress
Chellapp R (1997) Intermediaries in cloud-computing: a new computing paradigm. Presented at
INFORMS Meeting, Dallas, 1997R. Chellapp, Intermediaries in Cloud-Computing: A New
Computing Paradigm, Presented at INFORMS Meeting, Dallas
Cloud Industry Forum (20019). 8 Criteria to ensure you select the right cloud service provider.
https://www.cloudindustryforum.org/content/8-criteria-ensure-you-select-right-cloud-service-
provider. Accessed 14 July 2019
Schulze H. Cloud Security Report-2018. https://pages.cloudpassage.com/rs/857-FXQ-213/images/
2018-Cloud-Security-Report%20%281%29.pdf
Coram R (2002) Boyd: the fighter pilot who changed the art of war. Brown and Company, Little
DevOps@Logicworks (2019) ArchOps vs. DevOps: Foundation and Automation. https://www.
logicworks.com/blog/2015/03/aws-devops-archops-automation/. Accessed 14 July 2019
Erl T, Khattak W, Buhler P (2016) Big data fundamentals: concepts, drivers & techniques.
Prentice Hall
Erl T, Cope R, Naserpour A (2017) Cloud computing design patterns, 1 edn. Prentice Hall
Fehling C, Leymann F, Retter R, Schupeck W, Arbitter P (2014) Cloud computing patterns:
fundamentals to design, build, and manage cloud applications. Springer
Grassé PP (1959) Insectes Sociaux VI 79
Hohpe G (2019) The architect elevator—visiting the upper floors. https://martinfowler.com/
articles/architect-elevator.html#ArchopsBuildAVerticalArchitectureTeam. Accessed 14 July
2019
Holland JH (2006) Studying complex adaptive systems. J Syst Sci Complexity 19(1):1–8
Igbe D (2017) Defense in breadth or defense in depth? https://www.cloudtechnologyexperts.com/
defense-in-breadth-or-defense-in-depth/. Accessed 14 July 2019
Jansen W, Grance T (2011) Guidelines on security and Privacy in public cloud computing.
National Institute of Standards and Technology. Special Publication 800-144. https://nvlpubs.
nist.gov/nistpubs/Legacy/SP/nistspecialpublication800-144.pdf. Accessed 14 July 2019
Jastrow (2006) Bust of aristotle. Marble, Roman copy after a Greek bronze original by Lysippos
from 330 BC; the alabaster mantle is a modern addition. Wikimedia website. https://commons.
wikimedia.org/wiki/File:Aristotle_Altemps_Inv8575.jpg. Accessed May 26 2020. Public
domain
Karthikeyan SA (2017) Azure automation using the ARM model: an in-depth guide to automation
with Azure resource manager, 1st edn. Apress
Kern J (2018) Agile in the context of a holistic approach. https://www.infoq.com/articles/agile-
holistic-approach/. Accessed 14 July 2019
Lee J, Wei T, Mukhiya SK (2018) Hands-on Big Data modeling: effective database design
techniques for data architects and business intelligence professionals. Packt Publishing
Levy N (2019) Amazon and Capital One face legal backlash after massive hack affects 106 M
customers. https://www.geekwire.com/2019/amazon-capital-one-face-lawsuits-massive-hack-
affects-106m-customers/
Loboz CZ (2010) Cloud resource usage—heavy tailed distributions invalidating traditional
capacity planning models. J Grid Comput. https://doi.org/10.1145/1996109.1996112
Louridas P, Spinellis D, Vlachos V (2008) Power laws in software. ACM Trans Soft Eng Method
18(1):1–26. Article 2. https://doi.org/10.1145/1391984.1391986
Mell PM, Grance T (2011) The NIST definition of cloud computing. Special Publication (NIST
SP)—800-145. September 28, 2011. https://www.nist.gov/publications/nist-definition-cloud-
computing
Meyer B (2014) Agile!: The good, the hype and the ugly. Springer
Morris C (2019) A Goldmine for identity thieves: unprotected database puts 65% of American
Households at risk. April 29, 2019. http://fortune.com/2019/04/29/security-gap-personal-
information-breach/
656 J. Thomas and P. Mantri

Newman-Toker D (2013) Diagnostic errors more common, costly and harmful than treatment
mistakes. https://www.hopkinsmedicine.org/news/media/releases/diagnostic_errors_more_
common_costly_and_harmful_than_treatment_mistakes. Accessed 14 July 2019
Niklas KJ (1994) Plant allometry: the scaling of form and process. University of Chicago Press
O’Sullivan D (2018) Cloud leak: how a Verizon partner exposed millions of customer accounts.
https://www.upguard.com/breaches/verizon-cloud-leak
Parunak HVD (2006) A survey of environments and mechanisms for human-human stigmergy.
Environ Multi-Agent Syst II: 163–186
Passmore E (2016) Migrating large-scale services to the cloud: a master checklist of everything
you need to know to move to the cloud. Apress
Puik E, Ceglarek D (2018) Application of axiomatic design for agile product development. In:
MATEC web of conferences 223, 01004 (2018). ICAD 2018. https://doi.org/10.1051/
matecconf/201822301004
Richards C (2004) Certain to win: the strategy of John Boyd, applied to business. Xlibris
Ross R, McEvilley M, Oren JC (2016) Systems security engineering considerations for a
multidisciplinary approach in the engineering of trustworthy secure systems. NIST Special
Publication 800-160. https://doi.org/10.6028/NIST.SP.800-160
Roumeliotis G (2018) U.S. blocks MoneyGram sale to China’s ant financial on national security
concerns. https://www.reuters.com/article/us-moneygram-intl-m-a-ant-financial/u-s-blocks-
moneygram-sale-to-chinas-ant-financial-on-national-security-concerns-idUSKBN1ER1R7.
Accessed 14 July 2019
Roussey B (2019) 10 biggest 2018 data breaches—and what they mean for 2019. March 22, 2019.
http://techgenix.com/2018-data-breaches/
Siegel E (2013) Predictive analytics: the power to predict who will click, buy, lie, or die. Wiley
Snowden D (2006) Safe-fail or fail-safe. https://cognitive-edge.com/blog/safe-fail-or-fail-safe/.
Accessed 14 July 2019
Snowden DJ, Boone ME (2007) A leader’s framework for decision making. Harvard Bus Rev.
https://hbr.org/2007/11/a-leaders-framework-for-decision-making. Accessed 14 July 2019
Sonmez J (2014) Soft skills: The software developer’s life manual. Manning Publications
Sroczkowski P (2019) Cloud: IaaS vs PaaS vs SaaS vs DaaS vs FaaS vs DBaaS. https://brainhub.
eu/blog/cloud-architecture-saas-faas-xaas/. Accessed 14 July 2019
Statt N (2019) Apple’s cloud business is hugely dependent on Amazon. https://www.theverge.
com/2019/4/22/18511148/apple-icloud-cloud-services-amazon-aws-30-million-per-month.
Accessed 14 July 2019
Suh NP (1990) The principles of design, 1st edn. Oxford University Press, New York
Suh NP (2001) Axiomatic design–advances and applications. Oxford University Press
Suh NP (2005) Complexity: theory and applications, 1st edn. Oxford University Press, New York
Surowiecki J (2005) The wisdom of crowds. Anchor
The Open Group (2019). Cloud Computing Governance Framework. http://www.opengroup.org/
cloud/gov_snapshot/p3.htm. Accessed July 14 2019
Thomas J (1995) Archstand theory of design for innovation. Ph.D. Thesis. http://dspace.mit.edu/
handle/1721.1/11722. MIT. Accessed 14 July 2019
Thomas J, Mantri P (2015a) Axiomatic design/design patterns Mashup: part 1. In: 9th International
conference on axiomatic design (ICAD), Procedia CIRP 34:269–275. http://www.
sciencedirect.com/science/article/pii/S2212827115008501
Thomas J, Mantri P (2015b) Axiomatic design/design patterns Mashup: part 2. In: 9th
International conference on axiomatic design (ICAD), Procedia CIRP 34:276–283. https://
www.sciencedirect.com/science/article/pii/S2212827115008513
Thomas J, Mantri P (2018) Complex adaptive Blockchain Governance. MATEC Web of
Conferences 223, 01010 (2018) https://doi.org/10.1051/matecconf/201822301010
Thomas J, Zaytseva A (2016) Mapping complexity/human knowledge as a complex adaptive
system. 2016 Wiley Periodicals, Inc., vol 21, no S2. https://doi.org/10.1002/cplx.21799,
https://onlinelibrary.wiley.com/doi/abs/10.1002/cplx.21799
22 Axiomatic Cloud Computing Architectural Design 657

Thomas J, Mantri P (2019) Axiomatic Cloud Computing Architectural Design. In: the 13th
International Conference on Axiomatic Design (ICAD 2019). MATEC web of conferences
301, 00024 (2019). https://doi.org/10.1051/matecconf/201930100024
Thompson MK (2014) Where is the ‘Why’ in axiomatic design? ICAD2014. The Eighth
International Conference on Axiomatic Design. Campus de Caparica, September 24–26, 2014
Tollen DW (2016) The tech contracts handbook: cloud computing agreements, software licenses,
and other IT contracts for lawyers and businesspeople. Am Bar Assoc
Urquhart J (2012) Cloud is complex—deal with it. Jan 8, 2012. https://gigaom.com/2012/01/08/
cloud-is-complex-deal-with-it/. Accessed 14 July 2019
Vleck TV (2019) Project MAC. https://multicians.org/project-mac.html. Accessed 14 July 2019
Ward H (2016) How to hire. https://medium.com/eshares-blog/how-to-hire-34f4ded5f176#.
b04yy5bka. Accessed 14 July 2019
Weick KE (1976) Educational organizations as loosely coupled systems. Adm Sci Q 21(1):1–19.
Published by: Sage Publications, Inc
Weick KE (2001) Making sense of the organization, Oxford. Blackwell Publishers, England
Weick KE, Orton JD (1990) Loosely coupled systems: a reconceptualization. loosely coupled
systems: a reconceptualization. Acad Manage Renew 15(2):203–223
Weinman J (2008) The 10 laws of cloudonomics. https://gigaom.com/2008/09/07/the-10-laws-of-
cloudonomics/. Accessed 14 July 2019
Weinman J (2012) Cloudonomics: the business value of cloud computing. Wiley
Wikipedia (2019a) Borders Group. https://en.wikipedia.org/wiki/Borders_Group. Accessed 14
July 2019
Wikipedia (2019b) Power law. https://en.wikipedia.org/wiki/Power_law. Accessed 14 July 2019
Wikipedia (2019c). Intergalactic Computer Network. https://en.wikipedia.org/wiki/Intergalactic_
Computer_Network. Accessed 14 July 2019
Wikipedia (2019d). Timeline of virtualization development. https://en.wikipedia.org/wiki/
Timeline_of_virtualization_development. Accessed 14July 2019
Wikipedia (2019e). Horizontal gene transfer. https://en.wikipedia.org/wiki/Horizontal_gene_
transfer. Accessed 14 July 2019
Wikipedia (2019f) Markov property. https://en.wikipedia.org/wiki/Markov_property. Accessed 14
July 2019
Wikipedia (2019g). Stent. https://en.wikipedia.org/wiki/Stent. Accessed 14 July 2019
Wikipedia (2019h). Ironic process theory. https://en.wikipedia.org/wiki/Ironic_process_theory.
Accessed 14 July 2019
Wikipedia (2019i) SWOT analysis. https://en.wikipedia.org/wiki/SWOT_analysis. Accessed 14
July 2019
Wikipedia (2019j). Porter’s five forces analysis. https://en.wikipedia.org/wiki/Porter%27s_five_
forces_analysis. Accessed 14 July 2019
Wikipedia (2019k). OODA loop. https://en.wikipedia.org/wiki/OODA_loop. Accessed 14 July
2019
Wikipedia (2019l). Matthew 6:3. https://en.wikipedia.org/wiki/Matthew_6:3. Accessed 14 July
2019
Wiktionary (2019). Give a man a fish and you feed him for a day. https://en.wiktionary.org/wiki/
give_a_man_a_fish_and_you_feed_him_for_a_day;_teach_a_man_to_fish_and_you_feed_
him_for_a_lifetime. Accessed 14 July 2019
Wilensky U (1997). Ants. NetLogo Models Library. http://ccl.northwestern.edu/netlogo/models/
Ants
Winters GS (2016) Why Agile is failing at large companies. Ty yn Goch Forrest Publications
Wong G (2011) Jumping the S-curve. https://cognitive-edge.com/blog/jumping-the-s-curve/.
Accessed 14 July 2019
Zapf GW (1801) Gallerie der alten Griechen und Roemer. Bürglen, Augsburg, Austria. http://digi.
ub.uni-heidelberg.de/diglit/zapf1801/, public domain
Future Design Challenges
23
Nam Pyo Suh

Abstract
One thing that never ends as long as humanity continues to exist: the emergence
of new challenging problems that humanity must solve for their posterity
through creative design. In the 1950s, few people would have thought that the
invention of polymers with amazing properties would someday create new
problems for humanity to solve. In the early decades of the twenty-first century,
the pollution of the Pacific Ocean by the debris of solid plastics from countries
around the Pacific rim is a major catastrophe with significant negative
implications for the entire global community, especially for the people living
in the nearby islands. Somehow, we must solve this problem without banning
the use of plastics in the future, because the appropriate use of plastics fulfills
human needs. Similarly, a century ago, few ever thought that the replacement of
horse-drawn buggies with automobiles would result in global warming that
might change and threaten the future of humankind. In each case, a new
innovative technology begat new unanticipated technological or societal
problems for future generations to solve or deal with. Therefore, it may be
rational and reasonable to assume that current technologies and socio-political
norms would someday become the source of new challenges for future
generations to solve. In a way, it is the price humanity has to pay to make
continuing progress through innovation and discovery. Thus, the saga of
humanity continues. The natural progression always leaves new challenges for
the next generation to solve—rationally and creatively. Therefore, the need for
new knowledge and creative design will always be with humanity as long as
humans exist. In the early twenty-first century, humanity is facing a unique set of
challenges. Either people solve them in time, or the events will overtake
humanity’s ability to solve them, leading to instability in human and societal

N. P. Suh (&)
Cambridge, MA, USA
e-mail: [email protected]

© Springer Nature Switzerland AG 2021 659

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8_23
660 N. P. Suh

evolution. Some of these problems are results of the ever-expanding population

of the world, which has been increasing at the rate of 1–2% a year and will
exceed 8 billion people in the early part of the twenty-first century. An
associated problem is a wide variation in the rate of population increase
throughout the world, which results in migration and immigration of people from
highly populated to less populated, from politically unstable to stable regions,
and from poor regions to affluent areas of the world. The problems created by
this imbalanced population increase and distribution will lead to many
challenging design issues. If we cannot come up with the right design solution,
humans might again resort to wars to settle their differences, which should be
avoided at all costs. In this sense, the human ability to design will continue to be
critical and in high demand. When and if new solutions do not emerge to deal
with new problems, Nature may dictate the future of humanity, including a
catastrophic demise of the world and humankind.
A partial list of significant design challenges humanity is facing in the early
twenty-first century is related to, but not limited to, the following:
Energy needs;
Global warming;
Instability of weather patterns;
Need for portable water;
Pollution-free electric power generation;
Pollution-free transportation systems;
Protection of privacy in the information-intensive technological world;
Preservation of green plants;
Feeding of people,
Control of weather;
Prevention and elimination of contagious diseases;
The ability to maintain free society in a peaceful world;
Supplementary brain; and
Protection of fundamental human rights.
The solution to any one of these problems could consist of different kinds of
design solutions—technical, scientific, socio-political, and economic as well as a
combination of these fields. There could be more than one solution. Regardless
of the specific approach chosen, the method and the process of developing
design solutions are ecumenical, as described in Chaps. 1–3.
Furthermore, a solution to one of these problems may generate a set of new
issues that the next generation must solve. The responsibility of the current
generation is to educate the next generation well. That is the primary guarantee
that their posterity can deal with their own set of challenges well.
The human ability to solve many of the above-listed problems are
continuously improved because humans continue to invent new tools and
advance sciences. Recent advances made in several fields such as artificial
intelligence (AI), quantum computing, neuro-biological, brain sciences,
nano-scale materials, and other areas of science and technology are most
encouraging as well as promising.
23 Future Design Challenges 661

Every problem discussed in this book, as well as many that are not addressed,
will, ultimately, require “design solutions” similar to those discussed in the
previous 22 chapters. The one(s) who create solutions to major challenging
problems will richly be rewarded through various recognitions by humanity,
which may include financial, intellectual, political, and personal recognition and
esteem. Solve them we must, because the other option will only invite demise for
humankind.
Often, we do not know which problem will require new designs. Also, at the
beginning of the execution of design, one cannot have all the specific knowledge
in the relevant field(s) because we may not have the requisite expertise. Once we
understand the design task and associated questions (i.e., functional requirement
(FRs)), we can acquire the necessary fundamental knowledge of the related
fields quickly to develop design solutions in terms of DPs and PVs. Even the
most complex problems can be systematically solved through design by
following the steps outlined in this book. An exemplary design problem related
to water and weather is discussed that can have a significant worldwide impact if
a rational design solution can be devised.

23.1 Introduction: Continuing Human “Saga”

“Inputs to an operating system generate outputs, some desired, some unwanted, and some
even detrimental. Humanity must deal with the detrimental outputs of the system. They will
not disappear by themselves.”

When people invented the internal combustion engines near the end of the
nineteenth century, they thought that finally, they could have manure-free streets in
addition to having faster moving vehicles. Automobiles, with other technologies,
indeed led the industrial renaissance of the twentieth century. A century later,
however, people are discovering that it is a major source of the environmental
problem. The emission of carbon dioxide (CO2) from these powerful engines,
which have served humanity so well, is contributing about 30% of the CO2 gases
that are responsible for greenhouse effects and global warming that are threatening
the very survival of human society as we know it.
Similarly, when the electrification of the rural community was finally accom-
plished in the United States, a new era of modernization came to every corner of the
nation. It also enabled the modernization of many other countries. Electrification
raised the standard of living, bringing in high quality of life through the replace-
ment of human labor with machines and equipment powered by electricity. Now
more than a century later, we are discovering that the generation of electricity
through the combustion of fossil fuel is leading the world to global warming,
threatening the very existence of the infrastructure of modern society and the habitat
of animals and marine life. Similarly, through much research and development,
engineers and scientists designed nuclear power plants to replace the fossil-fuel
662 N. P. Suh

power plants to generate electricity without generating greenhouse gases. However,

after accidents at a few of these nuclear power plants, people are abandoning this
powerful technology for fear of new nuclear accidents and contamination, taking
actions that will accelerate global warming.
These examples illustrate how technologies must continue to evolve to seek the
next level of improvement and innovations to satisfy human needs and overcome
problems created by past deeds of human beings. The consequence of technological
advancement may generate new design problems that require unique solutions. In a
larger scheme of things, these by-products of technological progress will always be
there. It is a fundamental nature of systems, that is,

Old system þ design innovation ! new system þ new by-products ð23:1Þ

Equation (23.1) states that there are “new by-products” embedded in techno-
logical solutions (i.e., global warming with industrialization and invention of
internal engines). Some of these by-products cannot be ignored over a long period,
thus creating the need to design a new solution that can overcome the problem
created while solving an old problem. Equation (23.1) can be used to achieve many
positive results if humanity can deal with undesired by-products, if any, success-
fully and well. The conversion process indicated by Eq. (23.1) generates economic
activity, creates jobs, and continues to advance knowledge in all spheres of human
activity. The shortcomings of horse-drawn buggies created cars with internal
combustion engines, which have served humanity well for over 100 years. Still, a
new problem in the form of global warming emerged, forcing the creation of new
alternative transportation systems. This continuing evolution of human society is
part of the natural process.
So far, humanity has managed the process of evolution indicated by Eq. (23.1)
well, mostly through the design of the next advanced solutions as well as through
finding new solutions. Economic and financial investments in education, research,
and development enabled the continuing evolution and advancement of human
society since the Industrial Revolution, albeit a few calamitous blow-ups among
people. For this process to go on, we need the enlightenment of the large fraction of
people who support the dynamics of successful human, social, and economic
progress. When this process fails, humanity frequently discovered themselves in
ugly conflicts such as the First and Second World Wars. Humanity is often skirting
the boundaries of these disasters, a symptom of a poorly designed socioeconomic–
political system.
In this chapter, a few of the major problems that may require innovations,
solutions, and design will be discussed as examples of future design tasks that may
require the attention of the next generations of designers, leaders, and the society at
large. Predicting future events and functions is not an easy task—with many risks of
being wrong—perhaps best left to those with psychic minds. In the following
sections, randomly selected topics are discussed with no assurance that they will
indeed be the most important topics. For instance, if nuclear war engulfs the world,
23 Future Design Challenges 663

many things will change with irrecoverable dramatic ends of the world, which is
hard to imagine at this time. In that sense, we should do everything we can to
design a system that will prevent events that may culminate in such a disaster.

23.2 A Challenging Design Problem: Creating Forest

in North Africa

Living beings such as animals and vegetation cannot survive in the absence of
water, as well known. The availability (or scarcity) of water has determined the
politics and conflicts among nations, especially in the Middle East. The control of
water of the Jordan River (River Jordan) that flows through Jordan, Syria, and Israel
is politically and historically significant. In China, there are five or more large
deserts. The Gobi Desert, which is between China and Mongolia, is a vast and arid
region. The sandstorm from the Gobi desert reaches Korea, Japan, and parts of the
northern American continent every spring. People in Korea wear masks, and
sometimes, schools close during the peak season of the sandstorm. In North Africa,
the vast land area is mostly desert with limited human habitation. The world’s
largest Sahara desert covers about one-third of the African continent (about 3.5
million square miles) from the Red Sea on the west and the Nile River on the east
(extending about 3,000 miles) and from the Atlas Mountains of the south to the
Mediterranean sea of the north (about 1,000 miles).
Many theories have been advanced to explain how the deserts in North Africa
were created. It appears that the changes in the weather pattern created the North
African desert some seven million years ago. One theory is that when the North and
the South poles became extremely cold, the moisture condensed at these two poles,
thus depriving the moisture condensation in the hot regions near the Equator of
Earth, creating the vast desert of North Africa. There is also a theory that when
people and animals overglaze a particular area, the region becomes a desert,
especially in a hot climate. For example, the urbanization of the suburb near Cairo
and the Nile River created a small desert in the adjacent area between Cairo and the
Nile River. The population density of these desert areas is very low since the
climate is not very hospitable to people and other living beings, including
vegetation.
Some of the deserts in the world became habitable through massive irrigation.
The southern part of the State of California became thriving urban centers of the
United States with extensive agricultural business and high-tech industries thanks to
the water diverted from the Colorado River. Many other countries such as Israel,
Saudi Arabia, and the United Arab Emirates converted desert into valuable urban
centers by securing freshwater either through de-salination or divergence of rivers.
There is also a “theory,” cum speculation that the presence of green plants and
vegetation near the shore areas changes the weather pattern that brings in moisture
to the area, which condensate and rain in the region with forest and vegetation.
Assuming that these observations or theories are correct, can we design a solution
664 N. P. Suh

to convert some parts of the desert areas into habitable land for plants, animals, and
people? The first requirement for converting the desert into inhabitable land is to
bring more water to the region. In many areas, the de-salination processes, such as
the evaporation process or a reverse osmosis (RO) process, are used to generate
freshwater. The evaporation process and RO processes of producing freshwater are
energy-intensive processes, which are the major processes used for de-salination in
many countries in the Middle East, including Saudi Arabia and all neighboring
countries. These energy-intensive processes are expensive to operate. We should
design a more energy-efficient de-salination process by making use of solar energy
or wind power.
The following story illustrates how the desert can be converted into a modern
university–city complex with lots of trees and vegetation when sufficient financial
resources are available. A design challenge is how a similar transformation of the
desert can be achieved at a lower cost of construction and maintenance, using less
energy.
The KAUST Story:
About 40 miles north of Jeddah, which is a major city of Saudi Arabia on the Red Sea, there
used to be a small town called Thuwal. Its western border is the Red Sea, and the desert
surrounds it on the other three sides. Thuwal is also about 87 miles south of Mecca, the
holiest city of Muslims—the birthplace of the Prophet Muhammad and the Muslim faith.
Thuwal used to be a small fishing village until about 2009. In mere a decade, Thuwal has
become a well-known city in the world because it became the home for King Abdulah
University of Science and Technology (KAUST). King Abdulah of Saudi Arabia, a
visionary and benevolent king, established the university because he wanted to see Saudi
Arabia regain its eminence in mathematics and science.
KAUST was established in 2009 at a location that used to be a vast desert. Now its campus
consists of 3,600 hectares (8,900 acres) of land with more than 16,000 palm trees, 17,000
other trees, modern research buildings with advanced instruments and equipment, academic
buildings, a fire station, apartments for students, faculty housing, movie theatre, restaurants,
etc. Its graduate students do not pay any tuition, receives a stipend for living expenses, and
have access to some of the best-equipped laboratories. The student body is co-educational,
international, and highly selective. The goal of KAUST is to become one of the best
university of science and technology in the world. To achieve this goal, it has attracted
leading researchers and students from many parts of the world.
KAUST is a private university with the second-largest endowment in the world, next to
Harvard University. KAUST is a relatively small graduate school of science and technology
with approximately 1,200 graduate students and about 200 faculty in 2020. It may double
its size within ten years. Its physical facilities are most impressive. About two-thirds of the
students and faculty, both men and women, are from outside of Saudi Arabia. There are
many Saudi women students. KAUST is truly co-educational free from many restrictions of
Saudi Arabia. In many ways, it is comparable to the leading universities in advanced
nations on other continents. About half of the trustees of the university are non-Saudis.
KAUST survives and thrives because of the electric energy supplied by electric power
plants that burn oil in Saudi Arabia. All the buildings are air-conditioned with the electrical
energy provided by electric powerplants that combust oil. To generate freshwater, KAUST
pumps water in from the Red Sea to its de-salination plants on the campus. The de-salinated
water is used throughout the entire KAUST community. The water that keeps all the trees
23 Future Design Challenges 665

and vegetation alive is the re-circulated residual water after its primary use. Its annual water
consumption varied from a high of 9 million cubic meters in 2015 to about six million cubic
meters in 2019, decreasing every year since 2015. The electric power required for
de-salination varied from 51 million kilowatt-hours (kWh) in 2014 to 33.5 million kWh in
2019. The beautifully landscaped green area of KAUST is 1.1 million square meters (m2).
The total number of people that work and live in the KAUST campus is around 7,200. The
total power consumption in 2015 was 530 million kilowatt-hours (kWh), decreasing to 509
million kWh. The electric power used at KAUST for de-salination was from 9.6% in 2015
to 6.5% in 2019 of the total electrical energy consumed. On per capita basis, the total power
consumption at KAUST was about 73,000 kWh per capita and for de-salination about
4,000 kWh per capita. To put these numbers in a proper context, the electricity con-
sumption per capita of various countries in kWh are as follows: Saudi Arabia (9,400), U.S.
(13,000), Sweden (13,000), Korea (10,400), China (3,900), and Norway (23,000). A con-
clusion from this rough analysis is that to solve the water problem in the North African
desert (or any other dessert), we cannot only use the energy-intensive de-salination pro-
cesses such as evaporation or reverse osmosis (RO) process. These high energy-intensive
processes are needed when the rate of de-salination is high. However, in many situations,
we may not need such a high rate of de-salination. Thus, we may have the possibility to
design a new system of de-salination.
Saudi Arabia is exceptionally fortunate to be sitting on a vast reservoir of clean oil with the
least amount sulfur, etc., below the sand, which was discovered in 1938. Its production cost
is the lowest in the world, being as low as $3 a barrel of oil, which is only about 7% of the
cost of producing oil by fracking of rocks in the United States. The amount of oil reserves
in Saudi Arabia is difficult to estimate because of the yet-to-be-discovered fields. It appears
to be huge. Just a the Stone Age did not end because the world ran out of stone (a statement
made by a previous Oil Minister of KSA), Saudi Arabia may not be able to sustain its
economy based on oil. Saudi Arabia will not be running out of oil any time soon. Still,
because of global warming, the oil industry may, in the future, have a limited market
because the world cannot depend on oil as their primary energy source. Saudi Arabia is well
aware of this situation and is planning to change its economic structure to avoid being
entirely dependent on the petroleum industry because of limited global demand for oil as
the renewable energy sources take over the energy market. They are trying to create new
economic engines that can create jobs for a rapidly growing population of the Kingdom, a
goal of their 2030 economic plan.
The amazing thing about the modern world’s economy is that the value of natural resources
is minuscule in comparison to the value-added created by ideas that do not depend on
natural resources. For example, the vast annual revenue of the giant oil company, Saudi
Aramco, is much less than those of four companies in the United States, i.e., Microsoft,
Amazon, Facebook, and Google, that does not depend on natural resources for their rev-
enue. Thus, the need for the Kingdom of Saudi Arabia to restructure its economy is quite
apparent, i.e., to go from a nation that depends on natural resources in the ground to an
economy that is vastly different from the petroleum-based economy by finding new eco-
nomic engines.
In the twenty-first century, the irony for Saudi Arabia is facing is the following: it is one of
the most energy-rich countries in the world, not because of the oil under the ground, but
rather the immense solar energy it gets from the Sun every day of the year! Yet, they are
unable to tap into this vast energy source and capitalize on it. The following design task is
formulated to solve this problem.
666 N. P. Suh

The Design Problem for the Future Well-Being of the People Throughout the World:
Suppose you are a bright graduate student at KAUST, doing design research under the
guidance of a globally renowned KAUST professor Sandra Cavique-Foley. The design task
is to design de-salination plants to establish “forest” in Saudi Arabia and Egypt along the
coast of the Red Sea, using the uninterrupted supply of solar energy (except at night), the
availability of the water of the Red Sea, and vast desert area in these two countries.
If the water from the Red Sea can be de-salinated inexpensively using solar energy, it can
be used to grow plants in what is now desert. When a forest exists through this process,
some claim that the weather may change in these areas because the trees will transform the
weather pattern, generating additional condensation. The moist air will condensate, create
more rain, and eventually make the current desert area into livable and arable land with
trees and other vegetation.
State your FRs and develop ideas that can fulfill this dream. Then state the DPs that can
satisfy the FRs. What would be your process variables (PVs)?
As you embark on this design assignment, you should remember that, in general, slow
processes that require low levels of energy require larger equipment, which requires more
capital investment. This is the case when we try to make use of the tidal wave of the ocean
to generate electric power. This argument also applies to the case of electricity generation
with windmills. It requires large long blades to be economical. One has to consider the
financial viability of your idea as you decide on FRs, design parameters (DPs), and PVs.

23.3 Design Challenges Related to Renewable Energy

1. Issues associated with solar cells

a. One big problem with the use of solar cells is that the output of electricity
cannot be easily controlled or modulated quickly once installed. The output
depends primarily on the availability of sunlight, which cannot be regulated.
Thus, when the electricity generation by solar cells exceeds consumption, we
must store it for future use or dissipate it, neither of which is a trivial task.
Some countries may find it cheaper to give it away to their neighboring states
if they can use the electricity. We need to design solutions for storing excess
energy.
b. Various electric energy storage schemes have been considered. One idea that
has been promoted is to compress CO2 gas in a closed cavity such as
underground salt mine and generate electricity by decompressing CO2 when
electric power is needed. An idea of storing energy that is in use today is to
pump water to a reservoir at a higher elevation, e.g., mountains. Then use the
power of the waterfall by running a hydropower plant when needed. If such
clean electrical energy is abundantly available, we can create hydrogen and
oxygen through electrolysis, which can be used to power automobiles
without creating CO2.
23 Future Design Challenges 667

c. Another solution is to limit the power generation by solar cells to be always

below the demand level. When the demand exceeds the supply, supple-
mentary electric power generation by other means can be used, such as
conventional fossil fuel-burning power plants or through activation of
nuclear power plants. One of the design issues is the creation of control
system that can deal with a continuously varying demand and supply system.
d. The above-mentioned electric power transmission efficiency has other
implications. If the electrical power can be transmitted less expensively, then
electricity can be generated in hydrocarbon-rich and less densely populated
areas. Then the electricity can be transmitted to highly populated urban
regions. This transmission of power rather than crude oil will eliminate the
oil consumption associated with transportation, which should help in CO2
emission by ships. This transmission of electric power, especially wireless
transmission, is a challenging issue in design.
e. One current popular solution is to put solar cells on top of the roofs of
individual houses, which may alleviate the problems related to electricity
transmission in addition to creating a distributed power generation system.
Still, the electricity generated from the rooftops of these houses may not be
sufficient to supply reliable electric power for industrial operations and
transportation.

2. Wind Power

a. Wind power is another source of renewable electric energy. Windmills have

been installed in many countries, e.g., Denmark, the United States, and many
others. The power generation is proportional to the square of the blade
length, requiring tall towers for installation.
b. The availability of wind power is often highly location-specific. In some
countries, they are installed offshore or on top of mountains away from
densely populated areas for noise reduction and safety. It has to be located
away from urban areas to prevent the disastrous consequence of accidents
due to fatigue of materials and noise. In many countries, they are often
installed offshore, which solves many problems if the initial capital cost of
installation can be justified and is reasonable.
c. Wind power has the same issue as solar power in terms of the storage of
excess electric energy.

3. Nuclear Power

a. The nuclear power plant supplies a significant fraction of electricity in many

countries, e.g., France, South Korea, Japan, and China. In many countries,
they have operated these power plants for decades without any major
accidents.
b. The biggest threat to the use of nuclear power is the fear factor. The past
accidents that occurred, for example, in the Soviet Union as well as in Japan,
668 N. P. Suh

were truly horrendous. There are indications that the accident in Fukushima,
Japan, was due to the coupling of FRs at the system level. Building safe
electric power plants based on fusion is still a dream that requires creative
design for containing energy that is at extremely high temperatures.
c. The design challenge is to develop 100% accident-free nuclear power plants.
This means that the design must be thoroughly reviewed for any coupling at
the system level, i.e., in physical layout, software design, and operational
level.
d. Some countries have considered building small modular nuclear reactors that
are safer than large plants.
e. A more serious challenge in the use of nuclear power plants for electricity
generation is the disposal of the spent nuclear fuel, which must be stored
forever with 100% assurance. One way is to send the spent fuel toward the
Sun, which will be a challenging design task.

23.4 Design Problems Related to Transportation

Transportation of goods and people is both a necessity for economic activities and
fulfills the basic human desire to expand the sphere of human activities and
interactions. As the population continues to grow, both the issues and needs related
to transportation will continue to expand. Unfortunately, one major villain for
causing global warming is also the transportation of goods and people; the most
guilty one being automobiles, followed by airplanes and ocean-going vessels,
because ground transportation is the dominant mode of transportation of goods and
people. Roughly 30% of the CO2 emission is from the internal combustion engines
of automobiles and airplanes, another 30% coming from electricity generation.
Currently, the primary emission that concerns regulators is that of cars. In both
cases, alternative power sources consist of batteries, electric motors, and engines.
The most often-cited solution is to use lithium batteries that supply electric
power to electric motors. There are three problems associated with this solution: the
cost, weight, and safety of batteries. The weight of batteries can be as much as 30%
of the total vehicle weight. The other potential issue is the possible explosion of the
lithium batteries. As the number of accidents increases, the explosion of these
batteries during a collision can have a disastrous effect, especially if water seeps
into these batteries. These batteries contain both the oxidizer and the fuel, which
can be activated by water.
A successful alternative design is the use of wireless electric power transmission
to propel the vehicle. At the Korea Advanced Institute of Science and Technology
(KAIST), a university specializing in science and technology, a new of kind electric
vehicle was invented. It receives electric power wirelessly from power supply
systems embedded under the pavement and stores some of the energy in the
on-board batteries to use it to propel vehicles on roads without the embedded power
supply system. In 2019, the “On-Line Electric Vehicle (OLEV)” was running in
23 Future Design Challenges 669

five cities in Korea including at KAIST. The number of batteries on board is small
since the bus receives most of its electric power from the electrical power supply
system embedded in the underground. The size of the battery depends on the length
of the underground power supply system and the speed of the vehicle. The OLEV
vehicles are quiet, and the driver does not have to worry about the supply of fuel.
The OLEV system has been running in Korea since 2012, very reliably and quietly
without any significant issues. It may soon be installed outside of Korea. When the
electricity is generated from renewable sources and nuclear power plants, OLEV is
entirely emission-free. Its installation is simple and straight forward, especially
when installing new roads.
Also, there are possibilities of reducing fuel consumption of jet airplanes through
several concurrent means. The problem of current airplane design is that it has to
have powerful engines to propel the airplane to “lift-off speeds, about 160 miles an
hour” on a fixed-length runway. However, when the airplane reaches the cruising
altitude, the engine is too big and generates aerodynamic drag, which in turn
consumes more fuel. In other words, the need for “take-off” from ground deter-
mines the engine size, which at the cruising speed, is too big and adds to fuel
inefficiency.
There are several possibilities of reducing engine size. One is to reduce the
mixing section of the engine by coming up with faster means of mixing fuel with
oxygen, such as impingement of fuel and oxidizer rather than depending on the
slow diffusion process. Another method of reducing the engine size is by assisting
the airplane in reaching the lift-off speed by providing an externally assisted
take-off force. For instance, an external “train” can pull the plane forward to help
the acceleration of the airplane, similar to the way fighter planes take off from the
deck of naval carrier ships. The simplest solution is to extend the length of the
runways to enable aircraft to reach the take-off speed without having large engines.
All of these possibilities offer exciting design tasks.

23.5 Design Problems Related to CO2 and Methane Gas

(CH4) Emission

There have been many proposals for dealing with CO2. The most commonly dis-
cussed solution is to bury CO2 in-ground, dissolve in the ocean, or pump liquid
CO2 into ground or under the ground floor of deep sea. Another possibility is to
convert CO2 to other more useful materials. The difficulty is developing tech-
nologies that are economical and use less energy to transform. Becasuse CO2 is one
of the most stable forms of carbon products, it isn't straightforward to develop
inexpensive and environmentally acceptable conversion technologies. A research
center established at KAIST under the joint sponsorship of KAIST and Saudi
Aramco has been working on this problem for the past several years.
670 N. P. Suh

Methane (CH4) is another environmentally harmful gas. The primary sources of

CH4 are agricultural processes, degrading plants, and from cracking for oil. The
quantity of CH4 emission is about 9%, which is less than the emission of CO2. Still,
its environmental impact in the atmosphere can be substantial relative to its small
quantity. The diffused sources of CH4 gas are harder to control than the CO2
emission from internal combustion engines and electric power plants. However, as
the CO2 emission is reduced, the discharge of CH4 will receive more attention.
The free energy of CH4 is high, and therefore, it should be easier to convert it
into other useful products, the most stable reaction products of CH4 being is CO2
and H2O. Thus, CH4 can more easily be transformed into other useful materials
because CH4 can be reacted with other substances more efficiently than CO2.

23.6 Future Roles of Artificial Intelligence

(AI) and Quantum Computing in Design

Axiomatic Design (AD) teaches how to create an original design from scratch
correctly. Its axioms and theorems also explain how to find design flaws. This
process is going to benefit if the database on past designs, constitutive relationships,
various FR-DP-PV relationships, and past failures were readily available.
A promising approach is to generate plausible concepts by going through vast
databases to mine the data in the literature, hoping to find the right design solution.
This approach makes use of three development: the ability of computers to store a
vast database, process an extensive database rapidly at low cost, and the recent
advances in AI with the potential to extract the desired information quickly from the
vast database. A promising approach is to generate plausible concepts by going
through vast databases to mine the data in the literature, hoping to find the right
design solution. This approach makes use of three development: the ability of
computers to store a vast database, process an extensive database rapidly at low
cost, and the recent advances in AI with the potential to extract the desired infor-
mation quickly from the vast database.
AI can be used in many designed systems to make better decisions expeditiously
and to improve the effectiveness of the decision-making process. This use of AI is
possible because of the availability of high-speed computers, the vast memory
space available at low cost, and the development of sensors that can monitor the
behavior of the system. Already commercial software systems that incorporate AI
are being introduced. Such development will need a supervisory AI system that can
check for the correctness and accuracy of the AI decisions made.
Like many new technologies, one of the dangers is that AI may be used for
unsavory purposes. It can be used to cheat in elections for public offices, oppress
less fortunate people, and create systems that favor those with means at the expense
of democracy. If the history of technology provides a lesson, humanity eventually
finds positive and beneficial applications of technology. However, people were
concerned about the possible negative impact of new technologies on human
23 Future Design Challenges 671

beings, society, and individuals. So far, problems created by new technologies have
been solved through technologies, often advancing societal goals in addition to
technological advances. As long as we do not let technologies dominate human
decisions, we may be in good stead.
Many software and computer companies are active in these fields. Broadcom of
the United States (stock listing: AVGO) launched Automation.ai, an AI-based
software platform for supporting decision-making processes across different
industries. It deals with large volumes of data, which is challenging to do by digital
transformation, which can lead to slower decision-making. According to Broadcom,
Automation.ai is a platform designed to ease complications stemming from the
interference of diverse tools and data, and thereby facilitate informed
decision-making. It correlates and examines data to another software called Digital
BizOps to analyze the data and combine them to generate solutions to aid
decision-making. According to the company’s brochure, the technique harnesses
the power of machine learning, intelligent automation, and internet-scale
open-source frameworks to transform data. It is not clear if Automation.ai can
develop design solutions based on an original set of FRs.
Many other companies, such as IBM, have also been exploiting AI and quantum
computing, using their ability to store and manipulate vast data to improve
decision-making through cloud computing. IBM has been one of the first industrial
firms to work on quantum computing from many decades ago. Many of these tools,
including those of IBM and Broadcom, are based on their idea that there are vast
data available in the literature and companies that can be exploited and recombined
to create solutions to existing or new problems. The idea is to do what people can
do faster, utilizing the capability of machines to gather and sort out the data quicker
and more extensively. Ultimately, the hope is that the machines will also become
smarter than people in creating and designing new solutions. If this can be done, the
same thing will happen to scientists and engineers that happened to factory workers
due to automation using robotic technologies. Then, engineers and scientists must
work on a higher intellectual platform to do things that machines cannot do, which
will turn out to be a challenge. Computers are getting smarter faster than perhaps an
average person!
Some people are concerned that AI, the especially advanced form of AI, will
overtake the human ability to solve problems and create new solutions. In other
words, machine intelligence will be superior to human intelligence. That has been
the case for decades. For some tasks, machines have done a better job than human
beings ever since the Industrial Revolution. That is why we use many different
kinds of devices and tools. Now the computer can perform better than human
beings in some logical reasoning fields. This trend will continue. However, as
machine intelligence improves, so will human intelligence, often faster than
machines.
672 N. P. Suh

23.7 Design and Large Databases

The collection and management of the vast data that exist throughout the world is a
significant challenge. Cloud computing has enabled the creation and management
of an extensive database. Major corporations have jumped into the business of
managing the database and extracting the desired information. For most people, it
will soon be reasonable to assume that anyone can obtain personal information and
manipulate them by companies, individuals, and the government using facial
recognition. Regulating these activities will be extremely difficult. Individuals
should assume that it will be challenging to maintain personal secrets confidential
because the data have been automatically collected and stored through various
means by the government, merchants, schools, hospitals, and the like. Ultimately,
the government must regulate the use of these personal data, because no individual
can manage, protect, and restrict the use of personal data collected by machines and
organizations of various kinds.
The irony of the situation discussed above is that eventually, machines may
know more about us than we know ourselves! Under that circumstance, each
morning, we should consult the computer what we should do that day. Similarly, in
the evening, we should ask the machine to review what we have done during the
daytime and assess our performance and effectiveness, hoping that we can improve
our performance and efficiency the following day!

23.8 Design Issues Related to De-Salination

Today de-salination has become big business. Most de-salination processes in use
today are done using either reverse osmosis (RO) processes or various evaporation
processes. These processes are energy-intensive since these processes break the
atomic bonds between sodium chloride molecules and water molecules, both of
which are energy-intensive. Therefore, these are high-cost processes.
There are other possibilities for de-salination, such as de-salination by phase
separation. One of the techniques developed at MIT is to apply an electric potential
across the flowing brine water to create two streams: one salt-rich stream and the
other with low salt content. Then, the stream with high salt contents was separated
continuously from the lower concentration stream by draining off to a sink. By
continuing this process, what flows to the end of the stream is the water with low
salt concentration. The methods such as this need new design solutions.
Another possibility investigated at MIT and KFUPM is the use of graphene to
separate water molecules from other molecules. Making the graphene sheets
without flaws such as large holes is a challenging task. However, it is a concep-
tually promising means of making atomic-scale filters.
There may be other effective designs that are economical and reliable.
23 Future Design Challenges 673

23.9 Design of Software

Software is embedded in almost every product, in addition to substantial central

computing facilities that are maintained by all major corporations, universities, and
government agencies. Many software system developers often start coding their
software without first designing the software system, which may result in coupled
systems, requiring extensive revisions and testing.
One of the problems many software developers are facing is that they often build
new software systems on existing legacy software codes and policies. It isn't very
easy to know the intention of the designer of the original software unless they have
made thorough documentation on the system.
When large software systems are being developed, the design matrix must be
concurrently constructed to be sure that there is no coupling of FRs in the software
system. The current practice is to develop software codes and test before designing
the software system.

23.10 Control of the Weather Pattern

If we can control weather patterns, we can change the world for the better. Then, we
can improve many things such as agriculture, quality of life, prevention of forest
fire, conversion of dessert to useful land, low consumption of energy to heat houses
and factories, and many other beneficial things. Now unstable weather patterns are
causing significant problems in many parts of the world. For instance, a hurricane
that begins off the east coast of northern Africa due to high water temperature and
wind patterns creates vortices above the warm water. The vortices move toward the
southeastern area of the United States, where the temperature is lower, creating
hurricanes in the fall of every year and causing human tragedy as well as a lot of
economic losses. With global warming, the problem is going to get worse, more
tropical storms, the creation of more dessert, higher water levels, and unpredictable
weather patterns.
Hurricanes are caused by instability. When hot air forms a vortex motion on top
of the surface of the warm ocean water, it sucks up the water vapor, which
strengthens the vortex motion as it continues to move on top of the warm water.
This vortex motion generates strong wind as it grows. When the vortex moves
toward the colder surface of the land and the moisture condenses, it can unleash
rains with strong wind on the cold land surface. The damage done by these hur-
ricanes, wind, rain, surging ocean waves causes tremendous damage every year.
On the other hand, the weather pattern in Saudi Arabia and northern Africa is the
opposite. The whole region is hot throughout the year, and the Red Sea is relatively
small to develop unstable air-motion and create vortices. Therefore, the hot vapor of
the sea cannot be picked up and dumped on the top of the tropical land. Thus, the
region around Saudi Arabia is arid and became a dessert.
674 N. P. Suh

The design question is the following: Can we create and control vortices using
the instability phenomena of the circular motion of atmosphere by creating a large
number of small vortices artificially at different parts of the ocean surfaces to
control the weather pattern of Earth. If we can do that, we can change Earth to make
it more habitable, useful, and productive. The author of this chapter believes it can
and should be done.

23.11 Design of a Better Educational System

Human beings are created equal. However, the quality of education often ruptures
this “god-endowed” equality. The secondary variables such as family background,
financial resources, family stability, neighborhood, quality of education, and early
childhood education have deterministic effects on the well-being of the individual
as well as their society. Often, education has created a demarcation line between
those who will do well in society and those who would have problems in a
merit-based society. Therefore, parents in all countries are concerned about the
education of their children.
Similarly, the future development of a nation depends on the quality and
effectiveness of its educational system. Yet the educational system in many
countries is not effective and efficient for a variety of different reasons. In fact, in
many countries, educational systems are the primary reason for the lack of
advancement in the country. Ultimately, the educational system of most countries is
determined by design. In most nations, the emphasis is on teaching rather than
learning, i.e., teaching methods, teaching materials, teaching techniques, and less
on learning methods, learning efficiency, etc.
Many of these nations should redesign their educational systems. It should be
designed to bring out the best in their young people, must be rational, and must be
merit-based. It must emphasize ethical behavior, the concept of equality, the
importance of guarding justice, and doing one’s best in their profession. All these
features must be designed in the curriculum, institutional culture, and reward sys-
tems. The future of a nation depends primarily on its educational system, because
other socioeconomic-political factors are more difficult to change without
improving educational systems.

23.12 Design of a Democratic and Transparent

Government

Eventually, most nations will have a democratic form of government as a conse-

quence of significant advances in telecommunications, computers, and sensors that
can record most transactions and documents. It will be challenging to hide cor-
ruptive practices, illegal activities, dictatorial practices, and unsavory acts.
23 Future Design Challenges 675

Technologies will make it increasingly difficult to lie to the public for too long!
When technology is appropriately used, it will make society, politics, and gov-
ernment more transparent.
Recently, one of the most advanced democracies in the world has gone through
significant debates about the effectiveness, adequacy, and reliability of its election
system, which is the underlying lynchpin of democracy. Even after so many dec-
ades and centuries of maintaining effective democracy, people are finding that
modern technology can undermine the election system, especially when foreign
adversaries are interested in disrupting the system to bias the outcome in their favor.
In addition to the external interference, governments in many countries are tilted in
favor of those governing the nation. Such corrupt practices lead to both unfair and
undemocratic practices. We need an improved system for transparency in gov-
ernments, an effective governing structure, and a reliable election system to sustain
democracy. Truly democratic and fair practices can exist only with the citizens are
well educated, and laws and democratic principles rule governments. We will also
need continuous improvement in protecting the election system and in preventing
unlawful practices.
Democracy is a way of governing by following the majority opinion for a fixed
period and then reset the rule the group must adopt until the next period, i.e.,
functional periodicity. As Churchill said in 1947 in the House of Commons, “No
one pretends that democracy is perfect or all-wise. Indeed it has been said that
democracy is the worst form of government except for all those other forms that
have been tried from time to time….” It took about two thousand years from
Aristotle to the first English democracy. It changed the question from “who must
rule?” to “how to rule?” and “what to achieve?” To achieve this goal of a demo-
cratic nation, FRs must be established for the government and society through free
and democratic means based on fundamental governing rules and principles. FRs
are typically related to security, prosperity, welfare, health, aspiration, and freedom
of the people. Corrupting democracy leads to unpeaceful and conflicted societies.

23.13 Design of Improved Health Delivery Systems

The quality and cost of healthcare are two major issues that concern average
citizens in most countries. Gradually, every country is moving toward some form of
universal healthcare for all its citizens that can deliver the best care at the lowest
cost, which may be an oxymoron.
Many things must be done right to enact and maintain an affordable and effective
healthcare system. We need to design a healthcare system that has the following
qualities:

(a) High-quality hospitals with best trained medical staff, the right equipment, and
facilities,
(b) Efficient medical and pharmaceutical systems,
676 N. P. Suh

(c) Incentive system that rewards the medical skill, knowledge, efficiency and
dedicated work, and minimum bureaucracy.
Many countries have done much to improve their system, but none seem to satisfy
all these criteria.

23.14 Design of a Peaceful World

Almost everyone is seeking to live in a peaceful world. Yet, it has eluded

humankind for centuries. There are many reasons for it. Perhaps the most prominent
causes are human greed, a fundamental desire to protect one’s tribe and family,
even at the expense of others. Sometimes, national and regional interests, different
religious beliefs, limited resources to share, creeping dictatorship, and others tend to
override democratic tenents and a sense of justice. People have attempted to create a
peaceful and prosperous world through the establishment of the United Nations
(UN), the Worlds Health Organizations (WHO), the World Bank, and others. Their
effectiveness of these world organizations was checked by bureaucracy, lack of real
power to implement the best practices, limited incentives for good work, and
financial dependence on the member nations.
One significant danger is that many unstable people may have access to weapons
of mass destruction such as nuclear bombs, biochemical agents, disruption of the
electric power grid, and spreading of false information. We need to design means of
controlling and safeguarding these weapons of mass destruction.

23.15 Design of Better Drugs for Brains

We have many brain-related illnesses such as Alzheimer’s disease, autism, and

others. Yet, one of the least developed fields of medicine is those related to the
human brain. We have limited knowledge of the details of brain functions. Con-
sequently, we do not have a systematic means of diagnosis and fundamental
expertise in developing new pharmaceutical medicine that can deal with human
memory, enhanced brain functions, and the like. Can we design a system that can
assist brain functions through the use of embedded electronic microchips and the
like rather than depending on medication?

23.16 Design of Computer Assisted Brains (I.E.,

Supplementary Brain)

The brain is the most complicated and complex organ in human beings. The brain
performs various functions. Many of the workings of these functions of the brain
are yet to be fully understood.
23 Future Design Challenges 677

The brain retains information and analyzes them. It assembles and synthesizes
new information to create new ideas. Its logic circuits deduce conclusions from a
random set of data. Many other characteristics and functions of the brain we do not
yet fully understand. Some people retain and perform more of these capabilities
than others, e.g., Albert Einstein, Ludwig van Beethoven, Thomas Edison, Abra-
ham Lincoln, and others. We do not know whether these exceptional people had
more powerful brains or had used their brainpower more wisely and cleverly.
If we assume that human brainpower can be supplemented and complemented
through external means by attaching human-made nano-devices to human neuron
cells in the brain, we may be able to make “super-human beings.” The goal is not
only to provide more memory and logic capabilities to the brain but also to give
higher and faster reasoning and synthesis power to the person with this “supple-
mentary brainpower.” Then, human beings must make sure that the person with
such power does not become abusive to harm other people. Morality and ethics
must be built in such devices to be sure that we do not create a monster. Human
record on morality and ethics is not particularly reassuring.
Physiological functions of humans are affected by human thought processes,
although some people tend to minimize the effect. Human thought processes must
affect biological processes. The following true story supports this view:
Why Do People Sweat when They Eat Spicy Food?
In many countries, especially in the United States, the human palate has changed during the
past 50 years. More people eat spicy, hot food more than ever before. The meat and potato
culture of Americans has given away to spicy Mexican, Korean, Chinese, and Indian food
in major metropolitan areas of the United States. The massive immigration of people from
these countries has changed the palate of Americans gradually. Now the minorities in the
United States constitute a significant fraction of Americans, which is more multi-racial and
cross-cultural than ever before.
One of these hot dishes that have become very popular in the United States is “Kimchi,” a
Korean fermented cabbage, famous for its hot spice and garlic. Many people sweat when
they eat kimchi for the first time, especially if they have not had spicy food before. What is
interesting is that once people start eating this dish, they can no longer do without eating it
once in a while. Koreans eat kimchi almost every day with their main meals. Most people
have attributed their sweat when they eat kimchi to hot spices (primarily hot pepper, garlic,
salt) that somehow affected their biological and physiological systems.
One day an engineer who had researched mass production technologies was watching a
television program on mass production of kimchi in factories. He was fascinated to learn
that they were using a mass-production system similar to those used in making automobiles
to make kimchi. It was astonishing and interesting to learn that kimchi is no longer pro-
duced by housewives sitting around a table, which used to be the case it was made in
Korean households.
After watching the kimchi-manufacturing program for 20–30 min, he was surprised to
realize that he was sweating heavily. His shirts got wet, wiping away the sweat from his
face with paper towels, although he has not eaten anything spicy that morning! He called
his wife to see him sweating so heavily. She was surprised as well. This sweat was purely
psychosomatic. When learning of this experience, most people cannot quite believe what
they have heard. Some might have tried to replicate the experience.
678 N. P. Suh

It is abundantly clear that we have a lot to learn about the brain. What we know
and what we need to know are separated by a wall of missing information and
knowledge. Much investment will be required to close the gap. Future research
should seek a better balance between molecular-level scientific research and
hypothesis-based research to solve some of the urgent human sufferings such as
autism soon. These two different approaches will reinforce and complement each
other for faster advancement in this remaining frontier of human knowledge and
technologies.

23.17 Recycling of Materials

The world is rapidly becoming inundated with junk, toxic, and harmful materials.
When scientists first invented polymeric materials by creating long-chained mole-
cules, the resulting high molecular materials solved many problems that were
impeding the progress of societal functions. Only 70 years later, after the invention
of high molecular materials, the world is running out of space to store these
materials after using them because human beings consume so much of these
non-degradable materials. We need to design solutions for this waste disposal
problem. There may be several approaches: first, rapidly degrading these materials
to convert them to the original elements without harming nature. Second, recycle
waste into useful products. Third, levy enough taxes to finance recycling efforts.
Forth, forbid the use of some of these materials through international agreements
and legislation, and finally, incorporate, during the design stage, a life cycle sce-
nario to have a net-zero impact on the environment.

23.18 Design Problems Related to Self-Driving Cars

Many companies are developing self-driving automobiles. Their goals are admir-
able, but there will be many accidents and fatalities before a reliable system can be
installed unless they are correctly and adequately designed. Problems will arise in
the least expected situations, where machines are not pre-equipped to deal with for
lack of a database. Keeping everything else the same and only equipping cars with
sensors of many different kinds may be a wrong approach. We need to consider all
the issues of such a system and developing a systems solution involving “smart
cars,” intelligent roads, design of decision-making systems for a group of auto-
mobiles nearby, etc. We need to construct FRs, DPs, and the design matrix, con-
sidering the interaction of multiple vehicles, conflicting requirements of the vehicle
in the proximity, etc. The issues involved are not confined to the car, but the entire
system comprising a group of the vehicle with different goals and needs.
23 Future Design Challenges 679

23.19 Concluding Remarks

This chapter dealt with the challenging design problems of the future. The pre-
ceding chapters showed that if people can identify the problem, they can come up
with design solutions, following the steps outlined in this book. There may be
exceptions when the underlying science base is absent.
In many fields, humans have designed many incredible things. Various designs
and advanced technologies created since the Industrial Revolution attest to the
human ability to develop artifacts based on needs and goals. They invented things
to solve problems they identified to achieve specific goals. Chapter 1, through
Chap. 3, attempted to provide a logical structure for these designs to enhance,
expedite, and provide an intellectual framework for creating these designs. We
showed how to identify and define the problems, which were then transformed into
a set of FRs, followed by the development of DPs to satisfy the FRs. If some of the
early pioneers had known AD, they might have done a much better job of syn-
thesizing solutions more quickly! What people have achieved during the past four
centuries, at an ever increasingly faster rate, gives us confidence in what humans
can do to deal with future challenges.
The purpose of this last chapter is to speculate and consider what kinds of future
problems people will be called upon to solve through design. Some may be difficult
to address through design. However, what may be surprising to some is that the
basic AD methodology presented in this book will be equally applicable to all
future design problems. What would be equally remarkable to those learning design
for the first time is that the thought processes are similar, i.e., the same thinking
process and structure regardless of what we have to design. What was different in
designing the OLEV and the laminated coffee cup was the specific basic knowledge
involved in each field, such as physics of electromagnetic fields, materials behavior,
mechanics, thermodynamics, the need for relevant data, etc. That is why designers
need to learn the fundamentals of physics, chemistry, mathematics, and other
related subjects as much as possible. The same person who designed OLEV had
designed the coffee cup manufacturing systems and organizations in diverse fields
using the same process of design outlined in this book.
What the designer must know is the basic principles covered in this book and ask
questions to be able to define the design goals to proceed with the design task. One
can learn what one does not know to complete a design task. When one does not
have the necessary knowledge and cannot answer basic questions, then one should
acquire the required knowledge from appropriate books and colleagues, and pro-
ceed to design. The person who knows the “question to ask” can develop design
solutions! One should not be afraid of tackling a new problem, provided that we are
willing to acquire the missing knowledge to be able to ask field-specific questions
relevant to a given design task. As one executes more and more design tasks, it
becomes easier to be an ecumenical designer, who can function in many different
areas. It isn't straightforward to know everything from the beginning, but one can
always learn.
680 N. P. Suh

The preceding statements apply in many fields. The power of AD enables to

organize one's thought, quickly identify what needs to be done, identify the missing
knowledge, facilitate the quick acquisition of the tasks involved, and come up with
a conceptual design. When the design goals are clear, the designer can acquire the
necessary knowledge quickly. These steps are discussed and illustrated in the
preceding chapters. If one knows what one ought to know but does not have
detailed knowledge, the learning efficiency improves a great deal. Many students
spend a great deal of time to learn new subject matters. The effectiveness of
learning can be low when one does not know what the question is and what one
wants to achieve by learning the subject. To paraphrase it, “when people do not
know the question, they cannot find answers.” What AD enables the designer to do
is quick identification of missing knowledge when attempting to define the FRs,
DPs, and PVs.
There are two kinds of design tasks that are challenging and uniquely suited for
human designers. The first is the ability to define the design problem, for which
only human beings are uniquely qualified. The second is the ability to design
logically and rationally, following the steps outlined in this book. The cost of
creating something new based purely on experience and trial-and-error methods
may be too high with the uncertain and precarious outcome. (Question: Was the
failure of Boeing 737 MAX due to a coupling of FRs?) Creative solutions often
follow the identification of the most important and critical problem, followed by the
development of an uncoupled design.
Among the most critical problems, humanity must solve in the early twenty-first
century through design is global warming. Yet, the response of most nations to this
existential threat is so slow that we may not solve it in time. In comparison to what
many countries spend on defense, hardly any money is spent on global warming. If
all nations spend one-tenth of what they are spending on military defense to solve
the global warming problem, we may have a chance of preventing this impending
disaster before it is too late. The development of practical solutions to global
warming problems will require the human ability to design rationally, logically, and
without committing significant errors.
Another equally important issue that also requires “design thinking” is the
re-orientation of our thinking on “world peace” and “human prosperity and
advancement of human habitat.” Since the end of the Second World War in 1945,
many nations spent more financial resources to develop weapons and in defense
because of perceived external threats to their existence. Although many useful
technologies resulted from defense research (e.g., GPS, the Internet, new materials,
and many others), the world is yet to find peace. In retrospect, it is clear that if
leading nations had spent 50% of its defense-related resources for the “peace
dividend” for economic and educational development of developing countries, the
world today might be more secure, peaceful, productive, healthy, and more
prosperous.
Index

A deep learning, 589

Abduction, 612, 622 machine intelligence, 671
Academic machine learning, 589, 590, 592
administration, 355, 461, 468, 471 neural networks, 590
admission to leading colleges, 198 Ash tray design, 89
calendar, 582 Assembly, 57, 62, 79, 82, 92, 117–119, 129,
Acceptable range, 103, 137, 259 139, 140, 143, 159, 194, 276, 277,
Acclaro, 507, 587, 595 279, 280, 282, 466, 472, 481, 484,
Active redundancy, 293 487, 489–492, 500, 504, 521–524,
Adaptive 526–532, 534, 536, 537, 539, 540,
loose-coupling, 653 563, 571, 575, 587, 591, 645, 652
redundancy, 295, 296, 304 Atomic level, 246, 379, 382
Advanced Planning System (APS), 493 Augmented intelligence, 589
Aerospace Automobiles
airlines, 522 automatic driving system, 353
airplane control systems, 36 automobile cooling system, 386, 388–390
airplane/Jet, 522, 523, 527, 536 automobile suspension system, 390, 391,
Airport, 4, 31, 52, 173, 184, 328, 335, 368, 394
410, 566, 568, 573 collapsible steering column, 91, 242
fuselage, 521, 522, 524–531, 536 cost of steering column, 91
Aggregate, 174, 288, 374, 407, 560, 610, 611, Axiom
614, 615 independence, 2, 5, 13, 28, 35, 38, 41, 43,
Agile, 486, 487, 594, 598, 612, 619, 623, 626, 48–51, 55–57, 59, 62, 66–68, 72,
628, 637–639, 641, 652, 653 76–78, 82, 84, 85, 87, 90, 95, 106,
Alcohol, 440 110, 113, 115, 122, 178, 191, 192,
Allegory of the cave, 260 196, 201, 206, 209, 212–214,
Allocation of function to form, 371, 375, 376, 216–218, 227, 234–236, 238, 239,
378, 382, 401, 408 243, 244, 246, 256, 257, 278, 307,
Alternative redundancy, 295 310, 311, 315, 327, 329, 332,
Analytical path, 371 340–342, 347, 350, 351, 376, 377,
Analytic Hierarchy Process (AHP), 181 382, 390, 394, 404, 407, 408, 410,
Aneurysm, 547 434, 481, 484, 487, 556, 587, 597,
Apple 636, 639
apple orchard, 429 information, 2, 5, 28, 43, 48, 56, 62, 66, 68,
apple pie, baking an, 429 76, 77, 128, 191, 192, 205, 206, 216,
Arduino, 512 235, 236, 238, 274, 307, 310, 311,
Artificial Intelligence (AI) 323, 324, 327, 329–331, 339, 341,
artificial intelligence (AI) in axiomatic 342, 344–347, 433, 484, 487, 556,
design, 233 557, 597, 636, 639

© Springer Nature Switzerland AG 2021 681

N. P. Suh et al. (eds.), Design Engineering and Science,
https://doi.org/10.1007/978-3-030-49232-8
682 Index

Axiomatic design vs algorithmic design, 42 Common range, 311, 312, 316, 318, 320–324,
Axiomatic Maturity Diagram (AMD), 339, 342, 597
432, 437, 635, 639–642, 645, 653 Compatibility, 2, 62, 491, 605
Complexity
B combinatorial complexity, time-dependent,
Bankruptcy, 450, 606, 631 335, 336, 448, 473, 579
Bathroom furnishings combinatorial time-dependent, 335, 336,
bathtubs, 487 442, 448, 473, 579, 597
shower trays, 487, 491 Complex Adaptive Systems (CAS), 610,
Battle of Austerlitz, 294, 297 611, 647, 654
Battleship design, 73 complexity axiom, 327, 329, 330, 339, 340
Beverage can, 86, 87, 350 complex system, 122, 124, 328–330,
Big Data, 172, 175, 183, 187, 493, 588, 592, 332–334, 351, 367, 369–373, 378,
593, 610 408, 409, 428, 486, 550, 585, 586,
Biological 594–597, 599–601, 611
biologically inspired design, 177 complex vs complicated system, 333
cell, 563, 565, 568, 574, 584 imaginary time-independent, 331, 336
eyeball, 437 period time-dependent, 335, 336, 347, 448,
Biomedical 473, 545
drugs for brains, 676 real time-independent, 494
Braised lamb ribs, 433 theory of, 329
Brine, 429 Complicated, 25, 27, 36, 38, 40, 48, 65, 67, 71,
Broad knowledge base, 19, 21, 167, 192, 193, 78, 79, 84, 85, 142, 210, 217, 236,
202 243, 244, 251, 310, 329, 333, 334,
Brushing the back of teeth, 362 353, 369, 432, 453, 454, 481, 547,
561, 564, 577, 592, 624, 626, 627,
C 630, 642, 676
Cancer, 168, 178, 236, 544, 552, 556, 558 Compressor, 62, 141, 215, 379, 384, 385, 511
Carry-over cooking, 427 Computer-aided design (CAD), 121, 146, 277,
Cell/Cellular 318, 586, 587
biological, 563, 565, 568, 574, 584 Computer-aided engineering (CAE), 586, 587,
manufacturing, 526 591
Changability Computer-assisted brains, 676
changeable manufacturing system, 484, Conceptual design, 50, 148, 263, 264, 276,
487, 490 344, 396, 491, 587, 595, 600, 623,
Chaos, 52, 568, 582, 626 680
Chemical-Mechanical Polishing Conceptualization, 223
(CMP) machine, 200 Concurrent engineering, 263, 276–279, 282
Circadian cycle, 568, 572 Construction site, 579
Clock Containers, 26, 27, 199, 213, 394–398, 410,
function clock, 564, 569–572, 584 411, 514
OODA, 613, 627 Contextual information, 174, 184–186, 189
time clock, 563, 569–571 Control of the weather pattern, 673
Cloud Convertibility, 539
architecture, 607, 608, 632, 642, 652 Corollaries and theorems, 76
Cloud computing, 30, 236, 600, 605, 607–609, Coupling, 8, 11, 31, 38, 65, 66, 77, 78, 82, 84,
613, 618, 624, 626, 627, 631, 632, 85, 106, 115, 135, 145, 180, 191,
635, 643, 644, 646, 647, 650, 653, 193, 207, 210, 214, 216, 219, 236,
654, 671, 672 257, 258, 270, 273, 328, 343, 360,
Cold War, 583 388, 390, 394, 396, 398, 440, 458,
CO2level in atmosphere, 215 547, 550, 569, 605, 608, 610,
Collectively Exhaustive and Mutually 650–654, 668, 673, 680
Exclusive (CEME), 271, 274, 275 Creating forest in North Africa, 663
Index 683

Creation of Successful High Technology 288, 289, 298–302, 305, 310, 312,
Company, 112 314, 315, 317–319, 321, 322, 324,
Creativity 332, 376, 377, 382, 385, 389–391,
creative process, 3, 5, 18 480, 497, 498, 503, 516, 576, 587,
Criteria for selection of DPs, 212 627, 636, 650
CurcurPlate, 360, 361 definition of, 45, 181, 185
Customer designing a manufacturing process for a
customer attributes (CA), 12, 353, 473, 486 difficult-to-machine material, 247
customer need, 20, 31, 45, 128, 158, designing ecologically-minded food
167–169, 173–175, 182, 187–189, packaging, 418
241–244, 256, 263, 381, 390, 440, domains, 28, 129, 601
483, 486, 488, 495, 506, 508, 513, equation, 51, 52, 62, 63, 85, 121, 137, 227,
527, 528, 585, 596, 600 258, 261, 271–273, 278, 279, 286,
customer voice, 169, 171–175, 185, 189, 288–292, 296–298, 300, 303, 304,
352 315, 322, 332, 351, 353, 355, 356,
Customer-perceived design, 440 358, 365, 376, 529
Customer Relationship Management (CRM), for manufacturing, 276, 277
493 -it right, 78
Customization, 528, 539 machine, 67, 119, 235, 236, 586, 588, 595
Cyber-physical production systems (CPPS)”, matrix
484, 510–513, 515–517 master, 84
Cyber-physical Systems (CPS), 408–410, 493, methodology, 22, 110, 238, 423, 486, 521,
510, 514 522, 526, 527, 536, 540, 559, 679
Cynefin, 605, 612, 624–627, 630, 637, 641, of a government organization, 106, 244
642, 653 of a research university, 474
of an organization, 228, 447, 458
D of driverless car, 197
Database of government, 455
inn formulating FRs, 202 of government organization, 455
Data center, 286, 292 of high technology industrial firms, 111
De-carbonization, 302 of technological products, 85
Decision criteria, 263 of the NSF Engineering Directorate, 107,
Decision-making in an organization, 468 456
Decomposition process, 31, 120, 143, 149, of ultra-precision lithography machine, 104
152, 191, 192, 206–208, 215, 236, of water faucets, 49, 62, 64, 85
256, 263, 268, 278, 490, 528, 587, parameters (DPs), 2, 5, 10, 16, 19, 20, 22,
597 29, 32, 35, 39, 44, 45, 50, 52, 56, 62,
Deduction, 378, 612, 626 75, 80, 81, 86, 89, 97, 105, 118, 121,
Defense, 10, 23, 46, 60, 61, 107, 108, 202, 245, 169, 181, 182, 191, 192, 209, 213,
251, 454, 455, 583, 614, 630, 225, 241, 246, 250–252, 257, 280,
644–646, 654, 680 283, 310, 312, 315, 318, 322–324,
Definition of systems, 27 349, 350, 364, 370, 372–376, 379,
De-salination, 68, 663–666, 672 380, 398, 407, 408, 473, 483, 486,
Design 489, 522, 526, 528–530, 536, 540,
axioms, 3, 5, 28, 31, 43, 47, 48, 68, 78, 227, 544, 545, 548, 550, 552, 556, 557,
263, 277, 329, 473, 526 559, 560, 600, 609
constraints, 167–169, 181–185, 187–189, process, 3, 6, 16, 19, 28, 31, 32, 35, 46, 50,
545, 559 62, 67, 71–73, 91, 110, 113, 118,
context, 184–187, 189, 381 120, 121, 127, 128, 135, 140, 143,
coupled, 110 145–147, 152, 163, 188, 207, 209,
decoupled, 55, 59, 67, 78, 83, 93, 95, 101, 212, 215, 220, 221, 237, 238, 242,
113, 115, 122, 196, 201, 209, 210, 259, 262, 263, 265, 275–278, 282,
217–219, 233, 235, 239, 252, 259, 327, 330, 338, 339, 349, 379–381,
684 Index

385, 395, 400, 433, 479, 505, 543, Feedback, 123, 124, 246, 277, 377, 378, 384,
585–588, 594, 617, 622, 623, 628, 385, 409, 425, 595, 596, 600, 608,
636, 637, 639 611, 628
range and coupling, 115, 398 Feeling requirement, 419
representation purpose, 117–121, 126, 127, Finite Element Analysis (FEA), 79, 586
129, 131, 139, 141, 146, 158, 159, Flexibility, 55, 76, 244, 408, 473, 487–490,
263 522, 539, 608, 653
rule of thumb, 128, 372, 376 Floating keyboard, 170
science, 261, 367, 369, 408 Food
team, 65, 194, 260, 262, 265, 269–271, and drug administration (FDA), 548
273, 274, 276, 282, 288, 347, 370, braised lamb ribs, 433
372, 376, 377, 385, 396, 401, 547, butter, 436
550, 600 chicken soup, 427
theory, 2, 36, 41, 42, 181, 203, 265, 282, dough, 431
295, 324, 333, 398, 409, 410, 483, gluten, 440
518, 556, 628 hamburger, 421
Design Fields (DF), 4, 237, 409, 505 ingredients, 337
Design for Environment (DFE), 438 -related preparation, 418
Design Guidelines (DG), 506, 507, 510 sous-vide, 427
Diagnosability, 539 Forward-design, 376, 377, 401
Different solutions to the same design problem, Four domains of the design world, 11, 44, 45,
16 72, 106, 256, 307
Dispersion-strengthened copper alloy, 368 4-P method, 172
Diversity Fractal, 310, 344, 614, 619, 631, 649
diversity of internal and external Fraser MacGregor, 434
stakeholders, 369 Fully robust, 340, 346
Divide and conquer, 373 Functional
Domestic hot water, 290, 291 architecture, 374, 375, 378, 409
Drilling, 522, 523, 528–531 architecture domain, 369, 370, 372, 373
Dual hierarchy, 375, 378, 394, 407 domains, 11, 31, 32, 35, 39, 44–46, 50, 52,
56, 60, 65, 72, 106, 109, 114, 115,
E 167, 191, 192, 243, 251, 261,
Earth-friendly packaging, 438 264–266, 268, 274, 275, 280, 295,
Eastman Kodak Company, 196 317–320, 329, 334, 347, 554, 594,
Educational system, 582, 674 596, 597, 642
Electrical connector design, 224 hierarchy, 180, 181, 374, 408
Electronic Medical Record (EMR), 553, requirements (FRs), 2, 5, 6, 8, 10, 11, 16,
557–559 19, 20, 22, 23, 28–32, 35, 39, 42–46,
Emergency Room (ER), 24, 579, 626 49, 50, 52, 56, 62, 72, 76, 77, 86,
Energy 109, 118, 121, 124, 127, 167–169,
consumption, 295, 381, 484, 497, 500, 514 175–177, 179, 181, 182, 187–189,
storage, 199, 545, 666 191, 192, 209, 241, 251, 256, 257,
Enterprise Resource Planning (ERP), 493 260, 261, 311, 312, 314, 315, 318,
Ergonomic, 169 323, 324, 329, 349, 350, 352, 364,
Ethical, 8, 20, 205, 213, 229, 253, 264, 438, 369, 372–374, 376, 378, 381, 396,
450, 454, 546, 560, 674 406–409, 447, 470, 473, 483, 485,
Explicit FR, 23, 352 486, 489, 490, 495, 496, 501, 507,
513, 522, 529, 536, 540, 543–546,
F 548, 550, 554, 556, 559–561, 564,
Factors that affect our ability to choose a right 578, 579, 585, 596, 600, 609, 642
set of FRs, 204 Class FR, 404
Fan design implicit FR, 352, 360, 362, 364
bladeless, 355 instance FR, 407
Fasten, 528–531 tolerance, 268, 274, 280, 310, 323
Index 685

Functionality diagram, 327, 339, 341, 343, 544, 589, 599, 601, 619, 620, 627,
344, 346, 347 650, 661, 670, 678
Functional periodicity Importance of problem definition, 2, 22, 39,
function clock, 563, 564, 569–572, 584 167–172, 174–176, 180–182,
solar time clock, 563 184–189, 235, 307, 545, 621, 628
Function–behavior–structure ontology, 181 Induction, 103, 612, 623, 626, 637
Function-physical representations, 121 Industry 4.0 (I4.0), 187, 418, 484, 486, 491,
493, 504–506, 508–510, 512, 518,
G 585, 587, 594, 599, 600
Game-changing goals, 367, 368 Information content, 28, 43, 49, 66, 77, 78,
Game show, 441 115, 124–127, 137, 139, 206, 216,
Gaussian distribution, 65, 606, 614 255, 262, 263, 266, 269–271, 307,
Generation and selection of DPs, 218 309–315, 317, 318, 322–324, 331,
Geometric dimensioning and tolerancing, 279 481, 597, 609, 617, 630, 639
G{\'{\i }}sli Matt, 433 Innovation, 3, 4, 23, 38, 67, 109, 111, 169,
Global warming 170, 175, 188, 204, 227, 243, 248,
due to CO2 and CH4, 8 441, 458, 465–467, 472, 484, 585,
due to CO2and CH4, 669 587–589, 600, 626, 659, 662
Government organizations, 106, 244, 454, 455, Input constraints, 182, 183, 527, 528
457 Inspection, 276, 281, 282, 388, 591, 592
Graphical method, 317, 318, 324 Insurance, 221, 552, 558
Guidelines for choosing DPs and PVs, 236 Integrability, 539
Integration of DPs in a system, 221
H Integration team, 399
Healthcare, 7, 11, 25, 28, 61, 124, 163, 168, Interactions, 24, 57, 58, 121, 122, 147, 149,
221, 483, 543–560, 631, 675 152, 154, 174, 175, 183, 185, 228,
Health delivery systems, 675 329, 333, 372, 374, 375, 408, 440,
Heat exchanger system, 384, 385 486, 487, 502, 506, 512, 515, 549,
Herman against the others, 432 558, 586, 608, 652, 668, 678
Hetero-functional networks, 408, 409 Interdisciplinary background and choice of
Heterogeneity, 369, 544, 555, 557, 558 FRs, 206
High-energy imaging, diffraction and scattering Interfaces, 79, 80, 92, 124–126, 138, 149, 185,
beamline, 79 225, 249, 373, 396, 408, 496, 511,
Human intelligence, 589, 671 512, 514, 515, 539, 558, 588, 599,
645
I International Business Machines (IBM), 9, 21,
Icelandic delicacies 30, 67, 111, 168, 235, 454, 566,
blowtorch-sheared split sheep’s head, 135 607, 671
fermented Greenland shark, 437 Internet of Things (IoT), 175, 484, 486, 491,
fermented whale fat, 437 493, 613, 639, 645
Ideal design, 37, 55, 72, 77, 78, 82, 209, 212, Interoperability, 544, 555, 558, 559
216, 218, 225, 238, 278, 288, 289 Intrinsic redundancy, 295
Identification of DP Inverse problem, 29
based on material and mechanics, 429 IPod, 169
organizational design, 110, 227 ISO Standards
Identification of DP ISO 10218-1:2011, 502, 503
based on fluid mechanics, 231 ISO 10218-2:2011, 502
based on material and mechanics, 229 ISO 12100:2010, 502
Impact, 10, 91, 93, 94, 168, 262, 342, 345, 398, ISO 13849-1:2015, 502
410, 412, 413, 438, 456, 472, 488, ISO 13854:2017, 503
ISO 13855:2010, 502
686 Index

ISO Standards (cont.) Logistics and Supply chain Management

ISO 13857:2008, 503 (SCM), 592
ISO 14118:2017, 502 Long Tail model, 179
ISO 14120:2015, 502 Lukas Pfaff, 436
ISO 62046:2018, 502
ISO and ASME representations, 132 M
ISO TS 15066:2016, 502, 503 Maaike Dogan, 434
Machine or Production Data Acquisition
J (MDA/PDA), 493
James Watt, 3, 12, 13, 30, 38, 44 Machines that can design, 67
Jermain de Rozario, 436 Management, 7, 18, 65, 109, 111, 119, 146,
Julia Childs, 431 147, 150, 163, 169, 182, 187, 196,
197, 211, 214, 229, 244, 254, 257,
K 258, 262, 269–271, 276, 285, 287,
Kano Customer Satisfaction Model, 178 343, 351, 377, 381, 385, 400, 402,
K-cup, 442 450, 452–454, 457, 468, 472, 481,
Kidney stone, 87–89 483, 493, 498, 500, 506, 566, 592,
Kitchen, 188, 441, 487 593, 606, 608, 611, 612, 615, 627,
Kitchen Patrol (KP), 426 631, 633, 672
Knowledge Manufacturing
database, 592, 597, 599, 600 cell, 487, 526, 575
heterarchically hierarchical, 605, 612, 618 electric discharge machining (EDM), 577
knowledge-based expert systems (KBES), Henry Ford’s system for mass production,
593 466, 575
knowledge-based products, 605, 606 system design, 483, 485–487, 489, 575,
Korea Advanced Institute of Science and 592
Technology (KAIST), 13, 26, 66, systems, 7, 58, 211, 368, 407, 418,
100, 170, 193, 194, 395, 410, 668, 483–491, 500, 501, 505, 518, 526,
669 536, 563, 575, 576, 586, 594,
597–599, 679
L Manufacturing Execution System (MES), 493
Lack of regularities, 336, 337, 347 Mapping between the domains, 31, 72
Laplace demon, 287 Mass market, 620
Large Mass production of microcellular plastics, 248,
complex system, 367, 369–371, 373, 378, 249
408, 409, 417 Mathematical optimization and design, 82
data base, 672 Maturity diagram, 339, 635, 639, 640
fixed engineering system, 371, 373, 401, McDonalds, 421
407–409 Mechanical and Chemical Polishing Machine
flexible engineering system, 373, 406–410, (CMP), 142, 145, 200, 201, 207
559, 560 Mechanical drawing general rules, 131
large-sized, 358, 372 Mechanical product design
Laser from the design matrix to the mechanical
engraving, 511, 514–516 drawing, 129
Lead user theory, 173 Medical, 7, 25, 29, 30, 154, 168, 197, 265, 287,
Lean management, 484, 493, 505 342, 346, 364, 458, 544, 549–554,
Lean war room (Japanese: obeya), 493 556, 557, 559, 560, 621, 622, 626,
Learning factory, 484, 504–508 631, 675, 676
Legacy Medium-sized, 372, 472, 488
system, 370, 371, 544, 555, 559 Mental constructs, 370
Life cycle properties, 408 Mental-load, 375
Lifestyle, 60, 175, 350 Merger and consolidation, 466
Limitations of experience-based design, 4 Meta-design-parameters, 370
Limitations of purely experienced design, 599 Meta-problem-solving, 367, 369
Index 687

Minimum information content and the concept of systems, 20, 30, 66

of tolerances, 124, 126, 311, 323 Organizational design, 105, 107, 110, 228, 244,
Minimum number of FRs, 206, 258 448, 479
Mission to Moon, 198
Mitosis in biological cells, 565, 569, 574, 584 P
Mobile Pacemaker, 545, 546, 560, 561
app, 398, 496, 497, 499 Parallelogram, 318
harbor (MH), 26, 27, 66, 394–398, 401, Parameter value, 268, 269
410, 411, 465 Pareto efficiency, 179
Mobility, 96, 168, 398, 491, 512, 607 Passive redundancy, 290, 291, 293
Modularity, 491, 513, 539 Patent, 7, 41, 46, 61, 62, 214, 353, 413, 428,
Module, 118, 121–124, 126, 127, 131, 137, 465, 565
146, 147, 150, 157, 158, 355, 377, Peaceful world, 660, 676
385, 494–500, 529, 568–570, 572, Periodicity, 329, 336, 448, 452, 472, 473, 479,
617 563–569, 572–581, 598, 675
Module junction diagram and flow diagram, Periodic re-initialization of organization, 472
121 Pesticides, 438
Money circulation, 25, 60, 61, 66, 69 Phase transitions in biological cells
Mount type air conditioning system, 379, anaphase, 563
381–383, 385 interphase, 563, 574
Mutually Exclusive and Collectively metaphase, 563
Exhaustive (MECE), 351, 373, 374 prophase, 563
telophase, 563
N Photoresist, 200, 252, 567, 571
National Academy of Engineering (NAE), 187, Physical
367–371, 408, 409 architecture domain, 370, 372
National Laboratories in the United States hierarchy, 374–376, 407, 560
Argonne National Laboratory, 583 laws and thinking design machine (TDM),
Lawrence Livermore National Laboratory, 236
245, 583 Mock-Ups (PMU), 586
Los Alamos National Laboratory, 583 Physical hierarchy, 401
Oak Ridge National Laboratory, 583 Piping and Instrumentation Diagram (P&ID),
Sandia National Laboratory, 583 140
National Science Foundation (NSF), 107–111, Planning administration of admission exam,
194, 454–457, 583 355
Near Field Communications (NFC), 511 Plastic cups, 6, 211, 213, 439
NIST, 607–609, 645 Political, economic, social-cultural, and
Nominal dimension, 119, 133, 205, 211, 310 technological factors (PEST
Non-design, 262 Framework), 175
Non-FR, 642, 643, 649, 653, 654 Political systems
Not robust, 340 dictatorship, 579, 581
parliamentary, 579
O periodicity and re-initialization of, 579
On-line Electric Vehicle (OLEV), 13, 65, 66, presidential, 579
71, 103, 193, 194, 400, 402, 465, Polyhedron, 320, 322
547, 668, 679 Power law, 615, 618
On ranking of FRs, 206 Power take off/power take in PTO/PTI, 295
Ontology, 260, 285, 593, 597 Precision engineering, 245, 246
OODA loop, 605, 632 Presidential appointment, 106, 455
Operational blindness, 485 Privacy, 24, 610, 633, 655, 660
Operation of organizations, 253, 467 Probability Density Function (PDF), 311, 318,
Operations Research (OR), 257, 567 324, 559
Optimization Problem
688 Index

Problem (cont.) Re-intialisation

Abstraction-Analytic (PAa), 621 educational institutions, 565
Abstraction-Synthetic (PAs), 621 systems, 579
definition, 2, 22, 39, 167–172, 174–176, Relative ranking of the importance of FRs, 206
180–182, 184–189, 307, 545, 620, Renewable energy, 13, 302, 665, 666
621, 628 Repeatability, 261, 529
Perception (PP), 621 Repeated Stress Injury (RSI), 169, 170
Process Replication of PVs versus innovation, 243
architecture domain, 370 Research funding agencies of the U.S. Dept of
matrix, 277 Defense
planning, 591 Air Force Office of Scientific Research
variables (PVs), 2, 11, 22, 29, 32, 35, 45, (AFOSR), 583
52, 55, 57, 72, 86, 105, 178, 181, Defense Advanced Research Projects
191, 192, 222, 241, 242, 244, 246, Agency (DARPA), 251, 583
250, 252, 256, 281, 310, 486, 528, Office of Naval Research (ONR), 583
557 Research universities, 111, 458, 462, 464, 474,
Product 479, 480, 564, 583
design, 118, 121, 174, 183, 252, 263, 264, Retrospective coherence, 333
276–278, 280, 331, 337, 339, 341, Reverse engineering, 29, 205, 211, 275, 440,
342, 382, 394, 442, 472, 484, 485, 593
559, 586, 645 Reynolds number, 288
family design, 351, 364 RIM machine, 40–43, 232
Production design, 263, 264 Ripple, 394
Production management, 483 Robot, 87–89, 104, 185, 230, 398, 484, 500,
Production Performance Indicators (PPI), 497 502, 503, 514, 522, 529–531, 552,
Production planning, 579, 591 553, 557, 567, 570, 571, 591
PVs in organizational design, 252 Robust design, 126, 206, 323, 324
Robustness, 66, 115, 126, 182, 206, 323, 327,
Q 340, 342, 343, 433, 516
Quality definition heuristic, 421 Role of constraints in choosing DPs, 220
Quality Management System (QMS), 493 Role of database in selecting FRs, 202
Quantum computing, 481, 660, 670, 671 Roles of CEO, 451

R S
Radio Frequency Identification (RFID)”, 512 Salary raise system, 64
Random variation, 317, 318, 323 Scalability, 491, 514, 528, 539, 593, 607, 642
Real complexity, time-independent, 331, 442 Scheduling
Real problem, 40, 167, 328, 631 Airlines, 335, 568
Recipe, reverse-engineering , 418 Job shops, 576
Reconfiguration process, 407 production systems, 567, 571
Recycling of materials, 678 Scottish Haggis, 434
Re-design of an organization, 470, 473 S-Curve, S-Gap, and Vector delta (?), 474
Reduction Security, 10, 46, 61, 106, 173, 174, 188, 271,
of emission, 303, 484 291, 352, 449, 583, 607, 610, 614,
of materials consumption, 26, 94 624, 633, 642–645, 654, 675
Redundant, 77, 178–180, 209, 216, 238, Selection of
285–293, 295–305, 338, 382, 384, a right set of FRs, 111, 202, 204
404 of FRs, 192, 193, 202, 210, 227
Redundant Array of Inexpensive Drive Self-driving cars, 67, 180, 235, 678
(RAID), 286 Semiconductor
Redundant design, 77, 286, 288–291, 295, 297, chips, 21, 104, 112, 453, 454, 563
298, 301, 304 devices, 9, 27, 112, 113, 142, 200, 451,
Refrigeration cycle, 379 453, 467, 567
Index 689

wafers, 105, 569, 571 South Korea, miracle of, 581

Semiconductor fabrication facilities (FAB), Space life support system, 195
245, 567 Speech recognition, 588, 589
Shannon’s information theory, 328 Spidurky, 426
Shaped Magnetic Field In Resonance (SMFIR), Stabilizing the weather, 199
99–103, 193 Stakeholder, 107, 147, 168, 171–173, 182,
Shoe sole design, 53 183, 187, 188, 262, 355, 369, 381,
Shopfloor management, 484, 493–495, 498, 468, 471, 505, 506, 508, 513, 543,
499 545, 546, 558, 560
Shopfloor management software requirements domain, 373
data analytics, 497, 499 Start-up industrial firm, 452
dedicated trainings, 496, 498, 500 Stigmergy, 605, 608, 609, 653
digital value stream map, 497, 500 Stirling engine, 358
energy cockpit, 497, 500 Structural degrees of freedom, 407
issue tracking, 498, 500 Structural elements, 349, 526
location-based decision, 499 Student enrollment in a well-known academic
mobile app, 496, 497, 499 department, 75
myRole view, 497, 499 Stumbling block in learning AD, 22
online tutorial, 498, 500 Supply chain, 592
performance monitoring, 472, 494, 497, Sustainability, 125, 408, 442, 463, 484, 486,
499 493, 592, 601
predictive maintenance, 498, 499, 591 Sustainable packaging, 439
problem map, 496, 497, 499 Swiss Hay Soup, 436
smart info board, 494, 499 System
Shopfloor management software smart layout, architect, 65–67, 84, 194, 207, 210, 214,
495, 497, 499 372, 547, 632
Shoreline, 310, 344 architecture and design matrix, 207
Similitude of systems, 28, 73 boundary, 372, 543, 545–548, 552, 555,
Simplex method, 257 560
Simultaneous engineering, 342, 343 constraints, 182, 183, 406, 407
Small and Medium-sized Enterprises (SMEs), constraints matrix, 407
342, 346, 488, 490 decomposition, 367, 371, 378, 382, 408
Small-sized, 372, 487 design, 84, 148, 149, 163, 191–193, 195,
Smart manufacturing, 500, 587, 594 219, 235, 336, 352, 440, 473,
Smart Mini Factory Laboratory, 506 484–486, 505, 522, 526, 531, 536,
Society of Concurrent Product Development 539, 540, 543, 547–549, 554, 560,
(SCPD), 276 564, 575, 592
Software, 4, 7, 10, 19, 21, 28, 31, 36, 38, 42, form, 372, 548, 551–553
44, 48, 85, 117, 118, 121, 124, 127, function/behaviour, 372
146–152, 158, 163, 203, 207, 209, knowledge base, 407, 409
215, 221, 222, 241, 254, 322, 328, range, 66, 128, 177, 259, 310–312, 315,
352, 360, 458, 483, 484, 493–500, 317–321, 323, 324, 331, 334–336,
507, 518, 546, 549, 550, 567, 584, 338, 339, 342, 343, 481, 597, 598,
586, 587, 593, 594, 607, 608, 615, 645
617, 626, 639, 642, 668, 670, 671, systems-thinking, 367, 371, 373, 376, 378,
673 408
Software representation (UML), 118, 146–150,
152, 154, 158 T
Solar system, 373, 573 Target customer, 168, 170–173, 184, 185, 187,
Solunio Gmbh: Visual Shop Floor, 493, 494 188
Solution-neutrals Technical drawing of the brake disk and
environment, 8, 22–24, 39, 192–195, identification of DPs, 131
197–199, 204, 212, 256, 259 Thanksgiving, 429
690 Index

Theorem U
1 on selection of DPs, 219 UBER, 21, 183, 184
on rational design, 218 Umbrella that follows the owner, 358
R1, 298, 301, 302 Uncertainty and the information axiom, 66
R2, 298, 304 Unconventional problems, 169
R3, 299, 304 Universality, 404, 491
R4, 299, 304 Universal nature of design tasks, 72
R5, 299 University Research Laboratories funded by
R6, 300 the U.S. government
R7, 301 Applied Research Laboratory of Johns
R8, 302, 305 Hopkins University, 583
Theory of Inventive Problem Solving (TRIZ), Jet Propulsion Laboratory (JPL) of
423, 424 California Institute of Technology,
Thermodynamics, 4, 48, 358, 628, 679 583
Thinking Design Machine (TDM), 67, Lincoln Laboratory of MIT, 583
235–237, 586, 588, 595 USM corporation
3-D printing, 251 high pressure molding process, 249
Time delay, 563, 650 leasing versus selling, 565
Tolerance, 51, 58, 77, 115, 117–119, 125–127, shoe, 565
129, 133–137, 159, 205, 211, 224,
246, 257–259, 262, 270, 271, 274, V
279–281, 299, 310, 317, 319, 322, Virtual Mock-Up (VMU), 586
330, 332, 338, 344, 473, 556, 557, Virtual Reality (VR), 586
599
Track machine, 328, 563, 567–572 W
Trade-off, 259 Waterfall-model, 342
Transformation Water jet, 364
of a university, 466 Wireless electric power transmission system to
Transportation, 13, 16, 23, 24, 57, 61, 96, 113, a bus in motion, 95
173, 186, 188, 266, 410, 465, Workshop, 159, 495, 496, 505, 509, 512
551–553, 564, 567, 568, 615, 660,
662, 667, 668 Y
Tree-like large system, 368 Year 2520, 573
Trigger point, 599 Yoga ball chair, 170, 171
Tuning, 259, 269, 271, 298, 299, 302
Turkey Z
breast meat, 427 Zigzagging
brining, 440 between the domains, 52, 256
cooking a, 440 process, 52, 89, 180, 256, 312, 352,
dark meat, 440 376–378, 382, 401
deep-fat-frying, 428
drumstick, 426