Abstracts and Foreword for the Special Edition of the Journal of Engineering Valuation and Cost Analysis on Decision-Based Design

FOREWORD

Abstracts Available:
The Open Workshop on Decision-Based Design: Origin, Status, Promise, and Future
The Engineering Design Discipline: Is its Confounding Lexicon Hindering its Evolution?
A Toolkit for Decision-Based Design Theory
The Decision Support Problem Technique: Integrating Descriptive and Normative Approaches in Decision Based Design
On Decision Model Development in Engineering Design
Treatment of Uncertainties in Optimization-Based Design Decision Making
An Approach to Facilitate Decision Tradeoffs in Pareto Solution Sets
Pareto Sets in Decision-Based Design




THE NEED FOR A THEORY OF DESIGN

George A. Hazelrigg
Program Manager
Division of Design, Manufacturing, and Industrial Innovation
National Science Foundation1

Engineering design is important. It is what makes the engineering profession unique. Further, since the fall of the Berlin Wall, and with the increasing importance of markets for commercial products coupled with growing global competition, design has become even more important. Yet, engineers don't agree even on a definition of the word "design," let alone on a theory of design. Indeed, some people argue that a theory of design will never be found, nor is one necessary. I would like to argue against both points. First, I would argue that a theory of design is necessary and, second, that it is not only possible to develop such a theory but that we are well on the way to having one.

The fact is that we have today several methods that are used in support of the design process: Quality Function Deployment, Six Sigma, Voice of the Customer, Suh's Axiomatic Design, Analytical Hierarchy Process, Multi-Attribute Utility Analysis to name a few. But, curiously, in identical design situations, these methods all tend to make different design recommendations. More seriously, they can make widely diverse and clearly conflicting design recommendations. So it should be natural to ask, which if any of these methods is correct, when if at all is a method correct, and when should a method not be used? I assert that the only way to address these questions is through a theory of design.

Many researchers and practitioners in the field of engineering design assert that design methods can be tested through case studies. This is not true. Design is not a physical process, it is a logical process, and unlike models of physical phenomena, which are descriptive models, a design method cannot be verified and validated against physical data. Such validation would require that we repeatedly turn back time to try out alternative design choices to show that a prescribed process leads, at least statistically, to better design outcomes than all alternative choices. This obviously cannot be done.

Design methods are not descriptive, they are prescriptive. Thus, at least in intent, they are normative methods-they prescribe how design decisions should be made. The validation of a normative theory is a mathematical exercise. It begins with the statement and validation through consensus of a set of axioms. This set of axioms comprises a belief system within which the structure of a design logic must exist. From axioms, one derives theorems. Theorems are provable within the context of their supporting axioms. Then, a design method can be validated by deriving it from the resulting theorems, or invalidated by showing that it is inconsistent with the theorems.

Thus, the importance of an axiomatic basis for design is clear. Without such axioms, we have no basis thinking that any design method is valid, we have no means of validating a design method, and we have no means of deriving a valid design method. But, a set of axioms that is agreed upon, together with resulting theorems that enable their application, comprises a theory of design. Therefore, a theory of design is necessary.

It is more difficult to make a convincing argument that such a theory already exists, at least in the formative stage, but I will try. In the practice of design, we do principally two things: we generate design alternatives and we make a selection of a preferred alternative from among the set of alternatives so generated (the selection process includes the evaluation of alternatives). Furthermore, the selection of a preferred design alternative demands allocation of resources to the realization, marketing, field support and disposal of the resulting product. In accordance with the definition of "decision" by practitioners and researchers in the decision sciences, these activities comprise design decisions. The notion that design is decision making intensive is supported by ABET, the Accreditation Board for Engineering and Technology. Less recognized by ABET and others is the fact that engineering design comprises largely decision making under uncertainty. Fortunately, research on decision theory has been conducted for at least 300 years with many very fine results that can be drawn upon. Unfortunately, engineers have been slow to recognize and adopt these results. Yet we can conclude that what is needed is a theory that supports the notion that at least the selection of a preferred design alternative is a process of decision making under uncertainty. Some people also argue that the generation of design alternatives is also a process of decision making or at least of "value-focused thinking" [1].

An excellent set of axioms that provide a normative theory of choice under uncertainty was put forth by the mathematician John von Neumann and the economist Oskar Morgenstern in the 1940s. These axioms lead to a very useful set of theorems that provide a uniquely valid metric for design choice. More recently, I have incorporated these axioms into a comprehensive framework for engineering design [2]. Within the context of the von Neumann-Morgenstern axioms, we can be quite certain that this framework provides a logical, rational and self-consistent theory of design decision making. It is at least the basis for a theory of design. Furthermore, application of this theory can tell us which methods are valid, when they are valid, when they are not valid, and it can lead us to richer theories of design that include the entire product life cycle.

The theory of design that is now emerging from recognition of the role that decision making plays in engineering design is referred to as Decision-Based Design. It is a theory that needs to be studied carefully, for its conclusions and prescriptions are not obvious. But it is a theory that is rich in possibility.

This is a special edition of this journal, dedicated to papers that pursue this emerging theory of Decision-Based Design.

1The opinions expressed here are strictly those of the author and do not necessarily reflect the views of either the National Science Foundation or the Federal Government.

References 1. Keeney, Ralph L., "Value Focused Thinking; A Path to Creative Decision Making," Harvard University Press, Cambridge, MA, 1992. 2. Hazelrigg, George A., "An Axiomatic Framework for Engineering Design," Accepted for publication in ASME Journal of Mechanical Design.

Return to top

"The Open Workshop on Decision-Based Design: Origin, Status, Promise, and Future"

Wei Chen
University of Illinois at Chicago
Kemper Lewis
University of New York at Buffalo
Linda Schmidt
University of Maryland

ABSTRACT
The design perspective known as Decision-Based Design (DBD) has been under investigation via a web site-based workshop. During its first three years of operation, the Open Workshop focused on defining design from a DBD perspective and investigating the proper role of the DBD perspective on design. The workshop seeks to establish connections between theories in other disciplines, included and related to decision-making, and design. These relationships enrich our design theory foundations and provide insight for researchers. This paper summarizes the activities of the workshop as seen through the eyes of the organizers.

Key Words: Decision-Based Design, Open Workshop

Return to top

"The Engineering Design Discipline: Is its Confounding Lexicon Hindering its Evolution?"

Achille Messac, Ph.D.
Associate Professor
Northeastern University
Department of Mechanical Engineering
Boston, MA 02115
Wei Chen, Ph.D.
Assistant Professor
University of Illinois at Chicago
Department of Mechanical Engineering
Chicago, IL 60607

ABSTRACT
In this paper, we invite the engineering design research community to examine the current state of the engineering design lexicon. We expose the nature and the persuasiveness of practices that may hinder intelligible discourse within the engineering design literature. In particular, we show how such commonly used terms as criterion and metric are used sometimes as synonyms and sometimes not, often leading to material miscommunications. In our view, the engineering design discipline has reached a point in its evolution where clarity and conciseness of its lexicon should be a priority. Today's design activity takes place in a truly multidisciplinary environment, which often involves engineers of divers backgrounds. Written and oral design discourse among design researchers does not rely on a generally accepted and documented lexicon. These situations are symptomatic of a communication infrastructure that is not effectively facilitating the vigorous evolution of the engineering design discipline of recent years. In addition to detailing the outlines of the design lexicon deficiency, we also propose some avenues to a constructive and productive community-wide discussion on this subject, including conducting open discussions at the web site: http://www.eng.buffalo.edu/Research/DBD. We hope that this effort will be a catalyst for the development of an engineering design dictionary that will enjoy broad acceptance within the design community. A developed design lexicon will form a critical foundational component of Decision-Based-Design, the central topic of this special issue.

Return to top

"A Toolkit for Decision-Based Design Theory"

Beth Allen
Department of Economics
University of Minnesota
Minneapolis, MN 55455

ABSTRACT
This paper sets forth some mathematical techniques guaranteeing that an optimization problem has a solution. Attention is paid to the interpretation of these conditions within the framework of decision-based engineering design. Dependence of optimal solutions on underlying parameters that describe the model is also analyzed, so that the possibilities for robust optimal engineering designs are examined. Then the number of optional solutions is investigated under additional hypothesis, including some that ensure uniqueness.

Key words: Engineering design, decision theory, robustness, point set topology, convexity, constrained optimization, mathematical analysis.

Return to top

"The Decision Support Problem Technique: Integrating Descriptive and Normative Approaches in Decision Based Design"

Matthew Marston, Janet Allen and Farrokh Mistree
Systems Realization Laboratory
G.W. Woodruff School of Mechanical Engineering
Georgia Institute of Technology
Atlanta, Georgia 30332-0405

ABSTRACT
Decision-Based Design is in a pre-theory stage. Interest in the field is expanding rapidly as researchers recognize the inherent advantages of viewing design as a process of making decisions. However, much like the field of physics in the late 1890's, Decision-Based Design has no unifying theory or methodology. In this paper, we review the notion of decision making, our phrase for descriptive decision models, and utility theory in design. We then present some preliminary work into the use of game theory and an integration of these ideas into the Decision Support Problem Technique, our embodiment of Decision-Based Design. Our goal in writing this paper is to educate designers on the benefits and detractors from each approach and to show that we are pursuing an integration of both approaches.

Return to top

"On Decision Model Development in Engineering Design"

Xiao Tang
Graduate Research Assistant

Sundar Krishnamurty
Associate Professor

Department of Mechanical and Industrial Engineering
University of Massachusetts, Amherst, MA 01003

ABSTRACT
Decision analysis principles can provide valuable insights in advancing the state of knowledge on rational design decisions in engineering design and enable a better understanding of their consequences from an overall design perspective. From a practical point of view, a decision-based engineering design approach offers a formal strategy to reduce the multiple attributes in an engineering design problem to a single overall utility function in a probabilistic sense, which reflects the designer's intent and preferences under conditions of uncertainty. This paper focuses on some of the topics central to the development of decision models, and discusses them in the context of their use and implementation in engineering design. Along with a commentary on decision related issues that are specific to engineering design, a detailed study on the currently prevailing preference assessment methods is presented. Three different preference elicitation perspectives are introduced to emphasize the fundamental characteristics of value theory and expected utility theory based methods, as well as to highlight the differences between implicit and explicit articulation of priorities using normative and descriptive points of view. Their application to engineering design is demonstrated through a case study and the results are discussed.

Keywords: Engineering Design, Decision Making, Value Theory, SMARTS/SMARTER, Expected Utility Theory, Analytic Hierarchy Process, Multiattribute Utility Function

Return to top

"Treatment of Uncertainties in Optimization-Based Design Decision Making"

Li Chen
Assistant Professor

Design and Manufacturing Integration Laboratory
Department of Mechanical and Industrial Engineering
The University of Toronto
5 King's College Road
Toronto, Ontario M5S 3G8, Canada

ABSTRACT
This paper addresses the challenges of dealing with uncertainties for optimization-based design decision making. Centered on this theme, uncertainty-oriented design models are developed to account for design decision making in the presence of uncertainties. Relying on four uncertainty theories summarized, five categories of underlying uncertainty-oriented approaches are classified and reviewed. A variety of uncertainty modeling approaches are discussed for design decision making with uncertainties, in which a decision-based design problem is expressed using optimization formalism. Based on our research experience, design guidelines are provided as an alternate frame of reference for handling and manipulating the uncertainties. Two design examples are presented to illustrate applicability of the uncertainty-oriented design models. Keywords: Uncertainty Modeling, Uncertainty in Design, Design Selection, Decision-making, Optimization, Robust Design.

Return to top

"An Approach to Facilitate Decision Tradeoffs in Pareto Solution Sets"

Edward M. Kasprzak and Kemper E. Lewis
Department of Mechanical and Aerospace Engineering
University at Buffalo
Buffalo, New York 14260 USA

ABSTRACT
An approach to constructing Pareto sets and negotiating the decision tradeoffs within the Pareto set is presented. Points in the Pareto sets are generated by sampling the design space, and then a polynomial curve is developed to fit the points in order to approximate the Pareto set. A new method, called the scaling method, is used to automatically determine the appropriate weights for objectives of the problem based upon a selected design from the Pareto set. The Pareto set is then mapped to the design space to help designers make effective tradeoffs and decisions concerning the Pareto solution set. A vehicle dynamics design problem which has been developed with an industrial partner is used to demonstrate the usefulness of the approach.

Return to top

"Pareto Sets in Decision-Based Design"

Richard Balling
Dept. of Civil and Environmental Engineering
Brigham Young University
Provo, UT 84602

ABSTRACT
An approach to decision-based design involving Pareto sets is presented. Methods for generating Pareto sets using genetic algorithms are described. The paper presents a new fitness function which is a measure of Pareto-optimality in each generation. The methods are applied to a design example. Recommendations are made for developing tools to assist decision-makers in assimilating the information contained in the Pareto set.

Return to top