Position Paper - Decision Based Design

Dr. Georges M. Fadel

Clemson University

Who Am I?

I am a design faculty in the Mechanical Engineering department at Clemson University.

I was educated in Switzerland (Diploma in ME from ETH Zurich, 1976), then came to the US to study Computer Science because I felt that computers were an essential tool for the engineer, and I wanted to be an expert at the use of that tool. I received an MS in Computer Science from Georgia Tech, then worked overseas for several years using and developing simulation programs for energy conservation. I returned to Georgia Tech where I received my PhD in Mechanical engineering (1988), developing a simulation program for non-azeotropic heat pumps. Upon starting my teaching and research career at Georgia Tech, I decided to focus on design. I first started working in optimization since I had extensively used this tool, and wanted to quickly develop an expertise. I spent three months at NASA Langley developing an approximation for structural optimization applications. Upon my return, I introduced and taught classes in CAD, numerical methods, introduction to computers and optimization, and continued the research in optimization. In 1992, I left Georgia Tech and joined the faculty at Clemson. I have taught courses at the sophomore level in numerical methods and introduction to design, and at the senior level, a CAD course, senior capstone design courses, and an interdisciplinary design course involving marketing, finance, industrial engineering and architecture. At the graduate level, I introduced two courses related to design. The first one is a design methodology course that explores the various techniques described in the literature: PDS, creativity, decision matrices, axiomatic design, decomposition, QFD, Taguchi methods, DFA, DFX, concurrent engineering, utility theory, value analysis, etc. and applies some of these techniques to industry problems. The second is a graduate level engineering optimization course emphasizing the tool as an aid to design. My research has spanned various areas related to design from decomposition, functional design and design metrics to rapid prototyping, CAD, optimization and virtual prototyping. I spent a summer at the Wright Patterson Air Force base investigating the issue of affordability in the design of aircraft. My research targets the development of tools and methodologies to aid the designer and facilitate the decision process by providing earlier results, better results, and specifically more meaningful results.

My Position

Having been self educated in the design area, and having been influenced by the European/ German design process, I view design as the essence of engineering. Design for me is problem definition and formulation, creativity, decision making, and prototyping (virtual and physical). To facilitate this process, and to be able to teach design, all these aspects must be considered. Making decisions in design is one of the critical aspects that is very dependent on the definition of metrics, on the accuracy and repeatability of results, and on the uncertainty inherent at the various stages of the process.

I was trained to do a lot of hand drafting. I was also trained in the use of machine tools. Looking back, I recognize the importance of being able to communicate with manufacturing, and the importance of assessing what can and cannot be done. This is something that has eroded over the years with the practical elimination of drafting from many curricula, and the replacement of such a class with a computer based CAD class. This ability helped in the design process, but did very little in broadening the vision of the designer.

The use of CAD tools has facilitated the creativity aspect inasmuch as visualization can help in assessing feasibility and enabling intuition to see what needs to be done to "improve a design". Rough order of magnitude calculations and factors of safety played an extremely important role in design. These tools dealt with virtual prototyping and with the uncertainty issue. Their teaching has been significantly reduced in our curricula.

So what has replaced what we eliminated? The ability to communicate with manufacturing is slowly being integrated into CAD through the use of features. Drafting has been replaced by solid modeling, surface modeling and wire frame drawings. CAD can generate 2D plan views, but the ability to move to 3D representations and generate physical prototypes or directly interface with manufacturing machines might eliminate some of this need. Order of magnitude calculations have been practically eliminated with the need to produce more competitive designs where approximate results are not sufficient to give a company the edge. Analysis and optimization have become defacto tools used at various stages of the design to get that increase in performance that will give the product an edge. The results obtained aim at improving some performance parameter, but the issues of uncertainty of properties, the issues of multiple objectives or multiple decision makers have only recently become of interest to design researchers.

Furthermore, all these modern tools do not help in the selection of metrics to help in the design process. Value engineering, utility analysis, QFD and the house of quality, DFA and DFX tools and methodologies have attempted with various levels of success to provide the designer with ways to deal with the assessment issue to help in decision making. Many of the tools and methods do not consider the decision making at various stages of the design process, rather at a particular stage (conceptual, detail design) nor have they attempted to reconcile the marketing driven metrics with the engineer/designer driven metrics. Also, most of the metrics available apply at the feature or component level, very little has been done to study how metrics can be generalized for the complete system. The development of such methodologies is critical if design improvements at the global system level are to be achieved.

The designer is a visual person. Whether the vision is obtained by drawing or by imagining, visual feedback is often desired. To this end, the representation of the design space is a tool that might be of importance in the decision process. Marketers use tools such as perception maps to understand customers. We, the engineers and designers do not have such a tool. If we could provide designers with some illustration that conveys the difficulty to improve some performance index, or some illustration that represents the rate of value improvement as a function of effort achievable for a design, then decision making would be facilitated.

Finally, bringing analysis earlier into the conceptual stage, considering uncertainty and providing the designer with better methods to explore the design space and assess the designs are all additional components of decision based design.