(Drs. Farrokh Mistree and Janet Allen, co-advisors)
"Decision Support for Distributed Collaborative Design and Manufacture"
Traditional approaches to engineering design have relied heavily on the application of experience-based knowledge alongside trial and error exploration of a design space. A steady increase in the scope, distribution, and complexity of engineering projects, however, has rendered approaches that are based exclusively on insight highly ineffective. Both the number and difficulty of decisions involved in the realization of systems have reached levels where ad hoc evaluation and selection of alternatives is no longer viable and decision support is warranted. In this thesis, an approach for lending consistent decision support to designers engaged in Distributed Collaborative Design and Manufacture is proposed.
Adopting the view of Decision Based Design, in which the principal role of an engineer is taken to be that of making decisions, design can be viewed as the process of translating requirements for a product into design parameters characterizing it. This transformation is accomplished by making successive design critical decisions. These decisions are based on scientific, factual information as well as empirical, experience-based knowledge. More importantly, however, they are strongly influenced by preference. Often, preferences regarding the achievement of goals, meeting of targets, and satisfaction of constraints are conflicting; especially where multiple attributes are concerned. Thus, there is a need for decision support in all stages of a product’s life cycle and the nature and types of decisions that are being implemented determine the progression of a design. In this thesis, a quantitative method, anchored in utility theory, is presented for supporting designers in making selection decisions that are preference consistent and reflect decision-maker attitudes towards risk and uncertainty associated with potential outcomes. The effective structuring of dependent decisions is also addressed. The focus is on formalizing the interactions of sequential decision-makers, acting as stakeholders in a common design process, through the development of digital interfaces at the process level of interaction. Elements of the design space are modeled mathematically using the newly developed utility-based Decision Support Problem; preferences are assessed and communicated through multi-attribute utility functions. The resulting method is then applied to supporting designers engaged in the production of rapid prototypes and tools in a distributed product realization environment.