ABSTRACT
Design is a risky venture frequently involving a sequence of decisions under uncertainty (Chapter 1). The objective of a designer, who is also the decision-maker, is to select the product attributes in order to maximize the payoff of the venture. Designers need tools that will enable them to account for the uncertainty involved in design, and their own risk attitude. The process of managing uncertainty consists of four steps: 1) identify the important
uncertainties, 2) model uncertainties, 3) infer the payoffs of alternative courses of action, and 4) choose the best course of action. Figure 2.1 highlights these four steps in design decision-making under uncertainty. First, designers perform deterministic design in order to get a rough approximation of the final design. In this step, uncertainty is quantified using simple models such as design allowable values and safety factors. Designers identify important uncertainties by performing sensitivity analysis. For this purpose, they vary the uncertain variables in intervals, one at a time, and find the resulting intervals of the performance characteristics of the design. Tools, such as tornado diagrams, help designers identify the most important uncertainties (Howard, 1988). The second step is to model all the important sources of uncertainty. In the third step, a decision maker infers the uncertainty in the performance of each design alternative and the uncertainty in the payoff from building and selling each alternative design. The decision maker uses tools for propagating uncertainty through a system, such as probability calculus. Finally, the decision maker compares these alternatives and chooses the best one(s). This is an iterative procedure, as designers may want to create more alternatives, and obtain additional information from refined predictive models or tests. Designers need tools for modeling and propagating uncertainty through a system,
and for selecting the best design. This chapter presents an overview of the most important theories of uncertainty and the available tools for managing it. Each theory relies on ameasure of the likelihood that an outcome or set of outcomes of an uncertain event will materialize. The presentation in this chapter focuses on the philosophy of each theory and the interpretation of the measure of likelihood. Tools for modeling uncertainty, making inferences and choices are presented. Readers will learn that each theory is suitable for a different decision problem, and the choice of the best tool depends of the type of uncertainties, and the amount and type of the available information.
