Two Applications of Design As a Sequential Decision Process

Abstract

A promising approach to overcome the challenges of exploring a design solution space is to employ Set-Based Design tightly coupled to Model-Based Systems Engineering, and to treat the design process formally as a sequential decision process. In such a paradigm, designers start with an initial set of potential solutions, using lower fidelity models and analyses to winnow the set, with the process sequentially advancing through smaller sets of alternatives using models/analyses of ever increasing fidelity. The bounds around a solution are constricted until a candidate is chosen. This sequential decision process can itself be subjected to analysis, and optimized to most efficiently arrive the final choice. We have developed a formal model of Design as a Sequential Decision Process, and we illustrate the process through two examples. The examples are from two different design problem domains (finite element modeling with variable grid fidelity, bi-level optimization with combinatorial kernel) thus demonstrating the broad utility of the method. Ongoing and future work are also discussed.