Model-Based Support for Mutable Parametric Design Optimization

Abstract

Traditional methods for parametric design optimiza-tion assume that the relations between performance criteria and design variables are known algebraic func-tions with fixed coefficients. However, the relations may be mutable, i.e., the functions and/or coefficients may not be known explicitly because they depend on input parameters and vary in different parts of the design space. We present a model-based reasoning methodology to support parametric, mutable, design optimization. First, we derive event models to repre-sent the effects of the system’s parameters on the ma-terial that flows through it. Next, we use these models to discover mutable relations between the system’s de-sign variables and its optimization criteria. We then present an algorithm that searches for "optimal " de-signs by employing sensitivity analysis techniques on the derived relations.