A Multistage Decomposition Strategy for Design Optimization

Abstract

A basic multistage decomposition strategy for design optimization is described. Resolution of the independent variable vector is accompanied by a decomposition of the optimization problem equation system into sets of modular components and an integrating system component. The optimization subproblem of each modular component set must be explicitly solvable, and specifically the method of optimal design is used. Techniques are developed for handling interactions between the modular component sets. The integrating system component is the basis of control for an automated optimal design search process of reduced dimensionality. An example illustrates application of the developed algorithm.