Practical Implementation of Robust Design Optimization

Abstract

This paper focuses on practical implementation of multi-objective Robust Design Optimization (RDO). The main goal is to implement RDO such that it is intuitive; lexicon based and does not require extensive a priori knowledge of the available RDO formulations. It supports both single & multi-objective deterministic design optimization and single & multi-objective robust design optimization problem specification using single unified user interface. The implementation is based on wizards that guide the user through the optimization problem specification. The problem specification includes: input variable specification, objective and constraint specification, reliability and optimization method specification. Once the user specifies the design problem, the implementation automatically selects the suitable robust design formulation. The implementation provides options for the user to specify robustness bounds on inputs and select normalization of responses based on global sampling for weighted objective formulations. The intuitive nature of design problem specification is illustrated with an example of integrated knee bolster robust design optimization.