Surrogate-model based method and software for practical design optimization problems

Abstract

In a practical simulation-based design process most of the required simulations are carried out by using commercial software producing simulation responses but no gradient data. When applying mathematical optimization routines to such design process all the function evaluators must be considered as black-box solvers and the optimization can be integrated to design process efficiently using e.g. surrogate modeling techniques or heuristic optimization algorithms. In this paper we present a design optimization method that can be integrated very easily to any simulation-based design process. Special attention is paid on software implementation issues of the proposed method.