Managing surrogate objectives to optimize a helicopter rotor design - Further experiments

It is common engineering practice to use response surface approximations as surrogates for an expensive objective function in engineering design. The rationale is to reduce the number of detailed, costly analyses required during optimization. In earlier work, we developed a rigorous and effective scheme for managing the interplay between the use of surrogates in the optimization and scheduled progress checks with the expensive analysis so that the process converges to a solution of the original design problem. In this paper, we will report our latest numerical tests with a helicopter rotor design problem which has proved to be a fruitful laboratory for experimentation. The results given here support the use of an ANOVA decomposition on a DACE model to identify the most important optimization variables in an optimal design problem. Key Words: Approximation Models, DACE Mod- y Professor, Member AIAA Research Scientist. Research was supported by DOE FG03-95ER25257, NSF CRPC CCR9120008...