Multidisciplinary optimization of structures is characterized by a high computational effort to solve the coupled system equations. To reduce the computational cost for the optimization process approximations based on the response surface methodology are used instead of the exact analyses of the systems. A method is presented which allows to approximate major nonlinearities without having to use a large amount of costly system analysis. This is verified by practical examples. Grain parallelization of the analysis on a cluster of computers helps to achieve a huge reduction of computing time.     «

Multidisciplinary optimization of structures is characterized by a high computational effort to solve the coupled system equations. To reduce the computational cost for the optimization process approximations based on the response surface methodology are used instead of the exact analyses of the systems. A method is presented which allows to approximate major nonlinearities without having to use a large amount of costly system analysis. This is verified by practical examples. Grain parallelizati...     »