On variable screening and optimization of car body structures subject to multidisciplinary constraints

Besides eigenfrequency and other NVH related targets, vehicle structures are required to satisfy multidisciplinary design criteria. Efficient multidisciplinary design while taking into account conflicting structural requirements for car body structures comes with many challenges: the large number of parameters or design variables spans a vast high dimensional search space; the high computation cost for crashworthiness simulations strongly limits the number of available design evaluations; and the highly non-linear and non-convex simulation model responses complicate the search procedure for global optimization algorithms. There are various state of the art techniques that could be used to deal with these challenges separately, there are however only few significant guidelines available in the literature that help to select effective methods for the application to vehicle design problems which involve NVH, crashworthiness and light weight criteria, and the resulting previously mentioned challenges. Variable screening and sensitivity analysis methods can be applied in order to reduce the dimensionality of the search space by selecting the most relevant design variables, this comes however at additional function evaluation and computation cost. For the optimization of highly nonlinear and non-convex multidisciplinary structural problems many nature in-spired optimization algorithms have been proposed, the selection of efficient optimization algorithms remains however difficult. This communication presents an interdisciplinary review of state of the art on variable screening, sensitivity analysis, and optimization methods and comparisons on their performance to deal with some typical challenges that occur in the design of automotive structures. Since the design of car body structures deals with expensive simulations the performance of the methods are compared in terms of the computational effort and efficiency. Also the combined effects of variable screening based dimension reductions on the optimization efficiency will be assessed, to identify effective design strategies.