Optimization efficiency in multidisciplinary vehicle design including NVH criteria

An important challenge in the automotive product development process is weight reduction while satisfying the structural requirements on crashworthiness and NVH (Noise, vibration and harshness) criteria. The use of genetic algorithms and other heuristic search methods for multidisciplinary structural design optimization is widely spread in industry and research. Although there are many different optimization methods available, there are few comparisons or guidelines available to select effective algorithms for particular applications such as eigenfrequency optimization with nonlinear crashworthiness constraints. In this paper a selection of optimization algorithms is compared with respect to their performance efficiency for problems involving basic NVH criteria. The comparative assessment is made for several different optimization formulations that include multidisciplinary objectives and constraints related to: global vehicle eigenfrequencies, vehicle mass, and nonlinear crashworthiness responses. In order to perform sufficient optimization repetitions to achieve statistically significant results, meta-models have been used based on construction points of full vehicle simulation responses. The assessment is made using two distinct vehicle models. The significance of the assessment results are corroborated using results based on direct simulation workflow results of a third vehicle model.