Optimization and Robustness of Structural Product Families

This thesis concerns structural optimization and robustness evaluations, and new methods are presented that considerably reduce the computational cost of these evaluations. Optimization is an effective tool in the design process and the interest from industry of its usage is quickly increasing. However, the usage would probably have grown faster if the required number of computationally costly finite element analyses could be reduced. Especially in the case of product family optimization, the problem size can easily get too large to be solved within a reasonable time. This is sometimes also true for robustness evaluations. To enable the usage of optimization and robustness evaluations also for large scale industrial problems, two new methods are here presented, which require a considerably smaller number of finite element analyses. The first method concerns structural optimization of product families subjected to multiple crash load cases. Here, the number of required finite element analyses are considerably reduced by only considering the critical constraint in each iteration step. The second method is an approach to approximate the variable sensibility based on the distribution of internal energy in a structure. The method can be used to evaluate the relative robustness of different design proposals or for structural optimization. Since the method is independent of the number of parameters and design variables the computational cost of such evaluations is drastically reduced for computationally large problems