Robustness analysis of structural crash load cases at Daimler AG

In the virtual product development, crash analyses are an important part for the design of the car body. The minimization of weight and some times competing requirements from several load cases have to be adjusted as optimal as possible. Consequently, no high safety distances can be kept while maintaining all requirements. Therefore, the assurance of robustness of the optimized design against unavoidable scatters of crash test constraints, production constraints and material constraints in preferably early stages of product development becomes more and more important. Recently, numerical robustness analysis is recommended as method of stochastic analysis in virtual product development [1, 2, 3]. It is successfully used for tasks regarding NVH [4], passive safety [5] or forming simulation [6]. Robustness analysis of structural crash load cases has high demands for the methods of stochastic analysis, statistical measures of determination and correlation as well as the efficiency of post processing. Reason for this are the complexity of modeling, long calculation times, high non-linearities and the influences of numerical noise. In the last years, DYNARDO could increase the efficiency of optiSLang [8] and the post processor Statistics on Structure [9] so that since 2007, robustness analysis of structural crash load cases with LS-Dyna can be carried out efficiently at Daimler AG. Using the example of insurance load cases, the proceeding of robustness analysis will be described. Since the forecasting power of robustness analysis could be shown for a real phenomenon which up to now could not be imitated in the virtual product, robustness analysis is used for the numerical assurance of car bodies against scatters of material, sheet thickness and test constraints.