Robustness Evaluations in Virtual Dimensioning of Passive Passenger Safety and Crashworthiness

One of the most important tasks of vehicle development is the steady improvement of passenger passive safety systems. In reality significant result scatter can be observed when performing crash-tests. Cause of this scatter of important vehicle performance variables is scatter of variables concerning the dimensioning of passenger passive safety systems, vehicle structure and testing conditions. A computational robustness evaluation of important result variables can only be obtained by integrating stochastic simulation methods into virtual product design processes [1-2]. Primary result of the robustness evaluations is the calculation of the scatter of performance variables and of the connected probability of achieving safety goals. Secondary result is the investigation of the numerical stability of the models and identification of the input scatter, which is responsible for the output scatter. In this paper the fundamentals of computational robustness evaluation are explained compendiously and the experience gained by the systematical introduction of computational robustness evaluations at BMW AG [3] is discussed. Because of the complexity of finite element-models their robustness has to be examined with great care. Using a robustness evaluation for the front-crash load case of the ULSAB-study an approach for identification and quantification of numerical noise is presented.