Stochastic Crash Analysis of Vehicle Models For Sensitivity Analysis and Optimization

Within the 6th Framework Programme EU project APROSYS, the Sub-project 7 is fully devoted to virtual testing. The aim is to improve the quality of the crash simulation in order to be able to come to rating and qualification by virtual analysis. One of the main issues lies in the evaluation of scatter sources and the consequences of scatter on the results of the analysis. Therefore, great effort was devoted in the project to identify and quantify sources of dispersions, and to assess their relevance. To evaluate the influence of scatter on crash responses a series of stochastic models has been developed. Within the APROSYS project a series of generic car models was developed to perform this task. Generic car models are virtual vehicles, derived from the geometry, layout, and characteristics of the best-in-class models currently available on the market according to European New Car Assessment Program (EuroNCAP) ratings, generated to have commonly shared models to work out towards improvement in crash simulations. In this work a stochastic analysis developed by using one of these generic car model, called GCM4, a multi-purpose vehicle, will be reported. The stochastic model was generated by considering the stochastic variation of some parameters. In particular, the steel sheet properties were used as stochastic variables (input). Moreover, to evaluate the structure influence on the passenger behavior, a simpler stochastic passenger compartment model was developed. The simulation runs were managed by a specific tool, called ADVISER, developed within the APROSYS project and its antecedent ADVANCE. The results were analyzed by means of the postprocessor included in the same ADVISER tool. The results give further insight in the problem of the improvement of simulations for passive safety applications.