Use of Stochastic Analysis for FMVSS210 Simulation Readiness for Correlation to Hardware Testing

Abstract

The FMVSS210 regulation establishes requirements for seat belt assembly anchorages to be strong enough for effective occupant restraint. The belt separation from the vehicle structure in crash tests needs to be avoided. Federal government mandate requires use of Pelvic and Torso Body Blocks for testing belt anchor strengths for lap and shoulder belts respectively. The belt anchorages are expected to withstand loads of 13.34 kN if both lap and shoulder belts are used and 22.24 kN if only lap belts are used. The analytical simulation of the hardware test is done using explicit dynamic code LS-DYNA. Hardware testing is of quasi-static nature while the simulation uses the dynamic code. However the analysis could be made to approach the quasi-static test by adjusting some input parameters in the simulation. In addition some input parameters need adjustment for making the model robust and to make it correlate to the hardware test. This study involves the use of Optimal Symmetnc Latin Hypercube Design to explore the design space, and to develop a fast surface response model. This response model can be viewed as a surrogate model to the actual LS-DYNA simulation and is used in this work to rank the input parameters by the percent contdbution they make towards the variation of the desired output responses. After determining the fit of the response model, it is used to perform the stochastic simulation. The confidence interval for test correlation prediction can then be estimated. This technique can further be used to do design sensitivity studies and for optimizing the vehicle structure with respect to FMVSS210 regulation.