Robustness Study of an LS-DYNA Occupant Simulation Model at DaimlerChrysler Commercial Vehicles Using LS-OPT

The robustness properties of crash simulation models are emerging as an important criterion in today's simulation driven vehicle development process. We consider it relevant to any study of highly non-linear crash problems to obtain measures for the repeatability and reliability of both experimental and numerical tests. In 2003, we conducted a robustness study of a structural front impact model (JAPAN LS-DYNA Users Conference 2003) using the Meta-Model concept of LS- OPT. This study seemed to indicate that there is an inherent, residual randomness in crash problems. In the present paper, we present a new robustness study of a frontal sled test occupant model with a 50th percentile FTSS dummy. Our goals were: • Determine variations of typical occupant safety responses such as HIC, chest intrusion, and others due to uncertainties in the experimental setup, including dummy position, airbag mass flow, and acceleration. • Separate deterministic and residual, random variations of responses using the Meta-Model technique. • Get a feeling for random variations inherent in occupant studies. • Evaluate the generality of these robustness results through a convergence study varying the number of simulation experiments and variables for the Meta-Model.