Accuracy of a damage-based reconstruction method in NHTSA side crash tests.

OBJECTIVE

Delta-V (ΔV), the magnitude of the velocity change experienced by a vehicle during a crash, is widely used as a predictor of injury risk. The National Automotive Sampling System/Crashworthiness Data System (NASS-CDS) uses the WinSMASH computer code to reconstruct ΔV based on postcrash vehicle deformation. WinSMASH, a direct descendant of CRASH3, first uses vehicle damage to estimate absorbed energy and then applies momentum conservation to estimate ΔV. This study aims to determine the accuracy of WinSMASH ΔV reconstructions for NHTSA side crash tests.

METHODS

This study is based upon 168 dynamic side impact tests conducted by the National Highway Traffic Safety Administration (NHTSA). For each crash test the actual ΔV for the struck vehicle was first determined from test instrumentation. WinSMASH was then used to reconstruct the struck vehicle ΔV for each test. WinSMASH-reconstructed ΔVs were compared to measured ΔVs for each test to assess reconstruction accuracy.

RESULTS

WinSMASH predicts ΔV at maximum crush, before restitution occurs. WinSMASH predictions of struck vehicle ΔV at maximum crush were 20 percent high on average when using vehicle specific stiffnesses, with a great deal of inter-case variability in the error. When compared to the total struck vehicle ΔV at separation (including restitution), WinSMASH overpredicted ΔV by 11 percent. WinSMASH overpredicted the amount of energy absorbed in collisions by 40 percent, which is consistent with overprediction of ΔV. When forced to reconstruct tests using the amount of absorbed energy calculated from the test data, error in WinSMASH ΔV effectively disappeared.

DISCUSSION

When reconstructing NHTSA side crash tests, WinSMASH overpredicts maximum crush ΔV by about 20 percent on average. The primary factors determining the amount of ΔV overprediction appear to be overprediction of absorbed energy and the assumption of zero restitution. WinSMASH vehicle side stiffness parameters are calculated based on artificially high energies; this may explain the overprediction of absorbed energy. WinSMASH's assumption of zero restitution partially masks the effect of energy overprediction. When given accurate reconstruction inputs and energy estimates, WinSMASH is capable of accurately reconstructing maximum crush ΔV in NHTSA side crash tests but cannot compensate for restitution.