A Comprehensive Evaluation of NHTSA Rollover Test Data for Use in Computational Model Validation
Increased computational power and new software have brought occupant motion simulation into the mainstream for vehicle accident reconstructionists. Using programs available today, investigators are able to achieve numerical results that match actual physical results with a high degree of accuracy. It should therefore be possible to validate the performance of a software simulation using instrument data collected from a real vehicle test. For valid results, however, one must have valid data upon which to base the simulation. We attempted to validate MADYMO occupant motion simulation software by using data from the National Highway Traffic Safety Administration (NHTSA) vehicle crash test database for vehicle rollovers. In the course of our work, we discovered flaws in the NHTSA database that rendered it useless for both validating and disputing a computational simulation. These flaws included data that did not match the descriptions of vehicle travel in the written reports, entire channels of missing data, and others. NHTSA crash tests are often cited as reliable sources of data in vehicle crash situations. While not disputing the limited scientific value of these tests, this paper documents the problems with NHTSA test reports and concludes that the data contained therein can be unintentionally misleading and of little value for computational model validation of rollover simulations. This paper also presents testing improvement procedures that should allow a greater correlation of computational and testing data.