Evaluation of Seat Performance Criteria for Future Rear-end Impact Testing
In the past, EEVC WG12 and 20 have evaluated rear-impact dummies and reviewed associated injury criteria and assessment reference values for seat performance evaluations (Hynd et al. 2007 and Hynd and Carroll 2008). The BioRID II was recommended to be used in future legislative dynamic rear-end impact seat performance tests. Recommended injury criteria and assessment reference values to be used with the dummy are however still pending. This is mainly due to the incomplete understanding of the injury site and mechanisms responsible for the symptoms presented after such impacts. This lack of biomechanical data limits the possibility to evaluate any proposed injury criteria and associated reference values.
The aim of this study is to address these limitations by comparing crash test dummy parameter values from performed sled tests with real-life accident data. The results are expected to indicate the injury predictability of the complete sled test method, which includes performance criteria, the use of a generic sled acceleration pulse, the use of the BioRID II and its current positioning procedure.
Real-life injury risk was calculated for 32 individual car models and for 17 groups of similar seat designs from data provided by Folksam. When grouped data was introduced, i.e. by dividing applicable data into groups with similar seat designs, the reliability of the insurance data was raised, while the dummy measurements remained constant. The number of insurance cases ranges from 32 to 1023 for individual car models and from 132 to 1023 for groups with similar seat designs. Regression coefficients (r2) were calculated and the data presented graphically. Two types of injury risks were used in this study: those that had documented symptoms for more than one month and those that were classified as a permanent medical impairment as the consequence of a rear-end impact. These injury risks were compared to crash test dummy parameter values from sled tests performed with a BioRID II in 16 km/h medium Euro-NCAP pulse.
It was found that the analysis of groups of similar seat designs provided the most reliable results. Analysing individual data clearly showed that the insurance cases were too low per seat model to be used in an evaluation of seat performance criteria. In conclusion, the results obtained in the analysis of individual data did not invalidate the results obtained using grouped datasets. This conclusion was based on the observation that the correlations found in the analysis of grouped datasets could exist also for individual car model data.
When comparing groups of seats, the analysis showed that the Neck Injury Criterion (NIC), the maximum rearward Occipital Condyle x-displacements in a coordinate system that moves with the T1 and the maximum L1 x-acceleration were the parameters that best predicted the risk of developing permanent medical impairment, and symptoms for more than one month given that the occupant had initial symptoms following a rear-end impact. The maximum rearward head rel. T1 angular displacement, T1 x-acceleration and upper neck shear load (U.N.Fx, head r.w.) were parameters that also could predict the risk of permanent medical impairment and symptoms for more than one month. These results are supported by recent studies.
In comparison with a previous report, this study includes additional seat tests data which allowed additional data points to be included in the regression analysis. An expanded insurance claim database, about three times more insurance claims, was included in the analysis, which made the results more reliable. The insurance data was compensated for differences in the definitions of short term symptoms and permanent medical impairment during the accident data sampling period. This reduced errors that could have been introduced by the market share change during the sampling period for the various vehicle models included in this study.
In the future, a logistic regression including error estimation that covers all available insurance and test data should be carried out. The advantage of such an analysis would be that data could be included independent of the number of accidents. Another advantage of this is that a larger proportion of the data would be from tests and real life accidents with newer cars than those included in this study. Therefore the recommended parameters to use in seat evaluations would be more suitable for modern car seat systems.
real life data