4.a. Evaluation of Seat Performance Criteria for Rear-end Impact Testing: BioRID II and Insurance Data

Johan Davidsson (JD) presented a common study of Chalmers and Folksam on the evaluation of seat performance criteria based on insurance data (WCWID-01-03). The presentation was an update of GTR7-14-02. The study was performed within the EEVC work programme by EEVC WG12 (see report WCWID-01-09):

Approach for the study is to compare claim rates for different seats in the Folksam insurance data with BioRID II metrics from tests for the same seats.

The focus is on neck injury and very few thoracic or lumbar spine injuries/claims are contained in the Folksam database.

Two risk levels:
1. Symptoms for more than one month in case of initial symptoms
2. Permanent medical impairment in case of initial symptoms (with no financial incentive to claim this level of symptom)

The Chalmers/Folksam/EEVC study uses 150 cases per seat (IIHS used a minimum group size of 30 cases, which may explain the weak correlations that they reported).

In Sweden, the risk for a given vehicle has reduced over 10 years, indicating that the assessment of injury by the medical profession has become stricter. This has been compensated for in the study by normalising to the 2005 risk level.

The risk for different seats cf. the NIC was shown as an example. Philippe Petit (PP) asked if the differences were tested for significance, because the range is quite small. This was not the case. Norbert Praxl (NP) asked if confounding factors were controlled like e.g. delta-v?

JD replied that there was no information on delta-v, but ensured that the groups were not very light or very heavy vehicles. The Volvo and Saab had a step-change in seat design on a
vehicle platform of very similar mass and structure.

Koshiro Ono (KO) asked whether the patients’ medical records were re-reviewed by the study in order to confirm the diagnosis and the severity of WAD. This was not the case, but for the permanent injury rating there is a considerable effort that the patient has to go to in order to be classified, with little reward.

Overall, NIC showed the best correlation with risk, both at the symptoms > 1 month and the permanent medical impairment levels. OC-T1 x-axis relative displacement and L1 x-axis acceleration correlated with long-term injury risk. Neck extension and T1 x-axis acceleration may be candidates, but appear to be sensitive to inclusion or not of outliers.

Limitations of the study include:
• Changes to the dummy over time
• Changes since the seats in the study were tested
• Correlation coefficients were maximum 0.72.

PP noted that there is a very strong assumption that the pulse is the same over time, even for an identical vehicle. In France, it was observed that the average speed of fatal collisions
was increasing until automatic ticketing was introduced, when it dropped 5%. JD: 2009 pulses likely to be more severe than 1999, due to changes in the vehicle structures.

Agnes Kim (AK) noted that the study is based on Folksam data and that means Swedish drivers, who are more likely to sit in a ‘standard’ position than US drivers, who tend to have a
very poor, non-standard posture. BL: This is the data we need – the regulator can only assume that a driver is sitting in a good position with a correctly adjusted seat, and ensure that they are provided with a minimum level of safety in this case if they take this care. The seat-belt doesn’t protect you if you don’t wear it, but it is still required to be fitted. AK pointed out that in regulations to date we take due care to ensure that if an occupant is somewhat out of position they will still be protected, but the rear impact is somewhat different.

Annette Irwin (AI) asked whether the BioRID predicts that the risk is, as would be expected, lower in the heavier vehicles. JD replied that he can’t answer this question based on this data, but other studies have done this.