Evaluation of General Motors Event Data Recorder Performance in Semi-Trailer Rear Underride Collisions 2020-01-1328
The objective of this study was to analyze the validity of airbag control module data in semi-trailer rear underride collisions. These types of impacts involve unusual collision dynamics, including unusually long crash pulses and minimal bumper engagement.
For this study, publicly available data from 16 semi-trailer underride guard crash tests performed by the Insurance Institute for Highway Safety (IIHS) were used to form conclusions about the accuracy of General Motors airbag control module delta-V data in a semi-trailer rear impact underride scenario. These tests all utilized a 2016 Chevrolet Malibu impacting a stationary 53’ semi-trailer at a speed of 35 mph. Nine of these tests were full width overlap collisions, six were 30% offset, and one was 50% offset.
The IIHS test vehicles were equipped with calibrated 10000Hz accelerometer units which were used as a reference. The accelerometer data had been integrated to speed (ΔV) and filtered, allowing direct comparison. Event data recorder (EDR) data imaged post-accident from the test vehicles was compared to the reference IIHS instrumentation. Normalized root mean square error for each test was quantified and related to crash duration, revealing a general trend of decreasing EDR ΔV accuracy with increasing crash pulse duration, although overall error remained low. For each test, the percent error over time, and the difference between the event data recorder ΔV and the IIHS reference ΔV, were plotted. Eleven tests, all with airbag deployments, analyzed converged towards an average of 1.12% error at the end of the crash pulse, which were 150-250 ms. EDR recorded ΔV were in the range of 29.8-40.0 mph. Four tests, three of which were non-deployments, diverged to higher error percentages of 10-15% at an EDR ΔV of 31.8-40.0 mph. All the divergent tests were 30% offset and had the highest crash pulse durations of 240-300 ms. One fully overlapped test generated highly inaccurate EDR data due to rear underride guard double shear failure.
Analysis indicates that General Motors event data recorders record ΔV data with high accuracy in rear underride collisions with high or full overlap; within 5% even at the highest crash pulse durations studied. However, the accident reconstructionist should be skeptical of data generated during a non-deployment, low overlap event. In such a collision, the recorded V may underestimate the true value by as much as 15%. In a collision where both legs of the underride guard have failed in shear, the EDR recorded ΔV may be less than half of the true value.
Nicholas Famiglietti, Ryan Hoang, Edward Fatzinger, Jon Landerville