Predicting head injury metrics during low- to moderate-speed frontal collisions using computational simulations

Abstract

ObjectivesSubjective reports of mild traumatic brain injury (mTBI) are common following low-energy motor-vehicle collisions. Biomechanical analyses are useful in providing a quantitative means for determining the likelihood of sustaining mTBI. While occupant dynamics in low-speed rear impacts have been extensively investigated, peer-reviewed studies on occupant dynamics during low-speed frontal collisions are sparse. The objective of this study is to present a validated computational method to quantify the biomechanical response of the head in low- and moderate-speed frontal collisions. Data and methodsThis study used data from a previously-published series of four instrumented in-line front-to-rear staged collisions using 2014 Honda Accord sedans at closing velocities of approximately 7.4 kph (test L1), 12.7 kph (test L2), 21.7 kph (test L3), and 33.6 kph (test L4) kph. A model of the test vehicle occupant compartment was created using the MAthematical DYnamic MOdeling (MADYMO) software using methods previously described. Crash pulse data from L4 were applied to the MADYMO model. Seat belt parameters were optimized to achieve reasonable agreement between simulation results and test data for relevant head injury metrics (linear head acceleration [LHA], angular head acceleration [AHA], and HIC15). Crash pulses from the other tests in the series (L1, L2, and L3) were then applied to the model and peak values for LHA, AHA, and HIC15 were compared to the physical test data to demonstrate validation. ResultsThe optimization of seat belt and seat parameters within the MADYMO model resulted in accurate prediction of ATD dynamics demonstrated in Test L4. The simulation-predicted peak LHA was within 0.2 g of the test value, peak AHA was within 64 rad/s2, and HIC15 was within 0.46. When applied to the remainder of the tests (L1, L2, and L3), the optimized model showed excellent accuracy in predicting peak LHA and HIC15. When compared to the physical test data, the simulation-predicted values for LHA were within 0.4 g or less and the HIC15 values were within 0.4 or less across all tests. The model generally over-predicted AHA, particularly for the lower-severity collisions (L1 and L2). ConclusionsWe have demonstrated a reliable methodology for developing a biomechanical computational model to predict head injury metrics in low- to moderate-speed frontal collisions. This approach can be particularly valuable in forensic investigations of real-world crashes. Pre-existing crash test data can be used in conjunction with exemplar vehicle information to validate a MADYMO model. Appropriate crash pulse data from classical accident reconstruction techniques, event data recorders, or simulations can then be applied to the model to accurately predict head dynamics for real-world vehicle occupants without the need for full-scale staged crash tests.