Publication

The pelvis is a key load bearer in vehicle safety due to its relatively high load tolerance and shape, which is utilized to control occupant kinematics in accidents by engaging with vehicle restraint systems. However, epidemiological studies have shown that the pelvis is also a highly exposed structure, as pelvic fractures are common outcomes due to interaction with the vehicle interior and restraint systems during a crash. Furthermore, fracture risk is not equally distributed over the population and vulnerable sub-populations have been identified depending on the load scenario. In addition, future autonomous vehicles are expected to allow for a more relaxed occupant posture by reclining the seatback, which increases the risk, in frontal impacts, of the pelvis sliding under the lap belt, i.e. submarining. Together, this motivates a deeper understanding of the potential of the pelvis as a load bearing structure, as well as its interaction with the vehicle restraint systems across the entire population, in various crash scenarios.

While vehicle manufacturers try to minimize variability in product development, human individual variability is an intrinsic property that must be considered to capture the vulnerable population and maximize the efficiency of vehicle safety systems. Finite element human body models (FE-HBMs) are the most advanced tool available to use in the virtual design of restraint systems and they provide the opportunity to include both geometrical and material variability through population based models and assessments.

In this thesis, methods enabling inclusion of population variance in FE-HBMs were implemented for the pelvis. Key findings include that sex, age, stature, and Body Mass Index (BMI), only cover a limited part of the population variance in pelvic shape, which is relevant for state-of-the-art FE-HBM development, population based simulation studies, and post-mortem human subject (PMHS) experiments. In addition, pelvic shape was shown to be an influential factor for both pelvis response in side impacts and belt-to-pelvis interaction in frontal impacts, which warrants consideration in future safety assessments.

To conclude, this thesis advances the field of pelvis computational models for automotive safety assessment and enables a population based evaluation for future vehicle safety systems, which can result in more robust systems, reducing the risk of injuries in real-life accidents.