Pedestrian injury risk in sedan, SUV, and pickup impacts using an improved human body model

The objective of this study was to evaluate the biofidelity of the baseline Total Human Model for Safety (THUMS) AM50 pedestrian model as measured by CORA and develop further improvements of the model to improve injury prediction capability of the thorax and pelvis. Simulation of impacts against three vehicle types, sedans, SUVs and light pickups, calibrated for pedestrian impact, were conducted and vehicle shape and design effects on injuries in the thorax and pelvis were identified.

The THUMS AM50 pedestrian model was refined to improve joint biofidelity and validated against pelvic side-impact tests and full-body PMHS buck experiments. The model was then used in full-body simulations of pedestrian 40 km/h impacts with calibrated production vehicle FE models (sedan, SUV, and pickup) to investigate thoracic and pelvic injury mechanisms. A simplified buck model was modified to represent several vehicle configurations and evaluated against full-vehicle simulations to assess its capability for geometry-based injury prediction.

The refined THUMS model demonstrated improved CORA and injury prediction across component and full-body level validations. In pelvic side-impact validation, CORA ratings improved to a fair level, the refined model reproduced sacral fracture observed in 5 of 6 acetabular-impact specimens and SI joint dislocation without fracture observed in all 6 iliac-impact specimens, which baseline model did not. In pedestrian buck validations, local kinematic agreement improved substantially, with knee CORA scores increasing to 0.79, and overall injury prediction accuracy rising from 52% to 83%.In pedestrian full-vehicle impact simulations using representative models from three vehicle classes, distinct vehicle-dependent injury patterns were observed. The most severe thoracic injuries were produced in impacts with pickup, with a 100% risk AIS 3+ rib fracture driven by early bonnet leading edge (BLE) contact and chest velocities of 10.6 m/s. Pelvic forces were highest in impacts with SUV, exceeding injury thresholds and resulting in pubic and SI joint injuries, while pickup induced acetabular fractures via femur-driven loading. Sedans exhibited lower pelvic and thoracic injury severity. Abdominal injury risk was highest in pickup, with organ pressures exceeding AIS 3+ thresholds, whereas head injury risk remained moderate but increased at off-center impact locations in sedan and SUV due to structural stiffness. Simplified SAE buck models reproduced lower-extremity responses and injury trends but showed limitations in predicting pelvic force, thoracic, and head injuries.

Refinements introduced into the baseline THUMS model resulted in improved prediction of dynamic responses of the anatomical structures of the pedestrian and enabling more accurate injury predictions. The predicted injury patterns were influenced by vehicle front-end geometry, stiffness, and loading with increased probability of AIS 3+ thorax, pelvis, and internal organ injury for taller, blunt-front vehicle geometries. These findings have implications for the development of pedestrian subsystem tests in impacts involving aggressive vehicle front-end designs.