The existing head injury criterion (HIC) is based on translational accelerations of the center of gravity of the head, while the importance of the rotational motion has been discussed over decades. Although most previous studies to establish injury criteria have depended on cadaveric, human volunteer or animal experiments, ever-progressing computing techniques both in software and hardware are making such studies possible using virtual experiments by FE simulation. In reality, however, such simulation models must be fully validated against a real human body before they can be used for such studies, and these validations depend on the tests using animal and human material.

In this paper, another approach to validate a human head FE model is introduced. Two cases of pedestrian accidents were selected from the accident database of the Road Accident Research Unit of the University of Adelaide and were reconstructed using a combination of physical testing and a FE model of the pedestrian/vehicle collision. The results of the FE model of the head were compared with the neuropathology of the actual victims to see if such an index as maximum principal strain was a correlate of the location and severity of injury.

After a comparison between the results of the model and the neuropathology was made, a tentative application of the model was tried. A parametric study on translational acceleration and duration time was performed and the relationship between the simulated brain conditions and the existing head tolerance curve (WSUTC) were discussed. Finally, additional simulations where pure rotational motions were applied to the model showed the likelihood of injuries from these motions alone. From this, the need for a criterion that considers both translational and rotational motions was suggested.