Evaluation of Human Surrogate Models for Rollover

Anthropomorphic test dummies (ATDs) have been validated for the analysis of various types of automobile collisions through pendulum, impact, and sled testing. However, analysis of the fidelity of ATDs in rollover collisions has focused primarily on the behavior of the ATD head and neck in axial compression. Only limited work has been performed to evaluate the behavior of different surrogate models for the analysis of occupant motion during rollover. Recently, Moffatt et al. examined head excursions for near- and far-side occupants using a laboratory-based rollover fixture, which rotated the vehicle about a fixed, longitudinal axis. The responses of both Hybrid III ATD and human volunteers were measured. These experimental datasets were used in the present study to evaluate MADYMO ATD and human facet computational models of occupant motion during the airborne phase of rollover. Occupant motion predicted by the Hybrid III ATD computation models provided a good match to the temporal movement patterns and corridors of torso and head excursion measured in the volunteers. Differences in torso and head-neck posture were attributed to active muscle contractions in the volunteers. Simulations performed using the TNO human facet model, in the absence of muscle tone, predicted large head excursions and lateral neck and torso bending. These findings were attributed to the stiffer Hybrid III ATD neck and torso as compared to the spinal model incorporated in the human facet model. Although it is possible to model active muscle forces using the TNO human facet model, the appropriate control schemes for coordinating muscle activity in the rollover environment have not been established. Without the implementation of appropriate muscular controls, the TNO human model appears to be best suited to high-force environments or low-force environments where the occupant is unconscious or incapacitated. Our results indicate that among the currently available human computational surrogate models, the Hybrid III ATD provides the best prediction of occupant motion when compared to the available human volunteer data. These results have provided us the impetus to study future human models that incorporate active muscle control.