A Simulation-Based Calibration and Sensitivity Analysis of a Finite Element Model of THOR Head-Neck Complex

The THOR-NT dummy has been developed and continuously improved by NHTSA to provide automotive manufacturers an advanced tool that can be used to assess the injury risk of vehicle occupants in crash tests. With the recent improvements of finite element (FE) technology and the increase of computational power, a validated FE model of THOR may provide an efficient tool for the design optimization of vehicles and their restraint systems. The main goal of this study was to improve biofidelity of a head-neck FE model of THOR-NT dummy. A three-dimensional FE model of the head and neck was developed in LS-Dyna based on the drawings of the THOR dummy. The material properties of deformable parts and the joints properties between rigid parts were assigned initially based on data found in the literature, and then calibrated using optimization techniques. Frontal impact test data using head-neck THOR dummy complex and biomechanical requirements defined from volunteer test data were used to calibrate the neck FE model. A response surface methodology was used to determine the optimal stiffnesses of neck pucks and rear cables together with friction coefficients of cable-neck disks and cam-OC stoppers interfaces, and then to quantify their stochastic contributions to the neck response. A novel identification approach was used to determine the proper material property for the head skin based on the biomechanical requirements of both the head drop test and the forehead impact tests. A genetic optimization algorithm was utilized to identify the material parameters of a viscoelastic material model of head skin. To better understand the data variations observed in testing, a sensitivity analysis using both ANOVA and Sobol' indices were employed. The results suggest that the variations of initial velocities of head and impactor have the highest influence on the peak acceleration in drop test and the impact force in forehead impact test, respectively. While material characterization tests for deformable neck parts would be useful to check the calibration results obtained in this study, good performance of head and neck THOR FE models recommend its use in impact simulations intended to improve the design of new vehicles and their restraint systems. In addition, the computational methodology applied in this study to improve the performance of the THOR head-neck FE model and to investigate its sensitivity to the initial conditions in testing may be used for other improvements of other human or dummy FE models.