The use of Advanced High Strength Steels (AHSS) has greatly increased this last decade in the automotive industry. Because of their crash performance and their weight saving potential, these grades constitute a possible solution to achieve the safety and environmental regulations objectives. Nevertheless, the increase of tensile strength of these materials is generally associated with a loss of ductility compared to conventional steels. Thus, the prediction of their failure in crash loading conditions become of great importance for the design of vehicles. This paper proposes to calibrate for any AHSS, as Dual-Phase or Press Hardened Steels, different failure criteria, available in finite element software. First, specific tests and methodologies for strain measurement needed for models calibration are exposed. Second, an overview of these tested models and the procedure applied to take into account mesh dependency are provided. Finally, simulations results are compared with experiment on a real automotive component.

An improved physically based behaviour law for ferritic steels is presented in this paper, and used to perform crash simulations. Flow stress depends on three contributions: the friction stress, the work hardening at low strain rate and the effective stress derived from the viscoplastic potential. These three terms are in most existing approaches simply summed up, but an original alternative mixing solution has been proposed for bcc metals by Edgar Rauch. The viscoplastic potential comes from the most general definition of a thermally activated mechanism at the dislocation gliding scale. The law intrinsically contains a threshold stress and exhibits only two unknown physical variables (activation volume and energy). The athermal quasi-static behaviour is finally described thanks to a linear combination of a Swift and a Hockett-Sherby laws. The complete model is then validated on the available data for two commercial ferritic steel. The predictions show a very good agreement with experimental data, and allow a good description of the temperature effects with a small numbers of parameters.