During the last decades, numerical simulation of crash events has become one of the key topics in the reduction of costs for the phases of development of new automotive products. The former conception as a tool to provide qualitative support to designers has evolved up to the point of talking about “virtual testing” and about the feasibility of include it in standards and regulations. This evolution of the perspectives requires more and more predictive simulation models, leading to a continuous improvement in the mathematical reproduction of the physical reality.

Within this background, the correct numerical reproduction of the material behaviour has a critic role. The techniques for material characterization have also evolved from the use of simplified curves obtained from scarcely instrumented tensile tests, including strain rate dependency in a higher or lower degree, up to the use of complex yield surfaces obtained from the exhaustive analysis of the local phenomena that occur during the necking process in tensile tests, as well as the inclusion of other load cases different to the uniaxial tension.

The current paper reflects the results of some studies about the influence of different levels of material characterization on the correct reproduction of the material behaviour. The base case is the simulation of the characterization tests themselves, analyzing both local and global parameters for the validation of the models. Three different materials (one metallic and two plastics respectively) have been used in these studies, trying to deepen into their basic characteristics and requirements. Finally, a load case closer to a common energy absorption application has been chosen for the case of the plastics in order to illustrate and validate the hypothetical consequences of the use of the different material definitions.