Phenomenological driven modeling of joints

In the construction of automobiles different technologies are used to join sheets. Today the most common method for connections is resistant spot welding. There are thousands of spot welds in body-in-whites, which are determining the behavior of the structure under crash conditions. New research is striving after replacing these thermal joints by adhesives or rivets for optimization of the production process. It is essential to ensure a cost saving and time optimized car development to reduce the required number of experiments by using precise simulations. For the quality of such simulation models accurate reproduction of the mechanical joint behavior is necessary. Because of the fact that the element size is bound by the time step in explicit finite elements schemes, a detailed model for the joints is not applicable in typical body-in-white simulations under crash conditions. For this reason simplified models using beam or solid elements have to be used to represent the connection. In the majority of load cases current modeling strategies do not show the required accuracy needed for design decisions. Therefore new ways of spot weld modeling and approaches for rivets modeling should be investigated. For spot welds several strategies of modeling exist. They all are based on the effort to reproduce the spot weld behavior by using specially adopted constitutive or structural models. For these models parameters have to be determined by comparing the maximum load under tensile, shear and peel conditions with corresponding numerical investigations. In the present paper it will be shown that a model with a relative small number of elements driven by a rigorous phenomenological approach can achieve better results. The quality of the proposed model will be evaluated by comparison of tests with KS2-specimen. In the field of self-piercing rivets an established modeling technique doesn’t exist. In this paper firstly capabilities of several modeling techniques will be investigated and secondly they will be compared with a model based on the aforementioned approach.