Abstract

Contact mechanics plays a crucial role in many numerical simulations. Examples are sliding bearings, bolted and differently jointed areas, gear wheels or the contact between two rollers. Here, the contact forces depend on the one hand on the global deformations of the whole body, but also on the local deformations due to the local contact forces. These problems are therefore very challenging, so that usually the nonlinear Finite Element method is used. In order to keep the exploding computation times at least within tolerable limits, various simplifications, such as coarse meshes or quasi-static instead of dynamic simulations are common. In many cases, however, these simplifications lead to a reduced expressiveness of the result. This may lead to additional computational effort like stress reconstruction with a detailed FE mesh. In this presentation, we show an alternative approach that does not have all these drawbacks. The crucial point is, that instead of nodal degrees of freedom special contact modes are used. This is called model reduction. Those contact modes can be computed a priori, without a costly simulation of the unreduced system. If, in addition, stress modes are used for the reduction of the contact force computation, then it is called hyper-reduction. The above mentioned approaches have been implemented into multibody dynamics and successfully applied in academic and complex industrial examples like elastic crank drive simulations. With dramatically shorter computation times, the result accuracy of Finite Element method is obtained. At the beginning of the presentation, the basic ideas are explained. In the second part, case studies are shown that underline the strength of the approach. Those case studies are from the field of elastohydrodynamics and dry contact. The impressive reduction in computational times by orders of magnitude allows strategies to be applied to such problems that were not previously possible. One example is the consideration of nonlinear contact in optimization processes.