Hybrid discretizations in solid mechanics for non-linear and non-smooth problems

点击翻译按钮获取中文摘要

Modern finite element methods currently play an important role in the construction, design and development of new materials, innovative products and production processes. Despite successful research in the past, there are still many open problems, e.g., artificial stiffening effects, numerical instabilities and undesired mesh distortion sensitivity. Within this project, a special focus is on geometrical and material non-linearities, nearly incompressible, anisotropic and generalized materials as well as contact and interface models, since these fields are of great theoretical and practical relevance. Discontinuous Galerkin (DG) methods may be seen as generalizations of continuous methods, thus offering additional features and options for the improvement of numerical computations in the aforementioned fields. This comes at a cost, as DG methods require far more degrees of freedom and memory consumption than continuous discretizations on the same mesh. To improve this issue, we investigate hybrid discontinuous Galerkin methods allowing for a significant reduction of global degrees of freedom via static condensation.Our interdisciplinary group represents three research fields: Applied Mechanics, Numerical Analysis and Scientific Computing. Beyond the interdisciplinary research work within the team, we identified joint scientific goals with three different teams within the priority programme. Our aim is to explore the potential and the limitations of hybrid discontinuous Galerkin approximations in solid mechanics and to identify, develop and analyze related methods, which allow for an improvement of the performance in terms of convergence, robustness and stability without increasing the numerical effort.A first benchmarking of the hybrid methods showed promising results, which call for more investigations in the fruitful scientific environment of the priority programme. New symmetric hybrid DG methods shall be developed for the simulation of generalized material models in damage and plasticity as well as multi-scale problems. The DG concept opens up entirely new possibilities for adaptivity which shall be exploited based on proper error estimates. Within the field of interfaces, new promising discretization schemes for contact, delamination and non-matching meshes, being derived from the hybrid DG concept, will be developed and investigated. Existing knowledge about continuous methods will be exploited, whenever this is advantageous, to compare and transfer related technologies from continuous finite element and isogeometric methods to DG approximations and vice versa. For example, in contrast to DG methods, which allow for maximal discontinuity between the elements, isogeometric methods are based on maximal smoothness between the elements. Efficiency of isogeometric methods will be improved in a hybrid patch-wise approach. Within the patches maximal smoothness is used, while in between a discontinuous approach provides mesh flexibility.