Mini Symposium Details

16: Reduced Order Models for Multiscale and Multiphysics Problems

• Felix Fritzen, Efficient Methods for Mechanical Analysis, Institute of Applied Mechanics (CE), University of Stuttgart
• Ralf Jänicke, Division Material and Computational Mechanics, Chalmers University of Technology

The minisymposium addresses computational aspects of challenging engineering applications in technology and science. It is dedicated to strongly heterogeneous materials with multiple length scales and multiple interacting phases. The computational complexity of those problems may be driven by the consideration of real-data, for example arising from a X-Ray Computed Tomography or Field Ion Beam Tomography analysis, and/or by electro-, magneto-, hydro- or chemo-mechanical coupling phenomena.

Computational two-scale approaches employing a FE model at the RVE level for each point of the macroscopic problem have several limitations. First, the computing time is intense and the memory requirements are enormous. Second, systematic short-comings for nonlinear problems such as high number of iterations can occur and the robustness requirements for the RVE solver in nested finite element methods are demanding.

Reduced order models can help in several of the aforementioned aspects. They can reduce the computing time and the memory requirements on the one hand while they can also lead to new problem formulations on the other hand. Thereby, novel solution schemes can emerge as well as physical coupling can sometimes be understood in better way in the reduced framework.

Other promising applications for reduced order models are found in the many query context, e.g. for uncertainty quantification (UQ). The minisymposium welcomes contributions from various domains of model order reduction related to multiscale and multiphysics problems.