Prototyping of electrical machines in 3D is becoming increasingly important. It offers more design options for optimization. It also allows the simulation of more complex configurations. The workload grows tremendously when shifting from 2D to 3D simulations. This motivates the use of scalable domain decomposition approaches to divide and conquer the resulting large-scale problems.
New concepts for the simulation and optimization of electrical machines are studied in the German-Austrian Collaborative Research Centre/Transregio CREATOR – Computational Electrical Machine Laboratory (TRR361/F90). Mario Mally’s latest work in a recent Preprint explores dual-primal tearing and interconnecting (TI) methods. These techniques are applied to magnetostatic problems in 3D using isogeometric analysis (IGA). This combines the strong modeling power and flexibility of IGA basis functions with scalable computations from the TI context. One focus of this work is gauging techniques. These techniques ensure the solvability of local sub-problems. They must remain consistent with the overall global problem formulation of the numerical model.