Parallel and Explicit Methods for the Simulation of Eddy Current Problems

Project description

The development of efficiently designed electrical and electro-mechanical engery converters, e.g., transformers, electric motors, generators, actuators, eddy current brakes, etc., requires a detailed knowledge about the magnetic fields and fluxes in the respective devices. Numerical simulations of low-frequency electromagnetic fields and their interactions for computation and visualization (virtual prototyping) represent an essentiell and  cost-effective alternative to measurements.

In this research project, the fundamentals of noval numerical methods are to be developed or existing methods optimized in order to increase the efficiency of low-frequency magnetic field simulations. Goal of this project is to solve detailed models and the resulting large discrete problems in less time by exploting novel parallel proccessing architectures. In particular, the combination of discontinuous Galerkin finite elements methods including explicit time integration methods, such as the Runge-Kutta-Chebyshev scheme, are to be tested. This enables an outstanding utilization of upcoming multi-processing computer architectures, e.g., graphics processing units, since many parallel arithmetic operations can be executed with minimal data transfer.

Project-related publications