The goal of the project seminar is to study software for the simulation of electromagnetic field problems. Based on a few typical model problems, we want to assess the flexibility and performance of different methods and to study the relevance of, e.g., geometric modelling, mesh generation, adaptive mesh refinement, higher order approximations, and further advanced features.
Meetings
Problem 1: Inductance Computation
A detailed explanation of the problem can be found at www.femm.info/wiki/InductanceExample. A 3D version of the problem is available at gitlab.onelab.info/doc/models/-/wikis/Inductor. The goal of this example is to get familar with geometric modelling and finite element simulation, the formulations of 2D / 3D magnetostatics, and how to compute typical quantities of interest. This example might also serve to evaluate the effect of reduction to 2D, the truncation of the computational domain, the effect of corner singularities, etc.
Problem 2: Nonlinear Eddy current problems
We want to assess the eddy current losses in a bundle of magnetically conducting wires. The alternating fields/fluxes are generated by a copper winding. This problem can be solved in 3D/2D/1D. The goal of this tutorial is to get familar with linear/nonlinear eddy current problems and the experimental/numerical computation of hysteresis curves. Details about this example will be provided by Theodore.
Problem 3: Torque Computation
As a third test problem, we consider the computation of torque and other relevant quantities for a typical model of an electric machine. Details about the setup can be found at www.femm.info/wiki/RotorMotion. A key ingredient here will be the efficient handling of nonlinear materials, of moving meshes, and the torque computation.
A sequence of further typical test problems can be found at www.femm.info/wiki/Examples or gitlab.onelab.info/doc/models/-/wikis/home. Another source of more advanced test problems is www.compumag.org/wp/team/
Software
In addition to individual implementations, e.g., in Matlab/Python/C++ we also want to study and compare freely available software, e.g.
The idea is to group up in small teams and try to solve efficiently and reliably the test problems mentioned above. Every team should consider all test problems and also consider different software for their solutions. Apart from the specific solution of the problems, we are, in particular, interested to learn about pros/cons of the different approaches and tools available. In addition, we want to learn about the typical workflows, quantities of interest, etc. arising in different application contexts.