Efficiently Modeling Soft Magnetic Materials for Transformers, Actuators and Rotating Electrical Machines

The assemblage of individual soft magnetic laminations constitutes the magnetic core of an electrical machine, which is tailored to guide and amplify the magnetic flux distribution. This why the soft magnetic material significantly influences the operating characteristics and efficiency of electrical machines.

The complex flux waveforms in electrical machines exacerbate the development of appropriate methods for predicting the iron loss and the magnetic field distribution. There exist so far no macroscopic material model that considers simultaneously vector hysteresis, the laminated structure and the issue of parameter identification. As a solution, an energy-based vector hysteresis model with magnetic anisotropy and different methods to solve the diffusion problem inside the laminations are developed. This so-called "semi-physical lamination model" allows one to replicate the magnetization behavior of soft magnetic materials without limitation regarding the excitation waveforms by using quasi-static measured data for parameter identification. This is verified for sinusoidal magnetic polarization waveforms up to 1 kHz and for non-sinusoidal and PWM-like waveforms.

Direct incorporation of the model into 2-D FEA of electrical machines is rather technical and cumbersome and makes it necessary to modify significantly the FE code. For this purpose, a pragmatic two-step homogenization approach is developed, which describes the magnetization behavior averaged over the cross-section. This method turns out to be quite accurate and efficient in practice, and entails no implementation in the FE code.

Application examples of magnetic field and loss calculation using the homogenized model in a three-limb transformer and a PMSm conclude this thesis.