System model generation of traction machines using JMAG
Electromagnetics
PRESENTER:
Stefan Skoog
EXPERTISE:
Electromagnetics
Software:
JMAG
This presentation demonstrates a methodology on how to convert a single-case-setup advanced JMAG FEA-model to a high-fidelity system model ready to be used in commercially available system modelling tools such as Matlab/Simulink. The target application include advanced traction machines for road vehicles, where the typical usage is very dynamic in the speed-torque domain, and typically at least two quadrants of operation is expected in system modelling. Advanced FEA-modelling includes the analysis of higher-level physical phenomena such as slot harmonics and eddy currents in most parts of the machine. Axial Flux Machines with Soft Magnetic Composites is one example covered in this demo. In order to accurately capture the dynamics of several independent non-linear loss factors within the machine, the entire speed-torque operating region should be analysed through FEM. A full 3D FEM is typically associated with long simulations times even for single cases. This presentation outlines framework on how to automate pre-processing of a JMAG-model to enable accelerated analysis through parallelization and solve multiple FEM-cases within a reasonable timeframe. A comprehensive post-processing framework is also presented, where multi-dimensional, non-linear regression is automatically applied on a batch of FEM-cases with the goal of compiling a high-fidelity system model of the machine performance. The post-processing is done via a stand-alone graphical interface and offers analysis of individual FEM-cases, visualization of the machine performance over the entire operating region, and powerful tools to perform maximum-efficiency mapping including advanced losses.
