Evaluation of Real-Time Capable Thermal Simulation Models for Electrical Machines

This study investigates two advanced methods to derive real-time-capable thermal simulation models represented in state space and compares them in terms of their computational effort, result accuracy, and applicability. The first method involves an enhanced lumped-parameter thermal network using generic elements with an inner T-structure, while the second method utilizes finite element analysis and model order reduction techniques. Both methods are applied to analyze the thermal behavior of a stator tooth of an electrical machine with a single-tooth winding. It is found that both methods can provide sufficiently accurate temperature predictions with only minor differences in the computational effort, thus the advantages and disadvantages must be weighed depending on the specific application and available information.