Scalable Parameterisation of a Lumped Parameter Thermal Network Applicable for Multiple Electric Motor Variants for EV Applications

Abstract:
Estimating the temperature of permanent magnets is vital to generating high-accuracy torque by adapting the permanent magnet (PM) flux in Electric vehicle (EV) applications. Prior research explains the consequences of neglecting PM temperature estimation, followed by the PM temperature estimation methods. This paper investigates the scaling laws and describes their application on lumped parameter thermal networks (LPTN). When performing a geometrical scaling of machine design to determine its performance, especially with known parameters of LPTN for the reference motor variant, an individual can calculate LPTN parameters for the scaled-motor variant without investing much time and effort. Fundamentally, it consists of multiple and crucial scaling procedures: rewinding, axial and radial scaling. This paper considers only axial scaling for two motor variants of the same outer diameter. Additionally, it highlights model order reduction (MOR) methods using three different orders of LPTN. Simulation results from experimental data reveal that the proposed theory satisfies the preconditions for practical implementation. Moreover, it overcomes the cost, time, and drawbacks of implementing an LPTN for each motor variant separately by providing realistic and acceptable results.