Loss and thermal modeling of hybrid NRMM electric drives

Abstract

This thesis develops and experimentally validates a fast-running modeling chain that couples quasi-static loss models with lumped-parameter thermal networks (LPTNs) for electric drive subsystems in hybrid non-road mobile machinery (NRMM): a permanent magnet synchronous machine (PMSM), an induction machine (IM), and a two-level three-phase inverter. Loss models are validated separately, showing mean relative errors of 5.83 % for PMSM efficiency, 3.99 % for IM efficiency, and 1.98 % for inverter input power. Four LPTNs are implemented for different cooling architectures (stator-only water-jacket, direct liquid cooling, fanless, forced-air) and validated against experiments or reference models. The stator-only PMSM model predicts slot temperature within 5.56 % under IEC duty cycles S1–S3. The fanless IM model achieves 2.73 % to 2.74 % error for stator slot temperatures under sinusoidal and PWM supply. The DLC PMSM model agrees well with the stator but overestimates magnet temperature (16.6 %). The inverter thermal model shows 1.74 % error with resistor losses and 11 % to 11.4 % with SiC MOSFETs. The proposed chain provides temperature estimates in the millisecond range with sufficient accuracy for preliminary cooling system design of hybrid NRMM electric drives.