Dissertation – Leonardo Colombo
Thesis title: “Iron Losses, Measurement and Characterization in Stator Cores”.
Summary of the thesis
Driven by the global transition towards energy-efficient electrical machines, iron loss computation remains a complex challenge. Because material composition, manufacturing, and mechanical stress heavily influence ferromagnetic behaviour, accurate prediction is notoriously difficult. This thesis advances the field by investigating soft magnetic stator cores and related measurement techniques.
Statistical comparisons of different measurement techniques on specimens from the same batch reveal statistically significant differences. Tests on custom-manufactured toroidal cores reveal approximately frequency-independent core loss build factors up to 400 Hz, when compared to the material datasheet. The research also presents a custom-built core loss tester using a switch-mode, rather than linear, power amplifier, enabling higher-power, stable PWM excitation of inductive loads without impedance matching. It supports both sinusoidal-flux and PWM tests, achieving accuracy comparable to a commercial system for hysteresis loops and total core losses.
Experimental data is mapped using the Jiles-Atherton (JA) hysteresis model, optimized across multiple induction levels rather than a single saturation point as commonly found in the literature. While a dynamic JA formulation (integrating Bertotti’s loss separation) yields superior accuracy, the original magnetostatic formulation remains remarkably viable above quasi-static frequencies. Finite Element Method (FEM) simulations integrating the static JA model show strong agreement with the analytical optimization results. The application of the JA model in simulating electrical machines is exemplified by analyzing the peak-torque operating point of a permanent magnet machine at a low excitation frequency. For this case, a modified JA formulation capable of accurately modelling deep saturation magnetisation characteristics is introduced, yielding good agreement with alternative modelling approaches and aligning with the machine’s nominal datasheet values for peak torque. Insights on the influence of multi-level Pulse-Width Modulation (PWM) on iron losses are presented via reconfiguration of the in-house core loss tester to emulate cascaded H-bridge converters. Increasing the amplitude modulation ratio and switching frequency effectively dampens PWM-induced losses. Most notably, as the number of PWM voltage levels increases, characteristic minor hysteresis loops vanish, transforming into a high-frequency ripple. This confirms the in-house setup as a highly effective tool for initial investigations of how different converter architectures impact the losses in soft magnetic materials.