Design and Comparative Analysis of PM Traction Machines with Harmonic Current- Based Torque Ripple Reduction for Electric Truck Applications

Abstract

Permanent magnet traction machines for electric truck applications must combine high torque capability, wide speed operation, high efficiency, low material cost, and reduced dependence on rare-earth magnets. This thesis presents the design, finiteelement analysis, drive-cycle evaluation, sustainability assessment, and torque ripple reduction of three machine topologies: a hybrid-magnet permanent magnet-assisted synchronous reluctance machine (PMaSynRM), a ferrite spoke-type permanent magnet synchronous machine (PMSM), and a NdFeB V-type PMSM used as a reference. The comparison considers torque-speed and power-speed capability, field-weakening performance, loss and efficiency behaviour, magnetic saturation, demagnetization, mechanical rotor stress, material usage, cost, and environmental impact. The results show that all three machines can meet the required traction performance, but with different trade-offs. The NdFeB V-PMSM provides the most compact design and strong torque production, but it is the most expensive solution and has the highest CO2e per kilogram due to its strong dependence on NdFeB magnet material. The ferrite spoke PMSM reduces rare-earth magnet dependency and achieves the highest power capability in the field-weakening region by using ferrite magnets and flux concentration. It is also the cheapest machine and has the lowest CO2e per kilogram, but its larger and heavier structure leads to higher total emissions and increased core losses, especially at higher speeds. The hybrid PMaSynRM, which uses both ferrite and NdFeB magnets, reduces NdFeB usage while maintaining competitive torque-speed performance, low drive-cycle losses, and high average efficiency. Overall, it offers the best compromise between electromagnetic performance, rareearth magnet reduction, cost, efficiency, and sustainability. In addition to the motor design comparison, this thesis investigates steady-state torque ripple reduction using a map-based reduced-order model. A field-oriented control structure with feedforward voltage compensation and LQI current control is implemented, including digital controller sampling delays and zero-order hold effects. A linear time-varying Kalman filter is used to estimate flux components and torque, while a harmonic current optimizer modifies selected current harmonic components to reduce torque ripple without causing drift in the mean current or mean torque. The proposed optimizer achieves torque ripple reduction of up to 70-80% at steady state while respecting inverter switching constraints. The Kalman filter accurately estimates fluxes and mean torque but cannot fully reconstruct torque ripple caused by cogging and slotting effects.