Search

This thesis presents two different non-permanent magnet machine designs for belt-driven integrated starter-generator (B-ISG) applications. The goal of this project is to improve the machine performance over a benchmark classical switched reluctance machine (SRM) in terms of efficiency, control complexity, torque ripple level and power factor. The cost penalty due to the necessity of a specially designed H-bridge machine inverter is also taken into consideration by implementation of a conventional AC inverter. The first design changes the classical SRM winding configuration to utilise both self-inductance and mutual-inductance in torque production. This allows the use of AC sinusoidal current with lower cost and comparable or even increased torque density. Torque density can be further increased by using a bipolar square current drive with optimum conduction angle. A reduction in control difficulty is also achieved by adoption of standard AC machine control theory. Despite these merits, the inherently low power factor and poor field weakening capability makes these machines unfavourable in B-ISG applications. The second design is a wound rotor synchronous machine (WRSM). From FE analysis, a six pole geometry presents a lower loss level over four pole geometry. Torque ripple and iron loss are effectively reduced by the use of an eccentric rotor pole. To determine the minimum copper loss criteria, a novel algorithm is proposed over the conventional Lagrange method, where the deviation is lowered from ± 10% to ± 1%, and the simulation time is reduced from hours to minutes on standard desktop PC hardware. With the proposed design and control strategies, the WRSM delivers a comparable field weakening capability and a higher efficiency compared with the benchmark SRM under the New European Driving Cycle, where a reduction in machine losses of 40% is possible. Nevertheless, the wound rotor structure brings mechanical and thermal challenges. A speed limit of 11,000 rpm is imposed by centrifugal forces. A maximum continuous motoring power of 3.8 kW is imposed by rotor coil temperature performance, which is extended to 5 kW by a proposed temperature balancing method. A prototype machine is then constructed, where the minimum copper loss criteria is experimentally validated. A discrepancy of no more than 10% is shown in back-EMF, phase voltage, average torque and loss from FE simulation.