Search

Switched reluctance motors can be driven in a conventional or a modular way. In the conventional driving method, a number of stator coils are connected together and driven from one converter module. In the modular driving method, each stator coil is driven from one separate converter module. The conventional driving method provides better drive performance (i.e. efficiency, torque ripple, and power density) but narrower speed range at rated torque and less fault-tolerance capability. The modular driving method provides wider speed range at rated torque and much better fault-tolerance capability. Therefore, in low speed healthy operation, the conventional driving method is preferable while for high speed and faulty condition, the modular driving method should be used. In this article, a converter topology that can be reconfigured to operate in conventional and modular driving ways is proposed and analyzed. Using the proposed converter, the benefits of the conventional and the modular driving methods can be obtained by proper configuration of the converter. Also, a fault-tolerance strategy that results in symmetrical and high-performance operation of the drive is proposed. The performance of the converter is assessed by simulations and experimental measurements.

Integrated modular motor drives (IMMDs) have a compact and high fault-tolerant structure compared to their conventional counterparts. The most challenging issue in the implementation of these drives is the thermal management of the inverter modules due to the small space available for each module and its physical proximity to the machine windings with their high heat production. The electrothermal design of a discrete half-bridge inverter module for a radially stator-mounted (RSM) axial flux integrated drive is introduced in this article. The inverter modules are mounted on the stator, radially along the circumference. The converter is implemented using GaN technology, and the size is optimized. The structure of the presented integration topology from the case of the switches to the cooling ambient is optimized from the thermal point of view. The converter is modeled using steady-state and transient lumped parameter thermal networks to estimate the steady state and the instantaneous temperature of the switches. The converter parasitics are extracted using FEM simulations, and their influence on the switch loss and temperature is evaluated. The lumped parameter modeling is validated with computational fluid dynamics (CFD) simulations and experimental measurements.