Thesis

Advanced control of switched reluctance machines and electrical machine emulation

Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 2025
Thesis identifier
  • T17336
Person Identifier (Local)
  • 202154848
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • As society electrifies, there is a growing demand for electrical drive systems in the deployment of renewables and electric vehicles (EVs), along with the replacement of other traditional engine and hydraulic systems in order to reduce global emissions. For any drive system, there is a desire that the ideal candidate would have high efficiency, high power and torque density, low cost, fault tolerance and be relatively environmentally friendly. There are multiple electrical machine topologies in common use for these applications, such as the permanent magnet synchronous machine (PMSM), induction machine, synchronous reluctance machine and the switched reluctance machine (SRM). The SRM has received interest in recent times, especially in the electrification of passenger vehicles. It is an attractive machine topology, given it is low cost, simple design, fault tolerance, high torque density and no use of permanent magnet materials. Despite this, the SRM suffers from a lower efficiency compared to other candidate topologies and has an inherently high torque ripple (hence noise and vibration), which limits its use across a wider range of applications. In an SRM drive system, the drive converter can be a source of efficiency improvements. Dependent upon the device type used, the cost of the converter in terms of components price and volume can vary while also affecting the overall efficiency. An investigation is carried out into the power semiconductor devices used in SRM drives, where commonly used switching device types with near identical ratings are compared from a theoretical perspective, and then experimentally compared to gauge device losses. From this it is found that for the three variants (and models) of switching device, the Superjunction MOSFET outperforms both the Silicon Carbide MOSFET and Silicon IGBT in terms of losses in a limited use case scenario. Another source for efficiency improvements and the main source of the elimination of torque ripple is the control of SRMs. Using the torque ripple minimisation strategy of current profiling as a starting point, the theoretically optimal rms current for an SRM phase is established for a given load torque. A genetic algorithm is designed which uses this rms current as a target for optimisation, which produces optimally low rms current profiles which across the full rated speed range of a four phase SRM, an increase in rms currently only 4.3% above the theoretically optimal rms current is exhibited. Along with this, the algorithm design eliminates commutation torque ripple (<1%) across the full rated speed range of an SRM by utilising intentional three-phase or more overlap. This results in a current profiling control scheme which can apply this efficiency improvement and commutation torque ripple elimination to any SRM with overlapping phase torque capability. Along with improvements to drive systems, the design processes which create them can also be improved in terms of their testing. Typically, the verification of control software can be done using hardware-in-the-loop or dynamometers, while full power verification is only carried out using a dynamometer. Dynamometers have many known drawbacks such as being complex mechanical and electrical systems which are difficult to use and expensive. A potential alternative to dynamometers, known as power electronics based machine emulation, can be used to provide similar levels of verification using a fully electrical setup. A power electronics based emulator is proposed which aims to be an attractive testbench for full power verification in a commercial drive system design environment. The testbench maximises efficiency in terms of power supply usage being 2.7-3.5 times more efficient than an equivalent dynamometer, it is also modular and easily fabricated and uses a microcontroller architecture. This provides a significant cost saving compared to other power electronic emulators which typically use real time simulator (RTS) units which cost greater than £20000 at a minimum compared to Microcontrollers which can cost as little as £5 in this case. Along with this, ease of construction and use with no modification of commercial inverter circuitry required. The testbench is simulated using MATLAB/Simulink and is then verified experimentally by comparison against a dynamometer testbench, where the drive control inverters in each testbench are programmed with commercial sensorless position control, and tested with power levels up to 3.2kW in the three phase loop
Advisor / supervisor
  • Ahmed, Khaled H.
  • Williams, B. W.
Resource Type
DOI

Relations

Items