Multi-Objective Optimization Design of Permanent Magnet Vernier Motor for Direct-drive In-Wheel Electric Vehicle Powertrain

Abstract

With the increased awareness of pollution caused by Internal Combustion Engine (ICE) based vehicles, attentions have started towards Electric Vehicles (EVs) and Hybrid Electric Vehicles (HEVs). Unlike their predecessor, EVs rely on electric motors and energy storage units to drive the vehicle rather than ICE units and gasoline. This eradicates the hazardous gasses emission and has given rise to several research opportunities such as choosing the proper size of motors and their torque capability, deciding the capacity of the battery, studying the performance of the motors under different drive cycles and improving the efficiency of the powertrain. One recent solution that was suggested to improve the efficiency of the powertrain and overall performance of EVs lies in the concept of In-Wheel Motors (IWMs). As the name suggests, motors are placed within the wheels’ rims instead of using motors connected to the wheels through shafts. This reduces the mechanical components needed in the powertrain, allows better utilization of the available space and leads to an increase in efficiency and a reduction in mechanical wearing.

In recent years, Permanent Magnet Vernier Machines (PMVMs) have received increased interest due to their distinctive features such as high power density, high torque density, low torque ripple and the inherited magnetic gearing phenomena, which make them excellent candidates for various applications, especially for low-speed high-torque applications. The concept of PMVMs arises from Magnetically Geared Machines (MGMs). An MGM consists of a conventional electric motor with magnetic gearing capability integrated into it. The magnetic gearing is often accomplished by using a number of flux modulator steel blocks, which by properly selecting their number allows the gearing to occur. In PMVMs however, there is no need for the additional flux modulators, as the teeth on the stator are designed such that they perform a similar modulation. Hence, reduction in the complexity of the design is achieved. PMVMs can therefore be viewed as conventional Permanent Magnet Synchronous Machines (PMSMs) that obey certain conditions which enable the inherited magnetic gearing effect to occur. This trait qualifies them to be potential candidates for in-wheel motor application, in which the mechanical gears can be eliminated. One main drawback of PMVMs however, is the relatively low power factor due to large leakage in the magnetic flux and high inductive reactance; which stresses the power electronic drive circuits and raises the need for a bigger rated electric drive. A rigorous and comprehensive machine design procedure is therefore needed to overcome this challenge.

The main goal of this dissertation work is to design a high-performance in-wheel motor for an electric vehicle based on a PMVM design that scores highly on competing performance metrics such as torque density and power factor. This is accomplished by developing a detailed, comprehensive, computationally efficient and novel closed-form analytical model of a surface mounted (SM) PMVM that studies the many tradeoffs involved and captures with good accuracy the key performance metrics of the machine. The optimum machine design is then searched for through a rigorous and evolutionary multi-objective optimization design formulation, based on Genetic Algorithm (GA), with a large design space. This is done while taking into account a number of geometrical, electrical and magnetic constraints involved in in-wheel motor application.

The results achieved demonstrate the potential of achieving a design with relatively high power factor and torque density while still maintaining a simple machine structure that is easier to build compared to more complex designs suggested in literature. Validation of the electromagnetic performance and important mechanical constraints for in-wheel application is carried out via Finite Element Analysis (FEA), with experimental validation is carried out to validate some part of the design algorithm.

Date of Award	Dec 2022
Original language	American English
Supervisor	Jamal Alsawalhi (Supervisor)