An Experiment-Informed Methodology for the Thermal Design of Permanent Magnet Electrical Machines

Abstract

The continuous drive towards more electric technologies and the subsequent need for high-performance electric machines calls for a more accurate and reliable thermal design-analysis, where various design, manufacture and assembly factors are accounted for at the initial stage of the design process. The wide ranges of built factors have a significant impact on the overall thermal behaviour of the machine, and frequently require empirical techniques to validate the initial design assumptions. The conventional approach employed in empirical validation of thermal models of electrical machines makes use of thermal testing on a complete machine prototype. This is usually performed at the end of the machine manufacturing process to assess the measured thermal characteristics against the design targets and theoretical predictions. Thermal model validation from a complete machine assembly is frequently effective for experienced-based designs, where a designer has confidence in the pre-existing machine thermal design. In such a case, the initial machine assembly should closely match the specifications. However, for novel machine topologies, a designer typically needs to evaluate several design variants to realise an optimal solution. The machine fabrication for all design iterations would not be feasible due to the cost and time constraints. For these reasons, a method providing reliable loss and thermal predictions would significantly reduce the number of prototype iterations required for the development of a new machine. Therefore, time and cost associated with the new design would be substantially reduced. This thesis presents an experimentally-informed methodology for the high-fidelity thermal design of electrical machines. The methodology focuses on the stator-winding region, which is frequently attributed with the dominant power loss component within the machine body. The developed approach accounts for a variety of permanent magnet machine designs, where the majority of the heat produced in the machine is evacuated by conduction into the machine periphery. Three machine sub-assemblies have been selected to form the foundation of the proposed methodology, based on their complementarity and their capability to incrementally inform the machine thermal design. Efficacy of the proposed methodology is illustrated on a number of machine demonstrators with different build attributes. The proposed methodology aims to be a systematic and reliable tool for machine designers, guiding empirical calibration depending on available hardware, required fidelity and machine design focus.

Date of Award	25 Sept 2018
Original language	English
Awarding Institution	University of Bristol
Supervisor	Rafal Wrobel (Supervisor) & David Drury (Supervisor)