Desenvolvimento de nova metodologia para análise de desempenho de máquinas elétricas : uma contribuição ao método Maxwell-Fourier

Abstract

The advancements toward the energy transition, in which the energy sector is switching from a fossil fuel-based energy matrix towards a renewable solution, is establishing a favorable environment for technological development. This scenario brings forth a window of opportunity for the electrification of the transport, industry and building sectors. Therefore, it is clear the need to improve the design procedures of conventional electrical machines in order to meet more stringent efficiency requirements. Furthermore, the advancement of the electric drives into new applications requires further attention in the development of materials and new topologies that allow for a balance of cost and performance requirements. This increases the demands for the mathematical tool adopted to analyze and design such electrical machines. In this thesis, the use of the Maxwell-Fourier method is discussed. Based on the formal solution of Maxwell's equations, this method is suitable to characterize the electromagnetic performance of a wide range of electrical machine topologies. Firstly, this tool is used to analyze a high-speed synchronous machine. To characterize the torque oscillations, it is proposed a model capable of representing saturation in the retaining sleeve, an element used to ensure magnet mechanical integrity. The characterization of this phenomenon can help in the development of generator control strategies, as well as in future projects that wish to explore it. Then, the Maxwell-Fourier method is applied to design a vehicle regenerative braking generator. In this case, a design procedure is established to characterize the machine in electromagnetic and constructive terms. Thus, the proposed solution offers the designer greater sensitivity for evaluating the effects of the machine's parameters on its performance. Also, the procedure aims to reduce the search space of the problem to be addressed in later design stages, in which numerical methods are used with increased computational cost.