Élaboration de méthodes et d'outils logiciels pour l'optimisation topologique magnéto-mécanique de machines électriques tournantes

In the context of energy transition and the electrification of applications, improving the performance of electromagnetic actuators inevitably involves dimensioning optimization processes. Such methodologies have already been implemented but focus mainly on previously parameterized geometries, which limits the space of possibilities. This thesis aims to develop an efficient topological optimization methodology capable of optimizing the distribution of materials (iron, air, conductors, magnets) required to generate a synchronous machine in its entirety without parameterizing its geometry. To this end, a multi-material density topological optimization methodology has been developed. Its application to optimizing a three-phase stator highlights the importance of penalization, filtering, and control processes in the optimization algorithm. The procedure is then extended to the design of an entire machine: although efficient, the best structures obtained include flux barriers with no mechanical strength. After incorporating rotor stiffness constraints, the method produces high-performance, related structures in a reasonable computation time, demonstrating the relevance of this type of approach to the design of electromagnetic actuators. Eventually, integrating all the physics involved in specifications right from the preliminary phases will save time and money in designing innovative electrical machines.