Topology optimization of high-performance passively levitated bearingless machines

Details

Supervisors

Dehez, Bruno

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en génie de l'énergie , à finalité spécialisée

Abstract

enElectrodynamic Thrust Self-Bearing Machines (EDTSBMs) achieve fully passive magnetic levitation by using the electrodynamic forces generated by the interaction between permanent magnets and armature windings. However, the stator and rotor topologies currently employed in such machines remain relatively conventional, leaving potential for performance improvement through more advanced designs. This thesis applies topology optimization methods for the design of new, more efficient stator and rotor topologies for EDTSBMs, with the optimization objective aimed exclusively on the machine's bearing function. A first topology optimization was performed on the stator, using three density variables each representing the forward winding, the return winding, and the ferromagnetic material. Multiple stator design zones and rotor configurations were considered to assess the impact of ferromagnetic material on the bearing function. The results consistently showed that introducing ferromagnetic material into the stator provides no benefit, regardless of the configuration considered. Ferromagnetic material was therefore excluded from further stator designs. Based on the optimized winding density distributions, feasible winding topologies were identified and optimized using parametric optimization, which confirmed that certain winding configurations significantly outperformed others in terms of bearing stiffness. A second topology optimisation was defined on the rotor, by modelling the permanent magnets, the back iron yoke, and the ferromagnetic yoke surrounding the armature as independent density variables. The results demonstrated that the spatial arrangement of rotor components can be optimized to improve the electrodynamic stiffness of the machine. Feasible rotor topologies were then identified and validated through parametric optimization for each configuration studied. Finally, a combined parametric optimization, which simultaneously optimized the geometric parameters of the stator and rotor, was performed. This combined approach demonstrated that simultaneously accounting for the interaction between the stator and the rotor resulted in a further improvement in bearing performance, beyond what is achievable through sequential and independent optimizations.