Magnetic levitation has emerged as an attractive alternative to conventional mechanical bearings in high-speed rotating machines, as it eliminates contact-induced friction, wear, and thermal stresses. However, most levitated systems rely on active magnetic bearings, which require multiple sensors, power electronics, and control loops, leading to increased complexity and volume. To overcome these limitations, electrodynamic suspension offers a promising approach by exploiting the interaction between permanent magnets and motion-induced currents to generate restoring forces. This principle is implemented in the electrodynamic thrust self-bearing machine (EDTSBM), which combines passive axial levitation and torque production within a single compact structure.
This thesis advances the understanding and development of the EDTSBM through modelling, control, and design contributions. An extended electromechanical model was developed, accounting for position-dependent inductance variations and nonlinear flux evolution, thereby revealing additional force components and the influence of current injection on suspension behaviour. Building on these findings, novel stabilization strategies were proposed, including active damping through current injection and sensorless estimation of rotor axial position and velocity, enabling stable operation over the entire speed range without dedicated damping system. Finally, the thesis investigates alternative rotor topologies incorporating magnetic circuits and establishes a unified framework for their design and comparison.
Membres du jury :
Prof. B. Dehez (UCLouvain) - Promoteur
Prof. N.Moës (UCLouvain) - Président
Prof. E. De Jaeger (UCLouvain) - Secrétaire
Dr. J. Van Verdeghem (UCLouvain)
Prof. M. Bekemans (UCLouvain)
Prof. E. Severson (University of Minnesota)
Prof. W. Gruber (Johannes Kepler University of Linz)
Soutenance publique également accessible par visio-conférence via le lien (TEAMS) :
https://teams.microsoft.com/l/meetup-join/19%3ameeting_NzFlZDBmNTctOWRjNC00YWRlLTg4NWYtM2EzNjk2NWMzNGJl%40thread.v2/0?context=%7b%22Tid%22%3a%227ab090d4-fa2e-4ecf-bc7c-
