Advanced High-Power-Density Disc-Type Permanent Magnet Synchronous Machines for Aircraft Electric Propulsion


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date2 Aug 2021


Air transportation is a vital sector from numerous industrial applications in our society. It promotes that aircrafts not only regard electrical power as the secondary energy source, but also integrate electricity into propulsion systems.

Research on electric propulsion aircraft has been welcomed due to its technical and social merits of simple power architecture, low carbon emissions, environmentally friendly, and sustainability. Hybrid electric propulsion aircrafts reduce the power burden of aviation engines, while full electric propulsion aircrafts expect to replace engines with electric machines. Incorporating electrical energy into propulsion system, high-performance propulsion electric machine becomes the core component. This thesis proposes and investigates two novel brands of disc-type permanent magnet synchronous machines (PMSM), namely the slotless disc-type (SLDT) PMSM and radial-flux disc-type (RFDT) PMSM solution. The specific research work includes the aspects of machine topology design optimization, analytical modeling, winding analysis, high-performance control strategies, exact modeling in multi-physical domains.

Firstly, this article reviews the energy architectures of different aircrafts, analyzes the features and prospects of electric propulsion aircrafts based on commercial data and state-of-art, and explores the potential of emerging propulsion machine techniques.

Secondly, in the light of the features of SLDT-PMSM, a complete analytical model with respect to electromagnetic field is proposed, which does not rely on numerical computation. It provides guidance for the fast design of this type of machine. The analytical model can be used to accurately predict performances such as field distribution, no-load back-EMF, load torque, inductance, flux linkage, etc. Not only is this model suitable for various sizing, but also for different magnetization patterns and winding configurations. By measuring magnetic flux density and performance parameters of the slotless disc-type machine prototype, the model accuracy is verified. Besides, based on the two-level inverter and space vector control strategy, this thesis proposes a novel voltage modulation scheme and designs a deadbeat-based controller to actively suppress the current ripple of prototype with a multi-phase design. The control strategy under different working conditions and parameter variations is experimentally validated. It is also applicable to other multi-phase motors.

Thirdly, the thesis draws up a comprehensive design optimization methodology for RFDT-PMSM, including tooth coil winding analysis, parametric design analysis, optimization design analysis, and robust design analysis. Also, this paper proposes a novel model predictive optimization strategy that incorporates advanced machine learning algorithms for model construction. This strategy makes great contribution to reducing the calculation burden in optimization process and attaining human-machine collaborative design.

Finally, based on the RFDT-PMSM and the prototype, the thesis focuses on the electromagnetic force and vibration characteristics of stator mechanical system in the structural field. Methodologies of multi-physical-domain analysis together with five multi-physics models are proposed with consideration of accuracy and complexity. The following part provides qualitative and quantitative discussions of air-gap field calculation and exciting force calculation, by air-gap permeance model and magnetic reluctance network model, respectively. Both field prediction and force profiles are verified by FEA tools. Following the introduction of force mapping and characterization of inherent vibration patterns, the last part implements electromechanical vibration analysis and compares the results of different models. General guidelines of multi-physical-domain analysis will be eventually presented.