Impact of Flux Barriers Type and Parameterization in Reluctance Machine Torque Optimization

De Gréef, Christophe [UCL] Kluyskens, Virginie [UCL] Henrotte, François [UCL] Geuzaine, Christophe [Montefiore Institute] Dehez, Bruno [UCL]

This paper analyzes the impact of the flux barriers (FB) type and their parameterization on the torque optimization of synchronous reluctance machines. Four different types are considered: rectangular FB, round FB, hyperbolic FB, and finally Joukowski FB, which follow the natural flux lines in a solid rotor. A parameterization that simplifies the non-overlapping constraints is introduced, and, for each of these types, the num- ber of parameters is discussed. A multi-objective optimization algorithm is run to maximize the mean torque while minimizing the torque ripple. Evaluations are performed by a 2D nonlinear finite element magnetic model. The obtained Pareto fronts show that the optimization is able to reduce the torque ripple down to about 4% for all the types, with little impact on the mean torque. The round FB have the highest mean torque and a better convergence speed than rectangular and hyperbolic FB, whereas the Joukowski FB, with half as many parameters, reaches 98% of the round FB torque. Also, the proposed parameterization reaches the optimization convergence up to 3.5 times faster than a conventional parameterization, for all the FB types