Magnetic field analysis of electric machines taking ferromagnetic hysteresis into account

This thesis deals with the magnetic field analysis of electric machines by means of the finite element method taking the ferromagnetic hysteresis into account. The hysteresis is considered through a vector Preisach model, consisting of scalar Preisach models distributed along a finite number of angular directions. The incorporation of the vector hysteresis model into a two-dimensional time-stepping field solution in terms of the magnetic vector potential is accomplished by the Fixed-Point iterative technique. A combination of the field formulations and circuit equations of the windings presents a general voltage-driven solution with hysteresis, applicable to a 2D analysis of any electric machine. The time discretization is performed by using the Crank-Nicholson algorithm and the rotation of the rotor is modeled by moving the finite element mesh in the air gap. The verification of the scalar hysteresis model has been performed by comparison to dc-field measurements. The vector hysteresis model has been validated indirectly by the computation of the hysteresis torque and the associated losses in a rotor structure of an induction motor. The computation of the power balance in a simplified structure has been used to accredit the accuracy of the presented numerical techniques. The method of analysis has been applied to the magnetic field simulation and core loss computation of three individual cage induction motors. The computations have been carried out for the motors running at synchronous speeds and supplied from a sinusoidal voltage source. The computed core losses have been compared with the measured ones, yielding generally acceptable results. The influence of the time-step size and the number of scalar models included in a vector model has been studied. Many computations with the presented method, which takes account of the ferromagnetic hysteresis already when solving the field, have shown that although the technique is rather slow, it is robust, reliable and always convergent. These findings are the most important results of the thesis.reviewe