Conference Agenda

In this paper, a mesh-based magnetic equivalent circuit (MEC) modeling technique is developed for induction machines (IMs) under healthy and broken rotor bars conditions. This model is capable of obtaining the electromagnetic torque, rotational speed, and forces precisely, in various conditions, compared to a finite element model in various conditions. The air gap coupling is based on the Lagrange multiplier interpolation function between the stator and rotor flux densities.

The results show good accuracy of the MEC model for predicting the speed oscillations of the faulty IM in addition to the forces computation and the current harmonics. Moreover, the speed of the MEC model is more than 50 times faster than that of the FE.

In this paper, a second order approximation is proposed for nonlinear quasi-static eddy-current problems via parametric Cauer Ladder Network (CLN) method. Through the CLN method, an orthogonal sequence of electric and magnetic modes along with the equivalent circuit parameters are generated by magnetostatic finite element analysis. When there are nonlinear materials in the medium, the modes and their corresponding values in equivalent circuit may vary according to the core's saturation level. This paper introduces a new algorithm to generate second-order CLN (SO-CLN) based on intensity of the first and second magnetic modes. Additionally proper handling of circuit equations is also included. Numerical tests are carried out over a 2-D nonlinear inductor with bulk type conductive magnetic core to show the accuracy of the proposed nonlinearization method.

This paper presents a 3D magnetodynamic model for accurate magnetic loss calculation in laminated electrical machines, especially high-speed ones. In this model, a vector hysteresis model is employed to characterize the magnetic hysteresis effect in ferromagnetic materials. The excess field is described using a viscosity-based phenomenological model. A novel 3D formulation is proposed to model eddy currents including the skin depth across the lamination thickness. In this formulation, the governing electromagnetic equations are simplified by applying the thin lamination approximation to satisfy the coupling conditions in the strong form of the equations. The finite Element Method (FEM) is applied to solve the developed model. Finally, the model accuracy is validated by comparing the simulated magnetic losses to experimental data measured on a three-limbed magnetic core at a wide frequency range, where a maximum discrepancy of 5% is achieved.

A novel 3-D fixed-point harmonic-balanced finite element method with edge-based elements is presented and applied to nonlinear problems. This method is based on A, φ-A formulation with the relaxation factor to guarantee the convergence of the nonlinear computation. A new fixed-point magnetic reluctivity method, which selects the DC component of reluctivity as the fixed-point reluctivity, is also proposed and investigated under practical scenarios. By implementing a silicon laminated core model working under DC-biased conditions and a typical eddy current model, the effectiveness and efficiency of the new method can be verified. The numerical behavior is investigated exhaustively through the comparison with measurement and calculation results.

The numerical simulation of nonlinear magneto-quasistatic problems based on the Finite Element (FE) method can lead to huge computational times. Then, Model Order Reduction (MOR) approaches based on the Proper Orthogonal Decomposition (POD) coupled with an interpolation method enables to reduce the computational time, but can lead also to numerical instabilities. In this communication, we propose a methodology based on the Gaussian Newton Augmented Tensors (GNAT) method to construct a reduced model of a squirrel cage induction machine in the nonlinear case. The reduced model is evaluated in terms of accuracy on global and local quantities of interest and speedup, compared to a FE model.