Conference Agenda

This study focuses on the accurate numerical modeling and modal frequency analysis of a stator system in a permanent magnet synchronous motor (PMSM) to improve the prediction of electromagnetic vibration and noise. Accurate calculation of the stator’s natural frequencies is critical for understanding and mitigating NVH (Noise, Vibration, and Harshness) characteristics. To enhance the accuracy of these calculations, a high-fidelity finite element (FE) model of the PMSM was developed, incorporating the entire motor assembly and its support structure. The inclusion of the support structure allows for a comprehensive analysis of its influence on the system’s flexibility and vibration response. The FE model accounts for the anisotropic properties of the stator core and armature winding. Modal analysis was performed to determine the natural frequencies of the stator system and the support structure. To validate the computational results, experimental modal analysis was conducted on the same components. A strong agreement between the simulation and experimental results enabled the development of an updated FE model with high accuracy in predicting vibration behavior. This approach provides a reliable methodology for studying the coupled electromagnetic and structural dynamics in PMSMs.

In this work, we present a novel Finite Element Method (FEM)-based approach for modeling planar inductors using an equivalent two-dimensional (2D) formulation. The proposed methodology aims to significantly reduce computational costs while maintaining high accuracy in electromagnetic analysis. Traditional full three-dimensional (3D) models, though precise, are often computationally expensive and may become impractical for iterative design or optimization tasks. By leveraging the inherent symmetries and geometric properties of planar inductors, we derive an equivalent 2D model that accurately represents the electromagnetic behavior of the system, including inductance, losses, and field distributions.

The approach is validated through several case studies, comparing the 2D model's predictions against both full 3D simulations. Results demonstrate that the equivalent 2D model achieves computational savings of up to 90% while preserving an error margin below 5% in key performance metrics. This work provides a robust tool for the design and optimization of planar inductors, particularly in applications where computational efficiency is critical, such as in power electronics, RF systems, and wireless energy transfer.

The performance of Newton-Raphson method with A-formulation for superconductor applications is examined. While FEA with A-formulation has been one of the popular choices in today’s electrical designs, there is an argument that the combination of the A-formulation and Newton-Raphson has a convergence issue, especially for problems which consist of superconductors and non-superconductors with magnetic nonlinearity. Such a problem was solved with A-formulation. The results showed that although a bare Newton-Raphson method suffered from iteration deterioration, it was recovered by introducing a relaxation technique and showed enough performance for practical problems.

Topology Optimization (TO) presents a significant computational challenge due to its reliance on the design domain mesh, where finer discretizations allow for more detailed configurations but exponentially increase computational demands. This paper explores the application of Quantized Tensor Train (QTT) decompositions to mitigate these challenges in large-scale TO problems for electromagnetics. By efficiently approximating and compactly storing the dense matrices generated from the discretization of Integral Equation (IE) operators, QTT drastically reduces the computational burden. This innovative approach enables the practical use of high-resolution grids, opening new avenues for creating highly detailed and computationally efficient electromagnetic designs.

In reluctance machines, optimizing electromagnetic designs becomes more demanding with repeated evaluations of complex topologies for geometric parameter scaling. This paper presents a grey-box magnetic modeling approach for the electromagnetic scaling of variable flux reluctance machines (VFRMs). The proposed model integrates reduced-order finite element model (FEM) data into a semi-analytical harmonic model (HM) with mode-matching. This grey-box magnetic model combines the geometry-scaling flexibility

and computational efficiency of HM while accounting for magnetic saturation effects with FEM data. Gaussian process regression (GPR) is employed to create a reduced-order data merging via Fourier series (FS) coefficients of the airgap magnetic flux density. The resulting regression model serves as a post-processing tool for HM applied to various scaled geometries.

This paper proposes a method for reconstructing current sources with reduced computational cost. By employing current dot sensitivity analysis, the method eliminates the need for iterative finite element analysis (FEA) typically required in conventional approaches and enables direct current source design. To demonstrate the efficiency and robustness of the proposed method, a numerical example was conducted to non-invasively reconstruct current sources from magnetic flux density data.

This paper proposes an Incomplete Cholesky preconditioner with fill-in that improves the convergence of iterative methods and reduces memory usage for ill-conditioned symmetric linear equations. We developed an IC preconditioning where the fill-in level is a decimal and controls the number of fill-ins.

Our proposed method succeeded in finely controlling the number of fill-ins, reducing the memory usage for the preconditioning matrix by 22% and the computation time by 13% compared to the conventional methods.

This study introduces lumped elements into a fast three-dimensional numerical solver based on an electric field integral equation (EFIE) formulation. The efficient implementation of lumped elements is essential for the rapid analysis of reconfigurable metasurfaces (MTSs), one of the most promising technologies for the realization of reconfigurable intelligent surfaces (RISs). The lumped elements are modeled by a super element constructed as a linear combination of Rao-Wilton-Glisson (RWG) functions. The extended solver is capable of the fast re-computation of the problem for different impedance values, thus, enabling numerous evaluations of the field in optimization loops, for example. The implementation of the lumped elements are validated and run-time data are presented.

The continuous streamer discharge activity resulting from the degradation and failure of solid insulation systems leads to charge accumulation in the dielectric and distortion of the electric field. This paper presents an analysis of the charge accumulation and electric field distribution on the surface of an epoxy dielectric caused by continuous corona discharge activity using numerical analysis techniques. An additional experimental setup was designed to present the results of discharge current and surface potential measurements.

In this paper, the reduction method of the harmonic components in the back electromotive force (EMF) using the topology optimization is proposed. Because the waveform of the back EMF in the interior permanent magnet synchronous motor (IPMSM) conventionally includes the harmonic components, the torque ripple and other physical quantities tend to deteriorate. By changing the rotor structure of IPMSM using the topology optimization, the harmonic components of the back EMF were successfully reduced. Furthermore, the relationship between the eddy current losses in the permanent magnet and the reduction of harmonic components in the back EMF was also investigated.

The magnetization performance of the permanent magnets (PMs) using post-assembly magnetization is conventionally evaluated by fluxmeter which is implemented on the magnetizer. Although the surface magnetic flux is measured after its operation, it is not possible to directly estimate the magnetization states inside the PM. Then, the magnetization estimation method using leakage magnetic flux density (MELF) was proposed by some of the authors. The performance of MELF has been verified in the PMs magnetized with saturated magnetization. In this paper, the magnetization estimation of PMs with actual unsaturated magnetization in interior permanent magnet synchronous motor (IPMSM) was carried out.

In this study, a mode expansion method for estimating specific eigenmodes from measurement data and simulation results was proposed. As a test case, the estimation of propagation modes in a miter bend, where the presence of higher-order modes is a challenge, was performed. The electromagnetic field distribution within the miter bend was calculated through numerical simulations, and the validity of the proposed method was verified by applying the mode expansion technique. By calculating the proportion of the propagation modes, it was demonstrated that the input mode propagates as the dominant mode. Additionally, the presence of higher-order modes was confirmed, and their specific proportions were successfully calculated.