Academic Poster Session

The machine structure with slits inserted at the center of stator teeth is studied as the torque enhancement technique of permanent magnet synchronous machine. Previous study explains the torque enhancement mechanism from armature flux, but it is not explained from electro-mechanical energy conversion related to torque generation principle. This study investigates the mechanism by focusing on the energy conversion. Analysis result shows that sub-harmonic flux density contributes to torque.

Thin motors are highly demanded for application in robot joints, electric vehicles (EVs), and home appliances. Therefore, this study focuses on axial-gap motors, which are expected to achieve a thinner profile and higher efficiency than widely used radial motors. In previous research, variable field performance was implemented to expand the output range; however, large torque ripple caused significant vibration and noise issues. To address this, the present study introduces a fractional-slot structure and evaluates its torque ripple reduction performance using three-dimensional (3D) electromagnetic field analysis. As a result, under the condition of constant current density, it was confirmed through simulation that the average torque increased by 2.24 times and the torque ripple was reduced by 91% compared to the previous research. This paper reports the detailed results of this investigation.

Direct-on line (DOL) induction motors do not require an inverter and widely used in industrial applications because of the low cost and high robustness. However, the motor size is large because a distributed winding is applied, and the efficiency is low because of the large copper loss in the coil end. In order to solve this problem, we propose a 2-pole DOL induction motor with concentrated winding that the number of stator slots and winding factor are high, where the principle of the magnetic modulation is used. However, the slop under no-load condition is high. In this study, the mechanism of the large slip is shown. Finally, in order to decrease the slip, a skew is applied to the rotor, and the effectiveness of the skew is investigated with 3-D analysis.

This study proposes a manufacturability-aware topology optimization method for interior permanent magnet (IPM) motors utilizing JMAG and the image processing library OpenCV. Conventional topology optimization has faced the issue of generating unmanufacturable designs. Therefore, the proposed method extracts the optimized material distribution as an image using OpenCV and applies a geometric evaluation within the objective function, penalizing "separated cores," "thin bridges," and "unmanufacturable microstructures." Furthermore, Warm Starting CMA-ES was introduced to accelerate convergence by using individuals that satisfy manufacturability as the initial population. Numerical analysis results for the IPM motor confirmed that the proposed method can generate highly manufacturable geometries while maintaining excellent magnetic performance. Additionally, a prototype was fabricated using wire electrical discharge machining (Wire EDM) without any geometry modifications to the optimization results. As a result, while conventional methods resulted in degraded magnetic performance, the proposed method successfully limited the performance fluctuations to within 10%.

A spoke-type IPM motor is effective for achieving high torque density; however, leakage flux generated in the rotor prevents effective utilization of permanent-magnet flux. This study proposes a rotor structure in which the magnetic bridge is replaced with a non-magnetic resin unit to eliminate the leakage-flux path and reduce magnet volume. Type A employs resin fastening at the inner diameter, whereas Type B employs resin fastening at the inner diameter and outer diameter. Numerical analysis showed a magnet-volume reduction of 19.2% in Type A and 27.4% in Type B while maintaining torque performance comparable to that of the conventional motor. Furthermore, a prototype was manufactured, and the measured current–torque characteristic and speed–torque characteristic agreed with the analytical result, confirming the feasibility of the proposed structure.

Motor design often requires many objective optimization. While NSGA-II have traditionally been used as effective algorithms, efficiency of solution search decreases as the number of objective functions increases. To solve this problem, it is effective that the dimensionality of the solution space is reduced using principal component analysis; however, this method requires a huge amount of pre-obtained analytical data. In this study, generative AI is used to supplement the design knowledge lacking in users and to define low dimension objective functions.

A variable magnetomotive force memory motor (VMFM) can expand its operating range by changing the magnetization state of a variable permanent magnet. However, changing the magnetization state requires a pulse current several times larger than the rated armature current. This study proposes a method using magnetization windings placed in the stator to reduce the magnetization current flowing through each power semiconductor device in a VMFM. In addition, magnetization experiments were conducted using a prototype motor to verify the effectiveness of the proposed method. The results demonstrate that the proposed method enables the downsizing of power semiconductor devices and cost reduction while maintaining manufacturability.

Reducing the use of rare-earth permanent magnets while producing a high power at high efficiency is crucial. We have proposed a novel permanent magnet assisted reluctance motor that can produce high power with less rare-earth PM and operate over a wide speed range at high power by varying the magnetization. Furthermore, the proposed motor can naturally compensate and reduce itself to a significantly small short-circuit current by varying the magnetization state of the PM when short circuit trouble occurs. Results confirmed that the proposed motor can produce high torque at low speed and high power at high speed by varying the magnetization of PM while reducing the volume of rare-earth PM. In addition, results indicate that the short-circuit current naturally reduces itself due to demagnetization of PM to 42% of the rated current in case of short circuit trouble. It yields the suppression of excessive drag torque for motor drive systems.

Increasing rotational speed is an effective approach to enhancing specific power (kW/kg) of electric motors. However, ultra-high-speed operation also increases loss density, which must be limited to prevent excessive temperature rise. This presentation aims to achieve a specific power of 10 kW/kg under a loss density constraint of 1,000 W/L for an 80 kr/min, 80 kW bearingless motor. It also investigates the effects of winding materials and arrangements on winding AC loss.

Electric motors for automotive propulsion designed to be higher rotational speeds to achieve compactness and high-power density. However, high-frequency operation at high speeds leads to increased winding losses; addressing this issue is essential, particularly for the quasi-coreless motor investigated in this study. This presentation reports on a method for the separate evaluation of winding losses using JMAG-Designer and the effectiveness of winding transposition in reducing circulating current losses. An analysis method was developed to categorize and evaluate winding losses into three components: DC copper loss, winding eddy current loss, and circulating current loss arising between parallel strands. Circulating current loss was calculated for an aluminum winding consisting of 36 parallel strands using a circuit model that accounts for induced voltage differences between the strands. Additionally, an analysis was conducted for a configuration incorporating a 180° transposition, and the results were compared with a non-transposed configuration. The results confirmed that the 180° transposition reduced circulating current loss by 91.3%.

In induction heating (IH), a dedicated coil is required for each part geometry, and the burden of coil design increases for parts with greater variety and more complex geometries. In recent years, the practical application of metal 3D printers has made it possible to fabricate complex-geometry coils tailored to the workpiece. Such coils include winding turnarounds and lead sections, and their current paths are not axisymmetric; therefore, the heating characteristics must be predicted using the real three-dimensional geometry.
In this presentation, a coil with complex geometry is modeled in three dimensions with its real geometry, and a coupled electromagnetic-thermal analysis is performed using JMAG, taking into account the nonlinear B-H characteristics of the workpiece. However, the B-H characteristics of a steel workpiece change with temperature, and the workpiece almost loses its magnetism near the Curie point. A model of this temperature dependence has not been clearly established. Therefore, an induction heating system is constructed to measure the workpiece temperature, and the effect of the temperature dependence of the B-H characteristics on the heating characteristics is evaluated by comparison with the analysis. This clarifies to what extent the temperature dependence of the B-H characteristics should be considered in the analysis before prototyping.

High environmental regulation acerates the demand for electric mobilities, which requires high power density. To achieve this system, this presentation proposes axial flux type PCB motor with integrated motor drive. First, the design of PCB coil pattern and the integrated inverter circuit for realizing the proposed system are revealed. Then, the prototype and experimental test results of functional verification are demonstrated.

In recent years, the electrification of mobility has driven an urgent need for smaller and lighter motors and drive circuits. This research focuses on electromechanical integration using printed circuit boards (PCBs) to functionally integrate motors and inverters. Due to the extreme proximity of these components, it is essential to visualize and mitigate the effects of leakage flux on the circuits. This presentation focuses on a radial-flux PMSM, reporting on the PCB implementation of the coil-end and the verification of the pattern design through speed-torque (N-T) characteristics comparisons using a prototype. Furthermore, for the ring-shaped inverter board designed for integration, this study introduces a design process utilizing 3D-FEM to estimate the eddy current distribution and eddy current losses caused by leakage flux, demonstrating how inserting slits effectively reduces these losses.

Rising hardware procurement costs, driven by the rapid growth of AI-related demand and the weak yen, have made it a challenge in the CAE field to balance the cost-effectiveness and operational flexibility of computing infrastructure. This study focuses on mdx, a computing platform that offers industry–academia collaboration categories (such as Plan 3) and allows virtual computing environments to be built flexibly at low cost, and examines its effectiveness as an execution environment for JMAG parallel computing. As a reference, we use an on-premises server owned by our laboratory (Dell Precision 7920) with 52 physical cores and 192 GB of memory. The same analysis model is computed in both environments, and the parallel efficiency and the reduction in computation time with increasing core counts are quantitatively evaluated. We present the results of comparing the computational efficiency of the two environments and discuss the significance of mdx as a computing platform that universities and companies can use jointly. We also describe our future plan to apply the parallel computing environment to large-scale sampling and extend it to high-speed shape and control optimization using surrogate models.

Electromagnetic field analysis involves many steps, from modeling to solving and extracting results. We are developing an environment in which agents based on large language models (LLMs) are connected to analysis software through the Model Context Protocol (MCP) and used in analysis work. LLMs can be used in three ways: as an AI tutor, with which users learn analysis while carrying it out; for autonomous operation, in which design work proceeds by leaving the analysis to the LLM; and for developing extensions by creating DLLs. In each case, the challenge is how to provide the LLM with the knowledge needed to operate the software. In this presentation, we introduce an environment in which JMAG is connected to LLM agents via MCP, together with cases in which students have used it. The MCP server of this environment wraps Skills, which provide knowledge of JMAG procedures, and Lint, which checks JMAG analysis results, and makes them available to the LLM.

Optimization using a finite element method (FEM) is widely employed in electrical machine design to achieve the required performance. Optimization processes require many FEM analyses, and the total computational cost increases significantly, especially for large-scale FEM models. This study investigates surrogate model-assisted optimization for large-scale flux-modulated-type magnetic gears.

Surrogate model-assisted optimization using surrogate models generated from finite element method (FEM) analysis results can reduce optimization time. However, reducing computational cost and maintaining solution accuracy are competing objectives. In this study, the number of generations in which FEM evaluations are replaced by surrogate model evaluations is varied in genetic algorithm (GA)-based optimization, and the resulting optimization outcomes are compared and investigated.

The development of compact fusion reactors using high-temperature superconductors (HTS) are intensifying worldwide. Some remarkable progresses have been made in developing HTS conductors and coils intended for direct-current or quasi-static operations. Meanwhile, pulse currents and external magnetic field sweeps generate very large induced and screening currents in tokamak reactor components such as central solenoid (CS) coils. These currents result in large losses. In this study, we perform three-dimensional finite element analysis of stacked HTS conductors using JMAG to clarify the electromagnetic phenomena and losses within the conductors. The detailed results of the simulation will be presented at the poster.

The power density of traction motors can be enhanced through the use of concentrated windings, higher pole numbers, and shorter axial length. However, these design trends increase end-winding effects, leading leakage inductance and inducing eddy currents in motor housing. This presentation introduces a 3-D FEA method of housing eddy currents in a CW-SynRM.

This study focuses on the design and optimization of a single-phase flux reversal motor (FRM) for high efficiency and miniaturization. FRMs are characterized by large torque ripples and dominant permanent magnet eddy current losses. To address these issues, the effects of pole arc angles and cross-pole geometry were investigated to reduce torque ripples and permanent magnet eddy current losses. Furthermore, an AI agent was used to identify the optimal stator yoke thickness.

In recent years, the increasing power density of electric motors has created a growing demand for efficient stator cooling. In thermal analysis of oil drip cooling, it is necessary to appropriately define the oil-wetted area and heat transfer coefficient (HTC). However, accurately determining these parameters under actual oil dripping conditions remains challenging.
In this study, a fundamental investigation of an HTC estimation method using an aluminum cylinder is conducted toward the development of a thermal analysis model for stator oil drip cooling. First, the heating process of the aluminum cylinder is simulated using JMAG to evaluate its temperature response until steady-state conditions are reached. Next, oil drip cooling experiments are performed on the aluminum cylinder to measure the oil-wetted area, oil inlet and outlet temperatures, and cylinder surface temperatures. Based on the energy balance and Newton's law of cooling, the average effective HTC is estimated.
Through these investigations, this study aims to establish an experimental method for estimating the HTC and to apply the obtained results to thermal analysis of stator oil drip cooling in the future.