Conference Agenda

Overview and details of the sessions of this conference. Please select a date or location to show only sessions at that day or location. Please select a single session for detailed view (with abstracts and downloads if available).

 
Session Overview
Session
PA-M3: Optimization and Design
Time:
Tuesday, 16/Jul/2019:
10:50am - 12:40pm

Session Chair: Junwei Lu
Session Chair: Stephane Brisset
Location: Patio 44-55

Show help for 'Increase or decrease the abstract text size'
Presentations

Optimization of Ion Current Density Measuring Device and Its Application

Shuhong Wang, Ting Zhu, Shuang Wang, Song Huang, Naming Zhang

Xi'an Jiaotong University, China, People's Republic of

Ion current density is one of the important standards to measure the safety of HVDC transmission,and the Wilson plate has commonly been used in the measurement of ion current density. This paper is aim at optimizing the Wilson plate and improve the measurement accuracy of ion flow field. Firstly, the ion current density distribution of 800 kV DC transmission line is calculated based on finite element iterative method (FEIM). Secondly, the current distribution of the Wilson plate in ion flow field is calculated. Thirdly, the accuracy influence factors of ion current measuring device are studied, and the Wilson plate is optimized. Finally, the optimized device is used for measurement, and compared with the non-optimized device, which fully illustrates the effectiveness of the optimization.



Transfer Learning Through Deep Learning: Application to Topology Optimization of Electric Motor

Jo Asanuma, Shuhei Doi, Hajime Igarashi

Hokkaido university, Japan

This paper proposes use of transfer learning through deep learning to reduce the computing cost of the topology optimization of electric motors. It is shown that the training of the convolutional neural network (CNN) can be made effective using the transfer learning. Moreover, the recursion is realized by CNN. The computing cost is shown to be reduced by about 80% by the proposed method to obtain the pareto solutions of the multi-objective problem with respect to the average torque and torque ripple of a permanent magnet motor.



Optimal Design of Conductor-Air Interface in Eddy Current System Using Continuum Sensitivity Analysis and Level Set Method

Kyung Sik Seo, Il Han Park

Sungkyunkwan University, Korea, Republic of (South Korea)

The shape design using the continuum sensitivity analysis and level set method can be performed by solving a first-order Hamilton–Jacobi equation with the velocity field derived from the sensitivity formula, without additional boundary parameterization. This optimization method has been successfully applied to shape design in electro- and magneto-static systems. The continuum sensitivity analysis was also used to design conductive materials in eddy current systems. However, the optimization method involved classical boundary parameterization with a large number of design variables. This paper proposes an optimization method for the shape design of the interface between the conductor and air in eddy current systems by using the continuum sensitivity analysis and level set method. The sensitivity formula for an eddy current system is derived with respect to the velocity field Vn. The shape design is performed by solving a level set equation coupled with the sensitivity formula. The proposed optimization method is applied to a practical problem for the shape design of a conducting shield in magnetic induction tomography with multi-objective function.



A Study on the Optimal Design for Improvement of Efficiency of Wound Field ISG for Vehicle

Byung-Chan Kim1, Dong-Yeol Lee1, Jun-Young Kim2, Dong-Woo Kang1

1Keimying University, Korea, Republic of (South Korea); 2Hanyang University, Korea, Republic of (South Korea)

The ISG (Integrated Starter Generator) is a system that operates in the power generation mode while driving and charges the battery, and operates in motor mode that assists the engine shortly during acceleration. The ISG system largely consists of a wound field rotor and a permanent magnet rotor. In case of the wound field rotor, the price is lower than the permanent magnet rotor. However, since the current applied to the field side, causes additional copper loss of rotor so the efficiency is reduced. To satisfy the output voltage, the magnetic flux is increased by increasing the field current at the rated rpm. Since the Back-EMF(Electromagnetic force) increases in proportion to the speed, in order to reduce the Back-EMF at the high rpm, the field weakening control method that reduces the magnetic flux by reducing the field current is used. Automobiles are more likely to drive at rated rpm than high rpm. In case of the conventional wound field rotor, there is a problem that the field current is high to generate high flux when the rated rpm is driven. So the efficiency is reduced. Therefore, in this paper, we propose a PM-ASSIST structure using windings and magnets to compensate for the disadvantages of conventional wound field rotors. In the rated drive, it is driven in the same way as the consequent pole, which does not apply the field current but generate electromagnetic characteristics with only one pole magnet. In the high-speed drive, field-weakening control is performed to reduce the total magnetic flux by applying current to the wound field in a direction opposite to the magnetic flux of the magnet. As a consequence, we propose an optimal structure and control method that complements the disadvantages of the low efficiency of the conventional wound field Starter Generator.



Application of Conditional Variational Auto-Encoder to Magnetic Circuit Design with Magnetic Field Computation

Ryota Kawamata, Shinji Wakao, Noboru Murata

Waseda University, Japan

In the design optimization of electric machine, we have to take account of various features of the targeted device to be improved, i.e., device performances, shape with easy manufacturability, etc. However, it is generally troublesome to achieve the improvement of various features by adequately defining many kinds of constraint conditions as well as objective functions in the magnetic field computation. In recent years, representation learning using Deep Learning much attracts attention because it can acquire the features of data as a distributed representation, and accurately reproduce corresponding data. In this paper, utilizing machine learning technology, we propose an application of Conditional Variational Auto-Encoder (CVAE) to the design optimization of magnetic device in order to carry out the improvement based on automatically acquired characteristics of dataset. Finally, we show the possibility of novel optimization approach using machine learning.



Multi-objective Topology Optimization of DC-biased Reactor in Steady-state Time-domain with Magnetic Nonlinearity

Hiroshi Masuda1, Kohei Arase1, Yoshifumi Okamoto1, Shinji Wakao2

1Hosei University; 2Waseda University

Reactors implemented in electric circuits are typically driven under a DC-biased current. Therefore, DC-biased characteristics of reactors exhibiting a stable inductance with various DC-biased currents is desired. Topology optimization enables such a desired performance to be realized by changing the structure of electromagnetic apparatus significantly. In this paper, topology optimization is applied to a pot-type reactor in the steady-state time-domain electromagnetic field with magnetic nonlinearity to enhance the DC-biased characteristics. Furthermore, the eddy current loss occurring on the conductive plate installed outside the reactor is suppressed by multi-objective topology optimization. This method is used to derive the structure of the reactor having both high DC-biased characteristics and shielding performance in multi-DC currents.



Design of Linear Switched Reluctance Motor with Segmental Mover for Electromagnetic Launching Applications

Hao Zheng, Lin Wang, Xingfei Du, Daohan Wang, Xiuhe Wang

Shandong University, China, People's Republic of

Electromagnetic launching applications gain more and more attraction, which require high efficiency and force density. A new structure of Linear Switched Reluctance Motor (LSRM), Linear Switched Reluctance Motor with Segmental Mover (LSRMSM) has been studied to be a more significant alternative of electromagnetic launching applications. An overall design process of LSRMSM used in electromagnetic launching applications is proposed in this paper, which aims to achieve the requirements of launching application and enhance the efficiency both considering the dimensions of the LSRMSM and thermal analysis. The consequence of the process shows that the LSRMSM features excellent performance in the electromagnetic launching application as well as the higher efficiency and force density.



Estimation Method of Magnetization Vector Distribution in Permanent Magnet Using 2-D Fourier Series Expansion

Narichika Nakamura1, Yoshifumi Okamoto1, Kenta Osanai2, Satoshi Doi2, Tetsuya Aoki2, Keichi Okazaki2

1Hosei University, Japan; 2Denso Corporation, Japan

Because the magnetization distribution in the permanent magnet synchronous motor (PMSM) strongly affects the torque property, the identification of magnetization distribution in permanent magnet is essential for the design of less-vibration PMSM. The method to identify the magnetization distribution is based on the minimization of least square between the measured magnetic flux density at measured point and the flux density derived from the estimated magnetization distribution. However, since there is not uniqueness of magnetization distribution to realize the measured magnetic flux density, the estimated distribution of magnetization is unfeasible. To suppress the occurrence of unfeasible magnetization, the angle of magnetization in each magnet cell for Biot-Savart law is handled using Fourier series expansion with inequality constraint conditions of magnetization angle. The performance of proposed method is compared with that of the conventional method.



Adjoint Based Topology Optimization in Nonlinear Magnetostatics:Application to Hall Effect Thrusters

Rtimi Youness, Messine Frédéric

Laboratoire Laplace, Toulouse - France

In this paper, density based topology optimization (TO) is considered in nonlinear magnetostatics. Previously, TO in linear magnetostatics was formulated to design Hall-effect thrusters (HETs) and it was solved using gradient based solvers. In that previous version, the gradient information was provided by the continuous adjoint method. Herein, the adjoint approach is updated in orderto take into account the saturation aspect of materials. This new version of the adjoint method is implemented and compared to a finite difference approach. The comparison shows that the adjoint based gradient information in non linear magnetostatics is accurateand ready to be used in order to solve HET TO design problems.



Power Improvement of Spoke-type PMSM according to Permeance Changes by Applying a Novel Rotor Shape for Washing machines

Hyungkwan Jang1, Seung-Taek Oh1, Hyung-Woo Lee2, Jongsuk Lim1, Yeji Park1, Ju Lee1

1Hanyang University, Korea, Republic of (South Korea); 2Korea National University of Transportaion, Korea, Republic of (South Korea)

This study shows the novel rotor shape of the spoke-type PMSM to improve power by changes of permeance in the rotor and air gap. Applying a new shape to the rotor results in a permeance change in some areas of the rotor in which low magnetic flux density existing, which in turn changes the magnetoresistance in the air gap. As the magnetoresistance changes, the inductance of the rotor d-axis decreases and the q-axis inductance increases to improve the reluctance torque. We investigated to maximize the reluctance torque and air-gap magnetic flux density with a novel rotor shape. In other words, low magnetic flux density area which help rarely torque increase was subtracted. In addition, it also can change inductances of d-axis and q-axis. Accordingly, it was confirmed that average torque improves as some area of rotor where magnetic flux density is low, was removed.



Optimization Design of a Novel Dual-Stator Switched Flux Consequent Pole Permanent Magnet Machine with Unequal Length Teeth

Liangliang Wei, Taketsune Nakamura

Department of Electrical Engineering, Kyoto University, Kyoto 615-8510, Japan

This digest proposes optimization design of a novel dual-stator switched flux consequent pole permanent magnet machine (DS-SFCPM) with unequal length teeth. Compared with traditional dual-stator SFPM, it can significantly improve the torque density, and has good flux regulating capacity. Firstly, the basic principle of the proposed DS-SFCPM is introduced, and the influence of the unequal length teeth on the torque density and flux regulating capacity is analyzed. Then the different parameters of rotor-PMs, iron piece and dc of field coil are optimized. Finally, various FEA simulation, optimization and comparison studies of the proposed machine are performed. The results verified the effectiveness of the proposed structure.



Calculation of Active Material's Torque Contributions for a Flux Switching Machine

Haris Kurtović, Ingo Hahn

Friedrich-Alexander University Erlangen-Nuremberg, Germany

A new calculation method for the torque contribution of an arbitrary point or element in an electric machine's geometry is developed. The local torque contributions (LTC), in addition to informative visualizations, provide insights into the iron utilization and possible machine geometry improvements. Assigning flux tubes from the magnetic vector potential's separate equipotential regions, practically creates a set of minor or localized magnetic circuits (LMCs). Each of the LMCs, that crosses the air-gap, is assigned with a torque contribution from integrating the Maxwell stress tensor along its air-gap intersections. This torque contribution is mapped to each finite element belonging to this LMC, for each rotor position. The torque contribution data are briefly discussed and a frequency domain analysis is performed, focusing on the mean and dominant ripple components of the torque.



Automatic Design Optimization of a Permanent Magnet Assisted Synchronous Reluctance Motor with Field Line Shaped Barriers

Hongwei Xu, Jian Li

Huazhong University of Science and Technology, China, People's Republic of

This paper presents an automatic and effective optimization design of a Ferrite permanent magnet (PM) assisted synchronous reluctance motor (PMA-SynRM). As to maximize the rotor anisotropy and simplify the design degrees of freedom, the reluctance rotor with field line shaped (FLS) barriers is adopted and a convenient rotor parametric model is also considered. Hereafter, an automatic optimization program coupled with Non-dominated Sorting Genetic Algorithm II (NSGA-II) and computationally efficient finite element analysis (CE-FEA) is developed for improving the optimization efficiency. Based on that, the rotor geometry optimization of a PMA-SynRM with FLS barrier is automatically achieved. And the optimal structure with less PM usage could provide better performance in terms of motor torque output, torque ripple and efficiency, relative to the initial design.



Optimum Design of CFSM using a 2-Dimensional Equivalent Magnetic Circuit considering Axial Leakage

Kyoung-Soo Cha1, Young-Hoon Jung1, Myung-Seop Lim2, Jung-Pyo Hong1

1Hanyang University, Korea, Republic of (South Korea); 2Yeungnam University, Korea, Republic of (South Korea)

This paper proposes an optimal design using a 2 dimensional (2D) equivalent magnetic circuit (EMC) considering the axial leakage of a concentrated flux-type synchronous motor (CFSM). The CFSM is a typical motor using non-rare-earth magnets and is used in various industries. However, in CFSM design, a 3 dimensional (3D) finite element analysis (FEA) is needed to consider the axial leakage caused by the CFSM structure. The 3D FEA takes a long time to solve and is inconvenient for preprocessing. Therefore, to optimize the design in a short period of time, the 2D EMC which is faster and does not require pre-process compared to the 3D FEA, is deemed suitable. In this paper, the EMC is verified by comparing EMC and FEA results, and an optimum design process is presented. For future work, the optimal design model of CFSM is tested and compared with the initial model to demonstrate the validity of the proposed method.



Optimization Design of a Novel Partitional Stator Flux Modulated Memory Machine

Liangliang Wei, Taketsune Nakamura

Department of Electrical Engineering, Kyoto University, Kyoto 615-8510, Japan

This digest proposes the optimization design of a novel partitional stator flux modulated memory machine (PS-FMMM). Compared with traditional variable flux memory machine, it can significantly improve the torque density due to magnetic gear effect, and has good flux adjustable capacity with low coercive force (LCF) PMs. Firstly, the basic principle of the proposed PS-FMMM is introduced, and different combinations of stator poles and rotor poles are analyzed and compared. Then the torque regulating characteristics of the proposed PS-FMMM with different field currents is achieved. Various FEA simulation, optimization and comparison studies of the proposed machine are performed. The results verified the effectiveness of the proposed structure.



Analysis Method of Permanent Magnet Assistance Synchronous Reluctance Motor with Asymmetric Barrier

Hyunwoo Kim1, Seung-Taek Oh1, Chan-Bae Park2, Jongsuk Lim1, Yeji Park1, Ju Lee1

1Hanyang University, Korea, Republic of (South Korea); 2Korea National University of Transportation, Korea, Republic of (South Korea)

This paper establish an analysis method of permanent magnet assistant synchronous motor (PMA SynRM) with asymmetric barrier. In a general motor analysis method, the inductance is calculated using the dq-axis vector diagram. In addition, the characteristics of the motor are analyzed by separating the magnetic torque and the reluctance torque. However, in an asymmetric motor, the magnetic neutral plane (MNP) is shifted because the magnetic permeance is asymmetric. Therefore, it is difficult to analysis the characteristic of the asymmetric motor because it involves errors applying the general analysis method. In this paper, the magnetic property of the asymmetric motor is analyzed and the analysis method of asymmetric motor is proposed. To verify the proposed analysis method, PMA SynRM is designed as a conventional model. Furthermore, the magnetic torque and reluctance torque are separated through the proposed analysis method. The validity of the proposed analysis method is verified through finite element analysis (FEA) and manufacture of the conventional model.



A Novel Multi-Objective Optimization Method for the Optimal Design of Interior Permanent Magnet Synchronous Machine

Wenming Tong, Haiyang Wei, Shengnan Wu

Shenyang University of Technology, China, People's Republic of

A novel multi-objective optimization method is developed for design optimization of a V-shaped interior permanent magnet synchronous machine (IPMSM). This method is capable to find the Pareto optimal solution with less time for the given multi-objective problem. In this paper, the method is demonstrated to maximize the output torque of V-shaped IPMSM while minimizing the sinusoidal distortion rate of the air-gap flux density waveform. The process includes three steps: (1) Building the first level Kriging models to get several local areas which contain the local optimal points; (2) A trade-off between the Kriging response surfaces of different objective functions to get one or several local areas which contain global optimal points and using the selected areas to get the second level Kriging models; (3) Genetic algorithm can be used to obtain the Pareto Front based on second level Kriging models. Finally, the optimization method is verified by finite element method.



Utilizing Highly Nonlinear Materials in Electrical Machine Topology Optimization

Aino Manninen, Janne Keränen, Jenni Pippuri-Mäkeläinen, Tuomas Riipinen, Sini Metsä-Kortelainen, Tomi Lindroos

VTT Technical Research Centre of Finland Ltd., Finland

Nonlinear characteristics of soft magnetic materials need to be taken into account in topology optimization of electrical machines to achieve feasible results. However, use of nonlinear material models often causes slow convergence of the optimization. Furthermore, we aim at 3D printing the optimized geometry using Fe-Co-V material. This specific material has a highly nonlinear reluctivity curve, resulting in a high number of Newton-Raphson iterations for the FE model within the optimization scheme. In this work, we investigate the effect of different material models on the topology optimization and convergence. We present methods to simplify the material models and to formulate the topology optimization process for improved convergence.