Conference Agenda

This article proposed a method for analyzing the inductor based on a Cauer equivalent circuit. The circuit parameters are identified from the input impedance of an inductor. Based on the synthesized Cauer equivalent circuit, it is possible to perform time-domain analysis to evaluate the eddy current losses for arbitrary driving voltage waveforms. Moreover, the magnetic saturation of the core can be considered in the Cauer circuit. It is shown that the copper loss and current response in time domain simulated from synthesized Cauer circuit agree well with the results calculated from time-domain finite element analysis.

This paper proposes a hybrid analytical model (HAM) for predicting AC copper losses in hairpin windings. The HAM combines the nonlinear semi-analytical model (SAM) and the one-dimensional (1-D) AC copper loss computation model to solve the inaccurate boundary solutions of the 1-D AC copper loss computation model when magnetic saturation occurs in the stator teeth. This HAM is suitable for a variety of hairpin winding configurations. The finite element (FE) model validates the effectiveness of the HAM. The primary contribution of this research is to present a quick and simple approach for accurately predicting AC copper losses in hairpin windings.

A robust open-source workflow is developed for finite element (FE) data generation for active learning (AL) based surrogate modelling. Machine learning (ML) and FE simulations work iteratively together, ML in a local computer and open-source FE in a containerised cloud. We develop a surrogate model aiming for a lightweight induction machine steady-state torque and total loss prediction as a function of motor frequency, voltage, and slip. Results show how active learning improves the surrogate accuracy, especially in regions where the error was originally the largest. Special attention is given to making the cloud FE solution procedure as robust and fast as possible, e.g., by a special convergence criterion, reliable parallel computation and variable timestep length.

The paper "A Tensor-Analysis-Based Approach to Accelerate 3D Eddy-Current

Finite Element Methods" was moved to PO1.

This study investigates the effectiveness of a parallel-in-space-and-time finite-element method (PinSTFEM) in electric machine analyses taking account of magnetic hysteresis. The PinSTFEM is based on a combination of the domain decomposition method and parallel time-periodic explicit-error-correction (PTP-EEC) method. In addition, a magnetization history in each element is appropriately corrected based on time-series data of flux density vectors within the framework of the PTP-EEC method. The parallel performance of the developed method is verified in the 3-dimensional magnetic field analysis of a ring core specimen and a practical interior permanent magnet synchronous motor.

We propose a formulation of the regularised combined field integral equation (regularised CFIE) and its discretisation only with the Rao-Wilton-Glisson (RWG) method.

The CFIE is a formulation of integral equations, which avoids the so-called ficticious frequencies of integral equations.

The most typical CFIE, which is a linear combination of the electric field integral equation (EFIE) and magnetic field integral equation (MFIE),

is known to be ill-conditioned in terms of computational time.

The regularised CFIE is another formulation of the CFIE to solve the ill-conditioning of the original CFIE

by applying a regularising operator to the part of the EFIE.

In many previous studies the regularising operator is determined based on the Calderon preconditioning.

This regularising operator however takes much more computatonal time than the standard CFIE

since discretising the EFIE with the Calderon preconditioner requires the dual basis function.

In this article we propose a formulation of the regularised CFIE, which can be discretised with the Galerkin method without the dual basis function.

The UHVDC transmission line covers a wide area, and the climate along the line is complex and changeable, so rainfall is inevitable. In rainy days, the parameters such as corona onset field, ion mobility and ion recombination coefficient of the conductor will change compared with those in sunny days. In addition, the analysis of the charge capacity and force of suspended droplets, falling raindrops and other particles should be added. Taking the above factors into account and based on the upstream finite element method (UPFEM), this paper proposes a numerical simulation method for the ion flow field of UHVDC transmission lines in light rain environment, and validates it under laboratory conditions. The results show that the total electric field and ion current density have increased to a certain extent compared with sunny weather. The results calculated by this method is close to the measured data in the ±800 kV line in China.

3D eddy current calculation plays an important role in the design of electrical equipment. However, due to the special structure of some equipment, numerical calculation often faces great difficulties. For example, in the laminate core for transformers, the overall dimensions of the core can be about 104 times the thickness of the laminate. A large number of elements and nodes will be required using the finite element method in analyzing such equipment with laminates. The computational cost would be extremely high, which cannot be afforded. This paper introduces a multi-scale finite element method (MSFEM) that can effectively reduce computational costs. Using a special interpolation built with an enrichment function, the elements are not restricted by media or geometry. The enrichment function is presented through distance regularization level set evolution, which enables MSFEM to cope with complex 3D model. Taking Team Workshop problem 7 as a numerical example, the accuracy of this method and the saving of computational costs have been proved.

The solution speed of 3D ion flow field is generally very slow, which seriously restricts the application of this method in UHVDC transmission line projects. In this paper, based on the 3D upstream finite element method (UPFEM), an adaptive iteration factor for charge density update is introduced, which can automatically control the iteration speed according to the change degree of electric field strength, ensuring the rapid and stable update of charge density. At the same time, to solve the problem that the convergence rate of charge density and electric field strength is inconsistent after the introduction of iteration factor, a half-error term is added to the charge density update formula. This controllable disturbance is only half of the charge density error, so it has little impact on the convergence of the charge density, but can greatly accelerate the convergence of the electric field intensity. It solves the short board effect, balances the convergence of charge density and electric field strength, and improves the overall solution speed of the algorithm. This method is used to calculate the ion flow field between a 3D coaxial cylindrical electrode. The numerical simulation results are compared with the analytical solution. Furthermore, the iterative speed of the traditional method and this method is also compared, which verifies the accuracy and effectiveness of the method proposed in this paper. Finally, the method is applied to a real ± 800kV UHVDC transmission line in China. The results show that the method is in good agreement with the test results and has strong engineering practicability.

In order to calculate quickly and accurately the magnetic field of the rotating machine when the rotor is eccentric, and study the influence of rotor eccentricity on its operational performance, this paper proposes a hybrid method combing the proper orthogonal decomposition (POD) and the interpolation technique to establish a behavior model of a permanent magnet synchronous motor considering rotor eccentricity. By dividing the parameter space reasonably, the problem of the torque deviation caused by rotor eccentricity submerged in the error caused by POD is avoided. The results show that the proposed model can provide accurate dynamic time-domain response and torque spectrum of the motor, and the simulation time can meet the real-time requirement. It offers an effective method not only for fast design optimization, but also for real-time monitoring and predictive fault diagnosis of the motor.

This paper describes the development of an integrated method of high-frequency electromagnetic wave analysis using a parallel finite element method and an efficient Uncertainty Quantification (UQ) method, i.e. the Non-Intrusive Polynomial Chaos (NIPC) method, based on the assumption that input parameters are introduced as random fields. First, taking a heat transfer problem with the NIPC method, we investigate the computational complexity of each process appeared in the NIPC method, and clarify that the bottleneck process is the pre-processing eigenvalue analysis. Next, we propose a new method to reduce the computational complexity of the eigenvalue analysis by employing a coarser finite element mesh than an original mesh used in the deterministic analysis. Here we obtain eigenfunctions on the original coarse mesh, and then convert the coarse mesh to the fine mesh. Finally, by integrating the NIPC method with the proposed computational reduction method and the parallel finite element electromagnetic field analysis method, we successfully perform the UQ analysis of large-scale high-frequency electromagnetic wave problem, which has been difficult to achieve so far.

An original mesh refinement of Isogeometric model through $C^0$ isogeometric mortaring is proposed in this paper to address patch refinement propagation occuring when handling multipatch geometries, a particularly relevent side-effect in electromagnetic problems. The efficency of the proposed method is demonstrated on an academic problem.

The effect of AC loss is non-negligible for high-speed rotating machines with their associated PWM control. The same is true for the joule loss in the coil of the induction heating systems. To estimate those losses, solid conductors are used in 3-dimentional electromagnetic finite element analysis (FEA). However, the convergence characteristics of the iterative method deteriorates in FEA with solid conductors. The A-φ method and the application of appropriate initial state for each time step make it possible to overcome this challenge for transient analysis. We present the effectiveness of these two approaches by employing several practical analysis models.

This paper aims to propose an Improved Starting Point (ISP-) Newton method applied to vector potential A formulation for circuit coupled magnetostatic problems. These problems usually are known to vary greatly during the time. This impacts the quality of the initial guess of the Newton method and thus increases significantly the computational cost. The Newton method with vector potential formulation A has been analyzed. Numerical examples show the performance of our proposed ISP-Newton method.

The authors have been working in the development of dedicated computer based on dataflow architecture to achieve a portable, low power consumption, low cost and green high-performance computing (HPC) technology for electromagnetic fields simulation, which can be effectively used in industry applications. A dataflow architecture machine of 2-D finite integration technique (FIT) based on BiCG-Stab scheme matrix solver for magneto-static fields was proposed in the previous work. In this paper, we discuss a conceptual design of FIT dataflow machine for 2-D time domain eddy current fields simulation.

In this paper is presented an improved version of a general methodology for calculating the optimal solution of non-conventional winding distribution for applications in Wind Turbines based on a direct-drive permanent magnet synchronous generator. The proposed methodology is also well suited for different conventional windings in electrical rotating machines. Moreover, the aim of this paper is to compare the performance of wind turbines based on PMSGs with different types of windings such as concentric, integer and distributed. To achieve this, an exhaustive parametric study through 2D and 3D finite element analysis is carried out in order to measure variables such as the copper mass, losses, the power quality and, cogging and ripple torques. Finally, in a comparative table are described the main characteristics of each type of winding and the best performance for the purpose of using in wind turbines.

Modern electric drive systems consist of an electric motor drive connected to a mechanical system. Single-mass models are commonly used in electromechanical finite-element analyses to study various aspects of these systems such as the output torque of the electric motor or the transient performance of electric drives. Nevertheless, single-mass models are inherently not sufficient for the analysis of torsional vibrations. To inspect the behavior of the electric drives connected to flexible torsional systems, higher degree-of-freedom mechanical models should be employed. This paper presents a comprehensive finite element model of an induction machine coupled with a three-mass lumped-element model. The basic background for the model and some initial time-stepping results are presented in this paper. The results of the model in the frequency domain will be presented in the full paper.