Conference Agenda

This paper proposes a new model order reduction technique for the Darwin model of Maxwell’s equations. In this method, three symmetric non-gauged equations are solved to obtain the basis vectors. The reduced-order model is obtained via the congruent transformation, and the time-domain analysis is performed by means of the ordinary time-difference method.

In many technical applications at low frequency, wave propagation effects can be neglected, and Darwin models promise to accurately describe the electromagnetic fields, because they include resistive, inductive and capacitive terms. In a former work, we developed an efficient, frequency stable method to solve the full Maxwell model, i.e., without neglecting any effect. This method employs the electro-quasistatic gauge, which allows to cleanly separate the calculation of capacitive effects in a first step, from the calculation of inductive effects in a consecutive second step. The Darwin model in frequency domain and electro-quasistatic gauge can easily be derived from this method by neglecting just one term, namely the induced displacement current. Here, we compare this Darwin model with the full Maxwell model, for the example of an RLC-series circuit. We found that this Darwin model correctly describes the interaction between L and C in the RLC-circuit at most at very low frequencies.

This paper investigates an effectiveness of a parallel-in-space-and-time finite-element method based on a domain decomposition approach and a parallel time-periodic explicit-error-correction (TP-EEC) method, which is one of the parallel-in-time methods, in a large-scale analysis of a practical interior permanent magnet synchronous motor with concentrated winding fed by pulse-width modulation (PWM) inverter. The parallel performance in massively parallel computation is examined by utilizing over 10,000 processes.

This paper describes a novel and accurate implementation of Finite Element Method-Boundary Element Method (FEM-BEM) coupling based on Argyris Element (AE), for C^1 solution of axi-symmetric magnetostatic problems in open boundary domains. The aim of the work is twofold: (i) to provide novel, easy and accurate way to compute the basis functions following a "per-element" approach, rather than using the unit triangle (i.e. with vertices (0,0), (1,0), and (0,1)), and (ii) to provide a clear mathematical and numerical description of the FEM-BEM coupling involving C^1 basis functions. The first goal is achieved by simply shifting, in a suitable way, the physical triangular elements, and computing the basis function in the shifted coordinate system. Remarkably, the degree of accuracy of this approach is comparable with those given with the exact basis functions obtained using the unit triangle. Our FEM-BEM coupling is based on the evaluation of closed form analytical Green's functions for the poloidal flux and its partial derivatives of both first and second order. The method is tested on the so-called "snowflake configuration", an advanced plasma equilibrium magnetic configuration arising in Magnetic Confinement Fusion (MCF) in which a C^1 solution is fundamental. We show how accurate results are guaranteed, even for coarse meshes.

The Cauer Ladder Network (CLN) method enables to construct a reduced based circuit model of a Finite Element model. In this paper, we propose an estimator which provides an upper bound of the truncation error due to the CLN method. The error estimator is tested on an analytical model and a Finite Element model to validate the approach.