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).

Electromagnetic Modeling of Ferrites Using Shell Elements and Random Grain Structures

Paavo Rasilo^{1}, Joonas Vesa^{1}, Johan Gyselinck^{2}

^{1}Tampere University, Finland; ^{2}Université libre de Bruxelles, Belgium

We present a novel approach for stochastic finite element modeling of electromagnetic losses in ferrites, combining an existing thin shell model (TSM) for highly permittive grain boundaries with a Voronoi-tessellation-based geometry generation algorithm. The TSM is first validated in the case of a periodic grain structure in a linear 2-D time-harmonic case over a frequency range of 1 kHz – 1 GHz. It is then applied in a stochastic study for simulating the effect of varying grain structure on the losses and reactive power densities.

4:50pm - 5:10pm

Lattice H-matrices for Massively Parallel Micromagnetic Simulations of Current-induced Domain Wall Motion

^{1}The University of Tokyo, Japan; ^{2}Fujitsu Limited

This paper discusses parallel hierarchical-matrices (H-matrices) to compute a demagnetizing field, which is the most time-consuming part in the micromagnetic simulation of current-induced domain wall motion (CDWM). Although normal H-matrices exhibit high efficiencies for small numbers of Message Passing Interface (MPI) processes, the performance rapidly decays due to load imbalance and the MPI communication costs as the number of processes increases. We introduce lattice H-matrices to improve the parallel scalability, when using a large number of processes. The applicability of lattice H-matrices to CDWM simulations is confirmed and proper lattice block sizes and process grid shapes of the lattice H-matrices for H-matrix-vector products are investigated using practical data sets. Under appropriate settings, the lattice H-matrices show almost linear complexity in memory usage and calculation time of H-matrix-vector products. Our implementation continues to accelerate at least up to about 1,200 processes, even in a small problem with several tens of thousands of unknowns.

5:10pm - 5:30pm

Surrogate Model based on the POD combined with the RBF Interpolation of Nonlinear Magnetostatic FE model

Thomas Henneron^{1}, Antoine Pierquin^{2}, Stéphane Clénet^{1}

^{1}université Lille / L2EP, France; ^{2}Arts et Métiers ParisTech / L2EP, France

From solutions of Finite Element (FE) simulations, the Proper Orthogonal Decomposition (POD) is an interesting approach to express the solution into a reduced basis. In order to obtain a fast approximation of a FE solution, the POD can be combined with the Radial Basis Functions (RBF) method to interpolate the components of the solution associated with the reduced basis. In this communication, this POD-RBF approach is applied to a nonlinear magnetostatic problem and is used with a single phase transformer.

5:30pm - 5:50pm

Error Estimators for Proper Generalized Decomposition (PGD) in Time Dependent Electromagnetic Field Problems

Fabian Müller^{1}, Thomas Henneron^{2}, Stéphane Clénet^{3}, Kay Hameyer^{1}

^{1}Institute of Electrical Machines (IEM), RWTH Aachen University, Aachen, Germany; ^{2}EEA, Universit´e de Lille - Facult´e des sciences et technologies (FST), Lille, France; ^{3}L2EP Lille, Ecole Nationale Sup´erieure d’Arts et M´etiers, Lille, France

Due to fine discretization in space and time, the simulation of transient electromagnetic phenomena results in a large system of equations. To cope with this computational effort model order reduction techniques can be employed. To assess the accuracy of the solution of the reduced model, an error estimation is crucial. A commonly used approach consists of the evaluation of the deviation between the reduced and the full model. This yields a loss of the a-priori property of the PGD. To overcome this problem two a-priori criteria are presented in this paper.

5:50pm - 6:10pm

A Parallel Newton-Raphson Algorithm for Strong Coupling Modeling of Multi-Physics Problems

Amir Akbari, Dennis Giannacopoulos

McGill University, Canada

The Finite Element Gaussian Belief Propagation (FGaBP) method is an iterative algorithm with abundant parallelism making it an alternative for the traditional Finite Element Method (FEM), especially for large multi-physics problems. In this paper, we extend the FGaBP method to solve the coupled electrical-thermal problem that emerges in the modeling of radiofrequency ablation (RFA) of hepatic tumors. The strongest form of coupling algorithms which is the Newton-Raphson (NR) method is implemented in parallel using the localized computations of FGaBP. The parallel scalability of the FGaBP method is retained in the proposed algorithm by calculating local Jacobian matrices for each element and then updating the solutions for both electrical and thermal problems accordingly at each NR iteration.