Search results

Two sets of dual magnetodynamic and magnetostatic finite element formulations taking thin conducting magnetic shells into account are proposed. The abstraction of the thin region from the computational domain is performed by an appropriate treatment of the surface integral terms arising in the weak formulations. Results are presented for two three‐dimensional test‐problems.

The present paper deals with the fast multipole acceleration of the 2D finite element‐boundary element modelling of electromechanical devices. It is shown that the fast multipole method, usually applied to large 3D problems, can also lead to a reduction in computational time when dealing with relatively small 2D problems, provided that an adaptive truncation scheme for the expansion of the 2D Laplace Green function is used. As an application example, the 2D hybrid modelling of a linear actuator is studied, taking into account saturation, the voltage supply and the mechanical equation. The computational cost without and with fast multipole acceleration is discussed for both the linear and nonlinear case.

This paper investigates new technological devices to be utilised in future optical communications, by means of variational method (FEM) and multipole scattering approach (Rayleigh method). This last one provides interesting asymptotic results in the long‐wavelength limit. The so‐called photonic crystal fibres (PCF) possess radically different guiding properties due to photonic band gap guidance: removing a hole within a macro‐cell leads to a defect state within the gap. In the case of multi‐core PCF, the localised modes start talking to each other which possibly leads to a new generation of multiplexer/demultiplexers.

This paper is devoted to the presentation of a new finite element formulation for spectral problems arising in the determination of propagating modes in dielectric waveguides and particularly in optical fibers. As an example, we compute the coupling between two parallel optical wave guides. The originality of the paper lies in the fact that we take into account both the vector character of the problem (no weak coupling assumption) and the unboundness of the domain.

This paper deals with the two‐dimensional finite element analysis in the frequency domain of saturated electromagnetic devices coupled to electrical circuits comprising nonlinear resistive and inductive components. The resulting system of nonlinear algebraic equations is solved straightforwardly by means of the Newton‐Raphson method. As an application example we consider a three‐phase transformer feeding a nonlinear RL load through a six‐pulse diode rectifier. The harmonic balance results are compared to those obtained with time‐stepping and the computational cost is briefly discussed.

Connection boundary conditions are studied with the finite element method using different types of mixed finite elements, i.e. nodal, edge and facet elements of different shapes and degrees, used in both b‐ and h‐conform formulations. The developed associated tools are first applied to periodicity boundary conditions before being applied to the treatment of the moving band in 2D and 3D. This step by step approach enables their validation before pointing out the effect of the considered elements on the accuracy of the moving band method. A special attention is given to the consistency of these boundary conditions with gauge conditions and source magnetic fields.

The purpose of this paper is to deal with the correction of the inaccuracies near edges and corners arising from thin shell models by means of an iterative finite element subproblem method. Classical thin shell approximations of conducting and/or magnetic regions replace the thin regions with impedance-type transmission conditions across surfaces, which introduce errors in the computation of the field distribution and Joule losses near edges and corners.

In the proposed approach local corrections around edges and corners are coupled to the thin shell models in an iterative procedure (each subproblem being influenced by the others), allowing to combine the efficiency of the thin shell approach with the accuracy of the full modelling of edge and corner effects.

The method is based on a thin shell solution in a complete problem, where conductive thin regions have been extracted and replaced by surfaces but strongly neglect errors on computation of the field distribution and Joule losses near edges and corners.

This model is only limited to thin shell models by means of an iterative finite element subproblem method.

The developed method is considered to couple subproblems in two-way coupling correction, where each solution is influenced by all the others. This means that an iterative procedure between the subproblems must be required to obtain an accurate (convergence) solution that defines as a series of corrections.

To understand the behavior of the magnetization processes in ferromagnetic materials in function of temperature, a temperature-dependent hysteresis model is necessary. This study aims to investigate how temperature can be accounted for in the energy-based hysteresis model, via an appropriate parameter identification and interpolation procedure.

The hysteresis model used for simulating the material response is energy-consistent and relies on thermodynamic principles. The material parameters have been identified by unidirectional alternating measurements, and the model has been tested for both simple and complex excitation waveforms. Measurements and simulations have been performed on a soft ferrite toroidal sample characterized in a wide temperature range.

The analysis shows that the model is able to represent accurately arbitrary excitation waveforms in function of temperature. The identification method used to determine the model parameters has proven its robustness: starting from simple excitation waveforms, the complex ones can be simulated precisely.

As parameters vary depending on temperature, a new parameter variation law in function of temperature has been proposed.

A complete static hysteresis model able to take the temperature into account is now available. The identification is quite simple and requires very few measurements at different temperatures.

The results suggest that it is possible to predict magnetization curves within the measured range, starting from a reduced set of measured data.

This paper aims to model a three-dimensional twisted geometry of a twisted pair studied in an electrostatic approximation using only two-dimensional (2D) finite elements.

The proposed method is based on the reformulation of the weak formulation of the electrostatics problem to deal with twisted geometries only in 2D.

The method is based on a change of coordinates and enables a faster computational time as well as a high accuracy.

The effectiveness of the adopted approach is demonstrated by studying different configurations related to the IEC 60851-5 standard defined for the measurement of the electrical properties of the insulation of the winding wires used in electrical machines.

Finite element (FE) models are considered for the penetration of magnetic flux in type-II superconductor films. A shell transformation allows boundary conditions to be applied at infinity with no truncation approximation. This paper aims to determine the accuracy and efficiency of shell transformation techniques in such non-linear eddy current problems.

A three-dimensional H – ϕ formulation is considered, where the reaction field is calculated in the presence of a uniform applied field. The shell transformation is used in the far-field region, and the uniform applied field is introduced through surface terms, so as to avoid infinite energy terms. The resulting field distributions are compared against known solutions for different geometries (thin disks and thin strips in the critical state, square thin films). The influence of the shape, size and mesh quality of the far-field regions are discussed.

The formulation is shown to provide accurate results for a number of film geometries and shell transformation shapes. The size of the far-field region has to be chosen in such a way to properly capture the asymptotic decay of the fields, and a practical procedure to determine this size is provided.

The importance of the size of the far-field region in a shell transformation and its proximity to the conducting domains are both highlighted. This paper also provides a numerical way to apply a constant magnetic field in a given region, while the source, on which only the far-field behaviour of the applied field depends, is excluded from the model.