Conference Agenda

Generation of adapted finite element meshes in electromagnetic computations is one of the major challenges today. This work paper aims to introduce an adaptive remeshing procedure based on error estimation. This procedure will focus on the mesh size adaptation to distribute the error uniformly over a computational domain to control the accuracy of the solution. It shall also enable dealing with complex geometries for electromagnetic-coupled material processing applications. For this purpose, a quasi-steady state approximation of the Maxwell's equations in a time-domain formalism is considered. The automatic remeshing procedure is based on the following key steps: a posteriori error estimator to pinpoint the critical areas requiring refinement or allowing coarsening, anisotropic metric computation based on Hessian approximation. Both steps use a global field recovery algorithm in order to enable accurate gradient computation. The procedure is then validated on test cases.

The calculation of steady-state on-load operating points is important to properly start transient finite-element simulations of synchronous generators. Although the steady-state of the machine can be obtained from a magnetostatic approach, the field current and load angle are unknown. So, additional constraints are required for a given load at the stator terminals, actually making the problem a circuit-field coupled one. Several approaches have been proposed but most of them require multiple solutions. A single nonlinear solution methodology is proposed here using generalized two-axis winding vectors. The formulation has been coded and incorporated into our finite-element FORTRAN software. It is called FLD, which is able to solve general 2D electromagnetic problems that include circuit-field coupled problems. The availability of steady-state experimental data for a 150 MVA turbine generator allows the validation of our proposal. The study highlights the easiness of incorporating initial condition calculations within available finite element software.

This digest proposes a sensitive approach to identify transformer interturn faults by detecting the terminal current distortions caused by the zero-off effect of the fault arc. First, a coupled field–circuit model for interturn faults is developed, in which the fault current is calculated simultaneously with the leakage flux. Second, an improved arc model based on the arc diameter is used to calculate the time-varying arc resistance, whose differential equation is solved by user-defined functions (UDFs) in the transient coupled simulation. Third, the improved complete ensemble empirical mode decomposition with adaptive noise (ICEEMDAN)-based algorithm is compiled to extract fault characteristics. A 40 MVA/110 kV three-phase power transformer is modeled and subjected to interturn faults under different conditions, and detailed results indicate that faults shorting a low number of turns can be sensitively identified.

Unwanted arcing can have serious adverse effects on electrical and electronic equipment and systems. Multiphysics simulation of transient arc processes is important for understanding and predicting arc dynamics. Simulation results can be used to innovate and optimize designs prior to prototyping. However, reproducible numerical studies of transient arcs are very limited in the literature. To bridge the gap, this study aims to provide detailed results from experimental studies and numerical simulations. The results can be used to verify and validate computer code for simulating arcing and other magnetohydrodynamic problems.

This research proposed a coupled electromagnetic-thermal analysis of a permanent magnet synchronous motor (PMSM) for the application of fully electric ships. By considering the temperature change of the motor, the accuracy of motor characteristics prediction according to the driving cycle can be enhanced. Compared to other coupled analysis studies, the proposed method considers updating the electromagnetic-thermal properties at each time step to improve simulation accuracy. Additionally, the coupled analysis considers changing cooling conditions to incorporate the variation in the convection coefficient. This technology expands the limit of the motor during driving cycle research by enabling it to operate at different loads and cooling conditions. The experimental validation was conducted with an acceptable error of 5⁰C.

This paper presents the stresses and displacements of NI REBCO pancake coils generating ultrahigh magnetic field. We have investigated the effect of electromagnetic forces on the coil geometry in simulation, for different fixed boundary and BJR conditions. To reduce a hoop stress due to the radial electromagnetic forces, each turn of the coil moves in the circumferential direction separately. It is shown that NI REBCO pancake coils may be distorted by unbalanced electromagnetic forces.

This paper presents a new calculation method based on LPTN for the calculation of heat dissipation in axial flux permanent magnet synchronous motor. Fast analysis of transient temperatures and improved model accuracy compared to previous LPTN modeling. An accurate convective heat transfer coefficient is used, and the fluid temperature is used as the reference temperature instead of the ambient temperature. Experimental results show that the method has good accuracy and can greatly reduce the time required for thermal design of the motor.

This paper proposes to apply a multiphysics FEM code to model magnetoelectric composite rings by a 2D axisymmetric approach. The formulation model includes an anhysteretic simplified magneto-elastic Gibbs free energy model to consider the magnteostrictive non-linearity effect. This study provides a useful tool to study the ME as energy transducers for engineering devices.

Numerical dosimetry is commonly used to evaluate human exposure to electromagnetic fields. In the low frequency range the induced electric field in the human body is computed and compared with standardized basic restrictions. Basic restrictions are defined starting from actual stimulation thresholds, considering additional safety factors to take into account several sources of uncertainty. As a consequence, numerical dosimetry leads to a conservative estimation of the exposure. In this paper we propose a method to couple the field problem solved through numerical dosimetry with a circuit model of the peripheral nervous system. With this approach it is not necessary to look for the maximum induced electric field in the human body because the actual values affecting the nerves are used to drive a circuit model that makes it possible to understand if a stimulation occurs or not.

In this study, through a series of numerical electromagnetic simulations, we have quantitatively evaluated the RF heating caused by the Spinal Cord Stimulation (SCS) implant in the MRI environment. An adult human model was used with an implantable pulse generator (IPG) implanted, and local Specific Absorption Rate (SAR) values and temperature rising were quantitatively estimated at 1.5 and 3 Tesla (T). This method can significantly reduce the uncertainty of the temperature probe placement, as well as the number of devices that need to be tested. Consequently, it can significantly reduce the time and expense of testing performed to evaluate MRI-related heating.

Since deep brain stimulation has become increasingly relevant for clinical practice, computational models and strategies to validate them against experimental results are essential. However, more research dealing with validating these models is needed. We use impedance data from in-vitro and in-vivo experiments for validation. For that, we characterized the electrodes in a well-known saline solution with impedance spectroscopy before implanting them in a hemiparkinsonian rat model. Validating the computational model of the electrode against the impedance data from the in-vitro studies ensures that we take the uncertainties in the electrode design into account when modeling the in-vivo studies. Further, we show that these uncertainties in the electrode design lead to different impedance spectra during electrode characterization and significantly affect the prediction of the outcome of DBS treatments. The build-up of an encapsulation layer can be inferred from the experimental data, and the dielectric properties of the layer become accessible through the simulations.

Well-established forward modeling methods in electrocardiography (ECG) require fine meshes to calculate the electric scalar potential at the body surface with high accuracy. We introduce a fast fictitious surface charge method with local mesh refinement and smart calculations of elements interactions which improves the accuracy of the calculations and, at the same time, preserves the performance speed.

In this paper, finite element analysis was used to study and analyze the electrical conductivity profile of several stroke patients based on clinical data obtained using electrical impedance tomography. Weight factors used for optimization can be found by finite element analysis of a head model. The genetic algorithm was applied with the weight factors to identify the conductivity distribution inside the brain by minimizing the difference between the simulated and measured voltages under stroke conditions. The preliminary result shows that it might be possible to reconstruct an approximate electrical conductivity distribution based on clinical stroke data.

The fictitious surface charge method (FSCM) is enhanced by a fast multipole method (FMM) for the calculation of the induced

electrical field in transcranial magnetic stimulation. The method was embedded and optimized in Python. An element-wise Jacobi

method was combined with vectorized matrix operations to increase the parallelization capabilities and enable GPU computing. The

induced fields are compared against an analytical solution for a homogeneous sphere. The results show that a normalized root mean

square error of 0.12 % can be achieved with the integral-free FSCM-FMM formulation on low-performance hardware.

Evoked compound action potential or ECAP describes the electrically evoked response of the auditory nerves excited. ECAP is an important physiological signal in cochlear implants (CI) for observing compound neural responses. Current steering ECAP provides more information about the neural population of cochlear implant patients. This digest uses the electrical current steering method to stimulate ECAP using the volume conduction method from a pair of electrodes instead of the traditional single electrode. ECAP obtained from the sensing electrode was processed using the alternating polarities method. In addition, the relationship between current steering ECAP and different current levels were studied. Preliminary results show that when the stimulation current was increased, the ECAP amplitudes would increase correspondingly, which agrees with typical clinical findings. The full paper will present a quantitative comparison of the current steering ECAP modeling result and clinical measurements.

Computation of low frequency induced electric fields in the human body requires the knowledge of the source field. The source field can be reconstructed by measuring the magnetic field over a regular grid, or by identifying a model, among which multipoles. Spherical multipoles are easy to compute, but cannot be used close to the source. In this work it is proposed to use a new multipolar model which holds outside an arbitrary geometry -- not limited to a sphere.