Tadashi Naito

Time-periodic finite element method for nonlinear diffusion equations

A new numerical calculation method, the time-periodic finite element method, for the nonlinear diffusion equation is proposed. In the method, the equation is closed by the periodic-time of electrical characteristics and is solved by the numerical procedure which is used not for the initial value problem but for the boundary value one. The method is applied to the calculations of the voltage across the nonlinear corona shield region of a rotating machine. The results demonstrate that the method extensively reduces the computing time without losing the accuracy compared to the conventionally used methods.

Computation accuracies of boundary element method and finite element method in transient eddy current analysis

The computation accuracies of the boundary-element method (BEM) and finite-element method (FEM) in transient eddy-current problems are compared by using a slot-embedded conductor model and a diffusion model that can be solved theoretically. For computing the vector potential or magnetic flux density it is shown that larger time-step width can be chosen in the BEM than in the FEM method for the same accuracy. >

Three-dimensional analysis of eddy current and electromagnetic force in cold crucibles

Three-dimensional eddy current distributions in cold crucibles are analyzed by using an integro-differential method. The electromagnetic force is calculated from the obtained eddy current distributions. In this paper, computations of eddy current and levitation force of the molten metal in some cold crucible models are performed, and the influences of coil position, the number of segments of the crucible, frequency of the coil current and gap length between the segments on the levitation force are discussed. Furthermore, an experimental verification is presented. >

Reduction of vector unknowns using geometric symmetry in a triangular-patch moment method for electromagnetic scattering and radiation analysis

A simplification method for the boundary element model developed by H. Tsuboi et al. (1990) is applied to the triangular-patch moment method in order to reduce the number of vector unknowns. The simplification of the computational model is performed by a variable transformation using the spatial eigenmodes, which is similar to transformation used in the boundary element case. However, the unknown vectors have to be arranged according to the geometry of the rotational symmetry because of the varying directions of the unknown vectors. The applicability of the simplification method was verified by a flat square plate model and a TEM (transverse electromagnetic) cell model. >

A new approach to the transient analysis of magnetic field with variable transformation based on eigenvalues

A novel approach to the transient analysis of a magnetic field is presented that uses the formulation of a boundary-element model in two-dimensional problems. In order to solve the transient problems, state equations are formed using the boundary element model, and are solved using the variable transformation based on eigenvalues. The computational accuracy of the proposed method is evaluated by an infinite plate model and an infinite cylinder model in which the impressed magnetic flux densities are given by a unit step function and a sinusoidal function respectively. >

A simplification method for reflective and rotational symmetry model in electromagnetic field analysis

In this paper, a simplification method for reflective and rotational symmetry model, is proposed. Using spatial eigenmodes, the coefficient matrix of the final simultaneous equations for the time and storage capacity can be reduced. The proposed method can adapt to not only an integral-equation-method model but also finite element method model.

Simplification of symmetric boundary element model including asymmetric parts

The simplification method for the boundary element model with rotational symmetry is expanded and applied to the boundary element model including asymmetrical parts. The expanded simplification method is performed by a variable transformation which is similar to the variable transformation for the rotational symmetry model. The coefficient matrix of the final simultaneous equations is transformed into a bordered block diagonal matrix. The reduction ratios of the calculation time are shown using the number of symmetrical subregions, the numer of unknowns in the fundamental part, and the number of unknowns in the asymmetrical part. The proposed simplification method is effective for large-scale problems using a large number of unknowns. The applicability of the proposed method was verified by a numerical example: a three-core cable model including a symmetric part for 2-D eddy current analysis using the z-component of the vector potential. >