Improvement of the finite element method equations conditioning for the magnetic field-circuital problems

The presented systems with magnetically coupled windings are solved with the finite element method. If the issue of voltage supply is analyzed, a system of linear equations with a partially skew-symmetric sparse matrix is obtained. Iterative methods used to solve a system of equations are particular...

Full description

Bibliographic Details
Main Author: Gołębiowski Marek
Format: Article
Language:English
Published: Polish Academy of Sciences 2017-06-01
Series:Archives of Electrical Engineering
Subjects:
Online Access:http://www.degruyter.com/view/j/aee.2017.66.issue-2/aee-2017-0024/aee-2017-0024.xml?format=INT
Description
Summary:The presented systems with magnetically coupled windings are solved with the finite element method. If the issue of voltage supply is analyzed, a system of linear equations with a partially skew-symmetric sparse matrix is obtained. Iterative methods used to solve a system of equations are particularly effective for symmetric matrices. Resultant equations can be reduced to this symmetrical form by using the method known from the literature [1]. The ratio of the maximum to the minimum eigenvalue of the main matrix of this circuit, which is the condition number, is however very high. This means that the problem is ill-conditioned and leads to a very long iterative solution process. The method presented in the article allows for a direct solution of a system of equations on its part, corresponding to high eigenvalues of the system matrix. The remaining part of the system of equations is solved by iterative methods. This part has much better condition number, and therefore the computational process is fast. The proposed iterative process depends on multiplication of a sparse matrix by vectors. It is not necessary (and possible) to store the entire matrix. This is especially important for larger sizes of a matrix.
ISSN:2300-2506