Title :
Analysis of convergence in nonlinear magnetostatic finite elements problems
Author :
Neagoe, C. ; Ossart, F.
Author_Institution :
Lab. d´´Electrotech., CNRS, Grenoble, France
fDate :
9/1/1994 12:00:00 AM
Abstract :
Deals with convergence difficulties using Newton-Raphson method in nonlinear magnetostatic problems. It is shown use of magnetostatic scalar potential can lead to a very unstable iterative process because of the shape of the residual function which is to be cancelled. Such effects do not exist when the vector potential is used and Newton-Raphson method is much more efficient. A simple example points out the behavior of Newton-Raphson method for both formulations. A method for reducing the CPU time required for determining the relaxation factor used to insure convergence in the case of scalar potential is also presented
Keywords :
convergence of numerical methods; finite element analysis; iterative methods; magnetostatics; numerical analysis; Newton-Raphson method; convergence; finite elements problems; magnetostatic scalar potential; nonlinear magnetostatics; relaxation factor; residual function; unstable iterative process; Convergence; Equations; Finite element methods; Jacobian matrices; Lead; Magnetic analysis; Magnetostatics; Newton method; Shape; Vectors;
Journal_Title :
Magnetics, IEEE Transactions on