Title :
Field smoothing and adaptive mesh generation
Author :
Lowther, D.A. ; Rong, R. ; Forghani, B.
Author_Institution :
CAD Lab., McGill Univ., Montreal, Que., Canada
fDate :
3/1/1993 12:00:00 AM
Abstract :
A cheap smoothing process is described which both improves results from a standard first order solver and forms the basis of an effective adaptive solution system. The resulting system is demonstrated on several devices and is shown to develop accurate results in the presence of mixed media and with a very coarse starting discretization. The adaptive solver can be made to converge on a specified output result; either global, e.g., inductance or force, or local, e.g., the flux density in a particular region. While the criterion for measuring error seems somewhat simplistic, it can generate good results, particularly when combined with a convergence criterion based on the desired results of the analysis
Keywords :
convergence of numerical methods; electromagnetic devices; error analysis; iterative methods; magnetic flux; mesh generation; adaptive mesh generation; adaptive solution system; convergence criterion; error; field smoothing; flux density; force; global output result; inductance; iteration; local output result; magnetic device; magnetic field; mixed media; polynomials; very coarse starting discretization; Finite difference methods; Finite element methods; Integral equations; Laboratories; Magnetic analysis; Magnetic fields; Mesh generation; Performance analysis; Polynomials; Smoothing methods;
Journal_Title :
Magnetics, IEEE Transactions on