[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [FEMM] A DOT J Integral over laminated region
Thomas Wilson wrote:
I am using the detailed flux field from FEMM to compute lumped
parameter coefficients for an external model. The lumped parameter
model includes eddy current circuits so I am evaluating the mutual
inductance between a coil and the region in which the eddy current
flows. The FEMM calculation is axisymmetric and harmonic. The main
calculation for the lumped parameter mutual inductance between the
coils and eddy currents involve the A dot J block integral. Things
work as expected for continuous materials but when I specify
"laminated in plane" for the region in which the eddy currents flow
then FEMM calculates that the current density and the A dot J integral
are zero in that region. Laminations reduce eddy currents but it does
not make them zero. So this is not the result that I expect or can
use. Any help on how to correct the problem or explanations as to why
it correct as is?
The "laminated in plane" case corresponds to a radially laminated
configuration for axisymmetric problems. Although this is difficult to
construct in practice, I've seen some specialty machines that have been
built in this way. With a radial lamination direction, there is no net
current in the tangential direction--thus the zero result for the A.J
integral. All of the eddy currents are conserved inside each
lamination. However, the eddy currents in these laminations do affect
the field. The eddy currents make the laminated core appear to have a
permeability that is of a lower magnitude than the DC permeability, and
they also give a complex component to the permeability. The complex
part of the permeability is associated with eddy current (and/or
hysteresis) losses.
At any rate, a good way to take these effect into account is by their
effect on the flux linkage of the coil that drives the eddy currents.
The eddy currents (in both laminated and non-laminated regions) make the
flux linkage of the driving coil have a frequency dependency. You then
assume a lumped-parameter model structure that yields the observed
behavior (typically a chain of inductors and resistors like the one that
I used in http://femm.berlios.de/dmeeker/pdf/chop2.pdf) and then pick
the parameters to be a best fit to the magnitude and phase of the flux
linkage as evaulated at a variety of different frequencies.
Dave.
--
David Meeker
Senior Engineer
Foster-Miller, Inc.
350 Second Avenue
Waltham, MA 02451-1196
781-684-4070
781-890-3489 (fax)
dmeeker@xxxxxxxxxxxxxxxxx