Re: [femm] Non linear Anysotropic analysis

[David Meeker <dmeeker@xxxxxxxx>]

I think that what Daniele is really after is the ability to model grain-oriented
steels.  Laminations made out of grain-oriented steels have better magnetic
properties in the rolling direction and opposed to the transverse direction.
 Although you could sort of account for it by specifying linear materials
and iteratively changing the permeability, this isn't that satisfying because
you probably can't get that accurate an answer if there are only a small
number of blocks whose properties you have to tweak by hand.  

Anyhow, there are many papers that describe various approaches to modeling
anisotropic laminations in Transactions on Magnetics. FEMM doesn't currently
implement any of those approaches for modeling nonlinear anisotropic materials,
although it would be an interesting area for future work. 

What is included in the program is a model of "on edge" laminations,
which is a continuum model of laminations composed of isotropic material
with small air gaps between the laminations. (n.b. The reason that
I did this originally was to analyze a common way of designing permanent magnet
biased radial magnetic bearings, where a portion of the PM flux path is driven
across the "hard" lamination direction.)  This sort of lamination model isn't
really new--a quick search just now on IEEExplore turned up:

Nonlinear three dimensional field computation
methods in laminated iron cores under saturated conditions
Mohammed, O.; Demerdash, N.; Nehl, T.;
Magnetics, IEEE Transactions on 
, Volume: 19	
Issue: 5
,
Sep 1983	
Page(s): 2091 
-2093

which uses more or less the same scheme I'm doing.  However, it's sort of
an interesting idea to intentionally mis-use this to model a grain-oriented
steel in a magnetostatic simulation--that hadn't occurred to me before. Since
you only have one parameter to dictate the anisotropy over the entire range
of the B-H curve (i.e. the lam fill factor), there might not be enough functional
"richness" to match experimentally obtained B-H curves.  (Does anyone else
have a feel for the adequacy of this approximation?) As long as you are satisfied
with the dependency between the hard and easy directions that is implied
by the lamination model, this might be good enough for your purposes.

I put up a little note on mis-using the lamination model in this way at http://femm.berlios.de/aniso.zip
 This note has some thoughts about how you'd go about picking an appropriate
lam fill factor and looks at an example case.

Dave.




Dave Squires wrote:
> Sure. Just
> go into the properties on any
> of the materials and change the relative permeability
> for x or y directions and watch what happens.
> Say if you wanted to mimic grain oriented Si steel.
> Pick M-19 steel and change the mu for Y to 440 or
> so (10 times less relative mu).  Run a simulation
> with PM or EM flux coming in cross grain (Y direction)
> and poles offset.  Just do a simple one first to
> see the effect.
>
> Dave Squires
>
> brendon2077 wrote:
> > Hello, 
> > 
> > I' writing from Italy, and my problem is the following... I have to 
> > perform a non-linear simulation with Femm, but the material in use is
> > not isotropic (I think Femm doesn't allow this)...is there anyone 
> > that knows some "trick" about it? (for example, to assign a 
> > lamination that "simulate" the anysotropic behaviour)...
> > 
> > If in any case this analysis can't run with current version of femm,
> > I hope it will become developed in future...
> > 
> > Best regards, 
> > 
> > Daniele
> >


-- 
David Meeker
email: dmeeker@xxxxxxxx
www: http://femm.berlios.de/dmeeker