[femm] Re: lamination fill factor

[David Meeker <dmeeker@xxxxxxxxxxxxxxxxx>]

Tecnico wrote:

> Hallo Dave,
>
> there is something that does not work as I expected in planar problems.
> Suppose that I have a square lamination with a central round hole. Through the hole there is a current.
> Lamination are angularly shifted of 45 degrees on the Z axis so as to obtain a star.
> Now, I set lamination fill factor to 1 for the octagon and 0.5 for the triangles, but it does not work:
> flux lines appear to be correct, but I expect that in triangular areas there is almost the same B as in the octagon since I have less iron.
> To bypass the problem I have two ways:
> 1- create a new material with a different B-H characteristic (one half of the original) so as to preserve the geometry and saturation, but with the wrong B
> 2- modify the geometry to preserve B and saturation
>
> Is this a bug or does fill factor work in a different way?

This isn't a bug. The lam fill factor actually works sort of like your (1). The way that the apparent properties are computed is explained in appendix A2 of
the manual. Basically, for your problem with laminations stacked in the z-direction, flux is always flowing in the "easy" direction, to use the nomenclature
from the manual.

The reason that the flux would seem to be low is that the values reported by the postprocessor are the average flux density, in which the thickness of the air
is taken into account in computing the average. For example, it might say that the flux density is 0.5 T at some point in your laminated section. If the
section had 50% fill, almost all of the flux would be carried in the iron, rather than the air. So you get about 1 T in the iron averaged with about 0 T in
the air between laminations to get an average of 0.5 T through the bulk laminated section.

For this particular problem, however, there might be some additional things to think about. Since there is invariably a small gap in between laminations your
stack, flux will pretty much stay confined to an individual lamination. The bulk lamination model implies some ability of the flux to rearrange itself in the
z-direction in the immediate neighborhood of the interface. This is reasonable if you have an interface with a laminated material on air or some other solid
material, but perhaps not so reasonable in your case, in which the low reluctance path is for the flux to stay within an individual lamination, rather than
jump laminations to take advantage of the "corners" of the laminations above and below. You might get a more accurate answer if you modeled a single lamination
in isolation (i.e. just a square region with a hole in the middle).

> I also have another problem:
> As I told you, I have written my own class library to read, write, and work with .fem and .ans files.
> My problem is that I can not get the correct result in calculation of torque from the stress tensor.
> The result that I get is wrong of 1e6 (probably due to some scale factor) plus 20-30% (it means that there is an error somewhere).
> I do the following:
> - find the edges over which the path lays
> - find the elements that share the edge and compute B as the mean of the B on the elements and H from permeability and B.
> - compute the torque on every edge of the path using the formula (26) of the manual
> - sum the results
>
> is there anything wrong on the above?

Eq. (26) really is the equation that femm uses to compute forces. Make sure that you are using absolute, rather than relative, permeability when you are
finding H (that is, absolute permeability = relative permeability * 4 * Pi * 1e-07). Keep careful track of your units as well when you differentiate A to get
B, because it is easy to get factors of 100 or 1000. To eliminate possible problems, convert all lengths, node positions, etc., to meters before you do any
calculations. Also, remember that to get torque, you have to include the moment arm that the force is acting upon--eq. (26) is just for the differential force
(dTorque = cross product of r and dF, where r is a position vector from the center of the rotor to the point at which dF is evaluated).

Dave.



begin:vcard 
n:Meeker;David
tel;fax:781-890-3489
tel;work:781-684-4070
x-mozilla-html:TRUE
url:http://members.aol.com/dcm3c
org:Foster-Miller, Inc.;Electrical and Electronic Systems Group
version:2.1
email;internet:dmeeker@xxxxxxxxxxxxxxxxx
title:Senior Engineer
adr;quoted-printable:;;350 Second Avenue=0D=0A;Waltham;MA;02451-1196;USA
fn:David Meeker, Ph.D.
end:vcard