Re: [femm] Re: Incorrect force direction

[David Meeker <dmeeker@xxxxxxxxxxxxxxxxx>]

Rob wrote:

> --- In femm@xxxxxxxxxxx, "Greg Watson" <gowatson@xxxx> wrote:
> > One of the basics of my "seat of the pants" approach to estimating
> ferromagnetic
> > force direction is that a MMF source will be drawn toward a area of
> lower
> > magnetic reluctance.
> >
> > The area of the rails and volume of the air to the right of the
> magnet should
> > show lower magnetic reluctance than that to the left of the magnet
> due to the
> > larger amount of inward facing rail surface and larger air volume
> between the
> > parallel rails and thus the magnet should be drawn into a equal &
> balanced
> > magnetic reluctance neutral point in the horizontal centre of the
> parallel bars.
>
> I tend to agree with this, that is what my intuition says also, which
> is why this is an interesting problem. I think what the calculation
> might be saying is that the difference in reluctance between the 2
> sides becomes small enough to be down in the noise sooner than when
> your intuition might lead you to believe... I'm tempted to build this
> myself now and play around with it.

As you say, this is sort an interesting problem because it's simple and
does something that could be a priori unexpected. It's highlighted some
of the strengths of a finite element approach (doesn't assume much about
what the problem "should" be doing) as well as some of the weaknesses
(e.g. a FEA solution will always be only approximate, and FEA doens't
necessarily give any insight into "why" a configuration is behaving the
way it is).

Anyhow, the trick here is that the _total_ reluctance that the magnet
sees doesn't really vary with position once the magnet is inside the
bars. Since we've come this far, I might as well obsess about this
problem for a bit longer. I've posted a quick note in the egroups
"files" section that looks at a simple circuit representation of this
configuration and shows why I would only expect very small centering
forces when the magnet is within the bars. Hopefully, the circuit model
is simple enough to lend some intuitive feel. Check out:
http://www.egroups.com/files/femm/circuitmodel.pdf

In femm@xxxxxxxxxxx, "Greg Watson" <gowatson@xxxx> wrote:
> My concern was not that FEMM should report a large centring force but
that a
> program which I normally have found to be very realistic would report a

> repulsive & not a centring force, no matter how fine the mesh or how
much care I
> took in making sure the geometry of the mesh and / or of the model
didn't effect
> the result.

Again, this sort of odd result is not unreasonable because the "true"
force is smaller than the accuracy of the calculation. For example, if
the "true" result were, for example, 0.01 with an accuracy of plus or
minus 0.1, any result on the range from -0.099 to 0.11 is possible
without it being a "bug". You will _always_ be able to find an example
that has this behavior, because finite elements are only an approximation
to the "true" solution to a differential equation. That said, I'll look
into other methods for force calculation other than stress tensor or
gradients of coenergy that might have a better rate of convergence.

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