Re: Force Calculus Methods

["rob_strandau" <rstrand@xxxxxxxxxxx>]

laurent_vp_fr wrote:

Hello,
I'm trying to evaluate the force acting on a solid in a magnetic field
and I don't obtain similar results with the Maxwell tensor and the
energy (or coenergy) variation.
Is there someone who can explain that or who have meet this problem ?
Thanks,
Laurent VP.

The basic problem is force computation is sensitive to small changes
in the numerical results. There is a section in the manual which
gives some recommendations.

I suggest you try doubling your mesh density in the appropriate areas
to see if the results change significantly. If either of the results
change you should try doubling it again until you get a stable result.

For the co-energy method try using larger or smaller pertubations to
see if your result is stable. You can also try pertubing the system in
both a positive and a negative direction then compute:

F = (Wc(p+dp) - Wc(p-dp))/(2 dp)

This method is likely to be more accurate and less susceptable to
numeric problems (you can also use larger dp's).

For the Maxwell tensor method follow the guide-lines in the manual. 
If you still have problems try defining three close contours (similar
to Fig 27 in the manual). Define a fine mesh in the two regions
enclosed by the coutours and use the central contour for the
integration. Double and halve the mesh densities in these regions to
check stability of the results. If you are still having problems try
to moving the three contours away from surfaces which have rapid
changes in the field magnitude - I've found cases where increasing the
mesh had little effect and only this produced a good result.

Regards
Rob