Re: some proposals for future developments

["atreding" <atreding@xxxxxxxxxx>]

--- In femm@xxxxxxxxxxxxxxx, David Meeker <dmeeker@xxxx> wrote:
> Hi Dave - I'm new to the group but I've been watching the posts for 
a while now and I downloaded the "beta" version of Femm 4.0 - my 
question is since the lua scripts from 3.3 are no longer going to 
work with the 4.0 version anyway, have you considered moving to Lua 
5.0 for the new program? It would be a good time to do it since 
scripts need to be rewritten anyway.

P.S. - This program is great - fully the equal of commercial magnetic 
software I have used. Thanks!

Also, for everyone else - just a question that has been eating at me 
for a while... We have been using Femm as well as Ansoft Maxwell for 
modeling solenoid actuators. The results agree well with each other 
in general, and when we have actually built the devices for test the 
models do a pretty fair job of predicting forces. However, in 
solenoids where the iron is not very strongly magnetized (B field far 
from saturation, ie. 1T or less) the results overestimate by a large 
amount, as much as 30%. While for saturated designs, the accuracy can 
be 5% or even better. This is true of both of the modeling programs 
we have used - they both overestimate force at low saturation and 
they both do so to roughly the same degree - that is they both give 
similar wrong answers.

We have gotten around this to some extent by simply overengineering 
the designs when the saturation is low, as well as intentionally 
designing for saturation or near to it. But it is an irritating 
phenomenon and I was wondering if anyone else has ever seen this 
problem. Our thought is that at saturation large errors in H cause 
only very small changes in B, therefore the force calculation is more 
numerically robust when problems are solved at saturation, versus at 
peak permiability where a tiny shift in H could cause a huge shift in 
B. But this would seem to suggest that we should see large errors of 
underprediction as often as overprediction - and we don't. It always 
overestimates.

Anyone have any thoughts on this?
Thanks, Andy Reding