Re: [femm] motion-induced torque

[dcm3c@xxxxxxx]

In a message dated 5/24/00 4:18:29 PM Eastern Daylight Time, 
phenning@xxxxxxxxxxxxxxxxxxx writes:

> I was curious if there was anything miraculous about the ability to 
> compute motion-induced torques etc. using FEA, for example computing 
> the torque output of an induction motor at a steady-state slip. 
> Several commercial products do this with "rotating airgaps", etc. 
> Are these meshing and solution algorithms above and beyond the scope 
> of shareware, involving very new ideas, or has it just not been 
> gotten around to yet? As a novice I was just wondering.

Well, the techniques that one would use to include this sort of transient 
analysis with motion aren't that mysterious. It would just take quite a bit 
of time do (i.e., nobody has gotten around to it yet in a freely available 
program, that I know of). The trick is that you have to end up doing a 
spice-like simulation of the motor and the electronics that drive it (or at 
least, some simplified representation of the electronics). The sort of 
analysis that femm does would be just one step in that analysis--determining 
the magnetic couplings at each time step. Anyhow, the above comments are 
considering a problem that is basically magnetostatic, but with a geometry 
that changes due to motion. That is, the currents in the finite element 
model that can change over time have dynamics that are part of externally 
modeled circuits--the magnetic analysis at each time step is basically a 
magnetostatic analysis.

A related issue that actually is more difficult is the inclusion of 
motion-induced eddy currents. Instead of just solving series of relatively 
benign magnetostatic problems, motion-induced eddy currents mean that the 
problem is governed by equations with strong convective terms, which can be 
quite difficult to model accurately. I don't know of a commercial program 
that does really well in eddy current problems with high-speed motion.

Anyhow, femm often can be used as an aid to transient simulations. I often 
use it to obtain estimates of parameters for analytical transient machine 
models. The analytical model then can be used to answer questions about 
transient performance. However, this is not always simple to do--it requires 
a pretty good understanding to the analytical models and the relationships 
between the model's parameters and the results of finite element solutions 
(i.e. hard for beginners).

Dave.
--
http://members.aol.com/dcm3c