[femm] Re: lamination core losses

[Dcm3c@xxxxxxx]

In a message dated 3/8/00 5:37:02 PM Eastern Standard Time, 
pvallian@xxxxxxxxxxxxxxx writes:

> how can i calculate a motor's core losses??
> i have the specific losses for the material

Hmm--good question. An approximate way that would get you into the right 
ballpark is to note that if the materials were linear (perhaps with some 
hysteresis that could be described as a simple phase lag), the losses at any 
point would scale with B^2. If this were true, you could find the 
root-mean-square flux density in the core of your machine and pick the point 
of your spec sheet corresponding to Brms at the frequency of interest. They 
usually give losses in Watts/lb (or perhaps Watts/Kg in parts of the world 
that know how to handle SI units), so to get the total loss, just multiply 
Watts/mass by the mass of the stator. Like I say, this is sort of rough, but 
it ought to get you pretty close.

A second way is to define the lamination properties in the Block Properties 
dialog corresponding to your material fully. Pick a lamination fill factor 
(1 isn't realistic; 0.9-0.95 might be more so; it depends on your lamination 
thickness and construction methods) and lamination thickness. Define a 
non-zero conductivity, so that eddy current effects are represented. Pick a 
reasonable hysteresis angle to include represent hysteresis losses. Then, 
evaluate the problem, and take the block integral of "hysteresis and/or 
laminated eddy current losses" over the volume of the core. This includes 
the effect that the hysteresis and eddy currents have on the field 
distribution, which can change flux levels and therefore losses. The down 
side of this method is that it uses sort of an overly simplistic material 
model--linear BH curve and hysteresis represented by a frequency- and 
amplitude-independent phase lag. However, with a little head scratching, you 
may be able to back out values for permeability, hysteresis lag, and 
conductivity that reproduce the manufacturer's loss curves relatively well, 
implying that good results might be obtained by this method.

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