WhitneyAVHarmonic - low-frequency homogenization

General discussion about Elmer
Post Reply
spanda
Posts: 11
Joined: 19 Feb 2018, 14:24
Antispam: Yes

WhitneyAVHarmonic - low-frequency homogenization

Post by spanda »

Hi,

I see that in MagnetoDynamics module low-frequency and wide-frequency models are implemented for enabling homogenization of laminated stack. What is with post-processing? I can not find any piece of code in SUBROUTINE MagnetoDynamicsCalcFields that takes into account eventually homogenized field in finite elements that belong to laminated stack when calculating total eddy current loss in laminated stack.
Am i missing something?
It seems to me that eddy currents loss in homogenized stack can not be calculated in standard way by P=0.5Re{e * conjugate(j)} but it is needed to use following equation (from 1999. IEEE article "Calculation of Eddy Currents and Associated Losses in Electrical Steel Laminations", Gyselinck, Dular)
P=(1/6)*sigma*(pi^2)*(LamThick^2)*(freq^2)*(Bre^2+Bim^2)
or, written in another way
P=0.5*sigma*(omega^2)*(LamThick^2)/12*(rot(Are)^2+rot(Aim)^2)

I hope my question is clear enough. So, am i wrong?

English is not my mother tongue so sorry for eventual mistakes :)
Thank you in advance

Spanda
Takala
Posts: 186
Joined: 23 Aug 2009, 23:59

Re: WhitneyAVHarmonic - low-frequency homogenization

Post by Takala »

Dear Spanda,

it seems that no-one has been interested in these losses in detail before (the total losses can be extracted from circuits as well), thank you for pointing out this short coming. You are partly right, the equation that you mention indeed is good to go for the low frequency model. But for the wide frequency model one needs to derive the loss formula from the more generic form:
P = 1/T \int_0^T H \cdot \frac{\partial B}{\partial t} \mathrm{d}t

Also note that the low frequency model is also available in transient simulations (again the general formula should be used).

Now the details of the implemented wide frequency formula can be found from
IEEE Trans. Magn. 39(3), May 2003
but unfortunately the losses are not derived (AFAICT).

I think this should be done soon since it is not so difficult task.

/Eelis
Post Reply