XANSYS Message: 3738 [Go back to message list]
[bookmark on del.icio.us]
No rating yet
Rate item:

Subject: Re: Physical Movement in an EMAG Analysis
Author: Mike Yaksh
Date: 1999-08-04 18:05:00

Wei:

Thanks for your comments. You are right. In order to easily identify the
bugs, I only listed the cases for locked rotor condition in my attachment.
Actually, in a comprehensive report on induction motor analysis to our
customer, I included all the cases of ANSYS linear, non-linear harmonic and
transient analysis with circuit (by using CIRC124). The effect of closed
and semi-closed rotor bar was also discussed. For a closed rotor slot, it
requires non-linear calculation as high saturation over the top of the
rotor slots.

Besides two bugs reported in my previous attachment, I found the torque
(locked rotor) calculated from non-linear harmonic analysis is about 14%
higher than that from a non-linear transient analysis (steady value after a
few cycles).

For torque-speed curve calculation, I used similar approaches as you
described, ie assuming supply frequency with modified rotor resistivity, or
slip frequency with modified supply voltage amplitude. I think the paper
(Weerdt et al), published on IEEE Transactions on Magnetics, vol.33, no.2,
1997, page2093-2095, is a good reference.

Response:

In performing the nonlinear AC analysis, the field variation is still
considered to a "sine", even though it is not (and the nonlinear transient
would simulate the "non-sine" variation of the field). In using the AC
assumption of the sine variation, an adjustment of the output quantities
is required (the B included). If there are no other differences between
the two runs, the 14% is probably due to the adjustment being made to
account for this difference between the sine and the non-sine behavior.

This issue arose when I was doing demos for ANSYS, and we were given closed
bar rotor problem. In the 95 time frame, I wrote a macro to allow
individual elements to have an independent permeability, based on the peak
field value. The transient was a possibility, but the time duration of the
runs caused us to consider using an AC approach. We were trying to
develope a design tool. The AC method gave the right behavior at the top
of the bar and it ensured consistency of the field with the BH data. The
number of iterations, even for the most severe case of saturation usually
was limited to 7 or 8. The torque was not seen to be extremely sensitive
to convergence once the field above the bar reached a certain level, but
what we saw was that the torque was consistently underestimated. The issue
of computing an RMS torque for non-sine variations could probably use
some more work.

An example:

(See attached file: testdmac.ppt)

About the 97 IEEE reference. The one I was thinking of is from the 93-94
time frame. It is good to see it being supported by others. Thanks.

Posts possibly associated with message #3738AuthorDateScore
3598Physical Movement in an EMAG Analysisderry.birse@1999/07/28 
3616Re: Physical Movement in an EMAG AnalysisMartin Liddle1999/07/29 
3626Re: Physical Movement in an EMAG AnalysisBill Bulat1999/07/2910
3665Re: Physical Movement in an EMAG AnalysisWei Wu wei.wu@1999/08/01 
3668Re: Physical Movement in an EMAG AnalysisMartin Liddle1999/08/02 
3681Re: Physical Movement in an EMAG AnalysisWei Wu wei.wu@1999/08/02 
3682Re: Physical Movement in an EMAG AnalysisBill Bulat1999/08/02 
3683Re: Physical Movement in an EMAG AnalysisWei Wu wei.wu@1999/08/02 
3702Re: Physical Movement in an EMAG AnalysisMike Yaksh1999/08/03 
3706Re: Physical Movement in an EMAG AnalysisBill Bulat1999/08/03 
3718Re: Physical Movement in an EMAG AnalysisWei Wu wei.wu@1999/08/04 
3738Re: Physical Movement in an EMAG AnalysisMike Yaksh1999/08/04