Re: [FEMM] A little paradox with anti-periodic boundary condition and torque computation

[David Meeker <dmeeker@xxxxxxxxxxxxxxxxx>]

It's fairly hard to confuse the "weighted stress tensor" algorithm, but 
not impossible.  In this case, by selecting the part of interest as well 
as all the air around it, you haven't really left the program anywhere 
to put contours around the rotor in a programmatic way.  You can 
visualize the contours along which the integration takes place by 
selecting the "Show stress tensor mask" block in the View|Contour Plot 
menu in the postprocessor.  If you had selected either the windings and 
iron of the stator or the iron of the rotor by themselves (i.e. with no 
additional air), you'd come up with reasonable results in both cases.

As far as Ht, the smoothing scheme that femm uses yields values of B and 
H that are of lower order accuracy in the direction tangential to the 
boundary. This is exacerbated by a relatively coarse mesh in these 
examples.  Note that it is possible, if desired, to turn off the 
smoothing via View|Smoothing selection on the postprocessor main menu.

Dave.

Tsarafidy RAMINOSOA wrote:

> Hello all,
> I've attached two models of synchronous reluctance motor using 
> anti-periodic boundary conditions
> in order to reduce the area to be computed. So only a half part of the 
> motor is drawn.
> The second model is obtained from the first by rotating the rotor of 
> 180 degree angle.
> Since the rotor is symmetric, for the two models, the Bn and Ht-plot 
> along the airgap arcsegment
> in front of the rotor should be equal but with opposite sign.
> For Bn, it seems to be verified, but not for Ht.
> AJ integrals are the same for the two models but the Torques obtained 
> from weighted stress tensor are not.
> These torques was computed in taking the whole group n°2 (in the 
> models) as area of integration.
> So I'm wondering if there is any limitation in using anti-periodic 
> boundary condition or if it's in
> the torque computation that I should take some care.
>
> I got the following result :
> Model_1
> AJ integral =  1.390088e+002 Henry Amp^2 / m    
> Torque =  -3.040467e-002 N*m     /m
> Model_2         
> AJ integral =  1.389757e+002 Henry Amp^2  /m    
> Torque =  6.750423e+001 N*m  /m
>
> Could anyone explain me why Ht are not the same ?
> Any help would be appreciated.
> Best regards
>
> Tsarafidy
>  



--
David Meeker
Senior Engineer
Foster-Miller, Inc.
350 Second Avenue
Waltham, MA 02451-1196
781-684-4070
781-890-3489 (fax)
dmeeker@xxxxxxxxxxxxxxxxx