Re: [femm] SRM Modeling

[David Meeker <dmeeker@xxxxxxxx>]

Pradeepkumar Ashok wrote:

> Hi all,
>
> Has anyone used FEMM to model and evaluate the torque capacity of a
> switched reluctance motor? I need help
>
> My FEM model only gives me half the torque that my analytical
> calculations predict. For those interested, the plots and the machine
> dimensions are in the file called comparisons.doc . The analytical
> method I used was based on a paper by Radun Analytically Computing the
> Flux linked by a Switched reluctance motor phase when the stator and
> rotor pole overlap. This paper uses the magnetic co-energy method to
> calculate torque, while FEMM uses the stress tensor method. Could that
> be a reason for the difference? I dont think so&but I am not sure&
>
> In the models, I used triangle to mesh the air gap between the stator
> and the rotor poles (remodel.fem) . I also used a slightly different
> model with a finer mesh (.0005) between the stator and rotor poles
> (remodel2.fem) to see if this problem was due to my mesh being coarse.
> The airgap is .0125 inch and so I felt that .0005 was more than
> sufficient. I found the torque output from these two cases to be not too
> different. So I went ahead a created a Lua script to plot the static
> torque Vs Rotor position Vs Current. (Files inside the folder
> torque_sim),and that is how I got the plot shown in comparisons.doc
>
> Simply put my question is Have I modeled the machine correctly& If so,
> I can safely attribute this difference to errors in the analytical
> model. Can somebody help ?

I didn't run the script all the way through, but there were a few of 
things that I noticed:

-- Your script only changes the current density in the Imax material. I 
think that you also need to change the Imin material in an analogous way 
to get correct results.

-- It's best to draw the contour along which the stress tensor is taken 
through the center of the air gap, rather than to one side.

-- Make the arcs that compose the pole faces and integration contour 
have a finer discretization (like 1 degree or less). This will get rid 
of the cogging in your graphs of torque versus rotational position.

Dave.

--
David Meeker
dmeeker@xxxxxxxx
http://femm.berlios.de/dmeeker