Axis Symmetric post processing

[thalesmaoa]

Hello, I'm trying to produce some tutorials about ELmer for my students. I'm facing some problems regarding post processing calculations.

I've replicated the first tutorial of FEMM (https://www.femm.info/wiki/MagneticsTutorial) using Elmer.

The worflow is as follows:

FreeCAD Design and Mesh -> Elmer definitions -> Paraview post processing

At this moment, I can integrate variables using Paraview, but I can't get the same result as FEMM when I use Axis Symmetric. For planar (cartesian) the result matches.

To compare the results, I'm integrating over surface the magnetic potential.

I'm not sure if Paraview understand that the result is (r, rho,z) or if I must pre-treat it somehow.

I've tried

Code: Select all

Coordinate Transformation = String "cartesian to cylindrical"

but I think this is not what I should do.

FEMM result with A integral.


Elmer definitions (sif attached)


The result using paraview



For a general information: A = Potential

FEMM = 1.28184e-11
Elmer -> Paraview = 9.99684e-11

Any suggestions?
PS: I can't attach my mesh since it is bigger than 1Mb.

[kevinarden]

I worked through these before

https://github.com/mrkearden/mag2D3D

[raback]

Hi,

Your sif looks fine to me.

(No need to do the coordinate transformation. It is intended for very rare cases when you switch the system on-the-fly back and forth.)

What if you plot the vector potential from (0,0) to (Rmax,0) and compare them?

-Peter

[thalesmaoa]

Hi Kevin, appreciate your file. Unfortunately, your problem does not consider axis symmetry. When I use the planar approach, the results are the same. Thx

Hi Raback, I can indeed plot a line. However, from the pictures that I've posted, you can notice that the contour lines are already different. I mean, I'm not sure if Paraview understands that this is an axis-symmetric solution. For planar, they are alike.

[raback]

Hi

Yes, it seems that the vector potential with the coil in FEMM is constant?

I guess a one-dof model would be needed for Elmer to replicate that. I guess FEMM does it with a circuit? I guess we could add a similar circuit or a simple constraint. I might have a similar test case.

Looking at the equation you could approach constant vector potential at the coil using very small relative permeability, say 1.0e-3. Not a pretty solution though...

Btw, you maybe could get the mesh directly from FEMM? Did you try that? It would make the comparision smoother.

-Peter

[kevinarden]

Yes, sorry, I thought this was the Femm tutorial but it was another persons attempt.

[gforti]

Hello Peter,

I stumbled in this discussion and got a little confused with your explanation.

First: "Yes, it seems that the vector potential with the coil in FEMM is constant?"
I think you got mistakenly mislead by the green color. The green color appears when the region is selected. If you observe the contours, which are the "constant potential" curves, you can see that the potential varies within the Coil.

Thanks again for your explanations.

[gforti]

Also,

Is there any special attention when defining a circuit for this axis-symmetric simulation?
Please find my simulation files here.

[raback]

Hello Peter,

I stumbled in this discussion and got a little confused with your explanation.

First: "Yes, it seems that the vector potential with the coil in FEMM is constant?"
I think you got mistakenly mislead by the green color. The green color appears when the region is selected. If you observe the contours, which are the "constant potential" curves, you can see that the potential varies within the Coil.

Thanks again for your explanations.

Yes, I think I thought the green color indicated constant vector potential. I got very confused then by the results. My bad. Forget all I wrote

-Peter

[thalesmaoa]

I still believe that there is a problem with coordinates.
If I simulate with planar I get the same exactly result.

The way it is now is different.
Considering a axis Symmetric simulation, Z-axis should be 2*pi*r. Not sure if paraview understand as this. Any hints?