Periodic BC for 3D stator

[sbrialmont]

Hello,

I'm trying to model a stator winding in 3D in magnetostatic.
I'm using the static current solver to first obtain the source current in the winding and then, obtain the magnetic vector potential and flux density.
I want to use antiperiodic boundary conditions using Mortar projectors as recommended in many similar cases.

Globally, it seems to work but I'm a bit concerned regarding the boundary conditions. Indeed, the flux density exhibits a quite strange behaviour on the periodic surfaces as can be seen in the attached picture.
I just wanted to check if the antiperiodic BC are properly written in my sif file. Maybe the problem comes from somewhere else, I think the mesh is far from being optimal.

Here is the link to the mesh and sif files : https://we.tl/t-j4BV529eCQ

Thanks very much in advance,

Sebastien Brialmont

[raback]

Hi

Quickly looking it is suspicous that you have set "use piola Transform = true" in CalcFields. This is something for the primary solver and CalcFields automatically follows the element types as needed. Don't know if it has any effect.

Also you have a current density that might not be fully divergence free. That's why you find keywords like "fix input current density" in many sif files. Maybe it is launched automatically. If the current density has even small divergence then you see that often in poor convergence of the magnetostatics problem.

-Peter

[sbrialmont]

Hi,

Thanks for the prompt answer !

Indeed, I removed the 'Piola transform' line, I didn't affect the solution anyway.

It looks like a bad convergence indeed as the linear solver goes until the max number of iterations is reached so the convergence criterion is never met...

Moreover, the results look quite dependent on the max number of linear iterations. However, the flux density seems to behave badly on the periodic surfaces only as it is shown in the picture in which the current is applied in the central conductor. That is why I was suspecting a bad definition of the boundary conditions.

I tried to add the following line: "Fix Input Current Density = True " in both the MagDyn and the Static Current Conduction solvers but it doesn't change anything.

Is there something I missed to ensure that the current density is divergence free ?

Thanks a lot,

Sebastien

[raback]

Hi Sebastian,

The transfer has expired. Maybe you could upload the files again.

-Peter