WhitneyAVHarmonicSolver and Electric Current Density

[vencels]

Hello,



The current density can be set either as

Electric Current Density = Real 100.0

for current density distribution on the terminal or

Electric Current Density = -distribute 100.0

as the total current through the face.

At low frequencies, the solver converges well.
WhitneyAVHarmonicSolver: Solving harmonic AV equations with edge elements
50 0.1359E+00
100 0.3095E+00
150 0.1037E+01
200 0.2617E+00
250 0.7828E-03
300 0.4052E-05
350 0.3422E-08

At high frequencies the solver stuck at the residual about 0.15E-06
WhitneyAVHarmonicSolver: Solving harmonic AV equations with edge elements
50 0.1633E+03
100 0.1926E+02
150 0.7856E+01
200 0.3782E+00
250 0.1516E+00
300 0.9654E+01
350 0.3289E+00
400 0.4955E-01
450 0.1908E-02
500 0.5251E-04
550 0.5538E-06
600 0.1520E-06
..... continues until max iterations are reached.



The result seems to be converged, but the residual is stuck at a fixed value. I suspect that the current distribution does not fit the numerical solution. Ideas on how to treat this case?

[Rich_B]

Hello,

Loading your mesh in ElmerGUI immediately crashes ElmerGUI. From the log, the geometry started to load, with surfaces and volumes, then crashed.

Running this:

ElmerGrid 2 2 mesh -out mesh2 -autoclean

allowed ElmerGUI to load and display mesh2. The number of boundaries went from 55 in 'mesh' to 35 in 'mesh2'. Running both high and low sifs with mesh2 gave trivially zero solutions in both cases.

This may or may not be part of the issue.

Rich.

[fjimenez]

Hi,

As a follow-up to to Rich_B's comment, the zero solution is because you also need to modify the first line of the sif file, otherwise boundaries are not correctly assigned. I think, and I might be mistaken, that you cannot force the current density in this geometry to be constant due to skin effects. If you set the potential at the boundaries though, both models converge at the specified level.

Cheers,

[vencels]

Hello Rich,

The case runs without issues from a terminal. Elmer version - Rev: Release, Compiled: 2023-06-21

ElmerSolver highFreq.sif
ElmerSolver lowFreq.sif

Hello fjimenez,

I know about setting potential. Also, fixed current density can be set through circuits.
The current task is to find a solution for setting fixed currents without circuits. THat might require modifying either current distribution on the terminal or residual.

Thanks,
Juris

[fjimenez]

Hi Juris,

Sorry, I guess my comment sounded like I was suggesting something new or insightful. I assumed that you of course knew about it, and that you had a good reason to do it this way. The point I was trying to make was that the case converged when you don't use the current density as BC with the same geometry. I see now that you are following a similar strategy to the one used here. Also in this file, it is stated that copper's conductivity needs to be significantly reduced otherwise a very fine mesh is required. An interesting option, other than putting a profile as you mentioned, would be to set the total current as a boundary condition (don't know if it is possible).

Cheers,

[vencels]

Hi,

Thank you for the suggestion. I think the solution is either numerical (modifying residual), changing material properties close to terminals, or figuring out how to set up the current distribution.