Zirka hysteresis model

[fjimenez]

Hi all,

I was wondering if anyone with some experience with Zirka materials can let me know if this is a bug or I am missing something. This is a simple case of a coil and a piece of metal (Low frequency, no eddy currents, metal conductivity =0). Case with single BH curve works as expected, but when I switch to a Zirka material things get weird. Other than the overall BH curve not following the input curve, the evolution of the BH curve also seems weird. I used values from the elmer tests. I could trace some of the input parameters to the Zirka paper but others were not that easy to find. For instance, I am not sure why we need to specify a relative permeability (it crashes if it is not included)

anim1.gif (703.31 KiB) Viewed 461 times

anim.gif (471.38 KiB) Viewed 461 times

case_zirka.tgz

(91.11 KiB) Downloaded 31 times

[kevinarden]

The answer changes with
element="p:2"
which means mesh quality, density, and order matters.

[fjimenez]

Hi Kevin,

Thanks for the suggestion. I added the line to the sif file and results did change, but just a little bit. The outstanding issues remain. In my opinion, there are two:
1. The evolution of the BH curve is not correct

hysteresis_1.gif (77.84 KiB) Viewed 424 times

2. Shape is not close to input.

I am not sure why 1 is happening. For 2, the problem seems to come from CalcFields (if relative permeability is not defined in the material, as it is derived from the BH curve). MagnetDynamics2D runs well without a relative permeability input.

Code: Select all

 MagnetoDynamicsCalcFields: ------------------------------
MagnetoDynamicsCalcFields: Computing postprocessed fields
MagnetoDynamicsCalcFields: Using potential variable: az
MagnetoDynamicsCalcFields: Number of components to compute: 13
InitializeToZero: Initializing the linear system to zero
ERROR:: GetReluctivityC: Give > Relative Permeability < or > Reluctivity <  for material!
STOP 1 

I wonder if this is the reason (CalcField.F90, line 1790):

Code: Select all

IF ( (ASSOCIATED(MFS) .OR. ASSOCIATED(EL_MFS)) ) THEN
   ! Don't really know how to compute MFS with Zirka
   ! Skipping it will certainly cause errors since k becomes invalid
   !IF(.NOT. HasZirka) then

Cheers,

[kevinarden]

Just adding element="p:2" doesn't mean the mesh is converged. Depending on how you created the sif means that elmer could output an solution that is not a converged solution. Check especially if the non-linear iterations are actually converging.

[fjimenez]

Hi Kevin,

Thanks again for helping. Nonlinear iterations are converging, even without the relative permeability input as expected
if the BH curve is provided.

Cheers,