Relative permeability tensor in WhitneyAVSolver

[szewro]

Hi Matthias!

I simplified the script simply declaring the relative permeability tensor as:

Code: Select all

Material 2 ! nonlinear material
  Name = "Iron"
  Relative permeabiliy(3,3) = real \
				50000.1 0 0 \
				0 2.1 0 \
				0 0 2.1
  Relative Permittivity = 1
End

Solver works, but unfortunately, the material still react as isotropic.

Please see the flux density B distribution in the disc:

aniso_test2.jpg (57.41 KiB) Viewed 4231 times

I attached also .sif file.

I am sure we need help of somebody who understand WhitneyAVsolver code.

Thank you for your help and interesting discussion!

With best regards,

Roman

[szewro]

Hi Matthias!

line 1880 of elmerfem/fem/src/modules/MagnetoDynamics/WhitneyAVSolver.F90 is:

Code: Select all

REAL(KIND=dp) :: Aloc(nd), JAC(nd,nd), mu, muder, B_ip(3), Babs

As I understand, mu variable is magnetic permeability. It seems to be scalar.

With best regards,

Roman

[mzenker]

Hi,

just a hint before leaving for the weekend:
Since you have decided to delve into the code, you could search for "Relative Permeabiliy" there and see how it is read by the solver. Using the information from the solver manual (chapter 18) it should be possible to find out which variable it is assigned to and if it can be an array or not.

Good luck!

Matthias

[szewro]

Hi!

I tried also Reluctivity tensor (directly, not relative)

Code: Select all

Material 2 ! nonlinear material
  Name = "Iron"
  Reluctivity (3,3) = Real \ 
			1.5915e+01   7.9577e+14   7.9577e+14 \
			7.9577e+14   3.9789e+04   7.9577e+14 \
			7.9577e+14   7.9577e+14   3.9789e+04  

!Relative permeabiliy(3,3) = real \
!				50000.1 0 0 \
!				0 2.1 0 \
!				0 0 2.1
  Relative Permittivity = 1
End

It solves, but it also gives no anisotropy.

There is function GetReluctivityTensor in Utils.F90, however, it seems it is not working.

It is to sophisticated to me to verify all chain of Reluctivity/Permeability scalar/tensor.

With best regards,

Roman

[mzenker]

Hi,

we didn't notice a typo in your sif:

Code: Select all

  Relative permeabiliy(3,3) = real \

There is a missing "t" in "relative permeability".

I have seen in the code that the relative permeability should be read as array.

HTH,
Matthias

[mika]

Hi,

For this solver the magnetic permeability cannot be defined as a tensor-valued property, while it should be possible to have the magnetic reluctivity as a tensor-valued parameter. What might be misleading is that the magnetic reluctivity has to be defined with a different keyword depending on whether it is scalar or tensor-valued parameter. The keyword "Reluctivity" works for the scalar case but the tensor-valued property must be specified by using the keyword "Relative Reluctivity". I think both the keywords nevertheless define the magnetic reluctivity similarly in absolute terms so it would be better to change the code such that the keyword "Relative Reluctivity" would be replaced by "Reluctivity".

Until the code has been revised my suggestion is to specify "Relative Reluctivity" as an array-valued property in order to obtain an anisotropic material (and remember that there is nothing relative behind the definition of this parameter).

Best regards,
Mika

[szewro]

With very kind support of Mika, Juhani and Matthias I finally got anisotropic results.

First important note: "Relative Reluctivity" should be considered as "Reluctivity".

The results are: left side: calculations in Octave 4.2.2, right side: results.

results.jpg (103.26 KiB) Viewed 4209 times

Material definition is:

Code: Select all

Material 2 
  Name = "Iron"
  Relative Reluctivity (3,3) = Real \ 
			15.9 0 0 \
			0 39788.9 0 \
			0 0 39788.9
  Relative Permittivity = 1
End

In the attachment please find the working example .sif and .geo file.

Please note, that in the future the meaning of "Relative Reluctivity" and "Reluctivity" may be the subject of changes.

I would like to thank again to Mika, Juhani and Matthias for your help!

With best regards,

Roman Szewczyk

[szewro]

In Elmer github repository, in commit 4334 dated Aug 2, 2019 Mika replaced 'Relative Reluctivity' by 'Reluctivity', to avoid misinterpretation.

Thank you Mika for your efforts!

[mzenker]

Hi,

for the Windoze users among us, it would be great to have recent nightly builds...

Thanks,
Matthias