2D induction cold crucible with impedance boundary conditions

[Roland]

Hi Peter,
Could you just tell me how I can test our updated MagnetoDynamics2D file?
Must I just reinstall the latest Elmer version?
Thanks in advance for your help
Roland

[raback]

Hi Roland,

Yes, a version that has been compiled after yesterday. Then add these values at some BC:

Code: Select all

Layer Electric Conductivity = Real 
Layer Relative Permeability = Real 

You should see the effect in the solution of A_phi with and without them. The Joule heating for the boundary is not yet included.

-Peter

[Roland]

Hi Peter,
Ok I will install this new updated version within this evening and keep you informed. Very exciting and again many thanks for all your efforts!
After that there will be another thing to add for forcing the looping of the surface current density Js_phi on the cold crucible surface. This is naturally done in 3d (since the cold crucible has slits...) but not in 2d. So, in Comsol, we do that by adding an integral constraint which forces the integral of Js_phi to be zero all over the crucible section. This is done by adding an ODE with an additional fictive variable which is a phi component voltage "V" which acts a bit like a Lagrange Multiplier for forcing the integral(Js_phi) to be equal to zero. Do you think that this could be added? As soon as I have checked the IBC with the latest installed version, we can discuss that, if you are ok to spend a little more time for this.
Menu thanks and we keep in touch.
Roland

[Roland]

Hi Peter,
As discussed I just tried the cold crucible model with the IBC relations after having installed the new Elmer version.
Unfortunately it does not work for the moment and the solve gives a "SIGSEGV" error.
I sent you here attached the zipped GUI project. The model is very simple: the one turn inductor surrounding the rectangular cold crucible.
It would be very nice if you could take a look and see what is wrong.
Many thanks in advance!
Roland

[raback]

Hi Roland,

Really nice test case!

The error came in the postprocessing part. It cannot yet be considered. I added skipping of the not implemented part. There seems to be some effect particularly on "Potential Im". Maybe you can compare to reference.

-Peter

[Roland]

Hello Peter,
Yes ok I will check this asap and keep you informed.
Thanks!
Roland

[Roland]

Hi Peter,
As discussed I made 2 cases and send you here attached the corresponding zipped projects :
- One without IBC with a 1 mm thick cold crucible (CC_2D_No_IBC)
- One with IBC (CC_2D_IBC)
Here attached are also the results showing the "r*A_re" contours which are the magnetic flux density lines (say B_lines) in 2D_axi.
- Without IBC the result is correct and shows the B_lines which flow nicely between the inductor and the cold crucible and are tangent to the cold crucible surface.
-With IBC the results are quite different and not correct since, as I told you this morning, the B_lines point toward the cold crucible and are perpendicular to it (this jpg result is sent in the next post)
I let you take a look and see where the problem comes from. Don't hesitate to ask me further questions.
Many thanks in advance and we keep us informed.
Roland

[Roland]

Result for CC_2D_IBC (here attached)
Roland

[raback]

Hi,

There was just a wrong sign. Now things look good. Attached is comparison between the two models.

CalcFields also works. Ask for "Calculate Nodal Heating = True" when using IBC. That can be used as "Temperature Load = Nodal Joule Heating" when solving for temperature.

This was done in cartesian system. The cylindrically symmetric case might work out of the box but I won't guarantee...

-Peter

[Roland]

Hi Peter,
Whaou! This sounds very nice and realistic!
I will test this in cylindrical configuration after installing the latest updated release and keep you informed (probably within tomorrow).
And after that, as I told you in a former post, we will have to add an integral constraint for forcing the whole surface current density in the cold crucible cross section to be equal to zero (by adding a phi component electric voltage which is like a Lagrange multiplier and which is also an additionnal variable), which will force the surface current density on the outer surface to loop on the inner surface (which is not yet the case for the moment) This is just a trick in 2D for simulating the slits (which exist in 3D but not in 2D!) of the real sectorized cold crucible. And this makes that there is a non zero magnetic field inside the crucible thanks to which it is possible to heat a charge set inside it...
Thanks again for your huge efforts and we keep us informed for this next step. Very exciting and nice perspectives in view!!
Roland