Levelset method

[Roland]

Hi,
I would like to investigate the levelset method for setting up a phase change model for crystal growth processes.
For that I saw the Github fem/tests "levelset3" and "levelset3b" examples. I could solve them with Elmersolver, but cannot see the graphical results with Paraview.
It would be nice if somebody could bring me some help for plotting the graphical results (and especially the interface moving) with Paraview.
Thanks in advance!
Roland

[kevinarden]

levelst3
Solver 6 is set to never, and is not part of the equation. So I believe it never runs, so no vtu files are created. But rather than use solver 6 I deleted it and add
Post File = case.vtu
to the simulation section
and then the vtu files are created by default.

[Roland]

Hi Kevin,
I did exactly what you told, adding "Post File = case.vtu" in the Simulation section, but when opening it with Parkview it asks with what reader it should open the case.vtu file and suggested a list of many readers. Do you know which reader could open the case.vtu file?
Thanks in advance for your help!
Roland

[kevinarden]

I did not get that question. Is the output file named case.vtu?

[Roland]

Yes, the file is case.vtu in the beam repertory.
I attach it here in zipped form.
Can you please try to open it with Paraview and keep me informed?
Thanks !
Roland

[Rich_B]

Hello Roland,

Using the levelset3 test case, and simply adding 'Post File = case.vtu' to the simulation section, leads to no output file being created. If one changes the output level from 5 to 6, this message appears in the solver log:

MAIN: -------------------------------------
AddVtuOutputSolverHack: Adding ResultOutputSolver to write VTU output in file: case
WARNING:: AddVtuOutputSolverHack: ResultOutputSolver instance with VTU format already exists, doing nothing!

As Kevin stated, delete solver 6 entirely, or at least comment out solver 6, then the case.vtu files are properly created.

Note that when Paraview asks about readers, that is simply Paraview politely saying something is wrong with the vtu file.

Rich.

[Roland]

Hello Kevin and Rich,
It is ok now, I just replaced in solver 6, "Exec Solver = never" by "Exec Solver = always", and now I get the result vtu files which are opened by Paraview and show nicely the results fror the 10 time steps. I will try to better understand now this example and see if it could be applied to my crystal growth phase change model, since I still struggle to model the triple point with the Lagrangian method for the moving phase change interface (with the Transient Phase Change Solver) and for which I wait some help from Peter.
Still many thanks for your help!
Roland

[kevinarden]

In the test cases they typically suppress the output, the purpose of the tests is to check if the installation gets the expected answer.
You should either use the
Post File = name
in the simulation section
or the output solver.
Using both can lead to confusion.

[Roland]

Hi Kevin,
Ok thank you for that. It's just an idea to investigate in the levelset method, but I am convinced(even if it does not work for the moment...) that the phase change (and especially the triple point modeling which is the crucial point...!) must be possible in Lagrangian with the use of the phase change solver. Any help would be welcome for this!
Roland

[raback]

Hi Roland,

Levelset method is best suited for smooth interfaces or at least interfaces with not to different materials. Hence in crystal growth the solid-liquid interface is easier than the liquid-gas interface. If there is a lot happening at the interface one would like to have a nice mesh that follows the interface. Hence Lagrangian methods are superior if they are applicable, Eulerian methods are nicely generic for different shapes but you need large amount of elements at the interface. A metal forming problem with a mushy interface would be rather ideal for levelset. For Cz growth I really see no benefits.

-Peter