## Triple Point Modeling for Phase Change _ follow up

Numerical methods and mathematical models of Elmer
Roland
Posts: 220
Joined: 12 Apr 2018, 11:29
Antispam: Yes

### Triple Point Modeling for Phase Change _ follow up

Hi,
After the difficulties for modeling the triple point moving in the former Solid_Liquid phase change models (already presented and discussed in former posts) I made a very simple Solid_Solid phase change transient model which I would like to discuss here and get some help.
The model geometry (here attached) shows the lower Body 1 and the upper Body 2 which are both solid and separated by the Phase Change interface Boundary 7 , which ends on its right by the triple point (Node 7). As all the bodies are solid, their boundaries (especially the outer boundaries 6, 8 and 9) are fixed and must not deform. Only the phase change interface (boundary 7) can deform and move (up or down) with the phase change Stefan velocity. The coupled physics are Heat_Equation (HT), Mesh_Update (MU) for Bodies 1 and 2 and Transient_Phase_Change for the phase change interface (Body 3). For the Phase Change Solver I used the "Result Output Solver" in which I replaced the Procedure "ResultOutputSolve" "ResultOutputSolver" by "TransientPhaseChange" "TransientPhaseChange".
Unfortunately the results (see the here attached t=0s, t=5s and t=10s plots) show that the triple point moves only vertically down and this implies a deformation of the boundaries 8 and 9 which should normally not deform (as they are solid!), meaning that the triple point should nicely move downward by following the curved (circular) boundary 9 and then move horizontally to the right on boundary 8.
Here attached is also the zipped Elmer GUI project.
It would be very nice and wellcome if somebody could give some help for getting the correct move of the triple point, meaning an up or down move by following the outer boundaries 6, 8 and 9 without deformation of these boundaries which must remain fixed and undeformed.
Thanks in advance for any help!
Roland
Nota: the plot results t=5s and t=10s are attached in the next post
Attachments
t=0s.JPG
Phase_Change_Solid_Solid.zip
geometry.JPG
Roland
Posts: 220
Joined: 12 Apr 2018, 11:29
Antispam: Yes

### Re: Triple Point Modeling for Phase Change _ follow up

Plots t=5s and t=10s
Roland
Attachments
t=10s.JPG
t=5s.JPG
raback
Posts: 4707
Joined: 22 Aug 2009, 11:57
Antispam: Yes
Location: Espoo, Finland
Contact:

### Re: Triple Point Modeling for Phase Change _ follow up

Hi Roland

The key problem with the TransientPhaseChange solver is that there it is assumed that the movement is only in vertical direction. See eq (35.3) in:
https://www.nic.funet.fi/pub/sci/physic ... Manual.pdf

When this solver was coined it was assumed that a feedback mechanism is used to ensure constant radius of the drawn crystal.

I could imagine formulation for general free surface (operating in normal-tangential coordinates). Then the issue becomes also how to have the phase interface and free surface compatible. Ideally there would be just one solver for both so that continuity of solution at the triple point could be ensured.

-Peter

PS. There is also the entalhphy method can be found in PhaseChangeB. It is basically a levelset type of approach for phase change problems. You don't really need a separate levelset function as the temperature acts as a natural indicator where the field is.
Roland
Posts: 220
Joined: 12 Apr 2018, 11:29
Antispam: Yes

### Re: Triple Point Modeling for Phase Change _ follow up

Hi Peter,
Thank you very much for your answers on both level set topic and this one.
According to your suggestion could it be possible to imagine such a formulation for general free surface?
Concerning the other code you mention in the other "levelset" topic, and as I have already presented it, the triple point managing works nicely in my Comsol models, but I wonder how it is really managed and especially, as you say it, how this crucial problem of how the triple point manages the 3 BC's (1/ the crystal external surface BC, 2/ the liquid external surface BC and 3/ the phase change interface BC) to which it is submitted. Tell me how this could be invastigated in the Comsol model and I can take a look.
We keep us informed.
Best regards
Roland