Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Phase change of a metal with version 4.4 and version 5.0 does not match why ?

Please login with a confirmed email address before reporting spam

Hi

I am trying to make a simulation of a melting process of a square of Silicon 0.5m x 0.5m fixed temperature of 1941K to the left wall and 1687 K (melting point of silicon) to the right wall ( this with the intention to have a clear and observable liquid/solid interface). I was using the physics heat transfer in fluids (ht) and Laminar Flow (spf). I select (spf) because I wanted to see the effect of gravity on the melted material, then it suppose to show a nice melting front from left to right, but with a higher speed of melting in the upper part of my square of Silicon due to the clockwise flow of melted Silicon at the left side of the interface (analitical solution). However, no matter what it ALWAYS was showing error, error, error and error. One day I decided to disable the Laminar Flow node and finally it was working with a solution very nice and very similar to the analitical solution(upper part melting faster). After a few days the comsol software was updated to the latest version 5.0 and now it is showing a completely different results (uniform melting front).

Why they have different results ?
Why Laminar flow never reach a solution (convergence) ?

heat transfer by itself is enoguh to simulate that melting front or not ?


the last thing is....sometimes when I am running a model and it shows me error as result, I change the model for a model that I know perfectly works, but it still shows me the same error..... why ?


I appreciate any help on this topic that is very frustrating.


thanks........Michael........ file attached (properties are defined as variables in function on temperature by the equations in the analytic node)


now I am trying to attach the file, but the website is saying also error error.....

8 Replies Last Post Mar 19, 2015, 2:43 p.m. EDT

Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 13, 2015, 6:58 a.m. EDT
Hi Michael,

Your file is certainly too large
You have to make Edit -> clear all mesh + clear all results and attach your model.

Julien
Hi Michael, Your file is certainly too large You have to make Edit -> clear all mesh + clear all results and attach your model. Julien

Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 13, 2015, 3:15 p.m. EDT
You right. this is my file...cleared mesh and results.. Thanks
You right. this is my file...cleared mesh and results.. Thanks


Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 16, 2015, 3:30 a.m. EDT
Hi Michael,

I don't understand what you want to do. Do you want to see a phase change?
All temperatures defined in your file are higher or equal than the melting point of silicon (with thermal insulation boundary conditions).
If you want only see some natural convection in a silicon melt, no need to use heat transfer with phase change. Just heat transfert + spf. Have a look on Tin melting front tutorial...

Julien
Hi Michael, I don't understand what you want to do. Do you want to see a phase change? All temperatures defined in your file are higher or equal than the melting point of silicon (with thermal insulation boundary conditions). If you want only see some natural convection in a silicon melt, no need to use heat transfer with phase change. Just heat transfert + spf. Have a look on Tin melting front tutorial... Julien

Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 16, 2015, 4:02 a.m. EDT
Yes Julia, I want to see a phase change of Silicon from Solid to liquid. Initially, the entire PCM (phase change material) or Silicon in this case is solid at its melting temperature of 1687K( as you know a material at its melting point is still solid, just when the temperature is infinitesimally above of the melting point, it is liquid). I put that temperature at the right wall of the simulation in order to have a clear visualization of the Solid/liquid interface because when I put 298 (room temperature) to the right wall the Solid/liquid interface was not clear. On the other hand, I was studying the model of Tin melting front that you say. That model is very nice, but the equation that you use in the Deformed mesh node (solid /liquid interface) with the symbol T_lm (lagrange multiplier) is very nice. however, I don't understand it and there is no any explanation available. Finally, the "trick" that I am using in my model to see the solid/liquid interface without using Deformed mesh node is also used in a Comsol conference paper (Boston). Please see attached.


I would appreciate any help to this respect....I am stagnant on it.

Michael
Yes Julia, I want to see a phase change of Silicon from Solid to liquid. Initially, the entire PCM (phase change material) or Silicon in this case is solid at its melting temperature of 1687K( as you know a material at its melting point is still solid, just when the temperature is infinitesimally above of the melting point, it is liquid). I put that temperature at the right wall of the simulation in order to have a clear visualization of the Solid/liquid interface because when I put 298 (room temperature) to the right wall the Solid/liquid interface was not clear. On the other hand, I was studying the model of Tin melting front that you say. That model is very nice, but the equation that you use in the Deformed mesh node (solid /liquid interface) with the symbol T_lm (lagrange multiplier) is very nice. however, I don't understand it and there is no any explanation available. Finally, the "trick" that I am using in my model to see the solid/liquid interface without using Deformed mesh node is also used in a Comsol conference paper (Boston). Please see attached. I would appreciate any help to this respect....I am stagnant on it. Michael


Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 16, 2015, 7:50 a.m. EDT
I am Julien ;-)

ALE (Tin melting front) can't "create" new solid-liquid interface, in your case it's better to use heat transfert with phase change.
You can't see solid phase beacause you have defined a mushy zone of 3 K, at 1688 + 1.5 silicon is still solid. Your mesh is very coarse.
To start, try something more simple, for example, solid liquid interface not on an edge of the domain.
Time dependent study is easier to converge. Viscosity must be temperature dependent

Julien
I am Julien ;-) ALE (Tin melting front) can't "create" new solid-liquid interface, in your case it's better to use heat transfert with phase change. You can't see solid phase beacause you have defined a mushy zone of 3 K, at 1688 + 1.5 silicon is still solid. Your mesh is very coarse. To start, try something more simple, for example, solid liquid interface not on an edge of the domain. Time dependent study is easier to converge. Viscosity must be temperature dependent Julien

Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 16, 2015, 11:42 a.m. EDT
Sorry... Julien,


I can not understand why I can not see a solid phase due to the mushy zone (3K), below the mushy zone it is solid and above it will be liquid and I can play with the amplitud of the mushy zone for a balance between accuracy and computational capacity. Actually, my model has a extra fine mesh, but the model that I uploaded here doesn't have mesh, you told me to clear it for uploading. In addition, I am using time dependent study ( that is the only way to see the melting front on this PCM). Also Viscosity, density and heat capacity are temperature dependent variables in both phases (solid/liquid) as you can see the equations in the corresponding Analytic subnodes in the Basic subnode, in the Materials node.

Again

1. Why the results from 4.4 (that are very close to the analytical solution) are completely different for 5.0 (very different to analytical solution) ?

2. Heat transfer physics by itself can show me the circular flow of the melted phase (convection heat transfer) or the spf (laminar flow) node is absolutely necessary ?

3. Why laminar flow doesn't converge no matter what ?

Thanks in advance for any help

Michael




Sorry... Julien, I can not understand why I can not see a solid phase due to the mushy zone (3K), below the mushy zone it is solid and above it will be liquid and I can play with the amplitud of the mushy zone for a balance between accuracy and computational capacity. Actually, my model has a extra fine mesh, but the model that I uploaded here doesn't have mesh, you told me to clear it for uploading. In addition, I am using time dependent study ( that is the only way to see the melting front on this PCM). Also Viscosity, density and heat capacity are temperature dependent variables in both phases (solid/liquid) as you can see the equations in the corresponding Analytic subnodes in the Basic subnode, in the Materials node. Again 1. Why the results from 4.4 (that are very close to the analytical solution) are completely different for 5.0 (very different to analytical solution) ? 2. Heat transfer physics by itself can show me the circular flow of the melted phase (convection heat transfer) or the spf (laminar flow) node is absolutely necessary ? 3. Why laminar flow doesn't converge no matter what ? Thanks in advance for any help Michael

Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 19, 2015, 5:58 a.m. EDT
Hello Michael,

I have worked on phase change in pure silicon with different implementation of the phase change (enthalpy or ALE).
I have compared a 2D case with the open source Elmer.
Equations and results are given in attach file.

Hope it helps

Julien
Hello Michael, I have worked on phase change in pure silicon with different implementation of the phase change (enthalpy or ALE). I have compared a 2D case with the open source Elmer. Equations and results are given in attach file. Hope it helps Julien


Please login with a confirmed email address before reporting spam

Posted: 10 years ago Mar 19, 2015, 2:43 p.m. EDT
No, but thanks in anycase Julien.
No, but thanks in anycase Julien.

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.