Discussion Forum

Hello All,

I am a new user of COMSOL and dont know so much about it !! I would like define outwatd unit normal (N) and curvature (K) during the Phase field/level set formulation in phase change issue like : Melting / solidification.
I have tried by writing the usual math like, n=grad(phi)/abs(grad(phi)) and K=div(n) After the execution of COMSOL, it says that the expressions for outwatd unit normal (N) and curvature (K) are not correct and consequently, calculation diverges. If anybody knows it, please let me know.

With best regads

AH

Hi

do not forget that if phi is a dependent variable, you have the gradient defined as (phix,phiy,phiz) with the suffix x and the second derivative with a suffix xx,yy,zz,xy ....

see the doc

On the boundaries you have the normals already defined, but that does not fully match the level set method

--
Good luck
Ivar

Dear Mr. Yu,

Thanks for your reply.

>> Maybe we can discuss each other if you don't mind.

No problem. Lets discuss together either here or privately. It will be helpful for both of us !!

Actually, I am not doing purely melting and solidification process. My target is includes multiphase ( like : solid> liquid> gas) and melting is one of the intermediate stage. So, Phase field and level set approach, both can be used (literature says !!!) . However, melting is the dominating stage in my calculation and then vaporization ......

>>If you use the phase field or level set module, the unit normal and curvature are both calculated automatically by comsol.

Yes, you are right.


Since my target is not purely melting or solidification, I have to modify some source terms as

reported by previous researchers ..

Thanks once again for your reply.

With best regards

AH

Hi,
Are you simulating Melting/ Solidifcation using comsol? I am doing it now. Maybe we can discuss each other if you don't mind.
If you use the phase field or level set module, the unit normal and curvature are both calculated automatically by comsol I think, see the enclosed Users' Guide about the level set method.
By the way, which method do you use, phase field or level set ? I think the phase field module in comsol is for the simulation of multiphase flow, but not suitable for melting or solidification, because the governing equations are different, for the phase field in comsol, the governing equation is Cahn-Hilliard equation, while for the simulation of melting or solidification we need the Allen-Cahn equation. What's your opinion?

Dear Mr. Yu,

Thanks for your reply.

>> Maybe we can discuss each other if you don't mind.

No problem. Lets discuss together either here or privately. It will be helpful for both of us !!

Actually, I am not doing purely melting and solidification process. My target is includes multiphase ( like : solid> liquid> gas) and melting is one of the intermediate stage. So, Phase field and level set approach, both can be used (literature says !!!) . However, melting is the dominating stage in my calculation and then vaporization ......

>>If you use the phase field or level set module, the unit normal and curvature are both calculated automatically by comsol.

Yes, you are right.


Since my target is not purely melting or solidification, I have to modify some source terms as

reported by previous researchers ..

Thanks once again for your reply.

With best regards

AH

Hi,
Are you simulating Melting/ Solidifcation using comsol? I am doing it now. Maybe we can discuss each other if you don't mind.
If you use the phase field or level set module, the unit normal and curvature are both calculated automatically by comsol I think, see the enclosed Users' Guide about the level set method.
By the way, which method do you use, phase field or level set ? I think the phase field module in comsol is for the simulation of multiphase flow, but not suitable for melting or solidification, because the governing equations are different, for the phase field in comsol, the governing equation is Cahn-Hilliard equation, while for the simulation of melting or solidification we need the Allen-Cahn equation. What's your opinion?

Hello All,

I am a new user of COMSOL and dont know so much about it !! I would like define outwatd unit normal (N) and curvature (K) during the Phase field/level set formulation in phase change issue like : Melting / solidification.
I have tried by writing the usual math like, n=grad(phi)/abs(grad(phi)) and K=div(n) After the execution of COMSOL, it says that the expressions for outwatd unit normal (N) and curvature (K) are not correct and consequently, calculation diverges. If anybody knows it, please let me know.

With best regads

AH

Hello, Akter Hossain
May I ask you a question? when simulating melting or solidification, you need to calculate the velocity of the moving interface by the Stefan condition, i.e. rho*L*V=qs-ql, where rho is the density, L is the latent heat, V is the velocity, and qs-ql is the heat flux difference across the moving interface. How did you do about the calculation of "qs-ql" in comsol? I know the operators "down(), up()" can make it, but I'm not sure about the usage of them. I think two variables should be defined to represent qs and ql, for example, qs = up (ht. tfluxx), ql = down (ht. tfluxx) (considering 1D, in version 4, ht. tfluxx means total x component heat flux). But if just define these two variable like these, do comsol know that up() or down() is effective to which boundary?

Hello, Akter Hossain
May I ask you a question? when simulating melting or solidification, you need to calculate the velocity of the moving interface by the Stefan condition, i.e. rho*L*V=qs-ql, where rho is the density, L is the latent heat, V is the velocity, and qs-ql is the heat flux difference across the moving interface. How did you do about the calculation of "qs-ql" in comsol? I know the operators "down(), up()" can make it, but I'm not sure about the usage of them. I think two variables should be defined to represent qs and ql, for example, qs = up (ht. tfluxx), ql = down (ht. tfluxx) (considering 1D, in version 4, ht. tfluxx means total x component heat flux). But if just define these two variable like these, do comsol know that up() or down() is effective to which boundary?

Hello Xiaoyi Yu,

Your are right that Stefan condition, i.e. rho*L*V=qs-ql should be solved for melting/solidification !!! But I found that instead of solving Stefan condition for phase change, some authors introduces a discontinuity in the Specific heat (Cp) of the Phase change material (PCM) and this discontinuity is implemented by a smoothed Heaviside function with a continuous second order derivative (flc2hs >> ver 3.5a ) in COMSOL. and the melting of PCM is observed including mushy zone. It seems to me that it is very simple to implement. But authors did not say anything why Stefan condition is replaced by Cp modification. !! Actually, I dont know about `` up() or down()`` operators and their effectiveness to calculate the fluxes. I am sorry in this regards and moreover, still I am using Ver 3.5a & I dont know about ver. 4.0 & above ... Anyway, I think that the modification of Cp is very simple way to simulate the melting or solidification. If you wish then you can try this approach.

Good luck !!

AH