[steventay]

Hello All,

I am working on a model that coil a tube in a tank. The tank is filled with water and another fluid will pass thru the tube. The temp of the fluid passing thru the tube is very low (-40 degrees) thus it will slowly cools and freeze the water in the tank. I managed to get all other parameters converge except for the RMS P-Mass of the water... It will converge for a while then begin to diverge out of the convergence criteria. Do anyone know what I can do to improve the convergence of the RMS P-Mass of the water? Or is there any setting that I have missed out? Thanks and looking forward to some advises!

Steven

[ghorrocks]

What are you modelling? Just the coolant flow in the tube? Or the freezing process in the water? Or both?

[steventay]

Hi,

I am using the coolant in the tube to freeze the water in the tank. Thanks.

[ghorrocks]

How are you modelling the freezing process?

[steventay]

I created two materials, one is liquid(water) while the other is solid(ice) then join them together thru another homogeneous binary mixture.

[steventay]

Is it necessary to set expert parameters for this situation? The funny thing is that if I set the tank fluid as only water (liquid), it will work well, but when I set it as the mixture of water and ice, it will diverge after some time.

[ghorrocks]

Nice try.... But not even close. Modelling the freezing process takes a bit more than just setting a HBM. CFX does not have a freezing model built in (to my knowledge) so you are not going to be able to do this model including the freezing process.

Do you really need to model the freezing process? What are you expecting to understand from the analysis?

[steventay]

I need to do this freezing process to validate my experiemental results. Is there any resources that I can learn more about analysing a frezzing process? Do you have any advise on how I should move on from here? Thanks for your help and advise.

[ghorrocks]

You are proposing doing a very complex multiphase analysis. I hope you have plenty of time to do the necessary research (I am talking months if you experienced, years if you are not). To model the freezing process accurately is not trivial - depending on what regime you are in you need to account for subcooling, and you have to have a way of stopping the velocity in a cell (ie freezing it) which is compatible with the Navier Stokes continuity equation. You have already come across this last point and it is not easy to do properly.

Have a look in the open literature, there will be millions of reports of simulations of the freezing process. You may also have more luck with other CFD codes. Another code may have a freezing model built in. I don't know of any off hand.

[steventay]

Thanks for your advice. I will search for literature on the freezing process. Is melting process as complicated as freezing?

[ghorrocks]

Generally, yes. It all depends on how much of the physics you need to include.

[steventay]

Understood. Thanks for your time and advice.

[ckleanth]

just thinking outloud, if you know the rate of change of the phase from liquid to solid i.e. a Temp VS mass fraction function (or tabulated data) you can define an algebraic relation for the mass fraction in your multiphase setup hence express the phase change purely by cel.

[ghorrocks]

Yes, approaches like that can work. I was assuming he wanted to include the build up of ice on the tubes. If you don't need to model the movement of the ice front it is much easier. If you want to move the ice front then it is far harder. Depends what Steven wants to get out of the analysis.

[steventay]

Hello,

Thanks for the suggestions. Unfortunately, as what Glen says, I am looking for the build up of ice around the tube.

As I have explained earlier, I am using HBM method to create the PCM however, the freezing or melting isn't working correctly due to the movement of the cell. I have thinking whether I could put in an expression on the velocity of u, v & w of PCM instead of stating zero. This expression is to ensure that the velocity of PCM remains close to zero all the time. However, I am not sure how to do it. Anyone have any ideas?

I am also looking at another method stated by CFX help. That is using thermal phase change model method. By adding two material with different phase into a domain, it will recognise as a phase change model. I am still trying this method and yet to get it work well yet. Anyone tried this method before?

Many thanks for all the help and advises.

[ghorrocks]

There are a number of ways to model freezing. You can do a moving mesh approach with two domains and the mesh follows the interface, you can put source terms on stuff you deem to have frozen to stop it or you can increase the viscosity of the ice to a huge number to lock it in place. All of these approaches have disadvantages and advantages, you will have to read the literature to find the one most suitable for you - and you will probably find lots more methods too.

[steventay]

Hello, Thanks. I managed to reduced the velocity of the PCM drastically by putting a very large resistance via a sub-domain. I can see that the u, v and w of the PCM almost approaches zero. However, the P_mass of the PCM is still relatively high but it is much better than previous analysis. May I know what is P mass? What will affect the P mass? Thanks!

[ghorrocks]

What is PCM?

P_mass is the solution residual of the continuity equation.

[steventay]

PCM = Phase Change Material (water to ice and ice to water)

Thanks.