[JuPa]

Hi,

I want to know how accurately CFX can model condensation. At the moment I am trying to look for papers where the authors have used the phase change model to try and characterise condensation.

I want to do a lot of research first, and read up about condensation in CFX before I fully commit to condensation simulations.

Does anyone have any experience of condensation in CFX?

The sort of thing I am after is relatively simple: condensing dry saturated steam on a flat plate. I'd like to know if CFX can deal with the effects of condensate on the surface, and how it will effect condensation rates. Empirical data exists for the condensation of dry saturated steam on a vertical flat plate. It was also the starting point for Nusselt's condensation theory so it seems like a good starting point to model it using CFX.

Ideally what I want out of the simulation is to determine the condensing heat transfer coefficient either:
- Directly by reading heat transfer coefficient values in CFD Post
- Indirectly by reading a heat flux and wall temperature [Q = u (Tsteam - Twall) then u = Q/(Tsteam - Twall)]

Something tells me condensation is not going to be easy to model, and will involve unrealistic length scales and time steps.

Any comments or suggestions would be helpful.

Thank you

[ghorrocks]

The sort of thing I am after is relatively simple

Nothing is simple when it comes to phase change.

Well done for having a good look at this before committing to the simulations. If you understand what you are doing and the limitations of it before you start you have a much better chance of successfully completing your task.

What level of resolution of the condensation process are you looking for? One possible approach would be to use the wall film model to model the liquid on the surface, and a second approach would be to directly model the wall film with a free surface model. These are totally different models, and the first would be much easier to implement than the second I suspect. But it depends what you are trying to get out of the model for which is more appropriate.

[JuPa]

Hi ghorrocks, thanks for your reply.

Originally Posted by ghorrocks

Nothing is simple when it comes to phase change.

I know the theory pretty well!

Originally Posted by ghorrocks

What level of resolution of the condensation process are you looking for?

I'm not sure what you mean by this. What I want to know is if I have a flat vertical plate, and if I then condense dry saturated steam will CFX also model the condensate layer flowing down the vertical plate due to gravity drainage? Also, will CFX take the subcooling of the condensate film into consideration? Because as the condensate layer becomes thicker, the vertical flat plate becomes more and more flooded, leading to lower condensation heat transfer coefficients.

Hence what I want out of the simulations are:
1. How does the condensate behave. Does it behave as it should, based on theory and experimental data? If so then great!
2. According to CFX/CFD Post what is the condensing heat transfer coefficient, and how does this compare to theory and experimental data.

Originally Posted by ghorrocks

One possible approach would be to use the wall film model to model the liquid on the surface, and a second approach would be to directly model the wall film with a free surface model. These are totally different models, and the first would be much easier to implement than the second I suspect. But it depends what you are trying to get out of the model for which is more appropriate.

Based on the above points 1 and 2, what would you suggest? I obviously need to read about this a lot before I even think about switching the condensation model on. If I find any useful papers I'll update this thread.

[ghorrocks]

At the smallest level you have nulceation sites and sub-cooled vapour. At some point these nucleation sites trigger condensation and that changes the condensation of the remaining vapour as the nucleation is very different with liquid present. Then the liquid builds up and might fill cavities or eventually cover the surface - all these things will change the condensation behaviour. This constantly changing nucleation behaviour, combined with specific heat effects and the temperature variations everywhere make this a very challenging model to perform.

This is condensation at a very elementary level and few people need to model it at that level of detail. So it is a matter of what simplifications you are going to make. How macro scale do you want to go.

If you want to see if you can model the heat transfer from first principles (which is what you suggest in your point 1) then you need to model the condensate as a liquid and a free surface model. This will be an extremely challenging model, and even more challenging to get accurate. Also do not think that this approach is without empicism - you will have to tune many parameters to get this to work, the parameters are just lower level than other approaches.

The wall film model approach would be much easier, but introduces empirical correlations which you will have to tune to match experimental behaviour. But it does not sound like this is what you want to do.

[JuPa]

I'm assuming the free surfacel is a similar but reverse process of the RPI boiling model. If so one of the issues is I would not know (and I don't think anyone will know) the finer points such as the nucleation site density etc.

I was kind of hoping it would be as simple as telling CFX:
1. Here's some steam.
2. Here a vertical plate.
3. Condense it and tell me the results.

I should have known better.

I think in the first instance I should look into and read more about the simpler wall film model.

[ghorrocks]

I think you get the picture - any phase change stuff if modelled from first principles is feindishly complex.

My recommendation for the starting point is to really understand what you are trying to achieve with this model. Once you know what you want to learn you can decide an appropriate level of simplification.

[soumitra2102]

Originally Posted by ghorrocks

One possible approach would be to use the wall film model to model the liquid on the surface, and a second approach would be to directly model the wall film with a free surface model. These are totally different models, and the first would be much easier to implement than the second I suspect. But it depends what you are trying to get out of the model for which is more appropriate.

Do the two options differ in accuracy?

I want to accurately take into account the Conjugate Heat Transfer through the solid boundary to the other side of the condensate. I have experimental solid wall temperature data for comparison.

[ghorrocks]

Theoretically if you model everything (ie the free surface approach) it is the most accurate as little or no empirical correlations are required. But in reality the mesh and time steps required to do this require access to a super computer.

The wall film approach simplifies the situation greatly and makes it manageable, but relies on some simplifications.

Which is more suitable for your situation is not a simple question. All I can recommend is to do some research into the models, give them a few test goes and work it out for your application.