[andrea93]

Dear all,

I want to simulate the Nusselt flat plate condensation problem, but I have been struggling without success for two weeks.

At first, I tried to use the solver "reactingTwoPhaseEulerFoam", modelling the condensation through the Antoine model (I tried many models, such as constant saturation conditions, but Antoine seems to be less bad one - I guess I set the model up in the wrong way).
After that, I tried to use compressibleInterFilmFoam, but I have many problems in setting up the file "surfaceFilmProperies", which is in the "constant" folder.

I attach two folders: one is the reactingTwoPhaseEulerFoam case; the other one is about compressibleInterFilmFoam.

Please, could you help me to find a solution? I would definitely appreciate.
If you are still here, that means you read all the post, so thank you for your time!
Andrea Iovenitti

[mizzou]

Originally Posted by andrea93

Dear all,

I want to simulate the Nusselt flat plate condensation problem, but I have been struggling without success for two weeks.

At first, I tried to use the solver "reactingTwoPhaseEulerFoam", modelling the condensation through the Antoine model (I tried many models, such as constant saturation conditions, but Antoine seems to be less bad one - I guess I set the model up in the wrong way).
After that, I tried to use compressibleInterFilmFoam, but I have many problems in setting up the file "surfaceFilmProperies", which is in the "constant" folder.

I attach two folders: one is the reactingTwoPhaseEulerFoam case; the other one is about compressibleInterFilmFoam.

Please, could you help me to find a solution? I would definitely appreciate.
If you are still here, that means you read all the post, so thank you for your time!
Andrea Iovenitti

Any succeed? I also like to model Nusselt condensation.

Thanks,

[andrea93]

Originally Posted by mizzou

Any succeed? I also like to model Nusselt condensation.

Thanks,

Hi mizzou,

unfortunately I have no positive news to give you. I am still working on that. If I find some ways to solve it, I will tell you.

Andrea

[mizzou]

Originally Posted by andrea93

Hi mizzou,

unfortunately I have no positive news to give you. I am still working on that. If I find some ways to solve it, I will tell you.

Andrea

Thank you so much Andrea. I just started to work on that and if I get any succeed I'll let u know.

[andrea93]

Originally Posted by mizzou

Thank you so much Andrea. I just started to work on that and if I get any succeed I'll let u know.

Hi mizzou,

I want you to know that I made some progress.
First, I deeply modified InterPhaseChangeFoam in order to implement some "thermal models" (Merkle, Kunz and ScherrSauer face the cavitation problem, whose physics is completely different from ours).
After that, I made the same modifications in CompressibleInterFoam.

I want to be as clear as possible: I did not post my solver because I still have many problems and I want you to know them so that maybe you can help me (I don't like to share things which are not 100% correct). The problem is the following one.

I simulated the Nusselt problem using water at 646K and 219 bar. You might say: "Why these awful condition?". The answer is easy: when you are close to supercrical fluid, liquid and vapour properties are similar, so at the interface you don't have a sharp discontinuity. Conversely, at 1bar and 373.15K, thermophysical properties are very different each other. This fact has a dramatic input on the simulation: in particular, pEqn starts being iterated 1000 times and the Courant number explode (>10000), meaning that the interfacial velocity is badly calculated.

Because of this, I tried to simulate a single phase flow of water (it does not matter if it is liquid or vapour) using InterFoam, interPhaseChangeFoam and CompressibleInterFoam. In all these cases the simulation creashes. Of course I paid a lot of attention in setting the proper boundary conditions. As a consequence, I realized that, MAYBE, the Nusselt simulation at 1bar and 373.15K has problems which are not strictly related to my model. Indeed, there might be a something much more strange.

Hoping this might help.
Andrea

[binit kumar]

hello andrea,
I need help to model condensation on flat plate, i am using star ccm+ software to deal the condensation so can you help me to work out with condensation problem.
i am usinf vof model, eulerian phase, and user code for the phase change... the user code is same as lee model... but still my results are not acceptable.

another method which i applied was fluid film model but i have still not got any progress in it...

will you please help me...

[andrea93]

Originally Posted by binit kumar

hello andrea,
I need help to model condensation on flat plate, i am using star ccm+ software to deal the condensation so can you help me to work out with condensation problem.
i am usinf vof model, eulerian phase, and user code for the phase change... the user code is same as lee model... but still my results are not acceptable.

another method which i applied was fluid film model but i have still not got any progress in it...

will you please help me...

Hi binit,

first of all, during the last year I started working on thermal phase change with VOF using OpenFOAM 5.0. I never used Star Ccm+ for multiphase flows with condensation, but I know that software since I worked with in the past.

According to my knowledge, I would say that: the problem is inside VOF methods, because they are not mass conservative when dealing with phase change.

Basically, if you assume to have two phases, both immiscible and incompressible, you could get from mass conservations that:

div(U) = (dm/dt)''' (1/rhoL-1/rhoV)

where:
- U= one fluid velocity. It is the mixture velocity, that means: U=alphaL*UL+alphaV*U, alpha_i being the volume fractions;
- (dm/dt)''' is the mass transfer rate, expressed in (kg/(s m^3).

All the softwares I know assume that div(U)=0 (divergence free velocity), meaning that the phase change rate is zero. Starting from this assumption a way to track the interface is derived.

No matter if you are using level set, PLIC, MULES or whatever else to reconstruct the interface, you should always remember that: you do not have only advection. You also have a phase change and this implies that the one fluid does not follow an incompressible flow (which means div(U)=0). This causes a compression/dilatation rate at the interface (depending on the type of thermal phase change - i.e. condensation or boiling).

In short, the problem has one more piece of physics that must be taken into account, but equations rarely consider that.

Till now I just talked about the physics. So, let's now turn to the numerics.

In Comsol Multiphysics the level set multiphase solver implements an equation that is "not really physical", since it has some source/sink terms which are just related to numerics. More detail can be found here:

https://www.sciencedirect.com/scienc...21999105002184

Basically, even if you assume div(U)=0, you might have stability issues and this explains why additional terms are required. Obviously, if you have phase change the problem can be even worse.

What to do (in principle)? Start from zero, get the equation, perform the Von Neumann stability analysis, VERIFY MASS CONSERVATION. You could not imagine how many s**t you can find online; as an example, I developed a solver for condensation and boiling. When performing the verification procedure to the Nusselt problem, it perfectly works! However... mass is not conserved. What is the conclusion? The solver is wrong. Full stop. If you show verification tests, but not verify at least mass conservation, your solver is simply useless... Sorry for this digression, but internet is full of fake good works, as mine, but many times the authors do not mention these things.

However, this is not the case of Star, because you should pay for a license. Let's come back to our stuff.

Another issue related to numerics is the following one: both StarCcm+, Ansys Fluent and OpenFOAM have a colocated mesh. If you use PISO, PIMPLE algorithms, the splitting error due to the U-p coupling might in principle be enhanced by the interpolation procedure that is required to get the surface values. At those cells where you have the interface, you might also have the effect of div(U)!=0. Obviously, this is not a positive news.

In conclusion, I think that your results are not acceptabe because the equations that are implemented do not assume div(U)!=0. If they do that, the interface tracking method should be troublesome. You might verify things such as the film width, but mass conservation might be not verified because of the numerics related to the interfacial cells.

Hope this might help.
Andrea

[binit kumar]

Thanks Andrea for your kind help...
i need one favor from you... have you solved the same problem on your software?? if so then can you tell me the details of the physics and properties how you have considered for condensation process.. and if you have used any coding in that then can you supply me that... as i have got stacked in this problem...

so can you please me send me the details on mail or as you wish to share... even if you have used fluent software then i can handle that software too...
my mail id - binit.kumar17@gmail.com

[andrea93]

I used OpenFOAM 5.0. I should simulate liquid and vapor lithium, but I finally used quasi supercritical water. The reason is related to the fact that in the first case the huge thermophysical difference might lead to a maxDeltaT which is not feasible for my computer (dt<1e-9).

Unfortunately, I could not give you the code, because the research group I belong to has decided not to publish anything, unless it is 100% correct. At the moment, we are still working on that.

[binit kumar]

Thanks Andrea