[zhubohong]

Hello, everyone
I want to predict the fluaction of free surface, which is affected by bubble flow process, as shown in figure. I simulated three phases flow, top gas, injected gas and liquid steel. The top gas and liquid steel were treated as the continue phase dealed with the Free Surface Model, and injected gas was treated as the dispersed phase dealed with the Particel Model.
However, in the real process, the top gas is actually the injected gas. But in my model, I divided it into two phases, because I want to reflect the free surface flow and bubble column simultaneously.
In addtion, the gas will be pumped out of the reactor in the real process. But I can't simulate this process. So, I want to make the injected gas phase disappear when it meets the top gas phase (when the volume fraction of top gas is equal to 0.95 at the interface). I think it can be achieved by modify the source item in CFX. But I don't kown how to do. I need you guys give me some advices. Thanks!

[ghorrocks]

What is the physical size of the bubbles? What is the physical size of the vessel? How many bubbles are there (is it isolated bubbles, lots of bubbles or a foam of bubbles)?

[zhubohong]

Originally Posted by ghorrocks

What is the physical size of the bubbles? What is the physical size of the vessel? How many bubbles are there (is it isolated bubbles, lots of bubbles or a foam of bubbles)?

bubble size:11-28mm, vessel size : diameter 2.1 m, height 2.8.

volume fraction of gas bubble >10%

[ghorrocks]

In that case:

It would be just possible to do this as a free surface model, if you have access to a large computing resources. Then you can do a direct model of all the physics. You also don't need to split the bubble phase from the continuous top gas phase, it can be all one gas phase. But this will be a seriously big simulation and you should only attempt it if you are experienced CFD operator with a lot of computer resources available.

If your resources are smaller then look at Eularian or Lagrangian particle tracking. It is not clear to me at the moment whether Eularian or Lagrangian would be the better approach at this stage. I suspect they both could do it, in which case you should probably use the Eularian model as it is easier to use. Also I don't think you need to remove the bubble phase from the simulation when it reaches the top of the steel. It might be OK to allow the gas bubble phase to exist in the gas continuous phase - I suspect this will not cause any problems. But you better check this before you start using it.

[zhubohong]

Originally Posted by ghorrocks

In that case:

It would be just possible to do this as a free surface model, if you have access to a large computing resources. Then you can do a direct model of all the physics. You also don't need to split the bubble phase from the continuous top gas phase, it can be all one gas phase. But this will be a seriously big simulation and you should only attempt it if you are experienced CFD operator with a lot of computer resources available.

If your resources are smaller then look at Eularian or Lagrangian particle tracking. It is not clear to me at the moment whether Eularian or Lagrangian would be the better approach at this stage. I suspect they both could do it, in which case you should probably use the Eularian model as it is easier to use. Also I don't think you need to remove the bubble phase from the simulation when it reaches the top of the steel. It might be OK to allow the gas bubble phase to exist in the gas continuous phase - I suspect this will not cause any problems. But you better check this before you start using it.

Thanks for your reply!
I think my choice is right. I must divide the gas into two phases, because I should accuratly predict the bubble behavior. It means all of interphase forces must be considered. So only using the Free surface model can not meet my requirement.
In addition, if I use the Lagrangain approach in my model, it will be meet the same problem. The bubble can not float out of the reactor. Of course, I can set the density of top gas with a larger value, but it doesn't match the reality. So, I think the better way is that making the injected gas disappear by modifying the source item. However, I just have this thought, I don't know how to do. I didn't found the related CFX tutorials.

[ghorrocks]

So only using the Free surface model can not meet my requirement.

You did not understand my suggestion. I was saying you could model the bubbles with a free surface model as well. This model includes all relevant physics, and in fact will be more accurate than the 3 phase model you are proposing. But it will require enormous computing resources and if you don't have that it cannot be considered.

If you want the bubbles to disappear when they get to the continuous gas phase you can have a source term which is multiplied by the volume fraction of the continuous gas phase. Then it will automatically apply in the continuous gas phase region only.

Also - Don't forget you can do this model by Eularian bubble model or a Lagrangian bubble model. The Eularian is likely to be easier, so I would use that one unless you can suggest a reason it will not work.

[zhubohong]

Originally Posted by ghorrocks

You did not understand my suggestion. I was saying you could model the bubbles with a free surface model as well. This model includes all relevant physics, and in fact will be more accurate than the 3 phase model you are proposing. But it will require enormous computing resources and if you don't have that it cannot be considered.

If you want the bubbles to disappear when they get to the continuous gas phase you can have a source term which is multiplied by the volume fraction of the continuous gas phase. Then it will automatically apply in the continuous gas phase region only.

Also - Don't forget you can do this model by Eularian bubble model or a Lagrangian bubble model. The Eularian is likely to be easier, so I would use that one unless you can suggest a reason it will not work.

If I use two phases and select the Free Surface option, there will be have a problem that the virtual mass force can not be applied. This reactor is a bottom blown reactor, the effect of the virtual mass force is not significant. But, I will simulate the same type of reactor with the horizontal injecting. At that time, the virtual mass force is very very important, because the large density difference and relative velocity/acceleration.
In addtion, if I use two phases to simulate, the outlet boundary is very hard to make sure. The overflow problem is coming. So, I should build a very large model. I used half a month to bluid the present model using PARTICEL+FREE SURFACE model. In the present model, I set a large value of the top gas density for letting injected gas flow out. But, the problem is obviously that, the larger density could restrict the fluaction of free surface. So, I think my final target about my model is to conquer the above problem.
For the Lagrangian approach, actually, I have already used this method to simulate. I set the top gas density with a large value and let them flow out. Though the predicted results can be accepted, I still doubt whether this method is the right. This is because the high speed of injection gas and large volume fraction. So, another aim of my present work is to compare this two approaches, and discuss which approach may be more suitable for this chemical reactor. This is why I always emphasize I want to use the Eularian model at this time.
Very very thanks for your reply!!

[ghorrocks]

If I use two phases and select the Free Surface option, there will be have a problem that the virtual mass force can not be applied.

No. If you model the bubbles with a free surface model it will directly model the virtual mass effect.

In addtion, if I use two phases to simulate, the outlet boundary is very hard to make sure.

No, the top boundary is simply just a pressure boundary. Very easy. But if you are not going to do the big full free surface simulation the discussion is not important.

Did you see my last comment on your original question about making the bubble phase disappear at the gas continuous phase? I said "If you want the bubbles to disappear when they get to the continuous gas phase you can have a source term which is multiplied by the volume fraction of the continuous gas phase. Then it will automatically apply in the continuous gas phase region only."

[zhubohong]

Originally Posted by ghorrocks

No. If you model the bubbles with a free surface model it will directly model the virtual mass effect.



No, the top boundary is simply just a pressure boundary. Very easy. But if you are not going to do the big full free surface simulation the discussion is not important.

Did you see my last comment on your original question about making the bubble phase disappear at the gas continuous phase? I said "If you want the bubbles to disappear when they get to the continuous gas phase you can have a source term which is multiplied by the volume fraction of the continuous gas phase. Then it will automatically apply in the continuous gas phase region only."

Thanks, I will seriouly check your comments again! Very thanks! As shown in the below figures, the virtual mass force can not be selected. Did I neglect something?

Another question:
I am very make sure if I choose the interphase transfer with Particel Model, the predict results of bubble column is very nice. But if I change to use the Free surface model, it is suitable? As we know, the Free surface model makes the two phases have a stratification trend, it is suitable for bubble column?

Sorry, because I have no any experience about modifying the source, I don't sensitive to your this comment：“If you want the bubbles to disappear when they get to the continuous gas phase you can have a source term which is multiplied by the volume fraction of the continuous gas phase. Then it will automatically apply in the continuous gas phase region only.” I will consider it later, seriouly!

[ghorrocks]

As shown in the below figures, the virtual mass force can not be selected. Did I neglect something?

In the free surface model you do not have an option to include the Virtual Mass as the free surface model will directly model the virtual mass and you do not need an external model to include it.

What don't you understand about my suggestion for a source term to make the bubbles disappear? If you make the source term which makes the gas bubble phase disappear multiply by the continuous gas volume fraction then it will be zero in the steel (so it does not apply) and full value in the continuous gas region (so it applies). This appears to answer your original question of how to make the bubble gas phase disappear in the continuous gas phase.

[zhubohong]

Originally Posted by ghorrocks

In the free surface model you do not have an option to include the Virtual Mass as the free surface model will directly model the virtual mass and you do not need an external model to include it.

What don't you understand about my suggestion for a source term to make the bubbles disappear? If you make the source term which makes the gas bubble phase disappear multiply by the continuous gas volume fraction then it will be zero in the steel (so it does not apply) and full value in the continuous gas region (so it applies). This appears to answer your original question of how to make the bubble gas phase disappear in the continuous gas phase.

I have known what you mean, but I still don't know the specific method. The CFX official tutorial in this area is so weak. Could you tell me more details about modifying method, or give me some similar examples? Of cource, I also hope you can guide me step by step to acheive the final purpose.

[ghorrocks]

The "Free surface flow over a bump" tutorial shows the use of a free surface model.

Other tutorials cover lagrangian models, eularian models, phase change, and so on. The CFX documentation has extensive details of the available models and on the ANSYS Customer website there is training material on multiphase models which goes into a lot of depth and is highly recommended. If you think the official tutorials are weak then I suspect you have not seen all the available material - it is extensive and thorough.

I will not have time to write a step by step tutorial for you. If you don't understand what you are doing you should not be attempting a challenging multiphase simulation.

[zhubohong]

Originally Posted by ghorrocks

The "Free surface flow over a bump" tutorial shows the use of a free surface model.

Other tutorials cover lagrangian models, eularian models, phase change, and so on. The CFX documentation has extensive details of the available models and on the ANSYS Customer website there is training material on multiphase models which goes into a lot of depth and is highly recommended. If you think the official tutorials are weak then I suspect you have not seen all the available material - it is extensive and thorough.

I will not have time to write a step by step tutorial for you. If you don't understand what you are doing you should not be attempting a challenging multiphase simulation.

I will start with reading the all of CFX tutorials I have. Could you give me more tips? To make the gas phase disappear, does it involves the phase change? just like Steam Jet Model, or I can directly let it disappear, which does not involve the phase change.

[ghorrocks]

The source term approach is not a phase change application. It will just magically disappear, there is no physics behind it.

[zhubohong]

Originally Posted by ghorrocks

The source term approach is not a phase change application. It will just magically disappear, there is no physics behind it.

Just like the "simplified version" of degassing condition for the bubble phase?

[ghorrocks]

It is similar, I guess. In that case it assumes the boundary is the location of the free surface and allows the dispersed phase to cross it and leave the domain and does not allow the continuous phase to leave, it acts like a wall.

[zhubohong]

Originally Posted by ghorrocks

It is similar, I guess. In that case it assumes the boundary is the location of the free surface and allows the dispersed phase to cross it and leave the domain and does not allow the continuous phase to leave, it acts like a wall.

Ok, it still a wrong way. I don't need the bubble phase cross it(free surface), I need it disappears. It means the value of volume fraction of bubble change to zero suddenly. So, the problem is how to achieve it.
Is it possible to modify the property of injected gas? Just like, when the injected bubble meets the top gas phase, the density of it change to zero? If the density value of injected gas is equal to zero, is there the injected gas phase still in domain? Otherwise, I still need to find a way around modifying the source.

[ghorrocks]

You can't make the density of a fluid go to zero. It will crash the solver.

I have already said three times in this thread how to make a source term to remove the gas bubble phase in the continuous gas phase region. You define a source term to set the volume fraction of the gas bubble phase to zero, and multiply the source term by the volume fraction of the continuous gas phase.

If you don't understand my suggestion please tell me what you would like me to explain. If you continue to ignore my suggestions I will give up responding to you.

[zhubohong]

Originally Posted by ghorrocks

You can't make the density of a fluid go to zero. It will crash the solver.

I have already said three times in this thread how to make a source term to remove the gas bubble phase in the continuous gas phase region. You define a source term to set the volume fraction of the gas bubble phase to zero, and multiply the source term by the volume fraction of the continuous gas phase.

If you don't understand my suggestion please tell me what you would like me to explain. If you continue to ignore my suggestions I will give up responding to you.

Thanks for your time. I must firstly say that I never ignore your suggestions, and I just don't understand your suggestion. This is because I'm weak in this part of my knowledge. I can't build a relationship between Source term, volume fraction and the result of gas bubble disappearing. For example, I read a example of solidification simulation, after that, I think I know how to simulate it, because I can build a relationship between Source term, velocity and the result of solidification to achieve the purpose. But this time, I can't develop the relationship from your suggestion. This is also why I always can't get your point. Could you give me more explanations for help me build the relationship?