[sersunzo]

Hi everybody,

I have a question concerning twoPhaseEulerFoam.

I have a rectangular channel where gas is introduced from one of the walls. The left wall and the bottom are just walls, the right wall is the gas-inlet and the top is open. The gas escapes from the top, which is the free liquid surface. it is laminar. As boundary conditions at the top I tried to replicate what are something called 'degassing b.c.' in other codes. The top, for the gas is an outlet. The liquid, on the other hand, cannot escape the boundary and I impose a free velocity slip as it should be for a gas-liquid surface. The pressure at the top is fixed. The results seem logical. There is a transient when the gas gradually fills the channel and when it reaches a certain value it oscillates around this value.

Now, however, I have a doubt.

If the liquid cannot escape the channel, how can the overall void fraction change? I inject gas, but liquid cannot exit; how it is possible in an incompressible liquid? A real system would expand, but in my case the boundary is fixed. I tried to understand it from the code, but I am not sure I really understood it. I think the pressure is not maintained fixed but it changes in order to compensate for the lack of expansion. It is correct? If it is, do you know how exactly is the pressure changed? In the b.c., I imposed fixed pressure at the top, is this overridden by the code? And, if this is the case, what exactly happens: the pressure is maintained fixed for the gas phase, but changes for the liquid phase? How is it possible?

Thank you in advance and best regards

Alessio

[alberto]

Hi,

one question before trying to answer: is the channel filled of liquid, with phase fraction equal to 1 at the beginning of the simulation, and then you start injecting gas?

Best,

[sersunzo]

Originally Posted by alberto

Hi,

one question before trying to answer: is the channel filled of liquid, with phase fraction equal to 1 at the beginning of the simulation, and then you start injecting gas?

Best,

Yes, it is

[alberto]

OK. So your outlet BC actually lets the liquid go out during the expansion (most probable), or no gas could enter the domain.

One trick could be to patch the initial volume fraction with a fraction lower than 1 and close to the expected value, but it is cheating a bit ;-)

Best,

[sersunzo]

Originally Posted by alberto

OK. So your outlet BC actually lets the liquid go out during the expansion (most probable), or no gas could enter the domain.

One trick could be to patch the initial volume fraction with a fraction lower than 1 and close to the expected value, but it is cheating a bit ;-)

Best,

Hi Alberto,
thanks for your reply.
>So your outlet BC actually lets the liquid go out during the expansion

how it is possible? The b.c. I specify do not allow it. Are they, somehow overriden?

[alberto]

You fix a pressure value at the outlet and a slip condition for the velocity on the same boundary, if I understood it correctly?

Best,

[sersunzo]

Originally Posted by alberto

You fix a pressure value at the outlet and a slip condition for the velocity on the same boundary, if I understood it correctly.

The slip condition does not set the velocity to zero anywhere. It states that the velocity gradient normal to the boundary is zero, and lets the flow move freely in the direction tangential to the boundary.

Best,

Hi Alberto,

thanks for your answer again. Yes, the tangential velocity is not zero, but the liquid does not exit the channel because the normal velocity is zero. Moreover, I tested a case with v_liquid=0 (only the gas can exit) and it is almost the same. Imagine just a single big computational cell. At time t=0, this is completely full of liquid (alpha=0). Then, some gas is injected and let's say at time t1 alpha=0.1. My top boundary conditions does not allow the liquid to exit the computational cell, but now beta=1-alpha=0.9. Thus, if the liquid cannot exit, where is the 10% of the incompressible liquid missing?
They told me that, in this case, mass is not conserved and the code allows somehow some liquid to "disappear", but I do not understand how, numerically, this is handled. Do you have any information about this?

Thanks in advance

Alessio

[alberto]

Hi, sorry for my last reply, which contained an error, since clearly the normal component of U is set to zero with the slip condition, or the "wall" would be permeable.

Check the phase fluxes, and their change to see where the mass goes. The code enforces the total mass conservation.

Best,