Multiphase simulation of pipes

Fededea

Hello,

I want to try to simulate a flow rate in a pipe with immiscible multiphase (for example: air and oil or air and water). I want to see how the pipes filled with air fill with the other fluid. I have 1 inlet and many outlets (e.g., 4-6).

I used volumetric flow rate as the oil inlet and the outlets for ambient pressure with hydrostatic pressure. I also set the force of gravity.

When I start the simulation I get this warning, "*** WARNING: missing node/element BC for fluid boundary node 768282."

The residual and solution ratios are not stable and also by changing the time step the pipe does not fill with oil but remains full of air. I also get the inflow to outflow warning.

Are the boundary conditions correct? How can I solve this?

The real problem is a pipe that starts out empty with no oil, then a pump with constant flow starts and the pipe fills with oil to a steady state. During the whole phenomenon, the outlets discharge oil into a tank with ambient pressure. I want to see this transient phase to understand what the flow rate behavior is during the systems startup.

Thank you very much,

regards.

acupro

Fededea said:

I solved the problem of missing BC node/element warning. 

Yes, I used HyperWorks CFD to create the model (after importing the cad geometry from another software). I did not make any changes to the input file.

This is a good question. I think I can have it both ways, because the flow rate can vary from 1 l/min to 10-15 l/min, with an input velocity from 0.0000001 m/s to 0.0000021 m/s. I have seen some experiments with this circuit, and I believe that sometimes air can be dispersed, but I am not sure. In case of having a lot of churning, it is necessary to use which model?

Thank you,

regards

You would probably want to try immiscible (level set - maybe with enhance feature capturing), disperse multiphase (Algebraic Eulerian), and Eulerian (Eulerian-Eulerian).  They will all require a fairly dense mesh and small time steps.

acupro

Definitely a problem if you have exterior nodes of the domain without any boundary condition.

Did you do any manual editing of the input file after writing from HyperWorks CFD?  (Looks like that's where you created the model - but please confirm that, too.)

Do you expect a lot of churning in the pipe(s) with a fast oil injection, or do you expect a fairly stable interface between the fluids with a slow oil injection?  If you expect a lot of churning, immiscible will not be a good choice.

Fededea

acupro_21778 said:

Definitely a problem if you have exterior nodes of the domain without any boundary condition.

Did you do any manual editing of the input file after writing from HyperWorks CFD?  (Looks like that's where you created the model - but please confirm that, too.)

Do you expect a lot of churning in the pipe(s) with a fast oil injection, or do you expect a fairly stable interface between the fluids with a slow oil injection?  If you expect a lot of churning, immiscible will not be a good choice.

I solved the problem of missing BC node/element warning. 

Yes, I used HyperWorks CFD to create the model (after importing the cad geometry from another software). I did not make any changes to the input file.

This is a good question. I think I can have it both ways, because the flow rate can vary from 1 l/min to 10-15 l/min, with an input velocity from 0.0000001 m/s to 0.0000021 m/s. I have seen some experiments with this circuit, and I believe that sometimes air can be dispersed, but I am not sure. In case of having a lot of churning, it is necessary to use which model?

Thank you,

regards

acupro

Fededea said:

I solved the problem of missing BC node/element warning. 

Yes, I used HyperWorks CFD to create the model (after importing the cad geometry from another software). I did not make any changes to the input file.

This is a good question. I think I can have it both ways, because the flow rate can vary from 1 l/min to 10-15 l/min, with an input velocity from 0.0000001 m/s to 0.0000021 m/s. I have seen some experiments with this circuit, and I believe that sometimes air can be dispersed, but I am not sure. In case of having a lot of churning, it is necessary to use which model?

Thank you,

regards

You would probably want to try immiscible (level set - maybe with enhance feature capturing), disperse multiphase (Algebraic Eulerian), and Eulerian (Eulerian-Eulerian).  They will all require a fairly dense mesh and small time steps.

Fededea

acupro_21778 said:

You would probably want to try immiscible (level set - maybe with enhance feature capturing), disperse multiphase (Algebraic Eulerian), and Eulerian (Eulerian-Eulerian).  They will all require a fairly dense mesh and small time steps.

Thanks a lot for all the calrification.