[DarrenC]

Hi all,

I am working through this problem whereby I am simulating a oil-water-gas separator using the CFX-multiphase-inhomogeneous and I am having some very peculiar mass balance and overflow problems.

A rundown of my simulation : I have 3 inlets - oil, water and gas connected together by a sphere to induce some mixing of the 3 inlets. the outlet of the sphere is then piped to my oil-water-gas separator to simulate the separation of the 3 continuous phases.

When I run my simulation using the inhomogeneous model in steady state, I seem to be getting some mass balance problems whereby the total mass flow of the liquid phases at the oil and water inlets do not add up to the total liquid mass flow of the outlet of the sphere. The total liquid flowing out of the sphere is approximately half of what is going in. However the mass flow of gas at the inlet does equate the mass flow of gas out of the sphere. This problem I find perplexing.

Then I tried to run the problem in transient mode to see if the problem is solved but the problem is the same. To make matters worse in transient mode, after simulating the steady flow of the system in transient mode for approximately t=7.5s, I get this sudden huge mass flux of liquid going into the separator and the simulation sometimes overflows and crashes. Initially I suspected it is a timestep problem and I have tried turning down the timestep so that it gives a global courant number to approximately 100, but the sudden liquid influx problem still persists.

I am not sure what is causing the mass imbalance. I have tried using the double precision solver, running it in steady and transient, but to no avail. Im also not sure if both problem are independent or they are somehow connected? Anybody have an idea what is going on? Any help is greatly appreciated.

Thanks

Cheers
DarrenC

[ghorrocks]

Have you looked at what your flow is doing over time? Before the level gets to the sphere outlet the sphere outlet flows just air. I bet your sphere thing is filling up with water and oil, then suddenly a big glob of water/oil level goes down the sphere outlet.

I would recommend a smaller time step in this time when the sphere fills up. Also improved mesh quality would help. You can also accelerate things by starting with an initial condition where the sphere is partly filled.

[DarrenC]

Hi Glenn,

I think I have a rough idea what it going on. The 2 liquid inlets into the sphere are riser style inlets (i.e pipe is vertical and then turns 90 degrees and connects horizontally into the sphere). As the liquid flow rates are not high the horizontal portions of the inlets will not be filled fully with liquid. There will be a gas cap on top of the horizontal section of the inlet pipe. What is happening is the high flow rate gas is blowing on the surface of the liquid in the pipe, causing the liquid in the horizontal section to become 'wavy' and unsteady. At a certain point the liquid becomes so wavy that the gas 'scoops' a large amount of liquid from the vertical section of the liquid pipe also and accelerates it into the sphere and subsequently the outlet of the sphere. At this point the courant number as low as 10-20 increases rapidly to 999.00 and the simulation overflows shortly after.

The mass flow from the gas inlet is approximately 4 times the mass flow of the liquid inlet so i guess it is possible for this to happen physically. What do you think?

I have been using this inlet device design for quite a while and have never seen a problem like this until I switched from homogeneous multiphase to inhomogeneous multiphase.

Cheers
DarrenC

[ghorrocks]

Your explanation is feasible. Weird things like that happen in multiphase flows, that is why they are so challenging.

You would have to have a close look at the results and compare the homogenous and inhomogeneous models to understand the difference - and the important thing is to decide whether it is real or not.

[mohsen_ba65]

Originally Posted by DarrenC

Hi Glenn,

I think I have a rough idea what it going on. The 2 liquid inlets into the sphere are riser style inlets (i.e pipe is vertical and then turns 90 degrees and connects horizontally into the sphere). As the liquid flow rates are not high the horizontal portions of the inlets will not be filled fully with liquid. There will be a gas cap on top of the horizontal section of the inlet pipe. What is happening is the high flow rate gas is blowing on the surface of the liquid in the pipe, causing the liquid in the horizontal section to become 'wavy' and unsteady. At a certain point the liquid becomes so wavy that the gas 'scoops' a large amount of liquid from the vertical section of the liquid pipe also and accelerates it into the sphere and subsequently the outlet of the sphere. At this point the courant number as low as 10-20 increases rapidly to 999.00 and the simulation overflows shortly after.

The mass flow from the gas inlet is approximately 4 times the mass flow of the liquid inlet so i guess it is possible for this to happen physically. What do you think?

I have been using this inlet device design for quite a while and have never seen a problem like this until I switched from homogeneous multiphase to inhomogeneous multiphase.

Cheers
DarrenC

Hi Darren,

I am doing the same simulation and I ran to this problem. Could you finally manage your problem?

My simulation runs are smooth until 0.83 s. After that, I guess it is the time that a bubble of air wants to pass the water phase and leave the vessel, the Courant number increases, velocity at some point increases, and my simulation is no more stable.
I am using compressible model in transient condition.
Do you have any suggestions?

[ghorrocks]

I could see this happening it is being driven by a pressure boundary. Is that correct?