## Fluids

Topics relate to Fluent, CFX, Turbogrid and more

#### How to improve the phase separation efficiency in a VOF model

• Fabrizio24
Subscriber

I am trying to simulate a three phase (gas, oil and water) separator in ANSYS Fluent with a VOF multiphase stationary model. Gas as primary phase and oil and water as secondary phases. The viscosity model is Standard k-e with standard wall functions. To control the interface level of water and oil, I use mass flow as the input and output boundary condition.

Please I need to increase the separation efficiency of water and oil because it does not match with the separation capacity of the real separator ?

The pressure-velocity coupling scheme is coupled, for the spatial discretization Pressure: PRESTO, Momentum: First Order Upwind, Volume Fraction: Modified HRIC and Turbulence: First Order Upwind

• Rob
Ansys Employee

Please can you post some images?

How many inlets did you use? If just the one, please re-read the manual covering VOF.

• Fabrizio24
Subscriber

Hi, Yes, I have one inlet as a mixture (gas 0.000539 kg/s, oil 5.49 kg/s and water 60.13 kg/s) and three separate outles for gas, oil and water with 0.000539 kg/s, 8.23 kg/s and 57.39 kg/s respectively.

The next image shows the simulation. Initially patch the water and oil zones in each chamber of the separator

I understand that it is necessary to divide the inlet for each phase even considering the same velocity?

• Rob
Ansys Employee

OK, with VOF we track the free surface so a mixed boundary shouldn't add anything at all. For a mixed system that's going to separate (as above) I'd suggest reading up on the Eulerian or Mixture model.

Also, do not use all fixed flow boundaries. At least one boundary must be "free" to allow the pressure to stabilise. In a simple pipe, setting velocity in & and velocity out will result in vacuum or infinite pressure as the two flows mismatch by +/- 0.001% when running steady state: same for transient but it'll take longer.

• Fabrizio24
Subscriber

Thank you for the answer, I will try with an Eulerian or Mixture model