Hi,

I am modeling two immiscible liquids (oil-water) which flow in different velocities in slightly upward inclined pipe. The pipe is 1 meter in length and 14mm in diameter.

My interest is following: 1) Location of interface. 2) Velocity profile.

I did following for a startup.

Code: FLUENT 6 Problem: 3-D Model: VOF Laminar flow. Steady-state with implicit scheme.

Discretization

Pressure: body force weighted

Pressure velocity coupling: SIMPLE

Momentum: 2nd order upwind

Volume fraction: 2nd order upwind

Under-relaxation factor: 0.2

Operation condition: gravity added.

Boundary conditions: Two velocity inlet boundary conditions at one end of the pipe, one outflow boundary condition at the other end of the pipe. No slip condition on the pipe wall.

Grid: some 140,000 cells

The size of the cell is 0.3x0.3x5 mm.

After running the FLUENT, the solution was converged but there is a numerical dispersion in the interface and there is a big discrepancy between FLUENT result and the result in literature (velocity profile and holdup).

Following is my questions:

1) Is my VOF setting OK?

2) How can I get less dispersed interface between the phases (I increased the mesh number, but again I got dispersed interface)?

3) In order to read the location of the interface, I set the level of the contour to 2, and then read the interface location with the mouse or I write the volume fraction of both phase with their special coordinates to the file then I draw the curves and then take the number where both of the volume fraction curves intersect. Is there any good way to read the interface location correctly?

4) In the FLUENT manual (VOF), it says that in cases where large velocity differences exit between the phases, the accuracy of the velocities computed near the interface can be adversely affected. My other question is that what is the upper limit of these differences which course the VOF model works not well?

Thanks for your attention.

Be well,

Rey

whats the convergence criteria ? How do you know that the solution converged ? Why do you expect a steady state solution? Why didn't you try geo scheme (time dependent)which is most accurate?

DC

Hi, David.

Thanks for your reply and your suggestion.

My convergence criteria was the residuals of continuity, X, Y and Z velocities and volume fraction of water. The setting was 1.0e-06.

I also monitored (using surface monitor) the value of water volume fraction in different locations along the pipe in order to see how it changes during iteration. This value was changing at the beginning of the iteration, and the changes appear to be decaying. After long iteration all values reach an equilibrium position and stay the same and the values also agree each other. At the end, the machine indicated that the criterion for the convergence was satisfied.

FLUENT manual says that if there is distinct inflow boundary condition and the final solution is not dependent on initial flow condition, one can use steady-state solution. I followed these recommendation since my boundary conditions satisfies it and also I don't interested in the intermediate solutions.

Any way I will try your suggestion (unsteady case), let me see what happens.

Be well,

Rey

Hello Ray,

shouldn't you include surface tension forces to better track the interface between phases?

anna

Hi, Anna,

I did add the surface tension as another case but I got very smearing interface comparing to the case including the surface tension. Still, there was big gap between the FLUENT result and the results in literature.

Rey

Hi, Using Unsteady state and Geo-Reconstruct will certainly give you more distinct phase boundaries