CFD Online Discussion Forums

CFD Online Discussion Forums (
-   FLUENT (
-   -   Converge problem for multiphase flow (

Jen July 20, 2005 11:23

Converge problem for multiphase flow
I am modeling two phase jet flow using mixture model and k-epsilon turbulent model in 3D space. I can get nice convergence if I just calculate air(phase 1). After I add water (phase 2), the solution will never be converge. The velocity of the air is 70m/s, and 2m/s for water. I can get it converge if the air velocity is 130m/s, but I can't do it for 70m/s.

The quality of mesh is pretty good, there are more than 90% between 0 and 0.5. I have tried to decrease the relax factor of k and epsilon, but the residuals of epsilon and continuity are always high. The display contour of air and water velocity is good, but why it never converge? Is there anything I can do to make it converge?

Thanks very much.

Jen July 20, 2005 11:48

Re: Converge problem for multiphase flow
What I want to do is first using Mixture model to get the steady state solution, and then change it to unsteady Eulerian model, because I want to know the slip velocity between these two phases. But now, I have a question. My air inlet and water inlet are different, the cycular water inlet is in the center of annular air inlet. In this case, is the Multiphase model proper to use? Or I should use VOF? Any help will be appreciated.


edi ghirardi July 20, 2005 11:54

Re: Converge problem for multiphase flow
Don't use VOF, it is not able to handle slip velocity between phases.


Jen July 20, 2005 12:39

Re: Converge problem for multiphase flow
Thanks. Then you mean I have to choose multiphase model? What I want is use unsteady Eulerian model to calculate the slip velocity. According to the help menu, it says Eulerian model is difficult to converge, so you'd better use mixture first and then change to an Eulerian model, but now, even the steady mixture model does not converge. :( What could I do?

zxaar July 20, 2005 16:52

Re: Converge problem for multiphase flow
can you converge this on a very coarse mesh. if yes then get that results, and write it as initial guess. then read this initial guess in finer mesh and iterate, it should not trouble you then.

All times are GMT -4. The time now is 02:52.