Dear all cfders:

I use VOF method to simulate the mutiphase flows (air and water) base on the FEM method. I encountered a difficulty on setting the density. If the ratio of the parameter of air and water is smaller than 1:10, it converges, otherwise, it diverges. But, the actual case is 1:1000. In my procedure, the density is set to RhoA*(1-F)+RhoW*F, RhoA and RhoW are the density of air and water, respectively, and F is the volume fraction. Am i right? Moreover, in the single phase flows, the procedure works well. The model consists of the characteristic method+UZAWA type PCG method+VOF method. The problem is break dam, and all boundaries is slip type.

Thanks in advance!

Chunning JI

It is typical with high jump coefficient, so look at better solver for that case.

Dear Junseok Kim:

Thank you very much. Can you recommend me a better solver or something about it. In my procedure, a Uzawa type preconditioned conjugated gradient method is adopted. Also, I wondered whether the model for sigle fluid can be transplated to the one for multiphase by only using variable density and viscosity of the fluid. If it is not true, what should be done?

Chunning JI

Dear Junseok Kim:

In the multiphase flow, is the surface tension necessary? If it is omitted, would the result diverges?

thanks a lot.

Chunning JI

I am not sure about FEM version, but multigrid preconditioned conjugated gradient method is one good solver for finite difference.

In the free surace, i.e., the density and viscosity of one of phase is negligible, then you can use single phase solver with free boundary condition for pressure solver.

See the reference,

An investigation of the flow regime rsulting from splashing drops by David Morton, et al. Physics of fluids, vol 12 747-763.

If the surface tension is important, then it is necessary , but some cases, it is negligible.

Most of cases, surface tension is omitted, the solver is more stable.

Dear Junseok Kim

Thank you for you advise and I will try it later. In you opinion, the pressure condition p=0 should be set on the free surface. I wonder whether the computational domain above the free surface (i.e., the element with f=0, f is the fraction) is also needed to set p=0. If the pressure of the domain above free surface is settled to zero, whether the velocity is also be enforced to zero?

Thank you very much.

Chunning JI

I didn't use one phase simulation, so would you refer reference paeprs?