[jherb]

Hello,

I try to simulate the mixing of two fluids with different densities. I use the solver twoLiquidMixingFoam. After some (simulation) time the solver is not converging anymore (it seems the pressure equation is causing the problem).

Therefore some questions:
Is twoLiquidMixingFoam based on pimpleFoam? Then a maximal Courant number larger than 1 should be allowed?!?

In the pressure equation of twoLiquidMxingFoam there is no call to p_rgh.relax(). After I added this (following the example of pimpleFoam), the solver seems more stable but after some time the solver again does not converge. But it is not clear, what causes the problem. The Courant number is increasing constantly. Suddenly it gets much larger and after a few iterations it reaches infinity. There are some cells which show large values of p, U, k and omega at these time steps. But it is not clear if a certain equation is causing the problem.

(At the beginning) the mean Courant Number is about 0.06, the maximal value is about 15.

Reducing the Courant Number by a factor of 10 or more is not really an options because this would result in much to large computation times.

I also tried relaxing the equation for alpha1, but this does not help either. Also bounding the value of alpha1 between 0 and 1 does not help.

If alpha1 is set to zero at all inlets there are no convergence problems. So it seems somehow related to the alpha1 equation.

What other options are there to get an converging result?

(With CFX the problem can be simulated without these problems. The same mesh is used.)

Thank you for your help
Joachim

[gwierink]

Hi Joachim,

Could it be that your time step is a bit large? Yours is 0.04 and in the twoStream example in this thread it is 2e-4 ...

[jherb]

Originally Posted by gwierink

Hi Joachim,

Could it be that your time step is a bit large? Yours is 0.04 and in the twoStream example in this thread it is 2e-4 ...

Thank you for your answer.

Shouldn't the absolute value of the time step depend on the geometry/mesh resolution? I also thought that the pimple solver should provide the possibility to use larger time step sizes. The stability should be assured by the relaxation of the (temporary) results?!?

[gwierink]

True . Can you have a go with dynamic time stepping (and maxCo of 1 or so)? Maybe that will give you a sensible estimate for the time step ...

[luchen2408]

hello,Jherb, have you solve the problem? I have performed a simulation with twoliquidMixingFoam,but the simulation divergen, it seems the pressure equation problem. have you solve the problem? how? Thanks