Boundedness of alpha1 in twoLiquidMixingFoam and interMixingFoam
Hello all,
I tried using twoLiquidMixingFoam and interMixingFoam for simple 3d cases with inlet and outlet boundaries and I notice that the volume fraction (alpha1 and alpha2) becomes unbounded. I have the courant number as low as 0.25. The sample output from such a case: % diagonal: Solving for alpha1, Initial residual = 0, Final residual = 0, No Iterations 0 smoothSolver: Solving for alpha2, Initial residual = 0.00256416, Final residual = 1.96744e17, No Iterations 100 Air phase volume fraction = 0.505585 Min(alpha1) = 0.0177925 Max(alpha1) = 1 Liquid phase volume fraction = 0.00634079 Min(alpha2) = 0 Max(alpha2) = 1 diagonal: Solving for alpha1, Initial residual = 0, Final residual = 0, No Iterations 0 smoothSolver: Solving for alpha2, Initial residual = 0.00253923, Final residual = 3.84262e17, No Iterations 100 Air phase volume fraction = 0.505553 Min(alpha1) = 0.0174753 Max(alpha1) = 1 Liquid phase volume fraction = 0.00640491 Min(alpha2) = 0 Max(alpha2) = 1 % I am wondering if someone has similar experience with these solvers? I would really appreciate suggestions to improve the boundedness of the volume fraction. Thanks in advance, Vinay 
Vinay,
It looks like the max number of iterations is reached (100), which may indicate the solution never converged for your phase fields. You could try a different interface compression scheme to possibly improve stability. Another possibility is to try tweaking your phase related options in fvSolution. It might also help to have your test case so other people can reproduce your results. Kevin 
Thanks for the reply Kevin. In the case I have shown, I had used a very small tolerance to see if this helps, therefore the maximum number of iterations. I will try to post the case file as soon as possible.
Right now i am trying the same case with a finer mesh and it seems to be going in the right direction. What do you mean by tweaking phase related options in fvSolution? Best regards, Vinay 
In my experience, the default linear system solver options are usually sufficient. In fvSolution there is a section for PIMPLE solver settings (see below), the phase specific settings are in bold. The other settings affect general solver stability (velocity, pressure coupling). I'd probably start by increasing the number of alpha correctors and sub cycles.
Code:
PIMPLE 
Quote:

All times are GMT 4. The time now is 18:27. 