[cpadam]

I'm running a transient two-phase flow case with compressibleInterFoam. I have a slug of water in a pipe being accelerated by air at high pressure. The simulation blows up after about 200 iterations with a floating point exception. I am wondering if anyone has any insight into what may be crashing my simulation.

The output just prior to the crash is as follows:

Courant Number mean: 0.17185088 max: 0.50333978
Interface Courant Number mean: 0.00033905591 max: 0.32202209
deltaT = 4.1666667e-06
Time = 0.01817083333

PIMPLE: iteration 1
MULES: Solving for alpha.water
Phase-1 volume fraction = 0.040471844  Min(alpha.water) = -0.00010445997  Max(alpha.water) = 1.0000003
MULES: Solving for alpha.water
Phase-1 volume fraction = 0.04047171  Min(alpha.water) = -0.00010295604  Max(alpha.water) = 1.0000003
DILUPBiCGStab:  Solving for T, Initial residual = 9.7342668e-05, Final residual = 1.0175892e-09, No Iterations 1

And the error is:

MPT ERROR: Rank 94(g:94) received signal SIGFPE(8).
	Process ID: 201260, Host: r3i3n18, Program: /p/app/openfoam/OpenFOAM-v1906/platforms/linux64GccDPInt64Opt/bin/compressibleInterFoam
	MPT Version: HPE MPT 2.17  11/30/17 05:59:14

You can see that the min/max liquid volume fraction goes below 0.0 and above 1.0, which looks rather dubious to me.

I have also attached the volume fraction and p_rgh contours just prior to the crash. The p_rgh contours look like there's some numerical artifacts on the left edge of the liquid slug (where it is pushing against the air ahead of it).

I appreciate any suggestions on how to fix this simulation.

[cpadam]

The problem has been fixed by increasing the number of PIMPLE corrector steps (found in the system/fvSolution file). In my case, this was set to 2 correctors, which is probably insufficient to achieve convergence when large pressure gradients begin to form in the solution. Increasing the number of correctors to 10-20 has allowed the solution to run to completion.