[gtrt]

Hi, I am attempting to complete a transient 2 phase VoF simulation with evaporation-condensation (Lee Model) in 2D.

Model Properties:
The flow is internal in a closed volume, I am assuming a variable saturation temperature with pressure and ideal gas law for the vapor density. The flow is assumed to be turbulent and we are using the k-omega SST model. The P-V coupling in SIMPLE, the gradient is LSCB, pressure is PRESTO!, volume fraction is geo-reconstruct, and all other formulations are second order. We are running VoF explicit in order to get geo-reconstruct, the Fluent transient formulation is first order implicit.

Mesh Properties:
The mesh is 86k cells, the y+ is < 10. The mesh is also uniform and rectangular.

Transient Properties:
We are using adaptive time stepping with Courant number < .5. This puts our time step somewhere between 1e-6 and 1e-5s.

Computation Properties:
Running on 1 node with 24 cores - 187.397 GB RAM. This is on a cluster but we are essentially running as if it is a single machine.

Question:
I am noticing that the time to complete this simulation is projected to take almost 4-7 days for 1 second of flow time. I have tried under relaxation factors to get better convergence with higher time steps, but this will just delay overall steady state. I am wondering if there is anything I am missing that will help speed up this simulation? I have tried NITA but it makes the convergence very unstable, continuity does not converge with this method. Please let me know if anyone has any ideas or experience with this kind of set up. Thank you.