Learning Forum

[kattar]

I'm simulating a cough jet issuing into a room, the domain consists of a rectangular box with two walls and four pressure outlets. A circular surface near the top of one of the walls serves as the inlet, there is no gravity.

I ran this simulation on an LES model, with harmonic blending Rplus, Species with air as last ratio, energy enabled, Pressure and momentum urf at 0.6, PISO, PRESTO, momentum central differencing, transient formulation bounded second order implicit, inlet with synthetic turbulence generator with turbulence at 20%

However, I ran it three times on meshes near 2 million cells, and was able to get it to run without many issues to completion, but when I tried a finer mesh at 4 million cells, the continuity kept blowing up.
I have included the residual plot of the first few iterations of the 2 million cells (I wasnt able to save full case and data, HPC cluster disconnected), and the residual plot of the 4 million cell run.
I have no idea how to fix this.

 

1) 2 million cells: the solution converged normally after this, im pretty sure

 

2) 4 million cells: blows up, this was my second try, first try it ran for longer and stopped with a floating point error 

 

[Essence]

Hello,

How did you initiate the case? Did you run the RANS/URANS initially and then proceed to LES? How is the mesh quality (growth rate, orthogonal quality, sizing, y+ values)?