[jeff.freeman]

I just attempted to run a case using adjustTimeStep = "on" with maxCo = 0.3. The alarming thing about this case is that it recommended a time step of roughly 5e-10 even though I was able to manually set the time step to 5e-7 and get Courant numbers of roughly 0.5. Has anybody experienced this sort of behavior before?


More detail of the situation:
I am attempting a grid refinement study of a rudimentary converging-diverging nozzle using laminar viscosity and rhoPimpleFoam as the solver. I am using two blocks, one for converging and the other for diverging. The case is 2D planar and I am only modeling one half of the nozzle. All of the cases prior to the problem case had adjustTimeStep = "on" and didn't seem to be misbehaving.

The only changes I made when progressing to the problem case was doubling the number of cells in both the x and y directions (z is empty) and changing the grading from simpleGrading(1 1 1) to simpleGrading(1 0.25 1) to give a tighter mesh near the wall.

The solver claimed that the Courant number was around 0.3 and it needed a time step of 5e-10 to achieve this. After 1.5 hours of getting nowhere, I understood this to be inappropriate. I then turned off adjustTimeStep and hard-coded the time step to be 5e-7 to achieve a Courant number of ~0.5.

Best regards,
Jeff