[scleakey]

OpenFOAM uses implicit methods so I thought I would see how far I could push the Courant number before it crashes - isn't the point of implicit methods to get large Courant numbers?

I'm running a multiphase case with interIsoFoam on six processors. Keeping everything else the same, I tried different values for maxCo...

1.0
2.0
2.3
2.5 crashed at t = 0.02 after high residual for p_rgh
2.7 crashed at t = 0.02 after high residual for p_rgh
3.0 crashed at t = 0.02 after high residual for p_rgh
5.0
10.0
20.0 crashed at t = 1.31, not sure exactly why

I don't understand how it ran for maxCo = 10 (with very sensible results) but crashed for maxCo = 2.5? I realise there are lots of reasons that it might crash but I thought that decreasing the Courant number would make it more stable, not less

Has anybody else experienced this?

[Bloerb]

Implicit methods while allowing you to use Co>1 have a smoothing effect on the solution. So high pitched oscillations with a frequency higher than your time step will be smoothed out. This usually means a loss in accuracy. Imagine a sine wave where you only calculate the values between peaks. It is not unreasonable that the same effect applies to errors of said frequency. So depending on your problem this can increase stability in a certain range.

[scleakey]

Originally Posted by Bloerb

Implicit methods while allowing you to use Co>1 have a smoothing effect on the solution. So high pitched oscillations with a frequency higher than your time step will be smoothed out. This usually means a loss in accuracy. Imagine a sine wave where you only calculate the values between peaks. It is not unreasonable that the same effect applies to errors of said frequency. So depending on your problem this can increase stability in a certain range.

Thanks for your reply I think I understand what you mean