Question on OpenFOAM and implicit time discretization
 

Hello folks,

OpenFOAM has quite a few implicit time schemes which are used pretty heavily. But what I don't understand is that often in practice and a lot of advice on this forum says you should still try to get a CFL of 1, even though the implicit scheme shouldn't require that.

I have found, though, that in many instances without this, the code blows up. For my higher-speed cases, this can happen with dt=1e-4s. With much smaller timesteps, 1e-5s or 1e-6, the code doesn't blow up. Does this represent some kind of instability in the scheme, or am I missing something?

Thanks.

The time marching is stable, the system coupling is not.