[Clémence]

Hello everyone,

I have a serious question for my simulation, and a few time left to make it.

I am making a simulation with the implicit unsteady solver. For that solver, I have been told that the CFL condition were not needed. To find the time step, the following condition is needed:
dt < 0.5 * (mesh size)²/viscosity
So I could find that the time step has to be lower than 54.94. For now it is ok, I took 0.01s for the time step, and I could increase it during the simulation.

But I don't know what to choose for the current number. Do I have to follow the CFL condition which say that the CFL is lower than 10??? But in this case the time step cannot be determined the way I did...

Thank you for your help!!!

[SB123]

Originally Posted by Clémence

Hello everyone,

I have a serious question for my simulation, and a few time left to make it.

I am making a simulation with the implicit unsteady solver. For that solver, I have been told that the CFL condition were not needed. To find the time step, the following condition is needed:
dt < 0.5 * (mesh size)²/viscosity
So I could find that the time step has to be lower than 54.94. For now it is ok, I took 0.01s for the time step, and I could increase it during the simulation.

But I don't know what to choose for the current number. Do I have to follow the CFL condition which say that the CFL is lower than 10??? But in this case the time step cannot be determined the way I did...

Thank you for your help!!!

clemence i'm kinda looking for a similar answer, but your question is not clear. where did you get that relationship for your dt?
secondly if by current number you mean courant number, then that's what i'm looking for as an answer too, what should the courant number be when working in implicit unsteady RANS eqns?

i also have heard its not specifcally an issue like it is with explicit. I imagine its to be used more as an estimate as opposed to 'be all truth' as with explicit. .... any thoughts would be appreciated

[tiatabasco]

I think one should try with different time steps for the same mesh and come to a conclusion. But I am also wondering based on what I can conclude. My guess wud be check the courant number for each case and take the one which has courant greater than one but not very large say under 50. Anybody got an update on this one please let me know.