with kOmegaSST model, I am able to get a pretty robust simulation even when I have a high courant number, I was wondering if that will result in a correct answer. If yes, maybe I can get much faster results
OpenFoam uses implicit schemes, so in principle is possible to run at high Courant number. Though stability doesn't imply accuracy, that depends on the time step of simulation and the variation over time expected in your flow. With larger time steps you will miss for example some fluctuations or transient features.
OpenFoam uses implicit schemes, so in principle is possible to run at high Courant number. Though stability doesn't imply accuracy, that depends on the time step of simulation and the variation over time expected in your flow. With larger time steps you will miss for example some fluctuations or transient features.
Does that apply to all solvers? I thought that some of them use explicit schemes; for example, some solvers allow the dynamic adjustment of the time step according to the maximum courant number.
Dear Abdel,
Some numerical schemes are unconditionally stable and some are not. It is possible for the mathematical model to run with high courant numbers, although this can lead to a decrease in accuracy like Andrea suggested. The idea is to increase the time step (i.e. courant number) in order to shorten the computation time but keep the accuracy at an acceptable level. In my experience this can be achieved by calibration and verification of your model using some experimental data or analytical solution.
Best regards,
Zoltan
Because of the implicit discretization in OpenFOAM, the Co is no longer a critical measure of stability. Yet, it still serves as a measure of accuracy.
Roughly if Co>1, this tells you that your time step is very high (Co > 1) to the point that the fluid particles do not reside within a cell but jump and cross one or more cells over that (large) time step. This means a loss of accuracy, but the solution is still stable. This matter is meaningful if you are interested in a steady-state solution.
Correctly said. High Courant number may give stability but this necessarily does not mean that solution is accurate enough. The Courant number should be such that the physical and numerical domain coincide. Otherwise there is possibility that some transient flow features will be missed
I have a small doubt in this does Co should be strictly greater than 1 or we can have value somewhat like Max value 1.5 and volume average vale = 0.0137?
It is not that Co has to be always under 1. It is sensitive case by case
- Badges
- Science topic