Numerical instability using "small" time step
Hi All,
I am experiencing a numerical instability (I think) when the time step goes small as 5E-5 or 1E-5 seconds. I am simulating a gas explosion with XiFoam and I would like to keep the Courant number smaller than 1 thus my time step is 5E-5 - 1E-5 seconds for a 50 - 25 millimeters mesh modeling a 4x4x3 meters chamber. I would like to highlight that a so small time step is common in combustion but seems something, that I am not able to debug, causes a numerical instability. Velocity and pressure drop and the combustion stops at a certain time when the time step goes smaller than around 8E-5 seconds; this is the numerical instability I am experiencing. Currently I am trying to avoid this issue allowing a Courant number up to 3 and keeping the time step large enough. Please let me know if you experienced something similar also using different solvers. Thanks, Pierluigi |
Hi Pier84,
I would recommend you to use adjustable time-steps, i.e. time step control by openfoam. You can tell the solver to adjust the time steps such that a predefined value of the courant number is not exceeded. You can check the Breaking of a Dam example to look how time-step control is configured in openfoam, http://cfd.direct/openfoam/user-guide/dambreak/ Good luck! Lesh |
Hi Lesh,
Thank you for your reply. I am already using adjustable time-steps. Sorry I did not specify it explicitly. Adjustable time-steps is the reason why my time step decreases unfortunately until a small value that causes the "numerical instability" I mentioned. Thanks, Pierluigi |
Hi Pier84,
sorry, I didnt know that you're already using adjustable time steps :-). In this case I'm sure that the solver is running into a singularity for some reason, and that also explains why it frequently tries to decrease the time steps. You should check your boundary conditions and initial conditions. Also verify whether your problem is well defined, such that the solver is able to find a (hopefully) reasonable physical solution. Best, Lesh |
Hi,
It is something in the solver that cannot be fixed right now. Increasing the time step or the mesh size solves the issue. Cheers, PO |
Hello to all,
I know that this post is old. However, I am currently using Xifoam to simulate hydrogen combustion, more likely based on the same experiment as Pier84, and I faced similar issues. With certain grids when using Courant number equal to 0.5 the simulation fails: pressure drops abruptly and it is as if the flame front stops moving. It would be really helpful if you could please let me know if you managed to solve the issue or if anyone else have faced similar problem. Thanks |
Time Step gradually decreasing to very low value
Quote:
I am too having similar problem with time step gradually decreasing to very low value , mean CFL number as well. Here, If i want to lookup numerical settings, what would you check if someone has managed to overcome deltaT issues ? |
unstable at small time-step (Courant Number)
For what it's worth, I have not run XiFoam, so I'm afraid I can't help you there, but I have encountered what sounds like a very similar problem https://www.cfd-online.com/Forums/showthread.php?t=172144. Have you tried running reactingFoam or some other combustion solver? This problem may be more systemic than just one or two solvers.
|
All times are GMT -4. The time now is 01:36. |