CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Global Courant number diverging (http://www.cfd-online.com/Forums/fluent/91697-global-courant-number-diverging.html)

The_Architect August 19, 2011 09:01

Global Courant number diverging
 
Hi,
I have a 2D problem with a moving wall as one of the BCs. The problem is that the global Courant number diverges after a few hundred iterations. I have used time steps as low as 1e-05 s and tried the variable time stepping method as well as the Implicit update method for the dynamic mesh at the BC. The results are however the same and I can't get away from the Courant number problem.
I think this is really odd because the wall is moving at a really slow pace, say cm/s. My grid has an aspect ratio of around 2 near the moving wall. The cell height normal to the moving wall is 5mm. As it is right now I really can't make the time step any smaller, in fact it is already way too small. I can't really see a new more dense mesh as an option either because that would make the problem computationally too heavy .

All help would be highly appreciated!

_Denis_ August 22, 2011 05:20

Hi, Patrik

Maybe it's just me, but it seems that your question is insufficienlty detailed. (almost as if "there is something wrong with my solution. What is it?").
But, as I know what tutorial you've been following lately, I'd suggest to uncheck the "uniterative solution" box, and experiment with the number of iterations, starting with, say, 3.
It seems, you've got a realy dens mesh, and little timestep. Do you realy need it? Try to generate a somewhat coarser mesh, and specify bigger timestep (mine for the same problem is 1e-2 for now), at least until you get to the final solution.
Hope this helps.

The_Architect August 23, 2011 05:00

Thanks again for your reply Denis.
I have now been running multiple simulations with different meshes, but the end result is always the same, the courant number literally explodes (during 2 time steps it might go from 2 to 10^6, and the continuity residual might in the same time jump from 10^-2 to 10^10!). The last mesh I tried was basically the same as in the Fluent tutorial for wave generation, I just made the domain larger to better fit a bigger wave.
I have unsuccessfully tried to get away from this problem by trying the following things:
- Non-iterative time advancement or normal time advancement
- Implicit update
- Variable and fixed time steps
- Time step size from 10^-2 to 10^-4
- Solve volume fraction equations at every iteration or once per time step
- First and second order moment discretization
- Smaller URF

My BCs are, velocity inlet with an UDF to generate the wave, walls on the opposite side of the velocity inlet and on the bottom ,and pressure outlet above the basin.

I guess there is something fundamentally wrong with my problem set up as the different settings all give the same result of an unsuccessful simulation.

niravtm007 August 29, 2011 02:22

velocity nlet udf
 
hiii friends i am entirely new to using udf's in fluent. I have a cubical geometry, with two inlets on perpendicular adjacent faces. i have to implement udf to only one of the conditions.
i have two cases
1) parabolic velocity inlet ( space varient)
2) sinusodial velocity velocity inlet ( time variant)
can any one pls send me the udf ( codes) ...


All times are GMT -4. The time now is 03:01.