CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   FLUENT (https://www.cfd-online.com/Forums/fluent/)
-   -   Non-reflective Boundary Condition divergence (https://www.cfd-online.com/Forums/fluent/110680-non-reflective-boundary-condition-divergence.html)

 frodooon December 18, 2012 05:02

Non-reflective Boundary Condition divergence

Hi all,

I'm currently trying to run a 2D axisymmetric calculation of a simplified engine exhaust on Fluent 14.5.
I have thus one velocity inlet at 200m/s, one pressure outlet (gauge pressure = 0 Pa), and one pressure inlet (also 0 Pa).
I ran it steady kwsst ==> ok
Then SAS ==> OK
Then I'm trying to switch to 2d LES with the TUI command, which works fine, and doesn't diverge (but shows a very more chaotic comportment compared to SAS).
Now, all of these runs were done with classic pressure BC. I now want to switch ON the non-reflective BC on the pressure outlet, still with the 2d LES. It diverges each time I try.

My solver settings are the default ones when switching to that 2d LES.
I tried dividing the time step by 5 (going to 1e-04s), reducing the courant number, going into more complex discretization schemes : it diverges always, and from what I see on the contour plots : it comes from the outlet.

the mesh is like I said 2D, perfectly squared cells (10cm) everywhere, overall length of the model is about 40m and "radius" of the engine is about 2m.

The starting point of the LES calculation is the SAS calculation, which seems stable and nice.

I now switched the inlet BC to a "mass-flow" BC instead of velocity, but it doesn't look promising...

I don't have any more ideas to try make it work, so can someone help ?

thank you,

Arnaud.

PS : It is the pressure based solver, coupled.

 flotus1 December 18, 2012 05:50

I hardly think that running an axisymmetric case with LES is physically valid.
So dont bother if it doesnt run. Switch to a 3D geometry if you want to run a LES.

 frodooon December 18, 2012 05:57

Yeah, you're right, but I'm not trying to get information really from this calculation, just trying to understand a little bit more the different models, so I'd like to run it 2d for starter. But of course I plan on going 3D after that.

 flotus1 December 18, 2012 12:45

In that case, a planar 2d simulation instead of an axisymmetric one might work better. I am not sure if the non-reflecting boundaries are suitable for axisymmetric cases.

 All times are GMT -4. The time now is 02:21.