CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   Interfoam (OF 1.7) : pressure evolution, impact, 2D computation (http://www.cfd-online.com/Forums/openfoam/83155-interfoam-1-7-pressure-evolution-impact-2d-computation.html)

kassiotis December 16, 2010 06:07

Interfoam (OF 1.7) : pressure evolution, impact, 2D computation
 
5 Attachment(s)
Dear All,

I'm currently trying to perform a 2D computation for validation purpose with interFoam. The problem is quite simple : it is a modified dambreak problem with a wedge obstacle.

I'm using different meshes with around 7e3, 3e4, 1e6 and 8e6 cells. The two fluids are considered to be laminar, and the high density flow has a large viscosity (1e-2). I'm giving you the evolution of :

$\int_S p dS$

computed with libforces at the right wall (force-r.png) of the bassin and the left wall of the wedge (force-o.png). As you can see, before the impact on the right wall, the results are in accordance. The problem is at the impact ( t~0.6s ) : the pressure pic does not seems to converge. I'm even more worried by the fact that the impact on the right wall modify the pressure evolution on the left of the wedge.

This pressure problem does not seem to influence too much the following of the computation for coarse meshes, but leads to divergence for fine meshes. Moreover, I never had this problem with 3D computation.

My guess : the air captured under the tongue (t=0.4 to t = 0.7s for instance) can't escape in 2D, and has a two large pressure (see pressure at t=0.6, given in attachement). The same problem occurs with the dambreak example given as a tutorial in OpenFOAM

What do you think about it ? Do you now a way to avoid this pressure pic ?

I give you my sources as well in attachement (without the mesh that is too heavy).

kassiotis December 21, 2010 10:55

I tried :

- change the boundary condition at the bottom right
- reduce the maxCo number (to 0.05)
- change the pressure solvers (pcorr, p_rgh, p_rghFinal) and add relaxation (but this does not seem to be a good idea, as I can't find p_rgh.relax() anywhere in the interFoam directory with a short grep).

Nothing give for the moment satisying results. Does someone have the same problem ?

alberto December 21, 2010 16:09

Hi, sorry if I ask. You partially answered my question, but I have a doubt from your fvSolution.

Did you try to use a zero relTol for p_rgh, so that the actual tolerance is achieved?

If you want a correct transient solution, under-relaxing is a big no with PISO algorithm, since interFoam does not perform outer iterations ;-)

Best,


All times are GMT -4. The time now is 18:31.