CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Programming & Development (https://www.cfd-online.com/Forums/openfoam-programming-development/)
-   -   Wall modelling and boundary conditions: why specify value for turbuelnce quantities? (https://www.cfd-online.com/Forums/openfoam-programming-development/78633-wall-modelling-boundary-conditions-why-specify-value-turbuelnce-quantities.html)

 djbungee July 28, 2010 12:10

Wall modelling and boundary conditions: why specify value for turbulence quantities?

Dear all,

we had a discussion today in the CFD team in our department and I mentioned that it is usually not necessary to require a user to define boundary conditions for turbulence quantities on or in the vicinity of walls.

Usually, wall velocity and heat flux and/or temperature are sufficient.

Physical considerations or model equations require a certain value or asymptotic behavior of turbulence quantities such as turbulent kinetic energy, turbulent frequency or dissipation at or close to the wall. Therefore, you can either implement a fixed value on the wall (such as turbulent kinetic energy = 0) or set a solution dependent value in the cell center of a wall-bounded cell (which requires that you can identify these cells) and adjust the contributions of these faces or volumes to the matrix coefficients directly.

At least, that is how I remember we did it in our code at university.

In OpenFOAM you can or have to define wall values, and correctly setting k=0 does not even work.

Any comments? I posted this thread there also on http://www.extend-project.de in the 'Special Interest Group on Turbulence Modeling and Simulation Group Forum'.

Best regards, Ulf.

 chrizzl July 29, 2010 04:07

Dear Ulf,

with a closer look at the code of a lowReynolds-model like the “LienLeschzinerLowRe” it seems to be that there is a special explicit treatment of epsilon for wall adjacent volume-cells. As one can see in the “LienLeschzinerLowReSetWallDissipation.H” it depends (beyond other things) on k_, wall distance y_ and yStar_ (like y+). That is why I think a given patch type of epsilon doesn't seem to have any influence on the volume field for epsilon (in case of LowRe-k-eps-model).

Regard, Christian;)

 djbungee July 30, 2010 07:32

Wall modelling and boundary conditions: a choas?

Thus, as we see, it is model dependent whether setting k=0 on the wall works or gives divergence. And the value for epsilon in the LL k-epsilon model is overwritten for this model.

Therefore, it is implemented in an arbitrary way in OpenFOAM. The complete organization of turbulence models and the treatment of turbulence quantities in the code is not very well organized, I think.