code_saturne User's Forum

Hello Yvan,

Thank you for your feedback. Referring to the practical user’s guide I chose the two-scale model type of wall functions for the k-omega-SST model. I expected a logarithmic respectively linear modelling of the velocity field depending from the distance to the wall and the velocity to be nearly zero for y=0. Is this approach not appropriate?
I am a little bit confused by the values for y+ and if they are consistent. I am using ParaView 5.2.0 for post processing and without choosing any filters the value of y+ is equal to zero everywhere but on the boundary face of the cells at the wall. Here the value varies between 40 and 80 which led me to the assumption that the quality of the mesh should be okay.
I appreciate your help and thank you very much for your time.


Best regards,
Chris

Hello Chris,

I've launched your calculation with the setup you sent.
I found that y+ is between 20 and 40, that means you're using the wall model (wall function) in the wall cell. To have a non-modeled no-slip condition at the wall, you would need to refine more as you know. With the k-Omega SST, you could try to reach a y+ value around 1-2.

But anyway, y+ between 20 and 40 is OK, k-Omega SST being a "High-Low" Reynolds model.

On the boundary faces which are not walls in your case, yplus being based on the friction velocity
, which is zero at symmetries and inlets, it is expected that the post-processed value is zero.

Regards,
Erwan.