Using High y+ Treatment with a Well Refined Mesh
 

Hi,

I know that high y+ models are devised for y+>30 but am just wondering what if I use a high y+ treatment model like the St K-epsilon while the value of y+ is very low (less than 1.0)? Is this numerically sound?

Thanks,
A E

Hi
This will lead to wrong results because of the concept underlying the wall function used in high Re turbulence models. Acording to this strategy the thickness of first cell adjacent to wall is selected such that its y+ goes above 30 (depending on the turbulence model). From cell center to wall an analytic profile is used to describe the velocity gradients. So if you decrease the mesh height next to wall not only the resolution of your model does not increase but also the velocity gradients will be wrong. Because from the second cell the analytic calculation ia abandoned and everything is calculated numerically while in gour case some other cells lies in y+<30 in which the bigh Re turbulence models are not still valid.
If you can have a dense meshing at near walls you can use low Re models like sst.

Thank you very much for your reply. So, what I have understood is that using a high y+ treatment for y+<1 is physically incorrect.

Surprisingly, the k epsilon and Reynolds Stress models have the same characteristics regarding high and low y+ treatments.

Best,
A E

Hi,
The main difference between K-Epsilon and Re Stress Models is that the later does not use boussinesq approximation and also takes into account anisotropy of Re stresses at near wall regions.

And apparently they have the same damping characteristics when y+ values are small.