Dear all,

When wall function is used in turbulence models, it is suggested y+ is about 30. This point is easy to understand. But in "10.9.1 Near-Wall Mesh Guidelines for Wall Functions", the following statement is difficult for me to understand *It is important to have at least a few cells inside the boundary layer.*

The height of boundary layer is always not known beforehand for many complex turbulence flows. So I am quite confused about this. Any response is warmly welcomed.

Thanks a lot!

sarah

You may need to modify your mesh after performing initial runs. Assume a near-wall mesh density, and perform a solution. Check the results to see if you have >1 cell in the boundary layer. If no, modify your mesh and try again.

Try using the "Viscous Grid Spacing Calculator" found on the the CFD Resources Online page under Online Calculators. This should give you a rough idea of the grid spacing required.

To apply the wall function properly you have to use a mesh spacing which follows the known rules (i.e. 200 > y+ > 20). But the second criterion you mentioned has also to be fulfilled. You have to resolve the boundary layer with some nodes. 10 is a good estimate here.

For many flows it is impossible to fulfill both criteria because the boundary layer is too thin. The complete boundary layer or most of it may lay within the first, wall adjacent cell assuming that this cell fulfills the y+ criterion.

In such a case you should switsch to a low-Re turbulence model which does not use a wall function approach. The mesh has to be much finer. Depending on the turbulence model a y+ between 0.1 and 1 will be required.

Dear,

Excellent response!

The quesion is "How to determine the height of boundary layer"? As we know, for lots of complex geometry turbulence flows, there are no free flow like boundary layer flow.

Any response?

hehe

"The complete boundary layer or most of it may lay within the first' How could we know that? I.e. how to determine the bounday layer height even we finish the calculation and get a rough solution? thanks

sheng

A good guess would be just to look at the velocity profile near the wall. In the regions of large gradients you should have placed an appropriate number of nodes. This of course only possible after you have made a first simulation.

Dear all,

thanks a lot! I have learned a lot!

regards, sarah

You can also adapt the boundary, if needed. But becareful on your cell counts when doing so. For 3D cells, 1 cell can turn into 8 and 8 can turn into 64, and 64 for can turn into 512...

Dear Sir Hi, I want select y+ for k-epselon model in cross over around cylinder, please guide me. Best Regard Arash