Hallo all

What happens if y+ is too large or too small when using the wall function in Star-CD....

Too large the gradient of the velocity profile is too small => too small a wall shear stress. Too small the velocity gradient is too large => too large wall shear stress.

Is this correct or have I misunderstood the meaning of the wall function?

Thanks

Bo

Hi, Assuming that the y+ values are not within the recommended range for wall functions, I interpret things the same way you do. For example, too small cells near a surface - with too small y+ values as a consequence (say, about 5 to 10) makes an overprediction of wall shear stresses and also heat transfer using the wall-shear based theory of Jayatilleka (as implemented in Star-CD). On the other hand, if the y+ values are still within the recommended range (say, 20 to 100) I don't know how much wall friction and heat transfer will spread. Will they spread at all, or do the wall functions take adequate care of this? I never made a study on this, anyone who did?

Hallo Anders

Yes I have doen a small simple study of this of flow in a pipe. I have seen that the wall shear stress is more or less constant until y+ around 10 (from 100 and down). Below 10 a sudden increase is seen. I have not tried to go the other way - that is using y+ values higher than 100.

Regards

Bo

Hi. What code did you use for this study?

Star-CD ver. 3.100a

I suppose for overly big y+ values the cells you can refine but what about overly small values? I can only think of refining them even more and then activating the two-layer models for that stretch of wall... is there a better way?

Hallo Tune

In star-cd you also have the oppurtunity to merge cells. However, this is only feasible if it is a relatively limited number of cells as you have to do manually. You need to type in which cells you would like to merge.

I normally do y+ optimisation like this.

I make a mesh that is relatively rough near the walls to ensure that y+ values are above at least 30 (this might take a couple of tries to get such a mesh.

Then I make a simulation. Then I collect all the cells with y+ above a certain limiet (60). Add their neighbors and refine this cell set perpendicular to the wall. Then a new simulation and so on untill y+ between 30 and 50 is achived in the entire geometry. More or less as there is always areas of very low velocity where the y+ are too low. But the alternative is to have VERY large cells in these regions....

Regards

Bo

IMHO very low y+ in the areas of very low velocities can be a source of problems - we can have very low velocities near separation/reattachment points where precise pediction of y+ is critical.

If you want to rely only on wall-functions then I suppose you can't do anything more to the very low y+ wall cells. If there's memory and CPU power to spare, and then why not just calculate directly the near-wall region behaviour? Two-layer turbulence models allow the calculation of most near wall flows as a classical 'law of the wall' function and you can directly calculate the near-wall behaviour at selected regions (areas previously having low y+) -still very economical compared to purely using the low-Re turb model. As Anton pointed out, having a low y+ doesn't mean it can be ignored!

When y+ goes over the roughly 300 to 500 limit, the more it just resembles frictionless wall, i.e., the wall boudary condition reduces to frictionless wall with the only function of not allowing flow trough the wall.

hey guys,, how important Y+ for CFD analysis anyway. It is just for indication whether our analysis is correct or what?

Hallo Campbell

The y+ values are important if you are interested in near-wall features. If your are just interested in the bulk flow then I imagin that the y+ values are less important. There are several publications showing the difference between the different near-wall treatments and indirectly the importance of the y+ value.

I have shown in a very simple simulation the importance of y+ value when using a standard wall-function. If you like I can mail that to you.

Regards

Bo

Hallo Jensen.. Thanks for your advice...Can u email mail the material that u just shortly brief to me. I really want to know about Y+. I always working with external flow analysis. In my analysis i gave less attention to Y+.

Thanks Jensen...