CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLOW-3D (http://www.cfd-online.com/Forums/flow-3d/)
-   -   Shear Stress (http://www.cfd-online.com/Forums/flow-3d/124683-shear-stress.html)

Marcus.filho October 10, 2013 16:04

Shear Stress
 
1 Attachment(s)
Dear all,

Ive beeing try to evaluate the shear stress of a channel but ive found some issues. Ive attached a picture of the results to help you guys to understand.

the problem is that it just doesnt look right. The changes are very sudden and the values seem off chart. Does any have one have an idea of what might be happening?

My first thought is the my mesh is too course. Is there a "rule of thumb" of how dense should my mesh be in order to correctly estimate the shear stress? The mesh is around 0,5m in a 10m wide channel.

Thank you all.

JBurnham October 17, 2013 10:57

Yes, Marcus, the shear stress output is mesh-size-dependent. Try halving your mesh size until the shear stress output doesn't vary by more than 3% between runs: you can call that the 'converged' mesh size. Also, please consider upgrading to v10.1 if you haven't already: the shear stress calculations are much better in v10.0 and v10.1 than in earlier versions. - Jeff

Marcus.filho October 22, 2013 08:03

Thank you Jeff!

One more question. Is there a "rule of thumb" of how many mesh cells must I have in a certain depth in order to have some first acceptable results?

Marcus

JBurnham October 22, 2013 10:39

Yes. The rules of thumb are:

1) at least 3 to 4 cells across the narrowest flow width (usually a depth, but might be a gate) in order to get a model that works.

2) at least 10 to 12 cells across the narrowest flow width to get a model that's accurate.

The best approach is still to start with rule #1 above and successively refine the mesh two or three more times (total of three or four meshes) and demonstrate that the difference between the finest mesh results and the next-closest mesh results are so small that the model can be considered 'converged'.

I know that's not always possible, so the rules of thumb above can be a shortcut.


All times are GMT -4. The time now is 18:28.