CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (http://www.cfd-online.com/Forums/ansys-meshing/)
-   -   [ANSYS Meshing] Abrupt contraction and y+ (http://www.cfd-online.com/Forums/ansys-meshing/105430-abrupt-contraction-y.html)

Kirjain July 31, 2012 01:55

Abrupt contraction and y+
 
1 Attachment(s)
Hi all,

I'm calculating a case of a straight pipe with a sudden contraction (see attachment) and find it hard to get the local y+ near the contraction to an acceptable level. The average value of y+ might be somewhere around 50 or so but at the contraction it is 1000. How could I change this? Any insights are warmly welcomed!

Cheers.

flotus1 July 31, 2012 02:51

1 Attachment(s)
If you don't want to use an O-grid you should first apply a bias to the mesh in the wide sections of the pipe towards the contraction. Just like you did it in the contraction
This is also necessary because the volume change between the last cell in the wide section and the first cell in the contraction is very high in your first attempt.

Kirjain August 1, 2012 04:11

Thanks for the assistance! I got the y+ value down to about 200 or 300 but the results still seem a bit off (I'm comparing the results from FLUENT with analytic ones). Might the value still be too high? Another problem I am facing now is that on densening the mesh the number of elements with bad orthogonal quality goes up (still there are only few bad cells in the mesh). Could this have an effect on the outcome as well?

flotus1 August 1, 2012 05:41

Yplus-values of 300 are definitely too high to achieve accurate results.
Reduce the thickness of the first cell (of course while keeping a wall-normal growth rate below 1.1)
With adaptive wall functions, the Yplus-values should be in the range of 30-100.

Cells with a bad quality introduce an unnecessary error source to this simulation. The geometry is simple enough to produce a high quality mesh regardless of the cell size.
Am I right assuming that your simulation is axisymmetric?
In this case, I recommend using an axisymmetric model. This makes meshing a lot easier and you can run simulations even on very fine meshes.

Wait a minute...
If you say you compare your results to analytical solutions, does this mean you flow is laminar?

Kirjain August 1, 2012 08:09

Thanks again, I'll try to refine the mesh further using your advice!

Oh and sorry for the bad wording, the results I have are not analytical in the exact sense of the word as the equations have been determined experimentally, I am dealing with turbulent flow all right.


All times are GMT -4. The time now is 08:22.