CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Y plus for various turbulence models (http://www.cfd-online.com/Forums/cfx/107055-y-plus-various-turbulence-models.html)

taram September 16, 2012 13:58

Y plus for various turbulence models
 
Dear all,
I am working on the study of wall treatment for different turbulence models (fluid domain as cyclone separator). This simulation is carrying out flow simulation alone for better understanding. In the above mentioned work, I need to know the effective range of y plus for various turbulence models such as k-epsilon, k-omega, SST and RSM etc. I have searched but I could not get material which gives elaborated details. So, I wish to know the range of y plus has to be used or better if you give me link to come across the same.

Thanks in advance.

ghorrocks September 16, 2012 18:44

SST can handle any y+.
k-e should be >11
k-omega should be <1
RSM should be >11 if using wall functions or anything if automatic.

But the final decision on wall mesh resolution should always be determined by a sensitivity study in your case.

taram September 17, 2012 06:28

Hi Glenn,
Thank you for valuable details. I will check on each turbulence models and do for sensitivity study.

hamed.majeed September 18, 2012 02:48

Hi,
Does these limits change:
1. If flow is internal or external.
2. Using wall functions

Thank you

Regards
Hamed

ghorrocks September 18, 2012 19:41

1. No.
2. If you use wall functions then you need y+>11.

This is all basic application of the wall boundary condition theory. If you read up about the wall boundary condition and understand it then the reason for these answers become obvious.

cfdmms December 10, 2013 02:37

Quote:

Originally Posted by ghorrocks (Post 382046)
SST can handle any y+.
k-e should be >11
k-omega should be <1
RSM should be >11 if using wall functions or anything if automatic.

But the final decision on wall mesh resolution should always be determined by a sensitivity study in your case.

If you kindly explain why these values of yplus is vary over different turbulence models? and what about LES?

Thanks.

Anna Tian December 10, 2013 04:13

Quote:

Originally Posted by ghorrocks (Post 382046)
SST can handle any y+.
k-e should be >11
k-omega should be <1
RSM should be >11 if using wall functions or anything if automatic.

But the final decision on wall mesh resolution should always be determined by a sensitivity study in your case.

Are you sure that k-e should be >11? Does CFX have low reynolds number K-e model like Fluent which likes to have Y+ < 0.2? I thought these two software are almost same at wall treatment for different turbulence modeling.

Antanas December 16, 2013 01:11

Quote:

Originally Posted by Anna Tian (Post 465637)
Are you sure that k-e should be >11? Does CFX have low reynolds number K-e model like Fluent which likes to have Y+ < 0.2? I thought these two software are almost same at wall treatment for different turbulence modeling.

As far as I understand CFX uses scalable wall functions with k-e model. If mesh is too fine near the wall then CFX ignores this region by "mathematically" placing first mesh node at Y+ ~ 11.

oj.bulmer December 16, 2013 12:44

Quote:

Are you sure that k-e should be >11? Does CFX have low reynolds number K-e model like Fluent which likes to have Y+ < 0.2? I thought these two software are almost same at wall treatment for different turbulence modeling.
In original k-e model, the definition of turbulent lengthscales and eddy viscosity has eps at the denominator. Due to rapid changes to k and eps as we approach wall, and hence in the values of lengthscales and eddy viscosity within the viscous layer, there are stability issues (along with divide by zero etc) hence wall functions are not optional.

But the low Re k-eps models in FLUENT (Launder-Sharma, Yang-Shih etc) all use damping functions and redefined values of eps to regularise the extremities in k and eps near wall and hence integration to the wall is possible with Y+ values less than 1 etc.

There are w based models both in CFX and FLUENT which are which are generally favoured.


All times are GMT -4. The time now is 14:59.