wall y+ value in buffer layer. what to do?

ruzfactor · December 3, 2010, 19:07

I'm using K-epsilon model and I have converged thee solution using all three of the wall functions. But for all three cases the wall y+ falls in the buffer layer( 5<y+<30). It fluctuates between 15-25 for different wall functions. What can I do to improve the wall y+? please suggest.

Jonny6001 · December 3, 2010, 19:51

change the cell size.

ruzfactor · December 3, 2010, 21:10

@jonny6001: Is it possible to change the cell size in fluent? if so how? I didn't create the mesh. I'm using it because it's given to me for my project. Also don't know how create a mesh.

Jonny6001 · December 4, 2010, 06:38

Y+ is proportional to the normal distance of the cell centroid from the wall. To change the Y+ you would need to reduce the cell length perpendicular to the wall.
I have never needed to do this in Fluent so I am not sure if it's possible. There is a nice built in function in Fluent that adapts the mesh based on user inputted criterion but I have never user it for boundary layer meshes. If I can remember, it just splits the cell in half or even a quarter for refinement. Now it might be possible for you to use this if the mesh adaptor splits your boundary cells like you need it to to reduce your Y+.

It is certainly not how I would do it but it might work.

Good luck.

ruzfactor · December 4, 2010, 09:25

I'm posting a y+ curve for you to see whether its a wrong curve or not. There is a mesh refine option in FLUENT but the problem is there are plenty of options to be adapted. which one to choose?

dst · December 8, 2010, 10:47

In general, you can try to live with y+ 5..30
Depending on the simulation, k-e Realizable Scalable wall function can give a good results.

in the command line:
define-mod*-visc*-near*-sc*
(without starts - you just do not need to finish the command... you know)

I have compared this model with some others in a certain range of y+ wider than 5..30 with some empirical solutions (known as correct). And it gave results much closer to theoretical solution in comparison to k-e Realizable Enhanced w.t. and many others two-equation based models.

But... This was one case. I can't be sure about different...

December 3, 2010, 19:07	wall y+ value in buffer layer. what to do?	#1
ruzfactor New Member RUZ Join Date: Oct 2010 Posts: 25 Rep Power: 16	I'm using K-epsilon model and I have converged thee solution using all three of the wall functions. But for all three cases the wall y+ falls in the buffer layer( 5<y+<30). It fluctuates between 15-25 for different wall functions. What can I do to improve the wall y+? please suggest.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How to make boundary layer mesh moving with wall	wayne	FLUENT	3	June 12, 2008 00:23
Question about bcdefw.f for wall temperature bc.	Jimmy	Siemens	10	March 18, 2008 16:28
wall function	Alex	CFX	9	May 19, 2005 11:05
When I use the wall functions....!	maximus	Main CFD Forum	7	January 20, 2003 10:35
atmospheric roughness height	Alain BASTIDE	Siemens	5	December 30, 2002 06:26

December 3, 2010, 19:51		#2
Jonny6001 Member Jonny Join Date: Aug 2009 Posts: 72 Rep Power: 17	change the cell size.

December 3, 2010, 21:10		#3
ruzfactor New Member RUZ Join Date: Oct 2010 Posts: 25 Rep Power: 16	@jonny6001: Is it possible to change the cell size in fluent? if so how? I didn't create the mesh. I'm using it because it's given to me for my project. Also don't know how create a mesh.

December 4, 2010, 06:38		#4
Jonny6001 Member Jonny Join Date: Aug 2009 Posts: 72 Rep Power: 17	Y+ is proportional to the normal distance of the cell centroid from the wall. To change the Y+ you would need to reduce the cell length perpendicular to the wall. I have never needed to do this in Fluent so I am not sure if it's possible. There is a nice built in function in Fluent that adapts the mesh based on user inputted criterion but I have never user it for boundary layer meshes. If I can remember, it just splits the cell in half or even a quarter for refinement. Now it might be possible for you to use this if the mesh adaptor splits your boundary cells like you need it to to reduce your Y+. It is certainly not how I would do it but it might work. Good luck.

December 8, 2010, 10:47		#6
dst New Member Join Date: Oct 2010 Location: Germany Posts: 25 Rep Power: 16	In general, you can try to live with y+ 5..30 Depending on the simulation, k-e Realizable Scalable wall function can give a good results. in the command line: define-mod-visc-near-sc (without starts - you just do not need to finish the command... you know) I have compared this model with some others in a certain range of y+ wider than 5..30 with some empirical solutions (known as correct). And it gave results much closer to theoretical solution in comparison to k-e Realizable Enhanced w.t. and many others two-equation based models. But... This was one case. I can't be sure about different...