Help using an understanding y+

siw · September 30, 2009, 10:46

Hi,

I'm running some simulations for the external aerodynamics of a streamlined body. I've made a mesh in CFX-Mesh and set the first layer height of the surface prisms by imputting a y+ = 30 and the Reynolds number based on the body length. I ran the simulation with the SST model. Now in CFX-Post I've plotted surface contours of Yplus (as CFX-Post terms it) and the data ranges from about 6 to 72. Is this data even comparing the CFX-Mesh inputted y+ with the target y+ values all over the body?

So my questions are: Is this contour plot showing me data that states I need to remesh with a different y+ value? I've looked at various texts about y+ and all they say is the equations with shear stress, skin friction coeffcient etc and not how to realistically put numbers into.

Thanks.

ghorrocks · September 30, 2009, 20:34

You will need to do a sensitivity analysis on your simulation to determine what to do. Refine the mesh and see if the drag (or whatever the parameter of interest is) changes. If the refined mesh gives different results then you have to think about further refining the mesh. If not then it is OK.

siw · October 1, 2009, 05:05

Thanks,

So plotting Yplus in CFX-Post is of no use and I should only alter the CFX-Mesh Stretch variables until solution independence occurs?

BTW, from your experience what is an acceptable aspect ratio for the prism elements? Having specified a first layer height via the y+ option surface spacing in the x direction would be based on the aspect ratio limits for an initial mesh to get a good start before refinement.

ghorrocks · October 1, 2009, 08:10

I did not say no use, I just said purely basing mesh sizing off y+ is not a good idea. The y+ figure is a guide to give you a starting point for general simulations. However if you are integrating to the wall (eg using turbulence transition model) then it becomes more important and you need to make sure the critical boundary layer regions are not exceeding 1 by too much - but still a sensitivity study is the way to make sure you are right.

Aspect ratio can go quite high in boundary layer elements, 100 or even 500 is OK providing the ends are perpendicular to the wall and the long edge parallel to the wall. Elements with aspect ratios this high in the internal flow regions will cause problems. Again, this is problem dependant and you should do a sensitivity study to determine the mesh you need both normal and tangential to the wall.

Hashmi · October 4, 2009, 11:23

Actually y+ is a non-dimensional distance, The distance from the wall measured in viscous lengths - or wall units -
It is denoted by
y+ = distance from the wall / viscous length scale

viscous length scale = kinematic viscosity/(sqrt(wall shear stress/density))

close to the wall viscosity and shear stress are important parameters, therefore they are present in the definition of viscous length scale.

Just like Reynolds number which gives relative importance of inertial and viscous forces in order to characterize the flow (laminar or turbulent).

the y+ determine the the relative importance of molecular and turbulent viscosity closed to the wall, e.g The viscous contribution drops from 100% at the wall (where y+=0) to 50% at y+=12, and is less than 10% at y+=50.

for detail see the book Turbulent flows by Pope

SMH Hashmi

September 30, 2009, 10:46	Help using an understanding y+	#1
siw Senior Member Stuart Join Date: Jul 2009 Location: Portsmouth, England Posts: 742 Rep Power: 26	Hi, I'm running some simulations for the external aerodynamics of a streamlined body. I've made a mesh in CFX-Mesh and set the first layer height of the surface prisms by imputting a y+ = 30 and the Reynolds number based on the body length. I ran the simulation with the SST model. Now in CFX-Post I've plotted surface contours of Yplus (as CFX-Post terms it) and the data ranges from about 6 to 72. Is this data even comparing the CFX-Mesh inputted y+ with the target y+ values all over the body? So my questions are: Is this contour plot showing me data that states I need to remesh with a different y+ value? I've looked at various texts about y+ and all they say is the equations with shear stress, skin friction coeffcient etc and not how to realistically put numbers into. Thanks.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Understanding interFoam	sega	OpenFOAM Running, Solving & CFD	26	February 10, 2022 04:21
Understanding of channelOodles and oodles solver	fs82	OpenFOAM	1	September 23, 2009 11:36
Need Help Understanding RMS residual calculation	TWaung	CFX	1	September 22, 2009 19:21
understanding dual-time stepping	srinath	Main CFD Forum	1	August 11, 2008 12:05
LUSGS understanding	Gustavo	Main CFD Forum	0	February 14, 2007 11:12

September 30, 2009, 20:34		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	You will need to do a sensitivity analysis on your simulation to determine what to do. Refine the mesh and see if the drag (or whatever the parameter of interest is) changes. If the refined mesh gives different results then you have to think about further refining the mesh. If not then it is OK.

October 1, 2009, 05:05		#3
siw Senior Member Stuart Join Date: Jul 2009 Location: Portsmouth, England Posts: 742 Rep Power: 26	Thanks, So plotting Yplus in CFX-Post is of no use and I should only alter the CFX-Mesh Stretch variables until solution independence occurs? BTW, from your experience what is an acceptable aspect ratio for the prism elements? Having specified a first layer height via the y+ option surface spacing in the x direction would be based on the aspect ratio limits for an initial mesh to get a good start before refinement.

October 1, 2009, 08:10		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,872 Rep Power: 144	I did not say no use, I just said purely basing mesh sizing off y+ is not a good idea. The y+ figure is a guide to give you a starting point for general simulations. However if you are integrating to the wall (eg using turbulence transition model) then it becomes more important and you need to make sure the critical boundary layer regions are not exceeding 1 by too much - but still a sensitivity study is the way to make sure you are right. Aspect ratio can go quite high in boundary layer elements, 100 or even 500 is OK providing the ends are perpendicular to the wall and the long edge parallel to the wall. Elements with aspect ratios this high in the internal flow regions will cause problems. Again, this is problem dependant and you should do a sensitivity study to determine the mesh you need both normal and tangential to the wall.

October 4, 2009, 11:23		#5
Hashmi New Member Syed Mushahid Hashmi Join Date: Aug 2009 Location: Hamburg, Germany Posts: 9 Rep Power: 17	Actually y+ is a non-dimensional distance, The distance from the wall measured in viscous lengths - or wall units - It is denoted by y+ = distance from the wall / viscous length scale viscous length scale = kinematic viscosity/(sqrt(wall shear stress/density)) close to the wall viscosity and shear stress are important parameters, therefore they are present in the definition of viscous length scale. Just like Reynolds number which gives relative importance of inertial and viscous forces in order to characterize the flow (laminar or turbulent). the y+ determine the the relative importance of molecular and turbulent viscosity closed to the wall, e.g The viscous contribution drops from 100% at the wall (where y+=0) to 50% at y+=12, and is less than 10% at y+=50. for detail see the book Turbulent flows by Pope SMH Hashmi