CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   ANSYS (https://www.cfd-online.com/Forums/ansys/)
-   -   Wall functions Ansys Fluent (https://www.cfd-online.com/Forums/ansys/109944-wall-functions-ansys-fluent.html)

rm123 November 30, 2012 14:00
Wall functions Ansys Fluent
 
Hi,
Can someone please tell me how to decide on which wall function should i use for k-epsilon equation? I know it is something related to y+ and y* but I will appreciate a bit of detail.

Many thanks.

Far December 2, 2012 00:16
There are two type of wall functions available for the high Reynolds number K-epsilon model:

1. Standard wall Functions (Will give problems if mesh is refined below Y+ 30)

2. Scalable wall function. Applicable for the meshes with even Y+ 1. But still it is wall function. The only purpose of scalable wall function is to avoid problems of successive refinements in standard wall function meshes. The algorithm ensures that your solver Y+ (Y* in Fluent) is always greater than 11.225. Y+= 11.225 is the intersection of linear and log law profile.
It has function which chooses the maximum of (11.225, Y*).

You already know that the Y* is different than the Y+ somehow. I don't remember when they are different and when they are same, having to do with buffer and log layer (Hint: turbulent and viscous stresses). Check the related material on the Y* and Y+. But any how in background fluent uses the Y*. So it will choose the 11.225 if Y* is less than 11.225 and uses Y* if it is greater than 11.225.

For low Reynolds number K-epsilon model there are two methods available (Select option enhanced wall treatment).

1. If Y+ is less than 1, it uses the pure two layer model and you solve the viscous sub layer. This formulation is good when you have separation or you want to resolve the near wall effects e.g. drag prediction, otherwise this will wast your resources and takes more solution time due to excessive stretching of mesh in the boundary layer.

2. If Y+ is variable along the wall (which is always present in practical industrial problems) then it use the enhanced wall treatment. It is the method to implement the enhanced wall function (hybrid wall function see this http://www.cfd-online.com/Forums/flu...-function.html and http://www.cfd-online.com/Forums/flu...treatment.html ) for the two layer k-epsilon model) For this method you should ensure Y+<10 for best usage. Otherwise for the higher Y+ values this will gradually use the log layer i.e. wall function approach.

rm123 December 4, 2012 19:45
Hi Far,

Thanks a lot for your help.
Also I have one question is how to calculate total pressure drop between inlet and outlet in ANSYS?

Many thanks...

Far December 4, 2012 22:37
It is total pressure difference at inlet and outlet....

3ashoosh December 13, 2012 06:29
Particle velocity in solid-gas flow
 
Dear all,

I am running a 2D simulation for FCC riser in ANSYS Fluent 14 using Eulerian multiphase model for the solid-gas flow. I use the standard k-epsolon per phase turbulence model with the standard wall function. The geometry is so simple axisymmetric pipe. The inlet boundary conditions for both phases were coded in UDF .c file and interpreted in the fluent setting up.

When running the case, it converged after 100's of iterations but the outflow particle velocity profile is not parabolic (not fully developed). It decreases slowly from the centre then fall sharply near the wall. Some times has another peak (maxima) near the wall as if it is an annular core flow.
I used no slip conditions for both phases at the wall, also tried different specularity coefficients for the solid phase, but I got same results.

When plotting the wall Y+ value for the solid phase, it changed sharply from 25 to 300 at the inlet then remain between 150 and 160 for the rest of the metres along the wall. And for the gas phase it stays below 20.

I don't know what to do to get the parabolic profile at the out let?
I appreciate any kind of help,

Regards,
3ashoosh

mohw2002 December 27, 2012 02:58
Quote:
Originally Posted by Far (Post 395232)
There are two type of wall functions available for the high Reynolds number K-epsilon model:

1. Standard wall Functions (Will give problems if mesh is refined below Y+ 30)

2. Scalable wall function. Applicable for the meshes with even Y+ 1. But still it is wall function. The only purpose of scalable wall function is to avoid problems of successive refinements in standard wall function meshes. The algorithm ensures that your solver Y+ (Y* in Fluent) is always greater than 11.225. Y+= 11.225 is the intersection of linear and log law profile.
It has function which chooses the maximum of (11.225, Y*).

You already know that the Y* is different than the Y+ somehow. I don't remember when they are different and when they are same, having to do with buffer and log layer (Hint: turbulent and viscous stresses). Check the related material on the Y* and Y+. But any how in background fluent uses the Y*. So it will choose the 11.225 if Y* is less than 11.225 and uses Y* if it is greater than 11.225.

For low Reynolds number K-epsilon model there are two methods available (Select option enhanced wall treatment).

1. If Y+ is less than 1, it uses the pure two layer model and you solve the viscous sub layer. This formulation is good when you have separation or you want to resolve the near wall effects e.g. drag prediction, otherwise this will wast your resources and takes more solution time due to excessive stretching of mesh in the boundary layer.

2. If Y+ is variable along the wall (which is always present in practical industrial problems) then it use the enhanced wall treatment. It is the method to implement the enhanced wall function (hybrid wall function see this http://www.cfd-online.com/Forums/flu...-function.html and http://www.cfd-online.com/Forums/flu...treatment.html ) for the two layer k-epsilon model) For this method you should ensure Y+<10 for best usage. Otherwise for the higher Y+ values this will gradually use the log layer i.e. wall function approach.
Hello, I am using CFX and I want to get the values of y+, k+ and epsilon+ in the pipe for two cases one with perforations and without perforations.
Thanks in advance

3ashoosh December 27, 2012 09:26
Quote:
Originally Posted by Far (Post 395232)
There are two type of wall functions available for the high Reynolds number K-epsilon model:

1. Standard wall Functions (Will give problems if mesh is refined below Y+ 30)

2. Scalable wall function. Applicable for the meshes with even Y+ 1. But still it is wall function. The only purpose of scalable wall function is to avoid problems of successive refinements in standard wall function meshes. The algorithm ensures that your solver Y+ (Y* in Fluent) is always greater than 11.225. Y+= 11.225 is the intersection of linear and log law profile.
It has function which chooses the maximum of (11.225, Y*).

You already know that the Y* is different than the Y+ somehow. I don't remember when they are different and when they are same, having to do with buffer and log layer (Hint: turbulent and viscous stresses). Check the related material on the Y* and Y+. But any how in background fluent uses the Y*. So it will choose the 11.225 if Y* is less than 11.225 and uses Y* if it is greater than 11.225.

For low Reynolds number K-epsilon model there are two methods available (Select option enhanced wall treatment).

1. If Y+ is less than 1, it uses the pure two layer model and you solve the viscous sub layer. This formulation is good when you have separation or you want to resolve the near wall effects e.g. drag prediction, otherwise this will wast your resources and takes more solution time due to excessive stretching of mesh in the boundary layer.

2. If Y+ is variable along the wall (which is always present in practical industrial problems) then it use the enhanced wall treatment. It is the method to implement the enhanced wall function (hybrid wall function see this http://www.cfd-online.com/Forums/flu...-function.html and http://www.cfd-online.com/Forums/flu...treatment.html ) for the two layer k-epsilon model) For this method you should ensure Y+<10 for best usage. Otherwise for the higher Y+ values this will gradually use the log layer i.e. wall function approach.
Hi,
Is this also true for multi phase. I mean how can I adapt the Y+ values? When I click the adapt button then Yplus/Ystar adaptation window then on clicking compute for phase 1 (air) gave max of 43.33. Then when I put the min allowed 30 and max allowed 300 and click adapt it gave me an error, fata signal (ACCESS_VIOLATION)
Can you or any one help me? Thanks

Crank-Shaft January 13, 2013 00:31
Quote:
Originally Posted by Far (Post 395232)
There are two type of wall functions available for the high Reynolds number K-epsilon model:

1. Standard wall Functions (Will give problems if mesh is refined below Y+ 30)

2. Scalable wall function. Applicable for the meshes with even Y+ 1. But still it is wall function. The only purpose of scalable wall function is to avoid problems of successive refinements in standard wall function meshes. The algorithm ensures that your solver Y+ (Y* in Fluent) is always greater than 11.225. Y+= 11.225 is the intersection of linear and log law profile.
It has function which chooses the maximum of (11.225, Y*).

You already know that the Y* is different than the Y+ somehow. I don't remember when they are different and when they are same, having to do with buffer and log layer (Hint: turbulent and viscous stresses). Check the related material on the Y* and Y+. But any how in background fluent uses the Y*. So it will choose the 11.225 if Y* is less than 11.225 and uses Y* if it is greater than 11.225.

For low Reynolds number K-epsilon model there are two methods available (Select option enhanced wall treatment).

1. If Y+ is less than 1, it uses the pure two layer model and you solve the viscous sub layer. This formulation is good when you have separation or you want to resolve the near wall effects e.g. drag prediction, otherwise this will wast your resources and takes more solution time due to excessive stretching of mesh in the boundary layer.

2. If Y+ is variable along the wall (which is always present in practical industrial problems) then it use the enhanced wall treatment. It is the method to implement the enhanced wall function (hybrid wall function see this http://www.cfd-online.com/Forums/flu...-function.html and http://www.cfd-online.com/Forums/flu...treatment.html ) for the two layer k-epsilon model) For this method you should ensure Y+<10 for best usage. Otherwise for the higher Y+ values this will gradually use the log layer i.e. wall function approach.
Fantastic explanation Far. This is so much clearer to me now and although I haven't used the k-epsilon turbulence model lately, learning about the various features in a simplified and intuitive manner is definitely appealing.

Thanks for sharing your knowledge.

mohw2002 February 28, 2013 03:20
Translatio CFX-pre
 
Hi everyone
I am looking for the answer about the translation by CFX-pre. My geometry is quarter pipe (inlet-velocity is boundary condition for axial inlet) with perforation at upper surface (mass flow rate for radial boundary conditions) and outlet-pressure for outlet pipe at 120" long. Please I want to extent this pipe to 2 copies i.e. to be 3 domains (one total domain) the total long is 360".
Now for the new domain one axial inlet (inlet-velocity at the beginning of the pipe) and at the end of the total domain is (outlet-pressure). My question what I will put the boundary conditions for the interfaces between 1 and 2, 2 and 3?

Thanks

Kanarya March 8, 2014 08:01
multiphase wall functions
 
hi,
is there any modifications in wall functions in fluent for two fluid models?
thanks in advance!

Far March 19, 2014 04:43
i have no info about the two phase flow

siamakghh2000 February 11, 2016 14:01
Wall function for very rough surfaces
 
Hello all,

Does anyone know if there is a certain type of wall function out there used for very rough surfaces?

Yours, respectfully,
Siamak Gharahjeh

davidwilcox May 26, 2016 06:56
for rough surfaces, you need to consider the roughness reynolds number, h+. If you look at the plot for y+ vs u+ you will see the effect of h+ on the log layer. There is a sand-grain roughness model. Note that when roughness is involved, your first node away from the wall must adhere to certain geometric constraints. I think it is best if you read this paper by Blocken titled "CFD simulation of the atmospheric boundary layer:
wall function problems".
Your problem might not be the same as the atmospheric boundary layer but the information in that paper is very useful.

Shamoon Jamshed July 10, 2016 15:16
Quote:
Originally Posted by davidwilcox (Post 601923)
for rough surfaces, you need to consider the roughness reynolds number, h+. If you look at the plot for y+ vs u+ you will see the effect of h+ on the log layer. There is a sand-grain roughness model. Note that when roughness is involved, your first node away from the wall must adhere to certain geometric constraints. I think it is best if you read this paper by Blocken titled "CFD simulation of the atmospheric boundary layer:
wall function problems".
Your problem might not be the same as the atmospheric boundary layer but the information in that paper is very useful.
Can the k-w SST model be good in determining shear quantities while not as good in determining thermal quantities?

mjthomp3 July 18, 2016 13:07
Quote:
Originally Posted by Far (Post 395232)
There are two type of wall functions available for the high Reynolds number K-epsilon model:

1. Standard wall Functions (Will give problems if mesh is refined below Y+ 30)

2. Scalable wall function. Applicable for the meshes with even Y+ 1. But still it is wall function. The only purpose of scalable wall function is to avoid problems of successive refinements in standard wall function meshes. The algorithm ensures that your solver Y+ (Y* in Fluent) is always greater than 11.225. Y+= 11.225 is the intersection of linear and log law profile.
It has function which chooses the maximum of (11.225, Y*).

You already know that the Y* is different than the Y+ somehow. I don't remember when they are different and when they are same, having to do with buffer and log layer (Hint: turbulent and viscous stresses). Check the related material on the Y* and Y+. But any how in background fluent uses the Y*. So it will choose the 11.225 if Y* is less than 11.225 and uses Y* if it is greater than 11.225.

For low Reynolds number K-epsilon model there are two methods available (Select option enhanced wall treatment).

1. If Y+ is less than 1, it uses the pure two layer model and you solve the viscous sub layer. This formulation is good when you have separation or you want to resolve the near wall effects e.g. drag prediction, otherwise this will wast your resources and takes more solution time due to excessive stretching of mesh in the boundary layer.

2. If Y+ is variable along the wall (which is always present in practical industrial problems) then it use the enhanced wall treatment. It is the method to implement the enhanced wall function (hybrid wall function see this http://www.cfd-online.com/Forums/flu...-function.html and http://www.cfd-online.com/Forums/flu...treatment.html ) for the two layer k-epsilon model) For this method you should ensure Y+<10 for best usage. Otherwise for the higher Y+ values this will gradually use the log layer i.e. wall function approach.

Yep, I would say go with EWT for y+ < 1 and use RKE.....
if you have y+ >=30 go with NEWF and use RKE....


All times are GMT -4. The time now is 15:07.