No Fluctuatios in LES results for pipe flow!!!

Roohi · September 5, 2011, 16:46

Hello Guys!!
I am simulating a sigle phase (water) flow inside a pipe by LES in CFX. The pipe diameter is 2cm and it's length is 6*D. Also the Re# is about 5300.
My inlet BC is a predefined velocity profile (1/7 Power law profile) and outlet BC is zero static pressure. I am using a SST Solution (RANS steady solution) as initialization file. Since it is intialized by RANS I set the Expert parameter "apply ic fluctuations for les" to "t". I did everything based on the documetations of CFX.
my mesh is fine enough {(wall scale)y+=1... ( span wise)Z+=20...(streamwise) X+=50} and Courant # is always less than 0.8.
It is pretty much all about my problem definition.
However when I run the case I get no fluctuations in velocities at all. I monitor velocity of a point and all the time it goes smoothly.
Would be kind enough to tell me what is wrong with my case??
Thanks in advance.
Roohi

ghorrocks · September 5, 2011, 19:50

What turbulence model are you using? What differencing scheme? What time advancement scheme?

Roohi · September 5, 2011, 21:24

Glenn,
Thanks for the quick reply.
My Turbulence model is Smagorinskey Model with default Cs=0.1...
And also I used Central Differencing scheme, and Second order backward Euler for transient scheme (if you mean that by "time advancement scheme").
Thanks again,
roohi

ghorrocks · September 6, 2011, 20:05

Why do you say "my mesh is fine enough"? You quote y+ values but that means little in LES where you must get the mesh size appropriate compared to the turbulence structures you are modelling.

Roohi · September 7, 2011, 16:11

Hi Glenn,
I picked up those numbers for mesh sizes based on what exists in literature. Because they have done this simulation with those mesh sizes.
Are you telling me that this is caused by my mesh?
Should I go for coarser or finer mesh?
Any suggestions?
Thanks again,
Roohi

ghorrocks · September 7, 2011, 19:50

Well, if you are not getting the LES fluctuations then you have too much dissipation in your model. Dissipation comes from the differencing schemes - but you are already using central differencing and second order time so you have minimised that already. So if you still have too much dissipation the next thing to look at is using a finer mesh.

Of course the whole idea of LES is that the mesh size is linked to a filter, where turbulence length scales in the inertial region are filtered out and the subgrid model replaces them, but length scales larger than that are simulated and therefore the mesh needs to resolve these scales.

This is why I say you need to work out the turbulence length scale you wish to resolve, and then generate a mesh fine enough to resolve it.

September 5, 2011, 16:46	No Fluctuatios in LES results for pipe flow!!!	#1
Roohi New Member Roohi Shokri Join Date: Aug 2011 Posts: 3 Rep Power: 15	Hello Guys!! I am simulating a sigle phase (water) flow inside a pipe by LES in CFX. The pipe diameter is 2cm and it's length is 6*D. Also the Re# is about 5300. My inlet BC is a predefined velocity profile (1/7 Power law profile) and outlet BC is zero static pressure. I am using a SST Solution (RANS steady solution) as initialization file. Since it is intialized by RANS I set the Expert parameter "apply ic fluctuations for les" to "t". I did everything based on the documetations of CFX. my mesh is fine enough {(wall scale)y+=1... ( span wise)Z+=20...(streamwise) X+=50} and Courant # is always less than 0.8. It is pretty much all about my problem definition. However when I run the case I get no fluctuations in velocities at all. I monitor velocity of a point and all the time it goes smoothly. Would be kind enough to tell me what is wrong with my case?? Thanks in advance. Roohi

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LES transitional pipe entrance flow	hasanduz	Main CFD Forum	2	October 11, 2013 18:59
Gas flow in vortex tube - LES and grid resolution	bernarde	Main CFD Forum	1	March 26, 2009 11:50
Turbulent Flow in a Square Duct using LES	Hock Ming	FLUENT	0	February 7, 2009 21:25
good cold flow results but problem with hot flow,.	Rams	FLUENT	0	June 13, 2006 12:08
computation about flow around a yawed cone	Tylor Xie	Main CFD Forum	0	June 9, 1999 08:33

September 5, 2011, 19:50		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	What turbulence model are you using? What differencing scheme? What time advancement scheme?

September 5, 2011, 21:24		#3
Roohi New Member Roohi Shokri Join Date: Aug 2011 Posts: 3 Rep Power: 15	Glenn, Thanks for the quick reply. My Turbulence model is Smagorinskey Model with default Cs=0.1... And also I used Central Differencing scheme, and Second order backward Euler for transient scheme (if you mean that by "time advancement scheme"). Thanks again, roohi

September 6, 2011, 20:05		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Why do you say "my mesh is fine enough"? You quote y+ values but that means little in LES where you must get the mesh size appropriate compared to the turbulence structures you are modelling.

September 7, 2011, 16:11		#5
Roohi New Member Roohi Shokri Join Date: Aug 2011 Posts: 3 Rep Power: 15	Hi Glenn, I picked up those numbers for mesh sizes based on what exists in literature. Because they have done this simulation with those mesh sizes. Are you telling me that this is caused by my mesh? Should I go for coarser or finer mesh? Any suggestions? Thanks again, Roohi

September 7, 2011, 19:50		#6
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,871 Rep Power: 144	Well, if you are not getting the LES fluctuations then you have too much dissipation in your model. Dissipation comes from the differencing schemes - but you are already using central differencing and second order time so you have minimised that already. So if you still have too much dissipation the next thing to look at is using a finer mesh. Of course the whole idea of LES is that the mesh size is linked to a filter, where turbulence length scales in the inertial region are filtered out and the subgrid model replaces them, but length scales larger than that are simulated and therefore the mesh needs to resolve these scales. This is why I say you need to work out the turbulence length scale you wish to resolve, and then generate a mesh fine enough to resolve it.