A question about LES in 2D Fluent

tjushang · September 25, 2014, 05:48

Hello everyone,
As far as I'm concerned, large eddy simulation (LES) is mostly employed in 3D cases, there're very few research using LES in 2D models. The choice of LES in 2D Fluent is also hidden and can only be chosen by typing (rpsetvar 'les-2d? #t) in TUI.
I wonder the reason why LES is barely used in 2D cases. Is there accuracy problem in 2D LES? DId anyone use 2D LES in Fluent and achieve good results? What about choose LES in 3D Fluent but make my geometry very thin in one direction (nearly a 2D model along, say, xy plane if the thickness in z direction is very small) ?

sbaffini · September 25, 2014, 20:35

What do you expect by LES and why?

LES can, and is, used in 2D, but only when the flow dynamics you are trying to capture can be approximated in that way.

For example, it is largely used for atmospheric flows, because their large scale dynamics is essentially 2D. 3D effects, while present, are at much lower scales, which are not resolved.

Roughly speaking, LES means describing the large scales of your fow. If they are 3D (and they usually are) there is no point in a 2D LES. Notice that we are talking about the instantaneous physical flow scales, nothing to do with the expected average solution typical of RANS/URANS.

tjushang · September 25, 2014, 22:08

Thank you for reply, Lampitella.
My aim is to model combustion in engine. Even if it's a 3D case, the thickness of the cylinder (about 8mm) at TDC is small compared with its diameter (about 80mm). I wonder if I can use a 2D circle as my computational domain if I only calculate the combustion at TDC (fixed grid), in which case I could save my computational cost. In 3D case, especially in the case I study that has large amount of cells, I encountered insufficiency of computer's memory. It also cost too much time to finish a series of calculations.
Anyway, you enlightened me a lot, Thank you again.

sbaffini · September 26, 2014, 04:35

Dear Yibao,

i can only say that i've nver seen a 2D computation in combustion, no matter what.

If you think about my example (horizontal scale around 40.000 km, vertical scale around 10 km, aspect ratio = 4000), you clearly understand why.

Also, i want to stress again that a fundamental aspect in going 2D for atmospheric flows is that the threedimensionality is relevant only for the scales which in LES parlance are SGS and not resolved, thus would make little sense to actually solve them (still, this might be merely seen as a consequence of the 4000 aspect ratio).

September 25, 2014, 05:48	A question about LES in 2D Fluent	#1
tjushang New Member Yibao Shang Join Date: Jun 2014 Location: China Posts: 15 Rep Power: 12	Hello everyone, As far as I'm concerned, large eddy simulation (LES) is mostly employed in 3D cases, there're very few research using LES in 2D models. The choice of LES in 2D Fluent is also hidden and can only be chosen by typing (rpsetvar 'les-2d? #t) in TUI. I wonder the reason why LES is barely used in 2D cases. Is there accuracy problem in 2D LES? DId anyone use 2D LES in Fluent and achieve good results? What about choose LES in 3D Fluent but make my geometry very thin in one direction (nearly a 2D model along, say, xy plane if the thickness in z direction is very small) ?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
LES Filtering: how are the small and large scales equations solved?	atmcfd	Main CFD Forum	38	March 14, 2016 15:52
question for LES	colopolo	Main CFD Forum	12	March 23, 2009 10:54
FDS vs Fluent, what's the difference on LES?	Jim	FLUENT	0	August 12, 2006 16:32
Fluent 6.2 + LES	Sowjanya	FLUENT	0	August 23, 2005 11:59
LES using FLUENT	sarah_ron	FLUENT	7	October 22, 2004 18:36

September 25, 2014, 20:35		#2
sbaffini Senior Member Paolo Lampitella Join Date: Mar 2009 Location: Italy Posts: 2,195 Blog Entries: 29 Rep Power: 39	What do you expect by LES and why? LES can, and is, used in 2D, but only when the flow dynamics you are trying to capture can be approximated in that way. For example, it is largely used for atmospheric flows, because their large scale dynamics is essentially 2D. 3D effects, while present, are at much lower scales, which are not resolved. Roughly speaking, LES means describing the large scales of your fow. If they are 3D (and they usually are) there is no point in a 2D LES. Notice that we are talking about the instantaneous physical flow scales, nothing to do with the expected average solution typical of RANS/URANS.

September 25, 2014, 22:08		#3
tjushang New Member Yibao Shang Join Date: Jun 2014 Location: China Posts: 15 Rep Power: 12	Thank you for reply, Lampitella. My aim is to model combustion in engine. Even if it's a 3D case, the thickness of the cylinder (about 8mm) at TDC is small compared with its diameter (about 80mm). I wonder if I can use a 2D circle as my computational domain if I only calculate the combustion at TDC (fixed grid), in which case I could save my computational cost. In 3D case, especially in the case I study that has large amount of cells, I encountered insufficiency of computer's memory. It also cost too much time to finish a series of calculations. Anyway, you enlightened me a lot, Thank you again.

September 26, 2014, 04:35		#4
sbaffini Senior Member Paolo Lampitella Join Date: Mar 2009 Location: Italy Posts: 2,195 Blog Entries: 29 Rep Power: 39	Dear Yibao, i can only say that i've nver seen a 2D computation in combustion, no matter what. If you think about my example (horizontal scale around 40.000 km, vertical scale around 10 km, aspect ratio = 4000), you clearly understand why. Also, i want to stress again that a fundamental aspect in going 2D for atmospheric flows is that the threedimensionality is relevant only for the scales which in LES parlance are SGS and not resolved, thus would make little sense to actually solve them (still, this might be merely seen as a consequence of the 4000 aspect ratio).