LES with rough surface

mikeP · October 2, 2013, 12:33

Dear Foamers,

How do you define surface roughness in LES? I read that OpenFOAM allows one to use RANS wall functions also in LES? How good is it, and also how should the mesh be then? 30 < y+ < 100?

or should I use a constant shear stress on the wall?

vivekcfd · November 4, 2013, 17:35

does not make much sense to use roughness wall function for LES. Even wall function approach for standard smooth pipe LES is not straight forward and questionable. How can one think of being doing LES using rough wall functions.

ashvinc9 · April 22, 2014, 16:30

Hi,

In my opinion, one can use the wall-function for LES too. It will reduce near wall mesh-resolution ans thus the CPU time. For the smooth surfaces and for low Re (or Re_Tau) of course it may not seen useful. But how about for rough surfaces, e.g. atmospheric boundary layers flows? The surface roughness (z_0 or K_s are implemented only using the wall functions). For the rough surface, I would recommend "nutURoughWallFunction" available in OF-2.2 which is meant for RANS simulation (based on Ks: Sand grain-roughness) but works perfectly in LES too. I have tested several times this function.

-- Ashvin

vivekcfd · April 23, 2014, 04:06

Wall functions follow log-law of the wall. It is not valid for LES for two reasons:

1) mostly the flow is not attached or accompany separation zones where no wall functions is valid. Moreover in most applications, flow is not at all fully developed.

2) Secondly, the log-law is applicable to the mean flow field and not to the instantaneous flow velocity field. In other words, the instantaneous wall shear-stress and flow profile can be well away from the reality.

However, the good thing is that wall function provides the possibility to compute high-Re flows. People use Werner-Wengel wall functions for LES simulations which is not exactly same as the RANS wall-functions

If you are getting good results in your case then it signifies that in your applications:

1. the role of wall is not that important

2. The results which you are comparing or predicting may be unaffected by the flow anisotropy near the wall.

For very accurate simulations (which indeed anyway make sense for LES), I would not use wall functions. In case one is interested in turbulence fluctuations, in my opinion it is cheaper or may be better to employ IDDES using SST or other RANS model. However drawback 1) of wall-functions still can not be overcome. On the other hand wall-functions may still give you good results for some cases. Since the definition of good and bad is relative and depends very much on the type of application and the precision requirement from the simulation.

I hope this helps you.

October 2, 2013, 12:33	LES with rough surface	#1
mikeP Member Join Date: Jun 2011 Posts: 42 Rep Power: 15	Dear Foamers, How do you define surface roughness in LES? I read that OpenFOAM allows one to use RANS wall functions also in LES? How good is it, and also how should the mesh be then? 30 < y+ < 100? or should I use a constant shear stress on the wall?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
[Gmsh] Problem with Gmsh	nishant_hull	OpenFOAM Meshing & Mesh Conversion	23	August 5, 2015 03:09
killed "snappyHexMesh"	parkh32	OpenFOAM Pre-Processing	2	April 8, 2012 18:12
[ICEM] Automatic mesh generation script surface intersection problem	stuart23	ANSYS Meshing & Geometry	0	May 13, 2011 02:10
Recover surface from surface mesh in IcemCFD	Jerry Tanner	CFX	0	August 20, 2008 13:48
CFX 4.4 New free surface option	Viatcheslav Anissimov	CFX	0	April 3, 2002 07:27

November 4, 2013, 17:35		#2
vivekcfd Member Vivek Kumar Join Date: Mar 2009 Location: Switzerland Posts: 35 Blog Entries: 1 Rep Power: 17	does not make much sense to use roughness wall function for LES. Even wall function approach for standard smooth pipe LES is not straight forward and questionable. How can one think of being doing LES using rough wall functions.

April 22, 2014, 16:30		#3
ashvinc9 New Member Ashvin Chaudhari Join Date: Aug 2011 Location: Finland Posts: 23 Rep Power: 15	Hi, In my opinion, one can use the wall-function for LES too. It will reduce near wall mesh-resolution ans thus the CPU time. For the smooth surfaces and for low Re (or Re_Tau) of course it may not seen useful. But how about for rough surfaces, e.g. atmospheric boundary layers flows? The surface roughness (z_0 or K_s are implemented only using the wall functions). For the rough surface, I would recommend "nutURoughWallFunction" available in OF-2.2 which is meant for RANS simulation (based on Ks: Sand grain-roughness) but works perfectly in LES too. I have tested several times this function. -- Ashvin

April 23, 2014, 04:06		#4
vivekcfd Member Vivek Kumar Join Date: Mar 2009 Location: Switzerland Posts: 35 Blog Entries: 1 Rep Power: 17	Wall functions follow log-law of the wall. It is not valid for LES for two reasons: 1) mostly the flow is not attached or accompany separation zones where no wall functions is valid. Moreover in most applications, flow is not at all fully developed. 2) Secondly, the log-law is applicable to the mean flow field and not to the instantaneous flow velocity field. In other words, the instantaneous wall shear-stress and flow profile can be well away from the reality. However, the good thing is that wall function provides the possibility to compute high-Re flows. People use Werner-Wengel wall functions for LES simulations which is not exactly same as the RANS wall-functions If you are getting good results in your case then it signifies that in your applications: 1. the role of wall is not that important 2. The results which you are comparing or predicting may be unaffected by the flow anisotropy near the wall. For very accurate simulations (which indeed anyway make sense for LES), I would not use wall functions. In case one is interested in turbulence fluctuations, in my opinion it is cheaper or may be better to employ IDDES using SST or other RANS model. However drawback 1) of wall-functions still can not be overcome. On the other hand wall-functions may still give you good results for some cases. Since the definition of good and bad is relative and depends very much on the type of application and the precision requirement from the simulation. I hope this helps you.