Outflow boundary condition for LES/DNS

 Register Blogs Members List Search Today's Posts Mark Forums Read

 February 18, 2014, 16:06 Outflow boundary condition for LES/DNS #1 New Member   Harish Abubaker Join Date: Jun 2013 Posts: 6 Rep Power: 12 I came across a statement that use of zero gradient boundary conditions for LES and DNS would result in unphysical results. Can anyone please explain why.

February 18, 2014, 16:13
#2
Senior Member

Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,765
Rep Power: 71
Quote:
 Originally Posted by Harish050887 I came across a statement that use of zero gradient boundary conditions for LES and DNS would result in unphysical results. Can anyone please explain why.
zero gradient BC can be adopted for outflow provided that is fixed far from the main zone... it can be improved by higher approximations

 February 19, 2014, 04:55 #3 New Member   Harish Abubaker Join Date: Jun 2013 Posts: 6 Rep Power: 12 Thank you..but can you also explain why does this happen ?

February 19, 2014, 05:59
#4
Senior Member

Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,765
Rep Power: 71
Quote:
 Originally Posted by Harish050887 Thank you..but can you also explain why does this happen ?
because you arbitrarily fix some components of the velocity gradient to vanish. That can be quite acceptable for some conditions but not in general

 February 28, 2014, 11:40 #5 New Member   Harish Abubaker Join Date: Jun 2013 Posts: 6 Rep Power: 12 Had to ask one more thing, why is it then ok to use the zero gradient bc for RANS ?

February 28, 2014, 12:12
#6
Senior Member

Filippo Maria Denaro
Join Date: Jul 2010
Posts: 6,765
Rep Power: 71
Quote:
 Originally Posted by Harish050887 Had to ask one more thing, why is it then ok to use the zero gradient bc for RANS ?

RANS solves for the statistically steady state, d/dn=0 is much more acceptable

March 2, 2014, 16:17
#7
Senior Member

Join Date: Aug 2011
Posts: 272
Rep Power: 15
Quote:
 Originally Posted by Harish050887 I came across a statement that use of zero gradient boundary conditions for LES and DNS would result in unphysical results. Can anyone please explain why.
zero gradient may be acceptable if the flow is established before it crosses the outlet boundary. This would certainly be not true for turbullent flows (LES/DNS) or only if the computational domain is quite large or in case where you know that the turbulence is damped by the viscosity in this region. However for RANS methods since the flow is time averaged then this BC becomes suitable. For LES/DNS use Orlansky boundary conditions that is:
dphy/dt + U*dphy/dn=0 (derivatives here are partial derivatives and U is a velocity which may be defined by different means, (it can be the mean velocity for example). phy stands for each componnent of the velocity vector.

 March 4, 2014, 04:42 #8 New Member   Harish Abubaker Join Date: Jun 2013 Posts: 6 Rep Power: 12 Thank you very much guys. I get the idea.

 May 17, 2014, 01:31 Orlansky Boundary condition #9 New Member   The Beginner Join Date: May 2014 Posts: 1 Rep Power: 0 How can i put Orlanksy boundary condition in FLUENT ?

 January 16, 2019, 08:37 #10 Senior Member   Jianrui Zeng Join Date: May 2018 Location: China Posts: 157 Rep Power: 7 Does someone know the Orlansky boundary condition or convective boundary condition in Openfoam? I want use this boundary in DNS simulation. Thank you.

January 16, 2019, 09:41
#11
Senior Member

Lucky
Join Date: Apr 2011
Location: Orlando, FL USA
Posts: 5,670
Rep Power: 65
Quote:
 Originally Posted by calf.Z Does someone know the Orlansky boundary condition or convective boundary condition in Openfoam? I want use this boundary in DNS simulation. Thank you.

 January 16, 2019, 10:04 #12 Senior Member   Join Date: Dec 2017 Posts: 153 Rep Power: 8 Yes. it is called "advective" in openFoam and does: du/dt + Udu/dn=0, at the outflow boundary. For more theoretical aspects on this condition you can look at ferzinger and peric book, chapter 9.

 January 16, 2019, 10:10 #13 Senior Member   Jianrui Zeng Join Date: May 2018 Location: China Posts: 157 Rep Power: 7 Thanks to AliE and LuckyTran ! I will have a try later!