
[Sponsors] 
September 8, 2013, 04:45 
Problems of bad results with LES

#1 
Member
sqing
Join Date: Sep 2012
Location: Dalian
Posts: 77
Rep Power: 6 
Hi foamers,
I have several problems with my walljet case for monthes. I have calculated the 2D case both with RANS models and LES models. However, the LES results are not good as I expected, especially in the viscous area, the velocity from the LES result is much lower than that from RANS result and EXP data. For I have set the same boundary conditions for both RANS and LES case, so I don't know what on earth leads to the "bad" result of LES. What's more, I don't know how to set the boundary condition for B and nuSgs in 0 folder. For my case, I just modified the B and nuSgs files with the tutorial case, and I'm not sure they are right. Case describe: velocity: inlet_cool: use data from exp; inlet_hot: use data from exp; wall_down:fixed value (0 0 0); wall_up: symmetry Plane; outlet:zerogradient; pressure: inlet_cool: zeroGradient; inlet_hot: zeroGradient; wall_down:zeroGradient; wall_up: symmetry Plane; outlet:fixed value 0; Temperature: inlet_cool: 300k; inlet_hot: 310k; wall_down:zeroGradient; wall_up: symmetry Plane; outlet:zerogradient; This is the framework structure：20130908 16:08:04的屏幕截图.png This is the velocity results：U.jpg These are my 0 and system folders of LES:0_system_les.tar.gz These are my 0 and system folders of RANS:0_system.tar.gz Can anybody help me please? Best regards! 

September 8, 2013, 09:44 

#2  
Senior Member
Join Date: Jan 2013
Posts: 342
Rep Power: 7 
Hi Sunxing,
Why did you expect to get the qualitatively correct results using 2D results? In 2D case, the turbulence is not true. But 2D case is still meaningful for debugging the code. Quote:


September 8, 2013, 20:33 

#3 
Member
sqing
Join Date: Sep 2012
Location: Dalian
Posts: 77
Rep Power: 6 
I know that it's not suitable to simulate a 2D case with LES model. However for time efficiency, I hope the 2D case can give me a "good" enough result in a short time. And I think at least the velocity result shouldn't be so lower than the exp data in the viscous area.


September 9, 2013, 01:46 

#4 
Senior Member
Bernhard
Join Date: Sep 2009
Location: Delft
Posts: 790
Rep Power: 15 
Did you check y+ values at the wall? Also, did you check the effect of inlet conditions, it seems to me that you are applying a fixed value inlet condition, which is of course not true for turbulence.
You are using LES, but what subgrid model do you use? You are using a zero value pressure at the outlet, are you sure that is correct for these kind of boundary layer flows? Did you try to confirm your model with normal walljets from literature? I am convinced that there are a few out there. 

September 9, 2013, 02:05 

#5  
Member
sqing
Join Date: Sep 2012
Location: Dalian
Posts: 77
Rep Power: 6 
Quote:
For my case the y+ is lower than 1 at the wall. Yes, I have applied a fixed value from exp data with a fluctuationScale (0.03 0.01 0.01), I think this can be enough for turbulent income, am I right? I have used smagorinsky(with damping function), dynOneEqEddy and dynamic smagorinsky models. All of them give lower velocity in the viscous area. I think a zero value pressure at the outlet is right, because for the RANS model, I also set this condition and it turns out well. Regards! 

September 9, 2013, 02:10 

#6 
Senior Member
Bernhard
Join Date: Sep 2009
Location: Delft
Posts: 790
Rep Power: 15 
How can you impose a velocity fluctuation in the zdirection?
Anyhow, you will have to perform a 3D simulation at least once to prove that your approach is correct. By the way, are you using your RANS mesh for the LES? 

September 9, 2013, 02:11 

#7  
Member
sqing
Join Date: Sep 2012
Location: Dalian
Posts: 77
Rep Power: 6 
Quote:


September 9, 2013, 02:18 

#8  
Member
sqing
Join Date: Sep 2012
Location: Dalian
Posts: 77
Rep Power: 6 
Quote:
I'm performing a 3D simulation for the case for more than 10 days, but it's so time costly and I think it still need more than one month to get the result. The mesh for LES is more finer than that for RANS. 

December 24, 2013, 00:08 

#9  
New Member
Simon Zhang
Join Date: Dec 2013
Posts: 12
Rep Power: 5 
Quote:
what solver did you used for wall jet simulation? I am also simulating wall jet with temperature variation. And I changed standard buoyantBoussinesqPimpleFoam to capable with LES. But the results is too bad, even worse than RANS. So how did you make it? 

December 24, 2013, 01:57 

#10  
Member
sqing
Join Date: Sep 2012
Location: Dalian
Posts: 77
Rep Power: 6 
Quote:
I used pisoFoam and added temperature function Best regards Sunxing 

March 31, 2014, 05:11 

#11 
Member
Join Date: Feb 2014
Posts: 57
Rep Power: 5 
Hi Sunxing,
I am new Foamer and I implemented a very simple LES model as shown by a tutorial pisoFoam/pitzDaily. I am searching to use the dynamic Smagorinsky model for incompressible nonhomogeneous turbulent flow inside an injector. This model is suitable for my case? May you give more details of following files:  constant/LESProperties  0/nuSgs, 0/k and 0/epsilon Thanks in advance, vut 

Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
bad results in first totalPressure boundary condition iterations.  immortality  OpenFOAM Running, Solving & CFD  0  March 15, 2013 06:01 
problems with a LES incompressible calculation (pisoFoam)  laurencep  OpenFOAM  2  April 22, 2010 04:53 
Problems with generated results of 3D wing analysis  Shaoming  ANSYS  0  March 12, 2010 11:37 
Problems with repeating results  Lee  Siemens  4  May 26, 2006 03:39 
RANS results for LES  Li Yang  Main CFD Forum  5  June 17, 2002 05:34 