# LES

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

 April 22, 2009, 06:31 #161 New Member   sungho yoon Join Date: Mar 2009 Posts: 20 Rep Power: 10 Sponsored Links how can I upload a figure from my hardware? I don't have the URL for the image.

 May 19, 2009, 09:46 Problems with LES simulation of 2-D plane liquid sheet #162 Senior Member   Suresh kumar Kannan Join Date: Mar 2009 Location: Luxembourg, Luxembourg, Luxembourg Posts: 129 Rep Power: 10 Hello Everybody, I have been using lesInterfoam to run plane liquid jet simulations. My mesh is a 2-D mesh with only hex cells . In my mesh I have a channel which then opens into a chamber. I am applying pressure inlet boundary conditions at the channel and at the end of the chamber it is presuure inlet-outlet condition. Actually my simulation results are not quiet matching with the experiements as I can see that in the experiements there is no break up for my case but in my simulations i see that my jet is breaking up. I use one equation eddy model for LES. I found in one of the articles by J.Janicka "Large-Eddy simulation of the Primary Breakup of a spatially Developing Liquid Film" where it is specified that the smagorinsky coefficent C_s has to be increased to 0.2. Then i looked into the paper by Henry and G.Tabor for the relation between the C_s and C-K,C_e. Luckily I found a post which gives me the relation as C_s=sqrt(C_k*(sqrt(C_k/C_e)). I hope this relation is correct. Then i used this relation and used a C_k of 0.15 and C_e of 1.75 which gives a C_s of 0.2. I found from my simulation that the break up now is delayed , So it means by varying this parameter am I doing the right thing as I am controlling the dissipation coefficent. My reynolds number is around 2000 range. I have one more question about the initial value of k in my 0/ directory. I am specifying a k value of 100 at the inlet and outlet. Do i have to play with this value as well or should i keep it constant and only change the C_k and C_e value to obtain closer results to experiments. Any suggestions are welcome. bye with regards K.Suresh kumar

 August 19, 2009, 13:24 #163 New Member   Gabriela Bracho Join Date: Mar 2009 Location: Valencia, Valencia, Spain Posts: 14 Rep Power: 10 Hello I'm doing LES calculation for internal flow and results look fine. Now, I need to calculate the energy spectrum turbulence (E(K) vs K)... Is there any tool in OpenFoam available to calculate this?? Or somebody knows a way to obtain E(K) ?? I'll appreciate any hint... Gaby

 August 20, 2009, 08:37 #164 Senior Member     Daniel WEI (老魏) Join Date: Mar 2009 Location: Beijing, China Posts: 689 Blog Entries: 9 Rep Power: 14 No, you need to do it yourself. __________________ ~ Daniel WEI ------------- Boeing Research & Technology - China Beijing, China Email

 October 9, 2009, 15:54 #165 Member   Join Date: Mar 2009 Posts: 46 Rep Power: 10 Hi everybody, I need to calculate epsilon in the oodles.C file to use it in one of the equations. Is there a way to calculate or get epsilon in the main solver body? I tried to access the velocity fluctuations or instantaneous strain rates there to define epsilon directly, but it didn't work either. Is there a way to get epsilon in the main LES solver? Any help is appreciated very much, Yours sincerely, Maani

October 10, 2009, 11:47
#166
Member

Michael Roth
Join Date: Mar 2009
Posts: 46
Rep Power: 10
Quote:
 Originally Posted by mmahdinia Hi everybody, I need to calculate epsilon in the oodles.C file to use it in one of the equations. Is there a way to calculate or get epsilon in the main solver body? I tried to access the velocity fluctuations or instantaneous strain rates there to define epsilon directly, but it didn't work either. Is there a way to get epsilon in the main LES solver? Any help is appreciated very much, Yours sincerely, Maani
Within e.g. oodles you can access epsilon through sgsModel->epsilon()

Take a look at the source code for the various LES turbulence models to see how specifically it is calculated so that you know what you're getting from this e.g. ce_ * k() * sqrt(k()) / delta()

Mike

February 3, 2010, 03:50
#167
Senior Member

Jiang
Join Date: Oct 2009
Location: Japan
Posts: 186
Rep Power: 9
Dear all, My LES result has problem when comparing with experiment and RANS model.

The following is my simulation model.
wind comes from left to right. The inlet velocity is a profile got from experiment.
First I use k-Epsilon RANS model, and the result is good;
Second I use Smagorinsky model+ wall function, the mesh is 30,0000.
Third I use Smagorinsky+wall function+VanDiest damping function, mesh number is 50,0000.

I compare my result with RANS model and experiment data. The comparition line is in front of the building, an horizontal line 12 mm above the ground.Through comparing, you could see my LES result is not good .

The LES U_x velocity is much smaller than experiment and RANS result. Who could tell me what's the reason ?
Because the comparition line is near ground, I doult my LES result is affected by ground, then how could I deal with this ?
or may be my LES result is affected by symmetry side boundary condition ?
or my mesh is not fine enough near the ground ?

Who could tell me the reason ? thank you very much.
Attached Images
 building.jpg (76.0 KB, 70 views) 1.jpg (66.5 KB, 114 views)

 February 3, 2010, 03:58 #168 Senior Member     Daniel WEI (老魏) Join Date: Mar 2009 Location: Beijing, China Posts: 689 Blog Entries: 9 Rep Power: 14 What's your Re? And could you give us a picture of your mesh? __________________ ~ Daniel WEI ------------- Boeing Research & Technology - China Beijing, China Email

February 3, 2010, 07:36
#169
Senior Member

Jiang
Join Date: Oct 2009
Location: Japan
Posts: 186
Rep Power: 9
Dear lakeat, this is my mesh.
Re number base on building width is about 30000.

when I use wall function, the first grid to the ground is 5mm;
when I use vanDriest+ wall function, the first grid is 0.8mm;
I know this is not fine enough for LES, but I think this is not the reason. when I use LES, below the height of half building, my velocity transfer slowly,
whereas the RANS model transfer faster, and the RANS model result is close to the experiment data. for both RANS and LES case, I use the same inlet velocity profile, which is got from experiment.
So I don't know why LES result is not better than RANS.

This is my mesh:
Attached Images
 top.jpg (75.3 KB, 65 views) sideview.jpg (68.1 KB, 53 views) nearground.jpg (62.0 KB, 50 views)

February 3, 2010, 12:15
#170
Senior Member

Daniel WEI (老魏)
Join Date: Mar 2009
Location: Beijing, China
Posts: 689
Blog Entries: 9
Rep Power: 14
Quote:
 Originally Posted by panda60 Dear lakeat, this is my mesh. Re number base on building width is about 30000. when I use wall function, the first grid to the ground is 5mm; when I use vanDriest+ wall function, the first grid is 0.8mm; I know this is not fine enough for LES, but I think this is not the reason. when I use LES, below the height of half building, my velocity transfer slowly, whereas the RANS model transfer faster, and the RANS model result is close to the experiment data. for both RANS and LES case, I use the same inlet velocity profile, which is got from experiment. So I don't know why LES result is not better than RANS. This is my mesh:

First of all, I do not think LES is an almighty machine, the one who uses LES must have a good understanding of the idea before he can get good results.

As the Re increases, and as the wall approaches, the grids and turbulence structures will become inhomogeneous, and hence violates most traditional SGS model, I am not surprised if LES (incorrectly applied) results are worse than RANS'.

Your case is absolutely a wall affected flow where wall effect can not be simply treated with so coarse a mesh and van-driest treatment.

Just some thoughts.
__________________
~
Daniel WEI
-------------
Boeing Research & Technology - China
Beijing, China
Email

 February 3, 2010, 22:39 #171 Senior Member   Jiang Join Date: Oct 2009 Location: Japan Posts: 186 Rep Power: 9 Dear lakeat, Because one case of LES simulation need a long time. so I must desigh the model before using. Do you think I will get a better result when using Smagorinsky +VanDriest model with a finer mesh ?

February 3, 2010, 22:54
#172
Senior Member

Daniel WEI (老魏)
Join Date: Mar 2009
Location: Beijing, China
Posts: 689
Blog Entries: 9
Rep Power: 14
Quote:
 Originally Posted by panda60 Dear lakeat, Because one case of LES simulation need a long time. so I must desigh the model before using. Do you think I will get a better result when using Smagorinsky +VanDriest model with a finer mesh ?
Okay, here's my experience, you may disagree.

I used to use a bad mesh with some traditional SGS model plus some traditional wall treatment, yet the results are very bad. (It become worse as the Re increases)

For Massively separated external flows, where pressure gradient is so great, I think DES (not DES97) will work better, with a "coarser" mesh.

Though I have never tried "Smagorinsky+VanDriest", but I would not expect a good result until I use them with a good mesh.

Another issue, how did you do your averaging?

Regards,
__________________
~
Daniel WEI
-------------
Boeing Research & Technology - China
Beijing, China
Email

 February 4, 2010, 01:32 #173 Senior Member   Jiang Join Date: Oct 2009 Location: Japan Posts: 186 Rep Power: 9 Dear lakeat, I just use OpenFOAM 's own average method like this: fieldAverage1 { type fieldAverage; functionObjectLibs ("libfieldFunctionObjects.so"); enabled true; outputControl outputTime; fields ( U { mean on; prime2Mean on; base time; } ); }

 February 9, 2010, 14:26 #174 Senior Member   Claus Meister Join Date: Aug 2009 Location: Wiesbaden, Germany Posts: 241 Rep Power: 11 Hello Foamers! You guys seems very involved into LES in OpenFoam. I have a question about the anisotropic filter. I have to definied anisotropicFilteredSymmTensorField { } but I do not know what to write in. I do not even know for what that is. Can anybody help? Cheers, Claus

February 13, 2010, 12:57
#175
Member

Join Date: Apr 2009
Posts: 38
Rep Power: 10
Quote:
 Originally Posted by luiz eduardo bittencourt sampaio I think the quality of my mesh is fine, at least in terms of edge angles, transition, etc. It may be not sufficiently refined, as it is my first LES case, and I dont have any idea what are the mesh requirements for LES. Does any one know how to estimate the best grid spacing for bounded LES flows? I am using very small steps (2x10-5), to get a Co lower then 0.5. Actually it is now 0.27, but every iteration, it gets a little bit higher and higher. I think it would take to long to get a steady state... And eventually this increase in Co may lead to overblown. Do you think this is happening due to the coarseness of my mesh. (well, I am not so sure it is coarse, unless I knew what should be a frequency in the inertial range of spectrum) Thanks for helping me. Luiz
The grid cell size should be set up so that it takes much more than one time step for fluid to cross a cell - so it's related to mean and peak local velocity. The time step should be set up so that turbulent vortices are resolved in time - say 10 time steps for a turbulent vortex to circulate once around. This is only ballpark, but it provides some guidance.

 March 11, 2010, 04:39 #176 Senior Member   Jiang Join Date: Oct 2009 Location: Japan Posts: 186 Rep Power: 9 Dear all, how can I average u'T' ? fieldAverage1 { type fieldAverage; functionObjectLibs ("libfieldFunctionObjects.so"); enabled true; outputControl outputTime; fields ( U { mean on; prime2Mean on; base time; } p { mean on; prime2Mean on; base time; } T { mean on; prime2Mean on; base time; } C { mean on; prime2Mean on; base time; } ); } It seems that fieldAverage only export u'^2, v'^2, w'^2, u'v', u'w', v'w' and T'^2. But i want to get turbulent heat flux u'T', v'T', w'T', How can I modify fieldAverage to get these variable ? Thank you very much.

April 6, 2010, 13:41
Were you able to implement this?
#177
Member

Matthew J. Churchfield
Join Date: Nov 2009
Posts: 49
Rep Power: 10
Rolando,

Were you able to implement an LES wall stress boundary conditions successfully? I am trying the same thing. I reformulated a solver to have the SGS stresses explicitly in the U equation (rather than take the anisotropic viscosity approach) and then directly apply the proper stresses through a boundary condition; however, this approach is plagued with numerical stability issues.

Please let me know which approach you took.

Thank you,

Matt

Quote:
 Originally Posted by rolando Thanks Henry, my actual problem is to implement a LES wall function which should calculate the wall stresses not only depending on the gradient but also on other information (which is equivalent to an anisotropic wall viscosity). Following your suggestion IŽll implement it in the LESmodel itself. Rolando

April 6, 2010, 18:38
Implementation of divDevReff(U) in full RST models
#178
Member

Matthew J. Churchfield
Join Date: Nov 2009
Posts: 49
Rep Power: 10
Henry,

I am trying to use OpenFOAM for atmospheric boundary layer LES, and the subject of the message below has become very important to me. I want to specify the shear stresses at the ground, so I thought it might be a good idea to modify pisoFoam such that it carries a symmetric tensor field R = turbulence->R()-(2.0/3.0)*I*k, which is the deviatoric SGS stress tensor. In this way, I can make the divergence of SGS stress term fvc::div(R), and specify the stress tensor at the ground.

Although this sounded simple, it creates very undesirable high-frequency oscillation in the velocity field. It makes me think that the usual formulation of divDevReff(U) is important to maintain an oscillation-free solution. It now appears to me that the implicit Laplacian in divDevReff(U) is very important for stability.

I also tried what you suggested below and used the formulation from the LRR model. Only with a fairly large value of couplingFactor could I create a smooth solution, but it was too smooth--it seemed to smooth out the turbulence that should be there.

I would highly appreciate any thoughts or comments upon this subject. I am about to give up on this approach and do something more similar to the nuSGSWallFunction--I just wanted independent control of the R_ij components at the wall.

Thank you,

Matt

Quote:
 Originally Posted by henry Yes currently the sgs viscosity is isotropic. Most anisotropic models generate stresses directly rather than an anisotropic viscosity and this would typically be implemented as a correction on a simpler model which can be implemented implicitly in the momentum equation like the one currently implemented. Take a look at the implementation of the LRR or LaunderGibson RANS models and you will see what I mean. > Maybe by deriving the wall function of > "fixedValueFvPatchVectorField"? No, it's not that simple. If you want to use an anisotropic sgs viscosity model you will have to use an tensorial-viscosity form of the laplacian in the momentum equation or use the correction approach I suggested above.

December 6, 2010, 19:37
#179
Member

Bernhard Grieser
Join Date: Mar 2010
Location: Zurich, Switzerland
Posts: 30
Rep Power: 9
Quote:
 Originally Posted by panda60 Dear all, how can I average u'T' ? . . . It seems that fieldAverage only export u'^2, v'^2, w'^2, u'v', u'w', v'w' and T'^2. But i want to get turbulent heat flux u'T', v'T', w'T', How can I modify fieldAverage to get these variable ? Thank you very much.
I'm interested in the answer to that question as well. Does someone know?
Did you find out, panda60?

 May 30, 2011, 14:14 #180 Senior Member   n/a Join Date: Sep 2009 Posts: 198 Rep Power: 9 Did anyone get to modify fieldAverage to calculate u'T,v'T, and w'T' ? Deji

 Thread Tools Display Modes Linear Mode

 Posting Rules You may not post new threads You may not post replies You may not post attachments You may not edit your posts BB code is On Smilies are On [IMG] code is On HTML code is OffTrackbacks are On Pingbacks are On Refbacks are On Forum Rules

All times are GMT -4. The time now is 22:00.