# LES of channel flow: data, case files, technical report.

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

 August 13, 2019, 05:46 #21 New Member   Join Date: Jan 2014 Posts: 20 Rep Power: 11 Nice! Thank you!

May 31, 2020, 12:35
#22
Senior Member

Arijit Saha
Join Date: Feb 2019
Location: Singapore
Posts: 132
Rep Power: 5
Respected Sir, it was so nice of you to share this topic here at CFD Forum. I am a newbie in OpenFoam and trying my best to learn most out of it but most of the time I suffer in finding the proper answer to my doubt in google.
Anyway I was doing some research with Channel 395 case with some modification like the boundary condition. In an unchanged channel flow the U and p folder look like the attachment shared along with this post. From my obserservation what I depicted is as follows:-
1.Bottom and top wall is having a no-slip boundary condition.
2. The solution has been provided with some initial condition i.e for 60000 cells they have provided some initial values.
3. The flow that is coming out of outlet is again fed to inlet and in this way the cycle goes on.
What I didnt understand is what is the significance of the side using it as a cyclic boundary condition?
Once I m clear with this part then I ll aim for my next target that is feeding some surface value to the inlet patch using mapped field and run the simulation. So in what way can I modify the bc in order to feed the inlet patch with some predefined value instead of the flow coming out from the outlet?
Attached Files
 O.zip (1.2 KB, 1 views)

June 1, 2020, 06:08
#23
Senior Member

Timofey Mukha
Join Date: Mar 2012
Location: Stockholm, Sweden
Posts: 118
Rep Power: 12
Quote:
 Originally Posted by ari003 Respected Sir, it was so nice of you to share this topic here at CFD Forum. I am a newbie in OpenFoam and trying my best to learn most out of it but most of the time I suffer in finding the proper answer to my doubt in google. Anyway I was doing some research with Channel 395 case with some modification like the boundary condition. In an unchanged channel flow the U and p folder look like the attachment shared along with this post. From my obserservation what I depicted is as follows:- 1.Bottom and top wall is having a no-slip boundary condition. 2. The solution has been provided with some initial condition i.e for 60000 cells they have provided some initial values. 3. The flow that is coming out of outlet is again fed to inlet and in this way the cycle goes on. What I didnt understand is what is the significance of the side using it as a cyclic boundary condition? Once I m clear with this part then I ll aim for my next target that is feeding some surface value to the inlet patch using mapped field and run the simulation. So in what way can I modify the bc in order to feed the inlet patch with some predefined value instead of the flow coming out from the outlet?

The idea is that the domain is infinite in the wall-parallel directions. We simulate that with cyclic boundaries. If you have walls on the side, that is going to lead to different results, since you'd have associated velocity gradients etc.

Regarding changing the boundary to some sort of inlet, it depends on what you intend to "feed". In any case, you put some sort of Dirichlet bc for U and zeroGradient for the pressure at the inlet, and the outlet set pressure to 0 and zeroGradient for velocity.

June 1, 2020, 06:23
#24
Senior Member

Arijit Saha
Join Date: Feb 2019
Location: Singapore
Posts: 132
Rep Power: 5
Quote:
 Originally Posted by tiam The idea is that the domain is infinite in the wall-parallel directions. We simulate that with cyclic boundaries. If you have walls on the side, that is going to lead to different results, since you'd have associated velocity gradients etc. Regarding changing the boundary to some sort of inlet, it depends on what you intend to "feed". In any case, you put some sort of Dirichlet bc for U and zeroGradient for the pressure at the inlet, and the outlet set pressure to 0 and zeroGradient for velocity.
I also realized that once I remove the initial condition for all the cells from non-uniform to uniform the simulation is not running properly. But, for my use of timeVaryingMappedFixedValue at inlet I ve to make the initial condition as uniform otherwise how is it possible? Basically the cyclic bc runs when I put initial condition to all the cells but how can I run the same by feeding face-center value to the inlet patch?
Also attached with this post is the boundary field which I m trying to feed to the inlet, u can have a look.
Attached Files
 boundaryData.zip (23.2 KB, 2 views)

June 1, 2020, 07:12
#25
Senior Member

Timofey Mukha
Join Date: Mar 2012
Location: Stockholm, Sweden
Posts: 118
Rep Power: 12
Quote:
 Originally Posted by ari003 I also realized that once I remove the initial condition for all the cells from non-uniform to uniform the simulation is not running properly. But, for my use of timeVaryingMappedFixedValue at inlet I ve to make the initial condition as uniform otherwise how is it possible? Basically the cyclic bc runs when I put initial condition to all the cells but how can I run the same by feeding face-center value to the inlet patch? Also attached with this post is the boundary field which I m trying to feed to the inlet, u can have a look.

If you'll have an inlet and an outlet then initial conditions don't matter that much at all, they will be flushed out of the domain pretty fast anyway.

 June 1, 2020, 09:56 #26 Senior Member   Arijit Saha Join Date: Feb 2019 Location: Singapore Posts: 132 Rep Power: 5 Do you have any idea how can I make it working?

 June 4, 2020, 04:26 #27 Senior Member   Timofey Mukha Join Date: Mar 2012 Location: Stockholm, Sweden Posts: 118 Rep Power: 12 I'm sorry but I don't really understand what the problem is.

 December 28, 2022, 13:13 #28 Senior Member   chandra shekhar pant Join Date: Oct 2010 Posts: 216 Rep Power: 15 Hi @Tiam, Thanks for your paper and the codes that you shared, it's a very old post, so not sure, if it's still active. I am also trying to perform LES of Re_tau = 180 for channel flow in a little bit bigger domain, by imposing the pressure gradient using fvoptions and calculating the pressure gradient as Code: u_\tau^2/h , and calculating u_\tau using the Re_\tau value, since h is known, viscosity I predefined for my simulations. Unfortunately I am unable to see any turbulence in my simulations, could you please share your 0 folder? I assume that when you started your simulations it could have been started from 0 time. Any comment or suggestion would be welcomed, many thanks in advance.

December 28, 2022, 14:15
#29
Senior Member

Timofey Mukha
Join Date: Mar 2012
Location: Stockholm, Sweden
Posts: 118
Rep Power: 12
Quote:
 Originally Posted by chandra shekhar pant Hi @Tiam, Thanks for your paper and the codes that you shared, it's a very old post, so not sure, if it's still active. I am also trying to perform LES of Re_tau = 180 for channel flow in a little bit bigger domain, by imposing the pressure gradient using fvoptions and calculating the pressure gradient as Code: u_\tau^2/h , and calculating u_\tau using the Re_\tau value, since h is known, viscosity I predefined for my simulations. Unfortunately I am unable to see any turbulence in my simulations, could you please share your 0 folder? I assume that when you started your simulations it could have been started from 0 time. Any comment or suggestion would be welcomed, many thanks in advance.

180 is difficult to make turbulent. I recommend that you change viscosity and run at a higher Re for a bit, just to get turbulence. Than u can switch tje viscosity back. Also, be ready that when u set the p gradient instead of Ub, it takes a lobger time to reach a statistically steady state.

 December 29, 2022, 04:48 #30 Senior Member   chandra shekhar pant Join Date: Oct 2010 Posts: 216 Rep Power: 15 Hello Good morning, many thanks for your prompt reply, but is there anything I can do to trigger the turbulence by boundary condition or initial condition? To save time for getting the statistically stationery state, I am using simpleFoam to get converged results and then mapping the fields to LES solver, it this fine ? Any further help would be very much appreciated, thanks in advance.

December 29, 2022, 05:19
#31
New Member

Join Date: Jan 2014
Posts: 20
Rep Power: 11
Quote:
 Originally Posted by chandra shekhar pant Hello Good morning, many thanks for your prompt reply, but is there anything I can do to trigger the turbulence by boundary condition or initial condition? To save time for getting the statistically stationery state, I am using simpleFoam to get converged results and then mapping the fields to LES solver, it this fine ? Any further help would be very much appreciated, thanks in advance.
Dear Chandra,

you should artificially introduce turbulence to the flow. There is an utility written by Eugene de Villiers that modifies the initial velocity flow in order to trigger turbulent motion in the channel. Basically, the utility will modify the initial velocity field near the wall boundary to mimic the coherent vortex-like structures based on the paper mentioned in the utility code. I tested it and it works just fine. But have in mind that you should probably modify code depending on the distribution of OpenFoam you use in order to compile it.

Best regards.
Attached Files
 perturbU.zip (5.9 KB, 4 views)

 December 29, 2022, 12:49 #32 Senior Member   chandra shekhar pant Join Date: Oct 2010 Posts: 216 Rep Power: 15 Thanks a lot Camel for your wonderful and helpful comment. I was trying to install it, but after unziping your sent zip file, i could see that it contain Make directory, perturbU.c and perturbUDict, could you please elaborate how to install and use it. I am using OpenFOAM v1906 and/or v2112. Any further comments/suggestions would be very much appreciated, many thanks in advance.

 December 30, 2022, 06:15 #33 Senior Member   chandra shekhar pant Join Date: Oct 2010 Posts: 216 Rep Power: 15 Hi again Camel, Many thanks for your help in this forum, could you please help me to use the utility perturbU in the OF v1906/v2112. How to install/use this utility, I am still stuglling to get turbulence for the channel flow Re_\tau = 180, with specified pressure gradient condition. I managed to get the steady state solution from RANS, using that I am trying to run LES case. I used WALE scheme for LES but unfortunately seems like that is killing turbulence, thus now switching towards the KEqn model. If this utility could help, then it would save enormous amount of time, hope you understand and will help me, thanks in advance.

 Tags channel flow, les, swak4foam