CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   OpenFOAM Running, Solving & CFD (https://www.cfd-online.com/Forums/openfoam-solving/)
-   -   LES of turbulent pipe/channel flow (https://www.cfd-online.com/Forums/openfoam-solving/102567-les-turbulent-pipe-channel-flow.html)

 francesco_capuano May 28, 2012 17:31

LES of turbulent pipe/channel flow

Dear all,

I am trying to run a large-eddy simulation of turbulent pipe flow (which should not be conceptually different from a channel flow) by using channelFoam, and have encountered some problems. There are a couple of interesting threads on the forum about LES of pipes/channels, but none of them has solved my issues.

If one considers a fully-developed flow in a pipe of diameter D, I think the following steps must be performed (please correct me if I'm wrong):

1. Create a 3D mesh, with a length of about 5*D in the streamwise direction;
2. Set up cyclic boundary conditions for "inlet"/"outlet", and no-slip for wall;
3. Impose a fluctuating field for variables (i.e. U) to generate turbulence;
4. Run channelFoam;
5. Postprocess (e.g. average in time and over the circumferential direction).

My doubts are particularly concerned with steps 3 and 5. About step 3:

a. How to perturb the velocity field? The perturbU and perturbCylinder utilities seem not to work with OpenFOAM 2.1.0.
b. Is it really necessary to superimpose fluctuations? In principle, assuming a sufficiently high Reynolds number (Re > Re_tr ~ 1800), shouldn't LES be able to capture turbulence itself? (e.g., starting from instabilities of growing amplitude)

And about 5: is the postChannel utility useful for pipe flow?

Francesco

 ganeshv May 31, 2012 11:28

Reg. point 3, I think the best way to do it would be outside of OpenFOAM. I have done LES of channel flow in a rectangular channel, and I remember having imposed an initial condition with an approximate log law profile and desired variance distribution. If your mesh is large, you may want to go for this approach.

Reg. point 5, I think the best way to do it is to create cellZones with all cells at a constant radius. Then enable functionObjects to do the averaging over the cellZones. It may seem odd that you need about 50-100 cellZones, but you can script the process out and that should make it easier. The latest version of OpenFOAM can even do time averaging for you along with spatial averaging in the functionObjects.

 owayz October 12, 2012 16:41

Hallo All,
Is there any utility to know about that can be used to perform spatial averaging of fields in some specific direction.?

Regards,
Awais

 francesco_capuano October 12, 2012 17:12

Quote:
 Originally Posted by owayz (Post 386339) Hallo All, Is there any utility to know about that can be used to perform spatial averaging of fields in some specific direction.? Regards, Awais
Hi Awais,

give a look at the postChannel utility, it performs spatial averaging in the two homogeneous directions of a channel flow.

Regards,
Francesco

 All times are GMT -4. The time now is 03:37.