[grtabor]

Ive done quite a lot with a PhD student; using wavelet and fourier series with UO processes to synthesise inlet conditions; and to use the internal mapping techniques (already implemented in OpenFOAM) to do the same sort of thing. If you email me at

g.r.tabor@ex.ac.uk

I'll send you a couple of papers on our work.

Gavin

[marhamat]

Hi Ville

Thanks alot

Marhamat

[marhamat]

Hello evrybody

Did anybody simulate turbulent pipe flow or turbulent channel flow in OF using with LES?
I made some effort in this field .But the results aren't correct.
Please explan me your exprience in this field

Best Regards
Marhamat

[ville]

Hi,
I did some experimenting lately and my experience is that making a pipe flow simulation (getting a turbulent flow situation) depends on whether you want to drive the system into turbulent starting
from the boundary layer at the walls takes long
time (which is the physics). If you just want to produce turbulence i.e. for e.g. collecting data for making a boundary
condition then you can try to initialize the velocity field with some function that will develop into turbulent after a while i.e. pretty
soon. If you want to do the former one there
is no way to avoid long simulation times but
if you want to go a quicker way you can try for instance a cyclic square channel with length=6*diameter
and initialize the field properly. You can
try for instance setting the field according to some function in a region in the middle and 0 elsewhere. You can mail me
if you need extra instructions (I'm interested
in the topic as well!)

Regards,
Ville

[panara]

Dear all and in particular Eugene,

I have used perturbCyl with success in OF1.2.
Now I am trying to use it with OF1.3 and I cannot compile it..

Does anyone have a version of perturbCyl for OF1.3?

Thanks

Daniele

[eugene]

Here is a more generalised implementation for perturbing any ducted flows. I haven't tried it on cylinders though so please let me know if it works ok.

perturbU.tgz

You will need entries for dimentsionedVector Ubar and dimensionedScalar Retau in the transportProperties dictionary. The exact value of Retau (shear velocity based Re) is not critical though.

[maka]

generating 3D mesh for a pipe LES.

how can we generate a good quality mesh for 3D LES of a pipe flow. I'm familiar with blockMesh. Thanks.

Best regards,
Maka

[alberto]

There's a script released by Rasmus here:

http://www.cfd-online.com/OpenFOAM_D...es/1/3249.html

Regards,
Alberto

[maka]

Thanks Alberto.

Best Regards,
Maka

[cedric_duprat]

Hi all,
questions about the upper perturbU from Eugene.
what is the wallDistReflexion to define yDir?
it means the wall normal direction (because X is streamwise direction), with the (.n()) ?
cause I also want to use it for a pipe and , I would like to ckeck if we can use this tool for that.

an other question more general, which kind of numerics we can use not to dissip these fluctuations...I afraid that upwind scheme won't work there so ... centred one in OF?

thanks
Cedric

[eugene]

You want to use CD for convection of U (linear).

reflexVec.n() returns the direction from the cell centre to the closest wall.

I rewrote perturbU to be more general, but it is bugged atm so I wont post it.

[cedric_duprat]

Hi Eugene,
Yes, I plan to superimpose on my inlet profile a noise to my laminar inlet.
I notice that the initial law in your PerturbU is for channel (Schoppa & Hussain) so I want to test it to my axisymetric geometry.
If it doesn't work, I found other noise like in (Balarac Metais Phy. of Fluids 2005) which seems to be interresting and it was written for a turbulent jet flow.
well, Thanks for the .n(), it will be usefull and I will try to understand your code well now and re-write it for my case.
Thanks for replying,

Cedric

[cedric_duprat]

Hi all,
I am doing calculation with periodic channel which is working well.
Now, I try to do the same thing with a periodic pipe. I use a mesh from Gambit and createPatch to make cyclic boundary condition.
In my calculations, the Courant number and the velocity are still increasing.
I changed my viscosity to reduce the Re but, it is still increasing.
I also change the numerical schemes (to more diffusive) but ... still not working.
My initial field is a turbulent profile, I checked it with paraFoam.
Maybe, the cyclic Patch is the problem but, I'm not sure.
Well, now, I have no more idea. Maybe someone did this calculation before ....Thanks for helping

Regards,
Cedric

Hi Cedric,

Maybe I am answering too late but anyway...
Which solver are you using?
My suggestion would be, if you have not done it yet,
to run the case with a laminar flow just to make sure the BCs are ok. Then you could shift to a RANS turbulent model to see the impact of the Re.

Hope it helps

Pierre

[armin_h]

Hello all

would you please any one let me know how can i push flow into the pipe in openfoam?
I have created the geometry of the pipe and also the mesh.
I need to see the velocity profile which comes from the moving flow into the pipe.
I am new user of openfoam and i dont know how can i put the amounts for the u velocity and see the motion of the fluid inside the pipe.

I would appreciate any comments.

sorry if it is not an advance question.

Cheers
Armin

[johndeas]

Is it normal that perturbU leaves the y component of the laminar flow unaffected ?

[cedric_duprat]

Hi John,

In perturbU, the y component is the wallReflection vectors.
if you look at the code from Eugene or the main article of this method (Schoppa & Hussain, "Coherent structure dynamics in near-wall turbulence", Fluid Dynamics Research, Vol 26, p 119-139, 2000), we only add streaks in two direction; the streamwise and the spanwise perturbation.

Regards,

Cedric

[cedric_duprat]

Hi,

I've got questions about perturbU parameter to initialise a periodic LES pipe flow.
I'd like to know if someone try to do that. I did a first try with a 10% variation, and perturbations die out, then I encrease to 20% with the same result.
I didn't play with the wave number value yet but if someone succed I'll be greatfull to get his parameters' value.

I know that an other method is to mapFields a previous result but, there is no turbulent pipe flow result avaiable (I think). So, I started from the channel395 (channelOodles tutorial) without any success.
I guess, map field works with i,j,k intercoordinate. My mesh look like Rasmus code one (buterfly mesh) which may not work with mapField (only one fourth of the geometry).
Here again, if someone have tricks, they will be usefull for people, .... and me :o)

Regards,

Cedric

[cedric_duprat]

Good morning OpenFOAMers,

I over pass the previous problem. Here is the method :
I haven't tryed to use mapfield anymore and I started from a perturbed parabolic profile (perturbU) :o) so, it's longer but, it's working fine.(See pictures, streaks visualized by vorticity)



If you start directly with a fine mesh, the flow will become laminar and the fluctuations will disappear.
So I started from a coarse mesh, made it turbulent and mapped it into a fine one.

Now, I plan to postprocess the result and I'd like to know if there is a "postChannel" for periodic LES pipe ? or if a postChannel with one periodic direction will be enough ?
If some one has experience of this topic, I'm interesting to that.

Regard,

enjoy this sunny week-end,

Cedric

[Tarak]

Hii Ville,

Can you please let me know how did you provide the initial perturbations, like gaussian noise etc? I am looking for such methods as I need to use them for doing LES of rearward facing step.

Thanks,
Tarak