[sam.ho]

Hi Foamers
I want simulate the flow past circular cylinder with very high reynolds no of order 10000 and above. Could you please tell which solver can be used to handle these turbulent flows. ?????

[Artur]

That depends on what you want to study in your simulations, please be a bit more specific. Ideally, Have a look at this first:

http://www.cfd-online.com/Forums/ope...-get-help.html

Al the best,

A

[TReviol]

Hi,

I simulated such a case a few month ago.

And as expected, with RAS-Models the problem of the karman vortex street will result in relativ "bad" values. My intention was to show the phenomenon of this flow, so a RAS-Model was sufficient to me.

But if you want to compare your results with a test case, or you want to analyze drag-coefficients or something else, you should choose a better turbulence modelling. For example the DES- or even the LES-Model. but I have no experience with these models.

Have fun with the Karman vortex street :-)

Regards

Thomas

[Marty]

The standard solver can be found at: http://www.openfoam.com/features/standard-solvers.php

For turbulence models have a look at:
http://www.cfd-online.com/Wiki/Turbulence_modeling

[sam.ho]

Hi TReviol,

I would like to know which solver did u use to run RAS-models ?

[TReviol]

Hi,

I used pisoFoam. I needed different grids to find out an appropriate grid-distance. Further more the timeStep was difficult to estimate. I used an correlation of Reynolds No. and Strouhal No. (for an example follow this link). I reached good results with kOmegaSST - but only for "coarse" grids. For fine grids and kOmegaSST the flow phenomenon was overestimated.

At least I analyzed the time averaged values for p and U and compared them with an test-Case. I also repeated all my effort with fluent and CFX to demonstrate the failure off RAS-turbulence-modelling when calculating high turbulent detachment flow.

Here are my results: karman.jpg


Thomas

[sam.ho]

Hi TReviol,
I want to simulate the same case but the geometry is Circular Cylinder instead of Square cylinder for the Re no of the order 10,000 and greater and want to use simpleFOAM with k-epsilon turbulent model.
Does it give comparably good results ????

[TReviol]

Hi,
simpleFoam will not be sufficient, because simpleFoam is a steadyState solver. You have tho solve transient flow phenomenons.

Regards Thomas

[sam.ho]

Hi
I used simpleFOAM with standard k-epsilon model for Flow past circular cylinder .
Re = 10^6
Velocity = 10 m/s
Kinematic Viscosity = 10^-5
Walls = slip condition
But y+ value is going far beyond.
Can anybody help me how to get rite value of y+ ??

[sam.ho]

Hi
Kindly find the attached 0 file for reference

[TReviol]

Hi,

in my case, the front and back faces were set as a cyclic-conditions. The other initial conditions of my case look similar to your conditions. But maybe a better choice is necessary for better agreement with empirical Data (e.g. inletOutlet and totalPressure conditions).

Yor question about y+ values may be answerde with this link.

Regards

Thomas

[sam.ho]

Hi TReviol,

Could u please share your pimpleFOAM case set up ?

Regards,
Sam Ho

[sam.ho]

Hi Foamers,

I am stuck up with specifying Boundary conditions like
K, epsilon, nut, nuTilda for the flow past cylinder with following conditions...
Kinematic viscosity = 10^6 m2/s
Re = 2.5 x 10^5
free stream velocity = 0.25 m/s

Kindly help me

[TReviol]

Hi,

have a look at this link.

But choose your characteristic length carefully. I always initialize with the "Fluent" formulas. Fluent scales the characteristic length with a factor. I think the factor is 0.07, but I am not sure.

In many cases, you doesn't need to initalize the turbulent viscosity. So maybe you can delete the file nut (and even nuTilda, when you have choosen kEpsilon).

Thomas

[sam.ho]

Thank you TReviol
I will set up the case with these information and if any problem i will post on this link.
Kindly help me in that case

[sam.ho]

Originally Posted by TReviol

Hi,

have a look at this link.

But choose your characteristic length carefully. I always initialize with the "Fluent" formulas. Fluent scales the characteristic length with a factor. I think the factor is 0.07, but I am not sure.

In many cases, you doesn't need to initalize the turbulent viscosity. So maybe you can delete the file nut (and even nuTilda, when you have choosen kEpsilon).

Thomas

Hi
Please look in to the following link and help me to resolve the issue.
http://www.cfd-online.com/Forums/openfoam-solving/126061-turbulrnt-flow-past-cylinder-new-post.html