[Mirage12]

I would like to compare the Turbulence-models in OpenFOAM and Fluent.

I used the same mesh and the same boundaries conditions, for this kind of comparison :

The following models were used for Re=3900 :

SpalartAllmaras
kOmegaSST
kEpsilon


Please find the cases here : http://www.workupload.com/file/jTi8N8K


But the results are not at all matching !!!

in the attachment, you find the screen shot of the velocity contours using Fluent.
Could you please explain me, what is the reason of this big difference what is wrong ? what should I change ?

[Mirage12]

here are the screen-shoots of the velocity contours in OpenFOAM.

Left picture -->K-epsilon model
Middle picture -->kOmega model
Right picture -->SpalartAllmarasmodel

[Bernhard]

I have a few remarks:

From the way you present this, it is very difficult to spot the differences: Different colorscales, different limits for the scales, different order of figures. Try to plot some extracted lines in the same figure, this makes it a lot easier to compare.

You say that you are using the same mesh for both approaches, and I believe you.

You said you are using the same boundary conditions. I assume you used wall functions. Did you confirm that the ones you used in Fluent are identical to the ones currently implemented in OpenFOAM?

Did you perform RANS or URANS simulations? Some of the look like they got transient behavior.

Did you use the same discretization on all terms?

Does this case have a solution in literature that you can compare to? I.e, give simulation and ideally experimental results?

[Mirage12]

Hello Bernhard

Thank for answer

I used RANS for the simulations.

I did not use the same schemes, because Fluent has not OpenFOAM's Schemes.
But the schemes, are not the big Problem, because changing the schemes means modifying the precision of the solution.

I am trying the validate an physical Experience. The publication was already validated using URANS and i would like to get the same results using RANS.

In the attachment,you will find streamlines, vectors, velocity-contours of the OpenFOAM's and Fluent's simulations

Link for the paper : http://astfm.tul.cz/ladmin/soubory/casopis/File/pdf/480744honzejk_frana_turbulent_flow_past_a_cylinder .pdf

As you see : the results are not at all matching \

I need your help

Thank your for your support

[eRzBeNgEl]

Hi,

well here are just some general thoughts from my side:


@Fluent:
- ke Model too dissipativ,
- kw and spalart allmaras better results but also too dissipativ. I am expecting the Karman vortex street.
=> Guessing from my side here: Implicit schemes with too large time step.


@OpenFoam:
- Also implicit schemes with wrong time step.


Quick estimation:
Strouhalnumber with Re=3900 should be around 0.2.