# Different flow pattern between OpenFOAM and CFX

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

January 12, 2010, 07:08
Different flow pattern between OpenFOAM and CFX
#1
Member

Join Date: Sep 2009
Posts: 45
Rep Power: 16
Dear Foamers,

I'd like to validate OpenFOAM by using CFX. I use the same mesh and boundary conditions in both software. The same Reynolds number of 300,000. The LES Smagorinsky in CFX using the default damping function (with Cs=0.1) and Scalable as wall function; The LES Smagorinsky in OpenFOAM is shown as:

LESModel Smagorinsky;

delta cubeRootVol;

printCoeffs on;

cubeRootVolCoeffs
{
deltaCoeff 1;
}

vanDriestCoeffs
{
delta cubeRootVol;
cubeRootVolCoeffs
{
deltaCoeff 1;
}

smoothCoeffs
{
delta cubeRootVol;
cubeRootVolCoeffs
{
deltaCoeff 1;
}

maxDeltaRatio 1.1;
}

Aplus 26;
Cdelta 0.158;
}

smoothCoeffs
{
delta cubeRootVol;
cubeRootVolCoeffs
}

Where can I modify the Cs in OpenFOAM?
Moreover, I don't use any wall functions.
The inlet used in CFX is: Normal Speed 1m/s
The inlet used in OpenFOAM is:
INLET
{
type turbulentInlet;
fluctuationScale (0.01 0.01 0.01);
referenceField uniform (1 0 0);
alpha 0.1;
value uniform (1 0 0);
}
I created a cyclic patch in OpenFOAM to match the periodic BC of CFX.
I set pointSync to false and matchTolerance to 0.2.
Regarding the discretiztion schemes:
in CFX: Advection Scheme: Central Difference.
Transient Scheme: Second Order Backward Euler
in OpenFOAM: ddtScheme: backward
others: Gauss linear

The flow pattern is not the same: in CFX, no separation occurs on the top of the cylinder, which is good for a Reynolds number of 300,000.However, in OpenFOAM, a separation with no reattachment can be seen as if the Reynolds number was less enough from 300,000. I do not know where this difference can come from.
If anyone see something I should change in my setting, it would be great .

Thanks a lot,
AirS.
Attached Images
 case.jpg (9.2 KB, 132 views)