[CFDnewbie147]

Hello together,

I'm trying to simulate some kind of geometry in a wind tunnel with inlet, outlet and the walls of the wind tunnel.

Laminar simulations are working fine but turbulence modelling doesn't the way I like to have it. I'm using a modified rhoCentralFoam-solver with mach 0.8. For turbulence modelling I use the kOmegaSST model. My mesh is a very fine one with about 30 boundary layers around my geometry(except the walls of the wind tunnel). As I said, laminar simulations work fine and Cdrag fits with my reference value.

For k I use the following setting:

dimensions      [0 2 -2 0 0 0 0];
internalField   uniform 2.78;
boundaryField
{
    wind tunnel wall    {
        type            slip;
    }
    inlet    {
        type            turbulentIntensityKineticEnergyInlet;
        intensity       0.005;
        value           uniform 2.78;
    }
    outlet    {
        type            inletOutlet;
        inletValue      uniform 2.78;
        value           uniform 2.78;
    }
    geometry
    {
        type            compressible::kqRWallFunction;
        value           uniform 2.78;
    }
}

For omega I use the following setting:

dimensions      [0 0 -1 0 0 0 0];
internalField   uniform 55.57;
boundaryField
{
    wind tunnel wall     {
        type            slip;
    }
    inlet    {
        type            fixedValue;
        value           uniform 55.57;
    }
    outlet    {
        type            inletOutlet;
        inletValue      uniform 55.57;
        value           uniform 55.57;
    }
    geometry
    {
        type            compressible::omegaWallFunction;
        value           uniform 55.57;
    }
}

The values of k and omega are from the CFD-online-turbulence-modelling tool for a freestream velocity of 272.215m/s, 0.5% turbulence intensity level and a turbulence length scale of 0.03m.

Running this simulation i get a much higher Cdrag value as my reference value. I think the problem is the turbulence modelling at inlet and outlet. I only want to use the compressible wall function for my geometry and no turbulence modelling of the freestream around my geometry.

I've also tried some other boudary conditions at in- and outlet but most of them doesn't work. How can I solve this problem?
Does anybody understand my problem and has some ideas?
If there are any questions left, please reply and I will answer.

Best regards,
CFDnewbie147

[RodriguezFatz]

Do you resolve the boundary layer (y+=1) ?
If so, you need to set k to some low value instead of using a wall function. Maybe that is the problem.

[CFDnewbie147]

Hello Philipp,

thank you for answering.
Yes I resolve the boundary layer. So the boundary conditions for inlet and outlet are ok and I "only" have to set a low number of k for the geometry instead of the wallfunction? Low number means about 0.1 or 1 or what do you propose?

And what's about omega? How to set this value?
Best regards,
CFDnewbie147

[RodriguezFatz]

Hi,

I don't know if the inlet / outlet is fine, this depends on your case. But if you calculated these values by hand (!?) it should be ok.
I only know about incompressible boundary conditions, I don't know if the compressible differ. But in the incompressible case, omegawallfunction needs to be used for low and high-Re cases. It automatically switches from y+=1 to wall modelling. I think this also will be done in your case.
Normally k should be zero at the wall for your case, but it seems that some equation divides by "k", so it should not be zero. People in this forum usually use a very small number (such as 1e-12) for k which is effectively "0", but avoids the zero devision.
Good luck. And post again if it works.

[CFDnewbie147]

Hello again,

but this options for inlet and outlet mean that there is a modelled turbulence in the flow around my geometry, but I don't want this. I only want to use the kOmegaSST-model for the geoemtry and its boundary layers.

But I will try your suggestion.
Thank you again.
Best regards,
CFDnewbie147

[RodriguezFatz]

If you choose a turbulence model this will be used in your whole geometry. This model needs boundary conditions also at your inlet and outlet. I don't understand your problem.

[CFDnewbie147]

Ok, you're right at all.

Do you know what kind of condition is used in the background of the simulation when setting the boundary condition of k / omega to "slip"?

Best regards,
CFDnewbie147

[RodriguezFatz]

No sorry. I never used "slip", I only use fixedValue or zeroGradient.