[Gearb0x]

Hello,

I'm currently trying to run a simpleFoam case with & without the use of wall functions.

The problem is : I fix the pressure drop to match a certain mass flow. Here I fix it to 30 so : p_inlet = fixedvalue 30, p_outlet= fixedvalue 0

The simulation seems to run fine on one geometry but not on the other and I don't understand why...

The two geometries are : One with "short legs" (see pictures below) and one with "long legs" to have a developed flow.

The geometry with short legs give the same results with the LowRe model (without wallfunction) and the HighRe model (with wallfunctions) :


Pfield without wallfunction (LowRe)

http://yfrog.com/iypressurefieldp

Ufield without wallfunction (LowRe)

http://yfrog.com/i3velocityfieldp

Pfield with wallfunction

http://yfrog.com/jvpressurefieldp

Ufield with wallfunction

http://yfrog.com/i3velocityfieldcp

Whereas the geometry with long leg gives weird results (numerical values & flow geometry) for the pressure with the two models :

Pfield (LowRe)
http://yfrog.com/3mpressurefieldp
Ufield (LoweRe)
http://yfrog.com/58velocityfieldp

Pfield(HighRe)
http://yfrog.com/58pressurefieldp
Ufield(HighRe)




Can someone give me a hint where I could go wrong with my simulation and why do I get such changes from just changing the geometry?

Thank you very much.

[Thomas Baumann]

Hi GearB0x,

could you upload your boundary conditions? Espacially what are your velocity bc's at the inlet and outlet.

Regards Thomas

[Gearb0x]

Here are my Boundary Conditions for the LowRe model (without wall functions) :

For p :

dimensions      [0 2 -2 0 0 0 0];

internalField   uniform 0;

boundaryField
{
    inlet           
    {
        type            fixedValue;
    value        uniform 30;
    }

    outlet          
    {
        type            fixedValue;
        value           uniform 0;
    }

    fixedWalls       
    {
        type            zeroGradient;
    }

    frontAndBack    
    {
        type            empty;
    }
}

For u :

dimensions      [0 1 -1 0 0 0 0];

internalField   uniform (0 0 0);

boundaryField
{
    inlet           
    {
        type            zeroGradient;
    }

    outlet          
    {
        type            zeroGradient;
    }

    fixedWalls       
    {
        type            fixedValue;
        value           uniform (0 0 0);
    }

    frontAndBack    
    {
        type            empty;
    }
}

For nut :

dimensions      [0 2 -1 0 0 0 0];

internalField   uniform 0;

boundaryField
{
    inlet
    {
        type            calculated;
        value           uniform 0;
    }
    outlet
    {
        type            calculated;
        value           uniform 0;
    }
    fixedWalls
    {
        type            zeroGradient;
    }
    frontAndBack
    {
        type            empty;
    }
}

For k :

dimensions      [0 2 -2 0 0 0 0];

internalField   uniform 0.3328;

boundaryField
{
    inlet
    {
        type            fixedValue;
        value           uniform 0.3328;
    }
    outlet
    {
        type            zeroGradient;
    }
    fixedWalls
    {
        type            fixedValue;
    value        0.00000001;
    }
    frontAndBack
    {
        type            empty;
    }
}

for epsilon :

dimensions      [0 2 -3 0 0 0 0];

internalField   uniform 0.3676;

boundaryField
{
    inlet
    {
        type            fixedValue;
        value           uniform 0.3676;
    }
    outlet
    {
        type            zeroGradient;
    }
    fixedWalls
    {
        type            fixedValue;
    value        0.00000001;
    }
    frontAndBack
    {
        type            empty;
    }
}

Thank for your reply and your help

[Thomas Baumann]

Hi Gearb0x,
I think for the nut at the wall you should use calculated, too. But that shouldn't be the problem.
How good are the residuals?

[Gearb0x]

Hello,

I'll try with calculated and see if that changes anything, I'll let you know.

Residuals were around 10^-5 if I remember. How can I check/plot them? I tried to run the simulation for 3000, 6000, 10000 and it doesn't seem to reach a steady state for the Low-Re model. The High Re achieve steady state.

[Gearb0x]

I've tried with nut = calculated @wall

it still gives weird results, even wost because now my pressure field is totally symmetrical ... :s

Is it possible that the mesh is too/not enough fine? Could that give wrong results?

[Thomas Baumann]

Hi Gearbox,

start a simulation with standard bc's for turbulent incompressible flow.

Inlet: U,k,epsilon field fixedValues, p zeroGradient
Outlet: U,k,epsilon zeroGradient, p fixedValue

nut everywhere calculated for low-reynolds-models

and look if everything works here fine...to be sure your mesh is okay.

best regards

[Gearb0x]

Thanks for the hint!

Apparently the mesh was the problem since now I get "nice" results

In fact, the grading was wrong : it switched between "fine" and "coarse" part too abruptly.

Best regards

[randolph]

Hey,

Just curious which low Re k epsilon formula does this setting enable?

Thanks!