[bendel_boy]

Using settlingFoam I have solids entering a tank at 3,500 mg/l. In theory the solids should settle on the floor and concentrate, but this never happens. Every so often during the simulation the velocities shoot off to high values and (presumably) entrain solids, so that the post-processing view looks as if the solids have vanished.

Any suggestions on suitable boundary conditions that may control this, or any thoughts on what may lie behind this?

[bendel_boy]

Here is 0/U

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \\ / O peration | Version: 2.1.x |
| \\ / A nd | Web: www.OpenFOAM.org |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
version 2.0;
format ascii;
class volVectorField;
object U;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

// This is used to set the inlet velocity as well as the initial velocities.

// Inlet is 1 m high, 1/8 of tank radius, and radius is assumed to be 1 m
// So area is 1 x 2 PI x 1 / 8 = 0.78 m2
// Velocity = 96.5 / 0.78 / 3600 = 0.0341
dimensions [0 1 -1 0 0 0 0];
internalField uniform (0.031 0.0031 0);

// X Y Z
boundaryField
{
inlet
{
type fixedValue;
value uniform (0 0.0341 0);
}
outlet
{
// Imposing a uniform velocity
// Area = 2 PI 6.6 / 8 x 0.15 = 0.78
// Velocity = (96.5 - 54.7) / 3600 / 0.78 = 0.0149
type fixedValue;
value uniform (0 -0.0149 0);
}
sludge
{
// 54.7 m3 withdrawn through a sector 0.5 m deep, radius 1.5 m, of 1/8 radial
// So surface area = PI (1.5 x 1.5 - 1 x 1) / 8 = 0.49 m2
// Velocity = 54.7 / 3600 / 0.49 = 0.031 m/s
type fixedValue;
value uniform (0 0 -0.031);
}
defaultFaces // walls
{
type fixedValue;
value uniform (0 0 0);
}
surface
{
type slip;
}
}

// ************************************************** *********************** //

[cutter]

Hi,

I checked your velocity boundary conditions, I'm basically using the same setup. They work just fine for rectangular tanks (Did you check rectangular tanks already?!). I'm also observing similar problems for round tanks with cyclic meshes tough. The main problems might be the relatively high velocities in the inlet region, which unfortunately contains the worst cells (problems regarding CFL condition and mesh quality).

Maybe you can share your pressure file too. Another frequent cause for trouble is the turbulence model.

You may want to have a look at driftFluxFoam, which replaced settlingFoam in OF 2.3.1 http://www.openfoam.org/version2.3.1/.

I'm highly interested in this issue, I will dig into into it again as soon as there's more time...

Cutter

[bendel_boy]

Pressure field of 0/p_rgh

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \\ / O peration | Version: 2.1.x |
| \\ / A nd | Web: www.OpenFOAM.org |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
version 2.0;
format ascii;
class volScalarField;
object p_rgh;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
dimensions [1 -1 -2 0 0 0 0];
internalField uniform 0;
boundaryField
{
inlet
{
type fixedFluxPressure;
value uniform 0;
}
outlet
{
type fixedValue;
value uniform 0;
}
sludge
{
type fixedFluxPressure; // fixedValue;
value uniform 0;
}
surface
{
type fixedFluxPressure;
value uniform 0;
}
defaultFaces
{
type fixedFluxPressure;
value uniform 0;
}
}
// ************************************************** *********************** //

[bendel_boy]

I had a rectangular tank run fine, including some mods to include a sludge take-off.

I had a circular conventional tank run 'fine', except that the baffle seemed to retard solids exchange between the feedwell and the main tank.

I am now trying to look at this rather unusual tank, where the feed enters right at the floor, and the overflow is only on a small part of the periphery.

I have increased my grid mesh by a factor of 8 - doubld in X, Y, Z - and a running now. Early results look more promising, but I ran out of disk space so have t oclear things and rerun. At the moment I have yet to reach the fall-over point from last night.

The model predicts a strong rotational velocity. The site measurements hadea weak rotational velocity. That may mean the model needs longer to settle down - the site measurements showed a deeper, more uniform, sludge blanket than the model is predicting.

[bendel_boy]

For turbulence I have tried the RNGKEpsilon as well as Daniel Brennan's default, kEpsilon. No success.

I also tried switching the outlet solids concentration estimates from fixed value $internalfField to zeroGradient - again, didn't make things better.

I couldn't see what had changed in the driftFluxFoam model other than internal matters to simplify time-varying conditions. At the moment I am using a Windows port, so would need to see if I can find a Windows port of driftFluxFoam.

[bendel_boy]

More changes made:

fvSchemes - from Euler to backward

ddtSchemes
{
default backward; // Euler;
}

Implicit schemes usually (not always) have better stability properties

fvSolution

Set non-orthognal sweeps in PIMPLE, and reduced relaxation factors

PIMPLE
{
nCorrectors 3;
nNonOrthogonalCorrectors 3; // 0;
}

// try reducing the relaxation factors. Were 1.0
relaxationFactors
{
fields
{
}
equations
{
"Alpha.*" 0.1; // 1;
"U.*" 0.1; // 1;
"k.*" 0.1; // 1;
"epsilon.*" 0.1; //1;
}
}

Will post if the result seems more stable.

[RodriguezFatz]

Can you post some iterations of the log file and the fvSchemes file you use?

[bendel_boy]

fvSchemes file

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \\ / O peration | Version: 2.1.x |
| \\ / A nd | Web: www.OpenFOAM.org |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
version 2.0;
format ascii;
class dictionary;
location "system";
object fvSchemes;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

ddtSchemes
{
default backward; // Euler;
}

gradSchemes
{
default Gauss linear;
}

divSchemes
{
default none;
div(phi,U) Gauss linearUpwind grad(U);
div(phi,k) Gauss upwind;
div(phi,epsilon) Gauss upwind;
div(phiAlpha,Alpha) Gauss limitedLinear01 1;
div(phiVdj,Vdj) Gauss linear;
}

laplacianSchemes
{
default none;
laplacian(muEff,U) Gauss linear corrected;
laplacian(Dp,p_rgh) Gauss linear corrected;
laplacian(DkEff,k) Gauss linear corrected;
laplacian(DepsilonEff,epsilon) Gauss linear corrected;
laplacian(mut,Alpha) Gauss linear corrected;
}

interpolationSchemes
{
default linear;
}

snGradSchemes
{
default corrected;
}

fluxRequired
{
default no;
p_rgh ;
}


// ************************************************** *********************** //

[bendel_boy]

log file goes to screen.

Here is a capture

diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG: Solving for Ux, Initial residual = 1.49901e-005, Final residual = 1.89267e-009, No Iterations 1
DILUPBiCG: Solving for Uy, Initial residual = 7.66126e-006, Final residual = 1.87921e-010, No Iterations 1
DILUPBiCG: Solving for Uz, Initial residual = 7.28147e-006, Final residual = 8.4795e-011, No Iterations 1
DILUPBiCG: Solving for Alpha, Initial residual = 1.42366e-005, Final residual = 2.17146e-012, No Iterations 1
Solid phase fraction = 0.000283384 Min(Alpha) = 1.50339e-006 Max(Alpha) = 0.00711959
DICPCG: Solving for p_rgh, Initial residual = 4.62563e-005, Final residual = 4.3753e-007, No Iterations 5
DICPCG: Solving for p_rgh, Initial residual = 5.20592e-007, Final residual = 6.45966e-008, No Iterations 6
DICPCG: Solving for p_rgh, Initial residual = 7.08292e-008, Final residual = 7.08292e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 7.08292e-008, Final residual = 7.08292e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.58806e-013, global = 5.71781e-014, cumulative = 1.01284e-009
DICPCG: Solving for p_rgh, Initial residual = 8.99792e-006, Final residual = 8.82751e-008, No Iterations 3
DICPCG: Solving for p_rgh, Initial residual = 9.20341e-008, Final residual = 9.20341e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 9.20341e-008, Final residual = 9.20341e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 9.20341e-008, Final residual = 9.20341e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.51114e-013, global = 5.42644e-014, cumulative = 1.01289e-009
DICPCG: Solving for p_rgh, Initial residual = 4.2205e-005, Final residual = 3.3756e-007, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 3.44318e-007, Final residual = 7.11982e-008, No Iterations 4
DICPCG: Solving for p_rgh, Initial residual = 7.23193e-008, Final residual = 7.23193e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 7.23193e-008, Final residual = 7.23193e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.62275e-013, global = 5.74859e-014, cumulative = 1.01295e-009
DILUPBiCG: Solving for epsilon, Initial residual = 8.34881e-007, Final residual = 2.083e-011, No Iterations 1
DILUPBiCG: Solving for k, Initial residual = 2.79616e-006, Final residual = 1.68162e-011, No Iterations 1
ExecutionTime = 1204.07 s ClockTime = 1204 s

Courant Number mean: 0.0129684 max: 0.496445
deltaT = 0.8
Time = 17699.99999999998

diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG: Solving for Ux, Initial residual = 1.49893e-005, Final residual = 2.06898e-009, No Iterations 1
DILUPBiCG: Solving for Uy, Initial residual = 7.66121e-006, Final residual = 1.87621e-010, No Iterations 1
DILUPBiCG: Solving for Uz, Initial residual = 7.27917e-006, Final residual = 8.44669e-011, No Iterations 1
DILUPBiCG: Solving for Alpha, Initial residual = 1.42362e-005, Final residual = 2.17124e-012, No Iterations 1
Solid phase fraction = 0.000283397 Min(Alpha) = 1.50362e-006 Max(Alpha) = 0.00711982
DICPCG: Solving for p_rgh, Initial residual = 6.47292e-005, Final residual = 6.41627e-007, No Iterations 3
DICPCG: Solving for p_rgh, Initial residual = 6.97173e-007, Final residual = 8.94536e-008, No Iterations 6
DICPCG: Solving for p_rgh, Initial residual = 9.73816e-008, Final residual = 9.73816e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 9.73816e-008, Final residual = 9.73816e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.3535e-013, global = 5.60918e-014, cumulative = 1.01301e-009
DICPCG: Solving for p_rgh, Initial residual = 2.06915e-005, Final residual = 2.03682e-007, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 2.09947e-007, Final residual = 9.31045e-008, No Iterations 3
DICPCG: Solving for p_rgh, Initial residual = 9.26247e-008, Final residual = 9.26247e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 9.26247e-008, Final residual = 9.26247e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.41472e-013, global = 4.92273e-014, cumulative = 1.01306e-009
DICPCG: Solving for p_rgh, Initial residual = 5.0785e-005, Final residual = 2.41016e-007, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 2.57421e-007, Final residual = 7.44953e-008, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 7.48202e-008, Final residual = 7.48202e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 7.48202e-008, Final residual = 7.48202e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.44326e-013, global = 5.55206e-014, cumulative = 1.01311e-009
DILUPBiCG: Solving for epsilon, Initial residual = 8.35323e-007, Final residual = 2.09142e-011, No Iterations 1
DILUPBiCG: Solving for k, Initial residual = 2.7965e-006, Final residual = 1.68291e-011, No Iterations 1
ExecutionTime = 1204.2 s ClockTime = 1204 s

Courant Number mean: 0.0129684 max: 0.496452
deltaT = 0.8
Time = 17700.79999999998

diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG: Solving for Ux, Initial residual = 1.49886e-005, Final residual = 1.88993e-009, No Iterations 1
DILUPBiCG: Solving for Uy, Initial residual = 7.66025e-006, Final residual = 1.88532e-010, No Iterations 1
DILUPBiCG: Solving for Uz, Initial residual = 7.28453e-006, Final residual = 8.49764e-011, No Iterations 1
DILUPBiCG: Solving for Alpha, Initial residual = 1.42358e-005, Final residual = 2.17104e-012, No Iterations 1
Solid phase fraction = 0.00028341 Min(Alpha) = 1.50386e-006 Max(Alpha) = 0.00712006
DICPCG: Solving for p_rgh, Initial residual = 5.48198e-005, Final residual = 3.35715e-007, No Iterations 6
DICPCG: Solving for p_rgh, Initial residual = 4.46005e-007, Final residual = 8.90205e-008, No Iterations 18
DICPCG: Solving for p_rgh, Initial residual = 1.18803e-007, Final residual = 5.71436e-008, No Iterations 1
DICPCG: Solving for p_rgh, Initial residual = 5.73297e-008, Final residual = 5.73297e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.53393e-013, global = 8.99084e-014, cumulative = 1.0132e-009
DICPCG: Solving for p_rgh, Initial residual = 8.54944e-006, Final residual = 5.55412e-008, No Iterations 4
DICPCG: Solving for p_rgh, Initial residual = 5.96285e-008, Final residual = 5.96285e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 5.96285e-008, Final residual = 5.96285e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 5.96285e-008, Final residual = 5.96285e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.58265e-013, global = 9.27224e-014, cumulative = 1.0133e-009
DICPCG: Solving for p_rgh, Initial residual = 3.67089e-005, Final residual = 1.5646e-007, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 1.70786e-007, Final residual = 8.22425e-008, No Iterations 1
DICPCG: Solving for p_rgh, Initial residual = 8.24296e-008, Final residual = 8.24296e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 8.24296e-008, Final residual = 8.24296e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.55074e-013, global = 7.79002e-014, cumulative = 1.01337e-009
DILUPBiCG: Solving for epsilon, Initial residual = 8.35766e-007, Final residual = 2.09985e-011, No Iterations 1
DILUPBiCG: Solving for k, Initial residual = 2.79683e-006, Final residual = 1.685e-011, No Iterations 1
ExecutionTime = 1204.33 s ClockTime = 1205 s

Courant Number mean: 0.0129685 max: 0.496459
deltaT = 0.8
Time = 17701.59999999998

diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
DILUPBiCG: Solving for Ux, Initial residual = 1.4989e-005, Final residual = 2.01153e-009, No Iterations 1
DILUPBiCG: Solving for Uy, Initial residual = 7.6609e-006, Final residual = 1.8824e-010, No Iterations 1
DILUPBiCG: Solving for Uz, Initial residual = 7.27441e-006, Final residual = 8.65466e-011, No Iterations 1
DILUPBiCG: Solving for Alpha, Initial residual = 1.42353e-005, Final residual = 2.17084e-012, No Iterations 1
Solid phase fraction = 0.000283422 Min(Alpha) = 1.50409e-006 Max(Alpha) = 0.00712029
DICPCG: Solving for p_rgh, Initial residual = 6.05046e-005, Final residual = 5.07212e-007, No Iterations 4
DICPCG: Solving for p_rgh, Initial residual = 5.86544e-007, Final residual = 6.98369e-008, No Iterations 20
DICPCG: Solving for p_rgh, Initial residual = 8.74332e-008, Final residual = 8.74332e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 8.74332e-008, Final residual = 8.74332e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.59385e-013, global = 7.0763e-014, cumulative = 1.01344e-009
DICPCG: Solving for p_rgh, Initial residual = 1.44399e-005, Final residual = 1.04184e-007, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 1.09629e-007, Final residual = 7.29751e-008, No Iterations 1
DICPCG: Solving for p_rgh, Initial residual = 7.30355e-008, Final residual = 7.30355e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 7.30355e-008, Final residual = 7.30355e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.52062e-013, global = 7.57142e-014, cumulative = 1.01352e-009
DICPCG: Solving for p_rgh, Initial residual = 3.14938e-005, Final residual = 2.68487e-007, No Iterations 2
DICPCG: Solving for p_rgh, Initial residual = 2.75149e-007, Final residual = 9.70101e-008, No Iterations 3
DICPCG: Solving for p_rgh, Initial residual = 9.79657e-008, Final residual = 9.79657e-008, No Iterations 0
DICPCG: Solving for p_rgh, Initial residual = 9.79657e-008, Final residual = 9.79657e-008, No Iterations 0
diagonal: Solving for rho, Initial residual = 0, Final residual = 0, No Iterations 0
time step continuity errors : sum local = 1.55699e-013, global = 7.83489e-014, cumulative = 1.0136e-009
DILUPBiCG: Solving for epsilon, Initial residual = 8.36205e-007, Final residual = 2.10839e-011, No Iterations 1
DILUPBiCG: Solving for k, Initial residual = 2.79715e-006, Final residual = 1.68518e-011, No Iterations 1
ExecutionTime = 1204.52 s ClockTime = 1205 s

Courant Number mean: 0.0129686 max: 0.496466
deltaT = 0.8
Time = 17702.39999999998

[bendel_boy]

Not sure if it is the use of the backward integration scheme, or setting PIMPLE to use non-orthogonal correctors - the circular geometry means that the grid has to be non-orthogonal - but at long last I am getting something that does not show the periodic destruction (resuspension) of the settled solids.

The velocity field also looks more like what I measured.

[RodriguezFatz]

Hi, so the log file / residuals look fine, is that what you think, too?

[bendel_boy]

Now, yes.

Yesterday, I had one of two outcomes. Either the solids profile looked reasonable - but the solver failed as internal values climbed towards infinity. Or the solids profile periodically vanished (resuspended throughout the tank, when the velocity blipped to large values before returning to the lower 'normal' values), but the solver then failed only when I ran out of disk space.

[RodriguezFatz]

What means the "rho=0 residual" equation you solve for? Why do you solve an "all zero" equation at all?

[bendel_boy]

For rho the residual is zero, not the density. So the calculated density from one time step to another has zero errors, and hence no need for iterations to converge.

The solids field is changing with time, but as a volume-averaged and normalised result this could well result in a negligible change with time - I can see that the progression is slow, with a stable sludge blanket building up.

[RodriguezFatz]

Yes, I know, but all zero residuals (not nearly zero but exactly zero) only comes up, if the value doesn't change at all. Thus, for real constants. Is "rho" a constant value? If not, it might be initialized uncorrectly.

[bendel_boy]

rho should not be a constant value. I presume that rho is water density x (1 - volume fraction) + solid density x volume fraction, where water density is 996 and solid density is 1042.

I don't have the progress results sent to a log file, so I can't see if at the start the density equation was varying rapidly. I'm assuming that the small change in solids is why the predictor is regarded as exact, and the residual ends up as exactly zero. But how OpenFoam calculates the rho residual I don't know - I haven't looked, to be honest.

At the start of the simulation the tank contains only water, so rho @ t=0 = 996.

I could try increasing the solids density to 1300 and see what happens.

[RodriguezFatz]

In my experience, a small change results in residuals that are very low, but never in zero.

What I mean: See the SpalartAllmaras turbulence model for example. If you unintentionally set nuTilda to zero in your domain for initial conditions, you will never get any turbulence and a "residual=0" for all time steps. This is, because in nuTilda equation the production of turbulence is proportional to nuTilda. So if it is zero at t=0 it will remain zero for all time. Maybe this happens here?

[bendel_boy]

rho is not an explicit initial condition. alpha (volume fraction of solids) is the initial condition.

Here is my /0/alpha file

Looking at it, I can do with re-instating the old b.c. for sludge/outlet. I will try with the internal field set to non-zero.


FoamFile
{
version 2.0;
format ascii;
class volScalarField;
object Alpha;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

dimensions [0 0 0 0 0 0 0];
internalField uniform 0;

// alpha is volumetric ratio of solids.
// To convert to concentrations multuply by solids density. Gets solids in kg/m3 -> g/l.
// In mg/l units multiply by a further 1000. Assume 3,500 mg/l MLSS -> 0.00336 a a volume fraction, using
// solids density as 1.042 g/cm3
boundaryField
{
inlet
{
type fixedValue;
value uniform 0.00269;
}
outlet
{
type zeroGradient; //inletOutlet;
//inletValue $internalField;
}
sludge
{
type zeroGradient; //inletOutlet;
//inletValue $internalField;
}
surface
{
type zeroGradient;
}
defaultFaces
{
type zeroGradient;
}
}

// ************************************************** *********************** //

[RodriguezFatz]

Can you post a picture of your domain with some explanations?