[danvica]

Usually rasinterfoam is faster than lesinterfoam, isn't it ?

I'll give it a try. Can I keep the same solver/solution setup ?

[danvica]

Even rasInterfoam is giving bad results.
I tried with and without turbolence and also changing the div gamma schemes to gamma01 1 but the result is sooner or later a picosec simulation.

Apart from solver/solution/BC error could be some mesh problem.
It's a tet mesh unv-imported from Salome.

Can a too coarse mesh cause this problems ? Checkmesh reports "only" 14 severe non-orth faces.

Could a hex element mesh bring better results ?

Thanks to anyone.

[danvica]

I tried also reducing the inlet pressure from 1e5 to 1e4 an I've got the same results later in time.

It seems that changing BC from atmosphere to fixed pressure outlet things are going better (now in rasinterfoam without turbolence).

I'm still running the simulation so I'll see.

I would expect that atmosphere bc would be more appropriate, it's been used in the les tutorial for a similar case. Any hint ?

[hjasak]

You have messed up something else, probably boundary conditions. First, set the discretisation to first-order accuracy and see if the code runs. If it does not, you've got a clear problem (boundary conditions). When you fix it, you can then return to full LES setup.

The pico-second time-step you are seeing is the automatic time-step control trying to save you and failing. Switch it off (system/controlDict, and get the code to run - this will also show you where in the domain the failure occurs (you can dump the results every time-step and actually look at them).

Enjoy,

Hrv

[danvica]

Thanks Mr. Jasak,
Just to be sure. For first-order accuracy you mean the following (for rasinterfoam):

divSchemes
{
div(rho*phi,U) Gauss linear;
div(phi,gamma) Gauss linear;
div(phirb,gamma) Gauss linear;
div(phi,k) Gauss linear;
div(phi,epsilon) Gauss linear;
div(phi,R) Gauss linear;
div(R) Gauss linear;
div(phi,nuTilda) Gauss linear;
}

Isn't it ?
k and epsilon are disabked (turbolence off) at the moment. I don't know the use of R.

I'll try.

[danvica]

I tried with linear discretisation and fixed time step of 2us.

This is what i obtained:

Courant Number mean: 4.00383e-05 max: 4.14158
Time = 0.000628

MULES: Solving for gamma
Liquid phase volume fraction = 0.0026595 Min(gamma) = -1.79162e-36 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00265951 Min(gamma) = -6.05957e-36 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00265952 Min(gamma) = -1.79162e-36 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00265953 Min(gamma) = -4.27041e-37 Max(gamma) = 1
#0 Foam::error::printStack(Foam:stream&) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#1 Foam::sigFpe::sigFpeHandler(int) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#2 Uninterpreted: [0xb7fd8420]
#3 Foam::GaussSeidelSmoother::smooth(Foam::word const&, Foam::Field<double>&, Foam::lduMatrix const&, Foam::Field<double> const&, Foam::FieldField<foam::field,> const&, Foam::UPtrList<foam::lduinterfacefield> const&, unsigned char, int) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#4 Foam::GaussSeidelSmoother::smooth(Foam::Field<doub le>&, Foam::Field<double> const&, unsigned char, int) const in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#5 Foam::smoothSolver::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#6 Foam::fvMatrix<foam::vector<double> >::solve(Foam::Istream&) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"
#7 Foam::lduMatrix::solverPerformance Foam::solve<foam::vector<double> >(Foam::tmp<foam::fvmatrix<foam::vector<double> > > const&) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"
#8 main in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"
#9 __libc_start_main in "/lib/i686/libc.so.6"
#10 Foam::regIOobject::readIfModified() in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"

I check the very last step with Paraview but actually there's nothing strange (as far as I can understand: no value spike)

The inlet velocity is fixed to 0.4m/s on the y axis, so after 0.6ms the water is almost still unmoved from the initial position.

Just in case this are my BC

gamma:

internalField nonuniform List<scalar>
//cut//

boundaryField
{
inlet
{
type fixedValue;
value uniform 1;
}

atmos
{
type zeroGradient;
}

wall
{
type zeroGradient;
}
}

pd:

internalField uniform 0;

boundaryField
{
inlet
{
type zeroGradient;
}

atmos
{
type fixedValue;
value uniform 0;
}

wall
{
type zeroGradient;
}
}

U:
internalField uniform (0 0 0);

boundaryField
{
inlet
{
type fixedValue;
value uniform (0 0.4 0);
}

atmos
{
type zeroGradient;
}

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

nuTilda:

internalField uniform 0;

boundaryField
{
inlet
{
type fixedValue;
value uniform 0;
}

atmos
{
type zeroGradient;
}

wall
{
type zeroGradient;
}
}

If you have any idea I would appreciate, thanks

[danvica]

I'm really very sorry to insist but I'm stuck.

Even reducing the inlet velocity to approx (0,0,0) the solver stop after 6.7ms
Co value seems right.

This is the last step:

Courant Number mean: 8.72853e-06 max: 0.162675
deltaT = 9.5844e-05
Time = 0.000557064

MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -1.13282e-19 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -6.52214e-20 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -6.41784e-21 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -4.74896e-21 Max(gamma) = 1
smoothSolver: Solving for Ux, Initial residual = 0.0769033, Final residual = 8.24476e-07, No Iterations 4
smoothSolver: Solving for Uy, Initial residual = 0.0888143, Final residual = 9.62986e-07, No Iterations 4
smoothSolver: Solving for Uz, Initial residual = 0.0913065, Final residual = 1.17177e-07, No Iterations 5
GAMG: Solving for pd, Initial residual = 0.00890823, Final residual = 4.32939e-07, No Iterations 6
GAMG: Solving for pd, Initial residual = 0.00226642, Final residual = 2.57883e-07, No Iterations 5
GAMG: Solving for pd, Initial residual = 0.000503093, Final residual = 8.34177e-07, No Iterations 3
GAMG: Solving for pd, Initial residual = 0.00045618, Final residual = 6.16192e-07, No Iterations 5
GAMG: Solving for pd, Initial residual = 0.00016907, Final residual = 4.03424e-07, No Iterations 3
GAMGPCG: Solving for pd, Initial residual = 3.91883e-05, Final residual = 1.10265e-09, No Iterations 3
time step continuity errors : sum local = 3.88878e-14, global = 3.42051e-16, cumulative = 1.47553e-14
ExecutionTime = 242.19 s ClockTime = 243 s

Courant Number mean: 1.22137e-05 max: 0.217563
deltaT = 0.000115013
Time = 0.000672077

MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -7.54913e-21 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -1.04923e-20 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -5.72812e-21 Max(gamma) = 1
MULES: Solving for gamma
Liquid phase volume fraction = 0.00264396 Min(gamma) = -5.667e-21 Max(gamma) = 1
#0 Foam::error::printStack(Foam:stream&) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#1 Foam::sigFpe::sigFpeHandler(int) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#2 Uninterpreted: [0xb7f3a420]
#3 Foam::GaussSeidelSmoother::smooth(Foam::word const&, Foam::Field<double>&, Foam::lduMatrix const&, Foam::Field<double> const&, Foam::FieldField<foam::field,> const&, Foam::UPtrList<foam::lduinterfacefield> const&, unsigned char, int) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#4 Foam::GaussSeidelSmoother::smooth(Foam::Field<doub le>&, Foam::Field<double> const&, unsigned char, int) const in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#5 Foam::smoothSolver::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/lib/linuxGccDPOpt/libOpenFOAM.so"
#6 Foam::fvMatrix<foam::vector<double> >::solve(Foam::Istream&) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"
#7 Foam::lduMatrix::solverPerformance Foam::solve<foam::vector<double> >(Foam::tmp<foam::fvmatrix<foam::vector<double> > > const&) in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"
#8 main in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"
#9 __libc_start_main in "/lib/i686/libc.so.6"
#10 Foam::regIOobject::readIfModified() in "/home/caelinux/OpenFOAM/OpenFOAM-1.4.1/applications/bin/linuxGccDPOpt/rasInterF oam"

Thanks

[danvica]

I tried to reduce the complexity of the problem.

Now CheckMesh reports as follow:

Create polyMesh for time = constant

Time = constant

Mesh stats
points: 1743
edges: 9116
faces: 13509
internal faces: 11031
cells: 6135
boundary patches: 3
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 0
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 6135
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
fin 86 54 ok (not multiply connected)
fout 102 62 ok (not multiply connected)
wall 2290 1165 ok (not multiply connected)

Checking geometry...
Domain bounding box: (-0.01275 -0.0485 -0.01275) (0.01275 0.06 0.01275)
Boundary openness (-2.01305e-17 -1.76262e-17 5.19899e-17) OK.
Max cell openness = 1.30668e-16 OK.
Max aspect ratio = 6.73238 OK.
Minumum face area = 6.85693e-08. Maximum face area = 4.25408e-05. Face area magnitudes OK.
Min volume = 1.05176e-11. Max volume = 8.75739e-08. Total volume = 3.24662e-05. Cell volumes OK.
Mesh non-orthogonality Max: 62.5586 average: 21.1981
Non-orthogonality check OK.
Face pyramids OK.
Max skewness = 0.979387 OK.
Min/max edge length = 0.000327004 0.0117359 OK.
All angles in faces OK.
All face flatness OK.

Mesh OK.

The geometry consists of just some pipes with different diameters.

One fixed velocity inlet.
One fixed pressure outlet.
Wall

Solver: rasinterfoam
Scheme: all linear
Solvers: as tutorial rasinterfoam (mainly PBiCG, for pd uses PCG)

The inlet patch and a piece of the inlet pipe are gamma-1 filled.

After some iteration in which i can see (paraview) the liquid pushing the air phase into the outlet pipe it seems that something is happening at the liquid surface. Strange pressure spikes.

What am I missing ?
I changed the mesh,the solver,the scheme and the solution. It has to be related to BC then...

Thanks.

[danvica]

I also tried to exclude surface tension and interface compression so I set sigma=0 and cgamma=0.

Note: I use a coarser mesh in order to have faster results.
Check mesh reports as follow:

Create polyMesh for time = constant

Time = constant

Mesh stats
points: 2581
edges: 14500
faces: 22343
internal faces: 19349
cells: 10423
boundary patches: 3
point zones: 0
face zones: 0
cell zones: 0

Number of cells of each type:
hexahedra: 0
prisms: 0
wedges: 0
pyramids: 0
tet wedges: 0
tetrahedra: 10423
polyhedra: 0

Checking topology...
Boundary definition OK.
Point usage OK.
Upper triangular ordering OK.
Topological cell zip-up check OK.
Face vertices OK.
Face-face connectivity OK.
Number of regions: 1 (OK).

Checking patch topology for multiply connected surfaces ...
Patch Faces Points Surface
fin 43 31 ok (not multiply connected)
fout 44 32 ok (not multiply connected)
wall 2907 1471 ok (not multiply connected)

Checking geometry...
Domain bounding box: (-0.01275 -0.0485 -0.0127498) (0.01275 0.06 0.0127496)
Boundary openness (-4.75469e-18 4.09572e-17 3.83941e-17) OK.
Max cell openness = 1.3167e-16 OK.
Max aspect ratio = 4.22384 OK.
Minumum face area = 2.19126e-07. Maximum face area = 2.99902e-05. Face area magnitudes OK.
Min volume = 4.8208e-11. Max volume = 5.25136e-08. Total volume = 3.24903e-05. Cell volumes OK.
Mesh non-orthogonality Max: 47.8867 average: 16.8891
Non-orthogonality check OK.
Face pyramids OK.
Max skewness = 0.679106 OK.
Min/max edge length = 0.000657036 0.00919595 OK.
All angles in faces OK.
All face flatness OK.

Mesh OK.

With automatic time step adjust limiting Co to 0.4 here is what I obtain.





Better then before but after 21ms Co starts to increase so time step reaches ps !

Has anyone an idea of what is happening ? I really hope to have done a very stupid error, otherwise I'm a "little bit" worried of so problematic fine tuning of OF.

[danvica]

Sorry, I wanted to include some movies but they're too big.

These are the pictures of the latest time step.





U values seems correct. I've got an inlet 21mm diam. cylinder at 0.4m/sec and an outlet of 5mm diam.