[huangxianbei]

Hi,all:
I want to extract the data on a cutting plane. Grid 128*128, however, the output data contains 32768 points rather than 16384, it's quite inconvenient to find which value corresponding to the specific point listed in the output data. For example(xz plane), if you range the x value from small to large, than you will find 256 x values, however, each x corresponds to 128 z values. The same to z
So I'd like to know how can I output the value only at cell center or vertex.
Thanks

[huangxianbei]

Anyone can help? Any comment is OK, I'm looking forward for the way to solve it, I have stuck with this for several days

[alexeym]

Hi,

Post your samlpleDict as from the post it's not quite clean what sampling mechanism you're using. If it is cuttingPlane, then AFAIK it's usually triangulated, that's why there's a difference between number of points in mesh and in sample output.

And if you'd like to have a table with columns |x y value|, it's quite simple to implement utility for this output yourself (though maybe there is already one).

[huangxianbei]

Originally Posted by alexeym

Hi,

Post your samlpleDict as from the post it's not quite clean what sampling mechanism you're using. If it is cuttingPlane, then AFAIK it's usually triangulated, that's why there's a difference between number of points in mesh and in sample output.

And if you'd like to have a table with columns |x y value|, it's quite simple to implement utility for this output yourself (though maybe there is already one).

Hi,Alexey:
Thank you.This is my sampleDict

setFormat raw;

surfaceFormat raw;

fields
(
    UMean
    U
);

surfaces
(
    constantPlane
    {
        type            plane;    // always triangulated
        basePoint       (0 0.041 0);
        normalVector    (0 1 0);
    }

);

So if I want to get the value whose number equals to the grids, what should I do in cutting plane?

[huangxianbei]

Originally Posted by alexeym

Hi,

Post your samlpleDict as from the post it's not quite clean what sampling mechanism you're using. If it is cuttingPlane, then AFAIK it's usually triangulated, that's why there's a difference between number of points in mesh and in sample output.

And if you'd like to have a table with columns |x y value|, it's quite simple to implement utility for this output yourself (though maybe there is already one).

thank you very much, it's really helpful. I look into the sampleDict and find all are trianulated except type-patchInternalField.

nearWalls_interpolated
    {
        // Sample cell values off patch. Does not need to be the near-wall
        // cell, can be arbitrarily far away.
        type            patchInternalField;
        patches         ( "bottomWall" );
        interpolate     true;


        // Optional: specify how to obtain sampling points from the patch
        //           face centres (default is 'normal')
        //
        //  //- Specify distance to offset in normal direction
        offsetMode  normal;
        distance    0.0407;
        //
        //  //- Specify single uniform offset
        //  offsetMode  uniform;
        //  offset      (0 0 0.0001);
        //
        //  //- Specify offset per patch face
        //  offsetMode  nonuniform;
        //  offsets     ((0 0 0.0001) (0 0 0.0002));


        // Optional: whether to leave as faces (=default) or triangulate
        // triangulate     false;
    }

Now the result is what I want.

[huangxianbei]

Originally Posted by alexeym

Hi,

Post your samlpleDict as from the post it's not quite clean what sampling mechanism you're using. If it is cuttingPlane, then AFAIK it's usually triangulated, that's why there's a difference between number of points in mesh and in sample output.

And if you'd like to have a table with columns |x y value|, it's quite simple to implement utility for this output yourself (though maybe there is already one).

I'm sorry, but the offset seems not work at all. Since I specified the offset distance(y direction), y=0 all the time.Is this means that the output x,y,z is related to the patch I specified, while the values based on (x y z) such as U at (x y z) is the offset value? Could you please help me?

[alexeym]

Hi,

in fact at least in 2.2.2 triangulation of a plane is optional:

Foam::sampledPlane::sampledPlane
(   
    const word& name,
    const polyMesh& mesh,
    const dictionary& dict
)
:
    sampledSurface(name, mesh, dict),
    cuttingPlane(plane(dict.lookup("basePoint"), dict.lookup("normalVector"))),
    zoneKey_(keyType::null),
    triangulate_(dict.lookupOrDefault("triangulate", true)),
    needsUpdate_(true)
{
...

so you can say "triangulate false;" in plane dictionary.

What are you trying to do with the data? Maybe there's another way of solving your problem.

[huangxianbei]

Originally Posted by alexeym

Hi,

in fact at least in 2.2.2 triangulation of a plane is optional:

Foam::sampledPlane::sampledPlane
(   
    const word& name,
    const polyMesh& mesh,
    const dictionary& dict
)
:
    sampledSurface(name, mesh, dict),
    cuttingPlane(plane(dict.lookup("basePoint"), dict.lookup("normalVector"))),
    zoneKey_(keyType::null),
    triangulate_(dict.lookupOrDefault("triangulate", true)),
    needsUpdate_(true)
{
...

so you can say "triangulate false;" in plane dictionary.

What are you trying to do with the data? Maybe there's another way of solving your problem.

Thank you. In fact, I want to calculate the energy spectrum in channel flow. So I need the information on a plane. I think I should try OF-2.2.2 to verify the plane data I extract in OF-2.1.1. Have you ever tried to do it in OF-2.2.2? Is the triangulate false possible in all the plane dictionary? Or just in type patchInternalField as I pasted before?

[huangxianbei]

Originally Posted by alexeym

Hi,

in fact at least in 2.2.2 triangulation of a plane is optional:

Foam::sampledPlane::sampledPlane
(   
    const word& name,
    const polyMesh& mesh,
    const dictionary& dict
)
:
    sampledSurface(name, mesh, dict),
    cuttingPlane(plane(dict.lookup("basePoint"), dict.lookup("normalVector"))),
    zoneKey_(keyType::null),
    triangulate_(dict.lookupOrDefault("triangulate", true)),
    needsUpdate_(true)
{
...

so you can say "triangulate false;" in plane dictionary.

What are you trying to do with the data? Maybe there's another way of solving your problem.

I just tried in OF-2.1.1 the constant plane and specify the triangulate false, however, this makes no sence, the output is still 32768.

[alexeym]

Hi,

well, yes. Just checked 2.1.1 sources and there:

bool Foam::sampledPlane::update()
{
    ...
    if (selectedCells.empty())
    {
        reCut(mesh(), true);    // always triangulate. Note:Make option?
    }
    ...

So I guess, it became option only in version 2.2.x.

Just for post-processing of your case, you can change that true to false in src/sampling/sampledSurface/sampledPlane/sampledPlane.C and then run wmake in src/sampling folder. After this modification planes won't be triangulated at all (I think I've made this thing once while post-processing results on regular mesh).

What do you plan to use for calculation of energy spectrum? If it is external tool (ex. scipy, Matlab), you can interpolate that triangulated data on regular rectangular grid. If it is C++, why not do it using OpenFOAM, where you can iterate over the cells of the mesh getting values inside cells and coordinates of the centers.

[huangxianbei]

Originally Posted by alexeym

Hi,

well, yes. Just checked 2.1.1 sources and there:

bool Foam::sampledPlane::update()
{
    ...
    if (selectedCells.empty())
    {
        reCut(mesh(), true);    // always triangulate. Note:Make option?
    }
    ...

So I guess, it became option only in version 2.2.x.

Just for post-processing of your case, you can change that true to false in src/sampling/sampledSurface/sampledPlane/sampledPlane.C and then run wmake in src/sampling folder. After this modification planes won't be triangulated at all (I think I've made this thing once while post-processing results on regular mesh).

What do you plan to use for calculation of energy spectrum? If it is external tool (ex. scipy, Matlab), you can interpolate that triangulated data on regular rectangular grid. If it is C++, why not do it using OpenFOAM, where you can iterate over the cells of the mesh getting values inside cells and coordinates of the centers.

I want to verify my DNS calculation in different ways, such as Reynolds stress, energy spectrum and two-point correlation.
Yes, I do the post-process in Matlab. In order to simplify the process, I need to output only the point or cell center values. Although C++ is a better way, I find it's hard for me to use, for example, in structure mesh, I don't know the cell index and how to get each. If I want to do the average along a homogenous direction, I don't know how to realize this.