[lentschi]

Hello,

I've tried to decompose a case using Mixing-Plane in order to couple rotating and statinary part - but during decomposing always an error occurs. I know that's not possible to run the Mixing-Plane Interface parallelized so I have changed the decomposeParDict as follows (in order to keep all cells at the Mixing-plane on one processor):

------------------------------------------------------------------------

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \\ / O peration | Version: 1.6 |
| \\ / A nd | Web: www.OpenFOAM.org |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
version 2.0;
format ascii;
class dictionary;
location "system";
object decomposeParDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

numberOfSubdomains 3;

preservePatches (STAT-TO-ROT-SIDE-2 STAT-TO-ROT-SIDE-1);
preserveFaceZones (STAT-TO-ROT-SIDE-2_zone STAT-TO-ROT-SIDE-1_zone);

method metis;

metisCoeffs
{
processorWeights
(
1
1
1
);
----------------------------------------------------------------------

And the corresponding error:

----------------------------------------------------------------------
Processor 0: field transfer
Initializing the GGI interpolator between master/shadow patches: PERIODIC-ROT-SIDE-1/PERIODIC-ROT-SIDE-2
Initializing the GGI interpolator between master/shadow patches: PERIODIC-STAT-SIDE-1/PERIODIC-STAT-SIDE-2
Initializing the mixingPlane interpolator between master/shadow patches: STAT-TO-ROT-SIDE-2/STAT-TO-ROT-SIDE-1
Segmentation fault (core dumped)
--------------------------------------------------------------------

Calculating with single processor works!

Has anyone an idea how to fix this problem?

Thank you for your help in advance.

BR

[timo_IHS]

Hi,

try with globalFaceZones instead.

Best,
Timo

[lentschi]

Thank you for you quick reply.

I have already tested this (before ich use the preserve commands) but there is no differnence in result.
It seems to be a problem of the underlying cluster system not of foam itself?!

BR

[lentschi]

One additional problem is to run the "mixinPlaneCheck" parallelized - that seems to be not possible up to now!

Has there anyone different experiences?

BR

[mbeaudoin]

Hello,

Yup, it looks like the mixingPlane is no longer playing nicely with decomposePar. I am looking into it.

Martin

Originally Posted by lentschi

Hello,

I've tried to decompose a case using Mixing-Plane in order to couple rotating and statinary part - but during decomposing always an error occurs. I know that's not possible to run the Mixing-Plane Interface parallelized so I have changed the decomposeParDict as follows (in order to keep all cells at the Mixing-plane on one processor):

------------------------------------------------------------------------

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \\ / O peration | Version: 1.6 |
| \\ / A nd | Web: www.OpenFOAM.org |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
version 2.0;
format ascii;
class dictionary;
location "system";
object decomposeParDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

numberOfSubdomains 3;

preservePatches (STAT-TO-ROT-SIDE-2 STAT-TO-ROT-SIDE-1);
preserveFaceZones (STAT-TO-ROT-SIDE-2_zone STAT-TO-ROT-SIDE-1_zone);

method metis;

metisCoeffs
{
processorWeights
(
1
1
1
);
----------------------------------------------------------------------

And the corresponding error:

----------------------------------------------------------------------
Processor 0: field transfer
Initializing the GGI interpolator between master/shadow patches: PERIODIC-ROT-SIDE-1/PERIODIC-ROT-SIDE-2
Initializing the GGI interpolator between master/shadow patches: PERIODIC-STAT-SIDE-1/PERIODIC-STAT-SIDE-2
Initializing the mixingPlane interpolator between master/shadow patches: STAT-TO-ROT-SIDE-2/STAT-TO-ROT-SIDE-1
Segmentation fault (core dumped)
--------------------------------------------------------------------

Calculating with single processor works!

Has anyone an idea how to fix this problem?

Thank you for your help in advance.

BR

[lentschi]

Thank you!

[lentschi]

Are there any new knowledges regarding decomposition with mixingplane?I'm using Foam-extend-3.1.

BR,
Markus

[Jack001]

I am encountering the exact same issue with parallelizing a mixingPlane case. My decomposeParDict looks like:

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | foam-extend: Open Source CFD                    |
|  \\    /   O peration     | Version:     3.2                                |
|   \\  /    A nd           | Web:         http://www.foam-extend.org         |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      decomposeParDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

numberOfSubdomains 4;

//method               metis;
method    patchConstrained;


// All GGI interfaces should be listed here:
globalFaceZones
(
    PER11_ROTORZone
    PER22_ROTORZone
    PER11_STATORZone
    PER22_STATORZone
    SHROUD_ROTOR_TIP_GGI_SIDE_11Zone
    SHROUD_ROTOR_TIP_GGI_SIDE_22Zone
);

patchConstrainedCoeffs
{
    method            metis;
    numberOfSubdomains    4;

    // You can force the faces of a patch to be on the same processor
    // this is currently needed for mixingPlane
    patchConstraints
    (
        (OUTFLOW1_ROTOR 0)
        (INFLOW1_STATOR 0)
    );
}

metisCoeffs
{
    processorWeights
    (
        1
        1
        1
        1
        1
        1
        1
        1
    );
}


distributed     no;

roots
(
);

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

[HenrikJohansson]

Hi

Have there been any update with mixing plane and running in parallel?
My project works fine on one core but when decomposing it and running in parallel it fails.
I'm using foam-extend 4.0.

I get the following warning when decomposing.

--> FOAM Warning : 
    From function decompositionMethod::loadExternalLibraries()
    in file decompositionMethod/decompositionMethod.C at line 508
    Loading of decomposition library libscotchDecomp.so unsuccesful. Some decomposition methods may not be  available

My decomposeParDict looks like this

/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | foam-extend: Open Source CFD                    |
|  \\    /   O peration     | Version:     4.0                                |
|   \\  /    A nd           | Web:         http://www.foam-extend.org         |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       dictionary;
    object      decomposeParDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

numberOfSubdomains 8;

//method               metis;
method    patchConstrained;


globalFaceZones
(
    MixP_Stator_FaceZone
    MixP_Rotor_FaceZone
    Periodic_Left_Stator_FaceZone
    Periodic_Right_Stator_FaceZone
    Periodic_Left_Rotor_FaceZone
    Periodic_Right_Rotor_FaceZone
);

patchConstrainedCoeffs
{
    method            metis;
    numberOfSubdomains    8;
    patchConstraints
    (
	    (MixP_Stator 0)
	    (MixP_Rotor 0)
	    (Periodic_Left_Stator 1)
	    (Periodic_Right_Stator 1)
	    (Periodic_Left_Rotor 2)
	    (Periodic_Right_Rotor 2)
    );
}

simpleCoeffs
{
    n               (2 2 1);
    delta             0.001;
}

hierarchicalCoeffs
{
    n               (1 1 1);
    delta           0.001;
    order           xyz;
}

metisCoeffs
{
    processorWeights
    (
        1
        1
        1
        1
        1
        1
        1
        1
    );
}

manualCoeffs
{
    dataFile        "";
}

distributed     no;

roots
(
);


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

Any clues?

Solved my problem.
Turned out to be meshing and B.C. problems. Switched from cyclicGgi to 1:1 cyclic.

/ Henrik