[gallon]

Hello CFD, hope this is the right forum to ask the question.

The goal is to place an stl model and refine the mesh.
Everything goes fine in serial but since I want to refine up to 5-6 levels it takes too long.

So here's my script:

blockMesh
decomposePar [-force]

At this decomposePar knows nothing about WALL10 (the surface read from .stl by sHM).
OK, then I run

surfaceFeautureExtract
mpirun -np 2 snappyHexMesh -overwrite -parallel
mpirun -np 2 sonicFoam -parallel

The surface is read and refined in parallel OK. If a given processors's decomposed domain does not contain a part of the surface, it is still mentioned in processorX/constant/polyMesh/boundary as an empty boundary:

[v.galas@node1 tmp]$ cat processor0/constant/polyMesh/boundary
/*--------------------------------*- C++ -*----------------------------------*\
| =========                 |                                                 |
| \\      /  F ield         | OpenFOAM: The Open Source CFD Toolbox           |
|  \\    /   O peration     | Version:  v3.0.1                                |
|   \\  /    A nd           | Web:      www.OpenFOAM.org                      |
|    \\/     M anipulation  |                                                 |
\*---------------------------------------------------------------------------*/
FoamFile
{
    version     2.0;
    format      ascii;
    class       polyBoundaryMesh;
    location    "constant/polyMesh";
    object      boundary;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //

4
(
    OUTL2
    {
        type            patch;
        nFaces          22816;
        startFace       418376;
    }
    INLE1
    {
        type            patch;
        nFaces          0;
        startFace       441192;
    }
    WALL10
    {
        type            wall;
        inGroups        1(wall);
        nFaces          0;
        startFace       441192;
    }
    procBoundary0to1
    {
        type            processor;
        inGroups        1(processor);
        nFaces          1360;
        startFace       441192;
        matchTolerance  0.0001;
        transform       unknown;
        myProcNo        0;
        neighbProcNo    1;
    }
)

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

mpirun -np 2 sonicFoam -parallel

But now the solver run at the processors without WALL10 cannot find the BC for it.
I have tried mapFields and redistributePar but to no avail.

p.s. sHM (serial) -> decomposePar -> sonicFoam (parallel) works ok.
Any ideas?