[cyou]

I encountered a fetal error when running redistributePar before chtMultiRegionFoam.

Selecting decompositionMethod simple [1] (region BaseBoard)
From time 0 have objects:4(rho cellToRegion p T)
[3] [0] 
[0] 
[1] 
[1] 
[1] --> FOAM FATAL ERROR: (openfoam-2012[2] 
[2] 
[2] --> FOAM FATAL ERROR: (openfoam-
[3] 
[3] --> FOAM FATAL ERROR: (openfoam-2012)
[3] Cannot find a patch which is neither of type empty nor coupled in patches 
67
(
.
.
.
There has to be at least one such patch for distribution to work
[1] 
[1]     From Foam::label Foam::fvMeshDistribute::findNonEmptyPatch() const
[1]     in file fvMeshDistribute/fvMeshDistribute.C at line 353.
[1] 
FOAM parallel run aborting
[1]

I checked the source code and found the mappedPatch is seen as empty patch from OP-v2012.

    label nonEmptyPatchi = -1;

    forAllReverse(patches, patchi)
    {
        const polyPatch& pp = patches[patchi];

        if
        (
           !isA<emptyPolyPatch>(pp)
        && !isCoupledPatch(patchi)
        && !isA<mappedPatchBase>(pp)
        )
        {
            nonEmptyPatchi = patchi;
            break;
        }
    }

    if (nonEmptyPatchi == -1)
    {
        FatalErrorInFunction
            << "Cannot find a patch which is neither of type empty nor"
            << " coupled in patches " << patches.names() << endl
            << "There has to be at least one such patch for"
            << " distribution to work" << abort(FatalError);
    }

All surface of solid region in my case is setted as mappedWall.
I want to use redistributePar to assign solid region to only one processor.
Any suggestions for my problem?

[nholt]

I'm also having this problem. It is also preventing me from running CHT cases with an isolated and smaller fluid region, since the smaller fluid region is only distributed onto one processor and the 'pRefCell' SIMPLE quantity gives an error for processors which do not contain any part of that mesh.

Would it be best to change the patch types to 'patch' before redistributing and then changing them back to mapped afterwards?