meshBalancing

This contains a single library, dynamicRefineBalancedFvMesh which is based on dynamicRefineFvMesh but adds mesh balancing for parallel cases to the update() function.

Note: The redistributeParPlus function does NOT work. For some reason it does not redistribute the cellLevel and pointLevel fields (hexRef8.distribute crashes) so it is only redistributing the refinement history.

A workaround may be to reconstruct the cellLevel and pointLevel manually from the distributed refinement history.

OpenFOAM Source Changes (for 2.1.x)

1. [CRASH] Add guard in src/dynamicMesh/polyTopoChange/refinementHistory.C in refinementHistory::distribute at line 927 to catch when fully un-refined cells are transferred

   if( newVisibleCells[i] >= 0 ) { visibleCells_[constructMap[i]] = newVisibleCells[i] + offset; }

2. [ANNOYANCE] Add if (debug) guard around print statements in refinementHistory::countProc in the same file as edit #1

3. [CRASH] Add the following to src/finiteVolume/fvMesh/nearWallDist.C at nearWallDist::correct() inside the mesh_.changing() condition but before the patch sizes are set. If the total number of patches on a processor increases, there is a crash unless we increase the size of nearWallDist first.

   // If the number of patches on this processor increased, we need to // increase the number of patches in nearWallDist to match if( mesh_.boundary().size() != size() ) { //Resize nearWallDist to match mesh patches resize(mesh_.boundary().size());

      //Set any newly created (unset) patches
      forAll(*this, patchI)
      {
          if( !set(patchI) )
          {
              set
              (
                  patchI, 
                  fvPatchField<scalar>::New
                  (
                      calculatedFvPatchScalarField::typeName,
                      mesh_.boundary()[patchI],
                      mesh_.V()
                  )
              );
          }
      }
   

   }

4. [CRASH] Mapping of patches had an error for mixed patches. In src/dynamicMesh/fvMeshAdder/fvMeshAdderTemplates.C in function fvMeshAdder::MapVolField around line 263 there is a section where imported patch values are mapped to existing patches. This correctly maps the patch value, but omits mapping of the refValue, refGradient, and other stored fields in mixed patches. The refValue is then left set to a section of allocated but unset memory if all the original faces of that patch get moved to another processor. Remove the manual loop near the end of the function and use the rmap function with a reversed map by replacing

   forAll(newFld, i) { label oldFaceI = newToAdded[i];

      if (oldFaceI >= 0 && oldFaceI < addedFld.size())
      {
          newFld[i] = addedFld[oldFaceI];
      } 
   

   }

with

    labelList addedToNew(addedFld.size(),-1);
    forAll(newFld, i)
    {
        label oldFaceI = newToAdded[i];

        if (oldFaceI >= 0 && oldFaceI < addedFld.size())
        {
            addedToNew[oldFaceI] = i;
        } 
    }
    
    newFld.rmap(addedFld, addedToNew);

5. [CRASH] When using a chemistry solver, the DimensionedField deltaTChem in basicChemistryModel.H is not properly mapped/distributed for reasons unknown. An easy workaround is to change it to a GeometricField by making the following edits:

   a. In basicChemistryModel.H change DimensionedField<scalar, volMesh> deltaTChem_; to GeometricField<scalar, fvPatchField, volMesh> deltaTChem_; and add #include GeometricField.H and #include fvPatchField.H to the list of included headers b. In basicChemistryModelI.H change both instances of return deltaTChem_; to return deltaTChem_.dimensionedInternalField();

   Recompile the thermophysicalModels source directory and the mapping works.