Parallisation

[JanGaertner]

Problems arise at the boundary of processor patches within regions that require limiting of the WENO scheme. I assume this is due to the use of the halo cells and that maybe only the patch neighbour cells are available.

I have following proposal:

• Make use of OpenFOAM parallisation handling by making intBasTrans within the WENOBase class a GeometricField. This has the GeometricField::Boundary class which can handle the processor boundaries correctly. In practice this field will then be GeometricField<WENOBase::volIntegralType,fvsPatchField,surfaceMesh> Updating the boundary could be done then with `intBasTrans.correctBoundaryConditionsor directly withintBasTrans_.evaluate(). Accessing neighbour patch field values is then possible by accesing the boundary field and callingpatchNeighbourField` on a coupled fvPatch type.
  • Developing this feature is done in a parallelOF branch which will be merged into master after a review process.

It would be great if someone would be willing to help me check the code and to look over my changes. Especially to make sure that no new errors are introduced.

[JanGaertner]

The error lies not within the cells directly at the processor patch as these do have access to the halo cells but at the cells neighbouring these. To avoid this issue I propose to reconstruct the field around each processor, calculate all stencils and stencilIDs and then to store this information.

[JanGaertner]

Parallelization using a reconstructed mesh is now introduced into the development branch.