[ingojahn]

Hi,

I am currently trying to write a simple mesh converter from our in-house grid generator e3prep to OpenFoam.
e3prep, is a multi-block meshing tools, that generates the mesh as a multitude hexahedra, each of which has its own internal mesh structure. These blocks are then linked at the interfaces, through face matching.

Now I want to create the ability to convert these multi-block meshes into an OpenFoam mesh.

What I have done so far: (and this works)
For a mesh consisting of a single block, I have written an output function (in python) for our meshing tool, which directly writes files in the /Polymesh folder (boundary, faces, owner, points, neighbour). Boundaries are automatically defined as North, South, East, West, Top, Bottom.

What i want to do:
Now if I have a multi-block mesh, I want to combine all the blocks into a single OpenFoam mesh.

Option 1: (this works, but becomes tedious as number of blocks increases)
I run the following sequence of commands (bnnnn are names of respective blocks)
mergeMeshes b0000 b0001 -overwrite
createPatch (together with createPatchDict, to combine respective patches on the blocks)
stitchMesh internal1 internal2 -perfect -overwrite
Repeat the above until I have added all the blocks. Obviously this works for 2-3 blocks and could be scripted, but once you start to add 20-30 blocks this becomes tedious and error prone

Option 2: (not tried yet)
Rather than writing the individual files within the /PolyMesh folder, I could write a script to create a "blockMeshDict" file.
Effectively I would turn each individual cell from each of my block into an individual hex e.g.[hex (0 1 2 3 4 5 6 7)(1 1 1) simpleGrading(1 1 1)].
Then use the mergePatchPairs to connect the internal faces.


At the moment I am leaning towards option 2, as this is more universal.
However before proceeding, I just wanted to check that I haven't overlooked some obvious solution.

Thanks for your help.

[GerhardHolzinger]

I followed option 2 for my meshes, for which I did the deconstruction into blocks manually with pencil and paper. For domains with some sort of regularity this works pretty well, e.g. a largely axi-symmetric 3d domain.