[gerritgroot]

Dear Foamers,

I'm stuck with a doubt about the difference between patch matching and patch merging, a feature that I need as I'd like to create blockMeshDict files with a multiblock philosophy (because of the origins of each block, I prefer to create multiblock meshes).
So far, none of both seems to work well. I only know that patch merging deletes coinciding nodes and patch matching doesn't. The resulting grid is shown in the picture: "mergedandmatchedpatches.png".

I did a try out, connecting 2 blocks of one hex each, one by patch matching and one by patch merging.
The blockMeshDict files are attached to this question and are named: "blockMeshDict_mergedpatches.txt" and "blockMeshDict_matchedpatches.txt".

The merge variant yields 12 nodes, while the match variant yields 16 nodes, so far everything is as expected.

However, checkMesh-ing the merge variant gives an error on face connectivity, see "checkMesh_mergedpatches.txt", while the matched variant yields an errorless mesh after running checkMesh, see "checkMesh_matchedpatches.txt", but gives the comment:

  *Number of regions: 2
    The mesh has multiple regions which are not connected by any face.

From which I tend to conclude that the mesh is not being treated as a whole unique domain, but two separated domains instead.

In my examples, merging does not yield a unique domain, while matching (apart fromt he strange pyramid border between the blocks) yields a wrong mesh according to checkMesh.

My questions to ye foamers are:

• -- Most important of all: What would be the correct philosophy to connect different grid blocks to form one domain without a boundary between the diferent (in general more than two) blocks? (either or not generating each block in the same blockMeshDict file)
• -- Why do the specified boundaries: wall, inlet and outlet disappear as soon as I specify the adjacent patches, patch0 and patch1?
• -- In a post I can't find back anymore, it is said that patches are reshaped to a pyramid to prevent for hanging nodes, why does this happen for matched patches and not for merged patches? (In my opinion this shouldn't happen at all for two adjacent cubes).
• -- Would it be better to use "stitchMesh" instead? Do you need complete directory structures with "./0", "./system" and "./constant/polyMesh" for each separate mesh before stitching? I couldn't find any example of stitchMesh, is there any?

Well, these are a lot of questions, maybe too many for just one post. I hope to get some clue at least.

Thank you in advance,

Gerrit