[botman]

Hi

I am attempting a meshing strategy that involves

1. directional refinement of the blockMesh using topoSet and refineMesh, where the refined mesh does not satisfy the requirements for snappyHexMesh
2. separation of the refined blockMesh into an "inner" and "outer" region, where the inner region satisfies the requirements for snappyHexMesh (quads).
3. snappyHexMesh of the inner region around an STL geometry which resides at the center of the inner mesh
4. recombination of the inner and outer mesh using stitchMesh (perfect).

Defining the interface patches that were created in step 2 as outerMesh_innerBoundary and innerMesh_outerBoundary, what appears to happen is that during the snapping phase of step 3, the boundary coordinates of the "innerMesh_outerBoundary" patch seem to move a little, which means I can't get stitchMesh "perfect" to work when I try to recombine the patch with the "outerMesh_innerBoundary" patch in step 4. I was hoping someone might know if there was a quick trick to fix/constrain the "innerMesh_outerBoundary" coordinates during snapping?

The number of faces on "innerMesh_outerBoundary" patch has not changed during snapping, but it seems that snapping process relaxes the coordinates of the whole mesh. The interface patches are not smooth and have lots of "kinks" (like a castellated mesh); so, I don't think that the other stitchMesh options will work for this mesh (or at least I haven't managed to get them to work with a toleranceDict).

I know I could try the directional refinement now available in snappyHexMesh, but I would rather not go back to the drawing board if there was quick fix to constrain the coordinates.

Thanks in advance for any assistance.

[botman]

In the off chance this is of use to others, below is the workaround I am adopting using the python package Ofpp. The script just shifts the points of the innerMesh_outerBoundary patch to match those of outerMesh_innerBoundary. I can then use stitchMesh perfect. Note that I have merged the meshes before running the script, and the script only works if the interfacing patches are almost perfect.

import numpy as np
import Ofpp 

mesh = Ofpp.FoamMesh('.')
bi = mesh.boundary[b'innerMesh_outerBoundary']
bo = mesh.boundary[b'outerMesh_innerBoundary']
fList_i = np.array(range(bi.start, bi.start + bi.num))
fList_o = np.array(range(bo.start, bo.start + bo.num))
mpts_i = np.zeros((len(fList_i), 3))  
mpts_o = np.zeros((len(fList_o), 3))

# Compute the mid-points of the boundary faces
for k in range(len(fList_i)):
        mpts_i[k] = np.average(
            mesh.points[np.ix_(mesh.faces[fList_i[k]])], axis=0)
        mpts_o[k] = np.average(
            mesh.points[np.ix_(mesh.faces[fList_o[k]])], axis=0)

# Set the points on the inner patch to match the outer patch
for i, mi in enumerate(mpts_i):
        j = np.argmin(np.linalg.norm(mpts_o-mi, axis=1))
        point_i = mesh.points[mesh.faces[fList_i[i]]]
        point_o = mesh.points[mesh.faces[fList_o[j]]]
        point_o2i = point_o[[np.argmin(np.linalg.norm(
            point_o-pi, axis=1)) for pi in point_i]]
        mesh.points[mesh.faces[fList_i[i]]] = point_o2i

# ... then you just export the updated points matrix to constant/polyMesh and then "stitchMesh -perfect" should run ok