[shogan50]

I have an .stl volume which represents features outside the fluid domain. Fluid, for the most part is flowing around it. I have an .stl disk shaped surface, which represents and inlet patch. The outlet is one of the walls in the blockMesh, but this isn't convenient for the inlet. This is a decomposed case. I have successfully created an inlet patch using topoSet and createPatch, but it only applies to the pre decomposed case. It appears to be grabbing faces from the block mesh. Further, RecostructParMesh warns that it is experimental and not recommended except...

What is the preferred method to generate the inlet patch from .stl? I've tried to included it in Snappy... but it is really external to the snapped geometry and it ignores it. The .stl's are created in Solidworks if that matters. I'm relatively proficient at editing/combing .stl's.

topo snippet:

(
	{
		name fz_inlet;
		type faceZoneSet;
		action new;
		source searchableSurfaceToFaceZone;
		sourceInfo
		{
			surface searchableDisk;
			origin (0 0 -0.037);
			normal (0 0 -1);
			radius .05;
			//name inlet.stl;
		}
	}
);

createPatchDict snippet:

patches
(
    {
        // Name of new patch
        name inlet;

        // Type of new patch
        patchInfo
        {
            type patch;
        }

        // How to construct: either from 'patches' or 'set'
        constructFrom set;

        // If constructFrom = patches : names of patches. Wildcards allowed.
        //patches ("periodic.*");

        // If constructFrom = set : name of faceSet
        set fz_inlet;
    }
);

[shogan50]

I figured out that you have to apply whatever utility you are using to each processor folder.

#!/bin/sh
cd ${0%/*} || exit 1    # Run from this directory
...
...
runApplication surfaceFeatures
runApplication blockMesh
runApplication decomposePar
foamJob -parallel -screen snappyHexMesh -overwrite
FOLDERS=./processor*/
for f in $FOLDERS
do
	echo "adding inlet patch to $f"
	cd $f
	surfaceToPatch ../constant/triSurface/inlet.stl
	cd ..
done
cd ..

[shogan50]

This doesn't totally solve my problem. The utility doesn't seem to have an -overwrite option, so this script creates timesteps where it finds matching faces, leaving the processors out of sync in their timestamps. My initial question still stands. What is best practice here?

Thanks