Just a comment to the usage of snappy:

I usually generate two meshes seperately, when I use AMI/GGI.

Firstly I create the outer domain using blockMesh + snappyHexMesh. The volume where the propeller will be located later is cut out by snappyHexMesh, using a cylinder. For the cylinder I usually use a high res STL, I placed in paraview. Remember to use surfaceFeatureExtract on the cylinder, in order to get the edges.

Secondly I generate a mesh in a second case directory, using blockMesh, that has just the right size, that the cylinder fits in. Pay special attention to the fact, that cubic cells are essential for a good snapping. The edges of both the cylinder as well as the ones of the propeller must be extracted, using surfaceFeatureExtract. After having set up the snappyHexMeshDict, I start the meshing process.

This results in two meshes in two case directories and has the benefit of being able to optimize each mesh individually, with a comparatively short meshing time. The last step is to copy the first mesh's case to a third location and use mergeMeshes in that case, in order to merge the propeller mesh into this mesh. Unfortunately this results in just a single zone. Hence some topoSet commands must be executed to split the mesh into two cellZones.

If you need specific help with any of the above steps, just let me know. I am in a hurry right now.

Hope this helps,
Jens