[CFDUser29]

Hello, I currently have a problem with snappyHexMesh that I was not able to solve for days.
I start with a simple case prior to complexify my study case as I want to add a ship in my domain, but the snappyHexMesh is not working correctly, and I modified many parameters but nothing solves my problem.

1 ) I define my blockMesh as a rectangular box with 6 different patches for the 6 different faces.
2 ) blockMesh
3 ) surfaceFeatures
4 ) decomposePar -noFields (to Mesh in parallel on 8 cores)
5 ) srun --export=ALL -N 1 -n 8 snappyHexMesh -overwrite -parallel
6 ) reconstructParMesh -constant

When looking on paraview, SHMesh is doing strange things with patches from blockMesh, some are cut in two or kept as defined in blockMeshDict (as you can see on the picture)
paraview.jpg
We can see in the boundary dictionary that three patches from the original blockMesh still have faces, whereas they should have 0 faces after SHMesh.

Can anyone help on this, because all my research could not give me a solution ?

blockMeshDict.txt

boundary.txt

snappyHexMeshDict.txt

log checkMesh.txt

[CFDUser29]

I believe that the problem comes from the dimensions of the box for the block Mesh. It has to be bigger than the fluid domain. Now, it works for a simplified fluid domain which is a box.

I add a ship in my domain on a surface disc (that will enable me to simulate different attack angles). What was previously called "water" becomes "water_surface" and "turning_disc", where there is a whole in the turning_disc with the shape of the floating line of the ship. You follow me? Of course, the ship is also here as a surface.

The problem is that after SHMesh, there is bad interaction between the turning_disc and wallInf from the blockMesh patches, and instead of being ovrewritten, a part of wallInf subsists at the end with the shape of the turning_disc, and this is a big problem for me.

[CFDUser29]

There is what the former "water" surface now looks like

waterplane.PNG

[CFDUser29]

There is the solution I found if it is of interest for anyone.

On Rhinoceros 3D, the best way to have a good mesh when exporting as OBJ is to group the surfaces where you have problems in OpenFOAM.

When you are then building your boundaries on Blender (I found a tutorial on Youtube), you see that the mesh is very different without problems of surface discontinuity and non-merged points. Hope that helps.

The only problem is when you have two surfaces on the same plane, when you group them, you lose the geometry of each part, which means in my case that I no longer work with my turning disc.