Conformal Meshes and Single Physics

[CrocoDuck]

Hi there!

I am working on a simple Acoustic problem with the Helmholtz solver. It is a simple room with a sphere source inside.

Normally I model the room with FreeCAD, and then I cut out a sphere from its volume. I then export the geometry as step, open up it with Salome, and mesh it. Then I will open the mesh with Elmer, assign the boundary conditions on the walls and on the sphere source (normally 0 flux on the walls and uniform velocity flux on the sphere surface) and solve. This seems to work really well.

I wanted to try something different. I left the sphere inside the room volume this time, without cutting it off from it. Then I exported the room and the sphere inside it to the same step file. I then meshed it with Salome, making a conformal mesh. My problem is that when I set up my problem with Elmer I am really only interested in what is happening outside the sphere. So, I assign 0 flux boundary condition to the walls, a uniform velocity flux boundary condition to the sphere surface, and apply the Helmholtz equation only to the volume of air, and no equation at all to the sphere itself. The issue is that the solution I get is 0 everywhere.

I think I am getting something very fundamentally wrong here. Any ideas?

[CrocoDuck]

Oh! I think I might be onto something. I think that by just exploding the geometry in solids and faces, and then using those to create groups in the mesh, the result is that the surface of the sphere gets "assigned" to the sphere and acts as a boundary for that body only, and not as a boundary to the volume of air. So, any boundary condition applied to it does not apply to the volume of air, and the solution is trivially zero.

I then made a partition in Salome of the two shapes: the volume of air and the sphere inside it. Then exploded the solids in the partition. Now the surface seems to act as a boundary for both bodies.

I will post a link soon to the study (it will be on GitHub).

[CrocoDuck]

Here the link to my study description: https://computational-acoustics.gitlab. ... io-part-1/, it contains links to the GitHub repo where all the files are stored.