code_saturne User's Forum

Hello,

applying cs_mesh_refine_simple() on meshes with periodic boundaries throws a sigsegfault during the refinement process (CS 6.0.4). Probably it can’t handle the ghost cells, since the refinement on the same mesh without periodic boundaries works well.
Do you have a recommendation for using refinement on a mesh with translation periodicity?


If I am correct the postprocessing writers work on the original not the refined mesh. How could I visualize the refined mesh?

Kind regards,
Ralf

Hello Ralf,

Yes, maintaining periodicity across mesh refinements can be quite complex, and is not supported so far.

The best solution is to do the periodic joining after refining. I intend to change the way some preprocessing operations are handled, to to them in a sort of "user-defined pipeline", with the user choosing the order, but in current versions, joining and periodicity are done first. So work around this, you need to have a separate "refinement" run, and use that as an input for a computation with periodicity.

I recently added a cs_mesh_boundary_remove_periodicity utility function to remove periodicity information from a given mesh, in case you need that in user functions such as cs_user_mesh_modify.

Best regards,

Yvan