[Nek5000-users] Implementing magnetic boundary conditions

[nek5000-users at lists.mcs.anl.gov]

Hi everyone,
I am currently trying to implement the quasi vacuum boundary conditions (n x B=0) for an ellipsoidal shell. My understanding from earlier threads is that I can only set Bx=By=dBz/dz=0 via the string ‚ddn‘ , restricting me to boundaries orthogonal to the cartesian axis. Has anyone tried some workarounds to this problem? And are the same restrictions applying to the flux boundary conditions of the add. scalars? If not, it might be possible to use 3 additional passive scalars to model the magnetic field (instead of the standard implementation).  Or, as another idea, is it possible to use something like the conjugate heat transfer implementation for the magnetic field: Having some larger Elements on the outside with u=0, where only laplace B=0 is solved.
My current approach, which is not working yet, is to transfer the boundary conditions to Dirichlet conditions via an interpolation algorithm, but this seems a bit dubious to me, both from a numerical and a physical perspective.
Any suggestions would be appreciated.

Best regards,
Jan