Option to prescribe spherical normal and tangential velocties individually

[anne-glerum]

It would be wonderful to be able to prescribe either the normal or the tangential velocity component on a boundary of a spherical domain, e.g. a chunk. For box models it is possible to prescribe e.g. the x component of velocity and leave the tangential y and z components free.

Would it be possible to use deal.II's compute_nonzero_tangential_flux_constraints() and compute_normal_flux_constraints() for this?

[tjhei]

Can you be more precise what you are trying to achieve? If you want free tangential velocity and a given normal flux, then nonzero normal flux is the correct one to use in deal.II.

[anne-glerum]

My specific use-case would indeed be to prescribe a normal velocity on the lateral boundaries of a chunk, but leave the tangential components of velocity free. At the moment, I have to prescribe all components of velocity on a spherical domain's boundary.

In essence, I'd like the same functionality as for boundary velocities on box geometries, where one can specify (or not specify) the velocity magnitude per coordinate direction.
E.g. set Prescribed velocity boundary indicators = east phi: function.

At the moment, I can specify the boundary velocity in spherical coordinates, using spherical unit vectors, but, the way I understand it, still all velocity components are prescribed.

[tjhei]

Then you would need to use compute_nonzero_normal_flux_constraints() instead of compute_no_normal_flux_constraints(). Shouldn't be too difficult to do the change. The hardest part is to make a plugin system.
Do you want to give it a try?