Generalized (Blockwise or ExtraIDwise reductions, VPP, spatial UO) distributed conservative transfers

[GiudGiud]

Reason

In a meeting with a client, feature was requested for conserving energy in transfer between a global calculation and individual distributed subscale calculations. The quantity, related to the transfered field, should be conserved on a local per-app basis rather than simply on a global basis (as the system currently allows with PPs only)

Local per-app basis can be done currently using the NearestPointAverage user objects. This is a special case in the code that could be generalized to user objects

Design

A VPP on global app (main app most likely) for the quantities of interest in each block/id-ed region OR a spatial UO on the global app

A postprocessor with the same name in each distributed subapp OR a VPP on the subapp tied to block/extraIDs

The ratios between the VPP-indexed / spatial values and the PP for each subapp should be used to renormalize the transfer. The renormalization should be local

EDIT: we can currently do this two ways:

• we would need as many postprocessors as quantities we want to conserve. That could be 50,000 pps, so unmanageable
• we could use the nearest point averages as quantities we want to conserve. This can provide local conservation, but only on for a single axis. If we wanted axial and radial conservation we need more

Impact

More fine grained energy conservation Better convergence of multiphysics as a result

Tagging @friedmud @permcody @vincentlaboure who attended said meeting

[GiudGiud]

@jortensi and @sallustius have confirmed this is of interest

[GiudGiud]

This can be tackled in four parts:

• define mesh divisions (see https://github.com/idaholab/moose/issues/25901)
• add postprocessing based on mesh divisions, and if needed nest mesh divisions!
• add restrictions to transfers to match mesh divisions
• add a distribution over the conservation factors in the transfers

I ll do the first three for sure. If there's more interest, please ping me and I ll do part 4.

I can see @aprilnovak being interested in this for Cardinal so I'll tag you if you don't mind.