Organizing the linear FV solution scheme for additional equations

[GiudGiud]

Motivation

We have v, p, energy and passive scalars implemented right now, with (v,p,energy) in a nested loop and passive scalars on the side.
We have to integrate:

• phase mixture
• k, epsilon or omega
  How should we nest those?
  In the absence of executors, we have to make a decision on how much nesting each solution gets. It wont be customizable from the input file to a complete extent.

There are (at least) three options:

• external loop (no nesting) after the main iteration (this is only good enough if we decide to lag the effect of one of these quantities, as they are not "passive")
• in the main (p,v,energy) loop: this means while the linear system is solved, the solution to that equation is not fully converged (due to matrix relaxation) on every (p,v,e) fixed point iteration
• fully converged in the main (p,v,energy) loop: we resolve any nonlinear effects from that active scalar onto itself by iterating solving that equation, inside each (p,v,e) fixed point iteration

My hunch is that option 2 is where we will go. I'll tag the others once it's not Christmas anymore

Design

I am thinking we could nest energy, k, epsilon/omega and phase mixture in an "active scalar" loop, with v and p.

Impact

Guidance for contributing new equations to SIMPLE/PIMPLE