Preconditioning in miscible2pnc constraintsolver leads to different solution compared to 2p2c
This emerged from issue #407 (closed) .
In principal the 2p2c and 2pnc model do the same for 2 components. Even the different values in the priVarswitch (see #407 (closed)) does not change the solution.
Looking into that we found that the preconditioning of the matrix in the miscible2pncconstraintsolver is what causes the different solutions of the 2p2c and the 2pnc model. Points up to discussion are now (in my view)
- we need the preconditioning in the 2pnc due to possible very low concentrations. Do we want to include that in the 2p2c misciblemultiphaseconstraintsolver as well? Then we should have the same solutions for 2p2c and 2pnc
- do we want to change the priVarsswitch values in 2p2c to the 2pnc values for consistency or is that unnecessary (does not change solution)?