When multisegment wells are involved for RESTART, based on the recovered
segment-related quantities as Opm::data::Segment objects, the segment
phase rates and segment pressure are recovered as part of the
WellStateFullyImplicitBlackoil.
Forgotten in previous commit. This restores our ability to run
flow-legacy with the new WellStateFullyImplicitBlackoil::report().
Thanks to Kai Bao for alerting me to the issue.
This commit adds a new data member, seg_number_, that maps a linear
segment index (0 .. #segments - 1) to the appropriate segment number
(segment ID). This ensures that member function report() is able to
produce segment results in terms of user-assigned segment numbers
rather than linear indices internal to the simulator and its state
variables. This, in turn, decouples the well state object from the
output (summary/restart) code and makes the restart facility more
self contained.
Rewrite the unit test for WellStateFullyImplicitBlackoil to account
for the new indexing scheme when assigning the synthetic segment
results.
While here, also abide by the file's naming convention for data
members and locals.
This commit extends class WellStateFullyImplictBlackoil to report
segment-related quantities as Opm::data::Segment objects (included
in Opm::data::WellRates objects). All wells have at least a top
segment in the context of WellState FIBO, so there is a meaningful
value to report for each well.
We put the extraction of segment-related quantities into a new
helper function
WellStateFullyImplicitBlackoil::reportSegmentResults()
to avoid cluttering up the body of report() more than absolutely
needed.
The primary use-case for this is assigning appropriate values to
items 8 through 11 of restart vector RSEG. In turn, this will
enable restoring these quantities from a restart file.
Concretely this avoids having to patch eWoms by adding a generic
`Opm::transposeDenseMatrix()` template function instead of relying on
the dense matrix class to provide a `transpose()` method.
these parameters where introduced with support for the TUNIING keyword
in `ebos`. since `flow` implements its own time stepping these
parameters are unused and should thus be hidden from view in it.
the former is caught by `ebos`, while the latter isn't. Alternatively,
this can be fixed by deriving `LinearSolverProblem` from
`NumericalIssue`, if preferred.
