This commit adds support for recognizing additional inter-region
flow rate/cumulatives at the parser level. In particular, we now
have input support for the E300-compatible keyword strings
R[OGW]F(R[-+]?|T[-+])
(e.g., ROFR, RGFT-, or RWFR+). This, in turn, ensures that the
parsing process does not abruptly terminate when encountering such
summary keywords. This is also a necessary step towards supporting
actually calculating those summary vectors, but we presently do not
support doing so.
Expand the minimum, input-reading only, corresponding unit test to
include the oil, gas, and water varieties of the new keyword strings.
While here, also properly classify the per-phase gas cumulatives
RGFTG, RGFTL
as inter-region summary vectors. These were inadvertently listed as
per-region vectors (i.e., in the REGION_PROBE instead of in the
REGION2REGION_PROBE).
This commit ensures that the field, group, and well level cumulative
production curves of solution/free oil and gas are continuous in a
restarted simulation run.
Thanks to Torbjørn Skille for identifying the appropriate locations
in XGRP/XWEL for outputting these items.
This is a quick fix.
The problem that occurs is the following. After loadbalancing the
FieldPropsManager only has a non-nullptr to the FieldProperties on
the process with rank=0. Therefore most of the calls to it will
segfault. This has not been problem before changeset
2921838895. We did not make any calls to e.g.
FieldProperties::apply_schedule_keywords during a simulation.
Now we do via [eclproblem.hh#L1021-L1022](/OPM/opm-simulators/blob/3e4e62bc4f6f6f8a02eb2d2e4976dc2e2d956313/ebos/eclproblem.hh#L1021-L1022))
The maximum supply of lift gas and/or the maximum total gas flow
rate of a group may be defaulted in the context of lift gas
optimisation. In that case, the std::optional<> objects will be
empty and we must not access the contained object through the
value() member function. We output the sentinel value "-10" in this
case.
This is more self-documenting that numeric item indices. While
here, also switch to constructing the EQUIL record in place instead
of forming a temporary and moving it into the vector.
This commit ensures that we do not access the well's BHP reference
depth unless that value is well defined. If the input deck does not
define an explicit BHP reference depth (WELSPECS(5)), and if the
well does not have any active reservoir connections, then we are
supposed to output the usual sentinel value (-1.0E+20) as the datum
depth in 'SWEL'.
Account for this altered semantics in the restart reader too.
