* origin/master:
Bugfix: use correct object in sequential model.
Bugfix: properly handle two-phase cases.
Bugfix: pass both normal and extra data to output.
Update well data output integration.
Adapt to API change in opm-parser.
Ensure logging only on first rank.
Add isIORank_ member to avoid repeated calls.
Rename isRankZero() -> isIORank() for consistency.
Ensure only first-rank logging.
Add isRankZero() utility.
Bugfix: missing return.
do not set limits for prt log and set correct value for Note.
correct function call order and add whitespace.
supprot MESSAGES default vaule.
wellToState reads new opm-output data exchange
Write control to data::Wells. Missing ability to restore.
Avoid using buggy wellstate api
Stop report early if there are no wells
Update to interface change in opm-output
Includes the following changes:
- update to match API change in opm-output (vector not map for data::Wells::completions),
- restore WellStateFullyImplicitBlackoil::perfPhaseRates() from output,
- restore WellStateFullyImplicitBlackoil::currentControls() from output.
Remaining non-restored well-related data are:
- well potentials,
- the dynamic list of econ-limited completions.
the 'tempV' object currently only contains garbagee but the condition
is needed and will thus be fixed in a different PR.
thanks goes to @totto82 for the clarifications.
Keep track of whether it is a restart or not and invalidate the
intensive quantitiesCache in ebos when restarting the timestep due to
convergence issues.
note that I don't know if these changes are semantically correct (I
doubt it), but this patch fixes the valgrind complaints I saw for
SPE9_CP and on Norne.
Also, this makes the timing of flow_ebos for SPE9 determinisic between
runs: without this, I got some random time steps fail in a given run
and in the next run a completely different set of timesteps
failed. Since this was on the same same machine, without any
recompiles and no changes to the deck or any other input parameters, I
initially attributed the behavior to cosmic rays ;)
@totto82: could you have a thorough look on this?
once again, the culprit was an API change in opm-output. I suppose
this means that the frankenstein branch should be merged soon after
the release in order to avoid to continue breaking the flow_ebos build
twice a week.
* master: (42 commits)
Let only one rank write to step_timing.txt
Do not refer users to issue tracker if multiple procs log.
Remove unused variable.
Use vector instead of VLA, also add missing includes.
changed: bundle eigen3 in the original tarball for debian
update redhat6 packaging
Bugfix parallel computation of weighted pressure etc.
Fixed uninitialized bug, and added logging/comment
Removed superfluous std::move
Refactoring
Initial version of summary data
Do not store collective communication in the wells object.
Make sure that updateWellControls is called on each process.
Make WellSwitchingLogger work with DUNE 2.3
Schedule::getGroup returns reference, not pointer
Removed warning in WellSwitchLogger::calculateMessageSize
Correctly initialize MPI for multisegment wells test
Changed some names in WellSwitchingLogger
Use speaking name for bool in getCellData
Whitespace and other formatting changes
...
All ranks were still writing to step_timing.txt at the same time.
This made it unusable for parallel runs. With this commit only
one processes writes to this file.
This is currently still happening due to the implementation of
OPM_THROW whenever the linear solver does not converge. This
happens quite often and we might not want to get overwhelmed by
the issue tracker.
Tor Harald Sandve
Robert Kloefkorn
Andreas Lauser
Atgeirr Flø Rasmussen
Liu Ming
Jørgen Kvalsvik
Joakim Hove
Markus Blatt