OilfieldToolsNet/opm-simulators
4
0
Fork 0
mirror of https://github.com/OPM/opm-simulators.git synced 2025-01-15 22:51:55 -06:00
Code Issues Packages Projects Releases Wiki Activity
7,943 Commits 24 Branches 102 Tags 90 MiB
b9d7b69bd6
Commit Graph

3008 Commits

Author SHA1 Message Date
Andreas Lauser
58a1b7df1f adapt to the move of infrastructure from opm-common to opm-material 2018-02-07 13:32:52 +01:00
Markus Blatt
bf3132e9cf Fixes unused variable warning 2018-02-06 19:48:26 +01:00
Markus Blatt
cda0c22d0f Avoid negation in if clause 2018-02-06 15:36:23 +01:00
Markus Blatt
034b1840ae Fixed comment about smoother arguments needed for construction 2018-02-06 15:34:13 +01:00
Markus Blatt
4c590c17d6 Moved definition of CPRParameter to top of header as needed for constructor call. 2018-02-06 15:32:35 +01:00
Markus Blatt
fc9acf963e Spelling fixes in BlackoilAmg.hpp 2018-02-06 15:31:50 +01:00
Markus Blatt
14cde5bafc Made use of range based for for graph traversal. 2018-02-06 15:30:24 +01:00
Markus Blatt
a8dde44afc Removed dead code und commented code. 2018-02-06 15:28:10 +01:00
Markus Blatt
6bcfce1733 Prevent problems to due oversight in parallel AMG. 
Currently we run into issues with the parallel AMG if
redistribution happened on the coarsest level. That is
not detected by AMG and it will construct smoothers on
all processors present before the redistribution. Some of
them will get OwnerOverlapCommunication objects that have
an invalid MPI communicator in them (MPI_COMM_NULL) and an
MPI_ERROR will be raised as we communicate in the constructor.

With this patch we detect this situaton and set the pointer to
OwnerOverlapCommunication to null to prevent communication. A
sanity check that the matrix has zero rows has been added.
 2018-02-06 13:19:59 +01:00
Atgeirr Flø Rasmussen
532403c5fb
Merge pull request #1391 from totto82/RFIP 
Use FIP, EGRID and INIT output code in ebos.
 2018-02-06 12:54:01 +01:00
Markus Blatt
28e566ba02 [bugfix] Prevent indexing an array of zero size for matrices with zero rows. 2018-02-06 12:45:13 +01:00
Markus Blatt
cfa311e055 recover compilation support for dune versions <2.6 2018-02-05 22:09:37 +00:00
Markus Blatt
cf1cdbd375 Improved documentation 2018-02-05 22:37:01 +01:00
Markus Blatt
8fc20456c5 Allow user to request aggregating the pressure system before solving it. 
For this we added a new parameter cpr_pressure_aggregation. If it is true
we will aggregate the pressure already for the first system.
 2018-02-05 22:37:01 +01:00
Markus Blatt
e24d338ddc Resort to support the usual parameters (as for flow_legacy) 
These are solver_approach=cpr cpr_use_amg=(true|false), etc.
 2018-02-05 22:37:01 +01:00
Markus Blatt
c367382a59 Prepare to use ILU for pressure system. 2018-02-05 22:37:01 +01:00
Markus Blatt
abacbe2cfc Do not use expensive aggregation when forming pressure system unless requested. 
That means that for traditional CPR no aggregation is used and we simply extract
the pressure component.
 2018-02-05 22:37:01 +01:00
Markus Blatt
5d8da52679 Use decoupling strategy of Scheichl/Masson in the CPR preconditioner. 
Often it is claimed by CPR-AMG evangelists that this will make the pressure
system more elliptic. That may be the case. But even more important it also
decouples the pressure from the saturations.

Without this approach the pressure correction influences the smoothing of the
full system too much and e.g. for Norne CPR is worse than simple block ILU0.
 2018-02-05 22:37:01 +01:00
Markus Blatt
6d21214fa7 Switch to ParallelOverlappingILU0 for CPRPreconditioner. 
This seems to have been forgotten previously. Now the code int CPRPreconditioner.hpp
uses ParallelOverlappingILU0 instead of SeqILU[0n]/BlockPreconditioner which
makes the code more slim.
 2018-02-05 22:37:01 +01:00
Markus Blatt
e6e18a3aa9 Make Blackoil system amg / CPR (flow_ebos) DUNE 2.6 aware. 2018-02-05 22:37:01 +01:00
Markus Blatt
c089a59cba Fix compiler lookup of get method. 2018-02-05 22:37:01 +01:00
Markus Blatt
f5d81513da First version of a AMG for the Blackoil equations. 
The approach is inspired by Geiger's system-amg but we use dune-istl
aggregation AMG for it. On the fine level all unknowns attached to a cell
form a matrix block and are treated fully coupled. To form the first
coarse level system we use only the pressure component to guide the aggregation
and neglect all other unknowns on the fine level. All other level are formed
in the usual way by scalar aggregation.

Currently,it has to be requested for flow_ebos manually by passing
"linear_solver_use_amg=true amg_blackoil_system=true" to it.
 2018-02-05 22:37:01 +01:00
Arne Morten Kvarving
141186ad1d changed: opm/[core -> common]/utility/parameters 2018-01-30 16:33:45 +01:00
Arne Morten Kvarving
e9404486bb changed: opm/common/[->utility]/ResetLocale.hpp 2018-01-30 12:34:04 +01:00
Tor Harald Sandve
cdeefc3a34 Fix command line argument for double precision restart output 2018-01-29 08:56:55 +01:00
Tor Harald Sandve
84a8b8eca6 Pass empty regionSummaryData and blockSummaryData to output 2018-01-29 08:56:55 +01:00
Tor Harald Sandve
6d0c716d76 Clean-up FIP 2018-01-29 08:56:55 +01:00
Tor Harald Sandve
36f6b7ad00 Remove writInit() 
The EGRID and INIT files are written using ebos
 2018-01-29 08:56:55 +01:00
Andreas Lauser
386ade39f4 flow: let core ebos handle the output directory for the result files 
this adds a new parameter --ecl-output-dir=$FOO to the "virtual"
command line arguments of the ebos simulator.
 2018-01-29 08:54:30 +01:00
Tor Harald Sandve
137ff53ae7 Remove output of FIP 2018-01-29 08:54:30 +01:00
Arne Morten Kvarving
4945c9b2ed changed: pass the mcmg element layout as a parameter for dune 2.6 
the template parameter is deprecated
 2018-01-22 17:25:14 +01:00
Arne Morten Kvarving
2aa0043550 changed: use Dune::createScalarProduct for dune 2.6 
ScalarProductChooser is no more
 2018-01-22 17:24:58 +01:00
Arne Morten Kvarving
4315665b64 changed: expose category as member for dune 2.6 
the enum can only be used with a define, and that would
just be a temporary solution in any case
 2018-01-22 17:23:22 +01:00
Andreas Lauser
f34cfafc22
Merge pull request #1377 from totto82/useEbosEclOutput 
Start using ecl output from Ebos
 2018-01-10 15:01:27 +01:00
Atgeirr Flø Rasmussen
6c8b0d68ca Remove use of old implicit imcomp 2p transport. 2018-01-08 17:23:43 +01:00
Atgeirr Flø Rasmussen
aca95895c9 Fix: set precision of the correct stream. 2018-01-08 17:18:47 +01:00
Tor Harald Sandve
0e6fe26a61 Start using ecl output from Ebos 
The wells, FIP and initial output of NNCs is still handled
by code in opm-simulators. The plan is to move more of the
functionality to ebos.

All tests pass and MPI restart works
 2018-01-04 09:29:58 +01:00
Tor Harald Sandve
969d8f238d Use phase and comp info from FluidSystem 
TODO: The output, fip and restart still uses a mixture of old and
new phase indices. This needs to be adressed in future PRs
 2018-01-03 08:44:37 +01:00
Atgeirr Flø Rasmussen
9f14b63b82
Merge pull request #1370 from totto82/changeDpMaxDefault 
Change dp_max_rel default from 1.0 to 0.3
 2017-12-20 16:59:43 +01:00
Tor Harald Sandve
7567748f2d Adapt to changed initial FluidState type in ebos 2017-12-15 08:19:01 +01:00
Tor Harald Sandve
4097a07572 Change dp_max_rel default from 1.0 to 0.3 2017-12-14 11:16:16 +01:00
Arne Morten Kvarving
dfd91d162b quell signed/unsigned comparison mismatch warning 2017-12-05 16:50:58 +01:00
Arne Morten Kvarving
b702d3834b remove unused variable 2017-12-05 16:50:52 +01:00
Atgeirr Flø Rasmussen
03a0baf447
Merge pull request #1343 from GitPaean/single_phase_resv 
making the single phase RESV injection work.
 2017-12-05 14:06:44 +01:00
Tor Harald Sandve
095b82212c Use the initial solution from Ebos 2017-12-04 10:35:30 +01:00
Atgeirr Flø Rasmussen
65034250d1
Merge pull request #1356 from dr-robertk/PR/fix-minor-issue-gridview 
[bugfix] make code compile when GridView is not Grid::LeafGridView.
 2017-12-04 09:25:43 +01:00
Kai Bao
9317c1f023 removing the current argument in updateWellStateWithTarget 
and some other cleaning up.
 2017-11-30 17:14:29 +01:00
Kai Bao
ea3cbd1fe8 removing numComponents() from WellInterface 
which is dumplicated from BlackoilWellModel.
 2017-11-30 16:31:48 +01:00
Robert Kloefkorn
3a35a495a8 [bugfix] make code compile when GridView is not Grid::LeafGridView. 2017-11-29 23:26:38 +01:00
Atgeirr Flø Rasmussen
eef7b8ea94
Merge pull request #1348 from totto82/refactorBlackoilModelEbos 
Put the ebosSimulator start and end methods where it belongs
 2017-11-29 20:49:18 +01:00
Tor Harald Sandve
2d2f9285c5 Only read use_TUNING param once. 
Just to clean up the .DEBUG file.
 2017-11-28 14:25:25 +01:00
Kai Bao
7b61f3e692 moving polymer related to updateWaterMobilityWithPolymer 
in StandardWell.

And not calculating the shear effect for the injecting wells when the
injection wells do not inject polymer.
 2017-11-27 09:24:02 +01:00
Atgeirr Flø Rasmussen
46e958af43
Merge pull request #1337 from totto82/removeBlackOilPropsFromInit 
Adapt to the refactoring of the Equil initialization code
 2017-11-27 07:26:22 +01:00
Atgeirr Flø Rasmussen
f62275cb06
Merge pull request #1339 from blattms/write-ranks-without-nonactive 
Disregard non-active cells when writing MPI_RANK in writeInitial.
 2017-11-27 07:22:34 +01:00
Atgeirr Flø Rasmussen
401aa28c08
Merge pull request #1340 from totto82/solventAndResv 
Make it possible to combine solvent and RESV
 2017-11-27 07:20:24 +01:00
Atgeirr Flø Rasmussen
3fd27d9bd6
Merge pull request #1346 from GitPaean/removing_more_multisegment_legacy 
removing more legacy Multisegment Wells related.
 2017-11-24 15:30:39 +01:00
Tor Harald Sandve
9edb351d34 Put the ebosSimulator start and end methods were it belongs 
Move the beginEpisode/endEpisode call to ebos to beginReportStep/
endReportStep and beginTimeStep/endTimeStep call to ebos to prepareStep/
afterStep
 2017-11-24 15:16:03 +01:00
Kai Bao
f87e6d35db removing more legacy Multisegment Wells related. 2017-11-24 12:27:43 +01:00
Kai Bao
59722e552c making the single phase RESV injection work. 2017-11-24 11:44:53 +01:00
Tor Harald Sandve
ba358c0522 Only update TUNING when TUNING_CHANGE event 2017-11-23 14:24:40 +01:00
Tor Harald Sandve
dd97fc8908 Some clean up. 2017-11-23 11:15:50 +01:00
Tor Harald Sandve
c076ed6485 Move Rateconverter and pvtIdx to the wellInterface 2017-11-23 08:37:30 +01:00
Tor Harald Sandve
054361d537 Make it possible to combine solvent and RESV 
Compute the conversion factor for solvent using the RateConverter.hpp
 2017-11-22 15:21:21 +01:00
Markus Blatt
f32743cf31 Disregard non-active cells when writing MPI_RANK in writeInitial. 
This saves space and it is allowed and recommended (spacewise) to
use integer arrays with only values for active cells in writeInitial.
 2017-11-22 14:18:32 +01:00
Atgeirr Flø Rasmussen
796f6cd13c
Merge pull request #1338 from totto82/scheduleTUNING 
Support TUNING in schedule section.
 2017-11-22 14:03:58 +01:00
Andreas Lauser
ef76d0a8cc
Merge pull request #1335 from atgeirr/refactor-sim 
Removing unused code parts and minor refactoring
 2017-11-22 13:57:58 +01:00
Markus Blatt
9402f4bde1 Make CellOwnerHandle available without MPI 
and thus make compilation succeed in this case as using
it is perfectly find even if MPI is not there.
 2017-11-21 15:06:36 +01:00
Tor Harald Sandve
2a7d7b8ae5 Pass ref instead of shared pointer to initEquil 2017-11-21 14:29:08 +01:00
Tor Harald Sandve
31bfb22273 Adapt to the refactoring of the Equil initialization code 
Most noteably the blackoilState is not returned by the initialization
code anymore, instead the initialization class is returned.
 2017-11-21 13:52:07 +01:00
Markus Blatt
e7688a2ca0 Use MPI_RANK as ECL keyword (max length is 8) 
MPI_RANKS was one character to long and rose an exception with message
"Error: Program threw an exception: Keyword is too long."
 2017-11-21 11:17:40 +01:00
Tor Harald Sandve
c03dbc1693 Support TUNING in schdule section. 2017-11-21 11:12:16 +01:00
Markus Blatt
073f63ac4c Adapt writeInitial call to use integer vectors for printing ranks 
This adapts to the new function signature as of PR 216 in opm-output. It uses the
newly introduced map of integer vectors to print the MPI ranks in a parallel run.
 2017-11-20 15:30:50 +01:00
Atgeirr Flø Rasmussen
6e8adbbd7d Remove old "backup-restore" functionality. 
Was no longer working and would require some maintenance to work again.
 2017-11-20 14:31:13 +01:00
Atgeirr Flø Rasmussen
353f4a6265 Move creating local fipnum array to constructor. 2017-11-17 15:42:16 +01:00
Atgeirr Flø Rasmussen
33833eebe3 Align PRT message header. 2017-11-17 15:41:46 +01:00
Atgeirr Flø Rasmussen
ed263ba030 Remove unused function argument. 2017-11-17 15:41:23 +01:00
Atgeirr Flø Rasmussen
8bd441f431 Remove obsolete time and convergence output. 2017-11-17 15:28:52 +01:00
Alf B. Rustad
7dba2c4d70 Format fix-up 2017-11-17 08:25:22 +01:00
Alf B. Rustad
18419be734 Moving developer centric messages from PRT file to DEBUG file 2017-11-17 08:25:22 +01:00
Alf B. Rustad
ad68c05982 Change substep to time step in logging 2017-11-17 08:25:21 +01:00
Alf B. Rustad
2d2198a775 Change log output from Time step to Report step 2017-11-17 08:25:21 +01:00
Atgeirr Flø Rasmussen
68dab29aec
Merge pull request #1333 from GitPaean/avoid_use_BlackoilPhases 
replacing BlackoilPhases:: with Oil Water Gas in well model
 2017-11-16 21:49:30 +01:00
Kai Bao
8274fc9275 replacing BlackoilPhases:: with Oil Water Gas in well model 
and also WellStateFullyImplicitBlackoil.

There are more places to do so, while it might be easier to create a
header file for this.
 2017-11-16 12:08:58 +01:00
Atgeirr Flø Rasmussen
d8ad13c4b5 Move fip computation and unit conversion to separate method. 2017-11-15 16:31:06 +01:00
Atgeirr Flø Rasmussen
4256af6754 Move (local) fipnum creation to separate method. 2017-11-15 16:13:28 +01:00
Atgeirr Flø Rasmussen
0f210dcb6c Move FIP output to separate method. 2017-11-15 15:55:58 +01:00
Atgeirr Flø Rasmussen
3d0c6da513
Merge pull request #1320 from joakim-hove/missing-features 
Missing features
 2017-11-14 11:19:13 +01:00
Joakim Hove
a6ef70cd91 Added keywords to list of unsupported. 2017-11-14 08:56:06 +01:00
Joakim Hove
d04c52cebf Reformatted list of unsupported keywords. 2017-11-14 08:55:52 +01:00
Atgeirr Flø Rasmussen
60ee67804b
Merge pull request #1330 from akva2/remove_simulators 
Remove duplicate EBOS based simulators and legacy simulators that has been replaced
 2017-11-13 15:56:33 +01:00
Arne Morten Kvarving
e0907ddc49 remove 'flow_solvent' and associated files 
use 'flow'
 2017-11-13 14:21:10 +01:00
Arne Morten Kvarving
ad23d98726 remove 'flow_multisegment' and associated files 
use 'flow' with use_multisegment_well=true
 2017-11-13 14:20:47 +01:00
Atgeirr Flø Rasmussen
098e229d2e
Merge pull request #1328 from totto82/fixShearVel 
Fix size of perf radius and length vectors.
 2017-11-13 14:00:07 +01:00
Tor Harald Sandve
614f75d9b9 Fix size of perf radius and length vectors. 2017-11-13 12:16:39 +01:00
Atgeirr Flø Rasmussen
29277489e1 Make calling wsolvent() and wpolymer() always ok. 
With the existing code, it was a bug to call wsolvent() when the well
was not a gas injector (and similar for wpolymer and water injectors).
Since such calls could happen, this minor fix removes the assertation
failure in those cases.
 2017-11-13 10:40:09 +01:00
Atgeirr Flø Rasmussen
6591ca89f6
Merge pull request #1324 from totto82/moveApplyWellToRes 
Apply the well residuals to cell residuals in SolveJacobianSystem
 2017-11-10 09:57:42 +01:00
Atgeirr Flø Rasmussen
e7a6e8f2c5
Merge pull request #1322 from totto82/fixResetSolution 
BUGFIX Update the solution variable in ebos
 2017-11-09 10:35:38 +01:00
Tor Harald Sandve
5647485daa Apply the well residuals to cell residuals in SolveJacobianSystem 
This was moved to assemble() some time ago but according to my logic it
belongs in the solve method since this is only a trick to solve the well
equations simultaneously with the reservoir equations.

A buggy side effect of the currect implementation was that the well
residual was considered in the convergence test. I.e. this move changes
the convergence behaviour of the simulator.
 2017-11-09 08:40:08 +01:00
Tor Harald Sandve
a73208f41c Address issues raised in the review of the PR 2017-11-08 15:48:30 +01:00
Tor Harald Sandve
4bb5213f37 BUGFIX Update the solution variable in ebos 
Updating the solution variable in updateState() was conceptually wrong
and lead to incorrect results if the linear solver failed before the
updateState() method was called. i.e. in the first iteration.
 2017-11-08 14:21:31 +01:00