Commit Graph

26 Commits

Author SHA1 Message Date
Kai Bao
1a7b5571b6 trying to add an inner well iteration to speed up convergence. 2017-10-12 14:10:24 +02:00
Kai Bao
c3a368e58e adding two more parameters related to ms wells
to help to tune and improve the convergence.
2017-10-12 13:43:04 +02:00
Kai Bao
6fb459a797 introducing tolerance_pressure_ms_wells for pressure equations
for the multisegment wells. More detailedly, we should distinguish the
rate control equations, bhp control equations and ordinary pressure
equations.
2017-10-12 13:43:04 +02:00
Kai Bao
0dc041d46a adding a use_multisegment_well_
to determine whether we will treat mutlisegment wells with
MultisegmentWell well model. Currently, it will be faulse by default.
It might be changed to be true by default when considering the
multisegment well model is well tested.
2017-10-12 13:37:05 +02:00
Andreas Lauser
e3d0bbc144 specify the number of iterations where we use a "strict" Newton convergence criterion using a separate parameter
that parameter is called "max_strict_iter". This increases the
flexibility of this slightly and it avoids screwing up the default
value for the "max_iter" parameter in the future. The credits for this
patch go to [at]atgeirr for proposing it.
2017-06-14 14:22:22 +02:00
Andreas Lauser
0304183361 Newton: fail earlier
i.e. give up after 10 instead of 15 Newton iterations. since this now
also causes the CNV convergence criterion to be ignored after 8
instead of after 13 Newton iterations, and because CNV stagnation
seems to be the cause for the vast majority of the failures, it does
not lead to worse results or more time step chopping on Norne but
causes fewer wasted iterations.
2017-06-14 14:22:22 +02:00
Andreas Lauser
e58fed9956 ignore the CNV convergence criterion for the last few newton iterations
the cell based residual can cause problems for cells with
almost-singular shapes and/or small pore volume.
2017-06-14 14:22:22 +02:00
Arne Morten Kvarving
3c0cb9e950 adjust for changed ParameterGroup namespacing 2017-04-28 15:36:25 +02:00
Kai Bao
e0d96b3410 removing compute_well_potentials_ from BlackoilModelParameters 2017-04-11 16:51:16 +02:00
Kai Bao
4159542186 setting tolerance_wells to be 1.e-4.
for better convergence behavoir.
2017-01-31 10:50:10 +01:00
Tor Harald Sandve
e614c42818 Avoid hard-coded appelyard parameter in flow_ebos
Start using the parameters in the parameter file instead of the hard-
coded ones.
Unify the restriction of the drs and drv in black-oil and black-oil-
solvent updateState
2016-12-05 13:44:33 +01:00
Tor Harald Sandve
294ca31fc8 Fixes in the Appelyard in updateState and updateWellState
1) changes dp_max_rel default to 0.2
2) introduces a dbhp_max_rel paramter to restrict the bhp update in the
updateWellState() (instead of using the dp_max_rel) Default is set to
1.0
3) Restrict rs and rv between 0,and the satruation value
4) Set rs and rv to zero for the water only cases
5) Guard against zero rs and rv when calcuating the maximum allowed rs
and rv change.

Tested on norne, model 2 and model 2.2

Number of problems for the different models with and without this fix
Case
this PR master
Norne
10
45
Model 2
21
78
Model 2.2
200
248
2016-12-05 13:44:33 +01:00
Robert Kloefkorn
feea8c1753 Introduce parameter for time step in days when single precision should be used
in the linear solvers. Also, a parameter was introduced to toggle the use of AMG.
2016-11-21 17:18:24 +01:00
Andreas Lauser
b8a0bb6aea Merge pull request #926 from totto82/frankenstein_revertChanges
Revert som changes in default values
2016-11-17 13:46:37 +01:00
Tor Harald Sandve
b66bb27a65 Revert som changes in default values
Make sure flow in Frankenstein and flow in master uses the same default
values
2016-11-16 15:09:54 +01:00
Andreas Lauser
fc4dfea185 Merge remote-tracking branch 'origin/master' into frankenstein
* origin/master:
  checking the residual for the well control equations.
2016-11-15 12:27:04 +01:00
Tor Harald Sandve
739c0906ef Improvments in convergence for flow_ebos
- restrict pressure changes. Set default to 1.0 (this also effects flow)
- change default number of linear iterations to 150
- tell stabilized newton the residual occilates even if it occilates in
only one phase (this also effects flow)
- avoid problems realated to division on small numbers

Tested on SPE9, norne and Model 2 with significant improvments.
2016-11-14 13:26:38 +01:00
Kai Bao
ce8498c4d7 checking the residual for the well control equations. 2016-11-11 13:26:22 +01:00
Andreas Lauser
4d63a4b0ba BlackoilModelParameters: add the file name of the deck
this is needed by the ebos based model.
2016-08-09 12:23:29 +02:00
Atgeirr Flø Rasmussen
3776ccab9e Add use_update_stabilization parameter.
This makes it easier to run without, for example for debugging.
The default is 'true', preserving existing behaviour.
2016-06-17 10:44:53 +02:00
Tor Harald Sandve
e9b097cbf4 Only compute well potential if needed
A boolen user parameter is added to controll the computation of well
potential.

This is a temporary fix to assure that no extra computation time is used
on well potential calculation if it is not needed. The long term fix
will require a more thorough revising of the well group implementation.
2016-04-19 16:24:28 +02:00
Kai Bao
b1b9c171b3 update scaling factors for mass balance equations. 2015-09-08 10:57:18 +02:00
Tor Harald Sandve
fd48a6e004 Change defaults
Use the more stable interleaved solver instead of CPR as default
Solve well equation initially as default
2015-09-03 12:38:03 +02:00
Kai Bao
f8e35535fc using lower case for solve_wellEq_initially_ 2015-06-16 17:03:53 +02:00
Tor Harald Sandve
8a166ebbd6 Add option for solving the wellEq seperatly
The well equations is solved seperatly in order to provide more accurate
initinal values to the reservoir model.
2015-06-16 17:03:52 +02:00
Atgeirr Flø Rasmussen
92ab1d7974 Use separate files for model parameter struct. 2015-05-24 17:33:06 +02:00