The initial solution in ebos and in flowebos are different in cases where
swatinit is present. Pass the initial solution and recalculate the
intensive quantities make sure that the flowEbos initial solution is
used.
the most significant change is that only a "PhaseUsage" object must be
passed to its constructor instead of a full "fluid object". also, the
pointers to the vector of "active" phases has been converted into a
full object. (this helps to avoid potential use-after-free errors.)
* https://github.com/OPM/opm-simulators:
[bugfix] add defunct_well_names to BlackoilModelEbos.
[bugfix] fix ownerMask for parallel FIP code.
flow_ebos: tell the ebos in ourselves to not handle SWATINIT
[bugfix] Make initialization work in parallel for flow_ebos.
[bugfix] make flow_ebos work when no wells are present.
flow_ebos: only instantiate the grid once
this patch also includes code to make FlowMainEbos more autonomous
(i.e. it does not derive from FlowMainBase anymore). this allows more
flexibility how and what stuff gets created.
this changes the conversion of the results to hopefully make it work
with twophase simulations. Note that flow_ebos still does not work
because there is a crash in the well model that I don't understand
(and also, I've disabled an assertation in the well model plus I'm not
completely sure if the conversion code need to do something about the
primary variables in the twophase case).
mainly, the "linsolver_" attribute is called "istlSolver_" in the
BlackoilModelEbos class. Also, this problem possibly only occured if
MPI was enabled...
- 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.
I'm not really sure if the fix is semantically correct (Dune
communicators are strange beasts if you're new to them), but at least
this patch fixes the build without MPI and the resulting flow_ebos
binary seems to work fine.
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?
* remotes/totto82/frankenstein_mod:
Fix seg-fault for cases without wells
Some micro performance improvments and cleaning
Add THP support in the denseAD well model
Only solve the linear system when it is not converged.
Revert changes to NewtonIterationBlackoilInterleaved.cpp
add and use class wellModelMatrixAdapter
Remove unused code and remove Eigen vectors
New updateState
Some cleaning and small changes
almost all of them were caused by recent changes in the master
branch:
- there were methods added which depend on the types `V` and
`DataBlock`. these do not make much sense in the context of the
frankenstein simulator. Also, these types are defined globally for the
whole Opm namespace in `BlackoilModelBase_impl.hpp` (which should be
prosecuted as a fellony IMO)! Besides this, their names are useless;
'V' is the letter which comes after `U` in the alphabet and when it
comes to computers basically everything can be seen as a chunk of data
(i.e., a `DataBlock`).
- it seems like the new and shiny dense-AD based well model was never
compiled with assertations enabled, at least some asserts referenced
non-existing variables.
- the recent output-related API changes were pretty unfortunate
because they had the effect of tying the (sub-optimal, IMO) internal
structure of the model even closer to the output code: as far as I can
see, `rq` does only make sense if the model works *exactly* like
BlackoilModelBase and friends. (for flow_ebos, this could be
replicated, but first it would be another unnecessary conversion step
and second, most of the quantities in `rq` are of type `ADB` and much
of the "frankenstein" excercise is devoted to getting rid of these.) I
thus reverted back to an old version of the output code and created a
`frankenstein` branch in my personal `opm-output` github fork.
-- The jacobian and residual in the reservoir is updated directly
-- The sparsity pattern are provided to the well matrices.
-- Some cleaning in updateWellState()
-- isRS and phaseCondition is removed and hydroCarbonState in the state
is used instead
-- input of pressurediffs to computeHydrostaticCorrection() is changed
to double from Vector in WellHelpers.hpp
-- a new updateState is implemented based on dune vectors
-- the old is kept for comparision in this PR
-- the updateState is not identical.
Tested on spe1, spe9 and norne and it improves the convergence compares
to the old one.
- unused code is removed
- the scaled normed is stored in residual_norm_history for usage in
stabilized newton
- number of linear iterations is outputted
- linear solver tolerance is reduced to 0.01
- make compute wellFlux local
- rewrite ADB::V to std::vector<double>
remove the now unnecessary inclusion of
"NewtonIterationBlackoilInterleaved.cpp" (mind the .cpp extension!)
and include "dune/istl/solvers.hh" instead.
1) The wellsolution is stored in wellVariables as a vector of DenseAD
objects. The wellVariables is computed from the wellstate in
setWellVariables()
2) BHP and well fluxes are not stored directly but calculated based on
the wellVariables
3) The initial well accumulation term is stored as a vector of doubles.
4) addWellFluxEq() uses DenseAd flux and accumulation terms
5) computePropertiesForWellConnectionPressures() no longer uses state as
input. The wellstate is used to get the bhp pressure.
6) The current wellcontrol is set when updated in updateWellControls()
in order to use well_controls_get_current_type(wc) insted of
well_controls_iget_type(wc, currentIdx)
Tested on SPE1, SPE9 and Norne.
Effects the convergence but the results are identical.
Start using the well model desribed in SPE 12259 "Enhancements to the
Strongly Coupled, Fully Implicit Well Model: Wellbore CrossFlow Modeling
and Collective Well Control"
The new well model uses three well primary variables: (the old one had
4)
1) bhp for rate controlled wells or total_rate for bhp controlled wells
2) flowing fraction of water Fw = g_w * Q_w / (g_w * Q_w + q_o * Q_o +
q_g + Q_g)
3) flowing fraction of gas Fg = g_g * Q_g / (g_w * Q_w + q_o * Q_o +
q_g + Q_g)
where g_g = 0.01 and q_w = q_o = 1;
Note 1:
This is the starting point of implementing a well model using denseAD
The plan is to gradually remove Eigen from the well model and instead
use the fluidState object provided by the Ebos simulator
Note 2:
This is still work in progress and substantial cleaning and testing is
still needed.
Note 3: Runs SPE1, SPE9 and norne until 950 days.
for the gas-only case with vaporized oil enabled, the oil pressure was
used, but ebos uses the gas pressure as a primary variable in that
case. (the reason is that it is faster to check if oil appears if the
gas pressure is available in that case.) This patch corrects that and
does no longer use the PrimaryVariables::set$FOO_PV() methods because
these methods are hard to keep in sync with the indices for the
primary-variables-as-a-Dune::FieldVector feature (which is actually
quite important for the linear solver).
it uses ebos for linearization of the mass balance equations and the
current flow code from opm-simulators for all the rest. currently, the
results match the ones from plain `flow` for SPE1, SPE9 and Norne, but
performance is not optimal: on SPE9, converting from and to the legacy
data structures takes about a third of the time to do the actual mass
balance assembly. nevertheless `flow_ebos` is almost as fast as plain
`flow` for SPE9. (for Norne `flow_ebos` is about 15% slower, even
though the results match quite closely. the reason for this is that it
requires more iterations for some reason.)
