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
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.
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.
The wellModel is now persistent over the time steps,
with an update method called every reportStep/episode.
This allows the following simplifications:
1. move the wellState to the WellModel
2. add a ref to the ebosSimulator to the wellModel
3. clean up the parameters passed to the wellModel methods
4. move RESV handling to the WellModel and the rateConverter
5. move the econLimit update to the WellModel
this needs to merged before the SimulationParameter hack can be
removed from eWoms. the patch will cause spurious deprecation warnings
until then. (i.e., it is best to merge them together.)
After the restructuring of of the well model, keeping an extra class for
the "Dense" model is not needed. The only thing still left in
WellStateFullyImplicitBlackoilDense was some solvent related stuff, this
PR moves this to WellStateFullyImplicitBlackoil and removes
WellStateFullyImplicitBlackoilDense.
In addition to a cleaning code this PR fixes missing solvent well output.
Tested with SPE1.
TODO: with the current way, we are not outputting any well residual
information. We need to address what kind of residual information we
want to output with the new well model.
to update well state.
With this way, the BlackoilModelEbos does not need to know the data type
assocated with different well type.
It is not well tested yet.
1) Use the solution variable directly in RelativeChange(...)
2) Add a method in the RateConverter that takes the simulator instead of the state.
3) Pass the reservoir pressure directly to the well initialization.
4) Move convertInput(...) to SimulatorFullyImplicitBlackoilEbos.hpp.
This code is only used to convert the initial reservoir state.
5) Modify updateState(...). The solution variable is updated directly and adaptPrimaryVariable(...)
from ewoms is used to switch primary variables. An epsilon is passed to adaptPrimaryVarible(...) after a switch
of primary variables to make it harder to immediately switch back.
The following code used by flow_ebos still uses the reservoirState
1) the initialization
2) restart
3) output of the initial state
4) the step methods in AdaptiveTimeStepping and NonlinearSolver.
The reservoirState is not used by this methods, so after the initial step, an empty reservoirState is passed around in the code.
We do this by switching to output the global index of the cells.
In a first step the problematic cell indices are gather on process 0.
Then they are logged there.
This should prevent spurious ouput at the end of PRT and DEBUG files after the
simulation time is printed. This happened previously for some parallel
runs of model 2. Unfortunately, it seems these problems do not appear any
more for the current master. At least I could not reproduce them.
