This restores the performance to approximately the level it had before
the change to support non-diagonal well jacobians, for SPE9. All changes
are to the eliminateVariable() method.
- Explicitly compute and apply the inverse.
- Change loop ordering to apply inverse only num_eq - 1 times, instead
of (numeq - 1)^2.
- Use UmfPackLU instead of SparseLU.
1. The right hand side is solved only once
2. The solver is constructed directly with the matrix
3. const is added where it was missing
4. More commennts is added
5. Variable names are changed for clarification
The following comments has been adressed
1. An array is used in stead of pair
2. is not empty is used instead of size>0 to check if the well has been
initialized before
3. const_iterator is used instead of iterator
4. partial copy is removed
5. WellMapType is no longer mutable
If we need more than 150 linear iterations, it's probably something
wrong, and we may be better of by restarting with smaller time-steps.
TODO: make it possible for the user to specify this number.
The non-diagonal elements in the sub-matrices in the Schur complement is
no longer ignored. Instead of assuming the matrix do be diagonal, and
compute the invert of the sub-matrix, small linear systems are solved
using superLU.
Tested on SPE3 and Norne. (With this fix a slightly modified norne runs
until 3292 days)
The residual output is changed from max residual to
total mass balance residual. In this way the output
resemble what is actually used as convergence criteria.
With the introduction of shut wells the same ordering in the well states
can no longer be assumed. The well names is instead used to map the old
well state to the new. Also the partial_copying is moved into the
initialization.
Tested on SPE1, SPE3, SPE9 and Norne. (Do not change the SPEs and is
nessesary for the Norne)
The old rs and rv values are used in the phase transitions calculations.
The update of rs and rv to the state is therefore moved after the phase
transitions in order for the code to use the old values.
of the matrix internal allocators.
This fix also avoid the copying of the BCRSMatrix by providing a contructor that creates
the DuneMatrix for a given Eigen SparseMatrix.
