Feature/ms wells - part 2: Solving, straightforward option #5746
Conversation
|
jenkins build this failure_report please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
e19bdf6
to
8757d8c
Compare
|
jenkins build this failure_report please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
8757d8c
to
9c055ce
Compare
|
jenkins build this failure_report please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
2 times, most recently
from
10e3d24
to
8b9be1d
Compare
|
jenkins build this failure_report please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
4 times, most recently
from
6a8e50a
to
89530ad
Compare
|
jenkins build this opm-tests=1250 failure_report please |
1 similar comment
|
jenkins build this opm-tests=1250 failure_report please |
|
jenkins build this opm-tests=1250 please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
89530ad
to
dcef937
Compare
|
jenkins build this opm-tests=1250 please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
dcef937
to
b3130be
Compare
|
jenkins build this opm-tests=1250 please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
b3130be
to
9e36faa
Compare
|
jenkins build this opm-tests=1250 please |
…Test.sh via the flag -s
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
9e36faa
to
8e01771
Compare
lisajulia
changed the title
lisajulia
changed the title
|
jenkins build this opm-tests=1250 please |
|
jenkins build this failure_report opm-tests=1250 please |
| return mswellhelpers::applyUMFPack(*duneDSolver_, rhs); | ||
| } | ||
|
|
||
| template<class Scalar, int numWellEq, int numEq> | ||
| void MultisegmentWellEquations<Scalar,numWellEq,numEq>:: | ||
| recoverSolutionWell(const BVector& x, BVectorWell& xw) const | ||
| { | ||
| BVectorWell resWell = resWell_; | ||
| BVectorWell Bx(duneB_.N()); |
There was a problem hiding this comment.
I would recommend to use the old way, when the well is on one process here.
Maybe we can also factor this out similar as for standard wells, or refactor that code to reuse here.
I am a bit late, sorry.
There was a problem hiding this comment.
Thanks! I'll try that!
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
8e01771
to
8645660
Compare
|
jenkins build this failure_report opm-tests=1250 please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
8645660
to
056e2fe
Compare
|
jenkins build this failure_report opm-tests=1250 please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
2 times, most recently
from
496fcca
to
c9f26cd
Compare
|
jenkins build this failure_report opm-tests=1250 please |
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
c9f26cd
to
f694f9d
Compare
|
jenkins build this opm-tests=1250 please |
… true, do not throw when initalizing distributed multi-segment wells in WellState.cpp
Here we go from cells to segments, and everything concerning segments is stored globally.
…olutionWell Here we go from cells to segments, and everything concerning segments is stored globally.
… can do this on all processes
lisajulia
force-pushed
the
feature/ms-wells-solving
branch
from
f694f9d
to
4801a17
Compare
|
jenkins build this opm-tests=1250 please |
|
jenkins build this opm-tests=1250 serial please |
This PR is based on #5680.
This PR enables simulations with MSWell distributed across several processes. The PR #5680 covered the assembly of the system, this PR covers the straightforward option of solving.
This PR adds communication when multiplying x_well with the matrix B of the Schur complement as explained on slide 13 of 24-10-21-MSWells.pdf.
This approach is the easiest way to solve, using preconditioners for the well system is to be implemented in the future.
I've added two tests comparing the SMRY files of a parallel and a sequential simulation with distributet MSWells. These tests fail in the PR #5747 because they do not run through, for this PR they pass.