Split the operation duneB_.mmv(x, resWell) in recoverSolutionWell into its two parts

OPM / opm-simulators

OPM Flow and experimental simulators, including components such as well models etc.

http://www.opm-project.org

GNU General Public License v3.0

124 stars 122 forks source link

Split the operation duneB_.mmv(x, resWell) in recoverSolutionWell into its two parts #5741

Open lisajulia opened 1 week ago

lisajulia commented 1 week ago

This is needed to later add communication for the multiplication with the matrix duneB_, since here we go from cells to segments, and everything concerning segments is stored globally.

lisajulia commented 1 week ago

jenkins build this failure_report please

lisajulia commented 1 week ago

6 failing tests

Problems at least at:

01_9_3d_grpctl_msw_model2 WGIGR:INJ2
01_9_3d_grpctl_msw_model2 WGIP:INJ2
01_9_3d_grpctl_msw_model2 WWPR:PROD3
01_9_3d_grpctl_msw_model2 WBHP:PROD3

This might be related to the time step change as well, like here: https://github.com/OPM/opm-simulators/pull/5738#issuecomment-2475812339

No visible deviations at:

02_9_3e_grpctl_msw_model2
03_9_4d_grpctl_msw_model2
04_9_4c_grpctl_msw_model2
05_0a3_grpctl_msw_model2
06_wsegvalv

GitPaean commented 1 week ago

For the one has obvious deviation, you can check whether there is deviation if you use small time step, for example, maximum 1 day time step. If with small time step, it has a good match, then to a big extend, we can think it is related to the different time step lengths.

lisajulia commented 1 week ago

jenkins build this opm-tests=1248 failure_report please

GitPaean commented 1 week ago

@akva2 @hnil do you understand the reasoning here?

GitPaean commented 1 week ago

It is good to show this change might cause regressions, but I think this type of change can go in with the main PR together (including the PR #5738) so the picture is more clear.

lisajulia commented 1 week ago

It is good to show this change might cause regressions, but I think this type of change can go in with the main PR together (including the PR #5738) so the picture is more clear.

Yes, yet the PR https://github.com/OPM/opm-simulators/pull/5680 is for the assembly and I will open another one for solving briefly.

lisajulia commented 1 week ago

It is good to show this change might cause regressions, but I think this type of change can go in with the main PR together (including the PR #5738) so the picture is more clear.

Yes, yet the PR #5680 is for the assembly and I will open another one for solving briefly.

This is the one for solving: https://github.com/OPM/opm-simulators/pull/5746

lisajulia commented 6 days ago

jenkins build this please