opm-simulators Support opening of following well

Add support for item 9 in WECON

Depends on https://github.com/OPM/opm-common/pull/4507

Mar 06 '25 07:03 totto82

In the current code we sometimes check the wellstate and sometimes the well objects in schedule for the well status. With this (and maybe already with the introduction of WCYCLE) they are not necessary the same. If the approach suggested here is acceptable, we need to replace status checks based on the well objects in the schedule with the ones from the well state. Before doing that I would like some feedback. (Especially from @bska).

Mar 06 '25 07:03 totto82

This PR needs to be rebased.

Jun 18 '25 13:06 GitPaean

I think we should merge #6051 first. I will rebase that one.

Jun 18 '25 14:06 totto82

Test case in https://github.com/OPM/opm-tests/pull/1301

Aug 26 '25 09:08 totto82

This or upstream broke norne in serial, https://ci.opm-project.org/job/opm-simulators/2917/ and likely https://ci.opm-project.org/job/opm-nightly-pypi-test/732/

Sep 17 '25 06:09 akva2

Post-commit build is broken now. Either fix immediately or back this out.

Sep 17 '25 07:09 bska

I am working on it. At the failing timestep, the same well is both SHUT and OPEN. I will post a fix asap.

Sep 17 '25 08:09 totto82

This or upstream broke norne in serial, https://ci.opm-project.org/job/opm-simulators/2917/ and likely https://ci.opm-project.org/job/opm-nightly-pypi-test/732/

please provide some education, those are tests separately from the regression tests, right?

Sep 17 '25 08:09 GitPaean

yes, the post builder runs norne in serial and parallel. The pypi test is a bit more fuzzy as that is also executed as a regression test, so the cause there may or may not be the same.

Sep 17 '25 08:09 akva2

yes, the post builder runs norne in serial and parallel. The pypi test is a bit more fuzzy as that is also executed as a regression test, so the cause there may or may not be the same.

They looks like the same problem.

Sep 17 '25 08:09 GitPaean

yes, the post builder runs norne in serial and parallel. The pypi test is a bit more fuzzy as that is also executed as a regression test, so the cause there may or may not be the same.

norne takes three minutes on my computer to run serial, can be around one minutes with 4-mpi. Maybe we can have a running tests running part of schedule in our regular jenkins tests.

Sep 17 '25 09:09 GitPaean

it takes longer on a shared ci system.

jenkixs run norne norne_parallel pleasx

has always been available.

breakage happens. we pick it up relatively fast with the post-builder. i think it's a good compromise as-is.

Sep 17 '25 09:09 akva2