dont call beginTimeStep when restarting simulation after failing

[totto82]

We do not want to redo the well testing, the pre-solving of wells, gas-lift optimization, network balancing, etc, after failed simulations

For now, this PR adds an if around the outer beginTimeStep. We may need a more fine-grained approach. However, initial testing on selected models is promising.

[totto82]

jenkins build this please

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[bska]

jenkins build this failure_report please

It doesn't show up in the failure report, so just in case you didn't see it, the MSW_2D_VERT_02 test does not complete and times out (1500 seconds, 25 minutes maximum runtime). The transcript suggests that the well solution fails at some point, but someone who knows this part of the code better than I will need to analyse the underlying problem.

[GitPaean]

jenkins build this failure_report please

It doesn't show up in the failure report, so just in case you didn't see it, the MSW_2D_VERT_02 test does not complete and times out (1500 seconds, 25 minutes maximum runtime). The transcript suggests that the well solution fails at some point, but someone who knows this part of the code better than I will need to analyse the underlying problem.

It looks like MSW_2D_VERT_02 got problem at the first time step with singular matrix,

https://ci.opm-project.org/job/opm-simulators-PR-builder/7783/testReport/junit/(root)/mpi/compareECLFiles_flow_MSW_2D_VERT_02/

[totto82]

It doesn't show up in the failure report, so just in case you didn't see it, the MSW_2D_VERT_02

Thanks for letting me know. I didn't notice this. I will investigate and report back.

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[vkip]

Think you need to globally communicate any_closed_wells_this_step (preferably an int equivalent?) to ensure all procs do the same thing.

[totto82]

Thanks for noticing. Fixed.

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please

[GitPaean]

jenkins build this failure_report please

[GitPaean]

a rebase is needed for this PR.

[totto82]

jenkins build this failure_report please

[totto82]

jenkins build this failure_report please