Tor Harald Sandve comments

Results 368 comments of


                                            Tor Harald Sandve

Also solve std for every inner iterations as default

Some spikes in the results are the most worrying differences. I have tried to reproduce them locally: udq_wconprod: WOPR:OPL02. Both master and PR give the same results locally (no spike)...

Also solve std for every inner iterations as default

@GitPaean, what do you think? Should we merge this before the release? I think so, as it improves the convergence in most cases. If it doesn't, it points to other...

Also solve std for every inner iterations as default

jenkins build this failure_report please

Also solve std for every inner iterations as default

> Did you check the results related to UDQ_WCONPROD? It seems super-sensitive to small perturbations in the time-stepping. I will dig deeper to see if I can solve the underlying...

Also solve std for every inner iterations as default

I think the issue is related to control switching. https://github.com/OPM/opm-simulators/pull/6145 seems to fix it. However, so does restricting the timestep and reducing the maximum number of well switching per step....

Also solve std for every inner iterations as default

@vkip comment strengthens my suspicion that it is related to well-switching.

Also solve std for every inner iterations as default

Let's merge #5924 first.

Also solve std for every inner iterations as default

There are still some issues for Norne that need to be solved before this can be merged. I will keep it open, but remove the release tag.

Also solve std for every inner iterations as default

jenkins build this failure_report please

Also solve std for every inner iterations as default

jenkins build this failure_report please