Erik Hide Sæternes comments

Results 16 comments of


                                            Erik Hide Sæternes

Time step rejection if tolerance test fails

> please rebase on master. Done!

Fix for relative change (used in time step controllers)

Concerning this PR: It might make more sense to take the square root before returning the relative change. Also, would it be better to weight the contributions from each cell...

Fix for relative change (used in time step controllers)

> I think you should make the changes you think are correct. Very well! I'll make the necessary changes soon. When taking the square root, the `time-step-control-tolerance` parameter will most...

Fix for relative change (used in time step controllers)

I've now added square root and weighting by volume. Further: When using "pressure+saturation", only pressure is considered for the first time step, due to issues with very large change in...

Fix for relative change (used in time step controllers)

> Is it the case for your local running? Yes, it turns out that they crash for `time-step-control-tolerance=0.12`, but not for `0.11`, which I think is a bit strange; I...

Also solve std for every inner iterations as default

> I have started testing the Norne prediction cases with STW, and the results are not conclusive. In some cases, Newton's convergence gets worse. I will have to dig deeper...