lgr: unrefined twin deck for a field-totals comparison - #1558
Conversation
…efinement Refining part of the grid must not change the field totals. The new deck is SPE1CASE1_CARFIN1-3DCORNERPOINT_XYZ with the CARFIN block removed and nothing else changed; both gain the in-place summary vectors so the comparison has something to check. No reference data is involved: the two runs are compared against each other. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
bska
left a comment
bska
left a comment
There was a problem hiding this comment.
Very good. Please update the top-level description of the _NOLGR model to reflect the actual purpose of the simulation model.
| -- Test case for LGR grid, CARFIN and ENDFIN keyword are used. | ||
| --3D CORNER POINT CASE WITH NON-ACTIVE CELL, LGR on grids 6,6,1:3 and non-active cells are 6,5,1:3 |
There was a problem hiding this comment.
Tiny nit: This description is not accurate since there are no local grids in this model. Please update the description to reflect what's actually included here.
There was a problem hiding this comment.
Fixed - the header now says it is the unrefined reference for the CARFIN deck and why it exists, rather than the copied LGR text.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
|
jenkins build this please |
bska
left a comment
bska
left a comment
There was a problem hiding this comment.
Fixed - the header now says it is the unrefined reference for the CARFIN deck and why it exists, rather than the copied LGR text.
Greatly appreciated–thanks a lot for the updates. I'll merge this into the master branch now.
2 participants
Adds
lgr/SPE1CASE1_CARFIN1-3DCORNERPOINT_XYZ_NOLGR.DATA: the existing..._XYZdeck with theCARFINblock removed and nothing else changed. Both decks gainFOIP/FGIP/FWIP/FRPV/FHPV(the refined one already asked forFPR).Refining part of the grid must not change the field totals, so the pair can be compared against each other — no reference data, and nothing existing is perturbed (the
..._XYZdeck is not currently used by any test).The companion test is OPM/opm-simulators#7315. Measured with both decks on the fixes in #7245 + #7244:
On current master every one of these is 0 for the refined deck — all cell-based output is dropped when an LGR is present.