Does the WAR rig-day basis agree with the owner?

The repository contains 19 reference figures bearing on rig days, 13 of them attributable to the owner. Sorted by what they can actually prove, they collapse to a single wellbore.

Grouped by evidence tier. Every row cites the committed file it came from; none of these numbers were reconstructed, inferred or rounded to improve agreement. Where we cannot compute the same quantity, the row says so rather than substituting a proxy.

The by-code figures held in rig_days_summary.md are DRL 151, PND 49, COM 87, TA 21. We reproduce DRL 151, PND 49, COM 91, TA 17 — two codes exact, the total exact at 308, and exactly four days moved from TA to COM. It is a boundary-placement difference, not a counting difference.

The well is temporarily abandoned in February 2015 and the rig does not return until 2015-10-15. That return is reported as a 3-day partial WAR week (2015-10-15–2015-10-17) — the only place in the well's entire record where an activity-code change does not fall on a Sunday week boundary. We take the code change at the WAR record boundary, so COM begins 2015-10-18. The 87/21 split is reproduced exactly, and only, if that boundary is placed four days later at 2015-10-22: TA then runs a nominal full week (14 + 7 = 21) and COM runs 2015-10-22–2016-01-16 = 87 days.

The owner's drilling rule is calendar: “I have been taking the spud date – the td date as drilling days.” Spud 07/24/2014, TD 12/26/2014.

The 155/156 pair is one figure, not two. 156 = 155 + 1 exactly; the two documents state the same milestone under the exclusive and inclusive conventions. That halves the apparent anchor count and is why the ledger files 156 as repo-legacy.

The 4-day gap to WAR DRL is the reporting quantum, not an error. The last DRL week ends 2014-12-20; TD is reached 12/26/2014, which falls inside the following week — and that week is returned as PND, not DRL. The calendar rule charges the whole TD week to drilling; the WAR rule charges it to whatever code the operator filed. On a continuously drilled well the two differ by at most one WAR week. On a suspended or batch-drilled well they diverge without bound: the calendar rule keeps counting through months of rig-off-well time, the WAR rule does not. That is the entire reason #1063 exists, and it is exactly why agreement on this one continuously-drilled well cannot be read as agreement on the population.

All 22 bores on the owner's Stones list carry WAR activity and are computable on the new basis. Exactly one of them carries an owner figure to check against. The rest are shown so the change is visible — they are not validated.

Against the owner's own V30 field totals (1,457 drilling, 1,145 completion) the WAR basis returns 1,105 and 574. Both gaps are large, both are expected by construction, and neither is evidence either way: his totals are his calendar rule summed over the same bores, so the difference measures the definition change, not its correctness. Note also that the published extract totals 1,168 completion days against the 1,145 in his shipped workbook — a +23-day divergence that predates this basis change and is tracked as #846.

Three of the eighteen Stones wells carry a sidetrack. Rolling those up by summing per-bore days double-counts the single WAR week that reports both bores across the sidetrack transition; the module unions instead.

Seven days per sidetrack boundary, every time. On these three wells the double-counted week happens to be filed under a non-D&C code, so the D&C columns are unaffected and the two roll-ups agree — which is precisely why the defect is easy to miss. Where the straddling week is filed DRL or COM, summing over-reports D&C by a full week per sidetrack. Both grains are published so a consumer can never pick the wrong one by accident.

Ranked by how much each would move the evidence base. The first three are cheap for him and would take the independent anchor count from one bore to a population.