v1.3.6: breakout timing

Release date: 19 October 2026. The scientific page and README wording were accepted before publication. The release uses the annotated tag v1.3.6.

Results

The first sampled delay with an open whole-hole drilling window is 20 minutes at Soultz, with a closed sample at 15 minutes. At United Downs it is 15 minutes, with a closed sample at 10 minutes. The fixed scan has no interpolation or refinement. Every complete programme in the scan remains NO-GO.

The breakout-timing page explains the assumption and reports the remaining failures. Its target requirements are SG-equivalent at depth; whole-hole bounds are surface SG at 20 °C. The retained scan links the calculation settings and provenance to the result files.

The calculation is unchanged from v1.3.5, commit 6c0613610ae000d5ec173ac69c9a5b6779bb5de4. The summary and its PDF remain unchanged. Earlier result files and goldens are retained.

Reproduction

From the repository root:

python -m pip install -r requirements-lock.txt
python tools/breakout_timing.py --output ../borethm-timing-reproduction
python tools/breakout_timing.py --scan --reuse --output ../borethm-timing-reproduction
python -m pytest
python -m pytest -m slow

The recorded calculations use Python 3.14.4 and the locked libraries. The release checklist records verification. The GitHub release contains the publication notes.

Validation

The fast suite passed all 357 tests in 14.42 seconds. The slow suite passed all 106 tests in 4,960.63 seconds (1 hour 22 minutes 40 seconds), including the eight breakout-timing regressions and four fresh paired calculations.

The slow suite emitted 12 runtime warnings from model and SciPy calculations in the existing reference tests. Those tests passed. Checksum checks confirm that protected original files and the recorded timing JSON files are unchanged.