This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
The default was adopted on wing and eVTOL test sets, but the map's own conditions had been flown only for the wing.
What was tested
120 runs: eVTOL and multirotor in calm, 5 m/s wind and at 1,600 craft, with and without the look-ahead. The 24 older calm comparison runs were re-flown as a check, and the wing flew in steady wind of 0–5 m/s.
What was found then
Every condition improved. In calm, eVTOL losses fell from 350 to 68 per 12 runs and multirotor losses from 242 to 78; across the six conditions, 60–81 % fewer losses. The re-flown runs matched 24 of 24. For the wing, the look-ahead removed 59–71 % of losses at every wind from 0 to 5 m/s. With and without it, losses dropped sharply between 2 and 3 m/s of wind, read then as the look-ahead scaling the whole wind curve down.
What we learned later
The dense-grid loss was 8.5–10 % of the 1,600 craft in 30 min, not 'a third' as first written. 'Scales the whole curve' was withdrawn as unproven; an exploratory re-check flagged 2 m/s.
research/envelope.md:3368; research/envelope.md:3440
The wind curve used one wind direction; in stage 5, wind from 70° did not reduce losses consistently.
research/envelope.md:8592-8600
What changed
Nothing further.
Corrections retained in the record
Earlier wording: That is 65 to 81 percent fewer losses on these six cells
Record correction: The cycle's own table has the multirotor dense cell at 4,938 → 1,956, a 60 % cut, so the range is 60–81 % (the envelope paragraph repeats the 65–81 % slip).
research/data/oos/default_map.md:24-30
Numbers and source references
- 120
- runs in the other-types testresearch/envelope.md:3368
- 350 → 68, 106 → 26, 5,234 → 1,628
- eVTOL craft lost per 12 runs: calm, wind 5, 1,600 craftresearch/envelope.md:3368
- 242 → 78, 30 → 6, 4,938 → 1,956
- multirotor craft lost per 12 runsresearch/data/oos/default_map.md:30
- 60–81 %
- range of loss reductions on the six conditions (computed from the counts)research/data/oos/default_map.md:20-30
- 24 of 24
- re-flown comparison runs identicalresearch/envelope.md:3405
- 72
- new runs: 24 re-flown checks and 48 wing wind runsresearch/envelope.md:3405
- 532 / 554 / 490 / 156 / 108 / 102
- wing losses per 12 runs at wind 0–5 m/s, old behaviourresearch/envelope.md:3405
- 156 / 178 / 198 / 58 / 44 / 32
- the same with the new defaultresearch/envelope.md:3405
- 59–71 %
- share removed per wind (new/old ratio 0.29–0.41)research/envelope.md:3405