This is an earlier experiment. Read the amendments below before treating its findings as current.
The question
Slow-responding craft descend better than they climb, so their lanes sink. A dodge sized to what the airframe can do in each direction had been built but never judged.
What was tested
Six layouts on the busy mixed sky at several traffic levels and two cruise speeds, with two sizings: shrink the weak upward share, or the strong downward share. We also measured lane heights.
What was found then
Shrinking the upward share more than doubled collisions where express craft cut into lanes, and sank the lanes deeper. Shrinking the downward share cut the busiest case by a third (3.9 to 2.6 per 1000 km). But lanes rose at the same time: the northbound express lane ended 80–110 m above its set height, and double traffic got worse. The setting also changes sideways dodging, so the cause is not pinned down. Neither setting keeps a lane.
What changed
Nothing.
Numbers and source references
- 6
- seeds per configurationresearch/reviews/brief-c18.md
- 1× / 2× / 4×; 16 / 30 m/s
- traffic multiples and cruise speeds testedresearch/reviews/brief-c18.md
- 10 → 23; 9 → 41
- express-into-lane collisions at 2× traffic and at cruise 30, upward scalingresearch/reviews/brief-c18.md
- 3.9 → 2.6
- busiest condition, collisions per 1000 km, downward scaling (3.93 → 2.57)research/reviews/brief-c18.md
- 80–110 m
- median express lane height above its altitude, downward scaling (+81 whole run, +109 late)research/reviews/brief-c18.md
- 0.69 → 1.37
- 2× traffic collisions per 1000 km, downward scalingresearch/reviews/brief-c18.md