Move the lanes, move the danger.

The question

The previous LINK cycle found that almost all craft lost to buildings while flying inside their height band were on one lane: westbound 45th Street at 50 m, under the crossing avenue’s lowest level and over its own street’s other level. Is that lane dangerous in itself, or is it its place in the stack? Five explanations were registered before any run, each predicting which lane would carry the losses when the stack changes: the order of its neighbours, its height, its street, its street’s upper level, or the westbound corridor itself.

The flight

The LINK campaign kept the crossing, the parcel-class craft, an 8 m declared height allowance, cross-level speed-matching off, no gate hold, calm air and five simulated minutes, and changed only which direction flies at which of the four heights (30, 50, 70 and 90 m). Four stackings were flown: the plan’s own (45th eastbound, 45th westbound, the avenue northbound, the avenue southbound from the bottom) and three permutations. Each ran on two fresh sets of 96 layouts on two assumed radio tiers: L3, the worst radio in this test (5 updates a second, 10 per cent lost, 0.25 s late), and L4, the better radio (10 a second, 1 per cent lost, 0.1 s late). 1,536 runs, no simulator change. Checks before any row showed that the plan’s stacking set the new way flies exactly as before and that the recorders change nothing on any stacking. A loss counts as in-band when the craft was within 3 m of its band centre at the last moment it was flying normally.

What happened

On the plan’s stacking, westbound 45th again took 95 to 100 per cent of the in-band wall losses on both radios and both layout sets, so the previous cycle’s result holds on new layouts. Re-stacked, the danger moved: to eastbound 45th (99 to 100 per cent, at 70 m) on one stacking and to the northbound avenue lane (77 to 92 per cent, at 70 m) on another. A different lane each time, on both streets, at both middle heights and with both neighbour orders, so none of the five explanations fits; by the rule registered in advance the test is unresolved, on both radios and both layout sets.

The fourth stacking (the avenue northbound at 30 m, 45th eastbound at 50, 45th westbound at 70, the avenue southbound at 90) lost almost nothing: 13 to 16 craft to buildings per 96 runs on the worst radio and 1 to 2 on the better one, against 113 to 396 on the other stackings, and no in-band losses on the worst radio. It is the only stacking where the two middle levels belong to the same street. Fewer craft also passed through it in the five minutes (about 15,800 against 17,600 to 19,500), and its avenue craft spent about nine times longer inside their bands; both are printed, not explained. No craft-to-craft collisions in any run. Two of the registered predictions held, one missed and one was void.

The next question

Nothing is adopted and no layout is recommended. The previous cycle’s “the risk sits on westbound 45th” is now read as a property of the plan’s stacking, not of that lane. The next question is why the quiet stacking is quiet; a further stacking can separate the candidate reasons.