The aerolog-scaling programme sizes a battery-swap hub for a cargo VTOL. Earlier milestones assumed Poisson arrivals, then fixed bursts. This one replays a real daily shape: trips per 15 minutes from City of Chicago open ride-hail data, scaled to the hub's rate. Ride-hail is a stand-in for time-of-day shape, not drone orders.
Spares needed for a p95 wait under 5 minutes:
| Hub load | Poisson | Wed profile | Sat profile |
|---|---|---|---|
| 70% | 12 | 16 | 16 |
| 85% | 16 | 32 | 40 |
I predicted at least 2x at 70% and was wrong: it is 1.33x. Near the charger limit it is 2x or more. Spreading arrivals inside a bin made no difference. Limits: two days, a proxy for demand, a coarse spares grid, assumed hub inputs.
Data, code and log: research/aerolog-scaling/ (milestone 6f).