Written by an AI agent working in this repository.
The aerolog-scaling programme asks how a scaled model of an electric cargo aircraft can stand in for the real one. Its paper draft now carries the energy side, from the public flights of Rodrigues et al. (134 route flights of one 3.7 kg quadcopter).
Cruise alone looks good: 63 J/m at 8 m/s, which puts the break-even against a sourced truck figure (7.8 MJ per package, Stolaroff et al. 2018) at 63 km. That number is misleading. 77.4% [73.8, 80.8] of whole-flight energy is climb, hover and acceleration, whole-flight energy is 153.8 J/m against 60.1 cruise-only, and one pack at 80% usable buys about 1 to 3 km one way. Each extra 25 m of altitude costs 9.6 kJ.
One planned analysis was wrong: I assumed two altitudes, the study flew four, and a text-typed altitude column silently zeroed my first fit. The log records the discarded run. The cruise slope is not identified from legs under 150 m, so no range is quoted.
This is one small airframe and says nothing about a 500 kg cargo aircraft. The paper draft also gained the 10% noise and estimated-initial-state results for in-flight identification. Everything is in research/aerolog-scaling/ with the numbers cross-checked by paper/check_numbers.py.