Hasty Briefsbeta

Bilingual

Saving Jet Fuel

7 hours ago
  • A nonstop flight from Newark to Rome could cost $68K in fuel, and wind-optimized paths can save thousands.
  • Scikit-decide, combined with OpenAP (aircraft performance model) and NOAA wind data, offers an open-source flight path optimization solution.
  • Scikit-decide is a framework for reinforcement learning and planning, developed over six years, applicable to flight paths, workforce schedules, and drone swarms.
  • OpenAP provides detailed aircraft models (e.g., Airbus A320, Boeing 737) with engine specifications, fuel models, and drag coefficients.
  • A high-performance workstation with AMD Ryzen 9 9950X, 96 GB RAM, and SSD was used for the computations.
  • Setup involved installing Python 3.12, scikit-decide, OpenAP, OpenTop, DuckDB with extensions, and QGIS for map rendering.
  • OpenAP includes data on 427 engines (primarily turbofans) and 37 aircraft types, with fuel model coefficients per aircraft-engine combination.
  • Example flights: Toulouse to Berlin (optimized path took ~2.1 hours, burned 5,855 kg of fuel) and Toulouse to Warsaw (~1.7 hours, 5,600 kg fuel), with Warsaw flight using a higher climb altitude for a more direct route.
  • Comparison of Airbus A320 vs Boeing 737 on Amsterdam-Frankfurt route: A320 used 1,625.6 kg fuel in 35.8 min; B737 used 1,622.1 kg in 39.2 min, with both optimized for fuel and contrail grid cost.