Machine Learning Models of C-17 Specific Range Using Flight Recorder Data
Abstract
Fuel is a significant expense for the Air Force. The C-17 Globemaster fleet accounts for a significant portion. Estimating the range of an aircraft based on its fuel consumption is nearly as old as flight itself. Consideration of operational energy and the related consideration of fuel efficiency is increasing. Meanwhile machine learning and data-mining techniques are on the rise. The old question, "How far can my aircraft fly with a given load cargo and fuel?" has given way to "How little fuel can I load into an aircraft and safely arrive at the destination?" Specific range is a measure of efficiency that is fundamental in answering both questions, old and new. Predicting efficiency and consumption is key to decreasing unnecessary aircraft weight. Less weight means more efficient flight and less fuel consumption. Machine learning techniques were applied to flight recorder data to make fuel consumption predictions. Accurate predictions afford smaller fuel reserves, less weight, more efficient flight, and less fuel consumed overall. The accuracy of these techniques were compared and illustrated. A plan to incorporate these and other modeling techniques is proposed to realize immediate fuel cost savings and increase savings over time.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 2019
- Accession Number
- AD1074737
Entities
People
- Marcus A. Catchpole
Organizations
- Air Force Institute of Technology