Research on Fuel Economy of Takeoff and Climb Phase Based on QAR Data Cluster Analysis

Abstract

The objective of this paper is to present an evaluation method for Fuel economy based on flight data through the cluster analysis. The flight datasets are extracted from quick access recorder (QAR) data available in the Flight Operations Quality Assurance (FOQA) program, including hundreds of parameters. Through K-means and First Integer Neighbor Clustering Hierarchy (FINCH) cluster analysis of QAR data, we observe that 51 kg fuel on average was saved when the pilot use a higher rate of climbing speed to climb to the cruise attitude during the takeoff and climb phase. We also find that FINCH cluster is fully parameter-free and easily scalable to large QAR data at a minimal computational expense.