Schedule creep - In search of an uncongested baseline block time by examining scheduled flight block times worldwide 1986-2016

摘要

Based on a stratified random sampling of airlines' schedules for 200 heavily travelled directional nonstop airport pairs, this paper examines systematically how scheduled block times in non-stop flights have changed from 1986 to 2016. Three econometric analyses, by way of a 10th percentile quantile regression, 15th percentile quantile regression and ordinary least-squares regression, show that after accounting for the effects of air traffic growth, airport-specific congestion, flight delays, number of seat per flight, aircraft type, flight heading, airport slot policy, other airport-specific anomalies, airline-specific policies and changes in crude oil price, scheduled block times have been growing at a pace between 0.21 and 0.33 min per year depending on the regression model, or a total of between 6.2 and 9.8 min per flight from 1986 to 2016. Over-flying crowded parts of Europe contributes to an increase of 4.1 min of block time in 2016 compared with 1986, while over-flying crowded parts of China contributes to a corresponding increase of 8.9 min of block time. Slot-based practices at one end of an airport pair reduce the scheduled block times between 1.3 and 2.0 min, and this reduction can vary slightly over time in depending on the regression model. Regional influences add to the changes in scheduled block times. Those airport pairs within north-eastern U.S. have scheduled block times between 3.6 and 4.0 min longer than their counterparts in the rest of North America, which in turn grow at 0.10 min per year in the 10th percentile regression. Airline-specific policies also add to further changes to the scheduled block times, with some starting with longer times in 1986 and reducing theirs over the years while others starting with shorter times in 1986 and increasing theirs over the years. Hub-specific adjustments to scheduled block times by individual airlines are also observed. Airlines with increasing frequency share at the departure and arrival airports, or in the non-stop airport pair itself, are shown to reduce the scheduled block times of that route. The overall increase from the projected baseline block times in 1986 to the actual scheduled block times in 2016 is 19.2 min per flight from the sampled nonstop city-pairs, consistent with a previous study on buffers in flight schedules within the U.S. Overall, un-adjusted scheduled block times are not a reliable benchmark for determining true flight delays, but using a percentile statistics from past flight records to determine a minimum feasible block time is a reasonable estimate even if aircraft types are not explicitly accounted for.