Using the entropy of multi-aircraft departure time distributions to predict temporal deviations from on-time departure performance

In airport departure operations, some aircraft departure schedules are more challenging to achieve than others requiring identical numbers of aircraft to depart in a similar amount of time. In particular, some schedules are “stacked,” in that they require multiple departures within relatively short time intervals, in contrast to “uniform” schedules in which departure times are relatively evenly spaced throughout the departure window. Metrics capturing the degree to which departure distributions are uniform versus stacked (statistically, uniform versus nonuniform) may provide a quantitative measure useful for predicting on-time departure performance, or equivalently, for predicting departure time deviations. This paper describes how Shannon’s entropy metric was used for this purpose. The predictive validity of the measure was evaluated in a medium-fidelity ground control simulation in which human controllers in various experimental conditions were instructed to achieve a timely aircraft departure schedule among other task goals. Regression modeling showed that the entropy of a departure time schedule is a good predictor of the degree to which on-time departure performance was achieved. Correlations between the entropy of departure time distributions and deviations from on-time departure performance of 0.75 were found. The strengths and limitations of the entropy-based approach to providing a quantitative measure of predicted deviations from on-time departure performance as a function of departure time schedules are discussed.