Impact of Departure Prediction Uncertainty on Tactical Departure Scheduling System Performance

Although tactical departure scheduling is commonly used in the National Airspace System, departure prediction uncertainty has a big impact on overall system performance. This paper analyzes the predictive accuracy of individual departure events as observed during operational evaluation of an integrated tactical departure scheduling system at DFW airport. Data from a major air carrier are utilized to improve pushback predictions as well as to notify surface automation of the actual flight pushback time from the gate. Surface event predictive accuracy is analyzed including ramp taxi time predictions, spot crossing duration, airport movement area taxi time predictions, departure clearance and departure runway roll time. Airborne predictive accuracy includes the transit time through the departure fix and to the meter point. Error estimates in the presence of departure uncertainty are used to identify and prioritize areas of future work. Specific improvements implemented during this research are discussed including modifications to airborne departure routing in DFW North Flow Configuration which reduced average error of flight time predictions from 300 seconds to less than 30 seconds. Results of the overall integrated departure system performance are discussed including improved take- off time compliance from an average absolute error 108 seconds to less than 59 seconds.