Improving aircraft departure time predictability

Abstract

In this thesis, a forecasting model is described that improves departure predictions over those of Collaborative Decision Making (CDM), reducing error by up to 30% for a given day. This model propagates delay from incoming flights to outgoing flights by using minimum turn times calculated from Airline Service Quality Performance (ASQP) data. The model was run on data covering every day of March, April, and May of 1999, and produced departure predictions 6 hours, 4 hours, 2 hours, and 1 hour in advance of departure time. The greatest improvement on CDM predictions was achieved in the 4-hour predictions.