On the Probabilistic Modeling of Runway Inter-departure Times

This paper examines the validity of the Erlang distribution for runway service times. It uses high-fidelity surface surveillance data, for the first time, to model the probability distributions of runway service times and departure throughput, and to validate the Erlang service time assumption. The paper proposes several potential approaches to determine departure runway service time distributions from empirical data, and compares the results. In particular, it finds that a displaced exponential fit may be a better match to the empirical service time distribution than an Erlang distribution. However, the computational benefits offered by the Erlang service time distribution, its accurate reflection of the means and variances of the empirical service time and throughput distributions, and its ability to represent the tail of the service time distribution, make it attractive for use in queuing models of airport operations.