Date of Award


Document Type


Degree Name

Master of Science in Operations Research


Department of Operational Sciences

First Advisor

Dennis C. Dietz, PhD


This study analyzes departure delays at a major airport using probability models to represent this non-homogeneous process. The models developed in this study expand on the Markovian models presently used by employing the method of stages to represent some of the model processes. This technique improves the user's ability to achieve a close fit for the service time probability distribution while maintaining the advantages of the Markovian model. The three types of models developed and compared all assume a Markovian system entry process. The first model uses an exponential distribution to model the service process. The second uses an Erlang distribution. The third models employs a unique server absence process to explicitly represents the periods of time when the server is unavailable to service departing aircraft. All three models generate results which correlate well with the delays actually observed. However, the Erlang model is preferred. Its results have lower variability than the exponential service time model. In addition, it generates solutions much faster than a typical application of the absence model. This model should be useful for improving capacity estimation and take-off delay prediction.

AFIT Designator


DTIC Accession Number



The author's Vita page is omitted.