Enhanced Air Transportation Modeling Techniques for Capacity Problems

Effective and efficient air transportation systems are crucial to a nation's economy and connectedness. These systems involve capital-intensive facilities and equipment and move millions of people and tonnes of freight every day. As air traffic has continued to increase, the systems necessary to ensure safe and efficient operation will continue to grow more and more complex. Hence, it is imperative that air transport analysts are equipped with the best tools to properly predict and respond to expected air transportation operations. This dissertation aims to improve on those tools currently available to air transportation analysts, while offering new ones.

Specifically, this thesis will offer the following: 1) A model for predicting arrival runway occupancy times (AROT); 2) a model for predicting departure runway occupancy times (DROT); and 3) a flight planning model. This thesis will also offer an exploration of the uses of unmanned aerial vehicles for providing wireless communications services.

For the predictive models of AROT and DROT, we fit hierarchical Bayesian regression models to the data, grouped by aircraft type using airport physical and aircraft operational parameters as the regressors. Recognizing that many existing air transportation models require distributions of AROT and DROT, Bayesian methods are preferred since their output are distributions that can be directly inputted into air transportation modeling programs. Additionally, we exhibit how analysts will be able to decouple AROT and DROT predictions from the traditional 4 or 5 groupings of aircraft currently in use.

Lastly, for the flight planning model, we present a 2-D model using presently available wind data that provides wind-optimal flight routings. We improve over current models by allowing free-flight unconnected to pre-existing airways and by offering finer resolutions over the current 2.5 degree norm.