Modeling in Air Transportation: Cargo Loading and Itinerary Choice

We examine two problems as part of this dissertation. The first is a cargo loading problem. The aim is to load a set of containers and pallets into a cargo aircraft that serves multiple airports. Our work is the first to model cargo transport as a series of trips consisting of several legs at the end of which pickup and delivery operations might occur. This problem is crucial for airlines because in an attempt to reduce their costs, most airlines prefer to load as many containers as possible, even if all the loaded containers do not have the same final destination. Our results demonstrate that it is possible to quickly find near optimal or excellent feasible loading plans, and that our approach leads to substantial savings with respect to typical manual approaches currently used in practice.

The second problem we examine involves the estimation of itinerary choice models that include price variables and correct for price endogeneity using a control function that uses several types of instrumental variables. The motivation for developing these models is to demonstrate the importance of accounting for price endogeneity and to estimate different price sensitivities as a function of advance purchase periods. This is important as the airline industry can use our results to incorporate different customer segments as revealed through high-yield and low-yield booking curves when evaluating the profitability of airline schedules.

Results based on Continental U.S. markets for May 2013 departures showed that models that fail to account for price endogeneity overestimate customers' value of time and result in biased price estimates and incorrect pricing recommendations. The advanced models estimated (nested logit and ordered generalized extreme value (OGEV) models) are shown to outperform the baseline multinomial logit model with regard to statistical tests and behavioral interpretations. Additionally, results show that price sensitivities vary as a function of advance purchase periods, with those purchasing high-yield products being less price sensitive than those purchasing low-yield products (across any advance purchase periods) and those purchasing closer to departure being less price sensitive. Results also indicate that inter-alternative competition is strong for itineraries that share similar departure times.

Finally, as part of the itinerary choice model developed in this dissertation, we estimate highly refined departure time of day preferences. Results are intuitive and show that departure time of day preferences vary across many dimensions including the length of haul, direction of travel, number of time zones crossed, departure day of week, and itinerary type (i.e., outbound, inbound, and one-way itineraries). To the best of our knowledge, these curves represent the most refined publicly-available estimates of airline passengers' time of day preferences.