PhD Thesis Defense: Keji Wei

Tuesday, July 16, 2019, 2:30–4:30pm

Rm. 118, Cummings Hall

“Schedule Planning and Endogeneity of Travelers' Decisions in Congested Large-Scale Transportation Networks”

Abstract

The core task in passenger transportation systems planning can be described as one of matching the network-wide demand with adequate capacity by deploying resources over the corresponding parts of the network. A prime challenge inherent in this task is the uncertainty in demand and supply in transportation systems. Demand uncertainty manifests itself in the form of the passenger choice behavior. Supply uncertainty presents itself as delays and disruptions in resource availability. This thesis tackles a set of three related research questions within passenger transportation systems while explicitly incorporating the inherent uncertainty in both demand and supply during the planning phases of air and urban transportation systems.

Flight delay propagation results in enormous additional operating costs for the airlines, passengers and the aviation system as a whole. The first part of this thesis is focused on proposing, optimizing and validating a methodological framework for estimating the extent of crew-propagated delays and disruptions. We identify the factors that influence the extent of crew-propagated delays, and incorporate them into a robust crew scheduling model. We then develop a fast heuristic approach for solving the inverse of this robust crew scheduling problem to generate crew schedules that are similar to real-world crew scheduling samples. Along with various other findings, our results show that airlines avoid up to 80% of crew-related delays through advanced planning methods.

The second part of this thesis introduces an original integrated optimization approach to comprehensive flight timetabling and fleet assignment under endogenous passenger choice. The resulting optimization model is formulated as a mixed-integer linear program. We propose an original multi-phase solution approach, which effectively combines several heuristics, to optimize the network-wide timetable of a major airline within a realistic computational budget. Using case study data from Alaska Airlines, computational results suggest that the combination of this model formulation and solution approaches can result in significant profit improvements, as compared to the most advanced incremental approaches to flight timetabling. Additional computational experiments based on several extensions also demonstrate the benefits of this modeling and computational framework to support various types of strategic airline decision-making in the context of frequency planning, revenue management, and post-merger integration.

The popularity of ridehailing services like Didi, Lyft, and Uber has soared recently. While ridehailing services offer a very convenient mode of urban transportation for many passengers, these additional vehicles on the road contribute to urban road traffic congestion, and in some cases are held responsible for falling public transit ridership. A high quality public transit system is an effective means to alleviate urban road traffic congestion, but the interdependence between transit ridership, ridehailing ridership and urban road traffic congestion motivates the following question: can public transit and ridehailing co-exist and thrive to improve the overall social welfare? To answer this question, we develop a novel mixed-integer nonlinear optimization model, and a new set of solution algorithms to optimize transit timetables. Our model explicitly accounts for the impacts on road congestion and the passengers' mode-switching behavior between transit and ridehailing services.

Thesis Committee

For more information, contact Daryl Laware at daryl.a.laware@dartmouth.edu.