Exploring Throughput and Travel Time Reliability in Freeway Networks

Abstract

This dissertation examines congestion on urban freeway networks, highlighting the importance of understanding and managing traffic throughput and travel time reliability. Recognizing the growing complexity of urban mobility and the need for efficient freeway management, this research contributes significant theoretical and empirical advancements to the field.

Based on queueing theory, theoretical models for traffic throughput and travel time reliability are constructed. Employing deterministic queueing models, we introduce a spatial queueing model for the morning peak period characterized by distinct bottlenecks. The freeway stretch is segmented into three parts: freeflow, transition, and queued, ensuring spatial continuity of flow and density. We determine the expansion of the queued segment and the implications of vehicle spillback under varied bottleneck intensities. Furthermore, within the proposed model framework, key parameters in the single-day morning peak arrival flow function are extracted and sampled through the Monte Carlo method to capture daily fluctuations in traffic flow.

Beyond our theoretical insights rooted in queueing theory, empirically, the dissertation explores freeway throughput and travel time reliability using loop detectors and vehicle trajectory data across different geographical locations and time scales. It reveals that lane changes often do not yield significant speed benefits and identifies a bidirectional causal relationship between lane changing and congestion using the Granger causality test.

In summary, this dissertation advances the understanding of urban freeway dynamics through a blend of theoretical modeling and empirical analysis. The findings provide a foundation for developing more resilient and efficient freeway management strategies, with potential future research exploring the integration of emerging technologies like autonomous vehicles to enhance urban mobility.