System optimal dynamic traffic assignment in congested networks with advanced information systems

Growing concern over urban and suburban congestion on traffic networks has led to technologies broadly labeled as Intelligent Vehicle Highway Systems (IVHS), that integrate advances in telecommunications, information systems, automation and electronics to alleviate recurrent and non-recurrent congestion by augmenting the efficiency of the existing network. One class of IVHS technologies, Advanced Traveler Information Systems (ATIS), provide traffic control centers the ability to communicate with suitably equipped road users on a real-time basis and supply routing information and/or route guidance instructions.;Existing methodological and algorithmic constructs for real-time control in large-scale traffic systems do not adequately exploit the available sophistication in technological and hardware capabilities, as highlighted by the paucity of suitable dynamic route assignment algorithms and associated real-time network traffic simulation capabilities. This research addresses dynamic traffic assignment problems in the ATIS context, where a central controller with partial or complete information on time-dependent origin-destination (O-D) trip desires aims at achieving certain systemwide objectives by providing real-time routing information and/or route guidance instructions to suitably equipped vehicles. Conceptual and mathematical formulations are developed for the single user class (SUC) system optimal (SO) dynamic traffic assignment problem based on the information available to the controller.;A simulation-based solution algorithm is presented for the SUC SO dynamic traffic assignment problem in which the controller has complete a priori information on O-D desires for the planning horizon. It utilizes a traffic simulation model that evaluates the SO objective function and circumvents principal difficulties that have precluded analytical approaches from solving realistic formulations of the problem by obviating the need for link performance and link exit functions, implicitly ensuring the first-in, first-out property and precluding holding of traffic. The associated user equilibrium (UE) solution is detailed. Experiments are conducted to obtain SO and UE solutions under alternative network loading intensities. They provide key insights into the nature and extent of benefits obtainable under advanced information systems and characterize the quality of service from a network traffic flow theory perspective.;A multiple user classes (MUC) time-dependent traffic assignment problem that considers different user classes in terms of information availability, information supply strategy, and driver response behavior is formulated. A solution algorithm obtained by extending the SUC algorithm is implemented. The MUC solution algorithm is then embedded within a rolling horizon framework for real-world on-line implementation.