Network traffic management under disaster conditions.
详细信息
文摘
This dissertation focuses on the development and solution of supply and demand management models for network traffic management under disaster conditions, more specifically for evacuation planning. The concept of evacuation management and how it relates to disaster management concept is reviewed for different types of disasters; an evacuation traffic management methodology (ETAM) that outlines major steps and management options for evacuation planning is developed. For supply management, capacity reversibility (also known as "contraflow") is proposed on congested evacuation routes. Demand management actions include departure and destination choice optimization and evacuation zone scheduling. Both options are studied analytically for optimal traffic assignment principles, as well as heuristically for computationally less demanding approaches suitable for large urban networks.;Analytical supply management models include a static model using polynomial link performance functions, and a dynamic traffic assignment (DTA) model with a Linear Programming (LP) formulation and a modified Cell Transmission Model (CTM) for traffic flow propagation. Both models find system optimal (SO) capacity reversibility (CR) and route choice in a network. The dual of the dynamic model (SO-DTA-CR) is studied to find the optimal capacity reversibility assignment principle, which is later used in developing a tabu-based heuristic approach for urban networks. Variations of SO-DTA-CR model for different reversibility definitions, such as lane-based and total-or-no reversibility, are presented, as well.;A demand management model that optimizes departure time, route and destination choices is developed for a Fixed Evacuation Time (FET) SO-DTA assignment. Dual of the model is studied to find the optimal assignment principle. This model is further modified to find optimal scheduling of zone evacuation start times within a disaster region. As there is no currently available FET-SO-DTA software, two scenario-based demand management analyses are proposed for large scale networks.;To support heuristic models for urban network examples, a path-based SO-DTA model is formulated as a Variational Inequality, which is solved by an Inner Approximation approach using a time-dependent least marginal cost path (TDLMCP) algorithm. This model is coded into the VISTA (Visual Interactive System for Transport Algorithms) software and verified by analytical results.