A logit-based stochastic user equilibrium model for the integrated scenario-based hurricane evacuation framework
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
This dissertation introduces a new traffic assignment model for use in hurricane evacuation modeling. The logit-based stochastic user equilibrium (SUE) model is designed to capture the stochastic route choice behaviors during hurricane evacuation. The model starts from Dial’s algorithm with a logit model developed for hurricane evacuation and provides the probability distribution of route choice to simulate evacuation. Evacuees are loaded onto the road network using Monte Carlo simulation, with the route choice based on the precomputed probability distribution. Although the route choice logit model is static, the traffic on the network, i.e., the number of vehicles (flow) on each link, is updated every 15 minutes, so that the link travel time is adjusted according to the dynamic flows. This model is incorporated into the Integrated Scenario-based Evacuation framework (ISE) framework and implemented in a full-scale case study. ☐ The recently developed ISE framework aims to support the decision process that leads to official evacuation orders. It provides a strategic and robust evacuation order recommendation that integrates all projected scenarios of hurricane tracks and adapts the recommended order issuance as time elapses and the hurricane evolves. The traffic assignment model evaluates the performance of an evacuation plan by simulating the whole evacuation process and computing travel time and risk values. This performance evaluation helps the heuristic algorithm in the ISE framework to generate a better evacuation plan and helps to assess and compare different sets of evacuation plans. ☐ The purpose of the ISE framework is to recommend when and where to issue official hurricane evacuation orders in real-time, as a hurricane is approaching land. In this dissertation, a case study application is conducted to test the impacts of the logit- based SUE model on the ISE framework, relative to the traffic assignment model based on dynamic user equilibrium that was in the original version of the ISE framework. The eastern half of the state of North Carolina is used as the study area, and the evacuation simulation is applied for Hurricane Matthew (2016). The results of the ISE framework with the SUE model are compared to those with the DUE model. Results indicate that the SUE-based model tends to evacuate more zones than the DUE-based model does, and the SUE-based model also evacuates some zones earlier than the DUE-based model does. ☐ The major contribution of this dissertation is first to present the newly developed logit-based stochastic user equilibrium model for traffic assignment in hurricane evacuation situations. Second, the model is integrated into the ISE framework to test its impact on evacuation planning, with the implementation of a case study. Finally, the results of the ISE framework with the SUE model are compared to the results of the original ISE with the DUE traffic assignment model.
Description
Keywords
Dynamic traffic assignment, Hurricane evacuation, Integrated Scenario-based Evacuation Framework, Stochastic User Equilibrium model