The threat of terrorist incidents is higher than ever before and devastating acts, such as the terrorist attacks on the World Trade Center and the Pentagon, have left many concerns about the possibility of future incidents and their potential impact. Unlike some natural disasters that can be anticipated, terrorist attacks are sudden and unexpected. Even if sometimes we do have partial information about a possible attack, it is generally not known exactly where, when, or how an attack will occur. This lack of information posses great challenges on those responsible for security, specifically, on their ability to respond fast, whenever necessary with flexibility and coordination. The surface transportation system plays a critical role in responding to terrorist attacks or other unpredictable human-caused disasters. In particular, existing Intelligent Transportation Systems (ITS) can be enhanced to improve the ability of the surface transportation system to efficiently respond to emergencies and recover from disasters. This research proposes the development of new information technologies to enhance today's ITS with capabilities to improve the crisis response capabilities of the surface transportation system. The objective of this research is to develop a Smart Traffic Evacuation Management System (STEMS) that responds rapidly and effectively to terrorist threats or other unpredictable disasters, by creating dynamic evacuation plans adaptable to continuously changing traffic conditions based on real-time information. The intellectual merit of this research is that the proposed STEMS will possess capabilities to support both the unexpected and unpredictable aspects of a terrorist attack and the dynamic aspect of the traffic network environment. Studies of related work indicate that STEMS is the first system that automatically generates evacuation plans, given the location and scope of an incident and the current traffic network conditions, and dynamically adjusts the plans based on real-time information received from sensors and other surveillance technologies. Refining the plans to keep them consistent with the current conditions significantly improves evacuation effectiveness. The changes that STEMS can handle range from slow, steady variations in traffic conditions, to more sudden variations caused by secondary accidents or other stochastic factors (e.g., high visibility events that determine a sudden increase in the density of the traffic). Being especially designed to handle evacuation in case of terrorist-caused disasters, STEMS can also handle multiple coordinated attacks targeting some strategic area over a short time frame. These are frequently encountered in terrorist acts as they are intended to create panic and terror. Due to the nature of the proposed work, an important component of this project is the development of a simulation environment to support the design and test of STEMS. Developing analytical patterns for modeling traffic dynamics has been explored in the literature at different levels of resolution and realism. Most of the proposed approaches are either too limited in representing reality, or too complex for handling large networks. The contribution of this work consists of investigating and developing traffic models and evacuation algorithms that overcome both of the above limitations. Two of the greatest impacts of this research in terms of science are as follows. First, the new simulation environment developed for this project provides a test bed to facilitate future work on traffic evacuation systems. Secondly, although the models and algorithms developed for STEMS are targeted towards traffic environments and evacuation, their applicability can be extended to other environments (e.g., building evacuation) and other traffic related problems (e.g., real-time route diversion in case of accidents). One of the broader impacts of this research would be the deployment of STEMS in a real environment. This research provides a fundamental tool for handling emergency evacuation for a full range of unpredictable incidents, regardless of cause, origin and scope. Wider and swifter deployment of STEMS will support Homeland Security in general, and will also enhance the surface transportation system on which so many Homeland Security stakeholders depend.

Evacuation planning is an important disaster management tool. A large-scale evacuation of a region by automobile is a difficult task, especially as demand is often greater than supply. This is made more difficult as the imbalance of supply and demand actually reduces supply due to congestion. Currently, most of the emphasis in evacuation planning is on supply management. The purpose of this dissertation is to introduce and study sophisticated demand management tools, specifically, staging and routing of evacuees. These tools can be used to produce evacuation strategies that reduce or eliminate congestion. A strategic planning model is introduced that accounts for evacuation dynamics and the non-linearities in travel times associated with congestion, yet is tractable and can be applied to large-scale networks. Objective functions of potential interest in evacuation planning are introduced and studied in the context of this model. Insights into the use of staging and routing in evacuation management are delineated and solution techniques are developed. Two different strategic approaches are studied in the context of this model. The first strategic approach is to control the evacuation at a disaggregate level, where customized staging and routing plans are produced for each individual or family unit. The second strategic approach is to control the evacuation at a more aggregate level, where evacuation plans are developed for a larger group of evacuees, based on pre-defined geographic areas. In both approaches, shelter requirements and preferences can also be considered. Computational experience using these two strategic approaches, and their respective solution techniques, is provided using a real network pertaining to Virginia Beach, Virginia, in order to demonstrate the efficacy of the proposed methodologies. / Ph. D.