Date of Award
Doctor of Philosophy (PhD)
Taaffe , Kevin
Chowhury , Mashrur
Melloy , Brian
Kurz , Mary
Evacuating a large population from an at-risk area has been the subject of extensive research over the past few decades. In order to measure trip completion and total evacuation times accurately, most researchers have implemented some combination of simulation and optimization methods to provide vehicular flow and congestion data. While the general at-risk population comprises the majority of travelers on the road network, there are often specific groups to consider when assessing the ability to evacuate an entire population. In particular, healthcare facilities (e.g., hospitals) may require evacuation, and the trip times may become an important health issue for patients being evacuated. Emergency vehicles from these facilities will share the same roadways and exit paths that are used by the local community, and it becomes increasingly important to minimize long travel times when patient care must be provided during transport.
As the size of the area to model grows larger, predicting individual vehicle performance becomes more difficult. Standard transportation-specific micro-simulation, which models vehicle interactions and driver behaviors in detail, may perform very well on road networks that are smaller in size. In this research, a novel modeling approach, based on cell transmission and a speed-flow relationship, is proposed that combines the 'micro' and 'meso' approaches of simulation modeling. The model is developed using a general purpose simulation software package. This allows for an analysis at each vehicle level in the travel network.
In addition, using these method and approaches, we can carry out dynamic trip planning where evacuees decide their route according to current road and traffic conditions. By translating this concept to an actual implementation, a traffic management center could identify current best travel routes between several origins and destinations, while continuing to update this list periodically. The model could suggest routings that favor either a user-optimal or system-optimal objective. This research also extended the concept of dynamic traffic assignment while modeling evacuation traffic. This extension includes the utilization of Wardrop's System Optimum theory, where flow throughout the network is controlled in order to lower the risk of traffic congestion. Within this framework traffic flow is optimized to provide a route assignment under dynamic traffic conditions.
This dissertation provides a practical and effective solution for a comprehensive evacuation analysis of a large, metropolitan area and the evacuation routes extending over 100 miles. Using the methodologies in this dissertation, we were able to create evacuation input data for general as well as special needs populations. These data were fed into the tailored simulation model to determine critical evacuation start times and evacuation windows for both the community-wide evacuation. Moreover, our analysis suggested that a hospital evacuation would need to precede a community-wide evacuation if the community-wide evacuation does not begin more than 24 hours before a hurricane landfall. To provide a more proactive approach, we further suggested a routing strategy, through a dynamic traffic assignment framework, for supporting an optimal flow of traffic during an evacuation. The dynamic traffic assignment approach also provides a mechanism for recommending specific time intervals when traffic should be diverted in order to reduce traffic congestion.
Duanmu, Jun, "SIMULATION AND MATHEMATICAL MODELING TO SUPPORT COMMUNITY-WIDE EVACUATION DECISIONS FOR MULTIPLE POPULATION GROUPS" (2010). All Dissertations. 601.