An integrated methodology for optimal egress route assignment during population evacuation under an evolving emergency event
The primary focus of this research is to develop an integrated methodology for Adaptable Evacuation Planning (AEP). In case of regional evacuation caused by a hazardous event, one of main objectives of AEP is the optimal design and analysis of evacuation routes in transportation networks that will minimize total clearance time, traveled distance, and potential congestion on roads to ensure overall safety of the evacuated population. The problem under analysis is complex and challenging due to its multi-objective nature, potential congestion, blocking and queueing along routes. In addition, the hazardous event, which caused the evacuation may evolve and affect the population on egress routes, deplete the capacity of the road network and therefore make the initial assumptions of the evacuation policy invalid. To cope with the complexity of the problem, we consider it as an interaction of events in two overlapping and orthogonal networks. The first network represents a surface wild-fire propagation through a complex landscape. The second is a regional evacuation network for which route assignment optimization models are suggested. The first model utilizes a Delaunay triangulation to represent surface fire spread as movement of the fire event within the network. A data dependent procedure to construct the triangulation and estimate the rate of spread along the edges of the network is discussed. After the Delaunay triangulation is constructed, a two pass shortest path algorithm is incorporated to estimate the minimum travel time paths and fire event arrival times. In the next part of the dissertation, an integer programming (IP) formulation and model for optimal route assignment is presented, which utilizes state dependent queueing models to cope with congestion and time delays on road links. State dependent simulation software is used to evaluate performance measures of the evacuation plan: clearance time, total distance travelled and blocking probabilities. The resulting methodology allows a decision maker to adapt routing policies effectively, in case of change in hazardous event behavior, road infrastructure failure, or traffic incidents. The third model integrates the evacuation model and the fire event model and allows one to reroute the population dynamically. Finally, in the third model demonstration we illustrate proposed methodology with a case study, where regional evacuation for the Western Massachusetts is modeled.