Global Optimization Of Emergency Evacuation Assignments

informs
Vol. 36, No. 6, November–December 2006, pp. 502–513 issn 0092-2102 eissn 1526-551X 06 3606 0502

®

doi 10.1287/inte.1060.0251 © 2006 INFORMS

Global Optimization of Emergency Evacuation Assignments
Department of Civil and Environmental Engineering, University of Tennessee, 112 Perkins Hall, Knoxville, Tennessee 37996-2010, lhan@utk.edu PTV America, Inc., 1300 N. Market Street,Suite 603, Wilmington, Delaware 19801, fyuan@ptvamerica.com ORNL Center of Transportation Analysis, National Transportation Research Center, 2360 Cherahala Boulevard, Knoxville, Tennessee 37932 {chins@ornl.gov, hwanghl@ornl.gov}

Lee D. Han Fang Yuan

Shih-Miao Chin, Holing Hwang

Conventional emergency evacuation plans often assign evacuees to fixed routes or destinations based mainly ongeographic proximity. Such approaches can be inefficient if the roads are congested, blocked, or otherwise dangerous because of the emergency. By not constraining evacuees to prespecified destinations, a one-destination evacuation approach provides flexibility in the optimization process. We present a framework for the simultaneous optimization of evacuation-traffic distribution and assignment. Based onthe one-destination evacuation concept, we can obtain the optimal destination and route assignment by solving a one-destination traffic-assignment problem on a modified network representation. In a county-wide, large-scale evacuation case study, the onedestination model yields substantial improvement over the conventional approach, with the overall evacuation time reduced by more than 60 percent.More importantly, emergency planners can easily implement this framework by instructing evacuees to go to destinations that the one-destination optimization process selects. Key words: emergency planning; microscopic simulation; traffic assignment; network optimization; mass evacuation; special event operations; dynamic traffic assignment; intelligent transportation system. History: This paper wasrefereed.

H

eightened interest in emergency management and evacuation operations has impelled transportation professionals to focus on evacuation modeling and planning. Most evacuation-modeling research has used conventional transportation-planning models, such as the four-step process, as a basis. The design of these planning models assumes their use in day-to-day travel under normalsituations where origins and destinations of trips are easy to determine and, for practical purposes, remain somewhat static over time. Emergency evacuation, on the other hand, calls for expeditiously mobilizing and transporting a sizable population out of danger, overcoming temporal and spatial constraints in situations where they might encounter roads that are congested, blocked, or otherwisedangerous because of the emergency. Although the transient, and potentially chaotic, nature of evacuation makes it far more challenging to
502

manage, planning an evacuation has a unique advantage over planning normal traffic operations. Under normal conditions, a motorist takes a specific route to arrive at a specific destination; that motorist is more flexible under emergency-evacuation conditions.In other words, as long as an evacuee safely exits the evacuation zone quickly, the route taken is not important. Emergency planners do not clearly understand this flexibility in destination selection and its associated benefit to the planning process; therefore, they seldom exploit it. In fact, the common practice of rigidly assigning evacuees to designated routes, shelters, or destinations basedmainly on proximity often negates this flexibility. Experience suggests that a major problem in evacuation operations is that evacuation-zone exit routes are often limited in number and insufficient in capacity to handle the traffic surge during a large-scale

Han, Yuan, Chin, and Hwang: Global Optimization of Emergency Evacuation Assignments
Interfaces 36(6), pp. 502–513, © 2006 INFORMS

503...