Abstract— The analytical evacuation network model EVAQ enables forecasting traffic flow conditions on a road network for a wide range of emergency situations, such as hurricanes, bushfires and floods. The proposed model is innovative as it combines voluntary evacuation, with possible pre-trip and en- route decisions, and mandatory evacuation, with prescribed destinations, routes and departure times, within a single generic model framework. Furthermore, we introduce dynamic road infrastructure, in which characteristics such as speed limits, capacity and flow direction can be time-varying due to the hazard’s progress in space and time and prevailing traffic regulations and control measures. The evacuation model can function as a decision support system for regional authorities and emergency services supervising an evacuation or constructing an evacuation strategy for planning purposes. I. INTRODUCTION VER the last couple of decades, big storms, mudflows, hurricanes, bush fires, floods, and many other hazards have caused massive economic and social damage as well as loss of life. Many studies have concluded that both the impact and frequency of these natural disasters are increasing (e.g., see [1], [2]). The ability to appropriately respond therefore becomes more important. The one-sided approach - referred to as ‘avoidance’ in risk management - aiming to limit the probability of such disasters by reinforcing dams, raising dikes, or clearing fire control lines is not always the most efficient or effective way of dealing with the threat. Although the disaster cannot always be avoided, exposure to the hazard is often avoidable. In fact, [3] shows by explicit cost-benefit analyses that by accepting evacuation as an optional response to such hazards peoples’ lives can be preserved at much lower costs. However, relatively little is known about the process of evacuation impeding making founded decisions by using available insights, theory and models As is discussed in [4], this may result in officials being reluctant to order an evacuation due to the uncertain and unfamiliar conditions and the financial liability involved. Better decisions can be made and society will benefit if the Manuscript received June 15, 2008. A. J. Pel is with the Transport & Planning Department, Delft University of Technology, Delft, the Netherlands (corresponding author, phone: +31.15.2784981; fax: +31.15.2783179; e-mail: a.j.pel@tudelft.nl). M. C. J. Bliemer is with the Transport & Planning Department, Delft University of Technology, Delft, the Netherlands (e-mail: m.c.j.bliemer@tudelft.nl). S. P. Hoogendoorn is with the Transport & Planning Department, Delft University of Technology, Delft, the Netherlands (e-mail: s.p.hoogendoorn@tudelft.nl). process of evacuation is better understood. Since a significant part of this process consists of human transport, the fields of transport planning and traffic flow modeling and control are essential to this understanding. Due to the complexity of the emergency-evacuation problem, simulation-based traffic models are often applied to obtain a better understanding of the process at hand. The comprehensive overview of past evacuation models given in [5] concludes that the majority of the more recently developed evacuation models attempt to combine a traffic simulation model with geographic information systems to predict the spatial implications of an evacuation. Evacuation strategies can then be assessed and heuristically optimized. These mostly macroscopic models concentrate on traffic dynamics during the evacuation, such as speeds and traffic volumes. This enables identifying where bottlenecks are likely to occur and computing the expected evacuation time. Several of these models use dynamic traffic assignment (DTA) and simulate the changing network conditions. However, the weaknesses of these models are in the way in which travel behavior and road infrastructure is modeled. Travel choice behavior is inadequately modeled, since most models compute an equilibrium traffic assignment, thereby assuming that the evacuees are aware of (future) network conditions, which is very unlikely in an emergency situation. Furthermore, travel behavior is insufficiently modeled since the focus is on evaluating mandatory evacuations with full compliance. The second shortcoming of these models relates to road infrastructure being static, thus assuming sufficient time for complete evacuation. The situation that infrastructure may become inaccessible over time due to the hazard is often ignored. In this paper, we present a new evacuation network model, called EVAQ (Evacuation of Vehicles using Assignment with Queuing), which overcomes aforementioned shortcomings by in particular dealing with the dynamic interaction between hazard, authority, and evacuees. More specifically, the evacuees decide pre-trip on their departure time, destination and route based on the (perceived) conditions and the information and evacuation instructions (departure time window, evacuation destination(s) and route(s)) given by the authority, but may also make en-route decisions to change their route and destination (possibly due to traffic conditions or links that become inaccessible due to flooding). This allows modeling of varying levels of compliance, enabling assessment of both voluntary and mandatory evacuations, EVAQ: A New Analytical Model for Voluntary and Mandatory Evacuation Strategies on Time-varying Networks Adam J. Pel, Michiel C.J. Bliemer, and Serge P. Hoogendoorn O Proceedings of the 11th International IEEE Conference on Intelligent Transportation Systems Beijing, China, October 12-15, 2008 1-4244-2112-1/08/$20.00 ©2008 IEEE 528 Authorized licensed use limited to: Technische Universiteit Delft. Downloaded on February 4, 2009 at 05:57 from IEEE Xplore. Restrictions apply.