LAUSR

One of the actions usually conducted to limit exposure to a hazardous event is the evacuation of the area that is subject to the effects of the event itself. This involves modifications both to demand (a large number of users all want to move together) and to supply (the transport network may experience changes in capacity, unusable roads, etc.). In order to forecast the traffic evolution in a network during an evacuation, a natural choice is to adopt an approach based on Dynamic Traffic Assignment (DTA) models. However, such models typically give a deterministic prediction of future conditions, whereas evacuations are subject to considerable uncertainty. The aim of the present paper is to describe an evacuation approach for decision support during emergencies that directly predicts the time-evolution of the probability of evacuating users from an area, formulated within a discrete-time stochastic process modelling framework. The approach is applied to a small artificial case as well as a real-life network, where we estimate users’ probabilities to reach a desired safe destination and analyze time dependent risk factors in an evacuation scenario.