Evacuation based on spatio-temporal resilience with variable traffic demand

Abstract

The efficient evacuation of people from dangerous areas is a key objective of emergency management. However, many emergencies give little to no advanced warning, leading to spontaneous evacuation with no time for planning or management. For large emergencies, destinations become less certain, with traffic demand imbalanced and concentrated on a few oversaturated routes familiar to evacuees. Ultimately, this leads to rapid congestion and delay on some routes, while others remain barely used, extending clearance times with an accumulating population at risk. In this study we address these issues through incorporating spatio-temporal traffic resilience dynamics into a destination choice model utilizing the available capacity of the overall network. We validate our model through a post-concert egress event. The results suggest that our method can reduce total egress times and average travel time by 20%–43% over the no-guidance condition. Our method can be used to estimate and quantify emergency conditions to optimally guide destinations and routing choice for evacuees and/or autonomously moving vehicles during evacuations.