Disruptions in Megaregional Network Evacuations: Identifying and Assessing Critical Links

Abstract

Mass evacuations are a protective action to move large populations from hazardous areas to safety. However, even the best-planned evacuations can be slowed by unexpected disruptions, such as traffic incidents. Even minor disruptions can significantly slow evacuations, so it is critical to understand which links are most vital to the operation of the system. This paper describes a study to address that need by developing a method to evaluate large networks more efficiently to identify links that disproportionately increase network delay when affected by disruptive incidents. The study is unique because it examined the impact of individual link disruptions over a megaregional network covering thousands of square miles while drastically reducing the computation time necessary for a traditional full-scan analysis. In the research, link criticality was quantified by an index using factors such as alternative path availability, global maximum flow properties, modified betweenness centrality, and hazard exposure. Links with high indices established an initial “most-critical” list, then agent-based simulation was used to quantify the network-wide effects of disrupting these most-critical links. Results showed that links with the highest indices often had the fewest alternative paths to avoid them. Thus, while incident effects tended to be localized, findings suggest that networks with more path alternatives tend to have higher overall resilience to disruptions. By giving the ability to reduce computational efforts to evaluate large-scale networks, this methodology can be used in emergency planning to focus monitoring on the most important areas and allow them to be monitored for disruptions to maintain network efficiency.