ULNERABILITY ASSESSMENT OF EMERGENCY TRANSPORTATION ROAD NETWORKS BASED ON EIGENVALUE ANALYSIS

Abstract

Japan is exposed to high risks of natural disasters, such as large earthquakes, floods, debris flows or tsunamis. In case of such disasters, emergency transportation road networks (ETRNs) are expected to play important roles for quick evacuations, smooth relief supplies and so on. In order to improve the connectivity of road networks and enhance their resilience, it is necessary to detect and identify critical links of the networks in advance and take effective countermeasures. As quantitative assessment of probabilities and scales of such disasters is far difficult, one of the feasible approaches to detect fragile sections of a road network is a vulnerability assessment based on the geometry of a network. Even though, direct evaluation of the connectivities between all the nodes demands computational cost and time, especially for large-scaled road network cases due to large numbers of OD pairs, a second least eigenvalue (Fiedler vector) of Laplacian matrix of networks is proposed as an alternative in this paper. In addition, other methods, i.e., eigenvector centrality of adjacency matrix or betweenness centrality analyses are also studied. Case studies of the ETRN in the Hokuriku-Tokai region are performed and compared to discuss the applicability to road network analysis and the interpretations of results in the context of transportation engineering.
The proposed method shows numerical efficiency and its results are acceptable. In this study, only network geometry and a lengths of a link are considered in this study. For a future study, other features of a link, such as its capacity or specific risks based on the regular inspections of structures along it, should be included and discussed to brush up the proposed method as a practical tool.