turkmath.org

Designing and maintaining resilient and robust transportation systems require identification of links that are critical for the functionality of the network. Network scientists have suggested various topological measures to identify critical network components in multiple domains such as energy, water supply, telecommunication, social, and transportation. However, topological criticality measures usually do not conform to traffic flow dynamics (e.g., non-linear link performance functions) and they do not necessarily produce realistic results for road transportation networks. In order to overcome this issue, researchers incorporate traffic characteristics (travel time, link flow, etc.) into criticality metrics, mostly through traffic assignment. The studies that employ traffic assignment generally utilize a network performance measure (e.g., total system travel time) for the complete network, remove each link one by one, re-run traffic assignment and calculate the selected network performance measure for each link removal, and identify each link’s criticality based on the change in the network performance measure. Such full network scan approach creates computational burden, especially for large networks. Link removals can also cause network disconnectivity which makes it problematic to run traffic assignment. This talk will present the recently developed Link Criticality Index (LCI) which does not require link removals and can provide transportation link criticality rankings for large networks with a single User Equilibrium (UE) traffic assignment. The LCI scores are calculated based on the link flow changes and link performance functions during UE iterations. In order to take the alternative OD paths into account, the links scores are weighted based on the travel time differences between identified OD paths. Also, to reflect the higher importance of links which serve multiple ODs (or higher demand ODs), demand weights are used. The LCI was tested on well-known large transportation test networks and compared with other criticality measures. The findings showed that LCI successfully provides balanced link criticality rankings with respect to network connectivity and redundancy as well as the network flow conditions. Due to elimination of link removals, the LCI runtime is considerably shorter than comparable traffic assignment based approaches, i.e., in the order of the number of links in the network. The inclusion of network topology and demand characteristics, and shorter runtime make the LCI an appealing approach for transportation link criticality analysis.