Practice Friendly Metric for Identification of Critical Links in Road Networks

Abstract

Despite the important planning value of transportation link criticality, the existing methodologies are mostly in the academic domain, and require in-depth technical skills and extensive data. The most common approach to identify critical links in transportation networks is to remove each link iteratively, conduct traffic assignment, and assess the criticality of each link based on the consequences of its removal. Since conducting multiple traffic assignment is costly for large networks, the authors of this paper recently introduced the link criticality index (LCI). The LCI utilizes the iterations in the Frank–Wolfe solution of the user equilibrium (UE) problem to provide link criticality ranking within a single traffic assignment. The LCI was shown to provide balanced rankings with respect to alternative routes as well as the link flows. However, the LCI is not practice-friendly because of the technical knowledge and data needed to run traffic assignments. Accordingly, this paper introduces a practice friendly link criticality index (PF-LCI). PF-LCI relaxes some of the technical requirements and uses some expert knowledge input data to provide “top” link criticality rankings that are consistent with the LCI. PF-LCI utilizes the network flow instances at different times of day instead of iterations of UE assignment solution. Expert knowledge input is sought for the major origin–destination pairs (ODs) and the viable routes between the selected ODs. The method is implemented on a small sample network and the Sioux Falls network to test PF-LCI’s capabilities. Results show that PF-LCI produces accurate rankings for the top critical links that are most relevant to practitioners’ concerns.