Link criticality index : refinement, framework extension, and a case study

Abstract

The link criticality index (LCI) is a network performance-based measure that assesses criticality of network links based on flow fluctuations during traffic assignment. Unlike other network performance-based methods, LCI does not require link removal or multiple traffic assignments, and can account for topology, redundancy, congestion, and traveler behavior. However, the original LCI is based on a deterministic user equilibrium (UE) framework, which may lead to two issues: (i) a link's criticality index can be affected by origin-destination (O-D) pairs even if their demand is routed away from the link and (ii) identical links can be ranked unequally. In this paper, we suggest a refinement to the functional form of LCI to cater to analysts who prefer to work within the UE framework; and we extend the LCI measure to stochastic user equilibrium (SUE) and SUE with elastic demand (ED). These extensions resolve the issues related to UE-based LCI and add to the behavioral realism of the network model. To demonstrate validity of concept, the resulting LCI measure is applied to assess the criticality of bridges in Winnipeg, Canada. The obtained results are reasonable and consistent with the previous methods based on the full-network scan approach.