Criticality analysis in road networks with graph-theoretic measures, traffic assignment, and simulation

Criticality serves as a foundational concept in the evaluation of the impact of disruptive events on the operation of road networks supporting the identification of road links whose unavailability affects to the greatest extent the mobility of people and goods. Several methods and techniques have been suggested so far for the quantification of criticality in road networks drawing from complex network theory, strategic transportation modeling, and traffic simulation. The current article provides a concise yet comprehensive overview of the relevant literature. Subsequently, it formulates a series of hybrid graph-theoretic measures that aim to evaluate, in a computationally efficient manner, the criticality of a road network's links by respecting their topological attributes and incorporating the inputs/outputs of user equilibrium-based traffic assignment models, assuming that all links are available and operational. Moreover, it formulates a flow reduction-based criticality metric, quantified through microscopic traffic simulation based on the well-established theory on the Macroscopic Fundamental Diagram. The validity and application potential of both approaches are experimentally scrutinized in two tailor-made case studies. The results of the first case study reveal that traffic volume centrality and two variations of a hybrid measure estimating the demand weighted decrease in network efficiency or the increase in total travel time complement each other and sufficiently explain road link criticality, as assessed through two different metrics of increased computational intensity and accuracy. Building upon this finding, an algorithmic procedure for combining the previously mentioned measures is proposed, enabling the targeted sampling of critical road link candidates. The results of the second case study reveal that the macroscopic flow-density relationship may not provide valid insights into road link criticality due to the weak correlation of average network flow reduction with average travel time increase as well as the occasional and paradoxical increase in average network flow after certain road infrastructure disruptions. It is experimentally proven that this is attributed to the shifted route choice behavior of road users. It is also shown that the macroscopic flow-density relationship can adequately assess the impact of disruptive events when route choices are kept fixed.