The objectives of this project are to quantify the relationship between criticality and travel time reliability (TTR) in transportation networks to assess if one can serve as a proxy or an auxiliary measure for the other and to facilitate informed decision-making for transportation investments. In essence, the TTR and criticality topics are research fields of their own with distinct analysis approaches and metrics, even though both fields have commonalities starting with the aim of assessing the transportation network performance. TTR serves as an important indicator of the day-to-day performance of transportation networks whereas criticality focuses on the system performance during extreme events and/or disruptions. In addition, both TTR and criticality assessment of transportation networks suffer from data (in)availability and/or computational constraints, such as extensive (ideally probe vehicle) travel time data requirements for TTR and computational burden of running of traffic assignments (TA) as many as the generated disruption scenarios for criticality analysis. Given those challenges, a potential relationship between criticality and travel time reliability (TTR) in transportation networks can be used to infer about one-another when either of them cannot be properly calculated due to data or computational challenges. The relationship between these two concepts can support informed decision-making to simultaneously improve criticality, reliability, and overall performance of transportation networks.
A pre-cursor study to this project by PI Yazici indicates that TTR and criticality indeed exhibit correlations that warrant further investigation. Based on the analysis on 30 test networks, the majority (92%) of the calculated correlations between link criticality and TTR metrics are statistically significant. These results are based solely on correlation analysis of 30 test networks that cannot be generalized in a reliable manner and do not clearly identify the factors that contribute to the correlations such as network topology. Hence, this project pursues the research questions of:
Research Question #1: “To what extent there is a correlation between link TTR and criticality in transportation networks?”
Research Question#2: “What are the topological and operational factors that lead to higher correlation between TTR and criticality?”
Accordingly, this proposed project builds up on previous encouraging findings and focuses the following aspects for improvement and generalization: 1) Understanding the impact of network structure/topology on the correlation and 2) Establishing the generalizability and transferability of the results with larger set of real-life transportation test networks. For this purpose, novel neural networks (NN) models that can incorporate both the operational (e.g., link flows) and topological (e.g., betweenness centrality of the links) features of the transportation networks will be used.