Print

Sullivan, James L.

Civil Engineering

2009

MS

This thesis focuses on the advancement of the Network Robustness Index (NRI) for application to highway transportation systems, which had been developed in 2006. A comprehensive review of the scholarly literature related to the field of network-disruption analysis was conducted, of which the NRI is part. As expected, the field was found to have developed immensely since the late 1990s and now includes a wide variety of themes and approaches used to assess a network's response to a variety of disruptive events. Of particular relevance are those approaches which use repetitive link-and/or node-removal scenarios to develop a measure of a transportation-network's robustness or vulnerability (complementary concepts). Recently, these methods have begun to focus on the sequential application of equilibrium-based traffic assignments to measure the cost of a disruption to the network. It is crucial for these types of methods to handle the complexities of real-world transportation networks -one of which is the presence of isolated sub-networks, one main focus of this thesis.
Modifications to the procedure previously used to calculate the NRI are explored in this thesis to make the NRI work on networks with isolated subnetworks (sub-networks with only one link back into the main body of the network). This exploration revealed two important new findings -the first was that use of complete link-removal in the assessment of network robustness (as had been done previously for the IlRI and other indices) is not ideal and has limitations beyond not allowing the measure to be calculated when isolated sub-networks are present. Instead, a procedure that utilizes capacity-disruption (instead of complete link-removal) was found to be immune to the effects of poor connectivity and isolating links in real-world transportation networks. The study goes further to define a broad range of optimal disruption levels much lower than 100% which would make the procedure more effective. In addition, a new measure, the Network Trip Robustness (NTR), which allows inter-network comparisons across networks of varying sizes, is proposed and formulated.
The final analysis of the thesis explores the. modified NRI procedure further, by applying it to numerous hypothetical networks with isolated sub-networks. Through several unique evaluation methods, the results allow the range of optimal capacity-disruption levels to be narrowed to 70-90%. The NRI and the NTR are shown to work effectively on networks with isolated sub-networks when the modified procedure is used with a capacity disruption level in this range. This finding paves the way for a real-world application of these measures to a transportation network.