Identifying critical disruption scenarios and a global robustness index tailored to real life road networks

The ability to maintain functionality in transport infrastructure is critical during disruptions. To ensure operational robustness in transportation networks, it is necessary to identify the most vital or critical roads (or links), then reinforce them to increase their resilience. In the literature, conventional approaches to analyze road network robustness have involved efforts to first remove selected road segments (one by one, not collectively), then measure the impact of these changes. Based on these results, the levels of impact are ranked and links that demonstrate the most significant impacts are deemed to be the most critical. One of the most significant limitations of such approaches, however, is that they disregard the combined effect of road connectivity. This study advances the state of knowledge in transportation-based resilience analysis through the development of an approach to assess the impact of critical combination scenarios. The methodology involves a two-phase process. The first phase is based on the sensor (loop detector) location problem, within which, a selected number of high demand roads are identified as candidate critical links. Then, the second phase employs a series of discrete network design problem (DNDP) to find a variety of critical combination scenarios. The DNDPs are solved based on a system optimal relaxation method using Bender's Decomposition. Building further from these results, the extent to which a road network is robust (or fragile) is analyzed. The results of the DNDP solutions are demonstrated to be similar to a Lorenz Curve in which the area under the Lorenz Curve (in percentage) can be viewed as a global robustness index. This index can be used to compare and assess the robustness of different road networks and mitigation scenarios. To illustrate the practical utility of this method, this research applied the methodology to the Winnipeg, Canada road network. (C) 2016 Elsevier Ltd. All rights reserved.