Stochastic behavior of transportation systems with disruptions

Abstract

Transportation systems serve as one of the most fundamental infrastructure systems for the society, but are at risk from different types of disruptions. Due to the variety and unpredictability of disruptions, transportation systems have stochastic behavior, which refers to the change of the state of transportation systems. The understanding of stochastic behavior of transportation systems when facing disruptions is vital. This thesis studies the stochastic behavior of transportation systems with three representative levels of disruptions: single link disruption, multiple hazard events, and cascading disasters. Behavior is quantified by different metrics. The features of the transportation systems with different disruptions are analyzed and several specific problems are addressed. For the first level, the impact of single link disruption on the behavior of transportation systems is studied and the most critical links are identified. Particularly, the concept of “criticality” is interpreted from two perspectives, i.e. vulnerability and potential. The behavior of transportation systems in these two opposite scenarios is compared. A modified cell transmission model is proposed to measure the performance of urban transportation systems. For the second level, a mathematical framework is proposed for assessing the robustness of transportation systems against multiple hazard events. A Monte Carlo method is employed to analyze the impacts of simultaneously happened accidents. A heuristic algorithm is used to approximate intentional multiple attacks. For the third level, the stochastic behavior of transportation systems in disasters is investigated. Percolation theory from physics is introduced to analyze the connectivity of disaster-impacted traffic networks. An integrated model which combines localized attacks and random link failures is proposed to capture the features of localized natural disasters in traffic networks. Furthermore, an efficiency metric based on percolation theory is proposed to consider the degradation of road capacities. It is found that efficiency is the product of global connectivity, local connectivity and connectivity strength. The contributions of this thesis are three-fold. Firstly, it presents a better understanding of the stochastic behavior of transportation systems with three representative levels of disruptions. Secondly, it provides a reference for studying the stochastic behavior of other practical systems. Thirdly, the methods are applied to real-world networks with insights.