Disrupted transportation networks under different information availability and stochasticity situations

Disruptive events may result in severe performance degradations in transportation networks. Properly evaluating the performance deterioration of a disrupted network with taking individual's behavioral response into account is a critical step within many operational and planning decisions, such as emergency rescue, traffic management, evacuation planning, infrastructure reinforcement, and so on. The focus of this paper is on the development and analysis of a set of disrupted network evaluation methods, considering different disruption information availability conditions and further an adaptation of these methods into a stochastic traffic environment and stochastic disruption context. Specifically, we define two typical information availability conditions: 1) information globally available (IGA); 2) information locally available (ILA), in terms of different data collection and transmission capabilities and traveler response behaviors. The IGA condition presumes that travelers know the complete set of disruption information before they make trips, while the ILA condition, however, represents another extreme information attainment condition, implying that travelers can observe or notice the occurrence of a disruption only when they reach in person the proximity of the disruption site. A new dynamic adaptive routing process is constructed to model individual routing behaviors for the ILA condition. The above modeling framework implies that, the total travel cost in the IGA condition is always better than the ILA situation in the deterministic traffic environment, and this conclusion is also applicable to the total traveler surplus in the IGA and ILA conditions in the stochastic traffic environment. On the other hand, the resulting ILA network flow pattern in the stochastic traffic environment can be fully accommodated by a Markovian traffic assignment process. When disruptions probabilistically occur, the proposed network evaluation method resorts to a combinatorial number of network evaluations and poses a very time-consuming process. For tackling this computational complexity, we suggested a Monte Carlo simulation-based evaluation framework and found that it can very precisely conduct network evaluations in a much more efficient manner, especially when the number of independently occurring disruptions is large.