Model and Evaluation of a Centralized Non-recurrent Queue Management System for Road Events

Non-recurrent queues caused by road evens, such as accidents and anomalous slow/stopping vehicles, partially block roads, which will generate traffic congestion and unbalanced lane-level traffic. The large differences of speed at different lanes prevent lane changes of queued vehicles and further increase road congestion. This paper develops a centralized non-recurrent queue management (CNRQM) system to resolve non-recurrent queues as well as to minimize their congestion impacts. The system applies a game theoretic lane changing model to coordinate connected vehicles in queues and open lanes and provides speed and lane changing instructions for all connected vehicles in the control region to minimize the travel time delay of all vehicles. In addition, a centralized control is deployed to manage traffic at high market penetration rates (MPRs) of connected vehicles to overcome the drawback of the distribution control. The system is evaluated with microscopic traffic simulations under different MPRs of connected vehicles and demand levels. The results indicates that the centralized control can gain significant mobility improvements for both connected and non-connected vehicles even at very high MPRs. Moreover, the mobility benefits to each individual vehicles are evaluated to analyze the society impact of the proposed system to different types of road users.