Evaluating the Impact of CAV-Dedicated Lanes on Freeway Capacity and Safety: Findings from Simulation Experiments

Mixed traffic of connected and automated vehicles (CAV) and human-driven vehicles (HDV) can lead to potential traffic conflicts, dramatically reducing highway throughput. Dedicated lanes (DL) for autonomous vehicles could be a key solution to this issue. This study builds a modified simulation environment for optimizing the assignment and management policies of CAV DL in real freeway scenarios based on the third-party software SUMO. First, a continuous freeway network is constructed according to the 102 km bidirectional eight-lane Beijing-Xiong’an Freeway. Multiple vehicle types and heterogeneous driving behavior models are integrated to the simulation environment as well as the traffic demand distribution being generated with the LandScan population density data. Next, a multidimensional evaluation framework incorporating traffic efficiency, capacity, and safety is created. Further, a ranking model based on ideal solution similarity is used to evaluate the performance of the optimized CAV DL. Finally, five groups of simulation experiments are conducted to quantify the impacts of different assignment modes and management policies on CAV DL deployment in various scenarios, and the traffic evolutions in emergency situation are also examined. The experimental results show that, when the CAV penetration rate is below 20%, one DL is enough for the isolated operation of CAV; simultaneously, traffic origin-destination (OD) distribution significantly affects the efficiency of DL. When direct routes account for less than 60% of total demand, the DL should be placed as the most outside lane to mitigate the adverse impacts induced by the frequent on-ramp and off-ramp maneuvers of CAV; in emergency situations, compared to the outermost side, placing DL on the innermost side leads to greater disruption, as the congestion induced by the accident spreads more easily, with an average speed reduction of 30.77%. Above all, the proposed simulation environment can conveniently provide effective decision support for planning and designing CAV DL in freeway networks.