Environmental Impact Estimation of Mixed Traffic Flow Involving Cavs and Human-Driven Vehicles Considering the Non-Equilibrium State

Abstract

The impact of Connected and Autonomous Vehicles (CAVs) on the traffic network has been extensively investigated, including road capacity, safety, traffic stability, etc. While existing network-level research predominantly highlights the potential enhancement of road capacity by CAVs , it often overlooks their influence on other critical aspects, particularly traffic stability. This study addresses this gap by focusing on the dual benefits of CAVs: road capacity improvement and traffic stability enhancement. We introduce a novel Dynamic Network Loading (DNL) model designed for mixed traffic scenarios involving both CAVs and Human-driven Vehicles (HVs). To achieve this, we adapt an existing velocity gradient model and generalize its discrete form to road networks. Our model employs three key parameters to characterize the advantages of CAVs: CAV market penetration rate, car-following safety time gap, and the propagation speed of a small disturbance. Numerical studies showcase the capabilities of this novel model in several aspects: Firstly, the new model can capture queue formation and propagation dynamics like prevailing dynamic network loading models. Also, it successfully replicates the stop-and-go traffic phenomenon, a feature lacking in existing DNL models. Moreover, the model provides a more accurate estimation of vehicle speeds and emissions compared to the Cell Transmission Model (CTM), with simulation results benchmarked against SUMO. Furthermore, this research explores the relationship between CAV market penetration rate and aggregate network emissions, utilizing a dynamic user equilibrium model. Our findings reveal that higher CAV market penetration rates correlate with reduced emissions. Specifically, a complete transition to 100% CAVs yields a substantial reduction in network emissions, estimated at approximately 20%.

Keywords: Mixed traffic, CAV market penetration, traffic stability, stop-and-go wave, velocity gradient model, vehicle emission.