Two-region mixed autonomy network with park-and-ride: Macroscopic modeling, microscopic validation, and multi-scenario analysis

Urban transportation networks often face severe congestion and parking difficulties during peak periods due to demand–supply imbalances in traffic and parking. Park-and-ride (P&R) policies have been widely adopted to mitigate these issues by diverting vehicles to parking facilities near the boundary between central and outer regions of the city. However, the emergence of autonomous vehicles (AVs) fundamentally reshapes traffic dynamics and parking behaviors, thereby challenging the effectiveness of traditional transportation management polices. This study proposes a parsimonious two-region mixed autonomy network (MAN) model that incorporates self-driving-and-parking AVs and P&R dynamics to analyze their combined effects and mutual influences. The model is built upon an extended multi-pool representation tailored to MANs and employs an accumulation-based approach to capture the dynamic evolution of different vehicle families within the system. Discrete choice models and Monte Carlo simulation are further integrated to characterize the distinct parking behaviors of AVs and human-driven vehicles (HVs). The consistency analysis against a microscopic model confirms that the proposed macroscopic model accurately captures the temporal evolution of traffic network conditions. Additionally, experimental results demonstrate that while P&R remains effective for regulating HV inflow to central regions, its impact diminishes as AV penetration increases, owing to AVs’ ability to self-park. These findings suggest that AV-specific management policies may be required to complement traditional HV-oriented P&R strategies. Overall, the proposed model offers a structured approach to inform decision-making on various urban parking policies and services for better traffic management in the era of AVs.