Impacts of Increasing Autonomous Vehicle Penetration on Urban Traffic Performance: A Multi-City Simulation Study

Connected autonomous vehicles (CAVs) will be gradually introduced into road networks, creating mixed-traffic environments where CAVs and human-driven vehicles (HDVs) operate together. Understanding how increasing CAV penetration influences traffic performance is therefore essential. This study evaluates the impact of CAV adoption across three cities—Berlin, Chicago, and Shanghai—using 165 simulations conducted in the Eclipse MOSAIC framework. CAV penetration was increased from 0% to 100% in 10% increments, and performance was assessed using mean vehicle speed, waiting time, and trip completion time during a three-hour peak-traffic period. The results show that the benefits of CAV adoption vary considerably according to network topology. Berlin exhibited the greatest relative improvement, with mean speed increasing by 58.8%, waiting time decreasing by 45.0%, and trip completion time decreasing by 33.3%. Shanghai achieved the highest final mean speed, increasing by 34.5% to 17.70 m/s under full CAV penetration, while waiting time and trip completion time decreased by 28.6% and 24.4%, respectively. Chicago demonstrated consistent improvements across most penetration levels, with mean speed increasing by 30.6%, waiting time decreasing by 35.5%, and trip completion time decreasing by 24.2%. Statistical analysis confirmed a strong relationship between CAV penetration and traffic performance. Spearman’s rank correlation revealed strong positive associations between penetration and mean speed across all cities (ρ=0.97–1.00). These findings demonstrate that higher CAV penetration can significantly improve traffic flow, although the magnitude of the benefits depends strongly on the characteristics of the underlying road network.