A Co-Simulation Study to Assess the Impacts of Connected and Autonomous Vehicles on Traffic Flow Stability during Hurricane Evacuation

Hurricane evacuation has become a major problem for the coastal residents of the United States. Devastating hurricanes have threatened the lives and infrastructure of coastal communities and caused billions of dollars in damage. There is a need for better traffic management strategies to improve the safety and mobility of evacuation traffic. In this study hurricane evacuation traffic was simulated using SUMO a microscopic traffic simulation model. The effects of Connected and Autonomous Vehicles (CAVs) and Autonomous Vehicles (AVs) were evaluated using two approaches. (i) Using the state-of-the-art car-following models available in SUMO and (ii) a co-simulation study by integrating the microscopic traffic simulation model with a separate communication simulator to find the realistic effect of CAVs on evacuation traffic. A road network of I-75 in Florida was created to represent real-world evacuation traffic observed in Hurricane Irma s evacuation periods. Simulation experiments were performed by creating mixed traffic scenarios with 25, 50, 75 and 100 percentages of different vehicle technologies including CAVs or AVs and human-driven vehicles. HDV Simulation results suggest that the CACC car-following model, implemented in SUMO and commonly used in the literature to represent CAVs, produces highly unstable results On the other hand the ACC car following model, used to represent AVs, produces better and more stable results. However, in a co-simulation study, to evaluate the effect of CAVs in the same evacuation traffic scenario, results indicate that with 25 percentage of CAVs the number of potential collisions decrease up to 42.5 percentage