Advances in connected and automated vehicle technologies have resulted in new vehicle applications, such as cooperative adaptive cruise control (CACC). Using CACC, cars can follow each other more closely, with braking and accelerating done cooperatively and synchronously. Previous studies have shown significant increases in throughput as a result. The impact on larger networks as a whole, however, is unclear.
To address this information need, a research team developed a simulation model that allows users to predict the more regional, network-wide impacts of CACC on traffic congestion. “Our mesoscopic simulation model lets us analyze congestion not just on the lanes with CACC, but also on nearby arterials and other roads,” says Michael Levin, assistant professor in the Department of Civil, Environmental, and Geo- Engineering.
To test their new model, the team calculated the impacts from implementation of CACC-exclusive lanes on two networks in Texas. One was a 28-mile corridor of I-35 near Austin, with 220 nodes, 95 zones, and 315 links; all vehicles were assumed equipped with CACC. Results indicate that CACC reduced congestion significantly and consistently as demand increased. “At the highest demand scenario studied, CACC reduced travel times by more than 50 percent,” Levin says.
The other network studied was Round Rock, with 2,744 nodes, 716 zones, and 4,236 links. The modeling assumed all freeway links had one lane converted into a CACC-managed lane. This reduced freeway capacity for non-connected vehicles; connected vehicles (CVs) could choose to use either regular lanes or the CACC lane. Market penetration of connected vehicles was assumed at 50 percent. “The model found the CACC lanes would have significantly higher speeds and significantly higher reliability,” Levin says.
A surprising result for Round Rock, he notes, is that with the CACC implementation, average travel times for all vehicles were higher than without the CACC lane. Non-CV demand for arterial roads increased because of the reduction in available lanes on freeways, and CV demand rose on arterials connecting freeways. “In addition, even though connected vehicles had better travel times and higher speeds than non-connected vehicles, CACC still increased CVs’ overall travel times due to changing congestion patterns elsewhere in the network,” he says.
The team’s results also indicate that adding CACC lanes at low CV market penetrations is likely to cause congestion. “At 15 to 25 percent market penetration, CVs create enough demand to increase congestion for non-CVs,” he says. “Network analyses should be used to determine at which market penetrations, and at which locations, CACC lanes should be deployed to reduce city network travel times. Our model can be used in conjunction with microsimulation to analyze route choice on regional networks.”
The team’s work was published in a paper titled “Dynamic traffic assignment of cooperative adaptive cruise control” in the Transportation Research Part C 90 (2018) 114–133. Lead author was Christopher Melson (Louisiana State University); co-authors were Levin, Britton Hammit (University of Wyoming), and Stephen D. Boyles (University of Texas at Austin). The work was sponsored by the Data-Supported Transportation Operations & Planning Center and the National Science Foundation.