Influence of priority taking and abstaining at single-lane roundabouts using cellular automata

• A new cellular automata roundabout model is developed and calibrated with field data.
• Queue length and vehicle delay increase with higher rates of non-compliant driver behavior.
• Observed behavior percentages are influenced by arrival rate and turning movement balance.
• Jamming and priority reversal phenomena are replicated with a parsimonious model.
• Reinforces need for driver education and behavior change to improve the capacity of roundabouts in the United States.

Existing roundabout simulation models fail to consider all types of driver behavior which compromises their accuracy and ability to accurately evaluate roundabout performance. Further, these non-compliant driver behaviors, including priority taking and priority abstaining, are inconsistent with existing traffic flow theories. In this paper, a new cellular automata model, C.A.Rsim, is developed and calibrated with field data from five single-lane roundabouts in four northeastern states. Model results indicate that approximately 20% of the individuals in the driver population are inclined to priority taking and approximately 20% are inclined to priority abstaining behavior, though the observed levels of these types of behavior are naturally lower and vary with traffic volume. The model results also corroborate other research indicating that current models can overestimate capacity at higher circulating volumes, possibly a result of the jamming effect produced by priority taking behavior. The reduction in priority abstaining behavior, which is observed at older roundabouts, significantly reduces delay and queue length in certain traffic volumes. C.A.Rsim is also more parsimonious than many existing microsimulation models. These results provide insight on how variations in conflicting flow (i.e., traffic volume and turning movement balance) impact the amount of observed non-compliant behavior.