Roundabouts have become increasingly popular in recent years as an innovative operational and safety solution at both low volume and high volume intersections. And while tools are available for evaluating roundabout intersection operations, the answers provided by these tools can vary widely. This is particularly true for high-capacity roundabouts (that is, those with flared entry or double lanes). In the U. S., the benefits to installing single-lane roundabouts compared to signalized intersections have been demonstrated, but relatively few high-capacity roundabouts have been built. It is unclear how well the high-capacity roundabout will operate and under which circumstances it will perform better than a signalized intersection. This paper compares the capacity analysis suggested in the FHWA roundabout guidelines with the results of the analysis software packages RODEL, aaSIDRA, VISSIM, and Paramics. The macroscopic models RODEL and aaSIDRA apply formulas based on observed data from U. K. and Australia, respectively. These models use roadway geometry and/or driver behavior to estimate intersection capacity. The microscopic models VISSIM and Paramics simulate individual driver decisions in navigating the roadway network using a stochastic process. As a result, the microscopic model can be more closely calibrated to observed traffic conditions. The authors have found that the macroscopic models may not accurately measure multi-lane roundabout operations in all cases because these models lack sensitivity related to the effects of roadway geometry and gap acceptance. Microsimulation models were found to provide more accurate and reasonable results in this study, but required detailed calibration to accurately represent roundabouts with unique characteristics such as skewed approaches or closely-spaced intersections.
Abstract