A Performance Analysis of Congested Multi-lane Roundabouts: A Case Study of East Dowling Road Roundabouts in Anchorage, Alaska

Existing studies on modern roundabouts performance in the US are mostly based on data from singe lane roundabouts that are not heavily congested. The main objective of this study is to investigate the performance of multilane roundabouts under congestion. The study also assesses how well software packages including RODEL, SIDRA and VISSIM predict capacity, queue length, and delay for the congested roundabouts. This study used data from East Dowling Road Roundabouts in Anchorage, Alaska. The roundabouts are currently operating with extensive queues during the evening peak hours. The most unique aspect of the research was the use of multiple video camcorders to capture vehicle turning movements in the roundabouts and the progression of vehicle queues at the entrances. With the resulting video records, we were able to accurately count the number of vehicles in the queues at any single minute during the peak hours. No existing roundabout studies ever recorded and analyzed such extended queues with the spatial and temporal resolution the authors were able to achieve with the collected videos. The results of the study found that unbalanced entrance flow patterns (i.e., one entrance has significant higher flow than others) can intensify the queue and delay for the overall roundabout. The authors also found that all three software packages overestimated capacities. SIDRA and VISSIM underestimated delays and queue lengths. RODEL overestimated delays and queue lengths at most of the entrances.