In the summer of 2006, the city of Springfield, Oregon installed the first urban multi-lane roundabout in the state. It was hypothesized that after installation, speed variability on approaches to the intersection would decrease from the values with the previous signalized intersection. It was also hypothesized that the initially observed high incidence of driving errors associated with specific areas of the roundabout would decrease over time. Before and after speed recordings of approach roads to the intersection revealed a significant increase in mean speed, but no consistent change in speed variability. Some design features caused initial confusion amongst drivers negotiating the roundabout, but the number of observed incidences of confused behavior declined over the first six months of operation at a rate that fit a classic logarithmic learning curve. (Author/publisher)
Abstract