Traffic signals in the United States have evolved from fixed-cycle to vehicle-actuated operation to the present-day advanced signal systems and adaptive signal control technology (ASCT). An adaptive traffic signal adjusts its phase plan and signal timing in response to real-time traffic demand. Field evaluation of ASCT is very important in understanding the system’s contribution to traffic safety and performance improvement–and, hence, its effectiveness. The Illinois Department of Transportation (IDOT) is interested in in field evaluation of an ASCT on a corridor. Through a competitive bidding process, a Trafficware product called SynchroGreen® was selected for field implementation. Six intersections along Neil Street in Champaign, Illinois, were selected for this implementation. To evaluate the SynchroGreen system, the corridor’s performance prior to ASCT deployment was measured. The data are used as a basis to compare the performance of the system after it is deployed. This report presents the methodology and outcome of data collection, data reduction, and data analysis of the field conditions before implementation of SynchroGreen in Champaign. Traffic characteristics for four different time periods (AM peak, off peak, noon peak, and PM peak) were obtained from field videotapes. Those traffic characteristics include peak hours, hourly volume, saturation flow rate, signal timing, arrival type, field delay, and queue length. The field delay and queue length measured before implementation are used to evaluate the operational performance of the SynchroGreen system by comparing those characteristics after implementation. Those measures of effectiveness in the “before conditions” were also compared with estimations from the Highway Capacity Manual (HCM) to quantify the effects of volume changes and additional developments at Neil Street and Devonshire Drive through the course of the study. The HCM estimates of stopped delays were significantly different in 58.3% of the cases, representing overestimation in 73.5% and underestimation in 26.5% of the cases. On major streets of typical intersections, HCM delay estimates and field data were significantly different in 72% of the cases; in 91% of these cases, HCM overestimated the delay by an average by 69%. On minor streets of typical intersections, in 56% of the cases there were significant differences between HCM and field data; in 94% of these cases, HCM overestimated the delay on average by 52%. HCM estimates of 50th percentile queue length were significantly different in 61% of all cases, including overestimations in 56% and underestimations in 44% of the cases. For typical intersections, 52% of the cases had significant differences, including overestimations in 93% and underestimations in 7% of the cases. On the major streets of the typical intersections, in 68% of the cases, the HCM queue lengths were similar to those from the field. However, in 28% of the cases, HCM overestimated the queue length on average by 66%; in 4% of the cases, it underestimated the queue length on average by 42%. On the minor streets of typical intersections, in only 25% of the cases were the median HCM queue lengths similar to those from the field; however, in 70% of the cases, HCM overestimated the queue length on average by 44%, and in 5% of the cases, it underestimated it on average by 20%. In addition, a 95th percentile queue length comparison was conducted between HCM estimates and field data. In general, it was observed that trends in the 50th and 95th percentile queue length comparisons supported each other. The consistency between the results of stopped delay and the 50th percentile queue length comparisons for the 64 overlapping cases was analyzed. In 91% of the cases, the trend in delay and queue comparisons were either consistent with each other or did not have any significant conflicts. However, in 9% of the cases, significant inconsistencies in trends were observed. Thus, to save time one may compare HCM queue length estimates to field data to assess intersection performance, though the delay comparison is preferred. (Author/publisher)
Samenvatting