Evaluation of adaptive signal control technology--volume 1 : before-conditions data collection and analysis.

PDFTech ReportR27-127FHWA-ICT-17-008Adaptive controlTraffic signalsTraffic signal timingTraffic actuated controllersData collectionBottlenecksTraffic congestionIllinoisUnited StatesIllinois Center for TransportationShaik, Mohammed Abdul RawoofLiu, XueyingBenekohal, Rahim F.Illinois Center for TransportationIllinois. Dept. of Transportation. Bureau of Materials and Physical ResearchUnited States. Federal Highway AdministrationUS Transportation CollectionField evaluation of adaptive signal control technologies (ASCT) is very important in understanding the system\u2019s contribution to safetyand operational efficiency. Data were collected at six intersections along the Neil Street corridor in Champaign, Illinois, beforedeployment of SynchroGreen, an ASCT system. The field delay and queue length for \u201cbefore\u201d conditions were measured andcompared with estimates from the Highway Capacity Manual (HCM). The HCM estimates of stopped delay were significantly differentin 58.3% of the cases, representing overestimation in 73.5% and underestimation in 26.5% of the cases. On major streets of typicalintersections, in 72% of the cases, there were significant differences between HCM delay and field data; in 91% of the cases, HCMoverestimated delay by an average by 69%. On the minor streets of typical intersections, in 56% of the cases there were significantdifferences, and in 94%, the HCM overestimated the delay on average by 52%. The HCM estimates of 50th-percentile queue lengthwere significantly different in 61% of all cases, including overestimation in 56% and underestimation in 44% of the cases. For typicalintersections, 52% of the cases had significant differences, including overestimation in 93% and underestimation in 7% of the cases.On the major streets of typical intersections, 68% of the cases had 50th percentile HCM queue lengths similar to those from the field.However, in 28% of the cases it overestimated the queue length on average by 66%; and in 4% of the cases it underestimated it onaverage by 42%. For minor streets of typical intersections, in only 25% of the cases were the HCM queue lengths similar to those fromthe field; in 70% of the cases, HCM overestimated the queue length on average by 44%, and in 5% of the cases, HCM underestimatedit on average by 20%. In general, it was observed that the trends in the 50th- and 95th-percentile queue length comparisonssupported each other. Also, in 91% of the cases, the trends in delay and queue comparisons were either consistent with each other ordid not have any significant conflicts. However, in 9% of the cases, significant inconsistencies in trends were observed. Thus, to savetime, one may compare the HCM queue length estimates to field data to assess intersection performance, though the delaycomparison is preferred