Delay at Signalized Intersections considering non-stationary Traffic Flow

Capacity Manuals use vehicle delay to assess the level of service for signalized intersections. Average delay is typically derived from capacity and traffic volume during an analysis period using adapted approaches on the basis of queueing theory. Variability of traffic volume during the analysis period can influence the resulting delay and needs to be taken into account. The impact of non-stationary arrival flow rates is not yet explained completely by queueing theory, therefore a simulation study is performed. The analysis of measurement data shows that traffic volume during the peak hour can be distributed in very different forms. A set of abstract flow patterns is derived, representing simplified flow profiles of the peak hour, split into 15-min-intervals. Microscopic traffic flow simulations are performed for different signal control programs and volume-to-capacity ratios to determine the delay caused by the different flow patterns. It is shown that in most cases delay is higher for non-stationary flow than for stationary flow. Finally, a correction factor is developed for the delay computation method used in the German Highway Capacity Manual (HBS) to better reflect the volume distribution within the design hour.