The "Highway Capacity Manual" and other traffic operations manuals give no guidance on the magnitude and factors affecting variability in control delay. However, knowing the variability in control delay has several useful applications, such as providing a more complete and statistically significant comparison of different signal timing or roadway geometric improvements. The objective of this research is to provide an exploratory analysis of the variability in signalized control delay using real-world measurements on an arterial corridor in Cary, North Carolina. Data were collected with a portable, onboard emissions and engine diagnostics measurement device, from which second-by-second speed data were recorded. A new method for estimating intersection control delay from second-by-second speed data is presented that attempts to take into account over-capacity conditions and closely spaced intersections. The data collection focused on recording a similar number of runs on a corridor before and after signal coordination was implemented. The results show that the distribution of control delay generally has a single, large peak at low delays, when the mean control delay was in the level-of-service AB range, whereas the distribution becomes increasingly bimodal (two distinct peaks) as the mean control delay increases. Another finding is that there is a direct relationship between mean control delay and standard error of control delay, but there is still variability in the standard error that cannot be explained solely by the mean value. Finally, differences in the before and after runs showed that a real-world signal timing improvement can reduce both the mean and the standard error of control delay as well as alter the shape of the delay distribution.
Abstract