Accuracy of Stopped Delay Measured by Stopped-Vehicle Counts Method

The stopped delay is a primary factor for measuring the level of service at signalized intersections, and the 1994 Highway Capacity Manual proposed a method for measuring the stopped delay in the field based on direct observation of “stopped-vehicle counts.” This paper evaluates the expected error in delay measured by this method for various ranges of counting intervals and survey offsets. A mathematical formulation was derived to evaluate this error for different combinations of relevant parameters. Results indicate that the expected error may exceed ±100% of the deterministic delay, and significant errors are generally associated with shorter cycles, higher green to cycle ratios, lower degrees of saturation, and longer intervals. Significant errors are also reported when the interval perfectly divides the cycle length. A set of optimum offsets is determined for use with the 20-s interval and different combinations of other parameters. These offsets confine the expected error in the measured stopped delay to be within ±1%, a significant improvement regardless of whether the cycle length is divisible by the interval.