The Oversaturated Signalized Intersection---Some Statistics

This paper describes the examination of an oversaturated signalized intersection with no turning permitted. During the green phase, data on the time between arrivals of successive vehicles at an intersection with no downstream bottleneck have been collected and analyzed. Measurements were made of only the vehicles in queue, i.e., vehicles that were stopped during the red phase and those that were stopped during the green phase because the starting wave had not traveled back far enough. Measurements were made only on weekdays from approximately 7:00 a.m. to 8:45 a.m. in order to restrict the sample, as far as possible, to commuters. It is shown that shifted Erlang density functions fit the data satisfactorily and that all vehicles from the third on may be considered as members of the same population with different means. The mean arrival times decrease from front to rear with all from the seventh on being the same. In addition, our statistical analysis leads us to accept the hypothesis that all interarrival times are independent. Experimental techniques and statistical methods are described. Two uses of these results are (1) to provide information for Monte Carlo simulations of traffic phenomena in which oversaturated situations are being considered, and (2) to develop, as will be done in a subsequent paper, a renewal model that takes account of variability in the empyting process, and that may be used in formulating control policies for the oversaturated situation.