Two methods of ascertaining the level of service of signalized intersections are analyzed and a third is proposed. The 1965 highway capacity manual utilizes vehicles per hour of green as an indicator of the level of service, while a 1962 design procedure relies on the percentage of cycle failures, i.e., the relative number of cycles during which arrivals, in a Poisson manner, exceed the capacity for departures. It is shown that the two indexes have a varying relationship. Conversion charts and tables are developed. For a constant service level, a low approach volume will allow higher failure rates than may be tolerated for high volumes. In addition, the feasibility of employing average individual delay as an index of service level is examined. It is found that failure rate and individual delay also have a varying relationship. Using constant delay lines, higher failure rates were allowed for high volumes. A combined plot of service level and delay indicates an apparent divergence of lines, which is taken to show that although an intersection may satisfy HCM criteria, individual delay may vary considerably depending on arrivals, cycle length, and length of green time. A discussion of the paper by three other persons focuses on the apparent multiplicity of choices for evaluating and designing intersections. /author/
Abstract