Requirements for automating signal timing plan design

Abstract

The optimization of signal timing plans for isolated intersections is a complicated procedure. Most current techniques such as the Canadian Capacity Guide for Signalized Intersections and the Highway Capacity Manual rely on relatively simple analytical models to estimate the optimum cycle time and green time splits, and leave the determination of the optimum phasing scheme to the experience of the user. The main problem with current procedures is that, because of the large amount of human effort required, only a very limited subset of all possible solutions can be examined. In addition, none of the current procedures are entirely satisfactory in their ability to deal with complicated congested conditions. The interaction of shared lanes, permitted left turn movements, cycle times, and green time splits requires an iterative approach to the design of signal timing plans, if current methods are to be used. Owing to the large number of calculations that need be done, current procedures are, by and large, computerized. Using any of the procedures currently available, each time a further iteration is needed, a new set of saturation flow rates or volume allocations must be manually coded into the input files of the computerized procedure. Thus, the search for the optimum solution to a signal timing plan problem can require a large amount of engineering time. This article describes the development of a notation and vocabulary that permits the automation of the logical decisions that must be made in order to optimize signal timing plan design. As well, the information that is necessary to define the intersection geometry and volume demands, in order that an automated process can optimize a signal timing plan is presented. The rules that can be used to determine the phase of discharge of each lane and pedestrian demand volume, the lanes that are opposed by a conflicting movement, and the movements that oppose traffic attempting to discharge from a turning lane are also described. Finally, a brief description of the framework of Signal Expert, a computer model that automates all of the above, is provided.