This paper summarizes an evaluation of the benefits of optimizing traffic signal timing plans, coordinating traffic signal control, and implementing adaptive signal control at locations throughout the State of California. The signal timing optimization and coordination projects were carried out during the Fuel Efficient Traffic Signal Management (FETSIM) Program, between 1983 and 1993. This program involved 163 local agencies and 334 projects, improving 12,245 signals at a cost of $16.1 million, or $1,091 per signal. Timing plans were developed using the TRANSYT-7F software package. TRANSYT-7F also produced estimates of delay, number of stops and fuel consumption under the recommended signal timing plans, indicating the benefits of optimization. This created a database of the impacts of coordinated signal timing optimization under a variety of local conditions across the many projects implemented during the FETSIM program. Data from the Los Angeles Advanced Traffic Control and Surveillance System (ATSAC), allows an assessment of the performance of traffic responsive signal control at several intersections. The results presented in this paper are based on analysis of the accumulated statistics from the local agencies carrying out the signal improvement projects.

The TRANSYT-7F model estimates from 163 of the 334 FETSIM projects (49 percent), representing 55 percent of the total number of signals retimed, demonstrated positive results from signal retiming of coordinated signal systems. The study found an average 7.7 percent percent reduction in travel time, 13.8 percent reduction in delays, 12.5 percent reduction in stops and 7.8 percent decline in fuel use. This study excluded TRANSYT-7F results for oversaturated approaches, as the simulation overestimates the benefits of retiming under these conditions. Results of "before" and "after" floating car studies reported by some of the jurisdictions support the TRANSYT-7F figures. The average reductions reported by the floating car studies were: 7.4 percent in travel time, 16.5 percent in delay, and 17 percent in stops. The differences between these figures and the TRANSYT-7F values are due to the survey routes selected by the projects, which did not cover each link of the retimed signal network.

There was considerable variation in the improvement achieved by the retiming projects, for a variety of reasons:

• Quality of existing timing plans: When existing timing plans were already performing well, there was often little benefit in signal optimization.
• Network configuration: The largest savings were obtained on arterials, rather than signalized grid networks.
• Traffic Patterns: High volume systems with predominant through movements achieved the greatest improvements. Savings were marginal on systems containing several congested intersections in need of capacity improvements.
• Signal Equipment: Systems with actuated signals and easily modifiable control parameters achieved greater benefits.

"Before" and "after" field studies using floating cars assessed the performance of 76 of the projects that implemented coordinated signal control during the FETSIM program. These field data found an 11.4 percent travel time reduction, 24.9 percent delay reduction, and a 27 percent reduction in stops, when compared to the performance of the systems prior to coordinated control. Major benefits were reported for through traffic for signal spacing up to 0.5 mile and moderate to heavy traffic volumes (v/c ratio > 0.6). Signal coordination generally worsened performance on system entry links.

The paper reports that the estimated benefit/cost ratio for the cumulative impact of all the FETSIM projects, as determined by the recommended AASHTO method, was 17:1.

The adaptive signal control projects investigated by this study involved the application of Critical Intersection Control (CIC) at seven Los Angeles intersections. CIC uses detector volumes to generate green demand for each phase during a cycle based on volume data collected during the previous cycle. The level of service (LOS) at 4 of the 7 intersections improved for a significant percentage of the cycles assessed during the study, ranging from 27 to 55 percent. The only decreases in LOS for any of the cycles evaluated were 5 percent of the cycles at 2 of the intersections studies. The LOS for every other cycle at each of the intersections remained the same as it was prior to the implementation of adaptive control. The study found that the benefits of CIC were greatest for two-phase intersections with exclusive turning lanes, and/or unbalanced critical volumes. Benefits were considerably lower for multiphase signals due to the inflexibility of phase split lengths.