Performance-based management of arterial traffic signal systems

Christopher Michael Day, Purdue University

Abstract

The current state of the practice in traffic signal system management is highly dependent on modeling of signal operations based on historical field data. While the initial design phase is typically rigorous, once systems are installed and timed, signal timing updates are infrequent. As agencies compete for increasingly limited resources, enhanced quantitative performance data from traffic signal systems is urgently needed. As such data becomes available, it is highly desirable to make use of it to improve operations. This dissertation defines and demonstrates several tools for measuring the performance of traffic signal systems and improving operations using that information. A portfolio of performance measures is proposed, including an analysis of the intersection phase capacity utilization; progression quality for coordinated phases; and estimated motorist delay time. Techniques for aggregating this data on an arterial system level are presented, including prototype graphics and tables that would be useful for agency performance reports. A new visualization graphic is developed called the Purdue Coordination Diagram (PCD), allowing a qualitative appraisal of coordinated phase performance. When coupled with quantitative data, the PCD is a powerful tool for analyzing offset performance in a coordinated signal system. Using a simple model, the PCD is used to predict the impact of changes to arterial offsets; the prediction is verified by field implementation, with observed performance slightly superior to the prediction. Probe vehicle travel times decrease by 1.7 minutes (28%) in the northbound direction on the arterial system; the difference is statistically significant. Procedures are proposed that aggregate the high resolution data to improve the computational efficiency of offset optimization algorithms. A “Link Pivoting” algorithm for optimizing offsets on a linear arterial is introduced, which revisits the Combination Method of the 1960s-1970s. Link Pivoting is compared against a heirarchical heuristic search, Monte Carlo simulation, and TRANSYT-style hill climbing, using alternate objective functions (maximize arrivals on green or minimize delay). Link Pivoting yielded the best performing offsets out of the algorithms tested, and is comparable to TRANSYT in terms of computational complexity. A simple model is proposed that extends the offset analysis methodology to include changes to cycle length.

Degree

Ph.D.

Advisors

Bullock, Purdue University.

Subject Area

Civil engineering|Transportation planning|Urban planning

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS