Near major bus terminals, multiple bus arrivals per signal cycle and a convergence of buses from conflicting directions can make it impractical to apply signal priority logic that attempts to interrupt the signal cycle for each bus. This research explores signal control logic for reducing bus delay around a major bus terminal in Boston, Massachusetts, where the busiest intersections see almost four buses per signal cycle. With a traffic microsimulation to model a succession of signal priority tactics, a reduction in bus delay of 22 s per intersection was obtained, with no significant impact on general traffic. The general strategy was to provide buses with green waves, so that they are stopped at most once, coupled with strategies to minimize initial delay. The greatest delay reduction came from passive priority treatments: changing phase sequence, splits, and offsets to favor bus movements. Green extension and green insertion were found to be effective for reducing initial delay and for providing dynamic coordination. Dynamic phase rotation, from lagging to leading left, proved less effective. Cycle-constrained free actuation, in which an intersection has a fixed cycle length within which two phases can alternate freely, provided flexibility for effective application of early green and green extension at one intersection with excess capacity. Emphasis is given to the approach of providing aggressive priority with compensation for interrupted phases, highlighting the compensation mechanism afforded by actuated control with snappy settings and long maximum greens.
Decentralized, actuated traffic signal control has many advantages, but it lacks mechanisms for coordinating with other signals along an arterial. When an intersection along an arterial is near or at oversaturation, coordination can play an important role in preserving and utilizing capacity by preventing spillback and starvation. Rules that can be added to a base of decentralized, clock-free actuated control are proposed for managing queues during periods of oversaturation. These rules are part of a larger framework for developing logic that will make arterial traffic signals self-organizing rather than organized around a common signal cycle. Features of the proposed logic include green truncation in case of intersection spillback, early green and double realization for left-turn phases prone to pocket spillback resulting from a limited turn-bay length, and dynamic coordination for groups of signals spaced too close together to hold a normal cycle's queue. With dynamic coordination, green waves are scheduled for each cycle following the critical intersection's critical arterial through phase, with noncritical intersections adjusting offsets based on queue counts and a logic. The proposed dynamic coordination logic allows temporary spillback at upstream intersections to prevent starvation at the critical intersection and temporary starvation at downstream intersections to prevent spillback at the critical intersection. Simulation tests using a benchmark network showed 45% less delay than standard coordinated control and 4% less than an optimizing control method designed for oversaturated arterials. Simulation tests on two realistic networks also showed delay reductions of 8% and 35% compared with coordinated control.
Reducing bus delay beyond what can be achieved with conventional transit signal priority requires making and responding to longer-range predictions of bus arrival time, which include dwell time at an upstream stop. At the same time, priority decisions based on such uncertain predictions should be reversible if the dwell time should be much longer than expected. Rules for applying these concepts are proposed for application in the framework of self-organizing traffic signal control developed by authors Cesme and Furth. Predicted arrival time is based on a calculation of expected remaining dwell time and is compared with the earliest time the bus phase can be expected to return to green. One possible decision is to expedite return to green so that secondary extensions (a feature of self-organizing control logic) are inhibited. The other is to hold the green; however, this decision can be reversed if updated predictions of expected remaining dwell time indicate that the bus will arrive after the maximum green extension has expired. Simulation tests on a corridor with nine signalized intersections showed a 75% reduction in bus delay, to only 5 s per intersection, with only a 3% increase in general traffic delay.
Time within an actuated signal cycle can be decomposed into time that is fully used, which is the saturation headway multiplied by the number of passing vehicles, and time that is wasted or lost. Activity network modeling is used to show the interaction between signal timing events and traffic flow transitions. Seven components of generalized lost time are identified: those associated with start-up, minimum green, parallel queue discharge (for simultaneous gap-out), extension green, parallel extension (for nonsimultaneous gap-out), the passing of the critical gap, and phase end. Simple formulas can be used to estimate all of these components for many practical cases, allowing one to estimate average cycle length without iteration. The modeling framework accounts for the dual-ring structure with minimum green and maximum green constraints and on–off settings for recall and simultaneous gap-out. Experiments with microsimulation software verify the formulas developed. The formulas show the sensitivity of lost time, and therefore average cycle length, to parameters that a designer can control including detector setback, critical gap, gap-out settings, and number of lanes. They also show sensitivity to total demand and to the ratio of noncritical to critical phase volumes.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.