New York City Department of Transportation (NYCDOT) and Metropolitan Transportation Authority (MTA) with Greenman-Pederson Inc as prime modelling contractor.
Greenman-Pedersen, Inc. (GPI) simulated all traffic operations in the corridor in order to assess and recommend improvements for optimal TSP implementation, including changes to geometry, striping and signal timing.
Transit Signal Priority (TSP) is an important element of Bus Rapid Transit (BRT) that involves coordinated efforts between transit vehicle detection systems, traffic signal control systems, and communication technologies. In a nutshell, TSP means that buses signal their impending arrival at a signalized intersection and then receive the green light to drive straight through.
The New York City Department of Transportation (NYCDOT) and Metropolitan Transportation Authority (MTA) are embarking on an ambitious program to provide TSP to 6,000 buses in New York City. A key component of the project is New York City’s dedicated broadband wireless infrastructure (NYCWiN), which was created by the city’s Department of Information Technology and Telecommunications (DoITT) to support public safety and essential urban operations. Because NYCWiN supports the implementation of TSP without any additional hardware or infrastructure changes, this approach is particularly cost-effective and attractive for widespread implementation of TSP in New York.
The City’s Traffic Management Center in Queens, New York can use NYCWiN to process real-time messages from buses indicating their position and route, and then transmit wireless TSP instructions to local traffic signal controllers. These controllers can then expedite bus movements in one of three ways: by extending a current green signal; by cutting short a current red signal and returning early to green; or by queue jumping, that is, providing an advanced green signal at a specially configured near-side bus stop allowing only buses to jump the queue.
The first implementation of this system will serve buses on the M15 Select Bus Service (SBS) route in Lower Manhattan. SBS routes offer BRT features including low-floor three-door buses, special fare payment. TSP is provided for the 2.2-mile section of the M15 SBS route where an exclusive bus lane is not feasible, stretching from the Staten Island Ferry Terminal up through the Wall Street Financial District. It is an ideal test bed, characterized by slow-moving buses and heavy cross-street, pedestrian, truck and bicycle traffic with 34 signalized intersections.
NYCDOT selected TransCore to develop software for the TSP server. It also selected Greenman-Pedersen, Inc. (GPI) to simulate all traffic operations in the corridor and to assess and recommend improvements for optimal TSP implementation, which included changes to geometry, striping and signal timing. Simulations of morning, midday, and afternoon peak traffic conditions were performed to provide answers to critical TSP implementation questions. This allowed GPI not only to assess the benefits of TSP for M15 Select Buses and the effects of TSP on other traffic along the corridor and its cross streets, but also to identify which intersections might be able to successfully support TSP operation. Simulation was also the key to understanding how the system could handle competing calls for service. At the level of individual intersections earmarked for TSP, simulation helped to establish: the maximum green signal extension time for buses; the minimum time for cross-street phases; how far upstream calls for TSP service should be acted upon in each direction; whether queue jumping should be offered and how it should be implemented.
Aimsun software was used for this analysis, which called for the development of several custom software features including specific NYCDOT policies to maintain signal coordination, define vehicle queues and process subsequent TSP calls after servicing the first call in a cycle. The GPI team developed logic specifications representing each of these policies and TSS then created the corresponding custom Application Programming Interface (API) software to model them. Trafficware’s Synchro software was also used to help identify existing traffic problems, evaluate low-cost geometric and striping improvements, and develop optimal signal timings and coordination for the corridor. This effort was found to be critical to maximize the benefits of TSP by alleviating existing traffic problems and providing timing adjustments that facilitate TSP. This set of improvements is referred to as “Passive TSP” since it lays the foundation for TSP success without actual implementation. The Aimsun simulation analyses were conducted in three stages: First GPI created and validated a baseline model of current conditions for each peak period (7:30am to 10:30am, 12:30pm to 3:30pm and 4:00pm to 7:00pm). Secondly the team looked at passive TSP, evaluating the benefits of recommended traffic improvements and Synchro optimal signal timings relative to existing conditions. In the third and final stage, they looked at active TSP, evaluating the effectiveness of all improvements in addition to TSP and then studying and resolving the TSP implementation issues.
The complexity of the traffic environment in Lower Manhattan, as well as the need to simulate the operation of a new wireless TSP system (including communication time-lags) posed several challenges. Besides the custom software needed to model NYCDOT protocols for TSP operation, challenges included high pedestrian volume, and calibrating driver behaviors for lane changing and yielding at conflict zones. Manhattan’s notorious double-parking problem also necessitated repeated and extensive field observations to quantify double-parking activity and its effects on traffic delay. For the purposes of analysis, it was more important to accurately represent the delay caused by pedestrians to vehicular traffic rather than the actual number of pedestrians observed in the field, which included numerous mid-block jaywalkers. Similarly, calibration parameters for driver behavior in the corridor were particularly difficult to quantify. Therefore, GPI employed an iterative process to carefully adjust these parameters until model predictions of travel time, throughput and queues matched field observations of these traffic measurements within their normal weekday variation.
An important outcome of the analysis was the identification of intersections that were suitable for TSP implementation. Factors influencing this decision included the availability of time to shorten cross-street phases and meet minimum requirements for pedestrian crossing and also the ability to provide TSP without significant adverse effects on cross-street vehicular traffic. Since traffic volumes and signal timing plans vary during the day, these factors vary by analysis period so the set of intersections recommended for TSP has to be unique to each period. Intersections were not suitable for TSP while operating under control of human traffic agents. Even intersections recommended for TSP were not necessarily recommended for both directions of travel on the M15 SBS route. None of the candidate intersections exhibited both sufficient queuing and adequate real estate to provide queue jumping so TSP was not offered to buses on any approach with a near-side bus stop. Therefore, if an intersection selected for TSP had approaches both with and without a nearside bus stop, GPI recommended TSP for the approach without the near-side stop. After applying these factors to the 34 candidate intersections, GPI recommended TSP for 19 to 21 intersections for northbound buses and 22 to 24 intersections for southbound buses depending upon the time of day. The simulation model provides detailed animation displays that track a bus along its route both with and without the proposed TSP system. In addition, the model identifies the travel time savings due to TSP and the TSP action occurring for each intersection. Drive-through animation is particularly useful when examining the planned TSP operations and identifying refinements such as adjustments to the signal progression, maximum phase extensions or starting points for TSP actions.
Given the numerous issues involved in TSP implementation, and the impact of these decisions on TSP benefits, simulation modeling is essential to optimize TSP operations. In this case, a custom model was required to accurately represent all operating protocols for TSP recently established by the NYCDOT. This model is now suitable to represent any TSP implementation within New York City and could be adapted to represent TSP implementations elsewhere. GPI is currently applying this model and optimization process for the implementation of TSP along four other corridors in New York City.
Results from this model suggest that TSP will bring significant benefits to both the riders of M15 Select Buses and other traffic in the corridor: the combination of TSP and recommended geometric and signal timing improvements will reduce peak hour average travel time for M15 Select Buses by 7.4 per cent to 14.2 per cent and up to 3.2 minutes. It will also reduce peak hour delay for all traffic along the corridor by 11.9 per cent to 14.6 per cent and reduce peak-hour delay for all side street traffic by 3.5 per cent to 10.8 per cent. Total peak hour delay for all traffic in the study area will reduce by 8.4 per cent to 11.9 per cent.
TSP implementation with NYCWiN has occurred gradually over 2013 in accordance with the recommendations of this simulation analysis, followed by a comparison of actual travel time savings with the simulation predictions.