Traffic Signal Systems Solutions Toolbox: A Case Study from Pennsylvania

Abstract

The selection of the appropriate traffic signal systems solution (e.g., fully actuated [free] control, pretimed control, adaptive traffic signal control, traffic responsive plan) is critical to ensuring efficient traffic operations. The success of any selected solution is largely dependent on corridor characteristics and the nature of the experienced problems on the corridor. In addition, agency capability and constraints (e.g., staffing) along with infrastructure needs (e.g., detection) are other factors that are sometimes overlooked but can be critical during the selection process. So far, researchers have proposed tools, guidelines, and recommendations typically to select a single solution (e.g., adaptive) primarily by observing only corridor operational characteristics. Such practice leaves traffic engineers without an appropriate process/tool that can evaluate whether some other traffic solutions are more appropriate for given conditions. Therefore, the objective of this study is to develop a comprehensive traffic signal systems solutions toolbox that can incorporate corridor characteristics, operational objectives, agency capabilities, and agency constraints and offer a broad range of signal control strategies as solutions. The developed toolbox relies on data that are readily available to most agencies, such as annual average daily traffic and vehicle probe data, and incorporates a range of factors related to corridor characteristics and agency capabilities and constraints. The toolbox was applied to three corridors in Pennsylvania, U.S. Results from the case studies demonstrated that the toolbox could provide the appropriate recommendations based on unique circumstances. Additionally, the recommended solutions were generally in line with the expectations of agency representatives.

Keywords

operations adaptive arterial signal system signal timing signalized intersection traffic signal

Transportation agencies worldwide employ various policies, strategies, and tools to mitigate traffic congestion and improve the quality of transportation systems. Signalized intersections, in particular, are typically the primary capacity bottlenecks in urban and suburban areas ( 1 ). To address traffic congestion and improve the efficiency of signalized intersections, several traffic control strategies have been introduced by researchers and practitioners. Some of the common strategies include fully actuated (free) control, coordinated-actuated control, pretimed control, adaptive traffic signal control (ATCS), and traffic responsive plan selection (TRPS). In addition to control strategies, performance management tools, such as central traffic signal systems and automated traffic signal performance measures (ATSPMs), have recently emerged to monitor the performance of an intersection and ensure efficient signal operations and maintenance ( 2 ). However, the selection of the most appropriate strategy is critical to ensuring efficient signal operation, as its success is largely dependent on corridor characteristics as well as the nature of the experienced problems on the corridor. For example, fully actuated control with lane-by-lane detection and appropriate gap-out settings (along with volume-density and dynamic max features) typically outperforms other control strategies at isolated intersections, while ATCS could be a more effective solution along arterials with a certain degree of congestion and demand fluctuations. Similarly, pretimed operation—the most basic signal control method that does not include detection or any advanced technologies—can be the most appropriate solution for a grid network of signals to provide rigid coordination and better manage queues (or to provide maximum “walk” duration for pedestrians).

In addition to corridor characteristics and the identified problems, agency capability and maturity along with infrastructure needs (e.g., communication, detection) are other factors that are sometimes overlooked but can be critical during the selection process. Some of the proposed signal control strategies may require certain staffing resources and experience to properly maintain and operate the deployed system. Others may need traffic signal infrastructure improvements or maintenance funding which makes them unfeasible to deploy and maintain because of funding constraints. Therefore, it is important to develop a robust process to assist agencies with the selection of appropriate traffic signal system solutions while accounting for any mitigating factors in the selection process.

The process of selecting an appropriate signal control strategy has been tackled from several perspectives in the past. In some cases, agencies have relied on the use of guidance documents which provide insights into the selection but usually do not account for agency capabilities or the prevailing operational conditions on the network ( 3 ). Other researchers used microsimulation to evaluate the benefits of certain strategies during the selection process ( 4 , 5 ). Although beneficial, simulation also tends to focus exclusively on the operational elements and does not consider agency capabilities or funding constraints. Additionally, such an approach is typically labor-intensive and is not scalable to larger corridors or networks.

Decision-making tools were also introduced in the past, as these can help agencies decide whether a certain solution is a “go” or “no go” ( 6 ). However, existing tools and platforms, such as ATCS, are usually focused on a single specific control strategy ( 7 , 8 ). Therefore, current practice appears to be lacking tools that can comprehensively examine different control solutions and incorporate the previously discussed factors to help agencies in selecting the most appropriate traffic signal control solution.

To streamline and ensure efficient selection of traffic signal control strategy, the Federal Highway Administration (FHWA) introduced the “system engineering” (SE) process ( 9 ). The SE process is a good starting point and provides a framework for agencies to define their needs and specify system requirements upfront to guide the selection and deployment of the solutions. This approach ensures that the selected strategy is aligned with the agency’s overarching objectives and best suits their needs within the stated constraints (e.g., staffing, funding). While the SE process is beneficial when properly applied, its main limitation is that it can require a considerable expenditure of resources (e.g., staffing and data needs to apply the process). For this reason, some transportation agencies have developed their own forms and checklists as abbreviated versions of the SE process (e.g., PennDOT TE153) ( 10 ). However, as noted earlier, these forms are typically confined to a single system solution (e.g., ATCS) where the evaluation of other feasible solutions is lacking. FHWA began to address this shortcoming by expanding the model SE document to the broader field of central traffic signal systems ( 11 ). To the best of the authors’ knowledge, there were no attempts to make a comprehensive tool that will account for multiple signal solutions.

The objective of this research is to develop a comprehensive traffic signal systems solutions toolbox (Ts3T) that can incorporate corridor characteristics, operational objectives, agency capabilities, and agency constraints and offer a broad range of signal control strategies as solutions. The developed toolbox relies on data that are readily available to most agencies such as annual average daily traffic (AADT) and vehicle probe data. The toolbox includes a two-stage process where Stage I asks basic questions related to the studied network to provide an “early-exit” to a user when the solution is evident, while Stage II performs a more detailed assessment of the network conditions and agency capabilities and limitations when additional information is required. Details of Stage I and Stage II are provided later in the paper.

The remainder of the paper is organized as follows. First, a literature review is provided that describes various methods and tools that have been developed to identify appropriate signal control strategies. Next, the methodology followed during the development of Ts3T is presented. After that, there is a section detailing the application of the tool and then results from case studies are discussed. Lastly, conclusions and recommended future work are provided.

Literature Review

In the previous section, the authors mentioned that previous research tends to focus on the development of tools primarily related to the selection of ATCSs. Therefore, this section will first review efforts related to the selection of ATCSs. In addition, a review of tools and methods for the selection of other traffic signal control strategies is provided.

Although introduced in the 1970s and 1980s, ATCSs began to receive significant attention in the early 2000s in the U.S. ( 12 ). Early studies on ATCSs were related to the evaluation of these systems, especially in microsimulation, to determine whether specific ATCS brands would bring benefits if deployed in the field ( 4 , 5 , 13 , 14 ). However, because the development of microsimulation models requires extensive time and resources, researchers also developed guidance documents to facilitate decision-making about the selection and deployment of ATCS. For instance, Fehon et al. developed SE documents for ATCS ( 9 ). Similar to other relevant SE documents, the provided guidance serves to help users with the procurement of the equipment and software ( 9 ).

Several researchers developed various tools to support decision-making for ATCS ( 6 , 15 ). Mudigonda et al. developed a GIS-based decision support tool to evaluate and select ATCSs based on prototype analysis for Split Cycle Offset Optimisation Technique (SCOOT) and Optimization Policies for Adaptive Control (OPAC) ( 6 ). Wang et al. developed a Microsoft-Excel-based tool with simple simulation logic for SCATS, InSync, and ACS Lite ( 15 ). Ban et al. developed a decision-making tool to guide traffic engineers and decision-makers for the deployment of ATCS ( 16 ). Similarly, Mladenovic et al. suggested that the process of ATCS selection should be addressed using the multi-attribute decision-making technique, considering that various features are associated with ATCSs ( 17 ). Dobrota et al. proposed a dashboard tool for assessing the benefits of deployed ATCSs ( 7 ). More recently, Mitrovic et al. developed a data-driven tool for the selection of locations where ATCSs deployment is appropriate ( 8 ). Overall, while these studies provide some guidance for the selection of ATCS, as noted below, they only focused on a single solution. Additionally, some of these tools do not account for agency capabilities or funding restraints, which are shown to be crucial factor for the efficient operation of the system (6 –8, 15 , 16 ).

Studies were also conducted into the selection of non-adaptive control strategies, with a primary focus on actuated signal control, responsive traffic signal control, and emerging solutions related to real-time performance of traffic signals that can support signal operations ( 3 , 18 , 19 ). For instance, Kim and Corage provided design criteria for maximum green time settings for traffic-actuated control ( 20 ). Balke and Sunkari investigated the performance of various signal coordination features (force-off modes, transition modes, and coordination modes) in traffic signal controllers ( 18 ). Yun et al. evaluated the adaptive maximum feature in actuated traffic control and proposed recommendations for its applicability to various volume scenarios ( 21 ). Furth et al. modeled the relationship between sources of lost time (e.g., start-up lost time, minimum green, simultaneous gap) and control settings, traffic volume, detector settings, and other relevant factors for actuated control ( 22 ). Day et al. developed guidelines that warrant the deployment of fully actuated coordination ( 23 ). Wu et al. examined optimal unit extension time for actuated operations (24). Further, with regard to responsive traffic control, Abbas et al. proposed guidelines for the selection of robust and optimal TRPS system parameters and thresholds (3). In their follow-up study, the authors proposed a multi-objective plan selection optimization approach for TRPS control (Multiobjective Plan Selection Optimization for Traffic Responsive Control –https://doi.org/10.1061/(ASCE)0733-947X(2006)132:5(376). Lastly, it is important to document several tools/platforms that utilize emerging data sources such as commercial probe vehicle data, high-resolution detection and signal timing data, or a combination of these (i.e., crowdsource data), to develop software platforms that will support signal operations, primarily for the purpose of performance monitoring ( 19 , 25 –27). However, a key limitation of these platforms is that they lack functionalities that will support the decision-making process of selecting an appropriate type of traffic signal control.

Overall, multiple studies have been conducted to examine the benefits of one traffic signal control strategy over another or to provide recommendations and guidelines on the selection of a specific signal control strategy. However, to the best of the authors’ knowledge, there have been no attempts to develop a traffic signal control toolbox that can consider multiple solutions and incorporate non-operational factors. In essence, an effective tool should be able to assist agencies in the selection of the most appropriate traffic signal system that can meet their requirements while accounting for any of their constraints.

Traffic Signal Systems Solutions Toolbox (Ts3T) Architecture

The study approach for the toolbox uses a two-stage process for identifying the most appropriate signal system solution. The first stage uses basic informational elements related to signal systems that are readily available to agencies and is intended to conduct high-level screening of various signal systems solutions. As noted above, when the solution is obvious based on the provided information, the toolbox provides the most appropriate solution and offers an “early exit” to the user (i.e., the user does not go through the Stage II process). If more than one feasible solution emerges from Stage I, then the process takes the user to Stage II for more detailed information that is still usually available to agencies.

Description of Stage I Elements

Stage I starts with the general premise of using data from widely available sources to eliminate the need for an extensive project-specific data collection effort to conduct a preliminary analysis of the corridor characteristics. Stage I of the developed toolbox uses four informational elements and modules for identifying the most appropriate solution: 1) network characteristics, 2) traffic congestion level, 3) maintenance considerations, and 4) pedestrian activity. Figure 1 displays these informational elements and the developed flow chart and decision-making process in Stage I. Brief information related to the flow chart and decision-making process is also provided below.

Section 1A. Network Characteristics: This module identifies traffic network characteristics as the effectiveness of the proposed solutions is influenced by traffic patterns and the way traffic flows in a network. This module includes four network types for the user to select: 1) one-way grid, 2) two-way grid, 3) arterial, and 4) isolated intersection. If a user selects “isolated intersection,” then the tool gives an early-exit as illustrated with the stop sign in Figure 1. In this case, the tool proposes fully actuated control as the solution, along with actuated control features such as volume-density control, dynamic max, phase reservice (if queue spillback is an issue), and dilemma zone protection. If a user selects “one-way grid,” the tool also provides an early-exit and recommends using pretimed coordination (with time-of-day plans). In all other scenarios, the tool suggests proceeding with the remaining three informational elements for Stage I.

Section 1B. Traffic Congestion Level: This module conducts a preliminary analysis of the congestion state at the critical intersection. The critical intersection is likely to govern control decisions for the signal system because it has the least green time available based on its combination of volume, geometry, and signal phasing. For example, the critical intersection may have the highest minor street traffic volume, longest pedestrian crossings, more left-turn phases, or more phases because of additional approach legs. The tool comprises one input module, two default data modules, and a computational module. A description of each module follows.

User Input: The analyst must provide the road AADT at the critical intersection, along with their functional description (major/minor arterial, major collector, or minor road), and the peak directional factor occurring in each AM and PM peak period. The start and end hour of each peak must be specified. In addition, the user must enter the lane configuration on each road. The tool assumes that an exclusive left-turn lane is present on each approach (which is often the case at the critical intersection). Finally, the user can enter the maximum desirable critical lane volume (or capacity). A default capacity value is also available to the user.

Traffic Pattern Group (TPG) Defaults: Pennsylvania Department of Transportation (PennDOT) specifies a series of ten TPGs ( 28 ). A TPG depends on site location (urban/rural) and the roads’ functional classification (interstate, arterial, etc.). By selecting a TPG number, the tool automatically extracts its default hourly volume distribution. The tool can also accept any other desired hourly volume distribution as well.

Turning Movement Percent (TMP) Defaults: A TMP table is borrowed from the work of Iroanya, where the author developed multiple tables for estimating turning movement counts based on the two roads’ functional class, the approach lane configuration, number of legs, and other parameters ( 29 ). In the Stage I tool, the functional road class of both roads determines the fraction of left and right turns. Thus, at the completion of this module, each turning movement in each hour of the day is specified.

Capacity Rating: Here, the tool compares each hourly critical lane volume (v) with the desired input capacity (c). Then, in each hour, the capacity rating is categorized as “well below” capacity (when v/c ≤ 0.60), “near capacity” (0.60 < v/c ≤ 0.98), “some over capacity” (v/c > 0.98, but for less than 2 h), and “significant over capacity” (v/c > 0.98 for 2 h or more). Under certain capacity ratings, the signal timing objective may be best achieved with a single strategy, and the tool provides an early exit by recommending that strategy.

Once the capacity rating is computed, if the module estimates that the critical intersection is well below capacity, then the tool provides an early-exit and proposes revisiting timing plans to reallocate green times more efficiently, as there is available green time at the critical intersection. The tool also suggests reviewing ATSPMs (if available) and controller clock and drift for coordinated intersections to identify potential sources of inefficiency. If the tool estimates significant over capacity, the tool also gives an early-exit and proposes omitting some phases/movements in some plans or increasing cycle length with the caveat that very long cycles can increase queue interactions and negate the benefit of long cycles ( 30 ). This is because, in an over-capacity situation, the signal timing objectives shift to managing queues and throughput, and commercially available optimization algorithms are typically not designed for those objectives.

Section 1C. Preventive Maintenance Considerations: This module seeks to assess the way an agency performs preventive maintenance by identifying the time interval between the last two maintenances conducted. For agencies that have not conducted preventive maintenance in the last 9 months or only conduct during citizen concern calls, the tool proposes signal control strategies that do not require proactive maintenance of signal infrastructure (e.g., pretimed coordination or coordinated-actuated coordination with only stop bar detection on non-coordinated movements). For agencies that perform regular and proactive preventive maintenance, the tool suggests proceeding with the last informational element for Stage I.

Section 1D. Pedestrian Activity: The last module determines the extent of pedestrian activity in the network, as high pedestrian activity (which typically comes with pedestrian recall) can diminish the effectiveness of certain signal control strategies by reducing the amount of available green time in a given cycle. For this module, the user is expected to categorize the study corridor as “high,”“medium,” or “low” pedestrian activity. Note that range of pedestrian volumes are provided to the user in the tool to make this categorization less subjective. If low or medium pedestrian activity is selected, the tool would recommend proceeding to Stage II. If high pedestrian activity is selected, the tool still recommends conducting the Stage II analysis but excludes ATCS from the solution set. This is because ATCS may have limited ability to leverage its capabilities during high numbers of pedestrian crossings ( 8 , 31 ).

Description of Stage II Elements

Analysis within Stage II is designed to analyze the study network in greater detail with the goal of identifying the most suitable signal systems solution that passed the Stage I screening. To ensure the proposed solution is not only based on operational conditions but can incorporate other constraints, as previously noted, Stage II considers both agency capabilities (section 2A), operational objective (section 2B), and corridor characteristics (sections 2C–2F). The flowchart presented in Figure 2 shows the relationship between these elements within Stage II. Additional discussion is provided in the following paragraphs.

Section 2A. Agency Capabilities and Resources: In a relevant publication sponsored by FHWA, it was recognized that, rather than using quantitative variables (e.g., number of employees in the agency, education level, amount of budget) to describe agency capabilities, a more efficient way is to use archetypes ( 32 ). Generally, archetypes are defined based on multiple scenarios (e.g., “the agency has a proactive approach to traffic operations management” or “insufficient operations staff resources are compensated by effective capital expenditure on systems that work effectively for long periods without frequent systematic retiming”). For this toolbox, we defined three archetypes (based on definitions found in the original study) to assess agency capabilities and resources which are then used to tailor the proposed solutions to meet agencies’ unique needs and circumstances. Definitions of archetypes are:

Archetype 1: The staff undertake signal timing reviews by hiring outside consultants to collect turning movement counts at all intersections, reviewing all fixed intervals at each intersection, developing and programming signal optimization software for calculating optimized signal timings, evaluating signal timings using micro-simulation, and providing revised signal timings to the agency.

Archetype 2: Operations staff is limited but maintenance capabilities are good and capital programs are well funded. The limited operations staff are used to performing travel time analysis and systematically reviewing the network for unexpected queuing.

Archetype 3: Has limited operations and maintenance staffing and limited capital resources. The operations staff is small, but the agency has a well-developed operational objective of providing free-flow travel time or equitable operation when that is not possible, as stated in a mission.

In this section, users are asked to provide an answer on the frequency of retiming, dedicated staff for monitoring equipment health, and funding streams to improve signal operations (e.g., annual operating budget, capital funding, grants). The answers to these questions are then used in tool’s backend to correlate answers with specific archetypes. It needs to be noted here that information about staff knowledge was purposely omitted as it is very subjective, and it might encourage tool users to provide ingenious answers. While it is unfeasible to provide an exhaustive list of responses to these questions and the corresponding Archetypes, a few examples are provided below:

Agency 1: Agency has good maintenance capabilities, performs cyclical retiming (either agency-conducted or through consultants), but has limited capital resources Archetype 2

Agency 2: Small agency operating with constrained capital resources and limited staff capabilities and mostly using pretimed and semi-actuated, coordinated operation Archetype 3

Section 2B. Primary Operational Objective: The purpose of this module is to ensure the selection process is objective-driven. Users are asked to identify the primary operational objective for the corridor and the tool’s backend will (in addition to other factors) recommend solutions that help achieve this objective. Considering that operational objectives can be different for peak versus non-peak periods, users can also define different objective for non-peak periods. There are several objectives identified for users to choose from, such as “Serve all vehicle movements as efficiently as possible, including side streets and left turns,”“Reduce pedestrian and/or cyclist delay,”“Improve transit service and reduce person delay,” and “Prioritize mainline movements to reduce mainline vehicle delay and stops,” among others.

Section 2C. Diversion of Traffic from a Freeway: Signalized corridors intersecting or parallel to freeways are affected by diversion of freeway traffic resulting from congestion on the freeway. These diversions can occur on a recurring basis as a result of congestion, or on a non-recurring basis as a result of incidents (e.g., inclement weather, traffic incidents). When traffic diversions from a freeway are non-recurring, a large spike in demand can be observed on the signalized corridor in a short amount of time. This type of large change in demand caused by a non-recurring diversion from a freeway requires an immediate and significant change to the signal timing, rather than incremental changes to signal timing on a cycle-by-cycle basis, and it is generally considered that incident signal timing plans can outperform ATCS or other solutions and “absorb” sudden and large changes in traffic volumes ( 33 ). On the other hand, the diversion pattern with a recurring freeway bottleneck is typically smoother and can be accommodated with incremental changes to signal timing on a cycle-by-cycle basis. Therefore, when recurring diversions occur because of a bottleneck on the freeway, certain traffic signal solutions, such as ATCS, can help manage traffic and alleviate congestion. The tool recognizes these differences, enabling it to offer the solution that can address corridor conditions.

Section 2D. Special Event Generators: Special event generators can create a significant impact on traffic operations in the nearby area. Special events can also vary with regard to their characteristics such as the size of the special event, whether arrivals and departures occur at the same time, and whether the start and end time of these events is generally known to allow agencies to respond to the changes in traffic conditions. This module uses these characteristics of special events to better tailor recommended traffic signal solutions. Figure 3 shows a set of solutions based on the characteristics of special events (after relaxing other conditions in Stage II).

Section 2E. Level of Travel Time Reliability (LOTTR): This module calculates the LOTTR metric by calculating the ratio of longer travel times to a “normal” travel time using vehicle probe data. The LOTTR metric is used in this module as a proxy to demand fluctuations in the study area, since variation in demand is an important factor in the selection of the appropriate signal system solution. While direct calculation of demand fluctuations would be ideal to determine the most feasible solution, most agencies do not have access to continuous count stations on arterials (or systems such as ATSPMs with count detection) to obtain yearly volumes. Therefore, we proposed using the LOTTR metric as a proxy, since vehicle probe data is available to most agencies on corridors. Additionally, FHWA recommends using the ratio of the 80th percentile travel time to the 50th percentile travel time to calculate LOTTR and suggests using a threshold of 1.50 to determine whether a facility is reliable or not (uses the same threshold both for freeways and arterials) ( 34 ). In this research, we initially followed the same methodology and the threshold provided by FHWA. However, based on the testing of LOTTR on several corridors that are known to have reliability issues, the LOTTR results were generally much lower than 1.50. Based on these findings, we revised the calculation and used the 90th percentile instead of the 80th percentile and reduced the threshold from 1.50 to 1.30.

Section 2F. Intersection Complexities: Intersection complexities can substantially constrain the flexibilities in signal timing and thus limit the effectiveness of certain signal system solutions. Therefore, identifying those complexities in the selection process is crucial. For instance, previous studies indicated that the effectiveness of ATCSs diminishes considerably in the case of complex geometry or when there are frequent preemption events ( 8 ). Within this module, the following conditions were identified as intersection complexities: the presence of railroad (or ramp) preemption, exclusive pedestrian phases, complex overlaps (beyond typical right-turn overlaps), and the presence of complex geometry (e.g., 5-leg intersection). In the tool, if a user identifies one or multiple complexities, the tool rules out ATCS as a feasible solution.

Traffic Signal Systems Solutions Toolbox (Ts3T) Development

The tool was developed in MS Excel and is readily available software for agencies which can be obtained on request. The decision-making logic was coded using VBA scripting. For LOTTR calculation using probe data, an additional module is developed using Python. As shown in Figure 4, this component of the tool connects to the Regional Integrated Transportation Information System (RITIS) application programming interface (API) and calculates the LOTTR metric ( 35 ). LOTTR is calculated for a year worth of data, during four distinctive periods, AM, midday, PM peak hours (for typical weekdays, Tuesday, Wednesday, and Thursday) and Saturday midday peak. This information is then used to assess the reliability of the facility in the decision-making process as previously discussed. Lastly, a set of recommendations is generated and reported to the user. Figure 4 shows the general tool workflow.

An excerpt of the tool’s interface, Stage I and Stage II are also presented in Figure 5. Once users provide input for all sections within a certain stage, the developed tool provides a list of recommendations.

Application of Toolbox to Case Studies

To test the developed toolbox and ensure that the tool offers reasonable solutions under varying conditions, we developed case studies by applying the Ts3T to several real-world corridors in the Commonwealth of Pennsylvania, U.S. This section demonstrates the way the toolbox is utilized and the solutions it offers under varying conditions (please note that presenting an exhaustive list of Stage II solutions is beyond the scope of this paper). These corridors were selected by state DOT representatives because of various operational conditions observed. Note that maintenance-related constraints were relaxed in the tool both for Stage I and II to be able to test the responsiveness of the tool based on the operational conditions. The following subsections provide more details on the case studies and findings from the tool.

Case Study 1: McKnight Rd in McCandless Township

The first case study represents an arterial road consisting of five signalized intersections with a mainline AADT of 26,000, as shown in Figure 6. The intersection of McKnight Rd and Cumberland Rd represents the critical intersection because of the highest AADT reported on the crossing street (Cumberland Rd) and the number of lanes for the critical phases. Some other input parameters used for Stage I are:

Stage I: Input

○ Network characteristics: Arterial

○ Principal arterial (AADT 26,000)/Major collector (AADT 12,000)

○ Preventive maintenance: Frequent (assumed)

○ Pedestrian volume: Low

Stage I: Output

○ The critical intersection is operating below capacity; estimated v/c < 0.6 at all times → Does not require Stage II analysis.

○ Provided recommendations:

■ Revisit timing plans for more efficient green allocation.

■ Review ATSPMs (if applicable).

■ If coordinated, review the controller clock and drift.

Case Study 2: US 15 around Dillsburg

The second case study represents a section of US 15 that consists of four signalized intersections (shown in Figure 7) with an AADT of 35,000. The intersection of US 15 and York Rd (PA 74) represents the critical intersection. Some other input parameters used for Stage I are:

Stage I: Input

○ Network characteristics: Arterial

○ Principal arterial (AADT 35,000)/Minor arterial (AADT 8,500)

○ Preventive maintenance: Frequent (assumed)

○ Pedestrian volume: Little-to-no pedestrians at the intersection

Stage I: Output

○ The critical intersection is operating near capacity; estimated v/c is between 0.60 and 0.98 at all times → Requires Stage II analysis.

For Stage II analysis, the following inputs were used:

Stage II: Input

○ Agency capabilities and resources: The agency has dedicated staff for monitoring and maintenance, and secured funding streams to improve signal operations (assumed).

○ Operational objective: Prioritize mainline movements to reduce mainline vehicle delay and stops (during peak and off-peak periods).

○ Diversion of traffic from a freeway: Corridor is not used by diverted traffic from a nearby freeway.

○ Special event generator: There are no special events in the vicinity of the study corridor that generate traffic multiple times in a year.

○ LOTTR: Lower than 1.3 for all peak periods (AM, PM, midday [typical weekday])

○ Intersection complexities: Presence of preemption

Stage II: Output

○ Provided recommendations:

■ Cyclic (3–5 years) signals retiming

■ Consider reallocation of split time to better serve mainline movements.

■ Use floating force-off if the corridor is coordinated-actuated.

■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”).

■ Consider increasing cycle length while managing queues.

■ Review ATSPMs (if applicable).

■ Use ATSPMs for signal retiming such as reallocation of split times and offset adjustments.

Please note that “yellow trap“ refers to scenario where a traffic signal exhibits a yellow light towards one direction of traffic and a green light towards the opposite direction at a shared intersection which could lead to unsafe left turns. Note that ATCS was not recommended primarily because of reliable travel times observed along the studied corridor. Additionally, responsive signal control was not recommended because there are no significant/sudden traffic volume changes caused by special events or traffic diversions from nearby facilities.

Case Study 3: US 30 (Lancaster Avenue), Radnor Township

The third case study represents a section of US 30 that traverses Radnor Township and consists of six signalized intersections with an AADT of 26,850, as shown in Figure 8. The intersection of US 30 and Radnor Chester Rd represents the critical intersection. Some other input parameters used for Stage I are:

Stage I: Input

○ Network characteristics: Arterial

○ Principal arterial (AADT 26,850)/Minor arterial (AADT 9,450)

○ Preventive maintenance: Frequent (assumed)

○ Pedestrian volume: Medium

Stage I: Output

○ The critical intersection is operating with some overcapacity, estimated v/c > 0.98, but for less than 2 h → Requires Stage II analysis.

For Stage II analysis following inputs were used:

Stage II: Input

○ Agency capabilities and resources: The agency has dedicated staff for monitoring and maintenance, and secured funding streams to improve signal operations (assumed).

○ Operational objective: Prioritize mainline movements to reduce mainline vehicle delay and stops (during peak and off-peak periods).

○ Diversion of traffic from a freeway: Corridor is used by diverted traffic from a nearby freeway.

○ Special event generator: There are medium-sized special events in the vicinity of the study corridor that generate traffic multiple times a year, arrivals (departures) to the special events are not in large groups, and the beginning/ending of these special events are generally not known in advance.

○ LOTTR: Higher than 1.3 for at least one peak period (AM, PM, midday [typical weekday, typical Saturday])

○ Intersection complexities: There are no complex operations.

Stage II: Output

○ Provided recommendations:

■ An adaptive signal system is recommended based on the expected traffic fluctuations and selected operational objective.

■ Please refer to system requirements to make sure that the selected adaptive system can serve all movements efficiently.

Although the corridor exhibits high travel time unreliability and there are no sudden increases in traffic flows caused by traffic diversion from a nearby facility (regular diversion without abrupt increase) or because of planned special events, responsive control was eliminated. Note that, although an adaptive system is recommended, users are still advised to ensure that the deployed ATCS technology meets corridors objectives.

Table 1 summarizes the key input parameters used within these case studies as well as the recommendations proposed by the toolbox. More importantly, results were presented to agency representatives in an attempt to validate Ts3T’s outputs. It was found that, for all proposed case studies, the recommended solutions are in alignment with the judgement of the agency’s representatives based on their local knowledge. In some instances, agency representatives provided more insights into corridor operations to select the most appropriate input parameters; those were primarily related to the diversion of traffic from nearby facilities and types of special events.

Table 1.

Summary of Traffic Signal Systems Solutions Toolbox (Ts3T) Application Results for the Case Studies

Case study	Input parameters	Recommendation	In line with agency staff
1. McKnight Rd *	■ Arterial ■ Frequent maintenance (assumed) ■ Low pedestrian activity v/c < 0.6 at all times	■ Revisit timing plans for more efficient green allocation. ■ If coordinated, review the controller clock and drift. ■ Review ATSPMs (if applicable).	Yes
2. US 15	■ Arterial ■ Frequent maintenance (assumed) ■ No pedestrian activity ■ 0.6 < v/c < 0.98 at all times ■ Agency capabilities and resources are substantial. ■ The operational goal is prioritization of mainline movements (during peak and off-peak). ■ Without diversion of traffic from a freeway ■ Without a special event generator ■ LOTTR < 1.3 for all peak periods ■ Presence of preemption	■ Cyclic (3–5 years) signals retiming ■ Consider reallocation of split time to better serve mainline movements. ■ Use floating force-off if the corridor is coordinated-actuated. ■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”). ■ Consider increasing cycle length while managing queues. ■ Review ATSPMs (if applicable). ■ Use ATSPMs for signal retiming such as reallocation of split times and offset adjustments.	Yes
3. US 30	■ Arterial ■ Frequent maintenance (assumed) ■ Medium pedestrian activity ■ v/c > 0.98 for less than 2 h ■ Agency capabilities and resources are substantial. ■ The operational goal is prioritization of mainline movements (during peak and off-peak). ■ With the periodic diversion of traffic from a freeway ■ With unpredictable special event generators ■ LOTTR > 1.3 for some peak periods ■ Without complex operations	■ An adaptive signal system is recommended based on the expected traffic fluctuations and selected operational objectives. ■ Please refer to system requirements to make sure that the selected adaptive system can serve all movements efficiently.	Yes

Note: ATSPMS = automated traffic signal performance measures; LOTTR = level of travel time reliability; v/c = critical lane volume/desired input capacity.

Case study 1 required fewer inputs, as Stage I provided an “early exit” because of very low v/c ratio observed at the critical intersection. Therefore, Stage II parameters are not shown in the table.

Table 2 shows the results of the sensitivity analysis with different agency archetypes while maintaining the same corridor operational characteristics for the case studies presented above. As can be observed from the case study results, the tool recommends different signal systems solutions for the same corridor based on varying agency capabilities and resources.

Table 2.

Results of the Sensitivity Analysis with Different Archetypes for the Case Studies

Case study	Previous recommendation in Table 1 assuming Archetype 1	Recommendation with Archetype 2	Recommendation with Archetype 3
1. McKnight Rd	■ Revisit timing plans for more efficient green allocation. ■ If coordinated, review the controller clock and drift. ■ Review ATSPMs (if applicable).	■ Revisit timing plans for more efficient green allocation. ■ If coordinated, review the controller clock and drift.	■ Revisit timing plans for more efficient green allocation. ■ If coordinated, review the controller clock and drift.
2. US 15	■ Cyclic (3–5 years) signals retiming ■ Consider reallocation of split time to better serve mainline movements. ■ Use floating force-off if the corridor is coordinated-actuated. ■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”). ■ Consider increasing cycle length while managing queues. ■ Review ATSPMs (if applicable). ■ Use ATSPMs for signal retiming such as reallocation of split times and offset adjustments.	■ Cyclic (3–5 years) signals retiming ■ Consider reallocation of split time to better serve mainline movements. ■ Use floating force-off if the corridor is coordinated-actuated. ■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”). ■ Consider increasing cycle length while managing queues.	■ Consider reallocation of split time to better serve mainline movements. ■ Use floating force-off if the corridor is coordinated-actuated. ■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”). ■ Consider increasing cycle length while managing queues.
3. US 30	■ An adaptive signal system is recommended based on the expected traffic fluctuations and selected operational objectives. ■ Please refer to system requirements to make sure that the selected adaptive system can serve all movements efficiently.	■ Adaptive signal system is recommended based on traffic pattern, LOTTR, and operational objective. ■ Consider reallocation of split times to better serve mainline movements. ■ Use floating force-off if the corridor is coordinated-actuated. ■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”). ■ Consider increasing cycle length while managing queues. ■ Please refer to system requirements to make sure that the selected adaptive system can serve all movements efficiently.	■ Consider reallocation of split times to better serve mainline movements. ■ Use floating force-off if the corridor is coordinated-actuated. ■ Consider using lead-lag to increase mainline progression bandwidth (ensure to avoid “yellow trap”). ■ Consider increasing cycle length while managing queues.

Note: ATSPMS = automated traffic signal performance measures; LOTTR = level of travel time reliability.

Conclusions

In this study, a novel toolbox, Ts3T, is introduced that can assist agencies with the selection of the most appropriate signal systems solution based on traffic conditions, corridor characteristics, operational objectives, agency capabilities, and institutional constraints. Based on these factors, the toolbox offers a wide array of signal control strategies to agencies that can meet their needs. The toolbox was developed as a two-stage process where Stage I seeks basic information related to corridor characteristics and agency capabilities and limitations to provide an “early-exit” to a user when the solution is evident. If more than one feasible solution emerges from Stage I, Stage II performs a more detailed assessment of the corridor conditions and agency capabilities and limitations. The toolbox uses data readily available to transportation agencies (AADT, vehicle probe data) and is sensitive to inputs while identifying the most appropriate traffic signal systems solution.

The toolbox was applied to three case studies on corridors in the Commonwealth of Pennsylvania where various operational conditions were observed by state DOT representatives. In the first case study, the toolbox provided an “early-exit” from Stage I and suggested better utilization of existing infrastructure (e.g., revisit timing plans, review ATSPMs if available, check controller clock) as the corridor experienced a low level of congestion. Results from the second case study indicated that, while most factors favor the deployment of an ATCS, no travel time reliability issues were noted on the corridor (which is used as a proxy to demand fluctuations in this study). As a result, the toolbox recommended other solutions (e.g., consider reallocation of split time to better serve mainline movements, use floating force-off if the corridor is coordinated-actuated, review ATSPMs if applicable) to overcome existing issues. Lastly, results from the third case study showed that the toolbox suggested using ATCS on this corridor based on the operating factors. Overall, results from the case studies and sensitivity analysis to agency capabilities and resources demonstrated that the toolbox could provide the appropriate recommendations based on the unique corridor circumstances and agency capabilities. Additionally, the solutions recommended by the toolbox were generally in line with agency expectations based on their local knowledge and understanding of unique corridor conditions. Another strength of the developed toolbox is its ease of use for practitioners. While only three case studies were presented in this paper, the application of the toolbox to numerous case studies showed that it takes less than a few hours to apply the toolbox to a given study corridor.

The developed toolbox is intended to support the process of selection of appropriate traffic signal solutions and, generally, can be considered in addition to proven engineering principles and judgment. The authors of the study are aware that not all signal systems solutions and technologies within a particular system group (e.g., ADTS) have the same capabilities. Therefore, users of the tool are advised to conduct separate analyses of whether existing available solutions on the market can achieve their operational goals.

Footnotes

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: S. Gault, B. Cesme, S. Warchol, N. Dobrota; data collection: R. Rahman, N. Dobrota, B. Cesme, S. Warchol, N. Rouphail; analysis and interpretation of results: N. Dobrota, B. Cesme, S. Gault, N. Rouphail; draft manuscript preparation: N. Dobrota, B. Cesme, S. Gault, S. Warchol, N. Rouphail. All authors reviewed the results and approved the final version of the manuscript.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was in part supported by the Pennsylvania Department of Transportation.

ORCID iDs

Nemanja Dobrota

Burak Cesme

Shannon Warchol

Nagui Rouphail

References

Schrank

Lomax

Eisele

2019 Urban Mobility Report. Texas Transportation Institute. 2019. https://static.tti.tamu.edu/tti.tamu.edu/documents/mobility-report2019.pdf. Accessed July 15, 2023.

National Academies of Sciences, Engineering, and Medicine. Performance-Based Management of Traffic Signals. The National Academies Press, Washington, D.C., 2020. https://doi.org/10.17226/25875.

Abbas

Chaudhary

N. A.

Pesti

Sharma

Guidelines for Determination of Optimal Traffic Responsive Plan Selection Control Parameters. Technical No. Tx0-4421-2, Texas Transportation Institute, College Station, 2005. http://d2dtl5nnlpfr0r.cloudfront.net/tti.tamu.edu/documents/0-4421-2.pdf

Stevanovic

Martin

P. T.

Split-Cycle Offset Optimization Technique and Coordinated Actuated Traffic Control Evaluated Through Microsimulation. Transportation Research Record: Journal of the Transportation Research Board, 2008. 2080: 48–56.

Zhao

Tian

Applicability of Adaptive Traffic Control Systems in Nevada’s Urban Areas. Technical Report, Department of Transportation, NV, 2011.

Mudigonda

Ozbay

Doshi

Evaluation and Selection of Adaptive Traffic Control Strategies on Transportation Networks: Decision Support Tool Based on Geographic Information System. Transportation Research Record: Journal of the Transportation Research Board, 2008. 2064: 51–64.

Dobrota

Stevanovic

Mitrovic

Development of Assessment Tool and Overview of Adaptive Traffic Control Deployments in the US. Transportation Research Record: Journal of the Transportation Research Board, 2020. 2674: 464–480.

Mitrovic

Dobrota

Pereira

M. B.

Espino

Stevanovic

Data-Driven Decision Support Platform for Selection of Intersections for Adaptive Traffic Signal Control. Transportation Research Record: Journal of the Transportation Research Board, 2023. 2677: 567–581.

Fehon

Krueger

Peters

Denney

Olson

Curtis

Model Systems Engineering Documents for Adaptive Signal Control Technology Systems-Guidance Document. No. FHWA-HOP-11-027. FHWA, U.S. Department of Transportation, Washington, D.C., 2012.

10.

Pennsylvania Department of Transportation. Pennsylvania Adaptive Signal Control System Evaluation. 2012. https://www.dot.state.pa.us/public/PubsForms/Forms/TE-153.pdf. Accessed July 15, 2023.

11.

Brummond

Murillo

Ice

Model Systems Engineering Documents for Central Traffic Signal Systems. No. FHWA-HOP-19-019. U. S. Federal Highway Administration, Washington, D.C., 2018.

12.

National Academies of Sciences, Engineering, and Medicine. Adaptive Traffic Control Systems: Domestic and Foreign State of Practice. The National Academies Press, Washington, D.C., 2010. https://doi.org/10.17226/14364.

13.

Stevanovic

Kergaye

Stevanovic

Long-Term Benefits of Adaptive Traffic Control Under Varying Traffic Flows During Weekday Peak Hours. Transportation Research Record: Journal of the Transportation Research Board, 2012. 2311: 99–107.

14.

Stevanovic

Zlatkovic

Evaluation of InSync Adaptive Traffic Signal Control in Microsimulation Environment. Presented at 92nd Annual Meeting of the Transportation Research Board, Washington, D.C., 2013.

15.

Wang

Corey

Lao

Henrickson

Xin

Criteria for the Selection and Application of Advanced Traffic Signal Systems. Technical Report. Oregon Department of Transportation, 2013.

16.

Ban

Wojtowicz

J. M.

Decision-Making Tool for Applying Adaptive Traffic Control Systems. No. C-13-04. New York State Energy Research and Development Authority, 2016.

17.

Mladenovic

Stevanovic

Kosonen

Glavic

Adaptive Traffic Control Systems: Guidelines for Development of Functional Requirements. Proc., International Scientific Conference on Mobility and Transport (mobil. TUM), Munich, Germany, 2015, TUM, p. 12.

18.

Balke

K. N.

Sunkari

S. R.

Evaluation Procedures for Coordination Features in Traffic Signal Controllers Using Hardware-in-the-Loop Simulation. Presented at 85th Annual Meeting of the Transportation Research Board, Washington, D.C., 2006.

19.

UDOT ATSPM. Utah Department of Transportation, Automated Traffic Signal Performance Measures. 2012. https://udottraffic.utah.gov/atspm/. Accessed July 29, 2023.

20.

Kim

J.-T.

Courage

K. G.

Evaluation and Design of Maximum Green Time Settings for Traffic-Actuated Control. Transportation Research Record: Journal of the Transportation Research Board, 2003. 1852: 246–255.

21.

Yun

Best

Park

B. (Brian)

Evaluation of Adaptive Maximum Feature in Actuated Traffic Controller. Transportation Research Record: Journal of the Transportation Research Board, 2007. 2035: 134–140.

22.

Furth

P. G.

Cesme

Muller

T. H. J.

Lost Time and Cycle Length for Actuated Traffic Signal. Transportation Research Record: Journal of the Transportation Research Board, 2009. 2128: 152–160.

23.

Day

C. M.

Bullock

D. M.

Design Guidelines and Conditions That Warrant Deployment of Fully Actuated Coordination. Transportation Research Record: Journal of the Transportation Research Board, 2014. 2439: 1–11.

24.

Zhang

Jiang

Chen

Zhang

Zhou

Wang

Miao

Optimizing Unit Extension for Fully Actuated Traffic Signal Controls in Intelligent Transportation Systems. 2018. –https://ascelibrary.org/doi/10.1061/9780784481523.148.

25.

Waddell

J. M.

Remias

S. M.

Kirsch

J. N.

Young

S. E.

Scalable and Actionable Performance Measures for Traffic Signal Systems using Probe Vehicle Trajectory Data. Transportation Research Record: Journal of the Transportation Research Board, 2020. 2674: 304–316.

26.

Zheng

Liu

H. X.

Misgen

Performance Diagnosis Tool for Arterial Traffic Signals. Transportation Research Record: Journal of the Transportation Research Board, 2013. 2356: 109–116.

27.

Xia

Shaon

M. R. R.

Y. J.

Jiang

Arterial Signal Offset Optimization Using Crowdsourced Speed Data. Transportation Research Record: Journal of the Transportation Research Board, 2022. 2677: 1633–1642.

28.

PennDOT. Traffic Data Report. Pennsylvania Department of Transportation. 2022. https://www.penndot.pa.gov/ProjectAndPrograms/Planning/TrafficInformation/Pages/2022-Pennsylvania-Traffic-Data.aspx. Accessed July 15, 2023.

29.

Iroanya

R. O.

An Evaluation of Turning Movement Counts and Estimation of Initial Turning Proportions. Doctoral dissertation. Texas A&M University, College Station, 2020.

30.

Denney

R. W.

Jr. Curtis

Head

Long Green Times and Cycles at Congested Traffic Signals. Transportation Research Record: Journal of the Transportation Research Board, 2009. 2128: 1–10.

31.

The Impact of Pedestrian Activities in Adaptive Traffic Signal Control System Operations. Doctoral dissertation, University of Pittsburgh, 2014.

32.

Denney

R. W.

Hamilton

B. A.

Improving Traffic Signal Management and Operations: A Basic Service Model. No. FHWA-HOP-09-055. U.S. Federal Highway Administration, Washington, D.C., 2009.

33.

Abbas

M. M.

Abdelaziz

Evaluation of Traffic Responsive Control on the Reston Parkway Arterial Network. Virginia Center for Transportation Innovation and Research, Blacksburg, 2009.

34.

Margiotta

R. A.

Turner

Taylor

Chang

National Performance Measures for Congestion, Reliability, and Freight, and CMAQ Traffic Congestion: General Guidance and Step-by-Step Metric Calculation Procedures. No. FHWA-HIF-18-040. U.S. Federal Highway Administration, Washington, D.C., 2018.

35.

Regional Integrated Transportation Information System (RITIS). A Data-Driven Platform for Transportation Analysis, Monitoring, and Data Visualization. The University of Maryland. 2023. https://ritis.org/login?r=Lw==. Accessed July 29, 2023.