Traffic Signal Retiming Practices in the United States (2010)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

SUMMARY TRAFFIC SIGNAL RETIMING PRACTICES IN THE UNITED STATES Traffic signals that are not timed to coordinate efficiently with vehicular traffic can cause travel delays, increased accident rates, increased pollution from vehicle emissions, and increased fuel consumption, among other concerns. Although many studies have shown that retiming traffic signals is a cost-effective expenditure of transportation agency resources, few agencies have developed regular programs to carry out the retiming process. The Urban Mobility Report and National Traffic Signal Report Card publicize the need for and public benefit of traffic signal retiming: these reports indicate that almost half of the transportation agencies surveyed (43%) do not regularly collect and analyze traffic data for signal timing, and many existing traffic data collection programs do not assess the quality of data collected. As a result, even agencies that do make an effort to compile traffic signal information may be using faulty data to analyze and time their traffic signals. Transportation agencies may need to explore new approaches to the signal retiming process in order to improve the quantity and quality of the traffic signal data collected and to streamline the use of new and existing resources. These new approaches may include cooperative action with other regional agencies, often with the assistance or leadership of a metropolitan planning organization. Interagency cooperation may allow transit agencies to assign more resources for signal timing or to improve the use of existing resources. This synthesis reports on the practices that operating agencies currently use to revise traffic signal timing. It includes the planning needed to develop signal timing plans and the processes used to develop, install, verify, fine-tune, and evaluate the plans. The authors col- lected information for this synthesis through a literature review, a review of two large-scale and two narrowly focused surveys of transit agencies, and a series of project case stud- ies. For the case studies, the authors prepared an in-depth questionnaire to solicit detailed information not addressed in the prior survey. Of the 17 agencies solicited for the case studies, the authors followed up with the 7 agencies that responded and were able to acquire additional statistical and anecdotal information. The practices covered by this synthesis include the following: General signal control issues such as selecting intersections for coordination and • determining different classes of intersection users. Agencies generally emphasize vehicle and pedestrian safety, as well as minimizing delay. Data collection requirements and intersection analysis, including phasing, retiming • tools, coordination, and safety issues. Most agencies use standard retiming software to prepare timing plans, but the traffic movement data used in the software are usu- ally collected manually, which decreases efficiency and increases the costs of the sig- nal retiming process. Closed-circuit television is more frequently being used to assist in fine tuning, monitoring traffic, and determining timing needs, but other process improvements are required. Policy, management, and planning practices. Most agencies exceed the advised signal • retiming interval of 30 to 36 months, largely because of limited resources. Agencies generally employ from three to seven daily timing plans, with two to three separate

2 timing plans for weekends, holidays, and special events. However, existing methodolo- gies for identifying the number of timing plans needed and their periods of use are in need of improvement. Signal timing performance evaluation methods, such as travel time measurements, • observations at intersections, crash records analysis, and public feedback. Most agen- cies track performance measures through delays, stops, route travel time, accident rates, and emissions and fuel consumption. Simulation and traffic detectors are used to a lesser extent for evaluation purposes. Transit agencies may be able to use emerging technologies to reduce the labor required to conduct field evaluations. Signal coordination across agency boundaries. Major barriers to signal coordina-• tion and shared control include technical barriers, and institutional legal and liability issues. The literature and survey review and the case study evaluation produced several notable findings on the current state of traffic signal retiming. First, resource limitations are the most significant factor contributing to suboptimal signal retiming, even though many studies have shown signal timing to be a cost-effective expenditure of resources (with an average cost of $3,700 per signal and 26 person-hours of work for most agencies). Second, although current guidance states that traffic signals should be reviewed and retimed at intervals of 30 months to 3 years, most agencies exceed this interval but generally retime their signals within a 5-year period. Third, clearer and more detailed guidance is needed for several aspects of the signal retiming process, including coordination boundaries for traffic signal networks, interactions between controller timing plan parameters, and methods to establish the opti- mal number of timing plans to employ and the periods for use. Finally, further research to develop detailed design requirements for archived data user services and management sys- tems for signal timing would reduce the cost of evaluating existing signal timing plans, and would also improve their usefulness in helping to determine the need for retiming.

3 Signal Association (IMSA), the Intelligent Transportation Society of America (ITS America), U.S.DOT–FHWA, and many other organizations. The results in the report card are based on the 2007 Traffic Signal Operation Self Assessment released by NTOC in the fall of 2006. A total of 417 agencies responded, representing 47 states. Each agency self-assessed progress in areas of traffic signal management, signal opera- tions at individual intersections, signal operations in coor- dinated systems, signal timing practices, traffic monitoring and data collection, and traffic signal maintenance. Puget Sound Regional ITS Implementation Plan• (Puget Sound Regional Council et al. 2008) This report, sponsored by the Puget Sound Regional Council Traffic Operations Committee, provides leadership for regional traffic operations initiatives. It identifies ITS improvements for key multijurisdictional arterial corridors and works to provide capital to support better coordination among transportation agencies and create a more seamless transportation network. Data input for this plan includes a survey of 19 transportation agencies’ practices regarding traffic management and signal timing. Traffic Signal Operations and Maintenance Staffing • Guidelines (Gordon and Braud 2009) This report provides guidelines to estimate the staff- ing and resource needs required to effectively operate and maintain traffic signal systems. The research effort included a literature review and in-depth survey responses from 7 agencies of 34, including cities, counties, and state DOTs. Signal Timing Practices and Procedures• (Tarnoff and Ordonez 2004) This manual, published and distributed by the FHWA and ITE, documents the state of practice in traffic signal tim- ing. Results include the need for a comprehensive guide and are based on information obtained through a survey of more than 100 state, city, and county agencies responsible for traf- fic signal operations. A set of in-depth case studies was conducted under this project. The case study agencies were selected for in-depth study because each represents a different experience, including CHAPTER ONE INTRODUCTION AND RESEARCH METHODOLOGIES PURPOSE OF SYNTHESIS Traffic delays are widely reported to result in very large costs to the public in terms of traveler delay, excess fuel con- sumption, and excess emissions (Schrank and Lomax 2007). Many studies (e.g., Sunkari 2004; Benefits of Retiming Traf- fic Signals 2005) have shown that traffic signal retiming can significantly reduce these costs. This synthesis presents current traffic signal retiming practices as provided in the literature and describes the results of in-depth case studies conducted by this project. It addresses the traffic signal retiming process in the context of traffic signal design, signal timing objectives, requirements, and practices. It also discusses strengths and weaknesses in the signal retiming process. Although the synthesis focuses on signal retiming, it also covers the larger issue of the rela- tionship of signal timing to the management practices of agencies responsible for the signal timing process. STUDY METHODOLOGY The current practices described in the study were obtained from reviews of the literature, from several prior surveys, and from in-depth case studies conducted under this project. The following documents are available to provide com- prehensive guidance to traffic engineers on different aspects of signal timing practices: Signal Timing on a Shoestring• (Henry 2005) Traffic Control Devices Handbook 2001• (Pline 2001) Traffic Signal Devices Handbook• (Bonneson et al. 2009) Traffic Signal Timing Manual• (Koonce et al. 2008) The synthesis also employed the following recent surveys: National Traffic Signal Report Card• (National Transportation Operations Coalition 2007) The National Transportation Operations Coalition (NTOC) is an organization comprising transportation experts including ITE, AASHTO, the American Public Works Association (APWA), the International Municipal

4 ORGANIzATION OF SYNTHESIS The results of the surveys and in-depth case studies are included in chapters two through eleven. Chapter two discusses signal timing policy, management, and planning by responsible agencies, including objectives and policies. Chapter three describes timing requirements and strate- gies at the network level. The chapter covers signal retim- ing issues such as network boundaries, retiming priorities, retiming frequency, the number of timing plans, and the periods for which they are employed. Chapter four describes such retiming considerations as data collection, phasing, safety, detector placement, and sig- nal retiming tools and simulation. Chapter five covers the selection of actuated timing peri- ods, relationships among phase intervals, pedestrian issues, and the frequency of signal retiming. a range of city sizes with a range of number of traffic signals, types, and operational practices. As such, the in-depth case studies are not intended to be representative of the population. Each case study was initiated with a detailed questionnaire so that their signal timing practices could be examined in depth. Appendix A provides the questionnaire, summarizes the responses, and describes the in-depth case study meth- odology. Agencies were selected for participation based on diverse geographical locations, and agency type and size. Of the 17 agencies solicited, 7 provided detailed responses to the questionnaire. Where responses appeared to have the potential for unique information, they were followed up with additional questions and discussion. In some cases agencies provided documentation that was directly incorporated into the synthesis. Table 1 identifies key characteristics of the in- depth case study participants. The agencies contributing to the in-depth case studies are generally larger and have a greater level of resources than the average agency responding to the National Traffic Signal Report Card survey, resulting in generally better manage- ment practices. TABLE 1 KEY CHARACTERISTICS OF CASE STUDY RESPONDENTS Designator Type of Agency Approximate Number of Signals Location Approximate Population Signal System Type Employed Resources Per Intersection Appropriate for Retiming1 CL C A TB a County 500, 75% coordinated Florida 1 million √ √ √ √ 16 eng hours, 40 tech hours, $5,000 if contracted b City 100, 97% coordinated Illinois 150,000 √ √ 3 eng hours, 14 tech hours, $2,500 if contracted c State DOT (geographi- cal district) 700, 50% coordinated Northeast 1.9 million √ d County 700, 98% coordinated Michigan 1.2 million √ √ √ 3 eng hours, 0.5 tech hour, $4,200 if contracted e City 150, 98% coordinated New York 57,000 √ √ 8 eng hours, 7 tech hours, $2,000 if contracted f City 850, 91% coordinated Texas 660,000 √ $750 per intersection if contracted g State DOT (geographi- cal district) 3,000, 75% coordinated Midwest Not applicable √ √ √ $7,500 per intersection if contracted CL = closed loop system , C = central control system, A = adaptive control system, TB = time base coordination, eng = engineer, tech = technician. 1 The wide variation in contracting cost results, in part, from agencies contracting for some services and performing other services with their own forces. Figure 13 in chapter seven describes the apportionment of contracted costs among signal timing subtasks.

5 Chapter six discusses the installation of signal timing plans into field controllers, as well as verification and fine- tuning practices. Chapter seven describes agency staffing levels. It also covers the person hours and dollar costs for various tasks associated with retiming as determined by the review of lit- erature, the surveys, and in-depth case studies. Chapter eight discusses performance measures described in the literature and those used by agencies. Chapter nine presents practices for signal timing per- formance evaluation techniques used by agencies. It also describes emerging labor saving technology for obtaining such measures as delay. Other measures covered include safety, fuel consumption, emissions, and cost-benefit analysis. Chapter ten provides practices from literature, the surveys, and in-depth case studies on barriers to imple- mentation of signal timing plans, including resource limitations and institutional barriers such as signal coor- dination across agency boundaries and sharing of infor- mation devices. The limitations of available guidance and tools for signal timing as well as competing requirements are described. Chapter eleven describes survey results, in-depth case studies, and literature reviews pertaining to methods to increase the resources for signal timing. Chapter twelve provides conclusions and approaches for further research. The survey and in-depth case study results are included in chapters two through eleven. The references and a glossary are provided. Appendix A describes the in-depth case studies. Appendices B through E further discuss and provide examples of the material in the synthesis.

6 CHAPTER TWO SIGNAL TIMING POLICY, MANAGEMENT, AND PLANNING “Development of signal timing policies should be a collabor- ative effort between regional partners and community stake- holders, crossing jurisdictional boundaries, with the service and safety of the customer in mind at all times. Signal tim- ing policies should be clearly documented and thoroughly communicated within an agency to those who operate and maintain the signal system” (Koonce 2008). Figure 1 illus- trates the relationship of policy to the signal timing process. Operations and maintenance, along with planning, are key components of the management process. FIGURE 1 Relationship of policy to signal timing process. (Source: Koonce 2008.) This chapter describes the implementation of policies. REVIEW OF LITERATURE Planning Processes and Management Controls These activities may include: Concept of operation. • Mission statement.• Operations procedures.• Compliance with Regional ITS Architecture and • Standards. Plans for capital and operational upgrades.• Objectives, Policies, and Competition Among Objectives Policies and objectives, along with constraints, form the cornerstones of signal timing practices. In addition to defin- ing the objective, it is desirable for planning documentation to also describe one or more methodologies to achieve the objective, as well as performance measures and techniques to be used by the agency to obtain data to generate the mea- sures. Table 2 illustrates the relationships among traffic sig- nal timing processes, functions, measures of effectiveness, and possible measurement techniques. Where possible, it is desirable to include quantitative values when stating objec- tives. These values should be high enough to provide mean- ingful benefits and yet be feasible with the system features and functions provided. Objectives may conflict. For exam- ple, if transit priority is provided on an arterial, the objective of minimizing traveler delay may conflict with the objective of avoiding congestion and spillback (on the major cross- ing street). Resolution might require a policy that limits the aggressiveness of the priority provided. Policies and guide- lines describe the agency’s approach, and the strategies to be employed in signal timing and retiming relative to the priori- tization of objectives and to resolve conflicting objectives. Gordon and Braud (2009) indicate that it is appropriate for signal timing management activities and policies to empha- size outcomes. Attainment of satisfactory outcomes is accom- plished by management practices that include the following:

7 TABLE 2 EXAMPLES OF OBJECTIVES AND MEASURES FOR TRAFFIC SIGNAL SYSTEMS Objective Examples Mechanism to Achieve Objective Possible Measure Possible Measurement Technique Mobility, Fuel Consumption, and Emissions 1. Reduce delay and fuel con- sumption for normal traffic pat- terns. Signal retiming may improve delay and fuel con- sumption by a specific mini- mum amount each time retim- ing is performed. Improved signal timing.a. Improved level of traffic b. system control (e.g., traffic responsive, traffic adaptive). Real-time adjustment of c. timing by operator. Improved maintenance d. response time. Vehicle hours delay.a. Gallons fuel reduced.b. Monitoring and tracking of c. citizen complaints (provide regional 311 or equivalent phone number for reporting). Travel time and delay runs.a. Traffic system data.b. Simulation independent of c. signal timing programs. Use of traffic system data for input. Real-time performance d. monitoring. 2. Reduce delay and fuel con- sumption for incident condi- tions and special events. New signal timing plans to support these functions may improve delay and fuel consumption. Signal timing—items a, b, c a. in 1 above. Same as 1. Simulation. Support of incident b. management using CCTV and other information. Same as 1. Simulation with reduction in incident clearance time. 3. Reduce emissions. Same as 1. kg of CO, NOx, SO2, CO2, VOC, particulate matter. Derive from gallons saved. Safety 4. Reduce crashes resulting from signal outages. Improved maintenance response time. Crashes reduced. Analysis inference techniques using response time mainte- nance records. 5. Reduce secondary crashes resulting from incidents. Support of incident manage- ment using CCTV and other information. Secondary crashes reduced. Analysis inference techniques using reduction in incident clearance time. Communication with Public and Public Perception of Service 6. Improved public perception of signal and management cen- ter operations. Achieve other objectives.• Regular reporting to public.• Monitoring and tracking of • citizens’ complaints (provide regional 311 or equivalent phone number for reporting). Develop and provide outreach • material describing how traffic signals function and the benefits of active operations. Develop website to • disseminate information and reports, provide an online feedback and complaint database. Number of calls, complaints. Monitor number of calls, time to respond to calls, and outcome of complaints received. 7. Provide traffic information to public and private traffic infor- mation services. Make traffic, construction, spe- cial event, incident, weather data, and CCTV signals avail- able to traffic services, media, websites. Rating scale. Survey. Ancillary Functions 8. Serve as a diversion route for corridor operations. Goal established by stakeholders. Items b and c in 1 above plus availability of diversion timing plans. Same as 1 and 3. Corridor simulations. 9. Provide preemption for emer- gency vehicles and railroads. Goal established by stakeholders. Preemption equipment. It is appropriate to restrict the use of equipment so as not to cause unreasonable delay to general traffic. Number of critical mission emergency vehicles provided preemption. Time saved by emergency vehicles. Some preemption systems pro- vide logs of preemptions granted. TABLE 2 continues on next page

8 quently than less traveled locations. Depending on the policies and measures selected, agencies may provide more favorable green splits and traffic progressions to major arterials. Optimization emphasis—Competing policy objec-• tives may include maximizing bandwidth, emphasiz- ing safety, minimizing delays and stops, and control of queues under saturated conditions. Emphasis may be a function of the type of network (grid, arterial, iso- lated), street classification, or other factors. Area access deterrence—Agency policies may require • the avoidance of traffic assignments to certain areas such as residential areas. These are sometimes known as neighborhood traffic management programs. Signal timing strategies may play a role in these programs. Koonce (2008) indicates that it is appropriate to review signal retiming every 3 to 5 years and more often if there are significant changes in traffic volumes or roadway conditions. It is appropriate for objective-oriented operation (Gordon and Braud 2009) to drive the frequency of signal retiming and consider the level of resources that are available to main- tain target performance levels. A number of studies are iden- tified that indicate that it is appropriate that, at a minimum, signal timing be reviewed at least every 30 to 36 months. The review is a key element to implement objective-oriented operation as described in that document. Most of the litera- ture on previous surveys shows that the majority of the oper- ating agencies do not conform to this guideline. Koonce (2008) identified the following signal timing issues: Will all types of users (transit, freight, emergency • respondents, pedestrians, vehicles, bicycles, etc.) be treated equally or prioritized at the signalized intersection? Objective Examples Mechanism to Achieve Objective Possible Measure Possible Measurement Technique 10. Provide transit priority. Goal established by stakeholders. Priority equipment used for late vehicles. Traveler hours reduced. Varia- tion in passenger time of arrival at designated stops reduced. Transit records coupled with simulation with delay and delay variation criteria. 11. Support emergency evacua- tions. Goal established by stakeholders. Evacuation signal timing, phas- ing, and lane use plans. Availability of plans. Management 12. Projection of historical data for capital programming of future projects. Mining of detector data. Source: Adapted from Gordon and Braud (2009). CCTV = closed circuit television; CO = carbon monoxide; NOx = nitrogen oxide; SO2 = sulfur dioxide; CO2 = carbon dioxide; VOC = volatile organic compound. Assurance of qualified personnel in charge of signal • timing. Availability and updating of a mission statement and • reviews to determine how well the agency accom- plishes these functions. Annual review by management of strategic manage-• ment plans, signal timing performance, traffic system reliability, and improvement in evaluation measures. Appendix B provides an example of the signal timing policies, guidelines, and strategies that result in desired outcomes. In reality, all traffic signal control is policy based. The policy may be to minimize vehicle delay or vehicle stops. Policy-based traffic control can be better understood by using a hierarchy of priorities (Urbanik 2000). The hierar- chy structure is, in effect, a policy developed by the agency. These policies may be determined by the agency or may be influenced by other stakeholders. Signal timing policies may relate to the following: Timing parameters—Examples include maximum • cycle length and clearance time policies. Transit—If transit priority is to be provided, policies • to determine the specific intersections qualifying for priority and constraints on the priority (e.g., maximum green extensions, volume-to-capacity ratio limitations on priority grants) are important. Preemption—Aside from rail preemption, policies to • establish the emergency vehicle classes qualifying for preemption are appropriate. Pedestrian—Policies to establish walk speed and • exclusive pedestrian intervals and phases at appropri- ate locations are appropriate. Street classification—Retiming priorities may vary by • street classification (as defined by the governing local- ity). Many agencies retime major arterials more fre- TABLE 2 continued

9 How frequently will signal timing plans be reviewed • and updated? How will approaches with differing street classifica-• tions be treated? Will there be preferential treatment for certain move-• ments beyond the definition of the coordinated phase (will the coordination timing plan clear all queues dur- ing each cycle for left turns and side street through movements)? How will intersections with deficient capacity be • treated? What measures will be used to determine whether the • timing plan is effective (vehicle stops, network delay, arterial travel speed, estimated person delay, estimated fuel consumption, transit speed, etc.) and how will they be collected? Agencies may identify other policy issues that affect mobility and safety such as speed limit changes. SURVEYS Figure 2 describes the emphasis placed on signal timing approaches and practices by the agencies responding to the Puget Sound survey. Key items shown on the left of the fig- ure include establishing effective green-band progressions, and minimizing delay and corridor throughput. FIGURE 2 Signal timing approaches and practices in the Puget Sound area. (Source: Puget Sound Regional ITS Implementation Plan 2008.) Gordon and Braud (2009) advise an annual review by management of design, operations, maintenance, and train- ing. Approximately one-third of the responders to that proj- ect’s survey indicated that they did conduct annual reviews. The report also indicates the generally poor availability of a concept of operations, mission statement, and operational procedures. The 2007 National Traffic Signal Report Card indicates that: Almost half of the agencies (43%) reported having • few or no regular, ongoing programs for collecting and analyzing traffic data for signal timing. (Because the respondents to this survey included a large number of smaller agencies, this figure may be more representa- tive of this class of agencies.) Half of the agencies do not assess the quality of data • collected. As a result, agencies may use faulty data to analyze and time their traffic signals. The overall agency score in the management area is • “D.” IN-DEPTH CASE STUDIES Four of the seven in-depth case study agencies respond- ing to the information query are in conformance with the Regional ITS architecture and its standards, and four of the seven have documented plans for capital and opera- tional upgrades. Four of the seven also indicated that they had prepared documented objectives. The most frequently reported objectives in the in-depth case study survey include maintenance of safety, maintenance of progression on selected arterials, and delay. Three of the seven in-depth case study respondents con- ducted reviews of signal timing performance and retiming of signals if necessary within a period of 3 years. Three agen- cies performed these operations every 3 to 5 years, and one agency exceeded 5 years. Signal timing by its nature is the assignment of right-of- way to competing directions and travel modes. Signal timing often requires trade-offs between various modes at an inter- section, such as vehicles versus pedestrians and bicycles. These tradeoffs typically result in competing priorities, such as safe pedestrian crossing times versus maximizing auto- mobile capacity. The most important conflicts reported are: Balancing all users’ needs;• Cross-jurisdictional coordination issues;• Pedestrian issues; and• Disruptions caused by railroad and emergency vehicle • preemption. Less frequently reported conflicts include coordination requirements where major arterials cross, transit priority, and the use of pedestrian push buttons by bicyclists. All of the in-depth case study respondents indicated that they conducted performance evaluations for signal timing impacts on vehicle flow. Table 3 shows the frequency of use of common measures.

10 Reduced v/c ratio—Modified volume-to-capacity ratio • relative to the base capacity caused by queue spillback (queues that fill a short link) and starvation (green time that is not used to service vehicles because of imperfect coordination with the upstream intersection). STATE OF THE PRACTICE The surveys and in-depth case studies indicated that signal timing agencies have the following concerns: Many agencies do not review field performance data • to determine the adequacy of signal timing at intervals of 30 to 36 months, as advised for objective-oriented operation. Many agencies do not review design, operations, main-• tenance, and training practices annually as advised for objective-oriented operation. Many agencies do not have precise and clearly stated • policies that result from objectives at the level of detail shown in Appendix B. As more fully discussed in chapter ten, the Puget Sound Survey (Puget Sound Regional Council et al. 2008; Gordon and Braud 2009) and anecdotal data in the in-depth case studies performed under this project indicate that the combi- nation of resource limitations and gaps in effective manage- ment are the most significant factors contributing to less than effective signal timing practices. Delay, stops, accidents, and fuel consumption mea- sures are commonly used for cost versus benefit analysis. Recently, some signal timing programs have incorporated measures for congested conditions characterized by spill- back or spillover across intersections and from turning bays. One program reported using the following measures: Queue delay—Effect of queues and blocking on short • links and turning bays. TABLE 3 AGENCY USE OF EVALUATION MEASURES FROM CASE STUDIES Measure Agencies Using Measure, n = 7 Delay 6 Stops 6 Route Travel Time 6 Accidents 5 Emissions and Fuel 4 Reliability of Travel Time 2 Spillback capacity reduction—Reduction to the base • capacity caused by a short downstream link becoming filled up. Base capacity is unimpeded capacity. Storage capacity reduction—Reduction to the base • capacity caused when turn pockets cannot accommo- date queue lengths.

11 CHAPTER THREE NETWORk TREATMENTS REVIEW OF LITERATURE Relationship of Intersection Operation to Network Coordination Traffic signals may be operated as pretimed or actuated sig- nals. When using signals in networks there is a relationship between the way the signals are operated and the coordina- tion needs of the network. The FHWA Traffic Signal Timing Manual (Koonce et al. 2008) summarizes the types of inter- sections that may be employed, as shown in Table 4. Semi-actuated signals may be isolated or coordinated, whereas fully actuated signals usually operate as isolated signals. Day et al. (2008) showed that the use of fully actu- ated signals in a coordinated system may reduce delay by terminating the coordinated phase early during cycles when no continued demand is present. Network Functional Architectures Four levels of traffic control systems as defined by Gordon (2003) are discussed in this section. [This description builds Chapter two discussed issues at the agency level, including objectives, performance measures (covered in more detail in chapter eight), and policies. This chapter discusses signal timing issues as they apply to the control of traf- fic networks, including isolated intersections within the network. Networks (sometimes called sections) are composed of signals that are coordinated as a group. The basic coordina- tion relationship is the change of timing plans at the same time among signals in the network. Coordination generally emphasizes the use of a common cycle length or a submul- tiple of the cycle length (double cycling) and establishment of the offset relationship among intersections to facilitate progression of vehicles through a series of signals. Where no such relationships exist, the signal is said to be isolated. Isolated signals may be present within the physical bound- aries of a network. TABLE 4 RELATIONSHIP BETWEEN INTERSECTION OPERATION AND CONTROL TYPE Pretimed Actuated Type of Operation Isolated Coordinated Semi-actuated Fully Actuated Coordinated Fixed Cycle Length Yes Yes No No Yes Conditions Where Applicable Where detection is not available Where traffic is con- sistent, closely spaced intersections, and where cross street traffic is consistent Where defaulting to one movement is desirable, major road is posted <40 mph, and cross road carries light traffic demand Where detection is provided on all approaches, isolated locations where posted speed is >40 mph Arterial where traf- fic is heavy and adjacent intersec- tions are nearby Example Application Work zones Central business dis- tricts, interchanges Highway operations Locations without nearby signals; rural, high speed locations; intersec- tion of two arterials Suburban arterial Key Benefit Temporary applica- tion keeps signals operational Predictable opera- tions, lowest cost of equipment and maintenance Lower cost for high- way maintenance Responsive to changing traffic pat- terns, efficient allo- cation of green time, reduced delay, and improved safety Lower arterial delay, potential reduction in delay for the system, depending on the settings Source: Koonce et al. (2008).

12 The in-depth case studies indicate that most U.S. traffic signal systems operate in a time-of-day mode. Typically most systems utilize three to five weekday timing plans and up to six plans for other needs such as weekends, holidays, special events, incident responses, construction, and weather. These systems utilize communications between the field controller and either the TMC or a field master controller (that may, in turn, communicate with the TMC). Thus controller status information (e.g., phase status, failure status, detector infor- mation, and, in some cases, reports developed by the field controller) are available at the TMC. Systems may be programmed to alter phasing and phas- ing sequences as timing plans change, or to provide isolated intersection operation for some intersections for some peri- ods (typically under saturated flow conditions or in some cases for late evening or early morning operation). Level 2 systems may control signal timing in either of two ways: Provision of direct timing signals from the control • computer at the TMC for each controller phase. Backup timing plans are stored in the controller in the event of loss of communications. Systems of this type are sometimes called centrally controlled systems. Signal timing plans may be downloaded and stored • in the intersection controller. In some cases, the plans may be provided directly to the local intersec- tion controller from the TMC, and in other cases the intersection controller receives its timing plans from a field master controller. The field master controller may receive plans from the TMC, or these plans may be directly programmed into the field master control- ler. When control can be accomplished from the TMC and data feedback is available at that location, these systems are often called closed loop systems. Most of the current systems operating in the United States are of this variety. Level 3—Traffic-Responsive Control Level 3 systems incorporate a traffic-responsive control strategy. These strategies use system detectors to select a plan from a timing plan library that is most appropriate to the current traffic conditions. With an appropriate comple- ment of traffic detectors in place, both centrally controlled systems and closed loop systems are capable of providing traffic-responsive operation. Systems currently deployed in the United States generally incorporate one of the following classes of strategies. Signature Matching Strategy This strategy develops and stores a “signature” set for each timing plan to be used. The current detector data are matched to the library of signature on the discussion in Gordon and Tighe (2005)]. Systems operated by most U.S. agencies can operate at Levels 1, 2, and 3, as described here. Operation at Level 1 requires no equipment to establish offset relationships between local intersections. Level 2 requires wireline or wireless commu- nication with the local intersection, and Level 3 requires a complement of system detectors in addition to the Level 2 requirements. Level 1—Time Base Coordination Modern intersection controllers that are not physically interconnected or interconnected by wireless means have the capability to support coordinated signal timing plans through time base coordination. Progressions are achieved by relating the cycle times and offsets to a common time standard. Although time base coordination (TBC) entails a relatively low capital cost and is commonly used for backup of interconnected systems when the central computer or communication fails, it is appropriate for the system designer to consider the following limitations before selecting this approach for primary control. Equipment status is not provided. Consequently, equip-• ment failure or failure to display the appropriate signal timing cannot be automatically detected at the traf- fic management center (TMC) or in the maintenance facility. Although detector data may be recorded at the control-• ler, it is not available at the TMC in real time. Timing plans cannot be selected by the TMC. Selection • of a timing plan that was not previously scheduled requires a visit to the intersection. Section-wide, traffic-responsive operation cannot be • achieved. TBC requires an accurate common time reference • among the controllers in the network. Depending on the technology used to provide this reference, frequent power interruptions may significantly disrupt coor- dination during and after the interruption. Although provisions for automatically resetting the controller’s clock by means of global positioning system (GPS)- or radio-provided information are optionally available, most users report that it is best to perform periodic checks of the controller’s clock reference. Gerken (2008) indicates that under TBC, slight drift in the clock can produce a 23% increase in delay and signifi- cant drift produces a 34% increase in delay. Level 2—Time-of-Day and Manual Control of Timing Plans Level 2 systems are interconnected by means of wireline or wireless communications to a central facility. They provide signal control by means of time-of-day selection of timing plans or selection of timing plans by the operator.

13 between thresholds. Split and offset thresholds are similarly established. In some cases, traffic features such as directionality or • queue presence may be used in the selection of cycle, split, and offset. System detectors may be assigned to compute these features. Provisions are often made for the constraint of cycle, • split, and offset selections by time of day or by some other means so that the entire timing plan conforms to a plan developed by a signal timing program. Traffic Responsive Performance Abbas et al. (2008) indi- cated that considerable improvement may be obtained under certain conditions with traffic responsive operation. Nelson et al. (2000) described a simulation study comparing this type of traffic responsive algorithm (TRP) with time-of-day operation (TOD). Their conclusions are as follows: Where traffic patterns are highly repetitive, TRP oper-• ation lags TOD owing to the time it takes to recognize the need for a new timing plan. Where traffic volumes fluctuate substantially, TRP has • the potential of improving operation over TOD. Performance penalties resulting from timing plan • changes do not contribute significantly to TRP disadvantages. The greatest problem with TRP appears to be the occa-• sional selection of an improper timing plan. System detectors for Level 3 operation are located suffi- ciently upstream of the stop line so that queues during most time periods do not back up over the detector for most of the traffic cycle. Detectors are most commonly located at all approaches to major intersections in a traffic section. Often, a single detector is located in the critical lane for an approach. Detectors may also be provided in left turn bays. Additional detectors may be provided on the major arterial at some minor intersections. Level 4—Traffic Adaptive Control Detectors at local intersections are commonly used to adjust the length of traffic phases in response to vehicles approach- ing the intersection. This type of operation is termed local traffic actuated operation and is described in chapter five. Traffic adaptive control extends the concept of rapid response to traffic networks by using flow data from detectors consid- erably upstream of the intersection to predict arrivals at the intersection. Two subclasses of adaptive control for signals in networks are conventional adaptive control, and a variation that uti- lizes fewer detectors and is easily adaptable to conventional closed loop systems. Koonce et al. (2008) provide the follow- ing description in the Traffic Signal Timing Manual: sets, and the timing plan with the closest match to the cur- rently measured signature set is selected. A “signature” consists of the following representation for each system detector selected for the computation: VPLUSKO = V • K • φ Where: VPLUSKO = signature for that detector V = Volume K = Constant φ = Occupancy During operation, the signature sets developed by the detector data are matched to a subset of the stored signature sets that are “allowable” for that time period. If the closest match differs from the signature set that corresponds to the timing plan currently in use, then the system transitions to the timing plan corresponding to the allowable signature with the best match. The algorithm is designed to avoid back and forth transi- tioning between timing plans that are close to the selection boundary. Filtering of the volume and occupancy detector data and a “hysteresis” algorithm provide this protection. The effect is to provide a somewhat sluggish timing plan selection response to changes in the traffic demand. A more detailed description is available in Gordon and Tighe (2005). Parameter Selection Strategy Many traffic responsive systems assign specific detectors to identify particular parameter characteristics that are then used to select timing plans or timing plan characteristics. The following discus- sion is adapted from Gordon and Tighe (2005). The specific signal timing plan selection algorithms vary among system suppliers; however, they generally provide the following features: System detectors in a control section are assigned to • influence either the cycle, split, or offset parameter. A system detector may be assigned to one or more parameters. Selections of cycle, split, and offset are made based on • detector data. The cycle selection, for example, would typically depend on volume and/or occupancy lying between pre-established thresholds. A cycle length is associated for each range of detector values lying

14 The approximate cost to upgrade to adaptive control is • approximately $20,000 per intersection [Costs reported in Stevanovic (2010) are considerably higher.] No consistent data could be identified for the average • time per intersection needed to install and upgrade an adaptive system. 64% of the survey respondents indicated that the train-• ing effort for adaptive systems was greater than they expected. Of the 28 systems surveyed, 13 have been abandoned • or shut down. Saturated Intersections When the demand volume at an intersection approaches its capacity, significant queues begin to form. These queues may exceed the storage capacity of turning bays (thus impacting flow on the through lanes) or may spill back across upstream intersections. The in-depth case studies indicate that many agencies use special measures to control the saturated intersections. If all the major phases become saturated (or oversaturated if demand exceeds capacity) and progression becomes inhibited, many agencies remove the intersection from coordinated operation and use timing that is designed to minimize spillback and lengthy queues. Research reports describe a number of strategies for control. Strategies to Control a Saturated or Oversaturated Intersection The in-depth case studies indicate that many agencies use the volume/density property of intersection controllers operating as isolated controllers to perform this function. Specific strategies to control isolated intersections under saturated conditions are discussed in the literature including approaches such as such intersection utilization (Li 2002) and queue length control based on vehicle storage capability (Gordon 1969), but these strategies have not seen significant operational deployment. The UTCS First Generation critical intersection control software (Gordon et al. 1996), which is used in a number of operational traffic control systems, com- putes green time demand for each phase based on volume and occupancy measured some distance upstream of the sig- nal. Total available green time is then split by the ratio of the phase demands. Some adaptive systems have algorithms that apportion green time by equalizing the degree of saturation (similar to volume-to-capacity ratio). In addition, techniques have been developed to identify downstream congestion by reducing upstream green times (Smaglik et al. 2006; Beaird et al. 2006). NCHRP Project 3-90, Operation of Traffic Signal Sys- tems in Oversaturated Conditions, will provide guidance on the control of these intersections. Adaptive traffic signal control is a concept where vehicular traffic in a network is detected at an upstream and/or downstream point and an algorithm is used to predict when and where the traffic will be and to make signal adjustments at the downstream intersections based on those predictions. The signal controller utilizes these algorithms to compute optimal signal timings based on detected traffic volume and simultaneously implements the timings in real-time. This real-time optimization allows a signal network to react to volume variations, which results in reduced vehicle delay, shorter queues, and decreased travel times. Adaptive signal control autonomously adjusts signal timing parameters in real- time, to respond to actual, real-time traffic conditions. By adjusting the traffic control parameters to more closely align with traffic conditions, adaptive systems can reduce traffic delay, increase average speeds, improve travel times, and decrease travel time variability. Small rapid timing adjustments are typically performed on the basis of a control algorithm that is optimized for some function. Conventional adaptive control systems change the signal timing on a cycle-by-cycle basis or during the phases of a cycle. Adaptive systems such as these require considerably more traffic detectors than is the case for Level 3 systems, thus increasing the cost of equipment and its maintenance significantly. A detailed discussion of adaptive systems is provided in Stevanovic (2010). Koonce et al. (2008) indicated that conventional traffic adaptive systems may be most useful when the following conditions apply: Traffic conditions fluctuate randomly on a day-to-day • basis. Traffic conditions change rapidly owing to new or • changing developments in land use. Incidents, crashes, or other events result in unexpected • changes to traffic demand. Other disruptive events, such as preemption, require a • response. Stevanovic (2010) reported that most of the agencies employing adaptive systems found performance to be gen- erally improved, and reported a higher level of improvement for operation under oversaturated conditions. Abbas et al. (2004) indicated that adaptive systems are not used more extensively because of their significant addi- tional cost. Operation and Maintenance of Adaptive Systems A recent survey of adaptive systems operations (Selinger and Schmidt 2009) indicates the following:

15 not well supported either by theory or by reported research. Traffic systems generally provide the ability to define a net- work or section of signals whose timing plans are controlled as a group. A follow-up question to the in-depth case studies assessed the techniques used for determining the boundaries of coordinated networks. Although analytic techniques such as those described in Appendix C are sometimes used, the tech- niques mentioned most frequently are the distance between intersections, along with judgment and traffic volumes. Number of Timing Plans Another key signal timing issue related to networks is the iden- tification of the number of timing plans and their time of oper- ation. The general practice in the United States is described by the FHWA Traffic Signal Timing Manual (Koonce 2008) as follows: The process of selecting the number of timing plans needed and the times of day when they operate may be determined through a combination of reviewing traffic data along the corridor, such as 24-hour directional traffic counts, intersection turning movement counts and traffic engineering judgment. Ideally, the following steps are used to determine the number of different plans and appropriateness for time of day changes: For each section of the system (this is another reason • why sectioning is important), select a sample of representative intersections. These might be the most important intersections in the section (if some parts of the section are more important than others), or they may be a group of intersections whose traffic conditions are typical of those existing throughout the section. In either case, the set of representative intersections are advised to include only locations that are adjacent to each other. Note that this process will require the availability of hourly counts for the entire time period being analyzed. Prepare a graph that plots the traffic volume as a function • of time of day for the two or three most important intersections in the sample. Using the graph, identify the AM, PM and off-peak time periods for the sample. This step is executed using judgment in the same manner that a traffic engineer would manually determine the time periods for these conditions. Figure 3 shows an example of a graph developed by this process. FIGURE 3 Traffic volumes summary used to determine weekday time-of-day plans. (Source: Koonce et al. 2008.) Strategies to Restrict Traffic from Entering Saturated Intersections A somewhat different approach recognizes that demand vol- ume at the critical intersection may be sufficiently high so that signal timing at that intersection cannot satisfactorily service these volumes, and that the resulting queues are unacceptable. These strategies reduce the flow to the critical intersection or to the area in which that intersection is located by control of upstream signals. This restriction is accomplished by tim- ing other signals so that fewer vehicles reach the approach to the critical intersection. Lieberman et al. (2000) described the RT/IMPOST strategy. The policy’s principles are: (1) the signal phase durations “meter” traffic at intersections servic- ing oversaturated approaches to control and stabilize queue lengths and to provide equitable service to competing traf- fic streams; and (2) the signal coordination (i.e., offset) con- trols the interaction between incoming platoons and standing queues in a way that fully utilizes available storage capacity, keeps intersections clear of queue spillback, and maximizes throughput. Lieberman has shown that this strategy is more effective at reducing delay than conventional signal timing programs. At least one adaptive system contains a capability termed “gating” (Bretherton 2003). Gating can reduce traffic entry to links likely to become congested (trigger links) by controlling discharge flows on one or more links (restrained links). Trig- gers may be provided on a gradual basis. Signal Timing Requirements for Networks Signal Groups For the purpose of coordinating traffic signals, an agency’s signals may be grouped into networks or sections. A network is a subset of the agency’s signals whose timing plans change at the same time, usually for the purpose of establishing traffic signal progressions. Signal timing techniques are discussed in chapter four. Isolated signals (signals not operating according to the network’s cycle length and offset constraints) may serve to define the boundaries of the network or may operate within the physical boundaries of a network. Some of the techniques described in the literature to estab- lish network boundaries emphasize the distance between adjacent signalized intersections, whereas others include the comparison of through volumes with volumes that turn into the link-containing intersection. Still other techniques are based on the friction caused by adjacent land use and other factors, as well as the travel time between signalized intersections. Appendix C discusses these techniques. There appears to be no consistent approach establishing signal section bound- aries. In addition, a number of the techniques described are

16 with the detector located some distance upstream of the stop line.) This indicates a considerable reduction in speed. It is pos- sible that consideration of a measure of congestion or incipient congestion such as occupancy, when used in conjunction with volume, will result in an improved profile for establishing the number of timing plans and their periods of use. Table 5 shows an example of a set of timing plan periods that were developed by the use of both volume and occupancy data. STATE OF THE PRACTICE AND CONCLUSIONS Some progress has been made in establishing the ben-• efits of fully actuated controllers under coordinated conditions. Guidelines to establish the conditions for this application may assist practitioners. No consensus appears to have been established for • a methodology to establish traffic section boundar- ies. Research to develop a methodology would likely improve the effectiveness of coordination. The current guidance for establishing the number • of timing plans that a network requires and the time periods for which they may be employed does not take near-saturation or oversaturated conditions into account. Although current research provides a meth- odology to establish time periods, this methodology is complex. Practical methodologies for establishing the number and time periods for timing plans that utilize measures of both volume and congestion may be useful to practitioners. An analytical approach is described by Park et al. (2004). It uses a genetic algorithm (an efficient search process) in con- junction with a signal timing program and a traffic simulation program in an iterative fashion to develop the combination of timing plans and the periods for which they are to be used. TABLE 5 TIMING PLAN INITIATION SCHEDULE FOR WHITE PLAINS, NEW YORK Section Type Time of Day Section 1 Central Business District Grid Section 2 Arterial Section 3 Arterial Section 4 Arterial 24:00 (continued) TP1 TP1 TP1 TP1 07:30 TP2 TP2 TP2 TP2 08:45 TP3 TP3 09:15 TP3 09:30 TP3 11:00 TP4 11:45 TP4 TP4 TP4 15:30 TP5 16:00 TP5 16:15 TP5 18:30 TP6 TP5 TP6 TP6 20:30 TP1 20:45 TP1 TP1 21:30 TP1 Source: Gordon (2003). When volumes are sufficiently high, these approaches appear to lack the sensitivity to identify flow conditions that result in delays, flow instability, and the resultant difficulty in traveling the green band. In Figure 4, Berkow et al. (2009) provide a volume versus occupancy plot that shows stable flow until 7% to 10% of occupancy is reached, at which point flow breaks down and occupancy increases much more rapidly than volume. (The detector site for this plot is a three-lane approach FIGURE 4 Flow-occupancy diagram for an intersection approach in Portland, Oregon.(Source: Berkow et al. 2008.)

17 CHAPTER FOUR GENERAL SIGNAL TIMING AND RETIMING CONSIDERATIONS of the period in question and extrapolating the data to the full period as described by Maher (2007). Estimate turning movement counts by using tech-• niques that estimate turning movements from volume counts. This may be accomplished by an iterative vol- ume balancing procedure as described in Pederson and Samdahl (1982). Software to accomplish this is avail- able online (http://www.dowlinginc.com/). Low-cost retiming substitutes simple evaluative pro-• cedures such as driving the site to observe unexpected congestion patterns and observations of left turn bays for more formal evaluation procedures. FIGURE 5 Classical approach to signal timing (Source: Henry 2005.) This chapter discusses signal timing and retiming require- ments. Where signals have been previously timed, a general database structure for phasing and timing exists. As described in chapter two, good timing practices require these items to be revisited to ensure that they reflect current policies as well as the current environment. Figure 5 (Henry 2005) provides an overview of the signal timing process. Figure 6 (Day et al. 2009) emphasizes the feedback of information in the sig- nal timing design process. The timing plans provided by the signal timing software are evaluated initially to determine whether other parameters or settings are more appropriate. The deployed timing plans are evaluated by such means as field measurements, motorist complaints, and safety records. This is often a long-term process leading to changes in the design of timing plans. REVIEW OF LITERATURE Collection of Data for Signal Timing Data collection requirements as described in the FHWA Traffic Signal Timing Manual (Koonce et al. 2008) are sum- marized in Table 6. Safety issues may influence phasing and clearance intervals. In addition to the requirements shown in the table, most agencies also consider safety issues. Safety reviews and audits may consider a history of crashes to determine whether safety conditions have recently changed, and how the crash rate at the intersection relates to intersec- tion averages that are normalized for arrival rates. Some agencies have used existing detectors in conjunc- tion with the data compiled by traffic signal systems to obtain the 24-hour weekly volume profiles described in Table 6 (Using Existing Loops at Signalized Intersections for Traffic Counts 2008). Budget considerations may preclude the use of standard techniques for retiming signals. Henry (2005) provides a procedure that accomplishes retiming at a lower cost with some degradation in performance. Some of the measures described include the following: Make use of all existing information• Use “short counts” to obtain turning movement data. • Essentially this entails collecting counts for a portion

18 FIGURE 6 Procedural steps and feedback loops in signal timing design and maintenance. (Source: Day et al. 2009.) TABLE 6 DATA REQUIREMENTS FOR SIGNAL TIMING Type of Data Requirement and Application Traffic Volumes 24-Hour Weekly Volume Profiles Collected at critical locations in the corridor. Used to identify the number of timing plans, transition times, volume adjustment factors for developing turning movement counts, and directional distribution. Turning Movement Counts Turning movement counts are typically collected at each of the subject intersections under consideration for retiming. Depending on the traffic volumes and traffic patterns along the corridor, turning movement counts often only need to be conducted during peak periods, commonly the weekday morning, midday, and evening peak time periods. Daily traffic volume profiles may be used to identify the specific time periods for con- ducting the counts and to develop volume adjustment factors for the weekend and/or off-peak traffic volumes. At some locations, the traffic volumes may be higher or more critical during the weekend time periods, which could lead to performing turning move- ment counts for the weekend and the weekday time periods. Additionally, seasonal traf- fic patterns may need to be considered and incorporated in the scope of work. Proce- dures for conducting, recording, and summarizing traffic counts are described in the Manual of Transportation Engineering Studies (Robertson 2000). Intersection turning movement counts are collected for representative traffic periods. Traffic count may include a count of all vehicular traffic at the intersection, as categorized by intersection approach (i.e., northbound, southbound, etc.) and movement (i.e., left-turn, through, or right-turn movement), pedestrians, and vehicle type (including transit). If heavy vehicles are significant, the vehicular count may be further categorized by vehicle classification. The presence of special users at the intersection (e.g., elderly pedestrians, school children, bicycles, emergency vehicles, etc.) may also be documented. Vehicular Speed Data on vehicular traffic speed may be gathered to identify the approach speeds to the intersection. This is especially important in determining the signal phasing and clear- ance requirements. Travel Time Runs Travel time runs can be used to calibrate the existing analysis model and to compare the corridor operations before and after the new timings are implemented. Travel time runs are collected by driving the subject corridor and recording the delay, stops, and running time using electronic equipment. Intersection Geometry and Control A site survey may be conducted to record relevant geometric and traffic control data. These data include: number of lanes, lane width, lane assignment (i.e., exclusive left turn, through only, shared through and right turn, etc.), presence of turn bays, length of turn bays, length of pedestrian crosswalks, and intersection width for all approach legs. A condition diagram is an effective method for recording this information. Field Review Key elements to consider are location and operation of signal equipment, intersection geometry, signal phasing, intersection operations, vehicle queuing, adjacent traffic gen- erators, posted speed, and/or free flow speed. The free flow speed will assist with cali- brating the traffic model and establishing offsets that reflect the traffic conditions. Existing Signal Timing The existing signal timing helps the traffic engineer understand what currently exists in the field and provides a baseline for improvement. Source: Koonce et al. (2008).

19 These guidelines are sometimes superseded by the requirements of state Uniform Manuals on Traffic Control Devices. In addition, they do not provide guidance for the possible elimination of conflicts with pedestrians. Consideration of Safety Issues The in-depth case studies have rated safety issues as among the most important factors in the operation of intersections. Crash experience is a warrant in the MUTCD. Many agen- cies use accident modification factors (AMFs), also known as crash reduction factors (CRFs), to estimate the effect of a design change in the crash rate. Monsere et al. (2006) describe a number of these factors. NCHRP Project 17-25 Crash Reduction Factors for Traffic Engineering and ITS Improvements provides comprehensive guidance on these factors. An “x” indication in Table 7 (Koonce et al. 2008) identi- fies the relationship of crash characteristics to signal timing issues. Enhancement of pedestrian safety includes special pedes- trian intervals (leading and lagging) and incorporation of an exclusive pedestrian phase. Although leading pedestrian intervals are often cited as improving pedestrian safety, the literature also shows cases where pedestrian–vehicle con- flicts have increased when leading pedestrian intervals have been deployed (Hubbard et al. 2008). Although the MUTCD provides for a reduction in the speed for the pedestrian walk signal clearance interval from 4.0 feet per second to 3.5 feet per second, this lower speed had already been adopted by a number of the agencies responding to the in-depth case study questionnaire. Measures such as these to enhance pedestrian safety often adversely affect vehicle green time and cycle length. Hua et al. (2009) reported that the most effective pedestrian safety improvements for the locations ANALYSIS OF THE INTERSECTION Assessment of Phasing The Manual on Uniform Traffic Control Devices (MUTCD 2009) defines a signal phase as “the right of way, yel- low change, and red clearance intervals in a cycle that are assigned to an independent traffic movement or combination of movements.” Modern U.S. practice for signal control organizes phases by grouping them in a continuous loop (or ring) and separat- ing the crossing or conflicting traffic streams by the time they are allowed to operate, either by making the movements sequential or adding a barrier between the movements. The ring identifies phases that may operate one after another and are typically conflicting phases organized in a particu- lar order. For instance, it may be desirable to separate the traffic traveling through the intersection in the northbound direction from the southbound left turn movement. A change interval and clearance time is used to separate that move- ment in time. In Figure 7, a “barrier” is used to separate the east–west movements from north–south movements to avoid operating conflicting movements at the same time. It is also used to define a relationship between the rings to ensure compatible movements. The barrier represents a reference point in the cycle at which a phase in each ring has reached a point of ter- mination; both rings must cross the barrier simultaneously (Koonce et al. 2008). Figure 7 shows the most commonly employed phasing diagram for major intersections. Guid- ance on selection of phase sequences is provided by Koonce et al. (2008). Koonce has also summarized (Figure 8) prior guidance on determining the need for a left turn phase based on previous work (Orcutt 1993; Kell 1998). Other sources may use somewhat different sight distance data from that shown in Figure 8. FIGURE 7 Typical vehicular and pedestrian movements at an eight-phase dual-ring intersection. (Source: Koonce et al. 2008.)

20 DETECTOR PLACEMENT Placement of detectors depends on the local and system requirements for the intersection. Local requirements include the phases to be actuated, the approach speeds, and studied included the use of flashing beacons and push but- ton actuation, in-street pedestrian signs, vehicle detection to actuate signal timing, leading pedestrian intervals, portable changeable message speed limit signs, and “Turning Traffic Must Yield to Pedestrians” signs. FIGURE 8 Guidelines for determining the potential need for a left-turn phase. (Source: Koonce et al. 2008.)

21 standard data format known as the Universal Traffic Data Format (UTDF). There are six file formats specified by the UTDF (Signal Timing Process Final Report 2003): VOLUME stores volume counts • TIMING stores timing plan information that varies by • time of day PHASING stores timing plan information that doesn’t • change TIMOFDAY stores a weekly schedule of when to use • timing plans LAYOUT stores intersection locations and connections • LANES stores lane and fixed information. • Although HCS (Highway Capacity Software), which programs Highway Capacity Manual relationships, is not a signal timing program, agencies often use it as a basis for timing isolated intersections. Simulation Programs Simulation programs are sometimes used to predict and eval- uate the effects of signal timing optimizations. Macroscopic models often accompany signal timing programs and are the queues expected at the intersection. System requirements depend on the type of traffic responsive or traffic adaptive coordination strategy employed. Review of existing detector placement is critical whenever any of these factors change. Guidance is provided in Gordon and Tighe (2005), Klein et al. (2006), and Koonce et al. (2008). RETIMING TOOLS AND MODELS FOR THE NETWORk Retiming Tools Most agencies use a signal timing software program to mini- mize a delay-based objective function or to maximize arterial through progression. These programs require a database con- sisting of the geometrics of the signalized intersections and the physical relationships among intersections, lane use, and phasing information. Phase-specific parameters such as clear- ance periods, pedestrian walk periods, and actuated control- ler parameters are to be entered. Signal timing programs also generally require average turning movement volumes for the period in which the timing plan will be used. In many cases, migration of data to and from the signal timing program and the traffic control system is facilitated by the use of an open TABLE 7 SAFETY RELATIONSHIPS TO SIGNAL TIMING Collision with Another Vehicle Single-Vehicle Collision Signal Timing Change Angle Head-On Rear- End Sideswipe— Same Direction Sideswipe— Opposite Direction Collision with Bicycle Collision with Park- ing Vehicle Collision with Pedestrian Over- turned Run Off Road Provide Left-Turn Signal Phasing x x x x x Optimize Clearance Intervals x x Restrict or Eliminate Turning Maneuvers (including right turns on red) x x x x x x Employ Signal Coordination x x x Implement Emergency Vehicle Pre-emption x x x x x x x x Improve Traffic Control of Pedestrians and Bicycles x x x Remove Unwarranted Signal x x x Provide/Improve Left-Turn Lane Channelization x x x x x Provide/Improve Right-Turn Lane Channelization x x x x x Source: Koonce et al. (2008).

22 used to assist in evaluating the impacts of signal retiming. Microscopic models simulate the vehicle following charac- teristics for each vehicle. They are sometimes used to deter- mine the effects of retiming on flow characteristics that may vary from cycle to cycle (e.g., the behavior of queues in left turn bays). Dynamic visualizations are usually included in the simulation software. Microscopic models may be used to simulate the effects of such special functions as signal pre- emption and transit vehicle priority. Jeannotte et al. (2004) provide guidance on the selection of models and other tools.

23 CHAPTER FIVE REqUIREMENTS FOR SIGNAL RETIMING be reduced. For example, if the detection zone is 60 ft, the probability that a gap will exceed this value is only 5%. Thus the passage time can be reduced to zero or near zero with a very low probability that the next vehicle in the queue will not be detected. Maximum Green The maximum green parameter represents the maximum amount of time that a green signal indication can be dis- played in the presence of conflicting demand. Maximum green is used to limit the delay to any other movement at the intersection and to keep the cycle length to a maximum amount. It also guards against long green times resulting from continuous demand or broken detectors. Ideally, the maximum green will not be reached because the detection system will find a gap to end the phase, but if there are con- tinuous calls for service and a call on one or more conflict- ing phases, the maximum green parameter will eventually terminate the phase. A maximum green that is too long may result in wasted time at the intersection. If its value is too short, then the phase capacity may be inadequate for the traf- fic demand, and some vehicles will remain unserved at the end of the green interval (Koonce et al. 2008). Guidance for maximum green time settings include the following: A percentage increment over the average time to clear • the phase may be used to ensure that most of the vehi- cles will be serviced during the phase. Koonce et al. (2008) provides specific settings. The maximum green setting is based on the equivalent • optimal pre-timed timing plan based on delay minimi- zation. The minimum-delay green interval durations are multiplied by a factor ranging from 1.25 to 1.50 to obtain an estimate of the maximum green setting (Skabardonis 1998). Guidance for maximum green time • Gmax (in seconds) provided by Bonneson et al. (2009) is as follows: Major-Road Through Phase: Gmax, thru = Larger of: (30, Gmin, thru +10, 0.1 × V) Requirements for timing actuated signals, together with other requirements and agency practices, are discussed in this chapter. REVIEW OF LITERATURE Actuated Timing Periods Three key timing periods related to actuated phases are minimum green, passage time (vehicle extension time), and maximum green. Minimum Green Decisions on the minimum time that the green is displayed should be based on a number of factors. These factors, iden- tified by Koonce et al. (2008), are shown in Table 8. Guid- ance for the settings is provided by Koonce et al. (2008), Henry (2005), and Tarnoff and Ordonez (2004). Passage Time (Vehicle Extension Time) Passage time is used to find a gap in traffic for which to terminate the phase. Essentially it is the setting that results in a phase ending prior to its maximum green time dur- ing isolated operation. If the passage time is too short, the green may end prematurely, before the vehicular move- ment has been adequately served. If the passage interval is set too long, there will be delays to other movements at the intersection. The appropriate passage time used for a particular signal phase depends on many considerations: type and number of detection zones per lane, location of each detection zone, detection zone length, detection call memory (i.e., locking or nonlocking), detection mode (i.e., pulse or presence), approach speed, and whether lane-by- lane or approach detection is used. Ideally, the detection design is established and the passage time is determined to ensure that the “system” provides efficient queue service and safe phase termination for higher-speed approaches (Koonce et al. 2008). Kyte et al. (2009) show that long detection zones for vehicles can permit the values of passage time settings to

24 Koonce et al. (2008) provides guidance for a number of settings for volume-density control. Relationship of Phase Intervals The 8-phase dual ring controller shown in Figure 7 is used extensively. Its many options and sequences, although well defined individually, exhibit numerous complex relation- ships whose interactive effects may be difficult to visualize. Figure 9 shows one example of timing interval relationships. Head et al. (2007) developed a precedence graph model for better understanding these relationships. The complexity of this interaction in a timing plan with multiple phases, over- laps, time of day, traffic responsive, and preemption plans is a substantial challenge for nearly all agencies. Pedestrian Issues MUTCD 2009 reduces the speed used to time the pedestrian clearance interval from 4.0 ft/s to 3.5 ft/s. The MUTCD also indicates that the sum of the walk and pedestrian clearance intervals should not result in a speed exceeding 3 ft/s. Clark et al. (2006) indicated that the minimization of pedestrian–vehicle conflicts is critical to pedestrian safety. They propose the following classifications to measure the effect of conflicts and procedures for measuring these classifications: Non-Conflicting Pedestrian Crossing• : Pedestrian leaves the curb during the walk phase and arrives at far side curb by the end of the pedestrian clearance phase. During the crossing, no delay or evasive maneuver was needed. Compromised Pedestrian Crossing• : Pedestrian reaches the far side curb by the end of the pedestrian clearance phase but is delayed by turning vehicles or takes eva- sive action in response to a turning vehicle. Failed Pedestrian Crossing• : Pedestrian did not reach far side curb by the end of the pedestrian clearance. TABLE 8 FACTORS CONSIDERED WHEN SETTING THE MINIMUM GREEN INTERVAL Phase Stop Line Detection? Pedestrian Button? Considered in Establishing Minimum Green? Driver Expectancy Pedestrian Crossing Time Queue Clearance Through Yes Yes Yes No No No Yes Yes* No No Yes Yes No Yes, if actuated No Yes Yes* Yes, if actuated Left-Turn Yes Not applicable Yes Not applicable No *Slightly modified factors are provided by Bonneson et al. (2009). Source: Koonce et al. (2008). Where: V = peak-period volume per lane (vehicles per hour). Minor-Road Through Phase: Gmax, thru = Larger of: (20, Gmin, thru +10, 0.1 × V) Left-Turn Movement Phase: Gmax, left = Larger of: (15, Gmin,left +10, 0.5 × Gmax, thru) Nichols and Bullock (2001) advise the following:• Max 1 = 1.3 * morning peak split time Max 2 = 1.3 * evening peak split time Very long green times and cycle times may not provide the additional throughput expected by the reduction in lost time (Denney et al. 2009). Volume-Density Control A detector placed considerably upstream of the intersection can anticipate the number of vehicles requiring service dur- ing the next green period. Volume-density control provides for a variable initial period in place of the minimum green time to service these arrivals, thus more accurately fitting the green period to the arriving platoon. The effect on the initial timing should be to increase the timing in a manner dependent on the number of vehicle actuations stored on this phase although the signal is displaying yellow or red (NEMA 2003). This increment is termed added initial. With the detector considerably upstream of the stop line, the period for the vehicle to reach the stop line is consid- erably greater than with stop line detection. Gap reduction timing reduces the allowable gap between vehicle calls in a controlled way to enable preference to be given to vehicles waiting on an opposing phase.

25 FIGURE 10 Minimum green time at isolated intersections. (Source: Tarnoff and Ordonez 2004) When there is a pedestrian demand but no push button is provided, the minimum green is advised to include the times for the pedestrian walk and pedestrian clearance intervals. FIGURE 9 Green interval for volume-density operations. (Source: Pline 2001.) SURVEYS AND IN-DEPTH CASE STUDIES Actuated Timing Periods Minimum Green Figure 10 shows the use of minimum green times as reported by the Tarnoff and Ordonez (2004) survey. Factors for determining minimum green as reported by Tarnoff and Ordonez (2004) include: Driver reaction times; • Intersection width; • Time to move one car through the intersection; • Different fixed policies for main-street left, main-street • through, and cross street; Observation; • 5 s with stop bar detectors, 15 s without; • Grade and type of traffic; • Initial vehicle startup time; • Traffic volume; and • Minimum driver reaction time.•

26 Techniques for determining maximum green time as reported by Tarnoff and Ordonez (2004) include: Traffic demand; • 1.5 times the optimized green time; • Volume/capacity ratio; • Maximum vehicles per hour for the intersection; • Level-of-service; • Field observations; • Desired (optimized) split; • “Rules-of-thumb, delay, density, and backups”; • Proportional to demand using a very long cycle length; • Queue clearance; and • Citywide standard.• Some practitioners use occupancy to assist in determining maximum green time as relatively high occupancies indicate the presence of unsatisfied demand at the end of the phase. Intersection Priorities and Triggers for Retiming The in-depth case studies indicate that although some agen- cies retime signals periodically, others use triggers to initiate the retiming process. These triggers include: Significant changes in land use;• Requests from the public;• Traffic conditions such as saturation or spillback; and• Detector or CCTV information that indicates changes • in volume or congestion patterns. Respondents indicated a strong predisposition to time major arterials more frequently. Other factors mentioned include areas with high growth, critical street segments and intersections, and accident histories. A small number of respondents indicated that no preferential treatment was provided. Appendix D describes a systematic approach to signal retiming prioritization employed by one agency. Frequency of Signal Retiming Most of the in-depth case study participants indicated that they retimed signals at 3- to 5-year intervals. Several of the recent surveys addressed this issue but in slightly different ways. The Puget Sound Survey (Puget Sound Regional ITS Implementation Plan 2008) indicated the following: 22% of the agencies retimed signals at 2-year intervals• 67% of the agencies retimed signals at 3- to 5-year • intervals 11% of the signals retimed signals at intervals of 5 • years or greater Tarnoff and Ordonez (2004) also provide data in Table 9. This is a significant contributor to the wide variation in mini- mum green in Figure 10. Although the presence of a stop line detector theoretically may eliminate the need for a minimum green, most engi- neers believe that at least a short minimum green is needed to satisfy driver expectation. This time may vary among practitioners from 2 to 15 s (Bonneson and McCoy 1993). Koonce et al. (2008) provides guidance for this expectancy time based on intersection classification. Passage Time (Vehicle Extension Time) Factors for determining passage time as reported by Tarnoff and Ordonez (2004) include: Field evaluation;• Speed and volume; • Detector type and location;• Single standard value; • Fixed value (values of 2, 3, 3.5, and 7 s were reported); • Length of detection zone; • Time for vehicles to cross intersection minus 4 s; and • Time needed for slow moving vehicle to pass over loops • and cross the intersection. Maximum Green Figure 11 shows the use of maximum green times as reported by the Tarnoff and Ordonez (2004) survey. The relatively large percentage of long maximum green times may achieve less than expected benefits, as indicated earlier in the chapter. FIGURE 11 Maximum green time at isolated intersections. (Source: Tarnoff and Ordonez 2004.)

27 that promote smooth and efficient traffic movement along an arterial during periods of light traffic flow or at night. Timing Plan Use This section reports the results of in-depth case studies and surveys that indicate the number of timing plans that agen- cies commonly employ. Table 11 and the following discussion employs the com- bined results of the in-depth case studies and the survey con- ducted under Gordon and Braud (2009). TABLE 11 AGENCY USE OF DAILY TIMING PLANS Number of Daily Timing Plans Employed Percent of Agencies 3 13 4 56 5 13 6 6 7 13 Source: Gordon and Braud (2009). All of the agencies responding to both studies employed weekend/holiday/special event plans, usually two to three timing plans of this type. Seventy-six percent of the agencies provided timing plans for diversion, evacuation, or other emergencies. The in-depth case studies indicated that the most com- mon way of establishing time periods is by examining vol- ume and occupancy data. The results of the survey performed by Tarnoff and Ordonez (2004) showing the use of timing plans for various purposes are presented in Table 12. Minor Geometric Improvements In some cases traffic flow may be improved through the use of minor geometric improvements. Five of the seven in-depth case study respondents indicated that they reviewed intersec- tions for minor geometric improvements either periodically, as needed, or in response to complaints. All of the agencies retimed the signals after minor geometric modifications. STATE OF THE PRACTICE AND CONCLUSIONS Passage Time Recent research (Kyte et al. 2009) highlights a strong rela- tionship between detection zone length and the best passage time settings. TABLE 9 AVERAGE RETIMING INTERVAL Average Retiming Interval Percent of Respondents More Frequently than Every 3 Years 42 Around 5 Years 18 Around 10 Years 5 More than 10 Years 35 Source: Tarnoff and Ordonez (2004). Rephasing Techniques and Issues Phasing requirements and signal timing requirements are generally reviewed at the same time. The in-depth case stud- ies identified the relative importance of techniques and issues considered during phasing reviews as shown in Table 10. TABLE 10 TECHNOLOGIES AND ISSUES LEADING TO REPHASING Highly Important Less Important Collision Analysis Critical lane analysis Volume Warrants Separation of conflicting movements Avoidance of Pedestrian Conflicts Green band considerations Left-turn Observations Source: Case studies. Signal Retiming Methodologies All of the agencies participating in the in-depth case studies and almost all of the agencies responding to the information query utilize a signal timing program to develop the timing settings. Tarnoff and Ordonez (2004) also indicated that 52% of respondents to this survey used some sort of manual tech- nique. All of the in-depth case study participants and most of the respondents to the Tarnoff and Ordonez survey use a sim- ulation program to assist in pretesting signal timing plans. According to the 2007 Traffic Signal Report Card, nearly 60% of agencies perform a comparative analysis of cycle lengths, offsets, phase sequence, and other timing param- eters as part of the process for the evaluation and implemen- tation of signal timings. Sixty-one percent reported having strong or outstanding procedures for considering different signal phase sequences to minimize interruption of traffic progression during the evaluation of timing for a coordi- nated system. Agencies also reported using actuation and off-peak timing practices to improve flow during periods of light traffic. Two-thirds (68%) of the agencies reported having strong or outstanding procedures for timing actu- ated controllers. Sixty-nine percent reported having strong or outstanding procedures for using operational strategies

28 Long Maximum Green and Cycle Times Figure 11 shows that 23% of isolated signalized intersections are programmed for green times equal to or greater than 150 seconds. Denney et al. (2009) indicated that these very long green times may not provide the improved throughput that was expected from their use. Further research should include substantiation using a larger database and resultant guidance for these parameters. Interrelationships of Phase Intervals As described by Head et al. (2007), the effects of the inter- relationships of options for the phase settings of actuated controllers are complex and difficult to predict under some circumstances. Current information on these interrelation- ships may be of only limited assistance; further research to improve information for practitioners would be useful. Volume-Density Operation Although descriptions of operation are available along with guidance for timing, additional research might include infor- mation on the conditions for which volume-density opera- tion offers performance improvements. TABLE 12 TIMING PLAN USE Type of Plan Description Percent of Respondents a.m. Peak Weekday morning rush hour 100 p.m. Peak Weekday evening rush hour 100 Off Peak Typically a weekday plan used all other times 86 Incidents Special plan for diverted freeway traffic during incidents 33 Weekends Saturdays and/or Sundays 66 Holidays Special plans when holiday traf- fic differs from weekend traffic (e.g., heavy shopping after Thanksgiving) 57 Weather Plans for snow, fog, or other weather that affects traffic flow 19 Special Events Plans for sporting events or other events that have an impact on traffic flow patterns 56 Other 10 Source: Gordon and Braud (2009).

29 CHAPTER SIX METHODOLOGIES FOR FIELD IMPLEMENTATION OF TIMING PLANS Fine Tuning of Timing Plans The models used for timing plan development described in chapter four are generally macroscopic models; that is, they use average parameters such as volume and capacity along with link parameters to establish the signal timing values. Models of this type generally do not consider the cycle-to- cycle variations in traffic demand. These variations may result in conditions such as spillback from turning lanes and across closely spaced intersections, resulting in considerable disruption to traffic flow. A fine-tuning process is appropri- ate to ensure that the timings address real-world conditions. Koonce et al. (2008) and Henry (2005) strongly advise field reviews of signal timing plans in the form of intersec- tion observations and route driving runs, and appropriate timing adjustments. The following techniques may be used at the central facil- ity to assist in the fine-tuning process. Use of • microscopic simulations. These simulations model the behavior of each vehicle through car-following models and lane-change models. Statistically based traffic gener- ation models provide inputs to the traffic network, and the results of the modeling period may be obtained as com- puter reports. The simulated motion of the vehicles may be observed by the traffic engineer on a visual display. Recent technology has enabled actual field controllers to be used in the office in conjunction with the simulations to provide a more realistic model for field controller opera- tion. Examples of the use of this technology are provided by Yun et al. (2007), Smadi and Birst (2006), and Balke and Herrick (2004). Use of performance monitoring systems. As described • in more detail in chapter nine, performance monitor- ing systems (see Liu and Ma 2008, and Balke et al. 2005) use actual data collected by field controllers to provide an estimate of the intersection performance. They require an additional component in the field con- troller cabinet. Use of high-resolution data collectors to tabulate phase • volume-to-capacity ratios, progression quality, and intersection delay. Smaglik et al. (2007) show how a controller can be used to collect the appropriate data, and indicate that the necessary algorithms can be After timing plans have been developed in the office, they must be transferred to the field controllers. The actual field timing plans should be checked against the intended plans to ensure that they are being properly displayed. Because the processes and models for developing timing plans have limitations in how well they represent actual conditions, it is appropriate for the timing plans to be fine tuned in the field, particularly with regard to such conditions as satisfac- tion of left turn requirements and spillback across upstream intersections. REVIEW OF LITERATURE Methodologies for Transferring Timing Plans to Intersection Controllers The methods for transferring timing plans to field controllers depend on the architecture of the traffic system, the avail- ability of interconnect at the field controller, and agency practices. The methodologies may include the following: In-system architectures that control the timing by • directly providing a signal from the central computer on a second-by-second basis (central control), only the backup timing plans need be installed in the field con- troller at the intersection. Where a central computer is available, and controllers • and communication to the field controller are available (such as for closed loop systems), timing plans may be downloaded from the central computer into the inter- section controller. Many system architectures require a master controller in the field. In this case, timing plans are first downloaded into the field master controller and then into the intersection controller. The timing plans in the controller may be uploaded into the cen- tral computer to ensure that the installed timing plan is correct. In most cases, the traffic systems compare the uploaded plans to the programmed plans and identify discrepancies. Laptop computers, notebook computers, and PDA • devices are often available from traffic system sup- pliers to facilitate the download of timing plans into intersection signal controllers in the field. Timing plans may be entered into the field controller • through a keypad.

30 embedded in the controller. Detectors considerably upstream of the intersection are used to implement these algorithms. IN-DEPTH CASE STUDIES The frequency of usage of these techniques for inserting timing plans into controllers, as determined by this project’s in-depth case studies, is provided in Table 13. TABLE 13 FREQUENCY OF USE OF TECHNIQUES FOR INSERTING TIMING PLANS Technique Number of Agencies Reporting Use, n = 7 Direct Control From Central Computer 1 Download From Central Computer 3 Download From Laptop, Desktop, or PDA 5 Manually Insert 4 Source: Gordon and Braud (2009). All of the in-depth case study agencies indicated that they verified the timing plan settings by field observation and that they performed fine tuning using field observations (one agency indicated that relatively little attention was paid to minor intersections). Four of the agencies indicated some use of CCTV information. Two agencies reported using sim- ulation and field detector data to assist in fine tuning. CONCLUSIONS There appear to be no serious problems in properly installing and verifying timing plans in field controllers. Techniques for facilitating fine tuning by field processing of data col- lected by controllers are emerging and will assist in this process. Smaglik et al. (2007) indicated that a standard to provide for data collection software common to all suppliers would facilitate this process.

31 CHAPTER SEVEN PERSONNEL RESOURCES AND COST FOR IMPLEMENTATION OF SIGNAL TIMING PLANS TABLE 14 NUMBER OF SIGNALS PER TRAFFIC ENGINEER Reference Number of Signals per Traffic Engineer Case Studies Average: 185 All 19 agencies in survey 5 agencies >150 signals Average = 67 Average = 93 Median = 62 Median = 81 Source: Puget Sound Regional ITS Implementation Plan (2008). Figure 12 shows a breakdown for the number of person hours appropriate for signal timing tasks as determined by the Traffic Signal Timing and Operations Survey (ITE 2009). The in-depth case study questionnaire was structured to identify the components of the effort and cost to perform the signal timing tasks. Five of the in-depth case study agencies provided the detailed information for Figure 13, the aver- age person hours expended, and Figure 14, the approximate costs of contracted signal retiming. FIGURE 12 Person hours to retime an intersection (redrawn). Source: Traffic Signal Timing and Operations Survey (2009). Agencies may retime signals using their own personnel or may contract part of or all of the retiming effort to consul- tants. This chapter describes the personnel qualifications and resources necessary to retime signals. REVIEW OF LITERATURE, SURVEYS, AND IN-DEPTH CASE STUDIES Personnel qualifications The appropriate use of signal timing tools and the quality of the entire signal timing process depend, in large mea- sure, on the capabilities of the personnel responsible and the adequacy of staffing. Guidance for personnel qualifications and training is provided by Row and Tarnoff (2005) and by Gordon and Braud (2009). The latter document advises that at a minimum, signal timing and operations be supervised by personnel with Professional Engineer and Professional Traffic Operations Engineer qualifications. Staffing Levels and Cost The ability to retime signals according to the performance- based criteria of objective-oriented operation depends on the agency’s resources and staff. The following discussion provides various perspectives on the requirements necessary to accomplish this as well as the current practices. Because engineering staffs have various responsibilities in addition to signal retiming, it is unlikely that an understaffed agency will be able to perform high-quality retiming. A work load of 75 to 100 signals per engineer for agen- cies that operate a minimum of 150 signals is advised by Gordon and Braud (2009). Smaller agencies will likely require fewer signals per engineer because economies of scale may be difficult to realize. Giblin and Kraft (2000) similarly advise the assignment of 75 to 100 signals per engineer. Table 14 summarizes the traffic engineering staffing levels reported by the Puget Sound survey and the in-depth case studies. The average agency size for the in-depth case studies was considerably larger than for the Puget Sound survey: 62% of the responding agencies responding to the Puget Sound survey believed that their staff size was inadequate.

32 The person hours expended in retiming intersections var- ies considerably among agencies. A representative average value is 26 person hours per intersection. Although the cost of purchasing signal retiming services also varies widely (variations are functions of the number of timing plans and differences in the level of services utilized by agencies), most agencies obtain the retiming services that they require for $3,700 or less per signal. Other sources, including Tarnoff and Ordonez (2004) and W. Hood’s personal communication (2009), reported the aver- age cost of signal retiming to generally be $2,500 to $3,500 per intersection (note that the Tarnoff and Ordonez survey preceded the in-depth case studies by about 5 years). CONCLUSIONS Although the work load levels for the agencies in the Puget Sound survey generally fall within the advised guidelines, most of those agencies perceived that they were understaffed. FIGURE 13 Person hours appropriate for signal retiming tasks. FIGURE 14 Cost for contracted signal retiming tasks.

33 CHAPTER EIGHT PERFORMANCE MEASURES agement purposes such as resource allocation. When consis- tent measurement and computation techniques are used over a period of time, the values developed are useful to estimate the relative performance improvement provided by signal retiming. The 2010 Highway Capacity Manual (HCM) identifies level of service (LOS) as a measure for signalized inter- sections. LOS depends on control delay and whether the volume-to capacity ratio is greater than 1.0. Control delay includes delay associated with vehicles slowing in advance of an intersection, the time spent stopped at an intersec- tion approach, the time spent as vehicles move up in the queue, and the time needed for vehicles to accelerate back to their desired speed. Table 15 shows the LOS for signalized intersections. TABLE 15 HCM 2010 LEVEL OF SERVICE CRITERIA Control delay (s/veh) Level of Service by Volume to Capacity Ratio ≤1.0 >1.0 ≤10 A F >10–20 B F >20–35 C F >35–55 D F >55–80 E F >80 F F Source: 2010 Highway Capacity Manual. Commonly employed vehicle performance measures include volume, travel time, travel time reliability, delay, stops, throughput, queue length, and progression quality. Crashes are a key measure. Evaluations usually include the components of vehicle emissions. A number of traffic signal controllers incorporate the capability for measures of effectiveness reports. One sup- plier’s controller has the capability to provide the following parameters (Balke and Herrick 2004): volume, stops, delays, and utilization (the average time that the green interval lasts for each phase). This chapter covers the functions of performance measures, followed by a discussion of the different classes and appli- cations of performance measures for signal timing. The methods for obtaining these performance measures are also discussed. Research currently being conducted under NCHRP Project 3-79a, Traffic Signal Performance Mea- sures, will significantly enhance these findings. NCHRP Synthesis 311 (Shaw 2003) provides a detailed review of performance measures used for transportation. For the purpose of signal timing evaluation, most agencies use measures that are specifically oriented to that application as described in this chapter. FUNCTIONS OF PERFORMANCE MEASURES Performance monitoring• . This includes the effect of signal timing on users of the intersection, including private vehicle occupants, pedestrians, transit vehicles, and bicycles. From a signal retiming perspective, these measures usually relate to the objectives established by the agency for retiming purposes. Table 2 in chapter two provides examples of the relationships between objectives and measures. Delay for each user class and safety are usually key measures. Measures used to assist in retiming the signals• . Although related to performance monitoring mea- sures, they are intended to assist traffic engineers with signal retiming. Figure 6 (in chapter four) shows how performance mea- sures relate to the retiming process itself (FB1) and to per- formance monitoring (FB2). PERFORMANCE MONITORING MEASURES Vehicle Performance All of the techniques used to obtain values for the measures are subject to errors that result from the measurement tech- nique, or from the processes used to compute the measure from data that are collected. The presence of these errors should be considered when the measures are used for man-

34 Gordon and Braud (2009) indicated that objectives and performance measures are important to achieve quality traf- fic signal system operation. They describe the following cri- teria for signal timing measures: Span the traffic system and agency functions to encom-• pass the different requirements of all agencies. Be measurable or answerable using existing informa-• tion, information that can be readily obtained, or infor- mation that may take some effort to collect but that is vital to the determination of the capability of the traffic system relative to the objective-oriented operation cri- teria described in the report. Minimize the need for subjective judgment to accom-• plish evaluation. The Knoxville Regional Transportation Planning Orga- nization uses the following evaluation measures in their congestion management system (Congestion Management System). Volume-to-capacity ratio (present and future)• Roadway corridor and segment travel time (peak ver-• sus off-peak) Stopped delay at intersections (from travel time study)• All of the in-depth case study respondents indicated that they conducted performance evaluations for signal timing impacts on vehicle flow. The frequency of employment of common measures in the signal timing process is provided in Table 16. TABLE 16 AGENCY USE OF EVALUATION MEASURES FROM CASE STUDIES Measure Agencies Using Measure, n = 7 Delay 6 Stops 6 Route Travel Time 6 Crashes 5 Emissions and Fuel 4 Reliability of Travel Time 2 Pedestrian Performance Measures for pedestrians at signalized intersections have undergone significant changes between HCM 2000 (High- way Capacity Manual 2000) and HCM 2010. Because pedestrians are key intersection users as well, evaluation measures have been developed and used for this purpose. The pedestrian LOS measures provided in HCM 2000 are based on pedestrian delay as computed by the equation. Pedestrian delay = 0.5(C – g)2/C Where: C is cycle length and g is green time. Hubbard et al. (2008) indicated that this equation does not account for pedestrian/vehicle conflicts resulting from vehicles (primarily right turning vehicles) turning into the pedestrian crosswalk during the pedestrian interval. In response to this and other research, HCM 2010 devel- ops a LOS based on a pedestrian perception index for the intersection. This index is based on a number of factors, including intersection and crosswalk geometrics, vehicle volumes, and speed. Hubbard et al. (2008) proposed compromised pedes- trian crossings as a measure of pedestrian service level. A compromised crossing occurs if pedestrians are delayed by turning vehicles or change their travel path or speed. The measure is provided by: Compromise % Non-CBD = 0.040 (right turn flow rate in pedestrian interval in vph) Compromise % CBD = 0.026 (right turn flow rate in pedes- trian interval in vph). Signal Preemption Traffic signal timing for railroad signal preemption relates to the connection between the activation of the protection devices and the traffic signal clearances that ensure queue clearance from the tracks. Balke et al. (2005) discussed these issues and introduced a truncation exposure index to estimate the potential effect of unprotected periods. Emergency vehicle preemption increases delay to gen- eral traffic. A key parameter that determines the extent of the delay is the classes of vehicles for which preempt is pro- vided. All seven of the in-depth case study agencies provide preempt service to fire vehicles, four to ambulances, and two to police vehicles. Delay to general traffic also depends on the type of algo- rithm employed by the controller to restore coordination at the termination of the preempt state (Park et al. 2008; Bull- ock and Cantarella 1998).

35 MEASURES USED TO ASSIST IN RETIMING SIGNALS To provide a basis for improved signal timing, Balke and Herrick (2004) identified the following measures for the evaluation of signal timing plans: The average number of times a phase was activated in • a given evaluation period The average number of vehicles served per cycle dur-• ing a given evaluation period The average number of vehicles stored per cycle during • a given evaluation period (residual queue) The probability of a vehicle having to stop at an • approach during a given evaluation period The percentage of overloaded cycles (or cycle failures) • during a given evaluation period Bullock et al. (2008) provided examples of how the fol- lowing performance measures defined in the HCM 2000 may be used to improve signal timing: progression adjust- ment factor, volume-to-capacity ratio per phase, and compo- nents of intersection saturation. These measures are further discussed in Day et al. (2010). Recent signal timing programs have incorporated mea- sures for congested conditions characterized by spillback or spillover across intersections and from turning bays. As an example, the measures currently provided by one pro- gram include: Queue delay—Effect of queues and blocking on short • links and turning bays. Spillback capacity reduction—Reduction to the base • capacity caused by a short downstream link becoming filled up. Base capacity is unimpeded capacity. Storage capacity reduction—Reduction to the base • capacity caused when turn pockets cannot accommo- date queue lengths. Reduced v/c ratio—Modified volume-to-capacity ratio • relative to the base capacity.

36 CHAPTER NINE EVALUATION OF SIGNAL TIMING PERFORMANCE This chapter discusses the techniques used by agencies to evaluate the benefits obtained by retiming traffic signals, and how the economic and emissions impacts of these ben- efits are estimated. The measures in Table 17 are commonly used to evaluate the benefits associated with the implementation of revised signal timing plans. This chapter discusses the principal methods for obtaining these measures. TABLE 17 COMMONLY USED EVALUATION MEASURES Measure Type of Benefit Delay Traveler utility Travel Time and Travel Time Reliability Traveler utility Stops Traveler utility Crashes Safety Fuel Consumption Out of pocket cost Emissions Environment REVIEW OF LITERATURE Techniques Used for Travel Time and Delay Evaluation Observations of Vehicle Travel Behavior Methods include travel time studies for test vehicles and intersection observations of delays and queues. Compari- sons may be made before and after signal retiming. Meth- ods, sample sizes, and data collection procedures for route travel times and delays as well as intersection delays are, for example, provided by Robertson (2000). A number of states also provide procedures for collection and analysis of field data (Manual on Uniform Traffic Studies 2000). GPS is commonly used to facilitate the data collection and pro- cessing associated with floating vehicle studies. Although traditionally used by traffic engineers, these methods are generally labor intensive and are therefore often expensive to implement. Software is available to assist in the collection and pro- cessing of data from floating vehicles. The software enables the user to study the detailed movement of test vehicles. It computes evaluation measures such as travel time, delay, fuel consumption and emissions. Recent technology advances enable vehicles serving as probes to be used for travel time determinations. Tech- nologies such as toll tag systems, Bluetooth readers, license plate matching systems, and systems using cellular tele- phones and GPS may supply data to provide travel time to motorists and evaluate travel time performance. Evaluations using these technologies can be economically performed at more frequent intervals than floating vehicle studies. Using special test vehicles, Fontaine and Smith (2005) described the error properties of probe-generated data. Wasson et al. (2008) indicated that the use of unexpected routes by vehicles on surface streets requires special data processing techniques to eliminate these vehicles from the estimation. They also indicated that the use of a vehicle identification format known as media access control facilitates the data processing appropriate to identify vehicles for travel time estimation. Estimates of Delay and Other Measures Using Traffic Signal Systems With more extensive use of traffic detectors and centrally monitored traffic control systems it may be possible to col- lect the data necessary to develop signal system evaluation measures in an automated and less labor intensive way than that described earlier. However, data collected by conven- tional traffic systems are usually in the form of periodic values for volume, occupancy, and speed at the detector loca- tions. These data do not directly correlate to the commonly used measures shown in Table 17. One approach to obtain- ing these measures is to incorporate advanced software into the traffic controller, as discussed in chapter eight. Another approach is to use archived user data system technology. The National ITS Architecture provides for the establish- ment of the Archived Data User Service (ADUS) to control the archiving and distribution of ITS data. ADUS provides the Historical Data Archive Repositories and controls the archiving functionality for all ITS data. FHWA maintains an ADUS website (ADUS) that references ADUS standards and other background material. An archived data management system (ADMS) is an application that takes the data from

37 Simulation Estimates of Performance Evaluations may be conducted by the use of microscopic simulation. Because of the difficulty in relating simulation results to actual field performance, simulation is most often used to choose alternative timing plans or alternative control modes for one or more sets of traffic conditions. An example of such an approach is described by Wilson et al. (2006), who simulated a traffic adaptive traffic signal system and the field controller response to signal control and traffic actua- tions. A general microscopic traffic simulation was modified using an application’s programming interface to model the traffic system’s control algorithms and detectors. Using this simulation, Figure 17 shows the results for an arterial for three different signal system modes. The SCATS Isolated VA mode is an adaptive vehicle actuated mode that changes cycle length and split. The SCATS Masterlink Mode changes offset as well. It is seen from the upper figure (which shows delay under average volume conditions) that delay is fairly similar for all three modes when the actual volume is close to the volume used to develop the timing plans. The lower figure shows that when the traffic volume differs from the volume used to develop the fixed time plans, both the value the archive and analyzes them or formats them in a way to facilitate subsequent analysis. Several ADUS/ADMS systems have been devised to develop evaluation parameters based on the traffic control- lers’ phase indications, phase calls, and detector inputs. Balke et al. (2005) describe the Traffic Signal Performance Monitoring System, that develops measures for isolated inter- sections, whereas Liu and Ma (2008) report on the SMART- SIGNAL system. Figure 15 shows the SMART-SIGNAL system’s architecture. The figure shows the data processing as located at the University of Minnesota’s facility. The local data collection units are SMART-SIGNAL equipment that must be added to the controller cabinet. The parameters generated by the SMART-SIGNAL sys- tem include intersection delay, stops, LOS, queue, and cor- ridor travel time. Figure 16 shows an example of the type of evaluation that it is possible to develop, the delay on an arte- rial improved by signal retiming. Real-time delay estimation models using high-resolution controller data have also been developed by Sharma et al. (2007, 2008). FIGURE 15 The SMART-SIGNAL system architecture. (Source: Liu and Ma 2007.)

38 Fuel Consumption and Emissions Fuel Consumption Fuel consumption rates depend on the mix of vehicle types and on the mix of engine technologies employed. These parameters are expected to change in the future as recently revised fuel consumption standards take effect. The price of fuel affects travel demand and consequently influences fuel consumption. The following is a model for fuel consumption used at this writing by two signal timing programs: F = TotalTravel * k1 + TotalDelay * k2 + Stops * k3 k1 = .075283 – .0015892 * Speed + .000015066 * Speed2 k2 = .7329 k3 = .0000061411 * speed2 F = fuel consumed in gallons Speed = cruise speed in mph TotalTravel = vehicle miles traveled TotalDelay = total signal delay in hours Stops = total stops in vehicles per hour. of delay and its variation is considerably smaller for the adaptive modes. Safety Changes in signal timing are sometimes used to improve safety at intersections. Conversely, changes to improve traf- fic flow may have adverse effects on safety. The crash rate and its trend when changes are made in signal timing, phas- ing, and intersection design are key measures of safety and change in safety. Some agencies provide overall crash rate statistics for intersection type and crash cause as an ongo- ing process. Figure 18 is an example (NYSDOT 2009) that provides intersection accidents per million entering vehicles (MEV). Data such as these establish a baseline for the analy- sis of intersection safety. Actual intersection crash rates may be statistically ana- lyzed in conjunction with these data to identify high crash locations. The analysis may result in establishing priorities for remedial action (2008 Annual Evaluation Report). Comparison of intersection crash performance with aver- age overall rates and crash types may provide a basis for improvement of signalized intersection treatment. As shown in the matrix in Table 7, Koonce et al. (2008) correlate crash characteristics with the types of signal timing changes that may reduce crash rates. FIGURE 16 Delay averaged over 10 weekdays from 3 p.m. to 5 p.m. (redrawn). (Source: Liu and Ma 2007.)

39 (data provided by an unpublished letter to the FHWA from the Oak Ridge National Laboratory) are given by: CO = F * 69.9 grams/gallon NOx = F * 13.6 grams/gallon VOC = F * 16.2 grams/gallon F = fuel consumption (gallons). The anticipated use of a new emissions simulation program (MOVES 2009) may result in a change in these values. FIGURE 17 Delay for different control modes and volume variations. (Source: Wilson et al. 2006.) Because of the model’s assumptions concerning vehicle types and engine technologies, caution is advised in using the model. It is best used for a comparison before and after signal retiming. (This is only one of several models cur- rently in use and may not represent the latest data.) Emissions Emissions models used by signal timing programs often use fuel consumption as a key parameter. The emissions com- monly used as measures include carbon monoxide (CO), nitro- gen oxides (NOx), and volatile oxygen compounds (VOC). Emission rates currently used by two signal timing programs

40 used to express the results of the analysis. Many studies have shown that signal retiming generally provides signifi- cant reductions in delay, fuel consumed, and emissions; however benefit-to-cost ratios may vary over a wide range (Skabardonis 1994; Sunkari 2004). Cost and benefit analy- sis has the following limitations (Papacostas 1993): Not all impacts considered to be important can be • included in the economic analysis. Benefits and costs may be unevenly distributed among • highway users. Components incorporated into the cost and benefit analy- sis often include: Value of vehicle delay reduced. A dollar value for • traveler time is introduced. Kruesi (1997) provides a methodology for estimating this value. Kruesi responds to the question of how to value the small time savings that often result from signal retiming projects by indicating that “the present state of knowl- edge does not appear to support valuing small time savings at lower hourly rates than larger savings.” Cost of accidents reduced.• Dollar value of reduced fuel consumption.• Dollar value of reduced emissions.• Appendix E provides an example of a typical signal retiming evaluation methodology, as well as the benefits and costs for this case. Table 18 provides dollar benefits in terms of Year 2000 dollars that may be used for benefit versus cost analysis (Highway Economic Requirements System 2002). TABLE 18 AIR POLLUTION DAMAGE COSTS AND ADJUSTMENT FACTORS Pollutant Damage Costs $/Ton Adjustment Factor Urban Rural Carbon Monoxide 100 1.0 0.5 Volatile Organic Compounds 2,750 1.5 1.0 Nitrogen Oxides 3,625 1.5 1.0 Source: Highway Economic Requirements System (2002). Because of continuing emphasis on control of CO2 emis- sions, that parameter is increasingly being used as a measure for improvements related to signal timing. The EPA (Emis- sion Facts 2005) provides the following data: CO2 emissions from a gallon of gasoline = 19.4 lb/gal CO2 emissions from a gallon of diesel = 22.2 lb/gal Cost and Benefit Analysis Cost and benefit analysis is a key measure for evaluation of signal retiming. The benefit-to-cost ratio is commonly Note: ACC/MEV = accidents/million entering vehicles. FIGURE 18 New York State DOT average intersection accident rates for accident types. (Source: New York State DOT 2009.)

41 IN-DEPTH CASE STUDIES All of the in-depth case study agencies reported that they performed evaluations of signal timing performance. The evaluation techniques used are shown in Table 19. TABLE 19 SIGNAL TIMING PERFORMANCE EVALUATION TECHNIQUES USED BY AGENCIES Evaluation Technique Used Agencies Using Technique, n = 7 Travel Time Measurements 7 Intersection Observations 6 Records of Crashes 5 Simulation 3 Traffic Detectors 3 CONCLUSIONS Most agencies evaluate the benefits of signal retiming projects. They usually do this through field observation of vehicle travel time and delay and by analyzing safety, fuel consumption, and emissions data. Simulation may be used to assist in evaluation. In recent years the following tech- niques have emerged to assist in data collection for retiming evaluations: Incorporation of evaluation capability into traffic • controllers. Use of additional equipment in the controller cabinet, • field master cabinet, and at the transportation manage- ment center to develop evaluation measures.

42 CHAPTER TEN BARRIERS TO IMPLEMENTATION OF SIGNAL RETIMING PLANS among agencies. This is sometimes facilitated by co-location of traffic management centers or by agencies that coordinate information among operating agencies (such as TRANSCOM in the New York City metropolitan area or by the I-95 Corri- dor Coalition). The increasing use of the National Transpor- tation Communications for ITS Protocol (NTCIP) standards facilitates the mobility of this type of information. Gaps in Guidance or Tools Some issues relating to signal timing are underemphasized either because they are not well understood or because no commonly accepted technique is available for practitioners. Controller Option Interactions Most of the controllers available for purchase have the features defined (as a minimum) by the NEMA TS2 standard (NEMA Standards Publication TS 2-2003 v02.06 2003). Many control- lers have additional capabilities. Most of the earlier controllers commonly in use also contain many of these features. Although guidance is available for many of the parameters taken sepa- rately, because of the complexity of their interactions, it is dif- ficult to predict their operation under certain traffic conditions (see chapter five). Although tools exist to simulate or test these interactions, many agencies do not have the resources to use these methodologies for a significant number of intersections. Subdivision of Traffic Networks Appendix C describes a number of analytical techniques for establishing network boundaries. Chapter three indicates that these techniques are not consistent with each other. Although agencies reported that they use such data as spac- ing and volume, these data are not commonly used in con- junction with analytical techniques. Selection of the Number of Timing Plans and Their Deployment Periods Current guidance and commonly employed practice, such as that discussed by Koonce (2008) and described in chapter three, basically depends on judgments by the analyst, and no evidence is offered to indicate that the approach leads to the best selection of plans and their deployment periods. The analytic concept described by Park et al. (2004) provides a This chapter discusses the technical, institutional, and resource limitations that often result in signal retiming plans that are less effective than what is possible with the current state-of-the-art. REVIEW OF LITERATURE Signal Coordination Across Agency Boundaries Coordination of traffic signals across agency boundaries is a common requirement to optimize signal timing. Timbrook et al. (2002) describe a series of in-depth case studies for which signal coordination was achieved. The study showed that coordination across agency boundaries was possible, even if the equipment and traffic system communications used by these agencies differ. The most important factor cited in the report is cooperation and communications among the agencies. In three of the in-depth case studies, the regional government agencies [metropolitan planning organizations (MPOs)] have been instrumental in bringing the agencies together and developing working plans for coordination. To successfully implement cross-jurisdictional signal timing, sometimes each of the agencies is advised to adopt less than optimal cycles or offsets to achieve the common goal of a seamless transition across neighboring boundaries. At the same time, each of the agencies wants to be able to respond to the needs of its constituents. This requires open communi- cation between the agencies. In some cases a memorandum of understanding provides the basis for the arrangements, whereas in other cases it is done on an ad hoc basis. Incident conditions on other facilities such as freeways provide another opportunity for signal coordination across agency boundaries. Predetermined, jointly developed timing plans to respond to incidents may be called for by a transpor- tation management center (Carvell et al. 1996). The 2007 Traffic Signal Report Card indicated that one- third of the agencies provided little to no signal coordination across agency boundaries. Sharing of Information Devices With the increased use of variable message signs and CCTV on surface streets, it is desirable to share the use of these devices

43 more systematic approach, but it currently appears to be too complex for use by practitioners. Neither method makes use of congestion-related informa- tion. It is likely that if a parameter related to congestion is introduced, the number of appropriate timing plans and their periods of use will change. Use of Traffic Detector Data Although most currently used traffic systems collect data from traffic detectors, it is usually not provided in formats that facilitate its use for the interpretation of timing plan per- formance. Prototypes for tools to address this issue appear in chapter eight, which discusses incorporating improved mea- sures into traffic signal controllers. Chapter nine discusses the archived data user service technology, which provides improved data presentations that facilitate rapid and inex- pensive feedback on system performance. SURVEYS AND IN-DEPTH CASE STUDIES Resource Limitations The Puget Sound survey and the in-depth case studies indi- cate that resource limitations are a major barrier to improving signal retiming. The Puget Sound survey reported that 72% of the agencies interviewed had less than the desired resources. All of the in-depth case studies conducted under this proj- ect identified resource limitations as a barrier to improved signal timing. (The in-depth case studies provide the most recent information and reflect reductions in resources imposed on some agencies by recent economic conditions.) A somewhat different way of viewing these limitations is to consider the agency performance using the resources available to it. In the survey reported by Gordon and Braud (2009), agencies were rated on their ability to achieve objec- tive-oriented operation (OOO). OOO roughly corresponds to a grade of 5 on the National Traffic Signal Report Card. Achievement of OOO was given a grade of 5.0 in Table 20 for agencies responsible for traffic signal operation. Sharing of Signal Control Sometimes adjacent agencies may find it useful to • share signal control. For example, sharing control would allow monitoring if an agency’s traffic manage- ment center is not staffed on a round the clock basis. In some cases a signal may be better controlled by an adjacent agency because it is physically closer to the other signals operated by that agency and/or is more closely related to its coordination needs. TABLE 20 AVERAGE AGENCY SCORE RELATIVE TO ACHIEVEMENT OF OBJECTIVE-ORIENTED OPERATION Function Average Score (1 is lowest, 5 is highest) Management Planning Documentation 2.5 Collection, Analysis, and Review of Data 2.0 Operations Monitoring at TOC 3.3 Personnel Qualifications 1.4 Use of CCTV 3.1 Number of Timing Plans 4.0 Frequency of Timing Plan Update 3.8 Maintenance Notification of Critical Failures 2.9 Time to Respond after Notification 4.5 Training and Qualifications of Maintenance Staff 3.5 Source: Gordon and Braud (2009). Table 21 identifies the barriers to this type of control as reported by the Puget Sound survey. Table 22 provides the survey respondents’ comments on the barriers. TABLE 21 BARRIERS TO SIGNAL CONTROL SHARING Barrier Percent of Agencies Identifying Barrier as an Issue Institutional Issues 64 Technical Barriers 80 Legal/Liability 54 None Reported 10 Other (including staffing, trust, security) 20 Source: Puget Sound Regional ITS Implementation Plan (2008). Competing Requirements Intersection signal timing reflects the competing objec- tives and requirements of different classes of users. Examples include pedestrian access and safety, emer- gency vehicle preemption, transit priority, bicycle access and safety, interjurisdictional signal timing coordination, and neighborhood access needs. Also described are the competing requirements in descending order of impor- tance as identified by the in-depth case studies. The in- depth case studies may not fully reflect the priorities for major cities.

44 TABLE 22 SURVEY RESPONDENTS’ COMMENTS City of Auburn Personal, we’re completely capable City of Bellevue Internal: We’re understaffed, so extra work needed for interjurisdictional is low priority. City of Marysville Financial City of Redmond -If we can get communications that will allow another agency to implement certain functions, and we can maintain our own com- munications abilities, this is a no-brainer. -Trust that outside agency could do a better job of operating the signal. -Legal/liability: Sure, we need assurance that the other agency will secure the system and not do certain things, but that should be straightforward. King County The technical barriers include non- compatible controllers and central system issues. Each agency must have a work station to view and operate their signals that are connected into another jurisdiction’s central system. There is a lack of true center-to-center capabilities. Pierce County We currently share signal timing at one interchange with WSDOT. In the past, this sharing of time of day plans has always worked well for both parties. Since the cor- ridor affected is interconnected and oper- ated by the County, the County informs WSDOT of the timing to be used at their interchange signals to promote progression on the corridor while minimizing queues on the ramp. Source: Puget Sound Regional ITS Implementation Plan (2008). COMPETING REqUIREMENTS RELATED TO SIGNAL TIMING More Frequently Mentioned Conflicts• Pedestrian safety and access – Inter-jurisdictional issues – Emergency vehicle preemption – Less Frequently Mentioned Conflicts• Bicycle access at actuated phases – Neighborhood access – Transit priority – Concerning emergency vehicle preemption, Nelson and Bullock (2000) indicated that significant delays may occur when a considerable number of preemption calls are experienced. The example of signal timing guidelines provided in Appendix B prioritizes and qualifies a number of competing objectives. STATE OF THE PRACTICE AND CONCLUSIONS The 2007 Traffic Signal Report Card (NTOC 2007) summa- rizes resource limitations for signal timing as follows: Findings from the 417 agencies that collectively account for ownership of approximately 45 percent of the nation’s 272,000 traffic signals indicate that resource and management constraints limit the effectiveness of traffic signal operations. As noted throughout this report, agencies are forced into difficult choices about how to spend their limited resources. For many agencies, this simply means addressing the most critical issues on a daily basis. A proactive, integrated program management approach that includes the principles of continuous improvement, asset life-cycle costs and resource allocation for traffic signal operations is seldom seen as an option. Specific conclusions include the following: Precedence graphs such as those discussed in chapter • five provide a start in understanding the interaction of controller parameters. Further research should include developing priorities for parameter selection under the range of traffic demands expected at the intersection. Current guidance on the technique for determining • network boundaries appears to be inconsistent and may be conflicting. Further research to improve guid- ance in this area is desirable. Further research on methodologies for identifying the • appropriate number of timing plans and methodology to use a congestion-related measure such as occupancy would forward the state of the practice. Further research in developing measures for evaluating • signal timing performance might include additional software in the signal controller and archived data user service technology.

45 CHAPTER ELEVEN METHODS FOR INCREASING RESOURCES FOR RETIMING Some agencies make a significant effort to relate benefits to public expenditures in a proactive way. The following textbox shows an excerpt from the Delaware DOT website that performs this function (Delaware DOT website 2009). DESCRIPTION: Traffic signal installation, upgrade, or reconstruction at nine intersections: New Castle County: SR 2 & Albertson Boulevard; US 13 & Boulden Boulevard; US 13 & Memo- rial Drive; US 202 & Righter Parkway; and, Mill Creek Road & McKennans Church Road. Kent County: US 13 & Lepore Road; US 13 & SR 42; and, Frederica Fire Signal. Sussex County: SR 1 & West Way Drive. Cost: $1.5 million. Justification: Improvements will improve traffic flow, vehicular safety, pedestrian safety, and the reliability of the system. Proj- ects were initiated through a public complaint, through a study completed by our Traffic Studies Section, or through a request from our Signal Maintenance Section. Each intersection has a different scope of work, involving new signal installations, reconstruction of aging infrastructure, pedestrian signal enhancements, upgrade of curb ramps to current ADA standards, modification of left-turn phasing, and lengthening of left-turn storage bays. The survey conducted under Gordon and Braud (2009) indicated that 29% of the agencies maintained websites to communicate the functions and/or benefits of signal tim- ing and the role of the responsible agencies; 14% used other means of communication with the public. The in-depth case studies conducted under this project showed that relatively few agencies proactively communicate with the public. Many studies have shown that retiming traffic signals pro- vides significant benefits in terms of reduced delay, fuel consumption, and emissions, and that this technique is often underutilized. Examples of such reports include Trans- portation Infrastructure (GAO1994), and reports on the Fuel-Efficient Traffic Signal Management (FETSIM) pro- gram (Skabardonis 1994; Chien et al. 2006; Sunkari 2004). Although traffic signal equipment has improved since the earlier reports were issued, in most cases, and with the use of improved equipment and methodologies, periodic signal retiming can provide significant benefits. This chapter dis- cusses three methods that have been used to increase aware- ness of the benefits of signal retiming. RAISE AWARENESS OF BENEFITS On the national level, the annual Urban Mobility Report (Schrank and Lomax 2007) points to the extensiveness of congestion and its penalties, whereas the National Traffic Signal Report Card (2007) describes the level of deficien- cies in traffic signal timing and operation. These documents are widely reported by the press and media. Some states have implemented publicity campaigns. An example, provided by Maze et al. (1990), is described here. A publicity campaign was initiated to improve the public’s awareness of its receipt of restitution and possible energy savings through the signal improvements. Named “Iowa Signals Go” the campaign included a press conference, press packets sent to the media, a table-top display, a booklet, radio and television public service announcements, a 10-minute promotional videotape, newsletter articles, and discussions of the project on radio and television programs broadcast throughout the state. Programs in other states have not attempted to provide wide dissemination of information to the general public. Some local operating agencies raise public awareness by reporting improvements to the public through signal timing. For example, Figure 19 shows the first page of a five-page report posted on the Naperville, Illinois, website (Naperville 2007).

46 FIGURE 19 Comparison of before and after timing performance. (Source: Naperville 2007.) RESPOND TO FEEDBACk FROM THE PUBLIC Public support for increased resources for signal timing may be enhanced if the agency is perceived to be responsive to the public’s comments and complaints. Gordon and Braud (2009) indicated that 29% of the agencies conduct surveys and that 43% of the agencies have easy telephone access for this purpose. COOPERATE WITH OTHER AGENCIES In some cases agencies have found it expedient to cooperate to facilitate the planning and budgeting process. Cooperation may include traffic engineering technical assistance programs, shared databases, and regional databases. Estrella and Geor- gevich (2000) and Carvell et al. (1996) provide examples of this. MPOs sometimes take the lead in initiating budgeting for traffic signal programs. Figure 20 (SPC 2009) shows the cover letter for a solicitation from an MPO for traffic signal projects. FIGURE 20 Cover page for solicitation for traffic signal projects. (Source: SPC 2009.) Agencies in a region may share skills, experience, and training on either a formal or informal basis. Figure 21 pro- vides responses to the Puget Sound survey that addressed this issue.

47 FIGURE 21 Assistance sharing. (Source: Puget Sound Regional ITS Implementation Plan 2008.)

48 CHAPTER TWELVE CONCLUSIONS Currently used traffic controllers contain many signal • timing parameters and options, and considerable guid- ance exists for setting most of the parameters. Although the parameters, taken individually, are well understood by traffic engineers in responsible charge, because of their complexity it is sometimes difficult to anticipate how they will interact with each other. The available tools do not indicate how these parameters may be best collectively adjusted to improve performance with no adverse impact on safety. The development of timing plans for signal coordina-• tion is generally well understood by traffic engineers, and the tools to support this development are generally well suited to the task. Although several methodolo- gies to define the boundaries of the sections to be coor- dinated are available, they sometimes conflict. These methodologies appear to be underutilized by traffic engineers. There is little useful guidance for the identification of • the number of timing plans that may be effectively used, and the time periods for which deployment is appropri- ate. As a result, traffic engineers use techniques based on experience and a general review of volume profiles. Research to develop guidance based on volume, occu- pancy, and intersection saturation is appropriate. The existing guidance to establish the type of signal • control (pretimed, semi-actuated, fully actuated) as a function of traffic conditions is limited. Current guid- ance to establish the conditions for the use of fully actuated controllers under coordinated conditions is insufficient. Guidance for the use of volume-density operation is also insufficient. The cost of signal retiming is $3,700 or less per signal • for most agencies. Although most of the coordinated traffic systems in the United States operate on a time-of-day basis, alternatives exist for coordinated systems to respond to changing traffic conditions. Conventional traffic responsive systems• . These systems employ a relatively small number of system detectors. Although research has shown that these conventional systems are beneficial under some traffic conditions, lit- tle guidance is available for when this type of operation is best employed. In some cases, it is difficult to identify This synthesis of research examined the following: Signal timing policy, management, and planning• Network treatments• General signal timing and retiming considerations • Requirements for signal timing• Use of retiming methodologies and personnel • qualifications Methodologies for field implementation of timing • plans Personnel resources and cost for implementation of • signal timing plans Performance measures• Evaluation of signal timing performance• Barriers to implementation of signal retiming plans • Increasing resources for retiming• The synthesis concluded that although modern traffic controllers are complex and contain many options, traffic engineers in responsible charge were generally knowledge- able in their use. They generally emphasize pedestrian and vehicle safety as well as bandwidth maximization in prefer- ence to signal timing strategies that minimize vehicle delay. They are generally sensitive to a variety of competing needs for green time and respond appropriately. Respondents to several surveys cited lack of resources as the key obstacle to improved signal timing. The literature, surveys, and in-depth case studies indicate the following: Resource limitations are the most significant factor • contributing to suboptimal signal retiming. Most traffic engineers understand that competing pri-• orities among users and modes require servicing of all needs, with pedestrian and vehicle safety being the primary requirements. Although review and retiming of traffic signals at intervals of 30 months to 3 years is advised, most agencies exceed this interval somewhat but perform retiming within 5 years. Signal retiming may be conducted at periodic intervals or may be initi- ated as the result of such triggers as land use changes or observed changes in congestion patterns. In some cases detector data may indicate volume changes. There is little guidance for the use of detector data to systematically identify the need for retiming.

49 threshold parameters for the detector settings to be used for selecting timing plans. In other cases it is difficult to migrate timing plans into corresponding controller selection settings. These difficulties appear to discour- age the use of conventional traffic responsive operation. Traffic adaptive systems• . A number of traffic adaptive systems have been installed in the United States. Most of the agencies using these systems found performance to be generally improved and reported a higher level of improvement for operation in oversaturated condi- tions. Although they generally avoid the problems with traffic conventional responsive systems cited earlier, traffic adaptive systems are expensive, require a large number of detectors, and generally do not support equipment interchangeability among suppliers. Most agencies take considerable care in physically imple- menting timing plans in field controllers and checking to ensure that they are properly installed and functioning. Sim- ilarly, agencies generally expend considerable resources in fine-tuning timing plans and evaluating their performance. Performance evaluations often involve a labor-intensive process. Archived Data User Service (ADUS) coupled with Archived Data Management Systems (ADMS) technology is an emerging approach that may reduce the labor-intensive characteristics of evaluations and provide a basis for identi- fying retiming needs. Further research to address these gaps includes the following: Better understanding of the interactions of controller 1. timing plan parameters may include hardware-in-the- loop capabilities. Improved methodologies to establish the boundar-2. ies of coordinated traffic networks (sections) and to select intersections for isolated operation within or near coordinated networks. Research on the number of timing plans that are 3. appropriate, the best periods of operation, and the most appropriate timing plan transition techniques. This research would examine processes to use data developed by system traffic detectors to develop oper- ational profiles for such traffic scenarios as weekdays, weekends, holidays, and special events. Research related to Item 3 resulting in guidance to 4. practitioners can be used to establish the appropriate type of signal control (pretimed, semi-actuated, fully actuated) as a function of traffic conditions. Guidance would also be useful to establish traffic conditions for the use of fully actuated control with signal coordi- nation and for the appropriate use of volume-density operation. As discussed in chapter five, recent research high-5. lights a strong relationship between detection zone length and passage time settings. Additional research would enhance understanding of this relationship. Chapter five indicates that the use of very long cycle 6. lengths at isolated intersections is extensive and may be counter-productive. Research to further develop an understanding of this relationship would be useful. Because of its labor-intensive character, evalua-7. tion of system performance, although extremely important, is expensive as currently practiced. A more automated capability would not only reduce evaluation costs per se, but will also go far toward identifying when retiming is warranted. Although ADUS coupled with ADMS provides the general structure for data analysis, relatively few imple- mentations have been developed for signal systems, and these are not marketed in a convenient form for agencies to use. Similarly, some controllers have incorporated software that enhances the capability to provide measures that are useful for signal tim- ing evaluation. Research would be useful to review existing work to date, and to develop the design requirements for a cost-effective approach that may be widely deployed to provide common evaluation parameters such as delay.

50 REFERENCES 2008 Annual Evaluation Report, Highway Safety Improve- ment Program, Five Percent Report, New York State Department of Transportation, Albany, Aug. 2008. 2010 Highway Capacity Manual [Online]. Available: http:// prj.kittelson.com/hcm [accessed Sep. 7, 2009]. Abbas, M., N. Chaudhari, H. Charara, and Y. su Jung, Foun- dation for Self-Tuning Closed-Loop Progression System, Report FHWA/TX-05/0-4729-1, Texas Transportation Institute, College Station, 2004. Abbas, M., G. Pesti, N.A. Chaudhari, and L. Pengfei, Con- figuration Procedure of Traffic-Responsive Plan Selec- tion on an Arterial Network. Presented at the 87th Annual Meeting of the Transportation Research Board, Wash- ington, D.C., Jan. 2008. ADUS, Archived Data User Service [Online]. Available: www.f hwa.dot.gov/policy/ohpi /t ravel /adus.cfm [accessed June 15, 2009]. Average Accident Rates, New York State Department of Transportation [Online]. Available: www.nysdot.gov/ divisions /operating/osss /highway/accident-rates [accessed May 22, 2009]. Balke, K.N., R.J. Engelbrecht, S.R. Sunkari, and H. Charara, TTI’s Hardware-in-the Loop Traffic Signal Controller Evaluation System. Report FHWA/TX-05/5-1752-01-1, Texas Transportation Institute, College Station, 2005. Balke, K.N. and C. Herrick, Potential Measures of Assessing Signal Timing Performance Using Existing Technolo- gies, Report 0-4422-1, Texas Transportation Institute, College Station, July 2004. Balke, K., H. Charara, and R. Parker, Development of a Traf- fic Signal Performance Measurement System, Report 0-4422-2, Texas Transportation Institute, College Sta- tion, May 2005. Beaird, S., T. Urbanik, and D.M. Bullock, “Traffic Signal Phase Truncation in Event of Traffic Flow Restriction,” Transportation Research Record: Journal of the Trans- portation Research Board, No. 1978, 2006, pp. 87–94. Benefits of Retiming Traffic Signals. Institute of Transporta- tion Engineers, Washington, D.C., 2005. Berkow, M., M. Wolfe, C. M. Monsere, and R. L .Bertini. “Using Signal System Data and Buses as Probe Vehicles to Define the Congested Regime on Arterials,” Trans- portation Research Record: Journal of the Transporta- tion Research Board, No. 2129, Transportation Research Board of the National Academies, Washington, D.C., 2009, pp. 81–89. Bonneson, J.A. and P.T. McCoy, Manual of Traffic Detector Design, 1st ed., Institute of Transportation Engineers, Washington, D.C., 1993. Bonneson, J., S. Sunkari, and M. Pratt, Traffic Signal Opera- tions Handbook, Report FHWA/TX-09/0-5629-PI, Texas Transportation Institute, College Station, Mar. 2009. Bretherton, D., G. Bowem, and K. Wood, Effective Urban Traffic Management and Control: Recent Developments in SCOOT, Presented at the 82nd Annual Meeting of the Transportation Research Board, Washington, D.C., Jan. 12–16, 2003. Bullock, D.M., J.R. Sturdevant, and C.M. Day, “Signalized Intersection Performance Measures for Operations Decision-Making,” ITE Journal, Aug. 2008, pp. 20–23. Bullock, D. and A. Cantarella, “A Real-Time Simulation Environment for Evaluating Traffic Signal Systems,” Transportation Research Record, No. 1634, Transporta- tion Research Board of the National Academies, Wash- ington, D.C., 1998, pp. 130–135. Carvell, J.D., E.J. Seymour, C.H. Walters, and T.R. Starr, Dallas Area-Wide Intelligent Transportation System Plan, Texas Transportation Institute, Texas Department of Transportation, Federal Highway Administration, College Station, 1996. Chang, E.C., “Guidelines for Actuated Controllers in Coor- dinated Systems,” Transportation Research Record 1554, Transportation Research Board, National Research Council, Washington, D.C., 1996, pp. 61–73. Chien, S.I., K. Kim, and J. Daniel, “Cost and Benefit Analy- sis for Optimized Signal Timing-Case Study: New Jer- sey Route 23,” ITE Journal, Oct. 2006, pp. 37–41. Clark, Z.T., S.M.L. Hubbard, and D.M. Bullock, Quantita- tive Measurement Procedures for Pedestrian Service at Signalized Intersections: Case Study at Skewed Inter- section, Presented at the 85th Annual Meeting of the Transportation Research Board, Washington, D.C., Jan. 22–26, 2006. Congestion Management System. Knoxville Regional Trans- portation Planning Organization [Online]. Available: www.knoxtrans.org/plans/cmsdoc/chap2.htm [accessed Mar. 3, 2009]. Day, C., E. Smaglik, D.M. Bullock, and J. Sturdevant, “Quantitative Evaluation of Actuated Versus Nonactu- ated Coordinated Phases,” Transportation Research Record: Journal of the Transportation Research Board, No. 2080, Transportation Research Board of the National Academies, Washington, D.C., 2008, pp. 8–21.

51 Day, C.M., J.R. Sturdevant, and D.M. Bullock, “Outcome Oriented Performance Measures for Signalized Arterial Capacity Management,” TRB Paper 10-0008, Presented at the 89th Annual Meeting of the Transportation Research Board, Washington, D.C., Jan. 10–14, 2010. Delaware Department of Transportation website [Online]. Available: www.deldot.gov/information/projects/ recovery/pages/2009_traffic_sig/index.shtml [accessed June 8, 2009]. Denney, R.W., E.J. Curtis, and L. Head, “Long Green Times and Cycles at Congested Traffic Signals,” Transporta- tion Research Record: Journal of the Transportation Research Board, No. 2128, Transportation Research Board of the National Academies, Washington, D.C., 2009, pp. 1–10. Emission Facts: Average Carbon Dioxide Emissions Result- ing from Gasoline and Diesel Fuel, Environmental Pro- tection Agency, Washington, D.C., 2005 [Online]. Available: www.epa.gov/OMS/climate/420f05001.htm [accessed May 26, 2009]. Estrella, A. and J. Georgevich, “Putting it all Together for the 21st Century, Regional Efforts for Improving Arterial Traffic Operations and Management,” In Compendium of Papers, District 6 Annual Meeting, Institute of Trans- portation Engineers, Washington, D.C., 2000. Fontaine, M.D. and B.L. Smith, “Probe-Based Traffic Moni- toring Systems with Wireless Location Technology,” Transportation Research Record: Journal of the Trans- portation Research Board, No. 1925, Transportation Research Board of the National Academies, Washing- ton, D.C., 2005, pp. 1–11. Gerken, J., Aging Traffic Signal Systems Timing and Coordi- nation, 2008 [Online]. Available: http://haginconsult- i ngser v ices .com / Dow n loads / Gerken _ Agi ng _ Infrastructure-Traffic_Signal_Timing_and_Coordina- tion_Presentation.pdf [accessed May 26, 2009]. Ghaman, R.S., ACS_Lite FHWA Adaptive Signal Control Systems, Traffic Signal & Operations Working Group, Mar. 26, 2008, Washington, D.C. [Online]. Available: www.mwcog.org/uploads /committee-documents / aV5eVltZ20080910135031.ppt [accessed Mar. 25, 2009]. Giblin, J.M. and W.H. Kraft, Traffic Control Systems Oper- ations—Installation, Management and Maintenance, Institute of Transportation Engineers, Washington, D.C., 2000. Gordon, R.L., “A Technique for Control of Traffic at Critical Intersections,” Transportation Science, Vol. 3, No. 4, Nov. 1969. Gordon, R. and C. Braud, Traffic Signal Operations and Maintenance Staffing Guidelines, Report FHWA-HOP- 09-006, Federal Highway Administration, Washington, D.C., 2009. Gordon, R.L., R.A. Reiss, H. Haenel, E.R. French, A. Mohad- des, and R. Wolcott, Traffic Control Systems Handbook, Report FHWA-SA-95-032, Federal Highway Adminis- tration, Washington, D.C., Feb. 1996. Gordon, R.L., NCHRP Synthesis 307: Systems Engineering Processes for Developing Traffic Signal Systems, Trans- portation Research Board of the National Academies, Washington, D.C., 2003. Gordon, R.L. and W. Tighe, Traffic Control Systems Hand- book, FHWA Report FHWA- HOP-06-006, Dunn Engi- neering Associates, Oct. 2005. Head, L., D. Gettman, D. Bullock, and T. Urbanik II, “Mod- eling Traffic Signal Operations Using Precedence Graphs,” Transportation Research Record: Journal of the Transportation Research Board, No. 2035, Transpor- tation Research Board of the National Academies, Wash- ington, D.C., 2007, pp. 10–18. Henry, D., Signal Timing on a Shoestring, Sabra Wang & Associates, Inc., Report FHWA-HOP-07-006, Federal Highway Administration, Washington, D.C., Mar. 2005. Highway Capacity Manual, Transportation Research Board, National Research Council, Washington, D.C., 2000. Highway Economic Requirements System-State Version Technical Guide, Federal Highway Administration, Washington, D.C., 2002. Hua, J., N. Gutierrez, I. Banerjee, F. Markowitz, and D.R. Ragland, San Francisco PedSafeII Project Outcomes and Lessons Learned, In TRB 88th Annual Meeting Com- pendium of Papers, Washington, D.C., Jan. 11–15, 2009, 18 pp. Hubbard, S.M.L., D.M. Bullock, and C.M. Day, Opportuni- ties to Leverage Existing Infrastructure to Integrate Real-Time Pedestrian Performance Measures into Traf- fic Signal System Infrastructure, Presented at the 87th Annual Meeting of the Transportation Research Board, Jan. 13–17, 2008. Intelligent Transportation Systems for Traffic Signal Con- trol, U.S. Department of Transportation, Washington, D.C. [Online]. Available: www.its.dot.gov/jpodocs/ repts_te /14321_f iles /a1019-TSC_digital_n3.pdf [accessed May 26, 2009]. Jeannotte, K., A. Chandra, V. Alexiadis, and A. Skabardo- nis, Traffic Analysis Toolbox, Volume II, Decision Sup- port Methodology for Selecting Traffic Analysis Tools, Report AFHWA-HRT-04-039, Federal Highway Admin- istration, Washington, D.C., 2004.

52 Kell, J.H. and I.J. Fullerton, Manual of Traffic Signal Design, 2nd ed., Institute of Transportation Engineers, Washing- ton, D.C., 1998. Klein, L.A., M.K. Mills, and D.R. Gibson, Traffic Detector Handbook: Third Edition-Volume 1, Report FHWA- HRT-06-108, Federal Highway Administration, Wash- ington, D.C., Oct. 2006. Koonce, P., L. Rodegerdts, S. Quayle, S. Beaird, C. Braud, J. Bonneson, P. Tarnoff, and T. Urbanik, Traffic Signal Timing Manual, Kittelson & Associates, Inc., Report FHWA-HOP-08-024, Federal Highway Administration, Washington, D.C., June 2008. Kruesi, F.E., Departmental Guidance for the Valuation of Travel Time in Economic Analysis, U.S. Department of Transportation, Washington, D.C., Apr. 9, 1997. Kyte, A.M., A. Abdel-Rahim, M.P. Dixon, J. Li, and T. Urbanik II, Distance Gap as a Detection Design and Operations Tool, Presented at the 88th Annual Meeting of the Transportation Research Board, Washington, D.C., Jan. 11–15, 2009. Li, H., An Approach on Network Traffic Signal Control Under the Real Time and Oversaturated Flow Condition, Presented at the 81st Annual Meeting of the Transporta- tion Research Board, Washington, D.C., Jan. 13–17, 2002. Lieberman, E.B., J. Chang, and E.S. Prassas, “Formulation of Real-Time Control Policy for Oversaturated Arteri- als,” Transportation Research Record: Journal of the Transportation Research Board, No. 1727, Transporta- tion Research Board of the National Academies, Wash- ington, D.C., 2000, pp. 77–88. Liu, H.X. and W. Ma, Real-time Performance Measurement System for Arterial Traffic Signals, Presented at the 87th Annual Meeting of the Transportation Research Board, Washington, D.C., Jan. 13–17, 2007. Maher, M.J., “Estimating the Turning Flows at a Junction: A Comparison of Three Models,” Traffic Engineering and Control, Vol. 25, No. 11, 2007, pp. 19–22. Manual on Uniform Traffic Control Devices, Federal High- way Administration, Washington, D.C., 2009. Manual on Uniform Traffic Studies, Florida Department of Transportation, Tallahassee, 2000. Maze, T.H., N. Hawkins, J. Graham, and M. Elahi, “Iowa’s Statewide Traffic Signal Improvement Program,” ITE Journal, May 1990. Monsere, C.M., R.L. Bertini, A. Breakstone, C. Bonner, P. Bosa, Z. Horowitz, and D. Houssaye, Update and Enhancement of ODOT’s Crash Reduction Factors, Report FHWA-OR-DR-06-11, Oregon Department of Transportation, Salem, and Federal Highway Adminis- tration, Washington, D.C., June, 2006. MOVES (Motor Vehicle Emission Simulator), U.S. Environ- mental Protection Agency, Washington, D.C., 2009 [Online]. Available: www.epa.gov/otaq/models/moves/ index.htm [accessed Nov. 2, 2009]. MUTCD 2009, Manual on Uniform Traffic Control Devices, Federal Highway Administration, Washington, D.C., 2009. Naperville, Illinois website, 2007 [Online]. Available: www.naperville.il.us/emplibrary/SCAT%20Results2. pdf [accessed June 8, 2009]. National ITS Architecture Version 6.1, U.S. Department of Transportation, Washington, D.C., 2009 [Online]. Avail- able: www.iteris.com/itsarch [accessed June 8, 2009]. National Traffic Signal Report Card—Technical Report, National Transportation Operations Coalition, Federal Highway Admnistration, Washington, D.C., 2007 [Online]. Available: www.ite.org/REPORTCARD/ technical_report%20final.pdf [accessed June 8, 2009]. NEMA Standards Publication TS 2-2003 v02.06, National Electrical Manufacturers Association, Rosslyn, Va., 2003. Nelson, E., M. Abbas, G. Shoup, D. Bullock, and R. Gal- lagher, Development of Closed Loop System Evaluation Procedures, Report FHWA/IN/TRP-2000/5, Purdue University, West Lafayette, Ind., Dec. 2000. Nelson, E. and D. Bullock, “Impact Evaluation of Emer- gency Vehicle Preemption on Signalized Corridor Oper- ation,” Transportation Research Record: Journal of the Transportation Research Board, No. 1727, Transporta- tion Research Board of the National Academies, Wash- ington, D.C., 2000, pp. 1–11. National Traffic Signal Report Card, Technical Report, 2007, online 9/7/2010 http://www.ite.org/reportcard/techni- cal_report%20final.pdfNichols, A., and D. Bullock, Design Guidelines for Deploying Closed Loop Systems, Report FHWA/IN/JTRP-2001/11, Purdue University, West Lafayette, Ind., Nov. 2001. NYSDOT—New York State Department of Transportation Average Intersection Accident Rates, 2009 [Online]. Available: www.nysdot.gov/divisions/operating/osss/ highway-repository/TABLE_II_2008.pdf [accessed Aug. 15, 2009]. Orcutt, F.L., The Traffic Signal Book, Prentice–Hall, Engle- wood Cliffs, N.J., 1993. Papacostas, C.S. and P.D. Prevedouros, Transportation Engi- neering and Planning, Prentice–Hall, Englewood Cliffs, N.J., 2009.

53 Park, B., P. Santra, I. Yun, and D. Lee, Optimization of Time-of-Day Breakpoints for Better Traffic Signal Con- trol, Presented at the 83rd Annual Meeting of the Trans- portation Research Board, Washington, D.C., Jan. 11–15, 2004. Park, B., I. Yun, and M. Best, Evaluation of Pre-Emption and Transition Strategies for Northern Virginia Smart Travel Signal Systems (NVSTSS), Report No. VTRC 08-CR9, Virginia Transportation Research Council, Charlottesville, Feb. 2008. Pedersen, N.J. and D.R. Samdahl, NCHRP Report 255: Highway Traffic Data for Urbanized Area Project Plan- ning and Design, Transportation Research Board, National Research Council, Washington, D.C., Dec. 1982. Pline, J., Traffic Control Device Handbook 2001, Institute of Transportation Engineers, Washington, D.C., 2001. Puget Sound Regional ITS Implementation Plan, Puget Sound Regional Council, HDR, IBI Group, Aug. 7, 2008. Robertson, H.D., Ed., Manual of Transportation Engineer- ing Studies, Institute of Transportation Engineers, Wash- ington, D.C., 2000. Row, S. and P.J. Tarnoff, USDOT, FHWA Traffic Signal Training Assessment—Summary Report, Institute of Transportation Engineers, University of Maryland, Col- lege Park, June 29, 2005. Schrank, D. and T. Lomax, The 2007 Urban Mobility Report, Texas Transportation Institute, College Station, 2007. Selinger, M. and L. Schmidt, Adaptive Traffic Control Sys- tems in the United States, HDR Engineering, Inc., Sep. 2009. Sharma, A., D.M. Bullock, and J. Bonneson, “Input-Output and Hybrid Techniques for Real-Time Prediction of Delay and Maximum Queue Length at a Signalized Intersection,” Transportation Research Record: Journal of the Transportation Research Board, No. 2035, 2007, Transportation Research Board of the National Acade- mies, Washington, D.C., pp. 88–96. Sharma, A. and D.M. Bullock, “Field Evaluation of Alterna- tive Real-Time Methods for Estimating Delay at Signal- ized Intersections,” Proc., 10th International Conference on Applications of Advanced Technologies in Transpor- tation, Athens, Greece, May 27–31, 2008. Shaw, T., NCHRP Synthesis 311: Performance Measures of Operational Effectiveness for Highway Segments and Systems, Transportation Research Board, National Acad- emy of Sciences, Washington, D.C., 2003. Signal Timing Process Final Report, Sabra, Wang & Asso- ciates, Inc., Dec. 30, 2003 [Online]. Available: http:// ops.fhwa.dot.gov/arterial_mgmt/rpt/sig_tim_proc/ index.htm [accessed June 28, 2009]. Skabardonis, A., Evaluation of the Fuel-Efficient Traffic Sig- nal Management (FETSIM) Program: 1983:1993, Final Report, Report UCB-ITS-RR-94-11, University of Cali- fornia, Berkeley, 1994. Skabardonis, A., Progression Through a Series of Intersec- tions with Traffic Actuated Controllers, Volume 2 User’s Guide, Report FHWA RD-89-133, Federal Highway Administration, Washington, D.C., Oct. 1998. Smadi, A., and S. Birst, “Use of Hardware-in-the-Loop Traf- fic Simulation in a Virtual Environment,” Ninth Interna- tional Conference on Applications of Advanced Technology in Transportation, American Society of Civil Engineers, Reston, Va., 2006. Smaglik, E., D.M. Bullock, T. Urbanik, and D. Bryant, “Evaluation of Flow Based Traffic Signal Control Using Advanced Detection Concepts,” Transportation Research Record: Journal of the Transportation Research Board, No. 1978, Transportation Research Board of the National Academies, Washington, D.C., 2006, pp. 25–33. Smaglik E.J., A. Sharma, D.M. Bullock, J.R. Sturdevant, and G. Duncan, “Event-Based Data Collection for Gen- erating Actuated Controller Performance Measures,” Transportation Research Record: Journal of the Trans- portation Research Board, No. 2035, Transportation Research Board of the National Academies, Washing- ton, D.C., 2007, pp. 97–106. SPC, Request for Projects-Regional Traffic Signal Program, 2009 Cycle, Southwestern Pennsylvania Commission, Pittsburgh [Online]. Available: www.spcregion.org/pdf/ signals/rtsp_guideandapp.pdf [accessed June 28, 2009]. Stevanovic, A., NCHRP Synthesis 403: Adaptive Traffic Control Systems: Domestic and Foreign State of the Practice, Transportation Research Board of the National Academies, Washington, D.C., 2010, 104 pp. Sunkari, S., “The Benefits of Retiming Traffic Signals,” ITE Journal, Apr. 2004. Tarnoff, P.J. and J. Ordonez, Signal Timing Practices and Pro- cedures, Institute of Transportation Engineers, Federal Highway Administration, Washington, D.C., Mar. 2004. Timbrook, P., J. Trombly, and A. Gupta, Cross-Jurisdictional Signal Coordination Case Studies, Science Applications International Corporation, Report FHWA-OP-02-034, Federal Highway Administration, Washington, D.C., 2002. “Traffic Signal Timing and Operations Survey,” ITE Jour- nal, Vol. 79, No. 4, Apr. 2009. Transportation Infrastructure: Benefits of Traffic Control Signal Systems Are Not Being Fully Realized, Report

54 GAO/RCED-94-105, U.S. General Accounting Office, Washington, D.C., 1994. Urbanik, T., Policy Based Traffic Signal Control, In Com- pendium of Papers, Institute of Transportation Engineers 2000 District Annual Meeting, Washington, D.C. Using Existing Loops at Signalized Intersections for Traffic Counts: An ITE Informational Report, ITE Traffic Engi- neering Council Committee TENC-103-02, Institute of Transportation Engineers, Washington, D.C., Feb. 2008. Wasson, J.S., P.E. Sturdevant, and D. Bullock, “Real Time Travel Time Estimates Using Media Access Control Address Matching,” ITE Journal, Vol. 78, No. 6, June 2008, pp. 20–23. Wilson, C.J., G. Millar, and R. Tudge, Microsimulation Eval- uation of Benefits of SCATS-Coordinated Traffic Con- trol Signals, Presented at the 85th Annual Meeting of the Transportation Research Board, Washington, D.C., Jan. 22–26, 2006. Yun, I., M. Best, and B. Park, “Evaluation of Adaptive Maxi- mum Feature in Actuated traffic Controller: Hardware- in-the Loop Simulation,” Transportation Research Record: Journal of the Transportation Research Board, No. 2035, Transportation Research Board of the National Academies, Washington, D.C., 2007.

55 GLOSSARY Accident modification factors—A means of quantifying crash reductions associated with safety improvements. Actuated signal control—Timing intervals are called and extended in response to vehicle detections. Bandwidth—The maximum amount of green time for a des- ignated direction as it passes through a corridor at an assumed constant speed, typically measured in seconds. Bluetooth—An open wireless protocol for exchanging data over short distances. Centrally controlled system—A system that provides signals to the intersection traffic controller to directly control each phase or each interval of the signal cycle. Clearance time—The time in seconds between signal phases to transition between normally conflicting movements. Closed loop system—A three-level distributed signal con- trol system. It includes a central computer, field master controllers, and local controllers. Timing plans may be downloaded to field controllers and supervised at the cen- tral site. Controller or field controller—The devices that operate traf- fic signals by controlling the sequence and duration of the indications displayed. Coordination—The ability to synchronize multiple intersec- tions to enhance the traffic flow of one or more direc- tional movements in a system. Crash reduction factors—See “accident modification factors.” Cycle length—The time needed for a complete sequence of signal indications. Detection mode—Detectors may operate in the presence mode (continuous detection of vehicles within the sens- ing field of the detector) or the passage mode (detector indicates a pulse). Fine tuning—Adjustments to the basic signal retiming set- tings made by observing the interaction of vehicles with the timing parameters. Fully actuated signal—Signal timing on all approaches is determined by traffic actuation. GPS—Global Positioning System. Green band—The width of the band (in seconds) that indi- cates the amount of green time that traffic can flow with a selected speed through a number of coordinated intersections. Interval—The time duration when traffic signal indications do not change. An actuated traffic signal controller also has timing intervals (minimum green, passage time) that determine the length of the green interval. Isolated intersection—An intersection whose approach traf- fic flow is not significantly influenced by the signal tim- ing of upstream intersections; that is, vehicles do not arrive at the intersection in platoons. Isolated intersec- tions (intersections that are not coordinated with adjacent intersections) may be present within a network because they operate more effectively under certain conditions such as oversaturation. Level of service—Defined in terms of control delay. Control delay is the total elapsed time from when a vehicle joins a queue until it departs from the stopped position at the head of the queue. Maximum green (time)—The maximum length of time that a phase can be green in the presence of a conflicting call. Minimum green (time)—The shortest time for which the green indication will be given to the approach or approaches. MUTCD—Manual on Uniform Traffic Control Devices. Network—A set of signals whose signal timing is coordi- nated. Sometimes called a section the signals operate in the same control mode (e.g., time of day). New signal timing plans are provided to all signals in the section simultaneously. NEMA—National Electrical Manufacturers Association. NTCIP—National Transportation Communications for ITS Protocol. Offset—The time relationship between a defined point in the green interval (such as the start of green) and a defined master clock reference. Passage time—See vehicle extension time. PDA—Personal digital assistant. Pedestrian walk—A signal indication that allows pedestri- ans to begin to cross an intersection. Phase—A set of signal displays that control a specific com- bination of vehicular and pedestrian movements. Preemption—The transfer of normal operation of a traffic control signal to a special control mode of operation that facilitates the passage of trains and emergency vehicles by prohibiting conflicting operation of normal traffic flows.

56 Progression—Coordination of traffic signal timings that results in a high proportion of vehicles arriving during the green period (good signal progression) while travel- ing from one signal to the next. Red clearance—A clearance interval that may follow the yellow interval during which the terminating phase and all other phases display red. Ring—Set of phases that may operate one after another. A ring and barrier diagram is often used to depict these relationships. Saturation—The traffic flow condition when demand equals or exceeds capacity. When demand exceeds capacity it is sometimes called oversaturation. Section—See “network.” Semi-actuated signal—Signal for which timing on one or more, but not on all, approaches is determined by traffic actuation. Spillback—Condition in which a queue from a downstream intersection uses up all the space on a link and prevents vehicles from entering the upstream intersection on the green. Split —The time assigned to a phase (the greater of the green and yellow plus all red or the pedestrian walk and clear- ance times) during coordinated operation; may be expressed in seconds or percent of cycle. Traffic adaptive control—Real-time implementation of sig- nal timing plans for a network. Implementation periods may be on a cyclic basis or may be changed within a cycle. In addition, changes may be introduced based on mea- surements made over several cycles. Traffic management center (TMC) or traffic operations cen- ter (TOC)—A central control facility used to manage traffic signal systems. It may be combined with other traffic management functions such as freeway manage- ment systems or police operations management facilities. Traffic-responsive control—Selection of stored timing plans based on measurements of current traffic conditions. Vol- ume and occupancy are the most commonly used measurements. TMC—Transportation management center or traffic man- agement center. Vehicle green—The duration of the green indication for a given movement. Vehicle extension time—A preset time interval extending the green time for each vehicle detection in an actuated controller. It may be reduced using the volume-density features of a controller. Volume-density control—A form of actuated control in which the initial green interval and the vehicle extension interval are varied as a function of traffic conditions and time. Volume-to-capacity ratio—The ratio of the volume approach- ing an intersection to the approach capacity. Wireline—Transmission of signals using a physical form of interconnection cable such as copper cable or fiber optic cable.

57 APPENDIx A IN-DEPTH CASE STUDIES The synthesis employed the results of four surveys conducted under the following projects, as well as a set of case studies performed under this project. National Traffic Signal Report Card• (National Transportation Operations Coalition 2007) Puget Sound Regional ITS Implementation Plan• (Puget Sound Regional Council et al. 2008) Signal Timing Practices and Procedures• (Tarnoff and Ordonez 2004) Traffic Signal Operations and Maintenance Staffing • Guidelines (Gordon and Braud 2009) The case studies conducted under this project were intended to obtain in-depth information on traffic signal timing practices. Thus the questionnaire not only solicited statistical information, but also provided an opportunity for the type of anecdotal information leading to a better under- standing of the reasons for the choice of practices. Follow-up questions resulted in additional information. Seventeen case study solicitations resulted in seven in- depth responses. Two agencies provided documents that they had developed, and this material was incorporated into the synthesis. Some responses to the questionnaire suggested the fol- lowing set of follow-up questions to some of the participat- ing agencies: How did you select the sections for traffic responsive • operation? How do you establish the boundaries for the networks • (sections) for your coordination plans? How did you select the intersections to be placed on the • SCATS system? Requests for additional information resulted in Figure 19 in chapter 11 and Appendix B. Table A1 provides a summary of the responses to the case study questionnaire.

58 TABLE A1 SUMMARY OF CASE STUDY RESPONSES Agency a b c d e f g Case Study Question 1.2 Number of signals 535 90 720 1,300 151 850 3,000 a. Coordinated 403 87 360 1,270 148 775 2,250 b. Isolated 172 3 30 3 75 750 1.3 Type of Coordination a. Closed loop Yes Yes Yes Yes Yes b. Central Yes Yes Yes Yes c. Adaptive Yes Yes d. Time base Yes Yes Yes Yes Yes e. Other 2 Management 2.1 Agency procedures for a. TMC Operating Procedures Yes NA Yes No Yes No No b. Maintaining traffic signals Yes Yes Yes Yes Yes Yes Yes c. Retiming signals Yes No Yes IP Yes Yes No 2.2 Compatibility with Regional Architecture and Standards Yes NTCIP Yes Yes Yes No NTCIP 2.3 Documentation Yes Yes Yes No Yes No No 2.4 Review periods for operations and performance a. 1 year or less Yes b. 1 to 2 years Yes c. 2 to 3 years Yes d. 3 to 5 years Yes Yes Yes e. Over 5 years or not reviewed Yes 2.5 Intervals for phasing review 3 No defined As needed Note 1 5–10 yrs N As needed 2.6 Intervals for geometrics review 3 No defined As needed Note 1 N As needed 2.7 Retiming after geometric models Yes Yes As needed Yes Yes Yes Yes 2.8 Objectives a. Documented objective for signal timing No No Yes No Yes Yes Yes b. Describe objectives c. Describe conflicts d. Vehicle classes for preempt F PFA FA F F FA FAP TABLE A1 continues on next page

59 TABLE A1 continued Agency a b c d e f g e. ICM or interagency coordination objectives Yes Yes Yes Yes Yes Yes NA 2.9 % signals employing preempt 70 100 25 6 60 15 75 3. Basis for retiming 3.1 Need for retiming a. All signal at periodic intervals 1 1 1 b. Periodic using priority structure or roadway type 1 1 1 1 c. Field or CCTV observations 1 1 1 1 1 1 d. Detectors 1 1 1 e. How analyze data f. Other triggers Note 4 1 1 1 g. Other 1 1 3.2 Average retiming interval a. 1 year or less b. 1 to 2 years c. 2 to 3 years Yes d. 3 to 5 years Yes Yes Yes Yes Yes e. Over 5 years Yes 3.3 Does timing vary by street class Yes Yes Yes No Yes Yes No 3.4 If 3.3 positive explain approach 3.5 Approach for rephasing a. Critical lane anal Yes Yes Yes b. Separation of conflicting movements. Yes Yes Yes c. Volume warrants Yes Yes Yes Yes Yes Yes d. Pedestrian conflicts Yes Yes Yes Yes Yes e. Collision analysis Yes Yes Yes Yes Yes Yes Yes f. Consistency Yes Yes g. Green band Yes Yes Yes h. Left turn observations Yes Yes Yes Yes Yes I. Other 3.6 Number weekday timing plans 3 4 4 3-4 7 5 3-5 3.7 Other timing plans a. Incidents No No Yes Yes Yes Yes b. Saturdays Yes Yes Yes Yes Yes Yes Yes TABLE A1 continues on next page

60 Agency a b c d e f g c. Sundays Yes Yes Yes Yes Yes Yes Yes d. Holidays Yes Yes Yes Yes Yes Yes e. Weather Yes f. Special events Yes Yes Yes Yes Yes Yes g. Other Note 2 3.8 Total non-weekday plans 1 0-4 1-5 1-2 4-6 2-3 1-3 3.9 How determine time periods V, O V S, O V Note 3 3.11 Use ACS Lite No No No No No No No 3.12 Use traffic responsive NA No No NA Yes Yes Yes 3.13 Data for signal retiming a. Turning movements Yes Yes Yes Yes Yes Yes Yes b. ATC Yes Yes c. Traffic system data Yes d. Travel time/delay Yes Yes Yes Yes Yes e. Other 3.14 Pedestrian detectors Yes No Yes No Yes Yes Yes 3.15 Current walking speed 3.5-4 3.5-4 3 4 3.5 3.5-4 3-4 3.16 Future walking speed 3.5 3-4 3 3.5 3.5 3-3.5 3-3.5 3.17 Neighborhood traffic mgt. No No No No Yes No No 4 Retiming tools and personnel qualifications 4.2 Personnel qualifications person in charge of timing a. PE Yes Yes Yes Yes Yes Yes Yes b. PTOE Yes Yes Yes c. TSOS d. TOPS e. Courses Yes Yes Yes Yes Yes Yes f. Training suppliers Yes Yes Yes Yes Yes g. Years of experience 21 8 15 24 h. Other 5 Field implementation 5.1 Method a. Central control Yes b. Download from central computer Yes Yes Yes TABLE A1 continued TABLE A1 continues on next page

61 Agency a b c d e f g c. Download from laptop or desktop Yes Yes Yes Yes Yes d. Manually insert Yes Yes Yes Yes e. Other 5.2 Insurance of timing plans a. Computer checks only Yes? b. Verify in field Yes Yes Yes Yes Yes Yes Yes c. Other 5.3 Fine tune a. Do not fine tune b. Observe field parameters Yes Yes Yes Yes Yes Yes 1 c. CCTV Yes Yes Yes Yes d. Simulation Yes Yes e. Detector data Yes Yes f. Other g. Controller drift check needed Yes Yes Isolated Isolated 6 Resources 7.1 Performance measures for retiming a. Delay Yes Yes Yes Yes Yes Yes Yes b. Stops Yes Yes Yes Yes Yes Yes Yes c. Emissions and fuel Yes Yes Yes Yes d. Bandwidth maximization Yes Yes Yes Yes Yes Yes e. Variation in trip time or delay Yes Yes Yes Yes Yes f. Other 7.2 Strategies for saturation a. Balance delays Yes Yes Yes Yes b. Manage queues Yes Yes Yes Yes Yes c. Suspend coordination Yes Yes Yes d. Other Yes 8 Evaluation of performance 8.1 Do you evaluate Yes Yes Yes Yes Yes Yes Yes 8.2 Evaluation techniques used a. Travel time measurements Yes Yes Yes Yes Yes Yes Yes b. Intersection observations Yes Yes Yes Yes Yes Yes TABLE A1 continued TABLE A1 continues on next page

62 Agency a b c d e f g c. Accident records Yes Yes Yes Yes Yes d. Simulation Yes Yes Yes e. Traffic detectors Yes Yes Yes f. Other 8.3 Evaluation measures a. Delay Yes Yes Yes Yes Yes Yes b. Stops Yes Yes Yes Yes Yes Yes c. Accidents Yes Yes Yes Yes Yes d. Route travel time Yes Yes Yes Yes Yes Yes e. Reliability of travel time Yes Yes f. Emissions and fuel Yes Yes Yes Yes g. Other Notes for Response Table 1. Annually and when requested by a supervisory agency. 2. School start and end times. 3. By review of capacity analysis, hourly through and turning movement counts, and specific detector volumes and occupancies. Systems are then set with TRP plans and fine tuned and observed until optimized. Consultants doing this work are required to retime should the system break down within 5 years unless due to significant changes in access, geometry, or infrastructure. 4. New signals, major development or land use changes, geometric changes, major shifts in patterns. TABLE A1 continued

63 CASE STUDY qUESTIONNAIRE NCHRP SYNTHESIS PROJECT 20-05 TOPIC 40-10 TRAFFIC SIGNAL RETIMING PRACTICES IN THE UNITED STATES The Transportation Research Board’s National Cooperative Highway Research Program has commissioned a study on traffic signal retiming practices in the U.S. The goal of the research is to identify and provide information to practitioners regarding the techniques and resources appropriate for signal timing, to provide information on competing stakeholder objectives and to identify barriers to providing effective signal timing and the methods to overcome the barriers. As someone with experience in this area, we would appreciate your input on the subject. Please be assured that your responses will be kept in strictest confidence, to be aggregated with all other responses. 1. General 1.1 Name of agency and type (city, county, state) 1.2 Number of signals Coordinated signalsa. _____________________________________________ Isolated signals b. ________________________________________________ 1.3 What type of system coordination do you use (check all that apply): Closed loop (pretimed and/or traffic responsive)c. ______________________ Central (pretimed and/or traffic responsive second-to-second d. control provided by central computer) _______________________________ Adaptivee. ______________________________________________________ Time base coordinationf. __________________________________________ Other (please explain)g. ___________________________________________ 2. Management 2.1 Does your agency have documented procedures for: Traffic Management Center operating procedures a. _____________________ Maintaining traffic signalsb. ________________________________________ Retiming traffic signalsc. __________________________________________ 2.2 Is your traffic system compatible with the regional ITS architecture data flows and the standards that support them ___________________________________________________________ 2.3 Do you have a document that describes plans for capital and operational upgrades and for the implementation of signal retiming ___________________________________________________ 2.4 At which periods are signal operations and performance measures reviewed for the need for retiming and rephrasing ______________________________________________________ One year or lessa. ________________________________________________ One to two yearsb. _______________________________________________ Two to three yearsc. ______________________________________________ Three to five yearsd. ______________________________________________ Greater than 5 years or not reviewed e. _______________________________

64 2.5 At which intervals (if any) do you periodically review phasing _____________ 2.6 At which intervals (if any) do you periodically review the need for geometric modifications_________________ 2.7 Do you retime signals after performing geometric modifications_____ ______ 2.8 Objectives Does your agency have documented objectives relating to signal timinga. ____ If the response to 2.8.a above is positive please describe objectives or enclose documentation.b. Please describe conflicts in objectives which your agency has experienced in providing traffic flow efficiency c. with such other possible objectives as safety issues, bicycle safety and mobility, transit priority, emergency vehicle preempt, pedestrian issues, cross jurisdictional coordination, and neighborhood traffic management. If emergency vehicle preempt is provided, please identify the vehicle classes for which preempt is employed.d. Are Integrated Corridor Management (ICM) and/or inter-agency signal coordination among your agency’s e. objectives _____________________________________________________ 2.9 What percentage of your signals employ emergency vehicle preempt ________ 3. Basis for Signal Retiming 3.1 How do you establish the need for retiming (check all that apply): Retime all signals at periodic intervalsa. ______________________________ Retime at periodic intervals using a priority structure or roadway typeb. ____ Review of field observations or CCTV observationsc. ___________________ Review of traffic detector datad. _____________________________________ If 3.1.d above is positive please explain how you analyze the data.e. f.f. Other triggers (e.g., new traffic signals in area, new development, geometric changes) Other (please explain):g. 3.2 What is the average retiming interval employed by your agency 1 year or lessa. ___________________________________________________ From 1 to 2 yearsb. _______________________________________________ From 2 to 3 yearsc. _______________________________________________ From 3 to 5 yearsd. _______________________________________________ Greater than 5 years e. ____________________________________________ 3.3 Does the retiming interval vary by street classification ___________________ 3.4 If the response to 3.3 is positive, please explain your usual approach to this issue 3.5 Which techniques are employed for rephasing Critical lane analysisa. ____________________________________________ Maximum separation of conflicting movementsb. _______________________ Volume warrantsc. _______________________________________________

65 Avoidance of pedestrian conflictsd. __________________________________ Collision analysise. _______________________________________________ Retention of consistency (e.g., avoidance of phase changes as a function of time of day or timing plan) to avoid f. motorist confusion ______________________________________________ Green band emphasisg. ____________________________________________ Observing intersection during peaks for adequacy of left turn treatmenth. ___ Other (please explain):i. 3.6 How many normal weekday timing plans do you employ_____ 3.7 Which other timing plans do you provide (check all that apply): Incidents (special plan for diversion during incidents)a. __________________ Saturdaysb. _____________________________________________________ Sundaysc. ______________________________________________________ Holidaysd. ______________________________________________________ Weather (plans for snow or fog or other weather related effects) e. __________ Special eventsf. __________________________________________________ Other (please explain):g. 3.8 What is the total range in the number of non-weekday timing plans that your agency usually employs ________________ 3.9 How are the time periods for normal weekday plans determined (please explain): _______________________________________________________________ 3.10 Please identify systems, if any, used for adaptive operation (e.g. SCOOT, SCATS, RHODES, RT-TRACS, other) 3.11 Do you use ACS Lite ______________________ 3.12 Do you use system traffic responsive operation as provided by closed loop or central control (second by second control) systems. If so please explain the basis for establishing plan selection parameters. 3.13 Which of the following data do you use for data for signal retiming (check all that apply): Turning movement countsa. ________________________________________ Automatic traffic counters (if used, explain how used)b. __________________ Data collected by traffic system (if used, explain how used)c. _____________ Travel time/delay studiesd. _________________________________________ Other (if used, explain how used):e. 3.14 Do you use pedestrian detectors _____________ 3.15 What walking speed is currently used for signal timing __________________ 3.16 What walking speed will be used in the future for signal timing ____________ 3.17 Do you employ signal timing or phasing policies that limit traffic access to designated areas or provide neighborhood traffic management____________

66 4. Retiming Tools and Personnel qualifications 4.1 Which of the following retiming tools do you use: 4.1.1 Optimization SYNCHROa. ___________________________________________________ TRANSYT 7Fb. _________________________________________________ Passer IIc. ______________________________________________________ Passer IIId. _____________________________________________________ Passer IVe. _____________________________________________________ Manual time-space diagramsf. ______________________________________ Other (please explain):g. ___________________________________________ 4.1.2 Which of the following simulation techniques do you use: CORSIMa. _____________________________________________________ SimTrafficb. ____________________________________________________ Other (please identify):c. __________________________________________ 4.1.3 Which other techniques do you use: Highway capacity softwarea. _______________________________________ Other (please explain):b. ___________________________________________ 4.1.4 Personnel qualifications—Please identify the qualifications and experience for the person(s) in charge of signal retiming: PEa. ___________________________________________________________ Professional Traffic Operations Engineer (PTOE)b. _____________________ Traffic Operations Signal Specialist (TSOS)c. __________________________ Traffic Operations Practitioner Specialist (TOPS)d. _____________________ Courses in traffic engineeringe. _____________________________________ Training courses provided by software suppliersf. ______________________ Years of experience in signal retiming (please provide)g. _________________ Other (please explain):h. ___________________________________________ 5. Field Implementation of Timing Plans 5.1 What is the principal method used to implement timing plans in the field controller Use central control (second by second control) that times the signals for non-clearance intervals directly from a. the central computer ____________________________________________ Download from a central computerb. _________________________________ Download from a laptop or desktop computerc. ________________________ Manually insert timing plans into field controllerd. ______________________ Other (please explain):e. ___________________________________________

67 5.2 How do you insure that the plans delivered by the field controller are identical to those that were developed and proposed for use: Only use computer based checks (no field verification)a. _________________ Verify all plans using field observationsb. _____________________________ Other (please identify):c. __________________________________________ 5.3 How do you fine tune timing plans Do not fine tunea. ________________________________________________ Observe such field parameters as spillback from turning bays, slack time, spillback to upstream intersection b. and make adjustments accordingly _________________________________ Conduct observations such as in 5.3b by CCTVc. _______________________ Use of simulation modelsd. ________________________________________ Use of measures based on detector datae. _____________________________ Other (please explain):f. ___________________________________________ 5.4 Is it necessary to check internal controller clock drift_____ _______________ 6. Resources Appropriate for Retiming Agencies may perform some or all of the retiming tasks themselves or contract some or all of the tasks. The following is intended to identify the costs or time estimates for retiming on a per intersection basis. It may be more convenient for respon- dents to provide a total for each column rather than the components in the table. Task If performed with own staff If contracted Engineering hours Technician hours $ Selection of intersections for retiming Data collection Timing optimization including any office simulation for fine tuning Establishing timing plans in controller, field checking and field fine tuning After studies and report preparation Total Any other relevant resource information __________________________________ 7. Performance Measures for Timing Signals 7.1 Which measures do you use for retiming signals (check all that apply): Delaya. ________________________________________________________ Stopsb. _________________________________________________________ Emissions and fuel consumptionc. ___________________________________ Arterial bandwidth maximizationd. __________________________________ Variation in trip time or delaye. _____________________________________

68 Other (please explain)f. ___________________________________________ 7.2 Please identify the following which you use for timing of saturated phases or saturated intersections: Balance delays on all approachesa. ___________________________________ Manage queuesb. ________________________________________________ Suspend coordination and use alternative strategyc. _____________________ Other (please explain):d. ___________________________________________ 8. Evaluation of Signal Timing Performance 8.1 Does your agency perform evaluations of signal performance______________ 8.2 If your agency evaluates signal performance please identify the following techniques used: Vehicle travel time measurementsa. __________________________________ Intersection observations of queues, delay, saturation, etc.b. ______________ Review of accident recordsc. _______________________________________ Simulationd. ____________________________________________________ Measures provided by traffic system detectorse. ________________________ Other (please explain): f. __________________________________________ 8.3 Measures used for evaluation of field performance (please check all that apply): Delaya. ________________________________________________________ Stopsb. _________________________________________________________ Accidentsc. _____________________________________________________ Route travel timed. _______________________________________________ Reliability of travel timee. _________________________________________ Emissions and fuel consumptionf. ___________________________________ Other (please explain):g. ___________________________________________ 9. Management Issues 9.1 Please identify and comment on potential barriers to implementing improved signal timing performance that your agency may have experienced. Examples may include funding limitations, staffing/time limitations, lack of adequate tools or excessive complexity of tools, conflicting objectives. _______________________________________________________________ 9.2 Please identify and comment on activities that your agency may have undertaken to improve the resources available for signal timing. These may include raising public awareness or awareness on the part of decision makers on the efficiencies to be obtained through improved timing, encouraging public feedback, publicizing other reports such as FHWA and RITA reports on the effect of improvements. _______________________________________________________________

69 Thank you for completing this survey. Your responses will help provide insights into how to improve signal timing practices. If you have any questions regarding the survey, please contact Bob Gordon, rob.gordon3@verizon.net, (516)938-2498. If you have documentation that might be helpful it can be mailed to: Robert L. Gordon, P.E. Dunn Engineering Associates 36 Stauber Drive Plainview, N.Y. 11803

70 APPENDIx B SIGNAL TIMING POLICIES, GUIDELINES, AND STRATEGIES The following document, provided by Mr. Thomas Soyk of the City of White Plains, N.Y., illustrates the development of signal timing policies, guidelines, and strategies. SIGNAL TIMING GUIDELINES—CITY OF WHITE PLAINS, NEW YORk General The primary objective in the timing of signals is often con- sidered to be overall safety. There is particular concern for pedestrians and bicyclists but safety of all persons going through the intersection is always the primary objective. It is recognized that maintaining an efficient operation usually results in a safer operation. The City strives to maintain as efficient an operation as possible without any compromise in safety. Delay and Stops The objective is to reduce overall delay and stops as much as possible in order to reduce congestion and subsequent emis- sions. There does have to be some balancing to ensure that delays do not become intolerable on minor approaches or for pedestrians. The following guideline is provided: Utilize shortest possible cycle lengths. The typical 1. maximum cycle length is 100 seconds with some groups of intersections running at 110 seconds during peak periods only. Utilize, when possible, possible, half cycle or 2/3 2. cycles to reduce side street delay at intermediate sig- nals which have limited demand. Utilize free and flash operation during significant 3. off peak periods as long as progressions between closely spaced intersections on major arterials are not affected. For major street segments in the downtown grid, a 4. maximum of one stop per cycle for vehicles travel- ing through is an appropriate goal. It is best for arte- rial access roads to provide for no stops in the peak direction. Progressive Flow Providing consistent progressive flow on the main streets and arterials is the primary goal but this is best accomplished while maintaining reasonable speeds and not creating long side street or pedestrian crossing delays. The following are general guidelines: It is appropriate to use lead/lag phasing where neces-1. sary to maintain proper progressive flow. If turning phases are not protected-only, caution may be used to avoid creating traps. Where protected/permis- sive phasing is utilized, the proper signing to warn of opposing traffic having extended greens may be installed. Minimum, pedestrian and max 2 recalls may be used 2. as necessary to maintain progressive flow where a non-call could lead to an excessive delay or an increased number of stops along the main road travel route. Recalls may also be necessary for lagging left turn phases to maintain progression for the through movement. As much as possible, if there are no other conflicting 3. criteria, progression may be provided on minor and circulating streets. Plan transitions may be scheduled to avoid peak travel 4. times, shift changes at hospitals, school arrival and departure times, and 10 minutes prior to train arrivals at or near train stations. Transitions may occur on the hour, 20 minutes after, or 40 minutes after for both time-of-day and responsive mode plan selection. queue Management It is paramount to time critical intersections in a manner that does not create spillback and gridlock events. If necessary, police may be utilized at peak times and seasons to help pre- vent gridlock at key locations. Of particular concern are the following: Bank Street (Martine Av. to Hamilton Av.)1. Bloomingdale Road (Maple Av. to Westchester Av.)2. Hamilton Avenue (MLK Blvd. to Bank St.)3. Main Street (Battle Av. to So. Lexington Av.)4. Main Street (Mamaroneck Av. to No. Broadway)5. Mamaroneck Avenue (Main St. to Martine Av.)6. Mamaroneck Avenue (Livingston Av. to Maple Av.)7.

71 Maple Avenue (Bloomingdale Rd. to Mamaroneck Av.)8. Tarrytown Road (I287 to Battle Av.)9. Westchester Avenue (Paulding St. to White Plains Av.)10. Splits may be adjusted by creating different versions of the same plan (i.e., Plan 11 instead of Plan 1) to mitigate near “hot spot” locations at critical time periods. Such split altered or alternate plans are utilized near arenas, hospitals, schools, and the railroad stations. Temporary free operation may also be considered for such applications. Clearance and Gap Times Clearance, gap, and extension times are values that may be designed to match the need of each set of traffic conditions. Some guidelines to follow include the following: It is appropriate for pedestrian clearances to use 3.5 1. ft/s walking speed for calculation while including the time for the yellow interval. Where there is unusually high senior presence such as near senior housing or the Senior Center, a walking speed of 3.0 ft/s may be used if possible. Vehicle yellow intervals may typically be 3.2 s and 2. range from 3.0 to 4.5. The all-red interval may typi- cally be 2s and range from 1 to 3.5 depending on the intersection dimensions and prevailing speeds. Gap and extension times may typically be 3.0 s with a range of 2.0 to 5.0 allowed when either unusual traf- fic or failed detection equipment may require a change in settings.

72 INTERCOORDINATION DESIRABILITY INDEx. [Source: CHANG AND MESSER (1986).] The Intercoordination Desirability Index is described here: where: t = link travel time in minutes; qMAX = straight through flow from upstream intersection (VPH); qT = sum of traffic flow at the downstream approach from the right turn, left turn, and through movements of the upstream signals, divided by the number of arrival links at the upstream intersection; N = Number of arrival lanes feeding into the entering link of the downstream intersection. I may range from 0 to 1.0. Interconnection is advised when I exceeds 0.35. APPENDIx C TECHNIqUES FOR DETERMINING WHETHER TO COORDINATE SIGNALS AND ESTABLISHING NETWORk BOUNDARIES If signals are sufficiently closely spaced, coordinated sig- nals can minimize delay and stops by providing progressions or green waves for vehicle flows. Isolated (noncoordinated) operation may be preferred when: Signal spacing is too distant to permit flows to • progress. The cycle length needed by a key intersection is con-• siderably different from that best suited to the remain- der of the network. The intersection is saturated and considerations such as • control of queue length dominate the timing pattern. Limitations on the ability of coordinated signals to service the platoon of vehicles arriving at a downstream intersection result from the length of the travel time of the platoon between these intersections. In addition, the propensity of the platoon to dissipate (alter its shape from the pattern developed after it is released by the green signal at the upstream intersection) owing to traffic turning into the arterial and the “friction” of the roadway also limits the benefits of coordination. This is schematically shown in Figure C1 along with Table C1 that establishes the friction level (Robertson and Hunt 1982). Although simulation is probably the best way to establish the need for coordination and the coordination boundaries, a number of techniques have been described for providing guidelines for coordination including the following. FIGURE C1 Reduction in delay through coordination. (Source: Robertson and Hunt 1982.)

73 TABLE C1 REPRESENTATIVE VALUES FOR PLATOON DISPERSION FACTOR PDF Value Roadway Characteristics Conditions 0.5 Heavy friction Combination of parking, moder- ate to heavy turns, moderate to heavy pedestrian traffic, narrow lane width. Traffic flow typical of urban CBD. 0.35 Moderate friction Light turning traffic, light pedes- trian traffic, 11 to 12 ft (3.4 to 3.7 m) lanes, possibly divided. Typi- cal of well-designed CBD arterial. 0.25 Low friction No parking, divided, turning pro- visions 12 ft (3.7 m) lane width. Suburban high type arterial. Source: Fambro et al. 1991. CBD = central business district. COUPLING INDEx. [Source: HOOk AND ALbERS (1999).] The theory is based upon Newton’s law of gravitation, which states that the attraction between two bodies is proportional to the size of the two bodies and inversely proportional to the distance squared. The relationship is: CI = V/D2 where: CI = Coupling Index (unitless), V = Traffic volume for period analyzed (1,000 vehicles/ hour), D = Link distance (mi). Coordination is always advised when the coupling index exceeds 50, coordination is desirable when CI is between 1 and 50, and coordination is not needed when CI is less than 1.0. STRENGTH OF ATTRACTION INDEx. [Source: HOOK AND ALbERS (1999).] In the Strength of Attraction Index, the desirability of coor- dinating two adjacent intersections is based on the inter- section spacing (link distance), link traffic volume, link travel speeds, and platoon interference (i.e., on-street park- ing maneuvers, driveways, etc.). The strength of attraction between intersections, for a given time period, is calculated from the following formula: AF = I * V * (S/D)2 where: AF = strength of attraction (unitless), I = platoon interference (unitless), V = volume for the time period analyzed (vehicles/h), S = speed (mi/h), D = link distance (ft). Platoon interference is a unitless value describing inter- ference of the platoon as it progresses down the street. For simplicity, a platoon interference factor of 2.0 can be used for roadways without parking, 1.5 for roadways with paral- lel parking, and 1.0 for roadways with angled parking. The relative values between intersections may be considered in determining optimum signal groupings. There is no absolute strength of attraction at which coordination may or may not occur, however a natural breakpoint of approximately “1” appears to exist. COUPLING INDEx. [Source: bONNESON (2009).] The Coupling Index, CI, is provide by: CI = V/L where: V = 2-way volume (veh/h) L = segment length (ft). CI may be used to establish the potential for coordination according to the following: 0.3 or less unlikely to benefit from coordination• 0.3 to 0.5 segment likely to benefit from coor-• dination if mod-segment access point activity is low and turn bays are pro- vided on the major street at each sig- nalized intersection 0.5 or more likely to benefit from coordination.• These techniques may be used to establish section bound- aries. In many cases, particularly where arterials intersect, the traffic engineer’s knowledge of corridor travel patterns and land use patterns influence intersection assignments to traffic sections.

74 NETWORk DECOMPOSITION AND SIGNAL TIMING OPTIMIzATION Lieberman and Chang (2005) describe an approach for decomposing a grid network into its constituent arterial sub- systems for the computation of optimal signal timing plans and the integration of these plans to form a network-wide signal timing plan. A signal timing plan generator (RT/ IMPOST) is described. REFERENCES Bonneson, J., S. Sunkari, and M. Pratt. Traffic Signal Opera- tions Handbook, Report FHWA/TX-09/0-5629-PI, Texas Transportation Institute, College Station, Mar. 2009. Chang, E.C.P. and C.J. Messer, Warrants for Interconnection of Isolated Traffic Signals, Report 293-1F, Texas Trans- portation Institute, College Station, Aug. 1986. Hook, D. and A. Albers, Comparison of Alternative Method- ologies to Determine Breakpoints in Signal Progression, In Conference Proceedings, 69th Annual Meeting of the Institute of Transportation Engineers, 1999. Lieberman, E. and J. Chang, “Optimizing Traffic Signal Timing through Network Decomposition,” Transporta- tion Research Record: Journal of the Transportation Research Board, No. 1925, Transportation Research Board of the National Academies, Washington, D.C., 2005, pp. 167–175. Robertson, D.L. and P.B. Hunt, “A Method of Estimating the Benefits of Coordinating Signals by TRANSYT and SCOOT,” Traffic Engineering and Control, Vol. 23, No. 11, 1982, pp. 527–531.

75 APPENDIx D NORTH CENTRAL TExAS COUNCIL OF GOVERNMENTS RANkING MODEL This appendix, largely abstracted from the reference, pro- vides an example of a priority process for identifying candi- date intersections for signal retiming. The NCTCOG ranking model is based on the existing traffic conditions. The variables used in the model and their weights are discussed in this section. VARIABLES Total Delay Delay is the most frequently used measure of effectiveness for signalized intersections. Delay can be quantified in many different ways: stopped time delay, approach delay, travel time delay, and time-in-queue delay (Roess et al. 1998). Travel time delay is used in this research. Travel time delay of an individual vehicle is the difference between the mea- sured travel time and the travel time at the desired speed. Measured travel time is taken as an average of travel time in both directions of travel. The desired speed is taken as the posted speed. In this model, delay is used on an aggregate basis, and it is calculated here: DPV = delay/vehicle/intersection = (measured travel time – desired travel time)/ (number of intersections) (1) Total delay/ intersection = DPV × ADT ( 2 ) Where ADT is the average daily traffic. Number of Stops The number of stops is taken as the average of the number of stops counted in both directions of travel along the corridor. To get the aggregate value, this average value per intersec- tion is multiplied by the ADT. Number of stops per intersection = (Number of stops/number of intersections) × ADT (3) System Type There are three types of existing systems. A value of one indicates that all intersections are part of an existing inter- connected system with communications. A value of two indicates that some but not all intersections are part of an existing interconnected system with communications. A value of three indicates that there is no system (currently an isolated operation). Weightings The weighting for each factor is allocated by an expert group. The weightings are presented in Table D1. TABLE D1 VARIABLES AND WEIGHTINGS Variable Weighting Total Delay (DELAY) 50 No. of stops (STOPS) 30 System type (SYSTEMTYPE) 20 Calculation of Rank Order Using the weightings applied by the NCTCOG, the follow- ing equation is developed. Total Score(s) = (Delay/Max(Delay)) • 50 + (Stops/ Max(Stops)) • 30 + System_Type • 20 (4) Where System_Type = 1.0 for type 1 (all signals interconnected) 0.5 for type 2 (some signals interconnected) 0 for type 3 (all signals isolated). Quantitative variables DELAY and STOPS are normal- ized by dividing the maximum value from all of the candidate corridors, which precludes any single variable dominating the total score because of its magnitude relative to the other variables. After normalization, each variable is expressed on a zero-to-one scale, and the weights are an expression of the relative importance of each criterion. The maximum value of a variable in the given data is used for normalization. REFERENCES Pulipati, S.B., Regional Prioritization of Corridors for Traf- fic Signal Retiming, University of Texas, Arlington, 2006. Roess, R.P., E.S. Prassas, and W.R. McShane, Traffic Engi- neering, 2nd ed., Prentice–Hall, Englewood Cliffs, N.J., 1998.

76 APPENDIx E COST AND BENEFIT ANALYSIS The following publication, Chien et al. 2006 [© 2009 Insti- tute of Transportation Engineers, 1099 14th Street, NW, Suite 300 West, Washington, DC 20005 USA, www.ite.org (used by permission)], provides an example of a cost and benefit analysis. It describes the techniques employed as well as the results.

77

78

79

80