Improving Traffic Signal Management and Operations: A Basic Service Model

III. Case Studies

Agency Archetypes

Archetypes of agencies can be used to evaluate the effectiveness of agency practices and needs. In this section, we will lay out a range of possible agency archetypes, and then we will use these archetypes as the basis for a series of interviews with real agencies.

Archetype 1. High-Activity Response to Adequate (Staffing) Resources

Archetype One would be typical of a state or large metropolitan agency that has identified operations and maintenance as a core business practice. The business plan for that agency supports staffing and resources focused on delivering an operations program. The staff undertakes bi-annual signal timing reviews by hiring outside consultants to collect turning-movement counts at all intersections, reviewing all fixed intervals at each intersection, developing and programming signal optimization software for calculating optimized signal timings, evaluating signal timings using micro-simulation, and providing revised signal timings to the agency. The agency has an active program for maintaining the data and making sure it is installed and maintained in signal controllers and systems.

Before any signal timings can be implemented, they must be evaluated using microscopic simulation, both by consultant and agency staff. Agency staff are hired with extensive expertise in analysis and simulation, and encouraged to improve that expertise through regular training. Consultants are hired to work on site, under agency supervision and with no access to highly experienced consultants. The operations staff does not have the authority over basic timing parameters such as phasing, phase sequence, pedestrian accommodation, detector placement, and intersection lane usage. Those parameters are aggressively controlled by other departments within the agency. Because of the perceived workload resulting from high activity levels, operational personnel also are not encouraged to spend time in the field to observe the results of the operation. Using microsimulation was a response to this restriction (and also a cause of it). Improvements are reported on the basis of simulation.

This archetype sets a precedent for for detailed analysis. However, several key weaknesses in their situation prevent getting the best results:

Not being able to design phase sequence along with signal timing limits the versatility and optimality of the results. Often, the phase sequence decisions are made to protect as many movements as possible rather than to provide efficient operation.
Signal timing technicians and engineers do not have authority over final results, and are not encouraged to understand their analysis in terms of on-street performance.
Traffic engineering staff and operations staff are not mutually responsive, and some operation is controlled by one group while other operation is controlled by a separate group. This is quite common in real agencies, where a traffic signal system is operated by a different group than the basic signal timing staff that does non-systems operational design including phase sequence.
Befitting their abundant resources, the agency employs or contracts with a large pool of engineers. Those engineers, however, remain separated from maintenance technicians. The technicians and traffic engineers do not benefit from any expertise in operations, and the operations staff often work in the absence of relevant input from traffic engineers and technicians.
The timing approach is driven by software and the data required by software, rather than being driven by performance results on the street.
Most of the agency's considerable operations resources are consumed doing data collection and analysis, rather than doing on-the-ground performance measurement meaningful to the objectives of motorists.
The agency does not benefit from hiring consultants, using them as an extension of staff rather than a source of expertise. This is reflected by structuring their consulting projects to require on-site services, which makes it difficult for consulting companies to commit their most experienced experts, and by refusing the pay the rates of consultants who are most advanced in their careers. (This report does not attempt to address the problem that consultants are not consistently expert despite their rates or stated experience.)
The resources and software emphasis reduces the agency's ability to respond to special-event signal timing needs.
The agency responds to citizen complaints with descriptions of activities they are undertaking, rather than with the reasons for the operation being complained about. Complaints lead to studies, justifications for system enhancements, revised timings, and other activities, but operations staff are not compelled to be able to explain their decisions to citizens and their representatives.
The agency believes that it is undertaking an aggressive and successful program.

Archetype 2. Infrastructure-Rich Response to Abundant (Capital) Resources

Archetype Two might be typical of an agency that has committed to improving operations through capital programs that supports infrastructure improvements. The agency has not garnered support for increasing staffing resources to support active operations; which, effectively is a new activity within the agency. The agency has built an extensive centralized traffic signal system with downstream detectors at all intersections, adaptively controlled. Operations staff is limited but maintenance capabilities are good and capital programs well-funded. The limited operations staff are used to performing travel time analysis and systematically review the network for unexpected queuing. The agency has a defined operational mission of providing smooth flow and minimizing congestion, the technicians are trained to observe and evaluate the system in accordance with those objectives. The strong maintenance capability keeps the extensive infrastructure in good working order. Well-founded design standards provide consistent infrastructure (that provides the features needed to achieve their operational objectives) and pedestrian accommodation, but without much downstream review after construction.

The agency reviews signal timings yearly by scheduling and reviewing split monitoring in the signal controllers, for each arterial system. This allows them to identify non-coordinated movements that consistently max out, suggesting the possibility of residual queuing. This is accomplished using the two operations technicians on staff. The two staff members work a split shift, with one technician working during the morning rush and the other during the afternoon rush. The morning technician observes system data for an arterial system Tuesday through Friday mornings, and the afternoon technician collects similar data Monday through Thursday afternoons. The technicians handle adaptive parameter and new intersection data maintenance duties in the middle of the day, when their shifts overlap. Using this approach, they are able to systematically review all 175 arterial systems plus a downtown grid each year. The split monitor and adaptive data results are summarized and reviewed by the engineer in charge of the signal program during the same week. Unexpected queuing or anomalies lead the engineer to schedule a site visit and a review of the arterial. If a field evaluation cannot rectify the problem, a more detailed analysis is conducted using data collection and analysis.

The focus of agency staff on actual performance observation allows them to be responsive to citizen complaints. Complaints are handled by the engineer in charge with input from the technicians who review that location. The signal system provides real-time monitoring capabilities that allow operations staff to respond substantively to citizen complaints that are caused by malfunction. The review technicians are also efficiently used for actual observation and assessment rather than for activity-based strategies, such as data collection and analysis, which may not always serve their objectives efficiently.

This agency does not score well on the Report Card because of limited systematic data collection and retiming and management that is not focused on real-time human interaction. But it shows strong results because:

Insufficient operations staff resources are compensated by effective capital expenditure on systems that work effectively for long periods without frequent systematic retiming.
The agency is focused on performance measures appropriate to motorist objectives, as a validation of the ongoing adaptive control.
By including pedestrian needs systematically in the design process, the need to be responsive to pedestrian requests is reduced down to special cases.
The agency designed a system that made use of their strengths (strong maintenance) and minimized their weaknesses (limited operations staff).

Archetype 3. Well-Managed Response to Limited (Capital & Staffing) Resources

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

Don't make motorists stop, but if you must, delay them as little as possible.

The agency has difficulty maintaining extensive detection and does not build intersection detectors in coordinated systems except on minor-street approaches. The agency also focuses on efficient and versatile timing that will operate reasonably even when detectors are not working. For example, they limit controller max times based on engineered fixed time operation, and they keep cycles short and use the least number of protected phases possible. Design favors progression for the predominate through flows. Operations staff are limited but have authority over the full operation of the signal, including design and signal phasing.

The agency reviews signal timings yearly using floating-car studies on each arterial system. This is accomplished using the two operations technicians on staff. The two staff members work a split shift, with one technician working during the morning rush and the other during the afternoon rush. The morning technician collects travel time data on an arterial system Tuesday through Friday mornings, and the afternoon technician collects similar data Monday through Thursday afternoons. Five runs are made in each direction. In addition to collecting travel time data, the technicians review operation by having a time-space diagram available in the vehicle to ensure that the signals are operating as designed. The technicians handle timing database maintenance duties in the middle of the day, when their shifts overlap. Using this approach, they are able to systematically review all 175 arterial systems plus a downtown grid each year. The travel time studies are compiled during the middle of the day, and reviewed by the engineer in charge of the signal program during the same week. Unexpected queuing or other delay noted in the travel time studies lead the engineer to schedule a site visit and a review of the time-space diagram. If a field evaluation cannot rectify the problem, a more detailed analysis is conducted using data collection and Synchro optimization by consultants. Technicians require training in travel time data collection and in observing that signal timing matches time-space diagrams, but it does not require expertise in simulation or optimization modeling.

The agency also built signal systems that provide remote access to local controllers to minimize the need for field visits to respond to complaints, but without traffic-responsive or adaptive control, which requires infrastructure (both communications and detection) that would have prevented agency-wide system implementation. The agency focuses its limited design resources on setting clear standards for good design such that pedestrian accommodations, phasing, initial timing, and lane utilization are done right at the outset. For example, during design, a complete operational review is performed to make sure that each signal phase provided is needed, that lane utilization balances capacity with respect to demand, and that detection (both pedestrian and vehicular) critical to minimizing the impact of the signal on major movements is provided. The design principles favor detection where it serves the objective of not stopping cars on the main street, without, for example, the added cost of extensive detection on the main street. (A common alternative to these appropriate standards is the use of typical designs without evaluating their needs at a given intersections, including but not limited to always providing a left turn phases when left turn lanes are provided, not considering whether a through lane should really be a turn lane, automatic use of a specific phasing scheme—such as 8-phase operation with leading lefts—without considering the demand at the intersection, providing extensive dilemma-zone detection even on intersections in coordinated systems, etc.) Operational staff are limited but have full responsibility and authority for the operation of the intersection, including phasing, phase sequence, signal head arrangements, detection, and timing.

While Agency Three's funding is considerably less than Agency One's, they get better results, and further they can show that their results are better, because:

The agency devotes its limited resources to evaluating the operation based on a clear understanding of motorist objectives. The objective of not stopping traffic, for example, is directly measured by the low-cost travel-time studies, rather than the costlier collection of traffic volumes and attempts to model, based on those volumes, a range of performance measures that have not been clearly articulated and are therefore difficult to evaluate.
The agency developed and then consistently applies design standards consistent with a realistic assessment of their maintenance resources. Thus, they don't end up with some intersections designed at a high level and others that don't meet adequate standards, particularly in regards to detection.
The agency designed signal systems consistent with their maintenance and operations capability. Their strong, objective-based measurement and evaluation program was considered more useful in responding to motorist needs than in providing additional automatic traffic monitoring. Systems with more advanced detection and control would have exceeded their budget for city-wide installation, and they considered it better to have good field device management throughout the agency rather than advanced features in a smaller area and no capability elsewhere.
The agency is highly responsive to citizen input because of their motorist-relevant measurement approach and good access to field infrastructure.
The versatility of signal timings minimizes the need for additional detection and automated responsiveness.
Centralized authority over all aspects of design and operation provide the tools necessary for versatile signal timing.
Consultants are used effectively for situations where new timings are truly needed.

This agency might not score as well as Agency One on the Report Card, but they get better results with respect to their more limited or constrained resources. Those better results are reflected in operations consistent with their operational objective—rather than engaging in office and computer-based data collection and retiming efforts, they are on the street observing actual operation from the perspective of the motorists. The time spent by Agency One staff developing simulation models and other volume-based analysis is spent by Agency Three to evaluate actual operation in the field, based on measuring travel times directly. And flags for action are based on those travel time measurements, which will quickly reveal new congestion, rather than on an assumption of need implied by an arbitrary schedule.

Archetype 4. Insufficiently Managed Response to Limited Resources

This agency has the same resource limitations as in the previous scenario, but strives implement a program of activities without clearly articulating the objectives and outcomes they would like to achieve. These activities are accepted as common practice but without clearly mapping the activities to local objectives and performance measures the outcomes may not be improved. This agency might perform well on the Report Card but not experience the desired level of on-street performance for the traffic signal program. The agency opts to hire consultants allowing them to apply arbitrary standards for design and operation rather than focusing on their limited resources, establishing consistent and adequate design standards and operational objectives. The outcome of this approach is expensive, detector-rich installations, resulting in fewer intersections being built due to the high costs of construction and maintenance. This forces the agency to build some intersections at extremely low cost that do not comply with any principles of good design. Thus, some intersections are built with an abundance of detection, mast arms, and expensive signal head arrangements, while others are built on wood poles using span wire, the minimum number of heads, and no pedestrian accommodation. This leads to intersections that must operate poorly to satisfy MUTCD pedestrian accommodation requirements.

Signal timing is performed by consultants without clear objectives, resulting in timings that attempt to make the most out of the provided infrastructure. Thus, intersections with extensive approach detection use dilemma-zone schemes and long actuated cycles that depend on actuation to provide optimal operation. Max times are set long to allow actuation to work. Consultants based their design on data collected by the agency, because of limited budgets to pay consultants to do so. The two operations technicians on staff are used to set road-tube counts and conduct occasional turning-movement counts. Signal timings are developed for all arterials on a rotating, five-year program. To improve the effectiveness of signal timing, intersection detectors are tasked as system detectors, and arterial systems use traffic-responsive operation. Frequent detector failures greatly reduce the effectiveness of the operation, which falls back to programmed daily schedules, but without the agency's frequent review.

The large number of detectors at some intersections pose a maintenance challenge, and the agency cannot keep the detectors working. Thus, the intersections do not operate effectively, with detector failures resulting in long green times and cycles even when demand is low. Limited resources preclude the needed reviews of these intersections.

The only review mechanism is responding to citizen complaints. When a complaint is received, the agency attempts to review operation on the basis of detection, even if several intersections away. The engineer reviewing each complaint has no baseline travel-time-based performance data against which to compare current operation. Signal timings are tweaked to accommodate the complaint, but without a systematic review of the whole arterial, which must wait for the next signal timing update, those tweaks may undermine the basic versatility of the signal timing plan. For example, adjustments to improve intersection splits often undermine progression if the progression design isn't considered as part of the adjustment.

The results achieved by the agency are poor, because:

The agency has no performance measurement outside of citizen complaints, let alone measurement linked to motorist objectives.
The agency feeds analysis software rather than collecting performance data.
The agency is reactive rather than proactive.
Adjustments made between optimization projects are not made systematically within the context of the timing design.
The operational picture presented to motorists is inconsistent, with apparent inequities.
Operation is usually degraded by the failure of detection, mostly because the operation depends too much on the detection.

Because the agency collects data and performs analysis, it might score better on the Report Card than in the previous scenario, though the results will not be as good, because resources are not aimed at activities directly linked to performance that matters most to motorists. Those activities can be seen as one step removed from operation on the street, as an abstraction or representation of that operation, rather than as the definite example. Analysis might show a significant improvement in delay, for example, but motorists are more directly concerned with how smoothly and predictably they flow through the network. The temptation to depend on analysis more than observation overwhelms this agency.

Archetype 5. Special Issues with Highly Dispersed Infrastructure (Typical State Agency)

Archetype Five is typical of an agency that has limited operations and maintenance capabilities due to in adequate capita and staffing resources relative to the number and dispersion of traffic signals within its jurisdiction.

Most of the intersections built by the agency are in remote areas, and coordinated systems are on suburban or fringe arterial streets and not grid networks. The arterial networks frequently center or cluster around freeway interchanges, including diamond interchanges.

The key operational constraint on the agency is the distance from the office to the traffic signals. Often, a state DOT district office is located in a city that maintains its own traffic signals. And just as often, the agency's operations staff must divide their time with many other traffic engineering duties, including signs, pavement markings, and liaison with local agencies. Because of this, these agencies usually seek to make their intersections operate without their direct involvement to the extent possible. These limitations lead to a series of broad requirements:

The operation of the signal must accommodate changes in traffic at all time series, including daily, weekly, and seasonal variation in addition to long-term trends.
The signal must provide access from the operations staff for observing operation in response to citizen complaints, and for making adjustments and maintaining the signal timing database. This access is needed on an ad hoc basis.
Field reliability is paramount. This requirement is in conflict with the first requirement, because accommodating traffic variation over the long run requires detection, and detection is the least reliable component of traffic signal installations.
Design standards that focus on high-quality installations improve reliability and thus minimize the need for future involvement. This includes not only detection, but also the physical infrastructure of the signal to support a range of operation (such as using longer mast arms than necessary to ease the installation of future left-turn heads). An agency with more signals closer to their office might be able to adopt a model that requires greater on-street involvement.

The agency reviews signal timings yearly using floating-car studies on each arterial system. Unexpected queuing or other delay noted in the travel time studies lead the engineer to schedule a site visit and a review of the time-space diagram. If a field evaluation cannot rectify the problem, a more detailed analysis is conducted using data collection and Synchro optimization by consultants. Signal timing is designed expressly to provide the maximum potential for through-band progression. Given the fringe suburban setting of their systems, severe congestion is rare, but rapid growth means that variability is high and such versatility maintains acceptable operation in the face of such variability.

Agency Five uses the detection infrastructure, and the ability of their systems to give them access to the field equipment to minimize the effect of such large distances to the signals from their office. Thus, they get good results:

The agency devotes its limited resources to evaluating the operation based on a clear understanding of motorist objectives.
The agency applies design standards consistent with a realistic assessment of their maintenance resources. For this agency, that means high-quality installations to minimize maintenance visits, which for them is a more important measure than maintenance skill.
The agency designed signal systems consistent with their maintenance and operations capability. Their need to monitor conditions remotely, and to effectively manage the field infrastructure remotely is well accounted in their design.
The agency is highly responsive to citizen input because of their motorist-relevant measurement approach and good access to field infrastructure.
The versatility of signal timings minimizes the need for additional detection and automated responsiveness.
Centralized authority over all aspects of design and operation provide the tools necessary for versatile signal timing.
Consultants are used effectively for situations where new timings are truly needed.

Features of Successful Archetypes

The archetypes above that are successful present the following features:

Strong concept of basic service. In the face of limited objectives, the effective agencies place a high priority on providing good basic service first, with less emphasis on attempting advanced service models. These agencies have defined what basic service means for their agency, and they evaluate their operation in terms of those basic services to ensure that they succeed. The ease with which one can construct an agency archetype that spends a lot of resources without achieving good basic service demonstrates that even good basic service can be a challenging goal. And while the Report Card does not distinguish between basic and advanced service, the poor Report Card performance suggests that focusing on basic service is not as common as it might be. Most agencies have limited resources, and all agencies experience occasional resource reductions; the definition of and commitment to good basic service is highly recommended.
Clear evaluation of objectives. While minimizing travel time might be considered too broad by many practitioners, the fact that only Archetype Three responds to limited resources by focusing more tightly on that simple objective reveals too little attention to those broad objectives. This is closely related to the basic service concept, and the best agency models evaluate their effectiveness in terms of basic service objectives first. Archetype One conducted extremely detailed analysis, but without on-the-ground attention to simply stated broad objectives. Thus, it cannot really know how well it meets that objective.
Close coordination of design, operations, and maintenance resources and limitations. The successful archetype agencies avoid constructing infrastructure elements that cannot be maintained, and they avoid operation that demands such infrastructure and maintenance. They also avoid overdesigning some intersections at the expense of barely meeting minimum standards at other intersections. Or, they purposely design systems that demand more capital funding up front in order to respond to a limited operational capability of their staff. The relationship between design, operations capability, and maintenance resources is the key element of this feature. Each agency must develop that relationship for themselves based on the dynamic balance between resources and objectives. Increasing infrastructure sophistication may be a good trade-off in response to limited operations staff. Decreasing infrastructure sophistication may be a good trade-off in response to limited maintenance capabilities. An agency should not build what it cannot maintain and operate, but sometimes a more sophisticated design may reduce demand on maintenance and operation. For example, one successful archetype used adaptive control to reduce the operations resource requirements, but with the recognition that adaptive control placed a greater demand on the agency's maintenance capability. An agency limited in both maintenance and operations may achieve the best balance by focusing on simple and highly versatile coordinated signal timings rather than on detection-intensive designs that result in poor operation when the detection fails. And the choice to use simple and versatile timings may get more consistently acceptable results rather than depending on analysis-intensive timing design which narrows the optimal range and reduces versatility in return for marginally better measures of effectiveness as reported by the analysis tools.
Good understanding of measuring results. This is related to the strong concept of basic service, but it also leads to a better validation process for more advanced service. For example, an agency that is spread over a large geographic area (the limitation) may design systems that maximize remote monitoring capability (the response), on the basis of minimizing the number of required field visits (the measure). An agency that has limited staff resources but with high skill levels may more effectively invest in sophisticated automated analysis methods, while an agency with a larger staff but with lesser skills might need to keep their analysis methods simple, even though they require more effort to implement.
Commitment to staff development. The strong archetypes seem to better understand the capabilities of their staff and do their best to develop staff capabilities and reward progress. It should be noted that not all rewards must be remunerative. Agencies that provide clarity of objectives and direct methods for evaluating achievement in those objectives often observe higher staff satisfaction and enjoyment of their work. And it should also be noted that staff development includes a mix of activities, including specific training to improve knowledge, mentoring to improve skills and abilities, and support for external professional activities such as attending conferences and being involved in regional and national technical, policy, and research committees.

Basic Service Concept

In this section, the concept of basic service, and how basic service might be evaluated, will be further developed.

As stated in the introduction, any basic service concept must start with the expectations of the motorists to whom the service is being provided. Most motorists, when driving their cars (as opposed to when they render opinions in the public forum) have some variation on the following broad expectation:

I want to drive to my destination at my desired speed with the minimum of attention and interruption. In the absence of achieving that goal, I want to be treated fairly and predictably so that I can plan my day with the minimum of uncertainty.

While such characterizations seem ambiguous, they do lead to a number of clear objectives for an agency. These might be:

Field infrastructure reliability.
Signal timing that minimizes and balances congestion.
Signal timing that promotes smooth flow. This objective is an amalgamation of more narrowly cast objectives, such as delay, stops, and so on. Actions taken to achieve those narrow objectives should be evaluated in terms of broad objectives. Delay may be reduced numerically, but will the action result in smoother flow? This will usually require the application of experienced judgment, but not necessarily high levels of technical sophistication.
Signal operation that responds to conditions predictably and consistently.
Versatile signal timing that provides broad-banded solutions rather than being narrowly optimized for a specific condition.

Each of these agency objectives requires a series of fundamental agency requirements to support them. These requirements can form the basis for evaluating basic service.

Field Infrastructure Reliability

This objective imposes the following requirements:

Realistic assessment of maintenance capability. Most agencies are resource-limited because maintenance is usually funded by a jurisdiction's general fund, rather than by capital funds, and those resources are capped and subject to sometimes fierce competition within the local policy and political environment. General fund sources include local property and (rarely) income tax, with some programs supporting operations and maintenance from the Federal government. Capital funds are supported by relatively rich Federally funded construction programs and the sale of bonds. This reality, which is usually outside the control of the traffic signal professionals, nevertheless requires those professionals to consider how they manage their resources. The two main constraints of maintenance resources are technological and quantitative.
- Technological resource limitations affect the level of technology an agency can support on the basis of their salary structure, training environment, general work environment, and local labor pool. Of these facets of the problem, only the training environment can be modified by the agency on the basis of a policy decision at the level of the operational staff. Agencies therefore attempt to train their best technicians to handle more advanced technologies, but often then lose those technicians to more lucrative employment in often more pleasing work environments, especially if such skills are in demand in the local labor pool. An appropriate response to this limitation, other than articulating it in hopes of broader policy-level understanding and support, is to design infrastructure and adopt operational approaches within the technological capabilities of the agency's maintenance resources. Another approach might be to devise a maintenance concept that rents that capability from a contractor rather than trying to respond within the agency's forces.
- Quantitative limitations abound in traffic signal agencies. Most agencies of any size complain that each maintenance technician is spread too thin over too many traffic signals, and this complaint is heard even by the most well-funded agencies. One approach to responding to this limitation is to reduce the complexity of the infrastructure. This is distinct from reducing the technical depth of that infrastructure, and focuses more on the number of infrastructure elements built. For example, the most vulnerable infrastructure element in any traffic signal program is detection. One of the main reasons for the rise in popularity of video detection systems, despite questions of their accuracy in certain situations, is that they are perceived to be easy to maintain, particularly without having to block traffic, and easy to reconfigure. A system that can provide effective signal timings with less detection can also achieve this objective.
Designing within an agency's capabilities. In a forum on operations and maintenance held during the late 1980's, Anson Nordby, then the head of the City of Los Angeles ATSAC program, observed during a panel discussion that designers often treat each system project as an opportunity for a technological tour-de-force. He further observed that by doing so, the designers were often leaving the agency's maintenance capability behind. The wisdom in this observation suggests that one critical step in any signal and signal system design project is a careful evaluation of the agency's maintenance requirements. For example, signal designs that require bucket trucks that first-line maintenance technicians do not use will require failures of those systems to be responded to by second-line resources. This increases the time required to respond to a problem, which undermines the basic service objective. Detection again provides another example. Many agencies are unable to maintain their pavement as well as they would like, and poor pavement usually results in poor in-pavement detector longevity. It should be noted that the only way designers will understand maintenance limitations is to ask the maintenance staff.
Using capital resources to minimize maintenance impact. One approach to the problem mentioned above of detector failures in poor pavements is to improve the design and construction of the detectors with the poor pavement considered. The City of San Antonio, in response to a large number of failed detectors, and in addition to moving to operational approaches that minimized the need for detectors, started constructing them more durably. They used a 4" rock saw and made a 6"-deep cut. The loop detector wiring was enclosed in Schedule 80 PVC pipe, and the loop was placed in a sand bed at the bottom of this saw cut. The cut was backfilled with stabilized base (two-sack portland-cement concrete) and the wear surface was patched with asphalt. While this approach used more capital funds, those capital funds were relatively more available than maintenance funds, and the resulting trend was an increasing percentage of functioning detectors.

Minimizing and Balancing Congestion

One of the recommendations made in the final report for the FHWA project Signal Timing Under Saturated Conditions was to devote the agency's best resources to its worst congestion problems. Leading practitioners, as reported in that work, recommended:

Make use of every available feature, however esoteric, to improve green time efficiency in favor of the congested movements. This requirement might be seen as conflicting with a realistic understanding of maintenance limitations, but it need not be so. An agency can reconcile both these objectives by ensuring that the extreme operational efforts are spent wisely and not wasted on intersections that operate routinely and don't need such efforts.
Devote the agency's best operational experts to the most congested problems, and be prepared to support their extended observation and experimentation in pursuit of a solution.

That research also suggested a clear focus on operational objectives as conditions moved into the congested regime. During uncongested conditions, the agency should try to provide smooth, predictable flow in accordance with the expectations of motorists. When congestion occurs, smooth flow is no longer possible, and the agency should switch to an objective of maximizing throughput. When demand increases to the point where queuing is inevitable, the agency should operate the signals to minimize the damage done by queue formation in an attempt to keep the problem from cascading throughout the network.

Following these approaches will, in many cases, provide a balanced approach to congestion that most motorists will see as (at least) equitable, consistent with their expectation of good basic service.

Smooth Flow

Many agencies now use advanced analysis tools to optimize and evaluate traffic signal operation. But the optimizations and the evaluation measures are surrogates for the "measurements" made by motorists. For example, motorists do not separately measure time delayed and time not delayed. They measure total travel time, and travel time reliability. All motorists apply a greater penalty to time delayed than time not delayed. But motorists also evaluate the cause of the delay when assessing blame, and delay caused by congestion is counted less against the agency than delay caused by red lights, especially when the other movements of the intersection are perceived to be underutilized. These situations are a leading cause of complaint calls.

The most successful agencies do not blindly optimize based on a narrowly cast objective function, such as delay, stops, or some combination thereof. Nor do they optimize solely on progression bandwidth. Rather, they provide a mix of operational objectives designed to be seen by motorists as fair and predictable.

For example, progression optimization might find a 10-second band through 20 intersections on a four-mile arterial system when using the optimally resonant, say, 90-second cycle. Such a progression band might be invisible to most motorists, who fall out of the band at the slightest disturbance in the flow, and are then repeatedly delayed until the next band catches up to them. Most motorists would perceive better basic service if this arterial was broken up into two or three segments, as long as the breaks in progression were clean (such that arriving traffic arrives on a long-standing red and not just as the signal is changing to yellow). With such breaks, much greater progression bandwidth might be possible. Instead of a large percentage of motorists being stopped unpredictably at many intersections, a higher percentage are stopped predictably at two locations (in addition to the first).

Another aspect of good progression design with respect to smooth flow is ensuring that the largest entering platoons see the greatest service by the progression band.

Optimization approaches that minimize delay and stops are often based on simplistic models of those stops. If the stops cannot be realistically modeled, optimizing on them is likely to lead to distorted results.

By focusing on good basic service, an agency might forego an abundance of computer optimization and simulation, and the numerical data collection that such requires, in favor of simple and direct methods that ensure smooth flow, including time-space diagrams and on-street evaluation.

Predictable and Consistent Response

Most highly responsive systems run into the problem of keeping the signals in useful coordination while the basic operation is being changed. Adaptive systems control this effect by limiting changes to small increments, though this may cause the operation to wander into timings that reduce the potential effectiveness of the timings. For example, some adaptive control systems allow small changes in the cycle length, even though a given system may provide good progression on only one or two cycle lengths within the acceptable range (as reported by Steven Shelby in his paper Resonant Cycles in Traffic Signal Control, Transportation Research Record #1925, 2005). Traditional traffic-responsive systems invoke often damaging transitions that are usually constrained to occur outside the peak period to avoid causing more problems than they solve. This is particularly true of transitions that dwell only, with shortway and subtraction transition methods causing less disruption. The transition problem suggests that fewer changes of the signal timing plan might be better in many situations.

Most agencies design only four or five signal timing plans for their arterial street signal systems. The range of conditions in any typical day, however, runs the entire gamut from a nearly empty network to peak conditions, and from a strong inbound bias to a strong outbound bias. When overlaying the range of possible solutions onto this wide range of conditions, most agencies divide that solution space into only four zones: Heavy inbound flow, heavy outbound flow, heavy balanced flow, and light flow. That suggests that signal timings must cover a broader range of conditions than the narrowly defined optimums many computer optimizers would imply.

Dr. Carroll Messer, author of the seminal PASSER II progression-based optimizer, observed during a past conference that broad progression-based solutions provided a predictable operation. Motorists routinely stop at the same intersections and therefore see the effect of the timing solution clearly. When that experience is consistent day in and day out, they can adjust their behavior (possibly) or their expectations to that reality. Systems that continuously change the motorist experience should be carefully selected when the need for that variability outweighs the expectation of a consistent motorist experience.

Signal Timing Versatility

Closely related to a predictable and consistent response is the need for versatility in signal timing. If the solution space is divided into only four zones, and if that solution space overlays a wide-ranging condition space, then each plan in the solution space must provide acceptable operation even at the boundaries of the condition to which it is applied. In highly dynamic cases, more than four solutions may be required, but that increased dynamic nature still will demand versatile signal timing solutions.

But most computer-based analysis of traffic signals applied by practitioners, both for optimization and simulation, considers the traffic demand as a steady state. Thus, signal timing approaches are evaluated by their performance during the narrowly unique conditions usually characterized by a 15-minute period rather than by the range of conditions over the hours of the daily schedule for which those timings will be used, and for the months and years over which that daily schedule will be used. The best agencies address this by one of two approaches:

Implement a system capable of continuous optimization. That system may use normal traffic actuated features at the local controller, including gapping out, hold release, phase skipping, and volume-density features. Or it may use a more centralized or regionalized adaptive control based on network objectives. In all cases, the effectiveness of the system depends on detection.
Design for versatility. The best agencies are able to understand signal timing effectiveness over a range of conditions by looking at the breadth of the solution rather than by seeking a narrow optimum. For example, a progression-based solution that works for heavy traffic will also work for light traffic. Even one car in the network will move unimpeded. This is a variation on Dr. Messer's observation, that such solutions a.) provide a predictable target into which motorists will fit their behavior, and b.) work over the broadest range of actual conditions. Agencies that are constrained by resources on a range of fronts can still maintain good basic service if they consider versatility in their methods and solutions.

Appendix I contains more development of the signal timing versatility concept.

previous | next