Adaptive Traffic Control Systems: Domestic and Foreign State of Practice (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.

36 INTRODUCTION Many factors influence the costs of an ATCS deployment and achievement of the full ATCS benefits. This chapter identifies those factors that affect the costs of installing and operating ATCSs. The chapter also addresses users’ expectations and achieved benefits from the ATCS deployments. The costs of ATCS deployments are captured through the costs of system installations per intersection and comparison of maintenance costs for ATCSs and non-ATCSs. In addition to discussing the common evaluation studies to investigate the performance of ATCSs, this chapter reviews the benefits of various ATCS deployments. Finally, the chapter addresses public percep- tion on ATCS implementations and provides some examples of lessons learned in practice during the implementation of ATCSs. COSTS OF DEPLOYING ADAPTIVE TRAFFIC CONTROL SYSTEMS According to an earlier study (Hicks and Carter 2000), cost appears to be a major obstacle to widespread ATCS deploy- ment. The term cost here encompasses both the capital and the operations and maintenance costs of an ATCS. There is some disagreement whether over the long term ATCSs are more cost-effective than traditional signal systems because ATCS operations and maintenance costs are much lower than those associated with signal re-timing. Others argue that the esti- mates need not be simplified because an ATCS may have higher costs of physical maintenance such as the repair and replacement of detector loops. However, the answer, as always, lies somewhere in between. There are ATCSs that do not expe- rience higher detection maintenance costs than conventional traffic signal systems. Conversely, some ATCSs may have significant signal timing adjustment costs. Belief that ATCSs do not require any fine-tuning and that they can self-adjust their operations indefinitely is one of the biggest myths about these systems. A review of some recent ATCS deployments show that licensing costs to run such a system may contribute an addi- tional 10% to 15% to the overall installation costs. The licens- ing costs are usually not one-time costs because the licensing rights are sold separately for various intersection bundles. If an agency wants to expand the system it will likely need to purchase licensing rights for a larger intersection bundle. Although there was a indication that licensing costs for sys- tem expansion are too high, only 42% of all interviewed ATCS users agreed with this notion. Approximately 38% of the respondents found the licensing costs for the expansion of their systems to be affordable, whereas approximately 20% either do not have to pay any licensing fees or did not need to expand the systems and do not know what the licensing expansion costs would be. On average, the costs of installing an ATCS are approx- imately $65,000 per intersection. Figure 12 is a histogram of ATCS installation costs per intersection. The histogram shows that these costs can vary significantly among various ATCS users. The median and mode of the distribution (of ATCS installation costs per intersection) are approximately $45,000 and $40,000, respectively. These numbers are signif- icantly higher than estimates reported previously in the litera- ture (Hicks and Carter 2000), where similar estimates were between $20,000 and $25,000. It is important to note here that the reported costs often include more than just the installation of the adaptive component of the system. Replacements of the local intersection hardware and software (sometimes even installation of new communication infrastructure) often accompany installation of the adaptive algorithms. In spite of the survey’s attempt to separate pure ATCS installation costs from the infrastructure upgrade costs, which do not necessar- ily need to be conducted at the time of ATCS installation, the ATCS users were able to report only the total costs (per inter- section) of their system deployments. Once an ATCS is installed there are costs to operate and maintain both the hardware and software of the system, as well as the infrastructure whose maintenance may be more costly owing to the higher infrastructure needs required by an ATCS operation (e.g., more detectors or newer communications). The percentage of an agency’s annual budget that is spent on the physical maintenance of an ATCS is a good indica- tor of the cost-efficiency of maintaining these systems. To get an unbiased picture of the costs of ATCS maintenance we need to consider also the percentage of intersections that run under an ATCS. Figure 13 shows the correlation between the two percentages. The figure indicates that, in general, propor- tions of annual budgets that are spent on maintaining an ATCS are lower than proportions of intersections under the ATCS in the total number of intersections. The few outliers that were originally in the data set were removed to achieve a better co- efficient of determination (R2). Removal of the outliers did not change the overall relationship between X and Y data sets; CHAPTER SIX IMPLEMENTATION COSTS AND BENEFITS

37 however, the scatter of the data points is reduced. This chart shows that when it comes to the costs of the physical mainte- nance, ATCSs are more efficient than conventional traffic sig- nal systems, which is the opposite of a widely accepted notion that ATCSs are expensive to maintain (especially communi- cations and detection). The other components of costs for maintaining an ATCS operation include consulting costs and the costs of maintain- ing ATCS hardware and software. Although there is a notion that once an ATCS is set up there is no need for re-timing traffic signal timing plans, there are some costs of reconfig- uring ATCS parameters. These costs can be significant owing to inadequate in-house expertise to adjust ATCS parameters to meet new operational needs. Most ATCS agencies have diffi- culties comparing estimates for the per-intersection annual costs of maintaining optimal signal timings (e.g., consulting, hardware, and software costs) for ATCS and non-ATCS sig- nals. A variety of answers were collected from respondents: from costs of maintaining an ATCS being 10 times lower than a non-ATCS costs to non-ATCS costs being 4 times lower than ATCS costs. On average, the ATCS agencies found that maintaining “optimal” signal timings under an ATCS accounts for only 75% of what is spent to maintain comparable signal timings under non-ATCS signals. EVALUATION STUDIES Evaluation studies that compare the effectiveness of a pre- ATCS traffic signal system with the effectiveness of the ATCS usually follow any new ATCS deployment. Most of the time (in 53% of the cases) ATCS users hire outside consultants to measure improvements in traffic operations FIGURE 13 Correlation between ATCS shares in budget and operations. 0 1 2 3 4 5 6 7 20,000 30,000 40,000 50,000 60,000 70,000 More ATCS Installation Costs per Intersection (US $) Fr eq u en cy FIGURE 12 Histogram of ATCS installation costs. y = 0.6459x + 0.0232 R2 = 0.7269 0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00% 90.00% 100.00% 0.00% 10.00% 20.00% 30.00% 40.00% 50.00% 60.00% 70.00% 80.00% 90.00% 100.00% Percent of Intersections under ATCS Pe rc en t o f A nn u al B u dg et fo r A TC S

gained by the implementation of an ATCS. All other times (33%), these studies are done in-house by the ATCS agen- cies. Fourteen percent of the interviewed agencies never performed any evaluations to investigate the benefits of their ATCS deployment. ATCS evaluation studies are done either in the field or by using high-fidelity microsimulation models. Commonly, field evaluations are done as a set of before-and-after studies where the “before” study reflects field conditions with pre-ATCS traffic signal control and the “after” study reflects the perfor- mance of an ATCS. Before-and-after studies are done to com- pare all sorts of alternatives that can possibly bring benefits to traffic operations. When a new ATCS is deployed, this com- parison of old versus new traffic control strategies may have another dimension. If an ATCS is installed with the ability to implement various signal timing plans, evaluators of the new system can re-apply the old traffic control timings in the new conditions for the sake of a fair comparison of the two (new and old) traffic control strategies. Most ATCSs have the ability to turn off their adaptive control algorithms and imple- ment the TOD signal timings that were in effect before the ATCS was installed. In this way, both old and new traffic con- trol strategies are exposed to (approximately) the same traffic conditions, as opposed to the before-and-after study where traffic conditions could significantly change between comple- tions of these studies. This approach to evaluating new and old traffic control is often called an Off versus On study. Off here refers to new traffic control being switched off (and instead running old TOD plans) and On refers to new traffic control being implemented. Field evaluations of ATCSs have their limitations. The experimental designs of such evaluations often lack rigor because of idiosyncratic traffic patterns that are difficult to control. However, no less difficult are the requisites for vali- dating simulation models that also depend on field conditions. Field data collection is expensive and usually cannot be as comprehensive as simulation outputs even with extensive survey instrumentation. Field evaluations typically address limited sets of traffic conditions. Furthermore, unexpected traffic conditions are by their very definition tough to capture; however, ATCSs are known for providing good performance in such circumstances. When an installation deviates from the requirements owing to a compromise by the client agency or inadequate maintenance, the ATCS is then subject to an unfair comparison. 38 ATCS evaluations through microsimulation overcome these shortcomings. Traffic conditions can be controlled tightly; they can be replicated and varied stochastically. Incident-and event-based traffic conditions can be constructed and tested carefully. Installation can be simulated to be optimal with high-quality detector placement and rigorous management of global and local control parameters, such as timing constraints. However, field evaluations exceed microsimulation evalua- tions in other aspects. For instance, there is always a margin between microsimulation and reality. Data communication between the ATCS kernel, local traffic controllers, and traffic detectors has to be emulated, which may be a difficult task. Survey results show that field evaluations are still the major way of evaluating ATCSs. Table 13 shows how popu- lar various evaluation studies are with ATCS users. “On and off” studies refer to evaluations where ATCSs are tested with full adaptive logics turned on and then turned off (with TOD plans running in the background). When performances of an ATCS and a non-ATCS traffic signal system are comparatively evaluated a variety of perfor- mance measures can be compared. Figure 14 shows the most common performance measures in such evaluations. The lead- ing performance measure is travel time (or travel delay along the travel time segment) followed by the number of stops, inter- section delays, average speeds, and queue lengths, in that order. BENEFITS FROM ADAPTIVE TRAFFIC CONTROL SYSTEMS DEPLOYMENTS ATCSs are known to have several advantages over traditional traffic signal timing operations with TOD plans. Ideally, ATCSs work best in conditions with high levels of nonrecur- ring congestion, such as incidents and special events, and in areas with fluctuating traffic demand. As mentioned earlier, an ATCS is not necessarily “the answer” for any situation. It is important to understand that it should not be expected that an ATCS deployment can totally resolve all traffic congestion issues. Instead, ATCSs could be considered as tools that can help to reduce traffic congestion by promoting the operational control and management of the transportation network. The primary area of benefits that can be achieved by an ATCS deployment is operational efficiency, measured through the reduction of delays, stops, and other negative measures of traf- fic performance. ATCS deployment improves the safety of traffic operations only indirectly—through reduction of some Type of Evaluation Study Percent of Agencies Field evaluation 89 Evaluation in microsimulation 11 Before-and-after study 65 On and off study 35 Note: The first and the second evaluation types are mutually exclusive, as well as the third and the fourth; hence, total percentage equals 200. TABLE 13 ATCS EVALUATION STUDIES

39 efficiency-related performance measures, which highly corre- late with some safety metrics (e.g., a reduction in the number of stops reduces the chance of rear-end collisions). The benefits of ATCS deployments are reported in numer- ous studies published during the last 30 years, since the first practical applications of the systems. There have been studies that reported 40% (and higher) improvement in certain per- formance measures after the ATCS has been deployed and the others that did not find any improvements or found that the operations worsened after ATCS installations. Although ATCSs have been shown to provide benefits in most cases, it is difficult to provide a detailed overview of the benefits for any of the systems, as each technology works differently, and each implementation is unique and customized to that particu- lar deployment site. Figure 15 shows a summary of the results from evaluation studies conducted at deployment sites covered by the survey. Most of the ATCS users reported (based on their evaluation studies) that their ATCSs perform much better than their previous conventional traffic signal systems [e.g., TOD plans executed through fixed-time or actuated (coordinated or isolated) traffic control]. Thirty-two percent of the respondents found that ATCSs are better, whereas 14% reported that no benefits of the ATCS were observed. Finally, only 5% of the ATCSs have performed worse than the previous type of traf- fic control system, whereas 11% of respondents did not report any findings. When these generalized findings are disaggregated into various performance measures it is found that 60% of ATCS users observe a reduction in travel times/delays when such a system is deployed. Similarly, deployments of ATCSs reduce the number of stops, intersection delays, and queue lengths in 37%, 37%, and 23% of the cases, respectively. Increases in average speeds have been observed by 35% of the ATCS users. The benefit of an ATCS in oversaturated traffic conditions is one of the most controversial aspects of the system’s perfor- mances. Many ATCS users state that the systems do not help significantly in oversaturated traffic conditions, although others have stated the opposite. The survey indicates that only a very small percentage of the interviewed agencies (3%) rec- ognize that their ATCSs prevent or eliminate oversatura- tion. The majority of the interviewed agencies reported that their ATCSs reduce or eliminate the extent of the periods of oversaturation. Approximately 33% of ATCS users have found the systems to be counterproductive in oversaturated traffic conditions. Figure 16 shows two separate categories (Ques- tion 11 under points c. and e.) that were combined to obtain this percentage (comments under “Other, please describe” mostly report that ATCSs are not useful for oversaturation). Considering the responses on ATCS benefits in oversatu- ration, and that oversaturation mostly occurs in peak periods, Arterial travel times/delays; 33; 26% Number of stops; 26; 21% Intersection delays; 21; 17% Other, please specify; 4; 3% Not applicable; 5; 4% Queue lengths; 17; 14% Average speeds; 18; 15% FIGURE 14 Most common performance measures to evaluate an ATCS.

40 ATCS is much better; 17; 38% ATCS is better; 14; 32% Neutral; 6; 14% Previous traffic control was better; 2; 5% Not applicable; 5; 11% FIGURE 15 Comparison of performances: ATCS vs. other traffic control. ATCS prevents or eliminates oversaturation; 1; 3% ATCS eliminates or reduces the extent of the periods of oversaturation; 23; 59% ATCS adversely affects the traffic conditions during periods of oversaturation; 4; 10% Oversaturation is very rare on the corridors operated by our ATCS; 2; 5% Other responses; 9; 23% FIGURE 16 ATCS performance in oversaturated traffic conditions.

41 one would expect to see that an ATCS performance in peak periods is not so beneficial. However, the survey reported that 37% of the agencies found ATCSs to be the most effective during peak periods. Twenty-three percent of the agencies found the highest ATCS benefits during the off-peak periods, whereas 23% of the ATCS users found ATCSs to be the most beneficial at the shoulders of peak traffic periods. Another 19% selected combinations of the aforementioned options. These results are slightly counterintuitive when one considers findings about weak benefits of ATCSs in oversaturation. Most likely, oversaturated conditions experienced by the interviewed agencies do not extend over the entire length of the peak periods. Therefore, because oversaturated conditions are shorter than peak periods, ATCSs are considered to be beneficial during those parts of peak periods when there is no oversaturation. ATCS operations can degrade over time owing to changes in operational traffic conditions and the inability of some of the ATCS parameters to self-adjust to those changes. Some of the ATCS’ user manuals document parameters that are sub- ject to the “ageing” process. It appears that most of the ATCSs can cope well with degradation in their performance. Most of the ATCS users (79%) find that their ATCSs have achieved a sustained level of performance since the initial installations; another 21% believe the opposite. One could note that only a few of the ATCS users have done multiple evaluation stud- ies during the course of their ATCS deployment. Multiple evaluation studies can be used to objectively doc- ument that the level of ATCS performance has been sustained. The following list shows some examples of ATCS’ users’ responses on specific benefits and costs observed in operations of their ATCS deployments. • Advantages – Fuel consumption benefits – City council sees the increase in efficiency – Thirty-three intersections resulted in more than $1 million in fuel reduction – Just the normal obvious benefits with adaptive control – Reduction in air pollution – $583,996/$542,511 one-year ratio of benefits/costs – Engineer less exposed in field – Decreased time to develop signal timings – Public transport priority and emergency vehicle priority – Lower cycle lengths—better pedestrian response – Accommodates roadwork and special events. • Negatives – Cost of software maintenance too high for the city – Save in costs of conducting traffic counts—system does it for you – Extremely expensive to maintain – Requires substantial manpower to maintain system – Additional overhead for maintenance/licensing – Costs of communications and training – Extensive experience necessary to get most benefits. PUBLIC PERCEPTION Installation of an ATCS does not readily provide observable benefits to the traveling public. If the previous conventional traffic signal control was maintained properly it may be diffi- cult for an ATCS to achieve benefits higher than 10% to 15% for any of the performance measures. Indeed, a 10% to 15% reduction in delays and stops is something that an everyday traveler may not notice easily. Despite this some agencies decide to conduct a public education campaign in which they attempt to familiarize the traveling population with the new system and its working principles. One of the primary reasons for conducting public education campaigns is that agencies implementing ATCSs would like to justify, to the travelers, reasons for potential changes in traffic signal operations and thus reduce the potential number of com- plaint calls. Travelers who are convinced that, overall, the new system provides more benefits than the old one may not com- plain as much about certain operations that may become worse through ATCS deployment. However, in spite of the good intentions, the decision to conduct an education campaign may bring more problems than benefits. Actually, at times the number of complaint calls increases after installation of a new ATCS not because the performance of the new system is worse, but because travelers subjectively believe that things have worsened with the installation of the new system. For this or similar reasons many ATCS agencies do not conduct pub- lic education campaigns. Results from our survey show that approximately 58% of ATCS users never launched an educa- tion campaign about deployment of their system. Of those agencies that conducted public education campaigns there are none that reported that the campaign was effective. Sixteen percent of the interviewed agencies found that their campaigns were somewhat effective and another 25% reported that the campaign was either neutral (23%) or slightly ineffective. Interestingly, no agency reported that the campaign was very ineffective. One might also note that deployment of an ATCS represents an action (by an ATCS agency) that is done to ease traffic congestion and improve arterial traffic operations. As such, this action has some political weight, and it is important that its promotional benefits not be underestimated. Surveying travelers to investigate the impact of newly installed ATCSs on traffic performance is not a common way to evaluate the quality of an ATCS. However, some agencies perform such a survey expecting to catch potential ATCS benefits that might be missed by common traffic engineering studies (e.g., travel time, stop, and delay studies). Of all inter- viewed ATCS users only 33% conducted any public percep- tion survey to catch benefits of ATCS deployment. Results of those surveys were reported in Table 14.

SUMMARY This chapter identified the major costs and benefits associated with ATCS deployments. Licensing costs to install propri- etary software are only 10% to 15% of the overall installation costs and they do not appear to represent a significant cost to customers. The survey results showed that ATCS instal- lation costs per intersection (approximately $65,000) are higher than reported in the previous literature (approximately $40,000). Interestingly, ATCSs require less money than con- ventional traffic signals for their physical maintenance, when their shares in overall budget and overall operations are com- pared. This finding contradicts the common belief in the traf- fic signal community, where ATCSs are known for the costly maintenance of their detectors and communication systems. Most agencies prefer to hire outside consultants for ATCS evaluations, who mainly perform field evaluations through a set of before-and-after studies. Other alternative methods of performing evaluations are much less frequently represented. 42 The chapter also provided a list of the most common perfor- mance measures collected during the ATCS evaluations. The majority of the users’ evaluations (71%) reported that ATCSs outperformed conventional traffic signal systems. However, the benefits of ATCS deployments are not observable in over- saturated traffic conditions. Although ATCS users find that their systems delay the start of oversaturation and reduce its duration, ATCSs are not recognized as a cure for oversatu- rated traffic conditions. Most of the users do not perceive that the performance of their ATCSs degrade over time—a find- ing that is quite remarkable considering some of the user com- ments. Public education campaigns about ATCS deployments are not very common or effective, as indicated by most of the ATCS users. Also, not many of the ATCS agencies conduct public perception surveys. Those agencies that do reported that results from such surveys are supportive approximately 60% of the time. The next chapter focuses on lessons learned communicated through user perspectives on several questions from the survey. Results from Public Perception Survey Percent of Agencies Clearly supportive 21 Somewhat supportive 36 Neutral 36 Unsupportive 7 TABLE 14 PUBLIC PERCEPTION SURVEY ON ATCS DEPLOYMENTS