Analytical Procedures for Determining the Impacts of Reliability Mitigation Strategies (2012)

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.

1Project Background The fundamental objective of SHRP 2 Project L03 was to develop predictive relationships between highway improvements and travel time reliability. In other words, how can the effect of an improvement on reliability be predicted? Alternatively, how can reliability be characterized as a function of highway, traffic, and operating conditions? A variety of challenging issues have been confronted in addressing this objective. Significance of Travel Time Reliability in Transportation System Performance Reliability is important to travelers and transportation practitioners for a variety of reasons: • From an economic perspective, reliability is highly important because travelers must either bud- get extra time for their trips to avoid arriving late or suffer the consequences of being late. This extra time has value beyond the average travel time used in traditional economic analyses. Recent work has documented that reliability has value to travelers and influences their behavior (1, 2). • Because of the extra time required in planning trips and uncertainty about the amount of time actually needed for a trip, reliability influences decisions about where, when, and how travel is made. • Transportation planners and operators need to include the extra economic cost of unreliable travel on users in project planning, programming, and selection processes. This is particularly true of strategies that deal directly with roadway events (e.g., incidents). In the past, most assessments of these types of strategies have missed this important aspect of travel. New Concept of Travel Time Reliability Although use of travel time–based performance measures in planning and operations applica- tions has taken on greater significance in the past few years, travel time reliability—how consis- tent (or variable) travel conditions are from day to day—is a relatively new concept to which much of the transportation profession has had only limited exposure. Congestion has been growing nationwide, and planners increasingly have become involved in short-term activities such as performance monitoring, as well as operations and management strategies. These activ- ities have been elevated in importance by transportation agencies in order to be responsive to the demands of the public and state legislatures. Anecdotal reports and technical studies indicate that average congestion levels have grown, and continue to grow, in our cities. However, talking about typical or average conditions in a transportation system that experi- ences wide performance fluctuations tells only part of the story. The notion of travel time Executive Summary

2reliability has taken on increasing importance as variation in travel times is now understood as a separate component of the public’s and business sector’s frustration with congestion problems. Reliability is a major part of system performance and of travelers’ perceptions of performance. It has not been widely used to describe performance, but increasingly agencies are recognizing its value in assessing their own performance and in communicating performance to the public. Defining Travel Time Reliability The Future Strategic Highway Research Program (F-SHRP) defined highway travel time vari- ability as synonymous with reliability: . . . from a practical standpoint, travel-time reliability can be defined in terms of how travel times vary over time (e.g., hour-to-hour, day-to-day). This concept of variability can be extended to any other travel-time-based metrics such as average speeds and delay. For the purpose of this study, travel time variability and reliability are used interchangeably. (3) A slightly different view of reliability is based on the notion of the probability of failure, which is often used to characterize industrial processes. With this view, failure is defined in terms of travel times, and the number of times a given threshold is not achieved or exceeded can be counted. In recent years, some non-U.S. reliability research has defined the probability of failure in terms of traffic flow breakdown. A corollary concept, vulnerability, is a measure of how vulnerable the network is to breakdown conditions. This concept can be applied at the link or network level. Understanding Travel Time Reliability To understand travel time reliability, it is essential to understand the factors that cause travel times to be unreliable. L03 research built on what the original F-SHRP Reliability Research Plan identified as seven sources of congestion that cause travel times to be unreliable and contribute to total congestion: incidents, inclement weather, work zones, special events, traffic control device timing, demand fluctuations, and inadequate base capacity. These categories were devel- oped to move away from the more general recurring–nonrecurring nomenclature. Operational Strategies and Capacity Expansion This project studied operational strategies and capacity expansion projects, both of which were expected to affect reliability. Many operational strategies are aimed specifically at the factors that cause unreliable travel (e.g., incident management, work zone management). It is generally expected that adding capacity will affect reliability. Travel Time Measurements Travel time is the starting point for sound congestion measurement because it reflects the actual experience of system users. When measured directly, it is independent of theoretical capacity concerns, such as what happens in oversaturated conditions. Further, once travel time is obtained, a whole family of additional measures can be created using basic information (e.g., volume, free- flow speed) about the system. Delay is one example of the metrics that naturally derives from travel time measurements. Project Approach Data Collection The research team undertook an empirical approach based on their familiarity with the data used to characterize congestion and reliability and the sufficient quality and amount of data available. Reliability can only be characterized by a long history of travel times, and the use of

3automated equipment is the only feasible method of data collection. Because of the cost of col- lecting new data, the team relied on data already being collected by transportation agencies, primarily in support of operations programs. Figure ES.1 shows the distribution of travel times along a section of highway. This distribu- tion, and the statistics that describe it, are the basis for research. The statistics superimposed on the distribution in the figure represent the reliability metrics used in the research. P10, P90, and P95 are the 10th, 90th, and 95th percentiles, respectively, of the distribution. The remaining metrics are defined elsewhere in this report. A very large data set (Figure ES.2), most of which covered urban freeways, was assembled from various traffic management centers (Tables ES.1 through ES.3). A separate data set for urban freeways was compiled for the Seattle area for the congestion by source analysis. Data on the basic characteristics of incidents were available from three sources and were used to varying degrees, depending on the team’s assessment of the data sources for each city’s situa- tion. Incident data were available from a private vendor, Traffic.com. Incident and event data were provided to the project team by Traffic.com at no cost from their traveler information manage- ment system. This system provided a standardized source of information for traffic incidents, Figure ES.1. Reliability is defined by how travel times vary. Figure ES.2. The analysis data set fused data from a variety of sources.

4events, scheduled and unscheduled construction, and other events that could affect traffic condi- tions (e.g., severe weather or transit delays). Weather data consisting of hourly weather observa- tions (e.g., precipitation, temperature, wind, fog) at multiple points within the urban areas were obtained from the National Climatic Data Center of the National Oceanic and Atmospheric Administration. Geometric data were obtained from satellite imagery (lane configurations) and the 2007 Highway Performance Monitoring data. Operating and improvement data were obtained directly from state departments of transportation. The most important data in this category were those elements related to calculating capacity for each individual link. Table ES.1. Urban Freeway Study Section Summary City Number of Directional Study Sections Total Directional Mileage Houston, Texas 13 58.80 Minneapolis, Minnesota 16 62.63 Los Angeles, California 3 50.27 San Francisco Bay Area, California 4 19.98 San Diego, California 6 28.04 Atlanta, Georgia 10 54.66 Jacksonville, Florida 8 17.71 Total 60 292.09 Table ES.2. Signalized Arterial Study Sections Travel Time Data City Arterial From To Length (mi) Data Technology Period Orlando, Florida Section 1: SR 50 eastbound Florida Turnpike SR 408 West 6.85 Tag-based probe March 2008+ Section 2: SR 50 westbound SR 408 West Florida Turnpike 6.85 Tag-based probe March 2008+ Section 3: U.S. 441 northbound SR 417 SR 408 10.67 Tag-based probe March 2008+ Section 4: U.S. 441 southbound SR 408 SR 417 10.67 Tag-based probe March 2008+ Section 5: U.S. 441 northbound SR 408 N. John Young Parkway 4.35 Tag-based probe March 2008+ Section 6: U.S. 441 southbound N. John Young Parkway SR 408 4.35 Tag-based probe March 2008+ Los Angeles, California Santa Monica Boulevard I-405 N. Gardner Street 6.9 GPS probe (Inrix) 2006–2007 Phoenix, Arizona E. Camelback Road 44th Street Highway 51 4.2 GPS probe (Inrix) 2006–2007 Minneapolis, Minnesota Washington Avenue County Highway 153 U.S. 65 3.4 GPS probe (Inrix) 2006–2007 Miami, Florida U.S. 1 17th Avenue Le Jeune Road 3.8 GPS probe (Inrix) 2006–2007 Houston, Texas Westheimer Road W. Sam Houston I-610 6.9 GPS probe (Inrix) 2006–2007 Note: GPS = global positioning system. Probe tag technology provides direct estimates of travel time over the segment. Inrix-provided data are supplied as speed estimates by link (approximately 0.5- to 1-mile long). Only the Orlando sections were used in the analysis because of sample size limitations on the other sections.

5Analysis Approach The analysis was based on a conceptual model previously developed by members of the research team (Figure ES.3). As the model indicates, the sources of congestion interact to produce total congestion. Reliability, an aspect of total congestion, is greatly influenced by the complex inter- actions of traffic demand, physical capacity, and roadway events. The analysis proceeded with four tracks: 1. Exploratory analysis, which was used to improve the understanding of reliability and establish many of the research parameters; 2. Before-and-after studies on selected study sections that resulted in empirical measurements of the change in reliability; 3. Cross-sectional statistical modeling, which was used to develop statistically based predictive models of reliability as a function of traffic, operating, and geometric conditions. The cross- sectional modeling was extremely important because it allowed a study of all of the possible improvement types in the field using a before-and-after approach; and 4. Congestion by source analysis, which was a microlevel approach to separating daily congestion into its component sources. Table ES.3. Rural Freeway Study Sections Travel Time Data State Route From To Length (mi) Data Technology Period Texas I-45 Exit 213, Navarro County Exit 267, Ellis County 54.1 GPS probe (Inrix) 2006–2007 South Carolina I-95 South Carolina– Georgia border SR 68, Hampton County 38.2 GPS probe (Inrix) 2006–2007 No. Figure ES.3. A model of congestion and its sources.

6Findings Data Set Compilation and Usage The large and comprehensive data set included many levels of aggregation and summarization. Traffic data from urban freeways comprised the largest portion of the data set and included the original measurements from roadway detectors (5-minute intervals by lane), numbering in the hundreds of millions of records. The traffic data were also summarized at several spatial and temporal aggregation levels. The most-summarized portion of the data set was the one used for the cross-sectional statistical analysis: every record is an annual summary of traffic and reli- ability characteristics, with annual event characteristics and roadway features merged into it. The data processing included new procedures specifically created by the research team for the project. Exploratory Analyses A large variety of exploratory analyses were undertaken before the main analyses to test assump- tions, develop data processing methods, and more thoroughly understand the manifestation and ramifications of reliability. Recommended Reliability Metrics The Travel Time Index (TTI) is the ratio of the actual travel time to the ideal or free-flow travel time. Based on empirical tests, it was found that the performance metrics defined in the early stages of the research were sensitive to the effects of improvements. The 95th percentile TTI may be too extreme a value to be influenced significantly by operations strategies, but the 80th per- centile was more sensitive to these improvements. As a result, the 80th percentile TTI was added to the list of reliability performance metrics for the remainder of the research. The final set of reliability metrics, which also are appropriate for general practice, appears in Table ES.4. Travel Time Distributions Developing travel time distributions is the starting point for defining reliability metrics. Travel time distributions also allow for visualization of general congestion and reliability patterns for a Table ES.4. Recommended Reliability Metrics Reliability Performance Metric Definition Units Buffer Index Difference between 95th percentile TTI and average travel time, normalized by average travel time. Difference between 95th percentile TTI and median travel time (MTT), normalized by MTT. % Failure and on-time measures Percentage of trips with travel times <1.1 MTT and <1.25 MTT. Percentage of trips with space mean speed less than 50, 45, and 30 mph. % Planning Time Index 95th percentile TTI. None 80th Percentile TTI Self-explanatory. None Skew statistic (90th percentile TTI - median)/(median - 10th percentile TTI). None Misery Index (modified) Average of highest 5% of travel times divided by free- flow travel time. None

7highway section or trip. An examination of the distributions from the research study section reveals several characteristics: • The shape of the travel time distribution for congested peak times (nonholiday weekdays) is much broader than the sharp spike evident in uncongested conditions. The breadth of this broad shoulder of travel times decreases as the congestion level decreases. • Likewise, the tails of the distributions (to the right) appear more exaggerated for the uncon- gested time slices. However, the highest travel times occur during the peaks. • Despite the fact that peaks have been defined, a number of trips still occur at close to free flow; there are more of these trips in the peak period than in the peak hour (see discussion below of peak period and peak hour). This is probably because peak times actually shift slightly from day to day, as traffic demand can be shifted by events. Data Requirements for Establishing Reliability Because reliability is defined by the variability of travel conditions (travel time), it must be mea- sured over a substantial portion of time to allow all of the influences of random events to be exerted. Tests showed that an absolute minimum of 6 months of data is required to establish reliability within a small error rate in areas where winter weather is not a major factor. A full year of data is preferred. Supplemental Reliability Metric The Atlanta study (detailed in Chapter 5) raised doubts about the use of the Buffer Index as the primary metric for tracking trends in reliability. The problem comes from how the Buffer Index is calculated: it is the buffer time (difference between the 95th percentile and the mean) normal- ized by the mean. The Buffer Index is considered to be too erratic and unstable for use as the primary reliability metric for tracking performance trends or for studying the effects of improvements. However, as a secondary metric, it provides useful information; rather than being discarded, it should be included in a suite of reliability performance metrics. Defining Peak Hour and Peak Period Most studies of reliability and congestion have defined fixed time periods for the peak hour and peak period. However, the research team decided that the most appropriate method would be to define peak hour and peak period specifically for each study section. The team used a definition based on the most typical start and end times of continuous congestion. The resulting time slices were reviewed against local anecdotal knowledge and required very little adjustment. Estimating Demand in Oversaturated Conditions on Freeways Because the study took an empirical approach, the team had to deal with the thorny issue of how to measure demand given that measured volumes under congested flow are actually less than capacity on freeways. A method for assigning the demand stored in queues during periods of flow breakdown was developed and used, particularly in defining the demand-to-capacity ratio for the statistical modeling. Reliability Breakpoints on Freeways It was shown that travel time reliability on a freeway is not a function of counted traffic volumes until a breakpoint volume is reached. Once the breakpoint volume is exceeded, the decrease in

8travel time reliability (increase in the variance) is extreme and abrupt enough to suggest it is a vertical function, with a nonsingular relationship to further volume increases. The breakpoint volume varies significantly between facilities and even within the same freeway facility (by loca- tion and direction of travel on the same facility). The breakpoint in reliability generally occurs at a counted volume significantly lower than the theoretical capacity of the facility computed according to the method in the Highway Capacity Manual (HCM). But this peaking effect does not entirely explain the difference. The breakpoint volume is significantly lower than the theoretical capacity partly because most freeway sections are upstream of a bottleneck and, thus, are affected by downstream congestion backing up into the subject section long before the subject section’s HCM capacity is reached. Further, traffic- influencing events (especially incidents) effectively lower capacity when they occur, and over time they cause reliability to degrade. This effect manifests itself in lower breakpoint volumes than capacity related strictly to physical features. Finally, even for bottlenecks, the data suggest that the reliability breakpoint occurs long before the theoretical HCM capacity of the bottle- neck is reached. Sustainable Service Rates on Freeways Just as travel times vary over time, capacity is not a fixed value but also varies over time. The same factors that influence reliability affect capacity variability. The research did not specifically tease out all the factors, but they all are imbedded in the final capacity distributions. The team devel- oped a large set of capacity distributions that look roughly like travel time distributions in reverse: the tail of the distribution is skewed to the left (lower capacity values) rather than to the right. Because these distributions were developed from year-long data measurements, they include the effect of the influencing factors, resulting in capacity values that could be used in a stochastic framework to model congestion and reliability. Travel Time Distributions on Urban Freeways, Signalized Arterials, and Rural Freeways An analysis of travel time distributions for different time slices and congested levels revealed the following characteristics: • All distributions feature a tail that is skewed to the right (i.e., higher travel times). Most of these abnormally high travel times can be attributed to one or more of the sources of conges- tion; that is, they occur in the presence of an event(s) and/or high demand; • Uncongested periods are characterized by a sharp peak of travel time frequencies near the free-flow speed; • When congestion dominates the time slice (e.g., peak hour, peak period), the travel time dis- tribution becomes more broad and less peaked; • Travel time distributions on signalized arterials are uniformly broad in shape, even for rela- tively low levels of congestion; and • As trips become longer, the travel time distributions assume the typical uncongested shape. Vulnerability to Flow Breakdown Examination of the 5-minute data at individual stations (groups of detectors in a direction on a highway segment) reveals that just 20 to 45 minutes before the start of what is considered the normal peak period, there is an upsurge in the 95th percentile TTIs. This upsurge begins before the uptick in average travel times and indicates that this window of time is vulnerable to flow breakdown. These windows are extremely important for operators to focus on, as breakdowns during this time will strongly influence the duration and severity of the peak.

9Reliability of Urban Trips Based on the Reliability of Links For extended travel on urban freeways (trips of 10 to 12 miles in length), the reliability of the entire trip can be predicted as a function of the reliability of the links that comprise the trip. Although not specifically tested, it should be possible to construct trip reliability for trips that include other types of highways in addition to freeways, subject to the issue of time dependency for long trips. Before-and-After Studies on Selected Study Sections The primary goal of the research was to develop relationships for predicting the change in reliabil- ity due to improvements. The best way to accomplish this goal was with controlled before-and-after studies. However, such analyses offer a substantial challenge because of their data requirements: to establish reliability empirically, at least 6 to 12 months of data are required. The preferred data col- lection period is 12 months, including a long period of continuously collected data before and after the improvement. Instead of designing traditional before-and-after experiments, the team concen- trated on collecting continuous traffic data from areas with quality data, interesting congestion, and good records of event data. The team identified 17 cases of improvements that coincided with the identified data, although the types of improvements were somewhat limited: • Ramp meters—four; • Freeway service patrol implementation—two; • Bottleneck improvement—three; • General capacity increases—five; • Aggressive incident clearance program—two; and • High-occupancy toll (HOT) lane conversion—one. The analysis produced reliability adjustment factors that can be applied to the various improve- ments (Table ES.5). A global finding from the before-and-after analyses is that all forms of improvements, includ- ing capacity expansion, affect both average congestion and reliability in a positive way (i.e., aver- age congestion is reduced and reliability is improved). Conceptually, this makes sense: one of the seven sources of congestion and reliability identified earlier was the amount of base capacity. All things being equal, more capacity (in relation to demand) means that the roadway is able to absorb the effects of some events that would otherwise cause disruption. The size of this effect was greater than originally anticipated; that is, a large part of the benefits of capacity expansion projects greatly contributes to the value of reliability. Cross-Sectional Statistical Modeling Because only a limited number of before-and-after studies were possible, much of the effort for the study went into the creation of a cross-sectional data set from which statistical models could be developed. The final analysis data set for the statistical modeling is highly aggregated: each record represents reliability, traffic, and event data summarized for a section for a year, and reli- ability is measured as the variability in travel times over the course of a year. As such, the cross- sectional model is a macroscale model; it does not seek to predict the travel time for a particular set of circumstances, and it is not appropriate for real-time travel time prediction. For example, it does not suggest an expected travel time if incident and demand characteristics for a given day are known. Rather, it seeks to predict the overall travel time characteristics of a highway section in terms of both mean and reliability performance. It is appropriate for adaptation to many exist- ing models and applications that seek similar predictions, and it can serve as the basis for con- ducting a cost–benefit analysis.

10 Table ES.5. Summary of Urban Freeway Before-and-After Studies Case No. Urban Area Highway Covered Improvement Reliability Impacts (Peak Period) 1 Los Angeles I-210 Ramp metering: design, field implementation, and evaluation of new advanced on-ramp control algorithms on westbound I-210. Slight increases in average travel time and Planning Time Index (PTI) were observed. However, subsequent to this evaluation, the algorithms have been adjusted. 2 San Francisco Bay Area I-580 Ramp metering. 22% reduction in average travel time. 20% reduction in PTI. 3 Seattle SR 520 Ramp metering. 11% reduction in average travel time. 12% reduction in PTI. 4 Atlanta I-285, Northern Arc Ramp metering. 9% reduction in average travel time. 7% reduction in PTI. 3% increase in sustainable service rate. 5 Atlanta All freeways inside beltway perimeter Incident management: incentive program for reducing large-truck crash incident duration (90 minutes). 13% reduction in large-truck crash incident duration. 9% reduction in lane hours lost per large- truck crash. 6 Los Angeles I-710 Incident management: evaluation of pilot project to deploy towing service for big-rig tractor trailers. 10% reduction in average travel time. 20% reduction in PTI. 7 San Diego I-8 Incident management: expansion of the existing Freeway Service Patrol Beat-7 on I-8. 3% reduction in average travel time. 4% reduction in PTI. 8 San Diego SR 52 Incident management: expansion of the existing Freeway Service Patrol. 20% reduction in average travel time. 10% reduction in PTI. 9 Minneapolis– St. Paul I-94 Capacity expansion: add third lane in each direction. 43% reduction in average travel time. 46% reduction in PTI. 10 Minneapolis– St. Paul I-494 Capacity expansion: add third lane in each direction. 31% reduction in average travel time. 16% reduction in PTI. 11 Minneapolis– St. Paul I-394 Capacity expansion: add auxiliary lanes westbound. 35% reduction in average travel time. 38% reduction in PTI. 12 Minneapolis– St. Paul Highway 169 Capacity expansion: convert signalized inter- sections to diamond interchanges. 16% increase in average travel time. 11% reduction in PTI. 13 Minneapolis– St. Paul Highway 100 Capacity expansion: add third lane north- bound; add auxiliary lane southbound; convert Highway 7 interchange from a clover leaf to a folded diamond. 20% reduction in average travel time. 30% increase in PTI. 14 Seattle I-405 southbound Capacity expansion: addition of one general- purpose lane. 11% reduction in average travel time. 11% reduction in PTI. 15 Seattle I-405 northbound Capacity expansion: addition of one general- purpose lane. 42% reduction in average travel time. 35% reduction in PTI. 16 Seattle I-405/SR 167 interchange Capacity expansion: grade separation ramp connecting southbound I-405 off-ramp with southbound SR 167 on-ramp. 20% reduction in average travel time. 23% reduction in PTI. 17 Minneapolis– St. Paul I-394 HOT lane conversion. 8% reduction in average travel time. 30% reduction in PTI. Note: Long study segment = 16 miles; study section influenced by downstream bottleneck.

11 Two model forms were developed: simple and complex. The simple model form relates all the reliability metrics to the mean TTI for the three highway types studied (urban freeways, rural freeways, and signalized arterials). These relationships are convenient for many applications that produce mean travel time–based measures as output. Because the mean TTI developed from the research data included the effects of all the possible influences of congestion, which produced a mean value greater than model results that usually are for typical (nonextreme) conditions, an adjustment factor was developed to convert model output to the overall mean TTI so that the relationships could be applied. An example of the strong relationship between mean TTI and 95th percentile TTI is shown in Figure ES.4. A more detailed model form was developed that related reliability measures to the factors that influence reliability. A series of statistical predictive models was developed that related the reli- ability metrics over highway sections (multiple links, usually 4 to 5 miles long) to • The critical demand-to-capacity ratio (maximum from the individual links); • Lane hours lost due to incidents and work zones combined (annual); and • Number of hours during which rainfall was ≥0.05 inch (annual). Models were developed for the peak hour, peak period, midday, and weekday time periods. Guid- ance was developed from readily available data on how to estimate demand, capacity, and lane hours lost. Guidance was also provided on how improvements affect changes in the models’ independent variables. The model structure is flexible and can easily incorporate new research on the effects of transportation improvements on reliability. Congestion by Source An assignment of congestion causality was made for the measured delay in the Seattle data (detailed in Chapter 5). Taken at face value, these data support the common thinking that inci- dents and crashes cause between 40% and 60% of all delay. In reality, a considerable portion of the delay associated with incidents and crashes is caused by large traffic volumes. Therefore, the amount of delay caused by incidents is actually less than can be reasonably assigned by simply observing the occurrence of events. Numerous examples in the analysis data set of significant crashes and other incidents caused little or no congestion because of when they occurred. With- out sufficient volume, an incident causes no measurable change in delay. Figure ES.4. Section-level relationship for mean TTI and 95th percentile.

12 In the Seattle area, many incidents take place during peak periods, causing already existing congestion to grow worse as a result of the interwoven effects of incidents, bad weather, and traf- fic volumes on travel times. In addition, all types of disruptions to normal roadway performance (rain, crashes, and noncrash incidents) cause congestion to start earlier and last longer during the peak period, while increasing travel times during the normally congested times. Incidents and other disruptions also can cause congestion to form during times that are normally free from congestion. However, congestion only forms when the disruption lowers functional capacity below traffic demand. Thus volume, relative to roadway capacity, is a key component of conges- tion formation, and in urban areas it is likely to be the primary source of congestion. Disruptions significantly increase the delay that the basic volume condition creates. The fact that traffic volume is the basis of congestion also affects how various traffic disrup- tions affect travel patterns. Not only does traffic volume affect whether an incident causes con- gestion, but it affects how long that congestion lasts once the primary incident has been removed. The Seattle data showed that in the morning peaks, disruptions have a more noticeable effect on the timing of the end of the peak period, while in the evening the opposite is true. Analysis of 42 roadway segments in the Seattle area showed that a majority of travel delay in the region is the direct result of traffic volume demand exceeding available roadway capacity. Whenever they occur, incidents, crashes, and bad weather add significantly to the delays that can be otherwise expected. The largest of these disruptions plays a significant role in the worst travel times that travelers experience on these roadways. However, the relative importance of any one type of disruption varies considerably from corridor to corridor. In peak periods, incidents add only a marginal percentage increase to total delay, but they shift when and where those delays occur, as well as who suffers from those delays. That is, many inci- dents shift where a normally occurring bottleneck occurs, freeing up some roadway sections, while causing others to suffer major increases in congestion. But taken as a total, if a roadway section is normally congested, the added delay from incidents is modest (at least in Seattle) com- pared with the daily delay from simply too many vehicles for the available physical capacity. In congested urban areas, traffic incidents more often cause unreliable traffic patterns than increases in total delay. While the total delay value does goes up, the big change is often the shift in who gets delayed. For an individual severe incident, many travelers may value the extra (unplanned) delay highly, and are more likely to remember these extreme cases. However, some of that (total) delay is offset by other travelers who reach their destination early because their trip is downstream of the incident-caused bottleneck, and consequently volume on their down- stream trip segment has probably been metered by the upstream bottleneck. Significance of Demand for Reliability Estimation A major finding was that demand (volume) is an extremely important determinant of reliabil- ity, especially relative to capacity. Demand’s interaction with physical capacity is the starting point for determining congestion. The research team initially postulated that the effect of most events would be determined by the level of demand under which they occurred. For example, if an incident or work zone blocked a traffic lane, the impact would only be felt if volumes were high enough to be affected by the lost capacity. Although demand was not expected to have a strong effect, it emerged as a significant factor throughout the various analyses. The influence of demand is probably related not only to the sheer volume of traffic but also the volume’s characteristics. As volumes approach theoretical capacity, traffic flow becomes unstable and increasingly susceptible to breakdown from even small changes. These small changes can occur at a point substantially less than theoretical capacity; when they occur near potential bottleneck areas such as on-ramps, weaving areas, and lane drops, the team postulates that their effect is enhanced. In addition to variations in demand as a source of unreliable travel times, evidence exists that physical capacity also varies. The work of Brilon and preliminary research conducted by other

13 SHRP 2 contractors suggest that capacity varies even in the absence of disruptions (4). The team postulates that fluctuations in traffic conditions at a microscale are the most likely causal factors for variations in capacity. There are several implications of the findings that demand and capacity strongly influence travel time reliability: • Capacity additions and demand reductions will improve congestion on nearly all days. Strat- egies geared to disruptions (e.g., incident management) will only affect congestion when those disruptions appear; • Demand management strategies (e.g., pricing) also will lead to improvements in reliability; and • Accounting for volumes in relation to available capacity can provide a tool for efficiently allo- cating operations strategies, particularly incident management. Reliability As a Feature of Congestion The intertwined relationship between demand, capacity, and disruptions documented in the L03 research led to another major conclusion: reliability is a feature or attribute of congestion, not a distinct phenomenon. Any influence on congestion that leads to unreliable travel reliability can- not be considered in isolation. Reliability has generally been considered to be related primarily to disruptions and the operational treatments aimed at those disruptions. However, analysis showed that a substantial amount of variability in travel times exists for recurring (e.g., bottleneck-related) conditions. Therefore, the most inclusive view of travel time reliability sees it as part of overall congestion. Just as congestion can be defined by extent and severity, it can also be defined by how it varies over time. Operational treatments are effective in dealing with unreliable travel times, but so are other congestion-relief measures. References 1. Joint Transport Research Centre. Improving Reliability on Surface Transport Networks: Summary Document. International Transport Forum, Organisation for Economic Co-operation and Development, Paris, 2009. www.internationaltransportforum.org/jtrc/infrastructure/networks/ReliabilitySum.pdf. 2. Fosgerau, M., and A. Karlström. The Value of Reliability. Transportation Research Part B, Vol. 44, No. 1, 2010, pp. 38–49. 3. Cambridge Systematics, Inc., Texas Transportation Institute, University of Washington, and Dowling Associ- ates. Providing a Highway System with Reliable Travel Times: Study 3—Reliability. Final report, NCHRP Proj- ect 20-58(3). Transportation Research Board of the National Academies, Washington, D.C., 2003. http://trb .org/publications/f-shrp/f-shrp_webdoc_3.pdf. 4. Brilon, W., J. Geistefeldt, and H. Zurlinden. Implementing the Concept of Reliability for Highway Capacity Analysis. In Transportation Research Record: Journal of the Transportation Research Board, No. 2027, Transpor- tation Research Board of the National Academies, Washington, D.C., 2007, pp. 1–8. http://trb.metapress.com/ content/u700713ur834410r/fulltext.pdf.