Effectiveness of Different Approaches to Disseminating Traveler Information on Travel Time Reliability (2013)

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.

15 In the many cities where congestion on the transportation system is commonplace, drivers are accustomed to conges- tion and expect and plan for some increase in travel time, particularly during peak driving times. Many system users either adjust their schedules to avoid peak hours or budget extra time to allow for unexpected traffic congestion or inci- dents. However, problems arise when travel times are much higher than anticipated. Most travelers are less tolerant of unexpected travel time increases because those longer travel times cause travelers to be late for work or important meet- ings, to miss appointments, or to incur extra child-care fees. Moreover, shippers that face unexpected delays may lose money, experience disruptions in just-in-time delivery and manufacturing processes, and lose their competitive edge (Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006). Transportation professionals most commonly discuss travel time reliability in terms of historical average travel times cal- culated over periods of a year or longer, as illustrated in Fig- ure 2.1. A typical definition for travel time reliability is the following: The consistency or dependability in travel times, as mea- sured from day to day or across different times of the day. However, most travelers do not experience the same average travel time each day. As shown in Figure 2.2, travelers experi- ence and remember something much different than the aver- age throughout a year of commutes. Their travel times vary greatly from day to day, and they remember the few bad days they suffered through unexpectedly longer travel times. Research within the profession has shown that travel time reliability information can provide transportation system users with a more complete picture of the expected travel time along a particular route. The challenge is how to com- municate that reliability information effectively to road and transit system users so that they understand it clearly. Another example illustrating travel time reliability is shown in Figure 2.3, which presents travel time data from a major commuter route in Seattle, Washington. Without congestion along the route, travel times are about 12 min (e.g., see Presi- dent’s Day in the figure). On all other weekdays, the average travel time is 18 min. However, when traffic incidents and weather combine to cause unexpected congestion, travel times may be 25 min or more, or 39%, longer than usual. Com- muters who travel this route must plan for this variability if they want to arrive on time. If they plan their commute on the basis of the average travel time, they will be late half the time and early the other half of the time. In other words, com- muters have to build in a time cushion, or buffer, to their trip planning to account for the variability. If they build in a buffer, they will arrive early some days. That is not necessarily a bad thing, but the extra time is carved out of their day—time they could be using on pursuits other than commuting. Travel Time reliability Metrics The measurement of travel time reliability is an emerging practice. However, a few measures appear to have technical merit and are thought to be easily understood by nontechni- cal audiences. Most of these measures compare high travel time days with average travel time days. Four recommended measures are as follows (Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006): • 90th or 95th percentile travel time; • Buffer index; • Planning time index; and • Frequency the congestion exceeds some expected threshold. The 90th or 95th percentile travel time is a time identified for a specific travel route that indicates how long the delay will be on the heaviest travel days (Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006). These travel C H a P T e r 2 Literature Review

16 times are reported in minutes and seconds and are thought to be easily understood by commuters familiar with their trips. Therefore, this measure is ideally suited for traveler informa- tion. It has the disadvantage of not being easily compared across trips, because most trips have different lengths. Nor can this measure be used to easily combine route or trip travel times into a subarea or citywide average. Other reliability indi- ces presented below do enable comparisons or combinations of routes or trips with different lengths. The buffer index represents the extra time cushion (or buff- er) that most travelers add to their average travel time when planning trips to account for unforeseen delays and to ensure on-time arrival (Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006). The buffer index is expressed as a percentage, and its value increases as reliability worsens. For example, a buffer index of 40% means that for a 20-min average travel time, a traveler should budget an additional 8 min (20 min × 40% = 8 min) to ensure on-time arrival most of the time. In this example, the eight extra minutes is called the buffer time. The buffer index is computed as the difference between the 95th percentile travel time and average travel time, divided by the average travel time. The planning time index represents the total travel time that a traveler should expect or plan on when an adequate Source: Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006. Jan DecJuly Travel Time How traffic conditions have been communicated Annual average Figure 2.1. Average travel time used by professionals. Jan DecJuly Travel Time What travelers experience… Travel times vary greatly day-to-day … and what they remember Source: Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006. Figure 2.2. Traveler travel time experiences. State Route 520 Eastbound, Seattle, 5 to 6 pm 0 5 10 15 20 25 30 January February March April Weekdays in 2003 Tr av el T im e (m in ut es ) Martin Luther King Day President's Day 2 incidents with rain 3 incidents 1 incident with rain 4 incidents 1 incident rain Free-flow travel time = 12 minutes Average travel time = 18 minutes Source: Cambridge Systematics, Inc. with Texas Transportation Institute 2005. Figure 2.3. Example of commuters planned trips based on the worst days, not an average day.

17 buffer time is included (Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006). The planning time index differs from the buffer index in that it includes typical delay as well as unexpected delay. Thus, the planning time index compares near-worst-case travel time with a travel time in light or free-flow traffic. For example, a planning time index of 1.60 means that for a 15-min trip in light traffic, the total time that should be planned for the trip is 24 min (15 min × 1.60 = 24 min). The planning time index is useful because it can be directly compared with the travel time index (a mea- sure of average congestion) on similar numeric scales. The planning time index is computed as the 95th percentile travel time divided by the free-flow travel time. From a data perspective, continuous travel time data are the only way to establish reliability patterns empirically. Although predictive methods—such as the ones developed by the project team for the SHRP 2 L03 final report, Analytic Procedures for Determining the Impacts of Reliability Mitigation Strategies— may be used in a reliability monitoring system when the data are unavailable, only continuously collected travel time data can produce the actual travel time distribution from which all reliability metrics are derived. For example, the reliability met- rics being used in the SHRP 2 L03 project, as shown in Table 2.1, are all derivatives of the travel time distribution. At present, agreement is lacking within the professional field on the terms to be used and what the mathematical cal- culations of each term should be. If the professionals cannot reach consensus on the technical terms, then the general public certainly will not do so. The purpose of the L14 proj- ect was to discover what terms the layperson would use to refer to travel time reliability concepts and to encourage the use of those terms in communications with transportation system users. Importance of Travel Time reliability Travel time reliability is significant to many transportation sys- tem users, whether they are vehicle drivers, transit riders, freight shippers, or even air travelers. Good and consistent system reli- ability is a valuable service that can be provided on privately operated and publicly operated highways alike. Because reliabil- ity is so important for transportation system users, transporta- tion planners, operators, and decision makers should consider travel time reliability a key performance measure. Travel Time Reliability and Highway Travel Travel time reliability is valuable to traffic professionals because it better quantifies the benefits of traffic management and operation activities than simple averages. For example, con- sider a typical before-and-after study that attempts to quantify the benefits of an incident management or ramp metering pro- gram. The improvement in average travel time may appear modest, as shown on the left side of Figure 2.4. However, reli- ability measures will show a much greater improvement—as illustrated on the right side of Figure 2.4—because they show the effect of improving the worst few days of unexpected delay. For drivers, travel time reliability information can be valu- able when they are selecting a route. For example, the value of travel time reliability was assessed through a mail survey, trip diaries, and loop-detector data by Lam and Small (2001) soon after the first high-occupancy/toll (HOT) lane opened on State Route 91 in Riverside, California. The researchers found that, for women in this study, the value of travel time reliability was actually higher than simple travel time infor- mation. For men, the value of travel time was roughly 50% Table 2.1. Recommended Reliability Performance Metrics from the SHRP 2 L03 Report, Analytic Procedures for Determining the Impacts of Reliability Mitigation Strategies Reliability Performance Metric Definition Unit Buffer index (BI), mean-based The difference between the 95th percentile travel time and the average travel time, normalized by the average travel time Percent Buffer index, median-based The difference between the 95th percentile travel time and the median travel time, normalized by the median travel time Percent Failure or on-time measures, median-based Percentage of trips with travel times less than 1.1 × median travel time and/or 1.25 × median travel time Percent Failure or on-time measures, speed-baseda Percentage of trips with travel times less than 50, 45, and/or 30 mph Percent Misery index (modified) The average of the top 5% worst travel times divided by the free-flow travel time None Planning time indices 95th, 90th, and 80th percentile travel times divided by the free-flow travel time None Skew statistic The ratio of (90th percentile travel time minus the median) to (the median minus the 10th percentile) None Source: Cambridge Systematics, Inc. (2007). a Speed is the space-mean speed over the study section.

18 for a transit vehicle as being longer than an equivalent amount of time spent riding in the vehicle. Real-time information that allows transit riders to schedule their own arrival at a transit stop and/or to monitor the wait time remaining until the vehi- cle’s arrival increases rider confidence in the service (Perk et al. 2008). Transit passengers surveyed in two cities ranked knowl- edge of when their bus would arrive and knowledge that it would arrive on time as the two most important factors affect- ing their decision to ride transit (Peng et al. 2002). Travel Time Reliability and Freight In terms of economic value, reliability is probably more impor- tant to freight carriers and shippers than to personal travelers. With the rise in just-in-time deliveries (largely as a replace- ment for extensive warehousing), providing dependable (reli- able) service has become extremely valuable, while failure to provide dependable service can increase costs considerably (Cambridge Systematics, Inc. 2007). For example, improve- ments in transportation reliability play an important role in reducing inventory in the chemical supply chain for freight shippers. Because of the many nodes in the supply chain, upwards of one-third of all chemical inventory is in transit at any point in time. Inventory managers keep safety or buffer supplies to cushion against variability of inbound arrivals, and the amount of safety supplies increases with the degree of unreliability and the number of stocking locations (Cambridge Systematics, Inc. 2007). However, the capacity to receive chemical supplies is limited by the size of the liquid storage silos. Balancing capacity with demand is a challenge. As transportation reliability decreases, wait time, dead freight, and cost increase (Cambridge Systematics, Inc. 2006). Travel Time Information: State of the Practice Real-time travel time messages have been in use in the United States for well over a decade, ever since traffic monitoring and integration systems became reliable. The most commonly higher than the value of reliability information. The reasons for this difference were not clear from the data collected, though some have interpreted the data to indicate that women have more time-critical commitments related to child-care trips. For this study, the researchers defined travel time as the 90th percentile travel time minus the median travel time. The authors discuss further how the transponder usage records of participants showed that few drivers habitually used the HOT lane. Rather, people made the decision whether to pay for the HOT lane on a daily basis depending on trip purpose and traf- fic conditions. In applications such as HOT lanes, travel time reliability information may be most useful en route to help drivers make the purchase decision to use the HOT lanes. The influence of pre-trip and en route travel information on route decisions has been demonstrated in other studies: An evalua- tion of the Washington State DOT’s 511 travel information system in 2005 found that 21% of respondents changed their original travel plans on the basis of information they got from the 511 system (PRR, Inc. 2005). Drivers on an Orlando, Florida, toll road who stated that they used information from the state’s 511 service or from DMSs (which displayed esti- mated delay times for the road) were more likely to change their route in response to unexpected congestion. A review of research on travel time and travel time reli- ability conducted by the Center of Urban Transportation Research (University of South Florida) includes the finding that most travelers value trip time reliability at least as much as actual trip time. In fact, when travelers’ arrival and depar- ture times were inflexible because of the nature of the trip, the value of reliability was as much as three times that of trip time (Concas and Kolpakov 2009). Travel Time Reliability and Transit Studies of transit ridership have shown that trip time reliabil- ity (including the reliability of a rider’s wait time at transit stops) is more important to retaining riders than the trip and waiting times themselves. Wait-time reliability is particularly important, as transit riders tend to perceive time spent waiting 2003 20052004 Travel Time Small improvement in average travel times 2003 20052004 Travel Time Larger improvement in travel time reliability Before After Before After Average day Worst day of the month Source: Texas A&M Transportation Institute with Cambridge Systematics, Inc. 2006. Figure 2.4. Reliability measures capture the benefits of traffic management.

19 a very short time period, constrained by the available sight distance and design features of the CMS (Dudek 2001, 2004, 2006). Some transportation agencies use signs that show one part of the message in static form (e.g., locations) and the travel times in changeable form. These types of signs are placed upstream of major diversion points which have no need for fully changeable message capability. Although these signs likely contain more units of information than can be fully read by unfamiliar drivers, drivers who are familiar with the area learn the static messages and concentrate only on the changeable travel times. One study of this sign type found that drivers rated the combined static/changeable format as easier to process than information displayed on a traditional CMS (Lerner et al. 2009). The credibility of the message, in this case of the travel time provided, is also a concern. Real-time travel times posted on CMSs are actually historical travel times based on the past sev- eral minutes or hours of travel speeds that have been recorded; thus they may not reflect recent changes in traffic speeds. To alleviate the potential credibility problems that can result from a large difference between posted travel time and the travel time that drivers experience, some TMCs display a time stamp to indicate when travel speeds were last calculated; others dis- play a range of estimated travel time (Houston TranStar 2012; San Antonio TransGuide 2012). Reliability Information on DMSs The travel time ranges used by some TMCs in real-time travel time messages can be one way of including some travel time reliability information along with real-time travel time infor- mation, though studies have found that drivers tend to prefer single time values to ranges and to accept that the actual travel time may vary from the single value posted (Ban et al. 2009; Phoenix Tightens Travel Time Estimates 2008). The French Ministry of Transportation in cooperation with the City of Paris experimented with another format: a travel time message that included slanted up and down arrows to indicate to drivers whether travel times were increasing or decreasing from the posted estimate (A. Hedhli, interview by S. Chrysler, May 26, 2010). Another study of travel time messages using trend arrows found that drivers took longer to process the information on CMSs when trend arrows were added and that they were more confident in the travel time value provided without a trend arrow (A. Hedhli, interview by S. Chrysler, May 26, 2010). The Long Island Expressway uses the term average travel time on its combination static/CMS signs showing travel times to multiple destinations along a single route. That term is a slight departure from the Manual on Uniform Traffic Control Devices (MUTCD), which shows signs with the term estimated travel time. Both terms are associated with travel time reliabil- ity, but in the Long Island Expressway case, the travel times used media for these messages are DMSs and transportation agency websites; but the widespread use of cell phones and other mobile devices is prompting a growing number of trans- portation agencies and providers to offer real-time updates on transportation conditions and options via e-mails, text messages, and Twitter feeds. Real-time travel time estimates are most often provided for a particular roadway segment or a particular transit route on the basis of recent travel speeds or conditions. Some agencies also provide travel time comparisons among two or more routes or roadways to help travelers make decisions about the route or transportation mode to take. Recent and rare sources are the information sources that advise travelers about travel time reliability—that is, about the likelihood that the esti- mated travel time for a particular trip or trip segment can be relied on dependably. The following subsections provide a concise summary of the state of the practice regarding travel time information. A more detailed literature review of this information is provided in Appendix A. Real-Time Travel Time Information on Dynamic Message Signs Changeable message signs (CMSs), also known as dynamic or variable message signs (DMSs or VMSs), can be used to provide several types of travel time information to drivers: • Travel time information between specified locations; • Comparative travel times for alternate routes (e.g., “Airport via Route 1—20 min, I-94—35 min”); • Time saved by taking an alternate route (e.g., “Accident at exit 12; use Route 46—save 20 min”); • Delay on the freeway; and • Delay avoided by taking the alternate route (Dudek and Huchingson 1991; Dudek 2004). Displaying travel times on CMSs is not a universal practice. A 2008 survey of 100 traffic management centers in 40 states found that only 30% displayed travel times on some of the CMSs in their jurisdictions during peak traffic periods, and only 23% displayed travel times during off-peak periods. Rea- sons for not displaying travel times on CMSs include (a) the CMS not being located where travel time messages would be useful and (b) a lack of communications infrastructure and software to maintain up-to-date information and messages. The primary lessons learned by the TMCs were that travel time information must be accurate and that displaying accu- rate travel times during rapidly deteriorating traffic conditions (e.g., transition between off-peak and peak periods, occurrence of incidents) is difficult (Dudek 2008). To be effective, a CMS must communicate a meaningful message that can be read and understood by motorists within

20 • Washington State DOT’s website includes a similar table that also displays travel times for high-occupancy-vehicle (HOV) lanes, where applicable, and displays the 95th per- centile travel time for an input roadway segment (Seattle Area Travel Times 2012). • The table of travel times on the Gary-Chicago-Milwaukee travel information website includes links from each aver- age travel time estimate to a graph displaying detailed his- torical travel time data for the corresponding roadway segment (Travel Midwest Stats 2012). • Rutgers University provides travel time reliability infor- mation for public transit routes. While most transit pro- viders generate reliability information for their own use, few pass the information on to riders (Rutgers Department of Transportation 2012). A growing number of TMCs and transit providers are communicating travel time information and traffic alerts via e-mail, Twitter, or text message. Communicating Reliability Information in Nontransportation Fields Cognitive science has shown that most people are not good at understanding statistical concepts and applying them to every- day situations such as medical diagnoses, gambling odds, or travel time probabilities (Gal 2002). Qualitative terms for con- veying statistical concepts (e.g., probably, most likely, rarely) can be interpreted in different ways, so selecting a term that has a consistent enough definition to communicate the desired mes- sage about probability to the public can be difficult (Teigen 1988; Wallsten et al. 1993; Biehl and Halpern-Felsher 2001). Medicine and weather forecasting are two fields that deal with probabilities and statistical concepts and must find ways of effectively communicating those concepts to the public. Graphical depictions of probabilities have been shown to improve comprehension among study participants, both for the probability of rain in a weather forecast and for the prob- ability that a course of medical treatment will be effective (Schwartz 2009; BBC News 2007; Price et al. 2007). Paling (2003) recommended using numbers to supplement descrip- tive terms, expressing probabilities as frequencies (e.g., 19 out of 20) instead of percentages, and using a consistent denomi- nator or scale (e.g., expressing two hypothetical probabilities as “2 out of 10” and “9 out of 10,” rather than expressing those same probabilities as “1 out of 5” and “9 out of 10”) to com- municate probabilities to medical patients. displayed on the changeable portions of the signs are based on recent sensor data rather than longer-term historical data (FHWA 2004). Although symbols and pictograms on roadway signs have not been tested for use with travel time or reliability con- cepts, the results of research on their use on other types of road signs and CMSs indicate mixed results for driver com- prehension (Luoma and Rama 2001; Arbaiza and Lucas 2010; Knoblauch et al. 1995). Another category of graphical sign is a graphical route information panel (GRIP). GRIPs display part of a road network using color coding to display informa- tion such as the level of traffic congestion on various road- way segments, similar to the color-coded traffic maps that appear on many TMC traveler information websites. Studies of GRIPs have identified several advantages over word mes- sages, including the ability to convey more information about multiple roadways and to communicate with foreign trav- elers; a disadvantage is that drivers who have difficulty under- standing maps may also have difficulty understanding GRIPs (Schouten et al. 1998; Alkim et al. 2000; Techie-Menson 2001). GRIPs that contain travel time information along with the graphical traffic status information are in use in other parts of the world (not yet in the United States); their effectiveness has not yet been determined (Lerner et al. 2004; Task Group 09 2009). Reliability Information via Websites and Mobile Devices The types of travel time information offered on travel infor- mation websites vary, with real-time information much more commonly available than historical and reliability informa- tion. Several websites operated by state or local TMCs repro- duce the travel time messages displayed on DMSs; users can select a DMS location from a map of area freeways to check real-time travel times and conditions for a particular section of roadway. Color-coded freeway maps displaying either traffic speeds or congestion levels are also common. Other features available on some TMC maps show incidents and weather- related hazards such as snow and ice or flooding. At the time of the literature review, only a handful of travel websites offered reliability information: • The Wisconsin DOT website provides a table of current and normal travel times for highways in the Milwaukee area, with travel times that are 20% or more above normal shown in bold print (Kothuri et al. 2007).