Deployment, Use, and Effect of Real-Time Traveler Information Systems (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.

10 C H A P T E R 1 Background Introduction Traveler information (TI) systems are diverse and rapidly evolving. It will be important for agencies and TI vendors to align development with the needs and expectations of the traveling public. The National Cooperative Highway Research Program (NCHRP) Project 08-82, Deployment, Use, and Effect of Real-Time Traveler Information Systems has the main objectives of (1) developing a better understanding of agency practices and rationales for disseminating real-time TI, (2) assessing the effectiveness of these real-time TI systems as it relates to traveler perception and use, and (3) suggesting best practices for future facilitation of this information. Do travelers use real-time TI systems to make decisions that will result in improved operational efficiency, decreased congestion, mode shifts, and increased throughput? The majority of real-time TI system evaluations conducted to date, particularly those related to 511 systems, focus on outputs rather than outcomes (e.g., Swan, Baker, Hintz, & Trimble, 2004). For example, 511 systems are typically evaluated based on call volumes and patterns, call frequency, call durations, system menu selections, etc. In the instances where attempts are made to evaluate outcomes, they are based on levels of “satisfaction” with the quality of information provided, but not necessarily the specific decisions or actions that were made as a result of the TI provided. This research explores outcomes from a traveler perspective including access to, perception of, and need for real-time information. It also addresses outcomes from an information provider standpoint. Recurring and non-recurring congestion are significant problems in many areas where the construction of new or expanded roadways is not feasible. Therefore, it is of great interest to traffic management agencies to better manage demand to maximize system efficiency, reduce congestion, and increase driver satisfaction. Critical to the successful implementation of real-time TI systems is: (1) the collection of real-time transportation system status data and (2) the delivery of this data in the form of information that can influence traveler decision making and hence, impact system demand (e.g., Wang, Khattack, & Fan, 2009). Over time, the development and deployment of real-time TI systems has evolved from an entirely public-sector role to the point where the private sector has become an increasingly pivotal player. From a data collection perspective, real-time TI systems today make significant use of privately installed traffic monitoring devices in public rights-of-way. Most recently, private companies are becoming involved in the collection of real-time data using non-infrastructure based techniques such as vehicle probe technology. From a delivery perspective, the private sector continues to strive towards the creation of successful TI dissemination-based business models. The expanding role of the private sector in existing and future real-time TI systems is a key aspect of this research initiative. To assess the effect of real-time information on traveler behavior, we must first understand traveler behavior. Traveler behavior is the process of individual decision making about what trips to make, where to visit, when to depart, what mode of travel to utilize, and what route to follow (Schofer, Khattak, & Koppelman, 1993). In a constantly changing environment, individual needs and trip purposes change at a day-by-day/week-by-week basis. One could even argue that they change at a minute-by-minute/hour-by- hour basis. Pre-trip information is frequently available for most of the different modal alternatives. The challenge for real-time TI systems is to intervene in these behavioral processes both pre-trip and en-route,

11 providing information that is desired, used, and contributes to improved travel experience for individuals and their community. When en-route information is provided, then the travelers could implement as many of the following strategies as they deem necessary: route change, change of schedules, change of activities, and change in activity locations (Anderson & Souleyrette, 2002; Adler, 2001; Chen & Jovanis, 2003; Dia, 2002; Levinson, 2003; Mahmassani & Liu, 1999; Thakuriah & Sen, 1996). The TI system can be separated into two key parts, the user (i.e., the traveler) and the information. Attributes of travelers themselves, their households, and their situational constraints, are likely to affect the utilization of real-time traveler information systems (RT-TIS) and/or their effects. Similar to all kinds of information, RT-TIS could be evaluated by their content (e.g., parking availability at destination, congestion levels, etc.), type (e.g., qualitative vs. quantitative), format (e.g., map-based versus text messages), reliability (i.e., credible, accurate, and relevant), and prescriptive characteristic (i.e., whether it advises on alternate routes, modes, etc., or just provides information on current conditions). Several studies have evaluated the influence of each component of TI. Schofer et al. (1993) and Casas & Kwan (2007) provide a more detailed review of the literature in this area. As in any project evaluation, the go-to methodology is the cost/benefit analysis. Results from a study performed in Washington, DC showed that if RT-TIS deployments are evaluated purely on time-savings, the benefits of TI will likely be grossly underestimated (Wunderlich, Hardy, Larkin, & Shah, 2001). In addition, an evaluation of Utah’s RT-TIS showed that there was a higher level of awareness and use associated with en-route TI sources such as VMS and HAR, when compared to pre-trip TI sources such as the 511 phone system and the CommuterLink website (Martin, Lahon, Cook, & Stevanovic, 2005). DC’s case study showed that TI users improved their on-time travel reliability. The value of improved on-time reliability is not easily nor directly monetized, but it is clear that many types of travelers can benefit from RT-TIS. Commercial vehicle operations rely on just-in-time deliveries and manufacturing processes, decreasing inventory size, and therefore cost. Commuters who have jobs with firm arrival time requirements are also likely to benefit from more consistent arrival times and reduced stress. From this perspective, it can be argued that the benefit of improved travel reliability and predictability from RT-TIS may outweigh whatever returns are generated from the monetization of aggregate in-vehicle travel time reductions. Three approaches have been generally implemented to measure the effect of RT-TIS (Kristof et al., 2005; Hu, 2009): observational studies (e.g., measuring the differences in travel times between drivers with and without RT-TIS), surveys (e.g., survey studies that poll RT-TIS users to determine what qualitative and quantitative benefits they perceive) or a combination of both (e.g., simulation). Observational studies can provide more accurate accurate results compared to surveys because they present test subjects with real information and observe their responses. However, these studies are performed at a higher cost and most likely with a smaller sample size. In contrast, surveys and other exploratory studies have shown promise as effective, relatively inexpensive, and fairly accurate methodologies to evaluate hypothetical alternatives/products. Finally, simulation combines the best of observational and exploratory studies (i.e., show the potential benefits that could occur if RT-TIS were used at a certain location), but tends to demand a lot of computational effort making it a less desirable approach. Today, with the availability of GPS, Bluetooth, and other technological advancements, it is possible to examine the behavior of travelers in a more reliable and accurate manner, obtaining the level of granularity, accuracy, and timeframe needed to assess change at the individual level in real-time over a considerable period of time. The literature on methodologies to evaluate TIS does not suggest any universal methodology. Besides traditional methods (i.e., surveys and data collection), several numerical simulators have been developed: IDAS (ITS Deployment Analysis System), DynaMIT (Dynamic Network Assignment for the Management of Information to Travelers), VISSIM and HOWLATE (Heuristic Online Web-Linked Arrival Time Estimator). IDAS is a sketch planning tool created for the FHWA. It calculates the benefits and costs of deploying the specified ITS alternatives and reports outputs in terms of the incremental

12 change in performance measures (e.g., vehicle-miles traveled, vehicle-hours traveled, volume-to-capacity ratios, and vehicle speed) and the annual benefit/costs. VISSIM is able to assess TIS options such as VMS effects in a network by applying dynamic assignment features. Martin et al. (2005) stated, regarding VISSIM, that “the benefit of this approach is in the integration of off-line planning information and information derived from the physics of traffic flow, in addition to the detector data. This system, when calibrated and fully deployed, will make TIS evaluations more valuable.” DynaMIT combines real-time data from a surveillance system with historical travel time data in order to predict future traffic conditions and provide travel information and guidance through a TIS. HOWLATE uses simulated yoked pairs traveling between specified origin and destination (O-D) pairs, one of which has a TIS and one that does not, from which the method measures travel time reliability. As technology and computer performance advances, simulation exponentially grows in demand and potential for answering key questions. Its ability to assess different alternatives and provide reliable outputs based on limited data makes it, in all likelihood, the next step in project evaluation techniques. For examples on how simulation has been used, and improved, in the past two decades, the reader is referred to Jayakrishnan et al. (1993), Florian (2004), Kristof et al. (2005), and Chorus et al. (2007). Furthermore, the density of traffic data to support TIS as well as the impact of TIS is ever growing. Objective measurements of the impact of TIS were once highly improbable due to the level of data collection and coordination needed. This is rapidly changing. A case-study from the Maryland-DC metropolitan area is a good example. In 2010, enabled by the successful proliferation of outsourced traffic data, the Maryland SHA activated statewide travel times on signs on its network VMS, providing travelers with route travel times at key decision points in the network. In 2011 a full network of BluetoothTM As for the data needed for assessment, the U.S. Department of Transportation (USDOT) proposed six goals for ITS so that the benefits of the technologies can be measured (U.S. Department of Transportation, 2004). The metrics for these goals are listed in Table 1. As can be seen, these goals encompass a vast variety of metrics, some of them could be perceived as complex and costly to obtain. Furthermore, FHWA’s Office of Operations (2005) defined congestion performance measures that could be transposed to RT-TIS evaluation, incuding metrics for reliability, delays of various types, travel time, and throughput. Some overlap can be seen between these programs, which indicate an expected relationship between using RT-TIS and diminished congestion levels. Finally, 23 CFR 511, Subpart C – Real Time Systems Management Information Program (Section 511.309) provides the minimum requirements for traffic and travel conditions made available by RT-TIS by type of information (e.g., construction activities, lane blocking incidents, weather observations, etc.). This regulation also states that the establishment of a real-time information program for traffic and travel conditions on the Interstate highways system shall be completed no later than November 8, 2014 and on the State-designated metropolitan area routes of significance by November 8, 2016. sensors was deployed in multiple corridors in the DC metro area providing some overlap between the TIS and high-resolution measurement with Bluetooth traffic monitoring (BTM) deployment. In 2010, researchers at the University of Maryland began experiments at isolated locations to measure the impact of TIS delivered via the VMS on route diversion. Though in its infancy, the data assets to objectively measure the network impact of TIS (in this example delivered by a VMS system) are beginning to be deployed. For example, a VMS placed prior to the diversion of I-95 east and west on the DC beltway provides travelers information to best assess whether the inner loop (eastbound signed with I- 95) or the outer loop (westbound signed strictly as I-495) is the best route for through trips. Currently, the VMS sign at this location provides TIS based on the outsourced data feed. The BTM deployment provides an indication of the travel time and percent diversion eastbound or westbound at any time period, providing the opportunity to measure the impact of TIS. A systematic study utilizing systems of similar combined capability provides an opportunity to objectively assess TIS impact to a level of granularity not previously achievable.

13 Table 1. USDOT benefit metrics. Many studies of RT-TIS effectiveness focus on user satisfaction (e.g., “NCHRP Synthesis 399: Real- Time Traveler Information Systems” by Deeter, 2009). In a recent effort, we went beyond such measures by using a laboratory study to tie such subjective measures to what drivers actually report doing (e.g., “Driver Use of En Route Real-Time Travel Time Information” by Lerner, Singer, Robinson, Huey, & Jenness, 2009). Methods included measurements of response time and comprehension, and ratings of likelihood to divert in laboratory settings. In addition, driver logs were used to assess drivers’ uses of real- time travel time information in daily commutes and supplemented by focus groups and questionnaires in three regions of the United States. For the current effort, our approach uses a variety of methods to not only understand satisfaction, but also behavioral outcomes (through self-report logs) and other low-cost methods (including extant agency

14 data on effectiveness). This combination of methods creates a better assessment of effectiveness in real- time TI systems and practices. It also provides a better pool of information from which to develop guidance for future approaches to RT-TIS. The Westat-CATT team believed that the most critical aspect of this research, from a practitioner perspective, was to identify the potential “gaps” between the agency’s processes for disseminating TI and the public’s expectations and needs. In other words, is the information that an agency is providing to the traveler provided in a way that is actually useful to the traveler? If not, what are agencies doing wrong and what can be done to improve real-time TI gathering/delivering mechanisms? What gaps exist between what travelers say they really want and what is currently being delivered? Another critical output of this research: Agencies are in dire need of guidance on how to measure the performance/impact of their TI programs. Currently, agencies base evaluations of system performance by measuring output data such as number of webpage hits, 511 call volumes, etc. Unfortunately, these approaches do not measure outcomes in terms of assessing the effect of information on trip behavior. How can this outcome-oriented approach be woven into their existing (if they exist) TI performance monitoring initiatives? This is becoming more and more critical as agency Administrators, Directors, and even high level politicians (e.g., Governors) are becoming more intent on measuring outcome-based performance to justify investments in their programs (including TI system programs).