Guidelines for the Application of Crossing Solutions at Roundabouts and Channelized Turn Lanes for Pedestrians with Vision Disabilities (2016)

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.

NCHRP 3-78b: Final Project Report April 2016 5 2 LITERATURE REVIEW It has been well documented in past research that pedestrians who are blind can face significant accessibility challenges when crossing at modern roundabouts or intersections with channelized right turn lanes. As a result of a systematic and iterative research program, researchers (many of whom are on this team) have documented that significant access challenges exist for individuals with blindness when crossing at roundabouts (e.g. Guth et al., 2005) and channelized turn lanes (e.g. Schroeder et al., 2006). Roundabouts with multiple lanes have been shown to be particularly problematic. For example, Ashmead et al. found that blind pedestrians had greater difficulty than sighted pedestrians when tasked with distinguishing gaps in approaching traffic at a two-lane roundabout that were long enough to cross from those that were not (Ashmead et al., 2005). 2.1 NCHRP Report 674 Framework for Accessible Crossings In NCHRP Report 674: Crossing Solutions at Roundabouts and Channelized Turn Lanes for Pedestrians with Vision Disabilities (Schroder et al., 2011a), the research team conceptually divided the crossing task for a blind pedestrian into four distinct components: 1. Finding the crosswalk, 2. Aligning to cross, 3. Deciding when it is safe to cross, and 4. Maintaining alignment during crossing. The third task is the one that is likely to be most critical to safety. Therefore the focus of the human factors research presented in NCHRP Report 674 was on the third task. In this project, the team supplemented the primary focus on the third component from earlier work, with field studies on tasks 1, 2 and 4. The motivation for this was to obtain a more complete picture of the wayfinding challenges that roundabouts and CTLs may pose to persons who are blind. Crosswalks at roundabouts and CTLs are not where pedestrians who are visually impaired are accustomed to finding crosswalks at “conventional” signal or stop-controlled intersections, and, as a result, pedestrians with blindness or low vision may cross well outside the crosswalk, or may cross to the central island if they are unable to locate a crosswalk. Because there is no vehicular traffic parallel to crosswalks at roundabouts, and because this cue normally provides the most reliable information about the direction of the crosswalk, pedestrians who are visually impaired may not establish a heading to cross that is in line with the crosswalk. If they initiate a crossing without aligning accurately, or if they fail to maintain a correct alignment while crossing, they may be outside the crosswalk as they complete their crossing. If there is landscaping that impedes their ability to step out of the street when they arrive at the end of the crossing (e.g., the splitter island if crossing from curb to island, or the curb if crossing from splitter island to curb), they may find it necessary to remain in the vehicular way for some time while they search for an opening that permits them to step out of the street. The framework to evaluate and quantify crossing performance used a four-criterion assessment that conceptually separates the various components of the crossing task. 1. Crossing Opportunity Criterion: Are there sufficient crossing opportunities in the form of yields or crossable gaps? 2. Crossing Opportunity Utilization Criterion: Are the crossing opportunities detected and/or utilized by the pedestrian? 3. Delay Criterion: Is a crossing opportunity taken within a reasonable time? 4. Safety Criterion: Does the crossing interaction occur without a significant degree of risk with an acceptable level of risk?

NCHRP 3-78b: Final Project Report April 2016 6 Arguably, the fourth component, the safety criterion, is the most important aspect to ensure a safe and accessible crossing environment, while the delay criterion is a term typically used in traffic engineering assessments of (sighted) pedestrian Levels of Service. However, all four measures are important in evaluating the accessibility of a crossing and understanding the contributing factors to a decision that may have resulted in high risk or delay. The research protocol in the present research collected data on all four components, and further divided the first and second criteria into sub-components related to crossing opportunities when vehicles yielded upstream of the crosswalk, and crossing opportunities created by gaps when no vehicles were approaching from upstream. From this framework emerge key accessibility performance measures, which include the frequency of orientation and mobility (O&M) interventions (our key measure of risk), and the average and 85th percentile pedestrian delay. Prior research developed predictive models for pedestrian delay as a function of driver yielding, gap availability, and the utilization rates for gaps and yields (Schroder et al., 2011a; Schroeder and Rouphail, 2010). In follow-up work, team members further developed models that predicted driver yielding behavior and pedestrian gap acceptance at midblock crossings using logistic regression models (Schroeder 2008; Schroeder and Rouphail, 2011b; Schroeder and Rouphail, 2011c), and worked on expanding the yielding prediction models so that they could be applied to multi-lane roundabouts (Salamati et al., 2013). A major focus of this research was the calibration and validation of these models with data collected at new sites, the customization of models specific to access for The key underlying premise of this framework is that the effectiveness of treatments can be represented as a change to one or more of the explanatory variables in these predictive models. For example, a Rectangular Rapid-Flashing Beacon (RRFB) is expected to increase the rate of driver yielding, while a raised pedestrian crosswalk (RPC) is expected to primarily reduce vehicular speed, which has been linked in research to an increased propensity of drivers to yield (Geruschat and Hassan, 2005; Schroeder and Rouphail, 2011b) and reductions in frequency of interventions. Using the logic of the framework, the effectiveness of any given treatment can be evaluated relative to other factors (geometry, speed, volumes, etc.), while being calibrated using data from the research of this project. In the following sections, previous research on accessibility of roundabouts and CTLs will be summarized in more detail, including research on crossings by pedestrians who are blind, research on wayfinding, and research on crossing treatments intended to enhance accessibility. 2.2 Overview of Previous Research with Blind Participants Regarding Actual Street Crossings and Judgments About When to Cross Research conducted during the last 15 years has documented the challenges experienced by pedestrians who are blind when crossing at roundabouts and CTLs. Comparisons of totally blind and sighted pedestrians crossing at roundabouts and CTLs show that blind pedestrians miss more crossing opportunities, wait longer to cross, and are more likely to make risky crossings. For example, Guth et al. (2005) investigated blind and sighted participants’ road-crossing judgments at three roundabouts in Maryland. Blind participants’ judgments were more risky than sighted participants at two higher-volume multi-lane roundabouts, but not at the low-volume single-lane roundabout. At a single-lane roundabout in Tampa (Guth et al, 2013), blind participants’ judgments about when to cross were more risky when traffic volume was high compared to judgments made during lower volume times of day. Blind participants also were slower than sighted participants in making crossing judgments, and they accepted fewer crossing opportunities. Both groups made somewhat safer and more efficient judgments at locations farther from the roundabout than at the “typical” crosswalk location one vehicle length back from the circle (Guth et al 2013). Similar comparisons of blind and sighted pedestrians at channelized turn lanes demonstrated similar difference in performance by blind and sighted pedestrians and documented difficulties by blind travelers identifying crossing opportunities and in making safe crossing decisions (Schroeder et al, 2006).

NCHRP 3-78b: Final Project Report April 2016 7 Figure 2-1: Blind pedestrian crossing at multi-lane roundabout This figure shows two pedestrians crossing at a multilane roundabout, one with a white cane extended. Vehicles are yielding in each lane. Ashmead et al. (2005) reported the result of a study of six blind and six sighted pedestrians who crossed the street at a double-lane urban roundabout in Nashville under high and low traffic volumes. As in the earlier study in Maryland, blind participants waited longer to begin crossing than sighted participants. About 6% of the blind participants’ crossing attempts were judged dangerous enough to require intervention, compared to none for sighted pedestrians. Drivers yielded frequently to pedestrians waiting to cross the entry lanes, but significantly less so on exit lanes. The study showed that sighted participants generally accepted driver yields, whereas blind participants rarely did so, presumably in part because they could not hear the yielding vehicles, or because they weren’t comfortable crossing in front of a stopped vehicle. In recent years, accessibility research involving multi-lane roundabouts has largely moved from descriptive work that documents crossing behavior at various conditions and at various locations to a focus on treatment studies. For example, there have been several evaluations of pavement treatments such as sound strips and raised crosswalks (Hughes et al., 2011; Inman et al., 2005, 2006; Schroeder, 2011) and signalization treatments such as pedestrian hybrid beacons and rectangular rapid- flash beacons (Final Report: Oakland County, 2011; Schroeder et al, 2015). 2.2.1 Crossing in Gaps in Traffic Crossing in gaps requires relatively large gaps in traffic and a location with minimal ambient noise, neither of which is commonly available at a roundabout or CTL. The fact that long gaps often are infrequent contributes to the relatively long pedestrian delay times that have been reported in several studies (Guth et al., 2013; Schroeder, 2011). Wall et al. (2008) investigated the auditory detection of approaching vehicles by pedestrians who are blind or who have low vision. The study considered the level of ambient sound; the sound level and speed of approaching vehicles; and physical features of the environment, such as hills, bends in the road, trees, and obstacles. One of their findings is as follows:

NCHRP 3-78b: Final Project Report April 2016 8 “In trials on the straight roadway, as the ambient sound level increased, the detection times decreased. Once the ambient sound level is above approximately 50 dBA, it becomes virtually impossible to hear vehicles far enough away to know whether it is clear enough to be able to complete a crossing before the vehicles arrive.” At CTLs and roundabouts, the ambient sound levels will often be above 50dBA so the safety of crossing in detected gaps in traffic at most roundabouts and CTLs is questionable. In addition, the increase in quieter vehicles has resulted in the need for blind pedestrians to modify the gap detection crossing strategy and to be more conservative in auditory-based gap crossing decisions at uncontrolled crosswalks. While driver stopping sight distance is often used as a safety measure, its use includes an assumption that drivers can see the pedestrian, are paying attention, and will be able to stop in time for pedestrians in their paths. Pedestrians who are blind cannot easily monitor driver actions and vehicle speed changes (or lack thereof) and revise their speed or position to avoid potential conflicts. Many blind people are therefore unwilling, correctly, to step out and just assume that drivers will stop. A recent article discussed using critical gap time for pedestrian crossings as a more accurate measure (Koslow et al., ITE Journal, 2013) and is similar to Sauerburger’s timing method for detection of adequate gaps in traffic (Barlow, et al. 2010, foundations of O&M chapter on teaching street crossing). Because the sound of vehicles after they have passed the crosswalk can mask the sound of approaching vehicles, individuals who are blind may not recognize gaps for several seconds after they begin. It frequently takes three seconds or more before the noise of vehicles that have passed is diminished enough to allow detection of a following gap (Guth et al., 2013). In comparison, a sighted traveler oftentimes seizes a gap opportunity immediately after the vehicle has passed through the crosswalk (Schroeder et al., 2006). Figure 2-2: Blind pedestrian attempting to cross at three-lane roundabout exit This photo shows a pedestrian with long cane at three lane roundabout attempting to cross, with car crossing the crosswalk, not yielding Although crossing one direction of traffic at a time at roundabouts can be a benefit for sighted pedestrians, the sound of traffic on the other side of the splitter island is not usually masked by the island, so traffic moving in other lanes or in the circulatory roadway can also mask the sound of approaching

NCHRP 3-78b: Final Project Report April 2016 9 vehicles in the lane to be crossed. The same is true at CTLs, where the movement of vehicles in other lanes can mask the sound of closer traffic in the channelized lane, negating somewhat the safety benefits of a separated single-lane crossing for individuals who are blind. 2.2.2 Crossing When Drivers Have Yielded Crossing when vehicles have yielded requires first detecting the vehicles, then confirming that the vehicles have stopped, and that the vehicles are staying stopped (Long et al., 2005). At roundabouts and CTLs, both the masking sounds of other traffic in the intersection and the tendency of some drivers to yield 30 feet or more upstream of the crosswalk result in yields that are not detected and used by the pedestrians who are blind (Ashmead et al., 2005; Schroeder et al., 2011). Also, drivers generally wait only a few seconds (Inman et al., 2006) before deciding that the individual is not going to cross. Oftentimes, previously yielded drivers begin to move again, just as the pedestrian has detected that the driver is present and possibly yielding. This lack of understanding resulted in some risky decisions by pedestrians in research (Schroeder et al. 2015). The sound-masking challenge noted above for detecting gaps also can affect the use of yielded vehicles. The sound of one vehicle that has yielded just upstream of the pedestrian can mask sounds of other vehicles approaching in adjacent lanes of a multi-lane crossing. This situation is similar to the multiple threat pedestrian crash type that is discussed frequently in pedestrian safety literature (e.g. PEDSAFE), where a yielding vehicle in the near lane blocks the line of sight between the pedestrian and an approaching vehicle in the far lane. 2.3 Overview of Previous Research on Wayfinding by Blind Persons at Roundabouts and CTLs While anecdotal evidence suggests that finding crosswalks, aligning to cross, and maintaining alignment while crossing at roundabouts and CTLs can cause considerable difficulty, there has been very little research specifically addressing these tasks. To our knowledge, the only previous research that has specifically focused on wayfinding at roundabouts and CTLs is pilot research conducted in Raleigh, North Carolina (Bentzen, B.L. Barlow, J.M., Guth, D., Long, R., Scott, A., Cunningham, C., & Schroeder, B., 2012). The Raleigh pilot research documented the difficulty of locating the crosswalk for pedestrians who are blind. Participants passed the crosswalk without detecting it on 17.9% of trials, and participants aligned in a heading that would have resulted in their completing crossings within the crosswalk if they maintained that heading on only 52.1% of trials. Temporary installation of a prototype 24-inch-wide guidance surface of bar tiles extending across the width of the sidewalk, and in which the bars were aligned perpendicular to the direction of travel on the crosswalk, resulted in reduction in failure to locate the crosswalk to 2.4% of trials, and an increase in headings that would have resulted in completion of crossings within the crosswalk to 77.3% (both statistically significant). The guidance surface used in the pilot research in Raleigh was selected because, internationally, it is common to use a tactile walking surface comprised of raised bars to both indicate the crosswalk location and guide pedestrians who are blind to crosswalks. The raised bar surface is called a “guiding pattern” in technical standards developed by the International Organization for Standards (ISO 23599:2012, Assistive products for blind and vision-impaired persons – Tactile walking surface indicators). The guiding pattern surface is not typically used in the US at this time. In other countries, the guiding pattern is usually used in association with truncated domes (also known in the US as detectable warning surfaces), which are used to warn of hazards such as drop-offs at transit platform edges, level transitions between pedestrian and vehicular ways, or changes in direction. The guiding pattern is required to have a minimum effective width of 550 mm (21.65 in) when it is

NCHRP 3-78b: Final Project Report April 2016 10 intended to be detected by a person approaching at an angle (ISO 23599). International research has determined specifications for the truncated dome and guiding patterns to ensure that they are highly detectable when used on various walking surfaces, and that they are highly discriminable from one another (Bentzen, Barlow and Tabor, 2000). That research was the basis for the ISO specifications. While there is considerable international research on the detectability of these types of surfaces, there has been little research on their use in travel or the ability of blind pedestrians to follow or align with guiding patterns effectively. Internationally, guidance surfaces comprised of raised bars typically have the bars aligned parallel to the direction of travel (where bar tiles approximately 24-inch-wide alert travelers who are blind to the location of a crosswalk, and then guide them to the crosswalk, or to the top of the curb ramp leading to the crosswalk). In the Raleigh pilot research, the surface was used with the bars aligned perpendicular to the direction of travel so that the bars would also provide an optimal physical cue for alignment to cross the street. Research by Takeda et al. (2006) and by Scott et al. (2011a) has found that raised bars aligned perpendicular to the direction of travel provided significantly greater accuracy in alignment than raised bars aligned parallel to the direction of travel. Accurate initial alignment, however, does not guarantee that pedestrians who are blind will maintain that alignment as they cross streets (often referred to as “maintaining a heading”). Guth and LaDuke (1995) found absolute errors (veering) from a well-aligned starting point over a distance of 25 meters, in the absence of physical or acoustic guidance cues, ranged from 1.84 to 5.75 meters, and Kallie, Schrater and Legge (2007) found a comparable range of error. This translates into a rate of error of 7-23% per meter of crossing. For a 12-foot-long, one-lane crossing, this error range translates into an error up to 2.8 feet on either side of the crosswalk. In research to identify cues for alignment that would also result in maintaining an accurate heading across a crosswalk, Scott et al. (2011b) had participants align on a plywood “curb ramp,” and then cross a virtual street in a parking lot using five tactile or acoustic cues, including raised bars oriented perpendicular to the direction of travel on the crosswalk. Participants’ locations relative to a virtual six-foot-wide crosswalk were made at distances of 12, 36 and 72 feet from the starting curb ramp. At 12 feet, average distance from the centerline of the crosswalk showed that participants were well within the width of the crosswalk regardless of the cue. However, by 36 feet, participants were within the crosswalk only when the cue was a prototype guidestrip or guidestrips across the crosswalk or a prototype beaconing accessible pedestrian signal. The guidestrip(s) could be followed or “trailed” with the pedestrian’s long cane, and the beaconing signal provided an acoustic aiming point. The research also suggested that participants were generally not very good at aligning with the slope of the ramp itself or with detectable warnings alone without the presence of additional treatments. On the basis of this research, it appears likely that if blind pedestrians are accurately aligned at the center of the crosswalk when they begin to cross at a roundabout or CTL, they will be able to maintain their heading sufficiently to still be within a six-foot-wide crosswalk if they are crossing a single lane or potentially two lanes. However, individuals often align to cross on one edge of the sidewalk or curb ramp rather than at the center. Crosswalks that are more than two lanes wide may require an additional or different tactile or acoustic cue to enable pedestrians who are blind to reliably remain within the crosswalk. A tactile guidestrip or an audible beacon, as reported by Scott et al. (2013), was more recently found to enable individuals to remain within the crosswalk at complex signalized intersections in situations where vehicular sound is minimal or where vehicles are not moving on a parallel trajectory relative to the with the crosswalk direction (Barlow et al., 2013). Based on this research, it appears that tactile guidestrips might be considered when the goal is to promote a straight line of travel within the crosswalk at roundabouts and CTLs. The challenge of traveling in a relatively straight line in the crosswalk also has implications for the design and size of channelization and splitter islands, and is discussed in later sections.

NCHRP 3-78b: Final Project Report April 2016 11 2.4 Overview of Research on Treatments to Reduce Crossing Risk for Blind Pedestrians In order to inform the site selection and treatment evaluation, a summary of existing and prior research on treatment effectiveness on roundabout and channelized turn lane accessibility is provided below. 2.4.1 A Synthesis of Crossing Treatment Effectiveness From the completed NCHRP 674 report, two treatments emerged that showed particular promise: the Pedestrian Hybrid Beacon (PHB, also known as a HAWK signal or HAWK beacon) and a raised pedestrian crosswalk. One of the earliest implementations of the PHB was at midblock and intersection locations in Tucson, Arizona. It was documented to have high rates in yielding (see NCHRP Report 562, Fitzpatrick et al., 2006). A temporary implementation of the PHB at a two-lane roundabout was tested in Golden, Colorado, as documented in NCHRP Report 674 (Schroeder et al., 2011a), and a permanent installation now exists at roundabout in Oakland County, Michigan with both two-lane and three-lane approaches. In NCHRP Project 3-78A, the team also tested a raised pedestrian crossing (RCP) at the same two-lane roundabout where the PHB was installed in Golden, CO. Exhibit 1: Images of Sample Treatments a) Lacey, WA Overhead Flashing Beacon b) Tucson, AZ RPC at Channelized Turn lane c) Oakland County, MI PHB three-lane RBT d) RRFB in Olympia, WA

NCHRP 3-78b: Final Project Report April 2016 12 This exhibit shows pictures of four existing treatments including: (a) an overhead flashing beacon at a single-lane roundabout approach in Lacey, WA, (b) a raised pedestrian crosswalk at a channelized turn lane in Tucson, AZ, (c) a Pedestrian Hybrid Beacon at a three-lane roundabout approach in Oakland County, MO, and (d) a rectangular rapid-flashing beacon at a two-lane roundabout in Olympia, WA. NCHRP 3-78A further studied a sound strip treatment geared at sending distinguishable audible patterns of vehicles approaching in a channelized turn lane, which would then allow a blind pedestrian to isolate the CTL vehicles from adjacent through traffic. The sound strip treatment was tested in isolation and along an overhead pedestrian-activated flashing beacon geared at promoting driver yielding. Project 3-78A further tested a variety of sites in a no treatment condition, which includes all “before” conditions of the tested treatments, as well as data for three single-lane roundabouts without treatments. Since the completion of NCHRP 3-78A data collection, another treatment has emerged that is a viable consideration. The Rectangular Rapid-Flashing Beacon (RRFB) has demonstrated great potential as an effective treatment to generate high yielding rates by drivers for pedestrians. The RRFB was first implemented at midblock locations in St. Petersburg, Florida, and exhibited high, sustained vehicular yield rates (Shurbutt and Van Houten, 2010). In a response to the effectiveness of the RRFB at midblock locations, FHWA recently commissioned a 30-month study to specifically look at the viability of the RRFB as an accessibility treatment at multi-lane roundabouts (FHWA, 2011). Early results in that project suggest that an accessible crossing environment may be established by the RRFB treatment only if combined with high driver yielding and roundabout geometry that promotes low vehicle speeds. In a prior evaluation of RRFBs at a multi-lane roundabout in Oakland County, Michigan, significant accessibility challenges remained due to low driver compliance and high vehicular speeds, especially when tested at multi-lane exits. The following tables summarize the effectiveness of the various treatments in terms of orientation and mobility (O&M) interventions and the average delay experienced by blind travelers. Table 2-1: Summary of Results Pedestrian Hybrid Beacon. (Schroeder et al., 2011a and RCOC 2011) City Crossing Geometry Entry/Exit Study O&M Int. Avg. Ped. Delay (sec.) Golden, CO Two-Lane RBT Crossing Entry/Exit No Treatment* 2.4% 16.0 Golden, CO Two-Lane RBT Crossing Entry/Exit With Treatment 0.0% 5.8 Oakland County, MI Two-Lane RBT Crossing Entry Leg No Treatment* 1.9% 15.4 Oakland County, MI Two-Lane RBT Crossing Entry Leg With Treatment 0.0% 11.5 Oakland County, MI Two-Lane RBT Crossing Exit Leg No Treatment* 8.7% 19.0 Oakland County, MI Two-Lane RBT Crossing Exit Leg With Treatment 1.7% 11.2 Oakland County, MI Three-Lane RBT Crossing Entry Leg No Treatment* 7.7% 20.1 Oakland County, MI Three-Lane RBT Crossing Entry Leg With Treatment 0.0% 14.2 Oakland County, MI Three-Lane RBT Crossing Exit Leg No Treatment* 9.6% 22.3 Oakland County, MI Three-Lane RBT Crossing Exit Leg With Treatment 0.8% 11.7 Table Caption: The table shows a summary of research results of various studies performed on the Pedestrian Hybrid Beacon (PHB) at multi-lane roundabouts. Rows marked with (*) correspond to the “pretest” condition, which is equivalent to a no treatment case, with only the crosswalk markings and detectable warning surfaces provided.

NCHRP 3-78b: Final Project Report April 2016 13 Table 2-2: Summary of Results RRFB. (RCOC 2011, FHWA TOPR34 Study) City Crossing Geometry Entry /Exit Study O&M Int.+ Avg. Ped. Delay (sec.) Oakland County, MI Two-Lane RBT Entry No Treatment* 7.5% 20.8 Oakland County, MI Two-Lane RBT Entry With Treatment 0.0% 17.1 Oakland County, MI Two-Lane RBT Exit No Treatment* 23.8% 22.2 Oakland County, MI Two-Lane RBT Exit With Treatment 16.4% 18.8 Oakland County, MI Three-Lane RBT Entry No Treatment* 12.5% 35.2 Oakland County, MI Three-Lane RBT Entry With Treatment 7.6% 19.8 Oakland County, MI Three-Lane RBT Exit No Treatment* 23.2% 30.5 Oakland County, MI Three-Lane RBT Exit With Treatment 18.9% 24.8 Olympia, WA (4th) Two-Lane RBT Entry With Treatment 2.2% 4.3 Olympia, WA (4th) Two-Lane RBT Exit With Treatment 3.0% 2.8 Olympia, WA (Olympic Way) Two-Lane RBT Entry With Treatment 6.7% 4.5 Olympia, WA (Olympic Way) Single-Lane RBT Exit With Treatment 0.0% 2.9 Olympia, WA (14th Street) Two-Lane RBT Entry With Treatment 7.1% 2.3 Olympia, WA (14th Street) Two-Lane RBT Exit With Treatment 2.4% 2.9 Springfield, OR (Hayden Br.) Two-Lane RBT Entry With Treatment 2.2% 8.9 Springfield, OR (Hayden Br.) Two-Lane RBT Exit With Treatment 12.2% 9.3 Springfield, OR (Pioneer Pkwy) Two-Lane RBT Entry With Treatment 4.2% 5.7 Springfield, OR (Pioneer Pkwy) Two-Lane RBT Exit With Treatment 11.4% 10.4 Oshkosh, WI (Jackson St) Two-Lane RBT Entry With Treatment 2.1% 12.4 Oshkosh, WI (Jackson St) Two-Lane RBT Exit With Treatment 16.0% 17.3 Oshkosh, WI (Murdock Ave) Two-Lane RBT Entry With Treatment 0.0% 13.1 Oshkosh, WI (Murdock Ave) Two-Lane RBT Exit With Treatment 15.0% 17.0 Carmel, IN (Clay Terrace Blvd) Two-Lane RBT Entry With Treatment 3.8% 16.4 Carmel, IN (Clay Terrace Blvd) Two-Lane RBT Exit With Treatment 4.0% 13.3 Albany, NY (Fuller Road North) Two-Lane RBT Entry With Treatment 13.6% 9.8 Albany, NY (Fuller Road North) Two-Lane RBT Exit With Treatment 21.7% 28.2 Albany, NY (Fuller Road South) Two-Lane RBT Entry With Treatment 1.7% 8.5 Albany, NY (Fuller Road South) Two-Lane RBT Exit With Treatment 12.9% 10.2 Davidson, NC (Griffin St - East) Two-Lane RBT Entry With Treatment 4.3% 9.1 Davidson, NC (Griffin St - East) Two-Lane RBT Exit With Treatment 0.0% 10.1 Davidson, NC (Griffin St - West) Two-Lane RBT Entry With Treatment 0.0% 14.2 Davidson, NC (Griffin St - West) Two-Lane RBT Exit Wit Treatment 8.3% 10.7 Table Caption: The table shows a summary of research results of various studies performed on the Rectangular Rapid-Flashing Beacon (RRFB) at multi-lane roundabouts. Rows marked with (*) correspond to the “pretest” condition, which is equivalent to a no treatment case, with only the crosswalk markings and detectable warning surfaces provided.

14 NCHRP 3-78b: Final Project Report April 2016 Table 2-3: Summary of Results for Other Treatments. (Schroeder et al., 2011a and RCOC 2011) Crossing Geometry City Treatment Entry/Exit O&M Int. Avg. Ped. Delay (sec.) Two-Lane RBT Crossing Golden, CO No Treatment* Entry/Exit 2.4% 16.0 Two-Lane RBT Crossing Golden, CO RPC Entry/Exit 0.0% 5.8 CTL Crossing Charlotte, NC No Treatment* n/a 9.4% 26.2 CTL Crossing Charlotte, NC Sound Strips n/a 2.9% 18.5 CTL Crossing Charlotte, NC No Treatment* n/a 5.6% 23.4 CTL Crossing Charlotte, NC SS & Beacon n/a 1.4% 12.2 Single-Lane RBT Crossing Raleigh, NC No Treatment Entry Leg 2.1% 10.5 Single-Lane RBT Crossing Raleigh, NC No Treatment Exit 5.8% 11.6 Single-Lane RBT Crossing Charlotte, NC No Treatment Entry Leg 0.8% 26.6 Single-Lane RBT Crossing Charlotte, NC No Treatment Exit Leg 0.8% 24.0 Single-Lane RBT Crossing Golden, CO No Treatment Entry Leg 2.8% 10.9 Single-Lane RBT Crossing Golden, CO No Treatment Exit Leg 0.0% 13.0 Table Caption: The table shows a summary of research results of various studies performed on other treatments at multi-lane roundabouts, single-lane roundabouts, and intersections with channelized turn lanes. Rows marked with (*) correspond to the “pretest” condition, which is equivalent to a no treatment case, with only the crosswalk markings and detectable warning surfaces provided. The results in the tables show that the PHB was effective in reducing both interventions and delay in all studied conditions. The PHB reduced the rate of interventions to zero at the Golden, CO roundabout, as well as the two-lane and three-lane entry legs at the Oakland County, MI roundabout. For the two-lane and three-lane exit legs in Oakland County, some interventions remained even in the PHB post-test condition, although at a statistically significant reduction over the pretest, where intervention rates were very high. The PHB installations also had a consistent impact on the average pedestrian delay, which was reduced in all tested installations. In addition to the PHB evaluation, the exhibits show promise for other treatments, including the raised pedestrian crosswalk (zero post-test interventions in Golden, CO) and the rectangular rapid-flashing beacon (zero post-test interventions at two-lane entry in Oakland County, MI). However, with a limited number of studies, these results cannot be generalized at this time, and the data further point to some remaining accessibility concerns even with treatments like the RRFB (e.g. 16.4% interventions at a two-lane roundabout exit in Oakland County, MI). A large number of two-lane roundabout RRFBs is also under evaluation through a study performed for FHWA by members of this research team. For single-lane roundabouts, all studied sites show relatively low intervention rates, although some accessibility challenges remain as evident, for example, by a 5.8% exit-leg intervention rate at the Raleigh, NC site. Other research by this team on gap judgments at a single-lane roundabout in Tampa, FL by pedestrians who are blind further supports the notion that under high-volume and high-speed conditions, single-lane roundabouts can exhibit accessibility difficulties (Guth et al., 2011). Much less is known at the present time about ways to improve accessibility at single-lane channelized turn lanes, although research has demonstrated significant challenges. Tested treatments in NCHRP 3-78A proved only partially successful (intervention rates of 2.9% and 1.4% in post-test). While signalization may be a feasible treatment solution given existing signal controllers and power at most CTL locations, additional research on raised crosswalks and speed-reducing geometric alignment is highly desirable. 2.4.2 Opportunities Through Geometry and Geometric Treatments In the discussion of accessibility treatments, emphasis is typically placed on technology solutions, often in the form of signalization or beaconing treatments. Sometimes overlooked is an effort to enhance

15 NCHRP 3-78b: Final Project Report April 2016 accessibility through geometric treatments and modifications to the site in question. In particular, geometric treatments that encourage (a) low vehicular speed, (b) good visibility and sight distances of the crosswalk, and (c) a separation of conflict points to reduce cognitive load for drivers and pedestrians warrant further scrutiny and consideration. Research has long linked lower vehicle speeds to improved safety by increasing the opportunity to react by drivers and pedestrians and reducing the severity of collisions if they occur. Lower speeds are further demonstrated to be associated with a higher likelihood that drivers will yield, and good visibility (and no occlusion of auditory cues) intuitively facilitates driver and pedestrian reactions to the presence of the other mode. A separation of conflict points may further enhance yielding and increase driver awareness of the crosswalk and the pedestrian. The potential of geometric design to enhance accessibility appears particularly relevant at single-lane roundabouts and channelized turn lanes, but even two-lane roundabouts may be designed with an emphasis on pedestrian access. Through studies conducted in this research and supported by earlier accessibility work, this team seeks to identify what aspects of geometry contribute to enhanced accessibility and document these findings. Examples include the R1 through R5 radii of a roundabout as described in NCHRP Report 672, the relative location of the crosswalk to the circulating lane, the shape of the splitter island at a channelized turn lane, and the curve radii in that channelized turn lane. The potential benefits of a low-speed, pedestrian friendly environment may further be enhanced by geometric treatments like the raised pedestrian crosswalk (RPC). While any low-speed and geometric treatment needs to consider impacts on other modes at the intersection (particularly on heavy vehicles), an RPC performed mostly as well as a PHB in testing in NCHRP 3-78A at a fraction of the installation and maintenance cost. A review of the literature on RPCs showed significant guidance on the design of RPCs through ITE’s Traffic Calming: State of the Practice (ITE, 2012a) and an ITE Proposed Recommended Practice on Guidelines for the Design and Application of Speed Humps (ITE, 2012b), as well as various state and local agencies (City of San Diego, 2012; Delaware DOT, 2012). Research has further highlighted impacts on RPC design details like the vertical elevation, transition slope, and transition shape (flat, sinusoidal, or parabolic). Mohammadipour et al. even developed a speed prediction equation for RPCs as a function of these design variables (Mohammadipour et al., 2009). The authors further suggested narrowing the roadway in advance of the RPC, a practice that is reflected also in some local guidance (e.g. Placer County, PA – Pennsylvania Department of Transportation, 2012). In combination with models that predict driver yielding as a function of speed, these RPC speed prediction equations may greatly facilitate the development of guidance for RPC installation and design. Similarly, guidance in the FHWA Roundabout Guide (Rodegerdts, 2010) on predicting vehicular speed as a function of different radii may be used to make inferences about the implications on accessibility – if this research successfully identifies a clear a linkage between vehicular speed and accessibility performance measures. 2.4.3 Understanding and Mitigation of Induced Vehicle Delay Some concern has been raised in the engineering community about additional delay incurred by vehicles at multi-lane roundabouts if a pedestrian signal (or PHB) is installed. While some pedestrian- induced delay to vehicles always occurs when pedestrians are crossing at roundabouts, research has shown that the impacts can be reduced significantly if innovative signalization schemes are used. Through a microsimulation analysis, NCHRP Report 674 (Schroeder et al., 2011a) explored these impacts in a computer modeling environment using the VISSIM software. The authors tested impacts using three signalization strategies: 1. Use of a PHB over a conventional red-yellow-green signal, 2. Implementation of a two-stage signal scheme, where a pedestrian call only stops one direction of traffic at a time, and 3. Relocation of the crosswalk to a zig-zag (offset) and distal configuration.

16 NCHRP 3-78b: Final Project Report April 2016 The authors further tested varying pedestrian and vehicle volumes, as well as single-lane and multi- lane roundabouts. For this discussion, only some of the two-lane roundabout results are presented. The table below shows the pedestrian-induced increase in average vehicle delay at the roundabout for various strategies. All strategies are also compared in terms of their percent reduction in delay relative to the assumed base case of a single-stage, green-yellow-red signal at the standard crosswalk location (20-feet from circulating lane). Exhibit 2 summarizes the key findings from the effort. Exhibit 2: Results of Simulation Testing of Roundabout Signalization Strategies (Schroeder et al., 2011a) Crosswalk Location Signal Staging Signal Strategy Below Capacity Scenarios - Delay per Vehicle (s) Below Capacity Scenarios - % Change over Base At-Capacity Scenarios - Delay per Vehicle (s) At-Capacity Scenarios - % Change over Base Proximal Single-Stage Signal 14.2 Base 68.4 Base Proximal Single-Stage PHB 6.3 -56% 39.4 -42% Proximal Two-Stage Signal 4.1 -71% 24.4 -64% Proximal Two-Stage PHB 1.5 -89% 5.5 -92% Zig-Zag Two-Stage Signal 3.9 -73% 23.4 -66% Zig-Zag Two-Stage PHB 1.3 -91% 7.0 -90% Distal Two-Stage Signal 2.8 -80% 5.9 -91% Distal Two-Stage PHB 1.2 -92% 0.0 -100% Exhibit 2 Caption: This table shows the results of a simulation-based sensitivity analysis of different pedestrian signal strategies, including a standard pedestrian signal and a pedestrian hybrid beacon (PHB). Both strategies were tested in a single-stage application (stopping traffic long enough for pedestrians to cross entry and exit leg) and a two-stage application (stopping only entry or exit at a time). The analysis further varied between three different crosswalk locations: Proximal (20-foot distance from circulating lane), Zig-Zag (exit-leg portion of crosswalk moved to 60 feet from circle), and distal (entire crossing moved to 100 feet from circle). The results are reported as average vehicle delay (in seconds) and as the percent change of a scenario to the base case. The results indicate that the impacts on vehicular delay can be mitigated significantly by using the PHB signalization strategy (resulting in shorter vehicle solid red times), using a two-stage signal scheme (reducing the pedestrian crossing time and thus vehicular red), and by moving the exit portion of the crosswalk further away from the circulating lane (reducing vehicle queue spill-back potential into the circulating lane). Similar to the analysis presented above, it is critical in NCHRP 3-78B to weigh the impacts of the various accessibility treatments on vehicular traffic, as well as other modes of transportation (bikes, trucks, transit, other pedestrians) as applicable. While considering different solutions, the accessibility criterion needs to be treated as a fixed constraint to comply with ADA. But where alternatives exist, these vehicular delay considerations and impacts on other modes will be important. For example, the speed-reducing effect of a raised crosswalk is expected to add vehicular delay through an increase in vehicular headways (following gaps between vehicles) and thus a reduction in capacity. Raised crosswalks may also impact bicyclist performance and pose further difficulties for maneuverability of heavy vehicles. 2.4.4 Driver Noncompliance at Roundabout PHBs Past research has noted driver noncompliance at roundabout PHB installations, especially at the exit legs. At both studied PHB Roundabouts (Golden, CO and Oakland County, MI), a significant percentage of drivers were observed to proceed through the solid red indication without stopping. At the Golden, CO

17 NCHRP 3-78b: Final Project Report April 2016 installation, 12.6% of drivers observed during the solid red did not stop at the PHB (RCOC, 2011). Due to the relatively short duration of the solid red, the actual number of vehicles corresponded to 15 out of 119 vehicles. This behavior is a concern, as the solid red indication coincides with the pedestrian WALK indication. At the Oakland County, MI roundabout, similar behavior was observed. At the 2-lane approach, 4.6% of drivers violated the steady red at the entry leg and 12.9% at the exit leg. At the three-lane approach, the rate of noncompliance of those drivers observed during steady red was 5.6% at entry and 31.1% at the exit leg (RCOC, 2011). No risky events (defined by O&M interventions) were observed in Golden, CO, as most study participants appeared to be able to hear the approaching, non-complying motorists. In the Oakland County, MI study, a few intervention events were observed, although these were not consistently linked to driver violation of the red signal. Regardless, the observed driver behavior at PHBs warrants further consideration. It is unclear at the present time what factors contribute to the observed behavior. However, the generally higher observed noncompliance at the exit leg points to potential sight distance concerns for drivers exiting the roundabout, as well as the possibility that drivers are likely to be accelerating at the time they encounter the signal or PHB. At the entry, drivers tend to be decelerating and are generally on a straight approach with improved signal visibility. This hypothesis is consistent with driver yielding rates, which are generally higher at the entry leg than at the exit (e.g. Schroeder and Rouphail, 2011b; Geruschat and Hassan, 2005). It should be noted here that at the time of the two experiments, both PHBs had been installed relatively recently, on the order of four to six weeks prior to the experiment. While this allowed for some driver adaptation time, it is unclear how many drivers were familiar with the intended operations of the PHB system because of relatively low occurrence of pedestrian activity at both roundabouts. Since the PHB rests in “dark” for vehicular traffic, it is possible that most drivers had not previously encountered the PHB in an active state. Follow-up research investigating the long-term performance of PHBs at roundabouts is therefore strongly recommended. To further test driver compliance at multi-lane roundabout exit signals, members of this team performed a driver simulator study (Salamati et al., 2012). The authors found similar noncompliant behavior at the exit, but also that compliance improved as the signal was moved further way from the circulating lane. These findings support the sight distance and reaction time hypothesis, but more empirical work is needed to confirm these results.

18 NCHRP 3-78b: Final Project Report April 2016 --- This page intentionally left blank ---