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 3 METHODOLOGY This chapter summarizes the study methodology for the project. The discussion includes the field study protocols of the various components of the Accessibility Audit, and a discussion of site selection and prioritization of research needs that guided the experimental plan. 3.1 Data Collection Protocol For data collection under the auspices of NCHRP 3-78B, the team applied a newly developed study protocol that was enhanced based on the lessons learned in NCHRP 3-78A and in earlier research. The Accessibility Audit limits the use of human subjects to the assessment of gap and yield utilization, delay, and risk while employing other, more focused studies to assess parameters such as the propensity of drivers to yield. Project 3-78A gathered a great deal of data on crossings by human subjects, but it should be noted that there was high inter-subject variability due to differences in travelers’ skill levels and tolerance for risk, along with variability among sites and treatments. The reactions of blind participants to the treatments, their ability to detect yields adequately, and driver reactions can be confounded by these differences. The Accessibility Audit carefully isolates the different accessibility indicators in a resource-efficient manner. The proposed studies in the Accessibility Audit protocol represent those that maximize the expected results for analysis and model development while balancing resource needs. The five studies are as follows: 1. Indicator Study with blind Participants: Blind pedestrians standing at roundabout and CTL crosswalks made independent crossing decisions and indicated to researchers by raising their hand when they would cross, without actually crossing the street. These crossing judgments were later analyzed in regard to the speed and distance of approaching vehicles as judgments were made, and this in turn permitted the research team to quantify the degree of time-to- contact/risk inherent in the crossing judgments. The study also included the use of pre- and post-test participant questionnaires to obtain standardized feedback about the difficulty individuals experienced at various locations in making judgments, and to assess their crossing confidence. A record of vehicle volume during the indicator trials also was obtained. 2. Wayfinding and Alignment Study with Blind Participants: The team evaluated blind pedestrians’ wayfinding behaviors as they prepared for crossings and as they actually crossed the road. In this component of the Audit, which followed the indicator study described above, the team observed participants as they attempted to independently locate crosswalks at unfamiliar locations, align for crossings, and maintain their alignment during a crossing (after being instructed by an O&M specialist that it was safe to step into the roadway. For safety purposes, they were followed closely during the study by the O&M specialist). This study allowed the research team to document the wayfinding challenges of the entire crossing (i.e., both the entry and exit at roundabouts, and at all lanes at CTLs). 3. Yielding study: The research team gathered data on driver yielding behavior in response to pre-defined set of pedestrian behaviors. Members of the research team, sometimes wearing dark glasses and carrying a long cane, acted as pedestrians/participants in this study. During some trials, the research team member stood near the crossing point of the target intersections (i.e., near the curb ramp) and moved the long cane or took a small step toward the street as vehicles approached. Other variables regarding pedestrian-driver interaction also were studied. Controlling pedestrian behavior allowed the research team to compare results across sites, which would have been difficult to do if blind individuals were the participants in this study. 4. Free-flow speed study: The speeds of free-flowing vehicles entering and exiting the roundabouts and traversing the CTLs were calculated. The goal of the study was to gain insight on the vehicle speed patterns in the vicinity of the crosswalk, and the effectiveness of various design features to reduce speeds. 19

NCHRP 3-78b: Final Project Report April 2016 5. Site geometry audit: The team inventoried the geometric attributes of the various study locations and compiled a photo log diary of site conditions. The geometry audit included the inscribed diameter of the roundabouts, other design parameters (e.g., radii R1-R5 per the definitions in NCHRP Report 672) of the various sites, and the crosswalk location, presence of detectable warnings, landscaping features, provisions for bicycles, and other features where applicable. Similar in concept to a Pedestrian Road Safety Audit, the protocol was applied consistently to all sites in this study. A more detailed description of the five studies is given below. 3.1.1 Indicator Study Protocol This portion of the audit involved the participation of five to eight blind participants at each site who were recruited for the study by the research team. Participants were familiarized with the roundabout, including the treatments designed to them in making street-crossing judgments (if any). Participants then were guided toward the crosswalk by an orientation and mobility specialist (O&M specialist). Once positioned at the curb ramp and facing the street, they indicated when they would cross by raising their hand, but without actually stepping into the street. These indicator trials were repeated 10 to 12 times at both the entry and the exit leg of each of two crosswalks at each roundabout (or at each tested CTL). The “indicator trial” approach was previously used in multi-lane roundabout studies in Oakland County, MI (RCOC 2011) and in Nashville, TN. (Guth et al., 2005). It is also consistent with the protocol being used in recently-completed FHWA research on the effectiveness of Rectangular Rapid-Flashing Beacons (RRFB) at multi-lane roundabouts. The “indicator only” protocol is faster than a crossing study, and most importantly, less risky for participants, drivers and researchers alike. While data collection with participants with vision impairments is intensive and time-consuming, it is essential to making decisions about appropriate treatments to support safe and efficient pedestrian access. To streamline data collection, the team focused on the evaluation of pedestrian behavior after treatments were installed at the various crosswalks, rather than conducting traditional “before-and-after” treatment evaluations. The team was able to use data gathered in prior research as baseline information for the studies reported here, which allowed for streamlined data collection. For example, the team combined the information gathered in indicator trials with information on driver yielding behavior gathered in the naturalistic yielding studies and other studies. During each trial, the O&M specialist evaluated the safety of the crossing decisions by showing a hand signal to a computer operator to indicate whether the crossing decision likely would have resulted in an intervention if actual crossings had been initiated by the participant. It is these “risky crossings” that were quantified and evaluated for each site. Frequency of intervention was the primary safety measure used in previous crossing studies in Nashville and Tampa, FL and other studies, where the O&M specialist physically “intervened” by stopping the pedestrian. The indicator protocol used here assigns one of three categories to each crossing event: “no intervention,” “possible intervention,” and “expected intervention.” In addition to the O&M specialist, a second expert observer rated each crossing event as safe, possible intervention, or expected intervention, independent of the O&M specialist. These two independent risk indicators were later compared to arrive at an overall risk assessment of the crossing decision. All trials were videotaped so they can be further evaluated against more objective risk measures, such as the vehicle time-to-collision between the pedestrian decision and the arrival of the next vehicle at the crosswalk. The team is sensitive that any research involving human subjects needs to protect the safety, privacy, and well-being of the study participants. As was done in previous research, the team obtained Institutional Review Board (IRB) approval (from North Carolina State University and the National Academy of Sciences) for all components of the Accessibility Audit, but with special emphasis on this crossing indicator study. After review by both IRBs, the team was authorized to proceed with this research protocol. 20

NCHRP 3-78b: Final Project Report April 2016 3.1.2 Wayfinding and Alignment Study Protocol To determine if a crossing is usable by individuals who are blind or visually impaired, it is important to evaluate all the accessibility challenges, not just the determining-when-to-cross problem. Once the sites were selected for the indicator studies, the team selected a subset of these sites for the wayfinding studies. Site selections were based on maximizing the design diversity while considering the experimental and practical constraints imposed by the logistical aspects and budget constraints of the project. The wayfinding protocol required about 45 minutes of additional time per participant per site beyond that required for the crossing indicator study. The procedure was similar to the method used in previous wayfinding research in Raleigh (Scott et. al, 2014). After the crossing indicator trials, a short break for water and snacks, and completion of the debrief for the indicator trials, participants were asked to independently approach several crosswalks two or three times at the roundabout or CTL from either upstream or downstream direction (depends on configuration of crosswalks at the roundabout or CTL and availability of landscaping or other cues or no cues at various crosswalks at the location). Instructions to participants were to find the crosswalk and align to cross and to inform the O&M specialist when he or she had done so. Performance measures for this study include: • Time from a marked location a consistent distance from the pushbutton or crosswalk to being ready to cross; • Location within or outside crosswalk, toward or away from circulatory roadway or downstream roadway; • Alignment (heading) aimed within, outside crosswalk lines, toward or away from circulatory roadway or downstream roadway; and • If pushbutton, time to find pushbutton. After this initial alignment activity, additional measures of “maintaining heading while crossing” were gathered by having the orientation and mobility specialist tell the participant when to cross. As the crossing took place, the experimenters recorded the participant’s crossing heading accuracy. Participants also crossed to the island and aligned to cross the other roadway lanes. This gave the team information about wayfinding in relation to the configuration and features of splitter islands and triangular islands at roundabouts and CTLs. 3.1.3 Yielding Study Protocol Since an increase in the propensity of drivers to yield is likely to provide more crossing opportunities for pedestrians who are blind, this study is a critical component of the Accessibility Audit. If drivers in all lanes routinely stop for pedestrians, as is seen in some European countries, more crossing opportunities exist. While yielding studies can be purely observational, varying pedestrian volumes across sites and different pedestrian populations are likely to introduce bias when comparing the same treatment across various roundabouts. The study applied a protocol where a member of the research team approaches the crosswalk at random intervals (e.g. once per minute), and takes one step into the crosswalk, which is in accordance with most states’ yielding laws. A video camera and synchronized radar speed measurement system capture the position and speed of approaching vehicles, while a trained observer records the yield outcome. That observer further describes attributes of the interaction between pedestrians and driver including crossing leg (entry/exit), lane position (near/far), platooning (yes/no), vehicle type (truck/car), and others. The yield trials are performed in randomized order starting at either the entry or the exit leg. Half the trials further employ the use of a white long cane to simulate the arrival of a blind pedestrian and to test for an increased propensity of drivers to yield. In all trials, the pedestrian will take one step “into” the crosswalk in accordance with most states’ yielding laws. 21

NCHRP 3-78b: Final Project Report April 2016 The protocol is consistent with the evaluation protocol used in HRT-10-043 (FHWA 2010) and with many state laws on driver yielding (“in” vs. “at” the crosswalk). The “in” crosswalk yield results will be a direct comparison to the indicator studies, which have to be performed “at” the crosswalk to pass IRB muster. The team used the same threshold for determining whether or not drivers “would have been able to stop” using the ITE Signal Formula (ITE 2009) that was used in prior research. 3.1.4 Free-Flow Speed Study Protocol The research team conducted a vehicle speed study at the crosswalk when no pedestrians were present, consistent with methods described in the ITE Manual of Transportation Studies (ITE, 2010). This provided insight on the relationship between geometric parameters and actual crosswalk speeds (rather than those estimated from equations in for example the FHWA Roundabout Guide). Vehicular speeds are critically linked to injury severity in the case of a pedestrian-vehicle collision, but also have a strong impact on pedestrian and driver decision-making. Higher speeds are generally associated with shorter allowable reaction times for gap crossings and have been linked to a lower likelihood of driver yielding. It was hypothesized that geometric parameters of the roundabout that impact speed therefore have a key impact on accessibility. 3.1.5 Geometry Audit Protocol A team of trained observers documented critical geometric features of the roundabout including inscribed diameter, roundabout design radii R1-5, the crosswalk locations relative to the circulating lane, the presence and orientation of truncated dome detectable warnings, landscaping, and other factors. The documentation was in the form of a detailed and narrated photo diary of the site with special emphasis placed on features that have been linked through prior research to impacting accessibility. The objective of the geometry audit was to identify good and poor design aspects that may impact accessibility. The comprehensive results of all geometry audits and photo diaries will be a useful resource for practitioners. 3.2 Site Selection The site selection process was principally motivated to fill knowledge gaps identified by the research team. The site selection processes balanced two competing objectives: • The first objective was to study a broad range of treatments and combinations of treatments in order to gain insight in the accessibility performance across a range of conditions. Note that the term "treatment" in this case may also refer to geometric variations, rather than a technology solution per se. • The second objective was to obtain a sufficient sample size for model development. While a broad range of treatment is desirable, a statistically-robust sample of any one treatment is also highly desirable. As such, it was advantageous to perform multiple studies of the same treatment (e.g. RRFB), than covering a wider range of treatments (e.g. RRFB, standard beacon, overhead beacon, in-pavement beacon, etc.). Both objectives were considered in identifying a list of potential sites for data collection in Phase II of this project. As such, the team initially identified desirable site characteristics, which were then matched against the database inventory of available sites for a list of proposed locations. Other considerations played into the final site selection process as well, including proximity to the research team (to save on travel expenses for sites within driving distance), and, more importantly, the ability to economize data collection at multiple sites within the same trip. When weighing studies of different treatments, the team considered the following items: • Prior research, with a higher priority for treatments without extensive prior research; 22

NCHRP 3-78b: Final Project Report April 2016 • Treatment cost, with a higher priority for treatments with low capital cost; • Likelihood of success, with a higher priority to treatments that are likely to have a high benefit to accessibility; and • Boundary conditions, with a higher priority to treatments and sites that are expected to provide better insights on the boundary conditions between accessible and non-accessible. When weighing studies in support of different models the team considered the following items: • Expected impact, with a higher priority to models that are expected to have a key impact on accessibility (e.g. yield prediction, which has been linked to better accessibility); • Likelihood of success, with a higher priority to models for which the team anticipates being able to obtain adequate sample sizes for modeling, under consideration also of existing data and in light of inherent variability in decision-making (e.g. gap selection is expected to show clear correlation with crossing width and approach speeds); and • Sensitivity, with a higher priority to models expected to show high sensitivity to parameters within the realm of control of the designer (e.g. speed prediction as a function of design radii). In general, sites were selected based on the type of crossing treatment installed. While wayfinding features were important considerations of this project, the team proposed to evaluate most wayfinding accommodations "as is." Even without explicitly screening sites for wayfinding treatments, the team was able to obtain a broad sample of landscaping treatments and accommodations from the selected sites. 3.2.1 National Agency Outreach and Site Inventory The focus of this task was to perform a national outreach to transportation agencies with the goal of identifying existing sites with one or more treatments installed. At the same time, the team planned to identify potential locations of new installations if the list of existing treatments shows gaps for particular treatments that are identified as key research needs. An important starting point for the identification of additional sites was the inventory database of roundabouts across the United States maintained by Kittelson and Associates, Inc. (http://roundabouts.kittelson.com/). As part of a prior research effort, the team recently amended that database to include specific information about crosswalk geometry, roundabout design speed, and other site characteristics pertinent to this project. Unfortunately, a similar database does not exist for channelized turn lanes, but these intersections are also more frequently found across the country. The team performed additional outreach efforts in the form of an online survey. The survey was administered to national list serves (e.g. the list serve of the TRB Roundabout Committee, TRB Pedestrian Committee, ITE Traffic Engineering Council, Associate for Pedestrian and Bicycle Professionals, etc.). Through this survey effort, combined with sites already familiar to the team, a list of 53 potential sites was identified for consideration in the data collection phase. These sites are comprised of 24 channelized turn lane sites, 22 multi-lane roundabouts, and 9 single-lane roundabouts. Of these sites, 22 have previously been evaluated in studies conducted under NCHRP 3-78a, the NIH grant, TOPR34, or other studies. • For the 24 CTL sites, 11 have raised crosswalks, 6 have signals, 1 has an RRFB (CTL at a RBT), 1 is outfitted with a flashing beacon, and 1 has a stop sign prior to the crosswalk • For the 22 multi-lane roundabout sites, 4 have raised crosswalks (one also a PHB), 5 have signals or a PHB, 9 have RRFBs, 1 has an offset crosswalk, and 1 has speed humps prior to the crosswalk at the entry leg • For the 9 single-lane roundabout sites, 2 have raised crosswalks, and 1 has an RRFB Tables with site details are shown below. 23

NCHRP 3-78b: Final Project Report April 2016 Table 3-1: CTL Sites ID Type St1 St.2 City State No. Lanes Treatment Studied Before Prior Project 1 CTL N Wilmot Rd E. Speedway Bvld Tucson AZ 1 Stop Sign at CTL No n/a 2 CTL E Mann Ave E Gold Links Rd Tucson AZ 1 Large Islands No n/a 3 CTL East River Rd N/ Sabino Canyon Rd Tucson AZ 2 Signal No n/a 4 CTL N/ Craycroft Rd East River Rd Tucson AZ 1 Signal No n/a 5 CTL Grant Rd Campbell Ave Tucson AZ 1 RCW No n/a 6 CTL Camino Seco Wrightstown Blvd Tucson AZ 1 RCW No n/a 7 CTL Oracle Rd Grant Rd Tucson AZ 1 RCW No n/a 8 CTL Lee Hill Dr 28th St Boulder CO 1 Skewed 3-legged No n/a 9 CTL Jay Rd Denver Boulder Turnpike (28th) Boulder CO 1 Skewed 4-legged No n/a 10 CTL 28th St Diagonal Hwy Boulder CO 1 RCW- x4 No n/a 11 CTL 28th St Valmont Rd Boulder CO 1 RCW-x1 No n/a 12 CTL 28th St Pearl St Boulder CO 1 RCW- x3 No n/a 13 CTL 28th St walnut St Boulder CO 1 RCW-x2 No n/a 14 CTL 28th St Canyon Rd Boulder CO 1 RCW-x3, 1 without No n/a 15 CTL 27th Way Baseline Rd Boulder CO 1 RCW-x1, 2without, 1 two-lane Signalized No n/a 16 CTL 27th Way Broadway Boulder CO 1 1 without RCW No n/a 17 CTL Magnolia Dr Park Ave Tallahassee FL 1 Signal No n/a 18 CTL St Francis Cir W Portal Ave San Francisco CA 1 RCW No n/a 19 CTL Grand Ave Harrison St Oakland CA 1 Signal No n/a 20 CTL SW Taylors Ferry Rd SW Terwilliger Portland OR 1 Signal No n/a 21 CTL Providence Rd NC51 Charlotte NC 1 Beacon Yes 3-78a 22 CTL Hillsborough St Gorman St Raleigh NC 1 n/a Yes NIH 23 CTL SE Cary Pkwy Walnut St Cary NC 1 n/a Yes NIH 24 CTL at RBT Fuller Rd Washington Ave Albany NY 1 RRFB Yes FHWA 24

NCHRP 3-78b: Final Project Report April 2016 Table 3-2: Multi-Lane RBT Sites ID Type St1 St.2 City State No. Lanes Treatment Studied Before Prior Project 25 Multi-RBT Golden Rd Johnson Golden CO 2 PHB 1, RCW 1 Yes n/a 26 Multi-RBT Golden Rd Ford Golden CO 2 RCW 1 No n/a 27 Multi-RBT E Orange Ave Jim Lee Rd Tallahassee FL 2 Offset CW No NIH 28 Multi-RBT N 40th St E Hanna Ave Tampa FL 2 Low-Speed No n/a 29 Multi-RBT N 40th St E Yukon St Tampa FL 2 Low-Speed No n/a 30 Mulit-RBT MLK Jr Blvd N Central Ave Kissimmee FL 2 Entry Speed Humps Yes NIH 31 Multi-RBT Causeway Blvd Mandalay Clearwater FL 2 Signal, Offset CW No n/a 32 Multi-RBT Clay Terrace Blvd Carmel IN 2 RRFB Yes FHWA 33 Multi-RBT Maple Rd Drake Rd West Bloomfield MI 2/3 PHB Yes NIH 34 Multi-RBT Maple Rd Farmington Rd West Bloomfield MI 2/3 RRFB Yes NIH 35 Multi-RBT Jetton St Griffith St Davidson NC 2 RRFB* Yes FHWA 36 Multi-RBT Harbour Place Griffith St Davidson NC 2 RRFB* Yes FHWA 37 Multi-RBT Fuller Rd Washington Ave Albany NY 2 RRFB Yes FHWA 38 Multi-RBT Pioneer Pkwy Hayden Bridge Way Springfield OR 2 RRFB Yes FHWA 39 Multi-RBT Blvd des Allimettieres Blvd St. Joseph Gatineau QC 2 Signal 1, Beacons 3 No n/a 40 Multi-RBT Blvd des Allimettieres Rue Demontingy Gatineau QC 2 Signal 1, Beacons 3 No n/a 41 Multi-RBT Aldrich St Mueller Blvd Austin TX 2 PHB and/or RCW* No n/a 42 Multi-RBT 4th Ave Olympic St Olympia WA 2 RRFB Yes FHWA 43 Multi-RBT 14th Ave Jefferson St Olympia WA 2 RRFB Yes FHWA 44 Multi-RBT Longview Circle Longview WA 2 RCW No n/a 45 Multi-RBT Jackson St Murdock Ave Oshkosh WI 2 RRFB Yes FHWA 46 Multi-RBT River Park Dr Driveway Riverdale UT 2 RCW Yes NIH * Treatments at this site have not been installed yet, but coordination is ongoing 25

NCHRP 3-78b: Final Project Report April 2016 Table 3-3: Single-Lane RBT Sites ID Type St1 St.2 City State No. Lanes Treatment Studied Before Prior Study 47 Single RBT Golden Rd Ulysess Rd Golden CO 1 n/a Yes 3-78a 48 Single RBT Gulf Dr S Bridge St Bradenton Beach FL 1 n/a No n/a 49 Single RBT W. County Club Dr. North Blvd Tampa FL 1 n/a Yes n/a 50 Single RBT Pikea Ave Liloa Dr Kihei HI 1 RRFB No n/a 51 Single RBT Cherrywood Ln. Metro Access Dr Greenbelt MD 1/2 RCW No n/a 52 Single RBT Tienken Rd Sheldon Rd Rochester Hills MI 1 RCW No n/a 53 Single RBT 9th St Davidson St Charlotte NC 1 n/a Yes 3-78a 54 Single RBT Pullen Rd Stinson Drive Raleigh NC 1 n/a Yes NIH 55 Single RBT River Park Dr 900 W Riverdale UT 1 RCW Yes NIH 26

NCHRP 3-78b: Final Project Report April 2016 3.2.2 Overview of Studied Channelized Turn Lane Sites Through the process of identifying research needs for CTLs, combined with an inventory of CTL treatments across the U.S., the team conducted two clustered studies of CTL accessibility in two locations across the US, supplemented by two additional isolated locations. The motivation for the clustered approach is two-fold: (1) clustering of data collection allows the team to efficiently study several sites within the same trip and (2) clustering allows the team to control for pedestrian and driver behavior by testing the same participants interacting with the same driver population for multiple sites. Prior research on roundabout accessibility, specifically the FHWA evaluation of RRFBs at multi-lane roundabouts, showed that regional differences of driver behavior appear to contribute greatly to differences across sites, making it harder to isolate the effects of geometric differences between sites. Similarly, one of the main conclusions in NCHRP Report 674 was that the blind travelers involved in the field studies had a high degree of inter- participant variability, once again making it difficult to isolate differences across sites. The clustered approach controls for differences in driver behavior and pedestrian skill level by having the same participants cross at multiple locations in the same city. A drawback of this approach is that studies are more time-consuming per participant (accounting for travel between sites and overall more trials per participant). Another drawback is that, by spending more time at one location, fewer overall cities can be visited, while more sites can be included. From the inventory of CTL locations, the team identified clustered sites in two cities: Boulder, CO and Tucson, AZ. Both locations offer a host of channelized turn lane locations and, more importantly, offer a range of geometric configurations and treatments at these CTLs. For example, each city has at least some CTLs with a raised crosswalk, and Tucson has signalized CTLs. Both cities are also known for embracing innovative pedestrian treatments, including pedestrian hybrid beacons and RRFBs at midblock locations, which suggests that drivers in both locations are used to seeing “special” pedestrian accommodations. In addition, the team was able to study a CTL each in Cary, NC and Greenbelt, MD as part of a planned roundabout data collection trip. In total, 12 CTLs were studied in this research as shown in the table below. A summary of all sites is given in the following sections. Table 3-4: Listing of CTL Study Locations Site Location Type Treatment Wilmont at Speedway NW Tucson, AZ CTL Decel Lane, Stop Sign Sabino Canyon at Tanque Verde NE Tucson, AZ CTL Decel & Accel Lane Grant at Oracle NE Tucson, AZ CTL Decel, Raised CW, Yield Grant at Oracle SW Tucson, AZ CTL Decel, Raised CW, Yield Foothills at Baseline NE Boulder, CO CTL Decel & Accel Lane Foothills at Baseline NE Boulder, CO CTL Decel & Accel, Sound Strips Foothills at Arapahoe SW Boulder, CO CTL Decel & Accel, Raised CW 28th at Pearl NE Boulder, CO CTL Decel Lane, Yield 28th at Pearl NW Boulder, CO CTL Yield, Raised CW 28th at Canyon SW Boulder, CO CTL Decel Lane, Yield, Raised CW Kenilworth at E. Street NW Greenbelt, MD CTL Decel & Accel Lane Kildaire Farm at Tryon SW Cary, NC CTL Decel Lane, Yield 3.2.2.1 Tucson, AZ In April 2014, the team completed data collection at four CTL locations in Tucson, AZ. All four CTLs 27

NCHRP 3-78b: Final Project Report April 2016 featured different geometric and treatment combinations, as described below with aerial views of all four locations. 1. Wilmont Road at Speedway Boulevard – Northwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, stop sign at downstream merge point 2. Sabino Canyon Road at Tanque Verde Road – Northeast quadrant of intersection, urban location, deceleration and acceleration lane, no additional treatments 3. Grant Road at Oracle Road – Northeast quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk 4. Grant Road at Oracle Road – Southwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk Figure 3-1: CTL at Wilmont Road at Speedway Boulevard, Tucson, AZ 28

NCHRP 3-78b: Final Project Report April 2016 Figure 3-2: CTL at Sabino Canyon Road at Tanque Verde Road, Tucson, AZ Figure 3-3: CTLs at Oracle Road and Grant Road, Tucson, AZ 29

NCHRP 3-78b: Final Project Report April 2016 3.2.2.2 Boulder, CO In July 2014, the team completed data collection at six CTL locations in Boulder, CO. The study focused on three CTLs in an urban environment within the Boulder City Limits, as well as three CTLs under county jurisdiction. All four CTLs featured different geometric and treatment combinations as described below and accompanied by aerial views of all six locations: 1. 28th Street at Pearl Street – Northeast quadrant of intersection, urban location, deceleration lane, no acceleration lane, no additional treatments 2. 28th Street at Pearl Street – Northwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk 3. 28th Street at Canyon Boulevard – Southwest quadrant of intersection, urban location, deceleration lane, no acceleration lane, raised crosswalk 4. Foothills Parkway at Arapahoe Avenue – Southwest quadrant of intersection, suburban location, deceleration lane, acceleration lane, raised crosswalk 5. Foothills Parkway at Baseline Drive – Southwest quadrant of intersection, suburban location, deceleration lane, acceleration lane, sound strip treatment 6. Foothills Parkway at Baseline Drive – Northeast quadrant of intersection, suburban location, deceleration lane, acceleration lane, raised crosswalk Figure 3-4: CTLs at 28th Street and Pearl, Boulder, CO 30

NCHRP 3-78b: Final Project Report April 2016 Figure 3-5: CTL at 28th Street and Canyon Boulevard, Boulder, CO Figure 3-6: CTL at Foothills Parkway and Arapahoe Avenue, Boulder, CO 31

NCHRP 3-78b: Final Project Report April 2016 Figure 3-7: CTL at Foothills Parkway and Baseline Road, Boulder, CO 3.2.2.3 Greenbelt, MD and Cary, NC In September 2014, the team completed data collection at a channelized turn lane in Greenbelt, MD. In November 2014, the team collected data at a CTL in Cary, NC. The sites are described below and accompanied by a series of aerial views. 1. Kenilworth Avenue at East West Highway – Northwest quadrant of intersection, deceleration lane, no acceleration lane, no other treatments installed 2. Kildaire Farm Road at Tryon Road – Southwest quadrant of intersection, deceleration lane, no acceleration lane, no other treatments installed Figure 3-8: CTL at Kenilworth Avenue and East West Highway, Greenbelt, MD 32

NCHRP 3-78b: Final Project Report April 2016 Figure 3-9: CTL at Kildaire Farm Road and Tryon Road, Cary, NC 3.2.3 Overview of Studied Roundabout Sites The team collected data at a total of eight roundabout approaches at five roundabouts in four different states. The emphasis in the roundabout data collection was on the evaluation of treatments not previously evaluated in sufficient detail. These treatments include: (1) raised crosswalks, (2) raised crosswalk in combination with RRFBs, and (3) rumble strips in advance of the crosswalk. The summary of sites is provided in the table below, and accompanied by aerial views and a discussion in the subsequent sections. Table 3-5: Summary of Roundabout Study Locations Site Location Type Treatment Cemetery at Main St. - East Hilliard, OH 2-lane Exit Offset Crosswalk Cemetery at Main St. - West Hilliard, OH 2-lane Standard Crosswalk Maple at Farmington - East West Bloomfield, MI 3-lane Raised CW and RRFB Maple at Farmington - N/S West Bloomfield, MI 2-lane Raised CW and RRFB Cherrywood at Metro - West Greenbelt, MD 1/2 lane Raised CW, divided lanes State at Ellsworth - West Ann Arbor, MI 2-lane Rumble Strips, Exit offset CW Huron at Nixon - South Ann Arbor, MI 1-lane Rumble Strips Old Apex at Chatham - West Cary, NC 1-lane Standard CW, 3-legged 3.2.3.1 Hilliard, Ohio The team completed a study at a multi-lane roundabout in Hilliard, Ohio in May 2014. The study focused on the east and west approaches of the roundabout at Main Street and Cemetery Road. Both approaches featured two-lane entries and two-lane exits. The west approach featured a “standard” crosswalk 33

NCHRP 3-78b: Final Project Report April 2016 location, while the east approach showed an offset or “zig-zag” configuration. Both approaches had in-road yield to pedestrian signs installed. Figure 3-10: Roundabout at Cemetery Road and Main Street, Hilliard, OH 3.2.3.2 Oakland County, Michigan In August 2014, the team completed a study at the multi-lane roundabout at Maple Road and Farmington Road in Oakland County, MI. The roundabout was previously evaluated in a separate research project in (a) a “before” condition without treatment and in (b) an “after” condition with Rectangular Rapid- Flashing Beacons (RRFBs) installed. For this study, the roundabout was also outfitted with raised crosswalks at four test legs: 3-lane entry from east, 3-lane exit to east, 2-lane entry from south, and 2-lane exit to north. 34

NCHRP 3-78b: Final Project Report April 2016 Figure 3-11: Roundabout at Maple Road at Farmington Road, Oakland County, MI 3.2.3.3 Greenbelt, MD In September 2014, the team completed data collection at a roundabout with raised crosswalks. The roundabout was located at Cherrywood Lane at Metro Access Parkway, and it featured slip lanes and raised crosswalks. The team studied the entry and exit legs at the west approach. 35

NCHRP 3-78b: Final Project Report April 2016 Figure 3-12: Roundabout at Cherrywood Lane at Metro Access Parkway, Greenbelt, MD 3.2.3.4 Ann Arbor, MI The team studied two roundabouts in Ann Arbor, MI in October 2014. The study focused on one single- lane and one two-lane roundabout, and both featured similar treatments. The sites have a milled rumble strip treatment, as opposed to a raised sound strip treatment, such as the one used at one of the CTL study locations in Boulder, CO. The single-lane site is at Huron Parkway and Nixon Road. The site has single-lane approaches on all four legs and standard crosswalks approximately 20 feet from the circulatory roadway. A set of four rumble strips is milled approximately 50 feet prior to each crosswalk to alert drivers and provide an auditory cue for pedestrians. The two-lane site is at State Street and Ellsworth Road. The site is a true two-lane roundabout with an offset-left design (i.e. higher deflection at entry and relatively straight exits). The crosswalks at this location are staggered (zig-zag), with the exit leg moved approximately 40 feet further away from the circle. State Road has a speed limit of 35mph and a daily volume of 31,500 vpd north of the intersection and 17,600 vpd south of the intersection. Ellsworth Road has a speed limit of 45 mph and a daily traffic volume of 15,600 vpd east and 13,000 vpd west of intersection. 36

NCHRP 3-78b: Final Project Report April 2016 Figure 3-13: Single-Lane Roundabout in Ann Arbor, MI with Milled Rumble Strips Figure 3-14: Two-Lane Roundabout with Milled Rumble Strips in Ann Arbor, MI 3.2.3.5 Cary, NC The team collected data at a single-lane roundabout in Cary, NC in November 2014. The site at the 37

NCHRP 3-78b: Final Project Report April 2016 intersection of Old Apex Road and West Chatham Street is a single-lane location with three approach legs. The team studied the west approach on Old Apex Road. Figure 3-15: Roundabout Site at Old Apex Road and West Chatham Street in Cary, NC 3.3 Modeling Overview The following list summarizes the critical modeling needs identified in this research. Most of these models are part of the crossing assessment methodology, with some focus on wayfinding. 1. Safety and Risk Prediction Models for All Three Facility Types 2. Yield Model for Single-Lane Roundabouts and Channelized Turn Lanes 3. Free-Flow Speed Prediction Model of Vehicle Speeds at Crosswalks 4. Wayfinding Assessment Tools 5. Crossing Alignment Assessment Tools 6. Gap Utilization Models for All Three Facility Types 7. Yield Utilization Models for All Three Facility Types The models above are those that have been identified as key assessment steps and that have not been developed in prior research. Each model was developed in Phase II of this project based on empirical data at existing sites and new data collected in this project. Table 3-3 lists all models by facility types, including a count of existing and new study locations proposed for each modeling effort. The table also lists the data source from one or more studies of the Accessibility Audit protocol described above. Modeling results are discussed in detail in Chapter 5 of this document and are incorporated into the guidebook deliverable. 38

NCHRP 3-78b: Final Project Report April 2016 Table 3-6: Modeling Needs and Site Selection # Model Type Facility Type # of Existing Sites # of New Sites Data Source 1a Safety and Risk Prediction Model CTL 3 12 Indicator Study 1b Safety and Risk Prediction Model Single RBT 3 2 Indicator Study 1c Safety and Risk Prediction Model Multi RBT 10 6 Indicator Study 2a Yield Model CTL 3 12 Yielding Study 2b Yield Model Single RBT 3 2 Yielding Study 3a Free-Flow Speed Prediction Model CTL 3 12 Free-Flow Speed Study 3b Free-Flow Speed Prediction Model Single RBT 3 2 Free-Flow Speed Study 3c Free-Flow Speed Prediction Model Multi RBT 10 6 Free-Flow Speed Study 4 Wayfinding Assessment All 16 20 Wayfinding Study and Geometry Audit 5 Crossing Alignment Tool All 16 20 Wayfinding Study and Geometry Audit 6a Gap Utilization Model CTL 3 12 Indicator Study 6b Gap Utilization Model Single RBT 3 2 Indicator Study 6c Gap Utilization Model Multi RBT 10 6 Indicator Study 7a Yield Utilization Model CTL 3 12 Indicator Study 7b Yield Utilization Model Single RBT 3 2 Indicator Study 7c Yield Utilization Model Multi RBT 10 6 Indicator Study 3.4 IRB Approval The team requested and obtained approval for all studies necessary from the Institutional Review Board (IRB) of North Carolina State University and of the National Academy of Sciences. Both IRB boards approved the study protocols as described. 39

NCHRP 3-78b: Final Project Report April 2016 This page intentionally left blank 40