Enhancing Pedestrian Volume Estimation and Developing HCM Pedestrian Methodologies for Safe and Sustainable Communities (2022)

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.

157 Chapter 5. Conclusions and Suggested Research Conclusions Pedestrian Satisfaction with Roadway Crossings Uncontrolled Crossings This research developed a model to estimate pedestrian satisfaction and LOS at uncontrolled crossings based on the AADT of the street being crossed, how often pedestrians experience delay when starting the crossing (accounting for motorist yielding behavior), and the presence of specific pedestrian safety countermeasures (median islands, rectangular rapid-flashing beacons [RRFBs], and crosswalk markings). Although pedestrian satisfaction at uncontrolled crossings decreased with increasing speed limit, traffic speed does not appear in the final model. Instead, AADT appears to account for such factors as speed, number of lanes crossed, and traffic intensity, as streets with higher AADTs are frequently wider and have higher speed limits. No statistical relationship was found between trip purpose, trip length, frequency of using the crossing, or crossing the street in conjunction with travel to or from a transit stop. The modeling found that all of the specific safety countermeasures included in the study produced an increase in pedestrian satisfaction, over and above other effects such as reducing delay or crossing distance. RRFBs produced the greatest increase in pedestrian satisfaction, followed by median islands, and marked crosswalks. Although other countermeasures were not tested, their effects can at least be partially accounted for in the model by their documentable ability to improve motorist yielding (thus reducing pedestrian delay), shorten crossing distances, or both. This model can be used in conjunction with a pedestrian safety analysis that estimates the crash-reducing effect of a particular safety countermeasure. In cases where high-quality crash modification factors have yet to be developed by research for a particular countermeasure, this model can help support the case for implementing safety countermeasures by estimating the improvement in pedestrian satisfaction that the countermeasure would create. The model is recommended for inclusion in an update of Chapter 20 of the HCM 6th Edition. Draft HCM chapter text, an example problem illustrating its use, and a computational engine have been developed and will be provided to the TRB Standing Committee on HCQS, the body responsible for maintaining the HCM, at the conclusion of this project. Signalized Crossings The research compared pedestrian satisfaction at signalized crosswalks with LPIs to satisfaction at similar signalized crosswalks without LPIs. No significant difference in satisfaction was found. Pedestrians were generally satisfied with their crossing experience at both types of signalized crossings. The only field- measurable factors found to be significant in predicting satisfaction were the average volume of conflicting left-turning traffic during the pedestrian phase and the specific city (Portland or Chapel Hill) where the survey was taken. Neither traffic speeds nor number of lanes crossed were found to be significant.

158 It is cautioned that because the purpose of the study was to identify whether LPIs increased pedestrian satisfaction with their crossing experience, the control sites were selected to be comparable to the sites where LPIs had been installed—typically urban locations with relatively high pedestrian volumes. Developing a more generalized model for signalized crossings will require more data collection, in particular at sites with long cycle lengths (e.g., downtown Baltimore, suburban locations), wider and higher- speed streets (suburban locations), and channelized right-turn lanes (suburban and rural locations). Safety Countermeasure Effects on Motorist and Pedestrian Behavior The video observations were used to identify motorist yielding rates at each uncontrolled crossing site. Motorists at the study sites in Chapel Hill and Portland tended to yield at rates higher than the national average, as determined through a synthesis of the literature on motorist yielding. Average motorist yielding rates at treated sites (RRFBs, median islands, marked crosswalk only) were higher than those at corresponding unmarked crosswalks. The yielding rates found in this project have been combined with yielding rates in the literature to develop a proposed update of the HCM’s table of national default yielding rates, covering a broad range of safety countermeasures. At LPI locations, pedestrian signal compliance was higher (typically by a few percentage points) compared to signalized control locations without LPIs. In all cases except one, signal compliance exceeded 90%. The pedestrian delay data collected from the video observations at uncontrolled crossings were used to validate the proposed revisions to the HCM pedestrian delay method for uncontrolled crossings. A similar validation effort for signalized intersections was not included in the project work scope, but the data are available to perform a similar validation in the future. Naturalistic Walking Study No correlation was found between participants’ stress levels and individual crossing locations, including those crossings that were part of the intercept survey and video observation studies. Instead, stress was associated with conditions along a particular facility. Higher levels of stress were generally associated with walking in proximity to collector and arterial streets and in areas with industrial and mixed (e.g., offices, retail, residential) land uses. Stress levels were relatively low in lower-density residential land uses, as well as in forest, park, and university campus environments. Participants’ mean and maximum heart rates were elevated in land contexts with mixed and industrial uses, as well as along collector roads. Participants’ heart rates were lower when walking along paths and in environments with lower motor vehicle traffic (<4,000 AADT). While none of these findings are particularly surprising, they do provide a quantitative confirmation of findings from previous qualitative studies involving pedestrian satisfaction surveys. Data collection for the naturalistic walking study was hampered by problems with the portable GPS units that reduced the amount of data available, including shorter-than-expected battery life and the GPS vendor failing to remotely set the data collection interval to the desired 5 seconds. The research team had considered recording participants’ location using their phone’s GPS functionality, but decided not to because of concerns about battery drain on the phone and because the manufacturer-provided app used to connect to the biosensing wristband did not record location data. The latter issue meant that the participants would have to work with two separate apps to log their trips, with the attendant potential for user error and data loss. If such a study were to be repeated in the future, it is recommended that the researchers develop a custom app (usable on both Android and iOS phones) to provide both the wristband data-syncing and location-tracking functionality, and supply power banks to participants to address potential power-drain issues.

159 Revised HCM Delay Methodologies Uncontrolled Crossings Based on the findings documented in Chapter 4, the following conclusions are offered:  At sites that do not have a left-turn lane, the revised model can reliably predict the average pedestrian delay. The data collected confirm this conclusion for delays less than 15 s/p. The theoretic basis of the delay model formulation provides reasonable confidence that this predictive reliability extends to delay values that exceed 15 s/p.  At sites that have a left-turn lane, the revised model can provide an unbiased prediction of the average pedestrian delay. However, the uncertainty associated with the predicted value will be large because the true left-turn delay will depend on the left-turn volume, left-turn capacity, and whether the left-turn lane is used by pedestrians as a refuge so they can complete the crossing in two stages. The predicted delay for these sites should be used carefully and it should be confirmed by field observation whenever possible.  The reliability of the predicted delay is highly dependent on the values used for the crosswalk walking speed Sp and the “pedestrian start-up time and end clearance time” ts. Local values should be established for these two variables whenever possible. This research has shown that values of 4.7 ft/s and 0.0 s for Sp and ts, respectively, provided the most reliable delay estimates for the study sites. The HCM default values 3.5 ft/s and 3.0 s for Sp and ts are likely to provide conservatively high delay estimates (which may be appropriate for some design applications). It is recommended that additional research be undertaken to extend the revised HCM model so that it can be used to obtain reliable estimates of the predicted average pedestrian delay at sites where the crosswalk crosses a left-turn lane, a right-turn lane, or both. For the “left-turn lane present” case, this research should consider the effect of the following factors on pedestrian delay: left-turn volume, left-turn capacity, and whether the left-turn lane is used by pedestrians as a refuge so they can complete the crossing in two stages. Signalized Crossings The research team prepared proposed revisions to the HCM pedestrian delay method for signalized crossings that allow the estimation of delay for one-leg, two-stage (i.e., potentially waiting in the roadway median) and two-leg, two-stage (i.e., diagonal) crossings, based on peer-reviewed research by Wang and Tian (2010) and Zhao and Liu (2017), respectively. Data collected by the project could be used at a future date to validate the existing HCM method for one-leg, one-stage signalized crossings. The team also prepared an example problem illustrating the methods and developed a computational engine. The research team also prepared proposed extensions to the above methods to address delay associated with crosswalk closures, exclusive pedestrian phases (e.g., Barnes dance) and the computation of pedestrian delay associated with any stage of a pedestrian crossing. The team also prepared extensions to the methods addressing two forms of pedestrian-friendly pedestrian-actuated timing, based on previous work for the City of Portland. Although all of these extensions have a good theoretical basis, none has yet been validated through simulation or field data collection. Roadway Crossing Difficulty The research team prepared a proposed revision to the HCM pedestrian LOS method for urban streets to address the issue that diversion delay to the nearest signalized crossing will nearly always exceed the delay associated with waiting for a suitable gap to cross the street midblock. The revision draws from pedestrian perception research by Chu and Baltes (2001) that found that crossing difficulty is sensitive to segment length. The revised HCM method incorporates sensitivity to segment length and produces logical results

160 when the segment and intersection LOS scores are equal, while retaining the original method’s intent of lowering segment LOS when the pedestrian environment is otherwise good, but the street is hard to cross, and improving segment LOS when the pedestrian environment is poor, but the street is easy to cross. Pedestrian Network QOS Based on the proof-of-concept testing described in Chapter 4, the research team recommends a “connectivity island” approach using the Oregon DOT’s PLTS measure for both segments and intersections. GIS software would be used to first determine the PLTS of each segment and intersection within the study area based on information commonly found in roadway databases. Next, the GIS software would be used to identify the extent of each sub-network within the study area connected by segments and crossings of a specified PLTS or better (PLTS 2 to describe networks usable by many users and PLTS 3 for basic connectivity). These subnetworks form “connectivity islands” that can be visualized as shown in Figure 5-1, where each shade of green and purple indicates a separate connected sub-network providing the specified PLTS or better, and grey indicates portions of the network that do not provide the specified PLTS. Source: Twaddell et al. (2018). Figure 5-1. Extract from a Connectivity Island Map for Fort Collins, Colorado. Once the subnetworks have been identified, the GIS software can then be used to quantify the following components of network connectivity defined in the FHWA Guidebook on Measuring Multimodal Network Connectivity (Twaddell et al. 2018):  Network quality—miles or percent of the study area network providing the specified PLTS or better.  Network density—average miles per sub-network (connectivity island), or number of subnetworks.

161  Route directness—shortest path (if one exists) along the PLTS network between a given origin and destination, compared to the straight-line (air) distance.  Access to destinations—percent of specified destinations reachable along the PLTS network from a given origin. The final component of network quality, network completeness, can be assessed separately as the percentage of the planned network that has been completed. Suggested Research This section lists suggested future research related to the project objectives. It is divided into two parts: (a) follow-up research directly derived from the original research performed by this project and (b) research problem statements for gap-filling research activities that were considered for this project’s data collection phase but were ultimately not selected. Follow-up Research The following research activities would build upon this project’s pedestrian crossing satisfaction work:  Intercept surveys evaluating pedestrian satisfaction with additional types of pedestrian safety countermeasures at uncontrolled crossings. This project’s research found that the specific countermeasure(s) implemented at a location influence pedestrian satisfaction beyond their effect on motorist yielding rates and crossing distance. Conducting intercept surveys for additional countermeasure types would establish whether and by how much they affect pedestrian satisfaction. With the current model, the countermeasures studied by NCHRP 17-87 may appear to produce much better pedestrian satisfaction than other countermeasures, simply because they were studied and the others were not. See also the gap-filling research section below.  Intercept surveys evaluating a broader range of signalized intersections. NCHRP 17-87 focused on LPIs and control intersections with similar characteristics. Because of this focus, many other types of signalized crossings (e.g., crossing many lanes, crossing higher-speed streets, having long signal cycles, crossing channelized right-turn lanes) were not studied. Conducting intercept surveys at a broader range of signalized intersections would allow a generalized satisfaction model to be developed for signalized crossings, similar to the one this project created for uncontrolled crossings. The following research activities would build upon this project’s pedestrian delay work:  Validating the extensions to the new signalized delay method. This project would use simulation, calibrated with delay data already collected by NCHRP 17-87, to validate the proposed extensions to the signalized delay method, covering crosswalk closures, exclusive pedestrian phases, individual crossing stage delay, and pedestrian-friendly actuated signal timing. Simulation is suggested because each extension can be modeled at a given intersection, and a variety of intersections can be modeled to test different signal timing patterns and pedestrian volumes. In contrast, field data collection would require extensive travel to obtain sufficient sites and, in the case of crosswalk closures, would be unlikely to have many pedestrians performing a 3-leg crossing. Validating the extensions would increase their likelihood of being approved for inclusion in the HCM, as well as increase practitioners’ confidence in the results.  Additional delay data collection at uncontrolled crossings with turn lanes. The validation work for the revised HCM model developed by this project found greater variability in the estimate of delay, compared to measured delay, at sites where the crosswalk crossed a left-turn lane, a right- turn lane, or both. For the “left-turn lane present” case, this research should consider the effect of the following factors on pedestrian delay: left-turn volume, left-turn capacity, and whether the left-

162 turn lane is used by pedestrians as a refuge so they can complete the crossing in two stages. This work would lead to more reliable estimates of pedestrian delay and LOS at uncontrolled crossings where turn lanes are present. Gap-Filling Research The scope of NCHRP 17-87 covered pedestrian operations, pedestrian QOS, and the effects of pedestrian safety countermeasures on operations and QOS. The project developed 18 research problem statements for gap-filling research activities that could be performed during the data collection phase of the project. Not all of these activities could be performed within the available project resources and remain as research needs. The activities are summarized below, with research problem statements provided in Appendix D.  Additional pedestrian safety countermeasure intercept surveys. This activity, described above with follow-up research, would conduct intercept surveys at crossings with additional types of countermeasures and update this project’s uncontrolled pedestrian crossing LOS model based on the results. Two countermeasures in particular were rated in project interviews as being highly effective by practitioners who had implemented the countermeasure: curb extensions combined with parking restrictions, and pedestrian hybrid beacons. Other possible countermeasures to study include raised crosswalks, in-roadway YIELD TO/STOP FOR PEDESTRIANS signage, and crossing illumination.  Sidewalk and Intersection QOS. This group of research activities focuses on pedestrian satisfaction using different types of facilities, primarily sidewalks and crosswalks. – Urban sidewalk QOS. The current HCM urban street pedestrian LOS method has been criticized for being too data-intensive to be usable for large-scale planning, or even smaller- scale studies, for sometimes producing counterintuitive results, and for not incorporating environmental and aesthetic factors. A variety of alternative pedestrian QOS measures have been proposed in the literature, but have only received limited testing (e.g., in one city only). This study would collect new data to address these issues. – Rural pedestrian facility QOS. The HCM’s pedestrian methods were developed using data from urban and suburban areas. No guidance is provided on evaluating pedestrian QOS on rural pedestrian facilities. This study would investigate pedestrian QOS on rural pedestrian facilities, specifically: sidewalks, shared roads, advisory shoulders, and sidepaths. – LOS methods for intersection forms not addressed by the HCM. This study would develop perception-based LOS methods for up to six intersection forms not currently addressed in the HCM: intersections with channelized right-turn lanes, two-way stops (crossing the controlled street), all-way stops, roundabouts, free-flow ramp terminals, and alternative intersections. – Driveway crossings. Although the research that developed the current HCM urban street method found no significant effect of driveways on pedestrian facility ratings (developed from video labs), practitioners have found this result hard to believe. In addition, the HCM urban street LOS method assumes that pedestrians are unaffected when they must cross stop-controlled side streets, under the theory that side-street traffic must stop and give the right-of-way to crossing pedestrians. However, the influence of turning traffic from the major street is not considered. Consequently, further research is required on this topic.  Pedestrian Operations. This group of research activities examines the operation of pedestrian facilities including sidewalks, crosswalks, and traffic signals, in terms of such measures as pedestrian delay, pedestrian speed, pedestrian spillover out of the designated facility, and noncompliance with traffic signal indications.

163 – Basic crosswalk operations. Research by the New York City DOT has found that a number of factors affect pedestrian speeds in signalized crosswalks, but the current HCM procedure does not account for these effects. Vehicles interacting with pedestrians using the crosswalk may create operational effects (e.g., delayed entry into the crosswalk, longer crossing times, higher pedestrian densities as pedestrians avoid encroaching vehicles). The length of time that pedestrians enter the crosswalk after the start of DON’T WALK affects both pedestrian and motorized vehicle operations. These research needs were combined into one activity because the research team felt that they could be addressed from the same video data collected at specified locations. – High-volume pedestrian facilities. Studies by the New York City DOT found that the current HCM pedestrian measure was too insensitive to changes in pedestrian volume or effective width. The New York City DOT also found that shy distances reported in the literature were not based on actual observations and hypothesized that they may vary under different conditions. The HCM identifies that pedestrians will start spilling out of a pedestrian facility under conditions better than LOS F, but does not identify what those conditions are. These research needs were combined into one activity because the research team felt that they could be addressed from the same video data collected at specified locations. – Effects of pedestrian delay. Pedestrian noncompliance with traffic signals is a safety issue. Demonstrating that countermeasures to reduce pedestrian delay (e.g., by shortening traffic signal cycle lengths) improve compliance could be used to justify those countermeasures. The objective of this research activity is to collect data on the effects of pedestrian delay and low cross-street traffic volume on pedestrian noncompliance with traffic signals. – Pedestrian crossing spacing. The objective of this research activity is to collect data to answer the following research questions: (a) the relationship between the distance between marked crossing opportunities and crossing outside the marked crosswalks, and (b) the relationship between traffic volumes and crossing outside marked crosswalks. These statements were combined into one group because the research team felt that both questions could be answered using the same video data at specified locations. – Effects of building entrances and high-activity transit stops on pedestrian facilities. Building entrances and high-activity transit stops generate both sidewalk cross-flows and groups of people that block a portion of the sidewalk circulation area. However, these effects on sidewalk operations have not been well-studied.