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.
109 Based on the findings of this study, roundabouts appear to be successful in a wide variety of environments in the United States. The following sections summarize the major conclusions from this study. In addition, further research is recommended in a number of areas. Safety Performance With the exception of conversions from all-way-stopâ controlled intersections, where crash experience remains statistically unchanged, roundabouts have improved both overall crash rates and, particularly, injury crash rates in a wide range of settings (urban, suburban, and rural) and pre- vious forms of traffic control (two-way stop and signal). Both types of safety prediction models developed for this projectâ intersection level and approach levelâemploy simple model forms that are supported by the available data. These models are of a form that is intended to be suitable for eventual inclu- sion in the forthcoming Highway Safety Manual. Overall, single-lane roundabouts have better safety per- formance than multilane roundabouts. The safety perfor- mance of multilane roundabouts appears to be especially sensitive to design details. Operational Performance Currently, drivers at roundabouts in the United States appear to be somewhat tentative, using roundabouts less effi- ciently than models suggest is the case in other countries around the world. In addition, geometry in the aggregate senseânumber of lanesâhas a clear effect on the capacity of a roundabout entry; however, the fine details of geometric designâlane width, for exampleâappear to be secondary and less significant than variations in driver behavior at a given site and between sites. This finding was tested at two levels: at the microscopic level, in terms of the effect of geom- etry on critical headways and follow-up headways, and at the macroscopic level, in terms of the overall ability of the model (analytical or empirical regression) to predict capacity. The resulting recommended operational models are there- fore relatively simple and capture only those effects that the data could support. These models are incorporated into an initial draft procedure for the Highway Capacity Manual, which the TRB Committee on Highway Capacity and Quality of Service will continue to revise until its eventual adoption. The proposed models result in lower capacity predictions than have been typically used to date in the United States. Because driver behavior appears to be the largest variable affecting roundabout performance, calibration of the models to account for local driver behavior and changes in driver experience over time is highly recommended to produce accurate capacity estimates. In addition, because the LOS thresholds for other unsignalized intersections appear to be appropriate for roundabouts, similar design standards for all unsignalized intersections (stop-controlled and roundabout) may also be appropriate (e.g., LOS D or E and/or volume-to- capacity ratios of 0.90 to 1.00, depending on the jurisdiction). Standards that allow for higher volume-to-capacity ratios and/or higher delays at roundabouts should be accompanied by local calibration where possible to improve confidence in the capacity and delay estimates. Ideally, such standards should also reflect the need to balance the accommodation of peak hour and/or peak-15-min traffic flows with the safety and accessibility of the intersection for all users. Geometric Design Although this project was unable to establish a strong sta- tistical relationship between speed and safety, the importance of controlling speed in roundabout design is well established internationally. Using the current AASHTO-based speed pre- diction tools as a base, the application of acceleration and deceleration effects appears to significantly improve the abil- ity to predict 85th-percentile speeds entering and exiting a C H A P T E R 7 Conclusions and Suggested Research
110 roundabout. This will aid in estimating speeds in the cross- walk area of a roundabout, for example. The combination of the extensive field observations of crit- ical gap and the revised speed predictions may be used to refine the current intersection sight distance procedure presented in the FHWA Roundabout Guide (1). These findings should be considered interim until a more comprehensive study of sight distance needs at roundabouts can be completed. Anecdotal evidence suggests the importance of consider- ing design details in multilane roundabout design, including vehicle path alignment, lane widths, and positive guidance to drivers through the use of lane markings. Pedestrian and Bicyclist Observations The overwhelming majority of the roundabouts observed in this observational study showed very few problems for crossing pedestrians and traversing bicyclists. From a safety perspective, where safety is measured in terms of crashes or in terms of a surrogate such as conflicts, the roundabouts observed performed very well. Out of the 769 pedestrian crossing events and 690 bicyclist events observed from video recordings, there were no observed crashes and only eight observed conflicts (0.5%). The low observation numbers confirm what was found in the crash reports that were collected for this project. Crash reports collected from 139 legs at 39 roundabouts revealed a total of five reported pedes- trian crashes and eight reported bicyclist crashes across all sites over a mean reported crash history period of 3.8 years per site. Another approach to measuring the risk to pedestrians and bicyclists is to observe the interactions between pedestrians/ bicyclists and motorists. The majority of the analyses in this study were focused around these interactions. The major findings from these analyses can be summarized as follows: ⢠Exit lanes appear to place crossing pedestrians at a greater risk than entry lanes. Motorists were less likely to yield to pedestrians on the exit side (38% of the time) compared to the entry side (23% of the time). Pedestrians and bicyclists were also more likely to hesitate when starting to cross from the exit side versus the entry side. Approximately 22% of the pedestrians that started crossing from the entry side hesitated, compared to 31% of those that started from the exit side. Similarly, 21% of the bicyclists hesitated when starting from the entry side compared to 29% when start- ing from the exit side. ⢠Two-lane legs are more difficult for pedestrians to cross than one-lane legs, primarily because of the non-yielding behaviors of motorists. On one-lane legs, 17% of the motorists did not yield to a crossing/waiting pedestrian. On two-lane legs, the non-yielding percentage was 43%. The lack of yielding was perhaps reflected in the observed pedestrian behaviors. Single-lane legs resulted in hesitation crossings 24% of the time, while two-lane legs produced hesitations 33% of the time. ⢠Roundabouts result in the type of behaviors expected when compared to other types of intersections and levels of traffic control. Roundabouts, which are under yield control, produced motorist and pedestrian behaviors that were between the behaviors observed at crossings with no traffic control and those observed at crossings with signal or stop control. Motorists not yielding to pedestrians ranged from 48% at uncontrolled locations to 32% at roundabouts to 15% at signalized locations to 4% at stop- controlled sites. Crossings in which the behavior of the pedestrian was considered to be normal were 70%, 85%, 90%, and 100% for uncontrolled, yield-controlled, signal- ized, and stop-controlled locations, respectively. However, there was no practical difference in the crossing pace of pedestrians between the various types of traffic control. ⢠Bicyclists appear to have very few problems interacting with motorists and maneuvering through a roundabout. Bicyclists approaching a roundabout were most often posi- tioned near the edge of the travel lane (73%), while bicy- clists in the circulating lane most often possessed the lane (83%). The problems that were identified were the result of inappropriate behaviors on the part of the bicyclists. One of the conflicts occurred because a circulating bicyclist was riding on the outside of the travel lane as opposed to tak- ing the lane, which resulted in the bicyclist almost being clipped by an exiting vehicle. The other conflict involved a wrong-way rider who was entering the roundabout on the exit lane. In summary, the findings of this research did not find any substantial safety problems for non-motorists at round- abouts, as indicated by there being few reported crashes and a very small number of observed conflicts. At the same time, the findings have highlighted some aspects of roundabouts where pedestrian and bicyclist ability to use the roundabout may be compromised as use of the roundabout by all modes and their subsequent interactions are greater than studied herein or where such interactions increase over time (i.e., as vehicle traffic and/or pedestrian traffic increases). For exam- ple, care must be taken to ensure that vehicles yield to waiting or crossing pedestrians. An emphasis needs to be placed on designing exit lanes to improve upon the behaviors of both motorists and pedestrians. And multilane roundabouts may require additional measures to improve upon the behaviors of motorists, pedestrians, and bicyclists. The specific countermeasures required to change the observed motoristsâ behaviors may include changes in
111 design, changes in operations, and/or targeted enforcement and education. Design changes could include reduction in the exit radius, reduction in lane widths, and/or relocation of the crosswalk. Operational changes could include static warning signs, real-time warning devices that are activated when a pedestrian is present, and/or some form of pedes- trian-actuated signalization. Enforcement and education could focus on improving user compliance with existing rules of the road. Any implemented countermeasure should be evaluated to determine if it does indeed result in behav- ioral changes that increase safety and mobility for non- motorists. The results of such evaluations should be used to change the design guidance that is currently available. No exposure data were available for the condition before a roundabout was installed at each study location, so it is unknown whether pedestrians have altered their travel pat- terns because of the presence of a roundabout. This study did not address the accessibility of roundabouts for pedestrians with visual impairments. As noted in the introduction, a separate research project (NCHRP 3-78) was developed to specifically address this issue. In addition, the U.S. Access Board is continuing to be active in proposing guidelines for accessible rights-of-way; the reader is encour- aged to stay abreast of this developing area. Suggested Research Even though the scope of this study was broad, the research team uncovered a variety of topic areas that warrant further research beyond what could be completed in this study. In some cases, the research may simply involve further analysis of the data collected as part of this study. In other cases, addi- tional data collected at a later date may be needed to provide adequate sample size and diversity. Updated Operational Models The operational models developed under this study are believed to be the best analysis of the data available in the United States at the time of this study. However, the research was conducted in the full knowledge that few sites in the United States have reached capacity, and most of those are at capacity for only short periods of time. In addition, the diver- sity of multilane roundabouts operating at capacity is partic- ularly thin. Therefore, further study is needed at a future date to (1) collect data at more sites operating at capacity for longer durations, (2) determine whether capacities have changed over time, and (3) expand the diversity of sites included in such a study, particularly among multilane roundabouts. A draft research problem statement on this topic is included in Appendix N. Updated Safety Models Unsurprisingly, the sample of available data was found to be such that better safety models could not be calibrated. While what was accomplished in this regard is still quite use- ful, it would nevertheless be worthwhile to build on the data- base and modeling effort for this project. For example, for the approach-level models, several variables appeared to influ- ence roundabout safety, but a larger sample would be required to resolve the correlations that were quite evident among several of these variables. And it would be very desir- able to have a large enough sample to develop approach-level models for injury crashes. The larger sample could be built through a combination of assembling complete data for more roundabouts and assem- bling data for additional years for those roundabouts cur- rently in the database. In assembling data for newly added roundabouts, attention should be paid to ensuring that there is variation in the variables of interest. At the same time, the sample for evaluating safety before and after roundabout installation could also be expanded because, undoubtedly, most of the intersections added to the modeling database will have been converted from some other form of control. A before-after evaluation based on a larger sample would improve knowledge of the geometric and operational condi- tions that better favor roundabout construction and safety. Given the rate at which new roundabouts are being constructed, the research team recommends that the safety analysis, both the modeling and before-after evaluation, be revisited 5 years after the data for this project were collected, or approximately 1 year from the date of this report. To this end, care should be taken to preserve the safety data and analysis worksheets developed for this project, so that the task of building on them would be considerably easier than starting afresh. Intersection Sight Distance This project revised the gap acceptance parameters used in the modeling framework for intersection sight distance orig- inally established in the FHWA Roundabout Guide. However, this project did not have the resources to more fully study the entire sight distance methodology, including the fundamen- tal question of how much sight distance is appropriate and how much sight distance may be excessive. This question is relevant, as some international studies have suggested that excessive sight distance at roundabouts leads to higher crash frequencies (4). Therefore, a new research effort is recom- mended, similar in scope to that conducted for intersection
112 sight distance at conventional intersections, to address this fundamental parameter. Additional Topics A variety of additional topics that could not be covered within the scope and budget of this project are also worth exploring further: ⢠Relationship between vehicle speeds and roundabout safety ⢠Relationship between design details and safety at multilane roundabouts ⢠Relationship between safety and illumination ⢠The operational effects between nearby traffic control devices and roundabout operations