Roundabouts: An Informational Guide – Second Edition (2010)

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.

Chapter 5/Safety Page 5-1 Roundabouts: An Informational Guide CHAPTER 5 SAFETY CONTENTS 5.1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-4 5.2 PRINCIPLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-5 5.2.1 Vehicular Conflicts at Single-Lane Roundabouts . . . . . . . . . . . . . . . 5-6 5.2.2 Vehicular Conflicts at Multilane Roundabouts . . . . . . . . . . . . . . . . . 5-8 5.2.3 Pedestrian Conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-10 5.2.4 Bicycle Conflicts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-13 5.3 OBSERVED SAFETY PERFORMANCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-14 5.3.1 Comparisons to Previous Intersection Treatment . . . . . . . . . . . . . . 5-15 5.3.2 Crash Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-16 5.3.3 Pedestrians . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-19 5.3.4 Bicyclists . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-20 5.4 INTERSECTION-LEVEL CRASH PREDICTION METHODOLOGY . . 5-22 5.4.1 Methodology to Evaluate the Safety Performance of an Existing Roundabout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-22 5.4.2 Application to Network Screening . . . . . . . . . . . . . . . . . . . . . . . . . . 5-24 5.4.3 Estimating the Safety Benefit of a Contemplated Conversion of an Existing Intersection to a Roundabout . . . . . . . . . . . . . . . . . . 5-25 5.5 APPROACH-LEVEL CRASH PREDICTION METHODOLOGY . . . . . . 5-28 5.5.1 Evaluation of Approach-Level Safety Performance . . . . . . . . . . . . 5-31 5.5.2 Consideration of Approach-Level Model Results for HSM-Type Application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-32 5.6 REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-34

Roundabouts: An Informational Guide Page 5-2 Chapter 5/Safety LIST OF EXHIBITS Exhibit 5-1 Vehicle Conflict Points for T-Intersections with Single-Lane Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-6 Exhibit 5-2 Vehicle Conflict Point Comparison for Intersections with Single-Lane Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-7 Exhibit 5-3 Total and Injury Crash Experience for U.S. Roundabouts with Four Approaches by Number of Lanes and AADT . . . . . . . . . 5-8 Exhibit 5-4 Failing to Maintain Lane Position at a Multilane Roundabout . . . . 5-9 Exhibit 5-5 Entering Next to an Exiting Vehicle at a Multilane Roundabout . . . 5-9 Exhibit 5-6 Improper Turn Conflicts at Multilane Roundabouts . . . . . . . . . . . . 5-10 Exhibit 5-7 Vehicle–Pedestrian Conflicts for One Crosswalk at Signalized Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-11 Exhibit 5-8 Vehicle–Pedestrian Conflicts at Single-Lane Roundabouts . . . . . . 5-12 Exhibit 5-9 Comparisons to Previous Intersection Treatments in the United States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-15 Exhibit 5-10 Mean Crash Reduction in Various Countries . . . . . . . . . . . . . . . . . 5-16 Exhibit 5-11 Crash Types at U.S. Roundabouts . . . . . . . . . . . . . . . . . . . . . . . . . . 5-16 Exhibit 5-12 Comparison of Crash Types at Roundabouts . . . . . . . . . . . . . . . . . 5-17 Exhibit 5-13 Graphical Depiction of Crash Types at Roundabouts . . . . . . . . . . 5-18 Exhibit 5-14 British Crash Rates for Pedestrians at Roundabouts and Signalized Intersections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-19 Exhibit 5-15 Chance of Pedestrian Death If Hit by a Motor Vehicle . . . . . . . . . 5-19 Exhibit 5-16 Percentage Reduction in the Number of Crashes by Mode in a Dutch Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-20 Exhibit 5-17 British Crash Rates for Bicycles and Motorcyclists at Roundabouts and Signalized Intersections . . . . . . . . . . . . . . . . . 5-20 Exhibit 5-18 A Comparison of Crashes in France Between Signalized Intersections and Roundabouts . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-21 Exhibit 5-19 Intersection-Level Safety Performance Models and Validity Ranges—Total Crashes . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-23 Exhibit 5-20 Intersection-Level Safety Performance Models and Validity Ranges—KAB Injury Crashes . . . . . . . . . . . . . . . . . . . . . . 5-23 Exhibit 5-21 Calculation of Total Crashes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-25 Exhibit 5-22 Calculation of Injury Crashes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-26 Exhibit 5-23 Calculation of Expected Change in Crashes Converting an Intersection to a Roundabout . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-29 Exhibit 5-24 Entering–Circulating Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-31 Exhibit 5-25 Exiting–Circulating Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-31

Chapter 5/Safety Page 5-3 Roundabouts: An Informational Guide Exhibit 5-26 Approach Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-31 Exhibit 5-27 Base Conditions for Design Variables and AMFs Implied for Unit Change in Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-33 Exhibit 5-28 Calculation of Expected Frequency of Entering–Circulating Crashes Using AMFs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5-33

Roundabouts: An Informational Guide Page 5-4 Chapter 5/Safety 5.1 INTRODUCTION The use of roundabouts is a proven safety strategy for improving intersection safety by eliminating or altering conflict types, reducing crash severity, and causing drivers to reduce speeds as they proceed into and through intersections. Decreased vehicle speeds will also decrease the speed differentials with other road users. Understanding the sensitivity of safety of the various geometric design elements and traffic exposure will assist the designer in optimizing the safety of all vehicle occupants, pedestrians, and bicyclists. In addition, the use of safety models will facilitate the planning and design of roundabouts by evaluating their safety com- pared to other intersection types and by quantifying the safety implications of design decisions. Many studies have found that one of the benefits of the installation of a round- about is the improvement in overall safety performance. Several studies in the United States, Europe, and Australia have found that roundabouts perform better in terms of safety than other intersection forms (1–4). Recent research using data in the United States (2) found that with the exception of conversions from all-way stop-controlled intersections, where limited data suggest that crash experience remains statistically unchanged, roundabouts have reduced crash frequencies for a wide range of settings (urban, suburban, and rural) and previous forms of traffic control (two-way stop and signal). This is especially evident with less frequent injury crashes. The safety benefit is greater for small- and medium-capacity round- abouts than for large or multilane roundabouts (1, 2, 5). While overall crash frequencies have been reduced, the crash reductions are most pronounced for motor vehicles, less pronounced for pedestrians, and equivocal for bicyclists and motorcyclists depending on the study and bicycle design treatments (4–6). The reasons for the increased safety level at roundabouts are: • Roundabouts have fewer vehicular conflict points in comparison to conven- tional intersections. The potential for high-severity conflicts, such as right angle and left-turn head-on crashes, is greatly reduced with roundabout use. • Low absolute speeds generally associated with roundabouts allow drivers more time to react to potential conflicts, also helping to improve the safety performance of roundabouts. Low vehicle speeds help reduce crash sever- ity, making fatalities and serious injuries uncommon at roundabouts. • Since most road users travel at similar speeds through roundabouts (i.e., have low relative speeds), crash severity can be reduced compared to some traditionally controlled intersections. • Pedestrians need only cross one direction of traffic at a time at each approach as they traverse roundabouts (i.e., crossing in two stages), as compared with many traditional intersections. Pedestrian–vehicle conflict points are reduced at roundabouts; from the pedestrian’s perspective, conflicting vehicles come from fewer directions. In addition, the speeds of motorists entering and exiting a roundabout are reduced with good design, increasing the time available for motorists to react and reducing potential crash severity. While multilane crossings still present a multiple- threat challenge for pedestrians, the overall lower speed environment

Chapter 5/Safety Page 5-5 Roundabouts: An Informational Guide helps to reduce the likelihood of collisions. As with other crossings requiring acceptance of gaps, roundabouts present visually impaired pedestrians with unique challenges, as described in Chapter 6. NCHRP Report 572 (2) presents U.S. data used to develop safety prediction models for both intersection-level and approach-level analyses. The intersection- level models were developed for total and injury collisions; the latter included fatal and definite injury but excluded possible injury collisions (i.e., they include KAB collisions on the KABCO scale). The approach-level models were developed for all severities combined for several collision types: entering/circulating, exiting/ circulating, and approaching. These models are of a form that is intended to be suit- able for eventual inclusion in the Highway Safety Manual (HSM) crash prediction procedures, although they are not included in the first edition of that document. The intersection-level models can be used to evaluate the safety performance of an existing roundabout and in the estimation of the expected safety changes if a roundabout is contemplated for construction at an existing conventional inter- section. The approach-level models are presented as tools for evaluating design options or evaluating the safety performance of specific approaches. With respect to roundabout geometry, the following observations pertain to the U.S. data analyzed: • Entering/circulating collisions increase with an increased entry width. • Entering/circulating collisions decrease with an increase in central island diameter. • Entering/circulating collisions decrease as the angle between legs increases. • Exiting/circulating collisions increase with an increasing inscribed circle diameter. • Exiting/circulating collisions increase with an increasing central island diameter. • Exiting/circulating crashes increase with an increasing circulating width. • Approach crashes increase with increasing lane width. 5.2 PRINCIPLES The frequency of crashes at an intersection is related to the number of conflict points at an intersection, as well as the magnitude of conflicting flows at each con- flict point. A conflict point is a location where the paths of two motor vehicles, or a vehicle and a bicycle or pedestrian path, diverge, merge, or cross each other. Conflicts can arise from both legal and illegal maneuvers; many of the most serious crashes are caused by failure to observe traffic control devices. The following sections present a variety of conflicts among vehicles, bicycles, and pedestrians. Both legal conflicts (queuing at an intersection, merging into a traffic stream) and conflicts prohibited by law or by traffic control devices (failure to yield to pedestrians, running a stop sign) have been included for completeness. Even though traffic control devices can significantly reduce many conflicts, they Conflict points occur where one vehicle path crosses, merges or diverges with, or queues behind the path of another vehicle, pedestrian, or bicycle.

Roundabouts: An Informational Guide Page 5-6 Chapter 5/Safety cannot eliminate them entirely due to violations of those devices. Many of the most serious crashes are caused by such violations. As with crash analyses, conflict analyses are more than the simple enumera- tion of the number of conflicts. A conflict analysis should account for the following factors: • Existence of conflict point; • Exposure, measured by the product of the two conflicting stream volumes at a given conflict point; • Severity, based on the relative velocities of the conflicting streams (speed and angle); and • Vulnerability, based on the ability for a member of each conflicting stream to survive a crash. 5.2.1 VEHICULAR CONFLICTS AT SINGLE-LANE ROUNDABOUTS Exhibit 5-1 presents a diagram of vehicle–vehicle conflict points for a traditional three-leg (T) intersection and a three-leg roundabout. As the figure shows, the num- ber of vehicle–vehicle conflict points for roundabouts decreases from nine to six for three-leg intersections. Note that these diagrams do not take into account the ability to separate conflicts in space (through the use of separate left- or right-turning lanes) or time (through the use of traffic control devices such as stop signs or traffic signals). A four-leg single-lane round- about has 75% fewer vehicle conflict points and no crossing conflict points compared to a conventional intersection. Exhibit 5-2 presents similar diagrams for a conventional four-leg (X or cross) intersection and a four-leg roundabout. As the figure shows, the number of vehicle– vehicle conflict points for roundabouts decreases from thirty-two to eight with four-leg intersections. Conflicts can be divided into four basic categories, in which the degree of severity varies, as follows: 1. Queuing conflicts. These conflicts are caused by a vehicle running into the back of a vehicle queue on an approach. These types of conflicts can occur at the back of a through-movement queue or where left-turning vehicles Exhibit 5-1 Vehicle Conflict Points for T-Intersections with Single-Lane Approaches Roundabouts bring the simplicity of a T-intersection to intersections with more than three legs.

Chapter 5/Safety Page 5-7 Roundabouts: An Informational Guide are queued waiting for gaps. These conflicts are typically the least severe of all conflicts because the collisions involve the most protected parts of the vehicle and the relative speed difference between vehicles is usually less than other conflicts. 2. Diverging conflicts. These conflicts are caused by the separating of two traffic streams. Examples include right turns diverging from through movements or exiting vehicles diverging from circulating vehicles. If the speed of one movement is significantly different from the other movement, the resulting speed differential increases the risk of a rear-end collision. 3. Merging conflicts. These conflicts are caused by the joining of two traffic streams. The most common types of crashes due to merging conflicts are side-swipe and rear-end crashes. Merging conflicts can be more severe than diverging conflicts due to the more likely possibility of collisions to the side of the vehicle, which is typically less protected than the front and rear of the vehicle. 4. Crossing conflicts. These conflicts occur where of the paths of two traffic streams intersect. These are the most severe of all conflicts and the most likely to involve injuries or fatalities. Typical crash types are right-angle crashes and head-on crashes. As Exhibit 5-1 and Exhibit 5-2 show, a roundabout eliminates vehicular cross- ing conflicts for both three- and four-leg intersections. Separate turn lanes and traffic control (stop signs or signalization) can often reduce but not eliminate the number of crossing conflicts at a traditional intersection by separating conflicts in space and/or time. However, the most severe crashes at signalized intersections occur when there is a violation of the traffic control device designed to separate conflicts by time (e.g., a right-angle collision due to running a red light and vehicle– pedestrian collisions). Therefore, the ability of single-lane roundabouts to reduce conflicts through physical, geometric features has been demonstrated to be more effective than the reliance on driver obedience of traffic control devices. Exhibit 5-2 Vehicle Conflict Point Comparison for Intersections with Single-Lane Approaches

Roundabouts: An Informational Guide Page 5-8 Chapter 5/Safety 5.2.2 VEHICULAR CONFLICTS AT MULTILANE ROUNDABOUTS Multilane roundabouts have some of the same safety performance character- istics as their simpler single-lane counterparts. However, due to the presence of additional entry lanes and the accompanying need to provide wider circulatory and exit roadways, multilane roundabouts introduce additional conflicts not present in single-lane roundabouts. This makes it important to use the minimum number of entry, circulating, and exit lanes subject to capacity considerations. For example, Exhibit 5-3, prepared from crash models developed with U.S. data (2), illustrates that crash frequencies increase with the number of circulating lanes. However, injury crash rates are much lower for both one- and two-lane roundabouts. The number of vehicular and pedestrian conflict points in both conventional intersections and roundabouts increases considerably when there are additional approach lanes. The designer is encouraged to graphically determine conflicts for a particular location, as this information can raise awareness of design issues and may be useful in public presentations. Conflicts occur at multilane roundabouts that do not happen at single-lane roundabouts. These can be categorized into three basic types: • Drivers fail to maintain lane position (Exhibit 5-4), • Drivers enter next to an exiting vehicle (Exhibit 5-5), and • Drivers turn from the incorrect lane (Exhibit 5-6). While these conflicts may also be present at conventional intersections, they can be more prevalent with drivers who are unfamiliar with roundabout opera- tion. The first two types of conflicts, in particular, can be a result of improper roundabout geometry as discussed in Chapter 6, and the latter type of conflict can be a result of improper traffic control devices. Proper driver education may also help to reduce these types of crashes. Multilane roundabouts have some of the same safety per- formance characteristics as single-lane roundabouts but introduce additional conflicts. 0 1 2 3 4 5 6 7 8 9 10 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 Average Daily Traffic To ta l C ra sh es P er Y ea r Total Crashes (1 lane) Total crashes (2 Lanes) Injury Crashes (1 or 2 lanes) Exhibit 5-3 Total and Injury Crash Experience for U.S. Roundabouts with Four Approaches by Number of Lanes and AADT Incorrect lane use and incorrect turns are multilane roundabout conflicts not present in single- lane roundabouts.

Chapter 5/Safety Page 5-9 Roundabouts: An Informational Guide As with single-lane roundabouts, the most severe vehicular crossing conflicts are eliminated and replaced by less severe merging conflicts. The additional conflicts unique to multilane roundabouts are generally low-speed side-swipe conflicts that typically have low severity. Therefore, although the number of conflicts increases at multilane roundabouts when compared to single-lane Exhibit 5-4 Failing to Maintain Lane Position at a Multilane Roundabout The overall severity (and often number) of conflicts at multi- lane roundabouts is typically less than other intersection alternatives. Exhibit 5-5 Entering Next to an Exiting Vehicle at a Multilane Roundabout

Roundabouts: An Informational Guide Page 5-10 Chapter 5/Safety roundabouts, the overall severity (and often number) of conflicts is typically less than other intersection alternatives. 5.2.3 PEDESTRIAN CONFLICTS Pedestrian–vehicle conflicts can be present at every intersection, even those with minimal pedestrian volume. The following section examines pedestrian conflicts at signalized intersections and at roundabouts. At conventional intersections, a pedestrian faces four potential vehicular conflicts, each coming from a different direction: • Left-turn, through, and right-turn movements coming from the leg of the intersection that the pedestrian is crossing; • Through movements coming from the opposite side of the intersection; • Right turns from the cross street; and • Left turns from the cross street. The amount of exposure and level of severity for each of these conflicts depends significantly on the type of traffic control used: • Two-way stop control intersections. At TWSC intersections, the most signifi- cant pedestrian conflict is pedestrians crossing the major street who have potentially severe conflicts with through vehicles on the major street. They also experience less severe conflicts with vehicles making left or right turns from or to the major street. Pedestrians crossing the minor There are four vehicle– pedestrian crossing conflicts for each crosswalk at conven- tional intersections. Exhibit 5-6 Improper Turn Conflicts at Multilane Roundabouts

Chapter 5/Safety Page 5-11 Roundabouts: An Informational Guide street generally face less severe conflicts, but conflicts still occur. These include those caused by vehicle drivers making left turns from the major street. In these cases they are looking for gaps in oncoming vehicles and may not see pedestrians crossing the minor street. In addition, drivers turning on to or crossing the major street are focused primarily on vehi- cles on the major street and may not see pedestrians. This is especially true for drivers turning right from the minor street who may look only to the left for vehicles and not notice pedestrians coming from their right. • All-way stop control intersections. Because all vehicles are required to stop, pedestrian crashes are generally less severe at AWSC intersections. The pedestrian conflicts on the near side of entering drivers are fairly benign. However, as drivers accelerate through the intersection the danger to pedestrians is greater at the far side of the intersection, whether drivers are going straight through or turning. In addition, many drivers roll through stop signs at AWSC intersections once they have perceived that there are no imminent vehicle conflicts. These drivers may occasionally fail to notice pedestrians crossing at the intersection. • Signalized intersections. Traffic signals can potentially reduce the likelihood of pedestrian–vehicle conflicts through the use of signal phasing that allows only a few legal movements at any given time. However, there are four vehicle movements at signalized intersections that create potential conflicts with pedestrians under common signal phasing schemes (see Exhibit 5-7): – Red light running (illegal)—includes through movements, left turns, and right turns. These movements, particularly the through move- ments, have the highest potential severity due to high vehicular speeds and the potential for surprise to the pedestrian. Exhibit 5-7 Vehicle–Pedestrian Conflicts for One Crosswalk at Signalized Intersections Four vehicle movements at sig- nalized intersections can result in pedestrian crossing conflicts. Right turn on green conflict Red light running conflict Left turn on green conflict Red light running or right turn on red conflict

Roundabouts: An Informational Guide Page 5-12 Chapter 5/Safety – Right turns on green (legal). These movements offer the highest poten- tial for visibility between drivers and pedestrians, but drivers may occasionally fail to notice pedestrians crossing at the intersection. – Left turns on green (legal for protected-permissive or permissive left-turn phasing). This represents a significant risk for pedestrians, as a driver undertaking a permissive left turn from the major street is primarily looking ahead for gaps in oncoming traffic and may not see pedestrians crossing the minor street. – Right turns on red (legal in most of the United States and Canada). These have a moderate potential for severity due to the driver looking to the left for a gap and not seeing the pedestrian crossing in front of the stopped driver from the right. In some cases drivers turning right on red move into the crosswalk to improve their sight lines to the left and cause pedestrians to pass either in front or behind them. Pedestrians at roundabouts, on the other hand, face two conflicting vehicular movements on each approach, as depicted in Exhibit 5-8: • Conflict with entering vehicles and • Conflict with exiting vehicles. At conventional and roundabout intersections with multiple approach lanes, an additional conflict is added with each additional lane that a pedestrian must cross. Exhibit 5-8 Vehicle–Pedestrian Conflicts at Single-Lane Roundabouts The direction conflicting vehi- cles will arrive from is more predictable for pedestrians at roundabouts.

Chapter 5/Safety Page 5-13 Roundabouts: An Informational Guide 5.2.4 BICYCLE CONFLICTS Bicyclists face similar conflicts as motor vehicles at both signalized intersec- tions and roundabouts. However, because bicyclists typically ride on the right side of the road between intersections, they face additional conflicts when they need to merge into the flow of motor vehicle traffic or where motor vehicles cross their path. Conflicts unique to bicyclists occur on each approach to conventional four-leg intersections, and the typical vehicle–vehicle conflicts within the intersec- tion can be more significant for bicyclists. The conflicts experienced by bicyclists vary widely depending on how they choose to negotiate the intersection: • Many cyclists making through movements will continue to ride on the right side of the road as they enter a conventional intersection; this action results in a conflict point with motorists who are making right turns (the “right hook” crash type). • Experienced cyclists often merge into the flow of motor vehicle traffic prior to entering the intersection, reducing the likelihood of right-hook crashes and making themselves more visible to other drivers at the inter- section. This action results in a possible merging conflict point in advance of the intersection. Experienced cyclists prefer this conflict point because they are able to control the location and dynamics of the merge. • Cyclists making vehicular-style left turns need to merge into other vehicle traffic and sometimes across travel lanes if there are multiple through travel lanes and/or a left-turn only lane. This maneuver results in at least one and possibly multiple merging conflicts. • Some cyclists may choose to make pedestrian-style left turns on the road- way by riding straight through the intersection on the right side, stopping at the far corner, turning their bicycle 90° to the left, and traveling straight through as if they were coming from the right. These cyclists experience the typical vehicle–vehicle conflicts as well as two right-hook conflicts, one for each of their through movements. • Some cyclists (typically children or less-experienced adult cyclists) choose to travel through intersections by using the sidewalks and crosswalks. These cyclists experience the same conflict points as described for pedestri- ans in Section 5.2.3. If these cyclists choose to ride their bikes through crosswalks, the probability of a crash is generally higher due to their speed and reduced ability to react to a possible conflict. These cyclists also experi- ence potential conflicts with pedestrians on sidewalks or in crosswalks. Similarly to conventional intersections, the conflicts experienced by bicyclists at roundabouts are dependent on how they choose to negotiate the roundabout. The primary issue is whether cyclists choose to travel through the roundabouts like other vehicles or like pedestrians. Some roundabouts include design features that make it easy for cyclists to make this choice. If bicyclists travel through a roundabout as a vehicle, they experience several conflicts unique to bicyclists: • A merging conflict occurs at the point where the bicyclist merges into the traffic stream. At both conventional inter- sections and roundabouts, the type and number of conflicts experienced by bicyclists are dependent on how they choose to negotiate the intersection.

Roundabouts: An Informational Guide Page 5-14 Chapter 5/Safety • At multilane roundabouts, it is recommended that cyclists travel through the roundabout in the same manner as other vehicles. Therefore, cyclists making left turns may encounter multiple merging conflicts as they change into a lane designated for left-turn movements. • At roundabouts where a right-turn only lane or a right-turn bypass lane is present, cyclists making through movements or left-turn movements may also experience additional merging conflicts. • Cyclists should not choose to travel on the outside part of the circulatory roadway, even at multilane roundabouts. However, some bicyclists may choose to ride in this position past a roundabout exit, where they face a potential conflict with exiting vehicles. • When circulating at a roundabout, bicyclists are less visible and therefore more vulnerable to the merging and exiting conflicts that happen at multi- lane roundabouts. This is especially true if cyclists hug the curb as described above, because motorists further to the right are more out of the primary sight lines of entering drivers. In addition, since cyclists typically travel slightly slower than other vehicles in roundabouts, it is possible for motorists to pass cyclists and cut them off when exiting. • As described above, bicyclists should make left turns in the same manner as vehicles. However, their slower speed makes it possible for motorists to pass them on the right, resulting in another possible conflict when left- turning bicyclists exit the roundabout. If bicyclists travel through a roundabout like pedestrians, then they experi- ence the typical pedestrian–vehicle conflicts as described in Section 5.2.3 as well as several conflicts unique to bicyclists: • A bicycle–pedestrian conflict occurs at the point where the bicyclist gets onto the sidewalk or shared-use path. • On shared-use paths or on sidewalks at roundabouts, if bicyclists con- tinue to ride, additional bicycle–pedestrian conflicts occur wherever bicycle and pedestrian movements cross. • If bicyclists choose to ride their bikes through crosswalks, the probability of a crash is generally higher due to their speed and reduced ability to react to a possible conflict with vehicles. • A merging conflict exists with other bicyclists and possibly motor vehicles at the point where the bicyclists reenter the roadway after traveling through the roundabout as a pedestrian. 5.3 OBSERVED SAFETY PERFORMANCE This section summarizes the overall safety performance of roundabouts in the United States and the detailed collision types experienced. Pedestrian and bicycle crash statistics are discussed separately using international data. Bicycle–pedestrian conflicts can also occur on sidewalks or shared use paths adjacent to the roundabout.

Chapter 5/Safety Page 5-15 Roundabouts: An Informational Guide 5.3.1 COMPARISONS TO PREVIOUS INTERSECTION TREATMENT The most up-to-date knowledge on the safety effects of roundabout conver- sions in the United States is summarized in NCHRP Report 572 (2). Before-and-after conversion data were collected for 55 locations with variations in previous inter- section treatment (two-way stop, all way stop, or signal control), environment (urban and rural), and number of circulating lanes. Exhibit 5-9 presents the results of this study for both total and injury accidents, including the expected percent reduction and associated standard error. Injury accidents are defined as those involving definite injury or fatality. In other words, property damage only (PDO) and possible injury accidents (C on the KABCO scale) are not included. Results are shown separately for various logical groups for which sample sizes were large enough to facilitate a disaggregate analysis. The percent reduction can be applied to the expected crash frequency prior to conversion to estimate the expected crash frequency of a contemplated roundabout or the expected reduction in crashes following conversion. Exhibit 5-9 Comparisons to Previous Intersection Treatments in the United StatesControl Before Sites Setting Lanes Estimate of the Percent Reduction in Crashes (and Standard Error) All Injury + Fatal All Sites 55 All All 35.4% (3.4) 75.8% (3.2) Signalized 9 All All 47.8% (4.9) 77.7% (6.0) 4 Suburban 2 66.7% (4.4) Sample too small to analyze 5 Urban All Effects insignificant 60.1% (11.6) All-way stop 10 All All Effects insignificant Effects insignificant Two-way stop 36 All All 44.2% (3.8) 81.8% (3.2) 9 Rural 1 71.5% (4.0) 87.3% (3.4) 17 Urban All 29.0% (9.0) 81.2% (7.9) 12 1 39.8% (10.1) 80.3% (10.0) 5 2 Sample too small to analyze Sample too small to analyze 10 Suburban All 31.8% (6.7) 71.0% (8.3) 4 1 78.2% (5.7) 77.6% (10.4) 6 2 19.3% (9.1) 68.0% (11.6) 27 Urban/ Suburban All 30.8% (5.5) 74.4% (6.0) 16 1 56.3% (6.0) 77.7% (7.4) 11 2 17.9% (8.2) 71.8% (9.3) Overall, there is an observed reduction of 35% and 76% in total and injury crashes, respectively, following conversion to a roundabout. These values are consistent with results from international studies, as shown in Exhibit 5-10. The findings of these studies all show that injury crashes are reduced more dramatically than crashes involving property damage only. This is in part due to the configuration of roundabouts, which eliminates severe crashes such as left- turn, head on, and right angle crashes. Other conclusions specifically drawn from the U.S. study (2) are as follows: • Control type before. There are large and highly significant safety benefits of converting intersections with signals and two-way stop control to

Roundabouts: An Informational Guide Page 5-16 Chapter 5/Safety roundabouts. The benefits are greater for injury crashes than for all crash types combined. For the conversions from all-way stop control there is no apparent safety effect. • Number of lanes. The safety benefit is greater for single-lane roundabouts than for two-lane designs for urban and suburban roundabouts that were previously two-way stop-controlled. • Setting. The safety benefits for rural installations, which were all single- lane, were greater than for urban and suburban single-lane roundabouts. • Additional insights. Further analysis provided the following insights: – The safety benefit appears to decrease with increasing AADT, irrespective of control type before, number of lanes, and setting. – For various combinations of settings, control type before, and number of lanes for which there were sufficiently large samples, there was no apparent relationship to inscribed circle diameter or central island diameter. – The reduction in all types of crashes and injury crashes is particularly notable in rural environments where approach speeds are high. 5.3.2 CRASH TYPES It is instructive for designers to examine details of crash types and location at roundabouts. Exhibit 5-11 shows the percentage of the main crash types found in U.S. data in an analysis of 39 roundabouts where detailed crash reports were reviewed (2). As can be seen from the exhibit, over half of the crashes are two- vehicle crashes involving entering or exiting the roundabout. Further distinction with entering–circulating and exiting–circulating crashes can be made between single-lane and multilane roundabouts. For single-lane roundabouts, 80% of these Exhibit 5-11 Crash Types at U.S. Roundabouts Crash Type Percent Entering–Circulating 23 Exiting–Circulating 31 Rear-End on Leg 31 Loss of Control on Leg 13 Pedestrian 1 Bicycle 1 Source: (2) Exhibit 5-10 Mean Crash Reduction in Various Countries Country Mean Reduction (%) All Crashes Injury Crashes Australia 41–61% 45–87% France - 57–78% Germany 36% - Netherlands 47% - United Kingdom - 25–39% United States 35% 76% Source: (7 ), France (8), U.S. (2)

Chapter 5/Safety Page 5-17 Roundabouts: An Informational Guide Exhibit 5-12 Comparison of Crash Types at Roundabouts Crash type France Queensland, Australia United Kingdom1 United States Single- Lane Double - Lane 1. Failure to yield at entry (entering-circulating) 36.6% 50.8% 71.1% 13% 17% 2. Single-vehicle run off the circulatory roadway 16.3% 10.4% 8.2% 2 50%2 28%2 3. Single vehicle loss of control at entry 11.4% 5.2% 2 2 2 4. Rear-end at entry 7.4% 16.9% 7.0%3 34% 19% 5. Circulating–exiting 5.9% 6.5% 4% 6. Pedestrian on crosswalk 5.9% 3.5%4 4%5 7. Single vehicle loss of control at exit 2.5% 2.6% 2 8. Exiting–entering 2.5% 1% 9. Rear-end in circulatory roadway 0.5% 1.2% 10. Rear-end at exit 1.0% 0.2% 11. Passing a bicycle at entry 1.0% 12. Passing a bicycle at exit 1.0% 13. Weaving in circulatory roadway 2.5% 2.0% 14. Wrong direction in circulatory roadway 1.0% 15. Pedestrian on circulatory roadway 3.5% 4 16. Pedestrian at approach outside crosswalk 1.0% 4 Other collision types 2.4% 10.2% 2% 3% Other sideswipe crashes 1.6% 24%6 Notes: 1. Data are for “small” roundabouts [curbed central islands >13 ft (4 m) diameter, relatively large ratio of inscribed circle diameter to central island size] 2. Reported findings do not distinguish among single-vehicle crashes. 3. Reported findings do not distinguish among approaching crashes. 4. Reported findings do not distinguish among pedestrian crashes. 5. Reported findings combine pedestrian and bicycle crashes. 6. Reported findings do not distinguish among sideswipe crashes. Sources: France (10), Australia (11), United Kingdom (1), United States (2) two types of crashes were entering–circulating, with 20% exiting–circulating. However, for multilane roundabouts, the opposite type of crash is predominant: 64% were exiting–circulating, with 36% entering–circulating. Additional data compiled by the IIHS (9) provides a summary of crash types at 29 single-lane roundabouts and 9 multilane roundabouts in Maryland. The study represents 149 crashes at the single-lane sites and 134 crashes at the multilane sites for which at least 2 years of data was available. Six of the single-lane roundabouts accounted for 59% of all crashes at single-lane roundabouts studied, and 2 round- abouts accounted for more than 80% of all crashes at the multilane roundabouts. Crash-type results from the IIHS study are presented in Exhibit 5-12 along with international data for comparison. Exhibit 5-13 illustrates the crash types

Roundabouts: An Informational Guide Page 5-18 Chapter 5/Safety Exhibit 5-13 Graphical Depiction of Crash Types at Roundabouts (Numbers correspond to those in Exhibit 5-12.) Source: (12) identified in Exhibit 5-12. Exhibit 5-12 shows that a variety of distinctive crash types can take place at roundabouts. A designer should be aware of these crash types when making decisions about alignment and location of fixed objects. These crash types are suggested as conflict types for reporting crashes at roundabouts and conducting traffic conflict analysis.

Chapter 5/Safety Page 5-19 Roundabouts: An Informational Guide 5.3.3 PEDESTRIANS As described previously, vehicular injury crashes normally decrease when roundabouts are installed at an existing intersection. The safety benefits of round- abouts have been found to carry over to pedestrians as well, as shown in the British statistics of Exhibit 5-14. This may be due to the reduced speeds at round- abouts as compared with the previous intersection forms. Exhibit 5-15 Chance of Pedestrian Death If Hit by a Motor Vehicle Intersection Type Pedestrian Crashes per Million Trips Mini-roundabout 0.31 Conventional roundabout (older designs) 0.45 Flared roundabout (newer designs) 0.33 Signals 0.67 Source: (1, 14) Exhibit 5-14 British Crash Rates for Pedestrians at Roundabouts and Signalized Intersections Source: (13) For pedestrians, the risk of being involved in a severe collision is lower at roundabouts than at other forms of intersections due to the slower vehicle speeds. Likewise, the number of conflict points for pedestrians is lower at roundabouts than at other intersections, which can lower the frequency of crashes. The splitter island between entry and exit also allows pedestrians to resolve conflicts with entering and exiting vehicles separately. For pedestrians, speed plays a significant role in whether a vehicle–pedestrian crash will result in a fatality. Exhibit 5-15 shows that a pedestrian is about 8 times more likely to die when struck at 30 mph (50 km/h) than at 20 mph (32 km/h)— a difference of only 10 mph (13). Therefore, the difference in design speed is criti- cal to all users who are not within the protective body of a motorized vehicle. The minor additional delay or inconvenience to drivers of lower-speed round- about designs (as compared to higher-speed roundabout designs) is a trade-off for the substantial safety benefit to pedestrians (and bicyclists). Older drivers may ben- efit from the additional time to perceive, think, react, and correct for errors (as may all users). It should be clarified that there has been no specific research performed on older drivers, older pedestrians, and older bicyclists at roundabouts. It should also be noted that visually impaired pedestrians are not provided the audible cues

Roundabouts: An Informational Guide Page 5-20 Chapter 5/Safety from vehicle streams that are available at a signal-controlled intersection. For example, at roundabout exits it may be difficult to discern the sound of vehicles that will continue to circulate from those exiting the roundabout. Therefore, infor- mation needs to be provided to these users through various appropriate design features for them to safely locate and navigate the crossings at roundabouts. A Dutch study of 181 intersections converted to roundabouts (4) found reduc- tions of 73% in all pedestrian crashes and 89% in pedestrian injury crashes. In this study, all modes shared in the safety benefits to greater (passenger cars) or lesser extents (bicycles), as shown in Exhibit 5-16. Bicyclists experience more problems at roundabouts than any other road user. Intersection Type Bicyclist Crashes per Million Trips Motorcyclist Crashes per Million Trips Mini-roundabout 3.11 2.37 Conventional roundabout 2.91 2.67 Flared roundabout 7.85 2.37 Signals 1.75 2.40 Source: (1, 14) Exhibit 5-17 British Crash Rates for Bicycles and Motorcyclists at Roundabouts and Signalized Intersections A risk analysis of 59 roundabouts and 124 signalized intersections was carried out on crash data in Norway between 1985 and 1989. Altogether, 33 crashes involv- ing personal injury were recorded at the 59 roundabouts. Only one of these crashes involved a pedestrian, compared with the signalized intersections where pedestrians were involved in 20% of the personal injury crashes (57 of 287 injury crashes) (15). 5.3.4 BICYCLISTS Safety studies on bicyclists at roundabouts have mixed findings. As shown in Exhibit 5-17, in Britain, bicyclists fare worse in terms of crashes at roundabouts than at signalized intersections. A French study (5) compared the crashes in 1988 in 15 towns in the west of France at signalized intersections and roundabouts, as shown in Exhibit 5-18. The conclusions from the analysis were: • There were twice as many injury crashes per year at signalized intersec- tions than at roundabouts. • Two-wheel vehicles were involved in injury crashes more often (+77%) at signalized intersections than at roundabouts. • People were more frequently killed and seriously injured per crash (+25%) at roundabouts than at signalized intersections. • Proportionally, two-wheel vehicle users were more often involved in crashes (+16%) at roundabouts than at signalized intersections. Furthermore, the consequences of such crashes were more serious. Exhibit 5-16 Percentage Reduction in the Number of Crashes by Mode in a Dutch Study Mode All Crashes Injury Crashes Passenger car 63 95 Moped 34 63 Bicycle 8 30 Pedestrian 73 89 Total 51 72 Source: (4)

Chapter 5/Safety Page 5-21 Roundabouts: An Informational Guide All European countries report that a more careful design is necessary to enhance bicyclist safety. The type of bicycle crashes depends on the bicycle facili- ties provided at the roundabout. If there are no bicycle facilities, or if there is a bike lane on the outer area of the circulatory roadway, crashes typically occur between entering cars and circulating bicyclists as well as between exiting cars and bicyclists that are circulating around the outer edge of the circulatory roadway. Improperly placed signs on the splitter island may also be a contributing factor. As a result, most European countries have policies: • To avoid bike lanes on the outer edge of the circulatory roadway; • To allow bicyclists to mix with vehicle traffic without any separate facility in the circulatory roadway when traffic volumes are low, on single lane roundabouts operating at lower speeds [e.g., up to 8,000 vehicles per day in the Netherlands (4)]; and • To introduce separated bicycle facilities outside the circulatory roadway when vehicular and bicycle volumes are high. These separated bicycle facilities cross the exits and entries at least one car length from the edge of the circulatory roadway lane, adjacent to the pedestrian crossings. In some countries (e.g., Germany), bicyclists have priority over entering and exiting cars, especially in urban areas. Other countries (e.g., Netherlands) prefer to give priority to car traffic, showing a yield sign to bicyclists. The latter solution (i.e., separated bicycle facilities with vehicular traffic priority at the crossing points) is the standard solution for rural areas in most European countries. Extrapolating European bicycling experience to the United States should be done with caution since drivers in Europe are more accustomed to interacting with bicyclists. Speed is a fundamental risk factor in the safety of bicyclists. Typical on-road bicyclist speeds are in the range of 12 to 20 mph (20 to 30 km/h), and designs that constrain the speeds of vehicles to similar values will minimize the relative speeds and thereby improve safety. Design features that slow traffic are considered safe treatments for bicyclists (16). These may include tightening entry curvature and entry width and radial alignment of the legs of a roundabout, such as with the Exhibit 5-18 A Comparison of Crashes in France between Signalized Intersections and Roundabouts Signalized Intersections Roundabouts Number of intersections 1238 179 Number of personal injuries 794 59 Number of crashes involving two-wheel vehicles 278 28 Personal injury crashes/year/intersection 0.64 0.33 Two-wheel vehicle crashes/year/intersection 0.23 0.13 Crashes to two-wheel vehicles per 100 crashes 35.0 40.7 Serious crashes/year/crossroad 0.14 0.089 Serious crashes to two-wheel vehicles/year/crossroad 0.06 0.045 Serious crashes/100 crashes 21.9 27.1 Serious crashes to two-wheel vehicles/100 crashes to a two-wheel vehicle 27.0 33.3 Source: (5) Typical European practice is to provide separated bicycle facili- ties outside the circulatory roadway when vehicular and bicycle volumes are high.

Roundabouts: An Informational Guide Page 5-22 Chapter 5/Safety urban compact design. In addition, multilane roundabouts should not be used where they are not needed for capacity purposes in the short term, as single-lane roundabouts are much safer for bicyclists. 5.4 INTERSECTION-LEVEL CRASH PREDICTION METHODOLOGY Intersection-level crash prediction models can be used to evaluate the safety performance of an existing roundabout relative to its peers and in the estimation of the expected safety changes if a roundabout is contemplated for construction at an existing, conventional intersection. A key proviso is a requirement that the models can be assumed as representative of the pertinent jurisdiction or can be recalibrated using data representative of that jurisdiction. For an existing roundabout, the safety-performance estimate can be used in a network screening process to examine the performance of that roundabout in relation to other roundabouts or other intersections. For roundabouts performing below par from a safety perspective, diagnostic procedures can be used to isolate any problems and develop corrective measures. The methodology provides a means to combine model predictions and observed accident frequencies into a single, refined estimate of the expected crash frequency so that the observed crash history of a site can be considered in the estimation process. This empirical Bayes (EB) methodology recognizes that the observed crash frequency, by itself, is a poor estimate of safety performance due to the randomness of crash counts. Intersection-level models for roundabouts in the United States are docu- mented in NCHRP Report 572 (2) and shown in Exhibit 5-19 and Exhibit 5-20. The dispersion parameter in these tables was estimated in the model calibration process and is used in the EB methodology as illustrated below. 5.4.1 METHODOLOGY TO EVALUATE THE SAFETY PERFORMANCE OF AN EXISTING ROUNDABOUT Step 1: Assemble data, including the number of approaches, number of circu- lating lanes, and the count of total and KAB injury crashes (i.e., excluding possible injury crashes) for the roundabout of interest for a period of n years (up to 10 years). For the same time period, obtain or estimate a total entering AADT representative of that time period. Step 2: Select the appropriate roundabout level model from Exhibit 5-19 or Exhibit 5-20 and use to estimate the annual number of crashes (P) that would be expected at roundabouts with traffic volumes and other characteristics similar to the one being evaluated. If the selected model can be assumed to represent the jurisdiction, it can be used directly. If, as may be more likely the case, the model cannot be assumed to represent the jurisdiction, a calibration multiplier must first be estimated using data (similar

Chapter 5/Safety Page 5-23 Roundabouts: An Informational Guide Exhibit 5-19 Intersection-Level Safety Performance Models and Validity Ranges— Total Crashes to data acquired in Step 1) from a sample of roundabouts representative of ones in that jurisdiction. At a minimum, a dataset for at least 10 roundabouts with a mini- mum of 50 crashes is needed. The recalibration multiplier is the sum of crashes recorded in this dataset divided by the sum of the crashes predicted by the model for this dataset. The model from Exhibit 5-19 or Exhibit 5-20 is then applied with the recalibration multiplier to estimate the annual number of crashes (P). Step 3: Combine the model estimate (P) with the count of crashes (x) in the n years of observed data to obtain an estimate of the expected annual number of crashes (m) at the roundabout. This estimate of m is calculated as: w P k nP 1 1 = ( ) + m w x w P= +1 2 Exhibit 5-20 Intersection-Level Safety Performance Models and Validity Ranges— KAB Injury Crashes Equation 5-1 Equation 5-2

Roundabouts: An Informational Guide Page 5-24 Chapter 5/Safety where m = expected annual crash frequency; x = total crashes observed; P = predicted annual number of crashes; n = years of observed data; and k = dispersion parameter for a given model (given in Exhibit 5-19 or Exhibit 5-20). Estimates can be obtained for crashes of all severities combined (total crashes) or for KAB-injury crashes only. 5.4.2 APPLICATION TO NETWORK SCREENING In network screening, the refined safety performance estimate, m, can be used to assess how well an existing roundabout is performing relative to other similar roundabouts or to intersections of other types. Comparisons may be made to the average expected crash frequency of a collection of other sites or to specific sites. If the other site(s) are also roundabouts, the appropriate models would be selected from Exhibit 5-19 or Exhibit 5-20 and recalibrated if necessary. If the other sites are other intersection types, then similar models specific to those site types need to be assembled. Many jurisdictions may have calibrated their own models for other intersection types; otherwise, models from other sources may be adapted by esti- mating a recalibration multiplier using the procedure outlined in Section 5.4.2.1. More detailed information on network screening methods is available in the HSM and in the documentation for FHWA’s SafetyAnalyst software (17). 5.4.2.1 Comparison to the Average Expected Crash Frequency of Similar Roundabouts Comparing the expected crash frequency of a particular roundabout to the aver- age expected frequency involves comparing that site’s EB estimate to the model estimate for roundabouts with similar numbers of approaches and circulating lanes. For instance, from Exhibit 5-21 an analyst could conclude that the safety performance for that roundabout is worse than that for similar roundabouts since its expected crash frequency, 3.94, is larger than the model estimate for similar roundabouts, 3.39. 5.4.2.2 Comparison to Other Specific Sites This comparison involves comparing the site’s EB estimate to the EB estimate for the other sites. These sites could be roundabouts or all intersections in a juris- diction (other roundabouts and other conventional intersections). A useful application of these estimates is to rank sites in descending order of the EB esti- mate of the expected crash frequency to prioritize the sites for a more detailed investigation of safety performance. An alternative method is to rank sites by the difference between the EB estimate and the prediction model estimate. Using Exhibit 5-21, the ranking measures described above would provide the following results: (1) for the first method, a value of 3.94 would be used, or (2) for the second method, a value of (3.94 − 3.39 = 0.55) would be used. w k k nP 2 1 1 = ( ) ( ) + Equation 5-3

Exhibit 5-21 Calculation of Total Crashes Chapter 5/Safety Page 5-25 Roundabouts: An Informational Guide These measures can be calculated for total crashes and for injury crashes and, from the differences between the two estimates, for non-injury crashes. A severity- weighted ranking measure can be derived by applying weights to injury and non-injury crashes that reflect their relative severity. (In estimating a severity ranking measure for the second method, a negative difference in crash frequency is converted to a value of zero.) Thus, continuing from Exhibit 5-21, using the EB weights on their own, the appropriate safety performance model and dispersion parameter k from Exhibit 5-20, given four approaches and one circulating lane, for injury crashes are as shown in Exhibit 5-22. 5.4.3 ESTIMATING THE SAFETY BENEFIT OF A CONTEMPLATED CONVERSION OF AN EXISTING INTERSECTION TO A ROUNDABOUT The objective of this procedure is to provide designers and planners with a tool to estimate the change in crash frequency expected with the installation of a roundabout at an existing controlled intersection.

Roundabouts: An Informational Guide Page 5-26 Chapter 5/Safety A safety performance model representative of the existing intersection is required. This, again, will require that one exist for the jurisdiction or that data are available to enable a recalibration of a model calibrated for another jurisdiction using the procedure outlined earlier. The model for the existing intersection would be used, along with the intersection’s crash history, in the empirical Bayes proce- dure to estimate the expected crash frequency with the status quo in place. This EB estimate would then be compared to the expected frequency should a roundabout be constructed to estimate the benefit of converting the intersection to a roundabout. The expected frequency should a roundabout be constructed is estimated from an intersection-level model. As before, this requires that it be possible to recalibrate intersection-level models or that existing models be deemed adequate for the jurisdiction. Where there is no applicable intersection-level model for the jurisdic- Exhibit 5-22 Calculation of Injury Crashes

Chapter 5/Safety Page 5-27 Roundabouts: An Informational Guide tion, an alternate approach can be used. In this, the results of the before–after study presented in Exhibit 5-9 can be applied as accident modification factors (AMFs) to the expected crash frequency with the status quo in place to find the expected benefit. More details on this alternate approach are provided in the HSM. The first approach is preferred to the alternate and is most convenient because a comprehensive set of accident modification factors for a large number of conditions, including AADT levels, which would be required for properly applying the second approach, is not likely to be available. 5.4.3.1 Overview of the Recommended Approach This example assumes that a stop-controlled intersection is being considered for conversion to a roundabout. Step 1: Assemble data and accident prediction models for stop-controlled intersections and roundabouts. 1. For the past 5 years (if possible), obtain the count of total and injury crashes for the stop-controlled intersection under review. 2. For the same period, obtain or estimate the average total entering AADTs. 3. Estimate the average annual entering AADTs that would prevail for the period immediately after the roundabout is installed. 4. Assemble required crash prediction models for stop-controlled intersec- tions and roundabouts for both total crashes and for KAB injury crashes. If the models cannot be assumed to be representative of the jurisdiction, a calibration multiplier must first be estimated using data from a sample of roundabouts representative of that jurisdiction and the procedure outlined earlier Step 2: Use the EB procedure documented in Section 5.4.1 with the data from Step 1 and the stop-controlled intersection model to estimate the expected annual number of total and KAB injury crashes that would occur without conversion (i.e., had the intersection remained stop-controlled). The EB estimate for non-KAB crashes is then derived as the EB estimate for total crashes, minus the EB estimate for injury crashes. Step 3: Use the appropriate intersection-level model from Exhibit 5-19 or Exhibit 5-20 and the AADTs from Step 1 to estimate the expected number of total and injury crashes that would occur if the intersection were converted to a round- about. The estimate for non-KAB crashes is then derived as the model estimate for the total minus the model estimate for injury. Step 4: Obtain, for KAB injury and non-KAB crashes, the difference between the stop-controlled EB estimate from Step 2 and the intersection-level model esti- mates from Step 3. Step 5: Applying suitable dollar values for KAB injury and non-KAB crashes to the estimates from Step 4, obtain the estimated net safety benefit of converting the intersection to a roundabout. A useful source of these dollar values is FHWA’s crash cost estimates (18).

Roundabouts: An Informational Guide Page 5-28 Chapter 5/Safety Step 6: Compare estimated net safety benefit from Step 5 against the annual- ized roundabout conversion costs, considering other impacts if desired, and using conventional economic analysis tools. How and whether this is done is very juris- diction-specific. Conventional methods of economic analysis can be applied after obtaining estimates of the economic values of changes in delay, fuel consumption, and other impacts. The results of the analysis above may indicate that roundabout conversion is justified based on a consideration of safety benefits. This result may be considered in context with other factors such as the following: • Other improvement measures at the existing intersection may be more cost effective. • Other impacts (delay, fuel consumption, etc.) may need to be assessed. • In a system context, other locations may be more deserving of a round- about. In other words, the results of the above analysis should be considered in the context of a more comprehensive, system-wide safety resource allocation process. See Exhibit 5-23 for an example calculating the expected change in crashes when converting an intersection to a roundabout. 5.5 APPROACH-LEVEL CRASH PREDICTION METHODOLOGY At the approach level, models are used for separately predicting three crash types (entering–circulating, exiting–circulating, and approach crashes) as a func- tion of AADT and design characteristics. A range of alternative models are available depending on the design features of interest. Entering–circulating crashes are computed using Equation 5-4. where parameters are defined in Exhibit 5-24. Exit-circulating crashes are computed using Equation 5-5. where parameters are defined in Exhibit 5-25. Approach crashes are computed using Equation 5-6. where parameters are defined in Exhibit 5-26. These models can be used for evaluating the safety at the approach level of existing roundabouts or alternative roundabout design options. There are two possibilities for application in this context: 1. Direct application for an existing roundabout: In this the model is used directly by substituting values of AADT and design characteristics to Crashes year a EnteringAADT ea b ApproachHa= ( )0 1 1 lfWidth( )[ ] Crashes year a ExitingAADT CircAADT ea a b= ( ) ( )0 1 2 1 51 5Var b Var( )+ + ( )[ ]. . . Crashes year a EnteringAADT CircAADT ea a= ( ) ( )0 1 2 b Var b Var1 51 5( )+ + ( )[ ]. . .Equation 5-4 Equation 5-6 Equation 5-5

Exhibit 5-23 Calculation of Expected Change in Crashes Converting an Intersection to a Roundabout Chapter 5/Safety Page 5-29 Roundabouts: An Informational Guide

Roundabouts: An Informational Guide Page 5-30 Chapter 5/Safety Exhibit 5-23 (cont.) Calculation of Expected Change in Crashes Converting an Intersection to a Roundabout

Chapter 5/Safety Page 5-31 Roundabouts: An Informational Guide evaluate the safety performance of an existing roundabout with respect to the three crash types. 2. Application for a roundabout being designed or redesigned: In this, similar in principle to the predictive methodologies in the HSM and the Interactive Highway Safety Design Model (IHSDM), models with AADT as the only variable (Model 1 in Exhibit 5-24, Exhibit 5-25, and Exhibit 5-26) are con- sidered as base models for average design conditions, and AMFs are applied for design features that are different from average conditions. Coefficients from models that include design variables are used in devel- oping these AMFs. Details of these applications are provided below. 5.5.1 EVALUATION OF APPROACH-LEVEL SAFETY PERFORMANCE While the approach-level models have been developed to assist with design decisions, they can also be used in an EB procedure to estimate the expected safety performance at an approach or a number of approaches to an existing round- about, provided as before that the models can be assumed as representative of the pertinent jurisdiction or can be recalibrated using representative data from that jurisdiction. This would be used to compare the safety performance of the subject roundabout approach to that of other similar approaches. For entities performing below par Exhibit 5-26 Approach Models Model No. Multiplier a0 Entering AADT a1 Circ. AADT a2 Entry Radius (ft) b1 Entry Width (ft) b2 Central Island Diameter (ft) b3 Angle To Next Leg (deg) b4 1/Entry Path Radius (1/ft) b5 1 0.00000176 1.0585 0.3672 — — — — — 2 0.00000216 0.9771 0.3088 0.0099 — — — — 3 0.00000474 0.9217 0.2900 — 0.0582 -0.0076 — — 4 0.00000213 1.0048 0.3142 0.0103 — -0.0046 — — 5 0.00015668 0.9499 0.2687 0.0105 — — -0.0425 — 6 0.00073488 0.7018 0.1321 — 0.0511 — -0.0276 — 7 0.00012735 0.8322 0.1370 — — — — 138.096 (Shaded row indicates preferred model.) Model No. Multiplier a0 Exiting AADT a1 Circ. AADT a2 Inscribed Circle Diameter (ft) b1 Central Island Diameter (ft) b2 Circ. Width (ft) b3 1/Circ. Path Radius (1/ft) b4 1/Exit Path Radius (1/ft) b5 1 0.00044631 0.3413 0.5172 — — — — — 2 0.00000846 0.2801 0.2530 0.0222 0.1107 — — 3 0.00001308 0.3227 0.3242 — 0.0137 0.1458 — — 4 0.02215926 0.2413 0.5626 — — — 372.8710 — 5 0.00005363 0.6005 0.7471 — — — — -387.729 (Shaded row indicates preferred model.) Exhibit 5-24 Entering–Circulating Models Exhibit 5-25 Exiting–Circulating Models Model No. Multiplier a0 Entering AADT a1 Approach Half-Width (ft.) b1 1 0.0034961 0.6036 — 2 0.0057838 0.4613 0.0301 (Shaded row indicates preferred model.)

Roundabouts: An Informational Guide Page 5-32 Chapter 5/Safety from a safety perspective, diagnostic procedures can then be used to isolate any problems and to develop corrective measures. The application of the EB method at the approach level would be identical to the procedure presented and illustrated earlier for the intersection level in Section 5.4. The models to be used would be those indicated by the shaded rows of Exhibit 5-24, Exhibit 5-25, and Exhibit 5-26. 5.5.2 CONSIDERATION OF APPROACH-LEVEL MODEL RESULTS FOR HSM-TYPE APPLICATION The HSM documents an accident prediction algorithm that enables the number of total intersection-related accidents per year to be estimated using Equation 5-7: where Nint = predicted number of total intersection-related crashes per year after application of accident modification factors; Nb = predicted number of total intersection-related crashes per year for base conditions; and AMFi = accident modification factors (AMF) (i = 1 to n) for various intersec- tion features different from base conditions. Each AMF is adjusted for observed features different from base conditions using Equation 5-8: where AMFi = accident modification factors (AMF) (i = 1 to n) for various intersec- tion features different from base conditions, AMFbase = AMF computed for the base condition value (see Exhibit 5-27), x = observed value for the variable, and xbase = base condition value for the variable (see Exhibit 5-27). For the HSM, base condition models and AMFs are provided for conventional stop- and signal-controlled intersections. A panel of experts selected the AMFs after a review of relevant research findings, including calibrated prediction mod- els, the estimated coefficients of geometric variables in these models, and the results of before–after safety evaluation studies. A similar methodology can be considered for roundabouts at the approach level. For this potential application, Model 1 (with AADT as the only variable) in Exhibit 5-24 is considered as the base model. And, as noted above, the estimated coefficients for geometric features in the approach-level models can be considered in developing AMFs. The AMFs directly related to geometric variables and base condition values for these variables are shown in Exhibit 5-27. Using the above equation, the effect of a design change can be identified by applying the appropriate AMF. However, caution is advised because many of the variables are correlated, resulting in model-implied effects that may not reflect AMF AMFi base x xbase = −( ) N N AMF AMF AMFb nint = × × ×( )1 2 . . .Equation 5-7 Equation 5-8

Chapter 5/Safety Page 5-33 Roundabouts: An Informational Guide Exhibit 5-27 Base Conditions for Design Variables and AMFs Implied for Unit Change in Variables Variable Base Condition Value Entering– Circulating AMF Exiting– Circulating AMF Approach AMF Entry Radius 76 ft 1.010 Entry Width 20 ft 1.052 Approach Half Width 18 ft 1.031 Inscribed Circle Diameter 134 ft 1.022 Central Island Diameter 69 ft 0.992 1.014 Circulating Width 23 ft 1.117 Angle To Next Leg 93 deg 0.973 reality. The correlation matrix provided as Table 3-14 of the NCHRP Report 572 (2) should therefore be considered before vetting these AMFs for formal application. See Exhibit 5-28 for an example of the calculation of entering–circulating crashes using AMFs. Exhibit 5-28 Calculation of Expected Frequency of Entering– Circulating Crashes Using AMFs

Roundabouts: An Informational Guide Page 5-34 Chapter 5/Safety 5.6 REFERENCES 1. Maycock, G. and Hall R. D. Crashes at Four-Arm Roundabouts. TRRL Laboratory Report LR 1120. Transport and Road Research Laboratory, Crowthorne, England, 1984. 2. Rodegerdts, L., M. Blogg, E. Wemple, E. Myers, M. Kyte, M. Dixon, G. List, A. Flannery, R. Troutbeck, W. Brilon, N. Wu, B. Persaud, C. Lyon, D. Harkey, and D. Carter. NCHRP Report 572: Roundabouts in the United States. Transportation Research Board of the National Academies, Washington, D.C., 2007. 3. Brilon, W. and B. Stuwe. “Capacity and Design of Traffic Circles in Germany.” In Transportation Research Record 1398. TRB, National Research Council, Washington, D.C., 1993. 4. Schoon, C. C. and J. van Minnen. “Accidents on Roundabouts: II. Second Study into the Road Hazard Presented by Roundabouts, Particularly with Regard to Cyclists and Moped Riders.” R-93-16. SWOV Institute for Road Safety Research in the Netherlands, 1993. 5. Alphand, F., U. Noelle, and B. Guichet. “Roundabouts and Road Safety: State of the Art in France.” In Intersections without Traffic Signals II (W. Brilon, ed.), Springer- Verlag, Germany, 1991, pp. 107–125. 6. Brown, M. TRL State of the Art Review: The Design of Roundabouts. London, HMSO, 1995. 7. Garder, P. The Modern Roundabouts: The Sensible Alternative for Maine. Maine Department of Transportation, Bureau of Planning, Research and Community Services, Transportation Research Division, 1998. 8. Guichet, B. “Roundabouts in France: Development, Safety, Design, and Capacity.” In Proceedings of the Third International Symposium on Intersections with- out Traffic Signals (M. Kyte, ed.), Portland, Oregon, University of Idaho, Moscow, Idaho, 1997. 9. Mandavilli, S., A. McCartt, and R. Retting. Crash Patterns and Potential Engineering Countermeasures at Maryland Roundabouts. Insurance Institute for Highway Safety, Arlington, Virginia, May 2008. 10. “Safety of Roundabouts in Urban and Suburban Areas.” Centre d’Etude des Transports Urbains (CETUR), Paris, 1992. 11. Arndt, O., “Road Design Incorporating Three Fundamental Safety Parameters.” Technology Transfer Forum 5 & 6, Transport Technology Division, Main Roads Department, Queensland, Australia, August 1998. 12. Bared, J. G. and K. Kennedy. “Safety Impacts of Modern Roundabouts.” In ITE Safety Toolbox, Institute of Transportation Engineers, 1999. 13. Leaf, W. A. and D. F. Preusser. Literature Review on Vehicle Travel Speeds and Pedestrian Injuries. Final Report DOT HS 809 021. National Highway Traffic Safety Administration, Department of Transportation, Washington, D.C., October 1999.

Chapter 5/Safety Page 5-35 Roundabouts: An Informational Guide 14. Crown, B. “An Introduction to Some Basic Principles of U.K. Roundabouts Design.” Presented at the ITE District 6 Conference on Roundabouts, Loveland, Colorado, October 1998. 15. Seim, K. “Use, Design and Safety of Small Roundabouts in Norway.” In Intersections without Traffic Signals II (W. Brilon, ed.), Springer-Verlag, Germany, 1991, pp. 270–281. 16. Van Minnen, J. “Safety of Bicyclists on Roundabouts Deserves Special Attention.” Research Activities 5, SWOV Institute of Road Safety Research in the Netherlands, March 1996. 17. FHWA. SafetyAnalyst. www.safetyanalyst.org/. Accessed August 2009. 18. Council, F., E. Zaloshnja, T. Miller, and B. Persaud. Crash Cost Estimates by Maximum Police-Reported Injury Severity within Selected Crash Geometries. Report No. FHWA-HRT-05-051. FHWA, Washington, D.C., October 2005.