Synthesis of Literature Relevant to Roundabout Signalization to Provide Pedestrian Access - Final Report

SIGNALIZATION OF ROUNDABOUTS

United States

Signalization of roundabouts is discouraged in the United States. The Federal Highway Administration’s roundabout informational guide (7) states “roundabouts should never be planned for metering or signalization.” However, the guide does concede that “unexpected demand” may require signalization after a roundabout is constructed. The FHWA guide goes on to describe three signalization alternatives to be considered should unexpected demand suggest the need for signals: (1) metering, (2) nearby pedestrian signals, and (3) full signalization of the circulatory roadway. It should be noted that the FHWA guide for Roundabouts was prepared at a time when the US had very little experience with modern roundabouts, and thus relied heavily on the experiences observed in other countries…primarily Europe.

Metering is the installation of a signal at entry yield lines. The principle and operation is the same as for meters that many jurisdictions use to meter freeway on-ramps. That is, where entry flows are high, meters are used to create gaps downstream. Thus, metering facilitates entry of vehicles downstream of an entry that has a high flow. The FHWA informational guide recommends a two-lens, yellow and red, signal head for metering. For example, a sign below the signal would read “stop on red”. A yield control would be in effect whether or not the meter was in use (i.e., the lenses were on).

Such metering may not benefit pedestrians because the meters are located downstream of entry crosswalks. Meters would have little or no effect on gaps at exit crossings. Meters might have the effect of slowing vehicles down, which some investigators believe makes yielding more likely (5). However, because the meters are downstream of entry crosswalks, they could cause traffic to backup into the crosswalks and thus obstruct crossings. Because metering is used when traffic volume is high, naturally occurring gaps in which vehicles do not arrive would be infrequent, and the noise of vehicles arriving and departing the crosswalk would be high.

The FHWA guide suggests that pedestrian signals placed upstream of roundabout entries and downstream of roundabout exits could be used to meter roundabouts that are experiencing operational problems because of high traffic volume. It suggests that these crossings need to be 65 to 165 ft from the yield line to avoid queues of exiting vehicles backing up from the crosswalk signal into the circulatory roadway. It also recommends careful analysis of each leg to determine a weighted aggregate performance measure of vehicle capacity. Although sighted and blind pedestrians might benefit from signalized crossings that are tailored to accomplish vehicle flow metering, such solutions cannot be trusted to provide access at all roundabouts, because unexpected demand, which would be the warrant for such signals, is intended to be the exception rather than the rule. Furthermore, if the purpose of the “pedestrian signals” is to meter traffic, then it is unlikely that signals will meet pedestrian needs. For instance, the placement of the crossing 165 ft from the intersection is intended to store vehicle queues that might otherwise back up into the roundabout and obstruct vehicle operations. Where pedestrian arrivals are infrequent, and crossing distances are short, a 165 ft setback might exceed the need for vehicle queue storage while inconveniencing pedestrians.

The third signalization alternative mentioned in the FHWA guide, full signalization of the circulatory roadway, is beyond the scope of the present literature review. According to the guide, fully signalized roundabouts must meet geometric design criteria that are very different from recommended designs for unsignalized roundabouts. Thus, most modern roundabouts in the U.S. could not easily be converted to fully signalized roundabouts. Furthermore, full signalization is required only where traffic demand far exceeds the demand at locations where modern roundabouts would be considered for new construction. According to the FHWA guide, a primary requirement for fully signalized roundabouts, is the capacity to store stopped vehicles in the circulatory roadway without creating queues that spill back to block upstream exits or entries. Unsignalized roundabouts that are built with diameters and roadway widths recommended by the guide could not be fully signalized without increases in diameter, lane width, and/or additional lanes.

In addition to the three warrants for signalization suggested in the roundabout information guide, the guide also points out that signals may be used where “disabled pedestrians and/or school children are present at high volume.” Such a warrant may not be compliant with existing civil rights law, which does not set high demand as the bar for accessibility, but rather requires usability by individuals. Nonetheless, the FHWA guide does recognize that signals may be warranted to accommodate pedestrians.

In the U.S., experience with pedestrian traffic signals at roundabout crosswalks has been limited. The team conducting NCHRP 3-78 (8) identified four signalized roundabout crosswalks in the U.S., one of which has been removed, and another one of which was not yet in operation when this report was prepared (9). The current installation at the University of Utah and the now dismantled installation in Clearwater Florida both follow typical mid-block crossing traffic signal practice. At both roundabouts, standard three-lens (i.e., green-amber-read) signals were used, and traffic in both directions of travel stopped for the entire duration of the pedestrian crossing phase. Experience at both locations was similar: the pedestrian phase backed traffic into the circulatory roadway during peak traffic periods. This was true, even though in the Utah case the crosswalk is 150 ft from the circular roadway. The two other signalized roundabout crosswalks identified by the NCHRP team were a location in North Carolina that is under construction and a location in Alpine City Utah where there are pedestrian activated amber warning flashers.

In summary, signalization of roundabout pedestrian crossings in the U.S. has been discouraged. Extant U.S. roundabout guidelines have not addressed roundabout pedestrian operational effectiveness and have not addressed the needs of all non-vehicle users. However, there is some recognition that some type of signalization might benefit both traffic and pedestrian operations under some unexpected circumstances. At the two locations where three-lens signals were installed, traffic operations were adversely affected. As will be seen, these adverse effects were predictable based on experience in the United Kingdom, and countermeasures to avoid these effects have proven successful in that country.

International

Signalization of roundabouts was reviewed for the following countries:

• United Kingdom

• France

• Australia (Victoria)

• The Netherlands

• Sweden

None of the reviews was exhaustive, but they did include personal communications for several of the countries and review of available English language documentation.

 United Kingdom (U.K.)

The U.K. has a variety of pedestrian crossing types. The most common crossing types that give priority to the pedestrian are the zebra, pelican, and puffin. All of these crossing types may be installed at roundabouts. Crossings that do not give priority to pedestrians do not have traffic signals and do not use the longitudinal (zebra) pavement markings. Davies provides a thorough review of pedestrian crossing types used in the U.K. (10)

The zebra crossing is the oldest mid-block crossing type currently in use that gives priority to pedestrians. The important features of a zebra crossing can be seen in Figure 2. These consist of wide longitudinal strips to delineate the crossing, zigzag striping at the roadway edges and between lanes to indicate that parking and passing are not permitted, and black and white striped poles topped by flashing amber globes (Belisha beacons). These crossings may have other features such as a raised crosswalk, bulb-out, median refuge, and bollards. Pedestrians in a zebra crossing have priority over vehicles, but unlike in the U.S. and some other countries, U.K. motorists usually respect that priority and often stop for pedestrians who are waiting at the curb (and thus do not yet have priority) (10). Pedestrians may claim priority by standing in the crossing, which is different from most U.S. states where pedestrians are expected to wait for a gap before entering the roadway at uncontrolled crosswalks. Once a pedestrian claims priority in a zebra crossing, all vehicles that can do so are required to stop before entering the crossing.

The pelican (PEdestrian LIght Controlled) crossing is reported to be the most common signaled-controlled crossing type in the U.K. (10). These crossings are similar to pedestrian-actuated signals that are used in the U.S., but there are several important differences between U.K. and U.S. signalized crossings. The most significant difference regards policy rather than design or operation. The pedestrian signal head is similar to that used in the U.S., having a green walking-man to indicate the walk phase, a flashing green walking-man to signal the end of the walk phase, and a red standing-man to indicate the no crossing phase. However, these phases are not regulatory – there is no offense for pedestrian violations of the do-not-start and do-not-cross phases. Williams (11) describes pedestrians who enter to crossing at the beginning of the flashing amber (standing-man) phase as “sheltered” by pedestrians already in the crosswalk. Three-color (red, amber, and green) signal heads are used to control vehicles at pelican crossings. The red phase is followed by a flashing amber phase that permits vehicles to proceed if pedestrians have cleared the crossing. Pelicans are both vehicle and pedestrian activated. Pedestrians press a button to request a walk phase. Sensors in the roadway hold the walk phase request until a gap in traffic is detected, or until the signal times out. As is typical at U.S. signal controlled crossings, pedestrian signal heads at pelican crossings are located on the far side of the crossing.

Puffin (Pedestrian User Friendly INtelligent) crossings are similar to Pelican crossings, but employ sensors to detect when pedestrians have cleared the crossing. This enables both shortening of the delay for traffic when pedestrians clear the crossing quickly, and extension of the walk phase when pedestrians need more time. Another difference between pelican and puffin crossings is that puffin crossings have the pedestrian signal head above the call button, which is on the near side of the crossing. Figure 3 shows a pedestrian signal call unit used in Dorset County, U. K. The pedestrian in the figure has his fingers on a rotating tactile cone that is intended to indicate the crossing phase to pedestrians who are blind, or deaf-blind. Because Puffin crossing visual indications are on the same side of the crossing as the pedestrian, pedestrians with low vision can often see these indications when they may not be able to detect or interpret far side signal indications found at Pelican crossings. (12) Puffin crossings may also have microwave, infrared, or pressure pad, detectors to sense pedestrians waiting to cross, and obviate the need for pressing a call button.

Roundabouts in the U.K. may have any of the mid-block crossing types described here. Other types, Pegasus to accommodate equestrians, and Toucan, which combines bicycle and pedestrian crossings are not described here because the added capabilities they provide do not appear to affect accessibility for pedestrians.

Guidance on the design of U. K. roundabout pedestrian facilities is flexible, and suggests that the type of facility should depend on expected volumes of traffic and pedestrians (13). Unmarked crossings (defined by curb ramps), zebra crossings, signal controlled crossings, and grade separated crossings (i.e., pedestrian overpass or underpass) are all options to be considered. Where at-grade crossings are included, they are to be placed away from flared entries or exits. Figure 4 shows an example of a flared entry and exit design. The flared design is intended to increase roundabout vehicle capacity. The flared design may be compared to the design without flare shown in Figure 1. Placing the crossings beyond the flares minimizes the length of the crossings itself, but may increase the distance pedestrians need to travel on the sidewalks. For usability to pedestrians with visual impairments, crosswalks away from the flared portion of the entry and exit would have at least two potential benefits. First, where a double-lane roundabout is at the intersection of otherwise single lane roads, the crossing would be across single lanes, which eliminates the risk of being struck by a vehicle that passes a yielded vehicle. Second, beyond the flare, the roadway curbs are usually parallel, and thus curb cuts for ramps would provide a reliable cue for aligning with the crosswalk. One disadvantages for pedestrians would be the additional distance that needs to be traveled to reach the crossing. Another disadvantage is that pedestrians who are not familiar with the intersection might have difficulty in orienting to the indirect path. Figure 4 illustrates the difference in travel distance between the typical U.S. design and the U.K. practice. The red line that shows the pedestrian track for the typical U.S. design can be seen to be considerably shorter than the green line that shows the path in the U.K. design.

Figure 4 also illustrates another recommended U.K. practice, particularly for signalized roundabout crossings. That practice is the offset of the crossings for the two directions of travel (14). On roundabout exit lanes, queued vehicles that are stopped for pedestrians may “block back” into the circulatory roadway (15). The block back can drastically reduce roundabout capacity, as has been observed at the two U.S. roundabouts where crossings were signalized. Moving the exit crossing 50 m (164 ft) from the edge of the circular roadway reduces the chances that the vehicle queue will block back into the circular roadway. When signals are installed, the length of the red phase can be reduced by more than half by signaling the two halves of the crossing separately. The shortened red phases also minimize the chance of block back. To prevent pedestrians from continuing across the splitter island and into the lane that is not stopped, guardrails are installed to force pedestrians to use the offset crosswalk. Although, audible APS devices may be installed at these crossings, it would seem that tactile APS would avoid confusion between the separate indicators at each offset crosswalk. Because the crossing distances are short, and the crossings are perpendicular to the curbs, audible signals are probably not needed to provide directional guidance.

Personal communications with Barry Crown (16), who is currently a consultant, but previously conducted research for the Transportation Research Laboratory (TRL), provided information on the history of signalized crossings in the U.K. that were not readily available in the literature. According to Crown, signals are not installed at roundabouts to provide pedestrian access. Because U.K. drivers gave priority to pedestrians at the zebra crossings that were standard in urban roundabouts, the high volume of pedestrian traffic was causing roundabouts to lock up. Rather than to provide pedestrian access, signals were installed at roundabout pedestrian crossings to improve vehicle operations. When signals were first considered, there was fear that standard threelens signals (i.e., red, amber, and green lenses) might be problematic on approach legs. The concern was that because the signal would be relatively close to the approach yield line, motorists might misinterpret the green ball as overriding the yield control. In practice, this fear was not realized. Crown believes that because U.K. drivers were accustomed to the priority of circulating vehicles at roundabouts, the addition of signals only 7 to 10 meters upstream of the yield line did not confuse drivers. However, Crown was not confident that 3-lens signals would work in the U.S., because roundabouts do not have the long history here that they had in the U.K. before signals were introduced. He suggests a two-lens (i.e., red and amber) signal head might be more appropriate for signalized roundabout crossings in the U.S.

Recently, Transport for London Street Management published the results of a before after safety analysis of roundabout signalization (17). Overall, it was found that signalization of “standard” roundabouts resulted in a 28 percent decrease in crashes, which was statistically reliable at the 0.01 level. Like their U.S. counterparts, standard roundabouts give priority to circulating traffic, and have deflection at entry. However, standard roundabouts may have larger diameters than typical U.S. roundabouts, so that signalization would typically include placement of signals in the circulatory roadway and at entries. The London study included 36 months of observation in both the before and after periods. The largest crash reductions were for bicycle involvement (58 percent) and for merge collisions (58%). The reduction in pedestrian involved crashes (23 percent) was not statistically reliable. However, the baseline number of pedestrian-involved crashes over three years was already quite small (22 crashes). The U.K. crash reduction result may not be applicable to the U.S., in part because the roundabouts that were signalized tended to be larger than the “modern” roundabouts that are typical in the U.S. Nonetheless, the U.K. experience should refute the argument that signalization of roundabouts per se will result in decreased safety, or that signalization per se will put pedestrians at greater risk. The available empirical evidence suggests that signalization may enhance the safety of roundabouts and, does not put pedestrians at greater risk.

France

There are now more than 25,000 roundabouts in France. The French policy is to favor roundabouts over signalized intersections and to replace signalized intersections with roundabouts where that is feasible (18). The French guidelines for the design of urban intersections allow for signalization of roundabouts for pedestrian safety, but do not recommend signalization for pedestrian safety. Where a need for signalized crosswalks is identified, guidance on implementation is provided. However, guidance is not provided on how the safety need is to be established. Recent efforts at roundabout signalization have focused on integrating roundabout operations with at-grade train crossings that are in or near the roundabout. Where roundabouts have been signalized to accommodate traffic or train operations, crosswalks have been moved, in some cases, to be adjacent to the circular roadway, as can be seen in Figure 5. Although such a configuration would probably be preferred by pedestrians who use traffic surges and movements to identify the walk phase and crosswalk alignment, this configuration would not be feasible at typical U.S. roundabouts, which do not have space for storing stopped vehicles within the circulatory roadway.

Note that the traffic signals shown in Figure 5 are mounted on posts at the roadway edges, not on mast arms. Because of the low speed environment, it may not be necessary to locate roundabout traffic signals over the travel lanes to achieve adequate conspicuity.

The new French guidelines for urban intersections, to be published in 2006 (19), indicate that roundabouts may be signalized for pedestrian safety under the following conditions (20):

• There is a splitter island that provides adequate pedestrian refuge between the two directions of vehicle travel.

• The crosswalks are at least 49 ft from the circular roadway.

• The two crosswalks are offset (same concept as illustrated in Figure 4, above).

• The new French guideline also specifies that if the pedestrian crossings are signalized, release of vehicles should be immediate upon the pedestrian(s) clearing the crosswalk. The guideline does not specify how the clearing may be detected or the immediate release may be accomplished.

Australia

Baranowski has reported a roundabout in Sydney, Australia, that was designed to interrupt pedestrian flows that were anticipated to be heavy during the 2004 Olympic Games (14). Figure 6 shows one of the crossings at that roundabout. The crosswalks on either side of the splitter island are not offset, as recommended in the U.K. and France. According to Baranowski, the two-lens (amber and red) signals first flash amber, then goes to solid amber, and then to red. When not active, the signals are unlit (off). At the roundabout entrance, this avoids the confusion drivers might experience, if they received a green signal only a few feet upstream of the yield control. We have requested further information on the signal timing from the Australian contacts, as well as whether the configuration in the photograph is according to a VicRoads standard, or if it is experimental.

The signals are post-mounted. Two of the signals are mounted on posts at the stop bar, and two additional signals are mounted at the upstream side of the crosswalk. The two signals at the stop bar would emphasize the desired stop location, whereas the two signals at the crosswalk are far enough from the stop bar so that drivers can see them when stopped at the stop bar. Note, however, that the vehicle in the picture appears to have stopped upstream from the stop bar, perhaps so that the driver could maintain line of sight to all four signals. If drivers stop or slow well upstream of the stop bar, they may adversely affect roundabout operations, particularly for roundabout exit crossings where stopped vehicles would be likely to queue back onto the circulatory roadway.

In the U.S., the MUTCD requires that traffic signals be placed 40 ft beyond the stop bar to ensure that drivers who are stopped at the stop bar can see them. Signal heads at the stop bar, as shown in Figure 6, would be permissible in the U.S., as long as there are also yoked signals 40 ft beyond. It appears from the figure that the VicRoads standard may be similar to that in the MUTCD, as the two pairs of signals appear to be about 40 ft apart.

The Netherlands

Roundabout crosswalks in the Netherlands are not signalized. According to Martijn de Leeuw, a traffic engineer for the Province of South-Holland, the policy for intersections there is as follows (21):

• Low volume intersections are normally single-lane roundabouts.

• If a single-lane roundabout does not provide sufficient capacity, then a turbo-roundabout is considered.

• If a turbo-roundabout does not provide sufficient capacity then a signalized intersection is considered.

• If a signalized intersection will not provide sufficient capacity, then a signalized roundabout is considered.

Signalized roundabouts are only constructed when there is sufficient space to store vehicles in the circulatory roadway.

According to Mr. de Leeuw, “We have no problems with persons with visual impairment and the turbo-roundabout. The speeds are low.” However, in urban areas speed are low everywhere in comparison to the U.S. The urban speed limit is 30 km/h (18 mi/h). The Netherlands, and Europe in general, does not have civil rights laws that protect persons with disabilities and independent travel by the disabled is not a legal objective.

Turbo-roundabouts are used in locations where double-lane roundabouts might be used in the U.S. (22). Figure 7 illustrates a typical turbo-roundabout design. This design addresses the problem of path overlap, which is described in Chapter 6 of Roundabouts: An Informational Guide (7), by using mountable curbs between lanes, and using a portion of the central island for “left-turn” lanes. This design uses geometric barriers to force drivers to select the proper lane before entering the roundabout, and to stay in that lane until they exit. Figure 7 shows typical bicycle crossings on the west and south legs of the roundabout. These crossings employ a jog in the bicycle path on the splitter island that is intended to force bicyclists to go slow and encourage them to yield to oncoming traffic. Figure 8 shows a jog in a bicycle path on a turbo-roundabout splitter island. Pedestrian crossings at turbo-roundabouts do not employ the jog in the path because it is assumed that pedestrians travel slower than bicyclists do and therefore have sufficient time to recognize vehicles to which they should yield.

At-grade pedestrian and bicycle crossings at turbo-roundabouts are used when pedestrian and bicyclist volumes are low enough that vehicle traffic flow will not be disrupted. At turbo-roundabouts, which are typically used in rural locations, pedestrians and bicyclists are required to yield to vehicles. Figure 9 shows typical pedestrian and bicycle crossings at the exit of a turbo-roundabout. In urban environments, at-grade pedestrian crossings are usually zebra crossings, which, as in the U.K., give priority to the pedestrian. Where pedestrian and/or bicycle volumes are high, grade separation is the preferred solution to conflicts between pedestrian/bicycle and vehicle traffic flow at roundabouts (23).

Sweden

We have requested additional information on Swedish policy and practice. At the time of this writing the available information is somewhat dated. The present information comes largely from a 1999 FHWA report on pedestrian safety in Sweden (24). Roundabouts are viewed as a safety enhancement for pedestrians because they serve to slow traffic and make driver yielding more likely. Nonetheless, studies of driver yielding behavior at roundabout zebra crossings suggest that, as in the U.S., most drivers do not yield to pedestrians (24).

Christer Hydén of the Lund Institute of Technology reports that there are very few signalized roundabouts in Sweden, with most of those in Stockholm. In those cases, only one leg is signalized. (25)

Per Garder of the University of Maine reports that the signalized roundabout crosswalks in Sweden flash amber until a pedestrian presses the button to call for a pedestrian crossing phase (26). Until the call button is pressed, the pedestrian signal head is either off, or indicates that a crossing is prohibited. When the button is pressed, two solid amber lenses are displayed to traffic. According to Garder, two solid ambers are required to call motorist attention to the end of the flashing amber phase. A walk phases is indicated to the pedestrian when the solid amber phase ends and a red signal lens is displayed to motorists.