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

GENERAL CONSIDERATIONS FOR TRAFFIC SIGNAL CONTROLLED ROUNDABOUT PEDESTRIAN CROSSINGS

Policy on Traffic Signals for Roundabout Crosswalks

Traffic signals at pedestrian crosswalks can be used in most of the jurisdictions surveyed in this report. Only in the Netherlands is there a policy against signalizing roundabout crosswalks except for fully signalized roundabouts (i.e., large roundabouts that have traffic signals within the circulatory roadway). Because the final decision on crosswalk signalization is left to Dutch provincial authorities, even in the Netherlands we cannot say with certainty that no roundabout crosswalks are signalized.

Among the jurisdictions surveyed, only the soon to be published French guideline suggests signalization may be considered for roundabout crosswalks to provide for pedestrian safety. Even so, the French guideline suggests that signals are not likely to make pedestrians safer, and that they might have the effect of making pedestrians overconfident and thus careless.

With the exception of France, the sole policy rationale this review identified for consideration of signalizing roundabout crosswalks is to improve traffic operations. No jurisdiction was identified that recommends traffic signals to improve roundabout access for pedestrians. The two operational rationales offered for signalization are: (1) to meter entering traffic when the volume is so high as to obstruct downstream entrances (e.g., the U.S.), or (2) to meter pedestrians so that traffic flow can be maintained (e.g., the U.K.). However, traffic signal Warrant 4 in the MUTCD (27) might be applied to specific roundabout crosswalks. That warrant requires that a traffic signal with pedestrian signal heads “shall be considered” if both of the following conditions apply: (1) the number of pedestrian crossings on an average day is at least 100 pedestrians in each of any 4 hours, or more than 190 pedestrians cross in any hour; and (2) there are fewer than 60 crossable gaps per hour during the peak times used to meet the pedestrian volume criterion.

The Effect of Roundabout Crosswalk Signalization on Vehicle Traffic Operations

In the U.S., in both cases where experience has been gained with signalization of roundabout crosswalks, the result was disruption of traffic operations, not enhancement. However, in both cases, the installations stopped both entrance and exit sides of the signalized leg at the same time. The result was the need for relatively long pedestrian crossing phases, and in turn, much longer vehicle queues than would have resulted if U.K. practice had been followed. In the U.K., exit and entrance crossing phases are separated in both time and space. The separation in time allows for short pedestrian crossing phases that result in shorter vehicle queues. For instance, consider a double-lane roundabout where there are four 12 ft wide lanes at the crossing, and the median is 12 ft wide. Assuming a minimum pedestrian phase for the entire crossing, the vehicles will have a red indication for (4 lanes + 1 splitter) × 12 ft ÷ 3.5 fps = 17 s. Whereas, if the U.K. guidance were followed the minimum red indication would be 2 lanes × 12 ft ÷ 3.5 fps = 7 s. Thus, if one vehicle per lane arrives every 3 seconds, in the former case 6 vehicles will queue up in each lane whereas in the latter case 2 vehicles will queue up in each lane.

The U.K. guidance to separate the exit and entrance crossings in space was originally intended to ensure that the pedestrians face into the oncoming traffic before crossing and to prevent pedestrians from continuing across the splitter island without checking traffic (16). As originally conceived, this would have located the exit crossing closer to the circulatory roadway than the entrance crossing. However, for operational reasons, the original configuration was reversed and the exit crossing was moved further from the intersection than the entrance crossing. Although this causes pedestrian to turn away from oncoming traffic when traversing the splitter, this design still achieved the goal of forcing pedestrians to make the crossing in two time-separated stages. Experience has shown that this configuration has not resulted in an increase in pedestrians stepping into the roadway without looking (16).

To allow traffic to queue up at the crossing without backing up into the circulatory roadway, the U.K. guidance suggests that the exit crosswalk be located about 164 ft from the edge of the inscribed circle. Such a location would allow for 4 to 8 passenger vehicles per lane to queue up before the queue would extend into the circular roadway. This amount of storage space may be appropriate where there is a high volume of pedestrians, as is the case where roundabout crossings are signalized in the U.K. However, at typical U.S. roundabouts the number of pedestrians crossing in a single phase is small (e.g., fewer than 10). Where storage requirements are less, the offset of the exit crossing could be lower, perhaps no more than 50 ft, which would allow a queue of about 2 to 3 passenger vehicles per lane. This contrasts with current-typical U.S. practice of locating the crosswalk 20 ft. from the circulatory roadway, which allows for storage of no more than one vehicle per lane.

Offset Crosswalk Considerations

One objection to moving the exit crosswalks 30 or more feet farther from the intersection is that pedestrians would be unlikely to walk the additional distance and might therefore jaywalk. In the U.K. jaywalking itself is not recognized as a traffic offense. So to some extent, if the jaywalking does not disrupt traffic, it is not a concern. However, jaywalking can disrupt traffic at times, and if a vehicle strikes a pedestrian, the disruption to traffic, as well as the life of the pedestrian, could be extreme. To limit jaywalking where crosswalks are offset, fences are often installed on the splitter island and on the sidewalks on both sides of the street. The fences steer pedestrians to the crosswalk. Slats in the fences are angled such that they do not obstruct the line of sight between drivers and pedestrians.

Another objection to moving the crosswalks further from the inscribed circle is that it increases pedestrian travel time. If all other things are equal, and a 3.5 fps walking speed is assumed, then moving the crosswalk 30 ft farther from the inscribed circle would add 8.6 s to the pedestrian travel time, or about 17 s if the pedestrian continues down the same street on the opposite side of the intersection. The U.K recommended 164 ft offset would add 47 s to a pedestrian travel time to the crosswalk, or 1.5 minutes if it is assumed that the pedestrian would then continue walking down the same street on the opposite side of the intersection. At exits where the crosswalk is not signalized, vehicle speed is likely to increase with increasing distance from the circular roadway. For instance, a vehicle is traveling 20 mi/h at the inscribed circle and accelerates according to the profile suggested by AASHTO design policy (28), then 164 ft from the inscribed circle the vehicle can be expected to be traveling at about 32 mi/h. At signalized crossings where vehicle and pedestrian behavior should be controlled by the signal, the speed increase that results from the increased travel distance may not be a major concern unless compliance with the signal is problematic.

Adding a travel distance of almost 300 ft and 1.5 minutes to travel time might be considered unreasonable. However, it should be noted that the offset is recommended when traffic signals are installed, and traffic signals themselves can add to pedestrian delay because the pedestrian must wait for a crossing call to initiate the walk phase.

Where pedestrian traffic is heavy, the crossing phase must be longer to accommodate all the pedestrians who need to cross. The longer the crossing phase, the greater the probability that the traffic queue will disrupt the operation of the roundabout and reduce intersection capacity. Thus, where pedestrian traffic is high, the exit crosswalks need to be farther from the inscribed circle, perhaps as far as the 164 ft recommendation in the U.K. guideline. The farther the exit crossing is from the circular roadway, the greater the number, and the longer, pedestrian walk phases that can be accommodated. Depending on the tradeoffs that are made between walk phase frequencies, walk phase duration, vehicle delay, and pedestrian travel distance, the longer crossing offsets might result in longer, shorter, or no change in pedestrian delay relative to keeping the crossing closer to the inscribed circle.

Traffic signals can reduce pedestrian delay under certain conditions:

• When the arrival of vehicles would otherwise be constant and drivers will not yield.

• When a pedestrian who relies on non-visual cues cannot detect available gaps without the assistance of an accessible traffic signal.

Inman et al. (29) reported that at a double-lane roundabout exit where the flow was about 800 vehicles per hour, it took a mean of 63 seconds before vehicles yielded in both lanes (range 6 to 243 s). Blind pedestrians took an average of 3 seconds to recognize that both lanes were blocked (range 0 to 27 s). Thus, without a traffic signal the average delay for blind pedestrians in that study was 66 s, exclusive of the time it may have taken those pedestrians to locate the crosswalk and orient to the intersection. Thus, with or without a traffic signal, and with or without added crosswalk offsets, some pedestrians may experience delay at roundabout crosswalks, particularly during peak traffic periods. Whether the installation of traffic signals with crosswalk offsets similar to those used in the U.K. would increase or decrease pedestrian delay would depend on a number of factors that include:

• The willingness of drivers to yield in the absence of signals.

• The volume of vehicle traffic at the intersection.

• How long it takes the pedestrian to accept an available gap, either from yielding vehicles or detection of a gap between arrivals.

• The delay between the time a pedestrian phase is called (e.g., the pedestrian pushes a button) and the onset of the walk phase.

• The walking speed of the pedestrian.

• The distance of the crosswalk from the edge of the circulatory roadway.

Traffic Signal Configuration Considerations

Standard three-lens traffic signals that operate full time could be used at roundabout crosswalks. This three-lens signal is the only one that currently conforms to the MUTCD standard, which may be found in section 4D of that publication (27). If the roundabout is to continue to give priority to circulating traffic, i.e., retain yield at entry control, then great care must be taken in installing signals at nearby crosswalks. Standard three-lens signals have been used successfully at U.K. roundabout crossings. However, Barry Crown, a leading authority on U.K. roundabouts, has warned that the successful implementation in the U.K. may not be applicable to the U.S., because in most areas of the U.S., drivers are not familiar with the priority rule at roundabouts. U.K. drivers had many years of experience with the roundabout priority rule before the first pedestrian traffic signal controls were installed. Despite the initial fears of U.K. engineers, U.K drivers did not misinterpret the green indication at crosswalks as overriding the yield control at roundabout entries. Such might not be the case in the U.S. where drivers are often naive to the priority rule.

In an apparent recognition of the potential for driver misinterpretation of green signals at a roundabout crosswalk, engineers in Victoria, Australia, installed two-lens signals that omitted the green indication. The Australian two-lens signal is blank until called by a pedestrian. When a pedestrian presses the call button, the amber light flashes to get drivers’ attention, then burns steady amber to indicate that a red light is imminent.

The Australian approach appears to be logical, easy to understand, and in the context of a roundabout entrance, more appropriate than a standard three-lens signal. However, the MUTCD only provides for two-lens signals at freeway onramps, and there, red and green lenses are prescribed, not red and amber. Furthermore, the MUTCD standard does not allow for the use of part-time traffic control signals. A signal that is blank until called by a pedestrian is not MUTCD compliant. The MUTCD calls for traffic signals that are not needed at all times of day to be in flash mode, when a traffic signal is not required. Leaving the amber light in flash mode when no pedestrians have requested a crossing-phase might meet the MUTCD standard for not having visible inactive signals, but it is not clear whether a flash mode would make the traffic signal more effective when it is needed. It is plausible that a normally inactive signal that begins to flash would be more effective in attracting driver attention, and in attaining subsequent compliance, than a light that constantly flashes and on rare occasions goes to steady amber. However, the Swedish approach of following one flashing-amber with two steady ambers may be a viable alternative to inactive signals.

Placement of the traffic signal heads needs to be considered for effectiveness, driver comprehension, and cost. The MUTCD standard calls for a minimum of two signal faces. This could be accomplished by placing signal faces over each lane of a double-lane roundabout, or signal faces on poles on each side of the road. The MUTCD standard requires that the distance between the stop bar and at least one of the signals be a minimum of 40 ft. On roundabout approaches where the typical crosswalk is about 20 ft upstream for the yield line, this would place a traffic signal face quite close to the yield line unless the stop bar were placed well upstream of the crosswalk. Placement of a traffic signal at or near the yield line could cause driver misunderstanding of the yield requirement, even if a green indication is not present. Therefore, at a minimum, a signal face should not be easily visible to drivers who are at or near the yield line.

Figure 10 shows a possible pedestrian signal configuration for double-lane roundabout crossings. The configuration follows the U.K. practice of offsetting entry and exit crossings. It also employs barriers to ensure that pedestrians observe the offset, places at least one signal 40 ft from the stop bar, shows pedestrian call buttons on the upstream (relative to vehicle traffic) side of the crosswalk threshold, and locates additional signal faces at the stop bars to emphasize stop bar location. The figure, which is roughly to scale, shows the exit crosswalk set back 50 ft from the circulatory roadway. The stop bar at the exit crosswalk is set close to the crosswalk to maximize the usable vehicle storage capacity. The stop bar at the entrance is set back away from the crosswalk to enable placement of signal faces on the side of the road 40 ft from the stop bar. To move the stop bar closer to the crosswalk would require either (a) moving the crosswalk further from the circular roadway, or (b) installing an overhead signal that would extend above the circular roadway and might be visible from the yield line.

Figure 10. Hypothetical pedestrian crossing signal configuration for double-lane roundabout crosswalk.

The pedestrian crossing-interval call button, shown in Figure 10 on the side of the crosswalk thresholds closest to the approaching traffic, follows the U.K. and Dutch practice of placing the buttons so that the pedestrian faces oncoming traffic when the button is pressed. In those countries, puffin style displays are used above the call button. These displays show the walk and don’t walk symbols on the call button housing, as illustrated in Figure 3. A device on the call button housing would signal the walk interval using vibro-tactile methods. Because the crossing of the entrance and exit portions of a roundabout leg would require time separated crossing intervals, it would be necessary to make audible indications for each half of the crossing (exit and approach) easily distinguishable. Offsetting the crosswalks facilitates compliance with MUTCD (27) guidance to ensure that call buttons with tones are separated by at least 10 ft. Bentzen and Tabor (30) provide an extensive review of the technical issues and available hardware for accessible pedestrian crossing signals. Issues discussed include placement of the call button, the nature of the crossing-interval notification, and the usability of the devices for visually impaired persons. More recently, National Cooperative Highway Research Project 3-62 produced a synthesis report on accessible pedestrian signal best practices which can be found at http://www.walkinginfo.org/aps/ (31). The discussion here is intended to highlight some of the issues that need to be addressed to provide accessible pedestrian signals at roundabouts.

The Cost of Signalization

Roundabouts have been shown to provide substantial safety and operational benefits. However, their accessibility to visually impaired pedestrians has been shown to be problematic. This review has shown that if traffic signals were installed to provide access, the signals would not necessarily disrupt intersection operations, and in some cases – when pedestrian volumes are high, or traffic volumes are unbalanced across approaches—might improve operations. Nevertheless, installation of roundabout intersections is not the default choice in most U.S. jurisdictions, although several states have policies to encourage the consideration of roundabouts ahead of other intersection alternatives (e.g., New York, Virginia, and Kansas). Engineers usually have alternatives to roundabouts for solving intersection problems, and cost is a factor in design tradeoffs. Although cost may not be a factor in determining whether a new or altered intersection is designed to be accessible, it may be a factor in determining how accessibility is addressed. It is conceivable that double-lane roundabouts with accessible traffic signals would provide superior access relative to a traditional signalized intersection that is designed to carry the same volume of traffic movements.

The cost of installing mast arm signals at traditional intersection is about $150,000. The cost of such an installation at a four-legged roundabout would be approximately $600,000 under the assumption that each leg operates independently as a conservative estimate. The configuration shown in Figure 10, which does not require mast arms, would cost less. Figure 11 shows a solar powered unit that would be off when not activated by a pedestrian call and would flash amber, and then show steady amber, and then red when activated. All the signals at a crossing could be controlled through radio communication. The solar powered unit would have low installation cost because it would not require trenching to supply power or to control the remote signal faces. The manufacturer of the solar powered unit provided an estimate of $2,775 for a unit with two eight-inch lenses, solar panel, batteries, control logic, and radio. This estimate does not include the accessible pedestrian signal call unit or signal, mounting pole, or installation. A Virginia Department of Transportation manager estimated that the cost of installation of similar devices that do not require trenching is about $1,500.

Figure 11. Two-lens solar powered pedestrian crossing signal.

Table 1 provides a rough estimate of what it might cost to signalize a multilane roundabout to provide access to pedestrians with visual impairments. The estimate includes the typical cost of installing solar powered signals at the roadside, a two-lens solar-powered signal, and an accessible call button unit (32). The lower right cell of the table shows the estimated cost for installing signals as depicted in Figure 10 at four legs of a roundabout. The table also provides estimates for signalizing one, two, or three legs, and for providing fewer that four signal faces at each crossing segment. The MUTCD requires a minimum of two signal faces. Four faces were included in the hypothetical intersection shown in Figure 10 under the assumption that this might be important to ensure detection and compliance.

The estimate does not include the cost of: (1) installing barriers to prevent pedestrians from crossing at locations other than the crosswalk; (2) drilling through concrete or other work that might be required to install mounting poles in difficult locations; or (3) maintenance or operations. Because ADA standards govern only new constructions or alterations, the estimate does not include the cost to modify an existing roundabout, such as the cost of moving crosswalks and ramps.

Research is recommended to determine whether four faces are required to maximize performance (i.e., detection and driver compliance). It is possible that equal performance could obtain with only two faces, those labeled B and D in the figure. If two signal faces are sufficient, then the cost of installation would be about 30 percent less than with 4 faces.