Supporting Material to NCHRP Report 674 (2011)

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.

APPENDIX E: Treatment and Site Descriptions This Appendix contains descriptions of the data collection sites for NCHRP 3-78a and details on treatment installation at these sites. 66

Appendix E: Treatment and Site Descriptions This appendix serves to explain specifics of the four treatments installed at two channelized turn lanes (CTL) and a dual lane roundabout as a part of this research effort: sound strips, flashing beacons, raised crosswalk, and a Pedestrian Hybrid Beacon (PHB). The sites are also briefly described. Sound Strips Sound strips were considered as a low cost treatment that could provide audible cues about yields or available gaps in traffic. Past research conducted by Fitzpatrick et. al. noted that the treatment was not effective at providing the necessary yield information when using a three strip configuration sounding a “clack…..clack-clack.” The single strip was intended to provide the pedestrian with cue that a vehicle was present. The two follow-on sound strips provided a different sound cue which was intended to provide the pedestrian with information that the vehicle did not stop and that they should not cross. Our team hoped to improve on this initial test by providing a series of strips equally spaced starting much further back from the crosswalk, starting at approximately 300 feet before the crosswalk. In addition, the materials used will be much different, utilizing a hard rubber-like material instead of a PVC- based material like that tested earlier. Last, should the sound strip prove insufficient in providing available yield and gap cues, another test of the material was supplemented with flashing beacons is completed and document in a later section. Site Selection The CTL at the intersection of Providence Road and NC 51 (Pineville-Matthews Road) in Charlotte, NC was chosen as the test site for sound strip installation. This site was also used to study the push button activated flashing beacon, which allowed economical testing of both treatments under similar traffic conditions with a common set of participants. Staff from the Charlotte Department of Transportation (CDOT) was very supportive of the research effort and were willing to pay for and install both treatments. Speeds on all approaches were posted at 45 mph. Land uses in the vicinity of the site included a good mix of office buildings, retail, and residential. Volumes at the northwest corner crosswalk are slightly higher than the southeast corner with approximately 6,000 vpd in the CTL and 18,400 vpd in adjacent through traffic. The downstream conflicting through flow was 22,400 vpd and the opposing left turn 2,700 vpd. Low pedestrian activity was observed at the site on the order of 20 pedestrians per day, primarily during the midday off peak period. The north-west quadrant of the intersection was utilized for the sound strip installation. Sound strips were also installed as a package treatment with the push button activated flashing beacon in the south- east quadrant. This package treatment is discussed further in the next section. An aerial photo of both legs prior to treatment installation is shown in Exhibit 1. 67

Appendix E: Treatment and Site Descriptions Exhibit 1: Aerial Photo of Charlotte, NC CTL Site Additionally, lane delineators were installed to prevent late vehicle merges into the CTL. This allowed the pedestrian to utilize audible queues of yielding or available gaps from each of the “clacks” (or lack of clacks) to attempt to make judgments about when it is safe to cross. Field Setup The sound-strip treatment was intended to distinguish the auditory pattern of turning vehicles from mainline traffic, thereby facilitating gap and/or yield detection at the crosswalk. To test the audible cues generated by the rumble strips, a pilot test was done by the research team in a nearby parking lot. Many different configurations were tested. In the end, it was decided that a total of six sound strips would be installed at an even spacing of approximately 50'. With constant spacing, the temporal separation of sound cues ("clack" noises) is also constant at a consistent speed (approximately 35 mph yields a 1 second sound cue). However, as vehicles decelerate to a yield, the time between "clacks" increases, thereby giving pedestrians additional auditory information about driver intent. In addition, if no “clacks” are present, this hopefully provides supplemental information about the possibility of a crossable gap. Exhibit2 shows the installation of the six rumble strips along with the delineators at the northwest corner of the intersection. Ph ot o by G oo gl e 68

Appendix E: Treatment and Site Descriptions Exhibit 2: North-west Quadrant - View of Sound Strips and Lane Delineators The sounds strips tested a t the CTL were an of f-the-shelf rumble s trip used for temporary applications. The strips were raised approximately 0.25 inches and could be cut to the specified length necessary to cover the entire lane width. Exhibit 3 shows a vertical profile of the sound strip used for our test. This height was j ust h igh en ough t o produce audible c ues while s till a llowing safe t raversal of al l m odes of traffic (especially bicycles and motorcycles). Although many colors were available, the team chose white which was consistent with the pavement markings and suggested by team members familiar with MUTCD guidance. Exhibit 3: Close-Up View of Sound Strip Profile 69

Appendix E: Treatment and Site Descriptions Costs The Charlotte Department of Transportation purchased the sound strip material along with the delineators for less than $1000. This treatment is a low cost treatment with the majority of the costs associated with field installation and removal. With a longer installation period, it would be expected that maintenance would need to be done on both the sounds strips and delineators as materials work their way lose or are hit rendering them useless. Installation Issues Addressed As discussed earlier, delineators were recommended to facilitate drivers entering the CTL in enough time to cross over all sounds strips. Delineators, while very practical for dividing lanes, are a maintenance issue as they frequently need to be replaced as they are inadvertently hit. Not only are they rendered useless if they are struck, they are not very aesthetically pleasing, especially when they are lying on the ground. With regards to the sounds strips, the installation of taller sound strips would be more audible; however, they would not be very practical to drivers of motorcycles or bicyclist. In addition, taller sounds strips would likely yield to public outcry to remove them to reduce noise pollution. Installing sounds strips of any sort near residential areas is not advisable. Flashing Beacon with Audible Pedestrian Signals Intro Site Selection The opposing CTL at the south-east quadrant of Providence Road at NC 51 (Pineville-Matthews Road) in Charlotte, NC was utilized for testing of a solar powered flashing beacon equipped with an audible pedestrian signal. The flashing beacon was supplemented with sound strips in the same pattern as the ones installed in the north-west quadrant. Also, delineators were installed similar to the opposing quadrant. This site was utilized because the Charlotte Department of Transportation was very supportive of our research efforts and even paid for the treatments. Also, by utilizing this site, we were able to use the same pedestrians from the sound strip only test (described previously) in the alternate corner of the intersection, thus allowing the effect of the flashing beacon to be determined on its own (the sound strip only effect could be accounted for directly from the opposing corner of the intersection). 70

Appendix E: Treatment and Site Descriptions Speeds on all approaches were posted at 45 mph. Land uses in the vicinity of the site included a good mix of office buildings, retail, and residential. Twenty-four hour volumes at the southeast corner were approximately 3,200 vpd in the CTL and 12,400 vpd in adjacent through traffic. The downstream conflicting through flow was 10,600 vpd and the opposing left turn 3,000 vpd. Low pedestrian activity was observed at the site on the order of 20 pedestrians per day, primarily during the midday off peak period. Field Setup The flashing beacon treatment was set up to be dynamically activated by an APS device. For our testing purposes, the flashing beacon supplemented sound strips by providing an additional cue to the driver that a pedestrian is attempting to cross the street. Two solar powered flashing beacons with APS devices were installed on each side of the road at the crosswalk. Exhibit4 shows the approach to the flashing beacon along with six rumble strips and the delineators at the south-east corner of the intersection. Exhibit 4: South-East Quadrant - View of Sound Strips, Delineators, and Flashing Beacons The flashing beacons tested at the CTL were off-the-shelf devices used for various applications such as advanced traffic signal warnings. They are almost always used as a static device which can often be overlooked by drivers who ignore them; therefore, the team recommended installing them as a package with an APS device. The beacons a pole mounted dual-head, wig-wag display which rested in a dark mode until activated via the push button. The APS message used at the crosswalk said “Cross with caution, cars may not stop” which played concurrently with the wig-wag signal display. 71

Appendix E: Treatment and Site Descriptions Costs While the "sound-strip-only" treatment is considered a low-cost solution, the addition of the flashing beacons adds some cost. However, the cost and associated impact to vehicle operations is still less than for a fully signalized crossing. A solar powered flashing beacon unit costs approximately $3000 per unit. The APS device was approximately $1000 per unit. In addition, the materials for the sound strip would also need to be considered if being used in combination with the flashing beacon. .Two units per CTL are necessary, totaling $8,000 for the entire set up at a single CTL. Installation Issues Addressed The installation of solar powered flashing beacons is a standard off-the-shelf treatment that is readily available. However, the inclusion of APS devices which have a locator tone every 1 second means more power consumption. For our test purposes, a larger battery and solar unit were necessary to keep the APS device functioning properly. In addition, the APS locator tone frequency was changed from 1 second to 2 seconds to keep battery drain to a minimum. Raised Crosswalk Raised crosswalks are sometimes used as a surrogate to standard crosswalks. The standard crosswalk marking is applied to a raised surface approximately 3 to 6 inches above normal road grade. The elevated surface is intended to attract driver attention, thus encouraging lower speeds approaching the crosswalk. The concept of a raised crosswalk is to encourage safe crossings for pedestrians since drivers need to yield prior to crossing the crosswalk. Site Selection The raised crosswalk was installed at the south-eastern quadrant of the double-lane roundabout at the intersection of South Golden Road/Jonson Avenue/16th Street in Golden, CO. The south-eastern leg was utilized for installation of the raised crosswalk. This site was also used for the Pedestrian Hybrid Beacon (PHB) installation on the alternate approach, which allowed the team to utilize the same subjects and limited the need for additional travel. Staff from the City of Golden was very supportive of the research effort, helping purchase and install the raised crosswalk. The raised crosswalk was installed as a low cost treatment in comparison to the PHB, which is quite costly. The raised crosswalk is intended to slow traffic, increasing the likelihood they drivers would yield to pedestrians wishing to cross. This leg was chosen because the geometry was very reasonable, traffic volumes were fairly high, and pedestrian crossings were fairly well established. In addition, a fair number of bicyclists were noted by the team on multiple visits. South Golden Road is the major approach and primarily consists of retail establishments with a small assortment of office and residential locations in the vicinity. Speeds at all the approaches are 45 mph. A photo of the south-east leg prior to installation of the raised crosswalk installation is shown in Exhibit 5. 72

Appendix E: Treatment and Site Descriptions Exhibit 5: South-East Quadrant - View of Sound Strips, Delineators, and Flashing Beacons Field Setup The raised asphalt crosswalk was installed at the south-east quadrant of the two-lane roundabout. In order to avoid drainage concerns, the city did not install the crosswalk flush with the sidewalk (which would have required a drainage pipe under the asphalt), but sloped the raised crosswalk upward from the curb-line. While this doesn't alter driver behavior, it makes for an uncomfortable pedestrian walking environment, as pedestrians have to first walk down the curb-ramp and then back up on the raised crosswalk. The O&M instructor practiced the crossing with all participants so that they would be familiar with the uneven pavement. The raised crosswalk was signed and marked consistent with MUTCD requirements. No other geometric changes were made to the roundabout or its approaches. Exhibits 6 and 7 show photos of the raised crosswalk installed in Golden, CO. Exhibit 6: South-East Quadrant – View of Raised Crosswalk 73

Appendix E: Treatment and Site Descriptions Exhibit 7: South-East Quadrant – View of Raised Crosswalk Costs The costs associated with retro-fitting a raised crosswalk was minimal compared to the installation of a PHB. It is estimated that a crosswalk such as this could be constructed and painted for less than $5000 in materials and labor for each leg. This assumes that drainage is not really accounted for by installing drainage inlets or a pipe under the crosswalk, but is instead allowed to flow down the original gutter line. In addition, because the crosswalk slopes back down to the gutter line to allow water to drain, the curb cuts were allowed left as-is. If drainage were accounted for, the curb cuts would have been filled in and reinstallation of detectable warnings at the crosswalk. Installation Issues Addressed The primary installation issue was how to account for drainage at the crosswalk. Our site was a temporary installation; however, it is recommended that the crosswalk not slope back to the curb cut; but instead be flush to the top of curb. In addition, the slope and height of this particular raised crosswalk were very low. It may be more effective if the crosswalk were installed with larger slope and height to force slower speeds coming into contact with the pedestrian. Pedestrian Hybrid Beacons Pedestrian hybrid beacons (commonly known as HAWK signals) have been in use in the United States since 2000, when the first one was installed in Tucson, Arizona (1). They are gaining acceptance nationally, and are proposed to be in the next edition of the MUTCD. Tucson’s HAWK signals are primarily installed at mid-block locations on wide arterials as either one-stage or two-stage operations. However, HAWK signals have been identified as a potential pedestrian crossing treatment to improve roundabout crossings for all pedestrians, including those with visual impairments, and appear to fulfill the Access Board’s proposed requirement that pedestrian crossings at roundabouts be signalized. 74

Appendix E: Treatment and Site Descriptions To further examine the viability and benefits of introducing HAWK signals at roundabouts, a temporary one was installed on one leg of a roundabout as part of this project. This was the first installation of a HAWK signal at a roundabout in the United States. Site Selection The double-lane roundabout at the intersection of South Golden Road/Johnson Avenue/16th Street in Golden, Colorado was chosen as the test site for a HAWK signal installation. This site was also used for the raised crosswalk treatment on an alternate approach, which allowed economical testing of both treatments under similar traffic conditions and with a common set of participants. Staff from the City of Golden was supportive of the research effort and willing to install both treatments. Double-lane roundabouts are generally more challenging to pedestrians than single-lane roundabouts. Pedestrian crossing distances are longer, vehicle speeds can be higher, and traffic volume can be higher. In addition, the draft Public Rights-of-Way Accessibility Guidelines require some form of signalization for multilane roundabout entries and exits. For these reasons, a double-lane roundabout was more suitable than a single-lane roundabout as a test site. The HAWK signal was installed on the northwest leg of the roundabout on South Golden Road. This leg was chosen because of reasonable geometry, moderate pedestrian volume, and physical conditions conducive to installing a temporary signal at reasonable cost. A photo of this leg prior to HAWK signal installation is shown in Exhibit 7. Entry Crossing Exit Crossing Exhibit 7 – HAWK signal test site prior to installation Field Setup HAWK signals are not addressed in the current (2003) edition of the MUTCD. The proposed amendments to the MUTCD, prepared in anticipation of the next edition, include a new chapter (4F) on “Pedestrian Hybrid Signals”, or HAWK signals (2). In this proposed MUTCD chapter, it is stated that, except as noted, pedestrian hybrid signals shall meet the provisions of [proposed] Chapters 4D and 4E, which address normal green/yellow/red signals and pedestrian control at such signals, respectively. 75

Appendix E: Treatment and Site Descriptions Many of these exceptions are related to the signal face and the sign placed next to the signal face explaining when to stop. There is only one mention of roundabouts in proposed Chapter 4F. It is noted that if a pedestrian hybrid signal is installed at a roundabout to facilitate crossings by visually impaired pedestrians, pedestrian signal heads may rest in dark if an engineering study determines that pedestrians without visual disabilities may cross safely without activating the signal. This is only an option at roundabouts; elsewhere the pedestrian signal shall rest with a steady upraised hand indication. To install the temporary HAWK signal, the City of Golden submitted and was granted a Request to Experiment to FHWA. This request, provided in Appendix F, includes signal design plans and signal timing plans. The HAWK signal was designed to conform with the 2003 MUTCD and the proposed amendments to the extent possible. Striping A 24” solid white stop bar was placed 4 feet prior to the crosswalk on the entry and the exit of the approach. All existing striping was left in place, including the continental-style crosswalk marking and the yield line at the roundabout entry. Signal Poles and Heads A total of four signal poles were used: two on each approach (entry and exit). Each pole was used to mount a pedestrian signal head, a pedestrian push button, and a vehicle signal head with 12-inch lenses. At the Golden roundabout, crosswalks are set back 30 feet from the circulatory roadway. Because of this, the placement of stop bars, poles and traffic signal heads presented challenges. Under a typical signal design, Section 4D.15 of the 2003 MUTCD specifies a minimum separation of 40 feet between the stop bar and traffic signal heads. On the exit leg, this 40 foot separation could be obtained in one of the following ways: • If the stop bar were placed 4 feet prior to the crosswalk, the signal heads would be placed approximately 25 feet beyond the far side of the ten foot wide crosswalk. This would result in signal heads being far removed from the pedestrian crossing and the roundabout intersection environment. Drivers might not see the signal. • If the signal heads were placed at the far side of the ten foot wide crosswalk, the stop bar could be placed approximately 30 feet prior to the crosswalk. This would place the stop bar in the circulatory roadway. On the entry leg, assuming the signal heads were located on the far side of the crosswalk, the stop bar would need to be located approximately 30 feet in advance of the signal. The language in proposed MUTCD Chapter 4F on pedestrian hybrid signals treats the entire section on signal head location as guidance, not a standard. In this Golden experiment, the site constraints require 76

Appendix E: Treatment and Site Descriptions a closer separation between the stop bar and signal faces, with the poles located at the far side of the crosswalk and the stop bar located 4 feet in front of the crosswalk. This creates a 14-foot separation between the stop bar and the signal face, which is mitigated by locating the bottom of the signal face 10 feet above the ground, five feet lower than the MUTCD-recommended minimum mounting height of 15 feet. Signal Timing and Controller Programming HAWK signals are intended to provide pedestrians with a safe means of crossing a roadway while minimizing the delay to vehicles that is created by doing so. On roundabout leg, this is preferably accomplished with a two-stage crossing. When a pedestrian activates the signal, vehicles are only stopped in one direction of travel. For example, if a pedestrian will be crossing the lanes entering the roundabout, only entering traffic is stopped. After making this crossing, the pedestrian will be in the splitter island, which is designed to be a refuge for pedestrians. The pedestrian can then activate the signal for the second crossing. This allows both the walk time and pedestrian clearance intervals to be as short as possible to minimize both vehicular and pedestrian delay. In addition to reducing vehicular delay, a two-stage crossing may result in increased vehicular compliance with HAWK signals. If, for example, a pedestrian were crossing the entry approach and the HAWK signal on the exit approach was activated, drivers on the exit approach could be less likely to stop since a pedestrian is not present. Controller and Timing Plan A NEMA controller (Econolite ASC2S Type II) was the design controller for the test site and is shown in the plans and specifications provided in Appendix F. This is the same controller used by the City of Tucson at their HAWK signals. Ultimately, the City of Golden installed a spare 170 controller, for which a timing plan was developed. The 170 plan in Appendix F is the initial plan; some changes were made in the field following installation and are discussed below. NEMA Setup The NEMA controller plan for the test site in Golden used specifications developed by Tucson staff for two-stage crossings. Tucson has adapted NEMA controllers and cabinets from control of green/yellow/red signals to control of HAWK signals largely by rewiring the controller cabinet into a custom configuration, with load switches being reassigned through the creative use of jumpers. For example, both the vehicular and pedestrian heads were assigned to pedestrian phases, which enabled the vehicular heads to display both solid and flashing indications. The two red indications were connected to separate load switches to make wig-wag flash possible. Limited software changes were necessary, including a change to the Econolite controller’s write-protected memory to allow flash don’t walk (FDW) intervals of less than seven seconds when FDW was used to flash vehicular signal indications. 77

Appendix E: Treatment and Site Descriptions A custom conflict monitor schedule was used to accommodate the extensive changes in phase assignments needed to create the HAWK timing plan. 170 Setup Unlike the hardware-oriented modifications needed for the Econolite NEMA system, the 170 system was modified to control a HAWK signal through use of the Command Box programming within the Wapiti W4IKS software that Golden uses in its 170 controllers. A custom Command Box program was written specifically for this project and tested with a briefcase tester prior to installation on the controller at the test site in Golden. In the 170 plan developed for this project, phases 3 and 4 are used in the first ring and phases 7 and 8 are used in the second. In the first ring, phase 3 amber is connected to the amber indication, phase 3 red is connected to one of the red indications, and phase 3 green is connected to the second red indication. Phase 4 is connected to the pedestrian signal head. The connections for the second ring are similar. Flashing and steady indications are created by Command Box program. Signal Timing HAWK signals operate with three phases – an activation phase, a pedestrian crossing phase, and a “green” time for vehicles. The final timing plan – installed on the 170 controller at the test site in Golden – is described below and shown in Exhibit 8. 78

Appendix E: Treatment and Site Descriptions Exhibit 8 - HAWK Signal Phases and Timing In the first phase, the signal is activated – the vehicular signal flashes yellow, and then displays solid yellow. Both the flashing yellow and solid yellow clearance intervals were set at 3 seconds. Throughout phase 1, the pedestrian signal heads continue to display a steady hand. 79

Appendix E: Treatment and Site Descriptions In the second phase, the pedestrian signal head displays a walk indication followed by a FDW and the vehicular signal head displays a steady red indication on both red lenses followed by a wig-wag flashing red. The interval for the walk and the solid red indications is 4 seconds, and the interval for the FDW and wig-wag flashing red is 7 seconds. The walk was set to the minimum of 4 seconds to minimize vehicular delay. The third phase functions as a minimum green time for vehicles. This prevents multiple pedestrian calls from causing excessive vehicular delay. There are no signal heads wired to the third phase, as a dark HAWK signal is “green” for vehicles. The minimum green time is set at 10 seconds and the yellow clearance interval is set at 5 seconds, resulting in an effective minimum “green” time of 15 seconds. The pedestrian signal heads display a don’t walk indication. If no pedestrian calls are placed, the signal will rest in this phase. The timing plan described above differs from the implementation designed for the Econolite NEMA controller in several ways: • The yellow flashing interval and the solid yellow clearance interval in the first phase were both decreased from 5 seconds to 3 seconds. These changes were intended to reduce delay and confusion for drivers and pedestrians. Three seconds of flashing was considered to be sufficient time to notify drivers that the signal is being activated, and 3 seconds of clearance time was considered sufficient for the speeds and geometry at the test site. Roundabouts generally limit vehicle speeds to 30 miles per hour, whereas mid- block locations could be much higher. For pedestrians, the shorter intervals decrease the time between placing a call and receiving a walk indication • The walk interval was shortened from 7 to 4 seconds. The 2003 MUTCD states that a walk interval should be 7 seconds, but that it may be as short as 4 seconds “if pedestrian volumes and characteristics do not require a 7-second walk interval”. At the test site, pedestrian volumes are low enough that persons crossing can be expected to be at the curb and not queued. Also, except when the minimum vehicular “green” has not been served, the walk indication will be given six seconds after a call is placed, when pedestrians will generally be expecting it to appear. Cost The City of Golden constructed much of the HAWK signal with spare materials already on hand, greatly reducing the experimental cost. If the signal had been built with new materials purchased specifically for this project, it is estimated that the cost would be about $53,000, as detailed in Appendix F. Approximately half of the estimated cost is for the controller cabinet and associated cabinet hardware, including the controller itself. If all crossings of a roundabout were to be signalized, costs could be significantly reduced by using one controller for the entire roundabout instead of one controller per leg. Installation and Turn On 80

Appendix E: Treatment and Site Descriptions The HAWK signal was turned on in mid-August of 2008. Photos of the installed signals on the entry and exit approaches are shown below in Exhibit 9. A photo of the two-stage crossing from a pedestrian’s viewpoint is shown in Exhibit 10. A photo of the 170 cabinet is shown in Exhibit 11. Entry Crossing Exit Crossing Exhibit 9 – HAWK signal test site following installation Exhibit 10 - Two stage, signalized pedestrian crossing 81

Appendix E: Treatment and Site Descriptions Exhibit 11 – 170 cabinet at test site in Golden, Colorado Lessons Learned and Recommendations for Future Study The groundbreaking nature of this design revealed several issues that will need to be further studied by the profession if HAWK signals are to be widely used at roundabouts. Visibility Issues At the test site, it was not practical to meet the 40-foot minimum separation between the stop bar and the signal heads that is recommended by the MUTCD for visibility purposes. Such a spacing would have either placed a stop bar in the circulatory roadway or placed signal heads far beyond the crosswalk. At the test site, a shorter spacing was used. Signal heads were pole mounted 10 feet off of the ground, making them visible to a design driver in a design vehicle stopped only 14 feet away. One way of obtaining a 40 or more feet of separation between the stop bar and signal heads would be to move the crosswalk further away from the circulatory roadway. The current recommendation of 25- foot spacing between the crosswalk and the circulatory roadway is primarily indented to create the safest possible unsignalized crossing. With a HAWK signal, a greater crosswalk setback, such as 50 feet, might be beneficial. Vehicle speeds and pedestrian visibility are less of a concern since the signal will be stopping vehicles. On the exit, the stop bar could be placed well in advance of the crosswalk and there could still be a space for at least one vehicle to stop and be fully out of the circulatory roadway. On the entry, a queue of two vehicles could be stopped and waiting to enter the roundabout without blocking the crosswalk. 82

Appendix E: Treatment and Site Descriptions Both of these designs should be further studied. Even if a 50-foot crosswalk setback was found to be optimal in most situations, a 25-foot setback (and shorter stop bar to signal head distance than currently recommended in the MUTCD) might still be optimal in some circumstances. These circumstances could include retrofits of existing roundabouts and locations where expanding the footprint of the roundabout would be prohibitively expensive. Moving crosswalks, and thus signal heads, further back from the circulatory roadway will help to mitigate this issue. It will not, however, entirely eliminate it. Modifications to the design and placement of the signal heads on the exiting lanes should be studied. One possibility would be to shield lenses. Another would be to change the mounting position of the signal heads. Different heights or angles should also be explored to see if visibility to vehicles entering on the opposite approach can be decreased while maintaining or increasing visibility to exiting vehicles (i.e. those being controlled by the HAWK signal). Obtrusiveness Issues HAWK signals are intended to provide pedestrians with a safe, controlled crossing of a roadway while minimizing vehicular delay. At the test site, a four second walk interval (the minimum allowed by the MUTCD) and a vehicle “green” time of 15 seconds were used. At a normal signalized intersection, it is common for pedestrians to wait much longer than 15 seconds to receive a walk indication after placing a call. At a congested roundabout, a minimum green time of more than 15 seconds may be appropriate and should be investigated. Clearance intervals and other aspects of the signal plan should be further investigated as well. Intersection-Wide Application Issues This experiment was limited in that a HAWK signal was only installed on one leg of the roundabout. If HAWK signals are adopted into the next edition of the MUTCD and become widely used at roundabouts, they would be installed on all legs at many locations. At a typical four-leg intersection, this would result in eight signalized crossings – one for each entry and one for each exit. Optimally, each of these crossings should operate independently to minimize vehicular delay. As previously noted, the controller cabinet and related hardware are estimated to comprise half the cost of signalizing a single leg. To minimize costs at a roundabout where crossings on all legs are signalized, a single controller should be used. This would require eight rings – more than are available on any commercial traffic signal controller software. Creating software capable of handling eight rings (or more, for cases of unusual geometry) should be a priority, as it will significantly reduce the cost of HAWK signals at roundabouts. Such software should also be designed specifically for HAWK signals. Currently, using software and controllers designed for normal green-yellow-red signals at HAWK signals requires extensive customization to do things such as display flashing and steady indications on the same signal head. 83

Appendix E: Treatment and Site Descriptions Off-the-shelf HAWK signal software, with eight or more rings and the ability to display all necessary flashing and steady indications should be developed for both NEMA, 170, and 2070 controllers. This would eliminate the need for extensive manual rewiring and/or reprogramming. References: 1. Nassi, Richard B., City of Tucson (retired). Presentation to Kittelson & Associates, Inc. staff, December 11, 2008. 2. Proposed Amendments to the MUTCD. <http://mutcd.fhwa.dot.gov/resources/proposed_amend/index.htm>. Accessed April 1, 2009. 84