Work Zone Speed Management (2015)

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.

35 chapter four ENGINEERING TECHNIQUES INTRODUCTION How a work zone is designed and organized can help create an environment that encourages drivers to select an appropriate speed. Typically, this involves providing visual or tactile cues (in addition to speed limit signs) that let drivers know they are entering a stretch of road where driving conditions are more difficult and speeds need to be reduced. A number of engineer- ing measures have been proposed to help reduce speeds in work zones. Researchers and transportation professionals in some regions of the United States use the term “work zone traffic calming” to refer to such methods. In other regions, the term “traffic calming” is strongly associated with residential streets and other low-speed environments, and the use of the term traf- fic calming may require clarification if applied in the context of speed reduction on freeways and other principal highways. Broadly speaking, engineering techniques can be classified as either physical or perceptual. Typically, physical measures involve reducing lane width or changing vertical or horizontal geometry so as to provoke a reduction in motor vehicle speeds. Perceptual measures generally involve creating the illusion that the roadway is narrower or the driving environment is becoming more difficult, and hence speed must be reduced. PHYSICAL REDUCTION OF LANE WIDTH Generally speaking, roads with wide lanes are easier to drive than roads with narrow lanes. As lane width increases there is an increased margin for error, so drivers will tend to increase their speeds. The physical width of travel lanes in work zones appears to have a moderate effect on traffic speeds: narrowing the lanes appears to reduce both speed and the capacity of the work zone. Right-of-way constraints often require lane width to be reduced to provide sufficient space for work operations. In some instances, lane width has also been deliberately reduced to achieve a speed reduction effect. From a human factors per- spective, reduced lane width means less lateral maneuvering distance and a reduction in the distance between the vehicle and roadside obstacles such as traffic control drums or barriers; this requires more driver attention, and drivers often compen- sate by reducing speeds. The 2010 Highway Capacity Manual (HCM) (TRB 2010) considers 12-ft lane width as the base case and provides values for reduction in free-flow speed because of narrow lanes on basic freeway sections and multilane high- ways. No guidance is provided on the impact of narrow lanes on speed reduction in work zones. Kemper et al. (1984) studied the safety effects of narrow lanes in construction zones. The study was conducted during the 17-month before and after period during the reconstruc- tion of bridge decks on the George Washington Memorial Parkway near Washington, D.C. The study found that the use of 9-ft lanes in Stage 1 of the reconstruction increased the crash rate, although crash severity decreased. Lower speeds were observed but the speed reduction was not quantified. Two 1980s studies evaluated the speed change effects of narrow lanes in two work zones (Richards et al. 1982; Kuo and Mounce 1985). At one location mean speeds were reduced by 3 to 8 mph, whereas there was only a slight reduction at the other. The study did not control for the influence of other fac- tors on speeds and, therefore, the reductions may not be fully attributable to reduced lane width. Another mid-1980s study evaluated the effect of lane width reductions at six work zones on rural and urban freeways in Texas (Richards et al. 1985a, b). Cones were used to reduce the lane widths to 11.5 and 12.5 ft in the work zones. The effect of the lane-width reduction varied widely across the different sites, from 0 to 8 mph. In other words, the effect varied from none to a 16% speed reduction. The study suggested that using drums or concrete barriers might result in greater reductions. A more recent study explored reductions in free-flow speed resulting from reduced lane widths and lateral clearances in work zones (Chitturi and Benekohal 2005). Traffic data were collected at 11 work zones on interstate highways in Illinois that normally had two lanes and were reduced to one open lane (Table 8). Four work zones had a period of time during which there was no work activity, police presence, flagger presence, or the presence of a flashing speed limit of 45 mph. All four sites were long-term work zone sites with a posted speed limit of 55 mph in the work zone and 65 mph outside the work zone, had no lateral clearance on either side of the travel lane, and were located in level terrain. The lane widths in the four work zones were 16, 12, 11, and 10.5 ft. The reductions in vehicular free-flow speed in the work zones were found to be greater than the reductions given in the HCM for basic free- way sections. The data also showed greater speed reduction for greater reductions in lane width. In addition, the reduction in the free-flow speeds of heavy vehicles was greater than the

36 NCHRP Report 476 cautions against this situation (Bryden and Mace 2002): It is often necessary to use two or more longitudinal rows of chan- nelizing devices to define both sides of a travel lane or for exit and entrance lanes. In these locations, it is essential to control the width between rows of devices to restrict traffic to the intended number of lanes while providing adequate width to accomplish turning reduction in the free-flow speeds of passenger cars. As shown in Figure 15, the study recommended that 10.0, 7.0, 4.4, and 2.1 mph be used for speed reduction in work zones for lane widths of 10, 10.5, 11, and 11.5 ft, respectively. A 1989 design manual published by Spain’s MOPU dis- cusses the use of “lateral obstacles” to control work zone speed by reducing the effective lane width (MOPU 1989). It appears that typical Spanish practice is to channelize the traffic using vertical panels arranged in pairs on each side of the roadway to create a “gateway” effect; longitudinal panels and tempo- rary barriers are also suggested for this purpose (Figure 16). As shown in Table 9, the manual suggests a nearly linear rela- tionship between lane width and the target traffic speed, but does not discuss relationships with the design standard, traffic environment, or ordinary speed of the roadway; as a result, it is unlikely that the effect of lane width alone is as strong as the MOPU manual suggests. Lane width reductions may require additional adminis- trative coordination for the routing and permitting of over- width trucks. In some cases it is necessary to move drums or other temporary traffic control devices while a wide load passes though the work zone (Shaw et al. 2014). Occasionally, there are work zone situations where a lane closure results in the remaining lane being very wide. Lane Width (feet) Free-Flow Speed Reduction (mph) 10 10.0 10.5 7.0 11 4.4 11.5 2.1 Source: Chitturi and Benekohal (2005). TABLE 8 RELATIONSHIP BETWEEN SPEED AND LANE WIDTH PROPOSED BY CHITTURI AND BENEKOHAL FIGURE 15 Restriction of expressway lane width, possibly initiated by a contractor attempting to reduce speeds adjacent to the activity area (Rizos 2014). (a) (b) FIGURE 16 Vertical panels: (a) Vertical panels used for channelization and lane narrowing in Spain (MOPU 1989), (b) U.S. MUTCD vertical panel (FHWA 2009). Desired Speed Width for Width for km/h mph One Lane (feet) Two Lanes (feet) 100 62 12.6 24.6 90 56 12.1 23.8 80 50 11.6 23.0 70 43 11.2 22.1 60 37 10.8 21.3 50 31 10.5 20.5 Source: MOPU (1989). TABLE 9 RELATIONSHIP BETWEEN SPEED AND LANE WIDTH PROPOSED IN WORK ZONE DESIGN GUIDANCE ISSUED BY THE MINISTRY OF PUBLIC WORKS AND URBANISM IN SPAIN

37 sure with traffic merging into the left lane followed by a lane shift to the right, as shown in Figure 18. According to Lorscheider and Dixon (1995), “this type of closure has [also] been successfully performed on three lane, one-way sections, although this use is limited because of its complexity. To perform this technique, two lanes are closed and the open lane is weaved across the center lane to the far side.” An example is illustrated in Figure 19. maneuvers required by the traffic pattern. Where excessive width is available, aggressive drivers may attempt to pass slower traf- fic, creating confusion and a potential conflict when the lane again narrows. Typical problem areas are locations where a single open lane shifts from one side of the roadway to the other, at on- and off- ramps. For tangent sections and flat curves, the normal lane width of 4 m (12 ft) is normally sufficient to permit smooth flow while discouraging high speeds. In shifts and other locations where some turning is necessary, especially if articulated vehicles are present, increasing the width between devices to 4.3 m or 4.9 m (14 ft or 16 ft) will smooth flow and reduce abrupt driver reactions. Except where severe alignment necessitates, more width, greater than 4.9 m (16 ft) between devices, may encourage drivers to form two lanes or attempt to pass. CHICANES In the absence of superelevation, vehicles must reduce speed when they encounter a horizontal deflection in the travelled way. Any roadway configuration that adds horizontal curves for the purpose of reducing traffic speeds can be referred to as a chicane. One example of the use of permanent chicanes is to reduce traffic speeds as a high-speed, two-lane highway enters a village, as shown in Figure 17. Various highway agencies have explored the use of chi- canes to reduce speed in the upstream portions of work zones. In the United States, the most common technique is the Iowa Weave or switchback, which typically consists of a lane clo- FIGURE 17 Permanent chicane intended to reduce speeds on highway D29 on a downgrade entrance to the village of Ernée, Mayenne, France (CERTU 2006, 2007). FIGURE 18 Iowa Weave for 2:1 closure with downstream left lane closed (Schrock et al. 2008). FIGURE 19 Iowa Weave for 3:2 closure with downstream left lane closed (See 2008).

38 ent situations, including two that are similar to the Iowa Weave (Direccion General de Carreteras 1997). In typical applications the speed limit is reduced from 100 km/h (62 mph) upstream to 80 km/h (50 mph) downstream. Older, more general, Spanish guidance cautions that, “to be safe and effective it is essential that [the chicane] is easily perceived and understood by the driver, and coordinated with signs and signals. At night, when traffic flow is low, it must also be clearly visible” (MOPU 1989). A Swedish study explored new methods for marking work zone closures. Figures 21 and 22 show the “conventional” double-chicane design that served as the study’s base- line (Nygårdhs 2007). Typical of current European practice, vertical panels with chevron markings are used for delinea- tion. Considerable speed reduction is intended in the Swedish design: the posted limit is stepped down in 20-km/h (12-mph) increments to 50 km/h (approximately 30 mph) in the chicane, rising to 70 km/h (approximately 45 mph) downstream of the median crossover. According to Nygårdhs, observed speeds at point M3 of Figure 22 were approximately 65 km/h (40 mph), and were at or below the posted limit of 50 km/h (30 mph) at point M4. Additional research on the capacity and safety of chicanes may be desirable to provide practitioners with guidance on their appropriate use. TEMPORARY TRANSVERSE RUMBLE STRIPS Rumble strips provide drivers with tactile and audible warn- ing in advance of a speed reduction. Proponents advocate the use of temporary transverse rumble strips for a wide variety of work zone applications, such as approaches to intersections, temporary traffic signals, and flagger stations. Interviews con- ducted for this project suggest that currently the primary field use of temporary rumble strips is for the approach to flagger stations on two-lane rural highways. Kansas DOT requires the use of rumble strips on state routes when work requires the closure of a lane on a two-lane highway (Meyer 2000). An experimental freeway application has also been attempted on I-35 in Texas. A study by Hallmark recommended that to increase effectiveness, signing and other traffic control devices should be used in conjunction with the rumble strip to give drivers an indication of the reason for the required speed reduction (Hallmark et al. 2007). Researchers have explored a number of temporary trans- verse rumble strip configurations (Figure 23). Various materials have been used including formed asphalt, exposed aggregate, thermoplastic, rubber, and multiple layers of pavement mark- ing tape. Studies have examined various combinations of num- ber, thicknesses, and color of the strips (Harwood 1993; Shaik et al. 2000; Morgan 2003). Typical rumble strip dimensions ranged from 4 to 8 in. wide and from 1⁄8 to 11⁄2 in. high. Sev- eral colors have been tested including black, white, yellow, and orange. These design differences (along with differences in the The Iowa Weave was developed in the 1960s by the Iowa State Highway Commission (now Iowa DOT) for use in the transition area of work zones on multilane roads and streets. It was originally designed as a method to reduce speeds prior to entering a work area where workers were present and is typically implemented using traffic control drums or cones. An early 1970s study on an arterial street found it highly effective, with more than 50% of all sampled vehicles traveling below the 30 mph posted temporary speed limit, compared with less than 20% in construction sites where the Iowa Weave was not used (Brewer 1972). Brewer concluded that there was no excessive driver confusion in the use of this pattern, although three drivers were found performing unusual maneuvers in the advance warning area. A mid-1990s study examined the effect of the Iowa Weave by evaluating the speeds at the advance of warning, end of taper, and weave and lane closure areas on both rural and urban freeways (Lorscheider and Dixon 1995). The research showed that the pattern is effective in reducing work zone speeds, but increases driver confusion in urban settings. The study therefore advised that the use of this chicane pattern be limited to rural freeways. In addition, Lorscheider and Dixon found that the speed reduction resulting from the chicane dis- sipated within 0.75 mile. The study’s authors also pointed to time-saving advantages for contractors, because the switch- ing from a right-lane Iowa Weave closure to a left-lane con- ventional closure can be accomplished quickly. A 2008 study reported on the use of the Iowa Weave for freeway projects in Arkansas, Iowa, North Carolina, and Tennessee (Schrock et al. 2008). The authors reported that the Tennessee Department of Transportation requires all interstate construction and maintenance projects to review and include the use of the merge left where lane closure is applicable. The following criteria are used to determine its use: • Projects on rural interstates should include the merge left concept. • Projects on urban interstates will be reviewed for merge left concept considering factors such as number of lanes, interchange spacing, and proximity to major splits. • Other controlled access facilities will be considered on a case-by-case basis. Another 2008 study evaluated the crash performance of 10 Arkansas work zones, with and without Iowa Weaves (see Schrock et al. 2008). Based on a paired comparison, the author found a 30% reduction in the crash rate when the Iowa Weave configuration was used, but a logistic regression model found that the crash severity differences between the Iowa Weaves and conventional right-lane closures were not statistically significant. Chicane designs are also used internationally to slow traf- fic at work zone approaches. As shown in Figure 20, a Spanish work zone design manual illustrates several designs for differ-

39 (a) FIGURE 20 Two-lane and three-lane chicane designs from a work zone design manual published by Spain’s Ministry of Public Works (speeds in kilometers/hour and distances in meters) (Dirección General de Carreteras 1997). (continued on next page)

40 (b) FIGURE 20 (Continued ).

41 mean speed by 6.9 km/h (4.3 mph) and the 85th percentile speeds by 9.5 km/h (5.9 mph). Variation of speeds was also improved by rumble strips. Specifically, the number of vehicles in the 15 km/h (9.3 mph) pace increased by an average of 6.4% and the standard deviation of oper- ating speeds was reduced by an average of 0.86 km/h (0.53 mph) (Copeland 1998; Hildebrand et al. 2003). • A study evaluated orange removable rumble strips and compared them with “standard” cold mix asphalt rumble strips (Meyer 2000). The orange removable strips were reported to have a significant effect on vehicle speeds (about 2 mph) as a result of their higher visibility. • A 2001 report focused primarily on rumble strip appli- cations for low-volume, two-lane roads in Texas with a 70 mph regulatory speed limit (Fontaine and Carlson 2001). The study’s authors stated that results for portable rumble strips were mixed, with passenger cars experiencing less than a 2 mph reduction in mean speed approaching the temporary traffic control zone. The impact of the rumble strips on trucks was more highway context, site conditions, and the upstream pavement surface condition) probably contribute to the wide variation in the results of field studies. Nearly all of the rumble strip research reported an increase in driver awareness; however, most of the studies did not report quantitative changes in average speed. As one research group explained, “tests indicate that there may be positive benefits from temporary rumble strips, but a behavioral response from drivers is not necessarily observable through objective obser- vations of vehicular speeds” (Horowitz and Notbohm 2005). The following are some of the findings: • An evaluation at four work zones on two-lane roads in New Brunswick reported that rumble strips reduced the FIGURE 21 Approach to chicane at a rural freeway work zone in Sweden (Nygårdhs 2007). FIGURE 22 “Conventional” chicane design from Swedish study (Nygårdhs 2007). (a) (b) FIGURE 23 Temporary transverse rumble strips (Pappe 2014).

42 The average speed of all vehicles deceased by 1 mph; nevertheless, the number of adjacent lane encroach- ments increased by 8.8%. • A Kansas study evaluated the effect of portable plastic rumble strips (PPRS) at three short-term maintenance work zones (Wang et al. 2011). The PPRS were placed in sets of four at 36 in., with two or three sets installed in advance of the flagger controlled work zones. Data were collected when no work activity or traffic control were present, when standard flagger traffic control was pres- ent and with PPRS in place. PPRS reduced car speeds by 4.6 to 11.4 mph. PPRS also reduced truck speeds by 5 to 11.7 mph on average, but only at two sites. About 5% of car and truck drivers swerved around the PPRS. The authors suggested that additional signing might be required. As with the academic studies, practitioners reported a range of experiences with temporary rumble strips. • VDOT recently established a typical application drawing for temporary rumble strips (D. Rush, VDOT, personal communication, 2014). The Virginia design calls for one set of three rumble strips to be placed 1,000 ft in advance of the flagger station; it is based on a proprietary prod- uct consisting of interlocking 36-in.-wide self-weighted rumble strip sections, black in color. The product vendor recommended using two sets of three strips; however, Virginia eliminated the second set after observing motor- ists attempting to drive around them. VDOT reports sat- isfactory performance, with the strips typically moving longitudinally approximately 3 in. during a typical workday. No difficulties with motorcycle safety have been reported. A gap in the strips allows bicycles to avoid driving over the rumble strips. • Saskatchewan’s Ministry of Infrastructure & Transpor- tation reports difficulties keeping rumble strips in the correct location, possibly because of incorrect installa- tion by contractors (M. Muhr, Saskatchewan MHI per- sonal communication, 2014). A new, easier-to-install version of the proprietary product is expected to resolve the problem. Driver acceptance has been high, with 79% of drivers who responded to an online panel survey conducted for the Ministry stating that they agree that “Rumble strips alerted me to important information.” The survey was conducted in December 2013 following the first construction season that the product was used in Saskatchewan (Insightrix 2013). EMERGENCY FLASHER TRAFFIC CONTROL DEVICE The Emergency Flasher Traffic Control Device (EFTCD) is intended to reduce traffic speeds on the approach to flagger stations at sites with two-way, one-lane operation. In a test at rural work zones in Kansas the technique was implemented as follows: as the first vehicle reached the flagger station and pronounced, with mean speed reductions approaching the temporary traffic control zone of up to 7.2 mph lower than normal traffic control. The percentage of vehicles exceeding the speed limit in the advance warning area was also reduced when the rumble strips were used. Erratic maneuvers near the rumble strips were observed, perhaps because the strips used for the study were bright orange that contrasted sharply with the asphaltic pavement. • A 2002 before and after study evaluated the perfor- mance of a proprietary product designed to be “quieter than a conventional rumble strip” at the approach to a work zone on a rural two-lane highway in Wisconsin (Horowitz and Notbohm 2002). The quality of sound was reported to be “distinctly different from a conventional rumble strip.” The combination of warning signs and the rumble strips resulted in a small (less than 1 mph) but statistically significant slowing of vehicles. • A 2005 report described results as “minor” (Lessner 2005). • An evaluation of two proprietary rumble strip products conducted in Wisconsin in 2005 concluded that speed was an important determinant of the amount of sound coming from a work zone rumble strip (Horowitz and Notbohm 2005). The 0.25-in.-thick product was “effective” at 55 mph, but not at 40 mph. The 0.75-in.-thick product was effective for vehicle speeds of 10 to 40 mph. • A 2009 Iowa report quantified the mean speed reduc- tions as being in the range of 1 to 2 mph (Fitzsimmons et al. 2009). • A Florida study evaluated removable rumble strips at construction work zones on a state route (McAvoy et al. 2009). Speed studies were conducted 600 ft and 5,500 ft upstream of work zones. At 5,500 ft upstream speeds were similar regardless of the presence of rumble strips. At 600 ft upstream rumble strips produced a speed reduc- tion of 8 mph when compared with locations without rumble strips. The authors suggest that placing rumble strips closer to the work zone and using several sets in succession might have contributed to the greater reduc- tion compared with previous studies. • A 2010 study tested portable rumble strips on a closed roadway in Kansas City and in a closed Park & Ride facility in Lawrence, Kansas (Heaslip et al. 2010). Por- table temporary rumble strips made out of steel with a rubber bottom and four generations of plastic rumble strips were tested. The most effective solution for most short-term work zones was found to be the fourth genera- tion of plastic rumble strips. Steel rumble strips were also found to be promising; however, their structural integrity needs to be addressed. • A Missouri study evaluated a nonadhesive temporary rumble strip near a two-way, one-lane operation on a low-volume road (Sun et al. 2011a). More than 10% of vehicles braked because of the rumble strips. Of the vehicles that braked, the average decease in speed was 3.7 mph and speed compliance increased by 2.9%.

43 Barriers, LLC (only the MBT-1 was in commercial produc- tion). Both systems are shown in Figure 24. They are typi- cally used for freeway maintenance projects such as shoulder repair, guardrail replacement, bridge deck repairs, bridge joint maintenance, and pavement patching. A 2013 Oregon study found that compared with a standard closure using traffic cones, deployment of the MBT-1 during night work generally increased traffic speeds passing the work zone, but work- ers were on the opposite side of the crash-tested barrier; the study’s authors concluded that this benefitted driver mobility (Gambatese and Zhang 2013). • Balsi Beam. The tractor-trailer Mobile Worker Protec- tion Device was developed by Caltrans in 2001–2003 (Caltrans 2004, 2007). The trailer consists of two tele- scoping steel beams whose width can be extended to 12 ft. Using hydraulic power, each beam can rotate to the left or right side as needed; with both beams stacked on the same side, the wall height is 3 ft. The trailer can be extended to provide a work area up to 30 ft long. In 2003, Caltrans conducted crash testing based on NCHRP 350 Test Level 2 procedures (43 mph). • MBT-1. The MBT-1 barrier consists of a 5-ft-tall smooth steel wall to protect the work zone from the side, com- bined with an impact attenuator and Portable Changeable Message Sign (PCMS) at the rear (Mobile Barriers LLC 2014). Wall sections can be added to increase the length of the work area from 42 to 102 ft. The device can be reconfigured for left or right side of road placement by swapping the positions of the semi-tractor and the rear wheels. In 2008, the system was certified as compliant with MASH/NCHRP 350 Test Level 3 (62 mph). GATEWAY ASSEMBLIES Gateway assemblies are intended to serve as a perceptual speed reduction measure, heightening the sense that the driver is approaching a constrained driving environment. At least two Canadian provinces (Manitoba and Saskatchewan) use the assemblies at work zone approaches and terminations came to a stop, a research assistant asked the driver to turn on the vehicle’s emergency flashers, thus increasing the con- spicuity of the flagger station. The same request was made as each subsequent vehicle stopped at the back-of-queue, fur- ther augmenting the visibility of the flagging operation and the location of the back-of-queue. Ideally, the drivers of all vehicles would consecutively illuminate their flashers until they reached the end of the work zone. An evaluation of the technique was conducted at three work zones on two-lane rural highways in Kansas (Bai and Li 2009, 2011). The sites had ordinary (nonconstruction) speed limits of 55 to 65 mph and traffic volumes of 750 to 5,000 vehicles per day. For all three sites combined, speeds were observed for a total of 110 vehicles with EFTCD in operation and 118 vehicles without EFTCD. At the 65 mph site, the average speed reduction measured 500 ft upstream of the flagger station was 5 mph; using combined data from the two 55 mph sites the average speed reduction measured 400 ft upstream of the flagger station was 2.5 mph com- pared with the speeds at the same locations when EFTCD was not used. Based on driver surveys, the EFTCD cap- tured the attention of 84% of the drivers and 60% of the drivers thought that the EFTCD signified a need for speed reduction. As currently conceptualized the EFTCD requires the pres- ence of a second traffic control person at the flagger station, whose duty is to ask drivers to activate their flashers. Further research on this technique may be desirable to compliment the small-scale Kansas study and evaluate the EFTCD in other driving environments. TRACTOR-TRAILER-TYPE MOBILE BARRIER SYSTEMS Mobile barrier systems are intended to protect workers by iso- lating short-duration work zones from live traffic. As of 2014, two tractor-trailer-type mobile barrier systems were in use in the United States, the Balsi Beam developed by California DOT (Caltrans) and the MBT-1 system developed by Mobile FIGURE 24 Balsi beam (left) and MBT-1 (right) (Caltrans 2004, Mobile Barriers LLC 2014).

44 FIGURE 25 Manitoba gateway assembly for freeway applications (approaching work zone) (Manitoba Infrastructure and Transportation 2013). FIGURE 26 Manitoba gateway assembly for freeway applications (leaving work zone). Sign dimensions are in millimeters (Manitoba Infrastructure and Transportation 2013). for long-term projects. Both provinces have large expanses of prairie terrain where the sense of wide-open space fosters high-speed driving. The assemblies also serve an advance- warning function, providing greater visibility at the work zone approach. The Manitoba and Saskatchewan designs use orange and black stripes on the approach side of the assem- bly (entering the work zone) and white and black stripes on the termination side (leaving the work zone). Figures 25 and 26 illustrate the freeway gateway (two center barricades are used on roadways with wide medians). Figure 27 illustrates the two-lane rural highway version (Manitoba Infrastructure and Transportation 2013). No information about the effectiveness of this treatment was found. CONVERGING OPTICAL DEVICES: OPTICAL SPEED BARS, CHEVRON PAVEMENT MARKINGS, AND RELATED TECHNIQUES Optical speed bars are specialized pavement markings intended to reduce vehicle running speeds. A number of marking pat- terns, such as transverse stripes and transverse chevrons, have been investigated as potential techniques for this purpose, but work zone applications have been rare. The concept of optical speed bars is perhaps most- effectively illustrated by Figure 28 from the British Traffic Signs Manual, which authorizes the use of a series of trans- verse bars at approaches to roundabouts. A series of 90 bars is used on main highways and 45 bars on exit ramps; each bar is 600 mm (24 in.) wide. The spacing of the bars decreases as the driver approaches the hazard, on the premise that the converging pattern will give the driver an increased percep- tion of speed and a corresponding inclination to slow down. For example, at the upstream end of the 45 bar configura- tion the specified spacing between bars is 7.7 m and this decreases progressively to 2.75 m at the downstream end. In the United Kingdom, all ordinary pavement markings are white; however, yellow is used in this application used to draw attention. A variety of optical speed bar patterns have been evalu- ated in permanent (nonwork zone) applications. Although most of the work on this topic has been done in Europe and Japan, two very different marking patterns have been tested in the United States: • A 2001–2003 study reported a 17 mph reduction in 85th percentile speeds on an urban freeway-to-freeway ramp in Milwaukee, Wisconsin, where a complex chev- ron pattern was deployed experimentally (Drakopoulos and Vergou 2003). The pattern, originally used in Japan, combines converging chevrons in the middle of the road- way with transverse bars straddling the wheel tracks, as shown in Figure 29. The study estimated that 3 mph of the reduction was the result of increased traffic volume and the remainder was attributable to the effectiveness of the device. The Milwaukee findings were contradicted by a 2012 evaluation of a similar treatment in Atlanta, which showed that the chevrons had a minimal long-term effect on vehicle speeds, with drivers adjusting back to their previous speeds as they acclimated to the treatment (Hunter et al. 2010). Although there was an initial reduc- tion in speeds, by the ninth month after implementation the speed reduction dropped to less than 1 to 2 mph for the mean speed and most vehicle speed percentiles in the Atlanta study.

45 on a visual simulation suggested that a chevron pattern was perceived as being more effective. • Within the study area three areas were identified, a “lead- ing pattern” with uniformly spaced bars, a “primary pat- tern” of converging bars, and a “work zone pattern” with intermittent uniformly spaced bars. Reductions in mean and 85th percentile speeds were observed by compar- ing speeds upstream of the pattern with speeds in the pattern; the magnitudes were small (about 1 mph), but statistically significant. There was no noticeable diminu- tion of the speed reduction effect over the three-month project duration. Insufficient contrast between the bars and the pavement surface may have contributed to the relatively low reduction in speeds. Speed variance was also reduced. • Both a warning effect and a perceptual effect were observed; that is, both the presence of the bars and the decreasing spacing (in the primary pattern area) con- tributed to the speed reduction, but it was not possible to separate the two effects. The intermittent work zone pattern was not effective in sustaining the speed reduc- tions downstream of the initial speed reduction. As a • A 2006 study explored the use of converging peripheral transverse bars at three sites (Rakha et al. 2006). This MUTCD-approved marking consists of relatively small bars (12 × 18 in.) that straddle the wheel tracks, as shown in Figure 30. The study’s authors found reductions of 5 mph in 85th percentile speeds at an exit ramp near Syracuse, New York. Results on two-lane rural highways were mixed, with a 1 mph reduction at a site in Missis- sippi, but a speed increase at a site in Texas. A 2008 study evaluated the use of peripheral transverse bar markings in a three-lane urban freeway curve located in Milwau- kee near the site studied by Drakopoulos and Vergou (Gates et al. 2008). The study’s authors found lane- specific, short-term 85th percentile speed reductions of 0 to 3 mph; after 6 months, a small additional reduction was observed in one travel direction but not the other. Only one study was identified that explored the use of opti- cal speed bars for work zone applications. It was conducted at a moderate-volume rural freeway site on I-70 in Kansas that had an ordinary speed limit of 70 mph and a work zone limit of 60 mph (Meyer 1999, 2004). A bar pattern was selected for the field deployment, although driver opinion surveys based FIGURE 27 Manitoba gateway assembly for undivided highways (Manitoba Infrastructure and Transportation 2013).

46 Manual stipulates that the skid resistance of the bars should not be less than 55 (DfT 2003). Another study exploring the perceptual effects of converg- ing visual devices was conducted in a New Zealand work zone (Allpress and Leland 2010). The study involved placing a series of 36-in.-tall traffic cones at a work zone approach, with either uniform or converging cone spacing. The work zone was on a two-lane rural highway with a traffic volume of more than 10,000 vehicles per day. The ordinarily posted speed limit was 100 km/h (62 mph); during construction, the speed limit was reduced to 50 km/h (30 mph). As shown in Figure 31, the work zone was hidden from approaching vehi- cles by a curve. At the work zone approach (just downstream of the curve) a total of 16 cones were placed: eight on each side of the lane. In the uniform configuration the cone spac- ing was 2 m (6.6 ft), whereas in the converging configuration the spacing ranged from 3.5 m (11.5 ft) at the upstream end result, the author concluded that the treatment was best suited for locations where a point-speed reduction is desirable. Examples of such locations could include the approaches to ramp terminals, sharp curves, or tempo- rary median crossovers. Three potential concerns have inhibited more widespread use of optical speed bars in the United States. For designs that span the full width of pavement, the possibility that the mark- ings will wear off in the wheel tracks is sometimes cited as a potential issue, although maintenance of the markings is not always necessary for short-term work zone deployments. A related objection is the potential for the markings to be scraped off by snow removal operations; in the work zone context this would apply only to long-duration projects in areas that receive snowfall. Skid resistance is also a concern. In general, these issues can be addressed through proper materials selec- tion and application. For example, the British Traffic Signs FIGURE 28 Optical bar markings for permanent hazard marking from the British Traffic Signs Manual (driving on left) (DfT 2003).

47 SEQUENTIAL AND SYNCHRONIZED WARNING LIGHTS The illusion of motion can be created when discrete stationary lights are flashed sequentially. A small number of studies have investigated the potential for deploying a series of lights along the roadway shoulder that flash at predetermined rates, with the intent of encouraging drivers to reduce speed. Evaluations of the effect on speeds have shown mixed results. Section 6F.63 of the 2009 U.S. MUTCD permits the use of a series of sequential flashing warning lights that may be placed on channelizing devices that form a merging taper in order to increase driver detection and recognition of the merging taper. The manual stipulates that successive flashing of the sequential warning lights shall occur from the upstream end of the merging taper to the downstream end of the merging taper and specifies a flash rate of 55 to 75 times per minute (0.92 to 1.25 Hz) (FHWA 2009). Flash rates in the range of 16–25 Hz must be avoided because of the potential for adverse health and safety effects on persons with photosensitive epilepsy; some individuals may be sen- sitive to rates as low as 3 Hz and as high as 60 Hz (Epilepsy Action 2014). A 2001 study tested a prototype warning-light system composed of a series of interconnected, synchronized indi- vidual flashing warning lights attached to the channelizing drums forming a lane-closure taper (Finley et al. 2001). The lights were timed to produce the perception of a light that “moves” repeatedly in a sequential manner from the begin- ning of the taper to the end of the taper. Two field tests were conducted, one on a low-volume, rural, farm-to-market road with a 65 mph night speed limit (which had already had a lane closure for 6 months) and one on I-10 West, an urban freeway near Houston with a 65 mph speed limit for cars and a 55 mph limit for trucks. In the nighttime field studies, the prototype warning-light system did not significantly affect the speed of vehicles at either test site, but the system encouraged earlier merging at the I-10 site. The study’s authors reported that the flashing warning-light system was perceived positively and was not confusing to the motoring public. A 2010 driving simulator study explored various light pat- terns: asynchronous, moving, and static. Random or steady lights were found to have little effect on driver speed, and to 0.5 m (1.6 ft) at the downstream end. The presence of the cones reduced speeds by 7.8 to 9.5 km/h (4.9 to 5.9 mph) compared with the without-cones condition. With the con- verging arrangement, speeds measured at the downstream end of the cone array were 1.6 km/h (1.0 mph) lower than with the uniform arrangement. Conversely, near the middle of the work zone at a distance 150 m (490 ft) beyond the end of the cone array, the speed with the converging arrange- ment was 1.5 km/h (0.9 mph) higher than with the uniform arrangement. Although the authors concluded that the results were statistically significant, the speed difference between the converging and uniformly spaced cone patterns may have been smaller than the measurement tolerances for the pneu- matic counters used to gather the speed data. FIGURE 29 Converging chevron pattern at a 2014 Wisconsin work zone. FIGURE 30 Peripheral transverse bar markings application diagram from the U.S. MUTCD (FHWA 2009).

48 2.2 mph and the 85th percentile speed decreased by 1.0 mph; however, the speed standard deviation (a measure of speed variation) increased by 0.91 mph. The study’s authors reported an improvement in overall speed limit compliance. With the lights the number of vehicles merging earlier (in the vicinity of the upstream end of the taper) increased at two rural work zones, but decreased at the urban site. synchronized lights did not show statistical significance in reducing driver speed (Khan 2010). A 2011 field study evaluated the lane taper approach speeds with and without sequential warning lights at three Missouri sites with 60 mph work zone speed limits (Sun et al. 2011b). With the sequential lights the mean speed decreased by FIGURE 31 Layout of work zone in New Zealand study (driving on left side of road) (Allpress and Leland 2010).