• Federal Highway Administration >
• Publications >
• Research Publications >
• Human Factors >
• 01103 >
• Ch4.Cfm >
• Highway Design Handbook for Older Drivers and Pedestrians

Highway Design Handbook for Older Drivers and Pedestrians

IV. CONSTRUCTION/WORK ZONES

The following discussion presents the rationale and supporting evidence for Handbook recommendations pertaining to these five design elements (A-E):

A. Design Element: Lane Closure/Lane Transition Practices

Table 42. Cross-references of related entries for lane closure/lane transition practices.

Minimum requirements for safely negotiating a lane closure include an awareness of a decrease in pavement width ahead, and of the direction of the lateral shift in the travel path; a detection of traffic control devices marking the location of the lane drop (beginning of taper); a timely decision about the most appropriate maneuver, taking other nearby traffic into account; and smooth vehicle control through maneuver execution. In the vicinity of a lane closure, the longer the information needs supporting these requirements remain unmet for the least capable drivers within the traffic stream, the less likely is a smooth transition through the work area for all drivers (Goodwin, 1975). The more time that is required for older drivers to prepare and initiate a merging maneuver, the more dense following traffic (including the adjacent lane) is likely to become; this, in turn, will make gap judgments and maneuver decisions at the point of a lane closure more difficult, and will increase the likelihood of erratic vehicle movements resulting in conflicts between motorists.

Relevant alterations in older adults' cognitive-motor processes include: failure to use advance preparatory information (Botwinick, 1965); difficulty in processing stimuli that are spatially incompatible (Rabbitt, 1968); initiation deficit in dealing with increased task complexity (Jordan and Rabbitt, 1977); and inability to regulate performance speed (Rabbitt, 1979; Salthouse, 1979; Salthouse and Somberg, 1982). Stelmach, Goggin, and Garcia-Colera (1987) found that older adults showed disproportionate response slowing when compared with younger subjects, when there was low expectancy for a required movement. When subjects obtained full information about an upcoming response, reaction time (RT) was faster in all age groups. Stelmach et al. (1987) concluded that older drivers may be particularly disadvantaged when they are required to initiate a movement in which there is no opportunity to prepare a response. Preparatory intervals and length of precue viewing times are determining factors in age-related differences in movement preparation and planning (Goggin, Stelmach, and Amrhein, 1989). When preparatory intervals are manipulated such that older adults have longer stimulus exposure and longer intervals between stimuli, they profit from the longer inspection times by performing better and exhibiting less slowness of movement (Eisdorfer, 1975; Goggin et al., 1989). Since older drivers benefit from longer exposure to stimuli, Winter (1985) proposed that signs should be spaced farther apart to allow drivers enough time to view information and decide which action to take. Increased viewing time will reduce response uncertainty and decrease older drivers' RT.

In focus group discussions consisting of 81 drivers ages 65 to 86, pavement width transitions were identified as sources of difficulty by 50 percent of the participants (Staplin, Harkey, Lococo, and Tarawneh, 1997). The drivers participating in these discussions suggested longer merging areas to give them more opportunity to find a safe gap and the use of multiple warning signs to allow them to plan their maneuver at an earlier point upstream. Use of multiple signs to give advance notice of downstream work zones and of required maneuvers was also offered as a desired change by older drivers participating in an earlier focus group (Staplin, Lococo, and Sim, 1990).

Lyles (1981) conducted studies on two-lane rural roads to evaluate the effectiveness of alternate signing sequences for providing warning to motorists of construction and maintenance activities that required a lane closure. The signs tested included a standard MUTCD warning sequence, the same sequence augmented with continuously flashing warning lights on the signs, and a sequence of symbol signs (WORKER and RIGHT LANE CLOSED). The most effective sign sequence was one that was flasher augmented; this treatment was twice as effective as similar signs with no warning lights in slowing vehicles in the vicinity of the lane closure.

The use of word messages on signs in highway work areas raises sign legibility issues for older drivers. In research conducted to improve the legibility of the RIGHT/LEFT LANE CLOSED and ROAD CONSTRUCTION series signs using test subjects in three age groups (18-44, 45-64, and 65 and older), Kuemmel (1992) concluded the following for black on orange (negative contrast) signs: (1) signs that increased both letter size and stroke width (SW) while maintaining or increasing the standard alphabet letter series resulted in the best improvement; (2) increasing letter size while decreasing the alphabet series (e.g., from C to B) reduces sign legibility, particularly at night; (3) the use of letter series E, with its 21-percent increase in SW-to-letter height over 200-mm (8-in) series C letters, appears to overcome the problems of irradiation (or overglow phenomenon) with high intensity retroreflective materials, thus increasing night legibility; (4) the legibility distance of the ROAD CONSTRUCTION signs can be increased by changing the word "construction" to "work," and increasing the letter size from 175-mm (7-in) series C to 200-mm (8-in) series C; and (5) for the RIGHT LANE CLOSED series, use of symbol signs will have to supplement word legend signs, and for the CENTER LANE CLOSED series, redundant signs will have to be employed if a 1,200-mm (48-in) maximum sign size is to be maintained. The author pointed out that the minimum required visibility distance (MRVD) for both signs is 101 m at 88 km/h, and 112 m at 104 km/h (331 ft at 55 mi/h and 369 ft at 65 mi/h). The legibility distances obtained in this study for the current standard construction work zone signs ranged from 198 m (650 ft) for the best observers to 43 m (140 ft) for the worst observers. In addition, 85th percentile values were closer to the minimum legibility distances than they were to the maximum legibility distances. This finding reinforces the need for redundant signing during the approach to a work zone.

Recent studies have centered on the use of fluorescent orange signs for work zone applications, particularly as their increased conspicuity may benefit older drivers with diminished visual capabilities by providing longer detection distances. Jenssen, Moen, Brekke, Augdal, and Sjøhaug (1996) state that flourescent materials have the potential to increase daytime conspicuity through increased contrast, while the nighttime brightness is sustained through a microprismatic retroreflective system. Burns and Pavelka (1995) explain that the high visibility of fluorescent materials is due to their ability to absorb energy in the near ultraviolet and visible region of the electromagnetic spectrum, and then to re-emit the energy as longer wavelength, visible light. Conventional colorants don't have this property. During the daylight hours from dawn to dusk, there is always sufficient solar energy to elicit light emission from fluorescent materials, irrespective of the cloud cover. Therefore, fluorescent colors maintain a significant daytime visibility advantage over ordinary colors in all types of weather.

Jenssen et al. (1996) conducted a controlled field study using 35 younger subjects (ages 18 to 25) and 44 older subjects (ages 55 to 75) to compare the detection distance, color recognition distance, and the legibility distance of fluorescent signs to traditional signs, under daytime and nighttime conditions. In this study, subjects sat in an open-ended container on a railway car that moved at a speed of 15 km/h (9 mi/h) along a set of unused railroad tracks. Subjects used a response form and were trained to look for specific signs. They activated a response lever that sent a signal to a distance measuring computer, and then recorded what they observed in the categories provided on their response forms for sign detection, shape, color, symbol, and text. The signs of interest for this discussion were those with an orange background and black text.

Signs with fluorescent orange Type VII retroreflective sheeting were compared to signs with standard orange Type VII retroreflective sheeting, signs with standard orange Type III high intensity grade retroreflective sheeting, and signs with standard orange Type I engineering grade retroreflective sheeting. In the daytime, only signs with Type VII optics were used. The Norwegian town names "LENSVIK," "LAKSVIK," or "LEKSVIK," appeared in randomized order on the signs. The height, angle, and distance of the signs relative to the track were adjusted according to standards for Norwegian two-lane roadways, to ensure realistic viewing positions. Signs were always placed on the right side of the track. For nighttime trials, original headlights for a Volkswagen Golf type 1 vehicle (placed on the railcar at the standard vehicle headlight orientation) were used.

Detection, shape recognition, color recognition, and content recognition were accomplished at significantly greater distances for fluorescent orange retroreflective signs than for the standard orange retroreflective signs, for both younger and older subjects under daytime conditions. The mean detection distance for all subjects during daytime conditions for the fluorescent orange retroreflective signs was 822 m (2,697 ft), compared to 783 m (2,569 ft) for standard orange retroreflective signs. This 40 m difference in detection distance was statistically significant. The difference in mean detection distance was larger for the older subjects (50 m) than for the younger subjects (22 m); however, both age groups demonstrated significantly longer detection distances when viewing the fluorescent orange retroreflective signs. The mean shape recognition distance across all subjects during the daytime was 744 m (2,441 ft) for the fluorescent orange retroreflective signs and 651 m (2,136 ft) for the standard orange retroreflective signs. Younger subjects were able to correctly recognize the shape of fluorescent signs at an average distance that was 100 m longer than for the standard signs, and older subjects demonstrated an average shape recognition distance difference of 59 m. Fluorescent signs also showed significantly longer correct color recognition distances (584 m [1,916 ft] across age groups) than standard signs (469 m [1,539 ft] across age groups). Color recognition distances were 130 m longer for younger subjects, and 106 m longer for older subjects when viewing the fluorescent signs during the daytime compared to the standard signs. In terms of legibility distances during daytime, across all subjects, the fluorescent signs significantly outperformed the non-fluorescent signs, with a difference in mean legibility distance of 13 m (43 ft).

At nighttime, there were no significant differences in detection, shape recognition, color recognition, or contents recognition distances between fluorescent orange retroreflective signs with Type VII sheeting and standard (non-fluorescent) orange retroreflective signs with Type VII sheeting, for either age group. However, comparisons between the three types of retroreflective sheeting indicated that the signs with Type VII sheeting produced detection distances that were 42 m (138 ft) longer than the signs with high intensity grade sheeting, and 62 m (203 ft) longer than the signs with engineering grade sheeting, for the older drivers. For the younger drivers, detection distances for the signs with Type VII sheeting were 19 m (62 ft) longer than those produced by the signs with high intensity grade sheeting, and 36 m (118 ft) longer than those produced by the signs with engineering grade sheeting.

The mean sign detection distance, shape recognition distance, color recognition distance, and contents recognition distance are presented in table 43, as a function of subject age group and lighting condition (day vs. night) for the signs with standard orange Type VII sheeting and for the signs with fluorescent orange Type VII sheeting.

Burns and Pavelka (1995) conducted a field study using 14 drivers ages 19 to 57 (median age: 40 years), to compare the visibility and conspicuity of durable retroreflective fluorescent sheetings (orange, red, yellow, and yellow-green) to the same color standard highway sheeting (orange, red, yellow, yellow-green, and green), at midday and at dusk. Circular targets with an area measuring 9.3 cm² (0.01 ft²) were viewed in pairs (one fluorescent and one standard highway color sign) against a 1.2-m by 1.2-m (4-ft by 4-ft) background. The background consisted of a complex camouflage pattern made up of 23 percent light green, 34 percent green, 25 percent brown, and 18 percent black. The targets were placed 0.3 m (1 ft) apart, and were viewed at four distances during the daytime (120 m [394 ft], 90 m [295 ft], 60 m [197 ft], and 30 m [98 ft]). At dusk (15 min before sunset, and 15 min after sunset), signs were viewed only at 30 m. Subjects viewed the target pairs while seated in a vehicle that had the headlights turned off. An electronic shutter system provided a viewing duration of 2 s. After each target pair was viewed, subjects provided responses to indicate: (1) number of targets detected (0, 1, or 2); (2) target location (left or right); (3) target color (left color and right color); and (4) attention-getting value (was one target perceived more easily, or did one target attract your attention more than the other?).

Table 43. Subject performance as a function of sheeting type (fluorescent orange Type VII vs. standard orange Type VII), as a function of subject age group and lighting condition.
Source: Jenssen, Moen, Brekke, Augdal, and Sjøhaug, 1996.

During the daytime, the durable fluorescent targets evaluated in the Burns and Pavelka (1995) study were detected with a higher frequency (close to 100 percent) and at greater distances than the standard highway colors. At midday (facing north on an overcast day) from a distance of 120 m, 93 percent of the subjects were able to detect the fluorescent orange targets; however, only 43 percent of the subjects could detect the standard orange targets at this distance under the same lighting conditions. At 90 m, 100 percent of the subjects detected the fluorescent signs, compared to 92 percent who detected the standard orange signs. At dusk (15 min after sunset) at a distance of 30 m, the fluorescent orange signs were detected by 96 percent of the subjects and the standard orange signs were detected by 85 percent of the subjects.

The fluorescent signs also showed greater color recognition than the standard highway signs at all distances. During midday (overcast facing north), the fluorescent orange signs were correctly identified by 61 percent of the subjects at 120 m, 58 percent of the subjects at 90 m, 86 percent of the subjects at 60 m, and 82 percent of the subjects at 30 m. By comparison, the standard orange signs were identified correctly by 7 percent, 23 percent, 64 percent, and 93 percent of the subjects at the same distances. At dusk (15 minutes before sunset) at 30 m, the fluorescent orange signs were correctly identified as orange by 74 percent of the subjects, compared to 62 percent of the subjects for the standard orange signs.

The fluorescent orange sign in each pair of viewings was subjectively determined to be more conspicuous (more attention-getting) than the standard orange highway sign, at 30 m, under all lighting conditions (midday, 15 min before sunset, and 15 min after sunset). Luminance measurements were taken of the targets and their backgrounds, so that a contrast ratio could be calculated. The fluorescent signs always produced a higher contrast ratio than the standard signs. The contrast ratios for the fluorescent orange and standard orange signs are shown in table 44 below, under the various, natural lighting conditions utilized in the study.

Table 44. Luminance contrast ratio ([L_t-L_b]/L_b) under different lighting conditions, for
fluorescent orange and standard highway orange signs.
Source: Burns and Pavelka, 1995.

The authors conclude that fluorescent orange signs are more conspicuous than standard highway orange sign colors during the daytime (as were the other fluorescent colors); are detected with higher frequency; and are recognized with greater accuracy at farther distances. Fluorescent signs provide a greater contrast with the background scene, and therefore should be considered as a countermeasure to address problems that older drivers have in the detection and recognition of traffic signs when viewed against a cluttered background.

Finally, Hummer and Scheffler (1999) conducted a field study at seven long-term work zones in North Carolina with left lane drops on multilane highways, to determine whether the increase in the conspicuity of fluorescent orange signs leads to positive operational changes in driver behavior. All seven sites were left lane drops on four-lane highways (with standard lane and shoulder widths), with the following sequence of orange signs (in pairs, with one sign on each side of the highway):

• Two fluorescent BEGIN WORK ZONE text message signs located 0.6 to 1.9 km (0.4 to 1.2 mi) from the taper.
• Two text message LEFT LANE CLOSED AHEAD signs located 0.4 to 1.1 km (0.25 to 0.68 mi) from the taper.
• Two symbol message LEFT LANE CLOSED AHEAD signs located 0.16 to 0.5 km (0.1 to 0.31 mi) from the taper.
• Two text message LEFT LANE CLOSED signs at the beginning of the taper.

Six sites had 90 km/h (55 mi/h) speed limits and wide grassy medians, and one site had a 105 km/h (65 mi/h) speed limit. Before the study was conducted, both treatment and control sites existed as work zones with standard orange signs, except for the first sign in the pair, which was fluorescent. In the "before" period, five operational measures were collected on this standard set of signs. In the "after" period, the standard orange signs were replaced with fluorescent signs (same size and message) at the "treatment sites,"and the same measures were recorded at these treatment sites, as well as at the "control sites" where the standard signs were left in place. Two weeks elapsed before data were collected in the "after" period, to eliminate novelty effects. The operational measures included: (1) traffic conflicts (one vehicle brakes or swerves to avoid hitting another); (2) percentage of all vehicles in the left lane; (3) percentage of trucks in the left lane; (4) mean speed; and (5) speed variance. These data were collected at the beginning of the taper, at the BEGIN WORK ZONE sign, and at the midpoint of the approach.

With regard to traffic conflicts, a reduction from 153 to 136 at the treatment sites (with fluorescent signs) was observed in the before and after periods, respectively, while an increase from 160 to 187 was observed at the control sites (with standard signs) in the before and after periods. This reduction in conflict frequency was statistically significant, when sites without potential confounding factors were removed from the analysis. Regarding the number of vehicles in the left lane, there was a significant reduction in the percentage of vehicles at the midpoint of the work zone approach at the sites with fluorescent signs (more than a 5 percent reduction, or 100+ fewer vehicles); similarly, an increase in the percentage of trucks that moved out of the left lane before the midpoint (16 more trucks, or 30 percent more than expected) and at the taper (12 more trucks than expected) was observed at the sites with fluorescent signs, compared to sites with the standard orange signs.

Differences in mean speeds were not statistically significant; speeds increased by approximately 1.6 km/h (1 mi/h) at the midpoint and taper of treatment sites, and decreased by the same amounts at the control sites. However, speed variance decreased at the midpoint and at the taper of the treatment sites (with fluorescent signs), relative to speeds monitored at the control sites (with standard orange signs).

Hummer and Scheffler (1999) state that the operational changes documented during their study would translate to fewer collisions in work zones that display fluorescent orange signs compared to those that display the standard orange signs. Although the overall reduction in traffic conflicts in this study was small (approximately 7 percent), they recommend that agencies use fluorescent orange sheeting on warning signs in work zones similar to those studied, as well as for work zones where warning drivers is as critical or more critical than it was in the current study. These would include long-term work zones where there is flagging, temporary traffic signals, sharp lane shifts, and changed merging patterns, as well as in many temporary and moving work zones. Hummer and Scheffler (1999) state that fluorescent orange sheeting costs only a few dollars more per sign installation than standard orange sheeting, and even if work zone collision frequencies declined by only one or two percent, the benefit-to-cost relationship would be substantial.

Next, a number of studies have been performed to determine the effectiveness and motorist comprehension of static signs and changeable message signs (CMS's)--also referred to as variable message signs (VMS's)--for lane closures. A general indication of the importance of CMS's to accomplish lane control in advance of work zones is provided by a field study on a four-lane section of I-35 in San Antonio conducted by Dudek, Richards, and Faulkner (1981) to evaluate the effects of CMS messages on lane changes at a work-zone lane closure. The measure of effectiveness used to evaluate the CMS was the percentage of vehicles that remained in the closed (median) lane as traffic progressed toward the cone taper. The results indicated that the CMS did encourage drivers to vacate or avoid the closed lane, compared with driver responses at the same site without use of the CMS. The percent volumes in the closed lane were significantly lower when a lane-closure message was displayed than during periods when the sign was blank. Specifically, there was a 46 percent greater reduction in the lane volume attributable to the CMS.

During the conduct of field studies for NCHRP project 3-21(2), the relative proportions of traffic in the through and closed lanes approaching construction lane closures were observed for a sample of more than 196,500 vehicles (Transportation Research Board, 1981). Data gathered in Georgia, Colorado, and California were used to compare these lane distributions between baseline (no CMS) conditions and various CMS applications. A fourth data set, gathered in South Carolina, was used to determine relative effects between certain CMS message alternatives (i.e., speed and closure, speed and merge, closure and merge advisories), and various placement configurations (i.e., one CMS at 610 m [2,000 ft] in advance; or one CMS at 1,207-m [3,960-ft] advance placement; or two CMS devices, one at each advance location; or one CMS placed at 1,207 m [3,960 ft] in advance of the taper and an additional arrow panel at the 610-m [2,000-ft] location). Findings indicated increased preparatory lane change activity, smoother lane-change profiles, and significantly fewer "late exits" (exit from a closed lane within 30.5 m [100 ft] of closure) in locations where a CMS was applied at the 1,207-m (3,960-ft) advance location and an arrow panel at the 610-m (2,000-ft) location.

Additional studies of flashing arrow panels at construction sites have shown that they are effective in shifting approaching traffic out of a closed lane (Bates, 1974; Shah and Ray, 1976; Graham, Migletz, and Glennon, 1978; Bryden, 1979; Faulkner and Dudek, 1981). These studies found that arrow panels were effective because they promote early merging into the open lane and fewer vehicles remained in the closed lane at the start of the lane-closure taper. A basis thus exists to assert that a CMS used to give advance notice of the need to exit a lane, followed by the application of an arrow panel, would be of clear benefit to drivers with diminished capabilities resulting from aging, inattentiveness, or transient impairment (e.g., due to fatigue, alcohol, or drugs). While the specific location of the arrow panel in this approach should be consistent with the signing sequence indicated in the MUTCD Part 6H (figure 6H-33 for divided highways), placement at the beginning of the taper is suggested by the findings reported above.

Mace, Finkle, and Pennak (1996) conducted a static and a dynamic field study to determine the minimum and optimum lamp intensities needed for arrow panel legibility (left arrow or chevron vs. right arrow or chevron presentation) during the day, and minimum and maximum intensity for nighttime operations to minimize glare effects. The authors cite the work of Faulkner and Dudek (1982), who found that sight distances to arrow panels (AP's) influences driver behavior, such that when AP's are used too far in advance of a lane closure, (e.g., 1,219 m [4,000 ft], drivers tend to return to a vacated lane. Also, if sight distance is less than 457 m (1,500 ft), an advance supplemental AP is desirable. While no data exist to document problems in the safe use of AP's by seniors, the following recommendations (see table 45) suggested by Mace et al. (1996) for arrow panel lamp intensity provide a useful reference for practitioners. These values ensure visibility for DSD's of 457 m and 283 m (1,500 and 930 ft), for high-speed and low-speed roadways, respectively.

Table 45. Recommended minimum on- and off-axis lamp intensities of arrow panels to ensure daytime visibility by older drivers, and the maximum intensity recommended for nighttime operations to ensure safe levels of discomfort and disability glare for older drivers, for high-speed (73 km/h [45 mi/h]) and low-speed (<73 km/h [45 mi/h]) roadways.
Source: Mace, Finkle, and Pennak, 1996.

A questionnaire also was completed during the conduct of NCHRP project 3-21(2), by 489 subjects ranging in age from under 20 to 80 to gather measures of driver detection, recognition, and comprehension of the CMS devices. Twenty percent of the drivers were age 60 and older. Five tested message conditions were: (1) speed and closure advisory (MAX SPEED 45 MPH/RIGHT LANE CLOSED); (2) speed and merge advisory (MAX SPEED 45 MPH/MERGE LEFT); (3) merge and closure advisory (RIGHT LANE CLOSED/MERGE LEFT); (4) speed advisory only (SLOW TO 45 MPH); and (5) closure advisory only (RIGHT LANE CLOSED AHEAD). Drivers consistently reported that the speed advisory and lane closure message combination was most helpful, was the easiest to read, best met their information needs, and would be most likely to cause them to change lanes early and reduce speed.

A recent human factors laboratory study was conducted to determine which CMS message alternatives would be most likely to enhance motorists' compliance with lane control messages in work zones (Gish, 1995). The subjects were divided into two age groups consisting of 24 subjects each: the youngest drivers had a mean age of 23.1 years (range: 16-33), and the oldest drivers had a mean age of 70.2 years (range: 65-84). The results of this study indicated that older drivers were more likely to reduce their speed and change lanes than the younger drivers, and that both older and younger drivers' compliance with lane-change messages was strongly influenced by surrounding vehicles and by the visibility of the lane closures themselves, which exerts a strong influence on message credibility. Other factors, such as traffic density, static displays, and merge arrows (arrow panels), influence driver compliance with CMS messages. To optimize lane-change compliance, Gish (1995) recommended that static displays, merge arrows, and other devices be used in addition to CMS messages. A need to study the long-term effectiveness on nonstandard messages was also indicated, and potential improvements in work-zone safety and operations through the use of condition-responsive (real-time) traffic control systems that provide continuously updated information to motorists (for enhanced credibility) were identified.

B. Design Element: Portable Changeable (Variable) Message Signing Practices

Table 46. Cross-references of related entries for portable changeable (variable) message signing practices.

The effectiveness of changeable message signs (CMS's), gauged in terms of observable driver behaviors that traffic management procedures are designed to elicit, rests upon a set of reasonably well-understood human factors. A motorist information system must be rational, relevant, and reliable. Driver sensory/perceptual and cognitive capabilities must be thoughtfully considered to ensure that a message will be acquired and then understood, recalled, and applied by the driver within a desired timeframe; the message must seem to clearly apply to the driver and to reflect current conditions to be credible; and it must be accurate in describing what the driver experiences downstream. The credibility of a highway advisory message certainly depends in part upon a presentation strategy that is "rational," but it also must be perceived to be relevant to the individual motorist, and reliable to the point of being virtually error-free. Reliability requirements--being dependent on real-time data on operations as input to the traffic control system--are most difficult to meet, but probably the most important if high rates of compliance in drivers' vehicle control decisions are ever to be realized.

A motorist's ability to use highway information is governed by: (1) information acquisition, or how well the source can be seen or heard; and (2) information processing, or the speed and accuracy with which the message content can be understood, and its ease of recall by the motorist after message presentation is completed.

In the acquisition of CMS information, a visual task, the key factors are: (1) its conspicuity, or "attention-getting value" to the motorist; (2) the size, brightness (contrast), stroke width-to-height ratio, and spacing of individual characters of text, which together determine the legibility of the message; (3) the placement of the CMS device--overhead versus one side versus both sides of the highway--which affects its likelihood of being blocked from a motorist's view by other vehicles, as well as the "eyes away from the road" time required to fixate upon the message; and (4) the exposure time, or available viewing time, of each message phase presented on a CMS.

Conspicuity is generally not a problem for any type of CMS under low traffic volumes, although under high volumes with a significant mix of heavy vehicles, a motorist may fail to notice a roadside device because of obscuration. Good conspicuity is achieved by overhead devices under all conditions. The attention value of a CMS display can be maximized by flashing operations, but this also works against information acquisition by reducing exposure time and legibility; this strategy is thus uniformly discouraged for an entire message. In rare circumstances, for a unit of information deemed particularly critical by the highway authority, the flashing of a single text element within a message at a slow rate may be justified. The use of flashing text may help bring the sign to the attention of an older driver who has a reduction in his/her useful field of view and may otherwise fail to notice the sign. If it is standard policy to leave the signs blank, then the mere display of a message will capture the driver's attention. However, if the CMS in question always has some type of message displayed, then slowly flashing (e.g., two cycles per phase) the problem statement line only may be warranted to attract attention. A preferred strategy under such circumstances would be to activate a flashing warning light separate from, though clearly attached to, the CMS.

The legibility of a CMS is influenced by the same factors influencing character and message legibility of static signs, including the key factor of driver visual performance capability. Letter acuity declines during adulthood (Pitts, 1982) and older adults' loss in acuity is accentuated under conditions of low contrast, low luminance, and where there is crowding of visual contours (Sloane, Owsley, Nash, and Helms, 1987; Adams, Wong, Wong, and Gould, 1988). In any event, the legibility for current CMS's is determined primarily by the technology and the device configuration (numbers of rows, characters per row, and number, size, and spacing of pixels per character) as fabricated by a given manufacturer, and for all practical purposes can be treated as a fixed factor--modified by environmental considerations--in considering whether a particular system as implemented in the field will meet motorists' needs.

For any given speed, older drivers' needs dictate a legibility distance that permits the entire CMS message to be read twice in its entirety. As a general rule, at least 305 m (1,000 ft) of legibility distance for a motorist with 20/40 visual acuity should be provided on a 88-km/h (55-mi/h) facility. Of the studies that assessed various character matrix forms (number of elements per character cell), most found a 7 x 9 element matrix to be necessary when using lowercase letters, because of the descenders and ascenders, but a 5 x 7 font was generally deemed acceptable with uppercase-only lettering. The MUTCD specifies a minimum legibility requirement of 200 m (650 ft) under both day and night conditions for Portable Changeable Message Signs. Given that the most common format for a portable sign is 450-mm (18-in) tall characters arranged in three lines of eight characters, this provides for a legibility distance of 0.44 m/mm (36 ft/in) of letter height. Thus, letter sizes of at least 450 mm (18 in) should be used to accommodate older drivers' diminished visual acuity. Other variables found to significantly effect CMS legibility for older observers are font, letter width-to-height ratio, contrast orientation, letter height, case, and stroke width (Jenkins, 1991; Mace, Garvey, and Heckard, 1994). The most consistent finding across studies evaluating CMS design elements was that the results found for older drivers were quantitatively but not qualitatively different from those of their younger counterparts. That is, if a manipulation of a variable resulted in an improved score for younger observers, it almost invariably improved older observer performance.

Garvey and Mace (1996) conducted several laboratory and controlled field studies to determine optimum legibility requirements of CMS's, particularly for older drivers. The laboratory studies included 24 "young" subjects ages 16 to 40 (mean age: 26.6); 25 "old" subjects ages 62 to 73 (mean age: 67.9); and 21 "old-old" subjects age 74 and older (mean age: 77.2). The first laboratory study used a CMS simulator that was programmed on a PC-compatible computer, simulating nighttime viewing conditions. Only positive contrast signs (light letters against a dark background) were used. The objectives were: (1) to determine the optimum width-to-height ratio (W:H) and stroke-width-to height ratio (SW:H); (2) to identify the CMS font that produced the smallest size legibility thresholds; and (3) to determine the effect of color on legibility. Six different sizes of each sign were evaluated. The dependent variable in the study was the threshold size at which a character became legible, which was converted into a legibility index (LI) reported in m/cm (ft/in). Seven combinations of CMS matrix size, W:H, and SW:H combinations were evaluated to determine the optimum character legibility, as shown in table 47.

Table 47. Variables evaluated by Garvey and Mace (1996) using a CMS simulator to determine the optimum character legibility.

Results indicated that for all conditions, the younger group performed significantly better (smaller letter size required for legibility) than both older groups, and the "old" group performed better than the "old-old" group. The authors indicated that, generally, what worked well for one age group worked well for all ages. Across all age groups, increasing the width-to-height ratio (W:H) of a character from 0.7 to 1.0 increased the legibility index (LI) by 0.84 m/cm (7 ft/in). This provides an advantage of 38 meters of legibility for the wider letter when using a 46-cm (18-in) letter height, or 1.5 s at 89 km/h (55 mi/h). A significant stroke-width-to height (SW:H) effect was also found. For the narrow letters (W:H = 0.8), a thinner stroke performed better than a wider stroke by 0.48 m/cm (5 ft/in). This effect was not significant with wider letters. There were no significant differences in legibility index as a function of matrix density. Therefore, for uppercase letters, increasing the number of elements beyond the standard 5 x 7 format did not improve legibility. The authors state that a typical CMS font with a W:H of 1.0 and a SW:H of 0.13 is optimal for the three age groups studied, from the median to the 85th percentile observer. Their data indicate that the 85th percentile old-old observer was capable of reading such a letter at the LI typically expected of CMS's (4.2 m/cm or 35 ft/in).

In another laboratory study using the same subjects and test apparatus, Garvey and Mace (1996) found that the fonts typically used by CMS manufacturers performed well, with the exception of "double stroke" characters within a 5 x 7 character matrix. A double-stroke font provided a LI of 5.2 m/cm (43 ft/in) for young observers compared to 6.6 m/cm (57 ft/in) for the typical CMS font. For old-old observers, the double stroke font provided a LI of 3.8 m/cm (32 ft/in) compared to the typical CMS font that provided 4.6 m/cm (38 ft/in). Across all age groups, the double-stroke font resulted in a decrement in LI of 1.2 m/cm (10 ft/in).

In the final laboratory study of CMS character legibility, Garvey and Mace (1996) found significant effects of contrast orientation on letter legibility. Positive-contrast stimuli (lighter colored letters on a dark background) produced a LI of 1.4 m/cm (12 ft/in) higher than negative-contrast stimuli (dark letters on a lighter background). This improvement is equal to an additional 67 m (220 ft) of legibility distance for a 450-mm (18-in) letter height, or 2.75 s at 89 km/h (55 mi/h). White-on-black signs performed similarly to yellow-on-black signs, except for the highest-percentile old-old group, where yellow-on-black signs were significantly better than white-on-black. Red-on-black signs performed as well as the other two colors for the young observers, but were found to be significantly less legible than yellow or white on black signs for both groups of older observers. The authors point out that the reduced performance of the color red for older subjects is likely due to its lower luminance, and as people age, they become more sensitive to changes in target luminance. The obtained LI by sign color and observer age and percentile is shown in table 48.

In a dynamic field study, Garvey and Mace (1996) employed older and younger drivers to evaluate legibility distance and detection distance of six portable CMS's. Participants included 33 "young" subjects ages 19 to 40; 25 "old" subjects ages 59 to 72; and 26 "old-old" subjects ages 73 to 82. Other independent variables included contrast orientation (positive or negative); character height (450 mm [18 in] or 1,050 mm [42 in]); lighting condition (backlit, frontlit, overcast, or rain); character luminance--day (350, 570, 850, or 1,200 cd/m²); character luminance--night (30, 80, 130, 200, 570, 1,200 cd/m2); inter-letter spacing--night (single or double); and sign lighting--night (internal vs external or backlight vs LED).

Significant findings in the field study included the following:

• At night, positive contrast messages (yellow on black) produced significantly longer legibility distances, representing a 29 percent improvement over negative contrast messages (black on yellow). The mean legibility distance for positive contrast messages was 152 m (497 ft), and the mean legibility distance for negative contrast messages was 118 m (386 ft). The "old-old" group showed significantly shorter legibility distances compared to the "young" and "old" groups, which were not significantly different from one another.

• Increasing luminance during daytime up to 850 cd/m² produced significantly longer legibility distances; however, increasing the luminance from 850 to 1,200 cd/m² did not significantly increase legibility distance. At night, the effects of increasing luminance were random, with the lowest and highest luminances both producing legibility distances of approximately 245 m (800 ft). Also, there was no significant interaction between character luminance and age group. Note: Important guidance on procedures for valid measurement of CMS character luminance is provided by Garvey and Mace (1996).
  

Table 48. Legibility index obtained by Garvey and Mace (1996) in a laboratory study of CMS
character legibility, as a function of driver age, sign color, and percent of drivers accommodated.

1 m/cm =8.33 ft/in

Next, the "target value," legibility, and viewing comfort of light-emitting diodes (LED's) and fiber-optic CMS technologies were compared with flip-disk and conventional overhead guide signs in a field study conducted by Upchurch, Baaj, Armstrong, and Thomas (1991). Younger (ages 18 to 31) and older (ages 60 to 79) subjects in this study demonstrated mean daytime target values for fiber-optic, LED, and flip-disk technologies that all were significantly better (longer) than the values for conventional overhead signs. Under nighttime conditions, however, the poorest performance (shortest distances) were demonstrated by both age groups for the flip-disk technology, falling below the conventional sign values as well. The fiber-optic and LED signs again exceeded the conventional signs, based on nighttime mean target value, with the fiber-optic technology showing a slight superiority for older drivers. Under backlight (sun behind sign) and washout (sun behind driver) conditions, target values for all sign types decreased substantially and the differences among sign types diminished, but the fiber-optic technology still resulted in the best overall performance, across age groups.

Legibility distance results tended to favor the conventional signs, followed by the fiber-optic signs, LED signs, and flip-disk technology. Mean daytime legibility distances for each sign type in this study were as follows: fiber-optic--0.74 m/mm (61 ft/in); LED--0.51 m/mm (42 ft/in); flip-disk--0.47 m/mm (39 ft/in); and conventional--1.07 m/mm (88 ft/in). Under nighttime conditions, the conventional signs again could be read at the longest mean distances, followed closely by the fiber-optic and LED signs, with the flip-disk technology showing the poorest performance. Backlight conditions favored the fiber-optic technology, and washout conditions favored the conventional signs; in both cases, however, the flip-disk technology resulted in the shortest legibility distances. Using a threshold for minimal acceptable legibility distance of 191 m (628 ft), the study concluded that flip-disk signs are deficient under all conditions except midday daytime viewing, LED signs are deficient under both backlight and washout sun conditions, and fiber-optic signs are deficient only with the sun glare present under backlight conditions.

Mean discomfort ratings were consistent with these patterns of results. Fiber-optic and conventional signs were assigned the best (lowest discomfort) ratings under daytime conditions, by younger and older drivers alike. LED signs caused slightly more discomfort for older subjects, and flip-disk signs resulted in the highest discomfort ratings, especially for older drivers. Under nighttime conditions, only the flip-disk technology resulted in high discomfort ratings. Discomfort ratings were more even, and much higher, across sign types under backlight conditions where the sun was behind the sign, though flip-disk signs still were rated the worst by both age groups. Under washout conditions, subjects reported little discomfort for either the fiber-optic or conventional signs, but much greater and roughly equivalent levels of discomfort with the LED and flip-disk technologies.

Table 49 contains legibility distances from the Upchurch et al. (1991) study. For older drivers, the legibility distances are lower due to the well-documented degradation of visual performance with age. Unfortunately, this is the only study that has assessed legibility distances for older observers. The legibility distances for conventional bulb matrix and LED/flip-disk hybrid CMS's were estimated from the results of the Upchurch data and data cited in Dudek (1991).

The older driver legibility distances in table 49 should be assumed to represent the legibility distances for the various types of technology represented. This ensures that the needs of older drivers have been met. The results suggest that flip-disk CMS's should not be used at night along roadways where average speeds reach or exceed about 88 km/h (55 mi/h).

Table 49. Day and night predicted legibility distances for various sign technologies.
Source: Upchurch et al., 1991.

* Legibility distance of this technology decreases over time, because as LED's age, they become less bright.

Although the bulb matrix CMS was assessed by Upchurch et al. (1991), no legibility distances for that sign were reported. Legibility distances for this type of CMS have been obtained; however, it is unknown whether any older observers have been used in assessing legibility distances. Dudek (1991) cited a study in which bulb matrix CMS's provided legibility distances of 244 m (800 ft) during the day and 229 m (750 ft) at night. These distances are similar to the legibility distances obtained by Upchurch et al. (1991) for LED-type CMS's using younger observers. Until psychophysical data can be obtained for older observers viewing bulb matrix signs, the legibility distances for older observers are assumed to be roughly 204 m (671 ft) during the day and 173 m (569 ft) at night. These estimates are based on applying the ratio of older-to-younger legibility distances for the LED-type display.

There are also a number of hybrid CMS's that were not included in the Upchurch et al. study. Hybrid CMS's apply various combinations of sign technologies listed in table 49 within a single sign. Product literature for one manufacturer's hybrid LED/flip-disk sign states that the sign provides 274 m (900 ft) of legibility distance during the day and greater than 274 m (900 ft) at night, using character heights of 450 mm (18 in). Unfortunately, the methods used to obtain these legibility distances are unknown. Since the sign uses the reflective flip-disk technology during daytime and the LED's at night, the legibility distances for older observers for the daytime flip-disk in table 49 (203 m [667 ft]) should be used as a more realistic estimate of legibility distance with LED/flip-disk hybrids. For nighttime viewing, use the nighttime LED legibility distance (183 m [602 feet]) in table 49.

CMS placement affects information acquisition under heavy traffic conditions where a center lane driver's view of a roadside device may be obscured for lengthy intervals. If a facility has more than two lanes, a consideration may be given to placement of a portable CMS in the median--space permitting and where glare from opposing vehicles is absent or minimal due to a large glare angle--rather than on the right shoulder, since lane control practices for heavy trucks are common throughout many corridors.

A motorist's reading time for a CMS message dictates the required exposure time at a given speed. Exposure time is the length of time a driver is within the legibility distance of the message. The minimum recommended exposure time per page (phase) for a three-line CMS is 3 s, aside from a consideration of any particular set of driver characteristics. However, while some jurisdictions have selected briefer exposure times, the increasing numbers of older drivers on limited-access highways makes an even stronger case for the 3-s minimum per page. Reading time is the time it actually takes a driver to read a sign message. In instrumented vehicle studies conducted in light traffic with familiar drivers on a rural freeway, reading times averaged

1 to 1.5 s per unit of information (Mast and Ballas, 1976). Reading times under "loaded" driving conditions would be higher, such as under extreme geometry, heavy traffic volumes, large volume of truck traffic, traffic conflicts, or poor climatological conditions. More recent field research using unfamiliar drivers has indicated that a minimum exposure time of 1 s per short word (four to eight characters) or 2 s per unit of information, whichever is larger, should be used (Carvell, Turner, and Dudek, 1978; Messer, Stockton, and Mounce, 1978; Weaver, Richards, Hatcher, and Dudek, 1978; Dudek, Huchingson, Williams, and Koppa, 1981). A unit of information is a data item given in a message, that can answer one of the following questions: (1) what happened? (2) where? (3) what is the effect on traffic? (4) for whom is the advisory intended? and (5) what driver action is advised? Thus, the exposure time for a three-line message could vary from 3 s to as much as 6 s, with each phase of a portable CMS at the lower end of this range and with each permanent CMS phase (page) at the upper end, due to differences in the number of characters per line. Reducing the exposure time per phase is warranted only when information is being repeated. For example, a three-line message may be displayed for only 2.5 s if it is a second phase of a two-phase message which repeats one or two lines from the first phase. If the second phase presents new information, the recommended minimum exposure time for both phases remains 3 s.

For a given operating speed, exposure will increase with increasing legibility distance. For example, an overhead sign message that is legible at 198 m (650 ft), will be exposed to drivers traveling at 88 km/h (55 mi/h) for approximately 8 s. With a legibility distance of 305 m (1,000 ft), the message will be exposed for about 12 s. Legibility distances for portable CMS's vary from the minimum of 200 m (650 ft) specified by the MUTCD, Part 6, and American Traffic Safety Services Association (ATSSA) to over 305 m (1,000 ft), depending on the technology. Permanent CMS's generally have legibility distances in the higher range of 274-366 m (900-1,200 ft). However, there is a point at which a sign becomes unreadable during a driver's approach to a CMS, which reduces the legibility distance, particularly for side-mounted CMS's. This unreadable distance, which is dependent on the number of lanes and the sign technology, as well as how far the sign is set back from the roadway edge or how high above the roadway it is mounted, ranges from 85 m to 128 m (280 ft to 420 ft). In an existing system, therefore, required exposure times dictate the maximum length of message that can be displayed, and in all cases, it is desirable that motorists be able to read the entire message on an (unobstructed) CMS twice.

The calculated maximum exposure duration of a message should not exceed 9 s. For two-phase messages, a separate requirement is needed to meet the needs of drivers. In this case, 3 s is added to the required exposure time because of the asynchrony between the time the driver can read the CMS and the onset of CMS phase displayed. In other words, the phase that the driver reads initially may have already been displayed for 2 s by the time he or she can read it. Thus, the driver will not have enough time to read this phase and will need to view that phase again. The net result is that 3 s needs to be added to the required exposure time to allow drivers to read the phase that first came into view a second time. Since the maximum recommended exposure time is 9 s, only 6 s of actual message reading time is allowed on a two-phase CMS, whereas the full 9 s can be used for a single-phase message. The important point here is that single-phase messages can more efficiently convey information to drivers. When use of a single-phase CMS is not possible because of message length, multiple devices with a single phase on each device will be superior to multiple phases on a single device. Part 6 of the MUTCD (section. 6F.52) refers the practitioner to table 6C-1 to determine separation distances between multiple portable CMS's placed on the same side of the roadway.

For these reasons, the maximum number of phases used to display a message on a permanent CMS should be two. The most effective format for CMS message presentation is a single phase which consists of a maximum of three units of information, but if two phases are required, each should be worded so that it can stand alone and still be understood. Portable CMS devices, though limited to fewer characters per line, should also be restricted to two phases. At high speeds (88 km/h [55 mi/h]), a driver may only have 2.8 to 4.6 s to read a message on a side-mounted CMS, depending on the available legibility distance. For this reason, messages should be restricted to one phase at high speeds.

The motorist's need for rapid understanding and integration of message components also focuses attention on the formatting of multiword text displays. The main concern is with "units of information"--i.e., where and how to divide phrases--and with the use of abbreviations and contractions in CMS messages. These formatting issues are discussed below.

Work zones constitute driving situations that require a high amount of controlled processing, and data show that cognitive ability scores that measure processing efficiency decline with age (Ackerman, 1987). In fact, sensory memory, working memory, and divided attention all show a decline with aging and must be considered in the display of messages on CMS's. This reinforces the conclusion that a message should be limited to a single phase, or certainty no more than two, because multiple phases will interfere with message comprehension. There is also considerable evidence that older adults have poorer working memory function than younger adults (Salthouse, 1991; Salthouse and Babcock, 1991). This indicates that message length should be limited to the fewest, most relevant units possible. Finally, older adults are particularly disadvantaged when they are required to use working memory to manage multiple tasks (Ponds, Brouwer, and van Wolffelaar, 1988). Van Wolffelaar, Brouwer, and Rothengatter (1990) found that there is a disproportionately greater problem for older adults in divided attention situations and directly linked this to a higher crash rate for older adults in time-pressured, complex traffic situations.

The minimum required information for traffic management includes: (1) a statement of the problem; and (2) the action statement(s)--i.e., a driver needs to know what to do and one good reason for doing it. Additional elements are included as needed for a specific situation. The key here is not to burden the driver with unnecessary information. Only about two-thirds of drivers are able to recall completely four pieces of information (problem, effect, attention, and action); however, 80-90 percent can recall the action message (Huchingson, Koppa, and Dudek, 1978). Two problems in message presentation must be avoided: (1) providing too much information in too short a time; and (2) providing ambiguous information that leaves either the intent of the message or the desired driver response uncertain.

The first problem does not refer solely to reading time difficulties, as discussed above; instead, it refers to the number of ideas, or "information units," contained in a message. Certainly, the number of words displayed on a sign is important, but so is the manner in which words are grouped. Units containing one word (DELAY), two words (DELAY AHEAD), or many words (MAJOR DELAY AT HIGH STREET) are equally difficult to remember when the display is no longer in sight. However, a series of, say, six units of information in a message displayed on a permanent CMS will be easier to remember if presented in two phases of three units each than if all six units are presented on a single phase. Studies have concluded that no more than three units of information should be displayed on one sequence when all three units must be recalled by drivers (Huchingson et al., 1978; Dudek et al., 1981; Gish, 1995).

Gish (1995) conducted a human factors laboratory study addressing the perceived timeliness, accuracy, and credibility of CMS messages using both younger (ages 16 to 33) and older (ages 65 to 84) test subjects. Results showed that correct recall of the first CMS phase (a downstream speed advisory) was nearly 100 percent for both age groups. However, successive phases of information (containing downstream delay and route diversion information) were recalled less accurately. For the delay information (second phase), correct recall for the younger subjects was about 82 percent, versus 60 percent for the older subjects. For route numbers (third phase), correct recall was 55 percent for the younger subjects and 19 percent for older subjects. These results reinforce the earlier recommendation that a maximum of two phases should be used.

When a message must be divided into two phases, it is desirable to repeat key words from the first phase on the second phase, to provide assurance that all drivers see the message at least once. This also allows information rehearsal, as provided by an additional "learning trial," which will facilitate message recall when the device is no longer in sight. A recommended standard practice is therefore to put the problem on line 1, the location on line 2, and alternate either the effect and action or diversion information on line three, repeating lines 1 and 2 on both phases.

The second type of problem can occur when an unfamiliar word or abbreviation is used, when a word is hyphenated or a phrase is divided inappropriately, or when an abbreviation or a word can mean different things in different word pairings or contexts. Ambiguity occurs, for example, when CENTER LANE is used on a freeway with four or more lanes in one direction. Another example is the use of LANE CLOSED versus LANE BLOCKED, to denote a prolonged closure for construction/maintenance versus a temporary blockage due to a crash or stall. To foster the most simple and consistent practice for motorists, LANE CLOSED is recommended under both roadwork and incident conditions, because at the time the message is displayed, the lane is effectively closed. Finally, neither FREEWAY BLOCKED nor FREEWAY CLOSED should ever be used when at least one lane is open to traffic.

Abbreviations also have the potential to be misunderstood by some percentage of drivers, exacerbating message comprehension problems for individuals with (age-related) diminished capabilities. It has been determined that certain abbreviations are understood by at least 85 percent of the driving public independent of the specific context (e.g., BLVD = boulevard). A second category of abbreviations are understood by at least 75 percent of the driving population but only with a prompt word, (e.g., LOC means "local" when shown with "traffic"). Other abbreviations are prone to be frequently confused with another word (e.g., WRNG could mean either "warning" or "wrong") and should be avoided. Following are lists of abbreviations in three categories, extracted from Dudek et al. (1981): (1) those that are acceptable (understood by at least 85 percent of the driving population) when shown alone (table 50); (2) those that are not acceptable and, therefore, should not be used (table 51); and (3) those that require a prompt word (table 52). Table 50 also includes abbreviations taken from the MUTCD, as well as common contractions used in the English language. The abbreviations in these tables have been incorporated into the MUTCD (FHWA, 2000), as tables 1A-1 to 1A-3 in section 1A.14; section 6F.52 of the MUTCD states that when abbreviations are used on CMS's, they should be easily understood, and refers the practitioner to section 1A.14.

Table 50. "Acceptable" abbreviations for frequently used words.
Source: Dudek, Huchingson, Williams, and Koppa (1981).

Table 51. Abbreviations that are "not acceptable."
Source: Dudek, Huchingson, Williams, and Koppa (1981).

Table 52. Abbreviations⁺ that are "acceptable with a prompt."
Source: Dudek, Huchingson, Williams, and Koppa (1981).

* Prompt word should precede abbreviation.

+ The words and abbreviations shown in normal type are understood by at least 85 percent of the driving population. Those shown in boldface type are understood by at least 75 percent of the driving population, and public education is recommended prior to their usage.

C. Design Element: Channelization Practices (Path Guidance)

Table 53. Cross-references of related entries for channelization practices (path guidance).

Channelization systems include the use of cones, posts, tubular markers, barricades, panels, drums, amber-flashing and steady-burn lights, and standard and raised/recessed pavement markings. They are used to direct motorists into the open lanes and to guide them through the work area. They must provide a long detection distance and be highly conspicuous under both day and night conditions. Using data collected by the police, it has been estimated that anywhere from 80 to 86 percent of the crashes in work zones can be attributed to driver error (Nemeth and Migletz, 1978; Hargroves and Martin, 1980). Hargroves and Martin (1980) found that crashes with fixed objects within a work zone account for a greater percentage than other crash types, such as rear-end or sideswipe. Nemeth and Migletz (1978) found that nighttime crashes are concentrated in the taper area. Humphreys, Maulden, and Sullivan (1979) identified the most significant problems with channelization in work zones as: (1) failure to use, or hazardous use of, temporary concrete barriers; and (2) inadequate or inconsistent use of devices and methods in closing roadways and establishing lane-closure tapers.

Older drivers, like alcohol-impaired and fatigued drivers, show reduced sensitivity to contrast. Olson (1988) pointed out that the brightness of a traffic control device is the main factor in its attention-getting capability: in a visually complex environment, the brightness must be increased by a factor of 10 to achieve conspicuity equivalent to that found in a low-complexity environment. A major problem at night is reduction in contrast sensitivity, which makes it difficult to see even large objects when they cannot be distinguished from their background. Older drivers also have difficulty processing information due to less effective scanning behavior and eye movements, diminished visual field size, difficulty in selective attention, and slower decision making. Inconsistent use of drums and traffic cones to delineate the travel path may be a particular problem for older drivers, especially when applied in the presence of remnants of old lane markings, because such inconsistency is confusing and older drivers (and inattentive drivers) are not able to react as quickly to conflicting traffic cues (National Transportation Safety Board, 1992). To compensate for their slower information-processing capabilities, their reduced visual capabilities, and their slower reaction time, older drivers often drive more slowly. Although driver age was not studied, Hargroves and Martin (1980) found that slow-moving vehicles were overrepresented in work-zone crashes. Older drivers also show reductions in lane-keeping ability, which is further compromised when they are required to attend to other tasks, in unfamiliar surroundings. Finally, steering abilities may be adversely affected by physical problems such as arthritis.

McGee and Knapp (1979) performed an analytic study to develop a performance requirement/standard for the detection and recognition of retroreflective devices (cones, drums, panels, and barricades) used in work zones. The performance standard developed in this study, presented in terms of visibility requirements (i.e., the distance at which motorists should be able to detect and recognize the devices at night) and established using the principles of driver information needs and the requirement for decision sight distance, calls for a minimum visibility distance of 275 m (900 ft) when illuminated by the low beams of standard automobile headlights at night under normal atmospheric conditions.

Pain, McGee, and Knapp (1981) evaluated the effectiveness of traffic cones and tubular markers, vertical panels, drums, barricades, and steady-burn lights in laboratory studies, in controlled field studies, and at actual construction sites. Two-hundred fifty-four subjects between the ages of 17 and 60+ participated; over half of the subjects were between ages 21 and 40, and 7 percent of the subjects were age 60 or older. Overall, there were no major differences between the device categories in the daytime. At night, barricades, panels, drums, cones, and tubular markers were also equivalent when the optimized cone and tubular marker retroreflectorization was used (two bands of retroreflective material for cones and one band for tubular markers totaling 96,800 to 129,000 mm² (150 to 200 in²), or roughly the amount provided by a 300- to 350-mm [12- to 14-in] collar) of retroreflective material with SIA of at least 250. However, tubular markers and cones with 150 mm (6 in) of collar resulted in diminished nighttime performance. The variables manipulated in the cone optimization study included amount of retroreflectorization (44,500, 89,000, 133,600, 178,100, and 222,600 mm² [69, 138, 207, 276, and 345 in²]), corresponding to single bands measuring 150-, 260-, 350-, 430-, and 500-mm (6-, 10-, 14-, 17-, and 20-in) wide; number of bands of retroreflective material (1, 2, or 3); 3 types of retroreflectorization plus 1 internally illuminated cone (polycarbonate Reflexite with SIA of 2000 at entrance angle -4 and observation angle 0.1, high intensity with SIA of 300 at entrance angle -4 and observation angle 0.1, and polycarbonate Reflexite plus vinyl Reflexite); color of retroreflectorization (white and yellow), 3 sizes (450-, 700-, and 900-mm [18-, 28-, and 36- in] tall); 3 device spacings (half, regular, and double-speed limit). The variables manipulated in the tubular marker study included amount of retroreflectorization (14, 28, 43, 57, and 71 percent of area covered, corresponding to bands measuring 150-, 300-, 450-, 600-, and 950-mm [6-, 12-, 18-, 24-, and 38-in] wide); number of bands (1, 2, 4, 6, or 8); the same retroreflectorization levels and colors as for the cone study, 3 sizes (450-, 700-, and 1,050-mm [18-, 28-, and 42-in]) tall; and the same device spacings as described for the cone study.

In comparing the meaning of chevrons versus stripes, Pain et al. (1981) found that diagonal, horizontal, and vertical stripes conveyed no consistent directional information; chevrons, though less easily detected than the stripe patterns, effectively and unambiguously indicated that a movement to the left or right was required. Since diagonal, horizontal, and vertical stripes conveyed no consistent direction information, Pain et al. (1981) concluded that there was no reason to have a diagonal stripe pattern for left and right "sidedness." They pointed out, however, that only one direction of diagonal should be allowed in an array so there is always a consistent pattern or image on devices.

In terms of device spacing, comparisons of regular speed-limit spacing (16.8 m [55 ft] in the test), half spacing (8.4 m [27.5 ft]), and double spacing (33.5 m [110 ft]) of Type I barricades and 200-mm x 600-mm (8-in x 24-in) panels showed that changes in spacing produced little impact on driver behavior. There were no significant speed or lateral placement differences between half, regular, and double speed-limit spacing during the day. At night, however, when devices were placed at half spacing, they produced a speed reduction, apparently from the illusion that the motorist was going faster than he or she actually was. Devices placed at double spacing tended not to perform as well as when they were placed at regular speed-limit spacing, as drivers made lane changes and detected arrays of traffic control devices sooner with shorter spacings. From these findings, Pain et al. (1981) recommended that: (1) all devices be placed at speed limit spacing for most conditions and, in all cases, along the taper or transition section; (2) if there is no construction work or hazard in the closed lane for a substantial length, or traffic delays, the spacing can be increased to no more than twice the speed limit; and (3) shorter spacings may prove to be useful where speed reduction is desired.

Device-specific findings by Pain et al. (1981) are as follows:

• Traffic cones. (1) They perform as well as other devices during daytime, with long detection distance and adequate lane-change distances. (2) Bigger is better: 900-mm (36-in) cones are more effective than 700-mm (28-in) cones; 700-mm (28-in) cones are better than 450-mm (18-in) cones (and 450-mm cones should not be used on high-speed facilities); (3) At night, 96,800 to 129,000 mm² (150 to 200 in²), or roughly the amount in a 300- to 350-mm (12- to 14-in) collar of highly retroreflective material (with specific intensity per unit area [SIA] of at least 250), is needed for effectiveness. Even higher brightness materials enhance driver response characteristics and are preferable. (4) Under both day and night conditions, the 2-band configuration outperformed the 3-band configuration, and both outperformed the 1-band configuration; therefore, two bands of retroreflective material are preferable on cones.
• Tubular Markers. (1) During daytime, 700-mm and 1,050-mm (28-in and 42-in) tubular markers are as effective as cones, but 450-mm (18-in) tubular markers are ineffective and not recommended for lane closures or diversions on high-speed facilities. (2) At night, tubular markers with at least a 300-mm (12-in) highly retroreflective band are equally as effective as cones. (3) The 1-band configuration outperformed the 2- and 3-band configurations for tubular markers
• Vertical panels. (1) Laboratory results showed that compared with the barricade, the vertical panel is more easily detectable. (2) Vertical panels are equally as effective (detectable) as Type I barricades, and vertical panels promote earlier lane changing than barricades. (3) The minimum width dimensions of the panel should be 300 mm (12 in) rather than 200 mm (8 in), especially when used at night and on high-speed facilities.
• Drums. (1) Drums are highly visible and detectable from long distances, during both day and night. (2) Drums promote lane changing further upstream of the taper than other devices. (3) Drums are associated with a speed reduction. (4) Drums are a dangerous object when hit.
• Barricades. (1) The Type I barricade is as effective as other devices. (2) The Type II barricade is no more detectable than the Type I barricade. (3) The 300-mm x 900-mm (12-in x 36-in) barricade is more conspicuous than the 200-mm x 600-mm (8-in x 24-in) barricade.

Other findings were reported for comparisons of steady-burn lights and Type II and Type III sheeting. The steady-burn lights provided the longest detection distances at night compared with all other materials, and they more than tripled the distance (or zone) in which lane changing occurred before the taper. In comparisons of Type II sheeting and Type III sheeting on cone and tubular marker optimization tests, Type III was significantly better at night on a flat road. Narrow-angle sheeting, even though offering high brightness, was not effective under certain sight geometry characteristics, such as hills and curves. Type III sheeting and steady-burn lights were comparable in terms of point-of-lane-change and array detection distance; however, the authors noted that the effect of vertical or horizontal curvature must be considered.

There have been mixed results regarding the effectiveness of steady-burn lights in highway work zones. The use of steady-burn lights mounted on channelizing devices has been shown to significantly influence driver behavior in some work-zone configurations, and they are particularly effective in left-lane closures (KLD Associates, 1992). Although drivers age 55 and older consistently showed poorer performance than younger drivers in all study conditions, evidence was found that the use of lights improved the performance of older test subjects. The variables manipulated in this study included work-zone configuration (left-lane, right-lane, and shoulder closures), device type (panels versus drums), and light placement (every device, alternate devices, no lights). Drivers of all ages were able to identify lane and shoulder closures from greater distances when lights were used on channelization devices, as opposed to when the channelizing devices were used alone. Steady-burn lights produced a higher percentage of correct responses (determining the direction the channelizing devices were leading) for all driver age groups when used in left-lane closures than in right-lane closures. Interestingly, the use of lights on every other drum or vertical panel (placement on alternate devices) generated more correct responses than the use of lights on consecutive devices. More generally, the literature suggests that in environments characterized by high-speed operations, compromised visibility due to inclement weather, and/or complex maneuvers required as a result of work-zone configuration, the deployment of steady-burn lights should be considered on all channelizing devices used for right-lane closures.

However, Pant, Huang, and Krishnamurthy (1992) obtained a different result when they examined the lane-changing behavior of motorists in advance of tapered sections as they drove an instrumented vehicle through work zones during the day, at night when steady-burn lights were placed on drums, and at night when the steady-burn lights were removed. They measured the traffic volume at several locations in each lane in advance of the taper. Results showed that the steady-burn lights had little effect on driver behavior in the work zones studied. It was concluded by Pant et al. that steady-burn lights have little value in work zones that employ drums with high intensity sheeting and a flashing arrow panel as channelizing devices.

Opiela and Knoblauch (1990) conducted laboratory and field studies to determine the optimal spacing and use of devices for channelization purposes in the taper or tangent sections of work zones. In the laboratory study, the recognition distances of eight different device types, spaced at the standard distance and at 1.5 and 2.0 times the standard distance, were measured for 240 subjects. Results indicated variability between the performance of most channelizing devices across the spacings tested. Right- and left-lane closures were then used at six actual work zones, to test the various device spacings under both day and night conditions. Field data were collected at four points equally spaced over 610 m (2,000 ft) before the work zone and the activity at the start of the taper for the lane closure, according to the premise that the most effective treatment would minimize the percentage of traffic in the closed lane at the start of the taper. Statistical analysis of 2,100 observation periods lasting 5 minutes each showed that neither type of device (round drums, oblong drums, Type II barricades, and cones with retroreflective collars) nor device spacings (16.8, 24.4, and 33.5 m [55, 80, and 110 ft]) had a significant effect on driver lane-changing behavior.

Cottrell (1981) also found that driver lane-change response was not strongly dependent on the channelizing device employed in a work-zone taper. The objective of this study was to evaluate the effectiveness of alternative orange-and-white chevron patterns on vertical panels and barricades that form an arrow pointing in the direction in which traffic is being diverted, compared with traffic cones, simulated drum vertical panels, and Type II barricades and vertical panels with standard orange-and-white striping patterns. The measure of effectiveness was the position of lane changing relative to the transition taper. Although the subjective evaluation revealed that chevron patterns were preferred over the presently used patterns because of their clear directional message, the positions of lane changing were similar for the stripes and chevrons. With respect to the spacing of devices, it was generally found that lane changes occurred more frequently at greater distances from the taper when the devices were spaced every 12 m (40 ft), as opposed to every 24.4 m (80 ft).

In a supplemental test, the effectiveness of the concrete safety-shaped barrier (CSSB), also referred to as a "Jersey" barrier in some jurisdictions, was compared with that of the channelizing devices studied (Cottrell, 1981). The barrier was marked with steady-burn warning lights and 150-mm (6-in) reflectors and had a slope of 16:1 for the 58.5-m (192-ft) taper. The CSSB was rated equal to the cone during the daytime and lower than all other devices based on the lane-change positions. It was recommended that a supplemental taper of channelization devices be used with the CSSB. In a study of concrete barrier visibility, Pain et al. (1981) found that retroreflectors were superior to retroreflectorized tape. Logically, the most conspicuous types of retroreflective devices, such as those containing cube-corner lenses, will be potentially the most effective in this regard.

Overall, Pain et al. (1981) concluded that most devices show relatively successful detection and path guidance performance. However, a major deterrent to effectiveness is not the device itself; instead, poor positioning, dirt, and overturned devices destroy the visual line or path created by the channelizing devices. Therefore, although use of appropriate devices is important, of equal importance is conscientious set-up and care of channelizing devices used in the work zones.

In consideration of the threat posed to drivers by passenger compartment intrusion or interference with vehicle control, or the threat to workers and other traffic from impact debris, plastic drums, cones, tubular markers, and vertical panels used as channelizing devices presented no hazards in full-scale vehicle crash tests (Bryden, 1990). However, Types I and II barricades and portable signs and supports formed impact debris, which was often thrown long distances through work zones, posing a threat to workers and other traffic. The American Traffic Safety Services Association (ATSSA) is opposed to the use of metal drums in work zones as channelizing devices, as they pose a hazard to motorists as well as workers in the zone (TranSafety, 1987). They suggest the use of plastic drums, which are safer. Riedel (1986) described studies showing that a substantial number of work-zone crashes occur in the taper and the crossover where channelization devices are located. The frequency of crashes involving drums has led to the use of forgiving devices such as plastic drums, which in tests have been shown to be safer than steel drums. Juergens (1972) noted that because barricades are inherently fixed-object hazards, they should not be used as primary delineation to guide traffic. Further, they should not be used unless the construction hazard the motorist may encounter is greater than the hazard of striking the barricades. A concern with the use of steady-burn lights mounted on channelizing devices was highlighted in full-scale vehicle crash tests evaluating the performance of work-zone traffic control devices, where warning lights attached to these devices were thrown free, posing a potential threat to workers and other traffic (Bryden, 1990).

D. Design Element: Delineation of Crossovers/Alternate Travel Paths

Table 54. Cross-references of related entries for delineation of crossovers/alternative travel paths.

Studies have established that: (1) a substantial proportion of construction work-zone crashes occur in the taper and the crossover, where channelizing devices are usually located; (2) darkness is associated with an increase in the frequency of crashes in these areas; and (3) construction zones are associated with increases in the incidence of fixed-object, rear-end, and head-on crashes (Graham, Paulsen, and Glennon, 1977). Nemeth and Rathi (1983), studying crash types in construction zones on the Ohio Turnpike, found that 52.4 percent of the crashes were with fixed objects, and 68.3 percent of the crossover crashes involved collisions with channelizing devices or other objects. In this study, 69.4 percent of the crashes at the first curve of a crossover occurred at night. Nemeth and Migletz (1978) found that 60.7 percent of single-vehicle fixed-object crashes were collisions with drums and 29.8 percent of all crashes involved collisions with drums. They also found that the proportion of crashes involving construction objects (drums) at night is significantly higher than the proportion of daylight construction object crashes. The results of these studies highlight the need for highly conspicuous and properly installed and maintained channelizing devices.

The relationships between functional capabilities of older drivers and their performance that are likely to be of greatest operational significance as they approach and negotiate a crossover in a work zone can be summarized as follows. Age-related declines in acuity (both static and dynamic) and contrast sensitivity will delay recognition of channelizing devices and pavement markings and will delay comprehension of the information provided by advance warning signs. This information loss in the early stages of the driver's vehicle control task will be compounded by attentional and decision making deficits shown to increase with increasing age, with age differences in performance magnified as serial processing demands for conflict avoidance and compliance with traffic control messages increase during the approach to the work zone. Age-related decrements in the "useful field of view," selective attention, and divided attention, and attention-switching capabilities will slow the initiation of a driver's response when a lane change is required prior to the transition zone, or maneuvering through channelization across the median. In addition, less efficient working memory processes may translate into riskier operations for older drivers in unfamiliar areas if concurrent search for and recognition of navigational cues is required, as such demands disproportionately tax "spare capacity" for lanekeeping and conflict avoidance for older operators. Finally, the execution of vehicle-turning movements becomes more difficult for older drivers as bone and muscle mass decrease, joint flexibility is lost, and range of motion diminishes. Simple reaction time, while not significantly slower for older drivers responding to expected stimuli under nominal operating conditions, suffers operationally significant decrements with each additional response to an unexpected stimulus, e.g., as required in emergency situations. In addition, older drivers' increased sensitivity to glare and reduced dark adaptation ability will compound the difficulties described above while driving at night.

The National Transportation Safety Board (NTSB) has expressed concern about the lack of positive separation of opposing traffic in work zones (NTSB, 1992). The NTSB uses "positive barrier," or "positive separation of traffic," to refer to the use of concrete barriers to separate traffic. (A number of States distinguish between these terms, using "positive separation" to describe various channelization treatments which do not necessarily involve use of a physical concrete barrier.) The NTSB (1992) emphasizes that, "Fatal crash rates increase significantly when an interstate highway is switched from a four-lane, divided operation to a two-lane, two-way operation (TLTWO) during construction work." Research bearing on the use of channelization and barrier delineation for TLTWO's is described below.

A crossover requires a change in direction and may require a reduction in speed. This requires adequate advance warning of the lane and speed reduction, conspicuous and unambiguous delineation/channelization in the transition zone, and conspicuous separation of opposing traffic the length of the TLTWO. One survey of drivers in Houston, TX and Dallas, TX by Hawkins, Kacir, and Ogden (1992) found that only half of the respondents correctly understood that they should turn before reaching the CROSSOVER sign (D13-1) when this device was shown in a field placement in an arterial work zone. Of course, the D13-1 sign panel is identified in the MUTCD as a device used in permanent installations on divided highways, not as a temporary device for use in construction zones. The poor comprehension of motorists for such an explicit message is alarming, nevertheless, and suggests the need for heightened conspicuity of guidance information in this situation. Hawkins et al. recommended that the spacing of channelizing devices be decreased in the vicinity of a crossover to reduce drivers' confusion.

Next, Pang and Yu (1981) conducted a study to verify whether concrete barriers were justified at transition zones adjacent to TLTWO's on normally divided highways, based on crash experience in several construction zone TLTWO's. They found that 34 of the 44 total crashes that occurred in TLTWO's were within the transition zone. Four head-on crashes occurred on two-way, two-lane segments away from the transitions. The transition zone was defined as the roadway section at which traffic flow was converted from a four- to a two-lane operation and vice versa. The absence of opposing traffic precluded the occurrence of head-on crashes during the study period; however, more than one-half of the crashes (56 percent) had the potential of becoming head-on collisions. The authors concluded that on relatively low-volume highways, delineation devices appear to be adequate at transition zones, assuming they are placed properly. A regression analysis provided by Pang (1979) indicated that as annual average daily traffic increases, the crash rate at transition zones also increases, with a concurrent increase in the head-on crash rate at the transition zone.

Project duration and approach speed are two other variables that appear to affect the head-on crash rate at transitions (Pang and Yu, 1981). Graham (1977) concluded that as project duration increases, the crash rate at the transitions decreases. Expectancy issues were highlighted as a plausible explanation. Pang and Yu (1981) reported that because the crash rate in the transition zone increases with shorter project duration, concrete barriers may be necessary for short-term projects. However, long-term projects are expected to have a greater number of crashes owing to a longer period of exposure. Thus, installation of concrete barriers would be more economically justified for long-term projects than for short-term ones. With regard to approach speed, it can be expected that as speed to the transition increases, the chances of a head-on collision would also increase, due to the tendency of vehicles to stray out of their lanes at curves such as those present in transition zones. Pang and Yu (1981) suggested that concrete barriers appear to be justified at transition zones where approach speeds are high.

The conspicuity of concrete safety shaped barriers (CSSB's) is an important issue. Their composition provides little contrast with the roadway pavement, making them difficult to see at night, particularly in the rain, and under opposing headlight glare conditions. Proper barrier delineation treatments will provide drivers with a defined path during darkness and adverse weather conditions. Standard barrier delineation treatments include Type C steady-burn warning lights on top of the barrier, retroreflective devices on the top or side of the barrier, vertical panels placed on top of the temporary concrete barrier, and retroreflective pavement markings on the side of the barrier. The results of studies of barrier delineation in work zones have been mixed (Ullman and Dudek, 1988). For instance, Mullowney (1978) suggested that delineation should be mounted on the top of the barrier so it will retain its reflectivity longer and require less maintenance. However, Ogwoaba (1986) recommended side-mounted concrete barrier delineation so that the delineators are not masked by oncoming headlight glare. The size and brightness of delineators is another controversial topic, with some studies suggesting the use of larger but less bright devices (Davis, 1983; Bracket, Stuart, Woods, and Ross, 1984; Kahn, 1985) and others recommending smaller, brighter reflectors (Mullowney, 1978; Ogwoaba, 1986). Kahn (1985) found that the delineation of portable concrete barriers improved considerably through the use of cylindrical reflectors on top and smaller units on the side of the barrier at 7.6-m (25-ft) intervals. Delineator spacings ranging from 7.6 m to 61 m (25 ft to 200 ft) have been recommended by various studies.

Ullman and Dudek (1988) conducted a study of five barrier delineation treatments, using observations of driver performance to determine how different delineator types, spacings, and mounting positions on the barrier affect nighttime traffic operating in the travel lane next to the barrier. An additional objective of the study was to determine how the visibility and brightness of different types of delineators deteriorate over time because of dirt and road film; in a controlled field study, drivers ages 18 to 56 were asked to provide subjective evaluations of delineator brightness. The study was not conducted at a work zone, but was conducted on an illuminated urban freeway with four lanes in each direction. The CSSB was located 0.3 m (1 ft) from the inside travel lane. The five delineation treatments were: (1) top-mounted cube-corner lenses at 61-m (200-ft) spacing; (2) side-mounted cube-corner lenses at 15.2-m (50-ft) spacing; (3) top-mounted retroretroreflective brackets at 15.2-m (50-ft) spacing; (4) side-mounted retroreflective brackets at 61-m (200-ft) spacing; and (5) top-mounted retroreflective cylinders at 15.2-m (50-ft) spacing. The cube-corner reflector (treatments 1 and 2) had a diameter of 81 mm (3.25 in). The brackets (treatments 3 and 4) were 75 mm (3 in) high and 106 mm (4.25 in) wide, and were covered with high intensity sheeting. The cylindrical tube (treatment 5) had a diameter of 75 mm (3 in) and was 150 mm (6 in) high, and was wrapped with high intensity sheeting. Before-and-after data were obtained for the following measures of effectiveness: lane distribution, lane straddling, and lateral distance from the left rear tire to the bottom of the CSSB.

Results of the driver performance data collected by Ullman and Dudek (1988) showed that the treatments had very little practical effect on lane distribution. Lane-straddling rates at all of the treatment segments were low during the higher volume nighttime hours; however, a significant increase in lane straddling occurred for treatment 2. The data suggested that the combination of close delineator spacing and the side-mounted position may make some drivers too apprehensive of driving near the barrier. Lateral distance data showed significant differences during the higher volume nighttime hours for treatment 4 and treatment 5. Lateral distance distributions shifted away from the barrier at treatment 4 and closer to the barrier at treatment 5. Subjective evaluations for clean delineators showed brightness rankings to be the same for all treatments. Treatments 1-4 received adequate ratings from at least 80 percent of the subjects, while treatment 5 was rated adequate by only 50 percent of the subjects. With respect to each treatment's relative effectiveness in helping drivers maintain a safe travel path next to the CSSB, the rankings did not differ significantly; however, treatment 5 again received the worst score. Subjects stated that side-mounted delineators were preferable to top-mounted delineators because side-mounted delineation provided a more direct line of sight, a better indication of the location of the wall, and a more realistic perception of the lane width. For dirt-covered delineators, treatment 2 was rated as brightest and most effective, while treatment 5 was rated as dimmest and least effective. Although further research was deemed necessary, the study authors recommended the use of cube-corner lenses for delineating CSSB's in narrow freeway median applications, because these delineators do not lose their reflectivity due to dirt and grime as quickly as those covered with high intensity sheeting. In addition, for situations with limited lateral clearance, as is common with TLTWO's, top-mounted delineation is recommended, because side-mounted close delineator spacing results in lane straddling if the barrier is located close to the travel lanes. Although subjects indicated a preference for close spacings, driver performance data did not show any differences between 15.2-m (50-ft) and 61-m (200-ft) spacing. The authors recommended that a 61-m (200-ft) spacing be considered maximum, and that closer spacings may be necessary for CSSB's on sharp curves. The recommendations were also deemed appropriate for CSSB's in work zones.

On divided highways with narrow medians, which are often created when barriers are used in crossover situations in work zones, drivers are subject to blinding glare from opposing vehicle headlights. This is particularly problematic for older drivers who have a reduction in their dark adaptation ability and increased sensitivity to glare. This results in reduced visibility of roadway alignment and channelization, and increases the possibility of crashes. Glare screens can solve the problem, as well as reduce rubbernecking and its associated problems. The Pennsylvania Department of Transportation discontinued the use of the standard glare-control mesh screen in 1976, based on maintenance difficulties, and has employed a paddle-type system in its place (Maurer, 1984). The system consists of plastic airfoil-shaped paddles, which when mounted resembles a picket fence. Results of a 5-year study have shown that the paddle-type system reduces headlight glare satisfactorily and is more cost-effective, both in terms of installation and maintenance, than metal mesh screen. The system was also found to be beneficial as a temporary control for channelizing traffic around a construction work zone, when screening was placed at the transition or the taper zone at the ends of the work zone (Maurer, 1984). Kelly and Bryden (1983) reported that a glare screen consisting of individual plastic louvers 900 mm (36 in) high, mounted vertically on a guiderail or median barrier spaced at 600-mm (24-in) centers, performed as expected in two safety improvement projects.

E. Design Element: Temporary Pavement Markings

Table 55. Cross-references of related entries for temporary pavement markings.

Preconstruction centerlines and edgelines that are not obliterated may confuse drivers about the exact locations of lanes. The National Transportation Safety Board (1992) has reported that although guidelines exist for proper signing and striping in construction areas, the traffic control techniques used in many jurisdictions are not in compliance with the guidelines. Lewis (1985) stated that if drivers are presented with conflicting information (as may be the case in a work zone), they will generally choose to follow the pavement, as the pavement itself is a primary source of information for drivers. This points to a need for unambiguous pavement delineation patterns in work zones, to provide clear guidance--particularly at night and under adverse weather conditions--and to accommodate drivers with visual limitations such as those associated with normal aging.

The research findings that have the greatest bearing on age differences in drivers' ability to acquire and use information provided by roadway delineation are a decline in spatial contrast sensitivity and acuity for older drivers, and a general slowing of responses because of deficits in visual search ability that slows discrimination of more important from less important information in a driving scene.

Discrimination of the boundaries of the traveled way often involves only slight differences in the brightness of the road surface versus the shoulder or surrounding land. The ability to obtain such "edge information" depends upon a driver's sensitivity to contrast. Age differences in contrast sensitivity, beginning at approximately age 40 and becoming progressively more exaggerated with advancing age, demonstrate significant decrements in performance for older persons (Owsley, Sekuler, and Siemsen, 1983). Under constant viewing conditions, older observers have lower contrast sensitivity especially in situations where there is a reduction in ambient light levels. A 60-year-old driver requires 2.5 times the contrast needed by a 23-year-old driver (Blackwell and Blackwell, 1971).

Age decrements in visual search and scanning capabilities are widely reported in gerontological research. Rackoff and Mourant (1979) measured visual search patterns for 10 young (ages 21-29) and 13 older (ages 60-70) subjects as they drove on a freeway under day and night conditions in low to moderate traffic. They reported that differences between young and older test subjects' performance were most apparent at night, and that older subjects required more time to acquire the minimum information needed for vehicle control. Thus, older drivers require delineation information that is optimal from the standpoints of both attention conspicuity and search conspicuity downstream, and that provides unambiguous path guidance cues for moment-to-moment steering control. Uncertainty about roadway heading and lane position has been cited by older driver focus group members as reasons for driving slower, for erratic maneuvers caused by last-second steering corrections, and for simply avoiding nighttime operations (Staplin, Lococo, and Sim, 1990). An exaggeration of the difficulties older drivers have in rapidly discerning the correct travel path may be expected in construction zones, where drivers must respond to temporary pavement markings that are often in competition with preexisting stripes and/or misleading informal cues provided by variation in the surface characteristics of the road, shoulder, or median.

These diminished capabilities must be considered in relation to specific information needs, when negotiating work zones, while also taking into account the time (distance) in which these needs must be satisfied. The information needs may be loosely contrasted according to the discrimination of continuous versus discrete roadway features, i.e., the perception and recognition of the boundaries of the traveled way, as opposed to a singular location which must be avoided (e.g., an island, barrier, or abutment) or where a path selection decision must be acted upon (e.g., a ramp gore, pavement width transition point, or intersection). Furthermore, delineation must provide information to a driver permitting roadway feature recognition both at "long" preview distances up to and sometimes exceeding 5 s travel time, and at the more immediate proximities (within 1 s travel time) where attention is directed for instant-to-instant vehicle control responses.

An investigation of age-related differences in the required contrast for pavement delineation showed that an older driver (ages 65-80) test sample required a level of contrast 20-30 percent higher than a young/middle-aged (ages 19-49) comparison group (Staplin et al., 1990). The differences became exaggerated with glare as an independent variable. An inevitable consequence of these age differences is an increased reliance on delineation elements for path guidance by older drivers under nighttime conditions, especially against oncoming glare. The "long preview" as well as the instant-to-instant steering control cues provided by pavement markings are critical to older drivers under these circumstances.

Raised pavement markers (RPM's) used for delineation of the centerline and edgelines in construction zones have been found to provide improved wet weather and nighttime reflectivity, and are particularly useful when lanes are diverted from their original path (Spencer, 1978). Davis (1983) reported that, compared with conventional pavement markings (e.g., paint), day-night/wet-night visible RPM's improved construction zone traffic performance significantly. In this study, the markers were associated with decreased lane-change frequency and night lane encroachments. In before-and-after comparisons of crash frequencies in two construction projects, the number of crashes and fatalities decreased as a function of RPM installation (Niessner, 1978). In a study investigating vehicle guidance through work zones, Shepard (1989) recommended that closely spaced RPM's should be used as a supplement to existing pavement striping in areas where the roadway alignment changes.

Dudek, Huchingson, and Woods (1986) conducted a study on a test track to examine the effectiveness of temporary pavement markings for use in work zones. Ten candidate treatments were tested during the day, and the most effective treatments were examined at night. All treatments were tested only under dry weather/dry road conditions. The candidate treatments are presented in table 56 and included patterns with stripes, RPM's, and combinations of stripes and RPM's. Treatment 1 was the control condition in the study.

Table 56. Temporary pavement marking treatments evaluated by Dudek, Huchingson, and Woods (1986).

*Treatments evaluated both day and night

1 ft = 0.305 m
1 in =25 mm

Results of both daylight and nighttime testing indicated that there were no practical differences between treatments when comparing measures of effectiveness developed from speed and distance measurements. Practical differences were arbitrarily defined as at least 6.5 km/h (4 mi/h) for speed measures and 0.3 m (1 ft) for distance measures. The greatest number of erratic maneuvers during daylight occurred for treatments 7 and 8, which consisted of 0.6-m (2-ft) stripes and long gaps. Drivers referred to 0.6-m (2-ft) stripes as dots. The subjective data indicated that treatments 5, 6, and 9 were preferred, under both daylight and nighttime conditions. Reasons given were that RPM's clearly identify curves, are highly visible at a great distance, provide noise and vibration when drivers cross them, and stand out more than tape markings. Of the treatments without RPM's, treatment 3 was the drivers' choice, for both lighting conditions, while treatment 2 was rated as least effective.

It should be noted that for temporary pavement markings, the MUTCD specifies in section 6F.66 that the same cycle length as permanent markings be used (9 m [30 ft]), with markings at least 0.6-m (2-ft) long, and that half-cycle lengths with a minimum of 0.6-m (2-ft) stripes may be used for roadways with severe curvature.

Because subjects tend to perform best when in a controlled, test track setting and because the range of performance measures are not always sensitive enough to discern small differences between candidate treatments, Dudek, Huchingson, Creasey, and Pendleton (1988) conducted field studies to compare the safety and operational effectiveness of 0.3-m (1-ft), 0.6-m (2-ft), and 1.2-m (4-ft) temporary broken line pavement markings on 12.2-m (40-ft) centers in work zones. The study was conducted at night on rural two-lane, two-way highways with 2.0-degree horizontal curvatures, level to rolling terrain, and average speeds between 80.5 km/h (50 mi/h) and 99.8 km/h (62 mi/h). In terms of speed, lateral distance, encroachment, erratic maneuver, and speed profile data for the sample of vehicles with headways of 4 s or more, there were no differences in driver performance between the 0.3-m (1-ft), 0.6-m (2-ft), and 1.2-m (4-ft) striping patterns. Analysis of subjective evaluations of the effectiveness of the markings found that the 0.3-m (1-ft) stripe was generally rated as poorest, but its mean ranking was not significantly different from that of the 0.6-m (2-ft) and 1.2-m (4-ft) stripes. Drivers generally preferred the longer stripes, but there was no evidence that the 0.6-m (2-ft) or 1.2-m (4-ft) stripes were superior to the 0.3-m (1-ft) stripe.

In a discussion of the conditions present during this research, Ward (1988) stated that all sites had 3.7-m (12-ft) lanes with 1.2-m (4-ft) to 3-m (10-ft) shoulders, the marking material was highly retroreflective yellow tape laid over very black new pavement overlays, and there were no edgelines; therefore, the drivers' focus was a "brilliant ribbon of yellow to follow," resulting in no difference in driver performance between the three stripe lengths. Most important was that none of the treatments were judged as extremely effective, although the 0.3-m (1-ft) stripe was rated as poorest, and there was a slight preference for the 1.2-m (4-ft) lengths. This is consistent with results obtained by Dudek et al. (1986), where subjects rated 2.4-m (8-ft) stripes with 9.8-m (32-ft) gaps as the best striping treatment (when RPM's were not available). In the Dudek et al. (1986) study, drivers preferred the treatments with longer stripes, shorter gaps, and RPM's. Hence, the results of the Dudek et al. (1988) study may be applicable only to pavement overlay projects on two-lane, two-way rural roadways, and may not translate to other highway work-zone situations.

Harkey, Mera, and Byington (1992) conducted a study to determine the effects of short-term (interim) pavement markings on driver performance under day, night, wet, and dry weather conditions. The three marking patterns tested included: (1) 0.6-m (2-ft) stripes with 11.6-m (38-ft) gaps and no edgelines; (2) 1.2-m (4-ft) stripes with 11-m (36-ft) gaps and no edgelines; and (3) 3-m (10-ft) stripes with 9.2-m (30-ft) gaps and edgelines. The measures of effectiveness included lateral placement of the vehicle on the roadway, vehicle speed, number of edgeline and lane line encroachments, and number of erratic maneuvers (e.g., sudden speed or directional changes and brake applications). For each operational measure, the 3-m (10-ft) markings resulted in better driver performance than either the 0.6-m (2-ft) or 1.2-m (4-ft) temporary marking patterns. Drivers traveled 1.2 km/h (0.76 mi/h) slower on segments with 1.2-m (4-ft) markings and 3.3 km/h (2.02 mi/h) slower on segments marked with 0.6-m (2-ft) markings than on segments marked with 3-m (10-ft) stripes and edgelines. In addition, compared with the 3-m (10-ft) pattern, drivers encroached over the lane or edgeline 66 percent more often in the presence of the 1.2-m (4-ft) temporary marking and 139 percent more often in the presence of the 0.6-m (2-ft) markings. These values increased dramatically under night and wet weather conditions. Comparisons of driver performance between the 1.2-m (4-ft) and 0.6-m (2-ft) markings showed the following: (1) the speed at which drivers traveled decreased as the length of the lane line decreased; (2) drivers positioned their vehicles closer to the center of the lane as the length of the line increased; (3) the variability of vehicle placement within the lane increased as the length of the lane line decreased; (4) the number of encroachments increased as the length of the lane line decreased; and (5) all operational measures were negatively affected by adverse weather conditions. Results provided evidence of significant decreases in driver performance associated with both the 0.6-m (2-ft) and the 1.2-m (4-ft) markings, but drivers performed better with the 1.2-m (4-ft) stripes compared to the 0.6-m (2-ft) stripes. The results suggested that while it may not be practical to place full markings (3.0-m [10-ft]) segments with 9.0-m [30-ft] gaps as specified by MUTCD Part 3A.06) on a temporary basis, measures should be taken to prevent reductions in driver performance which result in increased crash potential. Such measures include the use of longer temporary markings, the addition of RPM's for improved performance under adverse weather conditions, and the appropriate use of warning signs to indicate a change in the pavement marking pattern.

FHWA-RD-01-103