Guidance for the Design and Application of Shoulder and Centerline Rumble Strips (2009)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

7S E C T I O N 2 The purpose of shoulder and centerline rumble strips is to alert motorists that their vehicles have drifted out of their intended travel lane. The primary purpose of shoulder rum- ble strips is to reduce SVROR crashes, and for centerline rum- ble strips, it is to reduce head-on crashes, opposite-direction sideswipe crashes, and to some degree SVROR-to-the-left crashes. Shoulder and centerline rumble strips can be expected to have the greatest impact on crashes where drivers drift from their travel lanes because they are inattentive, distracted, drowsy, or fatigued. In these situations, the auditory (and pos- sibly tactile) stimuli generated while traversing the rumble strips can alert the inattentive, distracted, drowsy, or fatigued drivers to correct their vehicle trajectories. Shoulder or center- line rumble strips should not be expected to significantly impact those crashes where vehicles leave their intended travel lanes due to situations such as mechanical failures (e.g., tire blowouts), evasive maneuvers to avoid objects in the travel lane, or driver error due to medical conditions (e.g., heart attack or seizures). This section summarizes the magnitude and nature of the highway safety concerns related to SVROR, head-on, and fatigue-related crashes to put into perspective the extent of the problem and highlight the potential impact that shoulder and centerline rumble strips could have on reducing highway fatalities and injuries. SVROR Crashes Based upon a compilation of motor vehicle crash data from the Fatality Analysis Reporting System (FARS) and the Gen- eral Estimates System (GES), 39,189 fatal crashes, 1,816,000 injury crashes, and 4,304,000 property-damage-only crashes occurred on the U.S. highway system in 2005, totaling 6,159,000 crashes (5). Table 1 presents these crashes by crash type, relation to roadway, and crash severity. It shows that of the 39,189 fatal crashes, 12,340 (31.5 percent) were single- vehicle crashes that occurred off the roadway. An additional 2,431 fatal crashes (6.2 percent) occurred on the shoulder, and 1,022 (2.6 percent) occurred on the median. Thus, of the 39,189 fatal crashes, 15,793 crashes (40.3 percent) occurred off the roadway, on the shoulder, or within the median. Of the injury crashes, 21 percent (382,000) were single-vehicle crashes that occurred off the roadway, on the shoulder, or within the median. Of the property-damage-only crashes, 16.5 percent (710,000) were single-vehicle crashes that occurred off the roadway, on the shoulder, or within the median. These num- bers show that single-vehicle crashes that occur off the road- way, on the shoulder, or within the median account for a significant portion of all accidents (18 percent). It is also evi- dent from the higher percentage of fatal and injury crashes that these crashes typically cause severe injuries or fatalities. It is these single-vehicle, off roadway (and possibly shoulder and median) crashes that rumble strips placed on the outside shoulder or median shoulder have the greatest potential to impact. Table 2 presents 2005 crash data by first harmful event, manner of collision, and crash severity. It shows that of the 39,189 fatal crashes, 12,439 (31.7 percent) were single-vehicle collisions with fixed objects; while another 6,505 (16.6 percent) were single-vehicle collisions with objects that were not fixed. The single-vehicle collisions with fixed objects are potentially correctable by shoulder rumble strips, while it is not known what portion of the single-vehicle collisions with objects not fixed are potentially correctable by rumble strips because it is not known whether the collisions occurred on the roadway or off the roadway. It is also notable that the single-vehicle colli- sions with fixed objects are a high percent of the fatal crashes (31.7 percent). A final note concerning Tables 1 and 2, these tables do not indicate where these crashes occurred relative to junctions (i.e., whether the crashes should be attributed to an intersection or to a roadway segment). In most cases, the SVROR types of crashes that could be remedied by shoulder rumble strips occur along roadway segments, not at intersection junctions. In addition, shoulder rumble strips are typically discontinued at Magnitude and Nature of Highway Safety Concerns Related to Shoulder and Centerline Rumble Strips

Table 1. Crashes by crash type, relation to roadway, and crash severity (5). Table 2. Crashes by first harmful event, manner of collision, and crash severity (5). Relation to roadway Crash type On roadway Off roadway Shoulder Median Other/unknown Total Fatal Crashes Single Vehicle 6,507 12,340 2,431 1,022 353 22,653 Multiple Vehicle 15,647 297 302 198 92 16,536 Total 22,154 12,637 2,733 1,220 445 39,189 Injury Crashes Single Vehicle 154,000 320,000 14,000 48,000 28,000 564,000 Multiple Vehicle 1,235,000 7,000 1,000 7,000 2,000 1,252,000 Total 1,390,000 327,000 15,000 54,000 30,000 1,816,000 Property-Damage-Only Crashes Single Vehicle 328,000 598,000 31,000 81,000 277,000 1,314,000 Multiple Vehicle 2,957,000 11,000 3,000 14,000 5,000 2,990,000 Total 3,284,000 609,000 34,000 94,000 282,000 4,304,000 All Crashes Single Vehicle 488,000 930,000 48,000 129,000 306,000 1,901,000 Multiple Vehicle 4,208,000 18,000 5,000 21,000 7,000 4,258,000 Total 4,697,000 948,000 53,000 150,000 313,000 6,159,000 Crash severity Fatal Injury Property damage only Total First harmful event Number Percent Number Percent Number Percent Number Percent Collision with Motor Vehicle in Transport: Angle 8,119 20.7 586,000 32.3 1,185,000 27.5 1,779,000 28.9 Rear-end 2,118 5.4 513,000 28.2 1,309,000 30.4 1,824,000 29.6 Sideswipe 958 2.4 71,000 3.9 392,000 9.1 463,000 7.5 Head-on 3,970 10.1 62,000 3.4 57,000 1.3 123,000 2.0 Other/unknown 192 0.5 * * 4,000 0.1 4,000 0.1 Subtotal 15,357 39.2 1,232,000 67.8 2,947,000 68.5 4,195,000 68.1 Collision with Fixed Object: Pole/post 1,852 4.7 72,000 4.0 153,000 3.6 227,000 3.7 Culvert/curb/ditch 2,591 6.6 60,000 3.3 131,000 3.0 193,000 3.1 Shrubbery/tree 3,215 8.2 65,000 3.6 82,000 1.9 150,000 2.4 Guard rail 1,189 3.0 35,000 1.9 84,000 1.9 120,000 1.9 Embankment 1,444 3.7 25,000 1.4 28,000 0.6 54,000 0.9 Bridge 336 0.9 4,000 0.2 12,000 0.3 16,000 0.3 Other/unknown 1,812 4.6 65,000 3.6 165,000 3.8 232,000 3.8 Subtotal 12,439 31.7 326,000 18.0 653,000 15.2 992,000 16.1 Collision with Object Not Fixed: Parked motor vehicle 498 1.3 29,000 1.6 297,000 6.9 327,000 5.3 Animal 174 0.4 15,000 0.8 260,000 6.0 275,000 4.5 Pedestrian 4,520 11.5 59,000 3.3 1,000 * 64,000 1.0 Pedal cyclist 776 2.0 45,000 2.5 4,000 0.1 50,000 0.8 Train 204 0.5 1,000 * 1,000 * 2,000 * Other/unknown 333 0.8 8,000 0.4 41,000 0.9 49,000 0.8 Subtotal 6,505 16.6 158,000 8.7 603,000 14.0 768,000 12.5 Noncollision: Rollover 4,266 10.9 87,000 4.8 49,000 1.1 141,000 2.3 Other/unknown 564 1.4 12,000 0.7 51,000 1.2 64,000 1.0 Subtotal 4,830 12.3 99,000 5.5 100,000 2.3 205,000 3.3 Total **39,189 100.0 1,816,000 100.0 4,304,000 100.0 6,159,000 100.0 * Less than 500 or less than 0.05 percent. ** Includes 58 fatal crashes with an unknown first harmful event.

intersections, so shoulder rumble strips should not be expected to significantly impact SVROR crashes at intersections, such as those where a single vehicle strikes a signal pole. Neuman et al. (6) analyzed the extent of the SVROR prob- lem specifically related to two-lane, undivided, noninterchange, and nonjunction roadways using 1999 FARS data. On these roadways, Neuman et al. found that 24 percent of the fatal crashes were SVROR crashes. Figure 1 shows the distribution of SVROR crashes on two-lane roadways by roadway func- tional classification. Twice as many SVROR crashes occur on rural roads than on urban roads, partly due to higher speeds on rural roads and to the greater mileage. This suggests the expected safety effectiveness of shoulder rumble strips may likely be different on rural and urban two-lane roads due to the difference in crash distributions between rural and urban areas and probably on other roadway types as well. Neuman et al. (6) also investigated the distribution of SVROR crashes between tangent and curved sections of high- ways. For all roadway types, they found that 42 percent of the SVROR crashes occurred on curves and 58 percent occurred on tangents. For rural two-lane roads, the distribution of SVROR crashes is equally distributed between tangents and curves (i.e., 50 percent on tangents and 50 percent on curves). It is clear that SVROR crashes are a significant problem along both types of alignments. Head-On Crashes Centerline rumble strips are intended to reduce head-on crashes, opposite-direction sideswipe crashes, and to some degree SVROR-to-the-left crashes. This section focuses on head-on crashes rather than opposite-direction sideswipe crashes and SVROR-to-the-left crashes because head-on crashes are typically more severe than sideswipe crashes, and often it is difficult to distinguish between SVROR- to-the-right and SVROR-to-the-left crashes using elec- tronic crash data. Thus, this general discussion may under- estimate the potential safety benefits of centerline rumble strips. Neuman et al. (7) analyzed the extent of the problem of head-on crashes that could potentially be remedied by center- line rumble strips. Based upon 1999 FARS data, 18 percent of noninterchange, nonjunction fatal crashes were two vehicles colliding head on. In addition, the data revealed the following: • 75 percent of head-on crashes occur on rural roads, • 75 percent of head-on crashes occur on undivided two-lane roads, and • 83 percent of two-lane undivided road crashes occur on rural roads. 9 Figure 1. Distribution of SVROR fatalities on two-lane, undivided, noninterchange, nonjunction roads by highway type (6). Rural major collector, 26% Rural local road or street, 24% Rural minor arterial, 13% Rural minor collector, 9% Rural principal arterial, 8% Unknown rural, 2% Urban local road or street, 7% Urban minor arterial, 5% Urban collector, 3% Urban principal arterial, 3% Unknown urban, 0% Unknown, 0%

The high percentage of head-on crashes that occur on rural, undivided two-lane roads might suggest that many head-on crashes relate to failed passing maneuvers; however, of the 7,430 vehicles involved in head-on fatal crashes on two-lane, undivided roadways in 1997, only 4.2 percent involved a vehi- cle passing or overtaking another vehicle. This trend is consis- tent with two FHWA studies (8, 9). It might be thought that most head-on crashes occur along horizontal curves rather than tangent sections of roadway because vehicles would be expected to cross the centerline more frequently on curves. However, the majority of head- on fatal crashes occur on tangent sections. For all roads, 67 percent of the head-on fatal crashes occur on tangents, and 33 percent occur on curves. On rural two-lane roads, similar percentages are found; 63 percent of the head-on fatal crashes occur on tangents, while 37 percent occur on curves. This most likely reflects that the tangent sections account for a sig- nificant portion of the total miles of roadway. Still, based on the percentages, head-on crashes on both curves and tangents represent a significant safety problem, particularly on rural two-lane roads. Table 2 shows that head-on crashes account for approxi- mately 10.1 percent of all fatal crashes and 3.4 percent of all injury crashes. In particular, the table shows head-on crashes are often fatal. As noted above, Table 2 does not indicate where these crashes occurred relative to junctions (i.e., whether the crashes should be attributed to an intersection or to a roadway segment) so this table does not show a complete picture of those accidents that could be remedied by centerline rumble strips. Centerline rumble strips should not be expected to impact head-on crashes that occur at intersections, in part because centerline rumble strips are typically discontinued at intersections, and although head-on crashes at intersections might be related to driver inattention or distraction, the stim- uli generated by centerline rumble strips may not heighten the awareness of drivers in these situations. Drowsy and Fatigued Driving The discussion of SVROR and head-on crashes above does not illustrate a complete picture of the frequencies or propor- tions of crashes that are potentially remedied by shoulder and centerline rumble strips. What is missing is the driver behav- ior associated with these crashes. It is the SVROR and head- on crashes where the driver is inattentive, distracted, drowsy, or fatigued that have the greatest potential to be impacted by shoulder and centerline rumble strips. This section focuses on safety concerns related to drowsy and fatigued driving. Drowsy and fatigued driving leads to crashes because it impairs a driver’s performance. Drowsiness reduces reac- tion times and even small changes in reaction time can have a major repercussion, particularly at high speeds. Drowsiness also reduces vigilance and slows the driver’s ability to process information. These limitations, working in combination, lead to a high number of serious crashes each year. Several recent studies suggest that drowsy and fatigued driv- ing is a serious concern for highway safety. The 2002 Sleep in America Poll (10) found that 51 percent of drivers admitted to driving while drowsy, 17 percent admitted to dozing off while driving, and 1 percent reported having crashed due to dozing off or fatigue. A 2004 public opinion poll of Canadian drivers found that 20 percent of drivers admitted to falling asleep or nodding off at least once while driving in the past 12 months (11). In a study about factors associated with falling asleep at the wheel among long-distance truck drivers, 47 percent of the respondents admitted to falling asleep at the wheel of their trucks, and 25 percent had fallen asleep at the wheel in the past year (12). In the United States it is estimated that up to 20 per- cent of serious crashes may be due to drowsy or fatigued driv- ing and that fatigue likely contributes to between 79,000 and 103,000 crashes and approximately 1,500 fatalities annually. Typical crashes that are related to drowsiness/fatigue have the following characteristics (13): • Crashes occur during late-night hours—Most crashes involving drowsiness/fatigue occur from midnight to the predawn hours with a small peak in the middle of the after- noon. This is consistent with human sleeping patterns. • Crashes happen at high speed—Because more long trips occur on higher speed roadways, there is likely a higher pro- portion of drowsiness/fatigue crashes on roadways with speed limits of 55 to 65 mph (88 to 105 km/h). • Crashes are likely to be serious—Injury and fatality rates are higher for drowsiness/fatigue crashes than with other types of crashes. The higher rates could be a factor of the crashes happening at higher speeds. • Single vehicle leaves the roadway—A majority of drowsiness/fatigue crashes involve single vehicles leaving the roadway. Rear-end crashes and head-on crashes may also be increased due to drowsiness and fatigue. • No attempt to avoid crashes—Evidence of avoiding actions such as skid marks or brake lights are less likely in drowsiness/ fatigue crashes than in other types of crashes. • Driver is alone in vehicle—Drowsiness/fatigue crashes often involve single-occupant vehicles. Although no one is exempt from risk, the following three populations have a higher risk of being involved in a drowsiness/fatigue-related crash: • Young people, especially young men—Drivers under age 30 are four times more likely than other drivers to be 10

involved in a drowsy-driving crash. Men are five times more likely than women to be involved in a drowsy-driving crash. • Shift workers—The information concerning shift workers and drowsiness/fatigue-related crashes has come from self- reporting and interviews rather than crash reports; but due to changing sleeping patterns, loss of sleep, and more driv- ing done in the early morning hours, it is assumed that there is a greater risk of drowsiness/fatigue crashes among shift workers. • People with untreated sleep apnea syndrome and narcolepsy—The total number of drowsiness/fatigue crashes involving drivers with sleep disorders is low, but the risk is higher among drivers with untreated sleep dis- orders than among other drivers. Crashes and Heavy Vehicles There has been some debate on the impact that shoulder and/or centerline rumble strips can have on reducing the num- ber of crashes involving heavy vehicles. One issue is whether a sufficient amount of stimuli, either auditory or tactile, is gen- erated within the passenger compartment to alert a truck driver; the second issue is whether heavy vehicles are involved in the types of crashes that could be remedied by shoulder and/or centerline rumble strips. This section addresses the sec- ond issue. Based upon 2005 crash data, more that 94 percent of the 11 million vehicles involved in motor vehicle crashes were pas- senger cars or light trucks. Heavy vehicles accounted for 8 per- cent of vehicles involved in fatal crashes, 3 percent of vehicles involved in injury crashes, and 5 percent of vehicles involved in property-damage-only crashes. Of the 4,932 heavy vehicles involved in fatal crashes, 74 percent were combination trucks. Table 3 presents 2005 crash data for heavy vehicles by first harmful event and crash severity. Of the 4,932 fatal crashes involving heavy vehicles, 175 crashes (3.5 percent) were single- vehicle collisions with fixed objects, and of the 82,000 injury crashes involving heavy vehicles, 2,000 crashes (2.9 percent) were single-vehicle collisions with fixed objects. This suggests that very few heavy vehicles are involved in fatal and injury crashes that could be remedied by shoulder rumble strips. Data from the California DOT (Caltrans) indicate similar results (14). Using accident data for the period from 1997 to 1999, a total of 929 fatal SVROR crashes were identified. A small por- tion of these fatal crashes involved single heavy vehicles with 3 axles or more (i.e., 41 fatal crashes or approximately 4 per- cent) and the remaining 888 fatal crashes (i.e., 96 percent) involved passenger vehicles. Of the 41 fatal truck crashes, only 4 crashes were attributed to the driver falling asleep. The remaining truck crashes were due to primary causes includ- ing driving under the influence (DUI), alcohol, speeding, etc. From the 888 passenger vehicle fatal crashes, 54 involved drivers falling asleep. This analysis showed that the incidence of SVROR crashes for trucks is very low. Bucko and Khorashadi (14) thought this might be due, in part, to stricter requirements for licensing of commercial vehicle drivers, as well as restric- tions on the number of hours they are allowed to drive daily. It should be noted that the crash data contradict the drowsy and 11 Crash severity Fatal Injury Property damage only Total First harmful event Number Percent Number Percent Number Percent Number Percent Collision with Motor Vehicle in Transport by Initial Point of Contact: Front 2,309 46.8 29,000 35.3 74,000 20.8 105,000 23.8 Left side 417 8.5 12,000 14.7 50,000 14.0 62,000 14.1 Right side 212 4.3 12,000 14.5 55,000 15.6 67,000 15.2 Rear 740 15.0 14,000 17.6 58,000 16.5 74,000 16.7 Other/Unknown 33 0.7 * 0.3 * 0.1 1,000 0.1 Subtotal 3,711 75.2 68,000 82.3 237,000 66.9 309,000 69.9 Collision with Fixed Object: Subtotal 175 3.5 2,000 2.9 30,000 8.4 32,000 7.3 Collision with Object Not Fixed: Nonoccupant 405 8.2 1,000 1.2 * * 1,000 0.3 Other 119 2.4 1,000 1.4 65,000 18.3 66,000 15.0 Subtotal 524 10.6 2,000 2.5 65,000 18.3 67,000 15.3 Noncollision: Subtotal 522 10.6 10,000 12.3 22,000 6.3 33,000 7.5 Total 4,932 100.0 82,000 100.0 354,000 100.0 442,000 100.0 * Less than 500 or less than 0.05 percent. Table 3. Heavy vehicles involved in crashes by most harmful event and crash severity (5).

fatigued driving study of long-distance truck drivers, which suggests a good portion of these drivers are prone to falling asleep behind the wheel (12). Focusing on the crashes that could potentially be remedied by centerline rumble strips, Table 3 shows that of the 4,932 fatal crashes involving heavy vehicles, 2,309 (46.8 percent) where head-on crashes, and of the 82,000 injury crashes involving heavy vehicles, 29,000 (35.3 percent) were head-on crashes. Looking at total percentages, if a large truck is involved in a crash, 23.8 percent of the time the front part of the truck col- lided with another motor vehicle in transport. It is not known whether the truck collided with the front, rear, or side of the other vehicle. Not knowing this information, assume that a collision with a motor vehicle in transport by initial point of contact (front) is the best surrogate for a head-on crash. This suggests that trucks should potentially be considered in the design and application of centerline rumble strips. Again, Table 3 does not indicate where these crashes occurred relative to junctions (i.e., whether the crashes should be attributed to an intersection or to a roadway segment), nor does Table 3 indicate whether the heavy vehicle crossed over the centerline or whether the other involved vehicle crossed the centerline. Summary The primary purpose of shoulder rumble strips is to reduce SVROR crashes, and the primary purpose of centerline rum- ble strips is to reduce head-on crashes, opposite-direction sideswipe crashes, and to some degree SVROR-to-the-left crashes. For centerline rumble strips, the focus above was on head-on crashes rather than opposite-direction sideswipe crashes and SVROR-to-the-left crashes because head-on crashes are typically more severe than sideswipe crashes, and often it is difficult to distinguish between SVROR-to-the- right and SVROR-to-the-left crashes. Crash data show, in both frequency and proportion, that SVROR and head-on crashes are a problem on the U.S. high- way system and deserve attention. Similarly, the data on drowsy and fatigued driving suggest a significant portion of drivers are behind the wheel while they are drowsy and fatigued, leading to impaired performance. The data do not support the need to design and install shoulder rumble strips for heavy vehicles but do potentially support the need to design and apply centerline rumble strips to reduce the fre- quency of head-on crashes involving heavy vehicles. 12