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.

114 S E C T I O N 8 The minimum level of stimuli generated by a shoulder or centerline rumble strip able to alert an inattentive, distracted, drowsy, or fatigued driver is a key human factor issue for which there is little understanding. Without knowing this, it is difficult to recommend minimum or optimum dimensions for rumble strips. Based on the safety evaluations conducted as part of this research and from previous safety research, we know many rumble strip patterns generate sufficient stimuli levels to alert inattentive, distracted, drowsy, or fatigued drivers, but it is not known to what extent the dimensions of these rumble strips could be modified while still maintaining their effectiveness. The complexity of this issue rests at several levels. Current practice suggests rumble strips that generate at least 3 to 6 dBA above the ambient sound level are sufficient to stimulate an inattentive or drowsy driver. This is based on research that investigated just noticeable differences (jnd) in sound levels. In other words, an attentive person can distinguish the difference between 2 sound levels when the difference is at least 3 to 6 dBA. Research by Watts (76) also suggests a recom- mended duration for the sound level, but the focus of current practice is based on changes in magnitude, and not necessarily duration or frequency. Another level to this issue relates to the fact that rumble strips generate both vibration and sound. There is conflicting evi- dence that suggests the sound component may be more vital to alerting drivers than the vibration component. Bucko and Khorashadi (14) suggest that vibrations felt through the steer- ing wheel are negligible in their alerting properties compared with the noise level produced in the passenger compartment. On the other hand, Anund et al. (77) suggest that both sound and vibration contributed to drivers’ impressions of the rum- ble strips. Even though there is not necessarily agreement, current state of the practice focuses on sound levels generated by rumble strips because, even though the rationale for the recommended thresholds are difficult to determine, mini- mum sound level thresholds are provided in the literature (e.g., 3 to 6 dBA). However, similar minimum thresholds for vibration levels necessary to alert inattentive drivers do not currently exist in the literature and have not been applied in practice. In reality, it is probably a combination of both sound and vibration stimuli that provide the alerting properties of rumble strips. The weight of the contribution of either com- ponent (i.e., sound and vibration) to the alerting properties of rumble strips is unknown. To add to the complexity of the issue, the alerting properties of the sound component are likely a function of magnitude, frequency, and duration, whereas with the vibration component, the alerting properties are a function of the magnitude, frequency, direction, location, and duration. In previous research where vibration data have been collected, the focus has been on vibration magnitude and to a lesser degree on frequency and location. Location of vibration measurements is an important issue. Vibration levels of motor vehicles have been measured at numerous locations (e.g., steering wheel, right-front wheel, vehicle frame, and base of seat). Drivers experience the vibra- tion component generated from the rumble strip at their feet, seat-surface, back, and hands. Previous motor vehicle research has focused primarily on one or two of these components. With the exception of research conducted by Torbic (54) on whole-body vibration experienced by bicyclists, no efforts have been made to combine the vibration or sound compo- nents experienced by drivers of motor vehicles into a single- weighted value to rate the alerting properties of rumble strip dimensions. All of the vibration magnitudes reported in the literature are not directly comparable because of the different locations and directions of where the vibration levels were measured. Finally, research conducted by O’Hanlon and Kelley (50) suggests that the persistence of arousal following impacts with shoulder treatments was very brief. This indicates that it is not merely sufficient to establish minimum stimulation thresholds that simply arouse drivers for a short period of time, but rather it is desirable to alert a driver such that the driver’s arousal level is maintained for some extended period of time. Stimuli Levels for Effective Rumble Strips

115 As part of this research, the research team reviewed several studies that discussed the issue associated with minimum stimuli levels necessary to alert inattentive, distracted, drowsy, or fatigued drivers. The literature review did not provide definitive answers to this issue; however, the research team was not fully convinced that this topic has not been researched in other disciplines such as ITS crash avoidance, drowsy driver, and/or sleep deprivation research. There is a wealth of literature in these various disciplines, and so additional time and effort in this research was spent on reviewing the documented research related to ITS crash avoidance, drowsy driver, and sleep depri- vation to determine if a definitive answer on the minimum level of stimuli necessary to alert an inattentive, distracted, drowsy or fatigued driver has been provided elsewhere. The research team contacted agencies such as the FMCSA, the FHWA, and the National Sleep Foundation (NSF) to inquire about related research. The remainder of this section presents the results of this effort. Overview The purpose of shoulder and centerline rumble strips is to inform drivers as they inadvertently leave the travel lane that they are in danger of running off the road or colliding with oncoming vehicles. The information provided to drivers as they encounter a rumble strip comes in the form of vibra- tion. That vibration can be functionally separated into two physical sensations: auditory vibration (hereafter called noise) heard as an increase in sound magnitude (i.e., volume) and a change in frequency (i.e., pitch); and haptic or tactile vibration (hereafter called simply vibration) felt through the driver’s seat, foot pedals, floor, and steering wheel. The two types of vibration occur simultaneously and act in concert to attract driver attention. Research on in-vehicle lane departure warning devices, or so-called “electronic rumble strips,” has shown that drivers perform differently when exposed to rumble strip noise alone, vibration alone, and a combination of noise and vibration (78). However, as the noise and vibration produced by rumble strips that are part of the fixed roadway infrastructure will always operate jointly to alert drivers that they have left the travel lane, it is not ecologically valid to separately optimize the two vibration types. In other words, it does not make sense to determine the minimum noise level and the minimum vibra- tion level necessary to alert a driver, as the two vibration types will never occur in isolation, and their impact on driver percep- tion is not independent. The goal is to find the combination of vibration intensity level and noise magnitude and frequency that together will accomplish the following: • Optimize the probability that the driver will notice the rum- ble strip without causing a startle response; • Not result in damage to vehicles or infrastructure; • Not annoy residents in neighboring communities; and • Not cause problems for other highway uses (i.e., primarily bicyclists and/or motorcyclists). It must be understood that a driver’s detection of a rumble strip’s presence depends not on the absolute characteristics of the stimuli, but rather on the driver noticing a change in ambient sensation. To attract driver attention, the alerting stimuli must break through the ambient “noise” that the driver is experiencing (auditory and tactile). This ambient noise level will vary due to environmental characteristics, user character- istics, and mental states. The goal is really, therefore, to determine the combined stim- ulus characteristics of a rumble strip that, as Gustav Fechner (125) said over a century ago, “Lift the sensation or sensory difference over the threshold of consciousness.” Crossing the threshold of consciousness experienced by drivers as they encounter a rumble strip is a function of numerous variables, including the following: • Environmental variables: – Vehicle suspension, weight, and speed; – Pavement type; – Pavement profile characteristics (e.g., International Roughness Index [IRI]); and – Rumble strip dimensions (i.e., length, width, depth, and spacing). • User variables: – Adaptation, – Attention, – Hearing, and – Physical condition. Psychophysics Psychophysics is a subdiscipline of psychology dealing with the relationship between physical stimuli and their per- ception. Ernst Weber along with his student, Gustav Fechner, founded psychophysics while at the University of Leipzig in the mid-1800s. This field of study is concerned with deter- mining through experimentation how perception changes as a function of changes in physical intensity. For example, if something weighs (physical measurement) twice as much as another thing, is it perceived to be twice as heavy? For every physical stimulus there is a physical measure of intensity asso- ciated with a psychological perception for each sense modality (light intensity yields brightness; weight yields heaviness, etc.). Early work in this field resulted in the development of Weber’s law or the Weber-Fechner Law expressed as a very simple equation that can be used to determine the difference thresh- old (or difference limen—from which the term “subliminal”

116 or below threshold is derived) between two stimuli. This is what Fechner called the just noticeable difference (jnd). In Weber’s Law, the ability to notice a change in stimulus inten- sity is a function of the intensity level of the original stimulus: where “I” is the initial stimulus intensity, “ΔI” is the change in intensity or “difference threshold,” and “k” is the Weber fraction or Weber constant. This could be applied to the stimulus intensity of sound (expressed in watts/m2), but since sound as it is perceived is already commonly converted to a base 10 log scale reflecting human hearing (i.e., dBA), this is not necessary. “In fact, the use of the factor of 10 in the definition of the decibel is to create a unit which is about the least detectable change in sound intensity” (79). The change in loudness required to bring about a jnd holds nearly constant at about 1.0 dBA for moderate level stimuli, regardless of frequency (79). Sanfilipo (80) found empirical support for this in his review of human amplitude sensitivity. He wrote, “the minimum discernable changes by the human ear/brain mechanism I’ve seen in the research . . . ranged from about 0.5 dBA to 3 dBA, depending on a number of factors.” He concluded with, “I tend to use .75 dBA to 1 dBA when considering minimums.” For louder sounds above about 40 dBAs, however, research shows that the jnd can in fact drop to about 1⁄3 or 1⁄2 dBA (79) with sounds similar to those produced by rumble strips [75 or higher dBA, low-frequency sound (between 50 and 160 Hz) according to Higgins and Barbel (81)] having a jnd of about 0.5 to 0.6 dBA. This holds true if the ambient sound from the roadway is close in frequency to the sound of the vehicle driving over the rumble strip [a critical band of about 90 Hz for sounds below 200 Hz (82)]. Field measurement research discussed later indicates that this is indeed the case. FMCSA, FHWA, and NSF Interviews Representatives of the FMCSA and the FHWA were inter- viewed to determine the state of knowledge related to appropri- ate noise and vibration levels for rumble strips and to identify any current research projects that might be attempting to determine what those levels are. The FMCSA reported that no ongoing research on infrastructure, including rumble strips, was being conducted. None of the three FHWA contacts inter- ΔI I k= viewed knew of anything ongoing at FHWA related to vibration and noise levels needed to alert drivers. Contacts at the NSF did not reveal any new information to further understand the vibration and noise levels necessary to alert drivers. Field Data The following sections discuss field research that directly evaluated noise levels necessary to alert drivers and the char- acteristics of rumble strips that could produce these levels. Required Sound Levels Although he did not reference the source of the information, Outcalt (44) provided a table showing how a typical person perceives different amounts of change in sound level (Table 75). He stated that a change of 1 dBA would be imperceptible; a change of 3 dBA would be barely noticeable; a change of 6 dBA would be clearly noticeable; a 10 dBA change would be twice as loud; and a 20 dBA change would be perceived as four times as loud. Myer and Walton (83) wrote that while humans are capable of detecting changes in sound as low as 1 dBA under “ideal conditions,” for evaluating rumble strips, 3 dBA “is a more appropriate threshold for considering a difference to be practically significant.” Walton and Myer cite the “O’Hare Noise Compatibility Commission” as the source of the 3 dBA threshold. Spring (84) reported that a 4 dBA increase above ambient “is adequate to be recognized as a warning device.” Masayuki et al. (85) cited Chen (48) in stating, “warning drivers requires a sound [change] of more than 4 dBA.” Elefteriadou et al. (45) qualified these statements by concluding that “rum- ble strips which produce 4 dBA increases or above will be readily detected by motorists who are awake if the noise level is sustained for 0.35 seconds or longer.” In 2005, Mark Rosenker, Acting Chairman of the NTSB, testified before the U.S. House of Representatives about rail- road warnings, stating “if a sound is to be identified, the warn- ing signal must be 3 to 8 decibels (dBA) above the threshold of detection; if a sound is to reach the alerting level, the warning signal must be approximately 10 decibels above the ambient noise.” Similarly, Gardner et al. (43) also noted that through research on auditory perceptual factors influenc- ing the ability of train horns, Lipscomb (86) indicated that to Change in sound level (dBA) Change in apparent loudness elbitpecrepmIABd1 elbaecitonyleraBABd3 elbaecitonylraelCABd6 10 dBA About twice – or half as loud 20 dBA About four times – or one-fourth as loud Table 75. Approximate human perception of changes in sound level (Outcalt, 2001).

become aware of a sound and be alerted to the presence of that sound, the sound must typically rise 9 to 10 dBA above the sound of the environment. Green et al. (87), in a review of human factors literature associated with driver information systems, raised the amplitude above even that recommended by Rosenker by recommending a 15 dBA increase from ambi- ent for “non-speech” warning sounds as a guideline, while cautioning against absolute levels above 115 dBA to avoid approaching the pain threshold. This guideline is based on a compromise from five noise studies cited in the literature (88–92). Green et al. (87) also cited research by Berson (88) that reported sound changes above 15 dBA produce a startle reaction. In a 2002 study for Pennsylvania to evaluate the effect of shoulder rumble strips on bicycle comfort and safety, Zineddin et al. (93) examined the effect of rumble strip patterns that varied in sound level from 78 to 89 dBA at 55 mph (88 km/h). This represented increases from ambient road noise in the passenger compartment ranging from 13 to about 24 dBA. These researchers concluded that, “While the literature review uncovered research to help select rumble strip configurations capable of producing sufficient change in noise level to caution alert drivers [e.g., see Watts (76)], no data were found to indi- cate the noise level needed to arouse a fatigued, inattentive, or otherwise impaired motor vehicle operator.” They rec- ommended conducting rumble strip research using a driving simulator to test rumble strip noise and vibration “with sleep- deprived, distracted, or alcohol-impaired participants.” Rumble Strip Research Milled Versus Rolled Rumble Strips In a review of the literature, Spring (84) wrote that Perillo (23) reported a measurement of in-cab truck noise of 86 dBA for rolled rumble strips and 89 dBA for milled rumble strips at 40 mph (65 km/h). Spring stated that the 4 dBA increase was “a perceptible difference.” Spring (84) also reported that under different conditions milled shoulder rumble strips can result in 12.5 times higher vibration stimuli and 3.35 times higher auditory stimuli than rolled rumble strips. In a report on rumble strip practice and needs, Turochy (58) reported that milled rumble strips produce 3 dBA higher sound levels than rolled rumble strips. He also reported that milled rumble strips have become the preferred standard in Pennsylvania as this type of rumble strip gives contractors greater flexibility. In a review of shoulder rumble strip design for Michigan, Morena (21) stated that while both milled and rolled designs “can provide some outside noise to alert a drifting driver, the milled design produces a louder noise and adds to that a vehi- cle vibration that most certainly increases the potential for alerting the drowsy or distracted driver.” Rumble Strip Characteristics In a recent synthesis of centerline rumble strips, Russell and Rys (36) reported in-vehicle noise levels for seven test vehicles and 12 rumble strip designs at 60 mph (97 km/h). They stated that continuous 12 in. (305 mm) spacing patterns produced the highest average sound levels (80 to 94 dBA) followed by the alternating 12- and 24-in. (305- and 610-mm) spacing patterns, and that longer rumble strip patterns produced more noise. These researchers also reported on a Kansas study that surveyed driver perception of 12 and 24 in. (305 and 610 mm) spacing continuous and alternating rumble strip patterns and found that 36 percent of drivers stated that either pattern would be loud enough to get their attention. When asked about vibra- tion, only 10 percent of their subjects thought the alternating pattern produced adequate vibration; while 36 percent thought the continuous pattern had better vibration; 34 percent thought both patterns gave adequate vibration. Spring (84) recom- mended that Missouri adopt a 5-in. (127-mm)-wide rumble strip with 12 in. (305 mm) spacing, citing the Pennsylvania bicycle-tolerable rumble strip study, which suggested that this pattern was found to be preferred by bicyclists while also providing “more than adequate noise and vibration levels for motor vehicles.” Russell and Rys (36) reported that in part, depending on vehicle type, “continuous 12 in. (305 mm) on center spaced rumble strips” resulted in the greatest noise (from 80 and 94 dBA at 60 mph [97 km/h]). Russell and Rys (36) reported that a minimum of 0.315 in. (8 mm) rumble strip depth is necessary to create a “noticeable effect on tractor-trailers” and that 0.25 in. (6 mm) resulted in no noticeable change in noise or vibration. Other recent studies have demonstrated that 0.375 in. (10 mm) depth rumble strips produce sufficient noise to alert motorists. Masayuki et al. (85) concluded that the deeper the groove, the greater the noise inside the test vehicle, and the slim centerline rumble strips (0.625 in. [15 mm]) generated much more sound than did the conventional centerline rumble strips. In a recent study of the safety benefits of centerline rumble strips in Japan, Hirasawa et al. (42) found that a length of 14 in. (356 mm), width of 6 in. (150 mm), and depth of 0.5 in. (13 mm) was optimal (produc- ing in-vehicle sound level of 80 dBA) and that the deeper the groove, the louder the sound. This rumble strip pattern pro- duced sounds that were at least 15 dBA louder than the ambient pavement sound. Citing Chen’s (48) report that a minimum of 4 dBA is required to alert a driver, they concluded that their rumble strip pattern was “sufficient for warning.” Variation in In-Vehicle Sound Levels In an evaluation of the effect of rumble strip noise on local communities, Bajdek et al. (94) measured the sound level produced by various vehicles driven over an assortment of 117

118 rumble strip patterns. These researchers found increases of about 10 dBA to occur when drivers ran over the rumble strip and that the sound frequency was broadband, ranging from 125 to 1000 Hz, while frequencies on “standard pavement” typically range from 125 to 800 Hz. They found that speed, mass, and tire size all influence rumble strip sound amplitude and frequency. In a literature review Green et al. (87) found that [based on research by Potter et al., (95)]: “Interior noise [is] influenced by the state of the windows (a change of around 2 dBA at 30 mph, 5 dBA at 50 mph), use of snow or studded tires (increase of up to 8 dBA), road surface roughness (up to approximately 10 dBA), wet roads (up to 3 dBA increase), and use of the radio [which] can increase the ambient noise level on the order of 20 dBA. Aerodynamic and road/tire noise increases at a rate of about 12 dBA per doubling of vehicle speed. Engine/drivetrain noise increases at a rate of approximately 6 dBA per doubling of speed . . . Whatever the current validity of these results, interior noise levels are highly design-specific, and the acoustic environment should be determined on a case-by- case basis.” Green et al. concluded that “ambient sound levels should be tracked, and that the intensity of the auditory message should be adjusted accordingly, to be a specified amount above . . . threshold.” Summary of Key Findings After reviewing the literature and conducting interviews with several agencies, no conclusive evidence was found con- cerning the minimum stimuli levels needed to be generated by shoulder or centerline rumble strips to be effective in alerting inattentive, distracted, drowsy, or fatigued drivers. Several key findings related to this issue are as follows: • Several sources, not necessarily related to rumble strip research, indicate that humans can perceive a change in sound level intensity when the difference is as low as 1 dBA, or even lower. None of these sources suggest that a change of 1 dBA should be the minimum threshold level for the alerting properties of shoulder or centerline rumble strips. • Several sources suggest that if a sound is to reach the alerting level, then the noise should increase approximately 3 dBA, 4 dBA, 6 dBA, or 10 dBA above the ambient noise. Another source recommends a 15 dBA increase above the ambient is necessary for non-speech warning sounds. • At least one source reports that sound changes above 15 dBA could produce a startle reaction. Thus, although the primary objective of the literature review was to identify a minimum level of stimuli necessary to alert an inattentive, distracted, drowsy, or fatigued driver, the literature review revealed an upper threshold for design purposes. • The state of the practice is still focusing on designing rumble strips based on the noise levels generated by the rumble strips. This is consistent with the noise study con- ducted as part of this research. No efforts have been made to estimate the weight of a driver’s response to combina- tions of noise levels heard by the driver and vibration levels felt by the driver through contact points either at the seat, feet, or hands.