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.
6C h a p t e r 2 This chapter describes offset left-turn lanes, presents an over- view of the relevant literature, and discusses how NDS data were used to provide new insights into driver behavior at opposing left-turn lanes. Background: What are Offset Left-turn Lanes and how Do they Function? Left-turn lanes are used at intersections to provide a safe location for storing left-turning vehicles, out of the through traffic lanes, while their drivers wait for a suitable gap in opposing traffic to turn left. The provision of a left-turn lane minimizes the poten- tial for rear-end collisions with through vehicles approaching from behind the left-turning vehicle. The reduction in the risk of rear-end collisions provided by a left-turn lane also reduces the pressure on left-turning drivers to leave an exposed position and accept an inappropriate gap in opposing through traffic. Research for the Federal Highway Administration (FHWA) has documented that left-turn lanes reduce crashes by 10% to 44%, depending on the intersection type and area type (Harwood et al. 2002). These effectiveness estimates for left-turn lanes developed by MRIGlobal appear in the Highway Safety Manual (HSM) (American Association of State Highway and Transpor- tation Officials 2010). Highway medians, especially wider medians, are desirable in part because they generally have a positive effect on highway safety by providing greater separation between traffic travel- ing in opposite directions. However, wider medians may cre- ate safety concerns at intersections with conventional left-turn lanes, as vehicles in the opposing left-turn lanes may block one anotherâs views of oncoming through traffic. This type of site obstruction is illustrated in Figure 2.1. The upper portion of Figure 2.1 illustrates thatâfor a driver waiting to make a left turn in a conventional turn laneâthe view of opposing through traffic may be blocked by opposing left-turning vehi- cles. A vehicle stopped in the turn lane waiting to turn left has been referred to in the literature as an unpositioned left-turn vehicle. The lower portion of Figure 2.1 shows that, with a wider median, vehicles in the opposing left-turn lane waiting to turn left can block the view of oncoming through traffic even for a turning vehicle that has moved forward into the center of the intersection. Left-turning vehicles that have moved forward in this way have been referred to in the literature as positioned left- turn vehicles. At the intersections illustrated in Figure 2.1, the presence of opposing left-turn vehicles could cause one or both drivers to begin a left-turn maneuver without being aware of the presence of an opposing through vehicle. This could reduce the safety effectiveness of left-turn lanes, documented above. A geometric design solution for the sight obstructions that can occur as a result of opposing left-turn vehicles, like those shown in Figure 2.1, is to offset the left-turn lanes (i.e., to move the left-turn lane laterally within the median) so that the oppos- ing left-turn vehicles no longer block the sight lines of their drivers. The side-by-side drawings in Figure 2.2 illustrate inter- sections with negative offset, zero offset, and positive offset for opposing left-turn lanes. The length of gaps (in time) rejected and accepted by left- turning drivers was used as a surrogate safety measure for left-turn angle crashes. A gap is the time headway between suc- cessive vehicles, defined as the time between arrivals of the front bumper of successive vehicles at a common point, such as the center of an intersection. When a vehicle is waiting to make a left-turn, each gap in opposing through traffic is either accepted or rejected by the left-turning driver. Literature review Approximately 20% of all traffic fatalities occur at intersections (National Highway Traffic Safety Administration 2011). While only about 10% of intersections are signalized, one-third of the intersection fatalities occur at signalized intersections. Angle collisions involving vehicles crossing each otherâs paths tend to be some of the most severe crashes at intersections; they Background and Rationale
7Figure 2.1. Sight-obstructed regions for unpositioned and positioned vehicles at intersection without offset left-turn lanes. Figure 2.2. Illustration of intersection left-turn lanes with negative offset, zero offset, and positive offset.
8typically used striping treatments to narrow or shift the left- turn lane to provide offsets that were slightly less negative. The treatments in Wisconsin were found to reduce all crashes by 34%, injury crashes by 36%, and left-turn crashes by 38%, while the treatments in Florida and Nebraska showed no crash reduction. The study did not provide offset measurements at each intersection before and after the improvement, and no analysis was performed to evaluate the degree of offset that would result in improved safety (Persaud et al. 2009). Safety studies have traditionally used historical crash data to evaluate the effect of a given countermeasure by comparing crashes before and after its implementation. However, a desire to provide countermeasures proactively, rather than only after a crash pattern develops, requires that designers anticipate the likelihood of crashes before they happen. In addition, studies conducted over a short time period or at a small number of locations tend to have too small a sample size of crashes to draw conclusions, so other measures of safety must be used. Surrogate safety measures are used to give information about near misses and crash risk in simulations, and they can help predict potential crash issues before crashes occur. For left- turn maneuvers, gap acceptance is a common safety surro- gate: it stands to reason that the smaller the accepted gap, the less time there is between the clearance of the turning vehicle from the intersection and the arrival of an opposing through vehicle. When the time between those two events equals zero, one or both drivers must adjust their speed (or course), or a crash occurs. Several measures are used in the literature related to gap acceptance. Many studies evaluate the critical gap, which is the gap length equally likely to be accepted and rejected by a driver. Conditions that decrease the critical gap are those that encourage drivers to accept shorter gaps. A project conducted for FHWA investigated the potential for deriving surrogate measures of safety for existing traffic simulation models to support the safety evaluation of new countermeasures before construction and existing countermeasures in a more cost- effective manner. The study found that time to collision, defined as the time between the end of the encroachment of the turning vehicle and the projected arrival of the through vehicle if the vehicle continued along its path at a constant speed, and post- encroachment time, defined as the time between the departure of the turning vehicle from the conflict point and the actual arrival of the conflicting vehicle at the conflict point, were two of the best measures for the likelihood of a collision (Gettman and Head 2003). Chan defined left-turning conflicts by the difference in arrival time at the intersection between the left-turning vehicle and the next opposing through vehicle, which he called the trailing buffer. This is a similar measure to time to collision or postencroachment time used in other studies. Near misses were defined as occurring when the trailing buffer was 1 s, and close make up over 40% of fatal crashes at signalized intersections. About half of these crashes are left-turn crashes. FARS data for 2009 show that left-turn collisions constitute 8.5% of all traf- fic fatalities (National Highway Traffic Safety Administration 2011). A substantial proportion of these fatalities occur when a turning driverâs view of oncoming opposing through traffic is limited by the presence of another left-turning vehicle in the opposing left-turn lane. NCHRP Report 500, Volume 5: A Guide for Addressing Unsignalized Intersection Collisions, includes offsetting oppos- ing left-turn lanes as a âtriedâ strategy but provides no guid- ance on the desirable amount of offset or the effectiveness of implementing such a strategy (Neuman et al. 2003). NCHRP Report 500, Volume 12: A Guide for Reducing Collisions at Signalized Intersections, presents a strategy to provide or improve left-turn channelization, which includes a discus- sion of redesigning the intersection to provide positive visual offset as a way to improve left-turn lane geometry; but again, no specific guidance or effectiveness is provided (Antonucci et al. 2004). The AASHTO Green Book recommends that off- set left-turn lanes be used in medians wider than 18 ft, which is roughly equivalent to an offset of -6 ft, but gives no specific design guidance for the desirable reduction in negative offset (AASHTO 2004). A 1992 study provided guidance on the amount of offset required to provide unlimited sight distance for left-turning vehicles at 90-degree intersections on level, tangent roadway sections of four-lane divided roadways with 12-ft lanes. The required offset in these conditions is 2 ft when the opposing vehicle is a passenger car and 3.5 ft when the opposing vehicle is a truck (McCoy et al. 1992). Another study in the same year developed a model for determining the minimum offset requirement for ensuring adequate sight distance for left- turning drivers, which included factors for design speed and intersection geometry (Joshua and Saka 1992). These studies were simple physical models, based on intersection geometry, and did not evaluate the effect of offset on driver behavior or crashes. A study of five signalized intersections in Nebraska evalu- ated the degree to which narrowing the left-turn lane using wider pavement markings between the through and left-turn lanes affected the vehicle positioning of left-turning drivers within the lane, and therefore their view of oncoming opposing through traffic. Like the earlier studies, this study used physical models of available sight distance based on intersection geom- etry and vehicle positioning (McCoy et al. 1999). A more recent Empirical Bayes before-after study evalu- ating the safety effect of left-turn lane offset improvements at 117 signalized intersections in Florida, Nebraska, and Wisconsin showed mixed results among states. In Wisconsin, the treatment included major reconstruction to provide a positive offset; the implementations in Florida and Nebraska
9the gaps that are available, the time spent waiting to turn, light- ing, pavement and weather conditions, the turning driverâs perception of the speed and distance of the oncoming vehicles, the driverâs perception of his own and his vehicleâs capabilities, the driverâs familiarity with the intersection or similar inter- sections, and the driverâs tolerance for risk. To assist left-turning drivers in evaluating adequate gaps in oncoming traffic, researchers in California conducted a study of gap-acceptance behavior at five intersections to develop guidance for when an Intersection Decision Support System should alert drivers about gaps in oncoming traffic. The study found that no driver accepted a gap shorter than 3 s, while no driver rejected a gap greater than 12 s. Other studies agree with this range of gap length within which drivers must make a decision about whether or not to turn (Gattis and Low 1999; Madanat et al. 1994). The critical gap varied by intersection but ranged from 5.6 s to 7.6 s (Shladover et al. 2006). One study evaluated driver perceptions of the level of com- fort and degree of difficulty of making left-turn maneuvers at four intersections with left-turn offsets that ranged from neg- ative to positive. These measures were not found to improve with the increased sight distance provided by larger (i.e., more positive) offsets. Drivers were most comfortable with the -0.9-m (3-ft) offset, which the authors suggested was the most common offset found in the area, and least comfortable with the 1.8-m (6-ft) offset, which was least commonly used in the area. The authors suggest that familiarity might have had a stronger influence than available sight distance in this evaluation (Tarawneh and McCoy 1996). A study of factors that influence aggressive driving, which was measured in terms of start-up delay, gap acceptance, and acceleration or deceleration when facing an amber signal indi- cation or when changing lanes, was conducted at 10 major sig- nalized intersections near Washington, D.C. It found that the major contributor to aggressive driving was traffic operations. Being stuck in long queues, surrounded by heavy vehicles, and with increasing numbers of pedestrians and vehicles caused drivers to âlose their patienceâ (Hamdar et al. 2008). A study by Adebisi and Sama (1989) found that drivers who have to wait more than 30 s begin to take risks by accepting smaller gaps, indicating that drivers may accept smaller gaps when opposing traffic volumes are higher because fewer large gaps are available and the wait time for a suitable gap is longer. A more recent study evaluating the effect of wait time on driver left-turning behavior at a single intersection found that drivers become more aggressive as their search time for a suitable gap increases. The critical gap time was shown to decay in a linear fashion as the wait time increased (Zohdy et al. 2010). A study of the effect of weather on left-turn gap-acceptance behavior was conducted at a single signalized intersection over a 6-month period. More than 11,000 gaps for a permitted left-turn maneuver were observed; approximately 10% were encounters when it was 2 s. Depending on the site, 1% to 4% of left turns were considered near misses, but the percentage of close encounters was as high as 10%, depending on the intersection (Chan 2006). Measures related to gap acceptance not only serve as a sur- rogate for crash risk but also provide information about the operational performance of an intersection. Situations that cause drivers to wait for longer gaps reduce the number of left turns that can be made in a given time period. When sight distance for left-turning drivers is restricted, drivers may wait for longer than normal gaps on average, but they also have a higher likelihood of accepting a short gap because the oncoming opposing through vehicle cannot be seen. There- fore, the sight-distance issue that may be caused by opposing left-turn vehicles can decrease both safety and operational performance at an intersection. The AASHTO Green Book suggests using a critical gap time of 5.5 s for passenger cars making a left turn from a single lane on an undivided highway, adding 0.5 s for each additional lane crossed. This gap length was used to develop intersec- tion sight-distance guidelines (AASHTO 2004). The High- way Capacity Manual (HCM) has used a critical gap time of 4.5 for signalized intersections with a permitted left-turn phase (Transportation Research Board 2000). To identify the changes in driver behavior clearly associated with restricted sight distances, one study collected field data at a signalized intersection with the potential for sight-distance restrictions for left-turning drivers due to opposing left- turning drivers. A data set of 1,485 gap decisions was observed for 323 left turns. Of those turns, 218 were completed by drivers whose view was obstructed. Both linear regression and logis- tic regression models were developed to estimate parameters of gap acceptances. The results showed that sight obstruc- tion due to the opposite turning vehicles may contribute to significantly larger critical gaps (7.7 s versus 5.6 s) and mean accepted gaps (10.4 s versus 8.9 s). Follow-up time (the time between successive left-turning vehicles accepting the same gap) was also found to be longer when the driverâs view was obstructed. In addition, the authors evaluated gap acceptance by the gaps that were available to drivers. Drivers whose view was obstructed tended to wait for longer gaps than drivers who had clear sight lines when available gaps were larger than 4.8 s. However, when available gaps were smaller than 4.8 s, drivers with sight obstructions tended to take shorter gaps than drivers with clear sight lines, although the sample sizes for these comparisons were quite small. A review of erratic maneu- vers showed that eight of the 10 erratic maneuvers observed occurred when a left-turning driver with a restricted view of oncoming traffic took a short gap (Yan and Radwan 2007; Yan and Radwan 2008). The decision to accept a gap to complete a left-turn maneu- ver at an intersection is influenced by several factors, including
10 participants under age 30 allowed the smallest gaps, those over age 59 were the least consistent judges and were slower to clear the next lane when turning right. Older drivers may be at higher risk at intersections, especially when approaching traffic exceeds 100 km/h, through failure to detect approach- ing vehicles, poor speed and gap estimation once vehicles are detected, and slower lane clearance when turning (Parsonson et al. 1999). An FHWA study examined the effect of positive versus nega- tive offset at intersections in relation to safety of gap-acceptance behavior of older and younger drivers. A laboratory study using a video-based driving simulator was first conducted to exam- ine left-turn gap-acceptance behavior for drivers waiting to make a left turn, facing a green ball (permissive) signal. Drivers were found to be more cautious with increasing negative offset in terms of the least safe gap they were willing to accept. Older drivers (75+) accepted disproportionately higher numbers of unsafe gaps compared with younger drivers in the partial nega- tive geometry (Staplin et al. 1997). In a subsequent field study, left-turn performance of 100 sub- jects within three age groups (aged 25â45, 65â74, and 75+) was evaluated under normal driving conditions at four intersections of different left-turn offset configurations. The results showed that large negative offsets (more than 2.95 ft or 0.9 m) signifi- cantly increase the size of the critical gaps of drivers turning left and also seem to increase the likelihood of conflicts between left turns and opposing through traffic. Older drivers and women drivers were less likely than other drivers to position their vehicles within the intersection to see beyond vehicles in the opposing left-turn lane (Tarawneh and McCoy 1996). Another study used a driving simulator experiment for left- turn gap acceptance at a stop-controlled intersection to evaluate the effects of major traffic speed and driver age and gender on gap-acceptance behaviors. The experiment considered relation- ships among driversâ left-turn gap decision, driverâs acceleration rate, steering action, and the influence of the gap-acceptance maneuver on the vehicles in the major traffic stream. The experiment results showed that older drivers tend to wait for larger gaps than younger or middle-age drivers do. Male drivers appeared to accept smaller gaps than female drivers. The findings suggest that older drivers and female drivers are more conservative than the other groups. Older drivers also turned the simulator steering wheel more slowly than other drivers and were more likely to use slower acceleration rates (Yan and Radwan 2007; Yan and Radwan 2008). how NDS Data provide New Insights The needed research for establishing driver gap-acceptance behavior at intersections with and without offset left-turn lanes has not been done because the cost of collecting appropriate accepted gaps, while the remaining gaps were rejected. The study considered six combinations of weather and pavement conditions: no precipitation with dry, wet, icy, or snowy pave- ment surface; rain with wet pavement surface; and snow with snowy pavement surface. The critical gap measured for the dry weather, dry pavement condition was approximately 1.2 s shorter than for the snowy condition, which had the longest critical gap. It was also found that the critical gap for the dry weather with wet pavement condition was more than a sec- ond longer than for the dry pavement condition. The authors hypothesize that this may reflect that approaching vehicles are likely traveling at higher speeds than during other weather conditions and that left-turning drivers are more hesitant to turn in front of them on wet pavement. Critical gaps for all conditions and all three models evaluated ranged from about 6.2 s to 8.4 sâvalues substantially larger than those recom- mended in the Green Book or the HCM (Zohdy et al. 2011). The Zohdy et al. (2010) single-intersection study showed a linear increase in critical gap as rain intensity increased. Signal phasing and timing plans can also affect a driverâs gap- acceptance behavior. Left-turning drivers at intersections with protected/permissive phasing may be willing to wait for longer gaps knowing that if they do not find one, they will eventually be able to turn on a protected green arrow. Clearance inter- vals may affect left-turn decisions, as drivers waiting to turn left make assumptions about the likelihood that an opposing oncoming vehicle will slow or stop for a yellow or red light. One study found that even with modest traffic volumes, about 25% of the near misses occurred during signal transition. The author notes that âwhen signal transitions from green to amber and red, opposing traffic will mostly slow down to stop. Some drivers will initiate a left turn in front of these oncoming vehicles with the anticipation that the other vehicles will slow down and stop for themâ (Chan 2006). A driverâs age and gender have been demonstrated to play a role in gap acceptance behavior. A New Zealand study evalu- ated the effects of aging on driver behavior at T-intersections. Eighty drivers in four groups of 20 (10 males, 10 females)â respectively aged under 30 years, 40â59, 60â69, and 70 years and overâparticipated in research to identify factors contributing to rural T-intersection crashes involving older drivers. Partici- pants estimated safe gaps and speeds for traffic approaching from their right from a test vehicle parked at a right angle to the highway, simulating a T-intersection. Safe gaps for a right turn onto the highway were estimated using threshold (last possible moment) and single judgment procedures (go/not go). A laser device recorded traffic speed and distance. Each participantâs speed at turning right across the road also was tested. Drivers over age 59 had the most visual defects and the poorest neck articulation. All participants judged speed poorly, overesti- mating slower traffic and underestimating faster traffic. They used distance rather than speed in gap estimation. While
11 driverâs perspective. Other studies that have evaluated sight obstructions related to opposing left-turn lanes have sim- ply used the presence of an opposing left-turn vehicle as a surrogate for sight-distance restrictions, or they have used physical models to calculate sight distance based on assump- tions about each driverâs position. This study design allows the video data analyst to record whether an individual driverâs view was obstructed by an opposing driver during each accepted or rejected gap. NDS data also provide driver demographic information that is typically not available in fixed-camera studies. While it is available in driving simulator or recruited-driver studies, the behavior observations made during such studies are not of truly naturalistic behavior, since drivers are aware of the presence of an observer. The NDS data set provides more observations of driver behavior than a recruited-driver study typically would, records truly naturalistic driving behavior, and incorporates driver variables that would not be accessible in a fixed-camera study. In addition, the NDS data set provides the possibility of considering the interaction of effects of driver demographics and intersection geometry. Typically, recruited-driver studies can answer questions about how driver characteristics influ- ence behavior in a limited number of scenarios. Fixed-camera studies can gather large numbers of observations at a limited number of sites as well but cannot typically incorporate driver demographics. In this research, because of the availability of the NDS data set, the effect of both offset category and driver age on gap-acceptance behavior could be investigated. data at a sufficient number of intersections with various geo- metric designs through conventional methods would be quite high. Typically, studies of driver behavior at intersections use multiple cameras at fixed sites at each intersection. Placement of the cameras requires field personnel to be on site at each location. The video data must then be reduced by observers in an office setting, processed by data analysts, and then analyzed by statisticians. Within realistic budgets, field video studies with fixed-camera locations tend to provide a substantial amount of data for a relatively few sites. The NDS data provide an opportunity for the research team to bypass the expensive field data acquisition stage (since this will already have been done by SHRP 2 under other contracts) and go straight to data reduc- tion and analysis. In addition, video segments can be requested for only the time periods that include behaviors of interest (i.e., left turns completed by NDS drivers at intersection approaches appropriate for inclusion in the analysis), making video data reduction much more efficient than it would be for continu- ously recorded video data. Obtaining the NDS data from the Virginia Tech Transportation Institute (VTTI), which per- forms the queries and formats the data, requires an initial investment by the research project, but an important objec- tive of Project S08 is to demonstrate techniques for acquiring such data that can be applied in future research. The NDS data can be expected to provide data for more sites than would be possible in most fixed-site-camera field studies, given typical research budgets. In addition to the benefit of reduced costs, the NDS pro- vides the opportunity to see the traffic conditions from the
