Design Guidance for Intersection Auxiliary Lanes (2014)

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.

7 Review of Literature and State Design Guidance The following sections describe recent research efforts related to design elements and considerations for auxiliary lanes at intersections. Although much of the recent research investigated aspects of left-turn lanes as drivers approach inter- sections, there were also findings related to right-turn lanes and to acceleration lanes downstream of intersections. Literature Warrants Left-Turn Lane Installation Guidelines Although many procedures are in use by organizations to determine the need for left-turn lanes, several are either very similar or identical. The oldest research found on evaluat- ing the need for left-turn lanes at unsignalized intersections was that of M. Harmelink (3) in a paper published in 1967. His research provided the foundation for many current left- turn guidelines. Harmelink based his work on a queuing model in which arrival and service rates are assumed to fol- low negative exponential distributions. He stated that the probability of a through vehicle arriving behind a stopped left-turning vehicle should not exceed 0.02 for 40 mph, 0.015 for 50 mph, and 0.01 for 60 mph. He presented his criteria in the form of graphs, 18 in all. To use his graphs, the advanc- ing volume, opposing volume, operating speed, and left-turn percentage need to be known. Graphs for speeds of 40, 50, and 60 mph were given, as well as 5, 10, 15, 20, 30, and 40% left-turn volumes. Many variations of installation guidelines are based on Harmelink’s findings. For example, the 2004 edition of AASHTO’s A Policy on Geometric Design of Highways and Streets (commonly known as the Green Book) (4) contains a table (see Table 2-1) for use in determining the need for a left- turn lane on two-lane highways. Similar tables were also pres- ent in the 2001, 1994, 1990, and 1984 editions of the Green Book. The values in the tables are based on Harmelink’s work. In 1985, TRB published NCHRP Report 279: Intersection Channelization Design Guide (5). In that report, data from Harmelink’s work were used to establish guidelines for deter- mining the need for a left-turn lane. The following advice was provided for unsignalized intersections within new construction: 1. Left-turn lanes should be considered at all median cross- overs on divided, high-speed highways. 2. Left-turn lanes should be provided at all unstopped (i.e., through) approaches of primary, high-speed rural high- way intersections with other arterials or collectors. 3. Left-turn lanes are recommended at approaches to inter- sections for which the combination of through, left, and opposing volumes exceeds the warrants shown in a series of tables. 4. Left-turn lanes on stopped or secondary approaches should be provided based on analyses of the capacity and opera- tions of the unsignalized intersection. Considerations include minimizing delays to right-turning or through vehicles and total approach capacity. Many state DOTs also use a variation of Harmelink’s work in their design manuals, typically through reference to the Green Book or through specifically reproducing Green Book values. A review of state design manuals (6) revealed the following: • Nine state manuals either include the same table of criteria as the values included in the Green Book for determining the need for a left-turn lane or reference the Green Book. • Three of these states also include information from NCHRP Report 279 (5) or from Harmelink’s original paper (3). • Six states include the graphs available in NCHRP Report 279 along with some of the recommendations, while two more states include most of the recommendations from NCHRP Report 279, without referencing the graphs. C H A P T E R 2

8• The New Jersey manual references the Harmelink paper directly. FHWA’s Signalized Intersections: Informational Guide (7) states that in the absence of site-specific data, the designer should refer to the 2000 edition of the Highway Capacity Manual (HCM) (8), which indicates the probable need for a left-turn lane if the left-turn volume is greater than 100 vehi- cles in a peak hour, and the probable need for double left-turn lanes if the volume exceeds 300 vehicles per hour. The HCM also indicates a left-turn lane should be provided if a left- turn phase is warranted. The FHWA Signalized Intersections: Informational Guide presents several rule-of-thumb intersec- tion capacities for various scenarios where exclusive left-turn treatments may be required on one or both approaches to an intersection. In general, the guidelines adapted from NCHRP Report 279 say that exclusive left-turn lanes are needed when a left-turn volume is greater than 20% of total approach vol- ume or when a left-turn volume is greater than 100 vehicles per hour in peak periods (5). NCHRP Project 3-91 used a benefit-cost approach to determine when a left-turn lane would be justified. The steps included simulation to determine delay savings from install- ing a left-turn lane, crash costs and crash reduction savings determined from safety performance functions and acci- dent modification factors available in the Highway Safety Manual (1), and construction costs. Left-turn lane warrants were developed for rural two-lane highways, rural four-lane highways, and urban and suburban roadways. In addition, warrants for bypass lanes were developed for rural two-lane highways. A design guide on left-turn accommodations at unsignalized intersections was developed that discussed left- turn lane designs, traffic control treatments, and case study examples (9). The recommended left-turn treatment warrants developed based on that research from NCHRP Project 3-91 are provided in Appendix A; the three sets of warrants apply to • Rural two-lane highways (left-turn lanes and bypass lanes). • Rural four-lane highways (left-turn lanes). • Urban and suburban roadways (left-turn lanes). Technical warrants are an important element of the decision- making process; however, other factors should also be consid- ered when deciding whether to install a left-turn lane, including • Sight distance relative to the position of the driver. • Design consistency within the corridor. These factors should be considered in conjunction with the numerical warrants. For example, if volumes indicate that a left-turn lane is not warranted but there is insufficient sight distance at the location for the left-turning vehicles, then the left-turn lane should be considered along with other potential changes (e.g., remove sight obstructions, realign the highway). Right-Turn Lane Installation Guidelines NCHRP Report 279 (5) summarizes the then-current (mid- 1980s) practice in providing exclusive right-turn lanes (see Table 2-2). The report notes No specific warrants or guidelines are apparent for low speed, urban intersections. Engineers generally rely on capacity analyses Operating Speed (mph) Opposing Volume (veh/hr) Advancing Volume (veh/hr) 5% Left Turns 10% Left Turns 20% Left Turns 30% Left Turns 40 800 600 400 200 100 330 410 510 640 720 240 305 380 470 515 180 225 275 350 390 160 200 245 305 340 50 800 600 400 200 100 280 350 430 550 615 210 260 320 400 445 165 195 240 300 335 135 170 210 270 295 60 800 600 400 200 100 230 290 365 450 505 170 210 270 330 370 125 160 200 250 275 115 140 175 215 240 Source: A Policy on Geometric Design of Highways and Streets (2011) by AASHTO, Washington, D.C. Used by permission. Table 2-1. AASHTO installation guidelines for left-turn lanes on two-lane highways (4).

9 and accident experience when considering right-turn lanes. In rural areas, focus is primarily on a combination of through and right-turning volume. Hadi and Thakkar (10) investigated the use of through and right-turn movement volumes and speeds as criteria for installing right-turn deceleration lanes. To evaluate the need for right-turn lanes based on these criteria, research- ers used as a surrogate the percentage of through vehicles behind right-turning vehicles in the outside (right) lane that performed evasive maneuvers because of the presence of right-turning vehicles. However, this measure could not be estimated from traffic simulation models if these models were used to evaluate the need for right-turn deceleration lanes. They determined that the speed differential between through vehicles affected by right-turning vehicles and those not affected by these turns could be used to determine the need for right-turn deceleration lanes at unsignalized inter- sections; this was accomplished by using speed differential as a surrogate to safety and related to crash involvement. To determine the total speed differential caused by right-turning vehicles in the outside lane, two variables were needed: the number of through vehicles in the outside lane affected by right-turning vehicles and the average drop in speed of affected vehicles. Their research used these variables to determine criti- cal right-turn volumes that created a speed differential suf- ficient to necessitate installation of a deceleration lane based on a benefit-cost threshold. Potts et al. (11) used an economic analysis procedure to identify where installation of right-turn lanes at unsignalized intersections and major driveways would be cost effective. The researchers discussed results from their research with respect to right-turn deceleration lanes. They conducted a computer simulation study of motor vehicles and pedestri- ans at right-turn lanes to determine their operational effects. The researchers determined that right-turn maneuvers from a two-lane arterial at an unsignalized intersection or driveway could delay through traffic by 0 to 6 sec per through vehicle where no right-turn lane was present. Delays to through traf- fic due to right turns in the same situation on a four-lane arterial were substantially lower, in the range of 0 to 1 sec per through vehicle. They concluded that pedestrians at unsig- nalized intersections or driveways could have a substantial impact on delay to through vehicles as right-turning vehicles slow to yield to pedestrians, but provision of a right-turn lane could reduce pedestrian-related delays to through traffic by as much as 6 sec per through vehicle, depending on pedes- trian volume. The procedure from NCHRP Project 3-72 (11) indicated combinations of through volumes and right-turn volumes for which provision of a right-turn lane would be recommended. Researchers stated that their economic analysis procedure could be applied by highway agencies using site-specific val- ues for ADTs, turning volumes, accident frequency, and con- struction cost for any specific location (or group of similar locations) of interest. The procedure was used to develop plots that indicated combinations of through volumes and right-turn volumes for which provision of a right-turn lane would be recommended. Examples of such plots are presented in Figure 2-1. Deceleration Length Provision for deceleration clear of the through-traffic lanes is a desirable objective on arterial roads and streets and should be incorporated into design whenever practical. The Texas Urban Intersection Design Guide (12) states that the length of left-turn lanes depends on three elements: • Deceleration length. • Storage length. • Entering taper. State Condition Warranting Right-Turn Lane off Major (Through) Highway Through Volume Right-Turn Volume Highway Conditions Alaska NA DHV = 25 veh/hr Not provided Idaho DHV = 200 veh/hr DHV = 5 veh/hr 2 lane Michigan NA ADT = 600 veh/day 2 lane Minnesota ADT = 1500 veh/day All Design speed > 45 mph Utah DHV = 300 veh/hr ADT = 100 veh/day 2 lane Virginia DHV = 500 DHV = 40 mph 2 lane All DHV = 120 veh/hr Design speed > 45 mph DHV = 1200 veh/hr DHV = 40 veh/hr 4 lane All DHV = 90 veh/hr 4 lane West Virginia DHV = 500 veh/hr DHV = 250 veh/hr Divided highway Wisconsin ADT = 2500 veh/day Crossroad ADT = 1000 veh/day 2 lane Note: DHV = design hourly volume; ADT = average daily traffic; NA = not applicable. Table 2-2. Summary of state design practice in providing right-turn lanes on rural highways (5).

10 Deceleration length assumes that moderate deceleration will occur in the through-traffic lane and the vehicle entering the left-turn lane will clear the through-traffic lane at a speed of 10 mph slower than through traffic. Where providing this deceleration length is impractical, it may be acceptable to allow turning vehicles to decelerate more than 10 mph before clearing the through-traffic lane. Fitzpatrick et al. (6) developed recommendations for the approximate total lengths needed for a comfortable decelera- tion to a stop from the full design speed of the highway. These approximate lengths are shown in Table 2-3 and are based on grades of less than 3%. On many urban facilities, it is not practical to provide the full length of deceleration for a left-turn lane, and, in many cases, the storage length overrides the deceleration length. In such cases, a part of the deceleration may be accomplished before entering the left-turn lane. A 10-mph differential is commonly considered acceptable on arterial roadways. Higher speed differentials may be acceptable on collector roadways due to higher levels of driver tolerance for vehicles leaving or entering the roadway due to slow speeds or high volumes. Shorter left-turn lane lengths can increase the speed differential between turning vehicles and through traffic. Therefore, the no-speed-reduction lengths given in Table 2-3 should be accepted as a desirable goal and should be provided where practical. Torbic et al. (15) determined that drivers decelerate at rates lower than those identified in the AASHTO Green Book for freeway acceleration lanes, and they often compensate for that by beginning their deceleration in the freeway mainlanes. The result is that drivers do not consistently use the entire length of the speed-change lane for the purpose for which it is provided. It is unknown whether the same principles for deceleration lanes on lower speed freeways may also be appli- cable to deceleration lanes at intersections. Storage Length—Single Lane Long (16) reviewed the characteristics of intervehicle spac- ing for the purpose of auxiliary lane design. He concluded Economic warrant for right-turn lane (B/C=1) for four-leg unsignalized driveway or intersection on two-lane major street Economic warrant for right-turn lane (B/C=1) for four-leg unsignalized driveway or intersection on four-lane major street Economic warrant for right-turn lane (B/C=1) for three-leg driveway or intersection on two-lane major street Economic warrant for right-turn lane (B/C=1) for three-leg driveway or intersection on four-lane major street Source: Potts, I., J. Ringert, D. Harwood, and K. Bauer. Operational and Safety Effects of Right-Turn Deceleration Lanes on Urban and Suburban Arterials. In Transportation Research Record: Journal of the Transportation Research Board, No. 2023, Figure 4, p. 60 and Figure 5, p. 61. Reproduced with permission of the Transportation Research Board. Figure 2-1. Suggested right-turn lane warrants based on results from benefit-cost evaluations (11).

11 that the value of 25 ft per vehicle used by the CORSIM mod- eling software was a severe underestimation for determining queue lengths, as was the 3-ft distance between vehicles. He developed new models for estimating average queue lengths and maximum lengths at a given probability; those models were based on an intervehicle spacing of 12 ft, a passenger car length of 15 ft, a 65-ft length for combination trucks, and 30 ft for other vehicles. Lee, Rouphail, and Hummer (17) developed models to predict lane utilization factors for six types of intersections with downstream lane drops and to assess how low lane uti- lization affects the observed intersection capacity and level of service. They collected traffic and signal data at 47 sites in North Carolina. On the basis of 15 candidate factors, mul- tiple regression models were developed for predicting the lane utilization factor. They compared field-measured delays with delays estimated by the HCM with the use of regres- sion models for lane utilization. They stated that even with the new models for lane utilization, the HCM consistently overestimated delay for all types of lane-drop intersections with low lane utilization and suggested that a reassessment of the effect of lane utilization on capacity may be in order. The study also found that the downstream lane length and traffic intensity positively correlated with the lane utilization factor, existence of a two-way left-turn lane or midblock left- turn bay increased the lane utilization factor, lane drops due to lane usage change had more equal lane volume distribu- tion than the midblock taper lane drop, and some geometric variables at the approach may also influence lane utilization. Kikuchi et al. (18) examined the lengths of turn lanes when a single lane approached a signalized intersection and was divided into three lanes: left-turn, through, and right-turn. Their objective was to determine the appropriate length of each turn lane. From analysis of the vehicle queue pattern at the entrance to the turn lanes, they developed a set of formulas to compute the probabilities of the occurrence of turn-lane over- flow and turn-lane blockage. The recommended lane lengths were calculated so that the probabilities that a lane did not overflow and that the entrance of the lane was not blocked were greater than a threshold value of 0.95. Recommended turn-lane lengths, presented in a series of tables, were found to be shorter than those recommended by AASHTO. In a subsequent study, Kikuchi and Kronprasert (19) devel- oped analytical and computational processes for determining the length of the right-turn lane at a signalized intersection. They examined the factors that influenced length, reviewed available literature and practices, derived recommended lengths analytically, and developed a set of tables of recommended lane lengths as a function of approach volumes (right-turn, through-traffic, and cross-traffic volumes) and signal timing. Their analysis compared conditions when right-turn-on-red (RTOR) was not permitted and when it was permitted. Based on achieving desired probabilities of turn-lane overflow and turn-lane blockage, they calculated recommended lane lengths based on the number of vehicle spaces and described a procedure to convert that number to actual distance. They compared their guidelines that account for arrival rates of both right-turn and through vehicles to guidelines that only considered right-turn vehicles; as a result, they concluded their proposed lane lengths were different than those in exist- ing guidelines. Their recommended lengths for RTOR con- ditions were somewhat shorter than non-RTOR conditions when the right-turn arrival rate was greater than the arrival rate for through vehicles. Gard (20) conducted a study to develop a set of empiri- cal equations to accurately predict maximum queue lengths at unsignalized intersections. Using traffic data from a set of 15 intersections in California, Gard developed a series of Design Speed (mph) Deceleration Lengths (ft) from Following Sources: Deceleration Lengths from Other Manuals for Comparison Deceleration Lengths Determined Using 6.0 ft/sec2 Deceleration Rate 2004 AASHTO Green Book (4), page 714 TRB Access Management Manual (13), page 172 Florida Department of Transportation (FDOT) 2006 FDOT Design Standards (14) No Speed Reduction in Main Lanes 10-mph Speed Reduction in Main Lanes 30 170 160 170 80 35 145 230 120 40 275 275 155 290 170 45 340 185 370 230 50 410 425 240 (urban) 290 (rural) 460 290 55 485 350 550 370 60 605 405 650 460 65 460 770 550 Note: Blank cells = deceleration length not provided in reference document for the given design speed. Table 2-3. Deceleration lengths for left-turn lanes (6).

12 regression equations for the turning movements at an unsig- nalized intersection. Gard compared these equations to the procedures found in the 2000 Highway Capacity Manual, the monograph in the Institute of Transportation Engineers’ (ITE’s) 1988 Transpor- tation and Land Development, and the 2-min arrival method in the 2001 AASHTO Green Book. He found that, of the 70 data points for major-street left turns, his method correctly predicted 34% of the observations, and 84% were predicted within one vehicle. In contrast, the Green Book method pre- dicted 35% correctly and 71% within one vehicle; the other methods tended to underestimate queues, providing shorter lengths than needed to accommodate queuing traffic. In a comparison of methods similar to Gard’s, Lertworawa- nich and Elefteriadou (21) also developed a method to esti- mate storage lengths and compared it to the 2001 Green Book. The authors’ model, based on a Poisson arrival process, con- sidered service times of vehicles arriving at an empty left-turn lane and of vehicles arriving at an occupied left-turn lane but did not consider the effects of heavy vehicles. They created a series of tables of recommended storage lengths based on a threshold of the probability of overflow, and they compared their results to the Green Book. The authors found that in comparison to their Poisson model, the Green Book tended to overestimate the necessary storage lengths until the vol- umes approached capacity, while both the Green Book and Poisson methods underestimated the queue lengths. NCHRP Report 457 (22) developed suggested storage length values using a procedure similar to Harmelink’s work regard- ing storage length of left-turn bays at unsignalized intersec- tions. The storage length equation is a function of movement capacity, which is dependent on assumed critical gap and follow-up gap. Critical gap is defined by the Highway Capac- ity Manual as the minimum time interval in the major-street traffic stream that allows intersection entry for one minor- street vehicle. Thus, the driver’s critical gap is the minimum gap that would be acceptable. The time between the depar- ture of one vehicle from the minor street and the departure of the next vehicle using the same major-street gap, under a condition of continuous queuing on the minor street, is called the follow-up time. NCHRP Report 457 used a smaller critical gap (4.1 sec, as recommended in the Highway Capacity Manual, compared to the 5.0 or 6.0 sec used by Harmelink for two-lane and four- lane highways, respectively), which resulted in shorter values than those generated by Harmelink (3). The follow-up gap was assumed to be 2.2 sec as recommended in the Highway Capacity Manual. The assumptions made regarding critical gap and the resulting capacity for the movement used in these procedures can have a significant effect on the calculated storage length recommendations, as demonstrated by several researchers (20, 21, 23). The draft final report from NCHRP Project 03-91 pro- vides the following discussion regarding storage length for left-turn lanes. The left-turn lane should be sufficiently long to store the number of vehicles likely to accumulate during a critical period; the definition of that critical period can vary depending on the traffic conditions at the site. Regardless of the specific critical period, the storage length should be suffi- cient to avoid the possibility of the left-turning queue spilling over into the through lane. Kikuchi and Kronprasert (24) developed an analytical pro- cedure to determine the lengths of left-turn lanes at signal- ized intersections. They developed a general framework for determining the lengths of the left-turn lanes that prevented lane overflow and blockage of the entrance of the left-turn lane by the queued through vehicles. The framework consid- ered many factors: arrival rates and the sequence of left-turn and through vehicles, different signal schemes, and intersec- tion capacity. Signal schemes included • Non-exclusive left-turn phasing (Split-Phase). • Permissive-only (PmO). • Protected-only leading (PO-Leading). • Protected-only lagging (PO-Lagging). • Protected-permissive left-turn phasing (PPLT). The researchers stated that they identified all possible queue patterns (including the leftover from the previous cycle), and the probabilities of lane blockage and lane over- flow were obtained for different combinations of the param- eters. Using these parameters, the authors recommended lengths to prevent lane overflow and blockage of more than 95% of the cycles. Researchers (24) noted the effects of several additional variables in their comments associated with the proposed model: • Percentage of Left-Turn Volume: AASHTO’s guides con- sider only the volume of left-turn vehicles because they are concerned only about the lane overflow. The proposed model suggests the need to consider both through and left- turn volumes to prevent lane blockage as well as lane over- flow. For a large percentage of through volume, a length longer than what AASHTO suggests is recommended. • Protected and Permissive Left-Turn Signal Scheme: The permissive left-turn phase allows left-turning vehicles to clear during the green through phase. As a result, the left-turn lane for the permissive left turn is generally shorter by 10% to 40% than that for the protected left-turn movement. • Leading and Lagging Left-Turn Signal Schemes: The leading left-turn scheme requires a slightly shorter left-turn lane, 5% to 10%, than the lagging left-turn phase.

13 • Opposing Traffic for the PPLT Signal Scheme: The volume of the opposing flow has no significant effect on the left- turn lane length as long as the approach left-turn volume is small. When the derived lane length becomes substantially longer than what AASHTO recommends and the space for it is not available, the framework allows the basis to con- sider changing the signal scheme such that the increase of the length can be kept to a minimum. According to the Green Book (4), at unsignalized intersec- tions, the storage length, exclusive of taper, may be based on the number of turning vehicles likely to arrive in an average 2-min period within the peak hour. Space for at least two pas- senger cars should be provided; with over 10% truck traffic, provisions should be made for at least one car and one truck. Table 2-4 shows the recommended vehicle length by percent truck included in the TRB Access Management Manual (13). The 2-min waiting time suggested in the Green Book may need to be changed to some other interval that depends largely on the opportunities for completing the left-turn maneuver. These intervals, in turn, depend on the volume of opposing traffic, which the Green Book does not address. For additional information on storage length, the Green Book refers the reader to the Highway Capacity Manual (8). The first equation shown in Table 2-5 can be used to determine the design length for left-turn storage as described by the Green Book. For new construction, turning and opposing volumes are typically not mature; accommodation should be made for future growth, and a design year should be chosen that is appropriate for the purpose of the project. Many states use the Green Book method, a method based on work done by Harmelink (3) or a method based on work Percent Trucks Assumed Queue Storage Length (ft) per Vehicle in Queue ≤ 5 25 10 30 15 35 Source: Access Management Manual, 2003, Table 10-3. Reproduced with permission of the Transportation Research Board. Table 2-4. Queue storage length per vehicle (adapted from 13). Equation in TRB Access Management Manual Where L = design length for left-turn storage (ft); V = estimated left-turn volume, vehicles per hour (veh/hr); Nc = number of cycles per hour (for the Green Book unsignalized procedure, this would be 30 [V/N is the average number of turning vehicles per cycle]); k = factor that is the length of the longest queue (design queue length) divided by the average queue length (a value of 2.0 is commonly used for major arterials, and a value of 1.5 to 1.8 might be considered for an approach on a minor street or on a collector where capacity will not be critical; for the Green Book procedure this would be 1.0); and s = average length per vehicle, including the space between vehicles, generally assumed to be 25 ft (adjustments are available in several documents for trucks and buses, such as the TRB Access Management Manual [see Table 2-4]). Equations Used in NCHRP Report 457 Equations also used to generate values in Table 2-6 Where P(n>N) = probability of bay overflow; v = left-turn vehicle volume (veh/hr); N = number of vehicle storage positions; c = movement capacity (veh/hr); Vo = major-road volume conflicting with the minor movement, assumed to be equal to one- half of the two-way major-road volume (veh/hr); SL = storage length (ft); tc = critical gap (sec); tf = follow-up gap (sec); and VL = average length per vehicle, including the space between vehicles, generally assumed to be 25 ft (adjustments are available in several documents for trucks and buses such as the TRB Access Management Manual [see Table 2-4]). Table 2-5. Equations used to determine storage length (6).

14 by Jack E. Leisch and Associates (25) to describe their recom- mended storage lengths in their design guidelines. Others recommend that the designer assume that the intersection is signalized with a two-phase signal using a 40- to 60-sec cycle length, and then use the Highway Capacity Manual method to determine the expected storage length. The authors of NCHRP Report 348 (26) stated that the required storage length of a left-turn lane depends on the likely left-turn volumes during the peak 15 min of the design hour, which is typically but not always the morning or evening peak hour. The length for a stop-controlled lane should be adequate 95% of the time and can be estimated by using the cumulative Poisson distribution. It is generally recognized that a storage area should ade- quately store the turn demand a large percentage of the time (e.g., 95% or more, which means that the demand would exceed the storage length less than or equal to 5% of the time). A 0.5% limit was used for the major-road left-turn bay lengths in NCHRP Report 457 based on the recommenda- tion of Harmelink (3). This smaller limit reflects the greater potential for severe consequences when a bay overflows on an unstopped, major-road approach. The critical and follow- up gaps were assumed to equal 4.1 and 2.2 sec, respectively. Figure 2-2 shows storage length guidelines presented in NCHRP Report 457; the Internet version of that document also provides an interactive spreadsheet tool in which the designer can input the specific volume and gap variables to receive a recommended storage length for those conditions. NCHRP Report 457 assumed a 25-ft minimum storage length. Harmelink used larger values for critical gap (5.0 sec for two- lane highways and 6.0 sec for four-lane highways). When those gaps are used within the approach presented by Bonneson and Fontaine in NCHRP Report 457, storage lengths similar to those suggested by Harmelink are obtained. When the critical gap of 5.0 and 6.25 sec determined in NCHRP Project 3-91’s field studies are used, the storage lengths shown in Table 2-6 are generated. Each of the sources emphasized that the appropriate stor- age length is dependent on both the volume of turning traffic and the volume of opposing traffic. If volume data are not available for urban and suburban streets with lower speeds (e.g., less than 40 mph), the sources recommended that the minimum storage length be at least 50 ft to accommodate two cars; for high-speed and rural locations, a minimum storage length of 100 ft is recommended. Storage Length—Double Lanes Kikuchi, Kii, and Chakroborty (27) developed a method for estimating the needed length of double left-turn lanes (DLTLs). Their procedure first surveyed how drivers choose a lane of the DLTL in the real world and analyzed the rela- tionship between lane use and the volume of left-turn vehi- cles. Second, the procedure calculated the probability that all arriving left-turn vehicles during the red phase could enter the left-turn lanes (i.e., no queue spillback of vehicles from the DLTL and no blockage of the DLTL by the queue of through vehicles). This probability was presented as a function of the length of the DLTL and the arrival rates of left-turn and through vehicles. The adequate lane length was derived such that the probability of the vehicles entering the DLTL was greater than a threshold value. Third, the recommended ade- quate length was expressed in number of vehicles and then converted to the actual distance required based on the vehicle mix and preference between the two lanes. Resulting recom- mended lengths were presented as a function of left-turn and through volumes for practical application. The Green Book states that if double left-turn lanes are used, the length required for storage is approximately half that required for single left-turn lanes (4). Taper Two distinct tapers are commonly defined in many guide- lines: approach taper length and bay taper length. An approach taper provides space for a left-turn lane by moving traffic later- ally to the right on a street or highway without a median. The bay taper length is a reversing curve along the left edge of the traveled way that directs traffic into the left-turn lane. Illus- trations of the use of these tapers along with how the left- turn lane is added to the roadway are shown in the following figures: • Figure 2-3 shows a left-turn lane added within a median. • Figure 2-4 shows a left-turn lane added to an undivided two-lane highway where the through lane on the same approach as the added turn lane was shifted to the right 0 20 40 60 80 100 120 140 160 0 50 100 150 200 250 300 St or ag e Le ng th (ft ) Turn Movement Volume (veh/h) Adequate 99.5% of the time. Conflicting volume =1050 veh/h 850 650 450 250 Figure 2-2. Recommended storage lengths for left- turn lanes at uncontrolled approaches using a 25-ft minimum storage length along with a critical gap of 4.1 sec (22).

15 Left-Turn Volume (veh/hr) Storage Length, Rounded Up to Nearest 25-ft Increment (ft) Storage Lengths from Other Manuals for Comparison Storage Lengths Calculated from Equations b Documented in NCHRP Report 457 Using Revised Critical Gaps and 0.005 Probability of Overflow Green Book Procedure (k = 1)a Equation (k = 2)a Opposing Volume (veh/hr) 200 400 600 800 1000 Critical Gap = 5.0 sec, Follow-Up Gap = 2.2 sec (Represents the 50th Percentile Critical Gap Found in Field Studies) 40 75 75 50 50 50 50 50 60 50 100 50 50 50 50 50 80 75 150 50 50 50 50 50 100 100 175 50 50 50 50 75 120 100 200 50 50 50 75 75 140 125 250 50 50 50 75 75 160 150 275 50 50 75 75 100 180 150 300 50 50 75 75 100 200 175 350 50 75 75 100 125 220 200 375 50 75 75 100 125 240 200 400 75 75 100 125 150 260 225 450 75 75 100 125 175 280 250 475 75 75 100 125 175 300 250 500 75 100 125 150 200 Critical Gap = 6.25 sec, Follow-Up Gap = 2.2 sec (Represents the 85th Percentile Critical Gap Found in Field Studies, 85th Percentile Is Preferred for Design) 40 75 75 50 50 50 50 50 60 50 100 50 50 50 50 50 80 75 150 50 50 50 50 75 100 100 175 50 50 50 75 75 120 100 200 50 50 75 75 100 140 125 250 50 50 75 100 125 160 150 275 50 75 75 100 150 180 150 300 50 75 75 125 150 200 175 350 50 75 100 125 200 220 200 375 75 75 100 150 225 240 200 400 75 75 125 150 275 260 225 450 75 100 125 175 325 280 250 475 75 100 125 200 400 300 250 500 75 100 150 225 525 Note: This table assumes 25 ft per vehicle spacing. Table 2-4 provides other suggested spacing lengths based on percent trucks. a, b See Table 2-5 for equations. Table 2-6. Recommended storage lengths from Access Management Manual equation and NCHRP Report 457 equations with revised critical gap (6). Bay Taper StorageDeceleration Figure 2-3. Left-turn lane within a median (6).

16 of the full width of the turn lane. This condition is known as a fully shadowed left-turn lane. Where this configura- tion is used, it is important that designers follow typical guidelines for lane-addition tapers. • Figure 2-5 shows a partially shadowed left-turn lane where both through lanes are shifted to provide the needed space for the turn lane. With partially shadowed left-turn lanes, the offset created by the approach taper does not entirely protect or “shadow” the turn lane (5). • Figure 2-6 shows the condition when a lane is added to the outside edge of the approach and the through driver must change lanes to continue traveling straight; otherwise, the driver would be in the left-turn lane. This condition is also known as a bypass lane. Some agencies avoid this layout because of the mixed message to drivers between passing lanes and this condition. For passing or truck climbing lanes, the new added outside lane is for slower-moving traffic, and the inside existing lane is for faster-moving vehicles. For the configuration shown in Figure 2-6, the opposite situation is present; the new added outside lane is for the faster-moving traffic, and the inside existing lane is for the vehicles that are slowing and perhaps stopping while waiting to make the left turn. Tapers for Left Turns (Bay Taper) On high-speed highways, it is common practice to use a taper rate between 8:1 and 15:1 (L:T) (4). Long tapers approximate the path drivers follow when entering a left-turn lane from a high-speed through lane. However, long tapers tend to entice some through drivers into the deceleration lane—especially when the taper is on a horizontal curve. Long tapers also constrain the lateral movement of a driver wanting to enter the turn lanes. For urban areas, short tapers appear to produce better tar- gets for the approaching drivers and to give more positive identification of an added left-turn lane. Short tapers are pre- ferred for deceleration lanes at urban intersections because of slow speeds during peak periods. The total length of taper and the deceleration length should be the same as if a longer taper were used. This results in a longer length of full-width pavement for the auxiliary lane. This type of design may reduce Bay Taper Approach Taper StorageDeceleration Figure 2-4. Fully shadowed left-turn lane (6). Bay Taper Approach Taper StorageDeceleration Figure 2-5. Partially shadowed left-turn lane (6).

17 the likelihood that entry into the left-turn lane may spill back into the through lane. Municipalities and urban counties are increasingly adopting the use of taper lengths such as 100 ft for a single-turn lane and 150 ft for a double-turn lane for urban streets (4). Some agencies permit the tapered section of deceleration left-turn lanes to be constructed in a squared-off or shadowed section at full paving width and depth, particularly where a very short taper is applied. This configuration involves a painted delineation of the taper. The abrupt squared-off beginning of deceleration exits offers improved driver com- mitment to the exit maneuver and also contributes to driver security because of the elimination of the unused portion of long tapers. The design involves transition of the outer or median shoulders around the squared-off beginning of the deceleration lane. The Green Book provides advice regarding taper design. The recommended straight-line taper rate is 8:1 (L:T) for design speeds up to 30 mph and 15:1 (L:T) for design speeds of 50 mph. Straight-line tapers are particularly applicable where a paved shoulder is striped to delineate the left-turn lane. Short straight-line tapers should not be used on curbed urban streets because of the probability of vehicles hitting the leading end of the taper with the result- ing potential for a driver losing control. A short curve is desirable at either end of long tapers but may be omitted for ease of construction. Where curves are used at the ends, the tangent section should be about one-third to one-half of the total length. Tapers for Through Traffic (Approach Taper) Though left-turn lanes can be added so that the merge taper guides turning vehicles into the turning lane, certain locations instead use the taper to guide through traffic to the right of the turning lane, as shown in Figure 2-4 and Figure 2-5. Such treatments are often used in rural conditions where it is ben- eficial to provide added protection and/or guidance to turn- ing vehicles, particularly at isolated T-intersections, where there is no median in which to install a shadowed left-turn lane, and/or where right-of-way is limited. At these locations, through traffic is directed to shift its path, while turning traf- fic can travel straight into the turning lane. The approach taper is commonly estimated by one of the equations shown in Table 2-7. Table 2-7 also shows comparisons for various speeds and offsets. While the design guidelines described in the previous sec- tion pertain to the use of a bay taper for turning vehicles, similar principles apply when the taper is used to shift the path of through traffic. The taper should be short enough to provide sufficient visual clues to the driver that through traffic must shift, but be long enough to allow the shift to take place at the expected or prevailing operating speed of the road- way. Pavement markings and supplemental signs should be Note: Some agencies recommend against using this layout because drivers must change lanes to continue traveling straight; otherwise, the driver would be in the left-turn lane. Transition Taper Transition Taper Figure 2-6. Direct entry into left-turn lane (also known as bypass lane) (6). Design Speed (mph) Condition Equation Approach Taper Length (ft) 6-ft Offset 12-ft Offset 20 Typically used for low-speed approaches (e.g., 40 mph and less) L = WS2/60 40 80 30 90 180 40 160 320 50 Typically used for high-speed approaches (e.g., greater than 40 mph) L = WS 300 600 60 360 720 70 420 840 Note: W = width of offset (ft), S = speed (mph). Table 2-7. Typical length of approach taper to add left-turn lanes (6).

18 used to reinforce the action the driver is expected to take. It is important that appropriate vehicle storage length be pro- vided because, with no median protection, the excess queue will extend into the through travel lane. This is one reason why this treatment is more commonly found at isolated intersections with low turning volumes. For similar reasons, it is also important that sufficient deceleration length be pro- vided in the design of the turning lane. Deceleration and Storage Length for Right-Turn Lanes The design guide developed as part of NCHRP 3-89 (28) notes that speed-change lanes (both for deceleration and acceleration) may be provided to minimize deceleration and acceleration in the through travel lanes. The document also notes: There are no generally established criteria concerning where deceleration and acceleration lanes should be provided in con- junction with channelized right-turn lanes. The AASHTO Green Book does not give definitive warrants for the use of speed- change lanes, but identifies several factors that should be consid- ered when deciding whether to implement speed-change lanes: vehicle speeds, traffic volumes, percentage of trucks, capacity, type of highway, service provided, and the arrangement and fre- quency of intersections. The NCHRP 3-89 Design Guide (28) does not provide guidance regarding right-turn storage length. Lane or Median Width FHWA’s Highway Design Handbook for Older Drivers and Pedestrians (29) states that two factors can compromise the ability of older drivers to remain within the boundaries of their assigned lanes during a left turn. One factor is the diminishing ability to share attention (i.e., to assimilate and concurrently process multiple sources of information from the driving environment). The other factor involves the abil- ity to turn the steering wheel sharply enough, given the speed at which they are traveling, to remain within the boundaries of their lanes. Data sources cited by the handbook’s authors indicated that a 12-ft lane width provides the most reasonable tradeoff between the need to accommodate older drivers, as well as larger turning vehicles, without penalizing the older pedestrian in terms of exaggerated crossing distance. The handbook’s corresponding recommendation is for a mini- mum receiving lane width of 12 ft, accompanied, wherever practical, by a shoulder of 4 ft minimum width. Potts, Harwood, and Richard (30) investigated the rela- tionship between lane width and safety for roadway segments and intersection approaches on urban and suburban arteri- als. Their research found no general indication that the use of lanes narrower than 12 ft (3.6 m) on urban and suburban arterials increased crash frequencies. Researchers stated that this finding suggested that geometric design policies should provide substantial flexibility for use of lane widths narrower than 12 ft (3.6 m). They added that inconsistent results sug- gested increased crash frequencies with narrower lanes in three specific design situations: • Lane widths of 10 ft (3.0 m) or less on four-lane undivided arterials. • Lane widths of 9 ft (2.7 m) or less on four-lane divided arterials. • Lane widths of 10 ft (3.0 m) or less on approaches to four- leg stop-controlled arterial intersections. The researchers recommended that narrower lanes should be used cautiously in these three situations unless local expe- rience indicates otherwise. The width of auxiliary lanes should preferably match the width of the through lanes, although they should be at least 10 ft wide (4). If curbs are present, a curb offset of 1 to 2 ft from the edge of the travel lane to the face of the curb should be used. A suggestion by others, such as (30), is that when selecting the width of auxiliary lanes, designers should care- fully consider when a narrower lane may not be appropri- ate in combination with other factors; those factors could include high operating speeds, substantial truck volumes, or restrictive alignments. To accommodate a single left-turn lane, a median width of 18 ft—a 12-ft lane width plus a 6-ft divider—is recom- mended. The 6-ft divider may provide a refuge for pedestri- ans, depending on its design; however, it is not sufficient to fully offset the turn lane (discussed below). If double left-turn lanes are used, the median opening and crossroad should be sufficiently wide to accommodate both incoming lanes; a median width of 28 to 30 ft—11- to 12-ft lanes plus a 6-ft divider—is recommended (12). As part of NCHRP Project 3-72, Potts et al. (31) investi- gated the relationship of lane width to saturation flow rate on urban and suburban approaches to signalized intersections. Average headways were measured, and then saturation flow rates were calculated at signalized intersection approaches with lane widths of 9.5 to 14 ft. They concluded that satura- tion flow rate does indeed vary with lane width. Average satu- ration flow rate was in the range of 1,736 to 1,752 passenger cars (pc)/h/ln for 9.5-ft lanes, 1,815 to 1,830 pc/h/ln for 11- to 12-ft lanes, and 1,898 to 1,913 pc/h/ln for lane widths of 13 ft or greater. These measured saturation flow rates were gener- ally lower than those used in the Highway Capacity Manual. Furthermore, the percentage difference in saturation flow rate between sites with 9.5- and 12-ft lanes was about half the

19 value used in the HCM. Because data were limited to queue lengths between 8 and 11 vehicles, the research results did not directly address queue lengths longer than 11 vehicles. Simultaneous Left Turns Flexibility in signalization is provided if the left-turn movements are separated (12, 32). This separation, if suffi- cient, can allow concurrent double left-turn phases. Separate double left-turn phases eliminate the potential problem of overlapping vehicle paths in the intersection. Safety Harwood et al. (33) conducted a study to investigate the safety effectiveness of left- and right-turn lane treatments. The research team collected geometric design, traffic control, traffic volume, and traffic crash data at 280 improved sites in eight states and at 300 similar intersections not improved during the study period. The types of improvement proj- ects evaluated included installation of added left-turn lanes, installation of added right-turn lanes, installation of left- and right-turn lanes as part of the same project, and extension of the length of existing left- or right-turn lanes. Based on the results of the analyses, they concluded the following: • Adding left-turn lanes is effective in improving safety at signalized and unsignalized intersections. Installing a single left-turn lane on a major-road approach would be expected to reduce total intersection accidents at rural unsignalized intersections by 28% for four-leg intersections and by 44% for three-leg intersections, with corresponding reduc- tions of 27% and 33% at urban unsignalized intersections. At four-leg urban signalized intersections, installation of a left-turn lane on one approach would be expected to reduce accidents by 10%, and installation on both major- road approaches would be expected to increase, but not quite double, the resulting effectiveness measures for total intersection accidents. Table 2-8 shows a summary of the crash modification factors (CMFs), as presented in Bon- neson et al. (34). • Positive results can also be expected for right-turn lanes, with reductions in total intersection accidents of 14% at rural unsignalized locations and 4% at urban signalized locations for installations on single approaches. Installa- tion of right-turn lanes on both major-road approaches to four-leg intersections would be expected to increase, but not quite double, the resulting effectiveness measures for total intersection accidents. Table 2-9 shows a summary of the crash modification factors, as presented in Bonneson et al. (34). • In general, turn-lane improvements at rural intersec- tions resulted in larger percentage reductions in acci- dent frequency than comparable improvements at urban intersections. • Overall, there was no indication that any type of turn-lane improvement is either more or less effective for different accident severity levels. In NCHRP Project 17-21, researchers determined state and local agency design practices and policies related to unsignalized median openings for U-turns. After seven cat- egories of midblock and intersection median designs were identified, the research team assessed the designs’ effects on safety through field observation and crash data analysis for 115 unsignalized median opening sites with both crash and field data. This knowledge was transferred into design guidelines and a method for comparing the expected safety Intersection Type Number of Intersection Legs Number of Major-Road Approaches with Left-Turn Lanes Installed One Approach Both Approachesc All Crashes Severe Crashes All Crashes Severe Crashes Rural Signalized 3 0.85 0.86a not applicable 4 0.82 0.83b 0.67 0.69 Rural Unsignalized 3 0.56 0.45 not applicable 4 0.72 0.65 0.52 0.42 Urban Signalized 3 0.93 0.94d not applicable 4 0.90 0.91 0.81 0.83 Urban Unsignalized 3 0.67 0.65e not applicable 4 0.73 0.71 0.53 0.50 a Value estimated as 0.83 = 0.91 / 0.90 × 0.82 using urban intersection data from Harwood et al. (33). b Value estimated as 0.86 = 0.91 / 0.90 × 0.85 using urban intersection data from Harwood et al. (33). c CMFs for “both approaches” estimated as the square of the “one approach” CMFs. d Data not available from Harwood et al. (33). Value estimated using “all crash” data as 0.94 = 0.91 / 0.90 × 0.93. e Data not available from Harwood et al. (33). Value estimated using “all crash” data as 0.65 = 0.71 / 0.73 × 0.67. Table 2-8. Crash modification factors for adding a left-turn lane at various intersection types (34).

20 performance of different designs to enable engineers in set- ting policy, establishing project-level design, and discussing the impacts of medians with business and property own- ers. As documented in NCHRP Report 524 (35), researchers made the following conclusions: • As medians are used more extensively on arterial highways, with direct left-turn access limited to selected locations, many arterial highways experience fewer midblock left- turn maneuvers and more U-turn maneuvers at unsignal- ized median openings. • Field studies at various median openings in urban arterial corridors found estimated U-turn volumes of no more than 3.2% of the major-road traffic volumes at those locations. At rural median openings, U-turn volumes were found to represent, at most, 1.4% of the major-road traffic volumes at those locations. • Accidents related to U-turn and left-turn maneuvers at unsig- nalized median openings occurred very infrequently. The 103 median opening study sites on urban arterial corridors experienced an annual U-turn plus left-turn crash average of 0.41. Twelve median openings on rural arterial corridors had an annual average crash total of 0.20. Overall, at these median openings, U-turns represented 58% of the median opening movements and left turns represented 42%. Based on these limited crash frequencies, researchers concluded that there was no indication that U-turns at unsignalized median open- ings constituted a major safety concern. • For urban arterial corridors, median opening crash rates were substantially lower for midblock median openings than for median openings at three- and four-leg inter- sections. For example, the crash rate per million median opening movements (U-turn plus left-turn maneuvers) at a directional midblock median opening was typically only about 14% of the median opening crash rate for a direc- tional median opening at a three-leg intersection. • Crash rates at directional median openings on urban arte- rial corridors were lower than at traditional median open- ings, and conventional three-leg median openings had lower crash rates than corresponding four-leg openings. • Where directional median openings were considered as alternatives to conventional median openings, two or more directional median openings were usually required to serve the same traffic movements as one conventional median opening. Therefore, researchers concluded that design deci- sions should consider the relative safety and operational efficiency of all directional median openings in comparison with the single conventional median opening. • Analysis of field data found that for most types of median openings, most observed traffic conflicts involved major- road through vehicles having to brake for vehicles turning from the median opening onto the major road. • At urban unsignalized intersections, the research found that installation of a left-turn lane on one approach would be expected to reduce accidents by 27% for four-leg inter- sections and by 33% for three-leg intersections. • The minimum spacing between median openings then used by highway agencies ranged from 152 to 805 m (500 to 2,640 ft) in rural areas and 91 to 805 m (300 to 2,640 ft) in urban areas. In most cases, highway agencies used spac- ings between median openings in the upper end of these ranges, but there was no indication that safety problems resulted from occasional use of median opening spacings as short as 91 to 152 m (300 to 500 ft). Fitzpatrick, Schneider, and Park (36) conducted a study to determine variables that affected the speeds of free-flow turning vehicles in an exclusive right-turn lane and explore Intersection Type Number of Intersection Legs Number of Major-Road Approaches with Left-Turn Lanes Installed One Approach Both Approachesb All Crashes Severe Crashes All Crashes Severe Crashes Rural Signalized 3 0.96 0.91a not applicable 4 0.96 0.91 0.92 0.83 Rural Unsignalized 3 0.86 0.77 not applicable 4 0.86 0.77 0.74 0.59 Urban Signalized 3 0.96 0.91 not applicable 4 0.96 0.91 0.92 0.83 Urban Unsignalizedc 3 0.86 0.77 not applicable 4 0.86 0.77 0.53 0.50 a Harwood et al. (33) did not quantify CMFs for signalized intersections with three legs. They recommended the application of the “four-leg” CMFs to intersections with three legs. b CMFs for “both approaches” estimated as the square of the “one approach” CMFs. c Harwood et al. (33) did not quantify CMFs for urban unsignalized intersections. They recommended that the CMFs developed for rural four-leg unsignalized intersections can also be used for urban unsignalized intersections. Table 2-9. Crash modification factors for adding a right-turn lane at various intersection types (34).

21 the safety experience of different right-turn lane designs. Their evaluations found that the variables affecting the turn- ing speed at an exclusive right-turn lane included the type of channelization present (either lane line or raised island), lane length, and corner radius. Variables that affected the turning speed at an exclusive right-turn lane with island design included (a) radius, lane length, and island size at the beginning of the turn and (b) corner radius, lane length, and turning roadway width near the middle of the turn. The authors compared their study to previous research and found that treatments that had the highest number of crashes were right-turn lanes with raised islands; while in their analysis, they found this type of intersection had the second high- est number of crashes of the treatments evaluated in this study as well. In both studies, the “shared through with right lane combination” had the lowest number of crashes. They recommended that these findings be verified through use of a larger, more comprehensive study that includes right- turning volume. NCHRP Project 16-04 (37) was initiated to develop (a) design guidelines for safe and aesthetically pleasing roadside treat- ments in urban areas and (b) a toolbox of effective roadside treatments to (1) balance the safety and mobility needs of pedestrians, bicyclists, and motorists and (2) accommodate community values. In fulfilling the first of those objectives, researchers discussed auxiliary lanes as an item for consid- eration. They stated that although many auxiliary lanes have low volumes and may be included as part of a clear zone in the urban environment, higher speed auxiliary lane locations, such as extended length right-turn lanes, are common loca- tions for run-off-road crashes. A lateral offset of 6 ft from the curb face to rigid objects is preferred, and a 4-ft minimum lateral offset should be maintained. Offset Left-Turn Lanes The FHWA’s Highway Design Handbook for Older Drivers and Pedestrians (29) recommends that for new or reconstructed facilities, unrestricted sight distance, achieved through positive offset of opposing left-turn lanes, be provided whenever pos- sible. This recommendation is made in anticipation of pro- viding a margin of safety for older drivers who, as a group, do not position themselves within the intersection before initiating a left turn. Where the provision of unrestricted sight distance is not feasible, positive left-turn lane offsets are recommended to achieve the minimum required sight dis- tances appropriate for major roadway design speed and type of opposing vehicle. Vehicles in opposing left-turn lanes can limit each other’s views of approaching traffic. The restriction on the sight dis- tance is dependent on the amount and direction of the offset between the opposing left-turn lanes. The offset is measured between the left edge of a left-turn lane and the right edge of the opposing left-turn lane. Benefits of positive offset left-turn lanes include • Better visibility of opposing through traffic. • Improved unprotected left-turn phase. • Decreased possibility of conflict between opposing left- turn movements within the intersection. • Service for more left-turn vehicles in a given period of time (particularly at signalized intersections). The impact on pedestrian crossings of all roadways should be considered in the design of offset left-turn lanes. Greater right-of-way width is typically required to offset left-turn lanes, but research has shown that offset left-turn lanes can provide significantly greater sight distance for left- turn maneuvers, a particularly critical maneuver for older drivers (38). Guidelines were developed for offsetting oppos- ing left-turn lanes at 90-degree intersections on level, tangent sections of divided roadways with 12-ft lanes (39). They are applicable to left-turning passenger cars opposed by either another passenger car or a truck. The desirable offsets are those that provide opposing left-turning vehicles with unrestricted sight distances, and therefore, they are independent of design speed. The guidelines include minimum and desirable offsets for when both vehicles are unpositioned, and for when the left- turning vehicle is unpositioned and the opposing left-turning vehicle is positioned. Positioned vehicles enter the intersection to obtain a better view of oncoming traffic, while unpositioned vehicles remain behind the stop line while waiting to turn left. A previous study found that 60% of older drivers did not posi- tion their vehicle. Therefore, in areas with high percentages of older drivers, the guidelines based on both vehicles being unpositioned should be used. Likewise, in areas where there are high percentages of trucks, the guidelines based on the oppos- ing left-turning vehicle being a truck should be used. Increasing the width of the lane line between the left- turn lane and the adjacent through lanes can also improve the sight distance by encouraging the driver to position the vehicle closer to the median. McCoy et al. (40) developed a method for determining the width of the left-turn lane line. Two types of offset left-turn lanes are typically used: parallel and tapered. Parallel lanes may be used at both signalized and unsignalized intersections, while tapered lanes are usually used only at signalized intersections. Tapered offset left-turn lanes are normally constructed with a 4-ft nose between the left-turn and the opposing through lanes. This median nose can be offset from the opposing through traffic by 2 ft or more with a gradual taper, making it less vulnerable to contact by the through traffic. This type of offset is especially effective for the turning radius allowance where trucks with long rear overhangs, such as log- ging trucks, are turning from the mainline roadway. This same

22 type of offset geometry may also be used for trucks turning right with long rear overhangs (4). Parallel and tapered offset left-turn lanes should be separated from the adjacent through- traffic lanes by painted or raised channelization. Adequate advance signing is essential so that drivers recognize the need to enter the turn lane well in advance of the intersection. Results of a 1996 study by Tarawneh and McCoy (39) indi- cated that driver performance can be adversely affected by offsets that are much less (i.e., more negative) than -2.95 ft. Such large negative offsets significantly increased the size of the critical gaps of drivers turning left and also seemed to increase the likelihood of conflicts between left turns and opposing through traffic. Large negative offsets may be par- ticularly troublesome for older drivers and women drivers, who were less likely to position their vehicles within the inter- section to see beyond vehicles in the opposing left-turn lane. The same 1996 study had a somewhat counterintuitive find- ing. Driver perceptions of the level of comfort were not found to improve with greatly increased offsets. An offset of 5.9 ft was associated with a lower level of comfort and a higher degree of difficulty perceived by drivers than an offset of -2.95 ft, even though the latter provides less sight distance. The study’s authors speculated that this reaction might be because the -2.95-ft offset is more common than the 5.9-ft offset. While the literature supports the use of offset left-turn lanes, there has been little evaluation of the safety effective- ness of this strategy; a recommendation from the literature is that an investigation is needed to thoroughly evaluate the effectiveness of offset improvements for left-turn lanes in reducing crash frequency and severity at signalized inter- sections. The safety effectiveness of offset improvements for left-turn lanes is explored empirically in an FHWA Pooled Fund study (41, 42) to provide better support to the states when selecting safety improvements at signal- ized intersections. Geometric, traffic, and crash data were obtained for 92 installations in Nebraska, 13 in Florida, and 12 in Wisconsin, as well as for some untreated refer- ence sites in each state. To account for potential selection bias and regression-to-the-mean, an Empirical Bayes (EB) before-after analysis was conducted to determine the safety effectiveness of improving the offset for left-turn lanes. Researchers observed a large difference in effects among the three states, which they said may be explained, in part, by the wide variation in offset improvements. Florida and Nebraska employed pavement marking adjustments or minor construction to improve the offset, but while the offset was improved at each site, most improvements did not result in a positive offset. Wisconsin, however, reconfig- ured left-turn lanes through major construction projects, resulting in substantial positive offsets. Results in Florida and Nebraska showed little or no effect on total crashes, but Wisconsin showed significant reductions in all crash types investigated—total (34%), injury (36%), left-turn (38%), and rear-end (32%). As part of the Pooled Fund study (42), a disaggregate anal- ysis was conducted for Nebraska, the only state with enough installations to disaggregate the results. The analysis revealed that the percent reduction in crashes increased as the expected number of crashes increased. An economic analysis was con- ducted to identify the level of expected crashes that would yield a crash benefit to justify the construction cost. Based on this analysis, researchers concluded that offset improvement through reconstruction was cost effective at intersections with at least nine expected crashes per year and where left- turn lanes are justified by traffic volume warrants. This study did not address offset improvements at unsignalized inter- sections. While there are examples of offset improvements at unsignalized intersections in the United States, the authors noted that these results should not be extrapolated to that situation. Rather, they concluded that it is more appropri- ate to conduct a separate evaluation once there are sufficient installations at unsignalized intersections Offset Right-Turn Lanes Many transportation agencies have started using offset right-turn lanes (ORTLs) at two-way stop-controlled inter- sections in the hope of improving driver safety by providing intersection departure sight distance triangles that eliminate through-roadway right-turning vehicle obstructions. A proj- ect in Nebraska (43) was initiated to determine when con- struction of an offset auxiliary right-turn lane is cost effective. The researchers determined that actual ORTL installa- tions were scarce, leaving little opportunity to develop a robust set of observations. However, they did develop a data collection protocol and collected and analyzed data from a small number of study sites. Results of driver behavior stud- ies at existing locations of offset right-turn lanes indicated that drivers were not performing as expected at parallel-type ORTLs, rendering their presence useless. Taper-type ORTLs appeared to be much more intuitive to driver expectancy and appropriate for the three-dimensional characteristics of all vehicle types. Researchers identified specific negative driver behaviors and recommended appropriate Manual on Uniform Traffic Control Devices (MUTCD)-compliant traf- fic control devices to mitigate misleading visual cues and accentuate elements that reinforce the intended positive behavior at ORTL intersections for successful use of the lat- erally offset right-turn auxiliary lane. At the completion of their project, researchers made the following conclusions regarding ORTLs: • There are too few installations to allow safety studies. Ide- ally, this subject would be a good topic for an NCHRP

23 study because it could use multiple study sites across the nation to collect a large amount of data for a robust statis- tical analysis. • There are no geometric guidelines for designers to use when deciding key elements of three-dimensional features of the offset right-turn lane that can generate poor choices by through, right-turning, and left-turning drivers on major roads and stopped drivers at minor road approaches of two-way stop-controlled intersections exhibiting ORTLs. • Guidelines for typical auxiliary lanes do not appear to be transferrable to ORTLs. • Optimal guidelines for three-dimensional geometric road- way features evolve over time after the study of behaviors generated by drivers given unfamiliar features in an iterative manner. Intersection Sight Distance Yan and Radwan (44) developed sight distance geometric models for unprotected left-turning vehicles at parallel and taper left-turn lanes. For parallel left-turn lanes, they calcu- lated available sight distance as follows: 2 2 2 SD V D L m n g V V D n g e V m f t w f w ( ) ( ) = + + + − − − × + + + + − Where SD = available sight distance (ft); Vf = distance from the eye of the driver to the front of the vehicle (ft); D = distance between stop bars of opposing left lanes, which is composed of the width of pedestrian cor- ridors and the width of the minor road (ft); Lt = width of the opposing through lane (ft); m = width of the median (ft); n = width of the median nose (ft); g = distance from the left side of the left-turn vehicle to the left lane line (ft); Vw = width of the opposing left-turn vehicle (ft); and e = distance from the eye of the driver to the left side of the vehicle (ft). The authors produced a table that calculates available sight distance for common dimensions of parallel oppos- ing left-turn lanes. They also produced similar equations and tables for parallel lanes with offsets and for tapered left-turn lanes. The models and related analyses focused only on the unprotected phases of a signalized intersection, but the principles are comparable for sight visibility at unsignalized intersections. The authors cautioned that the absence of stop bars at unsignalized intersections discouraged a direct appli- cation of these models because left-turning vehicles’ posi- tions could be more flexible before crossing the opposing through traffic. Channelization FHWA’s Highway Design Handbook for Older Drivers and Pedestrians (29) recommends raised channelization with slop- ing curbed medians rather than channelization accomplished through the use of pavement markings, for the following oper- ating conditions: • Left- and right-turn lane treatments at intersections on all roadways with operating speeds of less than 40 mph. • Right-turn treatments on roadways with operating speeds equal to or greater than 40 mph. Where raised channelization is implemented at intersec- tions, FHWA also recommends that median and island curb sides and curb horizontal surfaces be treated with retro- reflectorized markings and be maintained at a minimum luminance contrast level of 2.0 with overhead lighting or 3.0 without overhead lighting. The design guide developed as part of NCHRP 3-89 (28) lists the following reasons for providing a channelized right- turn lane: • To increase vehicular capacity at intersections. • To reduce delay to drivers by allowing them to turn at higher speeds. • To reduce unnecessary stops. • To clearly define the appropriate path for right-turn maneu- vers at skewed intersections or at intersections with high right-turn volumes. • To improve safety by separating the points at which cross- ing conflicts and right-turn merge conflicts occur. • To permit the use of large curb return radii to accom- modate turning vehicles, including large trucks, without unnecessarily increasing the intersection pavement area and the pedestrian crossing distance. A channelized right-turn lane consists of a right-turning roadway at an intersection, separated from the through travel lanes of both adjoining legs of the intersection by a chan- nelizing island. The NCHRP Project 3-89 Design Guide (28) states: • Curbed islands are considered most favorable for pedes- trians because curbs most clearly define the boundary between the traveled way intended for vehicle use and the island intended for pedestrian refuge.

24 • Orientation and mobility specialists have a strong prefer- ence for raised islands with cut-through pedestrian paths because they provide better guidance and information about the location of the island for pedestrians with vision impairment than painted islands. • When right-turn volumes are high and pedestrian and bicycle volumes are relatively low, capacity considerations may dictate the use of larger radii, which enable higher- speed, higher-volume turns. Increasing the radius of a channelized right-turn roadway reduces right-turn delay by approximately 10 to 20% for each 5-mph increase in turning speed. • Small corner radii, which promote low-speed right turns, are appropriate where such turns regularly conflict with pedestrians, as higher speeds have been shown to result in a decrease in yielding to pedestrians by motorists. The align- ment of a channelized right-turn lane and the angle between the channelized right-turn roadway and the cross street can be designed in two ways (as illustrated in Figure 2-7): – A flat-angle entry to the cross street (island shaped like an equilateral triangle, often with one curved side). This design is appropriate for use in channelized right-turn lanes with either yield control or no control, such as locations with an acceleration lane, for vehicles at the entry to the cross street. – A nearly right-angle entry to the cross street (island shaped like an isosceles triangle). The nearly right-angle entry design can be used with Stop sign control or traf- fic signal control for vehicles at the entry to the cross street; yield control can also be used with this design where the angle of entry and sight distance along the cross street are appropriate. Effect of Skew Skew angles are an important design element for any approach to an intersection, but for certain auxiliary lanes (e.g., uncontrolled right-turn lanes) they are even more of a factor. Son, Lee, and Kim (45) developed a method for cal- culating sight distance available to drivers at skewed unsig- nalized intersections. The method considered that the sight distance may vary depending on (1) driving positions of the drivers and (2) different lines of sight given to drivers by dif- ferent types of vehicles with unique sight-line obstructions. They derived equations and nomographs for calculating available sight distance that included the influence of factors such as geometry (e.g., intersection angle, lane width, shoul- der width, and position of stop line), vehicle dimensions, and the driver’s field of view. They concluded that findings from their research provided evidence that a skew angle greater than 20 degrees should not be used in design when the design vehicle is a large vehicle or semitrailer because available sight distances were less than the required stopping sight distance, even with a low value of design speed for intersection angles less than 70 degrees. The Highway Design Handbook for Older Drivers and Pedes- trians (29) recommends establishing 15 degrees as a mini- mum skew angle as a practice to accommodate age-related performance deficits at intersections where right-of-way is restricted; at skewed intersections where the approach leg to the left intersects the driver’s approach leg at an angle of less than 75 degrees, the prohibition of RTOR is recommended. The handbook cites multiple studies documenting restricted neck movement in older drivers, making detection of and judgments about potential conflicting vehicles on crossing Channelized Right-Turn Lane With Flat-Angle Entry to Cross Channelized Right-Turn Lane With Nearly Right-Angle Entry to Cross Street Source: FHWA PEDSAFE, 2013 Guide, found at: http://www.pedbikesafe.org/PEDSAFE/ Figure 2-7. Typical channelized right-turn lane with differing entry angles to the cross street (adapted from 46).

25 roadways much more difficult. The Highway Safety Manual (1) includes crash modification factors for skew angle. A 2013 study (47) developed crash modification factors for three-leg and four-leg intersections and concluded that the minimum critical angle for intersections in roadway design policies should be revised to 75 degrees. Multiple Turn Lanes Several projects have documented experiences with mul- tiple turn lanes over the last 25 years, but a comprehensive study on optimal design features is not available. This section will summarize some of the experiences documented in the literature. Wortman (48) conducted a state-of-the-art review on double left-turn lanes for the Arizona Department of Trans- portation in the 1980s. In it, he cited work from Neuman (5) that suggested double left-turn lanes should be considered at any signalized intersection with high left-turn design-hour demand volumes and a general rule-of-thumb of 300 veh/hr or more as the appropriate demand volume for consideration of the double left-turn lanes. Other guidelines from Neuman include the following: • The throat width for the turning traffic is the most impor- tant design element. Drivers are most comfortable with extra space between the turning queues of traffic. Because of the offtracking characteristics of vehicles and the relative difficulty for acceptance of two abreast turns, a 36-ft throat width is desirable for acceptance of two lanes of turning traffic. In constrained situations, 30-ft throat widths are acceptable minimums. • Guiding pavement markings to separate the turning lanes are recommended. The then-current MUTCD recommended 2-ft long dashed lines with 4-ft gaps to channelize turning traffic. These channelization lines should be carefully laid out to reflect offtracking and driving characteristics. Wortman also cited some studies that showed that the capac- ity of either lane in a double left-turn lane approach was lower than a single left-turn lane (e.g., a lane in a double left-turn might have 80% of the capacity of a single left-turn lane) and that the inside (median-side) left-turn lane had a somewhat lower capacity than the outside (shoulder-side) left-turn lane. Brich (49) examined positive guidance pavement markings for double left-turn lanes in Virginia. In that study, he cited another finding from Neuman (5) that double left-turn lanes operate at approximately 1.8 times the capacity of a single left-turn lane. Hurley (50) developed a mathematical model, based on field data, to predict auxiliary lane use for an intersection that uses double left-turn lanes onto a freeway entrance ramp, where the right lane is dropped in advance of the merge area. He compared this condition with downstream lane reduction to the typical intersection configuration in the Highway Capacity Manual and found that the collected data indicated that lane utilization factors were consider- ably larger than the default value presented in the HCM. He added that this was not unexpected since the Highway Capacity Manual does not take downstream lane reduc- tion into consideration. However, for intersections with long turn and auxiliary lane lengths and high flow rates, he expected that at least some of the data would fall some- what close to the HCM default value. This did not hap- pen, even though outside left-turn lane lengths as long as 119 m, downstream auxiliary lane lengths as long as 463 m, and total left-turn flow rates as high as 1,288 veh/hr were encountered at the intersections studied. NCHRP Report 505 (51) discusses characteristics of intersection geometry to accommodate trucks. The authors referenced the existing Green Book guideline that the desir- able turning radius for a double left-turn lane is 90 ft, but they concluded that there was little else on the design of multiple turn lanes to account for turning trucks. They stated that the primary factor to consider in designing double left-turn lanes is vehicle offtracking or swept path width. When vehicles negotiate the turn side by side, the vehicles should not encroach on the adjacent travel lane. Because many factors affect the control turning radius of double left-turn lanes, they stated that it is necessary to provide guidance on the range of offtracking or swept path width of design vehicles for various turning radii. They determined offtracking and resultant swept path widths of several design vehicles for 90-degree turns with centerline turning radii of 50, 75, 100, and 150 ft using AutoTURN software. Based on their analysis, they recommended that an exhibit be included in the Green Book that indicates the swept path width of several design vehicles for centerline turning radii of 75, 100, and 150 ft, to provide flexibility in designing adequate turning paths for double left-turn lanes by allowing for interpolation of swept path widths for a range of turning radii. Cooner et al. (52) conducted research in Texas on design and operations of triple left-turn (TLT) and double right-turn (DRT) lanes. Previously existing guidelines were nearly non- existent, so the research was conducted to develop guidelines for elements such as geometric design, appropriate signs and markings, and signal timing. The field studies in Texas col- lected both static (e.g., lane widths, grades, pavement mark- ings, traffic signs, upstream and downstream conditions, and signal timings) and dynamic (e.g., volumes by lane, satura- tion flow, and critical events) data in order to evaluate design and operational performance. Researchers collected the data at five TLT and 20 DRT lane sites, primarily in the Dallas–Fort

26 Worth and Houston urban areas. They reported the following as key findings for TLT lanes: • Lane utilization patterns were varied for each of the five sites studied. • All sites were T-intersections with peak-hour volumes from 646 to 2,846 vehicles. • Lighted pavement markers used to delineate the lane lines between the TLT lanes were effective at reducing violations and well received by the public at one site. • Saturation flow rates in Texas were consistent with earlier published national values. They also reported these important operational findings for DRT lanes: • Most vehicles used the outside lane (closest to the curb) to make their right turns. • Peak-hour volumes ranged from a low of 200 to a high of almost 1,000 vehicles. • Lane utilization (inside vs. outside) was comparable when the right-turn volumes were high. • Saturation flow rates were higher in the inside lane (aver- age = 1,717 veh/hr versus the outside lane at 1,668 veh/hr) and also generally lower than those at TLT sites. • The impact of trucks in the inside lane was greater than when in the outside lane. A review of crash diagrams along with a field conflict study and comparison study led the authors to conclude that the TLT lanes they studied did not experience any major safety issues. They also concluded that, in general, a well-designed DRT lane does not cause significantly higher crash frequency or severity compared to single right-turn lanes. Some of the key recommendations based on the research include • TLT lanes should be considered when turning volumes exceed 600 veh/hr. • DRT lanes should be considered when turning volumes exceed 300 veh/hr. • Clear turning guide lines (i.e., dotted lines marking the turning path) are highly recommended for both sides of the inside right-turn lane when the intersection has a turn- ing angle greater than 90 degrees. • Narrow DRT lanes (i.e., turning roadway width less than 30 ft) with channelization should not be used. • RTOR is not advised for the inside lane when there are more than two receiving lanes. • Designers should avoid installing DRT lanes near access points (e.g., corner gas stations). • If an auxiliary receiving/acceleration lane is provided for the curb right-turn lane at channelized double turn lanes, its length should not be less than 150 ft. • For closely spaced intersections, if a downstream intersec- tion uses DRT lanes, the outside (curb) lane should not be aligned with any through lane at the upstream intersection. Researchers added that TLT and DRT lanes are not appro- priate for all situations and an operational analysis should support their use. They suggested that other techniques (e.g., grade separation, signal timing) might be better solutions for a particular site, especially when considering the effects of adjacent intersections, pedestrian/bicycle movements, and other key factors. Bypass Lanes NCHRP Project 3-91 (6) examined existing guidelines for installation of bypass lanes, also called blister lanes, at intersections on two-lane rural highways. Bypass lanes (see Figure 2-8) are provided to relieve congestion due to left- turning vehicles in the travel lane. This alignment requires the through driver to change lanes to continue through the intersection; it may be used where right-of-way is constrained but a left-turn lane is warranted. Researchers developed sug- gested left-turn treatment installation warrants based on results from benefit-cost evaluations for rural two-lane high- ways. Warrant information based on a benefit-cost ratio of 1.0 as a threshold is available in Fitzpatrick et al. (6). Though the bay taper length is not needed on a bypass lane, it is still necessary to provide the necessary approach taper to Transition Taper Transition Taper Dotted Line Solid Line Broken Line Figure 2-8. Example of bypass lane with markings (6).

27 guide through traffic around the left-turn lane. The typical length of the approach taper for a bypass lane is the same as that used for a shadowed or partially shadowed lane when through traffic is shifted to the right. Depending on the con- figuration of the intersection, it is also necessary for a bypass lane to have a departure taper as the through traffic is brought back into its original alignment. For the example shown in Figure 2-8, the departure taper for eastbound through traffic would be equal to the approach taper necessary to add the left-turn lane for the westbound turning traffic. The depar- ture taper associated with the intersection in Figure 2-6 is illustrated in Figure 2-8. A similar departure taper would be appropriate at a T-intersection where a left-turn lane is added in only one direction. In the case of Figure 2-8, if there is no south leg of the intersection, and an eastbound left-turn lane is installed, the same departure taper would bring the through traffic back into the two-lane alignment east of the intersection. Bypass lanes are more commonly used at T-intersections than at intersections with four legs because traffic is being diverted in only one direction with no cross traffic on the stem of the T. Use of bypass lanes at four-leg intersections must incorporate proper guidance to approaching drivers so that they do not continue through the turning lane into oncoming traffic. In addition, appropriate sight distance must be provided so that through drivers in the bypass lane have a clear view of through and left-turning cross traffic, and vice versa. Regardless of the number of legs at the intersection, it is often useful to provide additional guidance to through driv- ers that they need to change lanes at bypass lanes. This can be accomplished through broken pavement markings or dotted lines that have a much shorter stroke length and shorter spac- ing than standard markings that permit passing. This mark- ing is commonly called a “skip-stripe” and is illustrated in Figure 2-8. This marking reinforces the message that through drivers must change lanes but still permits left-turn drivers to travel straight into the turning lane. Passing Lanes An increasingly common treatment on rural two-lane highways is the addition of periodic passing lanes on corri- dors that may not be ready for expansion to a full four-lane alignment. These corridors are known by different names in different locations, but they are generally defined as two-lane rural highways in which periodic passing lanes have been added to allow passing of slower vehicles and the dispersal of traffic platoons. Passing lanes are typically provided in both directions of travel, either alternating from one direction of travel to the other or operating side by side within a section of roadway, allowing passing opportunities in both directions. In some parts of the United States, particularly in Texas, such corridors are known as “Super 2” corridors, while in Europe they are commonly known as “2+1” corridors. Super 2 projects can be introduced on an existing two-lane roadway where there is a significant amount of slow-moving traffic, there is limited sight distance for passing, and/or the exist- ing traffic volume has increased, thus creating the need for vehicles to pass on a more frequent basis. Although Super 2 corridors have demonstrated benefits in reducing delay and improving opportunities to pass, their operational performance at intersections is not as well known. A recent Texas project by Brewer et al. (53) examined Super 2 operations for corridors with higher volumes (ADT of 5,000 to 15,000 veh/day). Driveways were not a primary focus of the project, but researchers conferred with Texas Department of Transportation (TxDOT) engineers to deter- mine common practices, compared with findings from field studies and simulation. Researchers ultimately made only a general recommendation regarding the location of driveways, stating that where practical, designers should avoid substan- tial traffic generators such as state highways or high-volume county roads or driveways within passing lanes, or consider providing an auxiliary lane (for left turns or right turns, as applicable) if the traffic generator falls within the limits of a Super 2 passing lane. An NCHRP-sponsored scan tour looked at characteris- tics of 2+1 roads in several European countries to determine the potential applications of the design for use in the United States. The NCHRP authors (54) concluded that the benefits of 2+1 roads in Europe validated a recommendation for their use in the United States, to serve as an intermediate treatment between an alignment with periodic passing lanes and a full four-lane alignment. They also recommended that 2+1 roads were most suitable for level and rolling terrain, with installa- tions to be considered on roadways with traffic flow rates of no more than 1,200 veh/hr in a single direction. The authors recommended that major intersections should be in the buf- fer or transition areas between opposing passing lanes, with the center lane used as a turning lane. Alternative Intersection Designs Some alternate designs have been proposed and imple- mented to change the configuration of intersections to improve the efficiency and/or safety of turning movements. One such design is the crossover-displaced left-turn (XDL) intersec- tion, also called the continuous-flow intersection (CFI). The fundamental design principle of the XDL intersection involves displacement of the left-turn lane to the other side of the opposing through lanes several hundred feet upstream of the intersection. The displaced left-turn lanes are aligned parallel to the through lanes at the intersection. This design

28 results in the simultaneous movement of left-turning traffic with through traffic at the intersection. The key trade offs are the need for additional right-of-way to accommodate the displaced lanes and the creation of several smaller ancillary intersections around the primary intersection, which must also be maintained with signing and marking. This design is primarily intended for signalized intersections as an alterna- tive to grade separation, but there may be possible applica- tions for unsignalized intersections at high-speed locations. Jagannathan and Bared (55) modeled the performance of three sample XDL intersections in comparison to conven- tional intersections and found that average intersection delay, average number of stops, average queue length, and capacity all improved with the XDL. Roundabouts are growing in popularity in the United States, after having been developed in the 1960s in the United Kingdom and used at numerous intersections in other countries, primarily throughout other parts of Europe as well as Australia. Two key characteristics of the modern roundabout include a requirement for entering traffic to yield to circulating traffic and geometric constraints that slow entering vehicles. One result is that traditional left turns are eliminated as all intersection traffic travels around the circulatory roadway in the same direction. A recent NCHRP project (56) examined the safety and operation of roundabouts in the United States with the purpose of producing a set of operational, safety, and design tools cali- brated to U.S. roundabout field data. The researchers found that, with the exception of conversions from all-way-stop- controlled intersections, where crash experience remains statistically unchanged, roundabouts have improved both overall crash rates and, particularly, injury crash rates in a wide range of settings (e.g., urban, suburban, and rural) and previous forms of traffic control (e.g., two-way stop and sig- nal). Statistical analysis revealed a 35% reduction in crashes for all sites studied. Overall, single-lane roundabouts have better safety performance than multilane roundabouts. The safety performance of multilane roundabouts appears to be especially sensitive to design details, such as lane width. Multilane roundabouts have accessibility requirements that single-lane roundabouts do not have. The 2011 Proposed Guide- lines for Pedestrian Facilities in the Public Right-of-Way (57), submitted by the United States Access Board in its Notice for Proposed Rulemaking, contains a requirement that an accessible pedestrian-activated signal (e.g., a traditional traffic control signal or a pedestrian hybrid beacon) should be provided for each multilane segment of each pedestrian street crossing, including the splitter island. Such signals are also required to clearly identify which pedestrian street crossing segment the signal serves. The accessibility require- ments will have an effect on operations as well as construc- tion and maintenance costs. Rodegerdts et al. (56) further concluded that drivers at roundabouts in the United States appear somewhat tenta- tive, using roundabouts less efficiently than models suggest is the case in other countries around the world. In addition, the number of lanes has a clear effect on the capacity of a round- about entry; however, the fine details of geometric design— lane width, for example—appear to be secondary and less significant than variations in driver behavior at a given site and between sites. Although the project was unable to estab- lish a strong statistical relationship between speed and safety, the importance of controlling speed in roundabout design is well established internationally. Anecdotal evidence suggests the importance of considering design details in multilane roundabout design, including vehicle path alignment, lane widths, and positive guidance to drivers through the use of lane markings. Acceleration Lanes To increase the capacity of through traffic at signalized intersections, additional lanes with limited length—called auxiliary lanes—are added to the roadway on the approach and departure of the intersection. These are often added for vehicles turning right off the arterial at the intersec- tion, though they may also be used by through vehicles and by vehicles turning right and accelerating onto the arterial. Because of their limited length, as well as other factors, these lanes are not as fully used as other continuous through lanes. Tarawneh (58) conducted a research project to (1) observe and identify the level of use of auxiliary through lanes added at intersections of four-lane, two-way roadways and (2) study the effects of auxiliary lane length, right-turn vol- ume, and through/right-turn lane group delay on the level of their use. He collected lane use data during 1,050 saturated cycles at eight signalized intersections with different auxil- iary lane lengths, and all factors investigated—auxiliary lane length, right-turn volume, and stopped delay—were found to contribute significantly to the use of auxiliary lanes at the 0.01 level. The level of each factor’s contribution, however, was dependent on the level of the other two. Longer auxil- iary lanes, lower right-turn volumes, and excessive approach delays encouraged the use of auxiliary lanes by straight- through vehicles, and observed lane use by through vehicles ranged between one and seven vehicles per cycle. The range of lane utilization adjustment factors (fLU-factors) calculated from field data was 0.73 to 0.82, which was lower than the then-current 1997 Highway Capacity Manual (59) default value of 0.91 for a three-lane through/right-turn group. Other studies have reviewed characteristics of accelera- tion lanes and speed-change lanes, but those studies typically focus on freeway interchanges instead of at-grade intersec- tions. One study in particular, by Torbic et al. (15), revealed

29 that drivers neither accelerate at constant rates, nor at rates as high as those identified in the AASHTO Green Book. In addi- tion, drivers in free-flow conditions do not typically use the entire length of the speed-change lane when it is provided. It is not clear whether the same principles for lower speed acceleration lanes on freeways may also be applicable to accel- eration lanes for intersections. The design guide developed as part of NCHRP Project 3-89 (28) notes: Acceleration lanes provide an opportunity for vehicles to com- plete the right-turn maneuver unimpeded and then accelerate parallel to the cross-street traffic prior to merging. The addition of an acceleration lane at the downstream end of a channelized right-turn lane can reduce the right-turn delay by 65 to 85%, depending on the conflicting through traffic volume, and may be considered where right-turn delay is a particular problem. Channelized right-turn lanes with acceleration lanes appear to be very difficult for pedestrians with vision impairment to cross. Therefore, the use of acceleration lanes at the downstream end of a channelized right-turn lane should generally be reserved for locations where no pedestrian or very few pedestrians are present. Typically, these would be locations without sidewalks or pedestrian crossings; at such locations, the reduction in vehicle delay resulting from addition of an acceleration lane becomes very desirable. Design Tools Kindler et al. (60) described the development of an expert system for diagnostic review of at-grade intersections on rural two-lane highways. This system, the Intersection Diagnostic Review Module (IDRM), was developed as a component of the Interactive Highway Safety Design Model (IHSDM) to aid designers in assessing the safety consequences of geo- metric design decisions, particularly for combinations of geometric features. IDRM was developed to allow such problems to be identified and evaluated in an automated and organized fashion. IDRM identifies concerns by using models of the criticality of specific geometric design situa- tions. These include existing geometric design models—like sight distance models—as well as newly developed models. IDRM uses 21 specific models to address 15 high-priority issues related to the intersection as a whole and to individual approach legs. IDRM makes no attempt to select a particu- lar treatment as appropriate to the intersection. After further investigation, the IDRM user may select a particular treat- ment as appropriate on the basis of the available evidence and engineering judgment, or the user may conclude that no treatment is necessary and that the project should be built as designed. Thorough documentation of the development of IDRM is found in a separate FHWA report (61). The report focuses on documenting the knowledge base developed for the IDRM software. It also documents the software in that it identifies the knowledge structure, problem definitions, models, decision algorithms, formulas, and parameter values implemented in the software. Schurr et al. (62) developed a model to describe design speed profiles of vehicles traversing horizontal curves on approaches to stop-controlled intersections on two-lane two-way rural highways. They used the model to create a pro- cedure for designing horizontal curves that would accommo- date vehicles transitioning from high speeds to a stop. Based on speed profile data from 15 study sites in Nebraska, the researchers concluded that posted speed, median type, pres- ence of rumble bars, roadway surface condition, and degree of rutting did not significantly affect the vehicle speed profiles at these sites at a 95% confidence level. They also concluded that the intercepts of the regression lines for approaches with and without horizontal curves were significantly different in the case of heavy vehicles. The speed of heavy vehicles on tan- gent approaches was generally about 8 mph higher than on sites that exhibited horizontal curvature, though the rate of deceleration remained almost the same until vehicles were near the stop. Passenger cars exhibited no statistically sig- nificant difference between curved and tangent alignments. Researchers used the results of the study to develop a proce- dure for determining the minimum curve radius appropriate for a roadway alignment approaching a stop ensuring that (1) the visual expectations of the driver were met, (2) the comfort of the passengers within the vehicle was optimized, (3) the curve design used a simple curve with no spirals, (4) the vehicle speed within the limits of the curve was rea- sonable, (5) sufficient braking distance to the stop was avail- able, and (6) deceleration rates were reasonable. Although this project did not specifically focus on auxiliary lane design, the guidelines related to deceleration profiles of vehicles on stop-controlled approaches could be relevant to vehicles approaching a turning lane. Ongoing and Recently Completed Research Researchers reviewed some sources for additional research projects not complete during Task 2. A summary of selected research projects is provided in this section. Auxiliary through lanes beyond signalized intersections are recognized as an approach to increase the intersection capacity through their efficient use. The benefits from the use of auxil- iary through lanes at signalized intersections can be realized in the presence of equally distributed traffic over the lanes prior to the intersection. Use of auxiliary through lanes beyond a signalized intersection has been seen throughout the United States. Prior studies suggest that the length of the auxiliary lane beyond the intersection is a significant factor affect- ing the upstream lane usage and therefore the intersection

30 capacity. However, the conditions for their effective use and their effect on safety, operation, and the environment have yet to be documented. Thus, research is needed to provide tech- nical assessment of their use, document their effect on safety and operations, and develop guidelines including design crite- ria and placement. A recent NCHRP project was tasked with assisting traffic engineers in adopting criteria for the effective and safe use of intersection auxiliary through lanes. The objec- tives of the research were to provide guidelines and procedures to analyze, justify, and design auxiliary through lanes at sig- nalized intersections. The results of the research report were published in 2011 as NCHRP Report 707 (63) and NCHRP Web-Only Document 178 (64). Many transportation agencies use channelized right-turn lanes to improve operations at intersections, although their effect on safety for motorists, pedestrians, and bicyclists is not clear. The Americans with Disabilities Act (ADA) requires that all pedestrian facilities, including sidewalks and crosswalks, be accessible to pedestrians with disabili- ties. The U.S. Access Board has published draft rights-of- way guidelines requiring pedestrian signals at channelized turn lanes. Research in NCHRP Report 674 (65) addressed this issue. Regardless of the outcome of that research, some agencies may remove existing channelized right-turn lanes and avoid constructing new ones. Guidance is needed to help make these decisions based on reliable data on their safety impacts. NCHRP Project 3-89 (66) was initiated to develop design guidance for channelized right-turn lanes, based on balancing the needs of passenger cars, trucks, buses, pedes- trians (including pedestrians with disabilities), and bicycles. For this project, a channelized right-turn lane is character- ized by separation from the through and left-turn lanes on the approach by an island and separate traffic control from the primary intersection. The channelized right-turn lane may or may not have a deceleration lane entering it, and it may have a merge or an auxiliary lane at the exiting end. The revised final report has been received and will be published as an NCHRP report. A current NCHRP project (67) is examining existing access management policies and guidelines. These policies often include elements such as guidelines for auxiliary lanes and turning lanes at intersections. The objective of this study is to develop research-based guidelines for access management. This research will culminate in model access management and design guidelines and procedures for various roadway classifications and design types. It will also address how these criteria might vary in the context of different roadside envi- ronments. The resulting guidelines will be accompanied by detailed rationale on their benefits and application so they may be readily adapted and applied by state transportation agencies and local governments or metropolitan planning organizations (MPOs) through their transportation plan- ning and design processes. Review of Online Design Manuals As part of Task 2 efforts, the research team conducted a state-of-the-practice review of current design considerations. A review of state online design manuals was conducted to identify (1) what is being discussed at the state level and (2) current design criteria being used for intersection auxiliary lanes. The review focused on the following design elements and policy components: • Queue storage length. • Entering taper length. • Deceleration length. • Turn-lane width. • Channelization/island design. • Offset left-turn lanes. • Double (or dual) left-turn lanes. • Bypass lane for left turns. • Right-turn lanes. • Advice/warrants on when to install. • Pedestrians. • Free-flow or channelized right turn. • Reference to Green Book in policy. Researchers found 42 state manuals on line, 36 of which contained information on one or more of the items in the above list. Table 2-10 summarizes the guidance provided in the state manuals.

31 Qu eu e S tor ag e Le ng th En te rin g Ta pe r Le ng th D ec el er at io n Le ng th Tu rn -L an e W id th Ch an ne liz at io n/ Is la nd D es ig n O ffs et L ef t-t ur n La ne s D ou bl e (or D ua l) Le ft- Tu rn L an es B yp as s L an e (L eft T urn ) R ig ht -T ur n La ne s W he n to In sta ll A dv ic e Pe de str ia ns Fr ee R ig ht T ur n R ef er en ce G re en B oo k Green Book Alaska Arizona California Colorado Connecticut Delaware Florida Georgia Idaho Illinois Indiana Iowa Kansas Kentucky Louisiana Maine Massachusetts Minnesota Mississippi Missouri Montana Nebraska Nevada New Jersey New York North Carolina Ohio Oregon Pennsylvania South Dakota Tennessee Texas Utah Virginia Washington Wisconsin Note: No online manual found for Alabama, Arkansas, Hawaii, Maryland, New Mexico, Rhode Island, South Carolina, Wyoming. Online manual found, but with no information on auxiliary lanes, for Michigan, New Hampshire, North Dakota, Oklahoma, Vermont, West Virginia. Table 2-10. Summary of auxiliary lane design guidelines.