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.

77 Deceleration Field Study Background The objective of NCHRP Project 03-102 was to recommend improvements to the guidance provided in the AASHTO Policy on Geometric Design of Highways and Streets (commonly known as the Green Book) (2). Task 1 was to identify the state of design practice for auxiliary lanes at intersections by gathering and synthesizing information on existing practices and research. Task 2 determined issues that merit further study to validate, enhance, and expand current Green Book guidance. Task 3 produced the interim report and a Phase II work plan, which includes suggested research efforts for the remaining tasks. This chapter documents the method and findings for the decelera- tion study selected for Phase II of NCHRP Project 03-102. Green Book Review on Deceleration (Section 9.7.2) The 2011 Green Book discusses deceleration lanes in Sec- tion 9.7.2, and it does so in two ways: first by defining them and then describing what makes up each of their three com- ponent sections (i.e., deceleration length, storage length, and taper length). Table 6-1 lists potential research needs based on a review of Green Book material and related literature. The definition section is new to the 2011 Green Book and also discusses a fourth component: distance traveled during perception-reaction (PR) time. Item 1 in Table 6-1 could be addressed by referencing one or more previous studies on the subject. Item 2 deals with assumptions made on calculating appro- priate deceleration distance. Section 9.7.2 of the 2011 Green Book states that a desirable objective on arterial streets is the provision for deceleration clear of the through-traffic lanes, and Table 9-22 in the same section provides estimates of the needed distances; however, the Green Book states that the lengths in the table should be accepted as a “desirable” goal and should be provided where “practical.” The Green Book section presents justification for accepting a 10-mph speed differential between turning vehicles and through traffic on arterial roadways and states that higher speed differentials may be acceptable on collector highways and streets. The notes within Green Book Table 9-22 provide information that can be used to generate the values, but it was not known whether the 10-mph differential was realistic for current drivers or whether the other assumptions made in the cal- culations needed to be updated. Feedback from the project oversight panel and from the state-of-the practice question- naire also suggested that these questions merited further study. Item 3 is concerned with appropriate storage length val- ues. Several projects, some recently completed, would help to support information contained in the Green Book. Item 4, also related to storage length, refers to the capacity of mul- tiple turn lanes, which was a topic better addressed within the double left-turn lanes operational field study. Objective and Measures of Effectiveness The objectives for this research study were to determine the following: • The speed differential for turning vehicles upstream of and at the left-turn taper. • If speed differential varies based on taper length and posted speed limit. • Whether the deceleration rates described in the 2011 Green Book are representative of current left-turn drivers. The proposed measures of effectiveness (MOEs) for the field study were • Difference in speed of left-turning vehicles between a point upstream of the left-turn lane and the end of the left-turn taper. • Deceleration rates of left-turning vehicles. C H A P T E R 6

78 These variables were used to identify the effects of taper and deceleration length and posted speed limit. Literature AASHTO Policies on Deceleration into a Left-Turn Lane The 2011 Green Book (2) provides desirable values for deceleration lengths in Table 9-22 on page 9-126; that table is reproduced in this document as Table 6-2. A copy of the asso- ciated figure (Figure 9-48 in the 2011 Green Book) is shown in Figure 6-1. These deceleration length values are the estimated distances needed by drivers to maneuver from the through lane into a turn bay and brake to a stop. They are referenced to a 1998 publication (84), but the reference should probably be to the 2003 Access Management Manual (13, page 172), as that document has similar values and comments about deceleration rates for the portion of the maneuver where the driver is clearing the lane and when the driver is decelerating to a stop. The Green Book discussion also states that “at least part of the deceleration distance for an auxiliary lane assumes that an approaching turning vehicle can decelerate comfort- ably up to 10 mph before clearing a through lane” and that “a 10-mph differential is commonly considered acceptable on arterial roadways.” The 2004 Green Book (4, page 714) and the 2001 Green Book (85, page 718) (after corrections from an errata sheet) # 2011 Green Book Statements Potential Additional Research Needs Potential Source for Information 1 The components of a deceleration lane “consist of the perception- reaction distance, the full deceleration length (also called the maneuver distance), and the storage length (called the queue storage length).” The distance traveled during perception- reaction time is illustrated on a related figure, but it is not discussed elsewhere. How should this length be determined? References could be made to one or more previous studies such as (29, 80, 81) to discuss how to calculate PR distance. 2 “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 speed differential between the turning vehicle and following through vehicles is 10 mph when the turning vehicle ‘clears the through traffic lane’. … 5.8 ft/s2 deceleration while moving from the through lane into the turn lane; 6.5 ft/s2 average deceleration after completing lateral shift into the turn lane.” Are the assumptions for speed differential rate valid for the current driving population? Are deceleration rates, including using two different rates to calculate the dimension, valid? Study needed. 3 “The auxiliary lane should be sufficiently long to store the number of vehicles, or queue, likely to accumulate during a critical period. The storage length should be sufficient to avoid turning vehicles stopping in the through lanes waiting for a signal change or for a gap in the opposing traffic flow.” Are storage lengths appropriate? NCHRP 3-91 developed recommended storage lengths for unsignalized intersections based on data collected as part of that project (9, 82) and on equations available in NCHRP Report 457 (22). Information on right-turn lane storage is provided in the TRB Access Management Manual (13) and Stover and Koepke’s Transportation and Land Development (83). 4 “Where turning lanes are designed for two-lane operation, the storage length is reduced to approximately one-half of that needed for single- lane operation. For further information, refer to the HCM.” Is dividing the length needed by 2 for a two- lane turn bay appropriate? What about uneven queuing of vehicles? Data are available from previous studies, including Sando and Moses (74), regarding saturation flow rates for triple left-turn lanes. These studies; however, were not sensitive to factors of interest such as turn lane width, receiving leg width, and downstream friction points. Researching this question was an objective of the Task 4 study on double left-turn lanes. Table 6-1. Green Book material on deceleration lanes where additional research may be needed.

79 had slightly different deceleration lengths; however, the ref- erences were also to the 1998 course notes (84) referenced in the 2011 Green Book. The deceleration lengths are shown in Table 6-3. The 2004 Green Book and the 2001 Green Book also note that “an approaching turning vehicle can decelerate comfortably up to 10 mph in a through lane before entering the auxiliary lane.” The 2001 Green Book was the first policy to provide a value for the amount of speed reduction (10 mph) that can be considered prior to entering the auxiliary lane. The 1990 Green Book (86, page 828) has very different numbers than the more recent versions of the Green Book. Table 6-3 shows the values. The 1990 Green Book states that the total length “required is that needed for a safe and com- fortable stop from the design speed of the highway.” These lengths “exclude the length of taper which should be approxi- mately 8 to 15 ft longitudinally to 1 ft transversely.” The 1990 Green Book also notes that “on many urban facilities it will not be feasible to provide full length for deceleration” and that “in such cases at least a part of the deceleration must be accomplished before entering the auxiliary lane.” The 1990 Green Book does not state what part of the deceleration can be accomplished before entering the lane. The 1984 Green Book (87, page 874) and the 1973 Red Book (88, page 688) base the lengths on average running speed, which is a speed that is less than the design speed of the road. Table 6-3 shows the deceleration lengths for the average run- ning speed. Elsewhere in those documents is information about the assumed relationship between average running speed and design speed, so the assumed design speed for the given aver- age running speed is also shown in Table 6-3. The deceleration lengths include the length of taper. Similar to the 1990 Green Book, these policies note that it will not be feasible to provide the full length for deceleration on many urban facilities, and they do not provide a specific value for the amount of decelera- tion that can occur prior to entering the auxiliary lane. Deceleration Rates Previous sources for deceleration rates are listed below. • As part of the 1990s study on stopping sight distance (SSD) (89), deceleration during braking maneuvers was recorded. The SSD study measured stopping distances for 26 subjects Speed, mph Distance, fta 20 70 30 160 40 275 50 425 60 605 70 820 Notes: 1. The above full deceleration lengths are L2 + L3 in Figure 9-48 (see Figure in this document). 2. Assumes a turning vehicle has “cleared the through lane” when it has moved laterally approximately 9 ft so that a following through vehicle can pass without encroaching upon the adjacent traffic lane. 3. The speed differential between the turning vehicle and following through vehicles is 10 mph when the turning vehicle “clears the through traffic lane.” 4. 5.8 ft/s2 deceleration while moving from the through lane into the turn lane; 6.5 ft/s2 average deceleration after completing lateral shift into the turn lane. a Rounded to 5 ft. Source: A Policy on Geometric Design of Highways and Streets (2011) by the American Association of State Highway and Transportation Officials, Washington, D.C. Used by permission. Table 6-2. Desirable full deceleration lengths, from Table 9-22 of 2011 Green Book (2). Source: A Policy on Geometric Design of Highways and Streets (2011) by the American Association of State Highway and Transportation Officials, Washington, D.C. Used by permission. Figure 6-1. Illustration of deceleration lane, Figure 9-48 of 2011 Green Book (2).

80 using an initial speed of 55 mph. Test conditions included enabled and disabled antilock brakes, wet or dry pavement conditions, and two geometric conditions (tangent section and horizontal curve) for a total of 986 maneuvers. The deceleration rate used in the 2004 Green Book stopping sight distance procedure (11.2 ft/s2) was selected based on the results of the 1990s study. A maximum deceleration rate to an unanticipated object was also identified (24.5 ft/s2) from the SSD study. • Two studies conducted in the 1980s measured dry pavement deceleration characteristics to traffic signal change inter- vals. The study by Chang et al. (90) found mean decelera- tions of 10.5 and 12.5 ft/s2 at the two subject intersections. • Wortman and Matthias (91) found mean deceleration at six study sites of 7.0 to 13.0 ft/s2. The mean value for all observations from the six intersections was 11.6 ft/s2, a result consistent with Chang’s findings. • The 5th edition of the ITE Traffic Engineering Handbook (92) provided a summary of deceleration rates, including deceleration without brakes and representative maximum and comfortable decelerations. The discussion on deceler- ation without brakes references the same 1940 study used to form the basis of the 1965 Blue Book (93) values. The representative maximum braking data provided by ITE in Table 3-12 (92) is from a 1948 paper on skid resistance measurements on Virginia highways. It provided data for speeds under 40 mph for four types of dry surfaces and new and worn tires. The guidance in the ITE Traffic Engineering Handbook for the higher speeds was back-calculated from the 1994 Green Book stopping sight distance value. The comfortable deceleration advice references a Green Book figure similar to the figure in the 1965 Blue Book (which is based on the 1930s data). A single value of 10 ft/s2 was also provided as being “reasonably comfortable for occupants of passenger cars.” • The 6th edition of the ITE Traffic Engineering Handbook (94) provides an update to the guidance on maximum decelera- tion rates, based on more recent research. It states that the maximum deceleration rate is assumed to be 11.2 ft/s2, or 0.35 times the acceleration of gravity. Therefore, tire-friction coefficients are no longer provided for calculation of braking distances as they were in the 5th edition of the Handbook. The 6th edition contains the same guidance on normal decel- eration as the 5th edition, stating that deceleration rates up to 10 ft/s2 are reasonably comfortable for occupants of passen- ger cars and that vehicle clearance intervals at traffic signals are determined based on that rate. A study published in 2007 (95) evaluated the decelera- tion of vehicles between 2.5 and 5.5 sec upstream of sig- nalized intersections at the start of the yellow signal phase, typically considered the “dilemma zone” for drivers. They made video recordings of vehicles approaching signalized intersections and measured several factors for each last-to- go (n=435) and first-to-stop (n=463) vehicle in each lane during each yellow interval, The observed 15th, 50th, and 85th percentile deceleration rates were 7.2, 9.9, and 12.9 ft/s2, respectively. ITE’s recommended comfortable deceleration rate of 10 ft/s2 (92) represented the 52nd percentile for their data, and AASHTO’s deceleration rate of 11.2 ft/s2 for SSD represented the 68th percentile. Gates et al. further analyzed the deceleration rate based on approach speed, using two categories: low (≤40 mph) and high (>40 mph). The cor- relations between deceleration rate and approach speed are 2011 Green Book Deceleration Length Includes Taper Length 2004 Green Book Deceleration Length Includes Taper Length 1990 Green Book Deceleration Lengths Exclude Taper Length 1984 Green Book and 1973 Red Book Deceleration Length Includes Taper Length Speed, mph Decela Distb (ft) Speed (mph) Decel Dist (ft) Speed (mph) Decel Dist (ft) Taper Lengthc (ft) Average Running Speed [Assumed Design Speed] (mph) Decel Dist (ft) 20 70 20 [20] 160 30 160 30 170 30 235 96-180 30 [35] 250 40 275 40 275 40 315 96-180 45 340 40 [45] 370 50 425 50 410 50 435 96-180 55 485 50 [58] 500 60 605 70 820 a Decel = deceleration, Dist = distance. b See Table for notes. c Length of taper should be approximately 8 to 15 ft longitudinally to 1 ft transversely; values shown assume 12-ft lanes. Table 6-3. Comparison of deceleration lengths from different versions of AASHTO policies.

81 presented in Figure 6-2a and distributions for the decelera- tion rates of the two approach speed categories are plot- ted in Figure 6-2b. Figure a shows a strong upward trend between deceleration rate and approach speed. Figure 6-2b shows that drivers approaching at high speeds typically used greater deceleration rates than drivers approaching at low speeds. The 15th, 50th, and 85th percentile deceleration rates for stopping drivers approaching at speeds above 40 mph were 9.2, 10.9, and 13.6 ft/s2, respectively; the rates for stop- ping drivers approaching at speeds less than 40 mph were 6.4, 8.3, and 11.6 ft/s2, respectively. The 10-ft/s2 comfortable deceleration rate recommended by ITE for timing yellow intervals represented the 31st percentile for high speeds and the 74th percentile for low speeds. In other words, they con- cluded that 69% of stopping drivers approaching at speeds greater than 40 mph will use a deceleration rate greater than the recommended design value of 10 ft/s2, compared to only 26% of stopping drivers approaching at speeds less than 40 mph. They suggested that, based on their findings, design deceleration values used for determining the length of the yellow interval should be based on approach speed instead of a single default value. Deceleration Lengths 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. Like the Green Book and many other reference documents, 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. In many guidelines, deceleration length assumes that some 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. A com- mon design guideline is that, in locations 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. NCHRP Project 03-91 (6) identified recommendations for the approximate total lengths needed for a comfort- able deceleration to a stop from the full design speed of the highway. These approximate lengths are shown in Table 6-4 and are based on grades of less than 3%. Two sets of values from NCHRP Project 03-91 are presented in Table 6-4; one set assumes no speed reduction in the main lanes, while the other set accounts for a 10-mph reduction in speed. Researchers acknowledged that, on many urban facilities, it is not practical to provide the full length of deceleration for a left-turn lane, and in other cases, the storage demand con- sumes part or all of the deceleration length. In either scenario, the length available for deceleration is less than what manuals commonly recommend. Shorter left-turn lane lengths may increase the speed differential between turning vehicles and through traffic, as some deceleration may occur before enter- ing the left-turn lane. Because a 10-mph differential is com- monly described as acceptable on arterial roadways, this was included as an alternative in NCHRP Project 03-91, but the Source: Gates, T., D. Noyce, L. Laracuente, and E. Nordheim. Analysis of Driver Behavior in Dilemma Zones at Signalized Intersections. In Transportation Research Record: Journal of the Transportation Research Board, No. 2030, Figure 4, p. 35. Reproduced with permission of the Transportation Research Board. Figure 6-2. Effect of approach speed on deceleration rate (95).

82 authors recommended that the no-speed-reduction lengths given in Table 6-4 should be accepted as a desirable goal and should be provided where practical. Warren et al. (96) conducted a study on the influence of decelerating right-turning drivers in which they developed a series of equations to predict speeds during deceleration based on the distance from the right turn. In that study, they com- pared field data with the Green Book’s suggestion that crashes become much more common when the speeds in the traffic stream differ by more than 10 mph. By identifying the point at which the difference between the through and the right-turning speeds from the study exceeded 10 mph, they could infer the distance in advance of a driveway at which the full length of the auxiliary right-turn lane should be developed. The taper for entering the deceleration lane would be upstream of this point. Using speeds from their predictive equations, they calcu- lated the distance in advance of the driveway edge at which the 10-mph difference (based on average speeds of both through and turning vehicles) occurred was 108 ft for the group of study sites with average speeds near 38 mph, 128 ft for sites with speeds near 42 mph, and 151 ft for sites with speeds near 46 mph. Researchers’ review of the data showed that typi- cally the standard deviations of turning vehicle speeds did not exceed 4 mph. They then concluded that instead of designing for average speeds, it would be better to design for a 10-mph difference between the average speed of through vehicles and the average minus 4 mph of the turning vehicles. Those result- ing lengths were 213 ft for 38 mph, 232 ft for 42 mph, and 237 ft for 46 mph. The researchers noted that these lengths were only for the deceleration lane and did not provide any storage space. Study Matrix Table 6-5 shows the initial site selection matrix based on the key site selection variables. The 2011 Green Book specifies two common tapers: a ratio of 8:1 (L:T) for design speeds up to 30 mph and 15:1 for design speeds 50 mph and greater. Although these are defined as straight-line tapers, the lengths (96 ft and 180 ft, respectively, for a 12-ft lane) can be used to compare with other taper designs. Indeed, state design manuals describe a wide variety of designs and lengths, both longer and shorter than the Green Book lengths, many based on speed and others based on turning volumes. For site selection in this study, the equivalent lon- gitudinal length of the taper was measured, and that length was classified based on the values in Table 6-5. Using this matrix, the ideal outcome would be that the research team would collect two sites for each combination of variables, a total of 16 sites (4 speed limit ranges × 2 taper length categories × 2 sites), as shown in Table 6-5. The research team set a goal of having a minimum of 12 study sites, using Table 6-5 as a guide. Table 6-4. Deceleration lengths for left-turn lanes (6). 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 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 6-5. Potential site selection matrix for deceleration study. Posted Speed Limit on Major Road (mph) Taper Length Threshold (ft) for a 12-ft Lane Number of Sites Below Threshold At or Above Threshold 30–35 96 2 sites 2 sites 40–45 96 2 sites 2 sites 50–55 180 2 sites 2 sites 60–65 180 2 sites 2 sites

83 In addition to the main study factors (i.e., posted speed limit and taper length), the field study also controlled other impor- tant variables that might affect the operation such as those in Table 6-6 so as to limit their effects on the MOEs and to allow the effects of posted speed limit and taper length to be more easily identified. The variables expected to affect the operation, but not significantly change the effectiveness of deceleration lanes, do not necessarily need to be included in the analysis as study factors, but it is still useful to control them by holding them at a constant level. Table 6-6 lists the variables controlled during site selection. Additional geometric and traffic control device variables were collected in the field; researchers did not anticipate including all of these variables in the models for analysis, but they were available if additional investigation was needed to help explain the results for a specific site. Table 6-7 lists additional variables collected at each study site. Study Sites The research team used the characteristics in Table 6-5 through 6-7 to select study sites. By using these character- istics to guide the site selection process, a more efficient and economical data collection plan was developed while ensur- ing that important effects of key characteristics could be estimated. A tradeoff to using this approach was that con- siderably more resources were required to identify potential study sites, as opposed to choosing study sites arbitrarily. Some potential study sites were suggested by panel mem- bers or through the responses to the state-of-the-practice questionnaire. The research team considered Texas, Ohio, Illinois, California, North Carolina, Florida, Tennessee, Louisiana, Mississippi, and Alabama as possible focus states for study sites. The team contacted colleagues and practitio- ners in the selected states to obtain further information and identify additional potential deceleration sites. Reviewing these sites in aerial photographs available online, research- ers gathered preliminary information on site characteristics to select the sites for the study, taking into consideration the states that would enable the most efficient means of complet- ing data collection efforts. The goal was to collect data in a minimum of three states. Based on the information available, researchers decided to focus their efforts on Texas and a selection of states in the south eastern part of the country. A more detailed review of Variable Characteristics Intersection type 4 legs Area type Urban or suburban Surface condition Fair to good Conditions during data collection Collected only during daytime, daylight conditions, and not raining Intersection sight distance Adequate Traffic control Signalized; preferred protected-only operations, but sites with permitted turning were allowed Grade Preferred between −2 and +2%, but accepted between −4 and +4% Skew Approximately 90° in skew angle Major road number of lanes 4 lanes Minor road number of lanes Preferred 4 lanes, but accepted 2 lanes Left-turn lane assignment Exclusive left-turn lanes (i.e., no shared lanes) Approach horizontal alignment Nominally straight Departure horizontal alignment Nominally straight Adjacent parking lane None Nearest upstream friction point Outside area of influence for left-turn vehicles Shadowing of left-turn lane Full Median type Raised Posted speed limit on major road See Table Table 6-6. Controlled site selection characteristics for deceleration study. Variable Comments State A minimum of three states represented Left-turn lane width Preferred 12 ft, but accepted as low as 9 ft for low-speed sites Through lane width Lane width measured and noted Deceleration length Desired an equal number of sites with lengths greater than and less than those recommended by the Green Book, but most sites had shorter lengths Pedestrian activity Level of activity observed and noted Bicycle activity Level of activity observed and noted Heavy vehicles Only considered passenger cars Table 6-7. Additional variables in deceleration study.

84 potential sites in Alabama, Florida, Mississippi, North Caro- lina, Tennessee, and Texas was conducted to produce a list of viable candidate sites. More than 70 sites in these six states were considered during site selection, and a list of candidate sites was developed for researchers to visit in person and col- lect data at the most promising sites. Researchers ultimately visited 42 of these sites in Texas, Mississippi, Alabama, and Florida, choosing 15 of them for data collection. Table 6-8 shows the final distribution of those 15 sites. Two of the sites (one 50 mph, below threshold; one 65 mph, below threshold) had to be eliminated after site selection and initial data collection because the video files recorded at those sites did not produce images of the clarity needed to accu- rately obtain the necessary data. Another site (55 mph, above threshold) was eliminated because no counter data were recorded during data collection. Ultimately, data reduction and analysis were completed for 12 sites. Table 6-9 lists the key study site characteristics for the deceleration lane study sites. Key length dimensions in Table 6-9 and key locations in the data collection protocol are shown in Figure 6-3. Those key locations were along the length of the left-turn lane to establish fixed points in the site characteristics documenta- tion. The key locations used in post-processing and reducing the data from the video included A. Upstream of the turning lane (location of the automated traffic counter to collect free-flow or pre-deceleration, speeds). B. Beginning of the taper. C. End of the taper. D. End of the deceleration length, as recommended by the Green Book. E. End of the turning lane at the stop line. F. Other locations as needed. Data Collection The primary data collection method used was video recording, supplemented by GPS receivers, lidar emitters (i.e., laser speed guns), and automated speed counters. Video pro- Si te # C ity PS L Le n_ Ta pe r Ta pe r_ Th re sh Le n_ D ec el Le n_ LT L LW _T hr u LW _L T M ed _W id Si g_ O p AL-03 Mobile 50 186 Above 180 180 11.0 11.0 18.5 Pro/Per AL-08 Mobile 35 115 Above 230 245 12.0 11.3 18.0 Pro/Per AL-09 Mobile 35 61 Below 94 94 8.5 9.0 14.0 Per Only FL-03 Tallahassee 45 98 Above 365 380 11.5 10.5 4.0 Pro/Per FL-09 Tallahassee 35 80 Below 173 173 10.5 10.0 5.0 Pro/Per FL-10 Tallahassee 45 71 Below 216 216 11.5 9.0 5.0 Pro Only MS-03 Biloxi 45 93 Below 186 186 11.5 10.5 11.3 Pro/Per MS-05 Biloxi 35 125 Above 80 80 12.0 11.0 13.0 Per Only MS-08 Gulfport 45 119 Above 255 255 11.0 10.0 5.5 Pro/Per TX-21 Pflugerville 50 204 Above 115 115 12.5 11.5 29.0 Pro Only TX-28 Austin 65 191 Above 283 283 13.5 11.5 56.0 Pro Only TX-33 Austin 65 152 Below 312 312 13.0 11.5 56.0 Pro Only Note: PSL = posted speed limit (mph); Len_Taper = length of taper (ft); Taper_Thresh = whether Len_Taper is above or below the taper length threshold in Table ; Len_Decel = length of deceleration lane (ft); Len_LTL = length of left-turn lane from end of taper to stop bar (ft), includes deceleration length + storage length; LW_Thru = width of through lane (ft); LW_LT = width of left-turn lane (ft); Med_Wid = width of median (ft); Sig_Op = left-turn signal operation mode: (Pro)tected, (Per)mitted, or both (Pro/Per). Table 6-9. Site characteristics at study sites. Posted Speed Limit on Major Road (mph) Taper Length Threshold (ft) for a 12-ft Lane Number of Sites Below Threshold At or Above Threshold 30–35 96 2 sites 2 sites 40–45 96 2 sites 2 sites 50–55 180 1 site 3 sites 60–65 180 2 sites 1 site Table 6-8. Final site selection matrix for deceleration study.

85 vided a permanent recording of the conditions and events at the site, speed counters provided the vehicle speeds upstream of the site, and the GPS and lidar provided a way to obtain comparative speed data to use for quality control. Tradi- tionally, video is not as commonly used to collect this type of speed data as the other methods, but the research team conducted preliminary field tests and compared the relative benefits of each method in selecting the video method for this study. Table 6-10 provides a summary comparison of the data collection methods considered—video had several key advantages: it could provide a permanent visual record of events during data collection, it produced both speed and volume measurements, and it did not require researchers or equipment to be within the roadway. These advantages out- weighed the limitations of video. Ultimately, a combination of the four methods was used to collect the data, but video was the primary data source. Prior to beginning data collection, data collectors visited the site to determine the best locations to set up equipment, and they coordinated as needed with local authorities and Figure 6-3. Key dimensions of left-turn lanes at study sites. Table 6-10. Comparison of data collection methods in deceleration lane field study. Data Collection Method Advantages Limitations Video Provides a permanent record of conditions at the site Provides a method of obtaining accurate volumes Permits observation of all vehicles at the site, and a larger number of vehicles in general Can be inconspicuous and not affect traffic Requires a vantage point with sufficient field of view to capture the entire turning lane Precision is related to the frame rate, resolution, and perspective of the camera and recorder Trailer must be towed from site to site Automated traffic counter (piezoelectric or tube sensor) Records speeds and classifications of all vehicles Relatively easy to transport and install on collector street (but perhaps not high- speed arterials) Records speeds at only one point Installed in the roadway; drivers must travel over the sensors Lidar Records a speed-distance profile Can be used for all vehicles Easy to transport Proper positioning may be difficult; line of sight can be obstructed by other traffic, which is common when studying left-turn lanes Operator needs to be inconspicuous to avoid influencing driver behavior Instrumented vehicle Records a detailed speed-location profile using GPS Can obtain data on drivers at multiple locations Only a limited number of subject drivers can be accommodated within the time and budget available; related issues include: o Range of driver characteristics (age, experience, etc.) o Drivers must be scheduled, as well as compensated for their time

86 adjacent landowners. The length of the study period at each site was generally between 3 and 4 hr. Figure 6-4 illustrates the position of the video trailer and other equipment at an example study site. Video Trailer Researchers used a trailer with a telescoping mast that raised two cameras up to 30 ft above ground level, positioned with a field of view perpendicular to the left-turn lane. These cameras were connected to a digital video recorder, which created a time-stamped video recording of traffic conditions at each intersection during the study period. During the setup process at each study site, researchers identified the key locations shown in Figure 6-3. During the setup process, researchers temporarily placed traffic cones or other markers at key locations (B, C, D, E, and F in Figure 6-3) to provide visual cues on the video recording. At most study sites, the left-turn lane did not provide the full deceleration length as recommended by the Green Book, so point D was not available. As a result, a point F was created, typically 50 or 100 ft upstream of the stop line, to provide an intermediate point between points C and E. The video trailer was typically positioned on the road- side near the end of the taper, which was often the center of the field of view needed to capture the entire length of the left-turn lane and some additional distance upstream and downstream. The two cameras were aimed to capture side- by-side points of view, in effect creating a “landscape” image of the left-turn lane, as shown in Figure 6-5. In some cases, Figure 6-4. Schematic drawing of equipment setup at study site. Figure 6-5. Screenshot of left-turn lane video.

87 the length of the left-turn lane or the available roadside area to position the trailer mandated that the cameras be aimed to exclude a middle portion of the left-turn lane, typically between points C and F. Using this method, researchers were able to maintain sufficient visibility on the extents of the left- turn lane and key points while omitting a short length from the field of view that was not necessary for data reduction. Figure 6-4 shows the trailer positioned on the same side of the road as the left-turn lane, though site conditions often made it more suitable to position the trailer on the opposite side of the road to obtain the best vantage point. Automated Traffic Counter Researchers also installed an automated traffic counter upstream of the left-turn lane, to collect free-flow speed data on vehicles approaching the intersection. The counter system used pneumatic tubes on the road surface to make contact with the tires of approaching vehicles; the tubes were con- nected to the counter itself, which was placed on the roadside or in the median out of approaching drivers’ field of view. The location of the counter was chosen so that vehicles would be far enough upstream of the intersection to not be affected by actual or anticipated deceleration to enter the left-turn lane. Thus, the speed data collected by the counter would closely approximate not only the free-flow speed of vehicles on the major road, but also the expected speed of through vehicles as they approached the beginning of the left-turn taper. Quality Control Speed Data After the video recording equipment was set up and acti- vated, researchers collected additional speed data for use as quality control comparisons for the speed data that would be obtained from the video. First, the researchers used a GPS receiver connected to a laptop computer to record speed pro- files in the research vehicle, a white four-door Ford F-150 with the TTI logo on both front doors. The researchers com- pleted 20 trips through the study site in the research vehicle, 10 trips at the posted speed limit and 10 more trips at 10 mph below the posted speed limit. Half of the trips were termi- nated with a left turn, and half were conducted by continuing through the intersection in the left-hand through lane. This distribution provided various conditions under which speed data recorded with the GPS receiver could be compared to the speed data reduced from the video. Researchers established and maintained speed by using the vehicle’s cruise control on the approach to the intersection. After the GPS data were collected, researchers collected speed data using a lidar gun connected to a laptop computer. The lidar gun enabled collection of speed-distance profiles on vehicles other than the research vehicle. Researchers identified a location away from the left-turn lane to set up the computer and lidar gun to record a sample (typically between 10 and 20) of left-turning and through vehicles. This location was chosen to minimize any effects that the researcher’s presence might have on approaching drivers. As shown in Figure 6-4, the lidar position could be downstream of the intersection if sufficient lines of sight and setup space were available. The lidar position could also be established upstream of the left- turn lane if site characteristics were more conducive there. The lidar gun collected speed and distance data on vehicles approaching the intersection at a frequency of approximately 0.3 sec; the data were stored on the laptop computer in a text- based log file, which was later converted to a spreadsheet for use in data reduction. Site Characteristics Finally, the researchers documented conditions at each study site by completing a site characteristics worksheet, recording key measurements such as lane and median widths and distances from the stop line to points A through F, and taking photographs of key locations and features along the approach to and departure from the left-turn lane. Data Reduction Researchers reduced the data from the video, lidar, GPS receiver, and traffic counter to determine speeds of turning vehicles at the key locations at each of the study sites. Vehicle speed within the left-turn lane was determined by calcu- lating the time it takes for a vehicle to travel a known dis- tance. Key locations were identified within the turning lane as described previously in the Data Collection section, the distances between those points were measured and recorded, and a reviewer watching the video noted the times when sub- ject vehicles reached each of those locations, to calculate the elapsed time to travel those known distances. The critical characteristic of this study is that the known distances are very short, so researchers must be able to deter- mine the elapsed time with great precision; otherwise, calcu- lated speeds may be highly inaccurate. To use video effectively on this project, researchers recorded images of the entire left- turn lane within a single field of view, and a video record- ing system was used with sufficient resolution for the video reviewer to determine when a specific point on a subject vehicle (e.g., the front tire or the front bumper) reached a key location within the lane, based on the timestamp recorded with the video image. The recording system also could play back frame-by-frame to pinpoint when vehicles arrived at key locations. The research team developed a spreadsheet in which to document the details of left-turn deceleration events. The

88 reviewer then opened the video file and, using field notes recorded at the study site, identified the location of each key point on the video screen. Field technicians at each site walked across the roadway at each key point, which was recorded in the video. Technicians then noted when they walked across the roadway. Based on that information, the video reviewer watched the video at that time, and, using a transparency film overlaid on the video screen, drew lines corresponding to each key point, similar to the yellow lines superimposed on the image in Figure 6-6. The reviewer then watched the video to identify appropri- ate target vehicles. Target vehicles were passenger vehicles that were the first in queue and came to a complete stop before making the left turn. Vehicles that did not come to a complete stop before turning were not accepted as target vehicles. After a suitable left-turning vehicle was identified in the video, the video was reversed until the target vehicle was upstream of the first line at point B. Using the timestamp on the screen and the frame-advance feature in the playback software, the reviewer then identified the hour, minute, second, and frame in which the target vehicle reached point B. That information for the first target vehicle at a given site was recorded in the spreadsheet. That process was repeated for points C, F, and E. The equations in the spreadsheet calculated the average speed of the vehicle between successive points, based on the elapsed time to travel those distances. The video reviewer continued to identify and document the information on left-turning vehicles until 100 left-turning events were docu- mented or until 4 hr of video were reviewed. To add another level of quality control, the video files were reviewed a second time, by a second reviewer, using a second playback program that provided frame count instead of a clock counter. This provided an additional confirmation that the results obtained in the video review were accurate. The speed data from the second review was added to the spread- sheet, and equations in the spreadsheet calculated the differ- ence in speed values with the first video review. Researchers then reduced the files from the GPS receiver to include in the spreadsheet. While completing the GPS trips at each site, researchers added another trip to stop and mark the location of each key point in the GPS speed-distance profile. While reducing the GPS data, researchers used the latitude/longitude coordinates of the four key points in the speed-distance profile to identify those points in the other 20 trips. The cumulative distance for the GPS reading closest to each key point was identified, and the elapsed time and aver- age speed over those distances were calculated. The average speed values were entered into the spreadsheet, and differ- ences between GPS speeds and video speeds were calculated. Finally, researchers reduced the speed-distance data from the lidar gun for inclusion in the spreadsheet. The data from the lidar gun was recorded in a format similar to that of the GPS data, in that it contained speed and location of each read- ing. However, the lidar gun records at a frequency of approxi- mately 0.3 sec, so there was typically not a speed reading at the exact location of all four key points for a given target vehicle. Researchers interpolated the elapsed time between the key points using speed-distance readings from adjacent points. That interpolated elapsed time was used to calculate the aver- age speed over the known distance between key points. Those speed values were entered into the spreadsheet, and differ- ences between GPS speeds and video speeds were calculated. Researchers then processed the data from the traffic coun- ter files to add to the database. Each vehicle recorded by the traffic counters was recorded with a timestamp in addition to its speed and wheelbase characteristics. The timestamp and wheelbase information was compared to the video to iden- tify and match vehicles in the counter and video data files. Using the time and speed at which a particular type of vehi- cle crossed the counter, researchers could estimate when that B C D/F E Figure 6-6. Example of video showing locations of key points.

89 vehicle would be visible in the video. To identify a particu- lar left-turning vehicle documented in the video, research- ers established the presence of one or more nearby vehicles; focusing on identifying non-passenger vehicles (e.g., buses and semi-trailers), researchers could confidently establish the elapsed time between the counter reading and the appearance in the video. Researchers could then determine the number of vehicles in the video between that non-passenger vehicle and the left-turning vehicle previously documented. Knowing the likelihood that the same number of vehicles traveled through the counter in the same order upstream of the left-turn lane, researchers compared timestamps in the counter file for vehi- cles near the non-passenger vehicle and determined which counter vehicle corresponded to the left-turning vehicle in the video. The speed associated with that vehicle was then entered into the database as the upstream free-flow speed for the left-turning vehicle. For each vehicle, researchers indicated their confidence level that they had identified the same vehicle in both the counter and video. For about one-fifth of the vehicles in the video (200 of 684), researchers were not able to identify the corresponding vehicle in the counter file with high confidence, and upstream free-flow speeds for those vehicles were not included in the database. In 195 of those 200 cases, there was no counter record of the vehicle because the counter was inactive when the vehicle traveled through the left-turn lane. In the other five cases, no appropriately sized vehicles traveled through the counter within the probable timeframe that would cor- respond to the left-turn vehicle that appeared in the video; those vehicles likely entered the roadway from a driveway or cross street between the counter and the left-turn lane, so there was no counter record of those vehicles. Upstream speeds for the other 484 vehicles were identified and included in the spreadsheet for analysis. Quality Control That each of the speed data sources (i.e., two video reviews, GPS, and lidar) calculated speeds in a different way enabled thorough quality control checking, but it increased the likeli- hood that speeds from different sources would not be equal, so in the comparison of data, researchers set a threshold of 3.0 mph as the target for maximum difference in speeds. Given that each method of collecting data has its own margin of error, which could be cumulative when comparing data from multiple sources, a value of 3.0 mph seemed a reasonable goal. No left-turn event had data from every source, but each qual- ity control (QC) comparison had at least two sources of data. When each site’s data was compared side by side in a spread- sheet, researchers reviewed the data looking for differences larger than 3.0 mph. When they were identified, researchers reviewed the video data and the second (or third) source to identify any adjustments needed. In many cases, the difference in speed value was due to data entry (e.g., typing in the wrong second or frame for video data) or spreadsheet miscalcula- tion (e.g., an equation in Excel referred to the wrong value). In some cases, the output from the GPS or lidar produced an average speed higher or lower than all of the individual speed readings collected over that distance, indicating a calculation adjustment was needed. If the source of the discrepancy could be identified, it was corrected, but in some cases, the source was unknown. In those instances, researchers either used the average of the individual speed readings over that distance or they removed that data point from the comparison. Researchers also checked the speeds from the upstream counter in relation to the speeds in the taper and found that 74 of the 484 vehicles had a negative speed difference; that is, the vehicles increased speed between the counter and the taper. In reviewing the characteristics of those vehicles, researchers determined that those vehicles had likely entered the study site from a driveway or side street near the counter and had not reached free-flow operating speed when passing through the counter. Those 74 vehicles were removed from the analysis, leaving 410 vehicles at 12 sites to be considered. After all of the quality control process was completed, the speed data for all of the target left-turning vehicles obtained from the video were compiled into a single database for use in analysis. Analysis The 2011 Green Book provides desirable values for deceler- ation lengths in Table 9-22 (page 9-126); that table is adapted for this document as Table 6-2. The research team focused its analysis efforts on three key guidelines that the Green Book provides as notes in conjunction with Table 9-22: 1. The speed differential between turning vehicles and follow- ing through vehicles is 10 mph when the turning vehicle clears the through-traffic lane (see Note 3 in Table 6-10). 2. The values for deceleration length are based on a 5.8 ft/s2 average deceleration while moving from the through lane into the left-turn lane (see Note 4 in Table 6-10). 3. The values for deceleration length are based on a 6.5 ft/s2 average deceleration after completing the lateral shift into the left-turn lane (see Note 4 in Table 6-10). The purpose of the analysis was to determine whether those three guidelines represented the actual deceleration of left-turning vehicles observed at the study sites. Researchers used multiple approaches to analyze the deceleration data, performing basic analytical reviews on the data in a spread- sheet and in graphs to look for possible patterns and trends and then conducting a more extensive statistical analysis using the ANACOVA mixed model. The variables reviewed in these analyses are listed in Table 6-11.

90 Speed Differential of Left-Turning Vehicles The first research question in this study focused on left- turning vehicles’ difference in speed between the upstream traffic counter and the end of the taper. To determine whether the differential from Note 3 in Table 6-10 is rep- resentative of current drivers, researchers analyzed the dif- ferences in speed between the upstream traffic counter and the left-turn taper for the 410 left-turning vehicles observed in the video that had corresponding speed readings from the traffic counter. The speed at the upstream traffic counter provides an indication of the speeds of through vehicles in addition to enabling same-vehicle comparisons of speed and deceleration for the observed left-turning vehicles. Figure 6-7 shows the cumulative distribution of the speed differential data (Spd_DiffAC). About 46% of all observed vehicles had a speed differential less than the Green Book sug- gested value of 10 mph, but the distributions varied greatly by posted speed limit. Approximately two-thirds of the vehicles observed at posted speed limits of 35 and 45 mph slowed by 10 mph or less, compared to just 7% of vehicles on 50-mph roadways and 34% of those on 65-mph roadways. About 30% Variable Variable Type Description Spd_DiffAC Response A observed left-turning vehicle’s difference in speed (mph) between the upstream counter at point A and the end of the taper at point C in Figure 6-3. Decel_AC Response The deceleration rate of an observed left-turning vehicle (ft/s2) between the upstream counter at point A and the end of the taper at point C. Decel_CE Response The deceleration rate of an observed left-turning vehicle (ft/s2) between the end of the taper at point C and the stop line at point E. PSL_Major Site (Key) The posted speed limit (mph) on the major road at the approach to the left-turn lane. Len_Taper Site (Key) The length of the left-turn taper (ft); the distance between points B and C in Figure 6-3. Len_Decel Site (Key) The length of the deceleration lane (ft); the distance between points C and E in Figure 6-3. Len_T+D Site (Key) The sum of the taper and deceleration lengths (ft). LW_Thru Site (Secondary) The width of the through lane adjacent to the left-turn lane (ft). LW_LT Site (Secondary) The width of the left-turn lane (ft). Site Random The study site at which the data were collected. Spd_BC Individual Vehicle The average speed of the left-turning vehicle in the left-turn taper (mph). Spd_A Individual Vehicle The speed of the left-turning vehicle at the counter upstream of the left-turn lane (mph). Table 6-11. Variables included in deceleration data analysis. 0% 20% 40% 60% 80% 100% 0 5 10 15 20 25 30 35 Cu m ul ati ve D is tr ib uti on SpdDiff_AC (mph) 35 mph 45 mph 50 mph 65 mph Figure 6-7. Cumulative distribution of speed difference between upstream and taper.

91 Table 6-12. Speed differential by site. Site Taper Length (ft) Decel Length (ft) Posted Speed Limit (mph) Speed Differential (mph) Number of Vehicles Average Std Dev AL-08 115 230 35 7.4 3.4 17 AL-09 61 94 35 12.9 5.1 17 FL-09 80 173 35 4.7 2.7 33 MS-05 125 80 35 12.9 5.2 20 FL-03 98 365 45 10.2 5.4 38 FL-10 71 216 45 4.8 3.8 28 MS-03 93 186 45 9.4 4.5 24 MS-08 119 255 45 7.9 4.7 59 AL-03 186 180 50 15.4 4.4 44 TX-21 204 115 50 20.3 5.5 32 TX-28 191 283 65 18.1 6.8 55 TX-33 152 312 65 8.9 5.9 43 All Sites 11.4 7.0 410 Table 6-13. Evaluation of difference between upstream speed and speed at the end of the taper. Response Spd_DiffAC Summary of Fit RSquare 0.663367 RSquare Adj 0.658355 Root Mean Square Error 4.137866 Mean of Response 11.41751 Observations (or Sum Wgts) 410 Parameter Estimates Term Estimate Std Error DFDen t Ratio Prob>|t| Intercept -7.688556 23.07555 6.06 -0.33 0.7502 PSL_Major -0.096517 0.233616 6.129 -0.41 0.6936 Len_Taper 0.017966 0.061903 6.071 0.29 0.7813 Len_Decel -0.032595 0.020219 6.097 -1.61 0.1573 LW_Thru -2.167737 1.849077 6.159 -1.17 0.2844 LW_LT 2.4297784 3.197241 6.078 0.76 0.4757 Spd_A 0.5955358 0.041732 398.4 14.27 <.0001* REML Variance Component Estimates Random Effect Var Ratio Var Component Std Error 95% Lower 95% Upper Pct of Total Site 0.8811868 15.087623 9.0000753 -2.552525 32.727771 46.842 Residual 17.121935 1.2151922 14.969187 19.777535 53.158 Total 32.209558 100.000 -2 LogLikelihood = 2370.7763692 of the left-turning vehicles at the 50 mph speed limit and 21% at the 65 mph speed limit slowed by more than 20 mph. Table 6-12 lists the speed differentials by study site, ordered by speed limit. While posted speed limit is often used in comparisons to other variables, the difference in speed between the upstream counter and the taper could be influenced by some factors. To properly account for the combinations of these variables, researchers used the ANACOVA mixed model to analyze the data. The model that produced the most informative results is summarized in Table 6-13. Table 6-13 reveals that the speed differential was signifi- cantly and positively related to the upstream speed (Spd_A). This is intuitive, because one can expect the speed differen- tial to rise as the upstream speed rises, and a vehicle’s speed differential cannot be greater than its initial speed. A plot of the two variables is shown in Figure 6-8, and a summary of the data in 10-mph increments is shown in Table 6-14. All of the speed differentials greater than 30 mph and 45 of the 50 differentials greater than 20 mph occurred for vehicles with upstream speeds of at least 50 mph. Speed reductions between 10 and 20 mph occurred commonly throughout all initial speed ranges except those less than 30 mph or greater than 69 mph. Results Compared to Green Book Guidance Comparing the results in Figure 6-8 and Table 6-14 to the Green Book guidelines provides opportunity for some inter- esting insights. Supporting the desirable deceleration lengths

92 and notes in Table 9-22, the Green Book (2) adds the follow- ing text on page 9-127: Inclusion of the taper length as part of the deceleration dis- tance for an auxiliary lane assumes that an approaching turning vehicle can decelerate comfortably up to 10 mph before clearing a through lane. Shorter auxiliary lane lengths will increase the speed differential between turning vehicles and through traffic. A 10-mph differential is commonly considered acceptable on arterial roadways. Higher speed differentials may be acceptable on collector highways and streets due to higher levels of driver tolerance for vehicles leaving or entering the roadway due to slow speeds or high volumes. Two key items are discussed in this selection of text: the roadway classification and the length of the auxiliary lane. All of the sites in this study were arterials, so the speed dif- ferential on collectors cannot be tested in this study; however, the Green Book’s observation that higher speed differentials on collectors might be acceptable due to slow speeds does not appear to be consistent with what was observed on the lower- speed arterials in this study. In fact, the higher speed differ- entials in this study occurred for vehicles traveling at higher speeds. A key descriptor in the Green Book text is the word “acceptable,” which is not the same as “expected.” Based on the data from this study, one would not expect higher speed differentials on streets with lower speeds, even if it might be considered acceptable for purposes of design, operations, or safety. The first edition of the Access Management Manual (13) and the draft second edition (97) both cite research by Soloman (98) stating that crash potential increases as the difference in speed increases between vehicles in a traffic stream. In particular, the research concludes that a vehicle traveling 10 mph slower than other traffic (i.e., a vehicle with a 10-mph speed differential) is twice as likely to be involved in, or cause, a crash as when all vehicles are traveling at the same speed, and the likelihood of a crash increases exponen- tially with greater speed differentials. This research is used in the Access Management Manual as the basis to recommend deceleration lengths that include a 10-mph speed differential. Similarly, Soloman’s research and/or the Access Management Manual can also be referenced in the Green Book to support the 10-mph differential as “acceptable”; the existing text that describes the assumption of comfortable deceleration for the driver could be replaced with a description of the benefits of reducing the likelihood of a crash. 0 5 10 15 20 25 30 35 40 0 10 20 30 40 50 60 70 80 Sp d_ D iff AC (m ph ) Spd_A (mph) Figure 6-8. Plot of speed differential versus upstream speed. Upstream Speed (mph) Number of Vehicles with a Speed Differential of… 0–10 mph 10–20 mph 20–30 mph > 30 mph Total 20–29 7 2 0 0 9 30–39 47 21 1 0 69 40–49 93 54 4 0 151 50–59 38 72 26 1 137 60–69 4 22 13 3 42 > 70 0 0 0 2 2 Total 189 171 44 6 410 Table 6-14. Speed differential by upstream speed.

93 Referring to the length of the auxiliary lane, a relation- ship between shorter lengths and higher speed differentials, as stated in the previous Green Book quotation, is intuitive. Indeed, the two largest speed differentials are associated with sites that have the greatest differences in length compared to Green Book recommendations; however, a strong statisti- cal relationship did not appear in this dataset, as shown in Table 6-15 and Figure 6-9. There were various speed differ- entials for each deceleration lane in the study. Neither taper length nor deceleration length was significant at a = 0.05 in any of the statistical models explored, and the two length variables had coefficients with small values and opposite signs. Other models looked at taper and deceleration length together, and the result was that an increase of 100 ft in com- bined length would decrease the speed differential between 3.0 and 4.0 mph, though that relationship was not statistically Table 6-15. Evaluation of speed differential with combined taper and deceleration length. Response Spd_DiffAC Summary of Fit RSquare 0.663342 RSquare Adj 0.659175 Root Mean Square Error 4.137816 Mean of Response 11.41751 Observations (or Sum Wgts) 410 Parameter Estimates Term Estimate Std Error DFDen t Ratio Prob>|t| Intercept -24.42452 14.8144 7.15 -1.65 0.1423 PSL_Major 0.0313606 0.189744 7.391 0.17 0.8732 Len_T+D -0.037865 0.01932 7.111 -1.96 0.0902 LW_Thru -2.601044 1.780614 7.172 -1.46 0.1865 LW_LT 4.6833316 2.130006 7.079 2.20 0.0634 Spd_A 0.5981215 0.041689 401.1 14.35 <.0001* REML Variance Component Estimates Random Effect Var Ratio Var Component Std Error 95% Lower 95% Upper Pct of Total Site 0.8697232 14.890988 8.2216983 -1.223541 31.005517 46.516 Residual 17.121525 1.215134 14.968874 19.776991 53.484 Total 32.012513 100.000 -2 LogLikelihood = 2367.6496048 Figure 6-9. Plot of deceleration length versus speed differential. 0 50 100 150 200 250 300 350 400 0 5 10 15 20 25 30 35 40 Le n_ De ce l ( ft ) Spd_DiffAC (mph)

94 significant at a = 0.05 (see Table 6-15). Note that although Len_Decel was not significant at a = 0.05, the sign of its coef- ficient makes sense and the effect may become significant as the number of sites increases, so the general effect of decel- eration length appears to be consistent with the Green Book text, but its practical effect is not as apparent. Deceleration of Left-Turning Vehicles The other key research question in this study focused on the deceleration rate of left-turning vehicles. In the notes asso- ciated with desirable deceleration lengths, the Green Book describes deceleration rates of 5.8 ft/s2 while moving from the through lane into the turn lane and 6.5 ft/s2 after complet- ing the lateral shift into the turn lane. Researchers compared the upstream speeds and end-of-taper speeds of the observed left-turning vehicles and, using the time it took them to travel to the end of the taper, calculated deceleration rates between points A and C (see Figure 6-4). Similarly, speeds at the end of the taper were compared with the time spent decelerating to a stop at the stop line to calculate deceleration rates within the deceleration lane. The analysis of those deceleration rates is described in this section. Deceleration Upstream of the End of the Taper The first deceleration to be considered is that which occurred as vehicles prepared to enter the left-turn lane; this deceleration took place upstream of and within the taper. Table 6-16 lists the deceleration rates prior to the end of the taper at each study site, with the sites ordered by taper length. The Green Book deceleration rate of 5.8 ft/s2 is within the range of average rates at the study sites, but 8 of the 12 sites had an average rate higher than 5.8 ft/s2. In addition, the 15th percentile deceleration rate for all sites was 2.8 ft/s2, and two sites’ 15th percentile rate was equal to or greater than 5.8 ft/s2. For reference, each site’s taper rates are also shown in Table 6-16; the distribution of taper rates is similar to the Green Book’s suggested range of rates (i.e., 8:1 at lower speeds and 15:1 at higher speeds). Five of the eight sites below 50 mph had taper rates less than 10:1, and all four high-speed sites had rates greater than 13:1. Figure 6-10 shows the cumulative distribution of the decel- eration rates in and upstream of the taper (Decel_AC). Overall, about 40.5% of the observed left-turning vehicles had a decel- eration rate lower than the 5.8 ft/s2 described in the Green Book. The 5.8 ft/s2 rate was approximately the 46th percentile value for vehicles at sites with lower speeds and taper rates, and it was the 33rd percentile value for vehicles at high-speed sites. The 50th percentile rate was approximately 6.1 ft/s2 for low- speed sites and 6.7 ft/s2 for high-speed sites. The 15th percentile deceleration rates at the high-speed and low-speed sites were approximately 4.2 ft/s2 and 2.2 ft/s2, respectively. To better understand the relative effects of the site and vehicle variables on deceleration rate, researchers developed ANACOVA models that included many of the variables listed in Table 6-11. In reviewing the results of the various mod- els, researchers observed that the statistical significance of the deceleration length changed based on whether all speed limits were considered together or whether they were divided into lower speeds (35-45 mph) and higher speeds (50 mph or higher). Further investigation revealed that taper length was confounded with posted speed limit; this is logical, con- sidering that the design guidance recommends longer tapers for higher speeds. As a result, researchers focused their atten- Table 6-16. Deceleration rates of left-turning vehicles prior to the end of the taper. Site Taper Ratea Taper Length (ft) Decel Length (ft) Posted Speed Limit (mph) Deceleration Rate Prior to End of Taper (ft/s2) Number of Vehicles Avg Std Dev 15th %ile 85th %ile AL-09 6.8 61 94 35 11.7 4.9 2.9 14.8 17 FL-10 7.9 71 216 45 5.5 4.0 0.6 8.0 28 FL-09 8.0 80 173 35 3.9 2.2 1.4 7.0 33 MS-03 8.9 93 186 45 6.9 2.4 4.0 9.2 24 FL-03 9.3 98 365 45 7.8 3.8 2.5 12.3 38 AL-08 10.2 115 230 35 4.2 1.8 2.1 6.0 17 MS-08 11.9 119 255 45 5.7 3.2 1.7 9.2 59 MS-05 11.4 125 80 35 11.5 3.9 8.0 15.2 20 TX-33 13.2 152 312 65 5.9 3.6 2.0 8.3 43 AL-03 16.9 186 180 50 6.4 1.9 4.3 7.9 44 TX-28 16.6 191 283 65 8.1 2.5 5.8 10.3 55 TX-21 17.7 204 115 50 6.3 1.6 4.2 8.0 32 All Sites 6.7 3.6 2.8 9.8 410 All Low-Speed Sites (35–45 mph) 6.7 4.1 2.2 10.8 236 All High-Speed Sites (50–65 mph) 6.8 2.7 4.2 9.0 174 a Taper rate is the amount of longitudinal distance for each unit of transverse distance (i.e., the length of taper for each unit of lane width added).

95 tion on the models that considered speed limits in separate categories. Those models are shown in Table 6-17 and 6-18. Those results show that the speed at the upstream counter and the speed in the taper were significant in affecting the rate of deceleration in the taper. The deceleration length was also significant at low-speed sites; indicating that some driv- ers consider all available length when completing their decel- eration. The signs associated with the coefficients for each of these three variables are as expected. The signs for deceleration length and speed in the taper are negative, because as these are reduced, one expects to see an increase in deceleration rate. Conversely, deceleration rate through the end of the taper increases as the upstream speed increases. While all three of these variables (Len_Decel, Spd_BC, and Spd_A) are statistically significant at a = 0.05 for low- speed sites, the speed variables have coefficients with larger Figure 6-10. Cumulative distribution of deceleration rate prior to the end of the taper. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 2 4 6 8 10 12 14 16 18 20 Cu m ul ati ve D is tr ib uti on Decel_AC (ft/s2) 35-45 mph 50-65 mph Decel_AC = 5.8 Table 6-17. Evaluation of deceleration rate prior to the end of the taper (speed limit < 50 mph). Response Decel_AC Speed_Group=Low (PSL_Major=35, 45) Summary of Fit RSquare 0.949283 RSquare Adj 0.948405 Root Mean Square Error 0.945365 Mean of Response 6.6861 Observations (or Sum Wgts) 236 Parameter Estimates Term Estimate Std Error DFDen t Ratio Prob>|t| Intercept 0.1182101 2.278592 5.171 0.05 0.9606 Spd_A 0.7968167 0.015806 227.7 50.41 <.0001* Len_Decel -0.017306 0.00569 5.102 -3.04 0.0280* Spd_BC -0.591261 0.016227 230.3 -36.44 <.0001* Len_Taper -0.032898 0.021939 4.969 -1.50 0.1944 REML Variance Component Estimates Random Effect Var Ratio Var Component Std Error 95% Lower 95% Upper Pct of Total Site 1.9640842 1.7553315 1.142023 -0.483034 3.9936966 66.263 Residual 0.893715 0.0840813 0.7493649 1.0844395 33.737 Total 2.6490465 100.000 -2 LogLikelihood = 697.37297772

96 magnitudes. This is a function of the units in which the vari- ables are reported; a 100-ft increase in deceleration length is associated with a decrease in deceleration rate of about 1.7 ft/s2, while a 1-mph increase in upstream speed is associ- ated with an increase in deceleration rate of about 0.8 ft/s2. Thus, a 1-ft change in deceleration length does not have as great an effect on deceleration rate as a 1-mph change in speed, but deceleration length is important in predicting the deceleration rate at low-speed sites. Results Compared to Green Book Guidance The text in the Green Book (2) that describes taper design (pp. 9-127, 9-128) suggests the importance of both taper length and speed: On high-speed highways it is common practice to use a taper rate that is between 8:1 and 15:1 (longitudinal:transverse or L:T). Long tapers approximate the path drivers follow when entering an auxiliary lane from a high-speed through lane. However, with exceptionally long tapers some through drivers may tend to drift into the deceleration lane—especially when the taper is on a hor- izontal curve. Long tapers may constrain the lateral movement of a driver desiring to enter the auxiliary lanes. This situation primarily occurs on urban curbed roadways. For urbanized areas, short tapers appear to produce better “targets” for the approaching drivers and to give more positive identification to an added auxiliary 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 deceleration length should be the same as if a longer taper was used. This results in a longer length of full-width pavement for the auxiliary lane. This type of design may reduce the likelihood that entry into the auxiliary lane may spill back into the through lane. Municipalities and urban counties are increasingly adopt- ing the use of taper lengths such as 100 ft for a single-turn lane and 150 ft for a dual-turn lane for urban streets. . . . The taper rate may be 8:1 [L:T] for design speeds up to 30 mph and 15:1 [L:T] for design speeds of 50 mph and greater. The recommended taper rates and lengths in the Green Book are presented in conjunction with the speeds of the vehicles using the left-turn lane, so the design of the taper area leading into the full-width deceleration lane recognizes an influence of speeds, but the design guidelines in the text are not directly tied to the 5.8 ft/s2 value presented in Table 9-22 (Table 6-2 in this document). Also, the guidelines give the designer the flexibility to use engineering judgment to deter- mine the appropriate taper rate for design speeds between 30 and 50 mph. Many of the study sites provided a taper length consistent with the recommended taper rates. If a taper rate of 15:1 is considered for left-turn lanes at sites with posted speed lim- its of 50 mph and higher, only TX-33 did not have at least 100% of the corresponding taper length; that site had 88% of the recommended length for that taper rate. Applying a taper rate of 8:1 for the remaining sites, only two sites (AL-09 and FL-10) did not have the corresponding length; AL-09 had 85% of the length needed for an 8:1 taper rate, and FL-10 provided 99 percent. Deceleration data from the study sites indicate that drivers are regularly using a deceleration rate that is higher than 5.8 ft/s2 while navigating through the left-turn taper, even though many sites provided the recommended taper length. To better understand the importance of this finding requires a more detailed inspection of specific components in the pro- cess that the Green Book uses to determine recommended Response Decel_AC Speed_Group=High (PSL_Major=50, 65) Summary of Fit RSquare 0.901794 RSquare Adj 0.89947 Root Mean Square Error 0.857278 Mean of Response 6.800208 Observations (or Sum Wgts) 174 Parameter Estimates Term Estimate Std Error DFDen t Ratio Prob>|t| Intercept -2.539578 5.091504 1.009 -0.50 0.7048 Spd_A 0.4630709 0.013009 168.1 35.60 <.0001* Len_Decel 0.0019478 0.005474 1.046 0.36 0.7803 Spd_BC -0.299187 0.014917 168 -20.06 <.0001* Len_Taper -0.026365 0.022337 1.006 -1.18 0.4465 REML Variance Component Estimates Random Effect Var Ratio Var Component Std Error 95% Lower 95% Upper Pct of Total Site 0.3817164 0.2805331 0.4230742 -0.548692 1.1097586 27.626 Residual 0.7349256 0.080187 0.5999918 0.921376 72.374 Total 1.0154587 100.000 -2 LogLikelihood = 476.18761109 Table 6-18. Evaluation of deceleration rate prior to the end of the taper (speed limit >_ 50 mph).

97 lengths. The 5.8 ft/s2 value in Table 9-22 is used in calculat- ing what is called the “full deceleration length” in Figure 9-48 (Figure 6-1 in this document). That length includes both the taper (L2) and the provided length within the full-width decel- eration lane (L3). A comparison of Figure 9-48 with Note 3 in Table 9-22 could suggest that the 5.8 ft/s2 rate occurs solely within the taper, but Note 3 specifically states that the rate applies to the distance in which the turning vehicle moves from the through lane into the turn lane. That distance may or may not fully coincide with the taper length. Meanwhile, the recommended taper rate on pp. 9-127 and 9-128 (a value between 8:1 and 15:1) is given independently of the calcula- tions in Table 9-22; the purpose of that taper rate is to provide a path into the left-turn lane that approximates the path that a driver would (or should) follow for a given speed during the lateral movement into the left-turn lane. The end result is that these two components of the left-turn taper design process are not directly linked, but the process provides a great deal of flexibility for the designer to consider site characteristics (e.g., speed and available right-of-way) to produce a left-turn lane design that meets the anticipated needs of a particular intersection. Data from this study show that 85% of observed drivers at high-speed sites decelerated at a rate of 4.2 ft/s2 or greater up to the end of the taper. A design that accommodates deceler- ating at 4.2 ft/s2 during the lateral movement into the turning lane provides for a more gradual, controlled deceleration, but a higher deceleration rate (closer to 6.5 ft/s2 for half of the observed drivers or 10 ft/s2 for the most aggressive drivers) could be acceptable if site constraints or other factors dictate a shorter length. The tradeoff for the shorter length, however, would occur in one or both of the following forms: • Less aggressive drivers would begin their deceleration ear- lier, either through coasting or applying the brake further upstream of the beginning of the taper, increasing the speed differential between turning and through vehicles. • Some drivers would accomplish more of their deceleration after completing their lateral movement, leading to much higher deceleration rates approaching the stop line and/or the back of the queue. Deceleration within the Deceleration Lane A summary of deceleration rates within the full-width decel- eration lane for each study site is shown in Table 6-19, with the sites ordered by deceleration length. The Green Book deceleration rate of 6.5 ft/s2 is within the range of rates from the study sites, but 8 of the 12 sites had a higher average rate. In addition, the 15th percentile deceleration rate for all sites was 5.3 ft/s2, and two of the 12 sites had a 15th percentile rate greater than 6.5 ft/s2. Figure 6-11 shows the cumulative distribution of the deceleration rates downstream of the taper (Decel_CE), in which about 35.6% of the observed left-turning vehicles had a deceleration rate lower than the 6.5 ft/s2 described in the Green Book; the 6.5 ft/s2 rate was approximately the 51st per- centile value for vehicles at lower-speed sites and the 15th percentile value for vehicles at high-speed sites. The highest deceleration rate for both high-speed and low-speed sites was about 15 ft/s2. Results Compared to Green Book Guidance Researchers considered that a possible explanation for the difference between the Green Book value of 6.5 ft/s2 and the observed values is the fact that almost all of the study sites had deceleration lane lengths that were less than the Green Book’s recommended values for those speeds. The Green Book (2) acknowledges the existence of this practice by saying that at Table 6-19. Deceleration rates of left-turning vehicles in the deceleration lane. Site Taper Length (ft) Decel Length (ft) Posted Speed Limit (mph) Deceleration Rate in Deceleration Lane (ft/s2) Number of Vehicles Avg Std Dev 15th %ile 85th %ile MS-05 125 80 35 5.6 1.8 4.0 6.3 20 AL-09 61 94 35 5.0 1.2 3.7 6.1 17 TX-21 204 115 50 8.0 1.6 6.2 9.3 32 FL-09 80 173 35 7.4 1.7 5.0 9.3 33 AL-03 186 180 50 10.3 2.1 8.2 12.2 44 MS-03 93 186 45 8.1 2.5 5.6 9.8 24 FL-10 71 216 45 8.0 1.5 6.4 9.4 28 AL-08 115 230 35 6.0 1.5 3.8 7.7 17 MS-08 119 255 45 7.2 1.7 5.5 8.6 59 TX-28 191 283 65 7.8 1.9 5.7 10.0 55 TX-33 152 312 65 9.9 2.3 7.6 12.3 43 FL-03 98 365 45 5.3 1.4 4.0 6.5 38 All Sites 7.7 2.4 5.3 10.3 410 All Low-Speed Sites (35–45 mph) 6.7 2.0 4.6 8.8 236 All High-Speed Sites (50–65 mph) 9.0 2.3 6.5 11.4 174

98 many locations, “it is not practical to provide the full length of the auxiliary lane for deceleration due to constraints such as restricted right-of-way, distance available between adjacent intersections, and extreme storage needs,” which is why there is an allowance for deceleration in the through lane upstream of the taper. (p. 9-127) The shorter lengths could persuade drivers that a more pronounced deceleration is needed to come to a stop at the stop line, especially if relatively little deceleration took place in the taper. To further explore this theory, researchers developed ANACOVA models that would quantify the effects of decel- eration length and other variables. Table 6-20 displays the Figure 6-11. Cumulative distribution of deceleration rate within the deceleration lane. 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 0 2 4 6 8 10 12 14 16 18 20 Cu m ul ati ve D is tr ib uti on Decel_CE (ft/s2) 35-45 mph 50-65 mph Decel_CE = 6.5 Table 6-20. Evaluation of deceleration rate in deceleration lane. Response Decel_CE Summary of Fit RSquare 0.784076 RSquare Adj 0.780316 Root Mean Square Error 1.134194 Mean of Response 7.691544 Observations (or Sum Wgts) 410 Parameter Estimates Term Estimate Std Error DFDen t Ratio Prob>|t| Intercept -1.371806 5.304099 6.097 -0.26 0.8044 PSL_Major 0.0489858 0.053769 6.199 0.91 0.3963 Len_Taper -0.001642 0.014231 6.112 -0.12 0.9119 Len_Decel -0.01747 0.004671 6.251 -3.74 0.0089* LW_Thru -1.008134 0.426735 6.287 -2.36 0.0542 LW_LT 1.0671197 0.735762 6.14 1.45 0.1960 Spd_BC 0.3187896 0.013705 400.9 23.26 <.0001* Spd_A -0.005365 0.012702 398.1 -0.42 0.6730 REML Variance Component Estimates Random Effect Var Ratio Var Component Std Error 95% Lower 95% Upper Pct of Total Site 0.609386 0.7839113 0.4740314 -0.14519 1.7130129 37.865 Residual 1.2863953 0.0914063 1.1244849 1.486175 62.135 Total 2.0703065 100.000 -2 LogLikelihood = 1331.2432682

99 results of that analysis, in which deceleration length and the vehicle speed in the taper were found to be significant. Deceleration length is negatively related to deceleration rate, while speed is positively related; both of those relationships are as expected. As a result, the model suggests that, all other variables being held constant, a 10-ft increase in deceleration length will reduce the deceleration rate by about 0.2 ft/s2, and a 1-mph increase in speed in the taper (i.e., at the start of the deceleration length) will increase the rate by approximately 0.3 ft/s2. Researchers also investigated whether the deceleration rate in the taper might have significantly affected decelera- tion in the full-width deceleration lane. The theory was that if vehicles decelerate more sharply within the taper that might lead to a decreased rate within the deceleration lane. A model similar to the one shown in Table 6-20 was run, substituting Decel_AC for Spd_A; however, Decel_AC was found to be not significant. In the Green Book (2) text that discusses speed differential, the following statement is made on page 9-127: Shorter auxiliary lane lengths will increase the speed differential between turning vehicles and through traffic. This relationship was evident in the speed differential data for the lower-speed sites in this study and inconclusive for the higher-speed sites. These findings along with the decel- eration data findings support a conclusion that shorter auxiliary lane lengths increase the deceleration rate as turn- ing vehicles approach the intersection. Compared to the 6.5 ft/s2 rate noted in the Green Book, approximately two- thirds of the drivers observed making left turns at the study sites decelerated at greater rates to come to a stop at the stop line. Indeed, page 3-3 of the Green Book describes 11.2 ft/s2 as a comfortable deceleration for most drivers in recom- mending a deceleration threshold for stopping sight distance, so rates greater than 6.5 ft/s2 are not unique to this study. In terms of design guidelines, however, results from this study indicate that a designer could produce a left-turn lane design that is associated with a deceleration rate of 6.5 ft/s2 and it would accommodate the current behavior of 85% of left- turning drivers at high-speed sites. Suggested Changes to Green Book Based on the results of this study, a two-stage deceleration process that uses rates of 4.2 ft/s2 during lateral movement into the turning lane and 6.5 ft/s2 within the deceleration lane would accommodate most drivers. In locations with geometric constraints or high-speed approaches, a design that incorporates a higher deceleration rate could be offered as an alternative. Exhibit 14-4 of the draft second edition of the Access Management Manual (97) lists distances for “limit- ing conditions” of deceleration distance; while the manual states that these values are supported by the research con- ducted by Gates et al. (95), Chang et al. (90), and Williams (99), they are equivalent to a constant deceleration rate of 10 ft/s2 throughout the full deceleration length (i.e., taper and full-width deceleration lane). In addition, the Gates, Chang, and Williams studies focused on drivers traveling straight through an intersection and responding to a change interval at a traffic signal. While drivers may be able to negotiate the left-turn lane at higher deceleration rates, a design that accommodates lower rates provides a more conservative design that is less demanding on drivers and contains more provision for storage of queues of left-turning vehicles. That more conservative design could be accomplished through the use of a constant 6.5 ft/s2 rate throughout the full decel- eration length, which is very similar to the existing lengths in Table 9-22 of the Green Book. Substituting those values into Table 9-22, and including the intermediate speed val- ues that are multiples of 5, would produce a result similar to Table 6-21. Speed, mph Deceleration and Lane-Change Distance, fta Typical Constrained 20 95 70 25 140 105 30 195 150 35 260 205 40 330 265 45 410 340 50 500 415 55 595 505 60 700 600 65 810 700 70 930 815 Notes: 1. The above full deceleration lengths are d2a + d2b in Figure 6-10. 2. The speed differential between the turning vehicle and following through vehicles is 10 mph when the turning vehicle “clears the through traffic lane” (i.e., distance d2a in Figure 6-10). 3. Deceleration lengths for a typical installation are based on 4.2 ft/s2 deceleration while moving from the through lane into the turn lane (distance d2a) and 6.5 ft/s2 deceleration after completing the lateral shift into the turn lane (distance d2b). 4. Deceleration lengths for a constrained installation are based on a 6.5 deceleration throughout the entire length. ft/s2 5. Typical lengths should be used for new roadway projects and for reconstruction projects where sufficient right-of-way exists to provide the additional length. 6. Deceleration rates are based on deceleration on dry, level pavement. Designs for approaches on downgrades of more than 2% and intersections at locations prone to wet pavement should account for the additional length necessary for vehicles to decelerate to a stop in those conditions. 7. Access points should not be allowed in the deceleration areas. a Rounded to 5 ft. Table 6-21. Deceleration lengths.

100 Source: Access Management Manual (draft second edition), exhibit 14-2b. Reproduced with permission of the Transportation Research Board. Figure 6-12. Upstream functional intersection area with a turn bay (97). Figure 6-12 shows the second half of Exhibit 14-2 from the draft second edition of the Access Management Manual (97), illustrating the various key dimensions of a right-turn lane. Adapting this figure for use with left-turn lanes will help further illustrate the dimensions of Table 9-22 and relevant text in the Green Book. Figure 6-12 also shows that turning vehicles do not necessarily complete their lateral shift into the turning lane within the taper area. A shorter taper length, such as the 50-ft length used by the Florida Department of Transportation (100), provides an efficient way of adding the turning lane and allowing more length for the full-width deceleration lane.