Design Guidance for Freeway Mainline Ramp Terminals (2012)

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.

33 The objective of the observational study was to gather speed and distance information for a large number of merging and diverging vehicles covering a range of ramp types and merge types. From this data, speed profiles of vehicles could be devel- oped to reflect current driver behavior and vehicle perfor- mance capabilities. Data from these speed profiles and other corresponding measures were analyzed to compare accel- eration and deceleration characteristics of the current driver population and vehicle fleet to the assumptions inherent in the existing AASHTO models for determining minimum accelera- tion and deceleration lane lengths. This section describes the study locations, the field data collection procedures, the data reduction process, and the analysis results for both entrance and exit ramps. 5.1 Study Locations Observational studies were conducted at 20 freeway main- line ramp terminals in Kansas, Missouri, Pennsylvania, and Texas. Specific ramps were chosen for inclusion in the study based on the following criteria: • Include ramps from different geographical areas of the country. • Include both entrance and exit ramps. • Include both parallel and tapered SCLs. • Include both loop and straight (i.e., diamond) ramps. • Include ramps with a range of grades (0 to 6 percent upgrade/ downgrades). • Include ramps where the freeway conditions range from free-flow to constrained (or forced) throughout portions of the day. • Operations near the ramps should be independent of upstream/downstream ramps. In addition to these criteria, the ability to set up the data col- lection equipment was factored into the site selection process. Table 14 provides the location of each ramp included in the study. Figure 6 shows two straight ramps (i.e., one entrance and one exit) included in the study, and Figure 7 shows two loop ramps (i.e., one entrance and one exit) included in the study. Appendix A (available on the TRB website at http:// www.trb.org/Main/Blurbs/167516.aspx) provides aerial views of all ramps included in the study. The following site characteristics were gathered for poten- tial study locations during the site selection process, prior to data collection, either in the field or from construction plans/ profiles or aerial photos in the office: General Ramp Characteristics: • Type of ramp (straight/loop), • Type of merge (parallel/taper), • Number of lanes, • Average lane width, • Average width of right shoulder, • Average width of left shoulder, • Radius of controlling curve, • Vertical profile/grade, • Posted speed limit and/or advisory speed sign, • Area type (CBD, urban mixed-use, suburban residential or recreational, suburban commercial), • Ramp volume, • Year of ramp volume, • Ramp vehicle mix (percent trucks), and • Design speed of ramp. General Freeway Characteristics: • Number of lanes (by direction), • Average lane width, • Average width of right shoulder, • Alignment (curve/tangent), • Posted speed limit, • Distance to nearest upstream ramp, • Distance to nearest downstream ramp, • Freeway volume, • Year of freeway volume, S e c t i o n 5 Observational Study of Freeway Mainline Ramp Terminals

34 • Vehicle mix (percent trucks), and • Design speed of freeway. Entrance Ramps: • Length from controlling feature to painted nose, • Length from painted nose to end of SCL (not including taper), and • Length of taper. Exit Ramps: • Length of taper, • Length from end of taper to painted nose, and • Length from painted nose to controlling feature. 5.1.1 Ramp Design and Geometry Table 15 presents basic design elements for each entrance ramp. Of the 11 entrance ramps studied, nine were straight (i.e., diamond) ramps, and two were loop ramps. Four ramps had tapered merge lanes, as illustrated in Figure 8, and the other seven ramps had parallel merge lanes, as illustrated in Figure 9. The “controlling feature” column indicates whether ramp curvature or the crossroad terminal is the design ele- ment that controls when a vehicle can begin accelerating to the merge speed. For example, on a loop ramp, the horizontal curvature limits the vehicle’s speed, and full acceleration can- not begin until the vehicle has exited the curve. Speeds on Ramp ID State Metropolitan area Freeway Cross street Direction Entrance ramps 1 KS Kansas City I-435 Quivira Rd WB Entrance 2 MO Kansas City I-435 US 24 NB Entrance 3 MO Kansas City I-435 63rd St SB Entrance 4 MO Kansas City I-435 Gregory Blvd NB Entrance 5 PA Pittsburgh I-376 Ardmore Blvd EB Entrance 6 PA Pittsburgh I-376 Wm. Penn Hwy WB Entrance 7 TX Dallas I-20 Belt Line Rd WB Entrance 8 TX Dallas I-20 Carrier Pkwy WB Entrance 9 TX Dallas I-635 Midway Rd WB Entrance 10 TX Dallas SH 114 Esters Rd EB Entrance 11 TX Dallas SH 114 Freeport Rd EB Entrance Exit ramps 12 KS Kansas City I-435 Quivira Rd EB Exit 13 KS Kansas City I-635 Metropolitan Ave NB Exit 14 MO Kansas City I-70 US 40 WB Exit 15 MO Kansas City I-435 US 24 SB Exit 16 MO Kansas City I-435 Gregory Blvd SB Exit 17 TX Dallas I-635 Freeport Pkwy EB Exit 18 TX Dallas I-635 Freeport Pkwy WB Exit 19 TX Dallas I-635 Marsh Ln EB Exit 20 TX Dallas I-635 Plano Rd EB Exit Table 14. Study ramps. WB Entrance EB Exit Figure 6. I-435/Quivira Rd—Ramps 1 and 12. (Image Credit: Google EarthTM Mapping Service). NB Entrance SB Exit Figure 7. I-435/US 24—Ramps 2 and 15. (Image Credit: Google EarthTM Mapping Service).

35 Ramp Ramp type Merge type Controlling feature Radius of controlling curve if ≤ 1,000 ft (ft) Grade Distance from painted nose to controlling feature (ft) Speed- change lane length (ft) Taper length (ft) I-435/Quivira Rd Straight Parallel Crossroad terminal 1,000 2% upgrade 890 485 235 I-435/US 24 Loop Parallel Horizontal alignment 150 1% upgrade 0 640 250 I-435/63rd St Straight Parallel Crossroad terminal N/A 2% upgrade 1,795 575 200 I-435/Gregory Blvd Straight Parallel Crossroad terminal N/A 3% upgrade 725 420 205 I-376/Ardmore Blvd Straight Parallel Horizontal alignment 475 4% upgrade 275 325 560 I-376/Wm. Penn Hwy Straight Parallel Horizontal alignment 955 6% upgrade 0 325 200 I-20/Belt Line Rd Straight Taper Crossroad terminal N/A 2% upgrade 1,390 470 300 I-20/Carrier Pkwy Straight Taper Crossroad terminal N/A 2% upgrade 905 640 350 I-635/Midway Rd Straight Taper Crossroad terminal N/A 1% upgrade 700 145 530 SH 114/Esters Rd Loop Taper Horizontal alignment 120 1% downgrade 395 455 560 SH 114/Freeport Rd Straight Parallel Crossroad terminal N/A < 1% upgrade 1,850 675 280 Table 15. Entrance ramp design features. Figure 8. Entrance ramp with taper merge. Figure 9. Entrance ramp with parallel merge.

36 straight ramps may also be limited by horizontal alignment if the ramp curves or bends. For this study, the horizontal align- ment is only considered controlling if the radius of the curve is equal to or less than 1,000 ft, consistent with current guidance provided in the 2004 Green Book. Grade also has an impact on acceleration, especially for large trucks on steep upgrades. The entrance ramps observed in this research had grades ranging from a very slight downgrade to a 6 percent upgrade. Grade was measured in the field and was determined based upon the vertical profile from the painted nose to the end of the SCL. Table 16 presents basic design elements for each exit ramp. Of the nine exit ramps studied, four were straight ramps, and five were loop ramps. Five ramps had tapered exit lanes, as illustrated in Figure 10, and the other four ramps had parallel exits, as illustrated in Figure 11. For exit ramps, the “control- ling feature” column indicates whether ramp curvature or the crossroad terminal is the design element that controls vehicle deceleration. For example, on a loop ramp, the curvature of the loop will require a nearly immediate reduction in speed as the vehicle exits the freeway onto a curve. On a straight ramp, a vehicle may not have to fully decelerate (to a stop or yield) until reaching the crossroad terminal. Deceleration may also be limited by horizontal alignment on a straight ramp if the ramp curves or bends. As on entrance ramps, the horizontal alignment is considered controlling if the radius of the curve is equal to or less than 1,000 ft. Grade also has an impact on deceleration. The exit ramps observed in this research had Ramp Ramp type Diverge type Controlling feature Radius of controlling curve if ≤ 1,000 ft (ft) Grade Distance from painted nose to controlling feature (ft) Speed- change lane length (ft) Taper length (ft) I-435/Quivira Rd Straight Parallel Crossroad terminal N/A 2% downgrade 1,025 615 250 I-635/Metropolitan Ave Straight Taper Crossroad terminal N/A 6% downgrade 835 125 360 I-70/US 40 Loop Parallel Horizontal alignment 180 1% upgrade 255 315 90 I-435/US 24 Loop Parallel Horizontal alignment 180 2% downgrade 0 595 270 I-435/Gregory Blvd Straight Parallel Crossroad terminal N/A 3% downgrade 900 620 190 I-635/Freeport Pkwy Loop Taper Horizontal alignment 200 1% upgrade 410 110 160 I-635/Freeport Pkwy Loop Taper Horizontal alignment 200 1% upgrade 400 100 180 I-635/Marsh Ln Straight Taper Crossroad terminal N/A 2% upgrade 995 40 135 I-635/Plano Rd Loop Taper Horizontal alignment 445 1% downgrade 340 85 170 Table 16. Exit ramp design features. Figure 10. Exit ramp with taper diverge.

37 grades ranging from a 6 percent downgrade to a 2 percent upgrade. Table 15 also provides for each entrance ramp the distance from the controlling feature to the painted nose, the length of the SCL from the painted nose to the beginning of the taper, and the length of the taper. Table 16 provides the equivalent measures for exit ramps—the distance from the beginning of the taper to the end of taper, the length of the SCL from the end of taper to the painted nose, and the length from the painted nose to the controlling feature. In instances where the crossroad terminal is the controlling feature, the length from the painted nose to the controlling feature is the length of the ramp. In instances where the distance from the painted nose to the controlling feature is 0 ft, the horizontal curve with a radius equal to or less than 1,000 ft ends or begins at the painted nose. For an entrance ramp, the SCL begins at the painted nose of the ramp and ends when the lane becomes narrower than 12 ft (i.e., where the taper begins). The taper begins at the end of the SCL and ends where the lane has completely merged back into the freeway lane. For an exit ramp, the SCL begins when the exit lane reaches a width of 12 ft (i.e., where the taper ends), and ends at the painted nose. The taper begins where the exit lane first begins to diverge from the freeway lane and ends where the SCL begins (i.e., where the taper width has reached 12 ft). The painted nose is defined here as the location where the freeway edgeline and ramp edgeline meet. These design features are illustrated in Figures 8 through 11. 5.1.2 Ramp and Freeway Speed Information Table 17 presents speed information for each of the ramps and associated freeway segments included in the study, including posted speed limit, freeway design speed, and ramp design speed. For ramp design speed, the design speed of the controlling feature was used. Where the horizontal alignment was the controlling feature, this was the design speed of the controlling curve. Where the crossroad terminal was the con- trolling feature, the stop condition was considered the appro- priate design criterion for the ramp. 5.1.3 Comparison of Existing Conditions to Green Book Criteria Each study ramp was evaluated to determine if it met the 2004 Green Book design criterion for the minimum length required for acceleration or deceleration as given in Green Book Exhibits 10-70 (entrance ramps), 10-73 (exit ramps), and 10-71 (adjustment for grade). For each ramp, the minimum length recommended by the Green Book is compared to the sum of the length of the SCL and the distance from the painted nose to the controlling feature. The minimum length recommended by the Green Book was determined based on the design speed (rather than average running speed) of the freeway, the ramp design speed (as shown in Table 17), and the grade of the ramp terminal. Tables 18 and 19 present comparisons of the actual acceler- ation and deceleration lane lengths to AASHTO criteria. Only three of the 11 entrance ramps meet or exceed the 2004 Green Book criteria, while all nine exit ramps meet or exceed the 2004 Green Book criteria. 5.2 Field Data Collection Procedures The general field data collection procedures performed to gather speed and volume data at freeway mainline ramp terminals are described below. Data collection activities were Figure 11. Exit ramp with parallel diverge.

38 similar at both entrance and exit ramps and for straight and loop ramps. Generally, field data were collected at each study site for 6 hours during a weekday during the following time periods: • AM or PM peak period—2 hours (i.e., the heavier direc- tional peak period was chosen for data collection), • AM or PM non-peak period—2 hours, and • Noon peak—2 hours. Data were collected using laser guns, traffic classifiers, and video equipment. Laser guns were used to collect speed and distance data of subject vehicles as they traveled along the ramps and acceleration or deceleration lanes. The laser guns Ramp Posted speed limit on freeway (mi/h) Freeway design speed (mi/h) Ramp design speed (mi/h) Entrance ramps I-435/Quivira Rd 65 75 Stop condition I-435/US 24 65 70 25 I-435/63rd St 65 70 Stop condition I-435/Gregory Blvd 65 70 Stop condition I-376/Ardmore Blvd 55 65 40 I-376/Wm. Penn Hwy 55 65 55 I-20/Belt Line Rd 60 70 Stop condition I-20/Carrier Pkwy 60 70 Stop condition I-635/Midway Rd 60 65 Stop condition SH 114/Esters Rd 60 70 25 SH 114/Freeport Rd 60 70 Stop condition Exit ramps I-435/Quivira Rd 65 75 Stop condition I-635/Metropolitan Ave 65 75 Stop condition I-70/US 40 55 65 30 I-435/US 24 65 70 30 I-435/Gregory Blvd 65 70 Stop condition I-635/Freeport Pkwy 60 70 30 I-635/Freeport Pkwy 60 70 30 I-635/Marsh Ln 60 65 Stop condition I-635/Plano Rd 60 65 40 Table 17. Freeway design, posted speeds, and ramp design speeds. Ramp Controlling feature Actual acceleration length (ft) Green Book minimum acceleration length (ft) Grade adjustment factor Adjusted minimum acceleration length (ft) Actual length minus adjusted minimum length (ft) Design status I-435/Quivira Rd Crossroad terminal 1,375 1,790 1 1,790 –415 Actual LESS than design I-435/US 24 Horizontal alignment 640 1,420 1 1,420 –780 Actual LESS than design I-435/63rd St Crossroad terminal 2,370 1,620 1 1,620 750 Actual GREATER than design I-435/Gregory Blvd Crossroad terminal 1,145 1,620 1.8 2,916 –1,771 Actual LESS than design I-376/Ardmore Blvd Horizontal alignment 600 770 1.6 1,232 –632 Actual LESS than design I-376/Wm. Penn Hwy Horizontal alignment 325 370 2.75 1,018 –693 Actual LESS than design I-20/Belt Line Rd Crossroad terminal 1,860 1,620 1 1,620 240 Actual GREATER than design I-20/Carrier Pkwy Crossroad terminal 1,545 1,620 1 1,620 –75 Actual LESS than design I-635/Midway Rd Crossroad terminal 845 1,410 1 1,410 –565 Actual LESS than design SH 114/Esters Rd Horizontal alignment 850 1,420 1 1,420 –570 Actual LESS than design SH 114/Freeport Rd Crossroad terminal 2,525 1,620 1 1,620 905 Actual GREATER than design Table 18. Comparison of actual acceleration lengths to 2004 Green Book criteria.

39 were connected to laptop computers and their readings were automatically recorded in a separate spreadsheet for each vehicle at a rate of approximately three readings per second. The speed and distance data were plotted during post pro- cessing to develop speed profiles of individual vehicles along the ramps and acceleration or deceleration lanes. Depend- ing on the geometry of the ramp and available sight distance, either one or two laser guns were used at each site to be able to develop speed profiles for vehicles over as much of the ramp and acceleration or deceleration lane as possible. Laser guns were handheld and operated by a researcher inside of a vehicle parked off the roadway in a location chosen based on several criteria: • Safety of data collectors and equipment, • Minimal impact of presence of data collectors and equip- ment on driver behavior or desired speeds (Note: Subject vehicles were tracked from the rear as they drove away from the laser gun), • Visibility of as much of the ramp and merge/diverge area as possible, and • Minimized angle between the laser gun and the vehicles being tracked. During a 2-hour study period, speed and distance data were collected for as many ramp vehicles entering or exiting the free- way lanes as possible. Both free-flow and platooned vehicles were recorded as such, and vehicles were classified generally as passenger cars (i.e., including light trucks and sport-utility vehicles) or trucks (i.e., including SU and tractor-semitrailer trucks). In situations where two lasers were used, the operator of the first laser gun would track a vehicle and then commu- nicate by radio to the operator of the second laser gun the description (i.e., make, model, and color) of the vehicle being tracked. The operator of the second laser gun would track the subject vehicle as soon as it became visible until completion of the maneuver being recorded. The operators would con- firm with each other that the vehicle was successfully tracked before storing the data for post processing. During post pro- cessing, the separate data files were matched by file name and time stamps and combined to create a single speed profile for each vehicle. For entrance ramps, the operator of the laser gun began tracking a vehicle (i.e., pulled the trigger of the laser) at the earliest point possible along the ramp and stopped tracking it (i.e., released the trigger of the laser) when the driver-side front tires crossed into the rightmost through lane of the free- way. When the controlling feature of the ramp was a horizon- tal curve, the operator began tracking vehicles as they exited the controlling curve. The final data points for each vehicle provided its final merge speed and merge location. For exit ramps, the operator of the laser gun began track- ing vehicles (i.e., pulled the trigger on the laser gun) when the passenger-side front tires crossed from the through lane of the freeway into the deceleration lane (or taper) and continued tracking vehicles to the controlling curve or for as long as pos- sible along a straight ramp. Thus, the initial recordings for each vehicle provided its diverge speed from the freeway and its diverge location. The final recordings provided its final speed at the controlling feature (or as near as possible to the control- ling feature). Due to the nature of the diverge maneuver and the capability to lock a laser gun on a diverging vehicle, there is more uncertainty in the initial diverge speed and diverge loca- tion data presented later compared to corresponding merge speed and merge location data for entrance ramps. Ramp Controlling feature Actual deceleration length (ft) Green Book minimum deceleration length (ft) Grade adjustment factor Adjusted minimum deceleration length (ft) Actual length minus adjusted minimum length (ft) Design status I-435/Quivira Rd Crossroad terminal 1,640 660 1 660 980 Actual GREATER than Design I-635/Metropolitan Ave Crossroad terminal 960 660 1.35 891 69 Actual GREATER than Design I-70/US 40 Horizontal alignment 570 470 1 470 100 Actual GREATER than Design I-435/US 24 Horizontal alignment 595 520 1 520 75 Actual GREATER than Design I-435/Gregory Blvd Crossroad terminal 1,520 615 1.2 738 782 Actual GREATER than Design I-635/Freeport Pkwy Horizontal alignment 520 490 1 490 30 Actual GREATER than Design I-635/Freeport Pkwy Horizontal alignment 500 490 1 490 10 Actual GREATER than Design I-635/Marsh Ln Crossroad terminal 1,035 570 1 570 465 Actual GREATER than Design I-635/Plano Rd Horizontal alignment 425 390 1 390 35 Actual GREATER than Design Table 19. Comparison of actual deceleration lengths to 2004 Green Book criteria.

40 In addition to gathering data for the merging and diverging vehicles, traffic classifiers were used to record vehicle speeds and traffic volumes in the rightmost lane of the freeway at each study site. The traffic classifiers were positioned at three locations at each study site: • Entrance Ramps – Upstream of the painted nose: This classifier was posi- tioned approximately 750 to 1,000 ft upstream of the painted nose. – Just prior to the painted nose: This classifier was positioned approximately 100 ft upstream from the painted nose. – Downstream of the end of taper: This classifier was posi- tioned approximately 500 to 750 ft from the end of the taper. • Exit Ramps – Upstream of the taper: This classifier was positioned approximately 750 to 1,000 ft upstream of the beginning of the taper. – Just prior to the taper: This classifier was positioned approximately 100 ft prior to the beginning of the taper. – Downstream of the painted nose: This classifier was posi- tioned approximately 500 to 750 ft downstream from the painted nose. Average traffic speeds and volumes were recorded for 15-minute intervals continuously during the study. A video recorder system was used primarily to collect merge/diverge locations of vehicles and to record unusual or critical behavior (e.g., braking, swerving, or use of the shoul- der beyond the acceleration lane) in the vicinity of the study sites. A pan-tilt-zoom camera was mounted to a mast arm on a video trailer. The trailer was parked off of the roadway in a location where the camera could view a significant portion of the freeway and the merge/diverge area. A digital video recorder stored the video data, which was later viewed in the office. At each site, a data collector marked locations along the acceleration or deceleration lane to divide it into thirds. These locations were marked with paint either on the shoul- der or near the lanes. Orange traffic cones were placed at the markings for a few minutes prior to each study period. The locations marked by the paint and traffic cones served as reference points that could be distinguished on the video for data reduction purposes. In the office, the videos were viewed to record the locations of merge/diverge maneuvers along the SCLs (i.e., in the first, middle, or last third of the SCL) and to note unusual or critical behavior. Figures 12 and 13 illustrate examples of the general equipment setup for typical straight entrance and exit ramps. An exception to the data collection procedures described above occurred at the two entrance ramps in Pittsburgh. Rather than collecting speed and volume data for the freeway mainline using traffic classifiers, speed and volume data were collected using non-intrusive radar technology. Speed and volume data were collected at one location near the gore of the entrance ramp. 5.3 Data Processing The laser gun readings, classifier measurements, and the video data collection required additional processing to ensure data quality. The three types of data were tied together as nec- essary through synchronization of internal clocks within the equipment. Data Collection Equipment (Straight Entrance Ramp) LG1 – Laser gun 1 TC1 –Traffic classifier (upstream) LG2 – Laser gun 2 (as necessary) TC2 –Traffic classifier (at gore) VT – Video trailer TC3 –Traffic classifier (downstream) - Traffic cones VT LG2 LG1 TC1 TC2 TC3 Figure 12. Example of equipment setup at a typical entrance ramp.

41 5.3.1 Laser Gun Data Processing The laser gun readings were processed using extensive quality checks, extrapolated as needed, and then smoothed using a local linear smoother. This was done to ensure that the profiles analyzed best reflected true vehicle profiles and that sources of measurement variability in the individual profiles were minimized. The quality checks for each speed profile generated from the laser gun readings included: • Ensuring vehicle descriptions matched for ramps with multiple lasers, • Exclusion of erroneous/inconsistent data points due to potential interferences from other vehicles, • Deletion of profiles with large gaps in readings, and • Exclusion of data points with invalid distances based on ramp length. The quality checks were implemented by visual inspection of the speed-distance profiles for each vehicle run. The merge and diverge locations are based on the point in time when the laser gun measurements ended or began, respectively. In some cases, the laser gun may not have been able to find the vehicle speed at the beginning or ending of the tracking process, resulting in initial or final data points where a time stamp was recorded but without an associated vehicle speed. In these cases, the first or last speed measure- ment was extrapolated as a constant and the location of the merge/diverge maneuver was calculated assuming that the first recorded speed was constant from the first time stamp until the first valid speed measurement (in the case of exit ramps) or that the last recorded speed was maintained until the point in time when the final measurement was taken (in the case of entrance ramps). After the quality checks and endpoint extrapolation were implemented, a local smoother was applied to the results. The laser gun values are reported to the nearest whole num- ber, which results in distinct jumps in speed over time/distance. Since this is not representative of the reality of vehicle perfor- mance, it was decided that a local linear smoother would bet- ter represent changes in speed and acceleration over distance. Using this approach allowed for seamless transition between lasers and reduced the impact of any potential erroneous mea- surements. The smoothing was performed using the PROC LOESS procedure in SAS v 9.2 with a local linear regression and smoothing parameter of 0.5. An example of a smooth vehicle speed profile is provided in Figure 14. 5.3.2 Classifier Data Processing The traffic classifier data were summarized into 15-minute intervals for each classifier roadway position. The average vehi- cle speed and the volume were recorded for each complete time interval. For time intervals that did not cover complete quarter- of-an-hour time periods (i.e., those at the beginning and end of the data collection), volume measures were excluded and average speed was compared to previous or latter time periods for consistency. Incomplete intervals with inconsistent average speed measurements were excluded. The average speed and volume for the three classifier posi- tions were plotted against the time stamp for the intervals to compare the consistency in the measurements. As expected, for exit ramps the volumes were lower at the downstream position VT LG1 LG2 TC1 TC2 TC3 Data Collection Equipment (Straight Exit Ramp) LG1 – Laser gun 1 TC1 – Traffic classifier (upstream) LG2 – Laser gun 2 (as necessary) TC2 – Traffic classifier (at gore) VT – Video trailer TC3 – Traffic classifier (downstream) - Traffic cones Figure 13. Example of equipment setup at a typical exit ramp.

42 and for entrance ramps, the volumes were higher in the down- stream. No major anomalies in the average speed between locations were observed. The painted nose collection point was selected to be repre- sentative of the freeway conditions (volume and speed) at the merge or diverge location. The average 15-minute interval speeds were divided into three speed categories: • Free Merge/Diverge: 15-minute average freeway speed above 50 mi/h; • Constrained Merge/Diverge: 15-minute average freeway speed between 40 and 50 mi/h, inclusive; and • Forced Merge/Diverge: 15-minute average freeway speed less than 40 mi/h. 5.3.3 Video Data Processing Video recordings from each of the study periods were reviewed in the office primarily to collect merge/diverge location information, but also to record unusual or critical behavior such as braking, swerving, or use of the shoulder beyond the end of the taper. For each 2-hour study period, a 15-minute period was selected for detailed data collection. For the given 15-minute time period, the merge/diverge loca- tion was recorded for each ramp vehicle entering/exiting the freeway. For entrance ramps, merge locations were grouped into five categories: • Before the painted nose, • In the first third of the SCL, • In the middle third of the SCL, • In the last third of the SCL, and • Into or beyond the taper. Similarly for exit ramps, diverge locations were grouped into five categories: • Before or within the taper, • In the first third of the SCL, • In the middle third of the SCL, • In the last third of the SCL, and • Beyond the painted nose. The merge/diverge locations were estimated based on known distances/locations along the SCL marked with paint (and cones) during data collection. During the same 15-minute time period, any unusual or critical maneuvers in the vicinity of the freeway mainline ramp terminal were recorded. Several examples of maneuvers that would be considered unusual or critical include: Figure 14. Example of smoothed vehicle speed profile across distance.

43 • A merging vehicle comes to a complete stop within or beyond the taper during free-merge conditions. • A following (i.e., platooned) vehicle in the SCL stops or decelerates suddenly to avoid a rear-end collision with a slower lead vehicle in the SCL waiting to merge. • A vehicle merges into the freeway at a lower speed and a following vehicle in the freeway decelerates suddenly or changes lanes to avoid a rear-end collision with the merging vehicle. 5.4 Analysis Results The analysis considered four measures of effectiveness to evaluate operations near entrance and exit ramps: • Merge/diverge location, • Merge/diverge speed, • Vehicle acceleration/deceleration along the ramp and SCL, and • Unusual or critical maneuvers in the vicinity of the ramp terminal. Analyses were performed separately for entrance and exit ramps. For entrance ramps, the following questions were addressed: 1. Were vehicles able to merge into freeway traffic from a point within the SCL, or did they merge prior to the painted nose (i.e., early) or into or beyond the taper (i.e., late)? 2. What was the difference in speed between merging vehicles and freeway vehicles? Were merging vehicles able to enter the freeway traffic at an appropriate speed? 3. Were vehicle acceleration rates greater than, less than, or similar to those assumed in the AASHTO Green Book? For exit ramps, the following questions were addressed: 1. Did vehicles diverge from the freeway traffic at a point along the SCL, or did they diverge prior to or along the taper (i.e., early) or beyond the painted nose (i.e., late)? 2. What was the difference in speed between diverging vehicles and freeway vehicles? Were diverging vehicles reducing their speed before diverging onto the ramp? 3. Were vehicle deceleration rates greater than, less than, or similar to those assumed in the AASHTO Green Book? These questions are explored in depth in the following subsections, first for entrance ramps, and then for exit ramps. The answers to these questions are used to draw general con- clusions about how effectively the ramp design accommo- dated vehicles as they accelerated along the ramp and merged with freeway traffic, or diverged from freeway traffic and decelerated along the ramp. 5.4.1 Entrance Ramps At entrance ramps, the primary objective of the ramp design is to (1) allow sufficient distance for vehicles to accelerate com- fortably from the crossroad terminal (or controlling horizontal alignment, in the case of ramps with a curve radius < 1,000 ft) to freeway speeds by the time a merge is required, and (2) pro- vide sufficient area to allow vehicles to safely find a gap in freeway traffic into which to merge. The distance needed for a vehicle to reach freeway speeds by the time a merge is required is a function of the vehicle’s acceleration rate along the ramp and SCL. Ramp design guidelines that assume an acceleration rate greater than what is comfortable or possible for the vehicle fleet may provide lengths that are too short. If an entrance ramp and SCL (the full area over which a vehicle may accelerate before merging) are too short, this may be evident by observa- tions of vehicles waiting to merge beyond the SCL and into or beyond the taper area or by merge speeds that are substantially lower than the speed of the freeway traffic into which they are merging. Both of these measures are considered below. In addi- tion, an analysis is performed to consider the acceleration pro- files of the vehicles to determine whether the current vehicle fleet is accelerating as expected in current ramp design guide- lines described in the AASHTO Green Book. In summary, under ideal operating conditions along an entrance ramp, vehicles would accelerate to near freeway speed (i.e., within 5 mi/h of the operating speed of the freeway) before merging, and during the merging maneuver vehicles would uti- lize most of the acceleration lane prior to merging (e.g., vehicles would merge in the middle or last third of the SCL). The great- est concern arises when vehicles merge late and at speeds sig- nificantly below the operating speed of the freeway. This type of behavior could be an indication that insufficient length was provided to accelerate to near freeway speeds or that insuffi- cient length was provided for gap acceptance. Lesser concerns are when vehicles merge close to freeway speeds from within the taper area, or when vehicles merge early but at speeds signifi- cantly below the operating speed of the freeway. The analyses were designed to compare actual driving practices to these ideal operating conditions and identify areas of concern. 5.4.1.1 Merge Location For an entrance ramp, vehicles will ideally merge onto the freeway from a point somewhere along the SCL—that is, beyond the painted nose, but prior to the taper. Figures 15 to 20 divide the possible merge locations into five categories: before the painted nose (gore), in the first third of the SCL

44 (early), in the middle third of the SCL (mid), in the last third of the SCL (late), and into or beyond the taper (taper). The figures below combine the data gathered from the video data reduction and the laser gun data. In Figure 15, the merge location is considered for all vehicle types on all of the study ramps, and broken out by data type (video on the left and laser on the right), and by freeway speeds. The top section shows merge locations for all freeway speed categories combined, the next section shows merge locations during free-flow condi- tions on the freeway (i.e., freeway speeds > 50 mi/h), the next section shows merge locations during constrained conditions on the freeway (i.e., freeway speeds 40 to 50 mi/h), and the final section shows merge locations during forced conditions on the freeway (i.e., freeway speeds < 40 mi/h). The proportions of merge location for each speed category shown in Figure 15 are somewhat dissimilar between those obtained from the video data and those obtained from the laser data. For example, under the constrained-merge condi- tion, approximately 5 percent of the vehicles merged in the final third of the SCL or along the taper according to the video data compared to approximately 57 percent of the vehicles according to the laser data. In considering the accuracy and reliability of the data from both data collection sources, the merge locations obtained from the laser data are considered to be more reliable for two reasons. First, the merge locations were directly observed in the field and recorded by the data collectors upon release of the trigger on the laser gun. Second, it is expected that there is more room for error in recording the merge location from the video compared to direct obser- vation in the field due to the angular position of the cam- era and quality of the video, especially for longer distances between the camera location and the merge locations. Third, the merge location data observed in the field from a sample of vehicles should be more representative of the actual dis- tribution of merge locations throughout the day, compared to selecting merge locations of all vehicles during a given 15-minute period from each 2-hr study period. For these rea- sons, the remainder of this discussion and information pre- sented in this section includes only the merge positions based upon the laser data. Figure 15 shows that very few vehicles merged prior to the painted nose, and that for the free-merge and forced- 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 3,327 Vehicles 9 Ramps Late/Taper 17.0% Video Data P er ce n t Percen t All Freeway Speeds Free Merge Constrained Merge Forced Merge 3,225 Vehicles 11 Ramps Late/Taper 33.3% Laser Data 2,636 Vehicles 9 Ramps Late/Taper 16.6% 2,935 Vehicles 11 Ramps Late/Taper 32.4% 465 Vehicles 2 Ramps Late/Taper 4.5% 109 Vehicles 5 Ramps Late/Taper 56.9% 226 Vehicles 2 Ramps Late/Taper 46.5% 181 Vehicles 5 Ramps Late/Taper 33.7% 7 53 22 12 5 0 20 40 60 80 100 3 26 37 21 12 8 51 24 13 4 0 20 40 60 80 100 3 26 39 22 10 0 88 7 3 2 0 20 40 60 80 100 8 14 21 15 42 11 15 27 21 25 0 20 40 60 80 100 Position Gore Early Mid Late Taper 10 31 25 17 17 Position Gore Early Mid Late Taper Figure 15. Merge location—all ramps, all vehicle types, by freeway speed categories (all speeds combined, freeway speeds > 50 mi/h, freeway speeds 40 to 50 mi/h, and freeway speeds < 40 mi/h).

45 merge conditions, a majority of vehicles merge at the early or mid positions, while for the constrained merge, a majority of vehicles merge in the late or taper positions. This observa- tion is intuitive because under free-merge conditions, vehicles should have no trouble finding a gap and should be able to merge as soon as a comfortable merge speed is reached. Under forced-merge conditions, gaps may be less frequent, but merging may take place with a certain amount of order, with each vehicle in the right lane of the freeway allowing one merging vehicle into the lane. Additionally, because free- way traffic is slower, the merging vehicle does not need to accelerate as much and can reach appropriate merging speed earlier. Under constrained merges, however, gaps are likely more difficult to find because they may be infrequent and because freeway traffic has not become congested enough to create a slow, ordered merging process. Figure 15 shows that 42 percent of merges in constrained conditions take place in the taper or beyond, and that an additional 15 percent take place in the final third of the SCL. Figure 16 shows similar information to that presented in Fig- ure 15, but compares the merge location proportions of pas- senger cars to those of trucks. In general, the two vehicle types behave similarly, but constrained-merge conditions are more difficult for trucks than for passenger cars. While 41 percent of passenger cars merge in the taper area or beyond under con- strained conditions, 56 percent of trucks merge in or beyond the taper. However, caution should be taken when compar- ing merge locations between passenger cars and trucks under the constrained-merge condition because the percentages for trucks are only based upon nine observations. Figure 17 compares the merge locations of vehicles merg- ing from loop ramps (i.e., low-speed ramps) to those merging from straight ramps (i.e., high-speed ramps). It appears that those coming from loop ramps tend to merge earlier in the SCL than those merging from straight ramps when the freeway is operating under free-merge conditions. Only two loop ramps were observed, and the observations at those ramps were all taken during free-merge conditions. Therefore, the obser- vations regarding the difference in merge location between loop and straight ramps cannot be made for constrained and forced-merge conditions. Figure 18 compares the merge locations of vehicles on parallel SCLs to those on taper SCLs. For all speed condi- tions combined (i.e., free, constrained, and forced), merges from tapered SCLs occur later than those from parallel SCLs. The difference is most extreme under constrained- merge conditions where 61 percent of vehicles on taper SCLs merged at a location beyond where the tapered lane narrowed to 12 ft, while only 12 percent of vehicles on a parallel SCL merged that late. One possible explanation for this is that on tapered SCLs, drivers tend to follow the path of the right edge line as they travel into the freeway lane, therefore crossing into the freeway lane at the point where the taper has narrowed to the approximate width of the vehicle. Parallel SCLs, on the other hand, are marked more similarly to a lane with a more defined end point, encouraging drivers to find a merge location before the taper. For all speed conditions combined, drivers on paral- lel SCLs merged into the freeway lane within the full length the SCL 94 percent of the time, while only 74 percent of vehicles merged in the freeway lane prior to the taper lane narrowing to 12 ft. When comparing the merge locations of vehicles on parallel SCLs to those on taper SCLs, it has to be recognized that the distance from the painted nose to the taper is longer for parallel SCLs than for tapered SCLs. In Figure 19, ramps that meet or exceed Green Book cri- teria for ramp and SCL length are compared to ramps that do not. When all merge conditions are considered together, there appears to be very little difference in the distribution of merge locations between the two categories. This similar- ity holds true when just considering free-merge conditions and forced-merge conditions. However, clear differences are apparent between the two groups when looking specifically at the constrained-merge observations. Under these con- ditions, vehicles are much more likely to merge late from a ramp that does not meet Green Book criteria than from those that do. At ramps that meet or exceed Green Book cri- teria, 11 percent of merges observed occurred in the late or taper areas, while for the ramps that do not meet Green Book criteria, 67 percent occurred in the late or taper areas. This may indicate that ramp length is less important when gaps are easy to find or when freeway speeds are low, but that it becomes more significant when freeway speeds are moderate and merge opportunities are more difficult to find. Again, caution should be taken when comparing merge loca- tions under the constrained-merge condition because of a low number of observations (19) for vehicles on ramps that meet or exceed Green Book criteria. Figure 20 compares the distribution of merge locations for free-flow merges (i.e., the speed of the vehicle on the ramp is not impeded by a lead vehicle on the ramp) and platooned merges (i.e., merges made by vehicles following a lead vehicle closely enough that the speed, acceleration, and merge loca- tion of the following vehicle may be influenced by the lead vehicle). Considering all merge conditions together, very little difference is evident between the two distributions. However, under constrained-merge conditions, platooned vehicles are much more likely to merge beyond the end of the SCL than free-flowing vehicles (58 percent versus 3 percent in the taper area), and under forced-merge conditions, platooned vehicles tend to merge earlier than free-flow vehicles (73 percent versus 55 percent in the gore, early, and mid areas). This may indi- cate that when freeway speeds are moderate and merge gaps are difficult to find, the lead vehicle has an advantage in

46 Heavy truck Late/Taper 30.3% Passenger car Late/Taper 33.5% 11 Ramps 267 Heavy trucks 2,958 Passenger cars P er ce nt All Freeway Speeds Free Merge Constrained Merge Forced Merge Heavy truck Late/Taper 28.1% Passenger car Late/Taper 32.7% 11 Ramps 231 Heavy trucks 2,704 Passenger cars Heavy truck Late/Taper 66.7% Passenger car Late/Taper 56.0% 5 Ramps 9 Heavy trucks 100 Passenger cars Heavy truck Late/Taper 37.0% Passenger car Late/Taper 33.1% 5 Ramps 27 Heavy trucks 154 Passenger cars Vehicle type Heavy truck Passenger car 3 3 25 26 42 37 18 22 12 12 0 20 40 60 80 100 2 3 26 26 44 38 18 22 10 10 0 20 40 60 80 100 11 8 0 15 22 21 11 15 56 41 0 20 40 60 80 100 15 9 22 33 26 25 19 16 19 17 0 20 40 60 80 100 Gore Early Mid Late Taper Figure 16. Merge location—all ramps by vehicle type (passenger cars versus trucks).

47 Loop Late/Taper 16.0% Straight Late/Taper 36.9% 567 Vehicles on loop ramps 2,658 Vehicles on straight ramps P er ce nt All Freeway Speeds Free Merge Constrained Merge Forced Merge Loop Late/Taper 16.0% Straight Late/Taper 36.3% 567 Vehicles on loop ramps 2,368 Vehicles on straight ramps Loop Late/Taper 0% Straight Late/Taper 56.9% 0 Vehicles on loop ramps 109 Vehicles on straight ramps Loop Late/Taper 0% Straight Late/Taper 33.7% 0 Vehicles on loop ramps 181 Vehicles on straight ramps Type of Ramp Loop (2 ramps) Straight (9 ramps) 6 3 45 22 33 38 9 24 7 13 0 20 40 60 80 100 6 2 45 22 33 40 9 25 7 11 0 20 40 60 80 100 0 8 0 14 0 21 0 15 0 42 0 20 40 60 80 100 0 10 0 31 0 25 0 17 0 17 0 20 40 60 80 100 Gore Early Mid Late Taper Figure 17. Merge location—by ramp type (straight versus loop).

48 Parallel Late/Taper 24.7% Taper Late/Taper 52.5% 2,231 Vehicles on parallel ramps 994 Vehicles on taper ramps P er ce nt All Freeway Speeds Free Merge Constrained Merge Forced Merge Parallel Late/Taper 24.2% Taper Late/Taper 51.0% 2,043 Vehicles on parallel ramps 892 Vehicles on taper ramps Parallel Late/Taper 31.0% Taper Late/Taper 73.1% 42 Vehicles on parallel ramps 67 Vehicles on taper ramps Parallel Late/Taper 29.5% Taper Late/Taper 51.4% 146 Vehicles on parallel ramps 35 Vehicles on taper ramps Type of Merge Parallel (7 ramps) Taper (4 ramps) 3 5 31 15 42 27 19 27 6 26 0 20 40 60 80 100 2 4 31 16 43 29 19 28 5 23 0 20 40 60 80 100 5 10 19 10 45 6 19 12 12 61 0 20 40 60 80 100 10 9 34 23 27 17 14 26 15 26 0 20 40 60 80 100 Gore Early Mid Late Taper Figure 18. Merge location—by merge type (parallel versus taper).

49 No Late/Taper 36.2% Yes Late/Taper 24.0% 2,461 Vehicles on ramps not meeting criteria 764 Vehicles on ramps meeting criteria P er ce nt All Freeway Speeds Free Merge Constrained Merge Forced Merge No Late/Taper 35.1% Yes Late/Taper 24.2% 2,199 Vehicles on ramps not meeting criteria 736 Vehicles on ramps meeting criteria No Late/Taper 66.7% Yes Late/Taper 10.5% 90 Vehicles on ramps not meeting criteria 19 Vehicles on ramps meeting criteria No Late/Taper 33.7% Yes Late/Taper 33.3% 172 Vehicles on ramps not meeting criteria 9 Vehicles on ramps meeting criteria Ramp Meets Criteria? No (8 ramps) Yes (3 ramps) 3 3 23 36 37 37 22 21 14 3 0 20 40 60 80 100 3 2 23 36 39 37 22 21 13 3 0 20 40 60 80 100 2 37 9 37 22 16 16 11 51 0 0 20 40 60 80 100 9 22 32 22 25 22 16 33 18 0 0 20 40 60 80 100 Gore Early Mid Late Taper Figure 19. Merge location—existing conditions (ramps meet current criteria: yes versus no).

50 Free-flow Late/Taper 33.2% Platooned Late/Taper 33.4% 11 Ramps 2,102 Free-flow vehicles 1,123 Platooned vehicles P er ce nt All Freeway Speeds Free Merge Constrained Merge Forced Merge Free-flow Late/Taper 32.9% Platooned Late/Taper 31.3% 11 Ramps 2,005 Free-flow vehicles 930 Platooned vehicles Free-flow Late/Taper 31.3% Platooned Late/Taper 67.5% 5 Ramps 32 Free-flow vehicles 77 Platooned vehicles Free-flow Late/Taper 44.6% Platooned Late/Taper 27.6% 5 Ramps 65 Free-flow vehicles 116 Platooned vehicles Vehicle position Free-flow Platooned 3 3 25 28 38 35 22 20 11 13 0 20 40 60 80 100 3 2 26 28 39 38 22 22 11 9 0 20 40 60 80 100 16 5 16 13 38 14 28 9 3 58 0 20 40 60 80 100 6 12 17 40 32 21 23 13 22 15 0 20 40 60 80 100 Gore Early Mid Late Taper Figure 20. Merge location—by vehicle position (free-flow versus platooned).

51 taking the first available gap, while in forced-merge condi- tions, the slower freeway speeds may allow platooned vehicles to merge into freeway traffic at an earlier location than the lead vehicle. It should also be noted that under free-merge conditions, when about twice as many free-flow vehicles as platooned vehicles were observed while under constrained- and forced-merge conditions, the ratio was reversed, with nearly twice as many observations of platooned vehicles as free-flow vehicles merging. This makes sense, since forced and constrained merges tend to take place during peak demand. Figure 21 shows the percentage of vehicles that merge in the late or taper regions by length of the SCL. In general, a greater percentage of vehicles merge in the late or taper areas at ramps with a short SCL, and fewer vehicles merge later along the SCL when it is long. The most significant findings from the examination of merge locations are as follows: • Vehicles tend to merge later under constrained-merge conditions than under free- or forced-merge conditions. • The distribution of merge locations for trucks is very similar to the distribution of merge locations for passenger cars. • In free-merge conditions, vehicles tend to merge near the middle of the SCL. • In forced-merge conditions, vehicles tend to merge earlier in the SCL, but also the distribution of merge locations under forced-merge conditions appears to be the flattest when comparing the three merge conditions. In other words, merge location is most evenly spread among the five merge locations during forced-merge conditions. • A higher percentage of vehicles merge early on loop ramps compared to straight ramps. • Vehicles merge later at tapered SCLs than at parallel SCLs. This is most evident under constrained-merge conditions. • Whether the ramp length met Green Book criteria was most important under constrained conditions. • The distribution of merge locations for free-flow vehicles is very similar to the distribution of merge locations for pla- tooned vehicles, considering all speed categories combined. The greatest difference occurs under constrained-merge P er ce nt L at e/ Ta pe r 0 20 40 60 80 100 Speed Change Lane Length (ft) 100 150 200 250 300 350 400 450 500 550 600 650 700 All Freeway Speeds Free Merge Constrained Merge Forced Merge Figure 21. Percent of vehicles merging in the late or taper regions of the SCL by SCL length.

52 conditions, in which case a larger percentage of platooned vehicles merge later than free-flow vehicles. 5.4.1.2 Merge Speed The data presented in this section compare the speed of merging vehicles at the time they begin to merge to the speed of freeway traffic in the rightmost lane. Specifically, this is a comparison of final merge speeds of vehicles based upon the laser gun data compared to the average 15-minute free- way speed of the rightmost lane of the freeway at the time of the merge. The Green Book states that the “geometrics of the ramp proper should be such that motorists may attain a speed that is within 10 km/h [5 mi/h] of the operating speed of the freeway by the time they reach the point where the left edge of the ramp joins the traveled way of the freeway.” It goes on to state that the location described is where the right edge of the ramp is 12 ft from the right edge of the through lane of the freeway. Table 20 provides a summary of the mean speed differen- tial measured between merging vehicles and freeway traffic for Type of ramp Type of merge No. of ramps Vehicle type No. of vehicles Merge speed minus freeway speed (mi/h) P-value for t-test of mean less than - 5 mi/h Mean Std dev Min Max All freeway speeds Loop Parallel 1 Truck 31 -24.8 5.4 -36.6 -15.6 0.003 Passenger car 291 -18.0 6.0 -33.5 1.0 0.048 Taper 1 Truck 8 -10.3 5.1 -18.0 -2.7 0.031 Passenger car 237 -4.6 7.0 -27.9 13.8 0.559 Straight Parallel 6 Truck 194 -17.7 8.5 -55.6 11.5 0.001 Passenger car 1,715 -12.1 7.8 -52.0 23.8 0.015 Taper 3 Truck 34 -10.2 6.5 -20.0 9.1 0.005 Passenger car 715 -6.7 9.5 -53.5 16.9 0.029 Free merge: Freeway speed > 50 mi/h Loop Parallel 1 Truck 31 -24.8 5.4 -36.6 -15.6 0.003 Passenger car 291 -18.0 6.0 -33.5 1.0 0.040 Taper 1 Truck 8 -10.3 5.1 -18.0 -2.7 0.021 Passenger car 237 -4.6 7.0 -27.9 13.8 0.636 Straight Parallel 6 Truck 165 -18.7 7.2 -55.6 3.8 0.001 Passenger car 1,556 -12.6 6.8 -52.0 23.8 0.016 Taper 3 Truck 27 -10.6 5.4 -20.0 -1.0 0.001 Passenger car 620 -6.5 8.5 -53.5 13.5 0.009 Constrained merge: Freeway speed between 40 and 50 mi/h Loop Parallel 0 Truck Passenger car Taper 0 Truck Passenger car Straight Parallel 2 Truck 4 -12.8 5.6 -31.6 5.6 0.344 3 Passenger car 38 -15.2 13.9 -38.7 12.4 0.158 Taper 1 Truck 5 -1.6 7.1 -9.4 9.1 0.553 2 Passenger car 62 -11.5 8.2 -50.3 16.9 0.286 Forced merge: Freeway speed < 40 mi/h Loop Parallel 0 Truck Passenger car Taper 0 Truck Passenger car Straight Parallel 3 Truck 25 -9.6 9.5 -27.2 11.5 0.052 Passenger car 121 -7.1 10.6 -29.3 21.2 0.027 Taper 1 Truck 2 -16.5 0.5 -16.8 -16.1 0.116 2 Passenger car 33 -13.8 6.4 -28.1 -1.9 0.006 Table 20. Speed differential by ramp type, merge type, and vehicle type.

53 the available combinations of ramp type, merge type, vehicle type, and merge condition. For each ramp and merge type, the mean value is the average of the individual ramp averages. Also provided are the standard deviation (pooled across indi- vidual ramps), minimum, and maximum values measured for each condition. A negative mean speed in Table 20 indicates that the freeway speed is greater than the merge speed. For conditions where the mean speed differential is greater than or equal to -5 mi/h, vehicles are within the appropriate merge speed according to the Green Book guidance; or, stated in another way, whenever the mean speed differential is less than -5 mi/h (e.g., a mean value of –10 mi/h indicates vehicles are merging 10 mi/h below freeway speeds), vehicles are merging into the freeway lane at speeds below the assumed design con- ditions according to the current design policy, which assumes that merging vehicles will be traveling at a speed within 5 mi/h of the freeway speed. Considering only observations made during free-merge conditions, the speed differentials for all combinations of geo- metric design and vehicle types were significantly less than -5 mi/h, except for passenger cars entering the freeway on loop taper entrance ramps. This indicates that most vehicles did not merge at speeds within 5 mi/h of the average freeway operating speed. For all combinations of ramp and merge type, the difference between merge speed and freeway speed was less for cars than for trucks. Fewer vehicles were observed under constrained-merge and forced-merge conditions, and these observations were only made at straight ramps. Under these conditions, cars and trucks behaved more similarly than under free-merge conditions. To further evaluate the effect of different ramp charac- teristics on the difference between merge speed and freeway speed (i.e., speed differential), a mixed effects analysis of vari- ance (ANOVA) model was developed. The model included main effects for ramp type (straight versus loop) and merge type (parallel versus taper). The interaction effect between the two was not included based on comparison of the differ- ences of group means, and it would not have been estimable given the number of loop ramps. The model also includes a random ramp effect to account for individual ramp-to-ramp variability and the fact that repeated measurements were taken at each ramp. The between-ramp variability differs for paral- lel and taper ramps. The model was estimated using restricted maximum likelihood (REML), and estimates of the differences between ramp types and merge types were obtained using least squares means. The statistical degrees of freedom for evalua- tion of the parameters were based on the number of ramps, not the number of vehicles. The mean speed differential for each ramp and merge type combination was evaluated to determine if it was statistically less than -5 mi/h. The results are provided by vehicle type and freeway merge condition in the last column of Table 20. Although there is only one ramp type for each merge type for loop ramps, the ANOVA model was used to evaluate each of the combinations. For passenger cars under free-merge conditions, the following results were obtained from the ANOVA model: • The mean speed differential for straight ramps (both types of merge) is -10.06 mi/h and is statistically less than -5 mi/h (p-value = 0.01). For loop ramps, the mean of -8.27 mi/h is also statistically less than -5 mi/h at the 90% confidence level (p-value = 0.06). • For parallel merge ramps (both ramp types), the mean speed differential of -12.74 mi/h is statistically less than -5 mi/h with a p-value of 0.02. The taper ramps have an average esti- mated differential of -5.59 mi/h and are not statistically less than -5 mi/h (p-value = 0.13). • Loop ramps tend to have a slighter higher mean merge speed differential (1.97 mi/h) but they are not statistically different from straight ramps (p-value = 0.10). • Parallel ramps have a lower merge speed differential than taper ramps (-7.15 mi/h) that is statistically significant at the 90% confidence level (p-value = 0.07). Table 21 examines whether ramp lengths that meet or exceed Green Book criteria experience a lower speed differen- tial between merging vehicles and freeway traffic than ramps with lengths shorter than the minimum recommended Green Book criteria. At parallel SCLs, both cars and trucks merged at speeds closer to freeway speeds at ramps that met Green Book criteria than at ramps that did not meet the criteria. At tapered SCLs, the speed differentials for both cars and trucks were very similar at ramps that met criteria and those that did not. When combining merge types, vehicles merged at speeds much below the average operating speeds of the freeway on ramps that did not meet current Green Book criteria than on ramps that met current Green Book criteria. Figure 22 presents a box plot illustrating similar infor- mation to that presented in Table 20, but breaks the obser- vations down by merge location. For each merge condition, the first box represents merges that took place at the gore, early, or mid positions, while the second box represents merges that took place at the late or taper positions. For free- and forced-merge conditions, the mean is similar between the two categories of merge location; however, under constrained-merge conditions, vehicles that merged later merged at a much closer speed to that of freeway traf- fic than those that merged earlier. This was true for both passenger cars and heavy vehicles. Figures 23 through 26 show subsets of the data presented in Figure 22. Figure 23 presents observations made at loop ramps with parallel SCLs. For this ramp and merge type,

54 Ramps meet current criteria Type of merge No. of ramps Vehicle type N obs Mean Std dev Min Max No Parallel 5 Trucks 198 -22.0 8.1 -55.6 11.5 Passenger cars 1,522 -16.1 8.0 -52.0 21.2 Taper 3 Trucks 40 -10.0 6.4 -20.0 9.1 Passenger cars 701 -5.7 8.7 -53.5 16.9 Yes Parallel 2 Trucks 27 -10.4 8.2 -33.0 3.8 Passenger cars 484 -5.1 6.3 -34.4 23.8 Taper 1 Trucks 2 -10.7 2.7 -12.7 -8.8 Passenger cars 251 -7.6 9.5 -50.3 13.5 No Both merge types combined 8 Trucks 238 -17.5 7.9 -55.6 11.5 Passenger cars 2,223 -12.2 8.2 -53.5 21.2 Yes Both merge types combined 3 Trucks 29 -10.5 8.1 -33.0 3.8 Passenger cars 735 -6.0 7.6 -50.3 23.8 Table 21. Speed differential by existing conditions (ramps meet current criteria: yes versus no). Figure 22. Merge speed differential by vehicle type and merge location for all ramp types.

Figure 23. Merge speed differential by ramp type, merge type, vehicle type, and merge location (ramp type: loop; merge type: parallel). Figure 24. Merge speed differential by ramp type, merge type, vehicle type, and merge location (ramp type: loop; merge type: taper).

Figure 25. Merge speed differential by ramp type, merge type, vehicle type, and merge location (ramp type: straight; merge type: parallel). Figure 26. Merge speed differential by ramp type, merge type, vehicle type, and merge location (ramp type: straight; merge type: taper).

57 observations were only made during free-merge conditions. Trucks have a greater speed differential than passenger cars, and early mergers have a greater speed differential than later mergers. Figure 24 presents observations made at loop ramps with tapered SCLs, and only under free-merge conditions. Like loop ramps with parallel SCLs, trucks have a greater speed differential than passenger cars, and early mergers have a greater speed differential than later mergers, but for each group of observations, the mean is closer to zero at tapered SCLs than at parallel SCLs. Figure 25 shows the speed differentials for straight ramps with parallel SCLs. Observations were made under all three merge conditions, although many fewer observations were made under constrained- and forced-merge conditions than under free-merge conditions. The speed differential was greater for vehicles that merged later than earlier (both cars and trucks) under free-flow conditions, but less for late mergers than early mergers under constrained- and forced-merge conditions except for trucks under forced-merge conditions. Figure 26 shows the speed differentials for straight ramps with tapered SCLs. Observations were made under all three merge conditions, although many fewer observations were made under constrained- and forced-merge conditions than under free-merge conditions. Under all three merge condi- tions, late mergers merged at speeds closer to freeway traffic than early mergers. This was most evident under constrained- merge conditions. Figure 27 shows the speed differentials for each of the four ramp type and merge type combinations for only the free- merge conditions. For both straight and loop ramps, those with a tapered SCL experience merge speeds closer to the freeway speed than those with parallel SCLs. This observation holds for both passenger cars and heavy vehicles. The most significant findings from the examination of merge speeds compared to freeway speeds are as follows: • Vehicles merge closer to freeway speeds along tapered SCLs compared to parallel SCLs. • Vehicles merge at speeds closer to freeway speeds under constrained- and forced-merge conditions than under free-merge conditions. This is probably because merging gaps are smaller and less frequent under constrained and forced conditions and require the merging vehicle to be traveling near the freeway speed to be able to accept the available gaps. Figure 27. Merge speed differential by ramp type, merge type, vehicle type, and merge location (for free-merge conditions only).

58 • Passenger cars merge at speeds closer to freeway speeds than do trucks. This is expected as trucks do not accelerate as quickly as passenger cars and would be unlikely to reach the same speeds as cars by the merge location. • When comparing ramp lengths that meet or exceed Green Book criteria or are less than Green Book criteria, the largest speed differentials occurred at ramps with parallel SCLs. In other words, at parallel SCLs the speed differentials were substantially lower (for both cars and trucks) at the ramps that met Green Book criteria than those that do not meet the criteria; while at taper SCLs the speed differentials (for both cars and trucks) were very similar at ramps that met criteria compared to those that did not. • In general, late mergers merge at speeds closer to freeway speeds than early mergers. • The speed differential between merge and freeway speeds is about the same when comparing straight and loop ramps. 5.4.1.3 Acceleration Rate The third measure of performance considered is the accel- eration rate of vehicles as they travel along the ramp and SCL to the point where they merge. While the AASHTO Green Book does not provide a table of the assumed acceleration rates used to determine guidance for minimum acceleration lane length, these accelerations can be back calculated from Exhibit 10-70, using initial speed, speed reached (i.e., merge speed), and acceleration lane length (see Table 1). The accel- eration rates measured in the field can be compared to the Green Book accelerations to determine whether vehicles are performing as assumed in the Green Book. However, this com- parison is not straightforward because vehicle acceleration in the field is a function of several variables including vehicle performance capabilities, driver preference, acceleration of other vehicles, operating conditions on the freeway, and, per- haps most importantly, ramp length. Ramps longer than the minimum length needed to reach the desired merge speed at a comfortable acceleration allow the driver more flexibility in determining when and where to accelerate along the ramp. When ramps are long, drivers may accelerate to very near their merge speed early along the ramp and then reduce their acceleration as they enter the SCL and look for a gap. Accelera- tions measured near the gore and along the SCL may not be representative of vehicle capabilities or driver comfort levels and preferences, because vehicles may not necessarily acceler- ate along the ramp at a constant rate. Because it was not pos- sible to capture speed and acceleration along the entire ramp length, the behavior of vehicles near the crossroad terminal is not well documented here. Field-measured accelerations that fall below the assumptions in the Green Book may indicate that drivers were completing most of the required accelera- tion prior to the location where initial speed measurements were taken, rather than that the accelerations assumed in the Green Book are too high. For this reason, comparisons of field- measured acceleration to Green Book assumptions need to be interpreted carefully and considered in context of the ramp characteristics and merge conditions. The analysis of acceleration rates includes a general analysis of the acceleration rates measured in the field and comparisons of acceleration rates measured in the field to assumed acceleration rates from the Green Book criteria for level grades (i.e., grades of 2 percent or less) and for grades of 3 percent and greater. In addition, the measured speed profiles of trucks on grades of 4 and 6 percent are compared to speed profiles generated using the TSPM, which is based on vehicle performance equations. The TSPM estimates the speed profile for an unimpeded truck on any specified verti- cal alignment given the truck’s weight-to-power ratio. The analysis of acceleration rates is based on the measured speed profiles of 3,225 vehicles. General Analysis of Acceleration Rates. Figures 28 through 33 provide the acceleration profiles of observed vehicles at each entrance ramp evaluated in this research. Each figure includes two ramps—one on the left and one on the right. The topmost sections show the acceleration profiles for all observed vehicles at the ramp, while the three lower sections show the profiles under the three merge conditions (i.e., free, constrained, and forced). Empty sections indicate that no observations were made for that merge condition at that particular ramp. The acceleration profiles are shown with boxplots, and can be interpreted as follows: • The x-axis shows distance along the ramp relative to the painted nose, which is marked as zero. For each ramp, the x-axis begins at the location of the controlling feature (crossroad terminal or exit of controlling curve) in the case of straight ramps, or at the location where speeds were first measured in the case of loop ramps. • At each measurement location, a “box” with “whiskers” pro- vides a summary of the acceleration measured for all of the study vehicles. The colored box indicates where the middle 50 percent of measured acceleration values fall. The whiskers indicate the minimum and maximum acceleration values observed at that location. • The line running through the boxes indicates the mean acceleration values along the ramp. • The left-side y-axis provides acceleration values. At zero, the vehicle is maintaining a constant speed. Negative values indicate the vehicle is slowing, while positive values indicate the vehicle is speeding up. • The s-shaped curve overlaid on the boxplot shows the cumulative percentage of observed vehicles that have merged at each location. The percentage is shown on the

Figure 28. Acceleration profiles and cumulative merge curve by merge condition (NB entrance ramp at I-435/US 24 and WB entrance ramp at I-376/Wm. Penn Hwy). Figure 29. Acceleration profiles and cumulative merge curve by merge condition (EB entrance ramp at I-376/Ardmore Blvd and EB entrance ramp at SH 114/Esters Rd).

Figure 30. Acceleration profiles and cumulative merge curve by merge condition (NB entrance ramp at I-435/Gregory Blvd and WB entrance ramp at I-635/Midway Rd). Figure 31. Acceleration profiles and cumulative merge curve by merge condition (WB entrance ramp at I-435/Quivira Rd and WB entrance ramp at I-20/Carrier Pkwy).

61 right-side y-axis. As more vehicles merge, fewer observa- tions are included in the boxes. For example, at the hori- zontal distance (x-axis) where the s-shaped curve crosses the horizontal dashed line, 50 percent of vehicles have merged and the box and whiskers shown at this location represents only the remaining 50 percent of the original vehicles. Several general observations from the acceleration profiles (i.e., Figures 28 through 33) are as follows: • The acceleration rates change along the ramp and SCL. • The range of acceleration rates decreases proceeding along the SCL. • Negative accelerations along the ramp and SCL were observed and are most evident under constrained and forced conditions. Tables 22 through 25 compare ramp length, initial mea- sured speed, and acceleration rates to those recommended or assumed in the Green Book. The tables are organized as described below. Rows. Each row provides information about one entrance ramp. The ramps are organized first by controlling feature, with those controlled by horizontal alignment in the upper section and those controlled by the crossroad terminal in the Figure 32. Acceleration profiles and cumulative merge curve by merge condition (WB entrance ramp at I-20/Belt Line Rd and SB entrance ramp at I-435/63rd St). Figure 33. Acceleration profiles and cumulative merge curve by merge condition (EB entrance ramp at SH 114/Freeport Rd).

Ramp location Diff. between actual accel. lane length and GB criteria1 (ft) No. of vehicles Initial measurements Merge measurements Acceleration Location (ft) Average speed (mi/h) Assumed GB speed at controlling feature (mi/h) Average distance past painted nose (ft) Average speed (mi/h) Desired GB speed reached (mi/h) Assumed GB rate (ft/s2) Percentiles from initial speed measured (ft/s2) From controlling feature From painted nose 15th 50th Controlling feature: Horizontal alignment I-435/US 24 –780 195 0 0 37.0 22 241 44.6 53 1.77 1.76 2.66 I-376/Wm. Penn Hwy –693 162 0 0 45.1 44 231 47.0 50 1.65 0.10 0.75 I-376/Ardmore Blvd –632 200 0 –275 43.4 36 198 47.7 50 1.69 0.22 0.93 SH 114/Esters Rd –570 196 0 –395 44.9 22 196 55.1 53 1.77 1.20 2.01 Controlling feature: Crossroad terminal I-435/Gregory Blvd –1,771 228 205 –520 31.9 0 176 50.1 53 1.87 1.70 2.39 I-635/Midway Rd –565 109 200 –500 38.5 0 227 47.6 50 1.92 0.52 1.37 I-435/Quivira Rd –415 49 370 –520 40.0 0 347 52.1 55 1.83 0.93 1.34 I-20/Carrier Pkwy –75 215 325 –580 46.5 0 399 55.4 53 1.87 0.54 1.30 I-20/Belt Line Rd 240 172 1,050 –340 51.1 0 337 56.5 53 1.87 0.55 1.00 I-435/63rd St 750 130 795 –1,000 47.3 0 185 56.5 53 1.87 0.49 0.93 SH 114/Freeport Rd 905 179 1,310 –540 49.5 0 257 58.8 53 1.87 1.01 1.33 Note: GB = Green Book. 1 Negative values indicate the actual acceleration lane length is less than minimum recommended by the Green Book. Positive values indicate the actual acceleration lane length is greater than minimum recommended by the Green Book. Table 22. Observed acceleration rates for free-flow ramp vehicles (free merge and passenger cars only).

Ramp location Diff. between actual accel. lane length and GB criteria1 (ft) No. of vehicles Initial measurements Merge measurements Acceleration Location (ft) Average speed (mi/h) Assumed GB speed at controlling feature (mi/h) Average distance past painted nose (ft) Average speed (mi/h) Desired GB speed reached (mi/h) Assumed GB rate (ft/s2) Percentiles from initial speed measured (ft/s2) From controlling feature From painted nose 15th 50th Controlling feature: Horizontal alignment I-435/US 24 –780 92 0 0 35.9 22 208 42.2 53 1.77 0.96 2.49 I-376/Wm. Penn Hwy –693 139 0 0 38.3 44 213 41.0 50 1.65 0.16 1.10 I-376/Ardmore Blvd –632 210 0 –275 40.3 36 187 44.7 50 1.69 0.08 1.03 SH 114/Esters Rd –570 41 0 –395 45.2 22 240 55.1 53 1.77 1.06 1.69 Controlling feature: Crossroad terminal I-435/Gregory Blvd –1,771 47 205 –520 31.6 0 178 49.4 53 1.87 1.62 2.33 I-635/Midway Rd –565 27 200 –500 38.9 0 285 50.9 50 1.92 0.97 1.67 I-435/Quivira Rd –415 37 370 –520 38.7 0 318 48.7 55 1.83 0.39 1.20 I-20/Carrier Pkwy –75 45 325 –580 44.7 0 344 50.0 53 1.87 –1.01 1.18 I-20/Belt Line Rd 240 51 1,050 –340 48.8 0 343 53.6 53 1.87 0.18 1.06 I-435/63rd St 750 136 795 –1,000 44.1 0 185 53.1 53 1.87 0.40 0.85 SH 114/Freeport Rd 905 39 1,310 –540 50.0 0 204 59.0 53 1.87 0.96 1.34 1 Negative values indicate the actual acceleration lane length is less than minimum recommended by the Green Book. Positive values indicate the actual acceleration lane length is greater than minimum recommended by the Green Book. Table 23. Observed acceleration rates for platooned ramp vehicles (free merge and passenger cars only).

Ramp location Diff. between actual accel. lane length and GB criteria1 (ft) No. of vehicles Initial measurements Merge measurements Acceleration Location (ft) Average speed (mi/h) Assumed GB speed at controlling feature (mi/h) Average distance past painted nose (ft) Average speed (mi/h) Desired GB speed reached (mi/h) Assumed GB rate (ft/s2) Percentiles from initial speed measured (ft/s2) From controlling feature From painted nose 15th 50th Controlling feature: Horizontal alignment I-435/US 24 –780 287 0 0 36.7 22 231 43.8 53 1.77 1.60 2.58 I-376/Wm. Penn Hwy –693 301 0 0 42.0 44 223 44.2 50 1.65 0.13 0.90 I-376/Ardmore Blvd –632 410 0 –275 41.8 36 192 46.2 50 1.69 0.15 0.98 SH 114/Esters Rd –570 237 0 –395 45.0 22 204 55.1 53 1.77 1.12 1.95 Controlling feature: Crossroad terminal I-435/Gregory Blvd –1,771 275 205 –520 31.8 0 177 50.0 53 1.87 1.67 2.39 I-635/Midway Rd –565 136 200 –500 38.6 0 239 48.3 50 1.92 0.72 1.39 I-435/Quivira Rd –415 86 370 –520 39.5 0 335 50.6 55 1.83 0.63 1.34 I-20/Carrier Pkwy –75 260 325 –580 46.2 0 389 54.4 53 1.87 0.46 1.27 I-20/Belt Line Rd 240 223 1,050 –340 50.6 0 338 55.8 53 1.87 0.43 1.01 I-435/63rd St 750 266 795 –1,000 45.7 0 185 54.8 53 1.87 0.43 0.88 SH 114/Freeport Rd 905 218 1,310 –540 49.6 0 247 58.8 53 1.87 0.98 1.34 1 Negative values indicate the actual acceleration lane length is less than minimum recommended by the Green Book. Positive values indicate the actual acceleration lane length is greater than minimum recommended by the Green Book. Table 24. Observed acceleration rates for free-flow and platooned ramp vehicles combined (free merge and passenger cars only).

Ramp location Diff. between actual accel. lane length and GB criteria1 (ft) No. of vehicles Initial measurements Merge measurements Acceleration Location (ft) Average speed (mi/h) Assumed GB speed at controlling feature (mi/h) Average distance past painted nose (ft) Average speed (mi/h) Desired GB speed reached (mi/h) Assumed GB rate (ft/s2) Percentiles from initial speed measured (ft/s2) From controlling feature From painted nose 15th 50th Controlling feature: Horizontal alignment I-435/US 24 –780 28 0 0 32.2 22 205 36.7 53 1.77 0.98 1.55 I-376/Wm. Penn Hwy –693 13 0 0 37.2 44 209 36.7 50 1.65 –1.07 0.30 I-376/Ardmore Blvd –632 50 0 –275 39.8 36 160 42.7 50 1.69 0.13 0.58 SH 114/Esters Rd –570 8 0 –395 39.8 22 410 50.2 53 1.77 0.85 1.32 Controlling feature: Crossroad terminal I-435/Gregory Blvd –1,771 13 205 –520 26.2 0 146 39.0 53 1.87 0.73 1.42 I-635/Midway Rd –565 13 200 –500 29.8 0 191 43.5 50 1.92 1.09 1.58 I-435/Quivira Rd –415 10 370 –520 34.7 0 291 42.7 55 1.83 0.56 1.13 I-20/Carrier Pkwy –75 5 325 –580 42.6 0 565 53.1 53 1.87 0.67 1.04 I-20/Belt Line Rd 240 2 1,050 –340 47.1 0 308 51.3 53 1.87 0.58 0.68 I-435/63rd St 750 20 795 –1,000 34.7 0 173 46.1 53 1.87 0.42 0.95 SH 114/Freeport Rd 905 4 1,310 –540 46.1 0 262 56.8 53 1.87 1.15 1.21 1 Negative values indicate the actual acceleration lane length is less than minimum recommended by the Green Book. Positive values indicate the actual acceleration lane length is greater than minimum recommended by the Green Book. Table 25. Observed acceleration rates for free-flow ramp vehicles (free merge and trucks only).

66 lower section. Within the upper section, the first two ramps are loop ramps (where the controlling feature—horizontal curvature—ends at the painted nose), and the second two ramps are straight ramps with a horizontal curve that con- trols speed. Within each section, the ramps are organized in order of difference between actual length of the ramp and the Green Book recommended ramp length. Columns Difference between actual acceleration lane length and Green Book criteria—Shows the difference between the total length provided from the controlling feature to the painted nose plus the length from the painted nose to the start of the taper (SCL) and the ramp length provided in Exhibit 10-70 of the Green Book and adjusted based upon grade using Exhibit 10-71. The Green Book criteria length is found using the design speed of the freeway and the radius of the controlling feature. For ramps without controlling horizon- tal curvature, the stop condition was assumed for the initial speed. Negative values in this column indicate that the actual ramp length is less than the Green Book criteria, and positive numbers indicate that actual ramp length exceeds criteria. Number of vehicles—Total number of vehicle observations included in the acceleration analysis described in the table caption. Location from controlling feature (initial measurements)— Distance in feet of the initial speed measurement from the controlling feature. For ramps controlled by horizontal alignment, this value is zero because the initial speed mea- surement was taken as the vehicles exited the feature. For ramps not controlled by horizontal alignment, this is the distance from the crossroad terminal to the location where speeds were first captured. Location from painted nose (initial measurements)—Distance in feet of the initial speed measurement from the painted nose. For loop ramps, this value is zero because the hori- zontal curve ends at the painted nose, and this is where initial measurements were taken. For straight ramps, this is the distance upstream of the painted nose where speeds were first measured. Average speed (initial measurements)—The mean of all observed speeds at the location of initial speed measurement. Assumed Green Book speed at the controlling feature (initial measurements)—The assumed initial speed as shown in Exhibit 10-70 in the Green Book, which is based on the design speed of the controlling feature. For example, a ramp con- trolled by a horizontal curve with a design speed of 30 mi/h has an assumed initial speed of 26 mi/h. For straight ramps not controlled by horizontal alignment, the initial speed is assumed to be zero at the crossroad terminal. Average distance past painted nose (merge measurements)— Average distance in feet from the painted nose to the merge location of all observed vehicles. Average speed (merge measurements)—Average merge speed in mi/h of all observed vehicles. Desired Green Book speed reached (merge measurements)— The desired speed in mi/h reached at the time of merge as indicated in Exhibit 10-70 of the Green Book based on freeway design speed. For example, a highway with a design speed of 65 mi/h has a speed reached value of 50 mi/h. Assumed Green Book rate (acceleration)—Acceleration in ft/s2 calculated from the assumed Green Book speed at control- ling feature (initial speed), desired Green Book speed reached (speed reached), and minimum acceleration lane length from Exhibit 10-70 in the Green Book. 15th percentile from initial speed measured (acceleration)— Taken from the distribution of the acceleration of each vehicle calculated from the initial measured speed, the mea- sured merge speed, and the distance from where the initial speed was measured to the location of the merge. Eighty-five percent of vehicles exceed this acceleration. 50th percentile from initial speed measured (acceleration)— Taken from the distribution of the acceleration of each vehicle calculated from the initial measured speed, the measured merge speed, and the distance from where the initial speed was measured to the location of the merge. Fifty percent of vehicles exceed this acceleration. Table 22 presents the information described above for all observed free-flow passenger cars during free-merge condi- tions. Table 23 presents information for platooned passenger cars during free-merge conditions. Table 24 presents informa- tion for all passenger cars (free-flow and platooned combined) during free-merge conditions. Table 25 presents information for free-flow trucks during free-merge conditions. These tables show several important observations: 1. At every ramp except at I-376/Wm. Penn Hwy, the aver- age measured initial speed is greater than the initial speed assumed in the Green Book. At ramps controlled by hori- zontal alignment, this means that vehicles are exiting the controlling curves at higher speeds than what the Green Book assumes based on the design speeds of the curves. Because vehicles are exiting the curves at higher speeds, they require less distance from that point to accelerate to merg- ing speeds than what the Green Book criteria suggests. For ramps not controlled by horizontal alignment (i.e., straight ramps with no controlling curve), initial speeds are assumed to be zero at the crossroad terminal, but measured initial speeds are not available until several hundred feet along the ramp in most cases, so true speeds at the crossroad terminal are unknown. At these ramps, the measured ini- tial speeds range from 32 to 52 mi/h for passenger cars, indicating that a great deal of acceleration has occurred prior to the initial measured speed. Comparing Table 22

67 to Table 23, platooned passenger cars are traveling slightly slower than free-flow passenger cars at the location of ini- tial speed measurement, most likely because their speed is limited by the vehicles they are following. Table 25 shows that the initial speed measured for free-flow trucks is around 5 mi/h less than that measured for free-flow pas- senger cars in Table 22, which indicates that trucks travel at slower speeds around the controlling horizontal curve and accelerate at a lower rate from the crossroad terminal. 2. For free-flow passenger cars under free-merge condi- tions, the measured average speed at the merge location tends to be a few mi/h less than the desired speed reached assumed in the Green Book when the ramp length is less than the recommended, and a few mi/h greater than the Green Book desired speed reached when the ramp length is longer than recommended. 3. Acceleration rates calculated from the measured initial speed, measured merge speed, and distance between the two measurements are, in general, lower than the assumed acceleration rates in the Green Book. At the 15th percen- tile, all calculated accelerations are lower than assumed accelerations. At the 50th percentile, a few of the calcu- lated accelerations exceed those assumed in the Green Book, and they occur at ramps that are several hundred feet shorter than the recommended minimum Green Book criteria. 4. When comparing acceleration rates of free-flow passenger cars and platooned passenger cars, there is no consistent pattern across ramps. At some ramps, the free-flow pas- senger cars have higher acceleration rates, while at other ramps platooned vehicles have higher acceleration rates. 5. The acceleration rates of free-flow trucks are lower than for free-flow passenger cars, with the exception of two ramps (i.e., I-635/Midway Rd and I-435/63rd St). Comparison of Acceleration Rates on Level Grade. Table 26 provides a comparison of the acceleration rates derived from the Green Book recommended ramp lengths to accel- eration rates measured in the field for various vehicle groups under various conditions. Only acceleration rates of vehicles on ramps with flat grades of two percent or less were used in the development of Table 26. The table format is based on Green Book Exhibit 10-70, and the values for design speed, speed reached, curve design speed, and initial speed are all taken directly from that figure. Below the presentation of the Green Book–assumed acceleration rates are four categories of field- measured acceleration rates: Condition 1—This category shows the 50th percentile accel- eration rate for passenger cars that were not in a platoon on the ramp and entered the freeway when freeway speeds were greater than 50 mi/h. Platooned vehicles are excluded because it is assumed their speed and acceleration choices are constrained by vehicle(s) in front of them. Condition 2—This category includes the same population of speed profiles as the previous category, but the 15th per- centile speed is considered rather than the median speed, to show that 85 percent of vehicles are accelerating at a rate greater than what is shown. Platooned vehicles are excluded because it is assumed their speed and accelera- tion choices are constrained by vehicle(s) in front of them. Condition 3—This category looks at heavy vehicles rather than passenger cars, but again, only those that are not pla- tooned on the ramp and that are entering freeway traffic that is moving faster than 50 mi/h. Platooned vehicles are excluded because it is assumed their speed and acceleration choices are constrained by vehicle(s) in front of them. Condition 4—This category considers all passenger cars (both those in platoons and not in platoons on the ramp) that are entering the freeway when the freeway speed is between 40 mi/h and 50 mi/h. Platooned vehicles are included because constrained-merge conditions typically occur during peak traffic flow periods where a majority of vehicles on the ramp are in platoons. Available observa- tion data for free-flow vehicles in constrained-merge con- ditions is limited. Constrained-merge conditions were considered more crit- ical for evaluation than forced-merge conditions based on the findings for the other measures of effectiveness (i.e., merge location and speed differential). In both of those analyses, it was shown that the measures showed a bigger change between free and constrained-merge conditions than between free and forced-merge conditions. Table 26 compiles field-measured acceleration data for comparison with Green Book assumed accelerations. For each condition, the cells in the table were populated by developing a database of the speed profile of each vehicle as measured by the laser gun in the field. This provided vehicle speed at known distances along the ramp for all the recorded vehi- cles. For each speed profile, an acceleration rate was calcu- lated for the sections of the profile that correspond to a given cell in the table (identified by an initial speed and a speed reached). Since the database provides the location along the ramp where each speed is first reached, the length between initial speed and speed reached is known and can be used to calculate accelerations between the two points. For example, if a given speed profile had an initial speed of 28 mi/h and a merge speed of 42 mi/h, acceleration rates could be calculated for four cells in the table (initial speed of 30 mi/h to speeds reached of 31, 35, and 39 mi/h, and initial speed of 36 mi/h to speed reached of 39 mi/h). This process was repeated for each speed profile, so that each cell in the table included sev- eral observations. Only sections of speed profiles that did

68 Design speed (mi/h) Speed reached (mi/h) A cceleration length, L (ft) for entrance cur ve design speed (mi/h) Stop 15 20 25 30 35 40 45 50 Initial speed (mi/h) 0 14 18 22 26 30 36 40 44 2004 Green Book acceleration rates (ft/s2 ) (deri ve d from recommended minimum ramp lengths ) 30 35 40 45 50 55 60 65 70 75 23 27 31 35 39 43 47 50 53 55 3.18 2.81 2.88 2.36 2.28 2.08 1.99 1.92 1.87 1.83 2.57 2.62 2.76 2.27 2.17 1.98 1.91 1.84 1.81 1.77 - 2.73 2.55 2.21 2.12 2.03 1.85 1.79 1.77 1.79 – – 2.45 2.11 2.04 1.89 1.83 1.79 1.77 1.74 – – 2.57 2.12 2.03 1.89 1.82 1.76 1.71 1.68 – – – 2.19 1.92 1.86 1.77 1.73 1.68 1.62 – – – – 1.87 1.87 1.79 1.69 1.63 1.61 – – – – – 1.79 1.57 1.62 1.59 1.48 – – – – – – 1.64 1.65 1.63 1.51 Condition 1: Measured median acceleration rates (ft/s 2 ) for passenger cars (free-flo w and free-merge) 30 23 35 27 3.39 40 31 3.28 2.88 45 35 3.22 2.97 50 39 3.24 2.90 2.69 55 43 3.40 2.72 2.68 2.47 60 47 2.68 2.66 2.36 1.97 65 50 3.18 2.57 2.36 1.96 70 53 2.62 2.49 2.05 75 55 2.76 2.60 2.12 Condition 2: Measured 15th percentile acceleration rates (ft/ s 2 ) for passenger cars (free-flo w and free-merge) 30 23 35 27 1.33 40 31 1.60 1.54 45 35 1.79 1.51 50 39 1.92 1.65 1.53 55 43 1.97 1.55 1.43 1.25 60 47 1.58 1.45 1.22 1.19 65 50 1.55 1.47 1.28 1.19 70 53 1.48 1.32 1.21 75 55 1.50 1.43 1.26 Condition 3: Measured median acceleration rates (ft/s 2 ) for trucks (free-fl ow and free-merge) 30 23 3.90 35 27 4.14 2.27 40 31 2.00 1.66 45 35 1.97 1.75 50 39 1.69 1.79 55 43 1.60 1.62 1.41 60 47 1.58 1.33 1.48 65 50 70 53 75 55 Condition 4: Measured median acceleration rates (ft/s 2 ) for passenger cars (free-flow /platoon and constrained merge) 30 23 35 27 40 31 45 35 50 39 1.76 55 43 1.60 1.62 60 47 1.70 1.76 65 50 1.84 1.87 70 53 75 55 Table 26. Comparison of Green Book and field acceleration rates on level grades.

69 not include deceleration were included in the analysis. For example, if a vehicle’s initial speed was 43 mi/h, but then fell to 41 mi/h, this section of the speed profile was disregarded. A distribution of all of the individual vehicle accelerations for a given cell in the table was plotted in histogram form. These histograms are presented in Appendix B (available on TRB web site at http://www.trb.com/Main/Blurbs/167516.aspx) and show the number of observations for each cell as well as sum- mary statistics. The histograms indicate that the acceleration values are skewed to the left, with long tails toward the higher accelerations. For this reason, median, rather than mean, accelerations provide a better estimate for evaluation. Only values based upon 20 or more observations are presented in Table 26. Table 26 shows that in ideal conditions (i.e., where the merg- ing vehicle is not in a platoon and the freeway traffic is mov- ing at a speed greater than 50 mi/h), passenger cars can and do accelerate at rates greater than those assumed in the Green Book. The Condition 1 acceleration values exceed the corre- sponding Green Book acceleration values where measurements are available. Green Book accelerations generally decrease mov- ing to the right and downward through the table. Condi- tion 1 accelerations also decrease moving to the right through the table, but this pattern is not as clear moving downward. Condition 2 accelerations are considerably lower than those shown in Condition 1, and are only slightly below those assumed in the Green Book. This indicates that even the 15 per- cent of vehicles with the lowest rates of acceleration are accel- erating near the assumed values. It should also be noted that drivers may have many reasons for not accelerating at a higher rate, and that the vehicles with the least acceleration may reflect driver preference rather than capabilities. For example, a driver may choose to accelerate at a lower rate when more distance is available to increase speed (on longer ramps) or when freeway traffic is light and merge speed is less critical. Condition 3 considers acceleration rates for trucks. As expected, these rates are lower than those exhibited by pas- senger cars, although they are quite close to the 15th percentile values for passenger cars, and in general, are about 10 percent less than the Green Book assumed accelerations. Condition 4 considers passenger cars in constrained-merge conditions. Both free-flow and platooned vehicles are consid- ered because constrained conditions often occur during peak travel times when ramp traffic is heavy. For this reason, very few vehicles were not in a platoon and the number of obser- vations for these cars was quite limited. Even in constrained- merge conditions, vehicles accelerate at rates very close to those assumed in the Green Book. Comparison of Acceleration Rates on Grades of 3 Per- cent or Greater. Where grades are present on ramps, the minimum SCL lengths listed in Green Book Exhibit 10-70 are adjusted as a function of grade using factors listed in Green Book Exhibit 10-71 (see Table 2). The adjustment factors are provided as the ratio of length on grade to length on level. Adjustment factors are provided for grades of 3 to 4 percent and 5 to 6 percent and range in value from 1.3 to 3.0 for upgrades. Table 27 provides a comparison of measured acceleration rates for free-flow passenger cars on ramps with level grade and ramps with grades of 3 percent or greater. Median accel- eration rates are provided, and corresponding adjustment factors are calculated as the ratio of the median acceleration rate on level grade to that for the respective categories of grades consistent with the current format provided in Green Book Exhibit 10-71 for upgrades. Only values based upon 10 or more observations are presented for grades of 3 percent or more. None of the entrance ramps included in the study had downgrades greater than 2 percent, so insufficient data were available to further investigate acceleration rates on down- grades greater than 2 percent. Table 28 provides similar comparisons of measured accel- eration rates for free-flow trucks. Rather than calculating the adjustment factors based on the median acceleration rate on level grade for trucks, the adjustment factors are based on the median acceleration rates on level grade for passenger cars. Again, only values based upon 10 or more observations are presented for grades of 3 percent or more; as such, insuf- ficient data were available for acceleration rates of trucks on upgrades greater than 4 percent and on downgrades greater than 2 percent. Comparing grade adjustment factors from Green Book Exhibit 10-71 to those provided in Tables 27 and 28 yields the following general observations: • Many of the adjustment factors for passenger cars on upgrades of 3 to 4 percent are less than one, indicating that passenger cars can and do accelerate at a greater rate on upgrades of 3 to 4 percent than on level grade. This indicates that upgrades of 3 to 4 percent do not limit the vehicle performance capabilities of passenger cars. In addition, the calculated adjustment factors that are greater than one for passenger cars on upgrades of 3 to 4 percent are much less than the corresponding values in Green Book Exhibit 10-71. • The calculated adjustment factors for passenger cars on upgrades of 6 percent are all greater than one. Several of the values are comparable to corresponding values in Green Book Exhibit 10-71, while several others are considerably less than values in Green Book Exhibit 10-71. • Most of the calculated adjustment factors for trucks on upgrades of 3 to 4 percent are comparable to correspond- ing values in Green Book Exhibit 10-71, or slightly greater, with exception of one value which is considerably less than the corresponding value in Green Book Exhibit 10-71. However, this needs to be carefully interpreted because the acceleration rates of trucks on level grade are less than for

70 Design speed (mi/h) Speed reached (mi/h) Acceleration length, L (ft) for entrance curve design speed (mi/h) Stop 15 20 25 30 35 40 45 50 Initial speed (mi/h) 0 14 18 22 26 30 36 40 44 Measured median acceleration rates (ft/s2) for passenger cars (free-flow and free-merge) (0 to 2 % grade) 30 35 40 45 50 55 60 65 70 75 23 27 31 35 39 43 47 50 53 55 3.39 3.28 3.22 3.24 3.40 2.88 2.97 2.90 2.72 2.68 3.18 2.69 2.68 2.66 2.57 2.62 2.76 2.47 2.36 2.36 2.49 2.60 1.97 1.96 2.05 2.12 Measured median acceleration rates (ft/s2) for passenger cars (free-flow and free-merge) (3 to 4 % grade) 30 23 2.46 2.83 2.94 35 27 2.53 3.07 3.12 3.16 40 31 2.48 3.07 3.18 3.16 3.17 45 35 2.31 3.01 3.11 3.12 3.06 50 39 2.41 2.97 3.08 3.04 3.00 2.78 55 43 2.94 3.09 3.01 2.96 2.81 2.57 60 47 3.00 3.10 3.02 2.99 2.89 2.83 2.48 65 50 3.11 3.18 3.15 3.11 2.91 2.83 2.67 70 53 3.10 3.27 3.28 3.21 3.07 2.91 2.80 75 55 3.29 3.37 3.38 3.36 3.37 3.21 3.00 Measured median acceleration rates (ft/s2) for passenger cars (free-flow and free-merge) (6 % grade) 30 23 35 27 2.14 40 31 2.60 45 35 2.49 50 39 1.57 55 43 1.89 60 47 1.69 65 50 1.76 70 53 75 55 SCL adjustment factor as a function of grade (3 to 4% grade) 30 23 35 27 1.07 40 31 1.04 0.91 45 35 1.03 0.97 50 39 1.07 0.97 0.97 55 43 1.13 0.92 0.95 0.96 60 47 0.89 0.92 0.83 0.79 65 50 1.02 0.88 0.83 0.73 70 53 0.85 0.85 0.73 75 55 0.82 0.81 0.71 SCL adjustment factor as a function of grade (6% grade) 30 23 35 27 1.58 40 31 1.11 45 35 1.19 50 39 1.71 55 43 1.30 60 47 1.16 65 50 1.11 70 53 75 55 Table 27. Measured acceleration rates for passenger cars on grades and corresponding adjustment factors.

71 passenger cars. Thus, even though the grade adjustment factors based on acceleration capabilities of trucks are com- parable to values in the Green Book, this does not neces- sarily mean that by applying the grade adjustment factors for upgrades that the corresponding minimum acceleration lane lengths would accommodate the vehicle performance capabilities of trucks. Comparison of Truck Acceleration Rates on Grades Using the TSPM. A truck speed profile model (TSPM) that esti- mates truck performance on grades was developed by Harwood et al. (2003) primarily as a design tool for highway agencies to determine the need for, and to design, truck climbing lanes. The TSPM is a spreadsheet-based tool that applies vehicle performance equations for trucks to estimate truck speed profiles on specified grades. Inputs for the TSPM include both roadway and truck characteristics as follows: Roadway Characteristics • Vertical profile—percent grade for specific ranges of posi- tion coordinates. • Elevation above sea level (ft). Truck Characteristics • Desired speed (mi/h). • Initial speed of truck at beginning of analysis section (mi/h). • Weight-to-power ratio (lb/hp). • Weight-to-front-area ratio (lb/ft2). Measured speed profiles for vehicles traveling on two of the entrance ramps (i.e., I-376/Ardmore Blvd with a 4 per- cent upgrade and I-376/Wm. Penn Hwy with a 6 percent upgrade) were compared to speed profiles calculated using the TSPM. The vertical alignment of each ramp was input into the TSPM along with the initial speed at the painted nose and the desired speed (i.e., final speed) at the merge location Design speed (mi/h) Speed reached (mi/h) Acceleration length, L (ft) for entrance curve design speed (mi/h) Stop 15 20 25 30 35 40 45 50 Initial speed (mi/h) 0 14 18 22 26 30 36 40 44 Measured median acceleration rates (ft/s2) for trucks (free-flow and free-merge) (0 to 2 % grade) 30 35 40 45 50 55 60 65 70 75 23 27 31 35 39 43 47 50 53 55 3.90 4.14 2.27 2.00 1.97 1.66 1.75 1.69 1.60 1.79 1.62 1.58 1.41 1.33 Measured median acceleration rates (ft/s2) for trucks (free-flow and free-merge) (3 to 4 % grade) 30 23 2.22 35 27 1.99 2.02 40 31 2.17 1.98 45 35 1.80 50 39 1.45 55 43 2.22 60 47 65 50 70 53 75 55 SCL adjustment factor as a function of grade (3 to 4 % grade) 30 23 35 27 1.68 40 31 1.51 1.45 45 35 1.65 50 39 1.86 55 43 1.11 60 47 65 50 70 53 75 55 Table 28. Measured acceleration rates for trucks on grades and corresponding adjustment factors.

72 as measured in the field. Then a weight-to-power ratio was input and adjusted to obtain a profile that matched the speed profile measured in the field as closely as possible. Table 29 provides summary statistics of the number and types of vehicles for which speed profiles were compared. In total, the speed profiles of 127 vehicles classified as trucks were compared, including SU trucks, semitrailers, and buses. In many of the cases, the speed profiles from the field closely matched the profiles from the TSPM; however, in several cases a good match could not be found. Figure 34 illustrates a speed profile that was considered a good match to the TSPM data, while Figure 35 illustrates a speed profile that was con- sidered a bad match. Of the 127 vehicles, good matches were found for 105 of the speed profiles. Figures 36 and 37 illustrate the distribution of weight-to- power ratios for those speed profiles that closely matched the TSPM data. The upper portion of Figure 36 shows the assumed weight-to-power ratios based on TSPM estimates for 60 free- flow trucks, and the bottom portion shows the assumed weight-to-power ratios for 45 trucks under platoon conditions. Similarly, the upper portion of Figure 37 shows the assumed weight-to-power ratios for 19 free-flow semitrailers, and the bottom portion shows the assumed weight-to-power ratios for 11 semitrailers under platoon conditions. Based upon the TSPM estimates, very few of the trucks had weight-to-power ratios greater than 140 lb/hp. Only two of the trucks had a weight-to- power ratio of 200 lb/hp, considered to be the weight-to-power ratio for a typical heavy truck for design of upgrades. After efforts to calibrate the TSPM to closely match the speed- distance curves for acceleration/deceleration of a typical heavy truck of 200 lb/hp on upgrades and downgrades as illustrated in Green Book Exhibits 3-55 and 3-56, speed-distance curves were generated for four weight-to-power ratios of 140, 160, 180, and 200 lb/hp using the TSPM (Figures 38 through 41). A weight- to-power ratio of 140 lb/hp was selected for developing speed- distance curves based upon the distribution of weight-to-power ratios observed in the field at the two locations with grades of 4 percent and 6 percent. The other speed-distance curves for 160 and 180 lb/hp were developed as possible design tools if a lesser weight-to-power ratio is selected for design purposes than the typical 200-lb/hp design vehicle. Selecting a lower weight- to-power ratio for design may be desirable in some instances as a compromise to better accommodate trucks. Even though speed-distance curves for trucks with weight- to-power ratios as low as 140 lb/hp are provided, it should be noted that there are very few instances where a 140-lb/ hp truck can meet the design conditions for minimum accel- eration lengths. On a 0 percent grade, a 140-lb/hp truck can accelerate from 14 to 23 mi/h within 140 ft and from 36 to 39 mi/h within 130 ft. For the other design conditions with 0 grade, and for all upgrades, a 140-lb/hp truck cannot meet the design conditions as specified in the Green Book, even taking into consideration the grade adjustment factors for upgrades. Summary of Findings from Examination of Acceleration Rates. The most significant findings from the examination of acceleration rates at entrance ramps are as follows: • Accelerations of merging vehicles along the ramp and SCL are not constant. Total number of vehicles 127 Total number of free-flow vehicles 75 Total number of platooned vehicles 52 Total number of semitrailers 37 Total number of free-flow semitrailers 24 Total number of platooned semitrailers 13 Total number of free-flow vehicles (matched) 60 Total number of platooned vehicles (matched) 45 Total number of free-flow semitrailers (matched) 19 Total number of platooned semitrailers (matched) 11 Total number of vehicles under free-merge condition 109 Total number of vehicles under free-merge condition (matched) 102 Total number of vehicles under constrained-merge condition 2 Total number of vehicles under constrained-merge condition (matched) 0 Total number of vehicles under forced-merge condition 15 Total number of vehicles under forced-merge condition (matched) 3 Table 29. Summary statistics of truck speed profiles compared with TSPM.

73 • Merging vehicles sometimes reach a peak speed along the ramp and/or SCL and decelerate to slower speeds prior to merging onto the freeway. This behavior is most evident under constrained and forced-merge conditions, when freeway speeds are reduced. • In many cases, vehicles are exiting the controlling feature along an entrance ramp at speeds greater than assumed in the Green Book. • For free-flow passenger cars under free-merge conditions, the measured average speed at the merge location tends to be a few miles per hour less than the desired speed reached assumed in the Green Book when the ramp length is less than the recommended, and a few miles per hour greater than the Green Book desired speed reached when the ramp length is longer than the recommended. Vehicle Speed Profile From Painted Nose to Merge Completion 40 41 42 43 44 45 46 47 48 49 50 0 50 100 150 200 250 300 S pe ed (m ph ) Distance From Painted Nose (ft) Calculated Measured Figure 34. Good match between field and TSPM data. • As expected, acceleration rates of free-flow trucks are lower than for free-flow passenger cars. • The median acceleration rates for free-flow passenger cars under free-merge conditions are greater than the assumed acceleration rates in the Green Book. Based on the median acceleration rates of free-flow passenger cars measured under free-merge conditions, minimum acceleration lane lengths could be reduced between 16 to 46 percent and still be sufficient for the median vehicle. • The 15th percentile acceleration rates of free-flow pas- senger cars under free-merge conditions are less than the assumed acceleration rates in the Green Book. • The median acceleration rates for free-flow trucks under free-merge conditions are less than the assumed accelera- tion rates in the Green Book.

74 Vehicle Speed Profile From Painted Nose to Merge Completion 20 25 30 35 40 45 50 0 50 100 150 200 250 300 350 400 S pe ed (m ph ) Distance From Painted Nose (ft) Calculated Measured Figure 35. Bad match between field and TSPM data. • Under constrained conditions, acceleration rates of pas- senger cars are very close to the assumed acceleration rates in the Green Book. • The Green Book generally assumes acceleration rates decrease as initial speeds increase and also decrease as the final speed reached increases. In other words, considering Exhibit 10-70, acceleration rates decrease across rows and down the columns. For free-flow passenger cars under free- merge conditions, the pattern of acceleration rates decreas- ing as initial speeds increase (i.e., decreasing acceleration rates moving from right to left across rows) was observed, but the pattern of acceleration rates decreasing as speed reached increased (i.e., decreasing acceleration rates from top to bottom within columns) was not observed. • Upgrades of 3 to 4 percent do not limit the acceleration capabilities of passenger cars. This is illustrated based upon the calculated adjustment factors having values less than one. • For upgrades of 6 percent, measured acceleration rates of passenger cars are comparable to assumptions in Green Book Exhibit 10-71, or in some cases are slightly greater. • With very few exceptions, a truck with a weight-to-power ratio as low as 140 lb/hp cannot meet the current design conditions in the Green Book for minimum acceleration lane lengths for level or upgrades. 5.4.1.4 Critical or Unusual Maneuvers While reviewing the video recordings, very few critical or unusual maneuvers were observed in the vicinity of the entrance ramps. All of the critical maneuvers that were observed occurred at the same ramp during forced-merge conditions.

0 20 40 60 80 100 120 140 160 180 200 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 45 47 49 51 53 55 57 59 W ei gh t/P ow er R at io (lb /hp ) Truck # Weight/HP ratio (all trucks: free-flow) 0 20 40 60 80 100 120 140 160 180 200 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 W ei gh t/P ow er R at io (lb /hp ) Truck # Weight/HP ratio (all trucks: platoon) Figure 36. Weight-to-power ratio of all trucks (free-flow and platoon). 0 20 40 60 80 100 120 140 160 180 200 1 3 5 7 2 4 6 8 9 10 11 12 13 14 15 16 17 18 19 W ei gh t/P ow er R at io (lb /hp ) Truck # Weight/HP ratio (semis only: free-flow) Truck # 1 3 5 7 2 4 6 8 9 10 11 0 20 40 60 80 100 120 140 160 180 200 W ei gh t/P ow er R at io (lb /hp ) Weight/HP ratio (semis only: platoon) Figure 37. Weight-to-power ratio of all semitrailers (free-flow and platoon).

76 Acceleration on Upgrades and Downgrades 0 10 20 30 40 50 60 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 Sp ee d (m ph ) Distance (ft) 8% -5% -4% -3% -2% -1% 0% 1% 2% 3% 4% 5% 6% 7% Deceleration on Upgrades 0 10 20 30 40 50 60 70 80 0 5,000 10,000 15,000 20,000 25,000 Sp ee d (m ph ) Distance (ft) 9% 0%; 1 % 2% 3% 4% 5% 6% 7% 8% Figure 38. Speed-distance curves for acceleration/deceleration of a 140 lb/hp truck using the TSPM.

77 Acceleration on Upgrades and Downgrades Deceleration on Upgrades 9% 1% 2% 3% 4% 5% 6% 7% 8% 0% 0 10 20 30 40 50 60 70 80 0 5,000 10,000 15,000 20,000 25,000 Sp ee d (m ph ) Distance (ft) 8% -5% -4% -3% -2% -1% 0% 1% 2% 3% 4% 5% 6% 7% 0 10 20 30 40 50 60 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 Sp ee d (m ph ) Distance (ft) Figure 39. Speed-distance curves for acceleration/deceleration of a 160 lb/hp truck using the TSPM.

78 Acceleration on Upgrades and Downgrades Deceleration on Upgrades 8% -5% -4% -3% -2% -1% 0% 1% 2% 3% 4% 5% 6% 7% 0 10 20 30 40 50 60 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 16,000 Sp ee d (m ph ) Distance (ft) 9% 1% 2% 3% 4% 5% 6% 7% 8% 0% 0 10 20 30 40 50 60 70 80 0 5,000 10,000 15,000 20,000 25,000 Sp ee d (m ph ) Distance (ft) Figure 40. Speed-distance curves for acceleration/deceleration of a 180 lb/hp truck using the TSPM.

79 -5% -4 % -3% -2% -1% 0% 1% 2% 3% 4% 5% 6% 7% 8% 0 10 20 30 40 50 60 0 2 ,0 00 4, 000 6, 000 8, 00 0 1 0, 000 12, 000 14, 000 16 ,0 00 Sp ee d (m ph ) Di st an ce ( ft) Acceleration on Upgrades and Downgrades 0 10 20 30 40 50 60 70 80 0 5 ,0 00 10 ,0 00 15, 000 20, 000 25 ,0 00 Sp ee d (m ph ) Di st an ce ( ft) Deceleration on Upgrades 9% 1% 2% 3% 4% 5% 6% 7% 8% 0% Figure 41. Speed-distance curves for acceleration/deceleration of a 200 lb/hp truck using the TSPM.

80 Four instances were observed in which a following (i.e., pla- tooned) vehicle in the SCL passed the leading vehicle in the SCL using the right shoulder. In another instance, a following vehicle in the right lane of the freeway changed lanes into the SCL to pass slower vehicles in the freeway lanes. 5.4.2 Exit Ramps The purpose of an exit ramp is to provide an appropriate means of accessing the adjacent surface streets from a location on a given freeway; the two primary components of exit ramp design that accomplish this purpose are (1) providing sufficient area for vehicles to depart from the main lanes of the freeway and (2) providing sufficient distance for vehicles to decelerate at a comfortable rate from freeway speeds to a speed appropriate for the crossroad terminal or controlling horizontal alignment feature. Similar to the characteristics of an entrance ramp, the distance needed for a vehicle to reach speeds appropriate for the crossroad or controlling feature is a function of the vehicle’s deceleration rate along the SCL and the ramp. Ramp design guidelines that assume a deceleration rate greater than what is comfortable for the driver or physically possible for the vehicle may provide lengths that are too short. If the SCL and ramp (the full area over which a vehicle may diverge from the freeway and decelerate before reaching the crossroad or controlling feature) are too short, this may be evident by observations of vehicles diverging upstream of the taper area or by decelerating in the freeway mainlane before entering the SCL. Both of these mea- sures are considered in this analysis. In addition, an analysis is performed to consider the deceleration profiles of the observed vehicles to determine whether the current vehicle fleet is decel- erating as expected based on current ramp design guidelines described in the AASHTO Green Book. In summary, under ideal operating conditions along an exit ramp, vehicles begin decelerating after exiting the free- way onto the deceleration lane, and vehicles decelerate along the deceleration lane to an appropriate speed before reaching the controlling feature of the ramp, which could either be a controlling horizontal curve or a crossroad terminal. The greatest concern arises when vehicles enter the con- trolling feature of the ramp at too high of speeds. Vehicles decelerating in the freeway lanes prior to diverging is of lesser concern. The analyses were designed to compare actual driv- ing practices to these ideal operating conditions and identify areas of concern. 5.4.2.1 Diverge Location For an exit ramp, vehicles will ideally diverge from the free- way at a point somewhere along the SCL—that is, beyond the taper, but prior to the painted nose. Similar to the analysis of merge points for vehicles entering the freeway, this analy- sis divided the possible diverge locations into five categories: upstream of or within the taper (taper), in the first third of the SCL (early), in the middle third of the SCL (mid), in the last third of the SCL (late), and beyond the painted nose (gore). However, in this study, only 10 vehicles observed (less than 0.5 percent of the total vehicles observed) began their diverge maneuvers in the late or gore areas, so the results pre- sented here will be confined to the taper, early, and mid areas. The taper area is of particular interest because a driver beginning the diverge maneuver either before or within the taper suggests that one of two scenarios is likely: 1. The driver believes that the provided SCL and ramp length are not sufficient to comfortably decelerate to the appro- priate speed for the crossroad or controlling feature. 2. The driver wants to decelerate at a lower, more casual rate than that associated with the ramp design and decides to utilize the extra distance in the taper to accomplish this. The second of these two scenarios will be discussed in more detail as the results from the analysis are presented. Figure 42 presents the diverge location by vehicle type (i.e., passenger cars versus trucks) on all of the study ramps, based on the laser data, broken out by freeway speeds. The top section shows diverge locations for all freeway speed categories com- bined, the next section shows diverge locations during free- diverge conditions on the freeway (i.e., freeway speeds greater than 50 mi/h), and the bottom section shows diverge locations during constrained-diverge conditions on the freeway (i.e., freeway speeds between 40 and 50 mi/h). Of the 2,584 vehi- cles observed, 86 percent exited the freeway in the taper area, 12 percent exited in the first third of the SCL, and 2 percent exited in the middle third of the SCL. In addition, 97 percent of vehicles exited the freeway under free-diverge conditions, while the remaining 3 percent completed their diverge maneuvers under constrained-diverge conditions. Given that no vehicles diverged under forced-diverge conditions, this category is also omitted from the figures in this analysis. Figure 42 shows that 86 percent of both passenger cars and heavy trucks exited the freeway in the taper area, 12 to 13 percent of both vehicle types exited early, and 1 to 2 percent exited in the middle. In general, the distribution of diverge location is very similar for both vehicle types. Figure 42 also shows that the proportion of vehicles diverging in the first third of the SCL increased under constrained-diverge condi- tions. This is an intuitive finding because the decreased speeds under constrained-diverge conditions reduce the speed dif- ferential between the freeway and the controlling feature of the ramp, which in turn reduces the deceleration rate needed to safely and comfortably travel through the ramp. Under constrained-diverge conditions, some drivers may feel com- fortable diverging as early as possible in the taper to enter an

81 Figure 42. Diverge location—all ramps by vehicle type.

82 uncongested ramp, but the lower initial speed at the diverge point allows them to exit the freeway at a more casual deceler- ation rate. It is also important to note that because such a high proportion of the observed vehicles exited under free-diverge conditions within the taper, analysis of the entire population of observed vehicles is heavily influenced by this block. Figure 43 shows the distribution of diverge location by ramp type. Regardless of freeway speed condition, over 90 percent of vehicles left the freeway in the taper portion of straight ramps. On loop ramps, about 80 percent of vehicles left the freeway in the taper portion of the ramp. The most noticeable change in the distribution of diverge locations occurred at loop ramps between free- and constrained-diverge conditions. Under free- diverge conditions, 81 percent of the vehicles diverged in the taper area and 16 percent diverged early in SCL, while under constrained-diverge conditions, only 42 percent of the vehicles diverged in the taper area and 55 percent diverged early in the SCL. This could be the result of an effect from the ramp geom- etry; a loop ramp typically has a shorter distance to the con- trolling feature so during free-diverge conditions when freeway speeds and diverge speeds are higher, drivers diverge earlier to use all available deceleration distance, including the taper, to achieve the appropriate operating speed to negotiate the curve on the ramp. Figure 44 shows the distribution of diverge location by diverge type. Approximately two-thirds of the 1,106 observed vehicles on parallel exit ramps diverged in the taper area, while 99 percent of the 1,478 vehicles on taper exit ramps did the same. This is an intuitive finding because the SCL on a tapered deceleration lane is typically much shorter than that of a parallel deceleration lane, so it is reasonable that most drivers on a taper exit ramp would diverge within the taper. However, there is a marked difference between dis- tributions under free-diverge conditions and constrained- diverge conditions. The proportion of vehicles exiting on the taper is similar for free- and constrained-diverge conditions on tapered SCLs, but it is substantially lower for constrained- diverge conditions than for free-diverge conditions on paral- lel SCLs. Similar to the findings discussed above, this may be in part because vehicles do not need as much deceleration distance when traveling at lower freeway speeds prior to their exit maneuver, and may therefore exit within the longer SCL provided at parallel SCLs. Figure 45 summarizes the data for diverge location based on whether a vehicle began a free-flow diverge or was within a platoon. The figure shows that about 91 percent of pla- tooned vehicles diverged from the freeway within the taper, compared to 85 percent of free-flowing vehicles. The differ- ence between free-flow and platooned vehicles was more pro- nounced under constrained-diverge conditions, where only about two-thirds of free-flow vehicles used the taper area, compared to 91 percent of platooned vehicles. The most significant findings from the examination of diverge locations are as follows: • Over 99.5 percent of diverge maneuvers observed began either before or within the taper or within the first or mid- dle thirds of the SCL. • Vehicles tend to diverge later under constrained-diverge conditions than under free-diverge conditions. • The distribution of diverge locations for trucks is very sim- ilar to the distribution of diverge locations for passenger cars for free-diverge conditions. • A higher percentage of vehicles diverge early on straight ramps compared to loop ramps. • More vehicles diverge later at parallel SCLs than at tapered SCLs, particularly under constrained-diverge conditions. • A greater proportion of platooned vehicles began their diverge maneuver in the taper area than free-flow vehicles, particularly under constrained-diverge conditions. 5.4.2.2 Diverge Speed The data presented in this section compares the speed of diverging vehicles at the time they begin to diverge to the speed of freeway traffic in the rightmost lane. Specifically, this is a comparison of initial diverge speeds of vehicles based upon the laser gun data to the average 15-minute freeway speed of the rightmost lane of the freeway at the time of the diverge. The Green Book describes the conditions for which an exit ramp should be designed as follows: Vehicles should decelerate after clearing the through-traffic lane and before reaching the point limiting design speed for the ramp proper. The length available for deceleration may be assumed to extend from a point where the right edge of the tapered wedge is about 3.6 m [12 ft] from the right edge of the right through lane, to the point of initial curvature of the exit ramp (i.e., the first horizontal curve on the ramp). The length provided between these points should be at least as great as the distance needed to accomplish the appropriate deceleration, which is governed by the speed of traffic on the through lane and the speed to be attained on the ramp. Deceleration may end in a complete stop, as at a crossroad terminal for a diamond inter- change, or the critical speed may be governed by the curvature of the ramp roadway. Researchers analyzed the speeds of observed exiting vehicles and compared them with the prevailing speeds on the freeway at the time of their exit maneuvers, calculating the difference between the speed of an exiting vehicle at the point of divergence and the average speed in the right lane of the freeway during the 15-minute period when the exit maneuver took place. Table 30 summarizes the data on mean speed differential by ramp and diverge type, vehicle type, and diverge condition. The mean value is the average of the

83 Figure 43. Diverge location—by ramp type.

84 Figure 44. Diverge location—by diverge type.

85 Figure 45. Diverge location—by vehicle position.

86 individual ramp averages. Also provided are the standard deviation (pooled across individual ramps), minimum, and maximum values measured for each condition. A nega- tive mean speed indicates that the freeway speed is greater than the diverge speed. Table 30 illustrates that, in general, the speeds of diverg- ing vehicles were below the average speed on the freeway’s right lane. Typically, the difference was between 4 and 7 mi/h, though it was as much as 10.7 mi/h for heavy trucks on loop ramps with taper departures. Similarly, the mean speed dif- ferential for trucks is a lower value (i.e., greater negative dif- ference in speed) than the corresponding value for passenger cars in every category. Only under constrained-diverge con- ditions was the average speed of exiting vehicles higher than the average speed of freeway vehicles. The maximum differ- ential for trucks was lower than that for cars in every category, though the difference between minimum and maximum val- ues was typically larger for cars. To further evaluate the effect of different ramp char- acteristics on the difference between diverge speed and freeway speed (speed differential), a mixed effects ANOVA model was developed. The model included main effects for ramp type (straight versus loop) and diverge type (parallel versus taper). The interaction effect between the two was not included based on comparison of the differences of group means, and it would not have been estimable given the number of ramps in the study. The model also includes a random ramp effect to account for individual ramp-to- ramp variability and the fact that repeated measurements were taken at each ramp. The between-ramp variability was assumed to be equivalent for ramp types and diverge types based on comparison of the variability in ramp means. The model was estimated using REML, and estimates of the differences between ramp types and diverge types were obtained using least squares means. Statistical degrees of freedom for evaluation of the parameters was based on the number of ramps, not the number of vehicles. The mean speed differential for each ramp and diverge type combination was evaluated to determine if it was statistically less than 0 mi/h. The results are provided in the last column of Table 30. Due to the minimal number of vehicles and ramp conditions available for the constrained- diverge condition, no statistical analyses were performed for this operational condition. For passenger cars under Type of ramp Type of diverge No. of ramps Vehicle type No. of vehicles Diverge speed minus freeway speed (mi/h) P-value for t-test of mean less than 0 mi/hMean Std dev Min Max All freeway speeds Loop Parallel 2 Truck 46 -7.1 5.1 -18.0 6.1 <0.001 Passenger car 616 -4.2 4.6 -18.1 11.6 <0.001 Taper 3 Truck 50 -10.7 4.4 -24.9 -1.6 <0.001 Passenger car 901 -5.4 4.7 -25.2 13.3 <0.001 Straight Parallel 2 Truck 18 -5.1 3.3 -14.1 -0.4 <0.001 Passenger car 433 -1.6 4.6 -18.9 11.6 0.004 Taper 2 Truck 21 -7.5 6.3 -22.1 3.3 <0.001 Passenger car 523 -4.1 5.1 -20.8 11.3 <0.001 Free diverge Loop Parallel 2 Truck 43 -7.7 4.8 -18.0 1.1 <0.001 Passenger car 600 -4.4 4.4 -18.1 11.6 <0.001 Taper 3 Truck 50 -10.7 4.4 -24.9 -1.6 <0.001 Passenger car 875 -5.7 4.3 -25.2 11.4 <0.001 Straight Parallel 2 Truck 18 -5.1 3.3 -14.1 -0.4 <0.001 Passenger car 433 -1.6 4.6 -18.9 11.6 0.015 Taper 2 Truck 21 -7.5 6.3 -22.1 3.3 <0.001 Passenger car 485 -4.1 5.0 -20.8 11.3 <0.001 Constrained diverge Loop Parallel 1 Truck 3 1.0 4.3 -1.6 6.1 Passenger car 16 1.6 7.4 -15.3 11.5 Taper 1 Truck 0 Passenger car 12 4.6 5.9 -5.3 13.3 Straight Parallel 0 Truck 0 Passenger car 0 Taper 1 Truck 0 Passenger car 38 -5.3 5.4 -17.7 5.8 Table 30. Diverge speed and freeway speed differential by ramp type, diverge type, vehicle type, and speed category.

87 free-diverge conditions, the following results were obtained from the ANOVA model: • The between-ramp standard deviation in the mean diverge speed differential was estimated to be 1.09 mi/h. The with- in-ramp (vehicle-to-vehicle) standard deviation was esti- mated to be 4.47 mi/h. • The mean speed differential for loop ramps of -5.02 mi/h is statistically less than 0 mi/h (p-value < 0.001). For straight ramps, the mean of -2.85 mi/h is also statistically less than 0 mi/h with a p-value = 0.002. • Parallel and taper diverge ramps (both ramp types) have statistically significant differences of -3.03 mi/h and -4.85 mi/h respectively (both p-values <0.001). • Loop ramps tend to have a slightly greater mean diverge speed differential (-2.17 mi/h) as compared to straight ramps. The difference is statistically significant with a p-value of 0.03. • Taper ramps have a slightly greater mean diverge speed dif- ferential than parallel ramps (1.82 mi/h) that is statistically significant (p-value = 0.05). Figure 46 is a graphical representation of the data in Table 30, organized by vehicle type and diverge location. The upper portion of the figure shows the distribution of speed differentials for passenger cars, with the correspond- ing information for heavy trucks shown in the lower portion. As with the data in Figures 42 through 45, the data for pas- senger cars under free-diverge conditions exiting at the taper constitute the largest category of observed vehicles. Figure 46 shows that for passenger cars under free-diverge condi- tions the speed differential increased as the diverge locations approached the painted nose (i.e., vehicles that diverged earlier along the exit diverged closer to freeway speeds than vehicles that diverged later along the deceleration lane). This indicates that passenger cars begin decelerating within the freeway lanes even though they may not necessarily diverge from the freeway lanes at the earliest opportunity. Figure 46 also illustrates that trucks diverged from the freeway at speeds about 2 to 4 mi/h below the speeds of passenger cars, and the variability of speed differentials was less for trucks than for passenger cars. Figures 47 through 50 show subsets of the data presented in Figure 46. Figure 47 presents observations made at loop ramps with parallel SCLs. For this ramp and diverge type, for passen- ger cars under free-diverge conditions the speed differential increased as the diverge locations approached the painted nose. For trucks under free-diverge conditions, the speed differential basically was the same for taper and early diverge locations, and under constrained-diverge conditions, diverge speeds were Figure 46. Diverge speed differential by vehicle type and diverge location for all ramp types.

88 Figure 47. Diverge speed differential by ramp type, diverge type, vehicle type, and diverge location (ramp type: loop; diverge type: parallel). Figure 48. Diverge speed differential by ramp type, diverge type, vehicle type, and diverge location (ramp type: loop; diverge type: taper).

89 Figure 49. Diverge speed differential by ramp type, diverge type, vehicle type, and diverge location (ramp type: straight; diverge type: parallel). Figure 50. Diverge speed differential by ramp type, diverge type, vehicle type, and diverge location (ramp type: straight; diverge type: taper).

90 slightly greater than average freeway speeds for both passenger cars and trucks. Figure 48 presents observations made at loop ramps with tapered SCLs. For this ramp and diverge type, trucks that diverged at the taper diverged at speeds approximately 4 mi/h slower than passenger cars that diverged at the same location. Again under constrained-diverge conditions, diverge speeds were slightly greater than average freeway speeds for passen- ger cars. Figure 49 shows the speed differentials for straight ramps with parallel SCLs. For this ramp and diverge type, observations were made only under free-diverge conditions. Again for pas- senger cars under free-diverge conditions the speed differential increased as the diverge locations approached the painted nose. Figure 50 shows the speed differentials for straight ramps with tapered SCLs. For this ramp and diverge type, there was a greater speed differential for passenger cars that diverged at the taper during constrained-diverge conditions than during free-diverge conditions. Figure 51 shows the speed differentials for each of the four ramp type and merge type combinations for only the free- diverge condition. Again, the trend that is most evident across all combinations is that the speed differential increased as the diverge locations approached the painted nose. The most notable findings from the examination of diverge speeds compared to freeway speeds are as follows: • With few exceptions, the speeds of diverging vehicles were between 4 and 7 mi/h below the average speed on the free- way’s right lane. For every category, the mean speed of trucks was lower than that of passenger cars. These results indicate that observed drivers are completing some degree of their deceleration within the freeway lane, and the finding is consistent with previous findings that trucks frequently accomplish more of their deceleration in the freeway lane. • Speed differential increased as the diverge locations approached the painted nose. • Speeds of diverging vehicles under constrained-diverge conditions were closer to freeway speeds than under free- diverge conditions. 5.4.2.3 Deceleration Rate The third measure of performance considered is the decel- eration rate of vehicles as they travel along the SCL and ramp after the diverge point. As with SCLs on entrance ramps, the AASHTO Green Book does not provide a table of the assumed deceleration rates used to determine guidance for minimum Figure 51. Diverge speed differential by ramp type, diverge type, vehicle type, and diverge location (for free-diverge conditions only).

91 deceleration lane length. Section 3 of this report provides two methods for back-calculating deceleration rates based on Green Book Exhibit 10-73. One approach is based on a two-step process of deceleration, which includes decelera- tion during coasting and deceleration during braking, and a second approach is based on a constant deceleration over the entire deceleration lane length. The deceleration rates mea- sured in the field can be compared to the Green Book rates based on these two differing design conditions to determine whether vehicles are performing as assumed in the Green Book. Such a comparison, however, is not a truly direct com- parison, because vehicle deceleration in the field is influenced by the same factors as described for acceleration rates in Sec- tion 5.4.1.3. For example, ramps longer than the minimum length needed to reach the desired speed at a controlling fea- ture using a comfortable deceleration allow the driver more flexibility in determining when and where to diverge and decelerate along the ramp. When ramps are long, some driv- ers maintain speeds closer to the prevailing freeway speed in or adjacent to a portion of the SCL and then decelerate more quickly as they enter the SCL and approach the controlling fea- ture. Alternately, some drivers begin their deceleration earlier, but at a lower, more casual rate, and maintain that initial rate throughout the ramp. Therefore, decelerations measured at a given point on the SCL or the ramp may not be representative of vehicle capabilities or driver comfort levels and preferences, because vehicles typically do not decelerate along the ramp at a constant rate. Because it was not possible to capture speed and deceleration along the entire ramp length, the behavior of vehicles near the crossroad terminal is not well documented here. Field-measured deceleration rates that are less than the values calculated from the Green Book may indicate that driv- ers were completing much of the required deceleration in the freeway lane upstream of the beginning of the taper and prior to the location where initial speed measurements were taken, contrary to assuming that the deceleration rates assumed in the Green Book are too high. For this reason, comparisons of field-measured deceleration to Green Book assumptions must be interpreted carefully and considered in context of the ramp characteristics and diverge conditions. The analysis of deceleration rates includes: • A general analysis of deceleration rates. • Comparison of deceleration rates measured in the field to deceleration rates from the Green Book criteria for level grades, assuming a constant deceleration. • Comparison of deceleration rates measured in the field to deceleration rates from the Green Book criteria for level grades, assuming a two-step deceleration process. The focus of this analysis is comparing deceleration rates measured in the field to the assumed deceleration rates during primary braking (i.e., the second portion of the two-step process). • Comparison of deceleration rates on grades of 3 percent or greater. The analysis of deceleration rates is based on the measured speed profiles of 2,264 vehicles. General Analysis of Deceleration Rates. Tables 31 through 33 compare ramp length, measured diverge speed, and decel- eration rates to those recommended or assumed in the Green Book. The tables are organized as described below. Rows. Each row provides information about one exit ramp. The ramps are organized first by controlling feature, with those controlled by horizontal alignment in the upper section and those controlled by the crossroad terminal in the lower sec- tion. Within each section, the ramps are organized in order of decreasing difference between actual length of the ramp and the Green Book recommended ramp length. Columns Green Book (GB) Initial Speed—The value listed in the Green Book for the initial speed of an exit ramp. Conceptually this is the diverge speed of exiting vehicles. The average speed of the freeway is provided, rather than the design speed of the freeway. Average Diverge Speed—The average speed at the diverge point for the vehicles observed and recorded at the ramp. Average Diverge Distance—The average distance in feet upstream of the painted nose at which observed vehicles diverged from the freeway at the ramp. Deceleration Length—SCL length plus distance from painted nose to controlling feature. Difference from Green Book Deceleration Length—The differ- ence in feet between the total length provided from the end of the taper to the painted nose (SCL) plus the length from the painted nose to the controlling feature and the mini- mum deceleration lane length provided in Exhibit 10-73 of the Green Book adjusted for grade. The Green Book criteria length is found using the design speed of the freeway and the radius of the controlling feature. For ramps without control- ling horizontal curvature, the stop condition was assumed for the final speed. For example, a ramp on a highway with a design speed of 75 mi/h has an assumed diverge speed of 61 mi/h. For a ramp with a stop condition as its design speed, the Green Book recommends a deceleration length of 660 ft. The positive numbers in this column indicate that all exit ramps studied had ramp lengths longer than recom- mended by the Green Book criteria. Green Book Speed at Controlling Feature—The average run- ning speed listed in Green Book Exhibit 10-73 at the con- trolling feature for a ramp, based upon the design speed of the exiting curve. The horizontal alignment is considered controlling if the radius of the curve is equal to or less than

92 1,000 ft, consistent with current guidance provided in the 2004 Green Book. Average Speed at Green Book Distance—The average speed recorded for observed vehicles at the distance correspond- ing to the Green Book deceleration length: the average observed speed at the location where the vehicle has trav- eled the minimum Green Book distance from the point of diverge. Average Speed at Controlling Feature—The average speed recorded for observed vehicles at the controlling feature on each ramp. At some ramps, researchers were unable to collect speed data at the controlling feature (e.g., the final speeds of all profiles were upstream of the controlling feature); therefore, this cell has no value for those ramps. Average Final Speed—The mean of all observed speeds at the location of the final speed measurement. Assumed Green Book Deceleration Rate—Deceleration in ft/s2 calculated from the assumed Green Book speed at diverge (initial speed), Green Book speed at controlling feature, and minimum deceleration lane length from Exhibit 10-73 in the Green Book. This is an assumed constant deceleration rate over the minimum deceleration lane length recom- mended in the Green Book. Deceleration Rate over Green Book Distance—The 5th, 15th, and 50th percentile observed deceleration rates over the distance equal to that recommended in Green Book Exhibit 10-73 (and adjusted for grade). Taken from the distribution of the deceleration of all vehicles calculated from the measured diverge speed, the measured speed at the Green Book distance, and the time taken for each vehicle to travel that distance. The percentile values indi- cate the deceleration rate that is achieved or exceeded by that percentage of vehicles. For example, Table 31 shows that at I-435/Quivira, 5 percent of observed free-flow passenger cars decelerated at a rate greater than 1.83 ft/s2 over the distance equal to the corresponding Green Book minimum length. Deceleration Rate to Controlling Feature or Final Speed—The 5th, 15th, and 50th percentile observed deceleration rates from the point of divergence to the controlling feature, or to final reading when speed was not recorded at controlling feature. Taken from the distribution of the deceleration of all vehicles calculated from the measured diverge speed, the measured speed at the controlling feature (or the final measured speed), and the time taken for each vehicle to travel that distance. Tables 31 through 33 provide details on diverge speeds, diverge location, speeds at controlling feature, final speeds, and deceleration rates based upon observations from the cur- rent vehicle fleet and compared to the corresponding values in the Green Book. Separate tables are provided for free- flow vehicles (Table 31), platooned vehicles (Table 32), and both combined (Table 33) under free-diverge conditions. Because Green Book Exhibit 10-73 is also based strictly on passenger cars, the data in Tables 31 through 33 are for pas- senger cars only. Tables 31 through 33 show several important observations, which are summarized as follows: • The observed speeds at the diverge points are typically at or below those assumed by the Green Book. • The average speed at the diverge point is higher for free-flow vehicles than platooned vehicles on every ramp but one. • The average observed speeds at the controlling features are higher than the assumed Green Book speeds. • For all vehicle movement categories (free-flow, platoon, and combined), deceleration rates in the 5th, 15th, and 50th percentiles are all closer to zero for the longer ramps (those more than 400 ft longer than the Green Book recom- mendation) than for the shorter ramps (those within 100 ft of the Green Book recommendation), which is intuitive because the greater length allows for more casual braking when exiting the freeway. This is illustrated in Figure 52. • For all ramps and vehicle categories, the observed decelera- tion rates over the distance equivalent to the Green Book deceleration length and the rates over the entire observed exit maneuver are less than the constant deceleration rates needed to match Green Book conditions. • At every ramp, for both free-flow and platooned vehicles, the observed deceleration rates to the controlling fea- ture and to the point of final reading in each percentile are closer to zero than the corresponding deceleration rate over the Green Book distance. • There is no consistent trend in the comparison of decelera- tion rates between free-flow and platooned vehicles. In some cases, the deceleration rates for free-flow vehicles are greater, while in other cases, the decelerations rates for platooned vehicles are greater. • On the whole, observed deceleration rates on loop ramps are greater than those on straight ramps, and frequently by a noticeable difference. That difference is most pro- nounced in the 5th percentile rates for all cars and for free- flow cars (see Figure 53). The noticeable exception is I-635/ Metropolitan, which has the highest deceleration rates of all straight ramps in the study. On balance, it should be noted that the loop ramps all have deceleration distances within 100 ft of the Green Book distance, as does the ramp at I-635/Metropolitan, so the effect of deceleration length should not be overlooked when considering influences of straight ramps versus ramps with curves.

Ramp location GB initial speed (mi/h) Average diverge speed (mi/h) Average diverge distance (ft) Decel length 1 (ft) Difference from GB decel (ft) GB speed at controlling feature (mi/h) Average speed at GB dis t 2 (mi/h) Average speed at controlling feature (mi/h) Average final speed (mi/h) GB decel rate (ft/s 2 ) Deceleration rate over Green Book distance 3 (ft/s 2 ) Deceleration rate to controlling feature or final speed 4 (ft/s 2 ) 5 1 5 5 0 5 1 5 5 0 Controlling feature: Horizontal alignment I-70/US-40 5 5 5 1.7 –258 570 100 26 37.3 — 5 34.8 –5.38 –4.94 –4.35 –3.29 – 5.22* –4.62* –3.68* I-435/US-24 61 58.7 –666 5 95 7 5 26 51.0 45.6 38.9 –5.70 –3.12 –2.67 –1.73 –3.50 –3.06 –2.30 I-635/Plano 5 5 5 3.9 –269 425 3 5 36 47.1 39.3 39.2 –4.77 –3.87 –3.11 –1.93 –4.33 –3.80 –2.85 I-635 EB/Freeport 58 55.7 –295 5 20 30 26 47.4 35.5 38.7 –5.56 –3.52 –2.87 –1.90 –4.28 –3.83 –2.98 I-635 WB/Freeport 58 57.0 –348 5 00 10 26 49.4 — 5 38.8 –5.56 –3.47 –2.74 –1.67 –4.10* –3.60* –2.91* Controlling feature: Crossroad terminal I-435/Quivira 61 62.2 –732 1640 980 0 58.7 29.8 45.2 –6.06 –1.83 –1.26 –0.72 –2.90 –2.82 –2.55 I-435/Gregory 61 60.4 –644 1 520 782 0 55.9 29.0 29.9 –4.90 –1.77 –1.59 –0.92 –3.02 –2.74 –2.29 I-635/Marsh 55 5 0.6 –208 1035 465 0 46.8 — 5 36.0 –5.71 –2.05 –1.57 –0.83 –2.90* –2.40* –1.67* I-635/Metro 61 60.5 –466 960 69 0 48.2 — 5 34.6 –4.49 –2.88 –2.36 –1.59 –3.49* –3.18* –2.53* Note: GB = Green Book. 1 Deceleration Length = SCL length + distance from painted nose to controlling feature. 2 Observed speed at location where vehicle has traveled the minimum Green Book distance from the point of diverge. 3 Deceleration rate over minimum Green Book distance beginning at point of diverge. 4 Deceleration rate from point of diverge to controlling feature, or to final reading when speed was not recorded at controllin g feature. 5 Researchers were unable to collect speed data at the controlling feature; the final speeds of all profiles were upstream of t he controlling feature. * Value is based on final speed readings. Table 31. Observed deceleration rates for free-flow ramp vehicles (free-diverge conditions and passenger cars only).

Ramp location GB initial speed (mi/h) Average diverge speed (mi/h) Average diverge distance (ft) Decel length1 (ft) Difference from GB decel (ft) GB speed at controlling feature (mi/h) Average speed at GB dist2 (mi/h) Average speed at controlling feature (mi/h) Average final speed (mi/h) GB decel rate (ft/s2) Deceleration rate over Green Book distance3 (ft/s2) Deceleration rate to controlling feature or final speed4 (ft/s2) 5 15 50 5 15 50 Controlling feature: Horizontal alignment I-70/US-40 55 48.8 –244 570 100 26 34.0 31.2 –5.38 –4.25 –4.09 –3.71 –4.52* –4.28* –3.75* I-435/US-24 61 55.3 –641 595 75 26 46.4 42.3 36.3 –5.70 –3.33 –2.94 –1.83 –3.58 –3.12 –2.14 I-635/Plano 55 52.5 –256 425 35 36 44.8 36.5 37.5 –5.56 –3.92 –3.38 –1.91 –4.38 –3.99 –2.93 I-635 EB/Freeport 58 53.6 –309 520 30 26 45.7 35.0 37.9 –5.56 –3.29 –2.90 –1.93 –3.41 –2.90 –1.85 I-635 WB/Freeport 58 55.3 –353 500 10 26 46.7 36.0 –5.38 –3.91 –2.96 –1.97 –4.11* –3.68* –2.86* Controlling feature: Crossroad terminal I-435/Quivira 61 59.6 –718 1,640 980 0 54.7 42.3 –6.06 –2.15 –1.54 –0.88 –2.26* –1.88* –1.29* I-435/Gregory 61 60.9 –658 1,520 782 0 54.8 27.2 27.8 –4.90 –1.92 –1.46 –1.08 –3.17 –2.84 –2.47 I-635/Marsh 55 48.3 –196 1,035 465 0 429. 33.7 –5.71 –2.34 –1.79 –0.79 –2.97* –2.26* –1.34* I-635/Metro 61 59.8 –507 960 69 0 45.6 —5 —5 —5 —5 —5 27.3 –4.49 –3.90 –2.70 –1.70 –4.33* –3.88* –2.86* 1 Deceleration Length = SCL length + distance from painted nose to controlling feature. 2 Observed speed at location where vehicle has traveled the minimum Green Book distance from the point of diverge. 3 Deceleration rate over minimum Green Book distance beginning at point of diverge. 4 Deceleration rate from point of diverge to controlling feature, or to final reading when speed was not recorded at controlling feature. 5 Researchers were unable to collect speed data at the controlling feature; the final speeds of all profiles were upstream of the controlling feature. * Value is based on final speed readings. Table 32. Observed deceleration rates for platooned ramp vehicles (free-diverge conditions and passenger cars only).

Ramp location GB initial speed (mi/h) Average diverge speed (mi/h) Average diverge distance (ft) Decel length1 (ft) Difference from GB decel (ft) GB speed at controlling feature (mi/h) Average speed at GB dist2 (mi/h) Average speed at controlling feature (mi/h) Average final speed (mi/h) GB decel rate (ft/s2) Deceleration rate over Green Book distance3 (ft/s2) Deceleration rate to controlling feature or final speed4 (ft/s2) 5 15 50 5 15 50 Controlling feature: Horizontal alignment I-70/US-40 55 51.5 –257 570 100 26 36.6 34.5 –5.38 –4.93 –4.33 –3.29 –5.21* –4.60* –3.69* I-435/US-24 61 58.0 –661 595 75 26 50.0 44.9 38.3 –5.70 –3.24 –2.73 –1.73 –3.54 –3.09 –2.24 I-635/Plano 55 53.8 –267 425 35 36 46.9 38.8 39.0 –5.56 –3.87 –3.20 –1.92 –4.38 –3.85 –2.86 I-635 EB/Freeport 58 55.2 –298 520 30 26 47.0 35.4 38.5 –5.38 –3.46 –2.90 –1.91 –4.26 –3.81 –2.96 I-635 WB/Freeport 58 56.6 –349 500 10 26 48.8 38.2 –5.38 –3.58 –2.84 –1.76 –4.11* –3.62* –2.88* Controlling feature: Crossroad terminal I-435/Quivira 61 61.4 –728 1640 980 0 57.4 29.8 44.2 –6.06 –1.88 –1.42 –0.78 –2.90 –2.82 –2.55 I-435/Gregory 61 60.5 –646 1520 782 0 54.7 26.2 25.0 –4.90 –1.77 –1.57 –0.93 –3.07 –2.76 –2.34 I-635/Marsh 55 50.1 –205 1035 465 0 45.9 35.4 –5.71 –2.13 –1.64 –0.82 –2.92* –2.39* –1.60* I-635/Metro 61 60.5 –470 960 69 0 48.0 —5 —5 —5 33.9 –4.49 –2.89 –2.38 –1.60 –3.76* –3.22* –2.55* 1 Deceleration Length = SCL length + distance from painted nose to controlling feature. 2 Observed speed at location where vehicle has traveled the minimum Green Book distance from the point of diverge. 3 Deceleration rate over minimum Green Book distance beginning at point of diverge. 4 Deceleration rate from point of diverge to controlling feature, or to final reading when speed was not recorded at controlling feature. 5 Researchers were unable to collect speed data at the controlling feature; the final speeds of all profiles were upstream of the controlling feature. * Value is based on final speed readings. —5 Table 33. Observed deceleration rates for free-flow and platooned ramp vehicles (free diverge and passenger cars only).

96 -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0 100 200 300 400 500 600 700 800 900 1000 D ec el er at io n Ra te (f t/s 2 ) Length of Ramp Greater than Green Book Length (ft) 5% All 15% All 50% All 5% FF 15% FF 50% FF 5% Platoon 15% Platoon 50% Platoon Figure 52. Comparison of observed passenger car deceleration rates by ramp length. -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 I-70/ US-40 I-435/ US-24 I-635/ Plano I-635 EB Freeport I-635 WB Freeport I-435/ Quivira I-435/ Gregory I-635/ Marsh I-635/ Metro D ec el er at io n Ra te (f t/s 2 ) 5% All 5% FF 5% Platoon Loop Straight Figure 53. Comparison of observed passenger car 5th percentile deceleration rates on loop ramps and straight ramps.

97 Comparison of Deceleration Rates: Deceleration Rates from Green Book Criteria for Level Grades Assuming a Constant Deceleration. Table 34 provides a comparison of the deceleration rates derived from the Green Book recom- mended deceleration lane lengths for level grades to decel- eration rates measured in the field for various vehicle groups under various conditions. The table format is based on Green Book Exhibit 10-73, and the values for design speed, speed reached, curve design speed, and initial speed are all taken directly from that table. Below the presentation of the Green Book deceleration rates assuming a constant deceleration are three categories of field-measured deceleration rates: Condition 1—This category shows the average deceleration rate for free-flow passenger cars that exited the freeway when freeway speeds were greater than 50 mi/h. Platooned vehicles are excluded because it is assumed their speeds and deceleration choices are constrained by vehicle(s) in front of them. Condition 2—This category shows the average deceleration data for free-flow heavy vehicles (rather than passenger cars) exiting freeway traffic that is moving faster than 50 mi/h. Platooned vehicles are excluded because it is assumed their speeds and deceleration choices are constrained by vehicle(s) in front of them. Design speed (mi/h) Speed reached (mi/h) Deceleration length, L (ft) for design speed of exit curve (mi/h) Stop 15 20 25 30 35 40 45 50 For average running speed on exit curve (mi/h) 0 14 18 22 26 30 36 40 44 2004 Green Book deceleration rates (ft/s2) (derived from recommended minimum ramp lengths) 30 28 –3.59 –3.16 –2.91 –2.30 – – – – – 35 32 –3.93 –3.56 –3.59 –3.14 –2.50 – – – – 40 36 –4.36 –4.01 –3.95 –3.72 –3.60 –2.75 – – – 45 40 –4.47 –4.31 –4.22 –4.07 –3.98 –3.42 – – – 50 44 –4.79 –4.62 –4.50 –4.40 –4.30 –3.91 –3.06 –2.07 – 55 48 –5.16 –4.98 –4.84 –4.77 –4.61 –4.31 –3.80 –3.22 – 60 52 –5.49 –5.39 –5.33 –5.19 –5.07 –4.79 –4.33 –3.96 –3.44 65 55 –5.71 –5.63 –5.59 –5.47 –5.38 –5.19 –4.77 –4.51 –4.18 70 58 –5.88 –5.78 –5.74 –5.63 –5.56 –5.41 –5.06 –4.86 –4.52 75 61 –6.06 –5.97 –5.89 –5.80 –5.70 –5.67 –5.32 –5.18 –4.92 Condition 1: Measured average deceleration rates (ft/s2) for passenger cars (free flow and free diverge) 30 28 35 32 40 36 45 40 –1.13 –0.85 50 44 –1.67 –1.37 –0.96 55 48 –1.93 –1.64 –1.28 60 52 –3.19 –3.15 –2.47 –2.41 –2.10 –1.71 65 55 –3.67 –3.59 –3.30 –2.61 –2.81 –2.44 –2.01 70 58 –3.91 –3.42 –3.04 –2.58 75 61 –2.82 –3.07 –2.57 Condition 2: Measured average deceleration rates (ft/s2) for trucks (free flow and free diverge) 30 28 35 32 40 36 45 40 50 44 –1.65 –1.89 –1.90 –1.83 –1.73 55 48 –2.35 –1.90 –1.65 –1.75 60 52 –3.67 –3.05 –2.80 –2.57 –2.20 –2.03 65 55 –3.77 –3.83 –3.19 –2.59 –2.27 –1.81 70 58 –2.16 –2.09 –2.65 –3.12 –2.99 –2.72 –2.38 75 61 –2.32 –2.26 –2.13 –1.90 –1.69 –1.55 Condition 3: measured average deceleration rates (ft/s2) for passenger cars (free flow/platoon and constrained diverge) 30 28 35 32 40 36 45 40 –1.17 50 44 –1.30 –0.88 –0.52 55 48 –3.00 –1.63 –1.37 60 52 –2.12 –2.06 –1.84 65 55 –4.10 –3.30 –2.94 70 58 –4.47 –4.19 75 61 Table 34. Comparison of Green Book and field deceleration rates on level grades.

98 Condition 3—This category considers all passenger cars (i.e., both free-flow and platooned) that are exiting the freeway under constrained-diverge conditions, when the freeway speed is between 40 and 50 mi/h. Platooned vehicles are included because constrained-diverge conditions typically occur during peak traffic flow periods where a majority of vehicles on the ramp are in platoons. Table 34 compiles field-measured deceleration data for comparison with Green Book assumed deceleration rates. For each condition, the values in the table were populated by developing a database of the speed profile of each vehicle as measured by the laser gun in the field. This provided each vehicle’s speed at known distances along the ramp. For each speed profile, a deceleration rate was calculated for the sec- tions of the profile that correspond to a given value in the table (identified by an initial speed and a speed reached). Since the database provides the location along the ramp where each speed is first reached, the length between initial speed and speed reached is known and can be used to cal- culate decelerations between the two points. For example, if a given speed profile had a diverge speed of 44 mi/h and a final speed of 22 mi/h, deceleration rates could be calcu- lated for five values in the table (initial speed of 44 mi/h to speed reached of 40, 36, 30, 26, and 22 mi/h). This process was repeated for each speed profile, so that each value in the table included several observations. Only deceleration rates of vehicles on ramps with flat grades of two percent or less were used in the development of Table 34. Only values based upon 5 or more observations are presented in Table 34. Table 34 shows that in the ideal situation of Condition 1, the observed passenger cars decelerated at rates less than those assumed in the Green Book. The Condition 1 decel- eration values are lower than the corresponding Green Book deceleration values where measurements were available. The magnitude of Green Book deceleration rates generally decreases moving to the right and upward through the table; this is generally the trend in Condition 1 as well. Condition 2 deceleration rates for trucks follow the same trend as rates for cars and are also lower than the Green Book values. The magnitudes of Condition 2 deceleration rates are frequently greater than those shown in Condition 1 at lower diverge speeds (55 mi/h and below), but not necessarily so at high diverge speeds. Condition 3 considers passenger cars in constrained- diverge conditions. The magnitudes of deceleration rates in constrained-diverge conditions are almost always higher than the rates in free-diverge conditions, but even in constrained- diverge conditions, vehicles decelerated at rates less than those assumed in the Green Book. Comparison of Deceleration Rates: Deceleration Rates from Green Book Criteria for Level Grades Assuming a Two-Step Deceleration Process. This section compares deceleration rates measured in the field to deceleration rates based on Green Book criteria for level grades, assum- ing a two-step deceleration process. The analysis focuses on comparing deceleration rates measured in the field to the assumed deceleration rates during primary braking. Based upon results of the behavioral study (see Section 6), the first 1.0 s of data after the beginning of the diverge maneuver was excluded from this analysis. This was done to better cap- ture the deceleration activity that takes place during braking. Excluding the first 1.0 s of data helps ensure that the coasting that began in the freeway lane was complete. This analysis approach deviates slightly from the basic two-step process for establishing design criteria for deceleration lanes assumed by AASHTO. AASHTO assumes 3.0 s for the deceleration time without braking (i.e., coasting), but a coasting time of 1.0 s after initiating the diverge maneuver is more consistent with the data from the behavioral study. Only deceleration rates of vehicles on ramps with flat grades of two percent or less were used in this analysis. In general, drivers in the observational study decelerated in a manner equivalent to a constant rate of 2.73 ft/s2. This rate is lower than those in Green Book Exhibit 10-73 for all but a few combinations of initial speed and final speed, where rates are typically between 5.00 and 9.00 ft/s2. The following tables provide summaries of the character- istics associated with the deceleration rates observed during primary braking. Table 35 shows the difference in rates by metropolitan area, and indicates that there was not a substan- tial difference between results from study sites in Dallas and Kansas City. An analysis of deceleration rates by the location at which the diverge maneuver began showed that deceleration rates increased substantially as the diverge point moved closer to the painted nose. This is intuitive, because greater deceler- ation is needed to accomplish the same speed change in a reduced distance. Table 36 shows deceleration rates as divided by presence of leading vehicles. Free-flow vehicles had a somewhat higher deceleration rate than platooned vehicles, indicative of the higher speeds at which free-flow vehicles travel and the need to reduce speed by a greater amount. Location Number of vehicles Average deceleration rate (ft/s2) Dallas 1,275 –2.68 I-635/Marsh 313 –1.63 I-635/Plano 302 –2.89 I-635 EB/Freeport 232 –2.93 I-635 WB/Freeport 428 –3.17 Kansas City 989 –2.79 I-435/US-24 363 –2.91 I-435/Quivira 327 –1.41 I-70/US-40 299 –4.15 Total 2,264 –2.73 Table 35. Average deceleration rates by location.

99 Table 37 shows the distribution of deceleration rate by vehicle type. There was not a great deal of difference between rates for passenger cars and rates for heavy trucks over- all (–2.72 and –2.82 ft/s2, respectively), though the rate for trucks was slightly greater. Comparison of rates for a given initial speed also suggests that trucks decelerated at greater rates than cars. The reason that the overall average rates for cars and for trucks are so similar is likely a result of there being more trucks with lower initial speeds. It should also be noted that, at the lowest initial speeds, passenger cars actually accelerated (shown as positive values in Table 37), indicating that drivers entered the ramp at speeds lower than the operat- ing speed of the controlling feature, and they felt comfortable enough to increase their speed and still negotiate the ramp. Tables 38 through 40 show deceleration rates by ramp characteristics. Table 38 shows a noticeable difference in rates on loop ramps and straight ramps, with loop ramps having a greater deceleration by 0.6 ft/s2. Parallel SCLs had Lead vehicle status Number of vehicles Average deceleration rate (ft/s2) Free-flow 1,803 –2.83 Platoon 461 –2.34 Total 2,264 –2.73 Table 36. Average deceleration rates by presence of lead vehicle. Initial speed (mi/h) Passenger cars Heavy vehicles Average deceleration rate (ft/s2) Count Average deceleration rate (ft/s2) Count 24 2.77 1 29 0.28 1 30 0.59 1 31 0.00 1 32 0.21 2 34 –0.16 2 35 –0.36 1 36 –0.43 8 37 –0.39 5 –0.56 1 38 –0.77 4 39 –0.67 11 40 –0.94 10 41 –0.98 3 42 –0.87 7 –1.75 1 43 –1.12 18 –1.77 2 44 –1.87 19 –2.98 4 45 –1.67 22 –2.25 5 46 –1.92 36 –1.70 1 47 –2.10 54 –1.91 6 48 –2.44 59 –2.82 6 49 –2.19 81 –2.91 11 50 –2.55 95 –2.15 4 51 –2.56 94 –2.80 8 52 –2.69 126 –2.83 13 53 –2.71 147 –2.94 9 54 –3.02 145 –3.53 7 55 –2.92 154 –2.78 8 56 –3.01 138 –2.96 2 57 –2.98 138 –2.87 5 58 –3.18 157 –3.25 8 59 –3.09 100 –3.49 3 60 –3.19 90 –3.91 4 61 –2.98 100 –2.06 2 62 –3.08 86 –3.76 2 63 –2.96 60 64 –2.61 52 65 –2.92 27 66 –2.51 32 67 –2.80 17 68 –2.47 12 69 –2.22 11 70 –2.40 11 71 –2.70 7 72 –2.79 4 73 –2.85 1 74 –1.67 2 Table 37. Average deceleration rates by vehicle type.

100 Ramp type Number of vehicles Average deceleration rate (ft/s2) Loop 1,325 –2.99 Straight 939 –2.35 Total 2,264 –2.73 Table 38. Average deceleration rates by ramp type. Diverge type Number of vehicles Average deceleration rate (ft/s2) Parallel 662 −3.47 Taper 1,602 −2.42 Total 2,264 −2.73 Table 39. Average deceleration rates by diverge type. Ramp/diverge type Number of vehicles Average deceleration rate (ft/s2) Loop Parallel 363 −2.91 Taper 962 −3.03 Straight Parallel 299 −4.15 Taper 640 −1.52 Total 2,264 −2.73 Table 40. Average deceleration rates by ramp and diverge type. Diverge conditions Number of vehicles Average deceleration rate (ft/s2) Constrained 63 −1.85 Free 2,201 −2.75 Total 2,264 −2.73 Table 41. Average deceleration rates by diverge conditions. a substantially higher rate than tapered SCLs (1.25 ft/s2), as shown in Table 39. This finding was not initially as expected, but could be explained by drivers on parallel SCLs making a more focused deceleration action within the SCL that they could not perform with a tapered diverge; on parallel SCL drivers possibly completed more of their deceleration in the form of coasting on the freeway mainlane. Table 40 shows results for combinations of ramp type and diverge type. The disparity between parallel and tapered diverge types was most apparent on straight ramps, where the deceleration rate for the former was more than twice the latter. A similar explanation is plausible, in that drivers have more latitude to decelerate within the SCL on a parallel SCL than on a tapered SCL. In addition, deceleration can be more gradual on a straight ramp than on a loop ramp because the controlling feature is typically further downstream than on a loop ramp. The average deceleration rates on both diverge types for loop ramps were higher than the overall average and had much more similar values than the corresponding rates for straight ramps. Table 41 provides the comparison of deceleration rates by diverge condition. The results show drivers decelerate at greater rates within the deceleration lane during free-diverge conditions compared to during constrained-diverge condi- tions. This is probably most reflective of the higher initial diverge speeds observed during free-diverge conditions. Figure 54 shows the observed deceleration rates as com- pared to the initial speed at the beginning of the diverge maneuver. For the lowest initial speeds (i.e., speeds 30 mi/h and below), there was actually a small acceleration, reflect- ing that the driver was already at or below a speed deemed appropriate or comfortable for traversing the ramp when the diverge maneuver was initiated. Above 30 mi/h, the decelera- tion rate generally increased as the speed increased, though not on a linear basis. A simple trendline for the data shows a strong relationship with the natural log of initial speed. Initial speeds below 35 mi/h or above 70 mi/h were very rare, so the shape of the trendline at each end may be less reliable than for speeds where more data were available, but the general trend has a high level of confidence.

101 Comparison of Deceleration Rates on Grades of 3 Per- cent or Greater. Two of the exit ramp locations had grades that could not be considered in the level-grade category with the other nine ramps: I-435 at Gregory (3 percent down- grade) and I-635 at Metropolitan (6 percent downgrade). Because of the steeper grade, these two ramps were withheld from the previous analyses. Table 42 shows the average deceleration rates by location for the steeper ramps. It illustrates that at the I-435/Gregory exit ramp, the average deceleration rate was 0.32 ft/s2 less than the average deceleration rate at the I-635/Metropolitan exit ramp. The exit ramp at Metropolitan is steeper and shorter than the ramp at Gregory; the combination of influences of length and grade encouraged drivers to decelerate more sharply at Metropolitan to accomplish the desired speed change. Even with the increased grade, however, both ramps have average deceleration rates within the range of rates found at the nine exit ramps on level grade. The deceleration rate for the Metropolitan ramp is 0.02 ft/s2 below the average for the nine level ramps, and the rate for the Gregory ramp is greater than two of the level ramps. Summary of Findings from Examination of Deceleration Rates. The most noteworthy findings from the examination of deceleration rates are as follows: • Decelerations of diverging vehicles along the SCL and ramp are not constant. • In every combination of diverge speed and final speed, vehicles are decelerating at rates lower than those assumed by the Green Book. This is likely due in part to the study site ramps being longer than the Green Book minimum lengths and to vehicles decelerating in the freeway lane upstream of the exit taper. • Ramps more than 400 ft longer than the Green Book mini- mum have lower deceleration rates than ramps within 100 ft of the Green Book length. • Observed deceleration rates on loop ramps were greater than those on straight ramps. • There was little practical difference in deceleration rates by metropolitan area in the field study. • Deceleration rates increased substantially as the diverge point moved closer to the painted nose. • Free-flow vehicles had a somewhat higher deceleration rate than platooned vehicles. y = −1.449ln(x) + 2.3006 R² = 0.8981 -4 -3 -2 -1 0 1 2 3 24 30 32 35 37 39 41 43 45 47 49 51 53 55 57 59 61 63 65 67 69 71 73 A cc el R at e (ft /s2 ) Initial Speed (mph) Figure 54. Average deceleration rate by initial speed. Location Number of vehicles Average deceleration rate (ft/s2) Kansas City I-435/Gregory 133 −2.39 I-635/Metropolitan 237 −2.71 Total 370 −2.59 Table 42. Average deceleration rates by location for ramps on steep grades.

102 • There was very little difference between deceleration rates for passenger cars and heavy trucks overall; however, com- parison of rates for a given initial speed suggest that trucks decelerated at a greater rate than cars. • There was a noticeable difference between deceleration rates on loop ramps and straight ramps, with loop ramps having greater deceleration. • Parallel SCLs had a substantially higher deceleration rate than tapered SCLs. The disparity between parallel and tapered diverge types was most apparent on straight ramps, where the deceleration rate for the former was more than twice the latter. • For initial speeds above 30 mi/h, the deceleration rate gen- erally increased as the speed increased, having a strong relationship with the natural log of initial speed. 5.4.2.4 Critical or Unusual Maneuvers While reviewing the video recordings, no critical or unusual maneuvers were observed in the vicinity of the exit ramps.