Selecting Ramp Design Speeds, Volume 2: Conduct of Research Report (2021)

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.

8 Section 2. Literature Review and Survey of Practice This section summarizes the state of practice and other information related to ramp design speed selection. It is based on a review of published research studies, guidance documents, and design manuals; a survey of engineers, designers, safety professionals, planners, construction managers, operations managers, and others; and interviews with selected professionals. 2.1 Literature Review The published literature is summarized according to the following topics: • Current AASHTO policy on ramp design speed. • Summary of state departments of transportation (DOT) design manuals on ramp design speed. • Speed prediction models for ramps. • Interchange ramp design and safety performance. • Summary of other key resources. Current AASHTO Policy on Ramp Design Speed To begin the review of AASHTO’s current policy on selecting appropriate ramp design speeds, it is necessary to first review AASHTO’s definition of design speed and the basic concepts of design speed most pertinent to ramp design speed. At the start of this project, the 2011 edition of the Green Book was the most recent version, so this section refers to the 2011 version of the Green Book. Design Speed Section 2.3.6 of the 2011 Green Book states that, “Design speed is a selected speed used to determine the various geometric design features of the roadway. The selected design speed should be a logical one with respect to the anticipated operating speed, topography, the adjacent land use, and the functional classification of the highway. In selection of design speed, every effort should be made to attain a desired combination of safety, mobility, and efficiency within the constraints of environmental quality, economics, aesthetics, and social or political impacts. Once the design speed is selected, all of the pertinent highway features should be related to it to obtain a balanced design. Above-minimum design criteria for specific design elements should be used, where practical, particularly on high-speed facilities. On lower-speed facilities, use of above-minimum design criteria may encourage travel at speeds higher than the design speed. Some design features, such as curvature, superelevation, and sight distance, are directly related to, and vary appreciably with, design speed. Other features, such as widths of lanes and shoulders and clearances to walls and rails, are not directly related to design speed, but they do affect

9 vehicle speeds. Thus, when a change is made in design speed, many elements of the highway design will change accordingly.” AASHTO policy continues to explain that the selected design speed should be consistent with the speeds that drivers are likely to expect on a given highway facility and should fit the travel desires and habits of all drivers expected to use the particular facility. It is also desirable that the running speed of a large proportion of drivers be lower than the design speed. Ramp Design Speed Referring specifically to guidance on selecting a design speed for ramps, Section 10.9.6 of the 2011 AASHTO Green Book states that it is desirable for ramp design speeds to approximate the low-volume running speed on the intersecting highways, but this is not always practical. Thus, lower design speeds may be selected but should not be less than the lower range of speeds shown in Green Book Table 10-1. AASHTO policy provides further guidance on selecting appropriate design speed values from Green Book Table 10-1 based on various conditions and ramp types. The Green Book also states that the guide values for ramp design speed in Green Book Table 10-1 only apply to the sharpest or controlling ramp curve, which is usually on the ramp proper, and that the speed values in Green Book Table 10-1 do not pertain to the ramp terminals. The ramp terminals are to be properly transitioned and provided with speed-change facilities adequate for the speed of the highway being considered. On ramps for right turns, the Green Book states that an upper-range value of design speed is often attainable, while a value in the middle range is usually practical. A diagonal ramp of a diamond interchange may also be used for right turns. For diagonal ramps, a middle-range value for the design speed is suggested as practical. Upper-range values of design speed are typically not attainable on loop ramps, because higher design speeds (above 30 mph) require greater travel distance and larger radii, which require more land often unavailable in urban areas. The Green Book advises that minimum design speed values usually control the design of loop ramps, but the design speed should be no less than 25 mph for highway design speeds above 50 mph. Under less restrictive conditions, the design speed and radius of a loop ramp may be increased. Design speeds between the middle and upper ranges in Green Book Table 10-1 are recommended for semidirect connections and are typically in the 30 to 40 mph range. A design speed less than 30 mph should not be used for semidirect connections. For short single-lane semidirect connections, design speeds greater than 50 mph are not practical. For direct connections, design speeds between the middle and upper ranges in Green Book Table 10-1 are recommended, and the preferred minimum design speed is 40 mph. The Green Book states that for intersecting highways with different design speeds, the highway with the greater design speed should be the control for selecting the design speed for the ramp as a whole, with the provision that, as stated above, design speeds apply to certain portions of the ramp more than others. As a result, the portion of the ramp nearer the highway with the lower design speed could also be designed for the lower speed, particularly for ramps on an upgrade from a higher-speed roadway to a lower-speed roadway.

10 The Green Book also states that Green Book Table 10-1 is not applicable to the portion of ramps near at-grade intersections, because stop signs or signals typically control those locations. The sight distance along a ramp should be at least as great as the design stopping sight distance. Sight distance for passing on a ramp is not needed. Ramp grades are not directly related to the ramp design speed. However, the ramp design speed is a general indication of the quality of the ramp design. Generally, ramps with higher design speeds should have gradients flatter than ramps with lower design speeds. Table 3 specifies the desirable limiting upgrades for ramps based on the ramp design speeds. Steeper grades may be used where appropriate based on topographic conditions. For downgrades on one-way ramps, the limiting values in Table 3 still apply, but in special cases they may be 2 percent greater. Table 3. Desirable Limiting Upgrades for Ramps Based on Ramp Design Speeds (based on AASHTO, 2011) Ramp Design Speed Desirable Upgrade Limits 45 to 50 mph 3 to 5 percent 40 mph 4 to 6 percent 25 to 30 mph 5 to 7 percent 15 to 25 mph 6 to 8 percent The three segments of a ramp (exit terminal, ramp proper, and entrance terminal) should be evaluated in combination to determine appropriate design speeds and superelevation rates for the given ramp configuration. For a diamond ramp, the superelevation rate and curve radii should reflect a decreasing sequence of design speeds for the exit terminal, the ramp proper, and the entrance terminal. Loop ramps consist of a moderate-speed exit terminal connecting to a slow- speed ramp proper which connects to a moderate-speed acceleration lane. Direct and semidirect ramps usually consist of a high-speed exit, a moderate- or high-speed ramp proper, and a high- speed entrance. Freeway Mainline Ramp Terminals AASHTO defines a speed-change lane (SCL) as an auxiliary lane, including tapered areas, used for acceleration or deceleration of vehicles entering or leaving the through traffic lanes. The three terms—SCL, acceleration lane, and deceleration lane—apply broadly to the added pavement joining the traveled way of the highway or street with that of the turning roadway. A SCL should have sufficient length to enable a driver to perform an appropriate change in speed between the highway and the turning roadway in a safe and comfortable manner. In some situations, AASHTO policy uses a speed other than design speed to determine the various geometric design features. In the case of freeway mainline ramp terminals, minimum acceleration and deceleration lengths are based upon operating speeds along the freeway and the entrance/exit curve of the ramp. Two general types of SCLs are tapered and parallel designs. The tapered design provides direct entry or exit at a flat angle, while the parallel design has an added lane for changing speed.

11 Entrance Ramp Terminals A taper-type entrance, as shown in Figure 2A, merges into the freeway with a long, uniform taper. The geometrics of a taper-type entrance should be designed in a way to allow drivers to attain a speed within 5 mph of the operating speed of the freeway by the time they reach the point where the left edge of the SCL joins the traveled way of the freeway. AASHTO defines this point of convergence as the location where the right edge of the SCL and traveled way are 12-ft apart. The length required for the vehicle to accelerate to this point of convergence is governed by the speed differential between the operating speed on the entrance curve of the ramp proper and the operating speed of the highway. Green Book Table 10-3 (Table 4) shows the minimum lengths of acceleration distances for entrance terminals. Green Book Figure 10-69 (Figure 2) shows the minimum lengths for gap acceptance. Adjustment factors in Green Book Table 10-4 (Table 5) are used to increase the recommended acceleration lengths for ramps with positive grades and decrease acceleration lengths for ramps with negative grades. Figure 2. Tapered and Parallel Designs for Entrance Ramps (AASHTO, 2011).

12 Table 4. Minimum Acceleration Lengths for Entrance Terminals with Flat Grades of Two Percent or Less (AASHTO, 2011) Acceleration Length, L (ft) for Entrance Curve Design Speed (mph) Highway Stop Condition 15 20 25 30 35 40 45 50 Design Speed, V (mph) Speed Reached, Va (mph) and Initial speed, V’a (mph) 0 14 18 22 26 30 36 40 44 30 35 40 45 50 55 60 65 70 75 23 27 31 35 39 43 47 50 53 55 180 280 360 560 720 960 1,200 1,410 1,620 1,790 140 220 300 490 660 900 1,140 1,350 1,560 1,730 – 160 270 440 610 810 1,100 1,310 1,520 1,630 – – 210 380 550 780 1,020 1,220 1,420 1,580 – – 120 280 450 670 910 1,120 1,350 1,510 – – – 160 350 550 800 1000 1230 1,420 – – – – 130 320 550 770 1,000 1,160 – – – – – 150 420 600 820 1,040 – – – – – – 180 370 580 780 NOTE: Uniform 50:1 to 70:1 tapers are recommended where lengths of acceleration lanes exceed 1,300 ft. Table 5. SCL Adjustment Factors as a Function of Grade (AASHTO, 2011) Design Speed of Highway (mph) Deceleration Lanes Ratio of Length on Grade to Length on Level for Design Speed of Turning Curve (mph)a All Speeds 3 to 4% upgrade 0.9 3 to 4% downgrade 1.2 All speeds 5 to 6% upgrade 0.8 5 to 6% downgrade 1.35 Acceleration Lanes Design Speed of Highway (mph) Ratio of Length on Grade to Length on Level for Design Speed of Turning Curve (mph)a 20 30 40 50 All Speeds 3 to 4% Upgrade 3 to 4% Downgrade 40 45 50 55 60 65 70 1.3 1.3 1.3 1.35 1.4 1.45 1.5 1.3 1.35 1.4 1.45 1.5 1.55 1.6 – – 1.4 1.45 1.5 1.6 1.7 – – – – 1.6 1.7 1.8 0.7 0.675 0.65 0.625 0.6 0.6 0.6 5 to 6% Upgrade 5 to 6% Downgrade 40 45 50 55 60 65 70 1.5 1.5 1.5 1.6 1.7 1.85 2.0 1.5 1.6 1.7 1.8 1.9 2.05 2.2 – – 1.9 2.05 2.2 2.4 2.6 – – – – 2.5 2.75 3.0 0.6 0.575 0.55 0.525 0.5 0.5 0.5 a Ratio from this table multiplied by length of Table 10-3 or Table 10-5 gives length of SCL on grade. A parallel-type entrance, as shown in Figure 2B, provides an added lane of sufficient length to enable a vehicle to accelerate to near-freeway speed prior to merging. A taper is provided at the end of the added lane. The length required for the vehicle to accelerate in advance of the merge point is governed by the speed differential between the operating speed on the entrance curve of

13 the ramp proper and the operating speed of the highway. Similar to a taper-type entrance, Green Book Table 10-3 (Table 4) shows the minimum lengths of acceleration distances for a parallel- type entrance; Green Book Figure 10-69 (Figure 2) shows the minimum lengths for gap acceptance; and Green Book Table 10-4 (Table 5) provides adjustment factors to increase the recommended acceleration lengths for ramps with positive grades and decrease acceleration lane lengths for ramps with negative grades. For acceleration lanes, AASHTO policy assumes part of the ramp proper may be considered in the acceleration lane, provided the curve approaching the acceleration lane has a radius of 1,000 ft or more and the motorist on the ramp has an unobstructed view of traffic on the freeway to the left. Exit Ramp Terminals A taper-type exit, as shown in Figure 3A and 3B, provides a direct path to the ramp proper. The deceleration length is measured from the point where the lane is a 12-ft width on the ramp to the first horizontal curve on the exit ramp. Green Book Table 10-5 (Table 6) provides recommended minimum deceleration lengths for various combinations of design conditions for both taper- and parallel-type exit terminals. The minimum recommended deceleration lengths are to be adjusted in accordance with Green Book Table 10-4 (Table 5) where grades are present. Figure 3. Tapered and Parallel Designs for Exit Ramps (AASHTO, 2011)

14 Table 6. Minimum Deceleration Lengths for Exit Terminals with Flat Grades of Two Percent or Less (AASHTO, 2011) Deceleration Length, L (ft) for Design Speed of Exit Curve V’ (mph) Highway Design Speed, V (mph) Speed Reached, Va (mph) Stop Condition 15 20 25 30 35 40 45 50 For Average Running Speed on Exit Curve, V’a (mph) 0 14 18 22 26 30 36 40 44 30 35 40 45 50 55 60 65 70 75 28 32 36 40 44 48 52 55 58 61 235 280 320 385 435 480 530 570 615 660 200 250 295 350 405 455 500 540 590 635 170 210 265 325 385 440 480 520 570 620 140 185 235 295 355 410 460 500 550 600 – 150 185 250 315 380 430 470 520 575 – – 155 220 285 350 405 440 490 535 – – – – 225 285 350 390 440 490 – – – – 175 235 300 340 390 440 – – – – – – 240 280 340 390 A parallel-type exit terminal, as shown in Figure 3C, generally begins with a taper, followed by an added lane parallel to the traveled way. The length of a parallel-type deceleration lane is measured from where the added lane attains a 12-ft width to the point where the alignment of the ramp roadway departs from the alignment of the freeway. Again, Green Book Table 10-5 (Table 6) provides recommended minimum deceleration lengths for various combinations of design conditions for both taper- and parallel-type exit terminals, and Green Book Table 10-4 (Table 5) provides adjustments to the deceleration lane lengths where grades are present. Summary of State DOT Design Manuals on Ramp Design Speed Individual state DOTs sometimes have guidelines that differ from, or add to, the material found in the Green Book. To gain an appreciation for the differences that might exist, the research team conducted an online search of state DOT design manuals. Rather than reviewing design manuals for all 50 states, the research team reviewed a sample of design manuals from 20 states, including: • Arizona • California • Colorado • Florida • Georgia • Illinois • Kansas • Kentucky • Maryland • Massachusetts • Michigan • Minnesota • Missouri • New Jersey • North Carolina • Ohio • Pennsylvania • Texas • Virginia • Washington Of these 20 states, 16 had design manuals that contained sections or chapters corresponding to the relevant material in the Green Book. Of these 16 states, 11 provided guidance that was

15 essentially the same as the Green Book and/or specifically referred the reader to the Green Book. For the remaining five states, much of their guidance was also very similar to the Green Book but contained some unique features. For example, the ramp design speed values presented in the Illinois and Washington design manuals, shown in tables analogous to Green Book Table 10-1, are in multiples of 5 mph, while in the Green Book the values can be any integer (e.g., the mainline design speed of 55 mph has an upper-range ramp design speed of 50 mph in the state manuals but 48 mph in the Green Book). The ramp design speed table from the Washington manual is shown in Table 7; values different from Green Book Table 10-1 are shown in red. The Florida manual advises that minimum acceleration/deceleration lengths are provided with a minimum length of taper, but those values correspond to Green Book values. The California manual bases the minimum deceleration lengths on the radius of the controlling curve, and the Georgia manual states that ramp design speed should be no less than 10 mph below the design speed of the mainline. Table 7. Ramp Design Speed from Washington DOT Design Manual Exhibit 1360-4 (WSDOT, 2015) Main Line Design Speed (mph) 50 55 60 65 70 80 Ramp Design Speed (mph) Upper Range 45 50 50 55 60 70 Midrange 30 40 45 45 50 60 Lower Range 25 25 25 25 25 25 NOTE: Values in red differ from design values in Green Book Table 10-1. Speed Prediction Models for Ramps Speed prediction models can be used to develop an estimated speed profile of vehicles along the entire ramp or at certain locations along a ramp. Some speed prediction models estimate speeds for passenger vehicles, while others estimate speeds for trucks. Ramp Speed Trends—HSM Methodology For use in the Highway Safety Manual (HSM) (AASHTO, 2010), Bonneson et al. (2012) developed crash prediction methodologies for freeways and interchanges. The methodologies were recently incorporated into the HSM as a supplement (i.e., Chapter 18—Predictive Method for Freeways, and Chapter 19—Predictive Method for Ramps) (AASHTO, 2014) to the original three-volume edition published in 2010 (AASHTO, 2010). The general form of the safety performance function (SPF) for estimating the crash frequency for a ramp is as follows: N = Lr × exp [a + b× ln(c AADTr) + d (c ×AADTr)] (1) where: N = Crash frequency per year on the ramp. Lr = Ramp length (mi). AADTr = Average annual daily traffic volume on the ramp (veh/day). a, b, c, d = Regression coefficients. The SPF uses different regression coefficients for one-lane and two-lane ramps, for fatal-and- injury (FI) and property-damage-only (PDO) crashes, and for multiple-and single-vehicle

16 crashes. The crash modification factors (CMFs) developed for use with the SPFs account for the following factors on ramp segments: • Horizontal curvature • Lane width • Right shoulder width • Right-side barrier • Left-side barrier • Lane addition or drop • Ramp SCL • Left shoulder width For horizontal curvature, the base condition is a tangent ramp proper, and the CMF value is a function of the radius of curvature, the average entry speed for the curve, and the proportion of the ramp proper with a curvilinear alignment. The CMF value predicts an increase in crashes as the radius of curvature decreases, the average entry speed increases, and the proportion of the ramp proper with a curvilinear alignment increases. A curve speed prediction model, used with the horizontal curvature CMF, was based on data from five interchange loop ramp curves and 20 rural two-lane highway curves. The speed profile models included in the HSM methodology are applied in the direction of travel and account for the variables listed in Table 8. The speed profile models are implemented in the spreadsheet- based Enhanced Interchange Safety Analysis Tool (ISATe). Bonneson et al. note that these speed models were not developed for the purpose of predicting vehicle speeds in the context of operational or design analyses. Table 8. Input Data for Ramp Curve Speed Prediction Procedures in ISATe (Bonneson et al., 2012) Variable Description Default Value Applicable Procedure Xi Milepost of the point of change from tangent to curve (PC) for curve i1, mi None All Ri Radius of curve i2, ft None All LC,i Length of horizontal curve i, mi None All VFrwy Average traffic speed on freeway during off-peak periods of the typical day, mph Estimate is equal to the speed limit All Vxroad Average speed at point where ramp connects to crossroad, mph 15—ramps with stop-, yield-, or signal- controlled crossroad ramp terminals 30—all other ramps at service interchanges Entrance ramp, exit ramp, connector ramp at service interchange NOTES: 1 If the curve is preceded by a spiral transition, then Xi is the average of the TS and SC mileposts, where TS is the milepost of the point of change from tangent to spiral and SC is the milepost of the point of change from spiral to curve. 2 If the curve has spiral transitions, then Ri is equal to the radius of the central circular portion of the curve. When applied, the speed profile models included in the HSM methodology yield average entry and exit speeds for each curve on a ramp. Separate seven-step procedures were developed for entrance and exit ramps and are described in the following subsections. Entrance Ramp Procedure Step 1—Gather Input Data: The input data for this procedure are identified in Table 8.

17 Step 2—Compute Limiting Curve Speed: The limiting curve speed is computed for each curve on the ramp using Equation 2. vmax,i = 3.24 (32.2 Ri)0.30 (2) where, vmax,i equals the limiting speed for curve i, ft/s. The analysis proceeds in the direction of travel with the first curve encountered on the ramp designated as Curve 1 (i=1). The value of vmax is computed for all curves prior to, and including, the curve of interest. The value obtained from Equation 2 is an upper limit on the curve speed. Vehicles may reach this speed if the distance between curves is long enough or the crossroad speed is high. Step 3—Calculate Curve 1 Entry Speed: The average entry speed at Curve 1 is computed using Equation 3. vent,1 = ([1.47 VXRoad]3 + 495 × 5280 X1)1/3 ; ≤ 1.47 VFrwy (3) where, vent,1 equals the average entry speed for Curve 1, ft/s. The boundary condition of Equation 3 indicates the value computed (vent,1) cannot exceed the average freeway speed (VFrwy). Step 4—Calculate Curve 1 Exit Speed: The average exit speed at Curve 1 is computed using Equation 4. vext,1 = (V3ent,1 + 495 × 5280 LC,1)1/3 ; ≤ vmax,1 and ≤ 1.47 VFrwy (4) where, vext,1 equals the average exit speed for Curve 1, ft/s. The boundary conditions of Equation 4 indicate the value computed (vext,1) should not exceed the limiting curve speed (vmax,i) or the average freeway speed (VFrwy). Step 5—Calculate Curve i Entry Speed: The average entry speed at Curve 2 (and all subsequent curves) is computed using Equation 5. vent,i = ( V3ext,i-1 + 495 × 5280 [Xi – Xi-1 – LC,i-1] )1/3 ; ≤ 1.47 VFrwy (5) where, vent,i equals the average entry speed for curve i (i = 2, 3, ...), ft/s; and vext,i equals the average exit speed for curve i, ft/s. Step 6—Calculate Curve i Exit Speed: The average exit speed at Curve 2 (and all subsequent curves) is computed using Equation 6. vext,i = (V3ent,i + 495 × 5280 LC,i)1/3 ; ≤ vmax,i and ≤ 1.47 VFrwy (6) Step 7—Calculate Speed on Successive Curves: The entry and exit speeds for Curve 3 and successive curves are computed by applying Steps 5 and 6 for each curve. Exit Ramp Procedure Step 1—Gather Input Data: The input data for this procedure are identified in Table 8.

18 Step 2—Compute Limiting Curve Speed: This step is the same as Step 2 for the entrance ramp procedure. A lower curve speed than that obtained from Equation 2 is possible as deceleration may occur along the ramp as the driver transitions from the freeway speed to the crossroad speed. Step 3—Calculate Curve 1 Entry Speed: The average entry speed at Curve 1 is computed using Equation 7. vent,1 = 1.47 VFrwy – 0.034 × 5280 X1 ; ≥ 1.47 VXRoad (7) The boundary condition of Equation 7 indicates the value computed (vent,1) cannot be less than the average speed at the point where the ramp connects to the crossroad (VXRoad). Step 4—Calculate Curve 1 Exit Speed: The average exit speed at Curve 1 is computed using Equation 8. vext,1 = vent,1 – 0.034 × 5280 LC,1 ; ≤ vmax,1 and ≥ 1.47 VXRoad (8) The boundary conditions of Equation 8 indicate the value computed (vent,1) should not exceed the limiting curve speed (vmax,i) and should not be less than the average speed at the point where the ramp connects to the crossroad (VXRoad). Step 5—Calculate Curve i Entry Speed: The average entry speed at Curve 2 (and all subsequent curves) is computed using Equation 9. vent,i = vext,i-1 – 0.034 × 5280 (X1 – Xi-1 – LC,i-1) ; ≥ 1.47 VXRoad (9) Step 6—Calculate Curve i Exit Speed: The average exit speed at Curve 2 (and all subsequent curves) is computed using Equation 10. vext,i = vent,i – 0.034 × 5280 LC,i ; ≤ vmax,i and ≥ 1.47 VXRoad (10) Step 7—Calculate Speed on Successive Curves: The entry and exit speeds for Curve 3 and successive curves are computed by applying Steps 5 and 6 for each curve. Ramp Speed Trends—Other Sources A focused analysis of vehicle speeds on loop ramps was conducted by Torbic et al. (2016). Field data showed that the HSM methodology models tended to overestimate vehicle speeds on the controlling curves (i.e., sharpest curves) of loop ramps by the following magnitudes: • Entrance ramp: 2.6 mph at the midpoint, 1.8 mph at the PT (PT is the point of tangency— the ending point of the horizontal curve). • Exit ramp: 10.6 mph at the PC (PC is the point of curve—the beginning point of the horizontal curve), 2.2 mph at the midpoint. The aforementioned analysis was based on 15 entrance ramp sites and 13 exit ramp sites. The following models were developed to provide more accurate estimates of ramp speeds:

19 tkisosllMCcent IWWWRIv 333.4682.0912.0313.0040.0978.1359.8 2,, −+++++= (11) tkoslPTcent IWRIv 051.4079.1054.0444.1276.16 2,, −+++= (12) tkPCcext IRv 967.5090.0515.17,, −+= (13) wp drstkoslMCcext II IIIWRIv 334.4975.3 911.2551.3873.4008.1053.0241.1512.9 32,, ++ ++−+++= − (14) where: vent,c,MC = Average passenger car speed at the midpoint of the entrance ramp controlling curve, mph. vent,c,PT = Average passenger car speed at the PT of the entrance ramp controlling curve, mph. vext,c,PC = Average passenger car speed at the PC of the exit ramp controlling curve, mph. vext,c,MC = Average passenger car speed at the midpoint of the exit ramp controlling curve, mph. Il2 = Indicator variable for lane 2 (= 1 if predicting speed in the outside lane, 0 otherwise). Il2-3 = Indicator variable for lanes 2 and 3 (= 1 if predicting speed in the middle or outside lanes, 0 otherwise). R = Radius (measured to the inside of the traveled way), ft. Wl = Lane width, ft. Wos = Outside (left) shoulder width, ft. Wis = Inside (right) shoulder width, ft. Itk = Indicator variable for trucks (= 1 if predicting truck speed, 0 otherwise). Irs = Indicator variable for curve radius type (= 1 if simple, 0 if compound). Id = Indicator variable for drop SCL (= 1 if present, 0 otherwise). Ip = Indicator variable for parallel SCL (= 1 if present, 0 otherwise). Iw = Indicator variable for weaving SCL (= 1 if present, 0 otherwise). A RSPM was developed by Venglar et al. (2009) for the purpose of setting exit ramp advisory speeds. Their model is described as follows: ( )ZDCvc ln864.9758.0872.20 +−−= (15) where: vc = Average passenger car speed, mph. DC = Degree of horizontal curvature. Z = Distance to the first downstream at-grade signalized or stop-controlled intersection, ft. Venglar et al. (2009) suggested applying a multiplier of 0.95 to Equation 15 to estimate the average truck speed, and then using the average truck speed to set the advisory speed on an exit ramp. Studies have shown that speed consistency is more important than speed alone in influencing road safety. Guo et al. (2011) compared speed differentials to evaluate the level of safety at

20 freeway exits. In the process of this evaluation, Guo et al. developed speed prediction models and speed differential models along different segments of a freeway exit, and suggested using the computed speeds and differentials as a method of assessing design and safety in a manner similar to the design consistency framework suggested by Lamm et al. (1988). The freeway exit was divided into five segments: freeway mainline, transition zone, deceleration lane, upstream ramp curve, and downstream ramp curve. The following models were proposed to predict speeds at different areas of freeway exit ramps: ( )%6.91 += FW TZ vv (16) RPDLFWDL SLLvv 073.0041.0021.2901.38 ,85,85 +−+= (17) 1 ,85 210.2674232.35 R v UR −= (18) 2 ,85 180.2394279.33 R v LR −= (19) where: vTZ = Point speeds in the transition zone, mph. vFW = Point speeds on the freeway mainline, mph. v85,DL = 85th-percentile operating speeds on the deceleration lane, mph. v85,FW = 85th-percentile operating speeds on the freeway mainline, mph. v85,UR = 85th-percentile operating speeds on the upstream ramp curve, mph. v85,LR = 85th-percentile operating speeds on the downstream ramp curve, mph. LDL = Length of deceleration lane, ft. SLRP = Speed limit on the ramp, mph. R1 = Radius of the upstream ramp curve, ft. R2 = Radius of the downstream ramp curve, ft. Guo et al. also developed models for estimating the operating speed differences between the segments on the freeway exit; these are expressed as follows: DLFWDLFW LvV 046.0032.1252.3885 ,85 +−=Δ − (20) 11 193.1002.0957.1485 DRV URDL −−=Δ − (21) 21 11475.3217162.385 RR V LRUR −+=Δ − (22) where: 85ΔVFW-DL = 85th-percentile speed difference between freeway mainline and the deceleration lane, mph. 85ΔVDL-UR = 85th-percentile speed difference between deceleration lane and the upstream ramp curve, mph. 85ΔVUR-LR = 85th-percentile individual speed difference between upstream ramp curve and the downstream ramp curve, mph. D1 = Deflection angle of the upstream ramp portion, degrees.

21 Xia et al. (2012) conducted a partial correlation analysis of vehicle operating speeds and off- ramp geometric parameters under free-flow conditions and concluded that the circular curve radius, superelevation, and longitudinal gradient are the most influential variables. Speed prediction models were developed based on data collected at 12 exit ramp sites in China. The following models were proposed to predict operating speeds for passenger cars and trucks at the PC and the midpoint of the exit ramp controlling curve: 805.2ln342.9, += Rv PCc (23) 976.0ln872.7, += Rv PCt (24) 836.5631.105ln673.7, +−= μRv MCc (25) 419.2609.42ln119.7, +−= μRv MCt (26) where: vc,PC = Average passenger car speed at the PC of the exit ramp controlling curve, mph. vt,PC = Average truck speed at the PC of the exit ramp controlling curve, mph. vc,MC = Average passenger car speed at the midpoint of the exit ramp controlling curve, mph. vt,MC = Average truck speed at the midpoint of the exit ramp controlling curve, mph; R = Radius, ft. μ = Longitudinal gradient, %. Gattis et al. (2010) conducted an analysis of truck speeds and acceleration rates as trucks departed weigh stations to re-enter the highway mainline. The dataset included hundreds of trucks across five sites; each site was a ramp connecting a weigh station to a highway mainline. The sites represented a range of grades and ramp lengths. Gattis et al. categorized sites as downhill, level, or uphill and each truck as impeded or unimpeded. A truck was considered impeded if its headway with the preceding truck on the ramp was less than 7 seconds or if its headway with the leading or trailing vehicle on the highway mainline was less than 3 seconds. They developed six models to estimate the average truck speed including all trucks (impeded and unimpeded) for each grade category (downhill, level, or uphill) and unimpeded trucks for each grade category. They developed an additional set of six models using the same combinations of grade and trucks to estimate the 10th-percentile truck speed. Their model for the average speed of unimpeded trucks at a level site is as follows: 26 ,, 108.10169.0272.22 dxdv lua −−+−= (27) where: va,u,l = Average speed of unimpeded trucks at a level site, mph. d = Distance from end of scale, ft. Based on their model trends, Gattis et al. observed that accommodating truck acceleration will generally result in substantially longer entrance ramps than what are typically provided. In a detailed analysis of vehicle speeds at freeway entrances during off-peak periods, Ahammed et al. (2008) developed several models to estimate entering vehicle speeds at the merge point.

22 Their dataset included speed and acceleration profiles of hundreds of vehicles at 16 freeway entrances in Ottawa, Canada, where the SCL ended before reaching the next downstream exit ramp. They developed the following models to predict the 85th-percentile speeds of merging (entering) vehicles and vehicles in the right lane of the freeway mainline: SCLMerge AADTLv 0005.0013.0 809.34348.4285 −++= θ (28) SCLMerge QLv 007.0013.0 447.34523.4285 −++= θ (29) MergeRLRL vQv 8585 324.0009.0512.56 +−= (30) where: v85Merge = 85th-percentile speed of merging vehicles, mph. v85RL = 85th-percentile speed of vehicles in the rightmost lane of the freeway mainline, mph. θ = Angle of convergence between entrance ramp and mainline, degrees. L = Entrance ramp length, ft. AADTSCL = Annual average daily traffic volume on entrance ramp SCL, veh/day. QSCL = Equivalent hourly traffic volume on entrance ramp SCL, pc/h. QRL = Equivalent hourly traffic volume in rightmost freeway mainline lane, pc/h. Based on these models, Ahammed et al. observed that entering vehicles will merge onto the freeway mainline at a higher speed if a longer SCL or smaller angle of convergence is provided. Vehicle speeds will decrease as ramp volume increases. Additionally, they calibrated the following model to predict the 85th-percentile passenger car merging distance: 2 85 030.028.261 LD Merge += (31) where: D85Merge = 85th-percentile passenger car merging distance, ft. Based on this model, Ahammed et al. observed that increasing the SCL length will not necessarily result in improved merging operations, as drivers do not necessarily use all of the added length. They suggested that it is more effective to improve merging operations by selecting ramp geometry that will result in higher merging speeds. Based on their findings and those of others, the key geometric variables affecting merging speed are controlling curve radius and distance between the controlling curve and the entrance SCL. Interchange Ramp Design and Safety Performance Several researchers investigated the relationship of geometric design elements and safety performance of ramps. Several relevant findings are as follows:

23 • Exit ramps have higher crash rates than entrance ramps (Twomey et al., 1993; Khaoshadi, 1998). • Ramps show increasing crash rates with increasing degrees of curvature (Twomey et al., 1993). • When comparing four types of exit ramps (Type 1: parallel from a tangent single-lane exit ramp; Type 2: single-lane exit ramp without a taper; Type 3: two-lane exit ramp with an optional lane; and Type 4: two-lane exit ramp without an optional lane), the Type 1 exit ramp had the best safety performance in terms of lowest crash frequency and crash rate (Lu et al., 2010). Key geometric variables and site characteristics that influence the safety performance of interchange ramps and SCLs include (Bauer and Harwood, 1998; Khaoshadi, 1998; Yi and Mulinazzi, 2007): • Freeway volume. • Ramp volume. • Speed of right freeway lane. • Area type (urban or rural). • Ramp type (entrance or exit ramp). • Ramp configuration. • Length of ramp. • Length of SCL. Several recent studies developed safety prediction models related to freeways and ramps. Bonneson and Pratt (2009) examined freeway safety trends using Texas data and developed safety prediction models for freeway segments, including CMFs that account for the presence of ramp entrances and weaving sections. The ramp entrance CMF is described by the following equation: ( ) enrIenrenraggenr ePPCMF /9.152| 1 +−= (32) where: CMFenr|agg = Aggregated ramp entrance CMF. Penr = Proportion of freeway segment length adjacent to a ramp entrance. Ienr = Average ramp entrance length (ft). Equation 32 is applied to a freeway segment that may have one or more ramp entrances along its length, and it yields an aggregated CMF value based on all ramp entrances on the segment. The CMF is illustrated in Figure 4. As shown, the presence of a ramp entrance increases the crash frequency along a freeway segment, and the magnitude of the increase is greater if the ramp entrance length is shorter. A similar CMF was developed for weaving sections.

24 Figure 4. Aggregated Ramp Entrance CMF (Bonneson and Pratt, 2009) Bonneson et al. (2012) developed safety prediction models for both freeway mainline segments and ramp segments and developed chapters for inclusion in the HSM. The supplemental material for the HSM contains several CMFs for freeway mainline segments relevant to the issue of selecting ramp design speed (AASHTO, 2014). These CMFs include: • Lane-change maneuvers, accounting for the following ramp entrance and exit characteristics: - Presence of, and distance to, upstream ramp entrances and downstream ramp exits. - Ramp entrance or exit length. - Ramp entrance or exit side (connects to right or left side of mainline). • Type B weaving sections (which allow one weaving movement to be completed without a lane change but require at least one lane change for the other weaving movement). Even more relevant to the issue of selecting ramp design speed are the safety prediction models for ramp segments. These models include CMFs for the following ramp characteristics: • Horizontal curvature. • Lane width. • Shoulder width. • Longitudinal barrier presence. • Weaving section presence (e.g., on collector-distributor roads within cloverleaf interchanges). • Speed-change lane presence (at locations where ramps join freeway mainlines or crossroads). • Lane add or drop presence. The crash prediction methodology involves describing the characteristics of homogeneous ramp segments and estimating the crash frequency on those segments. This approach is more detailed than approaches in earlier sources, which often involved simply applying base crash rates for entire ramps based on their configuration (e.g., diamond, free-flow loop, non-free-flow loop, connector). In particular, the horizontal curvature CMF for ramp segments is influenced by vehicle speeds as depicted in Figure 5. The crash prediction methodology includes basic procedures for estimating vehicle speeds at different locations along ramps.

25 Figure 5. Ramp Horizontal Curve CMF for FI Crashes (Bonneson et al., 2012) In other research, Torbic et al. (2016) assessed how well the crash prediction models for ramps represent the safety performance of two ramp types with distinctly different geometrics: loop ramps and diagonal ramps. The HSM crash prediction procedures were applied to 235 loop ramps and 243 diagonal ramps in two-states – California and Washington – and the results were compared to five years of actual crash data for the same ramps. The results indicate that the HSM crash prediction can be applied to both loop and diagonal ramps, but to properly compare the safety performance of these two ramp types, separate calibration of loop and diagonal ramps is recommended. Summary of Other Key Resources This section summarizes other key resources not discussed in the previous sections. FHWA Ramp Design Speed Research In the late 1990s the Federal Highway Administration (FWHA) sponsored research on ramp design speed which resulted in multiple published reports and papers by Hunter and Machemehl (1997), Hunter et al. (2000), and Hunter et al. (2001). The objectives of the research were to (1) evaluate through literature review and analysis current ramp design speed criteria, (2) evaluate current ramp design speed criteria through a carefully designed sample of ramp operational data, and (3) provide evidence to validate current ramp design speed policy or to modify current procedures. The authors note that evaluation of entry ramp design speed criteria requires examination of assumptions regarding vehicle acceleration and deceleration rates and gap seeking and acceptance behavior, and should include examination of freeway driver activity and ramp driver actions. Traffic counts were collected; individual vehicles were tracked along merging areas; and times and locations where ramp vehicles merge into the freeway were recorded. The relationships between entry ramp geometric design features, speed, and other operational characteristics were investigated by comparing characteristics of ramps having “good” versus “poor” geometrics. The primary findings and recommendations from the research are as follows: 1.0 1.2 1.4 1.6 1.8 2.0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 Curve Radius, ft C ra sh M od ifi ca tio n Fa ct or . Two-Lane Highway Harwood et al. (2000) Def. Angle = 30 deg. Curve speed = 50 mi/h 40 mi/h 20 mi/h 30 mi/h Urban, Single-Lane Ramp Yates (1970)

26 1. Average ramp speeds on all observed entry ramps are consistently greater than 50 percent of the freeway design speed, even where the freeway design speed is 70 mph. Designing facilities for speeds lower than the speeds at which drivers typically operate would be inappropriate. Thus, the design criterion allowing a ramp design speed to be 50 percent of the freeway design speed should be deleted from AASHTO and TxDOT policy. 2. Speed-distance data on ramps with adequate sight distance and SCL lengths exhibit smooth profiles, having no abrupt speed changes. Speed-distance data for ramp-freeway facilities with vertical profiles limiting sight distance on ramps and marginal SCL lengths exhibit undulating waveforms, indicating significant speed changes on ramps. 3. Acceleration/deceleration rates for ramps with “good” and “poor” geometrics have patterns similar to speed histories. Ramps having adequate sight distance and SCL lengths produce small positive acceleration rates (0 to 2 mphps). Ramps with inadequate sight distance and/or inadequate acceleration lane lengths produce larger values of positive and negative acceleration (−4 to 4 mphps). Thus, the AASHTO acceleration rate model used to estimate acceleration lane lengths should not be changed. 4. Freeway right-lane speeds are not largely affected by ramp vehicles if the ramp has adequate sight distance and SCL lengths. Inadequate sight distance and/or acceleration lane lengths tend to cause significant reductions in freeway right-lane speeds, particularly under high freeway and ramp traffic volumes. 5. Freeway right-lane time headways tend to be influenced not by complex ramp design but by traffic volume. 6. Freeway right-lane volume is the factor that primarily seems to influence the size of the time gap accepted by merging ramp drivers. 7. During high traffic volume conditions, most ramp drivers travel to the end of a ramp with adequate sight distance and SCL length before merging. If a ramp has inadequate sight distance and/or acceleration lane length, this trend is much less pronounced, with some drivers aggressively merging at different locations to avoid being trapped at the end of the SCL. 8. Drivers tend to begin the merge acceleration process only after gaining a clear view of freeway right-lane traffic. If the ramp driver’s view is obstructed, acceleration does not begin until near the ramp gore where the view becomes unobstructed. Thus, acceleration lengths for taper-type entry ramps should include only the portions from which ramp drivers can clearly view the freeway right-lane traffic. AASHTO policy should clearly include this specification. The authors also noted that driver behavior upstream and downstream of the ramp gore may exhibit different characteristics. A policy that allows flexibility of choosing separate design speeds upstream and downstream of the gore may provide for overall superior designs. For example, the current 50 percent criterion produces a very desirable long acceleration lane but permits speed limiting horizontal and vertical alignment elements upstream of the ramp gore. Provision of a high design speed for upstream features and a low design speed for downstream features may provide an optimal design standard. Such a change in design philosophy should be considered.

27 ITE Freeway and Interchange Geometric Design Handbook This handbook presents fundamental concepts and practices related to freeway and interchange geometric design, often referencing the AASHTO Green Book (Leisch, 2005). This section summarizes information from the handbook relevant to the selection of ramp design speeds. Ramp Design Speed The design speed of ramps is related to the design speed of the intersecting highways to which the ramps connect. Drivers anticipate some speed reduction when exiting a highway. From a practical perspective, there is a need to limit travel distance and speed on ramps and to maintain a degree of compactness of the interchange, especially in urban areas. At the upper limit, ramp design speeds equivalent to 80 to 85 percent of design speeds on intersecting highways are considered appropriate. At the lower limit, ramp design speed equivalent to about 50 percent of the design speeds on intersection roadways are considered appropriate. AASHTO’s guide values for ramp design speeds as related to highway design speed are provided in Table 1 of this report. The higher ramp design speed allows the driver to exit the highway with little effort and close to the average running speed on that highway, while the lower ramp design speed requires deceleration by the exiting driver, accomplished through appropriate speed transition. For the entering driver, a proper ramp acceleration transition allows the merge to occur at the approximate speed of traffic on the highway. Ramp design speed depends largely upon the intersecting highways, the area types (urban or rural), and the form of interchange. Site controls and topography also have an effect. Direct connections that involve right-hand ramps for all system interchanges and some service interchanges, as Parclo-Bs and cloverleafs, call for design speeds in the upper part of the range. For system interchanges, ramp design speeds are usually 40 to 60 mph, and for service interchanges, ramp design speeds are usually 30 to 50 mph. Semidirect connections for major left-turn movements are designed close to the upper range of design speeds. Because of reverse-curve constraints, design speeds are generally limited to about 50 mph and may be as low as 40 mph in urban areas with constrained conditions. Loop ramps serve as the primary ramp configuration to accommodate left-turn movements for both service and system interchanges. Considering their spatial influence upon other ramps within the interchange, loop ramp configurations limit curve radii and ramp design speeds. Operational experience, including the aspect of extra travel distance, indicates that the practical radii for loop ramps generally range between 150 and 250 ft; approximately equivalent to design speeds of 25 to 30 mph. Table 9 presents several operational measures related to the effect of differences in loop size for curve radii ranging from 100 to 250 ft. Table 9. Loop Ramp Characteristics (Leisch, 2005) Radius (ft) 100 150 250 Design speed (mph) 20 25 30 Travel distance (ft) 630 950 1,600 Travel time (sec) 22 26 37

28 The ramp design speeds given in Table 1 as a function of the mainline highway design speed are applicable to ramps on arterials and collectors, as well as freeway ramps. For arterials and collectors, the design speed of the arterial or collector is used as the highway design speed. Recall the 2001 Green Book provided two sets of design criteria for horizontal curves: high- speed and low-speed design criteria. The low-speed design criteria for horizontal curves are based on higher net lateral acceleration values than the high-speed criteria, not to be confused with the “upper, middle, and lower ranges” of ramp design speeds. Table 10 illustrates that the choice between horizontal curve design criteria based on AASHTO’s earlier high-speed and low- speed classifications for a specific ramp should be based on the highway functional classification and design speed. Table 10. Horizontal Curve Design Criteria for Ramps as a Function of Design Speed and Functional Classification of the Mainline Highway (adapted from Leisch, 2005) Functional Classification Highway Design Speed (mph) 30 35 40 45 50 55 60 65 70 Freeway or expressway Not applicable HSD Arterial LSD HSD HSD HSD Collector LSD LSD HSD HSD HSD (high-speed design criteria for horizontal curves). LSD (low-speed design criteria for horizontal curves). Speed-Curve Relationship Curvature of ramps is predicated on design speed, superelevation, and an acceptable side friction factor. On ramps and turns at intersections, drivers tend to operate at higher speeds in relation to the curve radius than on open highway curves largely due to the acceptance by drivers of a higher side friction (i.e., net lateral acceleration) on turning roadways than on highways for through travel. It has been the practice to establish a single minimum radius for each design speed for ramps. This is logical since there is a range of design speeds that may be designated on ramps for a given highway approach. It would be an undue refinement to vary the controlling curvature by a relatively small amount for a selected ramp design speed due to possible use of differing superelevation rates. In deriving the controlling curvatures, rates of superelevation have been assumed that can be achieved in design, ranging from zero at crown slopes and at-grade ramp terminals to 0.10 for ramps at higher design speeds. The superelevation rates also reflect the condition that the sharper the curve, the shorter the length available for developing superelevation. The side friction factor varies from 0.38 for sharp turns to 0.15 at relatively high- speed ramps. The minimum radii for turning roadways for various design speeds are provided in Table 11. These are normally applied to the inner edge of the traveled way.

29 Table 11. Minimum Radii for Interchange Ramp Curves (Leisch, 2005) Design (turning) speed, V (mph) 10 15 20 25 30 35 40 45 Side friction factor, f 0.38 0.32 0.27 0.23 0.20 0.18 0.16 0.15 Assumed minimum superelevation, e/100 0.00 0.00 0.02 0.04 0.06 0.08 0.09 0.10 Total e/100 + f 0.38 0.32 0.29 0.27 0.26 0.26 0.25 0.25 Calculated minimum radius, R (ft) 18 47 92 154 231 314 426 540 Suggested minimum radius curve for design (ft) 25 50 90 150 230 310 430 540 Average running speed (mph) 10 14 18 22 26 30 34 36 NOTE: For design speeds greater than 45 mph, use values for open highway conditions. On the controlling curve at each design speed, the ramp should be fully superelevated if feasible. Ramps should be superelevated at no more than 0.08, particularly on long curves such as jughandles and loops. Ramp design speeds are often dictated by physical design constraints. The most critical design conditions for horizontal curves occur at lower design speeds, particularly at design speeds of 20 to 30 mph. Under worst-case conditions, a truck on a curve with a ramp design speed of 20 mph can roll over when traveling at 25 mph and may skid off the road under wet pavement conditions at 27 mph. Similarly, a truck on a curve with a ramp design speed of 30 mph can rollover at 38 mph and skid off the roadway at 30 mph. Passenger vehicle operations are less critical than for trucks and higher design speeds are less critical than lower design speeds. The following guidelines should be considered when selecting the design speed for an exit ramp. 1. Consider physical and economic constraints when selecting a tentative design speed for the ramp. Use the upper or middle range of ramp design speeds if possible. Avoid the lower range of ramp design speeds on ramps that will carry substantial truck traffic. 2. Identify the most critical curve on the ramp (usually, but not necessarily, the first curve downstream of the gore). 3. Develop a forecast of operating speeds at the most critical curve on the ramp based on actual speeds on existing ramps with similar mainline design speeds, mainline operating speeds, and similar geometrics for the SCL and portion of the ramp prior to the most critical curve. This forecast should be based on mainline design and operating speeds, but not on ramp design speed. 4. If the forecast ramp operating speed exceeds the design speed, raise the design speed. If the ramp design speed cannot be raised due to physical, environmental, or economic constraints, consider implementing speed-control measures. For exit ramps where the anticipated operating speeds exceed the maximum feasible ramp design speed, the following speed-control measures should be considered: 1. Provide signing with an appropriate advisory speed for the ramp. 2. Place the advisory speed signing so that drivers have sufficient time to slow down before the most critical curve. 3. Increase the length of the deceleration lane. 4. Realign the ramp to increase the distance from the gore to the most critical curve.

30 5. Supplement standard advisory speed signs to make the signing more conspicuous, increase the distance from the signing to the most critical curve, and draw attention of truck drivers to the signing. This can be accomplished by: a. Using more than one ramp speed advisory sign. b. Placing ramp speed advisory signing on the mainline highway in advance of the ramp. c. Incorporating an exit speed panel on the guide sign for the exit ramp. d. Using overhead signing. e. Using a TRUCK SPEED advisory sign. f. Using flashing beacons to call attention to the advisory speed. 6. Avoid designs where the critical curve on a ramp is not obvious (e.g., where a “tight” horizontal curve follows a larger-radius curve). 7. Consider the use of collector-distributor roads in the interchange as they may reduce ramp speeds. Transitions and Compound Curves Interchanges involve curvilinear alignment with relatively rapid changes in direction. While traversing ramps, drivers naturally follow transitional paths. Spirals in combination with circular curves are well suited for this purpose. Spirals on ramps may be shorter in length than on open highways as drivers tend to drive more aggressively on turning roadways. Spiral lengths for design of ramps are shown in Table 12. The suggested lengths are generally applicable to the ramp proper where speed operation is relatively uniform. Where significant speed changes take place, considerably longer spirals are required. Table 12. Suggested Lengths of Spirals for Ramps (adapted from Leisch, 2005) Design speed (mph) 20 25 30 35 40 45 50 Minimum radius (ft) 90 150 230 310 430 550 700 Assumed C (ft/s2) 4.0 3.75 3.5 3.25 3.0 2.75 2.50 Suggested length of spiral (ft) 150 150 200 250 250 300 350 Begin using spiral when radius is (ft) 700 900 1000 1100 1300 1500 1800 Suggested length of spiral (ft) 150 150 200 200 200 200 250 Grades and Profile A general guideline for maximum gradients on ramps is presented in Table 13. There is no direct relationship between maximum gradients and ramp design speeds, but lower gradients are more applicable to the higher ramp design speeds. Table 13. Guidelines for Maximum Grades for Design of Ramps (Leisch, 2005) Normal Conditions Heavy Truck Traffic Rugged Terrain (Special Cases) 4%-6% 3%-4% 6%-8%

31 Table 14 provides suggested design speeds for exit and entrance vertical curves based on highway design speeds. These values are 80 to 90 percent of the highway design speed. This provides the driver with a sufficient view of the ramp and smooth transition between the highway grade and ramp grade. Table 14. Suggested Ramp Exit/Entrance Vertical Curve Design Speed (Leisch, 2005) Roadway Design Speed (mph) Suggested Design Speed (mph) Ramp Exit/Entrance Vertical Curve 70-80 60-65 60-65 55 50-55 45-50 40-50 40 Cross Slope and Superelevation Superelevation should be predicated on the design speed of each segment of the ramp. The development of superelevation is the transition from a crown slope to the full superelevation of a curve. The rate of cross-slope change for the design of ramps is a function of the ramp design speed. Another control in developing superelevation is the pavement/shoulder cross-slope break. To maintain comfort and safety as drivers encounter the pavement/shoulder cross-slope break, the algebraic difference in pavement slopes should not exceed the values in Table 15. Table 15. Maximum Algebraic Difference in Pavement Cross Slope for Design of Ramp Terminals at Crossover Crownline (Leisch, 2005) Design Speed of Highway (mph) Freeway/Highway Ramp Terminal Ramp/Ramp Terminal 15-25 N/A 0.07 30-40 0.05 0.06 50-60 0.04 0.05 70-80 0.03 N/A Sight Distance Sufficient ramp sight distance should be provided throughout based on the ramp design speed. Where feasible, longer sight distances should be provided. Speed-Change Lanes Lengths of SCLs are predicated on free-flow operating speeds for the designated design speed of the highway and ramp. Lengths of SCLs are based on the performance capabilities of passenger vehicles. If SCL lengths were based on trucks, longer lengths would be required. For tapered exit terminals, the angle of divergence is typically 3 degrees (i.e., a departure rate of just under 20:1) for higher design speeds. Angles of 4 to 5 degrees may be used, but these generally pertain to lower design speeds (highways and ramps). For tapered entrance terminals, the

32 angle of convergence is 1 degree, or a departure rate between 50:1 and 70:1. For highway design speeds of 50 mph or less and ramps, a rate of convergence of 40:1 may be appropriate. Design of Two-Lane Loop Ramps The loop radius of the inner edge of the traveled way normally should be not less than 200 ft. A radius of 180 ft may be employed under restricted conditions or where the design speed of the approach roadway is 50 mph or less. A radius of 150 ft has been observed to satisfactorily operate with moderate volumes and sufficient width, and where the approach roadway design speed is 45 mph or less. Metered Ramp Design Metered ramps require the application of design criteria to reflect the stop condition on the ramp proper. Also taken into consideration are the number of lanes at the ramp meter stop bar, ramp traffic volume, ramp grade, and freeway design speed. All of these dictate the length and geometry of the ramp proper and the location of the ramp meter stop bar. Figure 6 shows two examples of metered ramp designs. The dimension (L2) from the ramp meter stop bar to the merging tip at the beginning of the merging area is for acceleration. The dimension (L1) is for vehicles to accelerate from a stop to the running speed of the freeways, assuming the controlling ramp curve radius approaching the ramp terminal is 1,200 ft or greater. If the radius is less, then the dimension should be increased to reflect the design speed of the controlling curve. Also, if the ramp grade is 3 percent or greater, the L2 dimension should be adjusted. Figure 6. Example Designs—Metered Ramp (Leisch, 2005)

33 NCHRP Report 730 (Design Guidance for Freeway Mainline Ramp Terminals) The objective of this research was to develop improved design guidance for freeway mainline ramp terminals based on modern driver behavior and vehicle performance capabilities (Torbic et al., 2012). The primary steps of the research included a crash analysis and field studies. Several general conclusions from this research most relevant to the selection of ramp design speeds are as follows: • Passenger cars should remain the principal design vehicle for freeway mainline ramp terminals, which is consistent with current AASHTO policy. The exception to this rule is at freeway mainline ramp terminals where the truck volumes on the ramps are substantial, in which case further consideration should be given to more fully accommodating trucks within the design. • Freeway mainline ramp terminals should be designed based upon free-flow conditions, consistent with current AASHTO policy. By designing for free-flow conditions, sufficient length is provided to accommodate merge and diverge maneuvers during more congested operating conditions. • Consistent with current AASHTO policy, freeway mainline ramp terminal design should be based upon average operating speeds of vehicles, rather than design speeds. Designs based upon design speeds would provide terminals that are over designed. Several conclusions specific to entrance ramps from this research most relevant to the selection of ramp design speeds are as follows: • Merging vehicles do not accelerate at a constant rate along the length of the ramp and SCL. • A clear view of the freeway and mainline traffic is important to accommodate merging vehicles. Drivers begin glancing at the freeway and mainline traffic prior to reaching the painted nose of the ramp, and while drivers are taking a glance at the freeway traffic, they continue to accelerate along the ramp and SCL. • Vehicles exit curves on ramps at speeds much higher than the values given for “initial speed” in Green Book Table 10-5, suggesting that vehicle performance and driver preferences have changed since the design values were determined. • In free-merge conditions, many drivers choose to enter the freeway at speeds much lower than the speed of freeway traffic. • Constrained-merge conditions appear to be the most difficult for drivers, as reaching freeway speeds becomes more critical since gaps are smaller and do not provide as much opportunity for accelerating in the freeway lane. Additionally, drivers have a more difficult task identifying appropriate gaps. Some drivers take the first available gap they find even if they have not reached an ideal merge speed, while other drivers use the full length of the SCL, and in some cases the taper, to reach a speed near the speed of the freeway traffic before merging. • Heavy vehicles accelerate at lower rates than passenger vehicles, and they merge onto the freeway at lower speeds. However, their merging behavior does not appear to negatively impact the overall operation of the ramp terminals.

34 • Vehicles are more likely to use the full length of a tapered SCL to accelerate to near- freeway speeds before merging in contrast to a parallel SCL, where vehicles may merge earlier along the ramp and at lower speeds. • Because most situations do not require that vehicles accelerate to the speeds assumed with the design, many drivers choose to accelerate at lower rates than assumed within AASHTO policy, and many vehicles are capable of accelerating at higher rates than the assumed acceleration rates used to determine minimum acceleration lane lengths for entrance terminals in the Green Book. • Upgrades as steep as 3 to 4 percent do not impact the acceleration capabilities of passenger cars, at least over lengths necessary for entrance terminals. As grades increase to 5 or 6 percent, the acceleration rates of passenger cars tend to decrease. • Assuming constant acceleration is a reasonable approach for determining minimum acceleration lane lengths for design. The current values provided in Green Book Table 10-3 are conservative estimates for minimum acceleration lane lengths, given the current vehicle fleet and driver population. They provide sufficient length for vehicles to merge onto the freeway under a range of freeway operating conditions. In situations where free- flow conditions are expected for the foreseeable future and constraints make it difficult to provide the recommended minimum acceleration lengths, the minimum acceleration lane lengths can be reduced by 15 percent without causing expected operational problems. Several conclusions specific to exit ramps from this research most relevant to the selection of ramp design speeds are as follows: • Most diverge maneuvers begin before or within the taper or within the first or middle third of the SCL, while few diverge maneuvers take place in the final third of the SCL or beyond the painted nose. • Vehicles that diverge earlier along the deceleration lane diverge closer to freeway speeds than vehicles that diverge later along the deceleration lane, closer to the painted nose. • Deceleration rates of exiting vehicles are greater for vehicles that diverge closer to the painted nose than for vehicles that diverge further upstream from the painted nose. • Deceleration rates of diverging vehicles along the speed-change lane and ramp are not constant. • Where the deceleration lane length is longer than the Green Book recommended minimum length, most vehicles decelerate at rates lower than those assumed by the Green Book. • Free-flow vehicles decelerate at greater rates than platooned vehicles. • When exiting the freeway, trucks decelerate at rates very comparable to those of passenger cars and typically diverge from the freeway at lower speeds than passenger cars. • Crash rates for trucks are higher on parclo, free-flow and “other” ramp configurations than at diamond; outer connection; direct or semidirect connection; and button hook, scissor, and slip ramp configurations.

35 • Drivers exiting on loop ramps tend to reduce their speed in the freeway lane more, and decelerate along the speed-change lane at a greater rate, than drivers exiting on straight ramps. This may be due the visual perceptions of drivers as they approach the loop ramp. • Parallel deceleration lanes generally lead to substantially higher deceleration rates than on tapered deceleration lanes. The disparity between deceleration rates is most apparent on straight ramps. • AASHTO policy assumes a two-step process for establishing design criteria for minimum deceleration lane lengths. Deceleration is accomplished first while coasting in gear without the use of brakes and then during the application of the brake. AASHTO assumes 3.0 sec for the coasting period. The 3.0 sec of coasting time is a valid assumption for describing the diverge maneuver, but 2 sec of the coasting time typically occurs prior to the diverge maneuver while in the freeway, and 1 sec of the coasting time occurs within the speed-change lane following the diverge maneuver. • The two-step process of deceleration is a reasonable approach for determining minimum deceleration lane lengths for design and provides conservative estimates for minimum deceleration lane lengths, given the current vehicle fleet and driver population. Vehicles decelerated at rates well within the capabilities of the vehicle fleet and driver preferences. This was in part due to some deceleration by diverging vehicles in the freeway mainline prior to the diverge maneuver. It is beneficial to have a conservative design process that does not assume vehicles begin decelerating in the freeway mainline. NCHRP Report 687 (Guidelines for Ramp and Interchange Spacing) These guidelines assist planners and designers considering ramp and interchange spacing and the feasibility of new or rebuilt interchanges and ramps (Ray et al., 2011). This report does not provide much guidance relevant to the selection of ramp design speed beyond indicating that the design speed on system interchange ramps vary from 25 to 70 mph, and that interchange spacing and ramp density influence a freeway’s estimated free-flow speed. As the ramp density increases, there is a corresponding decrease in free-flow speed on the freeway. The guidelines for ramp and interchange spacing address various issues, including geometry, operations, safety, and signing, and are based on balancing the competing needs for mobility and access. Acceleration Characteristics at Metered On-Ramps Yang et al. (2015) state that ramp metering has significant impacts on the design and operation of freeway on-ramps; however, the majority of the metered ramps in the United States are retrofitted to existing ramps. A ramp meter should be strategically placed to prevent on-ramp queue spillback as well as to provide enough acceleration length for a vehicle to leave the ramp meter and reach desired merge speeds. Chaudhary and Messer (2000) explained that the design of a metered entrance ramp consists of three length components: • Sufficient distance should be provided for vehicles upstream of the entrance ramp to decelerate and stop as they encounter the back of a queue. • Sufficient queue storage length should be provided for vehicles waiting to enter the freeway.

36 • Sufficient length should be provided for a vehicle stopped at the meter to accelerate to freeway merging speed. When metering is implemented on an existing ramp, it is necessary to verify that sufficient acceleration length can still be provided after a portion of the ramp is allocated for deceleration and queue storage needs. Additionally, Zheng and McDonald (2007) observed that the merging operation may become more difficult for the entering driver as well as drivers on the mainline when ramp metering is implemented. Their analysis was based on observations of speed, lane changes, and eye movements of human subjects who drove through metered ramps (with the metering on or off) in an instrumented vehicle. They observed that when metering is on, entering drivers exhibit increased eye movements, merge onto the mainline at a lower speed, and more frequently induce lane changes by drivers already on the mainline. Zheng and McDonald suggested that it may be necessary to locate meters farther upstream along entrance ramps to mitigate these complications. This suggestion is feasible if the ramp is long enough to provide the needed queue storage space after the meter is moved upstream. 2.2 Survey of Practice A web-based survey of state DOT designers and design consultants was conducted to determine the current practice pertaining to the choice of ramp design speed. Interviews with selected design professionals were also conducted. The results of the web-based survey and interviews are reported in Sections 2.2.1 and 2.2.2. When the survey was distributed, the most recent version of the Green Book was 2011, so the survey questions reference the 2011 Green Book, not the 2018 Green Book. Web-Based Survey A web-based survey of state DOT and consultants was conducted regarding their policies and practices for the determination of ramp design speeds. One survey was sent to each of the 50 state DOT agencies. This survey focused not only on practice, but also on agency policy. A second survey targeted planners, engineers, designers, safety professionals, construction managers, operations managers, and others in the private sector and academia or at local highway agencies. This second survey was shorter and contained some of the same (or similar) questions as the survey for the state DOT agencies, but it focused more on state-of-the-practice. Participants invited to complete the surveys were identified from the following AASHTO and TRB committees and subcommittees: • AASHTO Subcommittee on Design—Technical Committee on Geometric Design. • AASHTO Highway Traffic Safety Subcommittee on Safety Management. • AASHTO Standing Committee on Planning. • AASHTO Subcommittee on Construction. • AASHTO Subcommittee on Traffic Engineering. • TRB Committee on Geometric Design. • TRB Committee on Operational Effects of Geometrics. • TRB Committee on Construction Management. • TRB Committee on Transportation Safety Management.

37 The purpose of the surveys was to learn about the experiences of professionals related to ramp design speed strategies, including: • What factors most influence the choice of design speed for various ramp types. • To what extent design practices conform to or differ from Green Book guidance. • How policies and practices differ by ramp type or location along the ramp. • Experiences related to design speed that led to a consideration of a change in policy or practice. • Practices related to advisory speed signing on ramps. • Tort liability concerns related to ramp design speed. • Use of the HSM Chapter 19 (safety prediction methodology for ramps). • Use of design exceptions for ramp design speed (or horizontal alignment) and experiences resulting from those exceptions. • Identification of ramp design projects that could be evaluated in this research project. The survey to state DOTs included 16 questions related to ramp design speed practices and experiences. The survey targeting the private sector included 11 questions related to ramp design speed practices. Both surveys also asked respondents for current or recent ramp or interchange design projects that could potentially be used as case studies for this research, as well as for contact information so the research team could follow up regarding such sites or any previous responses. Information about the type of respondents and a summary of their responses to the survey are presented next. Respondents to Web-Based Survey An email explaining the overall objective of research and a link to the survey was sent to one design or traffic engineer at each of the 50 state DOTs. Instructions invited recipients to forward the email with the survey link to a person in their organization qualified to respond to questions related to ramp design speed selection. Twenty-two responses were received from 21 state DOTs. The survey developed for local agencies, university researchers, and private-sector practitioners was sent to 324 members of the TRB and AASHTO committees listed previously. Like the survey to state DOT officials, this survey allowed recipients to forward the email to others within their organization most qualified to provide answers. Thirty complete and two partial survey responses were received. Agency types (i.e., local agencies, university researchers, and private- sector practitioners) that responded to the survey included:

38 Organization type Number of responses City or County 3 Consultant 12 University—US 5 University—International 2 DOT (retired staff) 2 None given 8 Total 32 Summary of Responses to Web-Based Survey Responses to both surveys are summarized below. Each survey question is shown in italics, followed by a parenthetical indication of which survey(s) the question appeared. Responses to questions that were identical on both surveys are grouped together, but if there were notable differences in how the two groups of respondents answered, these differences are discussed. Questions that were similar, but not identical, between surveys are shown consecutively, with the question and response from the state DOT survey presented before the question and response from the local agencies, university researchers, and private-sector practitioners. The percentages of responses are provided in the tables, along with the number of respondents in parentheses. Nearly every question on the survey provided a text box for respondents to further clarify or explain their answer, or simply provide additional detail. Many of those comments have been included here and only edited slightly to improve readability where needed. Survey questions regarding potential case study projects and survey respondents contact information are not included in this summary. Question: What level of importance do the following factors have when your agency considers the selection of an appropriate ramp design speed? (both surveys) Exit Ramps Answer Options High Importance Medium Importance Low Importance N/A Ramp configuration (e.g., diamond, loop, directional) 66.0% (33) 24.0% (12) 2.0% (1) 8.0% (4) Design/operating speed of highway 82.0% (41) 14.0% (7) 0.0% (0) 4.0% (2) Design/operating speed of intersecting roadway 48.0% (24) 24.0% (12) 22.0% (11) 6.0% (3) Ramp grade 24.0% (12) 50.0% (25) 20.0% (10) 6.0% (3) Number of lanes of ramp 16.0% (8) 44.0% (22) 26.0% (13) 14.0% (7) Use of spiral transition into control curve 10.0% (5) 20.0% (10) 28.0% (14) 42.0% (21) Presence of ramp meter 11.8% (6) 11.8% (6) 13.7% (7) 62.7% (32) Type of freeway mainline ramp terminal (e.g., parallel or taper) 24.0% (12) 28.0% (14) 38.0% (19) 10.0% (5) Type of crossroad ramp terminal (e.g., signalized, unsignalized, free-flow) 38.0% (19) 32.0% (16) 20.0% (10) 10.0% (5) Degree of physical constraints at site 38.0% (19) 42.0% (21) 12.0% (6) 8.0% (4) Degree of environmental constraints at site 40.0% (20) 42.0% (21) 12.0% (6) 6.0% (3) Right-of-way costs 44.0% (22) 36.0% (18) 14.0% (7) 6.0% (3) Red text highlights answers where 75% or more of respondents indicated a combination of high or medium importance. Shaded cells highlight answers where 25% or more of respondents indicated a combination of low importance or not applicable.

39 Entrance Ramps Answer Options High Importance Medium Importance Low Importance N/A Ramp configuration (e.g., diamond, loop, directional) 56.3% (27) 27.1% (13) 8.3% (4) 8.3% (4) Design/operating speed of highway 77.1% (37) 18.8% (9) 0.0% (0) 4.2% (2) Design/operating speed of intersecting roadway 38.8% (19) 28.6% (14) 26.5% (13) 6.1% (3) Ramp grade 28.6% (14) 42.9% (21) 22.4% (11) 6.1% (3) Number of lanes of ramp 18.4% (9) 36.7% (18) 30.6% (15) 14.3% (7) Use of spiral transition into control curve 8.2% (4) 18.4% (9) 30.6% (15) 42.9% (21) Presence of ramp meter 18.0% (9) 24.0% (12) 18.0% (9) 40.0% (20) Type of freeway mainline ramp terminal (e.g., parallel or taper) 32.7% (16) 24.5% (12) 32.7% (16) 10.2% (5) Type of crossroad ramp terminal (e.g., signalized, unsignalized, free-flow) 28.6% (14) 24.5% (12) 36.7% (18) 10.2% (5) Degree of physical constraints at site 34.7% (17) 44.9% (22) 12.2% (6) 8.2% (4) Degree of environmental constraints at site 36.7% (18) 44.9% (22) 12.2% (6) 6.1% (3) Right-of-way costs 42.9% (21) 36.7% (18) 14.3% (7) 6.1% (3) Red text highlights answers where 75% or more of respondents indicated a combination of high or medium importance. Shaded cells highlight answers where 25% or more of respondents indicated a combination of low importance or not applicable. Freeway-to-Freeway Ramps Answer Options High Importance Medium Importance Low Importance N/A Ramp configuration (e.g., diamond, loop, directional) 66.7% (32) 18.8% (9) 8.3% (4) 6.3% (3) Design/operating speed of highway 77.1% (37) 16.7% (8) 2.1% (1) 4.2% (2) Design/operating speed of intersecting roadway 57.1% (28) 20.4% (10) 8.2% (4) 14.3% (7) Ramp grade 32.7% (16) 40.8% (20) 20.4% (10) 6.1% (3) Number of lanes of ramp 22.4% (11) 32.7% (16) 32.7% (16) 12.2% (6) Use of spiral transition into control curve 10.2% (5) 22.4% (11) 26.5% (13) 40.8% (20) Presence of ramp meter 10.0% (5) 16.0% (8) 10.0% (5) 64.0% (32) Type of freeway mainline ramp terminal (e.g., parallel or taper) 32.7% (16) 14.3% (7) 36.7% (18) 16.3% (8) Type of crossroad ramp terminal (e.g., signalized, unsignalized, free-flow) 22.4% (11) 18.4% (9) 18.4% (9) 40.8% (20) Degree of physical constraints at site 28.6% (14) 46.9% (23) 16.3% (8) 8.2% (4) Degree of environmental constraints at site 32.7% (16) 42.9% (21) 18.4% (9) 6.1% (3) Right-of-way costs 38.8% (19) 36.7% (18) 18.4% (9) 6.1% (3) Red text highlights answers where 75% or more of respondents indicated a combination of high or medium importance. Shaded cells highlight answers where 25% or more of respondents indicated a combination of low importance or not applicable. Twenty state DOT representatives and 32 local agencies, university researchers, and private- sector practitioners provided at least partial responses to the question. Some respondents replied for some factors but not for others. Responses between the two groups were similar and are reported jointly here, although key differences are noted. In addition, responses were similar, in most cases, across ramp types. Notable differences are discussed. The design speed or operating speed of the highway was most frequently selected as having high importance when determining ramp design speed for all three of the ramp types. Over 90 percent of respondents indicated it had high or medium importance. This was followed closely by ramp configuration, which was categorized as high or medium importance by over 80 percent of respondents. Other factors frequently reported as having high or medium importance included: • Design/operating speed of intersecting roadway. • Ramp grade. • Degree of physical constraints at site. • Degree of environmental constraints at site. • Right-of-way costs.

40 The factor most frequently identified as being not applicable to the selection of ramp design speed was the presence of a ramp meter, although ramp metering was considered to be slightly more important on entrance ramps than exit or freeway-to-freeway ramps. The use of spiral transition into the controlling curve was also considered to have little importance to the selection of ramp design speed by most respondents. The type of crossroad ramp terminal was generally considered to have high or medium importance for exit ramps and little to no importance on freeway-to-freeway ramps, and responses were split for entrance ramps, with about half of respondents considering it important and half not. The number of lanes on the ramp was more often considered to be high or medium importance to local agencies, university researchers, and private-sector practitioners (between 62 and 73 percent for all ramps types) than to the state DOT respondents (between 40 and 45 percent for all ramp types). The question provided a comment box labeled “other” to allow respondents to input other factors they may consider in the determination of ramp design speed. One state DOT respondent commented that “our guidelines do not specifically address or list requirements on the above considerations and each designer considers all of the above with different importance on each element depending on the project scope and conditions, highway, etc.” Comments from local agencies, university researchers, and private-sector practitioners included the following: • Ramp terminal speed change from freeway to ramp. • These factors, and others, must be viewed in terms of available trade-offs. For example, right-of-way costs must be viewed in terms of expected safety and operational outcomes, i.e., don't build something unsafe or which will operate poorly just to save money. Maybe the project should not be built if the necessary factors cannot be included. • There are significant differences in service interchange requirements and system interchange requirements for design. Also, depending on interchange form, location (urban, suburban or rural), and interchanging facility types. • Should consider needs of large vehicles. Question: What type of policy does your agency follow in selecting ramp design speeds? (State DOT survey only) Use the policy in the 2011 AASHTO Green Book 61.9% (13) Use our own agency’s design policy which is very similar to the policy in the 2011 AASHTO Green Book 38.1% (8) Use our own agency’s design policy which differs from the policy in the 2011 AASHTO Green Book 0.0% (0)

41 Question: Specifically, does your agency use the Table 10-1 from the 2011 AASHTO Green Book in selecting ramp design speeds? (State DOT survey) Yes 80.0% (16) No 20.0% (4) Five respondents provided further comment, as follows: • We use the state roadway design manual, but the table is almost identical. • We use the information on pages 10-89 and 10-90 in the Green Book and provide additional guidance for direct and semidirect connections. • The table is used as a guide, a range of values if you will, but is not stringently adhered to. For example, while design and posted speeds on rural freeways have increased in our state, the superelevation and radius for controlling curves we use on diamond exit and entrance ramps are still set at 55 and 50 mph respectively. • There is direction to use the table within our Design Manual; however, it is not clear what the table means or how to apply it. It is generally recognized that the design speed varies through the ramp and design elements are controlled by insuring they meet the acceleration/deceleration table values. The overall length of the ramp is checked, and then the lengths between the design features are checked (i.e., if a 40 mph horizontal curve is needed on an exit ramp, there must be enough distance to decelerate to 40 mph prior to the curve). Question: Is Table 10-1 from the 2011 AASHTO Green Book sufficient for selecting ramp design speed? (local agencies, university researchers, and private sector survey) Yes 46.7% (14) No 53.3% (16) Several respondents provided an explanation or further comment, including the following: • I feel that a ramp should not have a set design speed. Each end is controlled by the roadway that it is tying into—so it is really a transition from a mainline speed to a crossroad or between two mainlines. I do believe that all ramps should have a minimum design speed for the tightest curve on the ramp—say 50 mph for a directional and 30 for a diamond ramp. • Large truck turnover potential. • I never liked this table. There needs to be extensive description as to when the upper, middle, or lower ranges apply. Exit and entrance ramps should be handled separately. • It should take into consideration freeway-to-freeway vs freeway to secondary road configurations, since in the first case rather increased speed values are implemented. • The current table is adequate for non-loop ramps, but there is inadequate guidance for loop ramps as it is not possible to know how to design for lower design speeds that is

42 typical of an urban loop ramp. It would also be desirable to expand the highway design speeds shown to higher values, as many states now have a 75 mph speed limit. • Based on the relatively conservative nature of the AASHTO policy, designing a ramp based on the 50th or even 70th percentile speeds does not seem appropriate. Additional justification for using speeds within these ranges should be required before utilizing them in design. • See ITE Freeway and Interchange Geometric Design Handbook (Leisch, 2005), Chapter 6, pages 238-241. • The curve radius and superelevation should be the primary considerations for ramp design speeds rather than the highway design speed. • Information regarding the effect of ramp type on design speed would be helpful. In addition, the table provides a very wide range for each highway design speed. More guidance on when to select the upper range, middle range, and lower range design speeds would be helpful. Finally, the presence of a stop-controlled intersection at an exit ramp terminal is also an important factor. • Additional guidance of the recommended practice for choosing from the upper, medium, and lower ranges would be helpful. Either theoretical or empirical information would be helpful and deciding which range would provide the best fit for a particular project. • Some of the agencies we design for have strict design speeds (often higher in speed). They provide a minimum design speed, not a selection of speeds like this table shows. • This table provides no real guidance for different ramp configurations for example a tight loop. One could look at the table and say the upper range must be 60 mph or middle 50 mph when that may not be attainable because of other constraints. Most tight loops are 25-35 mph for a majority of their length. • Read the text and apply judgment for the specific interchange design. • It's sufficient as a guide. The engineer needs to consider the context of the site and use good judgment. • Grades and other geometric features need to be considered. • The table does not explain any factors involved. Question: To what portion of the ramp is the ramp design speed applicable? (both surveys) Answer Options State DOT Responses Local Agencies, University Researchers, and Private Sector Responses Combined Responses Freeway mainline ramp terminal 65.0% (13) 61.3% (19) 62.7% (32) Ramp proper 80.0% (16) 83.9% (26) 82.4% (42) Crossroad ramp terminal 25.0% (5) 41.9% (13) 35.3% (18) Between the two surveys, responses were similar for the ramp proper and freeway mainline ramp terminal. However, only 25 percent of state DOT respondents indicated that the ramp design speed applied the crossroad ramp terminal compared to 42 percent of respondents from the local agency, university, and private sector survey.

43 Additional comments from state DOT respondents include: • Ramp proper design speed should be from 50 to 85 % of mainline design speed (70%-85% is preferred). Loops are the exception - we use a minimum of 25 mph (36 degree curve); 30 mph is preferred (but rarely attainable) for exit loops. • Exit ramps are typically going from the freeway speed to a stop condition, while on the entrance ramp it is typical to assume to go from a 15 mph start speed to the freeway speed. Additional comments from local agencies, university researchers, and private-sector practitioners include: • The ramp proper is the only area not controlled by the roadway on either end of the ramp. Probably should just be a minimum for the type of ramp. Say 30 mph for a diamond ramp, for example. • Storage for freeway meter and crossroad stop. • The speed at the mainline/ramp gore areas is most applicable and important for both entrance and exit ramps. Sufficient distance to the terminal and corresponding design speed must be provided on entrance ramps for acceleration to main line and deceleration on the exit ramps to the crossroad. • It is assumed that the crossroad ramp terminal design speed is a function of the stopping distance needed and traffic control at the crossroad. The freeway mainline ramp terminal is a function of the deceleration distance needed to reach the design speed for the ramp. • I've seen ramp terminal merge points for 65 mph freeways that dump vehicles into mainline traffic with only the 30:1 taper length and no merging length. Vehicles need distance to get up to speed and then distance to merge as well. • The ramp speed is dependent on the function of the ramp. Freeway ramp terminals should have a design speed more indicative of the freeways speed, while the ramp proper of the same ramp may be reduced after allowing for appropriate distance to slow down. • The ramp design needs to be developed as a system of all the three elements above as well as the design of the roadway from which the ramp exits and onto which the ramp enters and the type of traffic control device. • Ramp speed should ideally apply throughout the whole ramp. Stop-controlled intersection at exit ramp may allow for lower design speed. • We verify the speed transition areas on the ramp are designed to the highest anticipated speed in that location. We establish a design speed profile to do this. The profile is based on both horizontal and vertical geometrics. One of our client's has a very specific procedure we are to follow to do this. We have applied this to other projects to improve our design. • The speed transition must be considered. We use different speeds for different portions of the ramp. i.e., Usually 10 mph less than mainline for the first curve, transitioning to 30 mph if there is a stop condition at the crossroad terminal. Maintain higher speed for system ramps.

44 Question: Does your agency’s policy for selection of ramp design speed differ between ramp types, ramp configurations, or some other factor? (State DOT survey) Answer Options Yes No Ramp types (exit ramp vs. entrance ramp vs. freeway-to-freeway ramp) 15% (3) 85% (17) Ramp configurations (diamond ramp vs. loop ramp vs. directional ramp) 50% (10) 50% (10) Other (Please specify in comment box below) 10% (2) N/A Seven respondents provided further comment, mostly related to design speed practices on loop ramps. Comments included: • Loop ramp design speeds are generally 25-30 mph. This does not meet our guidelines (50%-85% of mainline design speed). We use longer deceleration lengths (sometimes adjusted when there are heavy truck volumes) and longer spirals on exit loops. • Loop ramps utilize 25-30 mph design. • Reference the text rather than chart for loop ramps (min 25 mph). • Physical/environmental constraints and right-of-way can have an impact. • Loop ramp design speed selection is based more on engineering judgment and site conditions; diamond ramps generally adhere to standard layouts and design criteria. Standard exit and entrance ramps have a 5 mph design speed difference. Additionally, freeway-to-freeway ramp speed differential would be limited, whereas other types would allow for more difference between mainline and the ramp. • We always start with Table 10-1 then adapt as necessary as funding dictates. (hard to design a loop ramp at 60 mph). Question: Should the practice for selection of ramp design speed differ between ramp types, ramp configurations, or some other factor? (local agencies, university researchers, and private sector survey) Answer Options Yes No Ramp types (exit ramp vs. entrance ramp vs. freeway-to-freeway ramp) 69.0% (20) 31.0% (9) Ramp configurations (diamond ramp vs. loop ramp vs. directional ramp) 73.3% (22) 26.7% (8) Other (Please specify in comment box below) 20.0% (6) N/A Eleven respondents provided further comment, including: • Large truck turn over potential. • Exit and entrance ramps should be handled differently. Diamonds vs. loops vs. directional are clearly different in configuration and design speed expectation from motorists. • Only freeway-to-freeway cases since increased speed values are found. • Need to consider design guidance for non-traditional ramp designs, e.g., diverging diamonds.

45 • Freeway-to-freeway ramps are often intended to maintain mainline speeds, while entrance or exit ramps are near mainline speeds. Exit and entrance ramps also may have a perceived design speed much higher than intended, creating safety concerns. Ramps would benefit from having perceived design speeds lower or at most equal to the posted mainline speed. • The design of the ramp as previously mentioned also depends on the interchange form, location, and type of interchanging facilities. • Design speeds on the freeway and crossroad should also be considered. Traffic control on crossroad is also a factor. Freeway-to-freeway ramps should typically have higher design speeds. Ramp configuration helps to form driver expectations (for example drivers expect low speeds on a loop ramp). • Higher driver speed expectations for outer ramps as opposed to loop ramps. • Type of control at the crossroad. • Freeway-to-freeway is typically a fully directional movement with higher speeds. Loop ramps will typically have the lowest speeds. The geometry of a diamond interchange can allow for higher speeds; however, you need to consider the length, queuing, and stop condition. • Grades and other horizontal restrictions. Question: Does your agency’s policy specify a single design speed for a ramp as a whole or is the design speed allowed to vary along a ramp? (State DOT survey) Single design speed for entire ramp 30.0% (6) Design speed may vary along a ramp 60.0% (12) Depends on situation (Please explain below) 10.0% (2) Six respondents provided further comment, including: • The design speed of the initial part of the ramp (ramp proper) is to be 50%-85% of the mainline, and the speed of the ramp approaching the crossroad (the terminal curve) should be 50%-85% of the main ramp speed. The intent is to provide a “stepped down approach” on exits, where we think this is most crucial. • Physical/environmental constraints and right-of-way (ROW) can have an impact. • Diverging diamond interchanges use different design speeds for the on-ramps and off- ramps depending on alignments. • If approach road terminal is stop condition, vary from ramp proper to stop condition. • For example, at a stop condition at the end of a diamond exit ramp, an intermediate speed will be used between the controlling exit curve and the stop condition (as vehicles are decelerating) that would control the geometrics. This is a much more economical approach.

46 Question: Should a single design speed be specified for an entire ramp or should the design speed vary along the ramp? (local agencies, university researchers, and private sector survey) Single design speed for entire ramp 24.1% (7) Design speed may vary along a ramp 51.7% (15) Depends on situation (Please explain below) 24.1% (7) Twelve respondents provided further comment, including: • Design speed may differ, depending on the design speed of the termini. (e.g., free-flow termini should have high design speed, stopping termini may have lower design speed...the ramp design speed should be utilized to accommodate this transition). • The design speed refers to the main body of the ramp. At the beginning and ending sections, the design should be consistent in terms of speed. • For short ramps a single design speed may be ok, but for others it should be variable. • Both safety and operations are compromised when the design of a ramp near merge/diverge points are not close to the mainline operating speeds. Cloverleaf ramps create insufficient weaving distances for adequate design speeds. • Ideally one design speed is used, but site constraints and intersecting roadways will alter the selection. A freeway to local road ramp will require a higher design speed when exiting the roadway, but may have a reduced design speed near the terminus of the ramp that connects to the lower-speed local roadway. • The physical and environmental constraints may dictate the geometry of the ramp which may require a variable design speed. • For simplicity I would say that the design speed should be the same for the entire ramp. However, there are special cases such as a stop condition on a crossroad. Ramps are typically short in length so varying the design speed along the length does not seem beneficial. • In situations where drivers will be expected to alter their speeds (i.e., approaching a stop- controlled intersections), a progression in design speeds is advisable to encourage this behavior. • See previous comments about calculating design speed profiles along the ramp. • Transition curves are often used near the gore areas that have a higher speed; however, I would recommend a single speed for the ramp proper. • Length of ramp and geometric features may suggest two or more ramp speeds are appropriate. • Depends on length of ramp.

47 Question: Is your agency considering any change in its policy for selection of ramp design speed? (State DOT survey) Yes 5.0% (1) No 95.0% (19) Only one of the twenty respondents indicated that their agency is considering a change in its policy for selection of ramp design speed. That respondent indicated that the state DOT was updating its design policy to include ramps. No other specifics about the policy change were given. The survey included a follow up question that asked if there was any performance information or empirical data that has led the agency to consider changing its ramp design speed policy. All respondents answered “no”. Question: Do you believe any changes to the AASHTO Green Book policy for selecting ramp design speed are needed? (both) State DOT Responses Local Agencies, University Researchers, and Private Sector Responses Yes 35.0% (7) 64.3% (18) No 65.0% (13) 35.7% (10) Comments from state DOT respondents included: • Additional explanation regarding exit ramps and the need for using “stepped down” approach in selecting speeds. • There are minor conflicts between the table and the text so greater clarification and guidance is needed. • Better explanation of applicable design speed at terminals. Louisiana requires a minimum of 700 ft auxiliary lane (where ramp and mainline meet). We have seen operational problems for the 300-500 gap acceptance for entrance ramps. • The information in the Green Book could be clearer. • More considerations should be listed as guidance for considering alternative or lower speeds then are listed in Table 10-1. • Possible changes/clarity: Current publications are indicating that design speeds are not operating speeds see FHWA Memorandum 20151007, HIPA-20, Relationship Between Design Speed and Posted Speed. Do the acceleration/deceleration tables reflect highway speed or design speed?

48 Comments from local agencies, university researchers, and private-sector practitioners included the following: • Generally do not use this information from the Green Book. • I teach highway design, and have always had difficulty effectively conveying the ramp design speed procedure to students, largely because I’m not sure if I fully understand it myself. • Selecting design speed for freeway-to-freeway vs freeway to secondary roads ramps. • Additional notes and consideration should be included for weaving length and the actual ability to accelerate/decelerate to and from the design speeds. • Allow the designer more discretion to change design speeds along a ramp. • See ITE Freeway and Interchange Geometric Design Handbook (Leisch, 2005), Chapter 3, pages 99-103. • More clarity for how the ramp design speeds were calculated and how to go about addressing unique situations such as ROW constraints. • More guidance on when to select low, middle, or high range of values is needed. Special designs such as the diverging diamond interchange should also be considered. • I recommend investigating the design speed profile calculations in use. It would be interesting to see how a design speed profile on a ramp compares to operating speed profiles on a built ramp. • Ramp configuration and type should factor into design speed. • More guidance should be given on designing the ramp to fit the context of the area. • It appears there is inherent flexibility in design parameters, but it may be good to give the designer some discussion regarding other factors that may way their respective design decisions. • If there is recent research that justifies it sure; however, I think it will be difficult to be more specific and engineers need to consider the context and use the flexibility provided. • I feel that the “ramp” needs to be better described in the Green Book. What part of the ramp is controlled by the main lanes if any, and similarly for the intersecting street. Sometimes there is confusion on what curve radii to choose, and if it meets the design speed, based on where the curve is located along the ramp. Question: Does your agency’s policy for signing advisory speeds on ramps relate directly to the design speed of the ramp? (State DOT survey) Yes 33.3% (6) No 38.9% (7) In some cases 27.8% (5)

49 Comments included the following: • Sometime it is based on a field check. • We utilize the MUTCD and PA Publication 236. • Districts will ball bank the curves. • Advisory speed signing, if necessary, on our ramps is determined by running a ball bank indicator test and signing appropriately. • Signing for advisory ramp speed is based on the curve geometry of the ramp proper from the initial design, adhering to the Green Book Superelevation Tables (3-10b). If later in the life of the ramp the adequacy of the signing is in question, a ball bank indicator will be used to determine the appropriate advisory speed. • Typically use MUTCD guidance related to ball bank values and posting advisory speeds when needed. • May be based on ball-banking (e.g., loops). Question: Has your agency encountered any specific tort liability concerns on ramps whose design speeds were selected with your existing design policy that this research should address? (State DOT survey) All 20 respondents answered “no” to this question. Question: Are you aware of any specific tort liability concerns related to ramp design speeds that this research should address? (local agencies, university researchers, and private sector survey) Yes 16.7% (5) No 83.3% (25) The following additional comments were provided: • Change ramp design speed and ramp length. • Varying ramp speeds. • I am not aware of any. Some old (50 year or so) designs did not anticipate the volumes or speeds associated with today's traffic. • Use of improper superelevation. • I would try to address weave length as part of the ramp design speed. • Design speed vs. posted (advisory) speed. • Yes, not all of our clients have tort immunity or a low dollar cap. The design speed impacts the clear zones and where we can then place roadside objects. Some clients are requiring level III barrier warrant analysis using RSAP software (the prior version as the new windows based version has too many bugs) to verify the benefit cost has been explored and the option to be built will be defendable in court.

50 Question: Has your agency/organization used the safety predictive models from Chapter 19 of the 2014 Supplement to the AASHTO HSM in the design of ramps and/or the selection of ramp design speeds? (both surveys) State DOT Responses Local Agencies, University Researchers, and Private Sector Responses Yes 10.0% (2) 24.1% (7) No 90.0% (18) 75.9% (22) State DOT respondents reported the following: • Guidance on the applicability and use of the HSM is being developed. However, the HSM can be and has been utilized to analyze project specific decisions occasionally. • Design of ramps. Some limitations to HSM. • Not yet, but may in the future. Comments from local agencies, university researchers, and private-sector practitioners included the following: • Very good experience to date, but there are subtle issues to always consider what the HSM predictive models bring out. • Agency recently introduced HSM analysis into design. Awaiting early results. • Need additional research. • We have not used it in the design of ramps or selection of ramp design speeds, but we have used the HSM for a variety of other tasks. For some designers, the predictive method can be too complicated and/or difficult to follow. Learning the predictive method becomes too overwhelming so they choose not to apply it. • Yes, for some clients; however, the client we utilize design speed profiles and RSAP software to aid in the design of the ramps does not permit us to use HSM and its methodologies. I find this very odd. • The results are not always intuitive. Not taking grade into account is a significant issue with the crash prediction for ramps. Question: In the past 5 years, how many design exceptions for ramp design speed (or design exceptions for horizontal alignment that are equivalent to a design exception for ramp design speed) has your agency initiated and approved? (State DOT survey) In the past 5 years, how many ramp/interchange design projects has your organization been involved with that required design exceptions for ramp design speed (or design exceptions for horizontal alignment that are equivalent to a design exception for ramp design speed)? (local agencies, university researchers, and private sector survey)

51 Answer Options State DOT Responses Local Agencies, University Researchers, and Private Sector Responses None 38.9% (7) 65.4% (17) 1 or 2 38.9% (7) 11.5% (3) 3 to 5 5.6% (1) 11.5% (3) 6 to 10 5.6% (1) 7.7% (2) More than 10 11.1% (2) 3.8% (1) State DOT respondents provided the following additional information: • We are reconstructing a full cloverleaf interchange (in phases) into a partial cloverleaf with a flyover ramp. As part of the final phase, we are rebuilding an existing loop ramp with improved geometry. The connection to the loop uses a “stepped down” approach to transition from a 60 mph to 25 mph condition, including an exception from our standard superelevation. • Physical/environmental constraints and ROW have impacted. • We do not have that information statewide, but I have not seen any in my responsible area (about half the state). • This question is not able to be accurately answered because we do not have the data. • Historic Property Impacts. • Believe there have been 2-3 due to horizontal sight restrictions around barrier wall on inside. • Sight distance over the barrier rail on a flyover. Typically for on- and off-ramps we design parallel so we can add or subtract length as necessary. • Loop ramp where we had topography and geometric constraints. Existing ramp had multiple curve radius and superelevation. An effort was made to create a more consistent superelevation and curve radius that did not fit the Green Book values. • Specifically for ramp loops where there are existing ROW constraints. Comments from local agencies, university researchers, and private-sector practitioners included: • One of two issues: 1) An Interstate to Interstate ramp (system ramp) should be designed for 50 mph, but because of ROW constraints ends up being a 30 mph loop. 2) Most of the deviations are for diamond ramps where the curve closest to the crossroad cannot meet the 30 mph minimum set for all service interchange ramps. • None that I can recall. • Typically dictated by ROW constraints, which would have significant impacts to the overall cost and schedule of the projects. • System interchanges in an urban area where the agency had a guideline of 50 mph minimum design speed which was impossible to achieve.

52 • Our office has not been involved with projects that required a federal design exception from the AASHTO criteria for ramp design speed. Several projects have required exceptions to the project criteria particularly as it relates to horizontal stopping sight distance. • We do not want design exceptions for this. We avoid whenever possible and have been successful doing so thus far. • None that I am aware of. • Interchanges in tight urban areas have required design exceptions. We do not get design exceptions for design speed, we get design exceptions for each element that does not meet the design speed, i.e., horizontal curve, vertical curve, superelevation, etc. Question: Are there any other factors that affect the determination of ramp design speed that have not been addressed in this survey? (both surveys) Respondents from state DOTs provided the following feedback: • Strengthened discussion on considerations at depressed interchanges (exit ramp sight lines, crossroad sight lines, etc.,) would be beneficial. Also, there should be discussion of the effects of vehicle queues at exit ramps. • The design speeds are the same as the answers already provided, but maybe in other States the ramp operation (i.e., freeway-to-freeway direct connectors or “drop ramps” in the medians for HOV or HOT lanes) might have different design speeds. • For ramp termini other than free-flow the Green Book is not real clear on the design speed of ramps (e.g., stop controlled, signal, or roundabouts). Would like to see continuous deceleration and/or acceleration discussed or addressed. • We have moved to 80 mph posted speeds on much of our Interstate system. These speeds are off the ramp acceleration/deceleration charts. We are having questions as to whether the design speed of the Interstate and thus the ramp terminal should be moved to 80 mph. Note that the entire mainline route ball banks at 80 mph or greater. Should we be keeping the design speed of 70 mph for our main line or moving to 80 mph (which may give higher ball bank values and operational speeds)? Should we be doing design exceptions for routes that do not meet the 80 mph design values of the Green Book, but ball bank at 85+ mph? Does not meeting the ramp deceleration/acceleration lengths constitute a design exception for the design speed of the ramp? The grade factors seem questionable. Currently working through an IC design where entrance ramp that intersect a freeway section has a 5% grade. Green Book values are giving us ramp lengths of excess of 3000 ft to meet a 75 mph freeway posted speeds. Seems excessive in that the current ramps are about 1200 ft with no identifiable operational issues (i.e., mostly due to lower volume Interstate, 12000 ADT, crossroad at about 3000 ADT, and freeway speeds appear to be operating slower then posted speed). Truck rollover—it is rare here, but at what point does it become a determining factor in relationship to design speed?

53 Comments from local agencies, university researchers, and private-sector practitioners included: • Toll plazas. • Design exceptions. • There are non-technical factors (you have included two; environmental and ROW costs), such as political engineering (the other PE) where some developer/unit of government wants an interchange “squeezed in” where it should not be allowed. There are also public involvement factors (often personal interests) that may influence interchange design. • Urban vs. rural design considerations. More could be presented relating to conceptual design approaches especially for off-ramps. Revisit difference in taper length estimate process—Green Book vs. MUTCD. Many subtle ramp design issues cannot be readily evaluated in terms of safety performance using ISATe. Look at differences in stopping sight distance between underpass ramp alignments vs. overpass ramps. • Resulting weaving length and resulting speed determined by ramp length. • Innovative geometrics such as diverging diamond or roundabout interchanges should be considered. I answered this survey based on my prior 14 years of experience as a Highway Engineer. • Toll plazas need to be considered. They have deceleration and acceleration in advance and after the plaza. The plazas may also have differing speed lanes. Some lanes are 5 or 15 mph, and others are stop controlled. Some are just open gantries with all electronic toll collection and would not impact the speeds. We have some ramps that connect to collector-distributor (CD) roads instead of directly to the mainline. The CD roads are usually about 10 mph lower design speed of the mainline. We design the ramps to tie to the appropriate road's design speed (we would use the CD design speed in our design speed profile). I also find it interesting that some agencies specify an emax for the entire ramp, but other agencies mix emax rates. The one that mixes emax has a larger value near the higher-speed road and along the ramp proper, but as the ramp's curves approach a slower crossroad ramp terminal, the emax may drop by 2% to 4% for those curves. I don't know why some agencies do this and others do not. Making the switch to a lower emax would certainly help the designer meet rollover criteria in the gores. Some of our ramps have barrier along them. Ramps are usually curved so it may be difficult to get higher design speeds in the tighter curves due to sight distance issues with the sight line hitting the barriers. Same could apply to a ramp just downstream or starting just in advance of a bridge where the bridge support could be in the sight line. Are our ramps being over designed due to having better pavement surface and better tires (more friction) than what have been included as the basis for the design resources we follow? One more for you. With future autonomous vehicles, do we need as much room for the ramps? Can we eliminate shoulders and eliminate sight lines? Will the vehicles be able to sense through or around barriers? Will they know a back of queue is ahead? Can we shorten acceleration/merging distances because the vehicles can talk to each other and set gaps? • Sight distance. Lane drop or lane add. Climate. Terrain. • Crash history.

54 Discussion Points from Web-Based Survey Several key points that can be drawn from the web-based survey are as follows: • Committee members, local agencies, university researchers, and private-sector practitioners overwhelmingly think policy should differ by facility type and ramp configuration, but most states do not have policies that vary by those factors. • Most state DOTs use Green Book or similar policies and have no plans to change, but most academics and private-sector practitioners would like to see changes to the policy. • The factors most frequently selected as having high or medium importance when determining ramp design speed include: - The design speed or operating speed of the highway. - Ramp configuration. - Design/operating speed of intersecting roadway. - Ramp grade. - Degree of physical constraints at site. - Degree of environmental constraints at site. - ROW costs. • Over 50 percent of respondents indicated that Table 10-1 from the 2011 AASHTO Green Book is insufficient for selecting ramp design speed. • There is no consensus on what portion of the ramp to which the ramp design speed is applicable. • A majority of the respondents think that the practice for selection of ramp design speed should differ between ramp types and ramp configurations. • There is no consensus on whether a single design speed should be specified for an entire ramp or whether the design speed should vary along the ramp. • No state DOTs that responded indicated any specific tort liability concerns on ramps whose design speeds were selected with existing design policies. • Other factors to consider when providing guidance on the selection of an appropriate ramp design speed include: - Toll plazas. - 75+ mph speed limits. - Innovative geometrics such as diverging diamond or roundabout interchanges. Interviews with Selected Professionals As a follow up to the online survey, the research team interviewed four survey respondents by telephone to explore issues covered by the survey in greater depth. Key discussion items from the interviews are summarized as follows: • Green Book Table 10-1 is insufficient. Suggested modifications or changes included: - The table needs to be expanded to show highway design speeds of 80 mph and 85 mph, perhaps even higher as speed limits are being raised to 75, 80, and 85 mph.

55 - The table needs to be expanded to show different ramp types (exit, entrance, freeway- to-freeway) and ramp configurations (loops, diamonds). - The table does not detail the three main areas of the ramp and should be expanded to show design speed (or range of design speeds) for: • The ramp terminal at the crossroad by intersection control (stop condition, signal, roundabout, etc.). • The ramp terminal at highway or freeway. • The ramp proper based on interpolation of speeds between the two terminals. - Ramp metering should be a major consideration in designing entrance ramps, noting an existing metered ramp in Utah that is approximately 4,000 ft long with a 45 mph curve near the freeway entrance. The ramp was designed to accommodate queuing during ramp meter operations; but when the meters are not on, drivers perceive the ramp to have a much higher design speed due to the long, tangent ramp proper preceding the curve. - There should be a varying ramp design speed for service interchanges. Freeway-to- freeway ramps should have a single design speed equal to that of the facilities to which they connect. For service interchanges, the design speed at the freeway mainline ramp terminal should be 50 mph (approximately 20 mph lower that the design speed of the freeway). • Factors to consider when selecting a ramp design speed: - Additional guidance on ramp design for diverging diamonds and single point urban interchanges (SPUIs) is needed. Both normally utilize signalized control at the crossroad intersection. - Vertical alignment should be taken into consideration along with the type of design vehicle, especially trucks on significant up or down grades. - Concerns about use of adverse superelevation on loop ramps should factor into the selection of ramp design speed. - Additional guidance is necessary for weaving lengths at auxiliary lanes and on CD roads used in cloverleaf interchange design. Utah DOT is trying to eliminate all cloverleaf designs due to extremely short weaving lengths (500 ft for some), noting there are a high number of crashes on those short weaving lengths, and crash frequency diminishes greatly when weaving lengths are greater than 1,000 ft. Utah DOT uses a lot of SPUIs. - Illinois DOT is trying to eliminate all cloverleaf interchanges due to problems with large truck turnover potential. Where it is cost prohibitive to reconstruct the interchange and truck turnover crash data is high, they increase the superelevation rate on the loop ramp (up to 10% superelevation) to reduce those types of crashes. They considered using spirals, but generally there’s not enough ROW. Part of the problem is that the trucks disregard the posted advisory speed limit on the ramp. - The use of speed profiles is employed based on Illinois Tollway Traffic Barrier Guidelines procedure. These are completely different than what Illinois DOT uses.

56 The vertical alignments are also checked against the horizontal speed profiles to ensure consistency in vertical/horizontal design. • Ramp safety and use of the HSM: - When using the HSM in ramp design, a few anomalies have been identified that do not seem logical. In comparing the “no build” and “build” alternatives, it was found that when the design speed increased, the number of crashes or severity of crashes also increased (or vice versa). The HSM also showed an improved travel time which seemed odd because the ramp was such a small portion of the overall design. - Both Ohio and New Mexico DOTs use AASHTO’s HSM. - Several respondents noted using FHWA’s Interchange Safety Analysis Tool for interchange safety analysis. - Safety related to the speed differential between the ramp and mainline should be addressed. - The HSM is not used in designing ramps. The Green Book and the Highway Capacity Manual are used to design ramps, and the HSM may be used to check the design for federal funding purposes; but the HSM is idealistic and not really effective. • Design exceptions: - Design exceptions for ramp design speed are most often at the curve closest to the crossroad and are related to the superelevation of the curve. Ramp design superelevation is 8 percent maximum, and frequently there is not enough tangent distance to transition through the curve and back to the grade of the crossroad if the superelevation is set at 8 percent. To address this issue, the design criteria has recently been revised to allow for a 6 percent maximum superelevation rate and 25 mph design speed for the curve closest to the crossroad. - Designers should not seek too many design exceptions. It can lead to problems. For example, in the design of the first designated Interstate through an urban area in the late 50’s, Illinois DOT obtained design exceptions for several slip ramps and short ramps where ROW was limited. Increased traffic and speeds 10-20 years later caused crash rates to rise, and it was extremely costly and difficult to remedy the situation. 2.3 Summary of Key Issues and Limitations of AASHTO’s Current Policy on Ramp Design Speed Based upon the literature review, the survey results, and interviews with selected professionals, this section summarizes the key issues and limitations of AASHTO’s current policy on selection of an appropriate ramp design speed. The key issues and limitations are not presented in any particular prioritized order: • Green Book Table 10-1 is insufficient for selecting ramp design speed: Green Book Table 10-1 is the centerpiece of AASHTO’s current guidance on selecting an appropriate ramp design speed for the design of all ramp types (i.e., entrance, exit, and freeway-to- freeway ramps) and ramp configurations (e.g., diagonal, loop, semidirect, direct). However, more than 50 percent of respondents to the surveys indicated that Green Book

57 Table 10-1 is insufficient for selecting ramp design speeds. Rather than Green Book Table 10-1 addressing entrance and exit ramps separately and the various types of ramp configurations, the table provides upper, middle, and lower range values of ramp design speeds. AASHTO policy then provides further guidance regarding desired, practical, and/or minimum ramp design speeds corresponding to the upper, middle, and lower speed ranges for different ramp configurations. Suggested revisions to the table included the need to address different ramp configurations and entrance and exit ramps. • No consensus on the portion of the ramp to which the ramp design speed applies: Existing guidance states that the guide values for ramp design speed apply to the sharpest, or controlling, ramp curve, usually on the ramp proper, and the ramp design speeds do not pertain to the ramp terminals. However, there is no consensus among practitioners as to what portion of the ramp the ramp design speed applies. Approximately 82 percent of survey respondents indicated that the ramp design speed is applicable to the ramp proper; approximately 63 percent of survey respondents indicated that the ramp design speed is applicable to the freeway mainline ramp terminal, and approximately 35 percent of survey respondents indicated that the ramp design speed is applicable to the crossroad ramp terminal. • Little guidance is provided on how the selected ramp design speed is to be used to develop a balanced design among all components of a ramp: AASHTO policy states that the ramp terminals are to be properly transitioned and provided with speed-change facilities adequate for the speed of the highway being considered. First, let us consider how this statement applies to freeway mainline ramp terminals. In addition to Green Book Table 10-1, Green Book Tables 10-3 and 10-5 provide the minimum acceleration lane lengths for entrance ramps and minimum deceleration lane lengths for exit ramps, respectively. The minimum acceleration and deceleration lane lengths are tied to the highway design speed and initial speeds at the beginning of the SCL and final speeds at the end of the SCL. Consistent with the guidance provided for using Green Book Table 10-1, Green Book Tables 10-3 and 10-5 do not specify a corresponding design speed for freeway mainline ramp terminals, nor is there guidance tying the conditions specified in Green Book Tables 10-3 and 10-5 for acceleration and deceleration lane lengths to the ramp design speed values in Green Book Table 10-1 for the controlling curve. Beyond AASHTO policy stating that the ramp terminals are to be properly transitioned and provided with speed-change facilities adequate for the speed of the highway being considered, there is no real guidance on how this is to be achieved for freeway mainline ramp terminals. The other type of ramp terminal is the crossroad ramp terminal. For exit ramps, Green Book Figure 9-2 (Figure 7) provides some guidance regarding how the functional area of an intersection, consisting of three basic elements (i.e., perception-reaction decision distance, maneuver distance, and queue storage distance), relates to speed transitions approaching a crossroad ramp terminal. For entrance ramps there is not any corresponding guidance in the Green Book.

58 Figure 7. Elements of the Functional Area of an Intersection (AASHTO, 2011) The guidance in Chapter 10 implies the use of the vehicle performance curves for acceleration and deceleration distances for passenger cars for coordinating the design of the ramp terminals with the controlling curve on the ramp proper. However, use of the performance curves provided in Chapter 2 of the Green Book is not explicitly stated or recommended in Chapter 10 as it relates to the use of Green Book Tables 10-1, 10-3, and 10-5. Also, associated with the need to provide transitions between the controlling curve and the ramp terminals, there may be connecting portions of the ramp that are tangent or have a radius larger than the controlling curve that have to be accounted for in the design of the entire ramp. Little guidance is provided within the Green Book on how the ramp terminals and adjoining sections are to be properly transitioned to connect with the controlling curve over the entire length of the ramp for a balanced ramp design that meets driver expectations. • No consensus on whether the ramp design speed should vary along the ramp: Approximately 24 percent of survey respondents indicated that a single design speed should be specified for the entire ramp, and approximately 52 percent of survey respondents indicated that the ramp design speed may vary along the ramp. In many ways this relates to the need provide proper transitions between the ramp terminals, adjoining sections, and the controlling curve. • Important factors to consider when selecting ramp design speed: The factors most frequently identified by survey respondents as having high or medium importance when determining ramp design speed include: - The design speed or operating speed of the highway. - Ramp configuration. - Design/operating speed of intersecting roadway. - Degree of physical constraints at site. - Degree of environmental constraints at site. - ROW costs. • Guidelines needed for higher ramp design speeds: Many states are posting freeway speed limits as high as 75, 80, and 85 mph. Guidelines for ramp design speeds to address such conditions are necessary as the current guidelines only address highway design speeds up to 75 mph. • No specific tort liability concerns with current approach to selecting ramp design speed: None of the state DOTs that responded to the survey indicated any specific tort

59 liability concerns related to ramps designed using existing design policy for selecting ramp design speeds. These key issues and limitations were considered in developing the Phase II work plan for this research and the guidelines that were developed.