Design Guidance for Freeway Mainline Ramp Terminals (2012)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

8This section first summarizes the current AASHTO design policy for freeway mainline ramp terminals and then summa- rizes existing literature on geometric design and safety per- formance, vehicle performance characteristics, human factors considerations, and operational performance issues related to freeway mainline ramp terminals. 2.1 AASHTO Design Policies for Speed-Change Lanes and Freeway Mainline Ramp Terminals AASHTO defines an SCL as an auxiliary lane, including tapered areas, primarily for the acceleration or deceleration of vehicles entering or leaving the through-traffic lanes. The 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 and do not necessarily imply a definite lane of uniform width. An SCL should have sufficient length to enable a driver to per- form an appropriate change in speed between the highway and the turning roadway in a safe and comfortable man- ner. Green Book design values for freeway ramps rely heav- ily on research conducted in the late 1930s and early 1940s. The 1965 American Association of State Highway Officials (AASHO) Blue Book provides more information on the design guidelines of SCLs than the later versions of AASHTO design policies (AASHO, 1965; AASHTO, 1990; AASHTO, 1994; AASHTO, 2001; AASHTO, 2004). Basic components that make up the SCLs are described below. Acceleration Lane: An acceleration lane begins where the driver transitions from the ramp curvature to the flatter geom- etry of the SCL. The lane allows for acceleration and checking for gaps in the freeway traffic. The length of the acceleration lane is determined on the basis that merging vehicles should enter the through lane at a speed approximately equal to the running speed of the freeway. The acceleration lane ends at the beginning of the taper. The taper section begins where the width of the SCL becomes less than 12 ft and ends at the point where the SCL has fully merged with the freeway through lane. The taper section is not included in the length of the SCL. The 1965 AASHO Blue Book describes the factors for the calculation of the length of an acceleration lane: (a) the speed at which drivers merge with through traffic, (b) the speed at which drivers enter the acceleration lane, and (c) the manner of accel- eration. The acceleration rates are determined from the studies done in the late 1930s. AASHO notes that the acceleration rates are based on passenger vehicle operation, and that trucks and buses require much longer distances but that providing such lengths would be out of reason. It assumes that slower entry is unavoidable and is generally accepted by the public. AASHO also mentions that if there are a substantial number of heavy vehicles entering, then lengths may be increased or, if feasible, the entries may be located on downgrades. The 1965 Blue Book provides the most detailed information on how the current design values in the 2004 Green Book are determined, but it should be noted that for some combinations, the 2004 Green Book acceleration lane lengths differ slightly from the 1965 Blue Book values. Also, AASHTO’s current geometric design guidance is based on a single set of operational assumptions that can be characterized as representing free or unconstrained merging behavior. Deceleration Lane: A tapered section of roadway, where the pavement transitions from the freeway edge of pave- ment to a point where the width of the SCL is 12 ft, precedes a deceleration lane. From the end of the taper to where the driver transitions onto the ramp curvature, the deceleration lane provides a means for vehicles to decelerate from the freeway speed to a speed commensurate with the control- ling feature of the ramp. Similar to the acceleration lane, deceleration lane length is based on three factors: (a) the speed at which drivers maneuver onto the auxiliary lane, (b) the speed at the end of the decelera- tion lane, and (c) the manner of deceleration. The length of the deceleration lane is determined by the speed differential, S e c t i o n 2 Literature Review

9 which is the difference between the average running speed on the mainline and the speed on the sharp or controlling termi- nal curve on the turning roadway. The average running speed is used instead of the design speed, based on the assumption that the exiting drivers travel at the average running speed when highway volumes are low. All design lengths are based on passenger car operations. Even though the 1965 AASHO guide recognizes that trucks require longer distances to decelerate for the same difference in speed, AASHO indicates longer lanes are not justified because average speeds of trucks are lower than passenger car speeds. Two general forms of SCLs are the taper and parallel types. Below is the description of entrance and exit terminals for both types of SCLs. 2.1.1 Single-Lane Free-Flow Terminals, Entrances 2.1.1.1 Taper-Type Entrance A taper-type entrance as shown in Figure 1A merges into the freeway with a long, uniform taper with a desirable rate of taper of about 50:1 to 70:1 (longitudinal to lateral) between the outer edge of the acceleration lane and the edge of the through-traffic lane. The geometrics should be designed in a way that drivers attain a speed that is within 5 mi/h 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 sets this location as the point where the right edge of the SCL and trav- eled way are 12 ft apart. The length required for the vehicle to Figure 1. Tapered and parallel designs for entrance ramps (AASHTO, 2004).

10 achieve this speed is referred to as the acceleration length and is measured from the end of the governing curve on the ramp proper to where the right edge of the SCL and through lane are 12 ft apart. Adjustment factors are used to increase the rec- ommended acceleration lane lengths for ramps with positive grades and decrease acceleration lane lengths for ramps with negative grades. 2.1.1.2 Parallel-Type Entrance A parallel-type entrance as shown in Figure 1B provides an added lane of sufficient length to enable a vehicle to accel- erate to near freeway speed prior to merging. A taper is pro- vided at the end of the added lane. AASHTO recommends a taper length of approximately 300 ft for design speeds up to 70 mi/h. The distance needed for acceleration in advance of the merge point is governed by the speed differential between the operat- ing speed on the entrance curve of the ramp and the operating speed of the freeway. Green Book Exhibit 10-70 recommends minimum acceleration lane lengths for entrance terminals, applicable to both taper- and parallel-type entrances. Green Book Exhibit 10-69 provides minimum lengths for gap accep- tance. The larger value between the acceleration lane length and the gap acceptance length is used in the design of freeway mainline ramp terminals. In places where it is anticipated that the ramp and freeway will frequently carry traffic volumes approximately equal to the design capacity of the merging area, a parallel acceleration length of at least 1,200 ft plus taper is desirable. Where grades are present on ramps, minimum recom- mended acceleration lanes ideally will be adjusted in accor- dance to Green Book Exhibit 10-71. 2.1.2 Single-Lane Free-Flow Terminals, Exits 2.1.2.1 Taper-Type Exit A taper-type exit, as shown in Figure 2, provides a direct path to the ramp proper. The divergence angle is usually between 2 and 5 degrees. The length of the deceleration lane 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. 2.1.2.2 Parallel-Type Exit A parallel-type exit terminal, as shown in Figure 2, generally begins with a taper, followed by an added lane that is 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. Transition may be provided at the end of the deceleration lane if the ramp proper is curved. The taper portion of a parallel-type deceleration lane should have a taper of approximately 15:1 to 25:1. Green Book Exhibit 10-73 provides recommended minimum deceleration lane lengths for various combinations of design conditions for both taper- and parallel-type exit terminals. Where grades are present on ramps, minimum recommended deceleration lane lengths ideally will be adjusted in accordance to Green Book Exhibit 10-71. Key elements of the AASHTO Green Book model are described below. 2.1.3 Design Speed Design speed is the selected speed used to determine the various geometric design features of a roadway and should be consistent with the topography, anticipated operating speed, adjacent land use, and functional classification of the roadway. In some situations, AASHTO policy uses a speed other than design speed to determine the various geometric design fea- tures. In the case of freeway mainline ramp terminals, mini- mum acceleration and deceleration lane lengths are based upon operating speeds along the freeway and the entrance/exit curve of the ramp. In the case of acceleration lanes, AASHTO policy assumes that a vehicle merges onto the freeway at a speed of 5 mi/h below the operating speed of the freeway. For acceleration lanes, AASHTO policy notes that part of the ramp proper may be considered in the acceleration lane provided that the curve approaching the acceleration lane has a radius of approximately 1,000 ft or more. Similarly, with deceleration lanes, a portion of the ramp may be considered as part of the deceleration lane when the initial curve has a radius of 1,000 ft or more. Thus, AASHTO policy assumes that the operating speeds of vehicles on the ramp are affected by the horizontal alignment of the ramp when the curve radius is less than 1,000 ft. Figure 2. Parallel and tapered designs for exit ramps (AASHTO, 2004).

11 2.1.4 Sight Distance Sight distance along a ramp should be at least as great as the design stopping sight distance. Sight distance for passing is not needed. The sight distance on a freeway preceding the approach nose of an exit ramp should exceed the minimum stopping sight distance for the through-traffic design speed, desirably by 25 percent or more. 2.1.5 Grade and Profile Design A ramp typically consists of a ramp proper with an apprecia- ble grade, while the ramp terminals (at both the crossroad and the freeway) generally have flatter grades. AASHTO guidelines indicate that it is desirable to limit upgrades based on the design speed of the ramp as follows: • Design speed of 45 to 50 mi/h: 3 to 5 percent maximum upgrade, • Design speed of 40 mi/h: 4 to 6 percent maximum upgrade, • Design speed of 25 to 30 mi/h: 5 to 7 percent maximum upgrade, and • Design speed of 15 to 25 mi/h: 6 to 8 percent maximum upgrade. Ramp grades should be as flat as practical, but where topo- graphic conditions dictate, grades steeper than desirable may be used. One-way downgrades on ramps should be held to the same general maximums, but in special cases they may be 2 percent greater. 2.1.6 Vertical Curves Ramp profiles assume the shape of the letter “S” with a sag vertical curve at the lower end and a crest vertical curve at the upper end. If a crest or sag vertical curve extends onto the ramp terminal, the length of the curve should be determined by using a design speed between those on the ramp and highway. 2.1.7 Superelevation and Cross Slope AASHTO has established limiting values of superelevations and friction factors for different design speeds on open road- ways. It recommends the cross slope on portions of ramps on tangent normally to be sloped one way at a rate ranging from 1.5 to 2 percent for high-type pavements. 2.1.8 Gore Areas At an exit terminal, the ramp “gore” is the area downstream from where the shoulder of the ramp and the shoulder of the freeway intersect. The physical nose is a point upstream from the gore, having some dimensional width, occurring at the separation of the roadways. The width of the gore nose is typically between 20 and 30 ft, including paved shoulders, measured between the traveled way of the main line and that of the ramp. In a series of interchanges along a freeway, the gores should be uniform and have the same appearance to drivers, and the entire triangular area should be striped to delineate the proper paths on each side. The gore area and the unpaved area beyond should be kept free of obstruc- tions when possible to provide a clear recovery area. At an entrance terminal, the gore points downstream and is less of a decision area for drivers since the traffic streams are in separate lanes. 2.1.9 Ramp Traveled-Way Widths Traveled-way widths are governed by the type of operation, curvature, and volume and type of traffic. Design widths of ramp traveled ways are classified into three general design traffic conditions: • Traffic Condition A—mainly passenger vehicles, but some consideration to single-unit (SU) trucks. • Traffic Condition B—SU vehicles to govern design and some consideration to semitrailer vehicles. • Traffic Condition C—sufficient buses and combination trucks to govern design. 2.1.10 Left-Hand Entrances and Exits Left-side ramp terminals break up the uniformity of inter- change patterns and are also contrary to the concept of driver expectancy. Care should be exercised to avoid left-hand entrances and exits in the design of interchanges. 2.1.11 Ramp Terminal Profile Ramp terminal profiles should be designed in association with horizontal curves to avoid sight restrictions that will adversely affect operations. It is desirable to design a platform, at least 200 ft in length, on the ramp side of the approach nose or merging end. 2.2 Geometric Design and Safety Several studies investigated the relationship of geometric design elements and the safety performance of freeway main- line ramp terminals. Relevant findings are as follows: • 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).

12 • One study found the safety of entrance terminals is enhanced when 800-ft or longer acceleration lanes are provided (Twomey et al., 1993). Other studies indicate that a mini- mum SCL length of 1,230 ft, including the taper, is desir- able for comfortable merging and anything longer than 1,400 ft does not improve merging behavior (Hassan et al., 2006; Ahammed et al., 2006). • Drivers tend not to begin the acceleration process until they have a clear view of freeway right-lane traffic (Hunter et al., 2001). Even if a ramp is designed to allow vehicles to begin acceleration prior to reaching the painted nose, drivers typically do not begin the acceleration phase until they have a clear view of the mainline traffic (Hunter and Machemehl, 1999). • Entrance ramps with vertical profiles that limit ramp driver sight distance and had marginal SCL lengths exhibited sig- nificant ramp driver speed changes (Hunter et al., 2001). • Speeds in the right lane of the freeway were affected where there was inadequate sight distance and acceleration distance was short. This was particularly noticeable at high freeway and ramp volumes (Hunter et al., 2001). • Poor ramp geometry led to a more aggressive ramp merge beyond the painted nose (Hunter et al., 2001). • One study (Eustace and Indupuru, 2008) suggested a con- stant acceleration rate of 2.5 ft/s2 should be used in the design of acceleration lanes. Another study (Hunter and Machemehl, 1999) indicated that drivers accelerate from 0 to 2.9 ft/s2 on ramps with adequate length compared to undulating acceleration rates from ±5.9 ft/s2 on poorly designed ramps. Yet, another study (Hassan et al., 2006) concluded the 85th percentile maximum comfortable acceleration rate was 6.5 ft/s2. • Free-merge vehicles merge onto a freeway at arbitrary locations (Yi and Mulinazzi, 2007). • Platooned merge vehicles follow a natural smooth path and force themselves into the mainline traffic within a certain area of a merge lane (Yi and Mulinazzi, 2007). • For entrance ramps, a taper design causes vehicles to use a greater portion of the ramp than a parallel SCL of the same length (Fukutome and Moskowitz, 1960). • More ramp length is used to accelerate at low volumes than at high volumes (Fukutome and Moskowitz, 1960). • For ramps with adequate sight distance and SCL lengths, drivers typically have a smooth transition into the through- vehicle traffic. Conversely, ramps with vertical pro- files limiting the driver’s sight distance often cause abrupt speed changes and inhibit the flow of traffic (Hunter and Machemehl, 1999). • At locations with adequate ramp design and acceleration lane length, drivers tend to traverse the entire length of the SCL before entering the through traffic, but at sites where inadequate design is provided, drivers tend to merge more aggressively and do not use the provided space for fear of being trapped at the end of the dropped lane (Hunter and Machemehl, 1999). • Deceleration lanes of 900 ft or more reduce traffic friction on the through lanes and account for reduced crash rates (Twomey et al., 1993). • Taper designs allow the driver to clearly view the start of the deceleration lane (Garcia and Romero, 2006). • Even with long deceleration lanes, drivers start to decelerate before exiting the mainline. A reduction of 10.5 mi/h on the mainline was observed (Garcia and Romero, 2006). • Deceleration lane length is indirectly proportional to early exiting and directly proportional to overtaking. As the decel- eration lane length decreases, overtaking reduces and early exits increase (Garcia and Romero, 2006). • No relation was found between lane length and deceleration rate (Garcia and Romero, 2006). • Longer deceleration lanes are more likely to reduce injury severity (Wang et al., 2009). • 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 identified to contribute to the safety performance of interchange ramps and SCLs include the fol- lowing (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, and • Length of SCL. Several studies evaluated the adequacy of current policies for freeway-ramp design. Harwood and Mason (1993) concluded that current AASHTO policies for freeway-ramp design are adequate so long as drivers adjust their speeds to lev- els that are less than or equal to the design speed. Where problems are anticipated, geometric design changes may be appropriate in order to increase the ramp design speed or decel- eration distance available to the driver. Koepke (1993) reviewed both taper and parallel entrance and exit design and concluded that current design practices are acceptable for today’s driv- ing conditions. Lomax and Fuhs (1993) reviewed the design

13 practices of freeway entrance ramp meters and HOV bypass lanes and noted that the designs performed well operationally. Hunter et al. (2001) concluded that the AASHTO acceleration rate model used to estimate acceleration lane lengths should not be changed. Fitzpatrick and Zimmerman (2007) examined potential changes to the 2004 Green Book’s adjustment factors for entrance and exit terminals. They recommended potential adjustment factors to acceleration and deceleration lengths for different grades. In a similar study, Fitzpatrick and Zimmerman (2007A) assessed that the acceleration lane lengths contained in the 2004 Green Book should potentially be increased. 2.3 Vehicle Performance Vehicle performance characteristics affect the geometric design of roads. Truck performance characteristics, rather than those of passenger cars, often govern the design of a facility except when truck volume is extremely limited or it is cost pro- hibitive to design the facility based upon truck performance capabilities. Harwood et al. (2003) critiqued the design cri- teria for acceleration lane lengths from the 2001 AASHTO Green Book. In their critique, Harwood et al. (2003) assessed truck performance characteristics implied by the design cri- teria using a truck speed profile model (TSPM). Harwood et al. (2003) used the TSPM to calculate the minimum accel- eration lengths required to enable a 180-lb/hp vehicle to reach the given conditions as specified in the current design criteria. These minimum acceleration lengths were, on average, about 1.8 times greater than the minimum acceleration lengths given in the 2004 Green Book. After analyzing speed/acceleration data of trucks near commercial vehicle weigh stations, Gat- tis et al. (2008) suggested that acceleration lane lengths of the same order of magnitude as those provided by Harwood et al. (2003) should be considered at locations where signifi- cant volumes of trucks enter a freeway. In an unpublished NCHRP Project 3-35 (Reilly et al., 1989) report, a methodology was developed for determining acceler- ation lane lengths based on human factors, traffic flow charac- teristics, and vehicle dynamics. The recommended minimum design lengths for acceleration lanes from this research, which are based on truck operations, are significantly longer than the AASHTO design criteria. In their critique of the AASHTO 2001 design criteria for deceleration lane lengths, Harwood et al. (2003) note that as vehicles exit the freeway mainline, speed changes are normally made with controlled deceleration rates of which trucks are clearly capable, and there is no indication that driver choice of faster operating speeds is the result of short deceleration lanes or is correctable by using longer deceleration lanes. However, Ervin et al. (1986) recommended that trucks require decel- eration lane lengths on the order of 30 to 50 percent longer than those in the Green Book. Firestine et al. (1989) also rec- ommended that minimum deceleration lane lengths required to accommodate trucks are approximately 15 to 50 percent longer than those required for passenger cars. 2.4 Human Factors Several studies investigated human factor issues related to the design and/or operation of freeway mainline ramp terminals. Michaels and Fazio (1989) proposed that the merge process is a four-step sequential process: initial steering control, accelera- tion, gap search, and merge steering; and indicated that these steps work in an iterative fashion until a successful merge is completed. Reilly et al. (1989) proposed the merging process consists of five sequential decision components: (1) steer- ing control (SC) zone, which involves the steering and position- ing of the vehicle along a path by steering from the controlling ramp curvature onto the SCL; (2) initial acceleration (IA) zone, in which the driver accelerates to reduce the speed differential between the ramp and freeway vehicles to an acceptable level for completing the merge process; (3) gap search and acceptance (GSA) zone, during which the driver searches, evaluates, and accepts or rejects the available lags or gaps in the traffic stream; (4) merge steering control (MSC) zone, during which the driver enters the freeway and positions the vehicle in Lane 1, followed by (5) visual clear (VC) zone, which provides a buffer between the driver and the end of acceleration lane. Figure 3 shows the five components described in the merging model suggested by Reilly et al. (1989). In other research, Staplin et al. (1998) note, “age-diminished capabilities that contribute most to older drivers’ difficulties at freeway interchanges include losses in vision and informa- tion-processing ability, and decreased flexibility in the neck and upper body.” Lunenfeld (1993) indicated that the key human factors considerations in the design of interchanges and asso- ciated design features include sensory-motor attributes, visual capability of the driver, importance of sight distance, and meeting driver expectancy. The role of congestion on the freeway lanes in drivers’ merg- ing behavior is not clear. Because speeds are lower in congested conditions, one may argue that the merging task is simpler; however, merging into congested situations may require more cognitive effort as the driver adjusts his/her own speed and repeatedly checks for gaps. In addition, in a congested situation, drivers must make more social judgments regarding the inten- tions of both lag vehicles and of mainline traffic. Sivak (2002) pointed out that driver gap acceptance behavior is not always rational. A driver may set a criterion for gap size that is safe and reasonable, but as the amount of time the driver waits to find that gap increases, the acceptable gap may decrease. Michaels and Fazio (1989) point out another important aspect of the merging process—that it is a collaborative effort between merg- ing and mainline traffic. Sarvi et al. (2002) note that if no gap is

14 available, drivers may accelerate to create a gap or decelerate to wait for a later gap. Choudhury et al. (2009) note that a driver who decides to “force-in,” for instance, is likely to accept smaller gaps and accelerate to facilitate the merge and developed a merge model based upon merging plan choice, gap acceptance, target gap selection, and acceleration decisions of drivers. Kondyli and Elefteriadou (2009) indicate that a driver’s deci- sion to initiate a forced merge depends on traffic-related factors such as freeway speed, congestion, and gap availability, and in congested conditions, driver behavior displays less variability. 2.5 Operational Issues at Freeway Mainline Ramp Terminals The 2000 Highway Capacity Manual (HCM) (TRB, 2000) provides a methodology to assess the level-of-service (LOS) at ramp-freeway junctions. Three major components of the meth- odology for uncontrolled ramp-freeway junctions include: • Vehicle flow in the two rightmost lanes of the freeway imme- diately downstream/upstream of the merge/diverge area. • Likelihood of congestion based upon capacity values and demand flows (either existing or predicted). These capac- ity values include maximum total flow approaching a major diverge area on the freeway, maximum total flow departing from a merge or diverge area on the freeway, maximum total flow entering the ramp influence area, and maximum flow on a ramp. • Determine LOS based on the density of flow in the ramp influence area and estimate average vehicle speed within the influence area. Models of this methodology address the design of entrance and exit ramps with various lane configurations for the main- line and ramp junctions on both left- and right-hand sides of the freeway. The ramp influence area is defined as the area 1,500-ft downstream of the painted nose of an entrance ramp and 1,500-ft upstream of the painted nose of an exit ramp and includes the acceleration or deceleration lane and Lanes 1 and 2 of the freeway. The capacity of merging areas depends primarily on the capacity of the downstream mainline segment. The HCM indi- cates that neither the traffic turbulence created in the merge area nor the number of acceleration lanes affects this capac- ity, but that it is based solely upon the sum of the upstream mainline and entrance ramp traffic volumes. When predicting flows for Lanes 1 and 2, the HCM states that longer accelera- tion lanes encourage less turbulence as ramp vehicles enter the freeway traffic stream and therefore lead to lower densities in the influence area and higher flows in Lanes 1 and 2. When an entrance ramp has a higher free-flow speed, vehicles tend to enter the freeway at higher speeds, and approaching freeway vehicles tend to move farther left to avoid the possibility of high-speed turbulence. When determining the LOS of the ramp-freeway terminal in a merge area, acceleration lane length is an input variable used by the model to determine the density of the influence area. Increasing the acceleration lane length decreases the density of the merge influence area. This calculated density is then used to establish a LOS value. The length of the acceleration lane is also used in the model to determine the average speeds in the vicinity of freeway-ramp terminals. For exit ramps, failure at a diverge area is often related to the capacity of one of the exit legs, most often the ramp. When determining the LOS for a diverge area, deceleration lane length is an input variable for the model to determine the density of the influence area, and increasing the deceleration lane length decreases the density of the diverge influence area. In other research, Polus et al. (1985) conclude that SCLs are used as much for gap location recognition as they are for vehicle Figure 3. NCHRP Project 3-35 entrance model (Reilly et al., 1989).

15 acceleration. Lighter vehicles merge much sooner than heavy trucks, and average speed differences between the mainline traf- fic and entering traffic at the end of the acceleration lane range from 6.5 to 10.2 mi/h. Abella et al. (1976) found a difference of 5 mi/h between the speed of the mainline traffic and merging vehicles. 2.6 Summary Some general findings from the research highlight the gaps in knowledge but also provide direction. It is clear that under certain circumstances, current design policy does not sufficiently provide for the needs of all drivers and all vehicle types. The merging and diverging tasks are complex and are difficult to accommodate within design, especially given changing driver populations and vehicle fleets. On the other hand, there is only limited evidence that current design policy values produce substantive safety concerns for the normal range of conditions. Vehicle performance characteristics form the basis for the geometric design of roadways. Although research indicates the potential need to increase acceleration lane length to accommodate heavy trucks, no research was found to indicate that trucks had difficulty with acceleration lanes designed by the current criteria. Attempts to model the complex driver/vehicle actions asso- ciated with the driving task (and in this case, the entry or exit maneuver) invariably require a series of simplifying assump- tions or assumed performance characteristics. Many models describe the task as a sequential series of events. When all such models are assembled, they may result in findings that appear logical, but produce design dimensions that appear unreasonable, doubtful, or overly conservative. AASHTO’s current geometric design guidance is based on a single set of operational assumptions that can be character- ized as representing free or unconstrained merging behavior. Driver behavior may vary, depending on whether the main- line is congested or not.