Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 65
66 chapter seven Conclusions Background · Observation of driving behavior revealed that the strong- est indicator of operating speed was posted speed limit. From 2000 through early 2011 a significant amount of geo- Design speed appeared to have minimal impact on metric design-related research was conducted on a wide operating speeds unless a tight horizontal radius or a low variety of topics and issues. The objective of this study was K-value was present (Fitzpatrick et al. 2003a). to identify and summarize a sample of roadway geometric · Researchers investigated the possibility of selecting design literature completed and published during that time, a design speed based more heavily on the context of particularly research that identified safety, operations, and the environment in which the roadway was located. A maintenance impacts. A national literature review repre- primary area of concern, however, was how to define sented the vast majority of the effort for this synthesis study. the context to be considered (Garrick and Wang 2005; Wang et al. 2006). Summary of Findings The body of this report has five primary chapters, in addition Driver Characteristics to the introductory chapter and this concluding chapter. This section of the report will present a summary of the key find- · Designers and traffic engineers must examine the road- ings in the body of the report, categorized by topic. This is way environment for information conflicts that may an annotated summary of the findings from the research dis- mislead or confuse road users (Campbell et al. 2008). cussed in the body of the report; the recommendations listed · Designers and traffic engineers must also seek road envi- are those of the authors of the references cited. ronments that are self-explaining, quickly understood, and easy for users to act on (Campbell et al. 2008). It is important to note that the recommendations included in this list of findings from the literature are those of the Stopping Sight Distance authors cited. Before any revisions to AASHTO's Green Book were to be made on the basis of these recommendations, · New values for stopping sight distance (SSD) and new they would need to be considered on the basis of the rigor of design controls for vertical curves were recommended the research and logic that underlie them. No endorsement based on a perceptionreaction time of 2.5 s, a 10th per- of these recommendations is implied by their inclusion in the centile deceleration rate of 11.2 ft/s2, a 10th percentile listing of findings from the literature. driver eye height of 3.5 ft, and a 10th percentile object height of 2.0 ft (Fambro et al. 2000). Design Vehicles · Ramp control signals placed on the left side of a curve of a loop on-ramp (even with a radius greater than 300 ft) · Dimensions of commonly used trucks have changed in are more critical for accommodating SSD than those on recent years, prompting recommendations to revise the the right side (Wang 2007). dimensions of those vehicles in AASHTO's A Policy · The method of selecting SSD values deterministically on Geometric Design of Highways and Streets, com- yielded very conservative estimates of available and monly known as the Green Book (Harwood et al. 2003). required SSD, resulting in a very low probability (0.302%) · Along with the changes in dimensions have come of hazard (Sarhan and Hassan 2008). changes in performance; however, design guidelines are sufficient to accommodate their performance for many Passing Sight Distance design elements (Harwood et al. 2003a). · An analysis of observed passing maneuvers provided Design Speed support for the AASHTO passing sight distance (PSD) model, and the model provided reasonable results for the · Posted speed limit and anticipated operating speed were assumptions made. However, the model's assumptions frequently associated with the selection of design speed may need to be updated or accommodate more flexibility (Fitzpatrick and Carlson 2002). for speeds higher than 55 mph (Carlson et al. 2005).
OCR for page 66
67 · Increased consistency between AASHTO PSD design SSD. An alternative design procedure is recommended, standards and Manual on Uniform Traffic Control based on a new model that incorporated longitudinal Devices (MUTCD) pavement marking practices was friction and acceleration, which produced new recom- recommended, specifically accomplished by using the mended values for minimum lengths of crest and sag MUTCD criteria for marking passing/no-passing zones vertical curves (Hassan 2004). on two-lane roads in the Green Book's PSD design pro- · A weight/power ratio of 102 to 108 kg/kW (170 to cess. In addition to providing the desired consistency 180 lb/hp) would be appropriate for freeways in Califor- between PSD design and marking practices, two-lane nia and Colorado, and a weight/power ratio of 126 kg/kW highways could be designed to operate safely with the (210 lb/hp) would be more appropriate in Pennsylvania, MUTCD criteria (Harwood et al. 2008). as compared with the 120 kg/kW (200 lb/hp) value rec- ommended in the 2001 Green Book (Torbic et al. 2005). · The upward divergent headlamp angle used in the sag Horizontal Alignment curve design equation should be reduced from 1° to between 0.75° and 0.90° (Hawkins and Gogula 2008). · Erroneous perceptions by drivers approaching horizon- tal curves, as influenced by vertical curves, increased as (1) the sight distance increased, (2) the horizontal curve Allocation of Traveled Way Width radius increased, and (3) the length of vertical curve per 1% change in grade decreased. Drivers tend to drive faster · The benefits of 2+1 roads in Europe validated a recom- on horizontal curves in sag combinations and slower on mendation for their use in the United States, to serve as horizontal curves in crest combinations. Designers should an intermediate treatment between an alignment with establish the profile and predicted operating speed of an periodic passing lanes and a full four-lane alignment. alignment based on a three-dimensional model, rather Such 2+1 roads are most suitable for level and rolling terrain, with installations to be considered on roadways than a traditional two-dimensional model (Bidulka et al. with traffic flow rates of no more than 1,200 veh/hr in a 2002; Hassan et al. 2002). single direction. The use of a cable barrier as a separa- · For drivers on curves with radii greater than or equal to tor is discouraged, and major intersections should be 350 m (1,146 ft), as the deflection angle increased, speed located in the buffer or transition areas between oppos- measures (mean, 85th percentile, and 95th percentile) ing passing lanes, with the center lane used as a turning decreased; as a result, motorists may view a large change lane (Potts and Harwood 2003). in direction as a motivation to slow their speed. In addi- · Passing activity on 2+1 roads was greatest at the begin- tion, as curve length increased, speed measures increased, ning of the segments and the greatest benefits of decreased suggesting that drivers may become more comfortable platooning and increased safety occurred within the first at higher speeds because they have more time to adjust 0.9 mi of a passing lane segment (Gattis et al. 2006). their vehicle path to a constant radius. Grade has an influ- · Most passing on Super 2 passing lanes occurs within ence on the upper-percentage range of vehicle speeds, the first mile of a passing lane, so additional length may because the 85th percentile speed decreased as approach be less useful than additional lanes in a Super 2 corridor, grade increased (Schurr et al. 2002). particularly at lower volumes. Designers should avoid · A study of driver behavior and errors on a selection of intersections with state highways and high-volume horizontal curves led Lyles and Taylor (2006) to conclude county roads within passing lanes, consider terrain and the following: right-of-way in determining alignment and placement Drivers approaching curves routinely exceeded the of passing lanes, avoid the termination of passing lanes posted speed limit as well as the posted advisory on uphill grades, and discourage passing lane lengths speed, where applicable. longer than 4 mi (Brewer et al. 2011). Drivers had more errors at curves where they had · Two-way left-turn lanes could be used as a strategy to limited or no visibility of the curves when the TCDs reduce head-on collisions on two-lane roads (Neuman were first visible. et al. 2003b). Drivers made more errors on horizontal curves that were adjacent to vertical curves, particularly crests Lane Width that obscured a downstream horizontal curve. There were increased errors when curves were com- · There was no general indication that the use of lanes bined with other elements, especially intersections. narrower than 12 ft on urban and suburban arterials increased crash frequencies. Geometric design policies Vertical Alignment should provide substantial flexibility for the use of lane widths narrower than 12 ft (Potts et al. 2007). · Current North American design practices might yield · Lane widths of 11 or 12 ft provide optimal safety benefit segments of the vertical curve where the driver's view for common values of total paved width on rural two-lane is constrained to a distance shorter than the required roads. Although 12-ft lanes appear to be the optimal
OCR for page 67
68 design for 26- to 32-ft total paved widths, 11-ft lanes Any combination of a sloping-faced curb that is perform equally well or better than 12-ft lanes for 34- to 150 mm (6 in.) or shorter and a strong-post guardrail 36-ft total paved widths (Gross et al. 2009). can be used where the curb is flush with the face of the guardrail up to an operating speed of 85 km/h. Guardrails installed behind curbs are best not located Road Diet closer than 2.5 m (8.2 ft) for any operating speed in · The rate of road diet crashes occurring during the excess of 60 km/h (37.3 mph). period after installation was about 6% lower than that For roadways with operating speeds of 70 km/h of matched comparison sites. However, controlling for (43.5 mph) or less, guardrails may be used with possible differential changes in average daily traffic, sloping-face curbs no taller than 150 mm (6 in.) as study period, and other factors indicated no significant long as the face of the guardrail is located at least effect of the treatment. Crash severity was virtually the 2.5 m (8.2 ft) behind the curb. same at road diets and comparison sites. Conversion to Where guardrails are installed behind curbs on roads a road diet should be made on a case-by-case basis in with operating speeds between 71 and 85 km/h which traffic flow, vehicle capacity, and safety are all (44.1 and 52.8 mph), a lateral distance of at least 4 m considered (Huang et al. 2002). (13.1 ft) is needed to allow the vehicle suspension to · The effects of the road diet on crashes in Iowa, account- return to its pre-departure position. ing for monthly crash data and estimated volumes for At operating speeds greater than 85 km/h (52.8 mph), treatment and comparison sites, resulted in a 25.2% guardrails are used with 100-mm (4-in.) or shorter reduction in crash frequency per mile and an 18.8% sloping-faced curbs, and are placed so that the curb is reduction in crash rate (Pawlovich et al. 2006). flush with the face of the guardrail. Operating speeds above 90 km/h (55.9 mph) require that the sloping face of the curb must be 1:3 or flatter and must be no Shoulder Width more than 100 mm (4 in.) in height. · The "Safety Edge" treatment produced small but posi- · For horizontal curves on two-lane nonresidential facili- tive results in crash reduction at 56 of 81 treated sites. ties that have 3 degrees of curvature, the width of the lane For all two-lane highway study sites in two states, the plus the paved shoulder should be at least 5.5 m (18 ft) best estimate of the treatment's effectiveness was a throughout the length of the curve (Staplin et al. 2002). reduction in total crashes of approximately 5.7%. The · Wider lane and shoulder widths are associated with a results were not statistically significant, but they were reduction in segment-related collisions on rural front- generally positive (Hallmark et al. 2006). age road segments (Lord and Bonneson 2007). Roadside Rumble Strips · Where possible at curb locations, provide a lateral · Crashes at approximately 210 mi of undivided rural offset to rigid objects of at least 6 ft from the face of two-lane roads treated with centerline rumble strips the curb and maintain a minimum lateral offset of 4 ft were reduced by 14% and injury crashes were reduced (Dixon et al. 2008). by an estimated 15%. All frontal and opposing-direction · At lane merge locations, do not place rigid objects in an sideswipe crashes were reduced by an estimated 21%, area that is 10 ft longitudinally from the taper point. The and those crashes involving injuries were reduced by an lateral offset for this 20-ft section is consistent with the estimated 25%. All of the reductions were determined lane width, typically 12 ft (Dixon et al. 2008). to be statistically significant (Persaud et al. 2003). · A lateral offset of 6 ft from the curb face to rigid objects · Crash data on roads treated with centerline rumble strips is preferred for higher-speed auxiliary lane locations, or shoulder rumble strips revealed noticeable crash reduc- such as extended length right-turn lanes, and a 4-ft tions on all classes of roads (rural and urban two-lane minimum lateral offset should be maintained (Dixon roads and freeways). Shoulder rumble strips placed as et al. 2008). close to the edgeline as possible maximize safety ben- · At locations where a sidewalk buffer is present, rigid efits. The safety benefits of centerline rumble strips for objects are best not located in a buffer area with a width roadways on horizontal curves and on tangent sections of 3 ft or less. For buffer widths greater than 3 ft, lat- are for practical purposes the same (Torbic et al. 2009). eral offsets from the curb face to rigid objects are main- tained with a minimum offset of 4 ft. At these wider Shoulder Edge Treatments buffer locations, other frangible objects can be strate- gically located to help shield any rigid objects (Dixon · Plaxico et al. (2005) made the following recommenda- et al. 2008). tions on design guidelines for using curbs on roadways · Rigid objects are best not located in the proximity of with operating speeds greater than 60 km/h (37.3 mph): driveways, and care should be taken to avoid placing
OCR for page 68
69 rigid objects on the immediate far side of a driveway. In "Provide slow entry speeds and consistent speeds addition, objects are not to be located within the required through the roundabout by using deflection. sight triangle for a driveway (Dixon et al. 2008). Provide the appropriate number of lanes and lane assignment to achieve adequate capacity, lane volume balance, and lane continuity. Intersection Alignment Provide smooth channelization that is intuitive to drivers and results in vehicles naturally using the · Avoid approach grades to an intersection of greater intended lanes. than 6%. On higher design speed facilities (50 mph and Provide adequate accommodation for the design greater), a maximum grade of 3% should be considered vehicles. (Rodegerdts et al. 2004). Design to meet the needs of pedestrians and cyclists. · Avoid locating intersections along a horizontal curve of Provide appropriate sight distance and visibility for the intersecting road (Rodegerdts et al. 2004). driver recognition of the intersection and conflicting · Strive for an intersection platform (including sidewalks) users." with a cross slope not exceeding 2%, as needed for · Maximum entering design speeds are based on a theo- accessibility (Rodegerdts et al. 2004). retical fastest path of 20 to 25 mph for single-lane round- · Approach curvature can be used as a treatment to force abouts and 25 to 30 mph for multilane roundabouts a reduction in vehicle speed through the introduc- (Rodegerdts et al. 2010). tion of horizontal deflection at high-speed intersection · Roundabout alignment is described as "optimally located approaches, but it is discouraged at downhill approaches when the centerlines of all approach legs pass through (Ray et al. 2008). the center of the inscribed circle" (Robinson et al. 2000). · A skew angle greater than 20 degrees is not recom- · Common inscribed circle diameters for single-lane mended in design when the design vehicle is a large roundabouts vary from 90 to 180 ft, depending on design vehicle or semitrailer (Son et al. 2002). vehicle (Rodegerdts et al. 2010). · A minimum skew angle of 15 degrees will accommo- · Designers "should provide no more than the minimum date age-related performance deficits at intersections required intersection sight distance on each approach, where right-of-way is restricted (Staplin et al. 2002). [because] excessive intersection sight distance can lead to higher vehicle speeds that reduce the safety of the Auxiliary Lanes intersection for all road users" (Robinson et al. 2000). · Crash experience at selected intersections in the United · Adding left-turn lanes is effective in improving safety States indicates an overall reduction in crash frequency at signalized and unsignalized intersections, reducing at intersections converted to roundabouts (Rodegerdts crashes between 10% and 44% (Harwood et al. 2002). et al. 2007). · Positive results can also be expected for right-turn lanes, · Pedestrian refuge a minimum width of 6 ft will ade- with reductions in total intersection accidents between quately provide shelter for persons pushing a stroller or 4% and 14% (Harwood et al. 2002). walking a bicycle (Robinson et al. 2000). · A method was developed to identify where installation · At single-lane roundabouts, the pedestrian crossing is of right-turn lanes at unsignalized intersections and best located one vehicle-length (25 ft) away from the major driveways would be cost-effective, indicating yield line. At double-lane roundabouts, the pedestrian combinations of through-traffic volumes and right-turn crossing is best located one, two, or three car lengths volumes for which provision of a right-turn lane would (approximately 25 ft, 50 ft, or 75 ft) away from the yield be recommended. The economic analysis procedure line (Robinson et al. 2000). can be applied by highway agencies using site-specific · The "pedestrian refuge should be designed at street level, values for average daily traffic, turning volumes, acci- rather than elevated to the height of the splitter island. dent frequency, and construction cost for any specific This eliminates the need for ramps within the refuge area, location (or group of similar locations) of interest (Potts which can be cumbersome for wheelchairs" (Robinson et al. 2007). et al. 2000). · Ramps may be provided on each end of crosswalks to connect the crosswalk to other crosswalks around the Modern Roundabouts roundabout and to the sidewalk network (Robinson et al. 2000). · A series of projects during the decade led to the pub- · A detectable warning surface, as recommended in the lication of two FHWA Informational Guides contain- Americans with Disabilities Act Accessibility Guide- ing recommendations and guidelines for all aspects of lines, may be applied to the surface of the refuge within roundabout design. the splitter island (Robinson et al. 2000). · General overarching principles of geometric design of · Use of standard AASHTO island design for key dimen- roundabouts (Rodegerdts et al. 2010) include: sions, such as offset and nose radii, is encouraged. For
OCR for page 69
70 sidewalks, a setback distance of 5 ft, with a minimum Pedestrian and Bicycle Facilities at Intersections of 2 ft is advised (Robinson et al. 2000). · For nonmotorized users such as bicyclists, one important · Suggested strategies (Raborn et al. 2008) for modifying consideration during the initial design stage is to main- intersections to accommodate bicycles and pedestrians tain or obtain adequate right-of-way outside the circula- included: tory roadway for the sidewalks. All nonmotorized users Reducing the crossing distance for bicyclists. who are likely to use the sidewalk regularly, including Realigning intersection approaches to reduce or bicyclists in situations where roundabouts are designed eliminate intersection skew. to provide bicycle access to sidewalks, should be con- Modifying the geometry to facilitate bicycle move- sidered in the design of the sidewalk width. Recom- ment at interchange on-ramps and off-ramps. mended designs for single-lane roundabouts encourage Providing refuge islands and raised medians. bicycle users to merge into the general travel lanes and Grade-separated crossings. navigate the roundabout as a vehicle, explaining that · "Pedestrian facilities should be provided at all inter- the typical vehicle operating speed within the circula- sections in urban and suburban areas. In general, tory roadway is in the range of 15 to 25 mph, which is design of pedestrian facilities with the most challenged similar to that of a bicycle (Rodegerdts et al. 2010). users in mind--pedestrians with mobility or visual impairments--should be done, and the resulting design Innovative Intersection Designs will serve all pedestrians well. ADA requires that new and altered facilities constructed by, on behalf of, or for A number of new or innovative intersection designs were con- the use of State and local government entities be designed sidered during the decade; each of the following was described and constructed to be readily accessible to and usable by in one or more studies. individuals with disabilities" (Rodegerdts et al. 2004). · Practitioners should incorporate key elements that affect · Displaced Left Turns showed considerable savings in a pedestrian facility into their design (Rodegerdts average control delay and average queue length, as well et al. 2004): as an increase in intersection capacity, in one series of "Keep corners free of obstructions to provide enough microsimulation analyses (Hughes et al. 2010). room for pedestrians waiting to cross. · Median U-turns are typically a corridor treatment applied Maintain adequate lines of sight between drivers at signalized intersections but are also used at isolated and pedestrians on the intersection corner and in the intersections to alleviate specific traffic operational and crosswalk. safety problems (Hughes et al. 2010). Ensure curb ramps, transit stops (where applicable), · Median width of Restricted Crossing U-Turns is a crucial pushbuttons, etc., are easily accessible and meet design element to accommodate large trucks without ADA Accessibility Guidelines design standards. allowing vehicles to encroach on curbs or shoulders Clearly indicate the actions pedestrians are expected (Hughes et al. 2010). to take at crossing locations. · Quadrant Roadways are best designed so that the left Design corner radii to ensure vehicles do not drive turn with the highest demand is the one that receives the over the pedestrian area yet are able to maintain most direct path (Hughes et al. 2010). appropriate turning speeds. · Double Crossover Intersections are found to have greater Ensure crosswalks clearly indicate where crossings throughput than a conventional intersection, along with should occur and are in desirable locations. lower values for number of stops, average stop time Provide appropriate intervals for crossings and mini- per vehicle, average queue, and maximum queue length mize wait time. (Bared et al. 2005). Limit exposure to conflicting traffic and provide ref- · Arterial Interchanges have an overall capacity near uges where necessary. 75% of a four-lane freeway (Eyler 2005). Ensure the crosswalk is a direct continuation of the · J-Turn and Offset-T designs had reductions in crashes pedestrian's travel path. between 40% and 92% (Maze et al. 2010). Ensure the crossing is free of barriers, obstacles, and · Two-Level Signalized Intersections produced modeled hazards." results with the shortest delay times in most evalua- tion scenarios as well as the least sensitivity to varia- Transit Considerations tions in traffic volume compared with other innovative intersection types; however, delay increased when · General intersection design principles and guidelines flow was unbalanced between the two crossing roads for transit issues (Eccles et al. 2007) include: (Shin et al. 2008). "Provide simple intersection designs. · The additional right-of-way needed to construct each of Provide clear visual cues to make busway intersec- these innovative designs was mentioned as a potential tions conspicuous. drawback by every report and author that addressed the Maximize driver and pedestrian expectancy. issue of the intersection's footprint. Separate conflicting movements.
OCR for page 70
71 Minimize street crossings. higher rates of undesirable driving behavior. In con- Incorporate design features that improve safety for trast, the median opening in urban and suburban areas vulnerable users. can be as long as necessary (Potts et al. 2004). Coordinate geometric design features and traffic con- trol devices." · There are four types of busways found at intersections: Interchange Ramp Design median busways, side-aligned busways, separated right- · The desired distance between the cross street and freeway of-way busways, and bus-only ramps. Each busway type merge point is at least 400 m (1,312 ft) for ramps at which has unique characteristics that are considerations for metering is envisioned (Chaudhary and Messer 2002). guidance on safety issues, basic geometry (including · The source of the adjustment factors in the 2004 Green placement of bus stops), and traffic control, along with Book was provided in the 1954 AASHTO Blue Book, examples of appropriate intersections for each type of in which they first appeared as being based on applying busway (Eccles et al. 2007). "principles of mechanics to rates of speed change for level grades." Further review did not reveal a procedure Access Management at Intersections for determining adjustment factors. A new procedure contains an alternative set of adjustment factors for accel- · Right-turn-plus-U-turn could have better operational eration length and deceleration length, the latter of performance than direct left turns under certain traffic which is based on the ratio of the deceleration length conditions, implying that directional median opening on a grade to the deceleration length on a level surface. designs could provide more efficient traffic flow than Actual performance of vehicles on grades and on a level full median openings (Zhou et al. 2002). surface should be measured and compared with the sug- · U-turns at signalized intersections resulted in a 1.8% gested adjustment factors to determine the accuracy of saturation flow rate loss in the left-turn lane for every those factors (Fitzpatrick and Zimmerman 2007). 10% increase in U-turn percentage and an additional 1.5% loss for every 10% U-turns if the U-turning move- ment was opposed by protected right-turn overlap from Ramp and Interchange Spacing the cross street (Carter et al. 2005a). · Recent guidelines make a distinction to separately define · Recommended practices (Potts et al. 2004) for rural "ramp spacing" and "interchange spacing" and recom- unsignalized intersections include: mend ramp spacing values be the primary consider- Medians that are as wide as practical, as long as the ation in freeway and interchange planning and design median is not so wide that approaching vehicles (Ray et al. 2011). on the crossroad cannot see both roadways of the · Guidelines are presented based on four areas of emphasis: divided highway. geometric design, traffic operations, signing, and safety. Where the AASHTO passenger car is used as the Geometric design principles, as well as site-specific fea- design vehicle, a minimum median width of 25 ft is tures, dictate minimum lengths needed for ramps and other recommended. interchange components. Traffic volumes can necessitate Where a large truck is used as the design vehicle, increased spacing beyond the dimensions needed purely a median width of 70 to 100 ft generally is recom- for geometrics. Safety tradeoffs, which have rarely been mended. If such a median width cannot be provided, quantified until recently, can now be considered in project consideration should be given to providing a loon. decision making. Finally, signing and other human fac- · Recommended practices (Potts et al. 2004) for suburban tors issues are best taken into account at the earliest in the unsignalized intersections include: evaluation process when making choices about ramp and Median widths at suburban unsignalized intersections interchange spacing (Ray et al. 2011). generally as narrow as possible while providing suffi- · Spacing assessments indicate that ramp spacing of less cient space in the median for the appropriate left-turn than 900 ft is likely not geometrically feasible. That treatment. spacing value increases up to 1,600 ft for entranceexit Median widths between 14 and 24 ft will accommo- ramp pairs (Ray et al. 2011). date left-turn lanes, but are not wide enough to store a crossing or turning vehicle in the median. Medians wider than 25 ft may be used, but crossroad Alternative Interchange Designs vehicles making turning and crossing maneuvers may stop on the median roadway. · Design practices for the Diverging Diamond interchange Median widths of more than 50 ft generally should (Hughes et al. 2010) include: be avoided at suburban, unsignalized intersections. "The minimum crossing angle of intersection should · Keep median opening lengths at rural divided high- be 40 degrees. way intersections generally to the minimum possible. The radius design should accommodate between 25 Increases in median opening length are correlated with and 30 mph.