USF Libraries
USF Digital Collections



Material Information

Medians a discussion paper
Physical Description:
1 online resource (xiv, 104 p.) : ill. ;
Stover, Vergil G., 1933-
Florida -- Dept. of Transportation
University of South Florida -- Center for Urban Transportation Research
University of South Florida, Center for Urban Transportation Research
Place of Publication:
Publication Date:


Subjects / Keywords:
Median strips -- Design and construction   ( lcsh )
Roads -- Design and construction   ( lcsh )
arterial highways /urban
bibliography   ( marcgt )
non-fiction   ( marcgt )


Includes bibliographical references (p. 95-104).
Statement of Responsibility:
by Vergil G. Stover.
General Note:
Prepared for the Florida Dept. of Transportation.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 029364666
oclc - 748831529
usfldc doi - C01-00021
usfldc handle - c1.21
System ID:

This item is only available as the following downloads:

Full Text


MEDIANS Prepared for the Florida Department of Transportation by Vecgil G. Stover Center for Urban Transportation Research College of Engineering University of South Florida December \994


This page intentionally left blank. u


TABLE OF CONTENTS FOREWORD ....................... .... ........... :. . . . . . xi ACKNOWLEDGEMENTS . . . . . . . . . . . . . . . . . . . . lUll INTRODUCflON . . . . . . . . . . . . . . . . . . . . . . . 1 Left Turn Bays . . . . . . . . . . . . . . . . . . . . . . . 2 Rural Highways . . . . . . . . . . . . . . . . . . . . . . . 4 MEDIAN DESIGN . . . . . . . . . . . . . . . . . . . . . . . . 9 Median Width . . . . . . . . . . . . . .. . . . . . . . . . . 9 Curbed Mecltans . . . . . . . . . . . . . . . . . . . . . . . 9 Curbed Medians V Painted Medians . . . . . . . . . . . . . . . 10 Depressed Medians v Rltised Medians . . . . . . . . . . . . . . 14 CL TL V Raised Medians . . . . . . . . . . . . . . . . . . . 14 Rltised V CL TL V Undivided . . . . . . . . . . . . . . . . . 21 Michigan's Experience With U-tums . . . . . . . . . . . . . . . 25 RETROFITS TO EXISTING MEDIANS . . . . . . . . . . . . . . 29 Memorial Highway, Atlanta Metro Area . . . . . . . . . . . . . 29 Jimmy Carter Blvd., Atlanta Metro Area . . . . . . . . . . . . . 30 SAFETY BENEFfl'S OF IMPROVEMENTS TO EXISTING MEDIANS 33 Oaklaud Park Blvd., Fort Lauderdale, Florida . . . . . . . . . . . 34 US I, Stuart, Florida . . . . . . . . . . . . . . . . . . . . . 39 Comparison of Oakland Park and US 1 with Sunrise Blvd. . . . . . . 40 Kellogg Avenue, Wichita, KanAAS. . . . . . . . . . . . . . . . 41 Vancouver, British Columbia . . . . . . . . . . . . . . . . . . 41 OPERATIONAL BENEFITS OF MEDIAN IMPROVEMENTS . . . . 43 GUIDELINES FOR LEFT-TURN LANES . . . . . . . . . . . . . 47 Turn Bay Length . . . . . . . . . . . . . . . . . . . . . . SO Maneuver Distance . . . . . . . . . . . . . . . . . . . . . . 51 Queue Storage . . . . . . . . . . . . . . . . . . . . . . 55 Storage Length for Dual Left-Turns . . . . . . . . . . . . . 5 7 Storage at Unsignalized Intersections ... .-. . . . . . . . . . . . . 58 Ill


This is a blank page


LIST OF FIGURES I Comparison of Crash Rates on Georgia Roadways with Raised Medians and Two-way Left-Tum Lanes ..................... 18 2 Crash Rates for Raised Medians and Two -Way Left-Tum Lanes in Florida . . . .. . .. .. .. .. . . .. . .. . . .. .. . . . . 19 3 Schematic Dlustration of the Michigan Directional Crossover . . 26 4 Signing for the Michigan Directional Crossover . . . . . . . . 27 5 Median Design on Oakland Park Blvd., Fort Lauderdale, Florida . 36 SA Original Median Design .. . . . .. . .. .. .. .. .. .. .. . 36 5B Redesigned Median . . . . . . . . . . . . . . . . . 36 6 Comparison of Total Crash Rates on Oakland Park Blvd., US I and Sunrise Blvd. . . . . . . . . . . . . . . . . . . . . 40 7 Til Guidelines for Left-tum Lanes . . . . . . . . . . . . . 48 8 Comparison of Harmelink and Til Curves . . . . . . . . . . 49 9 Comparison of the 1TI C)JI'Ves and Colorado DOT Warrants . . 49 I 0 Boundary of Intersection .. .. . . . . . . . .. . .. . .. . . 50 II Elements of the Functional Area of an Intersection . . . . . . 51 12 Storage at Signalized Intersections . .. .. . . .. .. . .. .. .. .. . 56 13 Storage Guidelines for Unsignali7l'.d Intersections . . . . . . . 59 13ANomograph for 4-Lane Highways . . . . . . . . . . . 59 13BExample Nomograph for 2-Lane Highways . . . . . . . 59 v


This is a blank page


LIST OF TABLES Table Pace I Crashes by Frequency of Median Openings by Type of Crash . . . . . . .. . . .. . . . . . . . . . . . . . . 4 2 Crash Rates on Rural Highways Related to Design ............ 6 3 Comparison of Crash Rates on Rural Higbways ... ............ 7 4 Two-Year Driveway Crash Experience as a Function of Median Control ............................................... 10 5 Comparison of Crash Experience Between Streets with Curbed Medians and Painted Medians .. .. . . . .................. II 6 Before-and-After Crashes by Left-Turn Channelization at U l'edA p nstgna ccess omts ........................... .... 13 7 Effectiveness of Raised and CL TL's as a Function of ADT and Driveways Per Mile . . . . . . . . . . . . . . . . . 15 8 Estimated Crashes Per Mile on Continuous Two-Way Left Turn Lane Roadways . . . . . . . . . . . . . . . . 16 9 Estimated Crashes with Continuous Two-Way Left-Tum Lanes Compared to Actual Crashes with Channelized Left-Tum Bays . . 17 I 0 Michigan Statewide Crash Rates for Selected Types of Arterials, 1985 I 987 . . . . . . . . . . . . . . . . . . . 19 I I Midblock and Intersection Vehicular Crash Rates by Median Type . . . . . . . . . . . . . . . . . . . . . . . . . 20 12 Swrunary of Suburban Vehicle Crash Rate Severity by Median Type .. ............ ............................ 20 V11


This is a blank page


LIST OF TABLES (Continued) Table 28 US 1 Crash Data . . . . . . . . . . . . . . . . . . . . 39 29 Effects of Increased Spacing and Limitations of Movements . . . 43 30 Comparison of Average Operational Characteristics on Two Urban Arterials . . . . . . . . . . . . . . . . . . . . . . . 44 31 Comparison of Brake Applications on Sunrise Blvd. and Oakland Park Blvd. . . . . . . . . . . . . . . . . . 45 32 Calculated Upstream Maneuver Distances . . . . . . . . . . 53 33 Effect of Radius and Turning Speed on Deceleration Distance . . 54 34 Distance Traveled While Making Lateral Movement Into a Turn Bay 55 35 Example Calculation for Dual Turn Bay . . . . . . . . . . 58 36 Effect of Median Openings on Sales Volume . . . . . . . . . 64 37 Percent of Respondents Favoring the Changed Median on Oakland Park Blvd. . . . . . . . . . . . . . . . . . . . 66 38 Opinions Regarding Oosing Median Opening on US I, Stuart, Florida . . . . . . . . . . . . . . . . . . . . . 67 39 Pedestrian Accident Rates . . . . . . . . . . . . . . . . 70 40 Comparison of Pedestrian Accident Rates . . . . . . . . . . 70 41 Crashes Involving Pedestrians on Urban Arterials in Florida . . 71 IX


This page intentionally left blank. X


FOREWORD This document is intended to serve as a discussion paper for FDOT personnel and to provide general guidance as to the design of medians on major roadways. It summarizes the literature relating to median designs. Its focus is primarily on crash rates related to median type and design. In addition to the relatively large body of research in the recent literature, data are drawn from the significant but limited studies conducted in the 1950's and early 1960's. Very little empirical information exists relative to a before-and-after comparison of operational characteristics. Also, information on the impact of the spacing o f median openings is extremely limited. To date, the current NCHRP project 25-4 on the effect of median closures on businesses has provided little enlightenme n t on this topic. X I


This page intentionally left blank


ACKNOWLEDGEMENTS Funding for this project was provided by the Florida Department of Transportation. Mr. Gary Sokolow of the FOOT Systems Planning Division provided project oversight. His review and comment on the various preliminary drafts of this report is gratefully acknowledged. Xl1l


This page intentionally left blank. XIV


INTRODUCTION The problems of applying access control to a developed arterial pose some of the greatest challenges to today's traffic manager. Many studies have documented the damaging effects that closely spaced and/or inappropriately designed access points have on the quality of traffic flow provided by a roadway. Officials responsible for safe, efficient movement of traffic are certainly aware of increasing accident rates and reduced levels of service that occur with an increase in traffic, an increase in access points or in most cases, both. Medial access control measures include both modifications ofroutesegmentsand point improvements. Route segment changes which are applicable to major roadways include the following: 1 Installation of a non traversable median; 2. Replacement of a continuous two-way left-tum lane with a nontraversable median; and 3. Oosure or redesign of a median opening along an entire section of roadway Point improvements/changes include: I. Oosure of a median opening; 2. Redesigning a median opening so as to permit a selected movement(s) only; 3. Adding a left-turn bay at a median opening; and 4. Increasing the length of an existing turn bay to provide adequate queue storage and to reduce the speed differential between turning vehicles and through traffic. Implementation of median access control on existing roadways is commonly very difficult and expensive. Opposition by the owners of the adjacent properties and the affected businesses often make it difficult to obtain the necessary political acceptance Concerns may also arise over the safety of resulting u -turn and weaving movements and the effect on business sales. However, it has been demonstrated that medial access control results in a substantial reduction in the number of crashes together with a reduction in the associated social and economic costs of deaths, injuries, and property damage. Other benefits include time savings and reduced fuel consumption. Furthermore, air quality improvement can be obtained through the implementation of access management techniques which will reduce vehicular emissions by improving traffic flow and reducing idling delay. Medial access control is also an effective 1 November 1994


This is a blank page



This is a blank page


These data for multilane highways are swnmarized in Table 3 for ready comparison. This comparison table clearly shows the safety benefit of divided roadways and marginal a:s well as median access control. The total average crash rate for 4-lane divided highways was found to be only 58% of the 4-lane undivided roadway. The additional benefit of marginal access control shows that the divided controlled access highways bad a crash rate which is only 41% of the 4-lane undivided roadways and 58% of the 4-lane divided highways (i.e median access control but without marginal access control) The safety advantage of a median is evidenced by the relatively low head-on crash rates which are only 30'% of that for the 4-lane undivided highways. Also, i t is to be noted that the divided highways had much lower intersection crash rates than the 4-lane undivided highways The 4-lane divided highways experienced an average intersection crash rate which is 67"A. o f the 4-lane undivided. More interestingly, the intersection crash rate of the divided controlled access highways was a mere 24% of that for the 4-lane undivided highways and only 36% of that for 4-lane divided highways The fact that median access control results in reduced intersection crash rates has been verified by other research (10 II, 12, 15). Tbe fmdings that median access control reduces midblock crashes is to be expected because of the elimination, or restriction of left-turns. The lower midblock crash rate for the divided controlled access highways compared to the 4-lane divided rates (1.08 compared to 1.22) indicates that the control of marginal access in addition to median access provides an additional safety benefit This probably results from the larger separation of right in/right-out conflict area where marginal access control is exercised. 29 November 1 994 5


This is a blank page


Table 3 Comparison of Crash Rates on Rural Highways +lane I>\yjdcd Iliyjdqi Conttollm AS&$' Percent Percent Percent Type 4-Lalle or 4-Lalle of 4-Lalle of4-Lane of Undivided l.lndlvidec! Undivided Divided Cn!s,b Rate Rate Rate To!al Reported 4 .09 2 .91 o.58 1.69 0.41 0.58 Head-On &tween Intersections 0 .20 0.06 0.30 0 06 0 .30 1.0 Non Head-On lletw<:en l.otonecllons 0.99 0.71 0.78 O.S9 0.59 0.71 At l.otersections 2.52 1.69 0 67 0.61 0 24 0.36 To!al Exduding Intersections J.S7 1 .22 0.78 1.08 0 69 0 .88 Soun:e : Calculated fro111 Table 2 Cribbins attempted to fmd optimal median opening spacings along a four-lane divided highway as a function of accidents, intenSity of roadside development, and travel time. A preci.<;e :;pacing could not be derived. However, his re.

This is a blank page


MEDIAN DESIGN Median Width There are three primary reasons f o r requiring a minilllum median width: I) separate opposing traffic streams; 2 ) provide auxiliary lane(s) to decelerate vehicles and store left-turning vehicles and u-turners; and 3) provide pedestrian refuge These factors reduce congestion and increase the capacity of the arterial by limiting conflicts between through traffic, cross traffic, and turning vehicles Limiting these conflicts increases capacity and improves safety by minimizing the speed differential between through vehicles and turning vehicles. This enhances a constant speed along the arterial. As early as 1951 Telford (),1), investigated the advantages ofnarrowmed i anscompared to undivided roadways. A 4-foot median separating two 33foot roadways was installed on a major street through a central business district. Head on collisions were reduced 65% after the median installation. Both the total number and the severity of crashes were reduced. The median also provided a pedestrian ref uge area, as a result the pedestrian accident rate was reduced 70%. In I %4 Priest (2&), reported on the value of having a median of sufficient width to "shadow" .a left. turn or crossing vehicle on a major roadway. Crash frequency was found to have an inverse relationship to the median width and the magnitude of an exposure index. The exposure index is a measure based on arterial ADT. cross street ADT and the exposure time of a crossing vehicle Curbed Medians Wilson (20), evaluated 12 types of simple intersection improvements at 1,160 different locations in a 1967 Highway Research Board Special Report. He reported a significant crash reduction with curbed, nontraversable medians and intersection channelization. Box (21), analyzed the crash experience for a period at 1238 access points to streets in Skokie, Dlinois. The data from the 1967 report SIIIIIIIl3J'ize in Table 4 illustrate the value of nontraversable medians in reducing crashes occurring at driveways. An even earlier 1953 report by Hanna CUl. concluded that where medians in the urban area were not curbed, damage to grass trees. and shrubs was frequent. Also, the control o f parking was impractical, especially near churches and shopping centers. The occurre n ce of 1 November 1994 9


This is a blank page


TableS -Comparison of Crash Experience Between Streets with Curbed Medians and Painted Medians Annual Number Number en..bes Crashes Crash of of per per Location Crubes, Openings Oncninv Mile Crash Ratcu1 Intersections Curbed Median m 64 21 l.S 17 3 .23 Painted Median 1 105 14 3.8 35 5 .74 Midbloclt (Other than Driveways) Curbed Median m 19 8 1.2 5 0.96 Paiute

This is a blank page


In 1967 Wilson (20), also found a significant reduction in crashes where channelized left tum lanes were added at unsignalized medial access points (intersections and high-volume driveways). Before-and-after studies were made at locations where the left-tum lanes were delineated using raised bars, curbs, and paint. As shown in Table 6 all three methods produced a significant reduction in crashes. Painted channelization produce a 32"/o reduction whereas curbed and raised bars(rumblestrip) resulted in 59% and 67%reduction in crash frequency and 64% and 69"/o reductions in crash rates. Table 6-Before-and-Mter Crashes by LeftTorn Channelization at Unsignalized Ac:g:ss Points N"'"bc< Millioa Type ot Vdoido. T otal Propeny ClyAOc;!iDtjon Pmiq;tJ Copditjop Mi"' Cp'n Ljwy ..E&III. ..!&.. l!iG1. Pai.ou:d 27 be( ore 13<.5 157 &4 71 2 9& 51 an.. 134.1 106' 6>1 50' 2 sa 48 o/ch&oac -l2 4 -30 0 .... Cu

This is a blank page


intersections. Also, fora given population and ADT, increasingthenumberofdrivewaysfrom less than 4Q to over 60 results in an increase of about 30 crashes per mile. Although the grouping of the reported data mask the variation in individual routes, the data clearly show crash experience increases as ADT and the number of access points increases T able 7 E ffectiveness of Raiaed and CLTL's as a FUDCtio n o f ADT and Driveways Per Mile Conditions Level of Roadside QeycJopment !..ow 60 driveways Per Mile Source: Reference (26) Highway ADI !..ow <5,000 High >15,000 Annual Reductio n in Crashes Per Mile Raised Median Diyjd;r 2.2 31.2 Cootinuous Two-Way Left-Tum L ane 4 4 This research also obtained data on II sections of roadways with medians and channelized left-tum lanes in Texas. In view of the small number of observations, the researchers did not develop a regression equation for tbe type of design. However the equation for the continuous two-way left-tum Jane was used to estimate the number of crashes that might be expected if a CLTL were used instead of a raised median. The regression equation estimated that the number of crashes per mile w ith a CL TL would be substantially larger than the number of crashes actually occurring with the raised, non traversable median (see Table 9). 29 November 1994 IS


This is a blank page


Table 9-Estimated Crashes with Continuous Two-Way Left-Turn Lanes Compared to A ctual Crashes with Channelized Left-Turn Bays Number Euimared Error of CLTL Actual (Estimated Through Signals Driveways Crubeo .. CLTL Lanes ADT Population Per Mile Per Mile. Per Mile. Per Mile ActuaD 6 29562 407,000 4.17 39.6 145. 5 1 66 7 1.2 6 31134 4 .65 39. 5 153. 2 127. 9 +2S.3 6 32106 3 .13 84.4 164.3 253.1 +88 7 4 15483 0.0 16.1 67.1 41.9 +2S.2 4 13921 0 0 31.3 71.4 12..5 + 58.9 4 13591 0 0 0 0 55.4 9.4 +46 0 4 14477 0.0 81.8 97.3 6S. 9 + 31.4 4 14477 0 0 100.0 106. 3 76 3 +30.0 4 14477 2.1 62.5 1 07.0 64 9 +42.4 4 8323 283,700 0 0 17. 0 31.4 36 2 4 .8 6 13660 3.2 35.5 81.0 29.0 +52. 0 4 17 1 97 407,000 0 0 23.3 74.1 46.4 +27 6 2 13223 283,700 2 0 56. 0 78.9 66.0 +12.9 2 11367 2.9 5 9 59.2 35.3 + 23.9 Soun:e: Ref=cc (.Jll), 1978 Recent (1989) research sponsored by the Georgia DOT concluded that on high volume roadways, nontratersable medians have a l o wer crash experience than roadways with continuous two-way left-turn Janes Ul). As shown in Figure 1, this is true for both 4-lane and 6-lane facilities. The ayerage crash rate for 4-lane divided roadways was about IS% lower than those havinJ 4 traffic lanes plus a CLTL. The crash rate on 6-lane divided roadways was about 2S0.4 less than comparable facilities with aCLTL. Studies in 1993 and Michigan (W, 1983 also found that roadways with nontraversa.ble medians have a much better safety than those with CJ,.TL's. Again, this is true for both roadways with 4 throup lanes as well as 6 throup traffic lanes The Florida data, Fipre 2, show that the 4-lane sections with traversable medians have an average crash rate which is 25% lower than those with a CLTL. For 6-lane roadways, the crash rate is about 12".4 Jess. 29 November 1994 17


This is a blank page


5 4 3 2 1 0 Slane I 41ane Raised Median J Source: Adapted from 1993, Aorida DOT, Reference Q2) Figure 2-Crash Rates for Raised Medians and Two-Way Left-Turn Lanes in Florida Table 10 Michigan Statewide Crash Rates for Selected Types of Arterials, 1985-1987 Arterial Average Length Type API Miles 22,000 223 4-lane 17,000 286 divided ratio, 4-Jane divided/S-laDe 7-!ane0> 35,000 29 50 ,000 44 divided ratio, 6-lane dividcd/7-Jane 0>-rb.e odd lane is a continuous 2way lcfttum Jane Source: Reference (J.!) I 988 l November 1994 Reponed CrlUbes per 100 Million Ydlis;ic-Miles Total Injury Fatal 956 276 2.55 41J7 118 1.77 0.42 0.43 0.69 1107 357 3.75 563 166 0.94 0.51 0.46 0 .25 19


This is a blank page


Table 13 -Summary of Suburban Midblock Vehicular Crash RatesU l b y Crash Type Crub T)'l!' Rear-end Right Allgle Head-On Left Tum Other < >erasb rate in c:ra.hes per 100-milloa vehicle-mileo Souroc: Adapted from Refereace (W, 1994 Rajyrl CLTI. 80 .98 1 39 .61 35.05 63. 26 1.34 2.55 24 .35 52.50 47.52 53.45 The Georgia, Florida and Michigan data clearly demonstrate that 4-lane and 6-lane dividedroadwayswithnontraversablemedianshaveamuchbettersafetyrecord(loweraverage crash rates) than 5-Jaoe and 7-lane roadways where the odd lane is a CLTL. More recent research comparing roadways with raised medians with CLTL's with undivided roadways summarized in the next section substantiates these flndings Raised v CL TL -v-Undivided An extremely well designed and executed study by Bowman and Vecellio (W. 1994 compared the crash e:tperience on urban arterials in Atlanta, Georgia ; Phoenix, Arizona; and Los Angele&'Pasadena, California. Fifteen homogeneous sections each of raised median, continuous two -way left-turn lane (CL TL) and undivided roadways were studied. Most of the mileage (84 .7"/o) was l ocated in suburban areas. Theremainder(15.3%) were located in central business districts (CBD's). The number of miles by location, median type, and number of through traffiC lanes is summarized in Table 14. The mean pedestrian accident rate for CBD's was statistically significantly (95%confldence level) largerthanforsuburban areas fo r all three median types However, the vehicular crash rates were not significantly different. Bowman and Vecellio concluded that, in suburban areas, raised m edians are safer than C LTL's or undivided roadways (see Table 1 5 ) 29 November 1994 2 1


This is a blank page


These results indicate that in CBD areas. raised medians and CLTL's do not have significantly different vehicular crash rates. However, in suburban areas the raised median is significantly safer than either the CLTL or an undivided roadway. Table 16 presents data from a Florida study 00 which compared crash rates by severity. The fatal crash rates are not meaningful because the number of fatal crashes for most median types was very small Inspection of Table 16 reveals the following: Undivided roadways have sub6tantially higher crash rates (both total crashes and midblock crashes) than roadways with CLTL's The average total and midblock crash rates are about the same for CL TL's and flush paved medians Flush grass medians have the lowest crash rates The crash rates for restrictive medians are about one-third lower than for non restrictive medians These fmdings are compatible with previous research. As shown in Table 17 crash severity was reduced where left-turn lanes were added to existing facilities The percent reduction in crash rates is much larger at unsignalized median openings than at signalized intersections Table 16-Crash Rates per M illi o n V ehicle-Miles ofTravel on4-LaDe Urban Arterials Accident Severity All CaWs Total Fatal lDjury roo< Midblgs;k S:ti!ib=t Total Fatal lDjury pooro 1CL TL plus flush paved '-"Flush grass plus raised 0'Propeny damage only Undiyidrcd CLTI. 4 .44 3 :ro O.oJ 0 .02 2 36 1.73 2.05 1.45 2 43 1.66 0 .02 0.02 1.28 0 90 1.13 0.15 Source: Adapted from Refereno: (46). 1993 I November 1994 Flush !'jiVed 3 25 O.oJ 1.79 1.44 1.71 0.02 0.95 0 .7 5 23 Mt.diaD :rs Nou(l) Remjctiye 3 21 0.02 1.74 1.45 1.67 0.02 0 .90 0.75 Flush Bai 1.80 2.46 O.o2 0.02 1.01 1.35 0 .77 1.09 0.97 1.26 0.01 0.01 0.54 0.69 0.42 0.56 Rc;strict jve f!) 2 .09 O.o2 1.16 0.9 1 1.09 0.01 0.60 0 4 7


This is a blank page


Page 25 IS m1ss1ng


This is a blank page


_j . ,. ... Source: Reference (.ll) Figure 4 S igning fo r the Mic higan Directional Crossover I November 1994 27


This is a blank page


RETROFITS TO EXISTING MEDIANS Continuous two-way left-turn lanes (CL TL'S) on two major urban arterials (Memorial Highway and Jimmy Carter Blvd.) in the Atlanta, Georgia, metropolitan area were replaced with nontraversable medians. Memorial Highway, Atlanta Metro Area A CL TL on a 4 .34 mile section of Memorial Drive in Atlanta, Georgia, was replaced with a median. The general features are listed in Table 19. Six through traffic l anes were provided before and after the median retrofit. Construction was completed between the end of J uly 1989 and the end of September 1990. Table 19-Memorial Drive Features 4.34 mile section Beron; TWL T lane Replaced By Raised Median A&!: No Median Break at 7 Public Roads 14 Signalized tntersections with Public Row and Major Driveways Souroe : Reference (.ll) No median opening was provided at 7 minor street intersections when the raised median was installed. Fourteen signalized median openings were provided with public roads and major private access drives All except one provided foru turns The median design involved a 6-inch mountable curb and a 2-foot gutter; thus the face of the curb is 2 feet from the edge of the traffic lane. A before-and-after study was performed to assess the impact on safety. Table 20 sununarizes the changes in crash rates. The total crash rate was reduced by 37%; the injury crash rate dropped 48%. The fact that crash rates decreased at those intersections which remained open demonstrates that improved design and traffic control can result in lower rates in spite of the increased turning traffic at these openings. Total injury crashes decreased by I November 1994 29


This is a blank page


Page 31 IS m1ss1ng


This is a blank page


SAFETY BENEFITS OF IMPROVEMENTS TO EXISTING MEDIANS As shown by the data in Table 24, left-tum bays resulted in lower crash rates of both sig:naliml and unsignaliztd access points. It is to be noted that total ctashes as well as left-tum crash rates are lower when a left-tum bay is provided. The benefit ofbaving left-tum bays at unsignalized access locations is particularly evident. Table 24 Effea of Left-Tum Bays on Crash Ralcs Cryb Rates Umjmaljzgi SiiQljrtd No With No With Ld\ Left Ld\ Left Type of Tum Tum Turn Tum eram Lilli: Lw Left Turns 1.20 0.12 0.65 0.37 All Other 3.1S 0 .92 1.82 1.17 TOTAL 4.3S 1.04 2.47 I.S4 Sou=: (}ID, 1982 Bold numbers d

This is a blank page


Table 26Oakland Park Blvd. Features Source: Reference W), 1989 2.437 mile section ADT 46,200 in 1980 46,800 in 1988 45 mpb poot

This is a blank page


Inspection of Table 27 shows that crashes and crash rates were reduced by reconstruction of the median on Oakland Park Blvd. The reduction in in j ury crashes wer e insignificant However, the reduction in total crashes and property damage only (PDO)crashes declined by 22"/o and 37% respectively More signifu:antly, the total crash rate declined by 26% and the PDQ crash rate was reduced by 41%. Moreover, the reduction for both total crashes and PDO crashes is significant at the 95% confidence leve l. Rear-end, angle, left-tum and sideswipe crashes and crash rates all declined However, the decrease was statistically signifiCant, at the 95% confidence level, for rear-end and angle crashes only. The decline in rear-end crashes may be explainable by the decrease in the number of locations where conflicts occur between left-turning vehicles and through traffic and improved tum bay geometries. The decrease in angle crashes may be due to both the fewer number of median openings and the reduced number of conflict points at those median openings which remain. (A full median opening has 16 major and 16 minor conflict points, whereas a left-turnlu-tum opening has only one major and 6 minor conflict points.) I November 1994 37


This is a blank page


US 1 Stuart, Florida The median on a 2 627 mile section of US I in Stuart, Florida, was reconstructed (see FigureS) in the period between September 1983 and May 1985 The ADTwas about 41,000. The before and after crash rates are given in Table 28. Data were not available by rear-end, angle, and other types of crashes. The total crash rate declined by 25%. Property damage only accidents also declined by nearly 25%. These decreases in crash rates are very similar to that experienced on Oakland Park Blvd. The 28% reduction in tbe injury crasb rate is much different than the experience on Oakland Park Blvd. The analysis indicates that the crasb reductions were significant for all three categories ofcrasbes. Table 28 -US 1 Crash Data (I> Average Number of Crasbm Per Year B a1G1 m Percent TYP. Before Aller CbaDge Total :io1 170 15.4 Injury 94 77 -18 1 Propeny Damage 107 97 -14.0 < "Before period 9/80 8183; allcr period 6185 -S/88 l>cfoa; ii&t study period, yean 3.0 3.0 average ADT 28,630 3:2.510 exposure, million vehicle-miles 82.36 93.52 "'Crashes per million vchiclemileo Before 7 .32 3.42 3 91 "'Crash Reduction F""tor = cfoa;l (no crasbC3 after) (ADT after) crasbC3 prevented= N0 NA N0 = (crash rate before) (exposure after) N A = number of crashes after Significant at the 95% COilfidencc le 22.3 24.7 21.8


This is a blank page


Kellogg Avenue. Wichita, Kansas An undivided segment (length not reported) of East Kellogg Avenue in Wichita, Kansas was retrofitted with a nontraversable median (construction was completed in August 1968). The original design had 4 signalized intersections and one unsignalized median break Four unsignalized directional breaks at which left-turns only were permitted from Kellogg were subsequently added. Traffic volumes along KeUogg varied from an ADT of21 ,500 to 26,400 Total crashes decreased significantly after installation of the median. However, crashes at intersections increased (the increase was not statistically significant). Crashes on the residential streets in the vicinity of KeUogg did not increase. Vancouver, Britiah Columbia Arterial streets in Vancouver, British Columbia are spaced at approximately one kilometre intervals (.l6). The initial street system was constructed without left-tum bays. The city's engineering department developed a benefit/cost measure to evaluate and rank various tum bay projects. Each year the city spends about $2.5 million to construct 6 to 10 left-turn bays. These improvements have resulted in a 20% increase in through capacity and a 25% to 50"/o reduction in accident rates I November 1994 41


This is a blank page


OPERATIONAL BENEFITS OF MEDIAN IMPROVEMENTS Unfortunately, before and after data relative to traffic operations are not available regarding the effect of changes in median type spacing of median openings or the design of median openings. However, data are available from two studies conducted for FOOT One is a computer simulation The other is a comparison of two arterials having similar characteristics, where the median of one was modified while the other was not. The comparison of traffic operational characteristics on Sunrise Blvd. (which was not modified) and Oak l and Park Blvd. (which was modified), both in the Fort Lauderdale area, provides some insight on the benefits of increased median opening spacing and design (ll.J!). (The design of Sunrise Blvd. and the original design of Oakland Park Blvd. are shown in Figure SA; the redesigned Oakland Park median is sb,own in Figure 58. ) The value of and percent change in various operational measures are given in Table 30. Simulation usingTRAF-NETSIM indicated thataveragespeedisexpected to increase when a median of the design shown in Figure SA is replaced by one shown in Figure SB. Also, the average number of stops and t he delay per vehicle will decrease as the volume increases (see Table 29). Table 2 9 E ffects o f Increased Spacing and Limitati ons of Movements Aver;age Stope per Delay per VcbjcJC3{Hour Spml Vehicle Yebjclc 1000 + 4% 60'/o % 1500 + 10% -49% 51% 2000 +38% 58% -65% Source : Reference (.lJ) Figure 6 illustrates the benefits of reconstructing medians to increase the spacing of median openings and to limit movements at remaining openings Crashes per million vehicle miles decreased significantly on both Oakland Park Blvd. and US I, whereas crashes increased slightly on a compara ble arterial having a median similar to that on Oak l and Park and US I prior to their reconstruction. 29 November 1994 43


This is a blank page


due to other factors. For example, the effect of traffic signal timing and progression effects, driveway volumes, and congestion can have a substantial effect on average travel speed. These may mask the effect of the differences in median design. Brake applications were also observed on Sunrise and Oakland Blvds.; these data are slll11lllarizled in Table31. Only 3%ofthe brake application on Oakland Park were attributable to the median. Whereas resulted from traffiC using the median openings on Sunrise Blvd. This presents the clearest evidence that the longer spacing and left-tum/u-turn median is operationally superior to the closely spaced median breaks at which all m o vements are possible. Table 3 1 Comparison of B rake Applications o n Snnrisc Blvd. and Oakland Park Bl v d. Braltc appW:ations Anributablc to: Median Effects Traff'u: Coogostioa Tratru: Signals Pcdestrilllls '"92 ruas, 462 brake apptic:ations "'41 ruas, 229 brake apptic:atioDO Sourc:c: Adapred from Refereace Pen:eot SoR" Qak (and Parle'" 29% 3% 12% 25% SS% 72% ....!lli.. 100% 100% This research 03. also found that the traffic operation in the middle through lane ofOakland Park Blvd. is superior to that on Sunrise Blvd. Furthermore, it was also concluded that the Oakland Park Blvd. median design results in 62% less delay to left-turns and 38% lower delay to u-turns than the median design of Sunrise Blvd. The researchers gave three reasons for the substantially lower t urning delay on Oakland Park Bfvd. These are: "First, the number of conflicting movements has been reduced, thereby allowing the driver to concentrate specifically on choosing an acceptable gap in which to execute his maneuver. Second, the sight distance at the median opening locations has been improved. No opposing left-turning or u-turning or crossing vehicles can obstruct the view of the driver making the turning movement. 1 November 1994 45


This is a blank page


GUIDELINES OF LEFfTURN LANES Various guidelines or standards have been developed for left-tum lanes at median brealcs. Most notable are those proposed by Harmelink (W and modified by ITE Commiuee 4A-2 (i3) and the standards used by the Colorado DOT(!!). Harmelink 's work, and ITE, consider the turning and opposing volume. Recent research by Til (iS) considers the left-tum volume and the opposing volume as well as the advancing volume from which tbe left-turns are made. The guidelines given in Figure 7 are based on the following two criteria : I) M inimizing total vehicular delay; and 2) A 0.01 (1%) probability that a left-turning vehicle will interfere with a followin g vehicle. The horizontal lines at 325, 35 0 and 375 vph directional result from the conflict between a left-turning vehicle and a following through vehicle at a maximum probability ofO.O!. The research by Til considered various directional splits over a range of direction al volumes. This analysis indicated that the results are not sensitive to directional splits between 50150 and 70130. Therefore, it is sugg,ested that the a verage of the opposing and advancing volumes be used as the "directional" volume in Figure 7. This also simplifie5 the comparison with Harmel ink and the Colorado DOT curves which are for the advancing volwne only Comparison of the Til curves and Harmelink's (Figu re 8) shows that Til guidelines result in use of a left-tum lane at lower directional volumes. This is due to: I) The Harmel ink curve is based upon the probability of a gap in the opposing lane, and 2) The IT! curves consider whether a turning left from a through lane will affect a following advancing vehicle as weU as the opposing volume. The TTl curves also account for the fact that, under low advancing volumes, through vehicles can change lanes prior to slowing because of a left-turning vehicle on multilane roadways. The Til curves show that a left-tum lane should be provided at directional volumes of325, 350 and 37S vph or more, depending upon speed. Again, this is due to limiting the probability that a left-turning vehicle will interfere with a following advancing vehicle to O.Ql o r less. When compared to the Colora do DOT warrants (Fi gure 9), the Til curves are more liberal at low directional volumes (i.e., higher left-tum volumes are required). This is due to a combination of two factors. One, when the tum volume is high compared to the advancing volume, the change of a conflict with a following vehicle is small And two, at low advancing volumes, a driver of a following vehicle has ample opportunit y to change lanes to avoid a vehicle turning left from a through lane. The curves given in Figure 7, or similar guidelines (Harm elink) or standards (Colorado DOl) indicate when a left-tum bay is to be provided. Such curves are most applicable to rural I l'

This is a blank page


. 500 Oi' a Rarmtll.ok Curvo ., ,3 400 1-----, All Spetc!s !:1 300 0 .. 200 Q'tl :a 100 o 10 20 30 40 so 60 Left'IUrn Volume (Vehicles Per Hour) Source: Reference Figure 8 Com parison of Harmeliot and 1TI Curves 500 -.;-:; 400 "..::I I tl ""-.. 300 0 .. 200 13 Q'tl 100 0 0 --1TicuColorado DOT Warraou I\ .... ... -1 / -45MPH dloSSMPR ..... I 55MPB 10 20 30 40 LertTu= Volwne (Vehicles Per Hour) Source: Reference so 60 Figure 9 Comparison of the Til Cunes and Colorado DOT Warrants 1 November 1994 4 9


This is a blank page


Maneuver Distance The elements of a tum bay (both right and left) are illustrated in Figure I L As illustrated, the taper is included in the deceleration distance. The distance traveled during the driver's pen:eption reaction time adds an additional length to the total intersection maneuver distance. The tum bay should be designed so that a turning vehicle will develop a speed differential of 15 kmlh (IOmph) or less at the point it clears the through traffic lane. The length of the bay should allow the vehicle to come to a comfortable stop prior to reaching the end of the expected queue in the turn bay. o-.,_..., .. '--'--... ... rwt O.w'e -.Weft ...,... ..... ..... -!----Minimum Length---! 41 .... ,,-...--. u.-4t. -.............................. .. ,.....,. 4) .ltla'IM .,........ M '-oliet'CIIloft _.., ...... Y e -.. ., "- ..... ot """ b.lrft 01" M UtiC\Itecl ... __ Source : Reference Q2. 1!) Figure 11 Elements of the Functional Area of an Intersection l November 1994 51


This is a blank page


Table 32 Calculated Upstream Maneuver Distances Minimum Maneuver Distan<:e1 in Metres (Feet) '[)c,jta bls; Conditjon300 Limitios Cooditioos01 Sgm1 Dc;erlcmlion(ol) Tolal ( 1) T)r;rclcratjoa("> To tat" kmlll (mph) rneltc3 metres (feet) metres metrli$ so (30) 70 (225) 100 (325) so (170) 65 (2 15) 55 (35) 90 (295) 130 (425} 65 (220) 80 (270) 65 (40) liS (375) 1 60 (525) 85 (27S) 70 (335) 70 (45) 140 ( 465) 190 (630) lOS (340) 125 (405) 80 (SO) 170 (565) 230 (750) 125 (410) 145 (480) 90 (55) 205 (675) 265 (875) ISO (495) 170 (565) 95 (60) 240 (785) 305 (1005) 170 (565) 200 (655) (I) All values rou n ded to nearest 5 metres (5 feet) '"2. 5 second pen:eption-reaction time; 1.1 mis' (3. 5 fps2 ) averase deccleration while moving lateral l y into tum bay and an averasc 1.8 mps2 (6 f ps2 ) deceleration thereafter; 1 6 kmlh (10 mph) speed diff=tial 0'1.0 second per<:cptionreaction; 1.4 mls' (4. 5 fps1 ) deceleration while movinglatcrally into tum bay and an averase 2.7 mls1 (9 0 fps1 ) deceleration thereafter; 16 kmlh (10 mpb) speed diffe=tial 1'>oistu>ce to decelerate from s peed to a s top while maneuvering laterally into a left-rum or right-rum bay '"Deceleration distance plus diatanoe traveled in pen:eption-rea.:tion time Sou roe: Adapted from Refcrmce (W, 1993 As the distances in t h e last column (total distance) ofT able 33 indicate, the curb ret urn radius or the inside radius of a turning roadway of a channelized intersection has very little effect on the length of an auxiliary lane measured from the beginning of the ta per of the near driveway edge I November 1 994 53


This is a blank page


Page 55 IS m1ss1ng


This is a blank page


Besides the methods already noted for determining storage length, one other is noteworthy. This simply involves "rules of thumb." Rules of Thumb (I) Storage Length= I foot per vpb turning left during peak hour. Example: 300 vph, Storage Length = 300 feet. (2) Storage Length = (vpblnwnber of cycles per br) x (2) x (25 ft). Example: for 60 second cycle and 300 vph: 300 vpb/60 cycles per hour= 5 vehicles per phase (average) Storage Length = (5) x (2) x (25 ft) = 250ft. Example for a 120 second cycle: 300 vpb/30 cycles per hr = I 0 vehicles per phase Storage Length = (I 0) x (2) x (25 ft) = 500 ft. Evaluation Similar results are achieved when cycle lengths are short: 300ft. vs. 250 ft. However. the fll'St method (I ft. per vph) results in an extreme underestimate at long cycle lengths The second method gives results over a wide range of cycle lengths which are comparable to using queuing theory. Storage Length for Dual Left Turns The storage for a dual left -tum lane at a signalized intersection can be estimated by queuing analysis, or by the nomograph in Figure 12. The storage length is estimated for a dual left-turn bay by dividing this storage length by 1.8. This practice is suggested even though recent research ().I), 1986, has shown that the saturation flow rate for a dual left-turn bay is about the same as for two through traffic lanes. The use of the value 1.8 recognizes that the left-turn traffic is not equally distributed between the two tum lanes. Example calculations are given in Table 35. In unusual cases, the imbalance between dual tum lanes may be much greater. 1 November 1994 57


This is a blank page


i I f C.....-'IM1 !..dlowe a.-1-. ...... ,...... Figure 13ANomogaph f o r 4-Lane Highways "\ -..... ............... ... "' lolft " ......... ........,. ........ ... L t O Figure 138 Example Nomograph for 2 -Lane Highways Source : Reference (iZ), !967 Figure 13 Storage Guidelines for Unsignalized Intersections 29 November 1994 59


This is a blank page


FULL MEDIAN OPENINGS AND SIGNAL SPACING Full median openings should be allowed only at locations which are suitable for signalization Based on the need to operate a signalized system under both peak and off-peak conditions, it is recogniu:d that a uniform signal of0.804 km (0.500 mile) should be used (32. !!&. 82). This spacing allows traffic operations engineers to use long (120 second) cycle lengths in the peak period while maintaining a traffic speed, (48 kmlb., 30 mph) which provides reasonably high flow rates and good progression efficiency. This spacing also allows the traffiC operations engineer to achieve excellent progression using shorter cycles under lower volume situations. A rule of thumb has been that if a signalized intersection does not conform to the uniform spacing interval, the main street green must be increased by 1% of the cycle length for each I% the signal is out of position in order to maintain progression However, analysis of the informatio n contained in a recent paper by Stover, Demosthenes, and Weesner (llll) indicates that this is true at shorter cycle lengths (about 60 seconds). However, at long cycle lengths (120 seconds) the main street green must be increased by 2"/o of the cycle length for each I% the signal is out of position if main progression is to be preserved. Thus it is absolutely essential that arterial-to-arterial intersection spacings be an even (nearly exact) multiple of 0.804 km (0.5 00 mile) spacing. Minor cross-streets (i.e., arterial-collector intersections) can often be located with some deviation from the uniform interval. This is the case where the low volume on the cross-street approaches can be accommodated with less than 50% of the cycle length. At highenolumes, the cross-street can be "flared" to provide additional approach lanes. For example, a collector street which would normally have two approach Janes can be flared to have separate lanes for left-tum, through, and right-turn movements. Or it can even be flared to have two left-tum l anes in addition to separate through and right-tum lanes. Criteria for Signals Out of Position Whenever a cross-street, or signalized access drive cannot be located at the precise uniform 0.804 km (0.500 mile) interval, analyses should be required to determine if the proposed location can accommodate the expected traffic without interfering with progression on the major street. Criteria as to specific combinations of progressional speeds. cycle lengths. and minimum progression efficiency (progression bandwidth as a percent of cycle length ) need to be specified. The criteria as to minimum progression efficiency might become more liberal l November 1994 61 .


This is a blank page


EFFECfS ON BUSINESS The introduction of a raised median on an existing roadway in a developed area is often controversial. Highway officials recognize that crashes and delays will be reduced if the median is installed. The owners of roadside businesses with no direct opening opposite their entrance often believe that their businesses will suffer fmancially Before and after studies where nontraversable medians were installed on roadways in three Texas cities fotiDd that some businesses experienced a decline in sales while others had increases in sales after the median was installed. This occurred for both "traffic oriented businesses" (drive-in restaurants, service stations and motels) as well as "nontraflic oriented businesses. The rank correlation coefficient of 0.637 indicated that there is a positive correlation between rankings based on percent change in sales and percent change in left-tum volume. This relationship is statistically significant at the 99 percent confidence level. Construction of a non traversable median eliminated left-turns exceptatmedian openings; thus left turns after construction of the median were accomplished by marking "u-turns". Analysis of the sales after the installation of the raised median as a percent of the sales volume before the installation of the median reveals that businesses along the low-volume roadways suffered (see Figure 14). However, on the average, businesses along the high-volume route did not experience a decrease in sales; in fact a slight increase resulted. These data suggest that a nontraversable median is not detrimental to overall business activity when traffiC vohunes are high. 1 November 1994 63


This is a blank page


Attitudes The results of opinion surveys of interest groups using Oakland Park Blvd. are sununarized in Figure I 5. The fmdings reveal that all interest groups were of the opinion that safety was ilnproved, better traffic operations resulted, and had a favorable opinion regarding the change (see Figure 5 for the before and after conditions). Table 37 shows the percentage of different affected groups that favored the change in the median. It is interesting to note that those who were familiar with Oakland Park Blvd. before its reconstruction had a more fa vorable opinion of the change than those who were not familiar with the before condition. 90 80 70 Peroelvd Slllty Improvement TOTAL 65% S.ttar Trtdllc Operation on Through LaM. TOTAL 64o/. In Favor of Modified Median Source: Florida DOT, Reference CW Figure 15Attitudes Regarding Chmge In Median Design. I November 1994 65


This is a blank page


Most Business Owners Saw No Loss Of Profit Small Increase 5.2% Lrge Loss 12.5% No Answer 5.3% Small Loss 14.6% Source : Florida DOT, Reference Q2) Figure 16Reported Effect on Business Several closely spaced median openings on US I in Stuan, Florida, were also closed as a safety improvement strategy. This resulted in a 22"/o reduction in the number of vehicular crashes. !v; Table 38 shows, opinions regarding the median closures on this roadway were positive. Table 38-Opinions Regarding Closing Median Opening on US 1, Stuart, Florida Perception of smoother traffic flow 57% of the c:u.stomers favored the project in spite of iocooveaicnce 48% of the adjaa:ot residents favored the chang .. yet 63% rCj)Oned being incooveoienced majority of residents felt crime would be deterred 58'/o of trucken favored access control and only 25% felt they w.:re inconvenienced 92% of trucken felt safer Source: Florida DOT, Reference (.12) I November 1994 67


This is a blank page


PEDESTRIAN SAFETY IntrodU<:tion Measures to increase vehicular capacity such as a continuous two-way left-tum lane, intersection channelization, right-tum on red. multiple through lanes, al)dshort signal phases may result in potential problems for pedestrians. The pedestrian-vehicular conflict problem becomes increasingly acute as vehicular and pedestrian volumes increase. Medians are commonly used on major urban arterials to separate opposing traffic, to provide space for left-tum bays and restrict or prevent left-turns and/or crossing maneuvers of unsignalized intersections of public streets or private access drives. Medians also provide refuge for pedestrians and improved pedestrian safety. The medial refuge allows pedestrians to cross one traffic stream at a time. This in turn permits the use of shorter pedestrian clearance intervals. Pedestrian Accidents by Median Type Bowman and Vecellio (22) studied the pedestrian accident experience of arterial streets in Atlanta, Georgia; Phoenix, Arizona; and Los Angeles/Pasadena, California. Data for 15 street seginentsforeach of three median types (raised, continuous two-way left-tum lane and undivided) were analyzed Table 39 summarizes the accident rates and statistical comparison of the pedestrian accident rates Inspection shows that the mean average accident rates for raised medians is much less than for streets with continuous two-way left-tum lanes(CL TL's) and no median (undivided). However the difference between the pedestrian accident rates of raised medians and CLTL's was not statistically significant for streets located in CBD's or suburban areas The mean pedestrian accident rate on CL TL's in suburban areas was found to be comparable to that for undivided streets and twice the accident rate for raised medians. However the authors concluded that the difference between raised medians and CLTL's, was not statistically significant of the 95% confidence level. The accident rates for midblock locations at the intersections are shown in Table 40. Raised medians in suburban areas experienced much lowermidblock and intersection accident rates than CLTL's in CBD's. Perhaps this difference between the two types of areas is that CBD's have high pedestrian volumes and relatively very slow speed traffic flow. Whereas in 1 November 1 994 69


This is a blank page


Long, Gan and Morrison also investigated crashes invo l ving pedestrian s on urban arterials 00. The data show n i n Table 35 indicates tha t undivided 4-lan e urban arterials experi e nce a t otal pedestrian crash rate w hich is abo u t d oubl e that ofn onrestrictive medians and nearl y 5 times that of restrict i ve medians. Also, as may be ex pec ted midbloc k pedestrian crash rates are much lo wer with restrictive medians than undivided r oadways and those with CLTL's The data in Table 41 als o show total ped estrian cras h rates on 6lane arterials are slightly h i gher than on 4-lane arterials for nonrestric ti v e medians Except fo r CLTL's midblock cras h rates are abo u t double the rate on 4-lane roadways. Table 41 Crashes Involvin g Pedestrians on U r ban Arterials in Florida Mcdiaa !:La Type U ndivided0 ) CLTLm Flush Paved NonRestrictive"' Flush Grass Raised tllContinu oustwo-way le fl tum lan e "1:ndividual plus CL TL grass p l us raised Illlil 0 .18 0 .10 0.09 0 .10 O.Q3 0.04 0 04 "'There were no 6-lanc undivid ed u r ban ane riab Sou rce : Adap ted from Refcrena: o 1 993 I November 1994 Mjdblock 0 .11 0 06 0 04 o.os O.Q2 0.02 O.Q2 71 Lual M idblock NA 0 .11 0 0 7 0 .12 0 08 0 .11 O.Q7 O.OS 0 04 0 08 0 04 0 07 0 04


This is a blank page


SUMMARY The continuous two-way left-tum lane (CL TL) bas worked well as a retrofit in developed areas on existing sections of roadway having moderate traffic volumes and low driveway traffic. It is not applicable on high-volume, high-speed roadways with moderate to high driveway volumes Moreover, it is not compatible with the movement function appropriate for major arterial streets and highways. Roadways with nontraversable (raised, flush grass or depressed) medians have been found to have substantially lower crash rates than roadways having continuous two-way left turnlanes(CLTL's). This is true for botb+lanedivided -v-S-lane and6-lanedivided -v-7-lane roadways in both urban and rural areas. The research indicates that CL TL's in urban areas should be replaced by a raised median when traffic volumes exceed 24,000 vehicles per day (vpd). Thus, all new construction and reconstruction projects should incorporate a non traversable median where the projected traffic volume exceeds 24,000 vpd. Increasing the spacing ofunsignalized median openings and the redesign offull-median openings to allow leftturns/u-turns only also greatly reduces crash rates. Such improved median design decreases the crash rate at signalizt:d intersections as well as reducing the total crash rate. Crash rates involving pedestrians are also lower on roadways with restrictive medians (raised or flush/depressed grass) than on undivided roadways and roadways having CL TL's. This is true for both midblock and total crashes in which pedestrians were involved. Attitude surveys in the Fort Lauderdale area indicated that the majority of those affected favor improved median design even though they may be inconvenienced by the more restrictive design. Turn bays at signalized and unsignalized median openings improve traffic flow and greatly improve traffic safety, especially at high volumes and high speeds. I November !994 73


This is a blank page


Full M edian Opening Left-turn Ingress I November 1994 75 16 Major 0 16 M ino r 32 Total 1 Major o 6 M i nor 7 Total


This is a blank page


4. Unsignalized full median openings should be permitted in rural areas only. Care is to be taken that the median opening can be redesigned for directional movements should the areas urbanize in the future M

This is a blank page


Intended Maneuver Unintended Maneuver 1'-------------I Median Width Insufficient to Preclude Left-tum Ingress Intended "Wings' Owrlap . Movement 1-1 --1 Median Width Sufficient to Preclude the Left-tum Ingress 7 Median tum bay lengths (including bay taper) should be suflkiently long to allow a driver to move latterly into the turn bay and comfortably decelerate to a stop plus adequate queue storage. The minimum length will consist of two e!ernents: !) a !engl.b suflicit:Dt for a fell turning vehicle to move!Ilterally into the tum b/ly and dccelerate to a stop, and 2) the length required to store all lefi tumiDg vehicles with a very high prob/lbi!ity of success. The speed dillerential between a left turning vehicle and /.be through trallic sue:un should not exceed I 5 kmlb (10 mph) when the turning vehicle clear.; the through tra!Jic Jane. The queue storage should be suDicient to store all turning vehicles at fe3St 98% of the time during peak periods. The design /engl.b will be determined by /.be sum of /.be off-peak liJIUJeuver dist.wa: plus the off-peak queue storage or by /.be sum for /.be peak period maneuver distance plus /.be peak period queue storage, whichever is lo nger. In rural areas. the design length will be determined by tbe design speed, or posted speed, plus a minimum standard s(orage 29 November 1994 79


This is a blank page


9 The minimum spacing of unsignalized median openings on high speed (>45 mph, 72 kmlh) urban arterials should be at least 1320 (402.6 metres) feet. The. minimum spacing of unsigcalized median opeciDgs on slower speed ( <45 mph. <72 kmlh) multilane divided urban arterials should be at least 660 feet (20 1.3 metres). Priority .fOr mediao access must be giveJJ to sipalimJ infllrs:tioDS. the tum bays at these iotersectioDS Deed to be of suff!Cierlt ieDgth to aJJow dece!eratioD to a stop plus storage. At a major iotersection tbis is commonly several hundred feet. AUowiDg alefl-tum bay in the opposite directioo oftra vel wiD, 1) ioterferr: witb a future extel/Sioo of tbe turD bay for tbe sigtJali.zed intersectioo aodlor 2) result in ao excessivt! speed differential between vehicles using the UDSigiJalized mediao opeo.iog aodlor J) a Dll.tTOw $ shaped mediao. I 0. The minimum length of full width median between directionalleft-tum/u-turn bays serving traffic in opposite directions should be at least 50 feet. GO-width medi:J' I I I I Median 'nose' Median 'nose' A minimum length of full mediao widtb .is .oecessary to provide landscapiog so as to improve tbe visibility (aod aesthetics) ofthemediao. A IJaJTOW 2 to 6 feet (0. 6 to 1.8 metres) wick median ("spaghetti striog") whicb is several huodred feet /oog does oot provide good visual delioeJJ.tioo to drivers, especially at Digbt. 29 November 1994 81


This is a blank page


I 5. The width (face-to-face) of the median nose(median width between the left-turn bay lane and the opposing traffic lane) should be at least 6 feet (1.8 metres) in' urban areas. This is the commonly accepted millimum width where some pedestrians may be present. If moderate to higb pedestriu volumes may be encoUJJtered, a wider mediu nose will be needed. 16. The semi-drcular median and design should be used only where the width of median nose is less than 3 feet (0.9 metres). Tbe semi-circular design does not conform to vebicular tumiJJg paths. Tbe half bullet-nose better matcfies the tuming path ud substantially reduces vehicular hits. 29 November 1994 83


This is a blank page


APPENDIX A: MODELS Current Models Various researchers have developed regression models for calculating the expected nwnber of crashes per mile per year for different median treatments (11. l!2 .&8.. 2ll. 22). Walton and Machemehl (l!.8), and McCoy and Ballard (li2), developed models for two-way left-tum Janes (CLTL 's) only. Howard (2!1) and Squires and Parsonson (ll), developed models for crashes per mile per year as well as crashes per million vehicle miles. Parker first developed a prediction model in 1983 (aj), and a revised model in 1991 (.81). The several variables included in the various models are identified in Table A-1 (undivided roadways), Table A-2 (CL TL's) and Table A-3 (raised medians). Calculated crashes permileperyear are given in Table A-4 for the following conditions: I 00,000 population 50 driveways per mile 2 signals per mile 8 unsignalized intersections per mile 5%trucks commercial development adjacent to roadway 4 through lanes $250 crash report threshold 16-f oot median 40 mph speed limit It should be noted that the last three variables are unique to the Bowman Vecellio model. Inspection of Table A-1 reveals the following: I. The various models give quite different results for the same conditions. 2 The BowmanVecellio model consistently predicts fewer crashes on roadways with raised medians than on CL TL's and fewer crashes on roadways having a CL TL than on undivided roadways. 3. The raised median typically has tbe lowest predicted number of crashes The exceptions are the SquiresParsonson model for20,000 and 30,000 ADT's. And I No v ember 1994 85

PAGE 100

This is a blank page

PAGE 101

2 BowmanVecellio considered the number of signalized intersections per mile but found it not to be statisticaUy significant. Presumably because the number of signals per mile is correlated with other variables in the model such as the number of driveways and unsignalized intersections per mile and type of adjacent land development. McCoy-Ballard also found signals to be not significant for undivided on CL TL roadways as did Chatteljee, et a! for both raised medians and CL TL's and Squires-Parsonson for raised medians. 3. Crashes tend to decrease as speed increases. Bowman and Vecellio (22) explain this apparent conflict with logic by noting that higher speeds generally occur where development and traffic conflicts, are lower. It may be due to speeds being higher where there is good median and marginal access control. The BowmanVecellio model (Table A-2) appears to be the best by virtue of its logical and consistent results. Again. this may be due to its large, geographically diverse data base. It may therefore be the most The results produced by the Harwood model are not consistent with logic on other research (ll, 22. 23, 2.2, 1L 82, U .8!, .&.l). Similarly, some of the proposals regarding CL TL's and medians in NCHR P Report 330 (W also seem to be counter-intuitive and in conflict With other research. I November 1994 87

PAGE 102

This is a blank page

PAGE 103

00 \0 J -. Table A-2. Variables Included in CLTL Crash Prediction Models Cruhca Per Mile Mcdds VWbk SquitaA Cb&uajoc., Pu\.er 191) hn:ct 1991 ........... Clol APT X X X X X Pc:t Mik 0 0 0 0 Sipabta-W i k X X X 0 UoJijnali&cd Approtdlct Per M ile 0 Pubic Sttcet AfP'oadla Pet M ile X X X PcnxntTNCU Lcft T\Im Volume (k...efopncoc Type PteporUaa 'Thtcaho!d OfT: Lllld U te BIIUcu l.&Ad Ute Arco Tyj>< X Mcdiao Wdtb Foct CrCiaWf!CN; Pet N.ik YcanorCruh Data 3 )-4 Nu!Ub:r of Sccticma 11 s 42 12 Total Scctioo lcnalh. M i les 12.2 .. 62.S 19.7 lbroqh Ttaff'tc l..uel 4 4 4 a' 0 .7S 0.73 0 .60 0 6 S Jlccl tppr C &llm&OOr tt0l4.-oect at IIJJ'I.II&CCI Of Ultencc:ooiU Xnrial>k iodudcd ib model 0 vtri.ablc CCKI.sidc:tod by oot JtatistlcaUy .tipi(ICa!\1 a htl 1hw not indu.Jcd C Ytriab k is sianil".c:ant bot CDn'dttcd with a l)othu more ... ni(Qnl vtriablc Blank -variable 1)01 mttudcd na '11)1 8oMDae A Vcodio X X X X X 0 0 X 0 X ).S 171 SS.I 4&6 CtUbco pu MYM Mod
PAGE 104

This is a blank page

PAGE 105

"f lr '<> Table A-4 Number of C rashes Per Mile Per Year Predicted by Various Model$ ADT 10.000 20,000 Uodlrid )I lJ 110 II ao-.aa vecdlo ('2) ., <1 ll sa ........ ......... .., )9 n IM 60 Jl w to lowa'lln 31 )I Sl 16 Midb&ul:l.I!Uid lcr..Wu ()gi.J ""'""""(110) 26 :10 )6 ll 19 12 bt&ok t.Wt tMdc:l oot dcvciopocl rorlbb oriftck:pcadml varil.ble lt ouclidt of l'bcnn a c lliiCd IO develop ll:lf IIIQdd :10.000 CLTL R ollO! 5I 4 1 .. 71 J2.S "' 1110 121 7> 1110 n 11 12 ,. 10 19 71 40,0)0 UNIIYidcd CLTL Roi>od 11 19 101 94 ,. lJJ llO 101 2$J Ill &I 101 104

PAGE 106

This is a blank page

PAGE 107

Thr = accident reporting threshold $; Off = type of development adjacent to roadway (1 if office, 0 if other) ; Bus = type of development adjacent to roadway (I if business, o if other) ; Area = area type ( 1 if CBD, 0 if suburban) ; Med = media!l width, ft; Unsig = unsignalized intersection approach density, number of approaches/mile; Drv "' driveway density, number of driveways/mile; Cross = median crossover density. number of crossovers/mile; Spd = speed limit, mph; B, = regression coefficients for exposure variables; and C = regression coefficients for exp l anatory variables. Resean:h in Propess NCHRP Project 3-49 includes an evaluation of the various crash prediction models developed to date The project will also develop a safety model for the following IDidblock l eft tum treatments: Undivided Continuous two way l eft tum lane (CLTL) Raised median with left-tum bays NCHRP Project 3-49 is also to deve lop an operations model which will consist of the following 5 models : I Mainline volume adjustment models t o provide an initial check of left-tum volume to capacity 2 Mainlin e left-turn module calculate the capacity, delay and queue length of mainline left turns; also to calculate the impedance among overlapping left tum queues. 3. Mainline thru Jane modulesto evaluate the impact of leftturns on main line operations the moduie will calculate the delays and stops of through traffic and the through lane capacity. 4 Midblock sec t ion modules the t otal through traffic delay and stops as weU as a v erage travel speed in each direction will be calculated. I November 1994 93

PAGE 108

This is a blank page

PAGE 109

REFERENCES I. A Policy on the Geometric Desim ofHi&Jlways and Streets, American Association of State Highway and Transportation Officials, 1990 2. A Policy on the Geometric Desim of HiGhways and Streets, Americin Association of State Highway and Transportation Officials, 1984. 3. C. V. Wootan, H. G. Meuth N.J. Rowan, and T. G. Williams, "Median Study i.iJ Baytown, Texas," Research Report 8-I, Texas Transportation Institute, Texas A&M University, College Station, Texas, August 1964. 4. C V. Wootan, H. G. Meuth, N. J. Rowan, and T. G. Williams, "A Median Study in Pleasanton, Texas,";h Report 8-2, Texas Transportation Institute, Texas A&M Univ e rsity, College Station, Texas, August 1964. 5. C. V. Wootan, H. G. Meuth, N. J. Rowan, and T. G. Williams, "A Median Study in San Antonio, Texas," R!:$l'd!rch Report 8-3. Texas Transportation Institute, Texas A&M University, College Station, Texas, August I 964. 6. J. A. Azzeh, et al., "Technical Guidelines for the Control of Direct Access to Arterial Highways; Volume 1: General Framework for Implementing Access Control Techniques," RCl!Ort No. FHWA-RD-76-86, Federal Highway Administration, August 1975. 7. "Access Managemen t for Streets and Highways," RCl!Ort No. FHWA-IP-82-3, Federal Highway Administration, June 1982. 8. Karl Moskowitz and W. E. Schafer, "California Median Study," HRB Record 266, Highway Research Board, 1960. 9. J. L. Kay, L. G. NewdorfT, and F A. Wagner, "Criteria for Removing Traffic Signals," Technical Report, Report No, FHWA!RDI80-IQ4, Fmal Report, Federal Highway Administration, September 1980. 10. C. A Squires, "Criteria for Two-Way Left-Tum Lanes -vOther Medians, Vol. II: Accident Comparison of Raised Median and Two-Way Left-Tum Median Treatments," I Novembe r 1 994 95

PAGE 110

This is a blank page

PAGE 111

20. J. E. Wtlson, "Simple Types of Intersection Improvements, Improves! Street Utjljuujon Jhrou&h Traffic En&ineerin&, Special Report 93, Highway Research Board, 1967. 21. P C Box, "Driveway Accident and Volume Studies," Public Safety Systems, May/June 1967. 22. ]. I. Hanna, "Median Dividers on Urban Streets," Trnffic Ouarterly. Columbia University Press, Volume 7, October 1953. 23. W. A. Frick, '111e Effect of Major Physical Improvements on Capacity and Safety ; Traffic En&jneerini. December 1968. 24. K. Moskowitz, "Studies of Medians in Developed Areas," J>rns:clin&S of the Fouttrcmh California Street and Hi&hway Conference. University of California, 1%2, also Hi&hway Rl'i'i!'Ml;b News, No. 13, June 1964. 25. G. R. Garner and R. C. Deen, "Elements of Median Design in Relation to Accident Occurn:nce," Hi&hwi' JU::srtln;h Board Record 432. Highway Research Board, 1973. 26. J. C. GleMon, J. J. Valenta, B. A. Torson, J. A. Azzeh, and C. J. Wilton, "Evaluation of Techniques for the Control of Direct Access to Arterial Highways," Report No, FHwA-RD-76-85: "Technical Guidelines for the Control of Direct Access to Arterial Highways, Volume I: General Framework for Implementing Access Control Techniques," Report No. EHWARD-76-86: "Volume II, Detailed Description of Access Control Techniques," Report No. FHWA-RD-76-87, Federal Highway Administration, August 1975. 27. E. T. Telford, "A Study of Highways with a Narrow Median," Hi&hway Research Board BuUetin 35, Highway Researcli Board, 195 I. 28. R V Priest, "Statistical Relationships Between Traffic Volume, Median Width, and Accident Frequency on Divided Highway Grade Intersections," Hi&hway Research News. NoJ3. Highway Research Board, June 1964. 29. Brian L. Bowman and Robert L Vecellio, '111e Effect of Urban/Suburban Median Types on Both Vehicular and Pedestrian Safety," Paper presented at the 73rd AMual Meeting of the Transportation Research Board, January 1994 I November 1994 97

PAGE 112

This is a blank page

PAGE 113

41. V. G. Stover, Access Control Issues Related to Urban Arterial Intersection Design," Transportation Research Record. Number 1385, Transportation Research Board, Washington, D.C., 1993. 42 M.D. Harmelink, "Aspects of Traffic Control Devices: Volume Warrants for Left Turn Storage Lanes at Unsignalized Grade Intersections, Hi&)!way Research Record, No. 211, Highway Research Board, Washington, D.C., 1967. 43. ITE Conunittee 4A-2, "Design and Use of Two-Way Left Tum Lanes," ITE Journal, Washington, D.C., February, 1981. 44. "The State Highway Access Code," Department of Highways. State of Colorado. As amended by the Colorado Highway Commission. Denver, CO., August 15, 1985 45. P. E. Hawley, "Guide l ines for Left-Tum Bays at Unsignali:red Access Locations on Arterial Roadways," Thesis, Texas A&M University, August 1994. 46. G. D. Long, C. T Gan and B. S. Morrison, "Safety Impac1S of Selected Median and Access Design Features," Report to the Florida Department of Transportation, Transportation Research Center, University of Aorida, May 28, 1993. 47. Institute of Transportation Engineers, "Effectiveness of Median Storage and Acceleration Lanes for Left-Turning Vehicles," Information Report. 1986. 48 V. G. Stover, "Urban Street Design," Short Course Notebook, Texas Engineering Extension Service, 1994. 49. V. G Stover, S C. Tignor, and M. J. Rosenbaum, Aocess Control and Driveways," Synthesis of Safety Research Related to Traffic Control and Roadway Elements, Vol. I, Report FHWA-TS-82-232, Federal Highway Administration December 1982 50. G. F. Hagenauer, J. Upchurch, D. Warren and M J. Rosenbaum, "Intersections," Synthr$3 of Safety Research Related to Traffic Control and Roadway Elements, Vol. I, Report FHWA-IS-82-232, Federal Highway Administration, December 1982. 51. D. W Harwood "Effective Utilization of Street Width on Urban Arterials, NCHRP Report 330. Transportation Research Board, August 1990. 1 November 1994 99

PAGE 114

This is a blank page

PAGE 115

64. "Evaluation of Four Saf e ty Projects," Traffic and Safety Division, Department of State Highways, Lansing, TSD-G-207-2, 1972. 65. J. C. Ray, "Evaluation of a Two-Way Median Left-Turn Lane," Traffic En&jneerin&, March, 1961. 66. R B. Sawhill and D. R. Neuzil, Accidents and Operational Characteristics on Arterial Streets with TwoWay Median Left-Turn Lanes." Hi&llwa,y Research Record 31, HRB, National Research Council Washington, D.C., 1963. 67. R. D Lewis, et al, "A Study of Two-Way Left-Turn Lanes By Technical Counc i l Collllllittee No. 10, ITE, Technical Notes, December 1976. 68. M. R. HotTman. 'Two WayLeft-TumLanesWorkl",IxafficEo&ineerin&, Vol. 44,No. I I, August 1974, pp 24-27. 69. R. C. Thomas. "Continuous Left-Tum Channelization and Accidents," Traffic Engineering, December 1966. 70. P T. McCoy, J. L. Ballard, D. S Eitel and W E. Witt, "Two-Way Left Turn Guidelines for Urban Four-Lane Roadways," TrapspOflatjOJI RC!II:jlrch Record 1195, 1988. 71. Karl Moskowit2, "Accidents on Freeways in California," World Traffic Engineering Conference, I 961. 72. P. D. Cribbins : "Location of Median Openings on Divided Highways," Traffic En&ineerioa. Apri11967. 73. J. Emerson, "Speed of Car on Sharp Horizontal Curves," T.nlffic Eneioeerine and Control, July 1969. 74. J Emerson. "A Note on Speed Road Curvature Relationships," Traffic En!Poeerioe and Con troL November 1970. 75. R. D. Worrall and A. G R. Bullen, "An Empirical Analysis of Lane Changing on Multilane Highways." TRBRecord 303, Transportation Research Board. 1970. 76. J E Leish. "At Grade Intersections a design reference book, Jack E. Leish and Associates, undated. I November 1994 101

PAGE 116

This is a blank page

PAGE 117

87 M R. Parker, Jr. "Simplified Guidelines for Selecting an Urban Median TreatmentEngineer's Guide," Virginia Department ofTransportation, Riclunond, Virginia, July 1991. 88. C. M Walton and R B Machemehl, "Accident and Operational Guidelines for Continuous Two-Way Left-Turn Medians," Transportation Research Record 737 1979. 89. P T. McCoy and J L. Ballard, "Cost-Effectiveness Evaluation ofTwo-Way LeftTum Lanes on Urban Four-Lane Roadways; Report No. NE-DORR87-1, Nebraska Department of Roads, Lincoln, Nebras ka, August 1986. 90 D. W. Harwood, "Multilane Design Alternatives for Improving Suburban Highways." NCHRP Report 282, Transportation Research Board ; Washington, D C 1986. 91. D. W Harwood, "Effective Utilization of Street Width on Urban Arterials," NCHRP Report 330, Transportation Research Board, Washington, D.C., 1986. 92. B. L. Bowman and R. L. Vecellio, "Investigation of the Impact of Medians on Road Users," R eport No, fHWA-RD-93130, Federal Highway Administration, D.C. 1994. 93. P. T. McCoy, J. L. Ballard, D S Eitel, and W. E. Will, "Two -Way Left-Tum Lane Guidelines for Urban Four-Lane Roadways," Transportation Rese a rch Record I 125 Transportation Research Board, 1988. 94 D W. Harwood and J. C. Glennon, "Selection of Median Treatments for Existing Arterial Highways." Transportation Rt::;tarch Record 68l, Transportation Research Board, 1978. 95. J. B Rollins, J. L Memmott, and J L. Buffmgtoo, "Effects of Roadway Improvements on Adjacent Land Use: An Aggregate Analysis and the Feasibility of Using Urban Development Models," Reoort No. 225-22, Texas Transportation Institute, Texas A&M University, College Station, Texas, May 1991. 96. J. C. Glennon, J J Valenta, B A Torson, and J. A Azzch, Technical Guidelines for Control of Direct Access to Arterial Highways, Vol. I, "General Framework for Implementing Access Control Techniques," Beggrt No, FHWA-RD-76-86, Vol. II, l November !994 103

xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 2200349Ia 4500
controlfield tag 001 029364666
005 20110902142117.0
006 m o d
007 cr bn|||||||||
008 110902s1994 flua ob 000 0 eng d
datafield ind1 8 ind2 024
subfield code a C01-00021
1 100
Stover, Vergil G.,
d 1933-
0 245
h [electronic resource] :
b a discussion paper /
by Vergil G. Stover.
[Tampa] :
University of South Florida, Center for Urban Transportation Research,
1 online resource (xiv, 104 p.) :
Prepared for the Florida Dept. of Transportation.
Includes bibliographical references (p. 95-104).
Description based on print version.
Median strips
x Design and construction.
Design and construction.
arterial highways /urban/
Dept. of Transportation.
University of South Florida.
Center for Urban Transportation Research.
i Print version:
Stover, Vergil G., 1933-
t Medians.
[Tampa] : University of South Florida, Center for Urban Transportation Research, [1994]
w (OCoLC)32094756
Center for Urban Transportation Research Publications [USF].
4 856