0

TRANSPORTATION RESEARCH BOARD EXECUTIVE COMMITTEE 1995

OFFICERS
Chair: Lillian C. Borrone, Director, Port Department, The Port Authority of New York and New Jersey
Vice Chair: James. W. VAN Loben Sets, Director. California Department of Transportation
Executive Director: Robert E. Skinner, Jr., Transportation Research Board

MEEMEBERS
EDWARD H. ARNOLD, Chair and President, Arnold Industries, Lebanon, PA
SHARON D. BANKS, General Manager, AC Transit, Oakland, CA
BRLAN J. L. BERRY, Lloyd Viet Berkner Regental Professor & Chair, Bruton Centerfor Development Studies, University of Texas at Dallas
DWIGHT M. BOWER, Director, Idaho Department of Transportation
JOHN E. BREEN, The Nasser L Al-Rashid Chair in Civil Engineering, The University of Texas at Austin
WILLIAM F. BUNDY, Director, Rhode Island Department of Transportation
DAVID BURWELL, President, Rails-to-Trails Conservancy
A. RAY CHAMBERLAIN, Vice President, Freight Policy, American Trucking Associations, Inc. (Past Chair, 1993)
RAY W. CLOUGH, Nishkian Professor of Structural Engineering, Emeritus, University of California, Berkeley
JAMES N. DENN, Commissioner, Minnesota Department of Transportation
JAMES C. DELONG, Director ofAviation, Denver International Airport, Denver, Colorado
DENNIS J. FITZGERALD, Executive Director, Capital District Transportation Authority, Albany, NY
JAMES A. HAGEN, Chairtnan of the Board and CEO, CONRAIL ,
DELON HAMPTON, Chairman & CEO, Delon Hampton & Associates
LESTER A. HOEL, Hamilton Professor, Civil Engineering, University of Virginia
DON C. KELLY, Secretary and Commissioner of Highways, Transportation Cabinet, Kentucky
ROBERT KOCHANOWSKI, ExecutiveDirector, Southwestern Pennsylvania Regional Planning Commission
JAMES L. LAMMM, President & CEO, Parsons Brinckerhoff, Inc.
CHARLES P. O'LEARY, JR., Commissioner, New Hampshire Department of Transportation
JUDE W. P. PATIN, Secretary, Louisiana Department of Transportation and Development
CRAIG E. PHILIP, President, Ingram Barge Co., Nashville, TN
DARREL RENSINK, Director, Iowa Department of Transportation
JOSEPH M. SUSSMAN, JR East Professor, Civil and Environmental Engineering, MIT (Past Chair, 1994)
MARTIN WACHS, Director, Institute of Transportation Studies, University of California
DAVID N. WORMOLEY, Dean of Engineering, Pennsylvania State University
HOWARD YERUSALIM, Secretary of Transportation, Pennsylvania Department of Transportation

MIKE ACOTT, President, National Asphalt Pavement Association (ex officio)

ROY A. ALLEN, Vice President, Research and Test Department, Association ofAmerican Railroads (ex officio)
ANDREW H. CARD, JR., President and CEO, American Automobile Manufacturers Association
THOMAS J. DONOHUE, President and CEO, American Trucking Associations (ex officio)
FRANCIS B. FRANCOIS, Executive Director, American Association of State Highway and Transportation Officials (ex officio)
JACK R. GILSTRAP, Executive Vice President, American Public Transit Association (ex officio)
ALBERT J.- HERBERGER, Maritime Administrator, US. Department of Transportation (ex officio)
DAVE) R. HINSON, Federal Aviation Administrator, US. Department of Transportation (ex officio)
GORDON f. LINTON, Federal transit Administrator, U.S. Department of Transportation (ex officio)
RICARDO MARTMEZ, Federal Railroad Administrator, U.S. Department of Transportation (ex officio)
JOLENE M. MOLITORIS, Federal Railroad Administrator, US. Department of Transportation (ex officio)
DAVE SHARMA, Research and Special Programs Administrator, US. Department of Transportation (ex officio)
RODNEY E. SLATER, Federal Highway Administrator, US. Department of Transportation (ex officio)
ARTHUR E. WILLIAMS, Chief of Engineers and Commander, U.S. Army Corps of Engineers (ex officio)

NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM

Transportation Research Board Executive Committee Subcommittee for NCHR

LILLIAN Q. BORRONE, Port Authority of New York and New Jersey (Chair) ROBERT E. SKINNER, JR., Transportation Research Board
FRANCIS B. -FRANCOIS, American Association of State Highway and RODNEY E. SLATER, Federal Highway Administratiori
Transportation Officials JOSEPH M. SUSSMAN, Massachusetts Institute of Technology
LESTER A. HOEL, University of Virginia JAMES W. VAN LOBEN SELS, California Department of Transportation

Field of Design Area of General Design Project Panel C15-14

MARK A. MAREK, Texas Department of Transportation (Chair) NORMAN H. ROUSH, West Virginia Department of Traniportation
MARK J. BERNDT, Minnesota Department of Transportation JOHN SANFORD, Illinois Department of Transportation
JOHN C. GLENNON, John C. Glennon, Chartered, Overland Park KS BOB L.'SMITH, Kansas State University
KENNETH HINT2MAN, CALTRANS, Sacramento, CA ROBERT L. WALTERS, Arkansas Department of Transportation
LARRY KING, Falls Church, VA JUSTIN TRUE, FHWA Liaison Representative
PATRICK T. McCOY, University of Nebraska D. WM. DEARASAUGH, JR., TRB Liaison Representative

Program S
ROBERT J. REILLY, Director, Cooperative Research Programs RONALD McCREADY, Senior Program Officer
CRAWFORD F. JENCKS, Manager, NCHRP FRANK R. McCULLAGH, Senior Program Officer
LLOYD R. CROWTHER, Senior Program Officer KENNETH S. OPIELA, Senior Program Officer
B. RAY DERR, Senior Program Officer SCOTT A. SABOL, Senior Program Officer
AMIR N. HANNA, Senior Program Officer EILEEN P. DELANEY, Editor
KAMI CABRAL, Assistant Editor
NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM

Report 375

Median Intersection Desi'gn

-

DOUGLAS W. HARWOOD
Midwest Research Institute
Kansas City, MO
and
MARTIN T. PIETRUCHA, MARK D. WOOLDRIDGE, and ROBERT E. BRYDIA
Pennsylvania Transportation Institute
University Park, PA
and
KAY FITZPATRICK
Texas Transportation Institution

Subject Areas
Highw ay and Facility Design
Highway Operations, Capacity and Traffic Control
Safety and Human Performance

Research Sponsored by the American Association of State.

Highway and Transportation Officials in Cooperation with the
Federal Highway Administration

TRANSPORTATION RESEARCH BOARD

NATIONAL RESEARCH COUNCIL

NATIONAL ACADEMY PRESS

Washington, D.C. 1~95
NATIONAL COOPERATIVE HIGHWAY RESEARCH NCHRP REPORT 375
PROGRAM

Systematic, well-designed research provides the most effective Project 15-14(2) FY'92
approach to the solution of many problems facing highway admin-
ISSN 0077-5614
istrators and engineers. Often, highway problems are of local inter-
est and can best be studied by highway departments individually ISBN 0-309-05704-3
or in cooperation with their state universities and others. However,
L. C. Catalog Card No. 95-62224
the accelerating growth of highway transportation develops in-
creasingly complex problems of wide interest to highway authori-
ties. These problems are best studied through a coordinated pro- Price $25.00
gram of cooperative research.
In recognition of these needs, the highway administrators of
the American Association of State Highway and Transportation
Officials initiated in 1962 an objective national highway research
program employing modem scientific techniques. This program is
supported on a continuing basis by funds from participating member
states of the Association and it receives the full cooperation and
support of the Federal Highway Administration, United States De- NOTICE
partment of Transportation.
The project that is the subject of this report was a part of the National Cooperative
The Transportation Research Board of the National Research ffighway Research Program conducted by the Transportation Research Board with
Council was requested by the Association to administer the research the approval of the Governing Board of the National Research Council. Such approval
reflects the Governing Board's judgment that the program concerned is of national
program because of the Board's recognized objectivity and under-
importance and appropriate with respect to both the purposes and resources of the
standing of modem research practices. The Board is uniquely suited National Research Council.
for this purpose as it maintains an extensive committee structure The members of the technical committee selected to monitor this project and to
from which authorities on any highway transportation subject may review this report were chosen for recognized scholarly competence and with due
consideration for the balance of disciplines appropriate to the project. The opinions
be drawn; it possesses avenues of communications and cooperation
and conclusions expressed or implied are those of the research agency that performed
with federal, state and local governmental agencies, universities, the research, and, while they have been accepted as appropriate by the technical
and industry; its relationship to the National Research Council committee, they are not necessarily those of the Transportation Research Board,
the National Research Council, the American Association of State Mghway and
is an insurance of objectivity; it maintains a full-time research Transportation officials, or the Federal ~Iighway Administration, U.S. Department of
correlation staff of specialists in highway transportation matters to Transportation.
bring the findings of research directly to those who are in a position Each report is reviewed and accepted for publication by the technical committee
to use them. according to procedures established and monitored by the Transportation Research
Board Executive Committee and the Governing Board of the National Research
The program is developed on the basis of research needs identi- Council.
fied by chief administrators of the highway and transportation
departments and by committees of AASHTO. Each year, specific
areas of research needs to be included in the program are proposed
to the National Research Council and the Board by the American
Association of State Highway and Transportation Officials. Re-
search projects to fulfill these needs are defined by the Board,
and qualified research agencies are selected from those that have
submitted proposals. Administration and surveillance of research
contracts are the responsibilities of the National Research Council
and the Transportation Research Board.
The needs for highway research are many, and the National
Cooperative Highway Research Program can make significant con-
tributions to the solution of highway transportation problems of
Published reports of the
mutual concern to many responsible groups. The program, how-
ever, is intended to complement rather than to substitute for or NATIONAL COOPERATIVE HIGHWAY RESEARCH PROGRAM
duplicate other highway research programs.
are available from:

Transportation Research Board

National Research Council
Note: The Transportation Research Board, the National Research Council, the 2101 Constitution Avenue, N.W.
Federal Highway Administration, the American. Association of State Highway
Washington, D.C. 20418
and Transportation Officials, and the individual states participating in the National
Cooperative Highway Research Program do not endorse products or manufac-
turers. Trade or manufacturers names appear herein solely because they are
considered essential to the object of this report. Printed in the United States of America
FOREWORD This report describes the development of design guidelines for the selection of median
widths for at-grade intersections on divided highways with partial or no control of access.
BY Staff It includes, as an Appendix, recommended revisions to the 1994 AASHTO Green Book,
Transportation Research A Policy on Geometric Design of Highways and Streets. The contents of this report are
Board therefore of immediate interest to highway and facility designers; highway-operations,
capacity, and traffic-control'personnel; and others concerned with safety and human
performance. The report's conclusions are based on field observations of traffic operations
at selected divided highway intersections; a statistical analysis of relationships between
median width and intersection-accident experience using a statewide accident, geometric,
and traffic volume data base; and the use of existing traffic simulation models.

Operational design of median widths and geometric configurations at intersections

to meet the needs of rural and suburban traffic has long been a concern. As noted in
AASHTO's Policy on Geometric Design of Highways and Streets, the width of medians
at intersections on highways with partial or no access control is critical. to their operation
and safety.
Median-width research efforts have, for the most part, addressed freeway operations
only. These efforts have provided a significant amount of information on cross-section
design and optimum width. However, little or no research has been directed toward
developing guidelines for median widths at intersections with partial or no access control.
Median width may pose operational problems in the vicinity of intersections for (1)
left-turning traffic from the main roadway and (2) crossing or left-turning traffic from
crossroads. These problems may be created or compounded by other factors, such as the
expanse of pavement area, inadequate storage for turning or cros ' sing vehicles, restricted
sight distance at the intersection approaches, and violations of driver expectations for
traffic movements (for example, drivers tend to become confused about intended opera-
tional characteristics of the multiple intersections encountered).
Currently, the focus of many transportation agencies is on the construction or recon-
truction of multilane facilities having partial or no access control. Because current design
policy does not adequately address median widths or intersection treatments on these
types of facilities, Midwest Research Institute was awarded NCHRP Project 15-14(2) to
develop and recommend median-width parameters and design criteria for intersections
on rural and suburban highways with partial or no access control.
Guidelines were to be developed for new and reconstructed facilities, considering at
least the following: (1) Safety, (2) Traffic Operations, (3) Traffic Volumes, (4) Type of
Traffic Control, (5) Design Speed of the Facility, and (6) Traffic Characteristics. Second-
ary issues included intersection configuration and spacing. The ultimate goal of the
research was to develop recommended guidelines prepared in a format suitable for adop-
tion by AASHTO. The proposed guidelines can be found in the Appendix at the end of
this report; the report itself describes the research justifying the recommended guidelines.
 The research found that at rural unsignalized intersections both accidents and undesir-
able driving behavior decrease as the median width increases. Conversely, at suburban
signalized and unsignalized intersections, accidents and undesirable behavior increase as
the median width increases. A more detailed description of the findings is outlined in the
summary of findings at the beginning of the report, and the confirming research is detailed
throughout the report.
 CONTENTS

I SUMMARY

3 CHAPTER ONE Introduction and Research Approach

Research Problem Statement, 5
Research Objectives and Scope, 6
Research Approach, 6
Organization of This Report, 8
9 CHAPTER Two Findings
Data Collection and Analysis Overview, 9
Current Design Policies and Practices of Highway Agencies, 10
Safety Characteristics of Divided Highways and Divided Highway
Intersections,'22
Traffic Safety and Operational Effects of Median Width, 27
Traffic Control at Divided Highway Intersections, 53
Truck Considerations, 59
Left-Turn Treatments, 59
Intersection Sight Distance,.60
Human Factors/Driver Perception Considerations, 61
Driver Expectancy Issues, 63
67 CHAPTER THREE Interpretation, Appraisal, Application ,
Key Factors in Selecting the Median Width at Divided Highway
Intersection, 67
Advantages and Disadvantages of Various Median Widths, 67
Feasible Left-Turn Treatments for Various Median Widths, 69
Guidelines for Selecting Median Widths at Divided Highway Intersections,
71
Design and Marking of the Median Roadway to Minimize Undesirable
Driving Behavior, 75
Design Guidelines for Other Geometric and Traffic Control Features of
Intersections, 76
Recommended Revisions to the AASHTO Green Book, 81
82 CHAPTER FOUR Conclusions and Recommendations
84 REFERENCES

86 APPENDIX A Summary of Questionnaire Responses from State and Local

Highway Agencies
86 APPENDIX B Field Observational Studies at Divided Highway Intersections
86 APPENDIX c Evaluation of Accident Histories at Field Study Sites
86 APPENDIX D Evaluation of Statewide Accident Data for Divided Highway
Intersections
86 APPENDIX E Effect of Median Width on Traffic Operations at Signalized
Intersections
86 APPENDIX F Sight Distance Implications of Offsetting Left-Turn Lanes at.
Divided Highway Intersections
87 APPENDIX G Recommended Revisions to the AASHTO Green Book
ACKNOWLEDGMENTS
The work reported herein was performed under NCHRP Project Panel Fitzpatrick. Other TTI staff members who participated in the research
15-14(2) by Midwest Research institute (MRI), Pennsylvania Transporta- included Mr. Torstein Lineau and Dr. Vergil Stover. Mr. Harwood, Dr.
tion Institute (PTI), and Texas Transportation Institution (TTI). The work Pietrucha, Dr. Fitzpatrick, Mr. Wooldridge, and Mr. Brydia served as co-
was performed in MRI's Center for Technology and Regional Develop- authors of this final report.
ment, directed by Dr. David L. Bodde. The research team received valuable assistance from . the staffs of 10
Mr. Douglas W. Harwood, Principal Traffic Engineer at MRI, was state highway agencies who participated in the study: California, Illinois,
the principal investigator of the research. Other MRI staff members who Iowa, Kansas, Maryland, Missouri, New Jersey, Pennsylvania, Texas, and
participated in the research included Ms. Karin M. Bauer, Dr. William D. West Virginia.
Glauz, and Ms. Erin J. McGrane. The subcontract work at PTI was directed We are also grateful to traffic engineers in 63 state and local highway
by Dr. Martin T. Pietrucha. Other PTI staff members who contributed to agencies who responded to a questionnaire on their design practices and
the work included Mr. Mark D. Wooldridge, Mr. Robert E. Brydia, and traffic operational and safety experience with at-grade intersections on
Dr. John M. Mason. The subcontract work at M was directed by Dr. Kay divided highways.
 MEDIAN INTERSECTION DESIGN

SUMMARY The objectives of this research were to develop and recommend appropriate design
policies, on the basis of operational and safety considerations, for median widths at rural
and suburban divided highway intersections. The research scope addressed the design of
the median width and related features of the median roadway for at-grade intersections
on divided highways with partial or no control of access.
Median widths for divided highways are often selected primarily on the basis of safety
conditions between intersections. The research was undertaken because knowledge of the
effects of median widths on the operation and safety of the intersections along a divided
highway is limited.
A field observational study of traffic operations was conducted at 40 selected divided
highway intersections in 10 states. The selected intersections included rural, unsignalized
intersections; suburban, unsignalized intersections; and suburban, signalized intersections
all having various median widths. This study focused on specific types of undesirable
driving behavior that were observed at divided highway intersections: drivers lining up
their vehicles side by side rather than in single file; drivers stopping their vehicles in the
median at an angle to, rather than perpendicular to, the major road; and drivers stopping
with part of their vehicles encroaching on the through lanes of the major road. The
analysis of the field data ' examined the relationship between the median width and the
observed frequency of undesirable driving behavior at the intersections.
A statistical analysis was also conducted using a data base of accident, geometric, and
traffic volume data for intersections on the California state highway system. This analysis
was to determine the relationship of median width to accident frequency at the intersec-
tions while considering the effects of traffic volume and other geometric features of the
intersection. The statistical analysis used the techniques of Poisson regression and log-
normal regression.
The field - observational studies and the accident analysis provided similar findings on
the operational and safety effects of the median width at intersections. At rural, unsignal-
ized intersections, accidents and undesirable driving behavior decrease as the median
width increases. In contrast, at suburban, signalized and unsignalized intersections, acci-
dents and undesirable driving behavior increase as the median width increases. Thus, at
rural, unsignalized intersections, wider medians are generally preferable to narrower medi-
ans, unless signalization or suburban development are anticipated at a particular intersec-
tion. The median at a suburban intersection generally should not be wider than necessary
to accommodate the appropriate median left-turn treatment needed to serve current and
anticipated traffic volumes.
The operational effects of median width at signalized intersections were further exam-
ined by using existing traffic simulation models. It was found that at a divided highway
intersection ~t_which all approaches are controlled by a single signal, vehicle delay
increases (and thus, the level of service decreases) as the median width increases. Intersec-
tions with median widths of more than 31 m (100 ft) may require separate signals on the
two roadways of the divided highway. That signal configuration, which is equivalent to
the signal configuration of a diamond interchange, often results in substantially greater
vehicle delay than a single signal. Those operational findings reconfirm the finding of
the accident analysis that median widths at signalized intersections should be kept to the
minimum necessary to accommodate the current (or anticipated) left-turn treatments.

2
CHAPTER 1

INTRODUCTION AND RESEARCH APPROACH

Multilpme, divided cross sections are frequently used by high- sections. Median widths in the range of 9 to 15 m (30 to 50 ft)
way agencies for major, nonfreeway facilities in rural and subur- also provide protection for left-turning vehicles but do not pro-
ban areas (1,2). The multilane cross section enables the highway vide enough storage space for larger vehicles crossing the me-
to operate between intersections with a capacity.(under ideal dian. The concern expressed by the Green Book is that a 9- to
conditions) approaching that of a freeway (3). The median sepa- 15-m (30- to 50-ft) median width may tempt drivers of vehicles
rates the vehicles traveling in opposite directions and reduces longer than the median is wide, such as trucks and buses, to
accident rates below the levels found on undivided highways. cross the divided roadways; however, if such longer vehicles
At-grade intersections are a major source of traffic operational are forced to stop in the median, they may extend into the
and safety problems on multilane, divided highways. Turning through lanes and create operational and safety problems. The
volumes are generally larger at intersections than at driveways Green Book states that even with these potential problems, inter-
and midblock median openings, and accidents tend to cluster sections on divided highways with median widths in the 9- to
where the turning volumes are largest. 15-m (30- to 50-ft) range -generally operate well.
The 1994 American Association of State Highway and Trans- Concerns are also expressed in the Green Book about intersec-
portation Officials (AASHTO) A Policy on Geometric Design tions with median widths in the 15- to 24-m (50- to 80-ft) range.
of Highways and Streets (1), also known as the "Green Book," Such intersections provide enough storage area for most large
establishes geometric design policies for most highway features, vehicles (or for several passenger cars). However, some intersec-
including intersections. The AASHTO Green Book identifies tions of this type have apparently developed operational and
the width of medians at intersections on highways with partial safety problems because drivers have become confused about
or no access control as critical to their operation and safety. the correct path to follow on the median roadway. The Green
Figure I illustrates typical, four-leg, at-grade intersections on Book notes that medians wide enough to ensure that the intersec-
a four-lane, divided highway without and with left-turn lanes. tions of the crossroad with each of the divided roadways operate
Figure 2 illustrates a design with a Wide median, where the independently generally have worked quite well, but the Green
junctions of the crossroad with each roadway of the divided Book also suggests that research may prove that wider medians
highway may operate, in effect, as separate intersections. are not desirable for some kinds of facilities with at-grade inter-
Research on the safety effects of median width has focused sections. The Green Book emphasizes the potential for confusion
on the rate of cross-median -accidents between intersections (i.e., created by wider medians on suburban arterials.
accidents in which one or more vehicles traverse the median As explained above, the width of the median has obvious
and enter the opposing lanes), but only limited work has been implications for how a divided highway intersection will operate
done on the influence of the median width on the operation of for specific traffic volume levels and tuniing movement patterns;
at-grade intersections. Thus, in many cases, the median widths however, driver expectancy and other human factors also affect
of multilane nonfreeway facilities may be selected to meet the the safe operation of such intersections. Using a consistent geo-
safety needs of the highway links between intersections without metric design and traffic control scheme for divided highway
intersections that does not change markedly from site to site
focusing specifically on the role of the median width in intersec-
(especially along the same highway) can be effective in devel-'
tion design and operations.
oping safe operations. However, introducing unusual designs at
Chapter VII of the AASHTO Green Book (1) discusses the
a few locations may violate driver expectancy and lead to safety
selection of the median width for rural arterials. Median widths
problems. For example, if a median is so wide that approaching
generally range from 1.2 to 24 in (4 to 80 ft) or more. The Green
drivers cannot see both intersections at the same time, they may
Book states that on highways without at-grade intersections, the
become confused about the correct path through the intersection.
median may be as narrow as 1.2 to 1.8 in (4 to 6 ft) under very
That can lead to drivers running through Stop signs that they
restricted conditions, but that a median width of 20 in (60 ft) or
did not expect or turning the wrong way into a one-way roadway.
more should be provided wherever feasible. However, the Green In addition to the key geometric variable discussed above—
Book goes on to state that special concerns exist in cases where median width—other geometric variables affect the operational
at-grade intersections are provided. and safety performance of a divided highway intersection and
The Green Book suggests thaf, while median widths as narrow should be explored. The most important of these geometric vari-
as 1.2 to 1.8 in (4 to 6 ft) may be required at intersections ables are as follows:
under very restricted conditions, such narrow medians are not
desirable. Medians 3.6 to 9 in (12 to 30 ft) wide are preferred Roadway widths
because they provide protection for left-turning vehicles at inter- Shoulder widths
4

(a) Without Left-Turn Lanes

——————————
----------

(b) With Left-Turn Lanes

Figure 1. Typical divided highway intersections.
 5

Figure 2. Typical intersection on a divided highway with a wide median.

Number of lanes The current guidance on selecting median widths provided by

Left-turn lanes the Green Book, as related to intersection operations, is general.
Stopping sight distance (SSD) Because very narrow and very wide medians may lead to opera-
Intersection sight distance (ISD) tional and safety problems at intersections, highway agencies
Approach curvature need more specific guidance concerning the operating experi-
Approach grade ence, advantages, and disadvantages for the complete range of
median widths, as well as other geometric considerations in the
Understanding the effects of these variables is important for design of divided highway intersections; this report provides
the design of each intersection approach and, most especially, such guidance.
for the median roadway.
ne effect of the traffic control devices used at divided high-
way intersections is also a key consideration. Most rural divided RESEARCH PROBLEM STATEMENT
highway intersections have two-way stop control, with Stop
signs on the crossroad approaches. However, if the median is The NCHRP research problem statement for this research is
wide enough, there may be stop or yield control for the intersec- as follows:
tion with the major road at each end of the section of roadway
within the median. In addition, some rural intersections may Operational design of median widths and geometric configura-
operate with four-way stop control, with two-way or four-way tions at intersections to meet the needs of rural and suburban traffic
stop control supplemented by flashing beacons, or with traffic continues to be a concern. As noted in AASHTO's A Policy on
signals. Furthermore, the use of four-way stop control, flashers, Geometric Design of Highways and Streets, the width of medians
or signals is even more likely in suburban areas than in rural at intersections on highways with partial or no access control is
areas. The type of traffic control has a strong influence on the critical to their operation and safety.
operation of the intersection and, thus, on the operational and Median-width research efforts have, for the most part, addressed
safety effects of median width and other geometric elements. freeway operations only. These efforts have provided a significant
Traffic volumes, traffic characteristics, and vehicle character- amount of information on cross-section design and optimum width.
However, little or no research has been directed toward developing
istics are also important considerations in intersection design.
guidelines for median widths at intersections with partial or no
Key factors include approach volumes, turning volumes, ap- access control.
proach speeds, vehicle mix (particularly percentage of trucks Median width on these facilities may pose operational problems
and distribution of truck types), vehicle turning paths, and vehi- in the vicinity of intersections for (1) left-turning traffic from the
cle length. main roadway and (2) crossing or left-turning traffic from cross-
 roads. These problems may be created or compounded by other The following criteria were used to define suburban divided
factors, such as the expanse of pavement area, inadequate storage highways in this research:
for turning or crossing vehicles, restricted sight distance at the
intersection approaches, and violations of driver expectations for o Divided highway traffic volume of more than 7,000
traffic movements (for example, drivers tend to become confused
veh/day
about intended operational characteristics of the multiple intersec-
Speeds between 56 to 80 km/h (35 to 50 mph)
tions encountered).
Currently, the focus of many transportation agencies is on the
Spacing of at least 1/4 mi between signalized intersections
construction or reconstruction of multilane facilities having partial Direct driveway access from abutting properties
or no access control. Because current design policy does not ade- No curb parking
quately address median widths or intersection treatments on these Location in or near a populated area
types of facilities, guidance is urgently needed.
Divided highways outside of populated areas with volumes
lower than 7,000 veh/day or speeds higher than 50 mph would
RESEARCH OBJECTIVES AND SCOPE generally be defined as rural, although some 88-km/h (55-mph)
sites on developed arterials have been classified as suburban in
The objectives of the research were to develop and recom- this study. Divided highways with speeds below 56 km/h (35
mend median-width parameters and design criteria for intersec- mph), with spacings between signals of less than 1/4 mi or with
tions on rural and suburban, divided highways with partial or curb parking would be defined as urban and were outside the
no control of access. Guidelines will be developed, for new or scope of this research.
reconstructed facilities, that consider the following factors: Most existing design policies and guidelines address median
safety, traffic operations, traffic volumes, type of traffic control, widths on divided highways ranging from 1.2 to 15 in (4 to 80
design speed, traffic characteristics; and design consistency and ft). However, particular emphasis in this research was placed on
driver expectancy. Other issues that may be considered include the traffic operational and safety performance intersections with
intersection configuration and intersection spacing. medians wider than 24 in (80 ft). The maximum divided highway
The recommended guidelines have been developed in a form median width considered in the study was 92 in (300 ft).
suitable for adoption by AASHTO or individual highway The research focuses on divided highways with raised or de-
agencies. pressed (e.g., depressed landscape) medians. Flush medians are
Throughout this report the term "divided highway" is used to also used extensively on arterials, particularly in urban and sub-
refer to a multilane, nonfreeway facility with a median separat- urban areas. However, flush medians (e.g., flush-paved) are gen-
ing the through lanes in opposite directions of travel. Divided erally fewer than 6 in (20 ft) wide and have median-width re-
highways typically have at-grade intersections where median quirements that are much better defined than for raised and
openings are provided to allow vehicles to turn onto and off of depressed medians. Continuous two-way, left-turn lanes
the divided highway. Some divided highways are expressways (TWLTLs) are considered to be flush medians.
on which direct driveway access between at-grade intersections Two key terms used through this report are "median roadway"
is restricted; other divided highways may have no control of and "median area." The median roadway is defined as the paved
access between intersections. area in the center of the divided highway at an intersection
The primary intent of the research has been to provide design defined by the median width and the median opening length, as
criteria and guidelines for selecting the median width at intersec- shown in Figure 3. Much of the research focused on the traffic
tions on new divided highways that either are being constructed operational and safety problems that can occur on the median
on a new alignment or reconstructed on an existing two-lane roadway. The median area is defined as the median roadway
highway. The research has also provided insight into effective plus the major-road left-turn lanes (if any) at the intersection.
methods of alleviating operational and safety problems at ex- The median area is also illustrated in Figure 3.
isting divided highway intersections. The median width is defined in all cases as the total width
Although very few divided highways are being built on new between the edges of the through lanes, including the left shoul-
alignments, many state highway agencies are upgrading the der and the left-tum lanes (if any). The median opening length is
higher-volume portions of their rural two-lane highway system the distance parallel to the through lanes of the divided highway
to four lanes. Such projects involve choices of median width between the noses or ends of the median on either side of the
and other intersection design elements that have been examined intersection.
explicitly in the proposed study.
The research has addressed intersections on divided highways
in both rural and suburban areas. There were two primary goals RESEARCH APPROACH
for the research. The first goal was to develop rational criteria
for defining the appropriate median width and other geometric The general approach to this research was to investigate the
elements for divided highway intersections in rural areas. The effect of the dimensions of the median roadway on traffic acci-
second goal was to develop guidelines for handling the special dents and traffic operational problems at divided highway inter-
problems that may develop at intersections at which rural divided sections. The research included the following six major data
highways enter suburban areas. The additional development and collection and analysis activities:
traffic volumes present in suburban areas may result in safety
problems, installation of traffic signals, pedestrian demands, and * A review of current geometric design policies concerning
many other concerns that are not typical of the rural divided highway intersections and a current practices survey of
environment. state highway agencies
 7

Median Area -
(median roadway plus
major-road left-turn lanes)

Figure 3. Definition of median roadway and median area.

* A review of literature to document current knowledge of using a statewide data base of accident, geometric traffic control,
divided highway intersections and traffic volume data pertaining to California
- A field observational study of traffic operations at 40 se- e A traffic operational analysis of signalized intersections
lected divided highway intersections using existing simulation models
* An analysis.of the traffic accident history of the field obser-
vational study sites
* A traffic accident analysis of divided highway intersections The findings of these analyses were used to develop guidelines
8

for median widths and other design features at divided highway The appendixes present the results of the research in greater
intersections. detail than the main text. The results of a survey of state and
local highway agencies concerning their design practices and
operational experience with divided highway intersections are
presented in Appendix A. The results of the analysis of field
ORGANIZATION OF THIS REPORT observational study data for divided highway intersections are
presented in Appendix B. Appendix C presents an accident anal-
The remainder of this report, following this introduction, is ysis of the field observational study sites. Analyses of a Califor-
organized into three chapters and seven appendixes. Chapter 2 nia statewide accident data base are presented in Appendix D.
summarizes the findings of the study, including the findings of Appendix E presents the results of a traffic operational analysis
the analyses described above. Chapter 3 discusses the interpreta- of the effect of median width at signalized intersections. The
tion, appraisal, and application of the research findings, with sight distance implications of offsetting opposing left-tum lanes
emphasis on their design implications and the resulting design are addressed in Appendix F. Appendix G presents the recom-
guidelines. Chapter 4 summarizes the conclusions of the study mended changes in design policies, for divided highway inter-
and presents recommendations for future research. sections, that resulted from the research.
 CHAPTER 2

FINDINGS

This chapter presents the findings of the research and explains Field Observational Studies
the data collection and analysis that led to these findings. Details
of the data collection and analysis efforts conducted for the Field observational studies were conducted at 40 selected in-
research are in appendixes to this document. The recommended tersections on divided highways. The field sites included 20
revisions to the Green Book are repeated in Appendix G. rural, four-leg, unsignalized intersections; 8 suburban, four-leg,
ne issues addressed in this chapter include data collection unsignalized intersections; 6 suburban, four-leg, signalized inter-
and analysis overview, current design policies and practices of sections; and 6 intersections with special features. The intersec-
highway agencies, safety characteristics of divided highways tions with special features included T-intersections, intersections
and divided highway intersections, traffic safety and operational with tapered and parallel offset left-turn lanes, and intersections
effects of median width, traffic control at divided highway inter- with median acceleration lanes. The intersections were located
sections, truck considerations, left-turn treatments, intersection in 10 states, including states in all major regions.
sight distance, (ISD), human fact6rs/driver perception considera- Field studies were conducted at each of the 40 intersections.
tions, and driver ekpectancy issues. The field studies included video recording of traffic operations
at each intersection for 4 to 6 hr. Night studies of 2 hr each
were also performed at three intersections. The video data were
reduced in the office to determine the turning movement vol-
DATA COLLECTION AND ANALYSIS OVERVIEW
umes at the intersections; the frequency of undesirable driving
behavior (such as side-by-side queuing on the median roadway,
The Research Approach section of Chapter 1 identified the angle stopping on the median roadway, and encroaching by
six major data collection and analysis activities in the research. stopped vehicles on the through lanes of the divided highway);
The following discussion provides an overview of each of these the frequency of traffic conflicts between through and turning
data collection and analysis efforts. The findings of those analy- vehicles; and driver compliance with traffic control 'devices,
ses are presented later in this chapter. particularly Stop and Yield signs on the median roadway. These
data were analyzed to determine whether a relationship existed
between observed traffic operational problems, including unde-
sirable driving behavior, and the median width or median open-
Design Policy and Practice Review
ing length at the intersection.

A review of current geometric design policies and practices

for divided highway intersections was conducted as part of the Accident History of Field Study Intersections
research. This effort included a review of the AASHTO Green
Book (1), other related AASHTO policies, and state and local Traffic accident data were obtained from the participating
design policies concerning divided highway intersections. state highway agencies for each of the 40 field study intersec-
A mail survey was conducted to document the current design tions. These data were in the form of either printouts summariz-
practices of state and local highway agencies. Responses to this ing the characteristics of each accident or hard copy police acci-'
survey were received from 43 of the 50 state highway agencies dent reports. One state highway agency also provided computer-
and 19 of 51 local highway agencies, corresponding to response generated collision diagrams of the accident experience at the
rates of 86 percent and 37 percent for state and local agencies, study sites. The accident data generally covered a period of 3
respectively. The survey questionnaire and the survey results years for each site, although up to 5 years of accident data were
are presented in full in Appendix A and are summarized below. available for a few sites. On the average, 3.4 years of accident
data per site were available. During these periods, there were
642 traffic accidents at the 40 intersections.
These data were analyzed to determine whether a relationship
Literature Review
existed between the median width or median opening length at
the intersections and their accident experience. Collision dia-
A review of literature presenting current knowledge related grams were prepared and the dominant accident pattem(s) were
to divided highway intersections was conducted. The findings identified at each intersection; these accident patterns were re-
of this literature review appear throughout this chapter as spe- viewed to determine whether they were related in any way to
cific topics are discussed. the median width, the median opening length, or the design of
10

the median roadway. Analyses were also conducted to determine and median intersection design. Design policies at the national
whether trucks were overrepresented in accidents at divided level are based on the AASHTO Green Book (1). The presenta-
highway intersections. tion of state and local agency design policies is based on re-
sponses to the survey of highway agencies that is reported in
Appendix B. Most highway agency design policies are based on
Statewide Traffic Accident Analysis the AASHTO Green Book, although many agencies have their
own design manuals and have adapted the AASHTO policies to
A traffic accident analysis was conducted with a statewide their own needs.
data base of accident, geometric, and traffic control data for The following discussion extensively uses material from the
state highways in California. Data were extracted from this data AASHTO Green Book. The AASHTO design policies concern-
base and analyzed for rural and urban/suburban, three-leg and ing medians and related issues are spread throughout the Green
four-leg, unsignalized intersections and for urban/suburban, Book in chapters dealing with elements of design, cross-section
four-leg, signalized intersections on divided highways. These elements, and specific functional classes of highway. In the
analyses were intended to determine whether a statistically sig- following discussion, the various material from the Green Book
nificant relationship existed between median width and accident that deals with medians on divided highways has been combined
experience at these types of intersections. and is presented, together with a description of state and local
highway agency policies, as a comprehensive overview of cur-
rent median design policies.
Traffic Operational Modeling of Signalized
Intersections
Functions and Types of Medians
An analysis was conducted of the effect of median width on
traffic operations at signalized intersections. As the median at a A median is defined on page 367 of the AASHTO Green
signalized intersection becomes wider, more clearance or "lost" Book as ". . the portion of the highway separating the traveled
time is required in the signal cycle for vehicles to clear the way for traffic in opposing directions." The Green Book states
intersection at the end of each green phase for crossroad and that "A median is highly desirable on arterials carrying four or
major-road left-turn -traffic. Thus, for a given traffic volume more lanes."
level and tuniing movement pattern, the signal operations win According to Chapter IV of the Green Book (Cross Section
be less efficient as the median becomes wider. Signalized inter- Elements), the functions of a median include the following:
sections with median widths of 31 in (100 ft) or more may
require separate signals on each roadway of the divided highway. Minimize interference of opposing traffic
This configuration, which is equivalent to the type of signaliza- Provide a recovery area for out-of-control vehicles
tion provided at a diamond interchange, is less efficient than an Provide a stopping area in case of emergencies
intersection with a single signal. Ho' wever, these relationships Provide open green space
have never been formally quantified. Minimize headlight glare from opposing vehicles
Simulation modeling was conducted using the PASSER 11-
Provide width for future lanes
90, PASSER III, and Texas models (4, 5, 6, 7) to quantify the Provide space for speed-change lanes and storage areas for
effect of median width on traffic delay at signalized intersections
left- and U-tum vehicles
and the differences in delay between the single and diamond Restrict left turns except where median openings are
signal configurations. Two traffic volume levels were postulated:
provided
a moderate-volume level that corresponds roughly to Level of
Service B and a high-volume level that corresponds to traffic
There are three major types of medians: raised, depressed, and
operations on the boundary between Levels of Service C and
flush. Flush medians include painted medians and continuous
D. For each volume level, a baseline set of turning movement TWLTLs. The divided highways of interest in this study typi-
patterns (i.e., percentages of the total approach volume making
cally have either raised or depressed medians. However, some
each tuniing movement) and four variations of turning move- flush medians have some of the same intersection design consid-
ment patterns for that baseline were postulated. For each combi- erations as raised or depressed medians and some divided high-
nation of traffic volume level and turriing-movement pattern, an
ways may incorporate limited sections with flush medians.
optimal signal timing was determined and then that tin-Ling was
used in simulation modeling to estimate the total delay that
would be experienced by traffic at several levels of median Median Width
width. The results of this modeling effort represent quantitative
estimates of the effect of median width and signal configuration The following discussion addresses current highway agency
on the efficiency of traffic operations at signalized intersections design policies for median width. Median width considerations
on divided highways. between and at intersections are addressed separately.

CURRENT DESIGN POLICIES AND PRACTICES OF

HIGHWAY AGENCIES Median Width Between Intersections

This section presents the current design policies and practices AASHTO Policy: The median width between intersections is
of highway agencies related to highway medians, median width, defined as the distance between the edges of the through lanes
 11

in opposing directions, including the width of the left shoulders, Whenever possible, medians should be designed with suffi-
if any. cient width to avoid the need for a median barrier. Figure 4
The minimum median width permitted by the Green Book shows the median barrier warrants applicable to high-speed free-
for most highways is 1.2 in (4 ft). Raised or depressed medians ways and divided highways with partial control of access (ex-
fewer than 1.2 in (4 ft) wide are not practical, and a flush divider pressways) that are presented in the AASHTO Roadside Design
fewer than 1.2 in (4 ft) wide would not be considered a median. Guide (4). As shown in the figure, most medians less than 9 in
Although wider medians are desirable, the Green Book makes (30 ft) wide on highways with average daily traffic (ADT) of
clear that there is demonstrated benefit in any separation — more than 20,000 veh/day warrant a median barrier. The figure
raised or flush or depressed—even if the separation is as little shows that median barriers are optional for medians 9 to 15 in
as 1.2 in (4 ft). The only exception in the Green Book to the (30 to 50 ft) wide on higher-volume highways and for some
minimum 1.2-m (4-ft) median width is for multilane urban collec- medians less than 6 in (20 ft) wide on highways with ADT of
tor streets, where median widths as narrow as 0.6 in (2 ft) are less than 20,000 veh/day. Where median barriers are optional,
permitted. they should be used only if a history of cross-median accidents
Most divided highways have median widths in the range of exists. Median barriers are not appropriate for flat medians more
1.2 to 24 in (4 to 80 ft); however, median widths wider than 24 than 50 ft wide. However, if steep slopes or objects in the median
in (80 ft) have been used. AASHTO policies impose no limit cannot be removed, roadside barriers may be warranted for medi-
on maximum median width. ans of any width; the Roadside Design Guide provides proce-
The Green Book notes that medians should be as wide as dures to address the cost-effectiveness of barrier needs in such
feasible but in balance with other components of the cross sec- situations. Median barriers are less desirable on divided high-
tion. As far as the safety and convenience of motor vehicle ways without full control of access than on freeways because
operation between intersections are concerned, the farther the terminating the barrier at intersections may be difficult and be-
pavements are apart, the better. However, economic factors often cause the barrier may restrict ISD.
limit the median width that can be provided. Construction and In summary, the Green Book does not prescribe particular
maintenance costs increase in proportion to increases in median median widths for particular types of highway facilities. Instead,
width, but the additional cost may not be appreciable compared it summarizes the advantages and disadvantages of particular
with -the cost of the highway as a whole and may be justified median widths and permits designers to make choices case by
in view of the benefits derived. case.
Insofar as through traffic is concerned, the Green Book notes
that the desired ease and freedom of operation, in the sense of
physical and psychological separation from opposing traffic, is Highway Agency Policies: The highway agency survey pre-
obtained when medians are about 12 in (40 ft) wide or wider. sented in Appendix A found a range of highway agency views
With such widths, the facility is truly divided. The noise and on the appropriate median width for divided highways. State
air pressure of opposing traffic are not noticeable, and the glare highway agencies were asked to assess the minimum, desirable,
of headlights is greatly reduced. With widths of 18 in (60 ft) or and maximum median widths for rural and urban nonfreeways.
more, the median can be pleasingly landscaped in a parklike These assessments may reflect median width requirements be-
manner. Landscaping to achieve this parklike appearance need tween intersections and the effects on intersection operations.
not compromise the roadside recovery area. Table I summarizes the survey results on median width
The Green Book discussion of rural arterials notes that on requirements.
highways without at-grade intersections, the median may be as The minimum median widths for rural nonfreeways reported
narrow as 1.2 to 1.8 in (4 to 6 ft) under very restricted conditions, by state highway agencies varied greatly, from as little as 0.9
but that a width of 20 in (60 ft) or more should be provided in (3 ft) to as much as 20 in (64 ft). Approximately 42 percent
whenever feasible. One advantage of a wide median on a rural of the responses recommended minimum median widths of
arterial without intersections is that it allows the use of indepen- greater than 9 in (30 ft). Desirable median widths for rural
dent profiles. divided highways ranged from 5 to 26 in (18 to 84 ft). Approxi-
Roadside design is relevant in selecting an appropriate rhedian mately 62 percent of the responses indicated a desirable median
width. Wider medians are generally preferable because they width of more than 15 in (50 ft). Many state highway agencies
allow the use of flatter roadside slopes while providing adequate did not indicate a maximum median width, implying that the
drainage. median should be as wide as possible; those agencies that did
The Green Book also makes clear that there is a tradeoff respond indicated maximum median widths ranging from 8 to
between the median width and border width between the traveled 92 in (25 to 300 ft).
way and adjacent development. If right of way is restricted, a Another recent survey of state highway agencies concerning
wide median may not be justified if provided at the expense of median widths on rural divided highways without full control
narrowed border areas. A reasonable border width is required of access (9) found similar results to the survey conducted in
to serve as a buffer between the private development along the the present research. Table 2 summarizes the range of minimum,
road and the traveled way, particularly where zoning is limited desirable, and maximum median widths found in the survey.
or nonexistent. Space must be provided on the borders for side- Note that the range of desirable and maximum median widths
walks, highway signs, utility lines, parking, drainage channels found in the survey was not as broad as the range found in the
and structures, proper slopes, a roadside clear zone, and any current study.
retained natural growth. Narrowing the border areas may pro- For urban and suburban nonfreeways, the minimum median
duce obstacles and hindrances next to the roadway, similar to widths indicated by state highway agencies ranged from 1.2 to
those the median is intended to avoid. 9 in (4 to 30 ft). Approximately 56 percent of respondents indi-
12

Traveled Way Shoulder Traveled Way J

Median Width

rh ft
70
20--
60

15— 50

40

C :10-
30

20

10

0
0 20 40 60 80
Average Daily Traffic*
(Thousand)

Based on a 5 year Projection

Figure4.- Median barrier warrants for freeways and expressways (4).

TABLE 1. Summary of highway agency survey results on TABLE 2. Summary of results from a recent survey on rural
minimum, maximum, and desirable widths for divided highway divided highway median widths (9)
medians
Rural
Range Median value Range
Median width m (ft) m (ft) m (ft)
Minimum 1-20 (3-64) 8 (27) 0.3-9 (1-30)
Desirable 5-26 (18-84) 16 (54) 3-20 (9-64)
 13

cated a minimum median width of 3 in (10 ft) or less for urban The Green Book provides the following guidance on the
facilities. The desirable median widths indicated by highway choice of median widths on urban arterials:
agencies for urban/suburban conditions ranged from 2.7 to 20 in
(9 to 64 ft). Approximately 39 percent of state highway agencies Medians are a desirable feature of arterial streets and should be
provided where space permits.... Where right-of-way is limited,
indicated a desirable median width of greater than or equal to it is frequently necessary to decide how best to allocate the available
9 in (30 ft). The maximum median widths indicated by state space between border areas, traffic lanes, and medians. On the less
highway agencies ranged from 5 to 31 in (16 to 101 ft). important arterials the decision is often resolved in favor of no
median at all. A median only 1.2 in wide is better than none;
however, each additional foot provides an added increment of safety
and improved operation. At intersections where left-turris are made,
Median Width at Intersections a left-turn lane is always desirable from a capacity and safety stand-
point. The median width to accommodate left-ruming movements
should desirably be at least 3.6 in. Desirably, the median should
As is the case between intersections, the median width at an be at least 5.4 in wide for a 3.6 in median lane and a 1.8 ft medial
intersection is defined to include the entire distance between the separator. At restricted locations, a 3.0 in lane with a 0.6 in medial
edges of the through lanes in the opposing directions of travel. separator may be used (1).
Thus, at an intersection, both the left shoulders and any left-
tain lanes are considered to be part of the median width. This The urban arterials to which the guidelines given above apply
definition is consistent with the definitions of median width include higher-spe6d suburban arterials. Medians less than 7.2
presented in both the AASHTO Green Book and in Chapter 1. in (24 ft) wide on urban arterials are generally raised medians.
The Green Book notes that in suburban areas and elsewhere
where a median width of 7.2 in (24 ft) or more can be provided,
AASHTO Policy: The AASHTO Green Book policies on a flush or depressed landscaped median offers most of the advan-
median width at intersections are presented in the Green Book tages of a raised median with few of the unfavorable attributes.
chapters that deal with specific functional classes of highway: The Green Book goes on to repeat much of the discussion from
local roads and streets, urban collector roads and streets, and the section on rural arterials concerning the advantages and dis-
rural and urban arterials. advantages of median widths in the 9- to 15-m. (30- to 50-ft)
The most extensive discussion of the attributes of different and 15- to 24-m (50- to 80-ft) ranges.
median widths is found in the section on rural arterials on pages The Green Book notes on page 519 that, for urban arterials,
498 and 499 of the 1994 Green Book. The following material "Experience indicates that drivers prefer medians that are either
from that discussion summarizes what is known about the advan- obviously narrow or those that provide an adequate refuge area
tages and disadvantages of different median widths on divided to allow independent roadway crossing operation." This state-
highways: ment is made from the point of view of the driver on a minor-
road approach to a divided intersection who intends to cross or
When intersections are to be provided, special concern must turn left onto the divided highway. Drivers making such maneu-
be given to the width of the median. While medians as narrow as vers would prefer to have a median that is at least wide enough
1.2 to 1.8 in may be required under very restricted conditions, to store their vehicle (i.e., at least 8 in [25 ft] wide for a passenger
medians 3.6 to 9 in wide provide protection for left turning vehicles
car design vehicle). If a median of that width cannot be provided,
at intersections.
Median widths from 9 to 15 in should be carefully considered the median width might as well be as narrow as possible because,
from an operational standpoint at intersections. These widths do from the point of view of the driver crossing or turning onto the
not provide median storage space for larger vehicles crossing the divided highway, wider medians just mean additional travel time
median. Also, these widths may encourage the driver to attempt the and distance (and longer required gaps in traffic) without any
crossings independently leaving a portion of the vehicle unprotected
from through traffic. These widths, even with these problems, nor- benefit from a safe waiting area in the median. On the other
mally operate quite well and apparently are within the realm of hand, from the point of view of the driver turning left off the
normal operational expectations of the driver. Widths in the range divided highway, medians wide enough to include separate left-
of 15 to 24 in have developed accident problems in some areas as turn lanes at intersections are desirable.
the drivers apparently tend to become confused about the intended
operational characteristics of the multiple intersections encountered.
Medians wide enough to assure the driver that the intersection with
the two sets of lanes operate separately have worked quite well. Highway Agency Policies: Approximately 76 percent of state
Research may prove that wider medians are not desirable for some highway agencies indicated that they consider intersection oper-
facilities with at-grade intersections (1). ations in selecting the median width for a divided highway.
Approximately 50 percent of the responding highway agencies
While the preceding discussion identifies the potential advan- indicated that storage needs in the median area have influenced
tages and the possible operational problems of different median either their median width policy or their choice of median width
widths, it is not very specific. The designer is left without much for particular projects.
guidance in the use of median widths in the ranges of 9 to 15 Several highway agencies reported that they have selected a
m. (30 to 50 ft) and 15 to 24 in (50 to 80 ft) because the Green large school bus as the design vehicle for median width on rural
Book implies that medians in these width ranges operate well divided highways. Typically, such agencies use medians with a
in most cases .but may develop operational problems in some minimum width of 15 m (50 ft) to store the largest school bus
cases. An important objective of the present research is to obtain safely, which is 12 m, (39.5 ft) long and carries 84 passengers.
better information about the circumstances under which opera- Other highway agencies stated that they consider the expected
tional problems can occur, the causes of such operational prob- queue lengths of left-turning vehicles in selecting the median
lems, and how they can be avoided or eliminated. width.
14

One highway agency indicated that it had widened the 14-m lanes. Figure 5, which is Green Book Figure IX-74, illustrates
(46-ft) median of one particular rural divided highway to 21 in a typical median left-tum lane. The Green Book encourages
(70 ft) at intersections to facilitate truck movements onto and the use of median left-turn lanes because accident potential,
off of the divided highway. Another highway agency has adopted inconvenience, and considerable loss in efficiency are evident
a policy of widening divided highway medians to 46 in (150 ft) at intersections on divided highways where median left-turn
at major intersections while maintaining its standard 18-m (60- lanes are not provided.
ft) median at minor intersections. Where curbing is to be provided adjacent to the roadway, an
Ten state highway agencies, representing 24 percent of the appropriate curb offset should be provided. The Green Book
agencies that responded to the question, stated that they inten- states on page 348 that "For low speed street conditions, mount-
tionally design narrow medians so that vehicles entering from able curbs may be placed at the edge of a through lane, although
the crossroad will not attempt to stop in the median. If the it is preferable that the curbs be offset 0.3 to 0.6 in .... Barrier
median is not wide enough to store a vehicle, the vehicle must curbs introduced intermittently along streets should be off-
wait for a simultaneous gap in traffic in both directions of travel. set .... at least 0.3 in, and preferably 0.6 m."
However, several of these agencies indicated that they use nar- The length of auxiliary lanes consists of three components:
row medians to enhance the operational efficiency of signalized (1) deceleration length, (2) storage length, and (3) entering taper.
intersections. Ideally, the total length of the auxiliary lane should be the sum
On the other hand, 19 state highway agencies, or 45 percent of these three components. The Green Book notes, however, that
of the agencies that responded to the question, indicated that common practice is to accept a moderate amount of deceleration
they have encountered traffic operational or safety problems at within the through lanes and to consider the taper as part of
intersections with median widths that they consider to be too the deceleration length. The following minimum deceleration
narrow. Most of these problems were related to turning and lengths are recommended for auxiliary lanes on arterial roads
crossing maneuvers by trucks and buses. and streets with grades of 2 percent or less:
Twenty state highway agencies, or 47 percent of the respon-
dents, reported operational problems at intersections related to
Design Speed Deceleration Length
medians that were considered to be too wide. These problems knVh (mph) rn (ft)
included the following:
50(30) 70(235)
60(40) 100(315)
* Side-by-side queuing in the median area, with resulting 80(50) 130(435)
confusion about which vehicle is to proceed first
Conflicting movements on the median roadway These lengths exclude the length of the taper, which should
Inefficient signal timing because of long clearance interval be approximately 2.4 to 5 in (8 to 15 ft) longitudinally to 0.3
requirements at the end of particular signal phases in (I ft) transversely (i.e., 8: 1 to 15: 1). Where these deceleration
9 Lack of sufficient sight distance if drivers do not stop in lengths cannot be provided, some deceleration in the through
the median lanes will be required of left-turning vehicles.
* Increased potential for wrong-way movements at night At unsignalized intersections, the storage length (exclusive of
taper) may be based on the number of left-tum vehicles likely
Only six state highway agencies, or 15 percent of the respon- to arrive in an average 2-min period during the peak hour. As
dents, indicated that they have median width policies that differ- a minimum requirement, space for at least two passenger cars
entiate between the median widths used at signalized and unsig- should be provided; at locations with more than 10 percent truck
nalized intersections. This is a concern because operating traffic, provision should be made to store at least one passenger
experience has shown that intersections with wider medians are car and one truck. The 2-min storage period may need to be
difficult to signalize effectively. modified depending on the actual waiting time required to com-
Nine state highway agencies (23 percent of the respondents) plete a left turn at specific intersections.
indicated that they consider bicycle operations, and 18 agencies At signalized intersections, the required storage length de-
(46 percent of the respondents) indicated that they consider pe- pends on the signal cycle length, the signal phasing arrange-
destrian needs in selecting median widths for divided highways. ments, and the rate of arrival and departure of left-turning vehi-
cles. The storage length should usually be 1.5 to 2 times the
average number of vehicles that would store (i.e., arrive on a
Left-Turn Lanes red signal indication) per cycle. The same minimum storage
lengths apply as were discussed above for unsignalized
The following discussion addresses lanes provided for left intersections.
turns from a divided highway onto a crossroad. The Green Book states on page 783 that "median widths of
6 in (20 ft) or more are desirable at intersections with single
left-turn lanes in the median, but widths of 4.8 to 5.4 in (16
AASHTO Policy to 18 ft) permit reasonably adequate arrangements. Where two
median lanes are used, a median width of at least 8.4 in (28 ft)
AASHTO policy on left-tum treatments for at-grade intersec- is desirable to permit installation of two 3.6-m (12-ft) lanes and
tions is presented on pages 778 through 787 of the 1994 Green a 1.2-m (4-ft) separator. Although not equal in width to a normal
Book (1). The Green Book states that auxiliary lanes, such as traveled lane, a 3-m (10-ft) lane and a 0.6-m (2-ft) curbed separa-
left-turn lanes at intersections or median openings, should be at tor (or with traffic buttons, paint lines, or both), separating the
least 3 in (10 ft) wide and, preferably, as wide as the through median lane from the opposing through lane may be acceptable
 15

mm ow To

CA CCU 9fW

~~R

I _~

1A 5.4~ AND OVER

-A-

LEOM

k CCNTM UM
CONSMCUOUS

'40~ opt~

Figure 5. Key dimensions in median left-turn lane design (1).

where speeds are low and the intersection is controlled with I * Where the left-turn design volume exceeds 100 veh/hr in
traffic signals." the peak period
The Green Book on page 787 alludes to the potential for
offsetting left-turn lanes and angling the left-turn lanes toward
In addition, several state highway agencies use a series of
the opposing lanes.to improve visibility of opposing traffic, to
decrease the possibility of conflict between opposing left-turn nomographs to determine if left-tum lanes are warranted (10).
vehicles, and (at signalized intersections) to increase the satura- These nomographs —typically based on the work of Hannelink
tion flow rate for left-turn vehicles. As discussed below and (11) in the 1960s—use input parameters of speed, the opposing
in Appendix F, some highway agencies have developed more volume, (Vo), the advancing volume (VA), the left-turning vol-
extensive policies on improving safety and capacity by offsetting ume (VL), and the percentage of left turns in the advancing
left-turn lanes and angling them toward the opposing lanes. volume. The nomographs developed by Harmehnk are illustrated
in Figure 6.
Existing highway agency design guidelines for left-tum lanes
Highway Agency Policies are generally similar to those in the AASHTO Green Book. With
respect to left-tum storage length, some highway agencies permit
minimum designs with space to store only one passenger car,
Conventional Left-Turn Lanes: A key issue in the design
of intersections on divided highways is the location of left- in contrast to the Green Book, which requires storage space for
two vehicles. Other highway agencies require sufficient storage
turn lanes.
In their responses to the survey questionnaire summarized in length for two passenger cars or one truck. Still, other highway
agencies require a minimum of 15 m (50 ft) of storage length,
Appendix A and in preliminary discussions with the research
which is approximately equivalent to providing storage space
team, a few state highway agencies indicated that they provide
left-turn lanes at all median openings (intersections or drive- for two passenger cars.
ways) on rural divided highways where a left turn is possible
(i.e., everywhere except, in one direction of travel, at T-intersec-
tions). In some states, this practice is a formal policy; in other
states it is more a consistent procedure. Double Left-Turn Lanes: Double left-turn lanes are typically
Most states use guidelines or warrants to determine whether used only at signalized intersections, although the research team
a left-turn lane should be provided at a particular location. Some is aware of one location where they have been used at an unsig-
examples of warrants for left-turn lanes that are in current use nalized intersection. State highway agencies have reported use
and were identified by states responding to the survey question- of the following guidelines for determining where double left-
naire are presented below. Left-tum lanes are provided at the turn lanes should be provided:
following locations:

* Where the left-turn design volume exceeds 20 percent of * At locations where there is insufficient space to provide
the total directional approach design volume the necessary left-tum storage length in a single turn lane
 16

20 m
F/1111
'MOMWNIN~
vml -. M F I 2'LANE ROAD
I
EMINE
MN LI I

0
> P,M MEM ;

>

0 MINE z
0

LIME
0
Elks mmammmm

100 I .cc 01I

LI

'1111
NOR

VA ADVANCING VOLUME jvPH) VA ADVANCING VOLUME (VPH)

F
2-LANE ROAD

IM
LEFT.T RNTAEAT.CNT
WARRANTED MOW'. NO : WHEN Vo < 400 VPH Washed line). A LEFT-TURN LANE IS NOT
WARRANTED
TE NORMALLY
UNLESS THE ADVANCING VO LUmE (V.) IN THE SAME DIRECTION
LEFT-TURNING TRAFFIC EXCEEDS 400 VPH (V.1 400 ~PHJ
AS THE

SIX
lit E 2000

i sm
4- LANE
UNDIVIDED ROAD

0 MMIMPUL'Imm
z

Ell WEENE 1000

Ar/v/,,.ILELI I 0
2-

01
100
V. L11EM 0 500
0 0

5 UM 20 VPH I 25
VIL LEFT-TURNING
10 15 VOL E I

VA ADVANCING VOLUME JVPHl

Figure6 Nomographs for determining volume warrants for provisio n of left-tum lanes. (11)

0 At locations where the storage length needed for a single Offset Left-Tum Lanes: Wider medians generally have posi-
left-tum lane would be excessive tive effects on traffic operations and safety. However, wider
a At locations where the necessary time for a protected left- medians can create sight obstructions for left-tuming vehicles
turn phase for a single lane becomes unattainable to still meet when confronting opposing left-turn vehicles. This type of sight
level-of-service criteria obstruction is illustrated in the two diagrams in Figure 7. The
* At locations where the left-tum volume exceeds 300 veh/hr
upper portion of the figure illustrates an intersection with a
(or, in the case of a different agency, 330 veh/hr)
relatively narrow median in which opposing left-tum vehicles
block the view of approaching through traffic for a driver in the
Although not yet common, one highway agency has reported left-turn lane. The lower portion of the figure shows an intersec-
that it has begun to provide triple left-tum lanes at, high-volume tion with a wider median; this diagram shows that even when
suburban intersections on divided highways. a left-turning vehicle advances from the left-turn lane onto the
 17

1"m

Figure 7 Examples of sight obstructions caused by opposing left-turn vehicles (9).

median roadway, the opposing left-turn vehicles still block the conventional left-turn lanes with two.altemative configurations:
driver's view of approaching through traffic. parallel offset and tapered offset left-turn lanes. Offset left-turn
. The most common solution to this problem is to offset the lanes are most common at signalized intersections, but highway
left-tum lanes (i.e., to move the left-turn. lanes laterally within agencies have begun to use them to minimize sight obstructions
the median) so that the opposing left-tum vehicles no longer at unsignalized intersections as well.
block the line of sight of the two drivers. Figure 8 compares The survey responses summarized in'Appendix A indicate
 18

Cr5-

: F Left-Turn Lanes
(a)Co,,elional

---------------

----------

Parallel Offset Left-Turn Lanes

69 CM
CM CMCM
- - - - - - - - - CT-D - - - - - - - -
Cro

Tapered Offset Left-Turn Lanes

Figure& Alternative left-turn treatments for rural and suburban divided highways (9).

that 16 state highway agencies; or 37 percent of those re- left-turn lanes generally also require raised medians 7.2 m (24
sponding, have used offset left-turn lanes. Parallel offset left- ft) or more wide. Appendix F addresses the amount of offset
turn lanes with 3.6 m (12-ft) widths can be constructed in raised between the opposing left-turn lanes required to minimize the
medians with widths as narrow as 7.2 m (24 ft), and can be sight distance obstructions caused by opposing vehicles.
provided in narrower medians if restricted lane widths or curb . On the basis of the information obtained to date, it appears
offsets are used or a flush median is provided (9). Tapered offset that the Illinois Department of Transportation (IDOT) has the
 19

most extensive experience with offset left-turn lanes. IDOT has Acceleration lanes are not always desirable at stop-controlled inter-
sections where entering drivers can wait for an opportunity to merge
adopted a design policy of providing left-turn lanes at all median without disrupting through traffic. Acceleration lanes are advanta-
openings (i.e., both intersections and driveways) where left turns geous on roads without stop control and on all high-volume roads
off the highway are possible. Left-turn lanes are provided in even with stop control where openings between vehicles in the
both directions of travel on the divided highway at four-leg peak-hour traffic streams are infrequent and short (1).
intersections and in one direction of travel at T-intersections.
Tapered offset left-turn lanes are generally provided by IDOT Median acceleration lanes are generally constructed with a
where the median widths are 12 m (40 ft) or more, where the parallel design with a taper length of approximately 92 ni (300
current crossroad ADT (average of both approaches) is 1,500 ft). Median acceleration lanes can be used both at T-intersections
veh/day or more, and where the current left-turn design hour and at four-leg intersections. The use of a median acceleration
volume in each direction from the major road is more than 60 lane at a four-leg intersection probably changes the tuniing paths
veh/hr. At signalized intersections, tapered offset left-turn lanes and conflict patterns of opposing vehicles within the median
are used on the major road where only one left-turn lane in each roadway, but the extent of this effect is unknown. Figure 9
direction of travel is needed for capacity. illustrates a typical four-leg intersection with median accelera-
Several potential disadvantages of offset left-turn lanes have tion lanes.
been cited by highway engineers. These include the following: In wider medians, where the opposing left-turn lanes do not
overlap, median acceleration lanes can typically be provided
Lack of driver familiarity with the offset left-turn lane within the width used for the conventional left-turn lane in that
design same direction of travel. Thus, median acceleration lanes can
Potential confusion for older drivers be incorporated in many existing designs without widening the
Roadside safety concerns, if barrier curbs are used median. However, the presence of a median acceleration lane
Increased difficulty of making U-tums for both motorists reduces the amount by which the conventional left-turn lane for
and emergency vehicles the opposing direction of travel on the divided highway can be
- Difficulty of snow removal and deicing activities on the offset. Thus, providing an operational and safety advantage for
separate left-turn roadway left turns onto the divided highway may create an operational
* Potential for wrong-way movements by opposing direction and safety disadvantage for left turns from the divided highway.
vehicles entering the left-turn roadway A 1986 survey by the Institute of Transportation Engineers
(ITE) found that median acceleration lanes had been used by 13
of the 53 highway agencies (25 percent) that responded to the
IDOT has considered these potential disadvantages and dis- survey (12). Respondents to the survey were divided evenly for
counted them on the basis of its operating experience. IDOT and against the use of median acceleration lanes. ITE concluded
has found that problems with driver confusion associated with that median acceleration lanes appear to promote efficient left
offset left-turn lanes are minimal if proper signing and pavement turns onto major roadways and to reduce accidents and traffic
markings are used (i.e., advance signing and pavement arrows conflicts but that insufficient data are available to quantify their
on the entrance to the left-turn lane). No problems have een
traffic operational and safety benefits.
observed with drivers entering the left-turn roadway in the wrong On the basis of the guidelines used by state highway agencies,
direction. Although barrier curbs have been used in older de- acceleration lanes for left-turning vehicles from a crossroad onto
signs, roadside safety concerns can be minimized with the use the divided highway should be considered at locations where
of mountable curbs, and IDOT now uses mountable curbs. - adequate median width is available and the following are true:
though plowing and deicing of the left-turn roadway will require
a separate pass by the plow truck, curbing is often needed for
Limited gaps are available in the major-road traffic stream.
conventional left-turn lanes as well. Finally, on divided high-
Turning traffic must merge with high-speed through traffic.
ways with two or more lanes in each direction of travel, U-turns
There is a significant history of rear-end or sideswipe
by passenger cars (but not always by larger vehicles) can be
accidents.
completed within the width of the opposing lanes (see the figure
4 ISD is inadequate.
presented in the subsequent section on U-turn treatments . -
5. There are high volumes of trucks entering the divided
turns by snow plows or emergency vehicles can be made, if
highway.
necessary, directly from the through lanes onto the median road-
way without using the left-turn roadway.
One highway agency guideline stated that a truck volume of
Offset left-turn lanes offer a potential advantage at signalized
75 to 100 trucks per day would be sufficient to warrant a median
intersections: the signal operation can be made more efficient
acceleration lane.
by reducing the clearance time at the end of the left-turn phase.

Indirect Left-Turn Treatments

Median Acceleration Lanes
The AASHTO Green Book presents design policies dealing
Median acceleration lanes are increasingly used at intersec- with indirect left-turn treatments on pages 768 through 774. The
tions on high-speed divided highways. However, median accel- indirect left-tum treatments presented in the Green Book for left
eration lanes are not appropriate for all divided highway intersec- turns off a divided highway include the use of jughandle road-
tions. The AASHTO Green Book states the following on pages ways before the crossroad, loop roadways beyond the crossroad,
750 and 751: and directional median crossovers beyond the crossroad. Such
20

c9
------------

Figure 9. Typicalfour-leg divided highway intersection with median acceleration lanes.

treatments are typically used at major signalized intersections plus shoulders. In no case should the width of the median open-
on divided highways, often in suburban areas. Figure 10 illus- ing be fewer than 12 in (40 ft). Design of the median opening
trates the indirect left-turn treatments identified above. The fig- length is based on the path of a particular design vehicle turning
ure shows that each of these indirect left-turn treatments for left left at a speed of 15 to 25 km/h (10 to 15 mph). The Green
turns from the divided highway has a corresponding treatment Book provides tables of minimum median opening lengths based
for left turns onto the divided highway. on control radii of 12, 15, and 23 m (40, 50, and 75 ft).
Indirect left-turn treatments are relevant to the selection of Figure 12 and Table 3 illustrate the AASHTO criteria for
median widths because indirect left-turn treatments may permit minimum design of median openings on the basis of a 15-m
a narrower median than would otherwise be possible. In other (50-ft) control radius; this figure and table are analogous to the
words, at some locations, the need to provide left-turn lanes in figures and tables presented in the Green Book for the other
the median may be the controlling factor in the selection of the control radii.
median width, and the use of indirect left-tum treatments can The definition of the control radius was illustrated earlier in
minimize this factor. Indirect left-turn treatments have been used Figure 5. The control radius is selected as follows:
extensively on suburban arterials in Michigan and New Jersey.

U-Turn Treatments
A 12-m (40-ft) control radius will accommodate passenger
cars and an occasional single-unit truck.
A 15-m - (50-ft) control radius will accornmodate single-
Median width requirements to permit U-turns through the
median of a divided highway, at intersections or on directional unit trucks and an occasional VVB-12 vehicle with some
crossover roadways, are presented in Figure 11, based on Green encroachment on the adjacent lanes.
Book Figure IX-69. This figure is based on U-turns on a four- A 23-m (75-ft) control radius will accommodate the WB-
lane divided roadway, although the final row in the figure labeled 12 and WB-15 design vehicles with only minor encroach-
"inner lane to shoulder" can be applied to turns into the outer ment on adjacent lanes at the end of the turn.
lane of a six-lane divided roadway. The design vehicles on which
the figure is based are those presented in Green Book Table 11-1.
The Green Book also presents above-minimum median open-
ing design appropriate for control radii of 30, 50, and 70 in (90,
Median Openings 150, and 230 ft) (see Figure 13) on the basis of Green Book
Figure IX-64. Such designs can provide for design vehicles
Minimum Length of Median Opening larger than the WB-15 truck (which is a relatively small combi-
nation truck in today's fleet) and will permit VVB-15 and smaller
The Green Book states that the minimum length of the median trucks to turn at speeds higher than 15 to 25 km/hr (10 to 15
opening should be the width of the crossroad roadway pavement mph) without encroaching on adjacent lanes.
 21

Indirect Left Turns from a Divided Highway

Loop

Jughandle

Median Crossover

Indirect Left Turns onto a Divided Highway

Jughandle

Median Crossover
Figure,10. Examples of indirect left-tum treatments for divided highway intersections.

Median Nose Treatments wide experience no difference in operations as a result of the

median nose type. With median widths greater than 3 m (10 ft),
There are three distinct types of nose treatments for medians. the bullet nose is superior because it more closely approximates
The two most common types are the semicircular nose and the the path of the inner rear wheel; this requires less intersection
bullet nose. The third type of end treatment, a squared end, pavement and a shorter opening length.
generally is not used. One state highway agency reported that it had evaluated a
According to the Green Book, medians less than 3 m (10 11) squared median nose in comparison with a bullet nose. The
22

results of the study indicated that the bullet nose had several ized. For unsignalized intersections, the spacing requirement is
.
advantages over the squared end, including fewer accidents, such that exclusive left-turn lanes with the proper taper and
more intersection capacity, and the ability to provide for a full storage lengths can be provided.
range of signal phasing (13). In rural areas, one state highway agency recommends a mini-
mum spacing of 0.4 km (0.25 mi) between median openings and
a maximum spacing of 0.8 km (0.5 mi). Another state uses a
Spacing Between Median Openings minimum spacing of 0.8 to 1.6 Ian (0.5 to 1.0 mi) between
median openings on divided highways, unless existing intersec-
Very few state highway agency design policies were found tions require closer spacing. Still another state tries to maintain
to have formal provisions for the minimum spacing between an average 1.6-km (1-mi) spacing between median openings on
median openings. One state highway agency recommended that rural.divided highways with partial control of access on new
median openings in urban/suburban areas be spaced no closer alignments and an average spacing of 0.8 kin (0.5 mi) between
than 488 m (1,600 ft) apart if the intersections are to be signal- median openings along existing roadways.
In rural areas, median openings are normally provided at all
intersections with public roads. Most states do not provide me-
dian openings for all driveways, although openings may be war-
ranted for higher-volume commercial or industrial driveways.
However, at least one state highway agency reported providing
TVF1 OF ~IUWR median openings for all driveways along extended sections of
divided highways.

SAFETY CHARACTERISTICS OF DIVIDED

HIGHWAYS AND DIVIDED HIGHWAY
'0 7,4M,
INTERSECTIONS

Recent Data on the Safety Performance of Divided

Highways and Intersections

Figure 11. AASHTO minimum median widths to Multilane divided highways, in general, experience some of
accommodate U-turns (1). the lowest accident rates of any type. of nonfreeway facility.

EDW OF TUVWD WAY

WN OF WAY&W WAY -,
,,

OL 10AMAY
- - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - - L --- - - =_~
O.S. EDU OF MEMAN
EVA Of MUXAN -I"-

- - - - - - - - - - - - - - - - ----I -- - - - - - - - - - - - - - - -

- - - - - - - - - - - - - - - - --- r ----------------
WE-- K-15.

NOT&
WII-12 WRNING PATH
W545 TURNNG PAT"
W849 TUANING PATH - - -
WS-34 TURNING PATH

Figure 12. Minimum design of median openings (SU design vehicle, control radius of 15 m)(1).
 23

TABLE 3. Minimum design of median openings (SU design vehicle, control radius of 15 m) (1)

Width L = Minimum Length of Median Opening (m)

Median
M Semicircular Bullet Nose

1.2 28.8 28.8

1.8 28.2 22.8
2.4 27.6 20.4
3.0 27.0 18.6
3.6 26.4 17.4
4.2 25.8 15.9
4.8 25.2 15.0
6.0 24.0 13.2
7.2 22.8 12.0 Min.
8.4 21.6 12.0 Min.
9.6 20.4 12.0 Min.
10.8 19.2 12.0 Min.
12.0 18.0 12.0 Min.
15.0 15.0 12.0 Min.
18.0 12.0 Min. 12.0 Min.
21.0 12.0 Min. 12.0 Min.

- - - - - - - - --- - - - - -
- - - - -- -- -- -- --- - - - - -
..TROUGH
TRAFFIC
L
-8

I
CA MEDIAN

` M

- - - - - - --- - - - - -
THROUGH
R
-
nA;FIC

;2

ASSUMED R-15-
R2 -M/,

Figure 13. Above-minimum design of median openings (typical bullet nose

ends) (1).

For example, recent California data show that, for nonfreeway ways. In urban and suburban areas in California, the accident
facilities in rural areas, including both intersection and noninter- rates of multilane divided highways were lower than for multi-
section accidents, multilane divided highways have lower acci- lane undivided highways but higher than for two-lane highways
dent rates than two-lane highways and multilane undivided high- (14). Similar data show that accident rates on rural multilane
24

TABLE 4. Summary of accident rates by roadway type reported in the literature (14, 15)
Accident rate (per million veh-km) Accident rate (per million veh-mi)
Highway and area type California' Minnesota
b
Utah b
Californiaa Minnesota b UtaT
RURAL HIGHWAYS
Two-lane undivided highways 0.82 1.11 1.31 1.32 1.79 2.10
Multilane undivided highways 1.32 4.44 1.41 2.13 7.14 2.27
Multilane divided highways 0.71 1.47 1.38 1.15 2.37 2.22
URBAN HIGHWAYS
Two-lane undivided highways 1.35 2.17
Multilane undivided highways 2.23 3.59
Multilane divided highways 1.60 2.58
a
Based on 1991 data; includes only roadways with uncontrolled access.
b
Based on 1987 data from the FHWA Highway Safety Information System.

TABLE 5. Accident data for rural multilane divided highways in Minnesota and Utah by severity level and relationship to
intersection-1987 (15)
% of Accidents
Annual Number of Accidents by Relationship to Intersection
Relationship- to Inters ction F+I PDO TOTAL F+I PDO TOTAL

- Intersection
accidents 266 363 629 69.6 32.3 41.8
At-intersection 220 284 504 57.6 25.3 33.5
Intersection -related 46 79 125 12.0 7.0 8.3
Non -intersection accidents 116 760 876 30.4 67.7 58.2
TOTAL 382 1123 1505 100.0 100.0 100.0

divided highways in Minnesota are much lower than the accident accidents that occur outside the curbline limits of the
rates of rural multilane undivided highways, while in Utah the intersection.
accident rates of rural multilane divided and undivided highways The accident severity data for Minnesota in Table 5 show that
are about the same (15). This difference probably results from approximately 42 percent of intersection accidents involved a
higher levels of development in Minnesota than in Utah. These fatality or injury. By contrast, only 13 percent of nonintersection
data are summarized in Table 4 and include highways with accidents involved a fatality or injury. The finding that intersec-
partial access control and with no access control. tion accidents tend to be more severe than nonintersection acci-
Table 5 presents a breakdown of the Minnesota and Utah dents is also confirmed in the Utah data. This fmding is another
accident data for rural multilane divided highways by relation- indication of the importance of intersections in safety manage-
ship to intersection and accident severity (15). The Minnesota ment of divided highways.
data show that approximately 34 percent of the accidents on Table 6 presents the distribution of accident types for roadway
rural divided highways occur at intersections, while another 8 section accidents in Minnesota. The table classifies accidents by
percent of accidents are intersection related, even though they the number of vehicles involved, collision type, relationship to
occur outside the curbline limits of the intersection. Forty-two intersection, and location of first harmful event (on roadway/off
percent of divided highway accidents in Minnesota are related roadway). The table shows that 97 percent of at-intersection
to intersections, which illustrates that intersections obviously accidents and 78 percent of other intersection-related accidents
constitute an important issue in safety management of divided involve multiple-vehicle collisions on the roadway. In contrast,
highways. multiple-vehicle collisions constitute only 29 percent of nonin-
The Utah data in Table 5 show that accidents at intersections tersection accidents. Thus, the single-vehicle run-off-road acci-
constitute 34 percent of all accidents on rural divided highways, dent problem, which predominates safety management between
a statistic that is comparable with the Minnesota experience. intersections in rural areas, is not a major issue in intersection
Unfortunately, the Utah data do not identify intersection-related accidents. The goal of safety management at divided highway
 25

intersections should clearly be to minimize multiple-vehicle col- accident experience over periods of 6 months to 5 years at
lisions. This finding is generally confirmed by the Utah accident 150 at-grade intersections on 290 km (180 mi) of rural divided
type distribution data in Table 7, although the percentage of expressways on the state highway system in California. Most of
intersection-related accidents that involve multiple-vehicle colli- the intersections were two-way stop-controlled intersections,
sions is lower in Utah than in Minnesota. with stop control on the minor road and no control on the divided
highway, although a few signalized intersections were included.
The analysis of these California data found that no direct
Research Findings Concerning Safety Performance
of Divided Highway Intersections relationship existed between intersection accidents and the sum
of the entering traffic volumes; thus, the study did not support
Two published research studies-one in California and one the use of accidents per million entering vehicles as a safety
in Ohio -have addressed the safety performance of divided measure of effectiveness. On the other hand, the study also found
highway intersections. Both of these studies were performed a that it was inappropriate to use an exposure index based on
number of years ago, one in the 1950s and the other in the 1960s. the product of the intersecting volumes because the accident
A 1953 California study by McDonald (16) developed rela- experience at divided highway intersections was found to be
tionships between the number of accidents and traffic volume more -sensitive to the minor-road ADT than to the divided high-
at divided highway intersections. This study was based on the way ADT. The following relationship was found to best describe

TABLE 6. Distribution of accident types for rural multflane divided nonfreeways in Minnesota-1987 (15)
Roacdtw y s*ction accidents At-Intereaclion accidents I Other inteesection-Iitsd accidents TO AL
On-to dway Run- road ICombined On-co dway Run-offroad Combined Ion-r dway Run-offfoad ICombin*d ACC16ENTS
No. % No. I % No. % -No.- %_ No. % No. % No. % No. % No. % No. %
.......... M ~I .. ~00 mml~w MM a4f,A. I "'M
-E......0. '.
30 5.5 138 4
41.8 108 10.2 4 0~8 5 20.0 _.'Q IA 0 156.0 16 :. 8.0 1-187 ' 12.4
24 4.4 122 36.7 146 16.7 2 0A 5 20.0 7 1.4 1 0.9 9 50.0 10 8.0 103 10.8
Other noncollision 6 1.1 16 30
8
4.8
46 22 2.5 2 0A 0 0.0 2 0.4 0 0.0 0 0.0 0 0.0 24 1.6
Collision accidents 261 164 58.4 455 51.9 2.7 ~:~26 0.0 33 - ~8 ~%47.6 M.,'6 ~56.6 ~117 13.6 505 :33A
Coll. W/ Pedestrian 1 0.2 1 0.3 2 0.2 1 0.2 0 0.0 1 0.2 2 1.9 0 0.0 2 1.0 5 0.3
Coll. w/ bicycle 1 0.2 0 0.0 1 0.1 2 0.4 0 0.0 2 0.4 0 0.0 0 0.0 0 0.0 3 0.2
Coll. w/ animal 243 44.7 2 0.6 245 28.0 3 0.6 0 0.0 3 0.6 3 2.8 0 0.0 3 2.4 251 16.7
Coll. .1 parked -hicle 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 01 0.0 0 0.0 0 0.0
Coll. w/ fixed object 13 2.4 184 5
55.4
05 0197 22.5 7 1.5 20 80.0 27 5.4 3 2.8 81 44.4 11 8.8 235 15.6
Coll. w/ other object 3 0.6 4 1.2 7 0.~ 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 7 0.5
.
Other c.11i,on 0 0.0 3 0.0 3 0.3 0 0.0 0 0.0 0 0.0 0 0.0 1 5.6 1 0.8 4 0,3
M9ON MW Rom Raw SWA ~ S. ~WMWW
=02MOM ' MR Mi M MW milli MR __ .... ... ... '
= 75RUME
Head-on 11 2.0 0 1.3 0.01
2 111 0.4 0 0.0 2 0.4 1 0.9 0 0.0 1 0.8 14 0.9
Sid62WiP6 (SaM6 dir) 52 9.6 52
0 5.9 0.0
is 3.1 0 0.0 15 3.0 15 14.0 0 0.0 15 12.0 82 5.4
Side-ilt. (opp d,) 0 _0 11 1.3 10 0.2 0 0.0 1 0.2 3 2.8 0 0.0 3 2.4 15 1.0
Angle or turning 1 221 4.0 -0 0
0.0
' ~~22
'0 2.5 334~ IIJ 0 0.0 4 66.3 24 22.4 0 0.0 24 19.2 380 J
22
Rew-end 0 0.0 102 11.6 60 12.5 0 0.0 60 11.9 39 30.4 0 0.0 39 31.2 201 '3.4
.0
Other MV collision 0 0.0 W
55 6.3 50 10.4 0 0.0 so 9.9 16 15.0 0 0.0 16 ~&.8 121 8.0
2.1
5. - 0'N.N

TABLE 7. Distribution of accident types for rural multilane divided nonfreeways in Utah -1987 (15)
Roadway section accidents Intersection accidents TOTAL
On-roadway Run-off-road Combined On-roadway Run-off-road Combined ACCIDENTS
---
No. F % No. % No. % No.- % No. % No. % No. %

NOhcoIIIslofi a6cIdenis .4 5~ 72;6 20.0 S,. '_M -'A

Overturned 3 1.3 0 0.0 3 1.0 3 1.9 0 0.0 3 1.9 6 1.3
Other noncollislon 5 2.2 53 72.6 so 19.0 2 1.3 1 20.0 3 1.9 61 13.1
o 2 0 6 a6.0
Coll. w/ pedestrian 3 1.3 0 0.0 3 1.0 3 1.9 0 0.0 3 1.9 6 1.3
Coll. w/ bicycle 2 0.9 0 0.0 2 0.7 3 1.9 0 0.0 3 1.9 5 1.1
Coll. w/ animal 81* 34.9 0 0.0 at 26.6 1 0.6 0 0.0 1 0.6 82 17.7
Coll. w/ parked vehicle a 3.4 0 0.0 8 2.6 1 0.6 0 0.0 1 0.6 9 1.9
Coll. w/ fb(ed object 0 0.0 .20 27.4 20. 6.6 0 0.0 4 80.0 4 2.5 24 5.2
Coll. w/ other object 3 1.3 01 0.0 3 1.0 ~i 0. 0 0.0 0 0.0 3 0.6
Other collision 1 0.4 0 0.0 1 0.3 11 0.6 0 0.0 1 0.6 2 0.4

Head-on 4 0 1.7
0.0 4 1.3 2 1.3 0 0.0 2 1.3 6 1.3
Sideswipe (same d1r) 16 0 6.9
0.0 16 5.2 3 1.9 0 0.0 3 1.91 19 4.1
Sideswipe (opp dir) 6 0 2.6
0.0 6 2.0 0 0.0 0 0.0 0 0.0 6 11.3
Angle or turning 53 22.8 0 0.0 53 17.4 iiol1 711A 0 0.0 110 69.2 163 35.1
Rear-end
Other MV collision
7- 44 1 9.0
311~ 1.30
0 0.0
0.0
44 -7 14.4
3 1.0
24
1
15.6
0*6
01
01
0.01
0.0
24
1 15.1
0.6
68
4
14.7
0.9
M
R won
26

the relationship between the number of accidents and the traffic concentrating cross-traffic at a few locations, by closing low-
volume at divided highway intersections: volume-crossroad intersections and providing frontage roads,
VO.633 may effectively reduce the number of intersection accidents.
N = 0.000783 1~,0.455
d C A study very similar to the McDonald study was conducted
where in Ohio by Priest (17) in 1964. The Priest study included 3
years of accident data for 316 at-grade intersections on divided
N = expected number of intersection accidents per year highways with partial or no control of access. Most, if not all,
Vd = ADT volume entering the intersection from the divided of the intersections studied were evidently unsignalized intersec-
highway (veh/day) tions; however, the author does not explicitly state the type of
Vc = ADT volume entering the intersection from the crossroad traffic control used at the intersections studied. Priest concluded
(veh/day) that accident frequency at divided highway intersections in-
creases with the product of the intersecting traffic volumes.
It should be noted that, in this relationship, the number of
However, like McDonald, Priest also found that accident fre-
accidents at a divided highway intersection is proportional to
quency is more sensitive to the crossroad traffic volume than to
the product of the intersecting volumes, where each of the vol-
the divided highway traffic volume.
umes is reduced by being raised to an exponent of less than one.
Figure 15 shows the relationship between traffic volume and
The exponent for the crossroad volume is larger than the expo-
nent for the divided highway yolume, indicating the greater accident frequency developed by Priest, which appears very sim-
sensitivity of accident experience to crossroad volume. ilar *to the relationship for California developed by MQDonald,
Figure 14 illustrates the relationship between traffic volume although it covers a higher range of crossroad volumes and a
and accident frequency represented by equation 1. The figure lower range of Aivided highway volumes. The findings of the
can be entered with the ADTs of the divided highway and cross- Priest study concen-iing the safety effects of different median
road to determine the expected number of intersection accidents widths are presented later in the median width discussion.
per year. Table 8 compares the annual accident frequencies for divided
On the basis of the relationships in equation I and Figure 14, highway intersections predicted by the McDonald and Priest
McDonald concluded'that low-crossroad-volume intersections studies. Although the ADT ranges of the two studies have only
have higher accident rates per crossroad vehicle than do high- limited overlap, the accident frequency predictions within the
crossroad-volume intersections. This finding is evidence that ADT range common to both studies are mutually supportive.

2400

2200

2000
W

1400
0

E- 1400
e

1400

c 1200
&5
0

loco

goo

0 Goo

400

200

6000 BODO 10000 12000 14000 16000 10,000 20000 ZZ000

Average Daily Traffic Entering frarn Divided Highway

Figure 14. Average number of accidents per year related to traffic volume at divided highway
intersections— California (16).
 27

C13

co
2

P
0
0

.0 1,000 2,000 - 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 11,000 12,000

ADT of Divided Highway (veh/day)

Figure 15. Average number of accidents per year related to traffic volume at divided highway intersections —Ohio (17).

TRAFFIC SAFETY AND OPERATIONAL EFFECTS dent analysis of field observational study sites. AU statements
OF MEDIAN WIDTH
concerning statistical significance in this discussion are at the
95 percent confidence level, unless otherwise stated.
This section presents the findings concerning the analysis of
the traffic safety and operational effects of median width and Current Knowledge
median opening length at divided highway intersections. The
Effects of Median Width Between Intersections
discussion includes findings drawn from current knowledge in
the literature,.the accident analysis of divided highway intersec- Most past research on divided highway median width has
tions in California, the field, observational studies, and the acci- focused on the effects of median width between intersections
28

TABLE 8. Comparison of predicted annual accident frequency for divided highway intersections as a function of intersection traffic
volumes for California and Ohio (16, 17)

Minor Road EXPECTED ANNUAL NUMBER OF ACCIDENTS PER INTERSECTION

ADT Divided Highway ADT (veh/day)
(veh/dav) 30001 40001 50001 75001 100001 150001 20000

1001 --1 --1 --1 0.8 1.01 1.11 1.3

5001 2.3 f- 2.61 3.21 3.6
100 3.61 4.11 4.91 5.6
200 5.61 6AI 7.61 8.7
30001 --1
40001 --1 --1
500
600

1001 0.11 0.11 0.11 0.31 0.4.1

5001 0.51 0.61 0.71 2.21 2.81
i 10001
20001
0.91
4.31
2.01
4.51
2.81
5.01
3.81
5.21
4.81
7.41
--1
--1
30001 5.41 5.81 6.31 8.21 --1 _71-
40001 6.21 6.71 7.41
50001 7.01 7.61
60001 7.61

rather than at intersections. For example, it is known that the Effects of Median Width at Intersections
presence of a median on a divided highway provides a definite
improvement in traffic operations compared with an undivided Less information is published about the effects of median
highway. The multilane highway analysis procedure in Chapter widths at intersections than between intersections. The 1964
7 of the Highway Capacity Manual (3) estimates that the free- Ohio study by Priest (17), described in the previous section,
flow speed on a divided highway is generally 2.6 km/h (1.6 established relationships between the number of intersection ac-
mph) higher than on a comparable undivided highway. However, cidents per year and the exposure index for different ranges of
the median width itself has no documented effect on traffic median width. The exposure index used by Priest is the product
operations, unless a very narrow median necessitates the location of the ADTs of the intersecting roadways. These relationships
of median barriers or other lateral obstructions within 1.8 in (6 are illustrated in Figure 16. For any given value of the exposure
ft) of the traveled way. index, the differences in accident rate between the three median-
Providing medians between intersections is desirable to re- width groups shown in the figure represent differences in both
duce the likelihood that vehicles whose drivers are inattentive accident frequency and in accident rate, because they compare
or lose control will run into the opposing lanes of traffic. The accident experience for equivalent entering traffic volumes. The
safety effects of median width between intersections have been relationships in the figure show that, except at very low volume
addressed in a number of studies (18,19,20,jl,22 23). These levels, intersection accident rates decrease with increasing me-
studies generally show a decrease in accident rate between inter- dian width. However, the difference in accident rate between
sections with increasing median width of up to approximately medians less than 6 m (20 ft) wide and medians 6 to 12 m (20
12 m (40 ft). For median widths between intersections in the to 39 ft) wide is greater than the difference in accident rates
range from 12 to 24 in (40 to 80 ft), there are mixed results; between medians with widths of 6 to 12 in (20 to 39 ft) and
some studies show that accident rate continues to decrease with medians with widths of 12 in (40 ft) or more.
increasing median width, while others show no relationship. A 1977 Purdue University study by Van Maren (24) devel-
Almost no research has addressed the safety effects of medians oped relationships between geometric and traffic volume vari-
wider than 24 in (80 ft). ables and accident experience at divided highway intersections.
These findings suggest that minimum median widths of 12 in Van Maren found no statistically significant relationship be-
(40 ft) be provided between intersections when possible. Medi- tween median width and intersection accident rate. The author
ans wider than 12 m (40 ft) between intersections may be desir- speculated that this finding may have resulted because of the
able, but safety benefits of wider medians have not been conclu- limited range of median widths (9 to 18 in or 30 to 60 ft) that
sively demonstrated. were evaluated. However, this range includes most of the rural
 29

Median Width * Rural, three-leg, unsignalized intersections with stop con-

0 to 5.8 rn 6.1 to 11.9 rn 12 rn (40 ft) trol on the crossroad approach
(o to 19 ft) (20 to 39 ft) and over
Urban/suburban, four-leg, unsignalized intersections with
stop
* control on the crossroad approach
~i * Urban/suburban, three-leg, unsignalized intersections with
(D
stop control on the crossroad approach
- Urban/suburban, four-leg, signalized intersections

(D In general, this research focused on rural and suburban inter-

sections. The California data base did not provide an explicit
method to distinguish between urban and suburban intersections
in accordance with the definitions used in the rest of this re-
E search. Therefore, all urban and suburban intersections in the
Z California data base were analyzed together. However, most
divided highway intersections on state highways in California
are suburban, rather than urban.
Table 9 summarizes the accident data for the five intersection
types listed above. The five intersection types included a total
4 6
' 8 10 12 14 16 1'8 20 of 2,140 divided intersections with median widths of 4 in (14
2 ~2
ft) or greater. The accident analysis was restricted to multiple-
Product of Entering Traffic Volumes/106(veh2/day2) vehicle accidents within the curbline limits of each intersection.
This restriction was made because it appeared unlikely that me-
Figure,16 Variation of annual accident frequency at divided dian width would have any direct causal relationship to either
highway intersections as a function of median width and multiple-vehicle accidents that occurred outside the curbline lim-
exposure index (the product of the ADTs of the intersecting its of the intersection or to single-vehicle accidents. In a 3-year
roadways)(17). analysis period (1990 — 1992), the 2,140 intersections experi-
enced a total of 8,748 multiple-vehicle accidents. The rural,
divided nonfreeways that have been built by highway agencies three-leg, unsignalized intersections had the fewest number of
since the 1950s, including current practices. Van Maren found accidents (approximately 0.6 multiple-vehicle accidents per n-
that larger Stop signs, advance warning signs, and marked stop tersection per year), while the urban/suburban, four-leg, signal-
ized intersections had the most accidents (approximately 2.6
lines on the crossroad approaches decreased accidents. Accident
rate was found to increase as the length of the travel path across multiple-vehicle accidents per intersection per year).
the intersection, for a crossing or turning vehicle, increased. On The statistical techniques used to evaluate the effect of median
the basis of this finding, Van Maren recommended that the width on accident experience included one-way analysis of vari-
distance across the intersection be minin-Lized by reducing the ance, Poisson regression, and log-linear regression. One-way
distance between the stop line on the crossroad approach and analysis of variance was used to take a first look at whether the
the edge of the major-road traveled way, whenever possible; the apparent trends in accident rate with changing median width
author was careful to say that the distance across the intersection shown in Table 9 were statistically significant.
should not be shortened by decreasing the median width. Analysis of variance is a statistical technique to evaluate rela-
tionships between a continuous dependent variable (in this case,
Research by Radwan et al. (25,26,27) at Purdue University
in 1979 used field studies and simulation modeling to investigate accident rate per million entering vehicles) and a categorical or
the operational effects of median width and traffic control at discrete independent variable (in this case, median width divided
both signalized and unsignalized intersections on divided high- into the five levels shown in Table 9).
A major limitation of the one-way analysis of variance was
ways. On the basis of simulation modeling with an early version
of NETSIM applied at unsignalized intersections, Radwan et al. that only one independent variable — major-road median
generally found no statistically significant effect of median width width—was evaluated. Therefore, an alternative analysis method
on traffic delays and traffic conflicts over the range of median that could simultaneously consider the effects of several indepen-
widths from 9 to 18 in (30 to 60 ft). dent variables, including the major-road median width and other
key geometric elements of the intersection design, was consid-
ered. The consideration of multiple independent variables in the
Accident Analyses Using Statewide Data for same analysis is intended to ensure that an effect of one of these
Divided Highways In California other variables is not mistaken for an effect of median width or
does not obscure an effect of median width. Table 10 lists the
A traffic accident analysis of divided highway intersections variables from the existing Califorriia data base that were consid-
was conducted with a statewide data base of accident, geometric, ered in these analyses.
traffic control, and traffic volume data for state highways in In past research, predictive models for accident rate have
California. Data were extracted from this data base for five typically been developed using a variation of the general linear
specific types of divided highway intersections: model: multiple regression analysis in which all independent
variables are continuous and quantitative; analysis of variance
* Rural, four-leg, unsignalized intersections with stop control where all variables are categorical; and analysis of covariance
on the crossroad approach where both continuous and categorical variables are present.
30
TABLE 9. Accident rate for divided highway intersections in California for selected ranges of median width

Accident rate (per

Number of Number of Exposure (million million entering
Median width intersections' accidentSb entering vehicles) vehicles)
RURAL FOUR-LEG UNSIGNALIZED INTERSECTIONS
14 to 28 ft 66 206 845.6 0.24
30 to 50 ft 43 173 1261.9 0.14
52 to 80 ft 31 76 517.2 0.15
82 to 100 ft 6 2 39.6 0.05
over.100 ft 7 6 87.2 0.07
153 463 2751.5

RURAL THREE-LEG UNSIGNALIZED INTERSECTIONS

14 to 28 ft 99 154 1698.8 0.09
30, to 50 ft 35 102 1080.1 0.09
52 to 80 ft 9 12 189.2 0.06
8~ to 100 ft 4 1 26.9 0' * 04
over 100 ft 10 2 109.1 0.02
157 271 4808.1

URBAN/SUBURBAN FOUR-LEG UNSIGNALIZED INTERSECTIONS

14 to 28 ft 246 962 8009.6 0.12
30 to 50 ft 31 214 1022.7 0.21
52 to 80 ft 15 153 507.6 0.30
82 to 100 ft 0 0 0.0
over 100 ft 1 5 22.9 0.22
293 1344 9562.8

URBAN/SUBURBAN THREE-LEG UNSIGNALIZED INTERSECTION

14 to 28 ft 829 1541 31069.9 0.05
30 to 50 ft 57 128 1854.9 0.07
52 to 80 ft 19 33 674.4 0.05
82 to 99 ft 1 1 41.1 0.02
over 100 ft 1 0 75.7 0.00
907 1703 33716.0

URBAN/SUBURBAN FOUR-LEG SIGNALIZED INTERSECTIONS

14 to 28 ft 551 4166 27184.4 0.15
30 to 50 ft 46 355 1856.6 0.19
52 to 80 ft 29 411 1281.4 0.31
82 to 99 ft 0 0 0.0
over 100 ft 35 199.6 0.18
630 4967 30522.0

a Includes all suitable intersections on the California state highway system.

b
Includes reported accidents of all severity levels for a 3-year period (1990-1992); includes
only multiple-vehicle accidents that occurred within the curb line limits of the intersection.
c Total number of million vehicles entering the intersections in a 3-year period based on
the best available estimates of major-road and crossroad ADT.
 31

TABLE 10. Geometric, traffic control, and traffic volume variables available for divided highway intersections in the Caffornia data
base

Major-road average daily traffic volume (veh/day)

Crossroad average daily traffic volume (veh/day)
Major-road median width (ft)
Functional class of major road
Design speed of major road (mi/h)
Access control on major road (partial/none)
Number of lanes on major road
Number of lanes on crossroad
Average lane width on major road (ft)
Outside shoulder width on major road (ft)
Presence of left-turn lanes on major road (yes/no)
Presence of left-turn lanes on crossroad (yes/no)
Presence of right-turn channelization on major road (yes/no)
Presence of right-turn channelization on crossroad (yes/no)
Terrain (flat/rolling/mountainous)
Intersection lighting (yes/no)
Signal phasing (2-phase vs. multiphase) (for signalized intersections only)
Presence of signal mast arms on major road (yes/no) (for signalized intersections
only)
Presence of signal mast arms on crossroad (yes/no) (for signalized intersections
only)

These statistical methods were initiafly used because they were cluded that two methods were the most appropriate for analyzing
familiar to researchers and were justified on the basis that, for the intersection accident data in this study. Poisson regression
large samples at least, the various statistics should be approxi- was useful for analyzing the accident frequencies for all the
mately normal. The general linear model makes the assumption rural and urban/suburban unsignalized intersections, which in-
that the dependent variable—in this case, accident frequen- clude many intersections with no or very few accidents in a 3-
cies—follows a normal distribution. Researchers have now year period. The Poisson regression approach was unsuitable
found that these techniques based on the general linear model and for analyzing the accident frequencies at the urban/suburban
the normal assumption are not well suited to accident analysis, signalized intersections, which did not include many intersec-
because accidents are discrete and rare events whose distribution tions with very few accidents; a log-normal regression approach
is better described by a discrete probability distribution. was found to be more appropriate for analyzing the data for this
Several candidate alternative analysis methods were reviewed intersection type.
for application to the accident frequencies at the divided highway The results of these analyses, (see Tables D-5 through D-9 in
intersections in this study. For each of the five types of intersec- Appendix D) are presented as predictive models for multiple-
tions, the distribution of the 3-year total multiple-vehicle acci- vehicle accident frequency as a function of selected geometric
dent frequencies was examined. As explained in Appendix D, and traffic control variables, including median width. The find-
the accident frequency distributions for those intersections were ings of these analyses concerning median width are as follows:
found to be quite skewed. Most types of intersections experi-
enced either no accidents or one accident over a 3-year period.
A few intersections experienced 2 to 10 accidents yearly but At rural, four-leg, unsignalized intersections, accident fre-
rarely did an intersection experience more than 10 accidents quency decreases as median width increases. This result is
yearly. This observation is not true for urban/suburban, four- statistically significant.
leg, signalized intersections, where about 50 percent of the inter- At rural, three-leg, unsignalized intersections, no statisti-
sections experienced seven or more total multiple-vehicle acci- cally significant relationship exists between accident fre-
dents over the 3-year period, and only about 11 percent had quency and median width.
either no accidents or, one accident. At urban/suburban, four-leg, unsignalized intersections, ac-
Upon review of these accident data distributions, it was con- cident frequency increases with increasing median width
32

over the range of median widths from 4 to 24 ni (14 to 80 accidents would be expected in the range of 2 percent to 6.3
ft). This result is statistically significant. percent. For a umit increase of 1 ft in median width, the expected
At urban/suburban, three-leg, unsignalized intersections, percentage decrease in accident frequency would be 1.27 percent
there is a statistically significant relationship between me- and the 95 percent confidence interval would be from 0.61 to
dian width and accident frequency. The intersection accident 1.92 percent. It should be noted that, while these percentage
frequency increases with increasing median width. effects apply directly to accident frequencies, they also apply to
At urban/suburban, four-leg, signalized intersections, acci- accident rates per million entering vehicles, because the compu-
dent frequency increases as median width increases over tation assumes that all other features of the intersection- includ-
the range of median widths from 4 to 24 m (14 to 80 ft). ing the major-road and crossroad ADTs-remain unchanged.
This result is statistically significant. A positive sign for a coefficient in Table 11 indicates that
accident frequency at an intersection increases as the median
It should be noted that the results stated above were found width increases. Conversely, a negative sign indicates that inter-
after accounting for the effects of as many of the other factors section accidents decrease as the median width increases.
listed in Table 10 as possible. Figures 17 through 19 show these results in the form of graphs
Similar analyses using fatal and injury-producing multiple- of the relationship between predicted total multiple-vehicle acci-
vehicle accidents (rather than total multiple-vehicle accidents) dents yearly and median width for "typical" intersections of each
as the dependent variable provided similar results, except that type. Each figure shows the intersection traffic volumes and the
no statistically significant effect was found for fatal and injury geometric and traffic control features that were assumed for that
accidents at urban/suburban, three-leg, unsignalized intersection type in preparing the figure. The major-road and
intersections. drossroad ADTs chosen for each intersection type are approxi-
Table 11 summarizes the effect of median width on multiple- mately equal to the median ADTs for all intersections of that
vehicle intersection accident frequency. The table includes the type. The geometric and traffic control features selected for the
effect of median width on accidents, expressed as a percentage, "typical" intersections are characteristic of most intersections of
as well as upper and lower confidence limits for that effect. The that type. The models presented previously predict the total num-
effects in the table are the percentage change in multiple-vehicle ber of multiple-vehicle accidents in a 3-year period. For conve-
intersection accident frequency that will result from a unit ni'ence, Figures 17 through 19 show the results expressed as
change in median width. Unit changes in median width of both predicted multiple-vehicle accidents per year.
1 m and 1 ft are presented in the table. For example, the table The plot for rural, four-leg, unsignalized intersections in Fig-
shows that,at a rural, four-leg, unsignalized intersection, a I-m ure 17 makes clear that, regardless of the median width, this
increase in median width would be expected to decrease multi- type of intersection experiences relatively few accidents: typi-
ple-vehicle intersection accidents by 4.16 percent. Confidence cally less than one multiple-vehicle accident per year. Neverthe-
limits for this effect are from 2 to 6.3 percent; this indicates that less, median width can have a substantial effect on the accidents
95 percent of the time, the actual effect of median width on that do occur. Intersections with wide medians can experience

TABLE 11. Summary of median width effects from California accident analysis

Median width effect expressed as a percentage of accident frequency

Per unit change of 1 m in median width Per unit change of 1 ft in median width
Lower Upper Lower Upper
Estimate confidence confidence Estimate confidence confidence
Intersection type N limit limit N limit limit

TOTAL MULTI PLE-VEHICLE ACCIDENTS

Rural 4-leg Unsignalized -4.16 -6.30 -2.00 -1.22 -1.92 -9.61

Rural 3-leg unsignalized

Urban/suburban 4-leg unsignalized 5.67 4.23 7.08 1.73 1.29 2.16
Urban/suburbari 3-leg unsignalized 2.69 1.11 4.26 0.82 0.34 1.30
Urban/suburban 4-leg signalized 3.02 1.11 4.97 0.92 0.34 1.50
FATAL AND INJURY MULTIPLE-VEHICLE ACCIDENTS
Rural 4-leg unsignalized -4.43 -7.15 -1.67 -1.35 -2.18 -0.51
Rural 3-leg unsignalized
Urban/suburban 4-leg unsignalized 5.25 3.21 7.31 1.60 0.98 2.23
Urban/suburban 3-leg unsignalized
Urban/suburban 4-leg signalized 2.92 1.11 4.72 0.89 0.34 1.44
 33

CZ
a) Assumptions:
0.9 Major road ADT 12,500 veh/day
CL Crossroad ADT 450 veh/day
(n
0.8 Functional class of major road = Principal arterial
Major road: 4 3.6 -m (12 ft ) lanes
0
Left turn lanes present on major road
0 0.7
CZ
a)
0.6-
a)
0.5-

E 0.4-
0
L_ .
a) 0.3-
M
E
0.2-
a)
0.1 -

10 20 30 40 50 60 70 80 90 100
Median width (ft)
I I I I I I I _J
0 5 10 15 20 25 30 32
Median width (m)
Figure17. Predicted number of multiple-vehicle accidents per year as a function of median width for atypical rural four-leg
unsignalized intersection.

less than half as many accidents as intersections with narrow leg intersections and the sensitivity of accidents to median width
medians. is greater for four-leg intersections than for three-leg intersec-
At rural, three-leg, unsignalized intersections, median width tions. The effect of median width on multiple-vehicle accidents
has no statistically significant effect on accidents. It may be at a typical urban/suburban, three-leg, unsignalized intersection,
reasonable to assume that, because the design characteristics of as illustrated in the figure, is so small that it may not need to
rural, three-leg, and four-leg intersections are so similar, accident be a major factor in design. However, it is apparent that for a
frequencies would decrease as median width increased at three- four-leg intersection, a wider median can substantially increase
leg intersections, as well; however, the accident frequencies and multiple-vehicle accidents— this suggests that wider medians
rates for rural, three-leg intersections are so low and the effect should be avoided at urban/suburban, four-leg, unsignalized in-
of median width, if any, is so small that this hypothesis cannot tersections whenever possible.
be proven. Three-leg intersections have fewer conflict points Figure 19 shows the relationship between median width and
than four-leg intersections and also tend to have lower crossroad multiple-vehicle accidents for a typical urban/suburban, four-
traffic volumes than four-leg intersections. leg, signalized intersection. As in the case of urban/suburban,
Figure 18 shows the predicted accident experience as a func- four-leg, unsignalized intersections, multiple-vehicle accidents
tion of median width for typical urban/suburban, unsignalized can increase substantially as median width increases, and it ap-
intersections. For both three-leg and four-leg intersections, mul- pears that wider medians should be avoided on divided highways
tiple-vehicle accident frequency increases as median width in- with at-grade, signalized intersections in the urban/suburban
creases. The figure shows, however, that four-leg intersections setting.
typically experience more multiple-vehicle accidents than three- There is a marked contrast between the effects of median
34

D1
L _
C13
W
Assumptions:
Major road ADT 27,000 veh/day
5 Crossroad ADT 1,000 veh/day
W Major road: 4 3.6 m (112 ft) lanes with 2.4 m (8 ft) shoulders
C Main road design speed: 80 km/h (50 mph)
Left turn lanes present on major road
4 Level terrain
C13
4)

Four-leg intersections
M
75
E
%I-.
0
2
E
Three-leg intersections
C

a.

a 10 20 30 40 50 60 70 80 90 100
Median width (ft)
I I I I- I I I I
0 5 10 15 20 25 30 32
Median width (m)
Figure.18. Predicted number of multiple-vehicle accidents per year as a function of median width for atypical urban/suburban
unsignalized intersection.

width on accidents at divided highway intersections in I

- Six suburban, four-leg, signalized intersections with me-
urban/subtirban and rural areas. In rural areas, multiple-vehicle aian widths ranging from 5 to 63 m (16 to 207 ft)
accidents decrease as median width increases, while the opposite * Six intersections with special features, including T-intersec-
is true in urban/suburban areas. tions, intersections with tapered and parallel offset left-tum
lanes, and intersections with median acceleration lanes

Field Observational Studies

These 40 study intersections were located in 10 states: Califor-
Field observational studies were conducted to observe driver nia, Illinois, Iowa, Kansas, Maryland, Missouri, New Jersey,
behavior and any traffic operational or safety problems occurring Pennsylvania, Texas, and'West Virginia. Thus, the sample in-
at a selected sample of 40 intersections on divided highways. cluded sites in several major regions of the United States and
The sample included the following: had good geographic diversity. Table 12 summarizes the charac-
teristics of the 40 study intersections.
* Twenty rural, four-leg, unsignalized intersections with me- The 20 rural, four-leg, unsignalized intersections, in particu-
dian widths ranging from 9 to 44 m (30 to 144 ft) lar, included various traffic control practices on the median road-
, Eight suburban, four-leg, unsignahzed intersections with way (i.e., at the intersection of the median roadway with the far
median widths ranging from 5 to 18 in (18 to 60 ft) roadway of the divided highway). Of these 20 intersections, 7
 35

CO
W 3.5

CL
In
C: 3
(D

2.5

2
75
E
0 1-.5

2 0.5
'@I

n
0 1.0 20 30 40 50 60 70 80 90 100
Median width (ft)
I I I I I I I _J
0 5 10 15 20 25 30 32
Median width (m) -
Figure 19. Predicted number of multiple-vehicle accidents per year as a function of median width for a typical urban/suburban
signalized intersection.

had stop control, 7 had yield control, and 6 had no control on intersections likely to carry commuter traffic, the data collection
the median roadway. period included either the morning or evening peak hour.
Figure 20 presents photographs of several typical rural, unsig-
nalized intersections with various median Widths. Similar photo-
graphs for suburban, unsignalized and suburban, signalized inter- Data Collection Procedures
sections are presented in Figures 21 and 22, respectively.
The field data were collected by videotaping traffic operations
with two or more video cameras. Th e cameras were positioned
Data Collection Periods to obtain different views of traffic operations on the median
roadway. Figure 23 shows a typical setup for- a data collection
Field data were collected at each intersection for periods of site. At some intersections, the video cameras were set up in
up to 6 hr during daylight hours on weekdays. A total of approxi- adjacent quadrants on the same side of the divided highway as
mately 200 hr of field data were collected at the 40 sites, or an shown in the figure. At other intersections, camera locations on
average of 5 hr per site. At suburban intersections and at rural opposite sides of the divided highway were used. The camera
36

TABLE 12. Summary of field data collection sites

W102
NJ101 itftiongimd STOP YIELD 1 12 (36) 1 19 (611 1 R 1 2 1 1 1 1 1 1 1 0 1 0 1 8D (SM 1 — 90 38WDI 3W JT~~ pnd 1,* 46 ~h
U..".— I — __ 1 5 (w) j, ug I . 0 W (M) I i2ODO 220D
SUBURBAN SIGNAUZED INMRSECTtONS (G Ud—Ibm)
KS102 Signalmd Signatwd 11 (35) 20 (96) D 2 1 1 1 0 1 Be (56) 90 9M
_V0_103 Subu
S'M _ 5 (16) 32 (104) R 2 1 1 a
1 i.
1_ 12
ii (11)
PA101 S= n 1SE~=_ 1 6 (20) 41 (134) R 2 1 1 (5s)
TX102 Subuft- I q— M ( TT) 18 (59) D 2_ H 1 0 0 W (M) 1 7~ ~1 I
TX107 I SignMd sig~wd 63 rM7) V 0 - 0 V1 1 0 0 72 (45) 1 90 1 36W01 10M I

Abb, M- MWw Tpe: Dkecrm d Ske, Ang

MAL MWi- -Wdio. In. D Dp—W L Lft N~
POLTL P-1,1,ftelleft4u, L~ R Ralsed R RjUM10—d
TOLTL Tap,,W HW left-tum ta,e

locations depended primarily on local terrain; elevated camera recorded in the field and (2) classification of maneuvers ob-
positions were selected, whenever possible, to obtain the best served on the videotapes.
view of the intersection. The data reduction effort revealed that 348,788 vehicles en-
Observers at the intersections kept a log of maneuvers of tered and passed through the study intersections during the study
interest during the study period. This log was used during the period. Trucks constituted 8.8 percent of these vehicles. A key
data reduction effort. At some sites, the logs were particularly volume variable in the data analysis is the median volume, which
valuable in determining whether a vehicle in the median roadway is the total volume passing through the median roadway and is
encroached on the through lanes of the divided highway, because the sum of the volumes for eight individual movements:
that was sometimes difficult to judge from the video image
alone. Two major-road left-turn movements
Two major-road U-turn movements
Two crossroad through movements
Night Studies Two crossroad left-turn movements
Studies were also performed at three sites to observe whether The total median volume observed during the study period
drivers operated at night in accordance with the designer's inten-
was 46,460 vehicles (or 13.3 percent of the total entering vol-
tions. The three sites included one rural, unsignalized intersec- ume). In contrast, through vehicles on the major road constituted
tion with a particularly wide median (41 m or 134 ft); one rural, 78.7 percent of the traffic entering the intersections. More de-
unsignalized intersection with median acceleration lanes; and tailed summaries of the traffic volume counts at the study inter-
one suburban, signalized intersection with tapered offset left- sections are presented in Appendix B.
turn lanes. During each night study, an observer made a count of The videotapes were reviewed to characterize driver behavior
turning movement volumes through the median roadway while at the intersections and to evaluate any operational problems
noting any erratic maneuvers or other evidence of driver confu- that occurred. Every vehicle that passed through the median
sion. No videotaping was performed at night because of low area, defined as the median roadway plus the major-road left-
ambient-light levels. turn lanes (see Figure 3), was observed on the videotape. Data
on driver behavior were recorded for the vehicles involved in
Data Reduction each maneuver during which one or more of the following events
occurred:
Data reduction of the field observational data consisted of
two stages: (1) turning-movement counts from the videotapes 1. Two or more vehicles were present in the median area at
 37

-77 "N' i

IN.

9-m (30-ft) median width

12-m (40-ft) median width

13-m (44-ft) median width

Figure 20. Photographs of typical rural unsignali7ed intersections on divided highways.

I
38

16-m (52-ft) median width

_~W C I

41-m (134-ft) median width

Figure 20. Photographs of typical rural unsignalized intersections oil divided higlma.vs (con till lied).

the same time (whether moving or stopped) for unsignalized e Stopping by vehicles in the median roadway in such a
intersections or two or more vehicles were stopped on the position that they encroached on the through lanes of the ma-
median roadway at the same time for signalized jor road
intersections.
For each observed maneuver, it was noted whether any traffic
One or more stopped vehicles encroached on the through
conflict (i.e., brakin or swerving by one or more vehicles)
lanes of the divided highway.
resulted that could g be associated with some risk of a traffic
There was evident hesitation or confusion by one or more
accident. For left-turn maneuvers by opposing vehicles, it was
drivers.
also noted whether the vehicles turned left in front of or behind
Any other unusual or noteworthy traffic event was observed.
one another.
Using a procedure, described in Appendix B. the maneuver
Any undesirable driving behavior in the median roadway was
data were analyzed as interactions between pairs of vehicles
noted for each observed maneuver. Examples of undesirable
present on the roadway simultaneously. Interactions between
driving behavior that were observed include the following:
pairs of vehicles in the median area were classified on the basis
of the following four factors:
Side-by-side queuing of vehicles in the median roadway
Stopping by vehicles in the median roadway at an angle e Approach by which the first vehicle of the pair entered the
other than perpendicular to the major road intersection:
 39

5-m (18-ft) median width

16-m (54-ft) median width

Figure 21. Photographs of typical suburban unsignaliZed intersections oil
divided highways.

—Major road (J) e Approach by which the second vehicle of the pair entered
—Minor road or crossroad (N) the intersection:
Maneuver made by the first vehicle: —From the same approach as the first vehicle (S)
—Through movement (T) —From the approach opposite to the first vehicle (0)
—Left turn (L) —From the approach to the right of the first vehicle (R)
—U-turn (U) —Frorn the approach to the left of the first vehicle (L)
40

6-m (20-ft) median width

P
R"

iii!q
'

~1=k 7 --&,A W§W41

. .
'Aij

11 -m (35-ft) median width

Figure 22. Photographs of lypical suburban signalized intersections on divided highways

Maneuver made by the second vehicle: fication of interactions between vehicles extended only to vehi-
-Through movement (T) cles that passed through the median area.
—Left turn (L) If each of the four"approaches to a four-leg intersection is
—U-turn (U)
analyzed separately, a total of 64 combinations of these factors
are found; however, 32 of these combinations are mirror images
Right-turn maneuvers were not considered because the classi- of the other 32, in which the first of the interacting vehicles come
 41

Figure 23. Typical data collection setup using video cameras.

from opposite approaches. If these mirror-image maneuvers are Data Analyses

combined, there are a total of 32 unique combinations of maneu-
vers by two consecutive vehicles passing through the median. An analysis of the field observational data was conducted to
Figure 24 identifies these combinations of maneuvers. Each ma- determine what traffic operational problems were observed and
neuver combination is identified by a four-character code, repre- whether these problems were related to the median width or the
senting the four descriptors of the maneuver combination listed geometric design of the median roadway. 'Me issues that were
above. For example, a maneuver combination in which the first examined included tum-in-front/tum-behind behavior in oppos-
vehicle approached the intersection on the major road (J) and ing left turns, types of undesirable driving behavior observed at
turned left (L), while the second vehicle approached from the divided highway intersections, and the effects of median width
crossroad approach to the left of the first vehicle (L) and made and median opening length on the frequency of undesirable
a. through movement on the crossroad (T), would be classified driving behavior. The results of these analyses are presented
as maneuver code JLLT. below. The field observational data were also used to evaluate
Each of these 32 possible combinations of maneuvers was driver compliance with traffic control devices; the results of this
observed at least once in the field observational studies. Table analysis are presented in the subsequent section on traffic control
13 summarizes the frequencies of each of these maneuver combi- at divided highway intersections.
nations for all 40 sites as a whole. The table shows that 6,680
maneuver combinations were observed. It also shows that certain Turn-in-FrontlTurn-Behind Behavior
maneuver combinations were quite common (e.g., maneuver
codes JLOL, JLSL, NLLL, NILRL, and NLSL), while other ma- One issue of interest to the research was how drivers making
neuver combinations (typically those involving U-tum maneu- opposing left turns are influenced by the median width. Specifi-
vets) were quite rare. Tables in Appendix B break down these cally, it was hypothesized that, at intersections with narrow me-
maneuver data by site and study period. dians, drivers making opposing left turns tend to turn in front
In addition to the two-vehicle interactions in the median road- of one another and, at intersections with wide medians, drivers
way described above, operational problems related to traffic making opposing left turns turn behind one another. These two
conflicts between vehicles leaving the median roadway and types of turning behavior are illustrated in Figure 25. However,
through vehicles on the major road were also noted, as were no quantitative information exists on the median width at which
single-vehicle erratic maneuvers in which driver confusion or drivers cease to turn in front of one another and begin to turn
misunderstanding of the intersection geometrics was evident. behind one another.
42

_J:: H:: L H:: L L LLi

JLLL—Major-road left turn/crossroad

7
JLLT—"or-road left turn/crossroad
F~ F7
JLOL—Major-road left turn/major-road
I:F]7 F
through movement from left approach. through movement from opposite approach. JLOU—Major-road left turn/major-road
left turn from left approach. U-turn from opposite approach.

H:: L
~~ ~: _J: H:: L _1:: H:: L L1:: L
77 7F
11"F-1:: -:11,T1 F
77 1-1:: F
7
JLRL—Major-road left turn/crossroad JLRT—Major-road left turn/crossroad JLSL—Major-road left turn/major-road JLSU—Major-road left turn/major-road
from right approach. through movement from right approach. left turn from same approach. U-turn from same approach.

1':I HL 1:: H:: L L:: L

_1 F-I I F FF--]]:: F
JULL—Major-road U- turn/crossroad left turn JULT—Major-road U-turn/crossroad through
F Ff-
JUOL—Major-road U-turri/major-road JUOU—Major-road U-tum/major-road
from left approach. movement from left approach. left turn from opposite approach. U-tum from opposite approach.

_k: H::-
rr'W",
L 1:: H:: L _J: L:: L L1:: L
- 11:: F I:: F 7 F__
i11,1-4
JURT—Iftor-road U-lum/crossroad
7 F:,11: 7FI::] F- F
JURL—Major-road U-turnIcrossroad JUSL—Iftor-road U-tum/major-road JUSU—Major-road U-turn/major road
left turn from right approach. through movement from right approach. left turn from same approach. U-tum from same approach.

Figure 24. Two-vehicle maneuver combinations on the median roadway of divided highway intersections.

An analysis for rural, unsignalized intersections (provided in median widths of more than 15 m (50 ft), vehicles making
Appendix B) found that at intersections with median widths of opposing
' left turns tend to turn behind one another. Only about
less than 15 m (50 ft), vehicles making opposing left turns tend 10 percent of opposing left-turn vehicles at intersections with
to turn in front of one another. In contrast, at intersections with median widths of less than 15 m (50 ft) turned behind one
 43

-1; H::'
FT F 7::,F-]:: F F11 F 7 FT: F
NLLL- Crossroad left turn/major- NLLU-U_tum
Crossroad left turn/major- NLOL- Crossroad left turn/crossroad NLOT- Crossroad left
road left turn from left approach. road from left approach. left turn from opposite approach. turn/crossroad through movement
from opposite approach.

I
_J L L _J: LJ ::'IL _J: H': L
F1
-1':' F F! F __T FT: F I! FT: F
NLRL- Crossroad left turn/major-road NLRU- Crossroad left turn/major-road NLSL- Crossroad left turn/crossroad NLST- Crossroad left
left turn from right approach. U-turn from right approach. left turn from same approach. turn/crossroad through
movement from same approach.

H:: L L
L Ll:: L H:: L
7 FT F FT FT: F __T FT: F
NTLL- Crossroad through NTLU- Crossroad through NTOL- Crossroad through NTOT- Crossroad through
movement/major-road left turn movement/major-road U-turn from movement/crossroad left turn movement/crossroad through
from left approach. left approach. from opposite approach. movement from opposite approach.

H:: L _11 H:: L —J

:U :11

Ll:: L
F7 1:: F 11:: F! 1:: F 7 F7: F I::. FT F
NTRL- Crossroad through, NTRU- Crossroad through NTSL- Crossroad through NTST- Crossroad through
movement/major-road left turn movementtmajor-road U-tum from movemenVcrossroad left turn movement/crossroad through
from right approach. right approach. from same approach. movement from same approach.

Figure 24. Two-vehicle maneuver combinations on the median roadway of divided highway intersections (continued).
44
TABLE 13. Summary of frequencies of maneuver combinations for all fixed sites combined

Maneuver code Number (percent) of maneuvers observed

JLLL 217 (3.2)
JLLT 108 (1.6)
JLOL 978 (14.6)
JLOU 104 (1.6)
JLRL 187 (2.8)
ART 108 (1.6)
JLSL 1,720 (25.7)
ASU 98 (1.5)
JULL 29 (0.4)
JULT 1 (0.0)
JLJOL 108 (1.6)
JuOu 22 (0.3)
JURL 49 (0.7)
JURT 8 (0.1)
JUSL 99 (1.5)
JUSU 31 (0.5)
NLLL 414 (6.2)
NLLU 77 (1.2
NLOL 125 (1.9)
NLOT 96 (1.4)
NLRL 628 (9.4)
NLRU 34 (0.5)
NLSL 559 (8.4)
NLST 145 (2.2)
NTLL 94 (1.4)
NTLU 6 (0.1)
NTOL 55 (0.8)
NTOT 70 (1.0)
NTRL 251 (3.8)
NTRU 7 (0.1)
NTSL 148 (2.2)
NTST 104 (1.6)
6,680

another, and no opposing left-turn vehicles were observed to widths at which opposing left turns were observed, opposing
turn in front of one another at intersections with median widths left-turn drivers also displayed turn-in-front, rather than turn-
greater than 15 in (50.ft). This analysis was based primarily on behind. behavior. However, both of the intersections with wider
left turns by opposing major-road vehicles (maneuver type medians where tum-in-front behavior was observed had small
JLOL); howeve r, opposing left turns by crossroad vehicles (ma- channelizing islands between the left-turn'lanes and the major
neuver type NTLOL), while fewer in number, showed similar road; this design guided drivers into the median roadway at an
behavior. angle and encouraged opposing left-turn drivers to turn in front
A similar pattern was found for suburban, unsignalized inter- of each another. Thus, there were no suburban, unsignalized
sections with median widths of less than 15 in (50 ft): nearly intersections without channelizing islands to verify whether the
all the opposing left turns involved vehicles turning in front of same turn-behind behavior observed at the rural, unsignalized
one another. Unlike the rural, signalized intersections, at the two intersections with median widths of more than 15 in (50 ft)
suburban, uhsignalized intersections with 18-m (60-ft) median occurred at similar suburban, unsignalized intersections.
 45

Figure25. Alternative paths for left-turning vehicles: turning in front of one another and turning behind one another.

In summary, the data for the rural, unsignalized intersections approximately 15 in (50 ft) changes in that drivers making op-
demonstrate that, for median widths of up to 15 in (50 ft), posing left turns at narrower medians tend to turn in front of
opposing left-turn drivers tend to turn in front of one another, each another, and drivers at wider medians tend to turn behind
and at intersections with median ' widths of more than 15 in (50 each another. That finding may make 15 m (50 ft) an appropriate
ft), opposing left-turn drivers tend to turn behind one another. breakpoint in setting design polices for selecting median widths.
There is no implication that turn-in-front behavior is either more The findings concerning tum-in-front/tum-behind behavior at
or less desirable than turn-behind behavior; indeed, each type rural intersections could be only partially verified at suburban
of turning behavior generally appears appropriate for the range intersections, but it is reasonable to assume that the turning
of median widths under which it is observed. However, the behavior at intersections with wider medians is the same in
field data do suggest that driver behavior at a median width of suburban areas as in rural areas. It is possible that turn-behind
46

behavior could result in undesirable operations at some higher gap in traffic and complete his or her maneuver. In this situation,
volume urban or suburban intersections because it could lead to there is a concern that, as the median opening becomes longer,
gridlock; however, that was not observed in any of the field more drivers may be tempted to queue side by side.
studies conducted. Another undesirable driving situation occurs when a vehicle
stops on the median roadway at some angle other than perpendic-
ular to the through lanes of the divided highway. In some cases,
Observed Types of Undesirable Driving Behavior where the median is very narrow or a driver decides to cut a
on the Median Roadway comer, the driver of a single vehicle may stop at an angle to the
major road. Alternatively, when the median roadway is already
As explained above, vehicles turning in front of each another occupied by one or more vehicles in the same direction of travel,
or behind each another do not necessarily show undesirable a driver of another vehicle entering the median may find it
driving behavior. However, several types of driving behavior necessary to stop at an angle to avoid encroaching on the through
were observed at the field study intersections that clearly were lanes of the major road or to avoid blocking another vehicle. In
undesirable. Those types of undesirable driving behavior include either case, stopping at an unusual angle is undesirable because
the following: the vehicle may be hit by another vehicle from any of several
directions and because other drivers may be confused about the
9 Side-by-side queuing on the median roadway by vehicles intended path of that vehicle. In contrast, when a vehicle on the
in the same travel direction median roadway stops perpendicular to the through lanes of the
Stopping at an angle on the median roadway major road, with its turn signal activated (where appropriate),
Encroaching on a through lane of the divided highway other drivers are not likely to be confused about the intended
path of that vehicle.
These three types of undesirable behavior are illustrated by Finally, a vehicle on the median roadway may be stopped
the drawings in Figure 26 and by the photographs in Figures with the front or rear of the vehicle encroaching on a through
27, 28, and 29, respectively. lane of the divided highway. This maneuver occurs occasionally
When one vehicle is waiting on the median roadway for an when a vehicle on the median roadway misjudges the edge of
opportunity to cross or enter the far roadway of a divided high- the through travel lanes or stops after failing to see an oncoming
way and a second vehicle arrives in the same travel direction, through vehicle; in this case, the front of the driver's vehicle
the most desirable traffic -operations result when the second encroaches on the through lanes of the far roadway of the divided
vehicle stops behind the first. However, in some cases, the sec- highway. A more common situation occurs when a driver enters
ond driver may stop his or her vehicle beside rather than behind the median roadway when it is already occupied by one or more
the first vehicle. This side-by-side queuing is a concern because other vehicles. At times, the driver of the vehicle entering the
it can lead to driver confusion about which of the two vehicles median roadway may stop with the rear of his or her vehicle
is to proceed first and, thus, can lead to potential conflicts. Side- encroaching on a through lane of the near roadway of the divided
by-side queuing tends to occur in the following situations: highway. Some drivers entering the median roadway may choose
Where the median width is less than the length of two to encroach on a through lane, rather than queue in a side-
vehicles, it is most desirable that a second vehicle not enter the by-side position or stop at an angle; this maneuver can occur
median roadway if a first vehicle traveling in the same direction particularly if the driver perceives (correctly or incorrectly) that
is already there. It would be most desirable for the driver of the he or she will have to encroach on a through lane for only a
second vehicle to wait before crossing the near lanes of the short interval or that no potentially conflicting through vehicles
major road or wait in the major-road left-turn lane (if one is will be present on the major road. Finally, encroachment on a
provided) until sufficient space on the median roadway is avail- ,through lane by either the front or rear of a vehicle may occur
able. However, if the second vehicle enters such a narrow me- if the median width is less than the length of a vehicle and the
dian, the driver may find a choice between pulling alongside )driver enters the median when there is no available gap to cross
the first vehicle, stopping in the median at an unusual angle, or or enter the far lanes of the divided highway.
encroaching on the through lanes of the major road. In this Each of the types of undesirable driving behavior described
situation, the driver of the second vehicle may choose to stop above — side-by-side queuing, angle stopping, and encroaching
beside the first vehicle as the best of several undesirable choices. on a through lane of the divided highway—could be prevented
Where the median opening is sufficiently long, even if the if drivers did not enter the median area unless they were sure
median is wide enough to store two or more vehicles one behind their vehicles had enough room and if, once they entered the
the another, the driver of a second vehicle may be tempted to median, they stopped behind the other vehicle(s) traveling in the
pull beside the first vehicle to avoid delay. For example, if the same direction. However, human nature being what it is, some
first vehicle is waiting in the median to turn left onto the major undesirable driving behavior of this type is bound to occur.
road, a second vehicle intending to cross the major road may Table 14 summarizes the frequency of the three observed
pull beside the first vehicle. As long as the crossing vehicle types of undesirable driving behavior for specific intersection
stops to the right of the tuniing vehicle, this maneuver may be types and ranges of median width. More detailed tables in Ap-
executed safely, but there is always the potential for the drivers pendix B present a breakdown of these data by study period
to become confused about which vehicle is making which ma- for individual sites. Table 14 includes instances of undesirable
neuver and about which vehicle is going to proceed first. In driving behavior for all.maneuvers in which two (or more) vehi-
some cases, the side-by-side queuing may involve two drivers cles were in the median roadway at the same time. It should be
making the same maneuver; the second driver considers the first noted that the frequencies of the individual types of undesirable
driver too timid and doesn't wait for the first driver to accept a driving behavior in the table do not necessarily add to the total
 47

Side-by-side queuing Angle stopping

Encroachment on through lanes Encroachment on through lanes

Figure 26 Types of undesirable driving behavior on the median roadway of divided highway intersections.

frequency of undesirable driving behavior, because some maneu- the first vehicle of the pair entered it. Table 15 is analogous to
vers showed more than one type of undesirable driving behavior Table 14 but includes only those maneuvers showing undesirable
(e.g., both side-by-side queuing and angle stopping or both angle driving behavior with no other vehicles in the median. The col-
stopping and encroaching on a through lane). The rate measure umns headed Vehicle 1 include all maneuvers in which the first
used for undesirable driving behavior is the rate of maneuvers vehicle of the pair entered the unoccupied median area and
involving such behavior per 1,000 vehicles passing through the showed some type of undesirable behavior. The columns headed
median roadway. Vehicle 2 include maneuvers in which the first vehicle of the
The instances of undesirable driving behavior surrunarized in pair entered the median roadway without any undesirable behav-
Table 14 include behavior for some vehicle pairs in which the ior, but the second vehicle of the pair—often because of the
first vehicle of the pair entered the median when other vehicles presence of the first vehicle—did show some type of undesirable
were already there. To best determine the effect of particular behavior.
median geometrics on undesirable driving behavior, researchers An analysis was conducted of the maneuver types for which
decided to focus on those two-vehicle median-roadway maneu- one or more types of undesirable driving behavior were ob-
ver pairs in which no other vehicles were in the median when served. It was found that 64 percent of the undesirable maneu-
48

5,,

side-by-side queuing

,.~ j

Side-by-side queuing and angle stopping

Figure 27. Photographs of side-by-side queuing on the inedian roadway.

vers; involved vehicles traveling in the same direction through Median Width and Median Opening Length
the median roadway, while only 36 percent involved vehicles Effects
traveling in the opposite direction. This suggests that the most
common source of undesirable driving behavior on the median An analysis was conducted to determine the effect of the
roadway appears to be competition for space between vehicles geometrics (size and shape) of the median roadway on the types
traveling in the same direction through the median. of undesirable maneuvers observed above. The key descriptors
 49

~A'

Figure 28. Photograph of angle stopping on the inedian roadway.

Figure 29. Photograph of a vehicle on the inedian roadway encroaching on the

through lanes of the divided highway.

of the median roadway that were considered included the median unsi nalized intersections). The measures of correlation shown
width and the median opening length. Other descriptors of the g table are the Pearson correlation coefficients. The Pearson
in the
median roadway were also examined. including its area (the correlation coefficient is a measure of the linear association
product of the median width and median opening length) and between two variables and varies in value from 0.0 (represent-
its shape (the ratio of the median width and median opening ing no statistical association between the two variables) to
length), but these measures did not prove to be as useful as the 1.0 (representing perfect association between the two variables).
median width and median opening length. The sign of the Pearson correlation coefficient represents the
Tables 16 and 17 surm-narize the results of an analysis of sense of the relationship between the two variables. A positive
the correlations between the median width, median opening correlation coefficient means that as one variable increases in
length, and the rate of undesirable maneuvers per 1,000 vehicles value, so does the other~ a negative correlation coefficient
entering the median roadway (based on the data shown in indicates that as one variable increases in value, the other
Table B-9 in Appendix B by site and study period for decreases in value.
50
TABLE 14. Frequency and rate of undesirable driving behavior for all pairs of vehicles using the median roadway at the same time
(tabulated by site)

Total Type of U desirable Driving Behavior I Total Rate of

Vehicle Pairs Side-by- Encroachment I Frequency Median Undesirable
Site Observed Side Angle on ~ Undesirable Volume Maneuvers
Number in Median Queuing Stopping Major Road Maneuvers_ (vehs) (per 1000 vehs)
RURAL UNSIGNALIZED INTERSECTIONS
CA1 06 14 0 13 1 13 68 191.2
CA1 07 69 0 0 0 0 316 0.0
CA1 08 54 18 47 1 51 311 164.0
lAl 01 N/A 0 2 1 3 407 7.4
IL105 12 2 5 1 5 534 9.4
11-106 37 0 8 1 9 522 17.2
- KS101 311 22 87 0 100 1065 93.9
- MD101
1.18 25 37 4 53 509 104.11
- MD102
28 5 5 0 7 188 37.21
- MD103
92 16 16 0 30 571 52.5
- MD104 215 1 35 9 44 508 86.6
- MO104
75 0 1 0 1 662 1.5
M0105 31 0 23 0 23 363 63.4
TX101 14 0 3 0 3 204 14.7
TX103 143 10 92 0 95 844 112.6
TX104 666 186 479 7 519 1426 364.0
TX105 130 6 1 0 7 523 13.4
TX106 646 48 336 7 355 1417 250.5
WV1 01 99 2 46 0 47 807 58.2
WV103 219 4 1311 11 1321 8981 147.0
Subtotal 29731 3451 13671 331 14971 121431 123.3
SUBURBAN UNSIGNALIZED INTERSECTIONS
CA1 02 64 1 36 1 36 315 114.3
CA1 03 117 4 61 4 62 693 89.5
CA1 05 62 2 13 2 15 951 15.8
11-104 91 1 35 0 36 517 69.6
M01 01 227 12 100 0 109 828 131.6
M01 02 3931 81 2621 0 263 894 294.2
NJ101 5 0 4 0 4 106 37.7
WV1 02 141 0 0 0 0 990 0.0
Subtotal 1100 28 511 7 525 5294 99.2
SUBURBAN SIGNALIZED INTERSECTIONS
KS102 6 0 0 0 0 1416 0.0
M0103 5 0 0 0 0 3388 0.0
PA101 5 0 0 2 2 1091 1.8
TX1 02 0 0 0 0 0 976 0.0
TX107 964 0 5 52 551 50201 11.0
TX108 12781 01 0 14 141 28631 4.9
Subtotal 22581 01 51 68 71 1 147541 4.8
INTERSECTIONS WITH SPECIAL FEATURES
CA1 01 125 2 34 2 37 1274 29.0
CA104 33 0 9 2 11 250 44.0
11-101 6 0 0 0 0 1620 0.0
IL102 16 0 0 0 0 3383 0.0
IL103 28 0 0 0 0 7030 0.0
WV1 04 141 1 95 1 95 712 133.4
Subtotal 349 3 1381 51 1431 14269 10.0
TOTAL 6680 376 2021 1 1131 22361 46460 48.1

NOTE: Observed maneuver frequencies of specific types of undesirable driving behavior do not always
add to the total frequencies because some maneuvers exhibited more than one type of
undesirable driving behavior.
 51

TABLE 15. Frequency and rate of undesirable driving behavior for two-vehicle maneuver pairs with no other vehicles present in the
median area (summary by site)
T pe of Undesirable Driving Behavior by Type of Undesirable Driving Behavior by Total Rate of
Vehicle I Vehicle 2 Frequency Undesirable
Total Side-by- Encroachmen Side-by- Encroachment of Median Maneuvers
Site Maneuvers Side Angle on Side Angl on Undesirable Volume (per 1000
Number in Median Queuing ~ Stoppin9 Major Road Queuing in g
Stoppe Major Road Maneuvers (vehs) vehs)
RURAL UNSIGNALIZED INTERSECTIONS
CA106 1 6- 0 5 0 0 0 0 5 681 73.5 j
CA107 60 0 0 0 0 0 0 0 316 0.0
CA108 51 0 35 0 5 10 1 48 311 154.3
IAIOI N/A 0 0 0 0 0 0 0 407 0.0
IL105 11 0 3 0 2 2 1 5 534 9.4
IL106 34 0 4 1 0 3 0 8 522 15.3
KSI01 245 0 341 0 15 39 0 84 1065 78.9
MDIOI 95 0 22 0 13 10 3 44 509 86.4
MD102 23 0 2 0 1 1 0 3 188 16.0
MD103 77 6 10 0 8 4 0 27 571 47.31
MD104 169 0 0 7 1 24 2 33 508 65.0
M01 04 59 0 0 0 0 1 0 1 662 1.5
M0105 29 0 15 0 0 6 0 _21 ___363 57.9
TX101 43 0 16 0 0, 8 0 24 567 42.3
TX1 03 120, 0 63 0 5 11 0 77 844 91.2
TX1 04 3921 0 176 0 37 86 4 281 1426 197.1
TX105 951 0 1 0 2 0 0 3 523. 5.7
TX106 431 0 137 0 12 77 1 223 1417 157.4
WV101 89 0 31 0 0 42 8071 52.01
WV103 154 0 89, 0 1 11
14 1 6
01 103 8981 114.7;
Subtotal 2183 5 6431 8. 102, 3071 121 1032 125061 82.51
SUBURBAN UNSIGNALIZED INTERSECTIONS
CA102 531 0 26 0 0 5 0 31 315 98.4
CAI 03 97 0 41 0 0 10 1 52 693 75.0
CAI 05 61 0 11 0 0 2 1 14 951 14.7
IL104 69 0 20 0 1 5 0 26 517 5n.3
M0101 173 2 40 0 4 39 0 83 828 1 oo.i
M01 02 257 0 140 0 1 30 0 171 894 191.3
NJ101 5 0 3 1 106 37.7
WV102 1 116, 0 01 01
2 0
2 oi 01 91 9901 0.0
Subtotal 8311 21 2811 01 61 921 21 3811 52941 72.0
SUBURBAN SIGNALIZED INTERSECTIONS
KS1 02 51 0 0 0 0 0 0 0 1416 0.0
M0103 5 0 0 0 0 0 0 0 3388 0.0
PAI 01 5 0 0 0 0 0 2 2 1091 1.8
TX102 0 0 0- 0 0 0 0 0 976 0.0
TX107 870 0 0 17 0 3 8 28 5020 5.6
TX108 0 0 0 0- 0 2 2 2863 0.7
Subtotal 2097 0 01 171 0, 31 121 32 14754, 2.2
INTERSECTIONS WITH SPECIAL FEATURES
CAI01 1171 0 26 0 1 5 1 33 1274 25.9
CAI04 32 0 7 1 0 2 1 11 250 44.0
IL101 6 0 0 0 0 0 0 0 1620 0.0
IL102 16 0 0 0 0 0 0 0 3383 0.0
ILI 03 28 0 0 0 0 0 0 0 7030 0.
WV 104 111 0 66 0 0 10 0 76 712 106.7
Subtotal 310 0 99 17, 21 1201 142691 8.4
TOTAL 1 5421 1 0231 261 1091 4191 281 15651 468231 33.4

NOTE: Observed maneuver frequencies of specific types of undesirable driving behavior do not always add to the
total frequencies because some maneuvers exhibited more than one type of undesirable driving behavior.

Rural, Unsignalized Intersections: Table 16 shows that, at In contrast, the table shows that the observed rate of undesir-
rural, unsignalized intersections, the rate of undesirable maneu- able maneuvers increases as the median opening length in-
vers decreases as the median width increases. This suggests that creases. The correlation coefficient for this relationship is highly
as the median of a rural divided highway becomes wider, fewer significant. The median opening length at a divided highway
problems are observed at the intersections; however, this correla- intersection is largely dictated by the crossroad width and the
tion is quite weak and is only barely significant at the 90 percent design vehicle selected for left-turn maneuvers; however, these
confidence level. Table B-9 in Appendix B shows that the pre- findings suggest that, within these constraints, the median open-
dominant type of undesirable maneuver at rural unsignalized ing should not be made unnecessarily wide.
intersections is angle stopping, with a much lower frequency of A review of scatter plots of the data (shown in Figures B-7
side-by-side queuing, and very little encroaching on the and B-8 in Appendix B) suggest that the primary reason for the
through lanes. negative correlation between undesirable maneuvers and median
52

TABLE 16. Correlation between median roadway geometries and rate of undesirable maneuvers for rural unsignalized intersections

Correlation with
rate of undesirable Statisticall~ Significance Sample
Geometric variable maneuversa significant level size'
d 54
Median width —0.230 YES 0.095

Median opening length +0.503 YES, 0.001 54

Median area +0.131 NO 0.346 54

Slenderness ratio —0.393 YES 0.003 54

a Rate of undesirable maneuvers per 1,000 vehicles entering the median roadway as given in
Table B-9.
b
Statistically significant at 95% confidence level unless otherwise indicated.
c Number of study periods included in the analysis. Each study period is typically 2 hr in
length at a particular intersection.
d
Statistically significant at 90% confidence level.

TABLE 17. Correlation between median roadway geometries and rate of undesirable maneuvers for suburban unsignalized intersections

Correlation with
rate of undesirable Statisticall~ Significance Sample
Geometric variable maneuversa significant level size'

Median width +0.790 YES 0.001 21

Median opening length —0.718 YES 0.003 21

Median area —0.065 NO 0.781 21

Slenderness ratio +0.744 YES 0.001 21

a Rate of undesirable maneuvers per 1,000 vehicles entering the median roadway as given
in Table B-9.
b
Statistically significant at 95% confidence level unless otherwise indicated.
Number of study periods included in the analysis. Each study period is typically 2 hr in
length at a particular intersection.

width is that almost no undesirable maneuvers occur at medians The findings for suburban, unsignalized intersections indicate
more than 31 in (100 ft) wide. that medians should not be unnecessarily wide because the rate
of observed problems appears to increase with the median width.
On the other hand, median openings that are longer than neces-
Suburban, Unsignalized Intersections: Table 17 shows, for sary do not appear to be the concern in suburban areas that they
suburban, unsignalized intersections, the opposite relationship are in rural areas.
exists between the rate of undesirable maneuvers and the median
roadway geometrics from that found at rural, unsignahzed inter-
sections. A positive correlation exists between the rate of unde- Accident Analysis of Field Study Sites
sirable maneuvers and median width, and a negative correlation
exists between the rate of undesirable maneuvers and median An analysis of the accident data for the field study sites found
opening length. Although the correlations found are based on a no correlation between the intersection accident rates and either
relatively small sample size (traffic observational data for 21 2- median width or median opening length; however, given the high
hr study periods at eight intersections), they are highly signifi- variability of accident data and the small number of intersections
cant in statistical terms. As was the case at the rural, unsignalized available for analysis —20 rural, unsignalized intersections and
intersections, the predominant undesirable maneuver observed 8 suburban, unsignahzed intersections —it was expected that no
at the suburban, unsignalized intersections was angle stopping. statistically significant correlations would be found.
 53

Rural Intersections with Wide Medians The effect of median opening length on intersection operations
could only be evaluated from the field observational study data,
One concern underlying this research was that intersections
because median opening lengths were not available for the inter-
with wide medians could be confusing to drivers and, thus, might sections in the California accident data base. However, on the
cause many accidents. In fact, far from confusing drivers, the basis of the analysis of the field observational data, it appears
three rural, unsignalized intersections with medians more than
that the effect of median opening length is opposite to that of
31 m (100 ft) wide had some of the lowest rates of undesirable
median width; i.e., the frequency of undesirable driving behavior
driving behavior in the field observational studies. All three of
increases with median opening length for rural, unsignalized
these sites had both a relatively wide median and a relatively
intersections and decreases with median opening length for sub-
short median opening length. In all three cases, the median urban intersections.
opening length was 12 in (40 ft) or less; in one case, the median The findings on opposite effect of median opening length
opening length was only 9 in (30 ft). This design makes the
found at rural and suburban, unsignalized intersections are less
median roadway appear to drivers as part of the two-lane cross-
well supported than the related findings on median width, be-
road extended across the median. In fact, such median roadways cause the findings on median opening length are based solely
are typically marked with a double yellow centerline. The center- on the field observational studies and could not be evaluated in
line and the relatively narrow shoulder area on the median road-
an accident study. The authors are very comfortable with the
way help to remove the temptation for drivers to cut across the findings that short median openings are desirable at rural, unsig-
median and stop at an angle when turning left or to queue nalized intersections, because of the findings of the field obser-
side by side in the median. In other words, the geometrics and vational studies at rural intersections with wide medians dis-
pavement markings create the impression that there is not much
cussed above. The implication that, by contrast, longer median
choice in traversing the intersection except to follow the path openings are desirable at suburban intersections seems less well
that the designer intended. In addition, the wide median provides
supported, however, because of the potential for side-by-side
enough space to store a variety of vehicle combinations conven- queuing; therefore, they have chosen not to base any policy
tionally (i.e., one behind the other), minimizing the motivation recommendations on that finding.
for side-by-side queuing.

Summary of Traffic Safety and Operational Effects TRAFFIC CONTROL AT DIVIDED HIGHWAY
INTERSECTIONS
In summary, relationships were found between median width
and the safety performance of divided highway intersections, Current policies concerning traffic control devices, including
and these findings are supported by two independent analyses. signs, signals, and markings, are presented in the Manual on
Both the accident analysis of the Califori-da data base and the Unifonn Traffic Control Devices (MUTCD) (24). Little litera-
field observational studies suggest that, at rural, unsignalized ture exists on the effectiveness of different traffic control device
intersections, the frequency of accidents and undesirable driving schemes for divided highway intersections.
behavior decreases as the median width increases. At suburban,
unsignalized intersections, the opposite result was observed—
Unsignalized Intersections
indicating that the frequency of traffic accidents and instances
of undesirable driving behavior increases as the median width
Current Traffic Control Policies
increases. Traffic accidents also increase with median width at
suburban, signalized intersections.
The MUTCD presents three signing plans that are applicable
There is no obvious explanation for the different effects of
to divided highway intersections:
median width at rural and suburban, unsignalized intersections.
However, these findings appear to be well founded empirically
MUTCD Figure 2-3 (presented here as Figure 30) has Stop
because they are supported by both the accident studies and the
signs on the median roadway and is applicable to divided
field observational studies. Possible explanations for this effect
were examined, but none was conclusive. Specifically, while highway medians more than 9 in (30 ft) wide.
The lower portion of MUTCD Figure 2-3a (presented here
the average median volume for rural, unsignalized intersections
as the lower portion of Figure 31) shows an alternative
is lower than for suburban, unsignalized intersections, the range
configuration to Figure 2-3 with Yield signs on the median
of median volumes at rural, unsignalized intersections com-
roadway.
pletely overlaps the range of median volumes at suburban, unsig-
The upper portion of MUTCD Figure 2-3a (presented here
nalized intersections; thus, it does not appear that differences in
as the upper portion of Figure 31) shows no Stop or Yield
traffic volumes on the median roadway explain the differences
signs in the median and is applicable to divided highways
in the observed effects. Suburban, unsignalized intersections
with median widths of less than 9 m (30 ft).
tend to have lower ratios of crossroad to major-road ADT than
rural, unsignalized intersections, because the major-road through
volumes are relatively high. No one factor appears to account The MUTCD gives very specific guidance on the use and
placement of all these signs in Section 2A-3 1, Wrong-Way Traf-
for the observed differences. Instead, the observed differences
fic Control. The MUTCD states the following:
are probably a combination of differences in many factors be-
tween rural and suburban intersections, including major-road
Where roadways are separated by median widths of 30 ft or
through volumes; trip length; daily, weekly, and monthly traffic more, the intersections with the crossroad shall be signed as two
volume variations; and driver expectancy. separate intersections and One Way signs (Section 213-29) should
54

to
r7kVM =0

rTNIM
—11
'__1

rXVMMM3N"
0 zl !ur

EAY\
MM W"

rk7MM3 0N "

DIRECTION OF TRAFFIC
*

Figure 30. Locations of One-Way and turn prohibition signs at divided highway
intersections (28).

be visible to each crossroad approach on the near right-hand and on a divided highway at a location to be directly in view of a driver
far left-hand comers of each intersection with the directional road- making a wrong-way entry from the crossroad. Additional signs
ways as shown in Figure 2-3. However, when an engineering study may be placed where the median width is 30 ft or more (28).
has demonstrated that placement of One Way signs in the median
area may create confusion, the near right-hand signs in the median
may be omitted and One Way placed in the far-right quadrant of As median widths increase to the point that the intersection
the intersection. Figure 2-3a shows this alternate scheme with one functions more like two separate, closely spaced intersections,
pair of One Way signs in the median replaced by Yield signs. Turn drivers of larger vehicles may be tempted to enter the median
prohibition, Do Not Enter and Wrong Way signs may be used to
area between the two intersections, even if there is not enough
supplement One Way sign layouts in Figures 2-3, 2-3a, or 2-4
[NOTE: MUTCD Figure 2-4 does not relate to intersections on room in the median area to store their vehicles. That may occur
divided highways]. because the median roadway appears similar to a small "street"
One Way signs are not ordinarily needed at divided highway between the two intersections, which may suggest to some driv-
intersections with median widths of less than 30 ft. In cases where ers that the median is wide enough to accommodate larger vehi-
they are needed, combinations of One Way and/or Divided High-
way Crossing, Do Not Enter, or Wrong Way signs may be used to
cles. Additional signing and pavement markings could add to
improve operations at these intersections. this problem. As with the case described above, the MLJTCD
If used, Do Not Enter and Wrong Way signs should be placed will allow "street-like" traffic control when medians are wider
 55

He
: MEDIAN LESS THAN 30'

r7VX0"
3N0
OPTIONAL

.91.
OPTIONAL :3 . I rTZINS
=OPTIONAL

OPTIONAL
C l
MEMA"
6P—-DONAL

OPTIONAL
OPTIONAL

OPTIONAL -K
001I.I.: [~~'
OPTIONALF0
V
NOTE: DO NOT ENTER AND
WRONG WAY SIGNS OPTIONAL
Figure3l. Alternative One- Way signing for divided highway intersections (28).

than 9 m (30 ft). However, a 9-m (30-ft) median provides only divided highways is not stated in this MUTCD. It seems likely
enough storage space for one or two passenger cars, and a larger that this value is based on the operating experience of highway
design vehicle (e.g., WB-50) could not be stored in a 9-m (30- agencies rather than any particular piece of research.
ft) median without encroaching on the opposing lanes. Current highway agency practice appears to be to use no
The MUTCD does not distinguish clearly between the apphca- control on the median roadway for intersections with medians
bility of the upper and lower portions of Figure 2-3a. The upper less than 9 m (30 ft) wide and to use stop control for medians
portion of the figure is labeled as applicable to "medians less more than 25 rn (82 ft) wide. However, stop control, yield con-
than 30'." The lower portion of the figure is not labeled with a trol, and no control are each used by some highway agencies
specific median width range, which might lead one to suppose on the median roadway of intersections with median widths in
that it is also applicable to narrow medians; however, both the the range of 9 to 25 m (30 to 82 ft).
text quoted above and the equivalent figure in the 1978 edition The signing at a divided highway intersection has two clear
of the MUTCD (29) make it clear that the lower portion of functions:
Figure 2-3a is applicable to medians more than 9 rn (30 ft) wide.
The reason for the choice of 9 rn (30 ft) as the threshold * To establish the right of way for vehicles crossing or turn-
median width above which separate yield or stop control is ing onto each roadway
provided for both intersections between the minor road and the * To inform drivers on the crossroad approaches that they
56

OR

Entrance Ramp

L?250'—

r=—
IMI
t rIN V
_AJ
Y

KEY:
Barrier Lines or
cz~r --ow. Lane Use and Wrong-Way Arrows
Raised Median
Direction of Flow
Optional

Figure 32. Application of Do Not Enter and Wrong Way signs on a freeway exit ramp.

have reached a divided highway with two separate roadways of the roadway especially when the entering vehicle turns to
rather than an undivided roadway enter the restricted road. The Wrong Way sign is often used to
supplement the Do Not Enter sign on divided highways. The
The crossroad approaches are normally controlled with Stop MUTCD states that the Wrong Way sign should be placed far-
signs, as a minor road entering a through highway or street is ther from the crossroad than the Do Not Enter sign. The issue
one of the warrants for a Stop sign in MUTCD Section 213-5. of wrong-way movements at divided highway intersections is
The sight distance implications of these various choices of traffic addressed further in a later section of this chapter.
control devices are examined in the subsequent section on ISD. MUTCD Figure 2-3 (presented here as Figure 30) shows the
The remainder of the signing is intended to help crossroad driv- median roadway of a divided highway marked with stop bars
ers to recognize the major road as a divided highway and to and a double yellow centerline, much as a conventional roadway
reduce the potential for wrong-way movements. The signs that or street would be. This marking scheme appears to be quite
help reduce wrong-way movements are the following: effective in conveying the message to drivers that it is acceptable
to cross the near roadway of the divided highway and then wait
The One-Way sign (MUTCD sign R6-1 or R6-2)
in the median before proceeding. There is very little guidance
The Divided Highway Crossing sign (R6-3 or R6-3a)
in the MUTCD or other sources on delineation requirements at
The No Right Turn and No Left Turn signs (R3-1 and R3-2)
divided highway intersections other than that shown in MUTCD
The Do Not Enter sign (R5-1)
Figures 2-3 and 2-3a.
The Wrong Way sign (R5-9)
Most unsignalized, divided highway intersections have two-
Typical applications of these signs (except for the Do Not Enter way stop control; however, four-way stop control has been used
and Wrong Way signs) are illustrated in MUTCD Figures 2-3 at some divided highway intersections between major routes.
and 2-3a. There have been some studies on four-way stop control for
MUTCD Figure 2-22a (presented here as Figure 32)illustrates higher-speed intersections. One study, by Briglia (30), found
the application of the Do Not Enter and Wrong Way signs. The that accident rates declined by an average of 58 percent at 10
Do Not Enter sign is used to tell drivers that entering the roadway rural intersections in Michigan when two-way stop control was
where the sign is posted is prohibited. The sign is usually replaced by four-way stop control. The increased cost associated
mounted on the right side of the roadway where it would face with the vehicle delays, however, would offset any savings from
the traffic entering the wrong way. The MUTCD states that it the reduced accident levels except at intersections with low or
may also be appropriate to mount a second sign on the left side moderate mainline volumes.
 57

Analysis of Field Observational Data As reported earlier in this chapter, it was observed that vehi-
cles making opposing left turns tend to turn in front of each
An analysis was conducted to examine driver compliance with other at intersections with medians less than 15 m (50 ft) wide
traffic control devices on the median roadway and to ensure that and to turn behind one another at intersections with medians
a traffic control device effect was not mistaken for an effect of more than 15 in (50 ft) wide. No indication was found that
median width or median opening length. In particular, the analy- this tum-in-front/tum-behind behavior depended on the type of
sis focused on the traffic control device used on the median traffic control device used on the median roadway.
roadway at its intersection with the far roadway of a divided
highway.
Compliance with traffic control devices on the median road- Signalized Intersections
way was evaluated in two ways. First, special substudies of Stop
sign compliance were performed for four of the seven rural, Current Traffic Control Policies
unsignalized intersections at which Stop signs were used. These
studies classified each vehicle passing through the median road- There is little guidance in the literature or in traffic control
way as making a full stop or a rolling stop or running the Stop device policies about the use and placement of traffic signals at
sign. Similar studies were not performed for intersections with intersections on divided highways. At some minimum median
yield control or no control on the median roadway because width, it becomes desirable to have separate signals on the minor
vehicles encountering such control are not required to stop but road for each roadway of the divided highway; however, there
are required to yield -the right of way to vehicles on the major are no formal policies on what this minimum median width
road. The compliance or lack of compliance at intersections with should be. While little guidance is available on traffic signals
yield control or no control can be judged only by determining that is specifically applicable to divided highway intersections,
whether vehicles departing from the median roadway created a information is plentiful on the use of signals at intersections on
traffic conflict by forcing a vehicle on the major road to brake higher-speed facilities. In an accident-based safety study of traf-
or swerve. Therefore, the second analysis performed, which in- fic controls at rural, high-speed intersections, Agent (31) ob-
cluded rural, unsignalized intersections with all types of control served that signalized intersections with high accident rates had
on the median roadway, determined the number of conflicts problems with opposing left-turn (i.e., vehicles turning left from
observed between major-road vehicles and vehicles crossing or the major roadway) accidents. For these types of accidents, the
entering the major road from the median roadway. In other most common anecdotal reasons for the accidents were that the
words, conflicts involving a crossroad vehicle entering or cross- left-turning drivers did not see the oncoming vehicles or that
ing the near roadway of the divided highway were not counted, their vision was obscured. The sight distance requirements for
but conflicts involving a crossroad vehicle entering or crossing this type of left-tum maneuver are addressed later in this chapter.
the far roadway of the divided highway were counted. Such Numerous studies also exist on active warriing devices at
conflicts could involve four different types of vehicles using the intersections (32,33,34). These devices include flashing Red
median roadway: major-road vehicles making a left turn or a Signal Ahead signs, flashing Prepare to Stop When Flashing
U-tum and crossroad vehicles making a through movement or signs, flashing beacons, and red signal lenses with flashing
a left turn. strobes. Each of these device types appears to improve the safety
The traffic control device compliance data are presented in of signalized intersections on higher-speed roadways, and many
Appendix B. For the intersections studied, driver noncompliance of the locations where they have been applied are on divided
with Stop signs on the median roadway (failure to stop) ranged highways in rural or suburban areas. Most of the evaluation
from 11.6 percent to 42.7 percent of vehicles using the median studies, however, do not seem to account for novelty effects,
roadway. The percentage of drivers who made a rolling stop which may last only until drivers become familiar with the
ranged from 30.8 to 39.0. A full stop was made by 22.9 percent device.
to 50.0 percent of drivers on the median roadway. Although less
than 50 percent of drivers generally comply fully with Stop signs
on the median roadway, most of the drivers could have stopped Simulation Analysis of Signalized Intersection
if another vehicle were in the way. As shown in Table B-14 in Operations
Appendix B, no traffic conflicts resulted from these maneuvers.
In fact, as shown in Table B-14, only four traffic conflicts be- A simulation analysis was conducted to determine the effect
tween major-road vehicles and vehicles emerging from the me- of median width and signal configuration on the efficiency of
dian roadway were seen in the entire data set for rural, unsignal- traffic operations at signalized intersections. The approach to
ized intersections. These four conflicts all happened to occur at this analysis is described earlier in this chapter and is presented
sites with yield control, but this sample size- is too small to draw in greater detail in Appendix E.
any conclusions about differences between the types of traffic Figure 33 presents the effect of median width on total delay at
control. It is apparent, however, that whatever type of traffic signalized intersections for two sets of approach traffic volumes
control is used on the median roadway, conflicts between major- ("moderate" volumes corresponding to Level of Service B and
road vehicles and vehicles emerging from the median roadway "high" volumes corresponding to the boundary between Levels
are rare. of Service C and D) and a set of typical turning movement
There is no indication that the above findings concerning the patterns, presented in Appendix E, that has been termed the
effect of median width and median opening length on the rate baseline scenario. Separate relationships to median width are
of undesirable maneuvers in any way depend on the type of shown in the figure for a single signal installation controlling
traffic control used on the median roadway. both roadways of the divided highway and for separate signal
58

baseline scenario (:3)

50

0
D
40

0
0

kid
C
03
4-1
0
20

LO B
10

A
0
0 50 100 150 200 250 300
Median Width (ft)

0 15 30 45 60 75 90

Median Width (in)

High Volume Single -,&- Moderate Volume Single

High Volume Diamond )( Moderate Volume Diamond

Figure 33. Illustration of the effect of increased median width on total delay at signalized intersections.

installations on the two roadways of the divided highway (equiv- with increasing median width, but total delay is less sensitive
alent to the type of signalization provided at a diamond inter- to median width than for the single signal consideration. For
change). Simulation models were used to obtain delay estimates both the moderate- and high-volume scenarios, the slope of the
for the single signal configuration for median widths of 6, 18, delay compared with median width relationship over the range
and 31 in (20, 60, and 100 ft) and for the diamond configuration of median widths of 17 to 56 in (55 to 185 ft) is -0.034 sec/veh/m
for median widths of 17, 37, and 56 in (55, 120, and 185 ft); (-0.010 sec/veh/ft).
where the relationships in Figure 33 are extrapolated outside Figure 33 shows that, for the moderate-volume level in the
these ranges, they are shown as dashed, rather than solid, lines. baseline scenario, the total delay for the single signal configura-
The figure shows that, within the range of median widths tion remains less than for the diamond signal for median widths
present at most divided highway intersections, the single signal of up to approximately 85 m (280 ft). For the high-volume level
installation operates with less total delay than the diamond con- in the baseline scenario, the point of equal delay for the single
figuration. For the single signal configuration, total delay in- and diamond signal configurations occurs at a median width of
creases as median width increases. For the moderate-volume approximately 49 m (160 ft). These points of equal delay were
level, as the median width increases from 6 to 31 in (20 to 100 examined for each of the scenarios evaluated, as described in
ft), the total delay increases relatively little and remains within Appendix E. On the basis of these findings, a general recommen-
Level of Service B over this range; the slope of this relationship dation is made that the single intersection configuration be used
for the moderate-volume level is 0.12 sec/veh/m (0.036 for median widths of less than 31 in (100 ft) and the diamond
sec/veh/ft). For the high-volume level, delay increases with me- intersection configuration be used for median widths of more
dian width at a steeper slope of 0.28 sec/veh/m (0.086 than 46 in (150 ft). For intersections with median widths between
sec/veh/ft); over the range of median widths of 6 to 31 in (20 these values, an acceptable design can probably be accomplished
to 100 ft), the total delay increases from Level of Service C to using either strategy. Designers should be aware that for all
Level of Service D. single signal intersections with median widths of more than 18
For the diamond signal configuration, total delay decreases in (60 ft), it is possible that vehicles may become trapped on
 59

the median roadway at the end of the crossroad green phase, cent of the traffic that used the offset left-tum lanes to make
and the signal system should be designed to minimize this possi- left turns during the study period. In fact, one of the maneuvers
bility (e.g., with vehicle detectors on the median roadway). . in question did not occur during the study period.
In summary, it appears that from a traffic operational point The accident history of the three field sites with offset left-
of view, the median width of a signalized intersection with a turn lanes was reviewed. The two intersections with offset left-
single signal should be kept as small as possible to minimize turn lanes experienced more left-tum accidents in 3 years on the
delay. Furthermore, to minimize delay, the diamond signal con- crossroad approaches (which had conventional left-tum lanes)
figuration should be avoided except at intersections where a than on the major-road approaches (which had offset left-tum
wide median is needed for other reasons. lanes). None of the intersections with tapered offset left-tum
lanes experienced any problem with accidents related to wrong-
way movements through the offset left-turn lanes. The intersec-
TRUCK CONSIDERATIONS tion with parallel offset left-tum lanes experienced more left-
turn accidents in 3 years on each of the three approaches with
The field observational studies found no significant problems offset left-turn lanes than on the one approach with a conven-
related to truck operations at the field study intersections. In tional left-turn lane; however, the approach with the conven-
general, truck drivers exercised good judgment about the length tional left-tum lane also had the lowest left-tum volume of any
of their vehicles in relation to the median width. Furthermore, of the four approaches to this intersection. In summary, the
the accident evaluation of the field study intersections found available data do not indicate any particular accident problems
that trucks were generafly underrepresented in accidents at the with tapered or parallel offset left-tum lanes. However, the avail-
study sites. OveraIl, trucks constituted 9.2 percent of the major- able sample sizes of accidents and intersections are not sufficient
road traffic and 5.2 percent of the crossroad traffic at the study to indicate that any particular left-tum lane design is superior.
intersections, but trucks represented only 4.3 percent of the vehi- All three study sites with offset left-turn lanes were located
cles involved in accidents. at signalized intersections. Offset left-tum lanes have also been
used, on a more limited basis, at unsignahzed intersections.
However, in our search for field study sites, no unsignalized
LEFT-TURN TREATMENTS intersections were found with offset left-tum lanes and sufficient
traffic volume to make a field study productive.
Current design practices for left-tum treatments at divided It should be noted that the field observational studies found
highway intersections have been discussed earlier in this chapter. that most undesirable driving behavior at unsignalized intersec-
Most of the field study sites had conventional left-tum lanes, tions involved competition for limited space between vehicles
and no particular traffic operational or safety problems related traveling in the same direction through the median roadway.
to these conventional left-tum lanes were found. Several inter- Tapered offset left-turn lanes could potentially reduce such con-
sections with innovative left-tum treatments, including tapered gestion by removing major-road left-turn vehicles from the me-
offset left-tum lanes, parallel offset left-turn lanes, and median dian roadway. However, because tapered offset left-tum lanes
acceleration lanes, were included in the field observational stud- have not been used extensively at unsignalized, four-leg intersec-
ies. The findings of these studies are summarized here and are tions with higher traffic volumes, it is not known how effective
presented in greater detail in Appendix D. These types of left- they would be. An evaluation is needed to ensure that they are
turn treatments are illustrated in Figures 8 and 9. not confusing to drivers; furthermore, because major-road left-
turn movements and traffic movements through the median road-
way are not separated in time by signal phasing as they would
Tapered and Parallel Offset Left-Turn Lanes be at a signalized intersection, it is important to establish that
conflicts between these movements do not result in accidents.
The field observational studies included two signalized inter- Because of these concerns, no recommendations have been made
sections with tapered offset left-tum lanes and one intersection on the use of tapered offset left-turn lanes at unsignalized
with parallel offset left-turn lanes. Although the sample was intersections.
very Iiintited, no operational problems were observed related to
median width or driver understanding of the offset left-tum
lanes. In fact, during the daytime study periods, no undesirable Median Acceleration Lanes
driving behavior was observed at any of the three intersections.
One undesirable driver action, involving a driver who missed Four intersections with one or more median acceleration lanes
the entrance to a tapered left-turn lane and then backed up in a were included in the field studies. These intersections included
through lane to return to it, was observed during a break in the three four-leg, unsignalized intersections with median widths of
field studies. Another undesirable maneuver was observed in a 16, 22, and 44 m (54, 72, and 144 ft) and a three-leg, unsignal-
night study when a driver intending to turn left did not realize ized intersection with a median width of 11 in (36 ft). These
until the last moment the need to enter the tapered offset left- studies found that median acceleration lanes can enhance the
turn lane; the driver braked hard and swerved across the gore operation of intersections on divided highways. In particular,
area from the left through lane into the tapered offset left-tum median acceleration lanes reduce the likelihood that vehicles
lane but did not run onto the curb in the gore area. In both of making a left turn from a crossroad approach will need to stop
these maneuvers, no other traffic was present and, thus, no con- in the median; stopping in the median is not necessary to yield
flicts resulted. to through traffic on the divided highway but is necessary to
These two undesirable maneuvers represent less than 0.1 per- yield to other traffic on the median roadway. Median accelera-
60

tion lanes work best when the acceleration lane is long enough The AASHTO policy provides geometric design criteria for
to allow the accelerating vehicles to reach the speed of major- five ISD cases, which are defined briefly in this section. One of
road traffic before merging into the through lanes. the five cases has three subcases so there are, in effect, seven
Operational problems were observed at only one site where ISD cases. These seven cases are defined by the type of traffic
some left-tum drivers failed to use the median acceleration lane control present at the intersection and the types of maneuvers
and several traffic conflicts with through vehicles resulted. How- that the drivers intend to make. The seven cases are the
ever, it should be noted that the median acceleration lane did following:
not cause these problems. The traffic conflicts created by the
drivers who failed to use the median acceleration lane would * Case I—ISD for vehicles approaching intersections with
have occurred in any case (perhaps in greater numbers) if the no control, at which vehicles are not required to stop but may
median acceleration lane were not present. be required to adjust speed
* Case II—ISD for vehicles on a minor-road approach con-
trolled by a Yield sign
INTERSECTION SIGHT DISTANCE 9 Case IIIA—ISD for a vehicle on a Stop-controlled ap-

proach on the minor road to accelerate from a stopped position

ISD is an important design and operational consideration at and cross the major road
all intersections. ISD at divided highway intersections is compli- e Case M—ISD for a vehicle on a Stop-controlled approach
cated by the presence of the median on the major road, which on the minor road to accelerate from a stopped position and turn
may increase the ISD requirements at some intersections or may left into the major road
contain sight obstructions that reduce the ISD. * Case IIIC—ISD for a vehicle on a Stop-controlled approach
The AASHTO policy on ISD design is presented on pages on the n-Linor road to accelerate from a stopped position and turn
696 through 721 of the AASHTO Green Book (1). The Green right into the major road
Book considers ISD to be adequate when drivers at or'ap- Case IV—ISD for a vehicle on a signal-controlled approach
proaching an intersection have an unobstructed view of the entire Case V—ISD for a stopped vehicle turning left from a
intersection and of sufficient lengths of the intersecting highways major road
to permit them to anticipate and avoid potential collisions. Ade-
quate ISD requires unobstructed sight distance along both ap- Each ISD case and its applicability to divided highway inter-
proaches of both intersecting roadways, as well as across their sections is summarized briefly below.
included comers (the so-called "clear sight triangles"). Adequate
clear sight triangles are required both for drivers approaching
an intersection where they are not required to stop and for drivers ISD Case I—Intersections with No Control
who are stopped at an intersection waiting to proceed safely to
cross a major roadway or to turn left or right onto a major ISD Case I is not generaUy applicable to divided highway
roadway. intersections because such intersections are seldom operated
ISD requirements for crossing and turning maneuvers at di- with no control on any of the approaches.
vided highway intersections are generally increased with median
width until the median becomes wide enough to store a vehicle.
If the median is wide enough to store a vehicle, then the intersec- ISO Case 11—Yield Control on Minor Road
tion operates as two separate intersections, because drivers can
cross the near roadway and stop in the median, if necessary, The use of yield control on the crossroad approaches to a
before crossing or turning into the far roadway. In this case, divided highway intersection is unusual because an intersection
the sight distance requirements of the intersection with the two of a less important road with a main road is one of the MUTCD
roadways of the divided highway can be determined separately. warrants for use of a Stop sign. Thus, ISD Case II is not generally
The minimum median width that would generally be consid- applicable to the crossroad approaches to divided highway inter-
ered adequate to store a passenger car is 7.6 in (25 ft), on the sections. Furthermore, because the current AASHTO model for
basis of the 5.8-m (19-ft) length of the AASHTO passenger car ISD Case H is based on a stopping maneuver, rather than a
design vehicle. Wider medians would be required to safely store crossing maneuver, the ISD requirements for Case 11 do not vary
a 12.1-m (40-ft) bus or a tractor-semitrailer truck (currently up as a function of median width; however, if it is necessary for a
to 22.5 in [75 ft] long). The choice of the design vehicle is driver of a vehicle on a yield-controlled approach to stop because
important for ISD on divided highways because there obviously of traffic on the major road, wider medians may require more
is a range of median widths wide enough to store a passenger ISD for crossing and turning maneuvers as the vehicle acceler-
car but not wide enough to store a longer vehicle. ates from a stop. This is discussed further in ISD Cases M,
An extensive reexamination of existing ISD design policies IHB, and HIC below.
and practices was done in NCHRP Project 15-14(l). That re- Although yield control is not often used on the minor-road
search is expected to recommend revisions to current ISD design approaches to divided highway intersections, yield control is
policies, including policies applicable to divided highway inter- frequently used at the intersections of the median roadway with
sections. Because this related work is not yet complete, the the divided highway. Thus, there is a need to consider ISD
following discussion identifies the situations where ISD require- Case II for vehicles on the median roadway; however, the sight
ments for divided highway intersections may differ from undi- distance needs for vehicles facing a Yield sign on the median
vided highway intersections but does not recommend specific roadway do not depend on the median width, because the pres-
changes in ISD criteria. ence of a Yield sign in the median presupposes that enough
 61

space is in the median for a vehicle to stop safely and wait for policy for Case IV notes that the required sight distance for Case
a suitable gap in the major-road traffic. III should be provided at signalized intersections. This policy
applies to signalized intersections on both divided and undivided
highways. The differences in the underlying criteria for ISD
ISD Case IIIA—Stop Control on Minor Road: Cases M and I111C between divided and undivided highways
Crossing Maneuver were discussed above.

ISD Case UIA applies to the sight distance requirements for

a minor-road vehicle to 'cross a major roadway. On divided ISD Case V—Left Turns from the Major Road
highways with medians wide enough to store a vehicle, the sight
distance requirements for crossing each roadway are determined The 1994 AASHTO Green Book introduced a new ISD case
separately. Thus, for this situation, ISD requirements would not on the basis of the policy on sight distance requirements for left
be a function of median width. If the median is not wide enough turns from the major road that first appeared in the 1990 Green
to store a vehicle, however, then the crossing of both roadways Book (1,2). The required left-turn sight distance is defined by
and the median must be performed in a single maneuver—thus, the equation:
for narrower medians, the ISD that should be required for Case
IIIA increases as the median width increases. In this case, the SD,q = 0.28 V (J + t,) (2)
path length used to determine ISD should include the width of where
both roadways of the divided highway and the median width.
SD,q = sight distance required to safely complete a left turn off
a major road (m)
ISD Case IIIB—Stop Control on Minor Road: Left-
Turn Maneuver
V = speed of vehicle on a major road (knVh)
J = perception-reaction (PR) time (sec) (assumed: J = 2.0
sec)
ISD Case IIIB applies to the sight distance requirements to
turn left from a stop-controlled approach onto a major road.
t,, = time required for a turning vehicle to accelerate and
travel a length equal to d,, (sec)
Where the median is wide enough to store a vehicle, provision
of the required sight distance for Case HIC will also provide
d,, = length of vehicle's turning path to clear opposing lanes
(i.e., path length to curb line plus vehicle length) (m)
enough sight distance for vehicles to cross the near roadway of
the divided highway. Sight distance for Case M should be The time required to traverse the distance da can be determined
provided for vehicles to turn left from the median roadway onto from Green Book Table IX-8. At a divided highway intersection
the far roadway of the divided highway; however, the sight with a median that is not wide enough to store a turning vehicle,
distance requirements for this maneuver are no different from the path length used to determine d,, should include the vehicle's
those for a left turn onto an undivided highway and are not turning path across both the median and the opposing lanes of
influenced by the median width. the divided highway. Where the median is wide enough to store
Where the median is not wide enough to store a vehicle, a turning vehicle, the path length needs to include only crossing
Case IIIB sight distance should be provided on the crossroad of the opposing lanes, and the ISD criteria for left turns off the
approach. In this situation, the Case IIIB sight distance must major road become equivalent to the ISD criteria for the crossing
allow for the time to cross the near lanes and for the time to cross maneuver in Case IIIA.
through the median. Thus, in this situation, the sight distance The sight distance for left-turning vehicles at divided highway
requirement for Case IlIB is a function of the median width. In intersections may be limited by fixed sight obstructions on the
the current AASHTO model for ISD Case 111113, that can be roadside and by opposing left-tum vehicles. Appendix F presents
accomplished by including the width of the near lanes of the equations based on the work of McCoy et al. (35) that can be
divided highway plus the median width to the distance to be used to determine the offsets between opposing left-turn lanes
traveled by the turning vehicle. that will eliminate or minimize sight obstructions caused by
opposing left-turn vehicles.

ISD Case IIIC—Stop Control on Minor Road: Right-

Turn Maneuver HUMAN FACTORS/DRIVER PERCEPTION
CONSIDERATIONS
ISD Case HIC deals with the sight distance required to turn
right from a minor-road stop-controlled approach onto a major The review of the literature in this area focused on driver
road. The sight distance requirements for right turns are indepen- performance characteristics that affect the design of intersec-
dent of median width and are unaffected by whether the highway tions. None of the available information is specifically applicable
onto which the crossroad vehicle is turning is divided or to divided highway intersections.
undivided.

Perception-Reaction Time
ISD Case IV—Signal Control
Generally, perception-reaction time has been thought to be
ISD Case IV addresses the sight distance requirements for the time needed to perceive some stimulus, and, if necessary,
vehicles on signalized intersection approaches. The AASHTO the time needed to take some type of action in response. Numer-
62

ous studies have attempted to analyze the component parts of design has concentrated on the 102- and 153-mm (4- and 6-in)
what has simply been called perception and reaction. The general objects that have been used for SSD considerations (38,39,40).
concept of perception has been described as being made up Although many of these studies provide useful information re-
of sensation, detection, eye movement latency, eye movement, garding the stopping sight distance (SSD) problem, none of these
fixation, accommodation, perception, cognition, recognition, in- studies addresses the basic issues related to ISD requirements.
tellection, comprehension, decision, and emotion. Some of these Besides identifying other vehicles, drivers on the minor road
terms describe the same element or parts of the same element. at intersections on divided highways must also detect and recog-
The reaction component is thought to be made up of just a nize that there are two one-way roadways separated by a median.
simple reaction (or volition or action response) element. The Recognizing this situation is one of the cues that prevents wrong-
reaction can be a movement response of almost any type (e.g., way driving. Principally, the driver depends on a clear line of
for driving applications, foot from accelerator pedal to brake sight to the far roadway of the divided highway so that he or
pedal or foot off accelerator pedal). she can identify that roadway and recognize it as the remaining
One approach for calculating the time element related to hu- half of a divided highway. Problems occur when the design of
man operators of different systems has been to measure values the highway cross section makes it difficult for the motorist to
for several of the constituent elements and add up these time see the other roadway or intersection (41).
segments to determine an appropriate perception-reaction time.
Another approach has been to calculate percentile values for the
measurable elements and assemble these percentile values into Older Drivers
some single value or range of values for perception-reaction
time. These approaches have several potential disadvantages. In recent years, there have been several excellent studies of
One disadvantage is the implicit assumption that the elements older drivers that address their declining functional capacities
of the perception-reaction process act in series without any time particularly, visual, cognitive, and psychomotor performance.
overlap or parallel functioning. Another disadvantage is that it Reports by Staplin (42); Dewar, Templar, and Knoblauch (43);
is highly unlikely that an individual will consistently perform at and Shinar and Scheiber (44) summarize most of the important
or near specified percentile values for all the individual elements research findings related to older driver and pedestrian
of the percept. ion-reaction process (36). Other researchers have performance.
proposed perception-reaction models that are probabilistic. Such The performance of several different visual functions deterio-
models have similarities with signal detection theory and Bayes- rate with age. Older drivers have reduced contrast sensitivity.
ian decision theory; however, there is no one analytical approach Visual acuity changes begin at age 40. The decline is slow at
(including the additive factors approach described above) on first but accelerates after 60 or 70. Other visual problems include
which researchers generally agree (36). a loss in the size of the visual field area (the field of view).
Additional attempts to determine perception-reaction times Changes in the visual field can be measured as a reduction in
have taken the form of empirical studies of the entire perception- field area or as an elevation in detection thresholds at locations
reaction process. Given our analytical models of perception- within the field.
reaction time, studies that measure the entire perception-brake- Although it is important to note these various changes in
reaction time are usually preferred over those studies that attempt visual performance of the elderly, it is equally important to
to measure the individual components and then add them to- recognize that many studies have shown driving performance is
gether. In these studies, measurements are made from the onset not inextricably linked to visual performance. Many accidents
of the perception process through the completion of the reaction in which human error is cited also seem to have some cognitive
component and the onset of mechanical acceleration or braking; performance problem as a contributing factor.
however, many of these studies tend to be deficient in that the Changes in cognitive capabilities occur with aging. These
subjects are already alerted that their reactions will be tested. changes are associated with behaviors such as attention span,
visual search behavior, memory functions, and complex problem
solving. Older drivers are known to have problems with visual
Detection and Recognition of Objects and search capabilities. For example, when asked to search for target
Vehicles stimuli, older drivers decide more slowly and generally make
more mistakes in identification. Older people have difficulty
The principal problem for the driver in all sight distance situa- refocusing on a specific task when asked to perform other
tions is detecting and identifying other vehicles. Detection and tasks—this can cause problems when older drivers are per-
recognition is primarily a matter of visual search and target forming tasks such as locating destinations or looking for places
acquisition. Research on visual search has generally focused to make turns. Memory is very important in the synthesis and
on models of search time and the identification of factors that management of sensory information to solve problems. Memory
influence the speed of the search (36). Much of this research is interacts with decision and response functions. These functions
based, however, on subjects' abilities to search for and identify are critical for driving, because control, traffic, and alignment
letters or numbers from an array. Boff and Lincoln (37) (in information must be attended to and stored temporarily to give
Fambro et al. [38]) maintain that target acquisition is affected the driver information to control the vehicle and plan the next
by brightness (luminance), texture contrast (including color), maneuver.
texture gradients (boundaries), dimensions of the target, dynam- Reaction time slows with age; however, brake reaction time
ics, and environmental factors. is slowed by only 0.1 sec for a 75-year-old as compared with a
From an applied standpoint, most (if not * all) of the research 25-year-old test subject (45). Researchers examining actual on-
related to visual search and object recognition related to highway road behavior have not found any significant age effects related
 63

to brake reaction time (46,47). Reactions to simple, well-defined

stimuli (i.e., a situation corresponding to an SSD case) have
been found to be very similar for older and younger subjects;
problems, however, have been observed in. situations with haz-
ards that have a high cognitive element. If a hazard is not easily
recognized as. such or the choice of an appropriate maneuver is
difficult, the older driver may take 2 or more seconds longer
to react.
There are several things to consider when interpreting data
concerning the performance of older drivers. One consideration
is older drivers' self-imposed limitations on their driving behav-
ior. Older drivers will avoid many conditions that they consider
hazardous or situations in which they feel uncomfortable or
confused when driving. For example, many older drivers avoid
peak-hour conditions. Some will not drive at night. Others will
not drive on higher-speed, limited-access facilities. These behav-
ior patterns should be considered by researchers in drawing
conclusions on the basis of raw accident frequencies or other
measures.

Intersection Design Issues

TRB Special Report 218, Transportation in an Aging Society

(48), submits that one of the more pressing roadway design
issues related to older drivers is the left turn off the mainline.
It is.acknowledged that older drivers generally have problems
selecting gaps in oncoming traffic and estimating the speeds of
approaching opposing vehicles. The report proposes the develop-
ment of formal warrants and greater use of dedicated left-turn
lanes and protected left-tum signal phasing. Figure 34. Paths of left-tuming vehicles turning in front of
Although there is little evidence that older drivers react more one another,
slowly than other motorists, they do take more time to make
decisions. This decision-making time can be even longer if the
older driver is confronted with a particularly complicated geo-
metric or traffic control situation.
There is also some concern that older drivers may have a tations concerning the turning paths of potentially conflicting
reduced range of head and neck movements. A simulator study vehicles and the potential for wrong-way movements.
of decision times of old and young drivers with restricted neck
movement by Hunter-Zaworski has confirmed this concern (49). Driver Expectations Concerning Turning Paths
Although the younger subjects with restricted movement could
compensate for their impairment, all older drivers, with and A key issue in selecting median widths and designing intersec-
without movement problems, had longer decision times. One tions on divided highways is whether the design of the intersec-
finding of this study was that skewed intersections are particu- tion leads drivers to believe that left turns in the median area
larly troublesome for all drivers with restricted movement. should be made in front of or behind an opposing left-tuming
In a recent project related to older driver perception-reaction vehicle (see alternative turning paths in Figures* 34 and 35).
time for ISD and object detection, experimental results by Lerner Little information is in the literature on driver choices between
et al. have shown no difference in the performance of older and "turn-in-front" and "turri-behind" maneuvers. Therefore, much
younger drivers in perception-reaction time measures related to of the information reported here is based on anecdotal inforina-
several aspects of highway geometric design (46). In fact, this tion heard during visits to the various state highway agencies
series of experiments confirms that the 2-sec perception-reaction and field observations made during the research.
time used for ISD Case III in the AASHTO Green Book is The "turn-in-front/turn-behind" issue is driven by several geo-
appropriate for older and younger driving populations. This find- metric and operational factors. In relatively narrow medians with
ing adds weight to the idea that most intersection accidents are left-turn. lanes or narrow median openings, there is seldom any
related to higher cognitive functions such as attention or confusion regarding this matter, and, except for rare instances,
multitask processing lapses. the left-tuming vehicles turn in front of each other (see Figure
34). When medians are very wide, to the point where the median
DRIVER EXPECTANCY ISSUES area roadway functions like a small street between the intersec-
tions, there is usually little confusion about which turning path
Two primary issues related to driver expectancy should be is used, so the left-turning vehicles turn behind each other (see
noted in designing divided highway intersections: driver expec- Fikure 35). Driver decision making at wide-median intersections
64

right instead, travel until they reach a median opening, and

then make a U-tum to proceed in the desired direction. Either
by intent or by mistake, drivers occasionally make a left
turn at the minor access point and travel the wrong way on
the near roadway of the divided highway.

I Crossing the median— Occasionally drivers are inattentive

or fall asleep and drive across the median without realizing
it. They sometimes drive for some distance unaware of what
has happened.
Transitions —Drivers entering a transition from an undi-
vided highway to a divided highway sometimes drive down
the lanes on the wrong side of the divided highway.
U-turns—Drivers may make a U-tum on one roadway of
a divided highway and return in the direction from which
they came; however, this is less common on divided high-
ways without access control than on freeways (where drivers
may be tempted to turn around if they miss an exit ramp).

The first two types of wrong-way movements occur at divided

I highway intersections and bear a direct relationship to the geo-

metric design and the traffic control devices at the intersection.
A 1969 California study (51 ) quantified the relative frequency
of these types of maneuvers (see Table 18) on divided highways
without full access control. The table shows that 47 percent of
wrong-way movements on divided highways originated at access
points with median openings (primarily intersections). Similar
Figure 35. Paths of left-turning vehicles turning results from a 1972 Virginia study (52) (shown in Table 19)
behind one another at a wide-median intersection. found that 45 percent of wrong-way movements on divided
highways originated at intersections. Thus, intersection design
and traffic control may play an important part in minimizing
wrong-way movements on divided highways.
can be aided by. adding pavement markings in the median area The literature concerning wrong-way movements on divided
(e.g., center lines and stop bars). highways (including studies in California, Indiana, Texas, and
Because substantial uncertainties existed about the ranges of Virginia) documents the following findings:
median width at which turn-in-front and turn-behind behavior
would be observed, the field observational studies addressed this Wrong-way accidents generally account for less than 1.5
issue. As described earlier in this chapter, it was found that turn- percent (and often less than I percent) of a state's total
in-front behavior predominates for intersections with medians accidents (51,52,53).
less than 15 in (50 ft) wide, and turn-behind behavior predomi- Wrong-way accidents generally account for between 1 and
nates for intersections with medians more than 15 in (50 ft) wide. 2 percent of a state's fatal accidents. The data show that
accidents involving wrong-way movements are generally
more severe than other accidents (50,51,52,53).
Wrong-Way Movements an Divided Highways Wrong-way movements are far more numerous than
wrong-way accidents (51,52,53).
Wrong-way movements are a definite cause of accidents on A disproportionate percentage of wrong-way movements
divided highways, including freeways and divided highways occur on Friday, Saturday, and Sunday (51,52,53). This
without full access control (expressways and arterials). Six types may be attributable to the greater number of intoxicated
of wrong-way movements can lead to accidents on divided high- drivers on the road on weekends (48). Several studies have
ways without full access control (50): shown that more than half of all wrong-way drivers have
been drinking (50,52,54). It should be noted, however, that
I . Left turn into the near lanes at an intersection —Drivers drivers are also more likely to be traveling on roads with
arriving at a divided highway on a minor-road approach may which they are not familiar on weekends than on weekdays.
turn left by mistake into the near roadway of the divided More than half of all wrong-way movements occur under
highway. restricted visibility conditions, especially at night (50,51).
Right turn into the far lanes at an intersection —Drivers Artificial lighting would probably reduce the likelihood of
arriving at a divided highway on a minor-road approach may wrong-way movements at most divided highway intersec-
turn right by mistake into the far roadway of the divided tions (50).
highway. The peak hour for accidents involving wrong-way move-
Left turn at an access point with no median opening—Many ments is between 2 and 3 a.m. (51). This is consistent with
minor access points on divided highways do not have median the hypothesis that drunk driving plays an important role
openings—this forces drivers who want to turn left to turn in wrong-way movements.
 65
TABLE 18. Origins of wrong-way movements that resulted in accidents on divided highways
without full access control in California (51)
Percentage of
No. of.wrong-way wrong-way
Point of origin movements movements
Intersection with median opening 67 47.2
Transition to divided highway 7 4.9
Drove through median opening 35 24.7
(no intersection)
Drove across median 4 2.8
Made U-turn 26 18.3
Other (driveways) 3 2.1
142 100.0
Origin unknown 25
167

TABLE 19. Origins of wrong-way movements that resulted in accidents on divided highways
without full access control in Virginia .(51)
Percentage of
No. of wrong-way wrong-way
Point of origin movements movements
Intersections 70 45.2
Interchanges 10 6.5
Commercial driveways 38 24.5
Private driveways 14 9.0
Crossovers 16 10.3
U-turns 7 4.5
155 100.0
Origin unknown 19
174

Accident-involved wrong-way drivers have previous acci- likelihood of wrong-way movements on divided highways were
dent rates and traffic violation rates that are more than developed in an Indiana study (50):
twice the average (51).
Accident-involved wrong-way drivers have slightly less At the intersection of a divided highway with an undivided
driving experience than the average driver (51). highway, the elevation of the undivided highway should be
Most wrong-way movements tend to originate from areas equal to or greater than that of the divided highway. This
of less-developed land use (50). gives the approaching motorist on the undivided highway a
Wrong-way movements take place when traffic volumes clearer view of both directions of travel. -
are low; i,e., when very little traffic is present to make the Whenever possible, angles of intersection other than 90 de-
proper travel directions clear (50). grees, as well as unusual intersection layouts, should be
Signing at most intersections where wrong-way move- avoided. Such layouts are frequently coriftising and may
ments take place is adequate. Thus, wr6ng-way movements encourage wrong-way movements.
seem to occur despite signing that meets the MUTCD re- At intersections where median storage is not required, medi-
quirements (50). ans should be narrow but distinct. Narrowing the median
Geometrics that are complex and hard to understand are a makes the far lanes of travel more visible, improving the
factor in wrong-way movements (50). driver's visibility of the overall intersection. Narrowing the
median also- reduces the amount of unchannelized space that
The following geometric design guidelines for reducing the the driver must negotiate. The median should be distinct to
66

aid the driver in understanding the intersection layout and in wrong-way movements and accidents. Very little evidence of
function. Distinctiveness can be achieved by raising and this problem was found in the data collected during the research
coloring the median. and, in fact, intersections with wider medians were found to be
the safest divided highway intersections in rural areas. Only one
wrong-way movement was seen in the field observational studies
This Indiana study suggests that wider medians are more and only one accident involving a wrong-way movement oc-
likely than narrow medians to be confusing to motorists, thus curred during the study period at the field study sites. Neverthe-
increasing the hkelihood of wrong-way movements (50). The less, because of the historical experience reported in the litera-
study did not, however, present any quantitative guidelines on ture, the concern about wrong-way movements persists. It is
median width to assist designers. appropriate for design guidelines to stress the need to ensure,
One of the issues that prompted this research was a concern through geometric design and traffic control, that the nature of
that divided highways with wider medians, particularly in rural divided highway intersections with wider medians is apparent
areas, could lead to driver confusion that could, in turn, result to drivers.
 67

CHAPTER 3

INTERPRETATION, APPRAISAL, APPLICATION

The findings presented in Chapter 2 illustrate the traffic opera- following discussion addresses roadways with raised or de-
tional and safety effects of median width and related geometric pressed medians but does not address roadways with flush
features at divided highway intersections. These findings provide medians.
a basis for selecting appropriate median widths and other design Table 21 summarizes the advantages of increasing the median
features for rural and suburban intersections on divided high- width on divided highways. The AASHTO Green Book states
ways. This chapter summarizes the key factors in selecting the clearly that some of the advantages of using medians can be
median width at divided highway intersections, the advantages obtained from medians of any width, even medians as narrow
and disadvantages of various median widths, and the feasible as 1.2 in (4 ft). In most cases, however, these advantages increase
left-turn treatments for various median widths. The chapter then as the median width increases. The desirability of providing
presents recommended guidelines for selecting the median width wider medians at intersections is highlighted by the list of disad-
at divided highway intersections and guidelines for other design vantages of narrow medians presented in Table 22. These disad-
features at such intersections. The chapter also discusses poten- vantages generally apply to raised medians that are less than 4
tial revisions to the AASHTO Green Book to implement the in (14 ft) wide (and are thus too narrow to include a left-turn
recommendations in this chapter. lane) and raised medians between 4 and 7 in (14 and 24 ft) wide
(which may include a left-turn lane but may not be wide enough
to store a passenger car with adequate clearances to through
KEY FACTORS IN SELECTING THE MEDIAN traffic). A minimum median width of 7.6 in (25 ft) is generally
WIDTH AT A DIVIDED HIGHWAY INTERSECTION
considered adequate to store the 5.8 in (19-ft) AASHTO passen-
ger car design vehicle with adequate clearances to through traf-
Table 20 summarizes these key factors. The research findings fic. Passenger car lengths have been decreasing—For the 1990
have shown that the area type (rural versus urban/suburban) is model year, the maximum length of passenger cars (including
a key factor in selecting the median width. The traffic volumes
minivans) produced in the United States was 5.6 in (18.4 ft),
and vehicle mix at the intersection —most particularly the cross-
which is approximately 0.3 in (I ft) less than the comparable
road traffic volumes and the volume of turns from the divided
maximum length of passenger cars in the 1978 model year. Many
highway —are controlling factors in all aspects of the design.
passenger cars now have lengths of only 4.6 to 4.9 in (15 to 16
The turning volumes and vehicle mix are critical elements in
ft) (56). It is recommended, however, that median width continue
determining the appropriate design vehicle to be considered, the
to be based on a design vehicle 5.8 m (19 ft) long unless
type of traffic control to be employed, the left-turn treatments
to be used, and the crossroad width and cross section. In turn,
AASHTO decides to reduce the length of the passenger car
the selected design vehicle, traffic control type, left-turn treat- design vehicle.
ment, and crossroad cross section define the other required inter- The minimum median width required to store a crossing or
section geometrics. left-tun-ling vehicle safely in the median is a key factor in select-
Sight distance is another factor that may enter into the selec- ing the median width at an intersection. Of .course, this minimum
tion of an appropriate median width. Chapter 2 of this report median width varies with the design vehicle selected. Thus,
has explained the influence of median width on sight distance medians widths from 8 to 24 in (26 to 80 ft) are in an intermedi-
requirements at divided highway intersections. Sight distance ate range that may allow storage of some design vehicles with
may also influence the choice of left-turn treatments because of adequate clearance to the through traffic lanes but not others.
the potential need to offset left-turn lanes to minimize or elimLi- As raised or depressed medians become wider, the disadvan-
nate sight obstructions caused by opposing left-turn vehicles. tages in Table 22 can be reduced or eliminated, but other disad-
vantages may develop. Table 23 summarizes the disadvantages
of wide medians. One common cause of problems is driver
ADVANTAGES AND DISADVANTAGES OF confusion about the intended operation and turning paths at
VARIOUS MEDIAN WIDTHS wide-median intersections. In addition, intersections with wide
medians do not lend themselves to signalization where the me-
Current design policies and the literature review have shown dian width exceeds 18 m (60 ft). Separate signalization on the
some clear' advantages and disadvantages of various median two roadways of the divided highway may be considered, but
widths; however, the median width ranges to which these advan- this approach (which is equivalent to the signalization used at
tages and disadvantages apply have not been quantified well a diamond interchange) typically involves more delay than a
enough to provide much specific guidance for designers. The single signal installation; this approach also requires careful at-
68
TABLE 20. Key factors to consider in selecting the median width at a divided highway
intersection
Area type:
rural.
urban/suburban.
Crossroad through and turning volumes and vehicle mix.
Volume and vehicle mix for turns from the divided highway.
Appropriate design vehicle for crossing and turning movements.
Type of traffic control to be employed:
signalized intersection.
unsignalized intersection that may be signalized in the future.
unsignalized intersection that is unlikely to be signalized in the
foreseeable future.
Crossroad width and cross-section (including travel lanes, shoulder, and
median [if any]).
Left-turn treatment to be utilized.
Needs for U-turns on the divided highway.

TABLE 21. Advantages of increasing the median width on divided highways

Interference from opposing traffic is minimized.

A feeling of greater freedom is provided.
An increased recovery area is provided for out-of-control vehicles.
More open green space is provided and a parklike setting can be created.
Headlight glare from opposing vehicles is reduced.
Flatter slopes can be used in the median.
The need for a median barrier may be avoided.
Additional width is available for future construction of additional lanes.
Additional width is available for median left-turn and U-turn storage lanes.
Offset left-turn lanes can be constructed to minimize sight restrictions caused
by opposing left-turn vehicles.
Additional width is available for median acceleration lanes where needed.
U-turns are feasible for larger vehicles.

tention to the storage requirements on the median roadway to lane but not wide enough to store a crossing or turning passenger
avoid overflowing of vehicles from the median roadway onto car with adequate clearance to through traffic
the through lanes of the divided highway. * 8 to 13 in (26 to 44 ft); wide enough to store a crossing
At divided highway intersections with very wide medians, or turning passenger car but not wide enough to store a school
where crossroad drivers approaching one roadway of the divided bus with adequate clearance to through traffic
highway cannot see the other roadway, drivers may not realize * 14 to 24 in (46 to 80 ft); wide enough to store a school
that the roadway isdivided and may turn in the wrong direction. bus or several passenger cars but not wide enough to store a
Even if the other roadway is in view across a wide median, large tractor-semitrailer truck with adequate clearance to
crossroad drivers may not see it, particularly at night or where through traffic
the far roadway is at a lower'elevation than the near roadway * More than 24 in (more than 80 ft); wide enough to store all
(50). Although no major problems of this type were found in current AASHTO design vehicles other than longer combination
the field and accident studies conducted as part of this research, vehicles (LCVs) but potentially confusing to some motorists.
care should be taken in both intersection design and signing to
minimi ze such problems. Tbus, the design vehicle selected is a key factor in determining
The following ranges of median width for raised or depressed median width requirements at divided highway intersections.
medians are appropriate for particular vehicle types: The ranges of median width identified above are based on all
current AASHTO design vehicles except LCVs (e.g., turnpike
* 1.2 to 4 in (4 to 12 ft); not wide enough to provide a left- doubles and triple-trailer trucks), which do not frequently use
turn lane nonfreeway divided highways and even less frequently make
* 5 to 7 rn (14 to 24 ft); wide enough to provide a left-tum turning or crossing maneuvers on such highways.
 W9

TABLE 22. Disadvantages of narrow medians at divided highway intersections

Crossing and turning maneuvers may require simultaneous gaps in traffic in both direction of
travel if the median is not wide enough to store a vehicle.
Vehicles may encroach on the through lanes while wafting in the median if drivers stop and
wait in a median that is not wide enough to store their vehicles.
If the median is just wide enough to store one vehicle, more than one driver at a time may be
tempted to use ft. The second vehicle may encroach upon the through lanes or the two
vehicles may waft side by side in the median, creating confusion about which vehicle has
the right of way and potential conflicts.
Narrow medians make it difficult for trucks to make U-turns.
Vehicles turning left from the divided highway are required to waft in the inside (higher-speed)
lane for a gap in opposing traffic when the median is too narrow for a left-turn lane to be
provided.
Provision of single or double left-turn' lanes may not be feasible if the median is too narrow.
Increased costs for widening the median will be incurred to provide left-turn lanes when they
prove necessary.
Narrow medians may require installation of a median barrier which is difficult to terminate
properly at intersections.
Narrow medians may not provide an adequate pedestrian refuge area [generally a 1.2- to
1.8-m (4- to 6-ft) median is required as a safe refuge for pedestrians].
Narrow medians may not provide a safe refuge area for bicyclists.

TABLE 23. Disadvantages of wide medians at divided highway intersections

Drivers may become confused abut the appropriate path through large paved area typically
present on the median roadway of a wide median intersection.
Drivers may ignore or fail to see traffic control devices in the median intended to control
movements onto or across the far roadway of the divided highway. Drivers may fail to
stop before entering the far roadway when they are required to do so.
Drivers approaching from the crossroad who fail to see the far roadway may turn in the wrong
direction because they do not realize that they are entering a divided highway.
Crossing or turning vehicles may overflow the available storage area.
Sight distance for vehicles turning left from the divided highway may be restricted by opposing
left-turn vehicles.
If the crossroad also has a wide median, a very large and confusing intersection may result.
Wider medians require longer clearance times and, thus, longer and less efficient signal
cycles if the intersection is signalized.
Wide medians make ft diff icult to place left-turn signal heads so that they are still visible if
left4urning vehicles move forward toward the center of the intersection.
Long clearance times at intersections with wide medians make ft possible for crossroad
vehicles to be trapped in the median when the major-road green phase begins.
If the median is wide enough, the intersection may have to be signalized as two closely
spaced intersections; closely spaced intersections are less efficient than a single
intersection and are difficult to signalize properly.
Signal phasing and clearance times at intersections with wide medians can be inappropriate
for bicyclists who attempt to use the vehicle timing.

FEASIBLE LEFT-TURN TREATMENTS FOR 24 and 25 present the feasible left-turn treatments for specific
VARIOUS MEDIAN WIDTHS
median widths for intersections on divided highways with raised
or depressed medians. Table 24 shows the feasible allocations
The left-tum treatment selected for a divided highway inter- of median width for single left-turn lanes; double left-turn lanes;
section may be a key factor in selecting the median width. Tables
70

TABLE 24. Feasible allocations of available width for various median widths and left-turn treatments
No LTL __—Sinlite Left—Turn Lane Double Left—Turn Lanes P—Hol Oftet L.11—Tur. Ln. Tapered at Left— urn no
SMadial S Medial So,. to,]
Offset 1* Thnr—~n ModT iIC:
are'o, .tt*
Median Median LTL Curb operator Separator Separator Curb eparato Opposing LTL Curb p oft
L "
pp~sm.
pposing
Width Width Width Oftet
in) Width ~ Currb
Cu
W.dthOftzt
L
L Width ! ~ u _"or
Width LTL
Width Offset (it)
Width LTL Width Width Oft t [
;Width
ift
"— 'a' LTL
(fit M) M)o (it) I
M)
4

NOT FEASIBLE NOT FEASIBLE NOT FEASIBLE

10 NOTFEASIBLE
14
1 10 2 2
10 4 2 .10 2 2 0
1: 4 *10 2 2. 2F 2 10 2 4 —2
5 10 2 2 4 4 10 2 4
20
22 Q 7 10 2 3 4 a 10 2 4 0
2
24 '10 . 2. 2 7 12 12 2 4 2
2 10
2: 11 2 4
30

0
38 0.8
4
Q0
44
4:6
R
4
5so
2
554

60

64 V

7
70 IN adzr"

KEY TO N : LIL - I.On—Turn Lane

®r . .. IBM

Not Feasible This table applies only to intersections with raised or depressed medians
LTLs could be used in narrower median widths it a flush median — used.
Marginal

Feasible

parallel, offset left-tum lanes; and tapered, offset left-tum lanes. tion (i.e., left off the major road only, left onto the major
Table 25 summarizes the ranges of median width for which road only, U-tum only).
these left-tum treatments are feasible, marginal, or not feasible For parallel, offset left-tum lanes, the medial separator be-
for intersections with raised or depressed medians. tween the left-turn lane and the adjacent through lane (in
Tables 24 and 25 are based on the following assumptions: the same direction of travel) has a desirable width of 1.2 in
(4 ft) or more. A minimum separator width of 0.6 in (2 ft)
1 . In accordance with the Green Book, the minimum median was used where necessary in narrow medians, but such
width considered is 1.2 in (4 ft). designs were classified as marginal in Tables 24 and 25.
The maximum median width tabulated is 26 in (84 ft), which Parallel, offset left-tum lanes with lane widths of less than
is the largest desirable width for a median on a rural divided 3.6 in (12 ft) or offsets to the opposing left-tum lane of less
highway indicated by any of the highway agencies that re- than 1.2 in (4 ft) were also classified as marginal.
sponded to the survey conducted during the research. Parallel, offset left-tum lanes were aligned to have an offset
Left-turn lanes had a desirable width of 3.6 in (12 ft), plus of +1.2 in (+4 ft) or more, whenever possible, on the basis
a 0.6-m (2-ft) curb offset. Minimum left-tum lane widths of the offset criteria defined in Appendix F.
of 3.1 in (10 ft), plus a 0.6-m (2-ft) curb offset, were used Tapered, offset left-turn lanes were aligned to intersect the
where necessary in narrow medians, but such designs were median roadway with a 1.2-m (4-ft) separator from the op-
classified as marginal in Tables 24 and 25. Where there is posing through lanes. Thus, the angle of the tapered left-
a curb on 'one side of a left-turn lane, a curb offset of 0.6 turn lanes varies as a function of median width. The widths
in (2 ft) is assumed. Where there is a curb on both sides of of the medial and through-lane separators shown in Table 24
a left-turn lane, a total curb offset of 0.6 in (2 ft) is assumed; are measured along the median roadway; these dimensions
it could be 0.3 m (1 ft) on each side of the left-turn lane or change as one moves upstream, away from the median road-
0.6 in (2 ft) on the median side and no offset on the side way. Tapered, offset left-tum lanes with lane widths of less
toward the through lanes (assuming a mountable curb is than 3.6 m (12 ft) or offsets to the opposing left-tum lane
used on that side). of less than 1.2 m (4 ft) were classified as marginal in
At locations where the median is wide enough to accommo- Tables 24 and 25.
date a double left-turn lane, it is also wide enough to accom- Each specific left-turn treatment is discussed ,later in this
modate a restricted median opening at an unsignalized loca- chapter.
 71
TABLE 24. Feasible allocations of available width for various median widths and left-turn treatments (cont.)

Not Feasible This table applies only to intersections with raised or depressed medians
LTLs could be used in narromr median vvidths if a flush median "re used.
Marginal

Feasible

GUIDELINES FOR SELECTING MEDIAN WIDTHS found between median width and accident rate or undesirable
AT DIVIDED HIGHWAY INTERSECTIONS driving behavior. Although this indicates that traffic operational
or safety problems are much less likely at three-leg than at four-
The following section presents recommended guidelines for leg intersections, it is recommended that the design guidelines
selecting median widths at divided highway intersections. The based on the findings concerriing rural, four-leg intersections
design policy presented in the AASHTO Green Book specifies also be applied to rural, three-leg intersections.
a minimum width of 1.2 m (4 ft) for highway medians. Most The following guidelines are recommended for selecting the
medians are wider than this, and the guidelines presented below median width at rural, unsignalized intersections on divided
explain the factors to be considered in selecting the median highways:
width for divided highway intersections.

1 . At rural, unsignalized intersections on divided highways,

Rural Intersections wider medians generally operate more safely than narrow
medians. On the basis of traffic operational and safety
The key finding presented in Chapter 2 of this report concern- considerations alone, the median at each rural intersection
ing rural intersections is that accident rates at rural, four-leg, should be as wide as possible. In selecting median widths,
unsignalized intersections decrease as the median width in- however, highway agencies must also consider right of
creases. This finding is supported by the results of the field way and construction costs, the potential for driver confu-
observational studies, which found that the frequency of undesir- sion and wrong-way movements, the desirability of uni-
able driving behavior at rural, four-leg, unsignalized intersec- form design for particular corridors'and within particular
tions also decreases as the median width increases. Rural, three- roadway functional classifications, and the potential for
leg, unsignalized intersections were found to have very low traffic volume growth and development.
accident rates, and no statistically significant relationships were 2. Two key factors in selecting the median width at rural
72

TABLE 25. Feasible left-turn treatments for various median widths

intersections are the appropriate design vehicle for turning from the composition and characteristics of traffic on the
and crossing maneuvers at the intersection and the type of major road. The demand for U-turns is another factor in
left-turn treatment to be implemented. determining the design vehicle; if a divided highway has
An important step in designing a divided highway intersec- driveways used by larger vehicles at which median open-
tion is to select an appropriate design vehicle for turning ings are not provided, larger vehicles find it necessary to
and crossing in~neuvers at the intersection. This design make U-turns at other intersections or median openings to
vehicle normally should be deteriiiined from consideration go to or from those driveways. Thus, at some intersections,
of the current and design year traffic composition and char- the design vehicle for turning and crossing maneuvers may
acteristics of the crossroad, which may be quite different be based on U-tum demands rather than crossroad traffic
 73

volume or mix. The median width requirements for differ- Median opening lengths at rural divided highway intersec-
ent left-turn treatments have been addressed in the previous tions generally should be kept to the minimum possible.
section. This issue is addressed further in the subsequent section
The minimum median width at intersections on a rural on design and marking the median roadway.
divided highway generally should be 8 in (25 ft). This
median width allows storage of the 6-m (19-ft) AASHTO In summary, the research results tend to confirm the practice
passenger car design vehicle with 0.9 in (3 ft) of clearance of most states to provide rural medians with widths in the range
to through traffic at each end of the vehicle. Medians less of 14 to 18 in (46 to 60 ft). Such medians are wide enough that
than 8 in (25 ft) wide generally should be used in rural one school bus or two passenger cars making a turning or cross-
areas only where wider medians are infeasible or, at ex- ing maneuver can be stored on the median roadway. Medians
isting intersections, where narrower medians are already narrower than 14 in (46 ft) are adequate to store a passenger car
in place and are operating without major traffic operational (or two smaller passenger cars) but may not be adequate to store
or safety problems. the largest school bus. Medians wider than 18 in (60 ft) provide
At many rural divided highway intersections, the crossroad very safe operations at rural intersections and may be particularly
carries only minimal heavy truck traffic. At such locations, appropriate at locations with substantial volumes of large trucks
many highway agencies have chosen the school bus as the making turning and crossing maneuvers; however, highway
appropriate design vehicle. The maximum length of school agencies should consider limiting the use of wider medians at
buses in use is approximately 12 in (40 ft). With 0.9-m intersections that may require signalization or may undergo sub-
(3-ft) clearance to the through lanes at each end of the urban development in the foreseeable future.
vehicle, a 14-m (46-ft) median width is appropriate for As stated above, the choice of an appropriate design vehicle
storing a school bus. For this reason, many highway agen- for turning and crossing maneuvers on the basis of crossroad
cies use divided highway median widths in the range of and U-tum traffic mix is a key step in selecting the appropriate
14 to 15 in (46 to 50 ft). median width for a divided highway intersection. Site-specific
5. Where a large truck is used as the design vehicle, a median traffic analyses may lead, however, to the choice of different
21 to 31 in (70 to 100 ft) wide generally should be selected. design vehicles and, therefore, different minimum median
r-1 When a median width greater than 24 to 31 m (80 to 100 widths at intersections along a given divided highway corridor.
ft) is used at a divided highway intersection, designers Consistency in median width along such corridors is highly
should recognize the potential for wrong-way movements desirable because consistency in design improves driver expec-
if crossroad drivers fail to see both roadways of the divided tancy of geometric features. Thus, the practice of using a consist-
highway and to recognize the divided highway as such. ent median width along a corridor, on the basis of the largest
Signing should also be effective in helping crossroad driv- design vehicle for turning and crossing maneuvers for any of
ers recognize the major road as a divided highway. A the intersections in the corridor, should be encouraged. Even if
subsequent section of this chapter addresses methods to this practice results in medians that are wider than required at
minimize wrong-way movements. some intersections, it is not generally a problem because accident
The leading cause of undesirable driving behavior at unsig- rates have been found to decrease with increasing median widths.
nalized intersections is the competition for limited space The only locations at which this corridor-based approach to
in the median roadway by vehicles making turning and median width selection may not be desirable are intersections
crossing maneuvers. As turning and crossing volumes in- that are considered likely to require signalization. or undergo
crease, more space on the median roadway is needed and suburban development in the future.
wider medians may be appropriate. Field observations have found that vehicles making opposing
Although traffic operational and safety research indicates left-turns tend to turn in front of one another if the median is
that rural, unsignalized intersections generally should have less than 15 m (50 ft) wide and tend to turn behind one another
medians as wide as possible, designers generally should if the median is more than 15 in (50 ft) wide. It may be undesir-
avoid wide medians if the location is likely to undergo able to mix intersections with median widths that are less than
suburban development or to be signalized within the fore- and more than 15 in (50 ft) within the same corridor, because
seeable future. As discussed below, wider medians lead to driver behavior at such intersections differs.
higher accident rates and undesirable driving behavior at Consistency can also be achieved by using a standard median
suburban, unsignalized intersections. Thus, designers width on a statewide basis for all rural divided highways. This
should avoid creating intersections that might develop traf- may improve driver expectancy of median widths even more
fic operational or safety problems if they become more than adopting a consistent median width for a specific corridor.
developed. In addition, highway agencies have found that Each intersection, however, should be reviewed in the design
intersections with wider medians are difficult to signalize process to determine if a change in the standard median width
properly. The design guidelines for suburban intersections, would be appropriate: Medians that are wider than the standard
presented below, may be applicable to such cases. may be appropriate at some intersections with large volumes of
Increased ISD is required at divided highway intersections turning or crossing truck traffic, and medians that are narrower
where the median is not wide enough to store a turning or than the standard may be appropriate where future signalization
crossing vehicle. If the intersection is wide enough to store or suburban development is expected.
a turning or crossing vehicle, the sight distance require- Some highway agencies have started to implement a new
ments at the intersection of the median roadway with the practice of widening their standard 14 to 18 in (46 to 60 ft)
far roadway of the divided highway are the same as at any median to 46 m (150 ft) or more at selected higher-volume
other intersection. intersections. This practice is new and largely untested but may
74

contribute to safety because rural intersections with medians through vehicles and vehicles leaving the median roadway
more than 31 in (100 ft) wide are known to operate very safely. to cross the through lanes. Appendix F presents a method
Because this approach results in varying median width along for computing the offset required between opposing left-
a corridor, methods of changing driver expectancy by calling turn lanes and the minimum left-turn lane length needed to
particular attention to the intersections with wider medians are minimize such problems. Parallel or tapered offset left-turn
desirable. For example, extensive use of advance guide signs lanes can be used at unsignalized intersections to provide
for the intersections with wider medians might provide a cue to increased sight distance, although such treatments are less
approaching drivers that these wider median intersections are common than at signalized intersections; offset left-turn
somehow different from the rest. Particular attention also needs lanes generally increase the required median width at the
to be paid to signing for crossroad drivers to help ensure that intersection.
they recognize the major road as a divided highway; such signing Medians with widths of more than 15 in (50 ft) generally
is critical to discourage wrong-way movements. should be avoided at suburban, unsignalized intersections,
both because they could be expected to have higher accident
rates and because driver confusion could result from oppos-
Suburban Unsignalized Intersections ing left-turn drivers turning behind one another, rather than
turning in front of one another as they do at most other
The key research finding concerning suburban, unsignalized suburban intersections.
intersections is that the frequency of accidents and undesirable
driving behavior increases as the median width increases. Medi- In summary, medians at suburban, unsignalized intersections
ans clearly add to the safety and efficiency of traffic operations should generally be narrower than those used in rural areas
on suburban highways; however, the research finding stated because narrower medians generally result in safer operations
above implies that, where medians are provided on suburban at suburban intersections and because narrower medians win
highways, the median widths at unsignalized intersections result in more efficient operations if the intersection requires
should be kept as narrow as possible. The following design signalization at a later date. The minimum median width in
guidelines are applicable to suburban unsignalized. intersections: suburban areas generally should be set by the width requirements
of the left-turn treatment to be used. Increased ISD may be
I . Median widths at suburban unsignalized intersections genei- required if the median is not wide enough to store a turning or
ally should be as narrow as possible while providing suffi- crossing vehicle. Consistency in median width along a divided
cient space in the median for the appropriate left-turn treat- highway corridor is desirable; achieving consistency may require
ment. Most suburban intersections have left-turn lanes, considering appropriate median widths for both signalized and
which generally require minimum widths of 4 to 5 in (14 unsignalized. intersections.
to 18 ft) for raised medians.
Many highway agencies prefer to use medians between 4.2
and 7.2 in (14 and 24 ft) wide on suburban highways because Suburban Signalized Intersections
they are wide enough to provide left-turn lanes but are not
wide enough to store a crossing or turning vehicle in the The research findings reported in Chapter 2 indicate that the
median. Median widths in this range require through and accident rate at suburban, signalized intersections increases as
left-turn vehicles on the crossroad approach to wait for si- the median width increases. In addition, the increased clearance
multaneous gaps in both directions of travel on the major times required by wider medians increase traffic delays at the
road before completing their turning or crossing maneuver. signalized intersections and can, therefore, reduce the level of
This philosophy is consistent with keeping the median width service. The following design guidelines are applicable to subur-
as narrow as possible. It should be recognized, however, ban, signalized intersections:
that the sight distance requirements for turning and crossing
vehicles are higher for divided highway intersections of this Median widths at suburban, signalized intersections gener-
type than for intersections on undivided highways. ally should be as narrow as possible while providing suffi-
Medians wider than 7.6 in (25 ft) may be used at intersec- cient space in the median for the appropriate left-tum treat-
tions on suburban highways, but designers should anticipate ment. A variety of left-turn treatments are potentially
that crossroad vehicles making tuniing and crossing maneu- applicable At signalized intersections, including conven-
vers will stop on the median roadway, if necessary. The tional left-turn lanes, double left-turn lanes, and parallel and
length of the design vehicle(s) to be accommodated and the tapered offset left-turn lanes. The median width require-
demand for storage space on the median roadway should be ments of these left-turn treatments have been presented ear-
considered in selecting the median width. Designers should lier in this chapter.
avoid making the median wider than necessary, because Suburban signalized intersections can generally operate ef-
accidents and undesirable driving behavior increase as me- fectively with median widths of less than 7.6 in (25 ft)
dian width increases and because, if the intersection requires because there should be no need for crossroad traffic to stop
signalization in the future, total delay will increase as me- on the median roadway except during the crossroad green
dian width increases (see next section of this chapter). phase. It is undesirable for turning or crossing vehicles from
Some unsignalized intersections have experienced safety a crossroad approach to remain in the median during the
problems because vehicles waiting to turn left from the major-road green phase; clearance intervals are generally
major road block the view of opposing left-turn or crossroad provided in the signal cycle so that crossroad vehicles are
vehicles, which can lead to collisions between major-road not trapped in the median after the signal turns red. Medians
 75

wider than 7.6 in (25 ft) are not generally recommended at studies, however, suggest methods that can be used to minimize
suburban, signalized intersections unless required for the such undesirable driving behavior.
selected left-turn treatment. The field study results indicate that the undesirable driving
Signalized intersections with median widths of more than behavior is often related to the competition for limited space
18 in (60 ft) are difficult to signalize effectively so that on the median roadway between vehicles traveling through the
vehicles are not trapped in the median at the end of the median in the same direction. Thus, undesirable driving behavior
crossroad green phase. Some highway agencies have found is likely to increase as the median volume increases. Traffic
it necessary to install detectors on the median roadway to volumes are generally not high enough at most rural intersections
delay the beginning of the major-road green phase until all to lead to operational problems, but such problems have been
crossroad vehicles have cleared the median roadway. observed at particular sites. Traffic volumes using the median
When intersections with median widths of greater than 31 roadway are likely to be higher at suburban intersections. Where
in (100 ft) are signalized, it may be necessary to provide the median volume is high and the available storage space on
separate signals at each intersection of the median roadway the median roadway is inadequate, several alternatives should
with the through lanes of the divided highway. This type be considered, including the following:
of signalization, which is equivalent to the signal system
used at a diamond interchange, often involves greater delay * Increasing the median width or median opening length to
than the use of a single signal; however, the total delay to provide more storage space. The choice between increasing the
through vehicles decreases as the median width increases median width and the median opening length is addressed below.
and, at some relatively large median width, eventually be- - Reducing the median width to 7.2 in (24 ft) or less so that
comes less than the delay at a single signal for that same there is not enough room in the median to store a vehicle, and
median width. Diamond-type signal systems also require crossing and turning vehicles from the crossroad approaches
careful attention to the storage requirements on the median must wait for simultaneous gaps. This alternative is generally
roadway, which may be exceeded if the traffic volumes are more applicable to suburban intersections than to rural intersec-
too high or the median width is too small. Wide medians tions, where wider medians are desirable for safety.
that require diamond-type signalization generally should be e Closing the median opening and permitting only right turns
avoided but may be necessary in some circumstances such in and right turns out of the crossroad.
as staged freeway construction in which frontage roads are * Reducing the median volume through turn prohibitions or
built first and the freeway is constructed between them at channelization to eliminate particular traffic movements through
a later date. the median. For example, Figure 36 shows how channelization
Some signalized intersections have experienced safety prob- can be used to permit left turns off a divided highway while
lems because vehicles waiting to turn left from the major preventing crossing maneuvers and left turns onto the divided
road block the view of opposing left-tum or crossroad vehi- highway.
cles, which can lead to collisions between major-road * Signalizing the intersection to separate conflicting traffic
through vehicles and vehicles leaving the median roadway movements in time. If the median width is narrow enough to
to cross the through lanes. Appendix F presents a method permit the installation of a single signal, then vehicles from the
for computing the offset required between opposing left- crossroad approaches should not need to be stored in the median
turn lanes and the minimum left-tum lane length needed to during the major-road green phase; however, signalization
minimize such problems. The required offset can be pro- should be considered only where the MUTCD signal warrants
vided by either parallel or tapered offset left-turn lanes, are met and where signalization can be accomplished without
which generally increase the required median width at the interfering with signal progression on the major road.
intersection.
One of the accepted principles of intersection channelization
In summary, the median widths at suburban, signalized inter- (10) is that large, uninterrupted expanses of pavement within
sections should be as narrow as possible but are largely deter- an intersection can be confusing to drivers and should be
rnined by the type of left-tum treatment to be used. Left-turn avoided. The presence of a large expanse of pavement makes it
treatments anticipated for future use (e.g., double left-tum lanes) difficult for drivers to decide what path to follow and to antici-
should also be considered in determining the median width. pate what path other drivers will follow. It is generally better
Consistency in median width along a divided highway corridor for both traffic operations and safety to channelize an intersec-
is desirable; achieving consistency may require consideration of tion and guide drivers along specific paths. For this reason, it
appropriate median widths for both signalized and unsignalized is not desirable to provide additional storage space on the median
intersections. roadway by increasing both the median width and median open-
ing length, as that would create a large unrestricted pavement
area like that discussed above. It is better to increase the median
DESIGN AND MARKING OF THE MEDIAN width or increase the median opening length but not both. The
ROADWAY TO MINIMIZE UNDESIRABLE DRIVING research findings indicate that at rural, unsignalized intersec-
BEHAVIOR tions, it is better to increase the median width and keep the
median opening length to a minimum consistent with the width
Several types of undesirable driving behavior, including side- of the crossroad roadway and shoulders. At suburban, unsignal-
by-side queuing, angle stopping, and encroaching on the through ized intersections, the research findings indicate that it is gener-
lanes of the major road, were observed in field studies at divided ally more desirable to keep the median width to a minimum and
highway intersections. The results of the field observational increase the median opening length; at many higher-volume
76

Figure36 Median channelization to prevent movements through the median 'except left turns off the divided
highway.

intersections, an increased median opening length will occur At unsignalized intersections with median widths of approxi-
naturally because of multiple lanes and/or wider shoulders on mately 18 in (60 ft) or less, dashed pavement markings that
the crossroad, but median openings should not be made unneces- extend the left edge line of the divided highway across the
sarily long because this may result in undesirable side-by-side intersection can also be helpful to drivers. Such markings, illus-
queuing. trated in Figure 38, define the boundaries of the median roadway
Pavement markings can be very useful in helping to define for drivers and should help to minimize encroachment on the
vehicle paths at divided highway intersections. At rural, unsig- through lanes of the divided highway by vehicles stopped on
nalized intersections with median widths of more than 31 in (100 the median roadway. This practice has been used satisfactorily
ft), it is highly desirable to provide a double yellow centerline by at least two of the state highway agencies that participated
separating the two directions of travel on the median roadway. in the study.
This type of marking, illustrated in Figure 37, provides visual
continuity with the centerline of the crossro"ad approaches and
helps to define a desired path for drivers. For example, by creat- DESIGN GUIDELINES FOR OTHER GEOMETRIC
AND TRAFFIC CONTROL FEATURES OF
ing a relatively narrow marked path through the median road- INTERSECTIONS
way, the presence of a double yellow centerline on the median
roadway should minimize the temptation for drivers to queue This section presents design guidelines for left7tum lanes,
side-by-side or to cut over to the left side of the median roadway offset left-tum lanes, indirect left-turn lanes, U-tum treatments,
and stop at an angle when making a left turn. The three sites median acceleration lanes, traffic control on the median road-
evaluated in the field studies, with medians more than 31 in way, wrong-way movements, and pedestrian considerations.
(100 ft) wide and with markings of this type, had among the
lowest rates of accidents and undesirable driving behavior of
any of the intersections studied. Left-Turn Lanes
The AASHTO Green Book generally recommends a mini-
mum median opening length of 12 in (40 ft). In fact, two of the The need for left-turn lanes is one factor that influences the
intersections studied with median widths of more than 31 in choice of median width. Tables 24 and 25 show the ranges of
(100 ft) had median opening lengths in the range of 30 to 40 ft median width for which conventional single and double left-tum
(9 to 12 in) and operated very safely. If the minimum median lanes are feasible.
opening length recommended by the Green Book is used at such Two alternative policies for installation of left-turn lanes on
a wide-median intersection, the use of edge lines to delineate a divided highways have been used by highway agencies:
portion of the median roadway as a shoulder might be consid-
ered, to define the path for vehicles across the median to be as * Provide left-tum lanes where specific warrants, such as
much like a normal lane as possible. those presented in Figure 6, indicate they are needed
 77

>1
C cts

0
cc

Figure 37 Recommended marking of the.median roadway at a wide-median intersection

with a two-lane crossroad.

* Provide left-turn lanes at all median openings where left left-turn lane design as superior. Advance signing and pavement
turns are permitted markings at the entrance to the left-turn lane can assist drivers
in recognizing offset left-turn lanes. Lighting of intersections
The first alternative is consistent with the Green Book and with offset left-tum lanes should be considered, whenever possi-
has, historically, been the policy of most highway agencies; ble, to assist left-tuming drivers to recognize the proper path
however, several highway agencies have adopted the second at night.
policy in recent years, at least in designing rural divided
highways.
Current practice in designing left-turn lanes at divided high- Indirect Left-Turn Lanes
way intersections is presented in Figure 5.
Indirect left-turn lanes like those shown in Figure 10 have an
important role in enabling drivers to make left turns efficiently
Offset Left-Turn Lanes on divided highways with relatively narrow medians. Two state
highway agencies—the Michigan and New Jersey Departments
Parallel and tapered offset left-turn laiies like those shown in of Transportation— have used indirect left-turn. treatments exten-
Figure 8 are still not common but are being used increasingly sively and drivers in those states understand and expect them.
to reduce the risk of accidents caused by the sight obstructions Other highway. agencies have used indirect left-turn treatments
from opposing vehicles turning left. The survey results in Ap- less frequently. An important advantage of indirect left-turn lanes
pendix A indicate that 62 percent of state highway agencies and is that they may allow a divided highway to operate with a
42 percent of local highway agencies have used offset left-turn narrower median than would be required if left-turn lanes were
lanes. There are no national design guidelines for offset- left- placed in the median.
turn lanes. Parallel offset left-tum lanes are mentioned only
briefly in the Green Book, and tapered offset left-turn lanes are
not mentioned at all. Tables 24 and 25 show the ranges of U-Turn Treatments
median width for which parallel and tapered offset left-turn lanes
are feasible. The current Green Book policy on median width requirements
There are no generally accepted warrants for offset left-turn at U-tum median openings has been presented in Figure 11 in
lanes and no guidelines for choosing between the parallel and Chapter 2 of this report. This policy shows the median width
tapered designs. The criteria for installing offset left-tum lanes requirements for U-turfis from the inner lane (leftmost travel
used by one state highway agency are presented in Chapter 2, lane) in one direction of travel to the inner lane, outer lane, and
following the discussion of Figure 8. The research performed in shoulder in the opposite direction of travel. This ' same policy
this study found that most drivers use offset left-tum lanes prop~ could be applied to determine the median width requirements
erly, although a few problems were observed. The research, for U-turns at intersections. Because many intersections have
however, does not provide a basis for selecting any particular left-turn lanes within the median area, it may be appropriate to
78

Figure 38. Exten&ng edgelines across the inedian to better defitne the boundaries of the inedian roadway.

include, as well, the median width requirements for U-tums from acceleration lanes is that they allow vehicles turning left onto a
a median left-turn lane to the inner lane, outer lane, and shoulder divided highway to continue through the median roadway with-
in the opposing direction of travel. out stopping and merge onto the far roadway of the divided
highway. That allows drivers to cross the near roadway of the
divided highway without considering the availability of gaps on
Median Acceleration Lanes the far roadway, even if the median is not wide enough to store
their vehicles; however, even if a median acceleration lane is
Median acceleration lanes, like offset left-turn lanes, have provided, left-turning drivers must anticipate potential conflicts
come into use at divided highway intersections in recent years with other vehicles in the median area. Thus, the presence of a
but have been used by only a few agencies. A typical median median acceleration lane changes the decision making process
acceleration lane is shown in Figure 9. The advantage of median and the maneuvers made by crossroad drivers turning left onto
 79

the divided highway, but not by drivers crossing the divided because driver errors of this type can lead to severe accidents.
highway. The median acceleration lane should be long enough Wrong-way movements are a particular concern at intersections
to allow left-tuming vehicles to reach the speed of major-road with wider medians, particularly when the median is more than
traffic before merging. 24 to 31 in (80 to 100 ft) wide, because it may not be apparent
to all crossroad drivers that the major road is a divided highway.
Traffic Control on the Median Roadway The following design and traffic control guidelines are
applicable:
Highway agencies currently use Stop signs, Yield signs, and
no control at the intersection of the median roadway and the far Provide signing to help drivers on the crossroad approaches
roadway of a divided highway. At unsignalized intersections recognize that left turns into the near roadway of the divided
with medians less than 9 in (30 ft) wide, highway agencies quite highway are not permitted. This may include One-Way
consistently use no control on the median roadway, as specified signs, No Left Turn. signs, Wrong-Way signs, and Divided
in the MUTCD. For unsignalized intersections with median Highway plaques.
widths of 26 in (85 ft) or more, highway agencies are quite Whenever possible, drivers on the crossroad approaches
consistent in using Stop signs at the intersection of the median should be able to see both roadways of the divided highway.
roadway and the far roadway of a divided highway. For intersec- Clearing sight obstructions in the median that limit the view
tions with median widths between 9 and 26 in (30 and 85 ft), of the far roadway should be considered. The view of the
Stop signs, Yield signs, and no control are all used by highway traffic control device at the far end of the median roadway
agencies. Even within a given state, more than one type of (Stop sign or Yield sign) may provide an important cue that
control was observed with this intermediate range of median the driver has reached a divided highway intersection.
widths. Where the crossroad approaches the divided highway on a
Field observations found that less than 50 percent of drivers downgrade, the driver is afforded the best view of the entire
complied fully (i.e., made a complete stop) with Stop signs on intersection. Where the crossroad approaches the divided
the median roadway. The remaining drivers either made a rolling highway on an upgrade, drivers may have a poor view of the
stop or did not stop at all. On the other hand, no traffic conflicts intersection, and special care in designing other geometric
resulted from vehicles violating the Stop signs, and most of the features and signing may be needed.
drivers could have stopped if a conflicting vehicle were present. Whenever possible, divided highway intersections with
In short, most drivers treated the Stop sign as if it were a Yield wider medians should be lighted at night to assist drivers
sign. The use of Stop signs in this situation may be detrimental in seeing both sides of the major road and recognizing it as
because traffic control devices are not desirable in situations in a divided highway.
which they are ignored. Whenever possible, angles of intersection other than 90 de-
The field observational study found no discernable differences grees, as well as unusual intersection layouts, should be
in driver behavior at intersections with Yield signs or no control. avoided.
On the basis of the field studies conducted in the research, the
following traffic control policy for divided highway intersections
appears to be appropriate: Pedestrian Considerations

For intersections with median widths of less than 9 in (30

ft), use no control on the median roadway as specified in The following design guidelines, based on the work of
the MUTCD. - Knoblauch et al. (55) are recommended for pedestrian facilities
For intersections with median widths between 9 and 26 in at divided highway intersections:
(30 and 85 ft), use either Yield signs or no control on the
median roadway. Either Yield signs or no control should be
used consistently to help in developing driver expectancy. Basic Criteria
Stop signs should be used only if an intersection develops
a record of accidents involving vehicles crossing or turning 1 . Crosswalks should not be marked where crossing the street
into the far roadway of the divided highway. may be unusually hazardous (e.g., locations with high traffic
For intersections with median widths of 26 in (85 ft) or speeds, poor sight distance, or poor illumination; locations
more, Stop signs should be used on the median roadway. with inadequate median refuge space; or locations with in-
sufficient pedestrian crossing time in the traffic signal
Some traffic control device is definitely needed at such wide- cycle).
median intersections to help drivers perceive the presence of the In light of the installation and maintenance costs of pave-
far roadway of the divided highway. Stop signs are recom- ment markings, crosswalk markings should be located at
mended because they are used quite universally for this apphca- places expected to receive sufficient benefit. This suggests
tion; however, there is reason to believe that Yield signs, used that crosswalks with low vehicular volume and/or low pe-
consistently, could function equally well. destrian volume do not warrant markings. Determining min-
imurn pedestrian and vehicle volume thresholds is an impor-
Wrong-Way Movements tant part of establishing reasonable guidelines for crosswalk
markings.
Careful attention should be given to minimizing the potential Guidelines for installing crosswalks should include the type
for wrong-way movements at divided highway intersections, of pedestrians expected to be crossing the street. Lower-
80

200—
Basic Criteria

Speed limit 72 km/h (45 mph).

Adequate, stopping sight distance.
For midblock, preferred block length ~ 180 m (600 ft)
~4-lane without median or - Crosswalk adequately illuminated.
8-lane with median - Minimal conflicting attention demands.
150 -
4)
E
Locations with predominantly young, elderly or
> handicapped pedestrians
C 0 2.1ane, 3-lane, or
20 4- to 6-lane with median Othe r locations
100 —
o

Z' Install Crosswalk

%
0
X
50— Do Not N%
%%
Install Crosswalk % %

25—

lor-
——————————————————————
00
L 12,000 ' 14,000
2,000 4,600 '00
6,0 8,0
'00 10,000

Average Daily Traff ic Volume

If using only the peak hour, threshold must be increased by 1.5.

For streets with median, use one-way (directional) ADT volume.

Other notes: Minimum striping is 150-mm (6-in) lines. Consider bolder markings and/or supplementary
advance markings or signing at uncontrolled locations where speed limits exceed 55 km/h (35 mph).

Figure 39. Guidelines for crosswalk installation at uncontrolled intersection legs, midblock crossings, and signalized
intersections without pedestrian heads (55).

volume thresholds should be considered for areas where Installation Guidelines

there is a greater proportion of less experienced and less
agile pedestrians (e.g., near schools and/or elderly housing Crosswalk markings should be installed at the following:
areas).
4. Crosswalk markings in higher-risk crossing areas (higher All signalized intersections with pedestrian signal heads
traffic volumes and speeds) should be supplemented by All locations where a school crossing guard is normally
warning signs at the crossing and before the crossing, and stationed to assist children in crossing the street
in some cases, warning pavement markings. * All intersections and midblock crossings satisfying the min-
61 Crosswalks should be used selectively. Allowing a prolifera- imum vehicular and pedestrian volume criteria in Figure 39(As
tion of crosswalks reduces the overall effectiveness of each long as the basic criteria governing sight distance, speed limit,
crosswalk. and so forth are met, a crosswalk is deemed appropriate if the
6. Specific variables that should be considered when locating pedestrian and vehicular volumes place it above the appropriate
crosswalks include activities located nearby (e.g., schools, curve in Figure 39. Each crosswalk is analyzed by approach leg,
shopping), pedestrian volume, vehicular volume, sight dis- indicating that a crosswalk might be warranted on one side of
tance, vehicular speeds, street width, presence and width the intersection and not the other. Tbus, the guidelines might
of median reftige area, comer radii and turning roadway suggest that only one crosswalk needs to be marked at a given
channelization, one-way versus two-way operation, and the intersection. If each approach warranted a crosswalk, then all
feasibility of providing the necessary pedestrian crossing would be marked.)
time in the traffic signal cycle. * All other locations where there is a need to clarify the
 81
preferred crossing location when the proper location for crossing G. The recommended revisions include modifications to Chapter
would otherwise be confusing. IV (Cross Section Elements), Chapter VII (Rural and Urban
Arterials), and Chapter IX (At-Grade Intersections). These rec-
RECOMMENDED REVISIONS TO THE AASHTO
ommendations will be considered by the AASHTO Task Force
GREEN BOOK on Geometric Design, which is responsible for determining what
changes to the Green Book are appropriate.
Recommended revisions to the AASHTO Green Book to im-
plement the findings of this research are presented in Appendix
82

CHAPTER 4

CONCLUSIONS AND RECOMMENDATIONS

The major conclusions of the research are as follows: ble, on the grounds of operational efficiency. Where dia-
mond-type signalization must be used, however, the
At rural, unsignalized intersections, the frequency of both vehicle delay for any given traffic volume level decreases,
accidents and undesirable driving behavior (e.g., side-by- and the space available for storing'vehicles on the median
side queuing, angle stopping, and encroaching on the roadway increases as the median width increases. Thus, if
through lanes of the major road) decreases as the median a median wide enough to require diamond-type signaliza-
width increases. tion is necessary, increasing the median width further
At suburban, unsignalized intersections, the frequency of should improve the level of service.
both accidents and undesirable driving behavior increases Divided highway intersections with medians that are not
as the median width increases. wide enough to store a vehicle safely in the median require
At suburban, signalized intersections, the frequency of ac- more ISD for crossing and left-turn maneuvers than similar
cidents increases as the median width increases. intersections on undivided highways. Where vehicles can
The frequency of undesirable driving behavior increases stop safely in the median, however, the sight distance re-
as median opening length increases at rural intersections quirements of divided highway intersections are similar to
and decreases as median opening length increases at subur- undivided highway intersections.
ban intersections. Tapered and parallel offset left-turn lanes can be used at
Most undesirable driving behavior at divided highway in- divided highway intersections to ' minin-Lize or eliminate
tersections arises from the competition for limited space sight distance obstructions caused by opposing left-turn
on the median roadway between drivers traveling through vehicles. No major traffic operational or safety problems
the median in the same direction. were found at three signalized intersections with offset
Compliance with Stop signs at the intersection between left-tum lanes that were evaluated in the research.
the median roadway and the far roadway of a divided
highway is poor. Less than 50 percent of drivers make a
The following recommendations have been developed on the
fuH stop, but very few traffic conflicts between median
basis of these conclusions:
roadway and major-road vehicles result. In fact, drivers
tend to operate similarly at the intersection between the
median roadway and the far roadway of a divided high- At rural, unsignalized intersections on divided highways,
way—regardless of the type of traffic control used—Stop medians should generally be as wide as practical and cer-
sign, Yield sign, or no control. tainly should be wide enough to accommodate turning and
Vehicle delay at signalized intersections on divided high- crossing maneuvers by a selected design vehicle, as well as
ways increases as the median width increases. Thus, in- any needed left-turn treatments. In most cases, the appro-
creasing the median width at a signalized intersection re- priate design vehicle for rural, unsignalized intersections is
duces the level of service. This finding is based on traffic a large school bus or a large truck. Whenever possible, the
simulation analyses for traffic volumes of up to 1,000 median opening length should be limited to the crossroad
veh/hr on each major-road approach and up to 800 veh/hr pavement width plus shoulders, to better define the turning
on each crossroad approach; left- and right-turn percent- paths and avoid making the paved area in the median too
ages of up to 30 percent on each major-road approach and large.
of up to 40 percent on each crossroad approach; and for At suburban, unsignalized intersections, medians generally
median widths of 6 to 31 m (20 to 100 ft). should not be wider than necessary to provide whatever
Signalized intersections on divided highways with medians left-turn treatment is selected. Wider medians at suburban,
wider than 31 in (100 ft) may require separate signal instal- unsignahzed intersections are associated with increased ac-
lations on the two roadways of the divided highway. This cident frequency. At specific intersections where substantial
type of signalization operates similarly to the signalization turning and crossing volumes of large vehicles (such as
at a diamond interchange. Using separate signal installa- school buses or trucks) are present, highway agencies may
tions substantially increases vehicle delay, which substan- find it appropriate to select an appropriate median width to
tially reduces level of service, in comparison with a single store a design vehicle of that type safely in the median.
signal installation serving the same approach volumes and At signalized intersections, medians generally should not be
turning movement patterns. Therefore, signalized intersec- wider than necessary to provide whatever left-turn treatment
tions with wide medians should be avoided, where possi- is selected. Wider medians at signalized intersections are
 83
associated with increased accident frequency and in- For intersections with median widths of less than 9 in
creased delay. (30 ft), use no control on the median roadway as speci-
4. Highway agencies should consider limiting median widths fied in the MUTCD.
at rural, unsignalized intersections that are likely to require For intersections with median widths between 9 and
signalization or undergo suburban development in the fore- 26 in (30 and 85 ft), use either Yield control or no
seeable ftiture. Wider medians should operate well at a rural, control on the median roadway. Stop signs should be
unsignalized intersection but may operate poorly if the inter- used only.if an intersection develops a record of acci-
section becomes signalized and/or undergoes development. dents involving vehicles crossing or turning into the
5. At rural, unsignalized intersections with median widths of far roadway of the divided highway.
more than 31 in (100 ft), the use of a double yellow center- For intersections with median widths of 26 in (85 ft)
line to separate the two directions of travel on the median or more, Stop signs should be used on the median
roadway is recommended. roadway.
6. At unsignalized intersections with median widths of approx- 8. Particular care should be taken in designing and operating
imately 18 in (60 ft) or less, dashed pavement markings that wide-median intersections to ensure that the intersection is
extend the left edgeline of the divided highway across the properly signed to discourage improper left turns into the
intersection can be helpful to drivers in defining the bound- near roadway of the divided highway and that, whenever
possible, a driver on the crossroad approach to a divided
aries of the median roadway. Markings of this type should
highway intersection can see the far roadway of the divided
help to minimize encroachment on the through lanes of
highway. Both signing and visibility of the far roadway
the divided highway by vehicles stopped on the median
help to discourage wrong-way movements that can lead
roadway.
to accidents. Such intersections should be lighted at night,
7. The following recommendations are applicable to traffic whenever possible. Wrong-way movements were not found
control within unsignalized divided highway intersections in this research to be a major problem, but the potential for
at the intersection of the median roadway with the far road- accidents involving wrong-way movements always exists at
way of the divided highway: divided highway intersections.
84

REFERENCES

A Policy on Geometric Design of Highways and Streets. Highway Grade Intersections." Highway Research News,
American Association of State Highway and Transportation No. 13, HRB, National Research Council, Washington, D.C.
Officials, Washington, D.C.(1994). (June 1964).
A Policy on Geometric Design of Highways and Streets. Hurd, F.W. "Accident Experience with Traversable Medians
American Association of State Highway and Transportation of Different Widths." Bulletin 137, HRB, National Research
Officials, Washington, D.C. (1990). Council, Washington, D.C. (1956).
TRB Special Report 209: Highway Capacity Manual, 3rd ed. Moskowitz, K., and Schaefer, W.E. "California Median
TRB, National Research Council, Washington, D.C. (1994). Study: 1958." Bulletin 266, HRB, National Research Coun-
Wallace, C.E., Chang, E.C.P., Messer, C.J., and Courage, cil, Washington, D.C. (1960).
K.G. PASSER 11-90 Users Guide. Texas A&M University, Hutchinson, JW., and Kennedy, T.W. "Safety Considera-
College Station (December 1991). tions in Median Design." In Highway Research Record 162,
Fambro, D.B., Chaudhary, N.A., Bonneson, J.A., Messer, HRB, National Research Council, Washington, D.C. (1967).
C.J., and Arabie, L.J. PASSER 111-90 User Manual. Texas Garner, G., and Deen, R.C. "Elements of Median Design in
A&M University, College Station (March 1991). Relation to Accident Occurrence." In Highway Research
Lee, C.E., Rious, T.W., and Copeland, C.R. The Texas Record 432, HRB, National Research Council, Washington,
Model for Intersection Tralffic -Development. Center for D.C. (1973).
Highway Research, Austin, TX (December 1977). Foody, T.J., and Culp, T.B. "A Comparison of the Safety
Lee, C.E., Machemehl, R.B., Rious, T.W., and Inman, R.F. Potential of the Raised versus Depressed Median Design."
Texas Model Version Documentation. Center for Transpor- In Transportation Research Record 514, TRB, National Re-
tation Research, Bureau of Engineering Research, Univer- search Council, Washington, D.C. (1974).
sity of Texas (August 1991). Knuiman, M.W., Council, F.M., and Reinfurt, D.W. "The
Roadside Design Guide. American Association of State Effect of Median Width on Highway Accident Rates." Pre-
Highway and Transportation Officials, Washington, D.C. sented at the 72nd Annual Meeting of the Transportation
(1989). Research Board, Washington, D.C. (January 1993).
Bonneson, J.A., McCoy, P.T., and Truby, J.E. "Safety Im- Van Maren, P.A. "Correlation of Design and Control Char-
provements at Intersections on Rural Expressways: A Sur- acteristics with Accidents at Rural Multilane Highway Inter-
vey of State Departments of Transportation." In Transporta- sections in Indiana." Report No. FHWAIIN-77120, Joint
tion Research Record 1385, TRB, National Research Highway Research Project, Purdue University, West Lafa-
Council, Washington, D.C. (1993). yette, IN (December 1977) (revised July 1980).
Neuman, T.R. NCHRP Report 279: Intersection Channel- Radwan, A.E., Sinha, K.C., and Michael, H.L. "Develop-
ization Design Guide. TRB, National Research Council, ment and Use of a Computer Simulation Model for the
Washington, D.C. (November 1985). Evaluation of Design and Control Alternatives for Intersec-
Harmelink, M.D. "Volume Warrants for Left-Tum Storage tions of Minor Roads with Multilane Rural Highways: Se-
Lanes at Unsignalized Grade Intersections." In Highway lection of the Simulation Model." Report No. FHWA-IN-
Research Record 211, HRB, National Research Council, 79-8, Joint Highway Research Project, Purdue University,
Washington, D.C. (1967). West Lafayette, IN (July 1979).
Effectiveness of Median Storage and Acceleration Lanesfor Radwan, A.E., Sinha, K.C., and Michael, H.L. "Develop-
Left-Turning Vehicles. Institute of Transportation Engineers, ment and Use of a Computer Simulation Model for the
Informational Report (1986). Evaluation of Design and Control Alternatives for Intersec-
Traffic Engineering Evaluation: U.S. 60 Intersections, Mesa tions of Minor Roads with Multilane Rural Highways: Field
to Apache Junction, MP 189.0-1960. Arizona Department Studies and Model Validation." Report No. FHWAIIN-7919,
of Transportation, Traffic Engineering Section, Traffic Joint Highway Research Project, Purdue University, West
Studies Branch (undated). Lafayette, IN (July 1979).
1991 Accident Data on California State Highways (Road Radwan, A.E., Sinha, K.C., and Michael, H.L. "Develop-
Miles, Travel, Accidents, Accident Rates). California De- ment and Use of a Computer Simulation Model for the
partment of Transportation (1992). Evaluation of Design and Control Alternatives for Intersec-
Harwood, D.W., Mason, J.M., and Graham, J.L. Conceptual tions of Minor Roads with Multilane Rural Highways:
Planfor an Interactive Highway Design Model. Draft report Model Application." Report No. FHWAIIN-79110, Joint
prepared for Contract No. DTFH61-91-C-00091, Midwest Highway Research Project, Purdue University, West Lafa-
Research Institute, Kansas City, MO (June 1992). yette, IN (July 1979).
McDonald, J.W. "Relation Between Number of Accidents Manual on Uniform Traffic Control Devices for Streets and
and Traffic Volume at Divided-Highway Intersections." Highways. Federal Highway Administration, Washington,
Bulletin 74, HRB, National Research Council, Washington, D.C. (1988).
D.C. (1953). Manual on Uniform Traffic Control Devicesfor Streets and
Priest, R.V. "Statistical Relationships Between Traffic Vol- Highways. Federal Highway Administration, Washington,
ume, Median Width, and Accident Frequency on Divided D.C. (1978).
 85

Brigha, P.M. "4-Way Stop Signs Cut Accident Rate 58 Shinar, D., and Scheiber, F. "Visual Requirements for Safety
Percent at Rural Intersections." ITE Journal (November and Mobility of Older Drivers." Human Factors (1991).
1984). Salthouse, T.A. "Speed of Behavior and Its Implications for
Agent, K.R. Traffic Control and Accidents at Rural, High- Cognition." In Handbook of the Psychology of Aging (J.E.
Speed Intersections. Kentucky Transportation Research Pro- Binen and K.W. Schaie, eds.), Van Nostrand Reinhold, NY
gram, University of Kentucky, Lexington (March 1987). (1985).
Eck, R.W., and Sabra, Z.A. "Active Advance Warning Signs Lemer, N., et al. "Older Driver Perception-Reaction Time
at High-Speed Signalized Intersections: A Survey of Prac- for Intersection Sight Distance and Object Detection." Re-
tice." In Transportation Research Record 1010, TRB, Na- port No. FHWA-RD-93-168, Federal Highway Administra-
tional Research Council, Washington, D.C. (1985). tion, Washington, D.C. (January 1995).
Foody, T.J., and Taylor, W.C. "An Analysis of Flashing Lerner, N. "Giving the Older Driver Enough Perception-
Systems." In Highway Research Record 221, HRB, National Reaction Time." Experimental Aging Research, Vol. 20
Research Council, Washington, D.C. (1968). (1994).
Goldblatt, R.B. "Effect of Flashing Beacons on Intersection TRB Special Report 218: Transportation in an Aging Soci-
Performance." In Transportation Research Record 644, ety. TRB, National Research Council, Washington, D.C.
TRB, National Research Council, Washington, D.C. (1977). (1988).
McCoy, P.T., Navarro, UR., and Witt, W.E. "Guidelines for Hunter-Zaworski, K.M. "T-Intersection Simulator Perform-
Offsetting Opposing Left-Turn Lanes on Four-Lane Divided ance of Drivers with Physical Limitations." In Transporta-
Roadways." In Transportation Research Record 1356, TRB, tion Research Record 1281, TRB, National Research Coun-
National Research Council, Washington, D.C. (1992). cil, Washington, D.C. (1990).
Wickens, C.D. Engineering Psychology and Human Per- Scifres, P.N. "Wrong-Way Movements on Divided High-
formance. Charles E. Merrill Publishing Company, Colum- ways." Report No. JHRP-74-3, Joint Highway Research
bus, OH (1984). Project, Purdue University, West Lafayette, IN (February
Boff, K.R., and Lincoln, J.E. Engineering Data Compen- 1974).
dium: Human Perception and Performance. AAMRL, Tamburri, T.N. "Wrong-Way Driving Accidents Are Re-
Wright-Patterson AFB, OH (1988). duced." In Highway Research Record 292, HRB, National
Pambro, D.B., Fitzpatrick, K., Griffin, L.I., Kahl, K.B., Research Council, Washington, D.C. (1969).
Koppa, R.J., and Petzold, V.J. "Determination of Stopping Mills, J.P. "Wrong-Way Driving in Virginia." Paper pre-
Sight Distances." Interim Reportfor NCHRP Project 3-42, sented at the 1972 Summer Meeting of the AASHO Op-
erating Subcommittee on Traffic Engineering (cited in Ref-
Texas Transportation Institute, College Station, TX (Decem-
erence 50).
ber 1992).
Messer, C.J., Friebele, J.D., and Dudek, C.L. "A Quantita-
Olson, P.L., Cleveland, D.E., Fancher, P.S., Kostyniuk, L.P.,
tive Analysis of Wrong-Way Driving in Texas." Research
and Schneider, L.W. NCHRP Report 270: Parameters Af-
Report 139-6, Texas Transportation Institute, College Sta-
fecting Stopping Sight Distance. TRB, National Research
tion, TX (May 1971) (cited in Reference 50).
Council, Washington, D.C. (1984).
Estep, A.C. "Wrong-Way Driving on California Freeways
Hall, JW., and Turner, D.S. "Stopping Sight Distance: Can 1961-1972." Paper presented at the 1972 Summer Meeting
We See Where We Now Stand?" In Transportation Re-
of the AASHO Operating Subcominittee on Traffic Engi-
search Record 1208, TRB, National Research Council, neering (cited in Reference 50).
Washington, D.C. (1989). Knoblauch, R.L., Tustin, B., Smith, S.A., and Pietrucha,
Rackoff, N.J., and RockweU, T.H. "Driver Search and Scan M.T. Investigation of Exposure Based Pedestrian Accident
Patterns in Night Driving." In TRB Special Report 156: Areas. Federal Highway Administration, Washington, D.C.
Driver Visual Needs in Night Driving, TRB, National Re- (November 1986).
search Council, Washington, D.C. (1975). Parking Dimensions of 1990 Model Passenger Cars. Motor
Staphn, L., Lococo, K., and Sim, J. "Traffic Control Design Vehicle Manufacturers Association, Washington, DC
Elements to Accommodate Drivers With Diminished Capa- (1990).
bility." Report FHWA-RD-901055, Federal Highway Ad- SASISTAT User's Guide, Version 6, 4th ed. SAS Institute,
ministration, Washington, D.C. (May 1990). Inc., Gary, NC (1990).
Dewar, R., Templar, J., and Knoblauch, R. "An Analysis McShane, W.R., and Roess, R.P. Traffic Engineering. Pren-
of the Older Pedestrian's Task and Related Motor, Sensory, tice Hall, Englewood Cliffs, NJ (1990).
Perceptual, Cognitive, and Behavioral Factors." Task A re- Joshua, S.C., and Saka, A.A. "Mitigation of Sight-Distance
port for the study Older Pedestrian Characteristics for Use Problem for Unprotected Left-Tuniing Traffic at Intersec-
in Highway Design, Center for Applied Research, Great tions." In Transportation Research Record 1356, TRB, Na-
Falls, VA (1992). tional Research Council, Washington, D.C. (1992).
86

APPENDIXES A-F

Appendixes A through F as submitted by the research agency APPENDIX C Evaluation of Accident Histories at Field
are not published herein but are available for loan on request to Study Sites
the NCHRP. APPENDIX D Evaluation of Statewide Accident Data for.
Divided Highway Intersections
APPENDIX E Effect of Median Width on Traffic
APPENDIX A Summary of Questionnaire Responses from Operations at Signalized Intersections
State and Local Highway Agencies APPENDIX F Sight Distance Implications of Offsetting
APPENDIX B Field Observational Studies at Divided Left-Turn Lanes at Divided Highway
Highway Intersections Intersections
 87

APPENDIX G
RECOMMENDED REVISIONS TO THE AASHTO GREEN BOOK

This appendix presents recommended revisions to the design difnensien in balanee with ethef: eefopettents ef the ef:ess
policies in the AASHTO publication, A Policy on Geometric seefien.
Design of Highways and Streets, known as the AASHTO Green In determining median width, consideration should be
Book(1). The Green Book is the primary geometric design guide given to the possible need for median barrier. Where possible,
used by many highway agencies and other geometric design median widths should be such that a median barrier is not
practitioners. warranted. In gefter-al, the Fnedian should be as wide as pr-a
The recommended revisions are based on the findings of this eal-. The general range of median widths is from a minimum
research and, other than a few minor editorial suggestions, do of 1.2 in to 24 in or more. Economic factors often limit the
not address issues other than median width at intersections on width of median that can be provided. Cost of construction
divided highways. Sections of the Green Book potentially af- and maintenance increases in proportion to increases in the
fected by the research findings include portions of Chapter IV median width but the additional cost may not be appreciable
(Cross Section Elements), Chapter VII (Rural and Urban Arteri- compared with the cost of the highway as a whole and may
als), and Chapter IX (At-Grade Intersections). The remainder of be justified in view of the benefits derived.
the appendix presents the current text of the Green Book with At unsignalized intersections on divided highways in
recommended changes. Deletions of text are shown as sw~keeuis rural areas, the median should generally be as wide as
and additions to the text are shown in bold. practical. However, narrower medians appear to operate
The Green Book text presented in this appendix is based on better at unsignalized intersections in urban and subur-
the 1994 edition of the Green Book that, for the first time, ban areas; therefore, wider medians should be avoided
incorporates units in the Sl or metric system. in urban and suburban areas except where necessary to
accommodate turning and crossing maneuvers by larger
vehicles [reference this report]. Medians at unsignalized
GREEN BOOK CHAPTER IV (Cross Section intersections should be wide enough to allow selected design
Elements) vehicles to safely make a selected maneuver. The appro-
priate design vehicle should be chosen based on the actual
The following text shows the recommended revisions to the or anticipated vehicle mix of crossroad and U-turn traffic.
section on medians that appears in pages 367-369 in Chapter A consideration in the use of wider medians on roadways
IV of the 1994 Green Book. The recommended changes are other than freeways is the provision of adequate shelter for
intended primarily to incorporate the research findings that wider vehicles crossing at unsignalized intersections and at cross-
medians are desirable at rural intersections but not at urban overs serving commercial and private drives. These median
or suburban intersections or at intersections that may require crossings may need to be controlled as at-grade intersections
signalization. (See Chapter IX). Wide medians may be a disadvantage when
at gfade inter-seetiens afe signalization is required. The in-
creased time for vehicles to cross the median may can lead
151 13) 11 Ell ~ to inefficient signal operation.
If right-of-way is restricted, a wide median may not be
A median is highly desirable on arterials carrying four or justified if provided at the expense of narrowed border areas.
more lanes. A median is defined as the portion of a divided A reasonable border width is required to adequately serve as
highway separating the traveled way for traffic in opposing a buffer between the private development along the road and
directions. The median width is expressed as the dimension the traveled way, particularly where zoning is limited or
between the through-lane edges and includes the left shoul- nonexistent. Space must be provided on the borders for side-
ders, if any. The principal functions of a median are to sepa- walks, highway signs, utility lines, parking, drainage chan-
rate opposing traffic, provide a recovery area for out-of- nels and structures, and for proper slopes and any retained
control vehicles, provide a stopping area in case of emergen- native growth. Narrowing these areas may tend to develop
cies, allow space for speed changes and storage of left-turning obstacles and hindrances similar to those that the median is
and U-turning vehicles, minimize headlight glare, and pro- designed to avoid.
vide width for future lanes. Another benefit of a median in A depressed median is generally preferred on freeways for
an urban area is that it may offer an open green space. For more efficient drainage and snow removal. Median side
maximum efficiency, a median should be highly visible both slopes should preferably be 1:6, but slopes of 1:4 may be
night and day and contrast with the though traffic lanes. adequate. All drainage inlets in the median should at least
Medians may be depressed, raised, or flush with the pavement be designed with the top flush with the ground or culvert
surface. Medians should be as wide as fe- Asible but of an ends provided with traversable safety grates.
88

Raised medians have application on arterial streets where findings on median widths at rural, unsignalized intersections
it is desirable to regulate left-turn movements. They are also along with caveats for intersections that may become signalized
frequently used where the median is to be planted, particularly or may undergo suburban development.
where the width is relatively narrow. Careful consideration
of the location and type of plantings must be given. Plantings,
particularly in narrow medians, may create problems for main- Medians
tenance activities. Plantings, such as trees, in the median can
also cause visual obstructions for turning motorists if not On highways without at grade intersections, the median
carefully located. may be as narrow as 1.2 to 1.8 in under very restricted
Flush medians are commonly used te seme extent on eA conditions but a width ef 20 in er mer-e wider medians
types e urban arterials. When used on freeways, a median should be provided wherever feasible. A wide median allows
barrier may be required. The median should be slightly the use of independent profiles. Reduced frequency of cross-
crowned or depressed for drainage. In warmer climates the over accidents and reduction of headlight glare are safety
crowned type is frequently used because it eliminates the need features associated with a wide median.
for collecting drainage water in the median. However, the When intersections are to be provided, special concern
slightly depressed type is generally preferred with a cross must be given to the width of the median. Research has
slope of about 4 percent or with a minor steepening of the shown that most types of undesirable driving behavior
roadway cross slope. on the median roadway of divided highway intersections
The concept of converting flush medians into two-way are associated with competition for space on the median
left turn lanes on urban streets has become widely accepted. roadway by vehicles traveling in the same direction
This concept offers several advantages when compared to through the median [reference this report]. The potential
no median. Among the advantages are reduced travel time, for such problems is lindted where crossroad and U-turn
improved capacity, reduced accident frequency, particularly volumes are low but may increase at higher volumes.
of the rear-end type, more flexibility because the median lane Types of undesirable driving behavior observed include
can be used as a travel lane during closure of a through side-by-side queuing, angle stopping, and encroaching on
lane and public preference, both from drivers and owners of the through lanes of the divided highway. At rural unsig-
abutting properties. Median widths of 3.0 to 4.8 in wide pro- nalized intersections, the frequency of undesirable driving
vide the optimum design for two-way left turn lanes. Refer behavior and the frequency of accidents were observed
to the MUTCD (2) for appropriate lane markings. Refer to to decrease as the median width increased; that implies
Chapter H for additional discussion and details. that the median width should be as wide as practical. It
Where there is no fixed-source lighting, headlight glare was also observed that the frequency of undesirable driv-
across medians or outer separations can be a nuisance; particu- ing behavior increased as the median opening length
larly where the highway has relatively sharp curves. Under increased.
these conditions, some form of antiglare treatment should be While medians as narrow as 1.2 to 1.8 in may be fequir-ed
considered as part of the median barrier installation, provided necessary under very restricted conditions, medians 3.6 to 9
it does not act as a snow fence and create drifting problems. in wide provide protection for left-tuming vehicles at inter-
Insofar as through traffic is concerned, a desired ease sections. A.4-F-A&A—a M.A.41414A ffAfR 9 tO 15 A; hAllid 11A PM:Ahl
and freedom of operation, in the sense of physical and psycho-
logical separation from opposing traffic, is obtained when
medians are about 72 in or wider. With such widths the facility
truly is divided. The noise and air pressure of opposing traffic
is not noticeable, and at night the glare of headlights is greatly
reduced. With widths of 18 in or more the median can be
pleasingly landscaped in a parklike manner. Plantings used to
achieve this parklike appearance should need not compromise
the roadside recovery area. it faust be pointed etti, heweveF,
dtat There is demonstrated benefit in any separation, raised
or AlAPA flush. Wider medians are desirable at rural un-
signalized intersections, but medians as wide as 18 in may
not be desirable at urban and suburban intersections or
at intersections that are signalized or may require signal-
ization in the foreseeable future [reference this report].
gfade inter-Seefiens. In many cases, the median width at a
rural unsignalized intersection will be a function of the
GREEN BOOK CHAPTER V11 (Rural and Urban design vehicle selected for turning and crossing maneu-
Arterlals) vers. Where a median width of 7.6 m or more is provided,
a passenger car making a turning or crossing maneuver
The following text shows the recommended revisions to the will have space to stop safely in the median area. Medians
discussion of medians in the section on rural arterials on pages less than 7.6 in wide should be avoided at rural intersec-
498 and 499 in Chapter VII of the 1994 Green Book. The tions because drivers may be tempted to stop in the me-
primary intent of these changes is to incorporate the research dian with part of their vehicles unprotected from through
 89

traffic. Many highway agencies use the school bus as the than barriers are discussed in Chapter IV. Where right-of-
design vehicle to determine the median width at rural way is limited, it is frequently necessary to decide how best
divided highway intersections; the school bus is often the to allocate the available space between border areas, traffic
largest vehicle to use the median roadway frequently. The lanes, and medians. On the less important arterials, the deci-
selection of a school bus as the design vehicle results in a sion is often resolved in favor of no median at all. A median
median width of 14 to 15 m. Larger design vehicles, in- only 1.2 in wide is better than none; however, each additional
cluding trucks, may be used at intersections where enough feet meter of median width provides an added increment
turning or crossing trucks are present; median widths of of safety and improved operation between intersections. At
23 m or more may be needed to accommodate large trac- intersections in urban and suburban areas, median widths
tor-trailer trucks without encroaching on the through should be limited, whenever possible, to those required
lanes of the major road. to provide appropriate left-turn treatments for current
There was concern that median widths in the range of and future traffic volumes. At intersections where left turns
15 to 24 m at divided highway intersections could cause are made, a left-turn lane is always desirable from a-capacity
some drivers to become confused. Recent research has and safety standpoint. The median width to accommodate
not found any evidence of such confusion at rural inter- left-tuniing movements should be at least 3.6 in. Desirably,
sections [reference this report]. However, an intersection the median should be at least 5.4 in wide for a 3.6-in median
with a wider median may become confusing to some driv- lane and a 1.8-m medial separator. At restricted locations, a
ers if the median is so wide that a driver on the crossroad 3.0-m lane with a 0.6-in medial separator may be used. Other
approach cannot see the far roadway of the divided high- left-turn treatments, including double left-turn lanes and
way. Such designs should be avoided and, where neces- parallel or tapered offset left-turn lanes, may be needed
sary, signing should be used to discourage wrong-way at some intersections to serve current or future traffic
movements. volumes.
Median widths of more than 18 m are undesirable at Figure VII-5 shows various configurations for medians
intersections that are signalized or may require signaliza- that may be used on urban arterials. The type of treatment
tion in the foreseeable future. The efficiency of signal usually depends on local practice and available right-of-way
operations decreases as the median width increases, be- widths. The type selected should always be compatible with
cause drivers require more time to traverse the median drainage and street hardware requirements.
and special detectors may be needed to avoid trapping Median openings on roadways provided with depressed or
drivers in the median at the end of the green phase for raised medians should be carefully considered when a divided
traffic movements that pass through the median. Further- arterial is planned. Openings should be provided only for
more, if the median is so wide that separate signals are street intersections or for major developed areas. Spacing
required on the two roadways of the divided highway, between median openings must be adequate to allow for
delays to motorists will increase substantially and careful introduction of left-turn lanes and signal detection loops to
attention must be given to vehicle storage requirements operate without false calls.
on the median roadway between the two signals. Where intersections are widely spaced, e.g., 1.0 kin or
The subsequent section concerning median widths at more, the median width may be varied by using a narrow
intersections on urban arterials indicates that wider medi- width between intersections where necessary for economy
ans may increase accidents and lead to undesirable driv- and gradually widening on the approach to the intersection
ing behavior. Therefore, highway agencies may wish to to accommodate the left-turn lane. This solution is hardly
limit the use of wider medians at rural intersections that practical, however, and should never be used where intersec-
are likely to undergo urban or suburban development in tions are closely spaced because the curved alinement of the
the foreseeable future. lane lines results in excessive lane changes. It is far more
Research also indicates that undesirable driving behav- desirable that the median be of uniform width.
ior at rural unsignalized intersections increases as the A street with an odd number of lanes, perhaps three or
median opening length increases [reference this report]. five, may be used to advantage in providing a storage lane
The median opening length should be equal to at least for left-tuming vehicles. This lane is one form of utilizing a
the minimum described in Chapter IX, but median open- paved, flush median. Left-turn bays are marked in advance
ings at rural unsignalized intersections should not be un- of the intersections. The length of lane between left-turn bays
necessarily long. may be used for storage of vehicles making midblock left
See Chapter IV for information on median design. turns if this usage does not adversely affect traffic flow and
safety on the arterial; otherwise, midblock left turns may
The following text presents the recommended changes to the be prohibited. Under some conditions it is better to permit
section concerning medians on urban arterials on pages 516 midblock turns than to require that vehicles make U-tums at
through 520 in Chapter VII of the 1994 Green Book. The pri- intersections or travel around a block to reach a destination.
mary intent of these revisions is to discourage the use of wider In some cases the center lane is designated for "Left-Turn
medians at intersections on urban and suburban arterials. Only" throughout, without specially marked bays at intersec-
tions. This type of operation works well where the speed on
Medians the arterial highway is relatively low (40 to 70 kni/h) and
there are no heavy concentrations of left-turn traffic. Figure
Medians are a desirable feature of arterial streets and VII-8 is an example of a continuous left-turn lane. Refer
should be provided where space permits. Medians and me- to Chapter IV, section "Medians" and Chapter IX, section
90

"Continuous Left-Turn Lanes" for further discussions of this unsignalized intersections should not be wider than neces-
type of median. sary. This trend is opposite to that observed at rural
Where an arterial must pass through a developed area hav- unsignalized intersections.
ing numerous street and driveway intersections, and where The median at an urban or suburban unsignalized in-
it is impractical to limit left turns, the continuous left-turn tersection should be wide enough to accommodate the
median lane is often the only practical solution. Because all left-turn treatment selected by the designer. As at rural
left turns are thus protected, the interference to through lanes intersections, the appropriate design vehicle for turning
is minimized. Successful operation of a continuous left-turn and crossing maneuvers, based on the vehicle mix for
lane requires adequate lane marking. crossroad and U-turn traffic, should be considered in de-
A raised median with barTier-type curbing may be used termining the median width. At urban and suburban un-
on low and intermediate speed arterial streets. This type is signalized intersections, medians less than 25 ft wide can
desirable where it is necessary to prevent midblock turns. On be used effectively and discourage drivers from stopping
streets serving low-speed traffic this type will prevent most in the median, and discourage multiple vehicles traveling
cross-median accidents. Raised medians also provide a refuge in the same direction from using the median at the same
for pedestrians and a good location for signs, signals, and time. At locations with substantial crossing and turning
other appurtenances. In snow-belt areas the curbed median volumes of larger vehicles, such as school buses or trucks,
provides positive delineation, whereas a flush median be- it may be appropriate to provide enough width to store
comes indiscernible under the lightest of snowfall conditions. such vehicles in the median without encroaching on the
However, raised medians present disadvantages that should through lanes of the major road.
be considered in deciding the type to be employed. Urban and suburban unsignalized intersections with
On streets serving high-speed traffic the raised median median widths from 9 7.6 to 15 m generally experience
does not prevent cross-median accidents unless a median slightly higher accident rates than intersections with nar-
barrier is provided. If accidentally struck, raised medians may rower medians but appear to should be eaFefolly eensider-ed
cause the driver to lose control of the vehicle. They are &BIR &H OpeffifiefieA viewpeim. These widths de net alwwfs
difficult to see at night except when used in conjunction pfeyide fnedian ster-age .'ehieles er-esging
with a good system of fixed-source lighting or with proper QhP an. AAse, these widths may eneeufage !he dfivef: to
delineation. They cast a shadow from oncoming headlights affempt the er-essings independently and lew.'P. fhP. A.'PhiPle ROt
that not only makes the curb difficult to see, but also places fully pfeteeted fFAFA thP thfeugh tFaffle. These widths, eyea
most of the adjacent lane in shadow. Raised medians are of with these pr-eblems, operate quite well [reference this re-
little use as a place of refuge for disabled vehicles unless
port]. affd Such intersections apparently are within the realm
they are very low with flat sloping curbs. In some cases the
of normal operational expectations of the driver. However,
prevention of midblock turns causes operational problems
intersections with medians wider than 15 in have still
because of heavy concentrations of left-turning traffic at the
more accidents, and intersections with medians wider
intersections.
than 18 m are difficult to signalize properly.
The foregoing disadvantages of raised medians are largely
Widths in a range ef 15 te 24 fa hewe Eleyeleped seeident
eliminated when flush or low-profile mountable medians are
pfeblems iA qRA;P. Ar- F-, Aq as the MN,er-s may tend to be
used. However, flush-paved medians are also difficult to see
at night when the pavement is wet. Visibility can be improved eenfused about the intended epeFatienal eha—meeter-isties ef t
Multiple iAtPfqPPtis;;.q
by use of a contrasting texture such as a chip seal coat and
by proper delineation. Raised bars or blocks have proved to Experience shows that at unsignalized intersections driv-
be ineffective as a median treatment and should not be used. ers prefer medians that are obviously narrow or those that
provide an adequate refuge area to allow independent road-
way crossing operation.
At urban and suburban signalized intersections, re-
search has found that accident experience increases as the
Special consideration needs to be given to the width of median width increases [reference this report]. Therefore,
medians when at-grade intersections are to be provided. Re- median widths at urban and suburban signalized intersec-
search has shown that most types of undesirable driving tions should not be wider than necessary and should be
behavior on the median roadway of divided highway in- determined primarily by the space required in the median
tersections are associated with competition for space on for left-turn treatments. Median widths of more than 18
the median roadway between vehicles traveling in the m are undesirable at intersections that are signalized or
same direction through the median [reference this re- that may require signalization in the foreseeable future.
port]. Such problems are generally greater at urban and The efficiency of signal operations decreases as the me-
suburban intersections than at rural intersections, where dian width increases, because drivers require more time
volumes of turning and crossing traffic are lower. Types to traverse the median and special detectors may be
of undesirable driving behavior observed include side- needed to avoid trapping drivers in the median at the end
by-side queuing, angle stopping, and encroaching on the of the green phase for traffic movements that pass
through lanes of the divided highway. At suburban unsig- through the median. Furthermore, if the median becomes
nalized intersections, the frequency of accidents and un- so wide that separate signals are required on the two
desirable driving behavior were observed to increase as roadways of the divided highway, delays to motorists will
the median width increased. Thus, medians at suburban increase substantially and careful attention must be given
 91

to vehicle storage requirements on the median roadway The following revisions are recommended to a portion of the
between the two signals. discussion on pages 752 through 754 in Chapter IX of the 1994
Uncurbed narrow medians often present problems with Green Book concerning control radii for minimum turning paths.
turning movements at intersections in that vehicles tend to The intent of these changes is to introduce a recommended
run off the pavement edges. An operational measure that approach for reducing the potential for erratic maneuvering by
appears promising is to provide guidance in the form of small vehicles where median openings with large paved areas
edge fines to accommodate the turning paths of passenger are required.
cars, while providing sufficient paved area beyond the
edge lines to accommodate the turning path of an occa- The customary at-grade intersection on a divided high-
sional large vehicle. way does not have a continuous physical edge of pavement
A median barrier may be desirable on some arterial streets delineating the left-turn path. Instead, the driver has guides
with relatively fast-moving traffic. It permits a positive sepa- at the beginning and at the end of the left-turn operation: (1)
ration of traffic and discourages indiscriminate pedestrian the centerline of an undivided crossroad or the median edge
crossing. Where the median barrier is terminated at cross of a divided crossroad, and (2) the curved median end. For
streets and other median openings, it should be turned down the central part of the turn the driver has the open central
or flared to reduce the hazard. Reference is made to the intersection area in which to maneuver. Under these circum-
Roadside Design Guide (6) for further discussion on treat- stances for minimum design of the median end, the precision
ment of the end of barriers. Additional information on median of compound curves does not appear necessary, and simple
barriers and median treatments at intersection areas is con- curves for the minimum assumed edge of the left turn have
tained in Chapters IV and IX, respectively. The information been found satisfactory. The larger the simple curve radius
on medians and median barriers in Chapter IV is especially used, the better it will accommodate a given design vehicle,
pertinent to urban arterials because they require the most but the resulting layout for the larger curve radius will have
varied application of these features. a greater length of median opening and greater paved areas
than one for a minimum radius. These areas may be suffi-
ciently large to result in erratic maneuvering by small vehi-
GREEN BOOK CHAPTER IX (At-Grade cles, which may interfere with other traffic. To reduce the
Intersections) effective size of the intersection for most motorists, con-
sideration should be given to providing an edge marking
The following revisions to the general discussion of median corresponding to the desired turning path for passengers
opening design on page 751 in Chapter IX of the 1994 Green cars, while providing sufficient paved area to accommo-
Book are recommended. The primary intent of these changes is date the turning path of an occasional large vehicle.
to incorporate the principle that the design of the median opening
and median width is influenced by the area type (rural versus The following text presents recommended revisions to remove
urban/suburban). mention of a recommended 12-m minimum median opening
length on page 757 in Chapter IX of the 1994 Green Book.
The design of a median opening and median ends should
During the research, intersections (particularly those with wide
be based on traffic volumes, area type, and type of turning
medians) wereobserved operating properly with median opening
vehicles, as discussed in Chapter H. Cross and turning traffic
lengths as short as 9 in. Furthermore, the research found that
must operate in conjunction with the through traffic on the
the frequency of undesirable driving behavior increased as the
divided highway. This requirement makes it necessary to
median opening length increased.
know the volume and composition of all movements oc-
curring simultaneously during the. design hours. The design For any three- or four-leg intersection on a divided high-
of a median opening becomes a matter of considering what way the length of median opening should be as great as the
traffic is to be accommodated, choosing the design vehicle to width of crossroad roadway pavement plus shoulders and in
use for layout controls for each cross and tun-Ling movement, no ease less than 40 ft (er- less than the er-essfead payeEneat
investigating whether larger vehicles can turn without undue plus 8 ft). Where the crossroad is a divided highway, the
encroachment on adjacent lanes, and finally checking the length of opening should be at least equal to the width of the
intersection for capacity, and evaluating the potential for crossroad roadways plus that of the median plus 8 ft. The use
operational problems related to undesirable driving be- of a 4" minimum length of opening without regard to the
havior. If the capacity is exceeded by the traffic load, the width of median or the control radius should not be considered
design must be expanded, possibly by widening or otherwise except at very minor crossroads. Care should be taken not
adjusting widths for certain movements. The area type may to make the median opening longer than necessary at rural
influence the median width selected. Intersections in unsignalized intersections. The 404 minimum length of
urban/suburban areas have been found to operate more opening dees not apply te epenings for U-aims-,-as is discussed
safely with narrow medians, while unsignalized intersec- elsewhere.
tions in rural areas have been found to operate more
safely with wider medians [reference this report]. Traffic In addition, consideration should be given to revising Figure
control devices such as yield signs, stop signs, or traffic IX-69 on page 776 of the 1994 Green Book to include the
signals may be required to regulate the various movements median width requirements for U-turns from a median left-tum
effectively and improve the effectiveness of operations. Wide lane to the inside lane, outside lane, and shoulder in the opposite
medians may lead to inefficient traffic signal operation, direction of travel. It might also be useful to point out that Figure
however. IX-69 can be used to determine the median width requirements
92

for U-turns at an intersection as well as at median openings Offset Left-Turn Lanes

intended solely for U-turns.
Changes are recommended on page 786 of the 1994 Green For medians wider than about 5.4 m it is desirable to
Book to remove the discussion of parallel offset left-turn lanes align the left-turn lane so that it will reduce the width
from the section entitled "Median End Treatment" and place it of the divider to 1.8 to 2.4 m immediately before the
in a new section entitled "Offset Left-Turn Lanes." This section intersection, rather than to allgn it exactly parallel with
would also introduce the concept of a tapered offset left-turn and adjacent to the through lane. This alignment will
lane. A figure would be added to the Green Book to illustrate place the vehicle waiting to make the turn as far to the
parallel and tapered offset left-turn lanes. The text showing. the left as practical, maxin-dzing the offset between the oppos-
recommended modifications is as follows: ing left-turn lanes, and thus providing improved visibility
of opposing through traffic. The advantages of aligning
For medians wider than about 5.4 in as shown in Figure the left-turn lanes are (1) better visibility of opposing
IX-74, it is usually preferable to provide some offset between through traffic; (2) decreased possibility of conflict be-
the left-turn lanes in the opposing directions of travel. tween opposing left-turn movements within the intersec-
Offset left-turn lanes of this type are discussed in the next tion; and (3) more left-turn vehicles served in a given
section. edigo the left tufn lane ifi a ffiftARLOF th-At fe period of time, particularly at a signalized intersection.
the width of!he EhVidff tO 1.8 tO 2.4 fa iffffnediately in advane-e Parallel offset left-turn lanes may be used at both signal-
ef the inter-seetien, father than te aline it e*aedy pafallel with ized and unsignalized intersections. This left-turn lane
and adjaeent to the thfeugh lafte. This alinefaefif will plaee configuration is referred to as a parallel offset left-turn
the vehiele waiting te make the Pdm as faf te the left as lane and is illustrated in Figure IX-74A.
proetieal and thus pr-evide appr-epriate visibility of opposing An offset between opposing left-turn vehicles can also
dueugh tfaffle. The advantages ef a4igning the left sum ]R;;P,.q be achieved with a left-turn lane that diverges from the
afei (1) beaeF visibility ef eppesing thfough wiffie; (2) de through lanes and crosses the median at a slight angle.
eFeased pessibilify of eefiffiet betweeft eppeping left tum Figure IX-74B illustrates a tapered offset left-turn lane
fnevemems withift the ifiter-seetien; and (3) Faer-e left ~UFH of this type. Tapered offset left-turn lanes provide the
vehieles sefved in a given period ef fifne, pfffieulafly at a same advantages as parallel offset left-turn lanes in reducm
signalized interseetion. ing sight distance obstructions and potential conflicts be-
Any emeess afea widtin the fnedian te the right ef the left tween opposing left-turn vehicles and in increasing the
tum lane sheuld be payed, pFefer-ably with ee efficiency of signal operations. Tapered offset left-turn
Faent, 9F provided with painted ehannelizatien A Ed affiRi"n-a-ineal lanes are normally constructed with a 1.2-m nose between
911Rh with
I..-- the MftdWay suffeee as deseribed in the pr-eeeding the left-turn lane- and the opposing through lanes. Ta-
seetien. pered offset left-turn lanes have been used primarily at
signalized intersections.
For curbed dividers 1.2 in or more in width at the narrowed Parallel and tapered offset left-turn lanes should be
end, the curb nose can be offset from the opposing through separated from the adjacent through travel lanes by
traffic lane 0.5 in or more, with gradual taper beyond to make painted or raised channelization.
it less vulnerable to contact by through traffic, as shown in
Figure IX-73B. The shape of the nose for curbed dividers 1.2 Two figures would accompany the preceding material in the
in wide usually is semicircular but, for wider width, the ends Green Book. These figures, which would be designated in the
are normally shaped to a bullet nose pattern to conform better Green Book as Figures IX-74A and IX-74B, are presented here
with the paths of turning vehicles. as Figures G-1 and G-2.
 93

- - - - - - - - - - - - - - - - - - - - - - - - -cm
-
cco

- - - - - - - - - ffm - - - - - - - - - -

Figure G-1. Parallel offset left-turn lane (prq~osed Green Book Figure JX-74A).

- - - - - - --- - - - - - - - - - - -
cm am am
F
- - - - - - - - - -
7 cco

Figure G-2. Tapered offset left-turn lane (proposed Green Book Figure IX-74B).
THE TRANSPORTATION RESEARCH BOARD is a unit of the National Research Coun-
cil, which serves the National Academy of Sciences and the National Academy of Engineering. It
evolved in 1974 from the Highway Research Board which was established in 1920. The TRB
incorporates all former HRB activities and also performs additional functions under a broader scope
involving all modes of transportation and the interactions of transportation with society. The Board's
purpose is to stimulate research concerning the nature and performance of transportation systems, to
disseminate information that the research produces, and to encourage the application of appropriate
research findings. The Board's program is carried out by more than 270 committees, task forces,
and panels composed of more than 3,300 administrators, engineers, social scientists, attorneys,
educators, and others concerned with transportation; they serve without compensation. The program
is supported by state transportation and highway departments, the modal administrations of the
U.S. Department of Transportation, the Association of American Railroads, the National Highway
Traffic Safety Adniinistration,'and other organizations and individuals interested in the development
of transportation.
The National Academy of Sciences is a private, nonprofit, self-perpetuating society of distin-
guished *scholars engaged in scientific and engineering research, dedicated to the furtherance of
science and technology and to their use for the general welfare. Upon the authority of the charter
granted to it by the Congress in 1863, the Academy has a mandate that requires it to advise the
federal government on scientific and technical matters., Dr. Bruce M. Alberts is president of the
National Academy of Sciences.
The National Academy of Engineering was established in 1964, under the charter of the National
Academy of Sciences, as a parallel organization of outstanding engineers. It is autonomous in its
administration and in the selection of its members, sharing with the National Academy of Sciences
the responsibility for advising the federal government. The National Academy of Engineering also
sponsors engineering programs aimed at meeting national needs, encourages education and research
and recognizes the superior achievements of engineers. Dr. Harold Liebowitz is president of the
National Academy of Engineering.
The Institute of Medicine was established in 1970 by the National Academy of Sciences to
secure the services of eminent members of appropriate professions in the examination of policy
matters pertaining to the health of the public.- The Institute acts under the responsibility given to
the National Academy of Sciences by its congressional charter to be an adviser to the federal
government and, upon its own initiative, to identify issues of medical care, research, and education.
Dr. Kenneth I. Shine is president of the Institute of Medicine.
The National Research Council was organized by the National Academy.of Sciences in 1916 to
associate the broad community of science and technology with the Academy's purpose of furthering
knowledge and advising the federal government. Functioning in accordance with general policies
determined by the Academy, the Council has become the principal operating agency of both the
National Academy of Sciences and the National Academy of Engineering in providing services to the
government, the public, and the scientific and engineering communities. The Council is administered
jointly by both Academies and the Institute of Medicine. Dr. Bruce M. Alberts and Dr. Harold
I
Liebowitz are chairman and vice chairman, respectively, of the National Research Coun cil.