ISSN 1055- 1425
March 2009
This work was performed as part of the California PATH Program of the
University of California, in cooperation with the State of California Business,
Transportation, and Housing Agency, Department of Transportation, and the
United States Department of Transportation, Federal Highway Administration.
The contents of this report reflect the views of the authors who are responsible
for the facts and the accuracy of the data presented herein. The contents do not
necessarily reflect the official views or policies of the State of California. This
report does not constitute a standard, specification, or regulation.
Final Report for Task Order 6601
CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
A Comparative Safety Study of Limited
versus Continuous Access High Occupancy
Vehicle ( HOV) Facilities
UCB- ITS- PRR- 2009- 22
California PATH Research Report
Kitae Jang, David R. Ragland, Ching- Yao Chan
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
A Comparative Safety Study of
Limited versus Continuous Access
High Occupancy Vehicle ( HOV) Facilities
Task Order 6601
March 10, 2009
Prepared for:
California Department of Transportation
Prepared by
Kitae Jang, MS, UC Berkeley Traffic Safety Center ( TSC)
David R. Ragland, PhD, MPH, UC Berkeley Traffic Safety Center ( TSC)
Ching- Yao Chan, PhD, Partners for Advanced Transit and Highways ( PATH)
UC Berkeley Traffic Safety Center ( TSC)
Institute of Transportation Studies
University of California, Berkeley, CA 94720
Tel: 510/ 642- 0655
Fax: 510/ 643- 9922
and
Partners for Advanced Transit and Highways ( PATH)
Institute of Transportation Studies
University of California, Berkeley
Tel: 510/ 665- 3406
Fax: 510/ 665- 3537
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
ACKNOWLEDGMENTS
The University of California Traffic Safety Center ( TSC) and Partners for Advanced
Transit and Highways ( PATH) appreciate and acknowledge the contributions of the
Department’s Advisory Committee for this research project:
Jason Osman
District Traffic Safety Engineer ( District 12)
Chief, Office of Traffic Studies
Jerry Champa
HQ Division of Traffic Operations
Traffic Operations Liaison Engineer ( Districts 7 & 12)
Janice Benton
HQ Division of Traffic Operations
Chief, Office of Traffic Safety
Monica Kress
HQ Division of Traffic Operations
Chief, Office of System Management Operations
Fred Yazdan
HQ Division of Research & Innovation
Contract Manager
ii
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
GLOSSARY OF ACRONYMS AND TERMS
PDO Collision: Property Damage Only collision
HOV Lane: High Occupancy Vehicle lane
HOT Lane: High Occupancy Toll lane
GP Lanes: General Purpose ( GP) lanes
Left Lane: General Purpose ( GP) lane closest to HOV lane
Ingress/ Egress Area: A section of HOV lane open for exiting and entering HOVs
Shoulder: Area between median and traveling lanes. There are two shoulder areas on
both sides of the freeway. Since HOV lanes in California are generally on the median, the
present report indicates median ( i. e., inner or left) shoulder.
Buffer: Pavement markings separating the HOV and GP lanes, which exist only in
limited access HOV facilities.
Total Width: Consists of three parts: 1) shoulder width, 2) HOV lane width, and 3)
buffer width.
HCCL: High Collision Concentration Location
VMT: Vehicle Miles Traveled
VHT: Vehicle Hours Traveled
CRP: Continuous Risk Profile, a method for estimating continuous risks along a
roadway. 1
Continuous Access HOV Lane: Drivers may move in and out of the HOV lane at any
point.
Limited Access HOV Lane: Drivers may enter and exit the HOV lane only at
ingress/ egress areas.
1 Chung, K. and Ragland, D. R., Method for generating continuous risk profile for highway collisions,
Proceedings of 86th Transportation Research Board Annual Meeting, 2007.
iii
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
ABSTRACT
The report summarizes the findings from comparative studies of safety performance
between two different types of High Occupancy Vehicle ( HOV) facilities in California –
continuous access versus limited access. The findings show that HOV facilities with
limited access offer no safety advantages over those with continuous access, whether
measured by percentage of collisions, collisions per mile, collisions per VMT, or
collision severity. As part of the present research, the authors investigated the
relationship between HOV design features and safety performance of HOV facilities.
One key design feature is shoulder/ total width. The findings indicate that maintaining
adequate shoulder and total width is essential, and a quantitative estimate for the
relationship between shoulder and total width versus safety performance of HOV lanes is
provided. Additionally, findings from investigating other influential factors on safety
performance of HOV facilities, including design features of ingress/ egress section in
limited access HOV facilities, congestion, High Collision Concentration Locations and
etc., were also documented. While further research is needed, results to date suggest that
improvements in HOV facility performance can be achieved by improved HOV facility
design.
Keywords: High Occupancy Vehicle ( HOV) Lanes, Managed Lanes, Limited Access,
Continuous Access, Design Features, Shoulder/ Total Width, Safety, Traffic Collisions
iv
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
EXECUTIVE SUMMARY
SYNOPSIS
Limited access HOV facilities were designed to separate higher speed traffic in HOV
lanes from lower speed traffic in adjacent lanes in order to reduce the risk of collisions
caused by vehicles weaving between lanes of traffic traveling at different speeds. Using
data from California freeways, limited access HOV and left lanes were compared with
those of continuous access HOV facilities to evaluate the safety of each, and to determine
which characteristics could improve performance in either type of facility. Based on these
results, limited access HOV facilities do not appear to provide increased safety, whether
measured by percentage of collisions, collisions per mile, collisions per VMT, or
collision severity— the pattern actually seems to suggest the opposite. From a strictly
safety viewpoint, this suggests that constructing limited access facilities would not
achieve the goal of increasing freeway safety. However, the study recommends design
features that could maximize the safety of limited access facilities, such as maintaining
adequate total and shoulder width, adequate length of access segments, and adequate
distance between access areas and on/ off ramps. While further research is needed to
expand understanding of safety performance of HOV facilities, results to date suggest
that improvements in HOV facility performance can be achieved through refinements in
freeway design.
BACKGROUND
Two configurations for HOV lanes— limited and continuous— are prevalent in California.
Limited access HOV lanes ( predominant in Southern California), have specified
locations for ingress and egress HOV maneuvers, and are separated from other freeway
lanes by buffer zones demarcated by pavement markings or physical barriers. Such
separation is intended to facilitate smooth and safe operation of traffic flows, typically at
relatively high speeds, within HOV lanes. Concerns about limited access lanes include
possible impacts on traffic maneuvers due to: ( i) vehicle lane- changing concentrated near
ingress/ egress locations, and ( ii) extensive vehicle lane- changing between freeway ramps
and HOV access points within a fixed and often relatively short distance. Continuous
access HOV lanes ( predominant in Northern California) do not include a buffer zone,
allow vehicles to enter and exit at any location, and are in operation only during peak
hours ( generally, Monday– Friday, 5– 9AM, 3– 7PM).
One of the objectives of the California Department of Transportation ( Caltrans) is
continuous evaluation and improvement of safety and operational efficiency in all
facilities. A large number of HOV facilities have been implemented on California
freeways as one of the major demand management strategies to counteract continuously
increasing congestion in metropolitan areas. HOV facilities are an evolving part of
freeway infrastructure and induce significant complexity for driving tasks. Unlike
continuous access HOV facilities, limited access HOV facilities have demarcation
between HOV and GP lanes, and allow HOVs to enter and exit only within limited
v
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
sections of HOV lanes. Geometric features associated with limited access HOV facilities
often create weaving movements both upstream and downstream of ingress/ egress areas,
and also encourage concentrated and consecutive lane changes across lanes in the vicinity
of ingress/ egress sections. Such traffic movements can conflict with existing traffic flow
and cause high- risk situations. Therefore, there are strong incentives to ( i) investigate the
safety performance of the two major types of HOV facilities in California, and ( ii)
enhance the level of understanding about the characteristics influencing safety
performance in both types of HOV facilities.
A research project at the University of California, Berkeley was funded by Caltrans to
compare traffic collision patterns between limited access and continuous access HOV
lanes and, if any, investigate the attributes accounting for such differences. While some
facilities utilize an actual barrier between HOV and adjacent lanes, the current study
focuses only on facilities that are buffer- separated, meaning that the separation is
indicated solely by pavement markings.
DATA SOURCES
Data for the study was collected from the following sources:
Collisions
• Traffic Accidents Surveillance and Analysis System ( TASAS), 1999– 2003
collisions in traveling lanes: TASAS is a collision database which records
information associated with each collision that occurs within the California
state freeway system.
Traffic Data
• Freeway Performance Measurement System ( PeMS),
( https:// pems. eecs. berkeley. edu/): PeMS is a tool that processes and analyzes
traffic data collected by loop detectors and tags.
Geometric Features
• Highway Performance Monitoring System ( HPMS): This is a federally
mandated inventory system and planning tool, designed to assess the
nation’s highway system.
• California Department of Transportation Document Retrieval System
( DRS): DRS is a document database which enables users to search for,
view, and print documents including built plans and survey files by using a
browser on the California DOT intranet.
• California Department of Transportation Photolog ( http:// video. dot. ca. gov/
photolog/): This is a series of photos recorded in accordance with post mile
in the California state freeway system.
vi
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
• Aerial Photos: Google Earth ( http:// earth. google. com/): Google Earth is a
virtual globe program which maps the earth via the superimposition of images
obtained from satellite imagery, aerial photography, and GIS 3D globe.
HOV Facilities
• California Department of Transportation HOV reports, which contain
information regarding HOV lanes on the California state freeway system.
• California Department of Transportation HOV inventory ( 2005), which is an
inventory of HOV facilities within the California state freeway system.
METHODS
The safety performance of both the HOV lane itself and adjacent left lane are likely to be
affected by the type of access ( limited versus continuous). For HOV and left lanes, a
general analysis was conducted of HOV facilities constructed before 1999 ( to provide
sufficient collision data) and consisting of over 60 percent of all California HOV lanes.
All collisions ( fatal, injury, and PDO) that occurred within traveling lanes between 1999
and 2003 were included in the analysis. Since continuous access HOV lanes are in
operation only during peak hours ( generally, Monday– Friday, 5– 9 a. m., 3– 7 p. m.), the
comparison was limited to those hours. More detailed analyses were then conducted on a
subset of eight sites ( four continuous access and four limited access).
The following analyses were conducted for HOV and left lanes:
• Percentage of total collisions for HOV and left lanes compared with all
freeway lanes ( 62 sites, analysis replicated for 8 sites)
• Number of total collisions per mile per hour ( 62 sites, analysis replicated for
8 sites)
• Number of fatal and injury collisions per mile per hour ( 62 sites, analysis
replicated for 8 sites)
• Number of fatal and injury collisions per VMT ( 8 sites)
• Impact on congestion ( 8 sites)
The following analyses focused on HOV lanes:
• Shoulder width ( 13 sites)
• Total width ( shoulder plus lane plus buffer) ( 13 sites)
• Spatial analysis ( 4 sites)
• Type of collision analysis ( 8 sites)
vii
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
RESULTS FOR HOV LANES
Compared with continuous access HOV lanes, we observed the following characteristics
for limited access HOV lanes:
• Higher percentage of total collisions across all freeway lanes
• Higher number of total collisions per mile per hour
• Higher number of fatal and injury collisions per mile per hour
• Higher number of total collisions per VMT
• Higher number of fatal and injury collisions per VMT
These differences were statistically significant ( p< 0.05).
Additional Findings for HOV Lanes:
• The difference in safety performance was apparently not due to differences
in congestion.
• Shoulder width was a very strong predictor of safety performance in both
limited and continuous access HOV lanes. However, differences in shoulder
width between continuous and limited access facilities did not account for
the differences in safety performance between the two HOV configurations.
• A spatial analysis of collisions along the length of the freeway suggested a
difference in patterns of clustering collisions between the two types of
facilities.
• An analysis of collision types indicated a higher proportion of rear- end
collisions, and a lower proportion of side- swipe collisions in limited access
facilities.
In sum, the safety performance of HOV lanes was lower in limited access HOV facilities
compared with continuous access facilities. The spatial and collision analyses are
consistent with the interpretation that vehicles in limited access HOV lanes have less
room to maneuver in the event of bottlenecks within the lane.
RESULTS FOR LEFT LANES
Compared with continuous access left lanes, we observed the following characteristics
for limited access left lanes:
• Higher percentage of total collisions across all freeway lanes
• Higher number of total collisions per mile per hour
• Lower number of fatal and injury collisions per mile per hour
• Higher number of total collisions per VMT
• Lower number of fatal and injury collisions per VMT ( Statistically
insignificant)
viii
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
These differences— except those for the fatal and injury collisions per VMT— were
statistically significant ( p< 0.05). In sum, the results for left lanes are mixed. Generally,
limited access left lanes experience lower levels of safety performance for total
collisions, but slightly better performance when analysis is limited to fatal and injury
collisions.
RESULTS FOR HOV AND LEFT LANES COMBINED
Compared with continuous access HOV and left lanes combined, we observed the
following characteristics for limited access HOV and left lanes combined:
• Higher percentage of the total collisions across all freeway lanes
• Higher number of total collisions per mile per hour
• Higher number of fatal and injury collisions per mile per hour ( statistically
insignificant)
• Higher number of total collisions per VMT
• Higher number of fatal and injury collisions per VMT
These differences— except those for the fatal and injury collisions per mile per hour—
were statistically significant ( p< 0.05). In sum, most of these measurements indicate a
lower overall ( HOV and left lanes combined) level of safety performance for limited
access facilities compared with continuous access facilities.
CONCLUSIONS
Our results suggest that, compared with continuous access HOV facilities, limited access
HOV facilities do not appear to provide increased safety, whether measured by
percentage of collisions, collisions per mile, collisions per VMT, or collision severity—
the pattern actually seems to suggest the opposite. Potential differences in traffic volume,
number of lanes, shoulder width, lane width, or total width ( shoulder plus HOV lane plus
buffer) did not appear to account for these findings. Strictly from a safety viewpoint, this
suggests that construction of limited access facilities to achieve a safety objective is not
warranted.
However, our study recommends design features that could maximize the safety of
limited access facilities. One of these features is shoulder/ total width. Our findings
suggest that maintaining adequate shoulder and total width is essential, and we provide a
quantitative estimate for the relationship between shoulder and total width versus safety
performance of HOV lanes. Unfortunately, we do not currently have sufficient data to
analyze the tradeoff between shoulder width and buffer width. This would be a fairly
straightforward extension of our study and could be conducted by including additional
sites in our analysis.
Another potential safety enhancing feature is to optimize the length of the access section
and its location in relation to on/ off ramps. Based on our analysis of collision rates near
ix
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
access points in limited ingress and egress HOV facilities, it appears that collision rates
are higher in short access sections. We hypothesize that short access sections create
queues in the HOV lane and increase the incidence of rear- end collisions ( our findings
showed increased rates of rear- end collisions in limited access facilities) and intensify
weaving. Our findings also suggest that locating access areas in close proximity to on/ off
ramps should be avoided, since it may lead to intense weaving and hot spots for collisions
across freeway lanes. Additional analyses would be required to develop a quantitative
estimate for the optimal access segment length and distance of access segments from on-off-
ramps.
In general, our study demonstrates that HOV design features can have an impact on
safety performance. While further research is needed, results to date suggest that
improvements in HOV facility performance can be achieved by improved HOV lane
design.
x
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
TABLE OF CONTENTS
ACKNOWLEDGMENTS ................................................................................................ ii GLOSSARY OF ACRONYMS AND TERM 1ET. XA EBDCLEEUF TOIINFVI CET OISOUNNMTSEM ANANTRSD Y . B.. .. A...... C...... K...... G...... R...... O...... U...... N...... D...... ... S.............................................................................................................................. ............................................................................................................................... ixi i1vii 2. PREVIOUS STUDIES .............................................................................................. 2 3. STUDY OBJECTIVE ............................................................................................... 4 4. DATA SOURCES ...................................................................................................... 5 5. COMPARISON OF HOV COLLISION STATISTICS IN STATEWIDE
ANALYSIS ........................................................................................................................ 7 6. COMPARISON OF HOV COLLISION STATISTICS IN DETAILED STUDY
ROUTES ............................................................................................................................ 9 6.1. Comparison of HOV Collisions in Detailed Study Sites ( 8 Study Routes) .......... 9 6.2. Comparison of HOV Collisions accounting for traffic volume ( 8 Study Routes)
10 7. DETAILED HOV COLLISION ANALYSIS ....................................................... 12 7.1. Shoulder Width Analysis ( 13 study routes) ........................................................ 12 7.2. Total ( shoulder + HOV lane + buffer) width analysis ( 13 study routes) ............ 13 7.3. Congestion Analysis ( 8 study routes) ................................................................. 14 7.4. High Collision Concentration Location ( HCCL) Analysis ( 8 study routes)....... 15 7.5. Spatial Collision Analysis ( 4 study routes) ......................................................... 15 7.6. Ingress/ Egress Area Analysis ( 8 study routes) ................................................... 16 7.7. Collision Type Analysis ( 8 study routes)............................................................ 17 8. CONCLUSION AND IMPLICATIONS ............................................................... 18 9. FURTHER RESEARCH ........................................................................................ 20 9.1. Investigation of safety performance of HOV lanes at micro level .................... 20 9.2. investigation of role of shoulder and buffer widTH .......................................... 20 9.3. Design criteria for limited access openings and evaluation of effectiveness of
HOV flyovers ................................................................................................................ 20 9.4. Application to High Occupancy Toll ( HOT) Facilities ...................................... 21 9.5. Drivers’ Behavior in HOV/ HOT Facilities ......................................................... 21 ATTACHMENTS ........................................................................................................... 22
xi
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
xii
TABLE OF FIGURES
Attachment A Diagram of Continuous and Limited Access HOV Configurations ......... 22
Attachment B Location of HOV Facilities and Collection of Collision Data
for HOV and Left Lanes ............................................................................................ 23
Attachment C A Comparison of Collision Distribution in HOV and Left Lanes
( Statewide Analysis) .................................................................................................. 24
Attachment D A Comparison of Collisions Per Mile Per Hour in HOV and Left Lanes
( Statewide Analysis) .................................................................................................. 25
Attachment E A Comparison of Fatal and non- fatal injury Collisions
Per Mile Per Hour in HOV and Left Lanes ( Statwide Analysis) ............................... 26
Attachment F Detailed Study Routes for HOV Collision Analysis ................................. 27
Attachment G A Comparison of Collision Distribution in HOV
and Left Lanes ( 8 Study Routes) ............................................................................... 28
Attachment H A Comparison of Collisions Per Mile Per Hour In HOV and Left Lanes
( 8 Study Routes) ........................................................................................................ 29
Attachment I A Comparison of Fatal and non- fatal injury Collisions Per Mile Per Hour
In HOV and Left Lanes ( 8 Study Routes) ................................................................. 30
Attachment J Collision Rate ( Collisions Per VMT) ( 8 Study Routes) ............................ 31
Attachment K Fatal and non- fatal injury Collision Rate ( Collisions Per VMT)
( 8 Study Routes) ........................................................................................................ 32
Attachment L Statistical tests ( Statewide and 8 Study Routes) ....................................... 33
Attachment M Detailed Study Routes for HOV Collision Analysis ( 8 Study Routes) ... 37
Attachment N Relationship Between Shoulder Width and Collisions
Per Mile Per Hour ...................................................................................................... 38
Attachment O Relationship Between Total Width and Collisions Per Mile Per Hour .... 39
Attachment P The Relationship Between HOV Lane Occupancy
and Collisions/ Million VMT in Study Sites .............................................................. 41
Attachment Q Continuous Risk Profiles ( CRPs) in HOV and Left Lanes
( 8 Study Routes) ........................................................................................................ 42
Attachment R Cross- Sectional Distribution of Collisions Near Ingress/ Egress Areas
( Interstate 210 Eastbound, Los Angeles County) ...................................................... 43
Attachment S Cross- Sectional Distribution of Collisions Near Ingress/ Egress Areas
( Interstate 5 Northbound, Orange County) ................................................................ 44
Attachment T Relationship Between Collisions Per Mile and Distance to Nearest
Entrance/ Exit Ramp in Limited Access HOV Facilities ............................................ 45
Attachment U Types of Collisions in HOV Lanes Per Facility ....................................... 46
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
1. DEFINITIONS AND BACKGROUND
High Occupancy Vehicle ( HOV) lanes have been implemented on crowded urban
freeways to mitigate continuously growing traffic congestion and to improve overall
mobility within metropolitan freeway systems. HOV lanes are designed to enable
vehicles carrying more passengers to bypass the congested General Purpose ( GP) lanes,
while encouraging the use of carpools and public transportation to move more people per
lane with a fewer vehicles. In California, HOV lanes were first introduced in the San
Francisco Bay area as early as 1962 and near Los Angeles in 1971. In the 1990s, HOV
lanes were increasingly implemented in congested freeway segments in Southern and
Northern California metropolitan regions. As of 2005, HOV lanes comprised 1,305
( directional) lane- miles of freeway, with 895 lane- miles located in Southern California,
410 in Northern California, and 950 additional lane- miles of HOV lanes proposed.
Since their inception, two configurations for HOV lanes— limited and continuous— have
emerged in California ( Attachment A). Limited access HOV lanes have specified
locations for ingress and egress maneuvers, and are separated from other freeway lanes
by buffer zones, 2 demarcated by pavement markings or physical barriers. Such separation
is intended to facilitate smooth and safe operation of traffic flows, typically at relatively
high speeds. However, safety concerns include potential impacts on traffic maneuvers
due to: ( i) vehicle lane- changing concentrated near ingress/ egress locations, and ( ii)
extensive vehicle lane- changing between freeway ramps and HOV access points within a
fixed and often relatively short distance. Continuous access HOV lanes do not include a
buffer zone, and allow vehicles to enter and exit at any location.
Predominant in Southern California, limited access HOV lanes are in operation 24 hours
a day, seven days a week, while continuous access HOV lanes, which are predominant in
Northern California, are in operation only during peak hours ( generally, Monday– Friday,
5– 9 a. m., 3– 7 p. m.). 3
The differences between HOV configuration and operation throughout the state largely
reflect different freeway commute patterns at the initial stage of HOV lane
implementation. In areas in which periods of congestion last for many hours of the day,
full- time HOV operation with limited access has been favored. In areas in which
commute patterns consist of short peak periods and clear directional flows, continuous
access HOV operation during peak commute hours has been preferred.
2 Some facilities utilize an actual barrier between HOV and adjacent lanes. The current study focuses only
on facilities that are buffer- separated, meaning that the separation is indicated solely by pavement markings.
The buffer can vary in width.
3 Operation hours vary across routes depending on their specific commute and congestion patterns.
1
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
2. PREVIOUS STUDIES
Previous studies regarding the safety performance of HOV facilities have focused mainly
on comparing collision patterns before and after implementation of HOV facilities, and
identifying factors influencing collision occurrence. The studies have been based on data
from a limited number of study routes and consequently the results have been relatively
inconclusive due to data limitations in both quantity and quality. Moreover, there has
been lack of research comparing the safety performance of different types of HOV
facilities.
One recent research study conducted a before and after comparison of buffer- separated
limited access HOV facilities in two corridors in Dallas, Texas. The before and after
comparison of corridor crash rates showed a substantial increase in injury crash rates
after installation of the buffer- separated HOV lanes. The study also suggested several
factors that might have contributed to the increased crash risk: speed differential between
HOV and general purpose lanes, reduced width of general purpose lanes, loss of the
inside shoulder, and difficulty for vehicles in the HOV lane to find gaps in traffic when
entering the general purpose lanes. 4
Another study also compared the frequency and characteristics of collisions before and
after installation of an HOV lane without physical separation ( i. e., buffer- separated) by
converting the inner shoulder area to an HOV lane on State Route 91 in Los Angeles,
California. The study concluded that installation of HOV lanes did not have an adverse
effect on the safety performance of the corridor and that the changes in crash
characteristics were due to the changes in spatial and temporal attributes of traffic
congestion. 5
An additional study documented findings based on analysis of how HOV facility
operation affects the safety of selected California freeways. The study suggested that the
collision patterns showed no systematic differences in the lane locations of collisions or
other influential factors, but were characterized by the location of traffic congestion.
Localized traffic congestion results in the clustered collisions, HCCLs, during peak hours
with and without HOV lanes. 6
The study, funded by FHWA, conducted a before and after comparison for four different
types of HOV facilities: 0– 2 foot buffer, 3– 8 foot buffer, 8 foot buffer with 6 inch raised
barrier, and 13 foot ( full) buffer. The latter two designs did not appear to increase overall
collision rates, while the first design appeared to increase the collision rates when
compared with the pre- HOV collision rates. The results of the study for the second design
4 Cooner, S. A. and Ranft, S. E., Safety evaluation of buffer- separated High- Occupancy Vehicle lanes in
Texas, Transportation Research Record, No. 1959, 2006
5 Golob, T. F., Recker, W. W. and Levine, D. W., Safety of High- Occupancy Vehicle lanes without physical
separation, Journal of Transportation Engineering, Vol. 115, No. 6, 1989
6 Sullivan, E. C. and Devadoss, N., High- Occupancy Vehicle facility safety in California, Transportation
Research Record, No. 1394, 1993
2
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
type were inconclusive. 7 Since more than 70% of HOV facilities implemented in
California are based on the former two designs, however, the safety performance of HOV
facilities currently in place in California cannot be explained based on the results of this
study.
Unlike the previously described studies, another study compared three specific types of
HOV facilities in California: physically separated facilities, buffer- separated facilities
( full lane width), and contiguous facilities. In the study, the term “ contiguous facility”
referred to continuous access as well as limited access facilities in which the buffer width
was narrower than full lane width ( 13 ft). The study found that separated facilities were
superior to contiguous facilities. 8 During the past two decades since the study was
conducted, however, continuous and limited access HOV facilities, which were
categorized as a single group in the study, become the two predominant types of HOV
configurations in California. The findings of that study are not applicable to the questions
addressed in our study.
Hockaday et al. investigated collision patterns in three different types of HOV facilities;
contiguous, buffer- separated and barrier separated. For this investigation, the authors
used TASAS collision data from 1989 to 1991 and concluded that HOV facilities did not
show any significant and systematic differences when compared with non- HOV facilities
with comparable features. 9 However, this study did not specifically compare collision
statistics between different types of HOV facilities, but instead compared overall
collision statistics between HOV and non- HOV facilities.
The findings of previous studies are not directly applicable to the questions addressed in
our study. The present study compares collision patterns occurring in continuous access
HOV facilities with those of limited access facilities ( of various buffer widths), which
represent the two major types of HOV facilities in California.
7 Case, R. B.. The safety of concurrent- lane HOV projects, Hampton roads planning district commission,
Chesapeake, Virginia, 1995
8 Newman, L., Nuworsoo, C. and May, A. D., Operational and safety experience with freeway HOV
facilities in California, Transportation Research Record, No. 1173, 1988.
9 Hockaday, S., Sullivan, E., Devadoss, N., Daly, J. and Chatziiouanou, A., High- Occupancy Vehicle Lane
Safety. Submitted to the State of California Department of Transportation by California Polytechnic State
University. Contract Number 51P278, TR 92- 107. September 1992.
3
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
3. STUDY OBJECTIVE
One of the objectives of the California Department of Transportation ( Caltrans) is
continuous evaluation and improvement of all facilities. Therefore, there are strong
incentives to ( i) investigate the safety performance of the two major types of HOV
facilities in California, and ( ii) enhance the level of understanding about the factors
influencing safety performance in both types of HOV facilities. A research project at the
University of California, Berkeley was funded by Caltrans to compare traffic collision
patterns between limited access HOV lanes ( predominant in Southern California) and
continuous access HOV lanes ( predominant in Northern California) and, if any,
investigate the attributes accounting for such differences.
4
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
4. DATA SOURCES
Comparison of safety performance between the two different types of HOV facilities
involved multi- dimensional issues including geometric features, demand for each facility,
inherent collision features, and other factors. Diverse data sources were utilized to
maximize accuracy in comparison and explanation of differences in collision patterns
between the two different HOV configurations.
Collisions
• Traffic Accidents Surveillance and Analysis System ( TASAS), 1999– 2003
collisions on traveling lanes: a collision database which records information
associated with each collision that occurs within the California state freeway
system. 10
Traffic Data
• Freeway Performance Measurement System ( PeMS), ( https:// pems. eecs.
berkeley. edu/): a tool that processes and analyzes traffic data collected by loop
detectors and tags.
Geometric Features
• Highway Performance Monitoring System ( HPMS): a federally mandated
inventory system and planning tool, designed to assess the nation’s highway
system. 11
• California Department of Transportation document Retrieval System ( DRS):
a document database which enables users to search for, view, and print
documents including built plans and survey files by using a browser on the
California DOT intranet.
• California Department of Transportation Photolog ( http:// video. dot. ca. gov/
photolog/): a series of photos recorded in accordance with postmile in
California state freeway system.
• Aerial Photos: Google Earth ( http:// earth. google. com/): a virtual globe
program which maps the earth via the superimposition of images obtained
from satellite imagery, aerial photography, and GIS 3D globe.
10 California Department of Transportation, California DOT traffic manual Ch. 3. Accident and roadway
records, 2004.
11 California Department of Transportation Division of Transportation System Information, Highway
Performance Monitoring System ( HPMS) instruction for updates, 2007.
5
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
HOV Facilities
• California Department of Transportation HOV reports: these reports contain
information regarding HOV lanes in the California state freeway
system. 12,13
• California Department of Transportation HOV inventory ( 2005): an
inventory of HOV facilities within the California state freeway system.
12 California Department of Transportation, District 4, Office of Highway Operations, Bay Area HOV lanes,
1999- 2005.
13 California Department of Transportation, District 7, HOV annual report, 2005 and 2006.
6
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
5. COMPARISON OF HOV COLLISION STATISTICS IN
STATEWIDE ANALYSIS
A statewide/ general analysis was conducted using a large sample of HOV lanes
constructed before 1999 to allow sufficient time to observe collisions after
implementation. A total of 824 HOV lane- miles were examined, including 279 lane- miles
of continuous access, and 545 lane- miles of limited access HOV lanes. The selected
routes covered more than 60 percent of all existing HOV facilities as of 2005
( Attachment B). All collisions ( fatal, injury, and PDO) that occurred within traveling
lanes between 1999 and 2003 were included in the analysis. Since continuous access
HOV lanes are in operation only during peak hours ( generally, Monday– Friday, 5– 9 a. m.,
3– 7 p. m.), the comparison was limited to those hours.
In the TASAS database, collisions are recorded by lane ( i. e., HOV, left, interior, and
right) and location along the freeway. Thus, we were able to calculate the distribution of
collisions across lanes and as a function of lane miles. Collisions per mile per hour were
averaged by weighting operational hours and lane miles of the routes.
We observed the following characteristics in limited access HOV facilities compared
with continuous access facilities:
• Limited access facilities experienced a higher percentage of total collisions
in the combined HOV and left lanes ( 43% for limited, versus 33% for
continuous) ( i. e., differences in collision distribution across the freeway).
The same pattern was observed separately in HOV lanes ( 9% for limited,
versus 5% for continuous) and left lanes ( 34% for limited, versus 28% for
continuous) ( Attachment C).
• Limited access facilities experienced a higher number of collisions per mile
per hour in the combined HOV and left lanes ( 3.6 collisions for limited,
versus 2.9 for continuous). The same pattern was observed separately in
HOV lanes ( 0.8 for limited, versus 0.4 for continuous) and left lanes ( 2.8 for
limited, versus 2.5 for continuous) ( Attachment D).
• Limited access facilities experienced a higher number of fatal and injury
collisions per mile per hour in HOV and left lanes combined ( 0.88 collisions
for limited, versus 0.86 for continuous). The same pattern was observed for
the HOV lane separately ( 0.25 collisions for limited, versus 0.13 collisions
for continuous), but the opposite pattern was observed in left lanes ( 0.63 for
limited, versus 0.73 for continuous) ( Attachment E).
7
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
For the comparison between distributions of collisions across lanes and collisions per
mile per hour, a test statistic for each case was derived ( Attachment U). All the
differences except for the disparity in injury collision rates in combined HOV and left
lanes were statistically significant ( p< 0.05). In other words, only the difference in injury
collision rates in combined HOV and left lanes was not statistically significant.
Compared with HOV lanes in continuous access facilities, HOV lanes in limited access
facilities experienced ( i) a higher percentage of collisions compared with other lanes, ( ii)
a higher number of total collisions per mile per hour, and ( iii) a higher number of fatal
and injury collisions per mile per hour. Each of these differences was substantial and
statistically significant.
The pattern for left lanes was different than the pattern for HOV lanes. Compared with
left lanes in continuous access facilities, HOV lanes in limited access facilities
experienced ( i) a higher percentage of collisions compared with other lanes, and ( ii) a
higher number of collisions per mile per hour, but, ( iii) a lower number of fatal and injury
collisions per mile per hour. The differences were statistically significant.
The HOV lanes in this study appeared to have greater safety performance than left lanes
in all comparisons of collision statistics. Compared with other traveling lanes, however,
such advantageous safety performance could also be observed in any type of lane
adjacent to the median, either HOV or non- HOV lane. This could be due to the
interaction level between the lanes. Traffic in the median lane interacts with traffic on the
right side only, while traffic in other lanes interacts with traffic on both sides. For further
interpretation, HOV lanes need to be compared with non- HOV median lanes. However,
the present study is limited to the comparison of HOV and left lanes, and further
comparisons are outside the scope.
There are two general categories of explanations for the apparent superiority in safety
performance of HOV lanes, and to some degree, left lanes, in continuous versus limited
access facilities: ( i) differences in traffic volume, traffic congestion, or design features
not integral to the continuous versus limited access facilities, or ( ii) features inherent in
continuous versus limited access design. Some of these potential explanations are
addressed in the following sections.
8
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
6. COMPARISON OF HOV COLLISION STATISTICS IN
DETAILED STUDY ROUTES
6.1. COMPARISON OF HOV COLLISIONS IN DETAILED STUDY
SITES ( 8 STUDY ROUTES)
To understand the geometric factors and other components that may impact the
distribution of collisions across lanes, a detailed analysis was carried out with a subset of
HOV facilities recommended by the Caltrans Advisory Group: four facilities with limited
access and four with continuous access HOV lanes ( Attachment F). Fatal, injury, and
PDO collisions were included in the analysis. The average number of lanes in both
continuous and limited access HOV facilities in the analysis was similar ( i. e., 4– 4.5
lanes). Therefore, the number of lanes did not account for the difference between
continuous and limited access HOV facilities. To begin this detailed investigation, we
first replicated the analysis described in the previous section.
Since this was a subset of the sites included in the previous statewide analysis, the HOV
lanes were all constructed before 1999. For the eight sites, a total of 91.6 HOV lane- miles
were examined, including 40.7 lane- miles of continuous access, and 50.9 lane- miles of
limited access HOV lanes ( Attachment F). As with the statewide/ general analyses, all
collisions ( fatal, injury, and PDO) that occurred within traveling lanes between 1999 and
2003 were included in the analysis and the comparison was limited to peak hours
( generally, Monday– Friday, 5– 9 a. m., 3– 7 p. m.). Regarding the statewide/ general
analyses, we calculated the distribution of collisions across lanes. Collisions per mile per
hour were calculated by dividing the number of collisions by operational hours and lane
miles of the routes.
We observed the following characteristics in limited access HOV facilities compared
with continuous access facilities:
• Limited access facilities experienced a higher percentage of total collisions
in the combined HOV and left lanes ( 49% for limited, versus 29% for
continuous) ( i. e., differences in collision distribution across the freeway).
The same pattern was observed separately in HOV lanes ( 10% for limited,
versus 4% for continuous) and left lanes ( 39% for limited, versus 25% for
continuous) ( Attachment G).
• Limited access facilities experienced a higher number of collisions per mile
per hour in the combined HOV and left lanes ( 4.0 for limited, versus 2.7 for
continuous). The same pattern was observed separately in HOV lanes ( 0.8
for limited, versus 0.4 for continuous) and left lanes ( 3.2 for limited, versus
2.3 for continuous) ( Attachment H).
9
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
• Limited access facilities experienced a higher number of fatal and injury
collisions per mile per hour in the combined HOV and left lanes ( 0.92 for
limited, versus 0.80 for continuous). There was a higher number of fatal and
injury collisions per mile per hour in the HOV lane ( 0.29 for limited, versus
0.10 for continuous), but the opposite pattern in left lanes ( 0.63 for limited,
versus 0.70 for continuous). ( Attachment I).
The same statistical tests were conducted as for the previous analyses. All the differences
except for the difference in injury collision rates in left lanes were statistically significant
( p< 0.05). In other words, only the difference in injury collision rates in left lanes was not
statistically significant.
Collision patterns observed in the eight study routes are generally consistent with those
observed in the statewide/ general collision comparison, suggesting that the set of eight
study routes is representative of the larger set used for the statewide collision comparison.
6.2. COMPARISON OF HOV COLLISIONS ACCOUNTING FOR
TRAFFIC VOLUME ( 8 STUDY ROUTES)
Traffic volumes were fairly similar, on average, for both the continuous and limited
access HOV facilities, suggesting that differences in traffic volume did not account for
the different collision patterns. However, to further clarify the possible influence of
traffic volumes on the distribution of collisions across lanes, we controlled for traffic
volumes by calculating collision rates ( collisions per million vehicle miles traveled). The
same set of eight routes used in the previous section for which data from PeMS was
available was selected. Using data from PeMS, we were able to calculate collisions per
million vehicle miles traveled ( rate) by dividing collisions per mile per hour by traffic
volume across various types of lanes. Four of the HOV facilities were limited access and
four were continuous access. For the eight sites, a total of 91.6 HOV lane- miles were
examined, including 40.7 lane- miles of continuous access, and 50.9 lane- miles of limited
access HOV lanes ( Attachment F). As in the previous analyses, fatal, injury, and PDO
collisions were included in this analysis.
We observed the following characteristics in limited access HOV facilities compared
with continuous access facilities:
• Limited access facilities experienced a higher rate of collisions in the
combined HOV and left lanes ( 4.49 for limited, versus 3.12 for continuous).
The same pattern was observed separately in HOV lanes ( 1.43 for limited,
versus 0.59 for continuous) and left lanes ( 3.06 for limited, versus 2.53 for
continuous) ( Attachment J).
10
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
• Limited access facilities experienced a higher number of fatal and injury
collisions per million VMT in the HOV lane ( 1.10 for limited, versus 0.92
for continuous), but the opposite pattern in left lanes ( 0.49 for limited,
versus 0.16 for continuous). The number of fatal and injury collisions per
million VMT was higher for HOV and left lanes combined ( Attachment K).
A test statistic derived in the previous section was extended for use in the collision rate
comparison ( Attachment L). Differences in collision rates and fatal and injury collision
rates were statistically significant at the 95 percent level of confidence.
With traffic volume controlled for, it is notable that observed collision rates were much
lower for HOV lanes than for the adjacent lanes. This is true for both limited and
continuous access facilities, and is consistent with the finding that the percentage of
collisions and number per mile per hour is also lower in the HOV lanes. Although this
finding is very interesting, and should be explored further, it is outside the scope of the
current study.
The results controlling for traffic volume— the results for rates— were generally
consistent with the results above for the number of collisions per mile per hour. The rate
differences between continuous and limited access HOV facilities did not appear to be
accounted for by higher traffic volumes in limited access HOV facilities. HOV lanes in
limited access facilities appear to be associated with an increased number of collisions
compared with HOV lanes in continuous access facilities. For left lanes, the mixed
pattern was observed. These patterns are explored further in the next section.
11
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
7. DETAILED HOV COLLISION ANALYSIS
7.1. SHOULDER WIDTH ANALYSIS ( 13 STUDY ROUTES)
The primary purpose of the shoulder is to accommodate stopped or disabled vehicles so
that they do not disrupt traffic flow in traveling lanes, and to manage water accumulation
on the roadway by installation of drainage facilities. The shoulder also provides drivers
with lateral clearance where they can avoid direct contact with other vehicles, recover
from error, and resume normal driving in order to increase safety performance of
freeways. 14,15 Because in California HOV lanes are usually located next to the inner
shoulder, we were able to analyze the influence of shoulder width on the safety
performance of HOV lanes to determine whether shoulder width accounts for differences
between limited and continuous access facilities.
For this analysis we used 13 sites, which are a subset of HOV facilities recommended by
the Caltrans Advisory Group; nine facilities with limited access and four with continuous
access HOV lanes ( Attachment M). A total of 184 HOV lane- miles were examined,
including 50 lane- miles of continuous access, and 134 lane- miles of limited access HOV
lanes. All collisions ( fatal, injury, and PDO) occurring within traveling lanes between
1999 and 2003 were included in the analysis, and the comparison was limited to peak
hours ( generally, Monday– Friday, 5– 9 a. m., 3– 7 p. m.). Collisions per mile per hour were
calculated by dividing the number of collisions by operational hours and lane miles of the
routes.
Shoulder widths varied greatly in segments within these two sets of facilities, with a
minimum width of about 2 feet, and a maximum width of about 12.2 feet. The variation
in shoulder width in the data collected from our study sites made it possible for us to
analyze the impact of shoulder width on collision patterns.
An initial comparison of the 13 sites showed that average shoulder width was about the
same for both limited and continuous access facility sites. This suggests that shoulder
width was not a factor in the differences in collision patterns found between the two
HOV configurations. We then investigated the relationship between the average shoulder
widths of individual freeway segments with collision rates from those segments. In this
analysis, narrower shoulder width was associated with a higher number of collisions per
mile per hour, regardless of whether the HOV facility was limited or continuous access
( Attachment N). As shoulder width increased from 2 to 12.2 feet, the number of
collisions per hour per mile decreased sharply.
14 Hauer, E., Shoulder width, shoulder paving and safety, Unpublished, 2000
15 Gross, F. and Jovanis, P. P., Estimation of the safety effectiveness of lane and shoulder width: case-control
approach, Journal of Transportation Engineering, Vol. 133, Issue 6, 2007
12
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
These findings indicate that shoulder width did not account for the differences in
collision patterns between the two types of HOV facilities. Consistent with previous
findings, however, shoulder width appears to be an important safety performance factor.
This was true in both continuous and limited access HOV facilities. This outcome should
be studied further to determine whether this is true for other Caltrans HOV facilities, and
to identify the mechanisms.
7.2. TOTAL ( SHOULDER + HOV LANE + BUFFER) WIDTH
ANALYSIS ( 13 STUDY ROUTES)
Total width is the lateral space or degree of freedom allocated to drivers in the HOV lane.
The present analysis focuses on the safety performance of HOV facilities in accordance
with the consumption of spatial resources. Total width consists of three different
components: 1) shoulder, 2) HOV lane, and 3) buffer. 16 Among these three components,
shoulder widths observed in the study sites varied from 2 to 12.2 feet ( i. e., a range of 10.2
feet), lane width varied only between 11.5 and 13 feet ( i. e., a range of about 1.5 feet),
while buffer width varied between 0 and 5.2 feet ( i. e., a range of 5.2 feet). Continuous
access facilities, of course, have no buffer between HOV and left lane. Most of the
variation in total width was contributed by variation in shoulder width, followed by
buffer width. For this analysis we used the same 13 sites studied in the previous analyses,
and the outcome was collisions per hour per mile. As with the previous analysis, this
analysis was limited to HOV lanes.
A scatter plot of collisions per mile per hour relative to total width was constructed. A
trend line for each type of HOV facility was estimated, based on the scatter plot, to
determine the variation of collisions per mile per hour with respect to total width.
Narrower total width ( shoulder plus lane plus buffer) was associated with a higher
number of collisions per mile per hour in both types of HOV configurations ( Attachment
O). The scatter plot appeared to yield different curves showing the relationship between
total width and collisions/ mile/ hour, with the curve for limited access facilities shifted to
the right. For any given total width, the number of collisions per mile per hour was higher
in limited access HOV lanes than in continuous access HOV lanes. Additionally, the
vertical discrepancy between the trend lines of continuous and limited access HOV lanes
could be interpreted as the potential safety benefit of continuous access with respect to
limited access, while holding all other influential factors constant.
This analysis does not establish the separate influence of each component of total width
in the study. Additional sites would need to be studied to evaluate the individual
influence of shoulder, lane, and buffer widths in a statistically meaningful manner, and to
determine the optimal balance between shoulder and buffer width.
16 Since continuous access HOV facilities do not have buffer zones, and observed lane width variations
were small, only two of the components, shoulder and HOV lane width, are included in total width.
13
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
7.3. CONGESTION ANALYSIS ( 8 STUDY ROUTES)
It has been suggested that congestion affects occurrence of collisions in HOV lanes. The
HOV lane is intended to be less congested than the GP lanes so that vehicles meeting the
criteria for use of HOV lanes can save travel time by bypassing congested GP lanes. Due
to high HOV demand or traffic operational issues, however, HOV lanes often become
congested. 17,18 In the congestion analysis, the relationship between HOV lane congestion
and HOV collisions was studied.
There are various congestion measurements based on either density or speed. Density
directly measures the degree of crowdedness on the freeway as the unit of vehicles per
unit distance, while measurements based on speed quantify the degree of congestion
relative to a specific freely flowing traffic condition. Speed measurement is advantageous
in terms of measuring delay with respect to freely flowing travel time. Caltrans produces
the Highway Congestion Monitoring Program ( HICOMP) annual data compilation which
measures congestion occurring on urban freeways in California based on speed data or
estimates. The compilation only provides regional congestion levels and the data
collecting method varies across routes and districts. 19 Moreover, variations inherent in
the range of freely flowing traffic conditions cannot be measured using speed
measurements. Due to such data issues in HICOMP, it is more feasible to use congestion
data from loop detectors.
From the perspective of traffic safety, density can more accurately reflect risk by
measuring the number of vehicles in the vicinity, which may be related to the probability
of conflicts between vehicles. For this reason, density was adopted to represent
congestion in the analysis. Occupancy data, dimensionless measure of density, from
inductive loop detectors were analyzed to measure the density of traffic flow. 20,21
By displaying a scatter plot of HOV lane occupancy versus HOV collisions per mile per
hour, no significant relationship was observed ( Attachment P). The graph shows that
higher collision rates are not always associated with higher levels of congestion and vice
versa. Therefore, we conclude that congestion itself does not account for the difference in
collision rates between limited and continuous access HOV facilities.
17 Chen, C., Kwon, J. and Varaiya, P., An empirical assessment of traffic operations, Proceedings of
International Symposium on Transportation and Traffic Theory, 2005.
18 Federal Highway Administration ( FHWA), Safe, Accountable, Flexible, Efficient Transportation Equity
Ace: A legacy for Users ( SAFETEA- LU), 2005 ( accessed 02. 03. 08) ( http:// www. fhwa. dot. gov/ safetealu/
legis. htm)
19 California Department of Transportation, 2006 State Highway Congestion Monitoring Program Annual
Data Compilation, 2007.
20 Cassidy, M. J. and Coifman, B., Relation among average speed, flow, and density and analogous relation
between density and occupancy, Transportation Research Record, No. 1591, 1997.
21 3. Hall, F. L., The relationship between occupancy and density. Transportation Forum, Vol. 3- 3, 1986.
14
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
7.4. HIGH COLLISION CONCENTRATION LOCATION ( HCCL)
ANALYSIS ( 8 STUDY ROUTES)
Continuous Risk Profile ( CRP) is a method that can directly generate a variation of risk
measurement interpretable as the number of collisions or collisions per unit distance
along a freeway segment. 22 In the high collision concentration location ( HCCL) analysis,
CRPs were constructed along the HOV and left lanes of eight detailed study sites to
examine the concentration of collisions in the vicinity of the ingress/ egress areas. The
eight sites were the same as those analyzed previously in this report. Four of the HOV
facilities were limited access and four were continuous access. For the eight sites, a total
of 91.6 HOV lane- miles were examined, including 40.7 lane- miles of continuous access,
and 50.9 lane- miles of limited access HOV lanes ( Attachment F). Fatal, injury, and PDO
collisions were included in the analysis, and we calculated the number of
collisions/ mile/ hour.
For the eight routes examined, two exemplary routes are displayed ( Attachment Q). The
peaks in the profile represent HCCLs for which collisions per mile exceed 90 percent of
collisions per mile along the route. The pattern of HCCLs is different for continuous and
limited access facilities. In continuous access facilities, each of the peaks accompanies
peaks in adjacent lanes. This implies that the factors causing collision concentration
appear to have an equivalent influence on HOV and left lanes, and possibly on all lanes.
In contrast, in limited access HOV facilities, some of the peaks are observed only in the
HOV lanes. This distinguishable pattern for limited access and continuous access HOV
facilities was observable in all of the study routes.
7.5. SPATIAL COLLISION ANALYSIS ( 4 STUDY ROUTES)
As an example of a potentially more detailed analysis, the cross- sectional distribution
patterns of collision frequencies were analyzed along four freeway routes with limited
access HOV lanes. The number of collisions within each 0.05 mile segment on the
freeway was counted and the segment was classified into one of five categorical groups
and shaded in accordance with its recorded collision frequencies. Darker shading
represents higher collision frequencies while lighter shading represents lower collision
frequencies. The plots constructed in this way facilitated visual observation of collision
concentrations along the freeway and across lanes in conjunction with the locations of
other freeway facilities including on- and off- ramps and HOV flyovers. 23
22 Chung, K. and Ragland, D. R., Method for generating continuous risk profile for highway collisions,
Proceedings of 86th Transportation Research Board Annual Meeting, 2007.
23 HOV flyover is also called HOV direct connector. An HOV flyover is a structure providing uninterrupted
access between freeways that enables HOV vehicles to move directly from the HOV lane of one freeway to
the HOV lane of the second freeway, without leaving the commuter lane to exit one freeway only to merge
back across the next freeway into its HOV lane.
( http:// carpoolconnect. com/ glossary/ show/ HOV% 20Direct% 20Connector% 20Ramp)
15
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
In this figure, three sections of freeway in the near ingress/ egress area are displayed: two
with ingress/ egress areas only ( Attachment R) and one with ingress/ egress area and HOV
flyover ( Attachment S). The following patterns were observed:
• The freeway sections where merging and diverging HOV maneuvers were
expected were associated with HCCLs, particularly when these areas were
closely located to on- and off- ramp areas ( Attachment R).
• Some HCCLs in GP lanes did not overlap HOV access areas but were
located close to HOV access areas. Although these sections do not reside
within HOV access areas, they appear to be associated with anticipated
HOV maneuvers ( Attachment R).
• The freeway section in the vicinity of an HOV access area where HOV
maneuvers directly travel through HOV flyovers have lower collision
concentrations ( Attachment S).
The analysis provided potential explanations of the causes of HCCLs associated with
HOV ingress/ egress areas. More detailed investigation is required in order to understand
the influences of traffic movement induced by HOV facilities.
7.6. INGRESS/ EGRESS AREA ANALYSIS ( 8 STUDY ROUTES)
The above analysis suggests that collision patterns in limited access HOV facilities may
be related to ingress/ egress sections. There are two principle factors related to
ingress/ egress sections. The first factor is the distance to the ingress/ egress area from the
nearest on- and off- ramps. An ingress/ egress area a short distance from either an on- or
off- ramp forces intensive weaving on the part of drivers attempting to enter the freeway
and then enter the HOV lane, or those attempting to exit the HOV lane and then the
freeway itself. The second factor is the length of the section. A shorter ingress/ egress
section forces more weaving within a smaller length of freeway, potentially increasing
the probability of collisions. In addition, shorter ingress/ egress sections might create
bottlenecks in either the HOV or left lane as drivers attempt either to exit or enter the
HOV lane. In field visits to HOV facilities we have observed and video- recorded both of
these phenomena, but we have not yet assessed them quantitatively.
To address the first factor, collisions per mile per hour at 24 limited ingress and egress
sections along four limited access HOV routes were plotted relative to their distance from
nearby on- or off- ramps ( Attachment T). From these plots, there was no clear systematic
relationship between distance from the nearby on- or off- ramps and collisions. This may
be because the plot does not systematically control for other factors.
However, three locations showed significantly higher numbers of collisions per mile per
hour than the average collisions per mile per hour in limited access HOV lanes. These
three ingress/ egress segments were all within 0.3 mile of the nearest on- or off- ramp.
However, the three segments also had short access lengths ( 0.25 mile) and high traffic
volume in the HOV lane during peak hours ( 1000– 1200 vehicles per hour versus 700–
800 vehicles per hour on average).
16
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
The data from the 24 sites in this eight- route analysis were not sufficient to separate the
impact of distance from the nearest on/ off ramp, length of access segment, and volume.
These are critical components related to the design of limited access HOV facilities and
further research is recommended to determine how these separate factors operate
individually and/ or together to impact safety performance. Sufficient additional sites are
available for this analysis.
7.7. COLLISION TYPE ANALYSIS ( 8 STUDY ROUTES)
An analysis of collision types was conducted to determine the different collision patterns
of limited versus continuous access facilities. For this analysis, the set of eight study
routes was used, and all collisions were included. Rear- end and sideswipe collisions
together comprised just over 90 percent of all collisions in both limited and continuous
access HOV lanes. Limited access HOV lanes experienced more rear- end and side- swipe
collisions in absolute numbers. However, the major difference was in the distribution of
rear- end versus side- swipe collisions in limited versus continuous access facilities.
( Attachment U). In continuous access HOV lanes, 66 percent of collisions were rear- end,
and 26 percent were side- swipe collisions. In limited access HOV lanes, 75 percent were
rear- end, and 17 percent were side- swipe collisions.
The difference in type of collision pattern observed in limited versus continuous access
HOV lanes could be explained by traffic movements inherent to continuous and limited
access HOV facilities. Compared with HOVs traveling in limited access HOV lanes,
HOVs traveling in continuous access HOV lanes were more likely to be exposed to
continuous interaction with traffic in adjacent lanes, and thus there was a greater
occurrence of side- swipe collisions versus rear- end collisions. Meanwhile, HOVs in
limited access HOV lanes are prohibited from changing lanes except in ingress/ egress
areas, and tended to have more interaction with vehicles in the back or front than those in
adjacent lanes and thus experienced a greater number of rear- end collisions.
17
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
8. CONCLUSION AND IMPLICATIONS
HOV lanes have been used for decades to improve freeway capacity. Two different HOV
configurations have been widely implemented in the state of California over the past
several decades, and are defined as follows:
• Continuous access, in which HOV lanes are demarcated only by signage
and pavement markings. This configuration is predominant in Northern
California, and operates only during peak hours ( generally, Monday– Friday,
5– 9 a. m., 3– 7 p. m.).
• Limited access, in which pavement buffers separate HOV lanes. This
configuration is predominant in Southern California, and is in operation 24
hours a day, seven days a week.
Limited access HOV facilities, where signage and pavement markings are designed to
separate higher speed traffic in HOV lanes from lower speed traffic in adjacent lanes, are
designed to reduce the risk of collisions caused by vehicles traveling in different lanes at
different speeds. Using collision data, traffic data, and infrastructure data from California
freeways, limited access HOV and left lanes were compared with those of continuous
access HOV facilities to evaluate the safety of each, and to determine which
characteristics could improve performance in either type of facility.
Our results suggest that, compared with continuous access HOV facilities, limited access
HOV facilities do not appear to provide increased safety, whether measured by
percentage of collisions, collisions per mile, collisions per VMT, or collision severity—
the pattern actually seems to suggest the opposite. Potential differences in traffic volume,
speed differential, number of lanes, shoulder width, lane width, and total width ( shoulder
plus lane plus buffer) did not appear to account for these findings. Strictly from a safety
viewpoint, this suggests that construction of limited access facilities to achieve a safety
objective is not warranted.
However, our study recommends design factors that could maximize the safety of limited
access facilities. One of these factors is shoulder/ buffer width. Our findings suggest that
maintaining adequate shoulder and buffer width is essential, and we provide a
quantitative estimate for the optimal shoulder width. Unfortunately, we do not currently
have sufficient data to analyze the tradeoff between shoulder width and buffer width.
This would be a fairly straightforward extension of our study and could be conducted by
including additional sites in our analysis.
Another potential safety enhancing feature is the length of the access section and its
location in relation to on/ off ramps. Based on our analysis of collision rates near access
points in limited ingress and egress HOV facilities, it appears that collision rates are
higher in short access sections. We hypothesize that short access areas creates queues in
the HOV lane and increase the incidence of rear- end collisions ( we found increased rates
of rear- end collisions in limited access facilities) and intensify weaving. Our findings also
18
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
suggest that locating access areas in close proximity to on/ off ramps should be avoided.
This may lead to intense weaving and hot spots for collisions across freeway lanes.
Additional analyses would be required to develop a quantitative estimate for the optimal
access segment length and distance of access segments from on- off- ramps.
In general, our study demonstrates that HOV design factors can have an impact on safety
performance. While further research is needed, results to date suggest that improvements
in HOV facility performance can be achieved by improved HOV lane design.
19
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
9. FURTHER RESEARCH
9.1. INVESTIGATION OF SAFETY PERFORMANCE
OF HOV LANES AT MICRO LEVEL
For the present study, influential factors on the safety performance of HOV facilities
were identified and quantified at the macro level ( i. e. corridor level). However, in order
to maximize safety performance, these findings should be expanded to a more accurate
microscopic level of quantification. Based on the refined quantifications, the current
HOV implementation guidelines can be improved by the following:
• Quantify the influence of various design features on safety performance
• Propose new design criteria for constructing new HOV lanes ( improving the
current design criteria)
• Conduct a before/ after HOV lane retrofit study
9.2. INVESTIGATION OF ROLE OF
SHOULDER AND BUFFER WIDTH
Our findings suggest that wider shoulders and buffers are associated with increased safety
performance. However, data are not sufficient to determine the relative contribution of
shoulder and buffer width. This information is crucial when it is necessary to design facilities
for locations with limited space. With additional study sites it would be possible to:
• Quantify the influence of shoulder and buffer width separately in detail
• Quantify the relative contribution of shoulder and buffer width
• Quantify the relative importance of the Weave Lane to the operation of the
buffer- separated HOV lane, including an analysis to see if it would be more
beneficial to reduce shoulder widths to provide for the Weave Lane.
9.3. DESIGN CRITERIA FOR LIMITED ACCESS OPENINGS AND
EVALUATION OF EFFECTIVENESS OF HOV FLYOVERS
Our findings indicate that there are higher collision concentrations at locations where
HOV maneuvers conflict with freeway ramp traffic. Additionally, such conflicts
substantially limit freeway capacity, leading to excessive delays and congestion in
general purpose lanes. Therefore, it is necessary to properly locate the access openings so
that operational and safety disadvantages can be minimized.
• Determine optimal locations of access areas with respect to ramps
• Evaluate length of access areas
As a countermeasure to mitigate such conflicts, HOV flyovers are constructed to enable
HOV lanes to directly connect between major interchanges. For greater efficiency, the
effectiveness of HOV flyovers needs to be evaluated.
20
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
• Evaluate the effectiveness of HOV flyovers in terms of safety and operation
9.4. APPLICATION TO HIGH OCCUPANCY TOLL ( HOT)
FACILITIES
Recently, High Occupancy Toll ( HOT) facilities have emerged as an alternative means to
control the level of utilization of HOV lanes that are often under- or overused. Some
HOV facilities in the California State Highway System have been selected for conversion
to HOT facilities. The design features of HOT facilities are similar to those of limited
access HOV facilities. Therefore, applicability of HOV facility design criteria needs to be
investigated and further developed to accommodate the HOT facility adaptation.
• Study applicability of HOV design criteria for HOT facilities
• Further develop design criteria and adapt some of the features into HOT
facilities
9.5. DRIVERS’ BEHAVIOR IN HOV/ HOT FACILITIES
A substantial proportion of collisions are caused by drivers’ behavior, not by other
environmental factors. Previous research suggested that operational features inherent to
HOV/ HOT configurations induce unique driver behaviors including increased lane
changes, smoothing effects, and synchronized speed drops on HOV facilities ( inducing
degradation of HOV facility, SAFETEA- LU). In- depth understanding of drivers’
behavior might enable us to identify the causes of HOV collisions and operational issues
related to HOV facilities, leading to mitigation strategies for improving operation and
safety of HOV facilities.
• Study drivers’ behavior in relation to HOV facilities
• Develop mitigation strategies for HOV operation and HOV safety
21
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
ATTACHMENTS
Attachment A
Diagram of Continuous and Limited Access HOV Configurations
There are two configurations for HOV facilities in California. Drivers may move in and
out of the continuous access HOV lane at any point; drivers may enter or exit the limited
access HOV lane only at ingress/ egress areas.
22
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment B
Location of HOV Facilities and Collection of Collision Data for HOV and Left Lanes
Collision Data
TASAS, 1999– 2003, Weekdays ( Mon.– Fri.)
Traveling Lanes Only ( HOV, Left, Interior, and Right Lanes)
HOV Operation Hours ( Peak Hours, 5– 9 a. m. & 3– 7 pm.)
Continuous Ingress/ Egress: HOV report and Inventory
Limited Ingress/ Egress: 5– 9 a. m. & 3– 7 p. m.
HOV Location Data
HOV lanes constructed before 1999
HOV lanes existing before 1999 were identified to allow comparable analysis
Routes with unique HOV operation were excluded ( HOT, Bus- Only, Elevated, etc.)
HOV Reports ( D4 & D7), HOV Inventory ( as of 2005), Google Earth Aerial Photo
Study Site Summary ( Comparison)
Facility Type District Number of Route Segments Lane- Miles
Continuous
3 2 25
4 22 254
Limited
7 24 311
8 4 55
12 10 179
Total Study Routes 62 824
Total ( Entire HOV system) 1305
• Lane- mile calculation is based on HOV inventory as of June, 2005
23
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment C
A Comparison of Total Collision Distribution in HOV and Left Lanes
( Statewide Analysis)
5% 9%
28%
34%
67%
57%
0%
20%
40%
60%
80%
100%
Continuous Limited
Other Lanes
Left Lane
HOV Lane
24
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment D
A Comparison of Total Collisions Per Mile Per Hour in HOV and Left Lanes
( Statewide Analysis)
0.4
0.8
2.5
2.8
0.0
1.0
2.0
3.0
4.0
Continuous Limited
Left Lane
HOV Lane
25
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment E
A Comparison of Fatal and injury Collisions Per Mile Per Hour
in HOV and Left Lanes ( Statewide Analysis)
0.13
0.25
0.73
0.63
0.0
0.2
0.4
0.6
0.8
1.0
Continuous Limited
Left Lane
HOV Lane
26
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment F
Detailed Study Routes for HOV Collision Analysis
Site Selection
List of HOV facilities provided by Caltrans Advisory Group
Collision Data
TASAS, 1999– 2003, Weekdays ( Mon. – Fri.)
Traveling Lanes Only
HOV operation hours ( See table below)
Study Routes Included in Detailed Analysis
Start PM End PM
Contra Costa I- 80 E 0.0 10.0 10.0 Monday~ Friday, 5~ 10 AM, 3~ 7 PM
Contra Costa I- 80 W 0.0 9.8 9.8 Monday~ Friday, 5~ 10 AM, 3~ 7 PM
Alameda I- 880 N 13.5 20.9 7.4 Monday~ Friday, 5~ 9 AM, 3~ 7 PM
Santa Clara SR- 101 S 26.4 39.9 13.5 Monday~ Friday, 5~ 9 AM, 3~ 7 PM
Los Angeles I- 105 E 1.2 16.9 15.7 24 hours
Los Angeles I- 105 W 2.6 16.8 14.3 24 hours
Los Angeles I- 210 E 24.8 36.4 11.6 24 hours
Los Angeles I- 405 S 12.9 22.2 9.3 24 hours
Continuous
Limited
Facility Type County Freeway Length Operation Hr.
Postmile
Further analysis was conducted by incorporating traffic data in the calculation of collision
rates based on vehicle- miles. The study sites for this analysis were selected from the list
of study sites from the previous section, based on the loop detector coverage and
condition.
Data Description
Peak hour traffic volume data in one- month period ( June 2003) were downloaded from
Performance Measurement System ( PeMS).
Data only reported as “ good” by PeMS were selected for traffic volume estimation by
comparing with detector health data.
Collision rates were calculated based on the following equation:
Total Peak HourTrafficVolume Yr Lane Mile
Collision rate Number of Collisions
× ⋅
×
=
( 5 .)
106
Collision rates in continuous access HOV lanes were lower than those in limited access
HOV lanes. Meanwhile, collision rates in the left lane were slightly higher in continuous
access HOV lanes.
27
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment G
A Comparison of Total Collision Distribution In HOV and Left Lanes
( 8 Study Routes)
Four of the eight study routes were continuous HOV facilities and four were limited
access facilities. Our subset of eight routes closely replicates findings of the collision
distribution ( See Attachment C).
4%
10%
25%
39%
71%
51%
0%
20%
40%
60%
80%
100%
Continuous Limited
Other Lanes
Left Lane
HOV Lane
28
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment H
A Comparison of Total Collisions Per Mile Per Hour In HOV and Left Lanes
( 8 Study Routes)
0.4
0.8
2.3
3.2
0.0
1.0
2.0
3.0
4.0
5.0
Continuous Limited
Left Lane
HOV Lane
29
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment I
A Comparison of Fatal and injury Collisions
Per Mile Per Hour In HOV and Left Lanes
( 8 Study Routes)
0.10
0.29
0.70
0.63
0.0
0.2
0.4
0.6
0.8
1.0
Continuous Limited
Left Lane
HOV Lane
30
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment J
Total Collision Rate ( Collisions Per VMT) ( 8 Study Routes)
0.59
1.43
2.53
3.06
0.0
1.0
2.0
3.0
4.0
5.0
Continuous Limited
Left Lane
HOV Lane
31
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment K
Fatal and injury Collision Rate ( Collisions Per VMT) ( 8 Study Routes)
0.16
0.49
0.76
0.61
0.0
0.5
1.0
1.5
Continuous Limited
Left Lane
HOV Lane
32
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment L
Statistical tests ( Statewide and 8 Study Routes)
1. Statistical test for the differences between collision distributions
Methodology
Let CC and CL denote the numbers of collisions observed in a specific lane in two
independent sets of C and L Bernoulli trials ( i. e. total collisions across lanes),
respectively. In the analysis, all collisions that occurred in continuous and limited access
HOV facilities are considered to be C and L, where pC and pL represent the true collision
distribution associated with each set of trials ( i. e. total collisions across lanes). Let
C L
C C
p C L
e +
+
= and define
L
p p
C
p p
L
C
C
C
z
e e e e
C L
( 1 ) ( 1 − )
+
−
+
= ~ N( 0,1)
A test at • significance level against one- sided alternative;
H0 : pC = pL versus HA: pC < pL
The hypothesis H0 is rejected at the • level of significance level if Z ≥ z1-•, where •( z •)=
•. ( Larsen and Marx, 2006)
Application
Since the number of days during the study period did not change, average operation hours
per day and total lane mile in each HOV facility were estimated.
Statewide
Continuous access HOV Limited access HOV
HOV collisions ( total) 885 3424
Left collisions ( total) 5218 12198
Total collisions ( total) 18593 36105
Both differences reject the null H0 : pC = pL at 5% significance level. ( i. e. the differences
are statistically significant at 5% significance level)
Eight Study Routes
Continuous access HOV Limited access HOV
HOV collisions ( total) 124 343
Left collisions ( total) 804 1291
Total collisions ( total) 3248 3317
Both differences reject the null H0 : pC = pL at 5% significance level. ( i. e. the differences
are statistically significant at 5% significance level).
33
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
2. Statistical test for the differences between collisions per mile per hour
Methodology
Suppose that the collisions that occurred in continuous and limited access HOV facilities
follow two different Poisson processes. In the analysis, these two Poisson processes were
observed for fixed mile- hours sC and sL. Let CC and CL represent the number of collisions
observed in continuous and limited access HOV facilities, respectively.
CC ~ POISSON (• C)
CL ~ POISSON (• L)
Where, • i= si • • i ( i= C, L)
A test at • significance level against one- sided alternative;
H0 : • C = • L versus HA: • C < • L
The approximated normal test statistic mentioned above can be generalized to the
unequal mile- hour case. Under H0 : • C = • L we have, approximately,
[ ( )] 2
1
L C C L
C L L C
s s C C
Z s C s C
⋅ ⋅ +
⋅ − ⋅
= ~ N( 0,1)
The hypothesis H0 is rejected at the • level of significance level if Z ≥ z1-•, where •( z •)=
•. ( Sichel, 1973 and Shiue and Bain, 1982)
Application
Average operation hours per day and total lane mile in each HOV facility were estimated.
Statewide
Continuous access HOV Limited access HOV
Length ( mile) 279 545
Hour ( hr) 7.5 8
HOV collisions ( total) 885 3424
Left collisions ( total) 5218 12198
HOV collisions
( fatal and non- fatal injury) 277 1109
Left collisions
( fatal and non- fatal injury) 1522 2761
All four differences reject the null H0 : • C = • L at 5% significance level. ( i. e. the
differences are statistically significant at 5% significance level)
34
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Eight Study Routes
Continuous access HOV Limited access HOV
Length ( mile) 40.7 50.9
Hour ( hr) 8.5 8
HOV collisions ( total) 124 343
Left collisions ( total) 804 1291
HOV collisions
( fatal and non- fatal injury) 33 117
Left collisions
( fatal and non- fatal injury) 242 256
All four differences reject the null H0 : • C = • L at 5% significance level. ( i. e. the
differences are statistically significant at 5% significance level)
3. Statistical test for the differences between collisions per million VMT
Methodology
A methodology similar to that used for the comparison of collisions per mile per hour is
applicable to collisions per million VMT. Suppose that collisions that occurred in
continuous and limited access HOV facilities follow two different Poisson processes. In
the analysis, these two Poisson processes were observed for fixed mile- hours sC and sL. In
the analysis, we additionally consider the different numbers of vehicles in continuous and
limited access HOV facilities, nC and nL vehicles per hour. Therefore, we can conduct the
same statistical test by weighting sC and sL by the amount of nC and nL, respectively. Let
CC and CL be the number of collisions observed in continuous and limited access HOV
facilities.
CC ~ POISSON (• C)
CL ~ POISSON (• L)
Where, • i= ni • si • • i ( i= C, L)
A test at • significance level against one- sided alternative;
H0 : • C = • L versus HA: • C < • L
The approximated normal test statistic mentioned above can be generalized to the
unequal mile- hour case. Under H0 : • C = • L we have, approximately,
[ ( )] 2
1
C L L C C L
C C L L L C
n s n s C C
Z n s C n s C
⋅ ⋅ ⋅ ⋅ +
⋅ ⋅ − ⋅ ⋅
= ~ N( 0,1)
The hypothesis H0 is rejected at the • level of significance level if Z ≥ z1-•, where •( z •)=
•. ( Sichel, 1973 and Shiue and Bain, 1982)
35
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Application
Eight Study Routes
Continuous access HOV Limited access HOV
Length ( mile) 40.7 50.9
Hour ( hr) 8.5 8
Vehicles Per Hour ( HOV lane) 3950 3610
Vehicles Per Hour ( Left lane) 5980 6350
HOV collisions ( total) 124 343
Left collisions ( total) 804 1291
HOV collisions
( fatal and non- fatal injury) 33 117
Left collisions
( fatal and non- fatal injury) 242 256
All four differences reject the null H0 : • C = • L at 5% significance level. ( i. e. the
differences are statistically significant at 5% significance level)
36
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment M
Detailed Study Routes for HOV Collision Analysis ( 13 Study Routes)
Site Selection
List of HOV facilities provided by Caltrans Advisory Group
Collision Data
TASAS, 1999– 2003, Weekdays ( Mon.– Fri.)
Traveling Lanes Only
HOV Operation Hours ( See table below)
Start PM End PM
ALA 3.373 ALA 8.036
CC 0.000 CC10.043
ALA 3.8 ALA 8.036
CC 0.000 CC 9.76
ALA I- 880 N 13.51 20.876 7.366 Monday~ Friday, 5~ 9 AM, 3~ 7 PM
SCL SR- 101 S R 26.4 39.92 13.886 Monday~ Friday, 5~ 9 AM, 3~ 7 PM
LA SR- 91 W R 19.434 R 6.85 12.584 24 hours
LA I- 105 E R 1.164 R 16.864 15.7 24 hours
LA I- 105 W R 2.556 R 16.847 14.291 24 hours
LA I- 210 E R 24.784 R 36.407 11.932 24 hours
LA I- 405 S 12.925 26.298 13.373 24 hours
ORA SR- 55 N 7 R 17.825 10.927 24 hours
ORA I- 5 N 7 29 22 24 hours
ORA I- 5 S 7 29 22 24 hours
ORA SR- 57 S 11.083 R 22.551 11.468 24 hours
Limited
Monday~ Friday, 5~ 10 AM, 3~ 7 PM
Facility Type Operation Hr.
Monday~ Friday, 5~ 10 AM, 3~ 7 PM
Continuous
Postmile
Length
D12
District County
ALA/ CC
15.029
D7
E
I- 80 W
Freeway Direction
D4 13.87
ALA/ CC I- 80
37
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment N
Relationship Between Shoulder Width and Collisions Per Mile Per Hour
Facilities with wider shoulder width generally experienced fewer collisions per mile,
regardless of whether the facility was a limited or continuous access facility, based on an
analysis of 13 study routes.
38
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment O
Relationship Between Total Width and Collisions Per Mile Per Hour
Facilities with wider total width generally experienced fewer collisions per mile, and
collisions/ mile/ hour were higher in limited access HOV lanes than continuous access
facilities in our analysis of 13 study routes.
Total width = Shoulder width + lane width + buffer width.
The light gray line was constructed by using the data from four routes with continuous
access and the black line was based on the data from nine routes with limited access ( see
figure on following page).
Data Description
Shoulder and lane widths were extracted from Highway Performance Monitoring System
( HPMS).
Lane widths were estimated by averaging total width across all traveling lanes divided by
number of lanes.
Buffer widths were based on Caltrans Digital Photolog were provided by Caltrans
Engineers. The accuracy is fully dependent on the quality of images.
Shoulder width takes the largest portion of variation in total width.
Maximum difference in lane width across all detailed study corridors is 1.5 ft.
Maximum difference in buffer width across all detailed study corridors is 5.2 ft.
Maximum difference in shoulder width across all detailed study corridors is 12.2 ft.
Collisions/ mile/ hour in continuous access HOV lanes is lower than that in limited access
HOV lanes.
A wide total width provides better safety performance in both continuous and limited
access HOV lanes.
Total width = Shoulder width + lane width + buffer width.
The light gray line is constructed by using the data from four routes with continuous
access and the black line is based on the data from nine routes with limited access.
39
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
40
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment P
The Relationship Between HOV Lane Occupancy
and Collisions/ Million VMT in Study Sites
0
0.5
1
1.5
0 5 10 15
HOV Lane Occupancy (%)
Collisions / Million VMT
Continuous Access Limited Access
I- 405S
I- 210E
I- 105E
I- 105W
I- 80W
SR- 101S
I- 80E
I- 880N
41
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment Q
Continuous Risk Profiles ( CRPs) in HOV and Left Lanes ( 8 Study Routes)
A route for each type of HOV facility was selected for presentation. ( Interstate 880,
Northbound and Interstate I- 210, Eastbound)
24.8 26.8 28.8 30.8 32.8 34.8
24.8 26.8 28.8 30.8 32.8 34.8
13.5 14.5 15.5 16.5 17.5 18.5 19.5 20.5
KHOV( d), B( d- d0) = 10 collisions/ mile
13.5 14.5 15.5 16.5 17.5 18.5 19.5 20.5
KLeft( d), B( d- d0) = 75 collisions/ mile
Continuous Access Limited Access
I- 880N I- 210E
42
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment R
Cross- Sectional Distribution of Collisions Near Ingress/ Egress Areas
( Interstate 210 Eastbound, Los Angeles County)
28.05 28.1 28.15 28.2 28.25 28.3 28.35 28.4 28.45 28.5 28.55 28.6 28.65 28.7 28.75 28.8 28.85 28.9 28.95 29
2 2 9 2 0 7 8 2 0 4 2 1 0 0 2 0 2 0 0 3
0 2 11 0 2 1 8 8 11 2 2 0 2 3 5 0 0 4 0 2
0 0 3 6 2 8 14 10 4 6 11 0 2 3 4 6 7 4 2 4
0 0 3 4 0 3 2 9 3 0 0 0 0 0 2 2 3 6 0 2
POSTMILE
Ingress/ Egress
Direction of Traffic
OFF Ramp ON Ramp
HOV Lane
Left Lane
Interior Lanes
Right Lane
> 20
15 ~ 20
10 ~ 15
5 ~ 10
0 ~ 5
Collision
Frequency
34.6 34.65 34.7 34.75 34.8 34.85 34.9 34.95 35 35.05 35.1 35.15 35.2 35.25 35.3 35.35 35.4 35.45 35.5 35.55
0 0 0 2 3 0 0 0 0 3 0 4 0 0 2 3 0 2 1 0
4 2 6 6 3 5 2 0 4 2 0 4 4 10 12 11 0 6 0 0
0 5 2 2 6 2 2 5 6 3 8 2 2 4 5 14 2 9 2 6
0 1 2 2 2 1 0 1 0 1 0 2 2 0 6 10 0 1 0 2
POSTMILE
Ingress/ Egress
Direction of Traffic
OFF Ramp ON Ramp
HOV Lane
Left Lane
Interior Lanes
Right Lane
> 20
15 ~ 20
10 ~ 15
5 ~ 10
0 ~ 5
Collision
Frequency
ON Ramp
43
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment S
Cross- Sectional Distribution of Collisions Near Ingress/ Egress Areas
( Interstate 5 Northbound, Orange County)
20.95 21 21.05 21.1 21.15 21.2 21.25 21.3 21.35 21.4 21.45 21.5 21.55 21.6 21.65 21.7 21.75 21.8 21.85 21.9
0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 1 0 0 0 0
0 2 0 2 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1
0 2 1 4 1 4 1 0 0 0 1 0 1 1 0 1 1 0 0 3
0 3 1 0 1 1 0 3 0 0 0 0 0 0 0 0 0 0 0 1
POSTMILE
Ingress/ Egress
Direction of Traffic
ON Ramp OFF Ramp
HOV Lane
Left Lane
Interior Lanes
Right Lane
> 8
6 ~ 8
4 ~ 6
2 ~ 4
0 ~ 2
Collision
Frequency
An HOV flyover is located approximately 0.2 mile upstream from the right most segment
of the graph. The HOV direct connector carries a large proportion of HOV traffic volume
from I- 5 northbound to I- 405 northbound resulting in a relatively small number of HOVs
being expected to use the ingress/ egress area displayed in the graph. The aerial photo
covers the area in the vicinity of HOV flyover, which is displayed in the graph above.
HOV
ingress/ egress
section
HOV fly- over
Mainline connector
between two major
freeways
I‐ 405
I‐ 5
44
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment T
Relationship Between Collisions Per Mile and Distance to Nearest
Entrance/ Exit Ramp in Limited Access HOV Facilities
Three common features which may cause higher collision rates in ingress/ egress areas
were identified by investigating three HCCLs and are as follows: high peak hour HOV
volume, 1000– 1200 vph ( compared to an average of 700– 800 vph in general), short
access distance of a quarter mile, which is the minimum access length, and location
within 0.3 mile of ramps.
45
A Comparative Safety Study of Limited versus
Continuous Access High Occupancy Vehicle ( HOV) Facilities
Attachment U
Types of Total Collisions in HOV Lanes Per Facility
( a) Continuous access ( b) Limited access
46