ISSN 1055- 1425
February 2011
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.
PATH Research Report on Technical Agreement 65A0339
CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
Bicycle Detection and Operational
Concept at Signalized Intersections
Phase 2
UCB- ITS- PRR- 2011- 02
California PATH Research Report
Steven E. Shladover, ZuWhan Kim, Meng Cao,
Ashkan Sharafsaleh, Irene Li, and Scott Johnston
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
i
Bicycle Detection and Operational Concept at Signalized Intersections
Phase 2
PATH Research Report on
Technical Agreement 65A0339
Steven E. Shladover, ZuWhan Kim, Meng Cao, Ashkan Sharafsaleh,
Irene Li and Scott Johnston
DISCLAIMER STATEMENT
This document is disseminated in the interest of information exchange. The contents of
this report reflect the views of the authors who are responsible for the facts and
accuracy of the data presented herein. The contents do not necessarily reflect the
official views or policies of the State of California or the Federal Highway Administration.
This publication does not constitute a standard, specification or regulation. This report
does not constitute an endorsement by the Department of any product described herein.
For individuals with sensory disabilities, this document is available in Braille, large print,
audiocassette, or compact disk. To obtain a copy of this document in one of these
alternate formats, please contact: the Division of Research and Innovation, MS- 83,
California Department of Transportation, P. O. Box 942873, Sacramento, CA 94273-
0001.
Abstract
This project was created as a follow- on to PATH Task Order 6203, to extend the studies
of bicyclist signal timing that were conducted in that project to a wider range of
intersections and traffic signal control scenarios. This work is motivated by the legal
requirement, instituted by the California Legislature, that the road network provide equal
service to bicyclists as it does to motorists. Based on the preliminary findings from Task
Order 6203, Caltrans issued Traffic Operations Policy Directive ( TOPD) No. 09- 06
effective September 10, 2009, including guidance on signal timing to serve bicyclists.
Additional field measurement data on bicyclist intersection crossing behavior were
needed to verify that the preliminary findings from TO 6203 would remain applicable for
more diverse intersections in different parts of the state, with a full range of bicycling
populations and traffic conditions. Furthermore, because questions were raised about the
potentially adverse traffic impacts of providing longer minimum green times on all signal
phases to meet bicyclists’ needs, more extensive traffic simulations were needed to
quantify the traffic impacts of a variety of signal control strategies in coordinated
corridors, where the signal progressions could potentially be disrupted.
Keywords: Bicycling, traffic detection, traffic signal timing.
.
iii
Executive Summary
This project was created as a follow- on to PATH Task Order 6203, to extend the studies
of bicyclist signal timing that were conducted in that project to a wider range of
intersections and traffic signal control scenarios. This work is motivated by the legal
requirement, instituted by the California Legislature, that the road network provide equal
service to bicyclists as it does to motorists. Based on the preliminary findings from Task
Order 6203, Caltrans issued Traffic Operations Policy Directive ( TOPD) No. 09- 06
effective September 10, 2009, including guidance on signal timing to serve bicyclists.
Additional field measurement data on bicyclist intersection crossing behavior were
needed to verify that the preliminary findings from TO 6203 would remain applicable for
more diverse intersections in different parts of the state, with a full range of bicycling
populations and traffic conditions. Furthermore, because questions were raised about the
potentially adverse traffic impacts of providing longer minimum green times on all signal
phases to meet bicyclists’ needs, more extensive traffic simulations were needed to
quantify the traffic impacts of a variety of signal control strategies in coordinated
corridors, where the signal progressions could potentially be disrupted.
The PATH portable video data acquisition system was used to collect data about bicyclist
crossing times and speeds at five new intersections, to complement the data previously
collected at two intersections. The combined data from the seven intersections provide
considerable geographical diversity ( Urban, suburban and rural, including northern and
southern California and Central Valley), diversity of bicycling population ( commuters,
recreational and serious bicyclists), and diversity of intersection size and geometry. The
complete distribution of bicyclist cruising speeds was derived for all seven intersections,
and the start- up timing relative to the onset of the green traffic signal phase was derived
for six of the intersections. These characterizations of bicyclist behavior are expressed in
terms of the complete cumulative distributions, so that a user of the data can choose
which percentile of the bicyclist behavior they want to accommodate in the selection of
signal timing.
The data about bicyclist crossing times show clear influences of several factors that need
to be accounted for in selection of signal timing, in addition to the obvious importance of
street width. Since bicyclists are strongly affected by road grades, it is necessary to allow
additional clearance time for intersections with significant grades on the approaches. In
addition, the demographics and trip purposes of the bicyclists influence their crossing
times. Where there is a significant proportion of recreational bicyclists or families with
children, the crossing times are longer.
The data are compared directly with the timing recommendations in Caltrans TOPD 09-
06, showing that those recommendations appear to be generally suitable for serving the
needs of 85% of the bicycling population ( subject to additional adjustments needed for
intersections with special circumstances such as grades or a significant proportion of
children or recreational bicyclists). This provides confirmation of the validity of those
timing recommendations, but does not provide a sufficiently complete set of data to
iv
support development of a detailed handbook of timing guidelines for all combinations of
conditions.
The signal timing recommendations in TOPD 09- 06 would require increases in the
minimum clearance intervals for wide intersections in California, with minimum green
times significantly longer than the current 4 s minimum. In order to assess the
implications of these changes for vehicular traffic, a detailed traffic simulation was
conducted for a suburban arterial with signal progression. The bicycle- friendly signal
timings were substituted for the current signal timings and traffic was simulated under
moderate ( mid- day) density conditions and low density conditions ( 20% of the mid- day
volumes). In both conditions, the effects on travel speed and delays were negligible,
while the number of stops increased slightly. The same corridor was simulated with the
addition of pedestrian crossing phases, and the results showed that these had a much
larger impact on traffic speed, delay and number of stops than the retiming for bicyclists.
Since our prior research under TO 6203 already showed that the effects of signal retiming
for bicyclists were negligible under peak traffic conditions, it appears to be reasonable to
conclude that there should be no concerns about traffic impacts of implementing TOPD
09- 06, especially when the signal timing is re- optimized after the bicycle minimum times
are included.
The results reported here provide strong support for the application of the signal timing
recommendations in TOPD 09- 06 to accommodate the needs of bicyclists crossing
intersections. This should enable new signals to be timed for bicyclists from the start, as
well as enabling rational re- timing of existing signals. However, additional work will be
needed to produce an authoritative handbook that can provide detailed quantitative
guidance for traffic engineers regarding how to time signals for bicyclists under the full
range of conditions that they will encounter in practice.
v
Table of Contents
Abstract ii
Executive Summary iii
Table of Contents v
List of Figures vi
List of Tables vii
1. Introduction 1
2. Selection of Field Data Collection Sites 2
2.1 Polk at Sutter, San Francisco 3
2.2 Marina at Cervantes, San Francisco 4
2.3 Venice Boulevard at Beethoven, Los Angeles 6
2.4 Laurel Canyon at Chandler, Los Angeles 7
2.5 Davis, Anderson at West 8th Street 8
2.6 Davis, Cowell at Drew 9
2.7 Santa Monica, Main at Marine 10
3. Bicyclist Crossing Time Data Analysis Results 12
3.1 Data for Polk at Sutter in San Francisco 12
3.2 Data for Marina at Cervantes, San Francisco 16
3.3 Data for Venice Blvd. at Beethoven, Los Angeles 17
3.4 Data for Anderson at West 8th Street, Davis 21
3.5 Data for Main at Marine, Santa Monica 25
3.6 Comparisons of data from all observed intersections 29
4. Simulations to Show Traffic Impacts of Increased Minimum Green 33
4.1 Corridor and Scenario Description 33
4.2 Impact under moderate traffic flow conditions 35
4.3 Impact under low traffic flow conditions 37
4.4 Conclusions 40
5. Signal Timing Recommendations 41
References 44
vi
List of Figures
2.1 San Francisco, Polk at Sutter Data Collection Site 4
2.2 San Francisco, Marina at Cervantes Intersection 5
2.3 Pedestrian Interactions with Bicyclists at Marina at Cervantes 6
2.4 Los Angeles, Venice at Beethoven 7
2.5 Laurel Canyon at Chandler, Los Angeles 8
2.6 Anderson at West 8th St., Davis 9
2.7 Cowell Blvd. at Drew Ave., Davis 9
2.8 Main at Marine in Santa Monica 10
3.1 Standing- start bicyclist trajectories on Polk at Sutter 12
3.2 Histograms of rolling start speeds on Polk St. 13
3.3 Cumulative distributions of rolling start speeds on Polk St. 13
3.4 Histograms of standing start offset times at Polk at Sutter 14
3.5 Cumulative distribution of standing start offset times at Polk and Sutter 14
3.6 Histogram of final crossing speeds of standing start bicyclists at Polk and
Sutter 15
3.7 Cumulative distribution of final crossing speeds of standing start bicyclists
at Polk and Sutter 15
3.8 Histograms of crossing speeds for rolling start bicyclists at Marina and
Cervantes 16
3.9 Cumulative distribution of crossing speeds for rolling start bicyclists at
Marina and Cervantes 17
3.10 Trajectories of standing start bicyclists on Venice crossing Beethoven 18
3.11 Histogram of rolling start crossing speeds on Venice at Beethoven 19
3.12 Cumulative distribution of rolling start crossing speeds on Venice at
Beethoven 19
3.13 Histogram of standing start offset time for Venice at Beethoven 20
3.14 Cumulative distribution of standing start offset time for Venice at Beethoven 20
3.15 Histogram of final speeds for standing start bicyclists on Venice Blvd.
crossing Beethoven 21
3.16 Cumulative distribution of final crossing speed for standing start bicyclists on
Venice at Beethoven 21
3.17 Trajectories of southbound standing start bicyclists on Anderson Rd.
crossing West 8th Street 22
3.18 Histograms of rolling start speeds on Anderson Rd. 23
3.19 Cumulative distributions of rolling start speeds on Anderson Rd. 23
3.20 Histogram of standing start offset, southbound, on Anderson Rd. 23
3.21 Cumulative distribution of standing start offset on Anderson Rd.,
southbound 24
3.22 Histogram of final crossing speeds for standing start bicyclists, southbound on
Anderson 24
3.23 Cumulative distribution of final crossing speeds for standing start bicyclists on
southbound Anderson Rd. 25
3.24 Standing start trajectories for southbound crossing of Marine at Main St.
in Santa Monica 26
vii
3.25 Histogram of rolling speeds of bicyclists crossing Marine on Main 26
3.26 Cumulative Distribution of rolling start bicyclist speeds on Main at Marine 27
3.27 Histogram of offset times for standing start bicyclists on Main at Marine 27
3.28 Cumulative distribution of offset times for standing start bicyclists on Main
at Marine 28
3.29 Histogram of final crossing speed for standing start bicyclists on Main at
Marine 28
3.30 Cumulative distribution of final crossing speeds of standing start bicyclists
on Main at Marine 29
3.31 Cumulative distributions of speed observations for rolling starts at the
observed intersections 30
3.32 Cumulative distributions of offset times for standing start intersection
crossings 31
3.33 Cumulative distributions of final speeds for standing start bicyclists 31
4.1 Study Corridor 33
4.2 Intersection Turning Volumes 35
4.3 Mainline through movement green split comparison ( moderate flow
condition) 38
4.4 Mainline through movement green split comparison ( low flow condition) 40
5.1 Crossing Times as a Function of Street Width 42
List of Tables
2.1 Summary of Intersection Characteristics 11
3.1 Key percentiles of observed bicyclist crossing behaviors 32
4.1 Minimum split requirement 34
4.2 Network MOE Comparison ( moderate flow conditions) for different signal
timing scenarios 36
4.3 Comparison of green splits under moderate flow condition 36
4.4 Network MOE Comparison ( low flow condition) 38
4.5 Comparison of green splits under low flow condition 39
1
1. Introduction
This project was created as a follow- on to PATH Task Order 6203, to extend the studies
of bicyclist signal timing that were conducted in that project to a wider range of
intersections and traffic signal control scenarios. This work is motivated by the legal
requirement, instituted by the California Legislature, that the road network provide equal
service to bicyclists as it does to motorists. Based on the preliminary findings from Task
Order 6203, Caltrans issued Traffic Operations Policy Directive ( TOPD) No. 09- 06
effective September 10, 2009, including guidance on signal timing to serve bicyclists.
Additional field measurement data on bicyclist intersection crossing behavior was needed
to verify that the preliminary findings from TO 6203 would remain applicable for more
diverse intersections in different parts of the state, with a full range of bicycling
populations and traffic conditions. Furthermore, because questions were raised about the
potentially adverse traffic impacts of providing longer minimum green times on all signal
phases to meet bicyclists’ needs, more extensive traffic simulations were needed to
quantify the traffic impacts of a variety of signal control strategies in coordinated
corridors, where the signal progressions could potentially be disrupted.
TOPD 09- 06 specified that the signal timing should be based on an assumed bicyclist
cruising speed of 10 mph and an additional start- up time for standing starts of 6 seconds.
This was used to calculate the sum of the minimum green interval, yellow interval and
red clearance interval for signal controllers as a function of the intersection width ( where
that was defined based on the distance from the limit line to the far side of the last
conflicting lane, plus 6 feet for the length of the bicycle). The results were tabulated in a
table for widths from 40 feet to 180 feet in increments of 10 feet, producing required
minimum phase lengths ranging from 9.1 to 18.7 seconds. Since the default minimum
green time in California has been 4 seconds, this is likely to lead to significant increases
in some minimum green times, especially for wider intersections.
The promulgation of TOPD 09- 06 generated controversy among local traffic engineers in
California, leading to an alternate proposal by the City of Vacaville that was supported by
Orange County and several other jurisdictions. These traffic engineers were concerned
that the increased minimum green time requirement would produce adverse traffic
impacts in several ways: depriving large, heavily traveled arterials of green time in order
to serve smaller cross- streets with light traffic, not only during peak periods but also off
peak; requiring excessive green times for left turning phases at large intersections where
bicyclists rarely if ever make left turns; and requiring longer total cycle times to serve all
phases at large 8- phase intersections, where the minimum green time would have to be
increased for every phase. Vacaville proposed that the bicycle signal timing requirement
be based on a 15 mph cruising speed plus a 1 second perception- reaction time and the
time needed to accelerate to the cruise speed at a rate of 3 ft/ s/ s ( about 0.1 g). They also
suggested an option for intersections with a high proportion of young bicyclists, reducing
the cruising speed to 10 mph and the acceleration rate to 1.5 ft/ s/ s ( about 0.05 g).
2
2. Selection of Field Data Collection Sites
The original field data collection reported in the final report on PATH Task Order 6203
( PATH Research Report PRR- 2009- 37) was conducted at two intersections in Palo Alto
and Berkeley. When these results were reported to the California Traffic Control Devices
Committee ( CTCDC), they indicated the need to see data from a wider range of
intersections that would not just be in Bay Area suburban university towns, but would
represent more of the state. This meant that it was necessary to include data from
Southern California, the rural Central Valley, and at least one of the major metropolises
( Los Angeles or San Francisco). Therefore, sites meeting these criteria were sought for
the new data collection in this project. The project staff contacted traffic engineers and
bicycle coordinators in San Francisco, Los Angeles, Long Beach, Davis, Vacaville, and
Santa Monica to identify promising intersections that have a high volume of bicycle
traffic and a wide range of other important characteristics that could affect bicyclist
crossing times and speeds:
- bicyclist demographics ( young adult, mature adult, child)
- bicycling trip purposes ( commuting vs. recreational)
- local traffic conditions ( density and speed, especially on the cross street)
- intersection geometry ( approach widths and grades, crown on cross street).
The candidate intersections that were recommended for our consideration were as listed
below, and the places where we actually collected data are indicated in boldface:
San Francisco:
Polk at Sutter – bike lane with strong commute bicycling and significant grade
Marina at Cervantes – high volume of recreational and family bicyclists
Market at Valencia – large intersection with heavy left- turning commute bicycling
Church at Market – wide intersection with heavy commute bicycling traffic
Market and 5th Streets – heavy bicyclist commute volumes, but no good place to park the
data collection system
Los Angeles:
Venice at Beethoven – bicycle lane serving diverse and leisure bicyclists
Laurel Canyon at Chandler – extremely wide ( 180 ft), diverse population
Reseda Blvd / Oxnard St. – Near dedicated busway but not enough bicyclists
Balboa / Victory -- large intersection but not enough bicyclists
Van Nuys Blvd / Oxnard Blvd – Large intersection, but no bicycle lane or bicyclists
Chandler Blvd / Vineland – Adjacent intersection is very close; thus the collected data
would not be representative
Sunset Blvd. / Silverlake Blvd. ( Parkman Ave.) - Silverlake and Sunset do not meet but
they are connected through a downgrade ramp, thus difficult to park/ observe
Sunset Blvd./ Griffith Park Blvd. ( Maltman Ave.) - three roads ( Sunset, Griffith Park, and
Maltman ) meet in a non- typical way
Sunset Blvd. / Hyperion Ave. - the two roads meet with an angle and one side of
Hyperion is a high grade uphill.
3
Sunset Blvd. / Santa Monica Blvd. ( Sanborn Ave.) - Sunset, Santa Monica, and Sanborn
meet in a non- typical way and the South side of Sanborn is uphill.
Venice Blvd / Sepulveda Blvd. – large intersection with bicycle lane, but no good place
to park the data collection system
Venice Blvd / McLaughlin – bicycle lane serving diverse and leisure bicyclists
Venice Blvd / Inglewood Blvd. – bicycle lane serving diverse and leisure bicyclists
Venice Blvd / Centinela Blvd. – bicycle lane serving diverse and leisure bicyclists
Paseo del Mar / Weymouth or Patton or Gaffey - Paseo del Mar is an ocean- side scenic
drive road next to many parks. All three are T- intersections with no traffic signals.
Rose / Pacific – new bike lane
Davis:
Anderson at W. 8th St. – high volume of college student bicyclists, also expecting many
teen bicyclists because of nearby middle school
Cowell at Drew – high volume of college student bicyclists
Villanova at Anderson Road – high volume of college student bicyclists, also expecting
many teen bicyclists because of nearby middle school
Sycamore at Covell Boulevard – large intersection with bike lanes with college and
family bicyclists, but no good place to park the data collection system
F Street and E. 14th Street – T- intersection with bike lanes, high volume of teen
bicyclists because of nearby middle school and high school
Arlington at Shasta Drive – T- intersection with bike lanes, near a park with very young
bicyclists
Santa Monica:
Main Street at Marine – complicated urban traffic, mixed bicycling population, unusual
intersection geometry producing wide range of starting positions for crossing bicyclists.
Main Street at Hill
Main Street at Ashland
San Vicente Boulevard at 7th Street
Broadway at 7th, 11th or 17th Streets
Ocean Avenue at Colorado Avenue
California Street and Ocean Avenue – left turning bicyclists
Ocean Park Boulevard and Main Street – left turning bicyclists
The seven intersections where we collected bicyclist crossing data are described below.
2.1 Polk at Sutter St., San Francisco
This intersection was chosen because it is in a high- density urban setting with a
reasonably high volume of commuter bicyclists of diverse age and vigor and a significant
grade on the approaches ( 4.5%). The intersection itself is flat, despite the grade on the
approaches, and the cross- street ( Sutter) is one way, which simplifies the bicyclists’
responsibility to check the cross traffic status before proceeding into the intersection.
They also have very good visibility of the cross traffic. These factors are the likely
reasons that 60% of the standing start bicyclists at this intersection did not even wait for
4
the green signal, but started moving prior to the green onset. The crossing distance of 58
ft. was measured from the stop bar on the starting side of the intersection to the curb line
on the opposite side ( equivalent to the front edge of the pedestrian crosswalk).
The Google Earth view is shown in Figure 2.1, indicating the location of the data
collection trailer and video cameras, where they provided visibility of bicyclists traveling
in both directions along Polk St. The cross street, Sutter, has three lanes of heavy one-way
traffic, with a posted speed limit of 25 mph, and parked cars on both sides.
Figure 2.1 San Francisco, Polk at Sutter Data Collection Site
2.2 Marina at Cervantes, San Francisco
This site was chosen to get recreational bicyclists of diverse demographics, especially
including families with children, because of its location in the tourist- heavy Marina
district of San Francisco. Indeed, one Saturday of observation time yielded a large
number of bicyclist samples, although many of them could not be tracked effectively
because they were surrounded by high density pedestrian traffic in the crosswalk. In
some of these cases, the pedestrian density was so high that it impeded the bicyclists’
movements and would have corrupted the data – in these scenarios it is reasonable to
assume that a pedestrian call would have been issued to the signal controller and the
bicyclists would not be depending on a vehicle detector based actuation. In other cases,
where the pedestrian traffic provided only limited interference with the bicyclists and/ or
5
the pedestrians were running rather than walking, the data were retained for analysis.
This intersection and its approaches are flat and the cross traffic is slow and benign
( entering and leaving the waterfront parking lot).
The Google Earth view of this intersection is shown in Figure 2.2, indicating the location
of the data collection system and its view of the bicycle traffic. Figure 2.3 provides
examples of the video data in a scenario with pedestrian congestion impeding bicycle
movement ( red circled bicyclists in right- hand image) and with pedestrian density low
enough that the bicycle timing data were judged to be valid and useful for this study ( blue
circled bicyclists in left- hand image and outside the pedestrian crossing in right- hand
image). The video observations of the traffic signal were troublesome at this intersection,
and in some cases it was not possible to distinguish the green onset time. This limited the
number of samples for which we could estimate the start- up offset time.
Figure 2.2 San Francisco, Marina at Cervantes Intersection
6
Figure 2.3 Pedestrian Interactions with Bicyclists at Marina at Cervantes: Acceptable
interference for valid data ( blue circles) and unacceptable interference for valid data ( red
circles)
2.3 Venice Boulevard at Beethoven, Los Angeles
Venice Boulevard was recommended by the City of Los Angeles because of its bicycle
lanes and an expected high volume of bicyclists. We were also expecting to get a good
percentage of school children because of a nearby school and of recreational bicyclists
accessing Venice Beach. However, the bicyclists we observed here were actually the
strong, hardy young adult commuters. We believe that this is because this is an
intimidating route for bicyclists, with fast and aggressive vehicular traffic along Venice
Blvd. and relatively long distances to travel to get to and from origins and destinations of
interest. The intersection and its approaches are flat.
The Google Earth view of this intersection is shown in Figure 2.4, indicating the data
collection van location and our view of the eastbound bicyclists along Venice Blvd.
7
Figure 2.4 Los Angeles, Venice at Beethoven
2.4 Laurel Canyon at Chandler, Los Angeles
Laurel Canyon was recommended by the City of Los Angeles because of its bicycle
lanes, and the intersection at Chandler was particularly interesting because of its great
width ( 180 ft), which would allow us to get a data point for one of the widest streets we
are likely to encounter in California. Unfortunately, the bicycle traffic at this intersection
was extremely low, and after more than a full day of observation we were only able to
observe 36 standing start bicyclists and 18 rolling start bicyclists. Since it would be
necessary to have many more samples than this in order to support any statistically valid
analysis, we determined that we could not justify the large additional investment of time
and effort that would have been needed to obtain a usable data set at this intersection.
8
Figure 2.5 Laurel Canyon at Chandler, Los Angeles
2.5 Davis, Anderson at West 8th Street
It was very difficult to find locations with high bicyclist volumes in the rural Central
Valley except in Davis, which is a well- known bicycling Mecca. So, we contacted the
City of Davis for recommended locations. We were particularly interested in locations
where we could collect data on school children bicycling to and from school, to
understand how different their timing needs are from those of adults. We chose this
intersection because of its proximity to an elementary and a middle school, but in the end
the bicyclists that we observed were predominantly U. C. Davis students going to and
from the campus rather than school children. There was a strong commute pattern,
southbound in the morning and northbound in the afternoon, requiring slightly different
alignment of the video cameras as shown in Figure 2.6. This intersection and its
approaches are flat. The width of the crossing is 60 feet, representing three lanes of
traffic ( two though lanes, one in each direction, and a left turn lane), plus residential
parking along the curbs. The speed limit is posted at 30 mph, with very light cross traffic
and excellent visibility of the cross traffic by the bicyclists.
9
Figure 2.6 Anderson at West 8th St, Davis
2.6 Davis, Cowell at Drew
The physical characteristics and bicycling population at this intersection turned out to be
very similar to those at Anderson at West 8th Street, but we had a lower volume of
bicyclists here and could only observe one direction of travel. In order to conserve
project resources, we decided to defer processing this set of data until we had a
sufficiently diverse collection of data sets from the other sites, to make sure that we
would be able to capture the widest possible variety of bicyclist crossing scenarios. This
intersection is seen in Figure 2.7.
Figure 2.7 Cowell Blvd. at Drew Ave., Davis
10
2.7 Santa Monica, Main at Marine
This intersection provided us with a high- density urban setting in Southern California,
with complicated traffic patterns and a diverse mix of bicyclists. Because of the unusual
geometry of the intersection, with an offset side street, bicyclists tended to stop at a wide
variety of locations within the intersection rather than all stopping near the stop line. The
traffic density and speed were moderate and the intersection flat.
Figure 2.8 Main at Marine in Santa Monica
The characteristics of the data collection sites are summarized in Table 2.1 below.
11
Table 2.1 Summary of Intersection Characteristics
Palo Alto
Park at El
Camino
Berkeley
Russell at
Telegraph
Davis
Anderson
at West 8th
S. F.
Polk at
Sutter
S. F.
Marina at
Cervantes
Los Angeles
Venice at
Beethoven
Santa Monica
Main at Marine
Width
Traffic lanes
125 ft,
7 lanes
84 ft,
4 lanes
60 ft,
3 lanes
58 ft,
3 lanes
63 ft,
4 lanes
63 ft.
2 lanes
48 to 84 ft.
2 lanes
Speed
Limit
40 mph 25 mph 30 mph 25 mph 25 mph 25 mph 25 mph
Cross
traffic
Heavy Moderate Very Light One- way,
heavy
Light
( Driveway)
Very Light Moderate
Intersection Crowned Flat Flat Flat Flat Flat Flat
Visibility Limited Better Best Best Best Very good Depends on
starting point
Approach
grades
Flat - 3.4%,
+ 2.5%
Flat +/- 4.5% Flat Flat Flat
Bike traffic Evening
commute
All day Commute All day ( Weekend)
Recreation
All day All day
Bicyclists Young
adults
Diverse College
students
Diverse Tourists,
families
Half experts Mix of tourists
and experts
12
3. Bicyclist Crossing Time Data Analysis Results
The video images of the bicyclists crossing the intersections were analyzed using the
method that was already described in the technical report on our previous project, UCB-ITS-
PRR- 2009- 37. The trajectories were extracted from the video sequences using the
video tracker software and these trajectories were then characterized in terms of their
slopes ( representing cruising speed) and the offset time from the green onset until the
cruising- speed slope intersected the starting location. This provided for two parameters
to fully characterize standing- start crossings and one parameter for rolling- start crossings.
The data for each intersection are first presented individually, and are then combined so
that the similarities and contrasts can be seen.
3.1 Data for Polk at Sutter in San Francisco
At this intersection, we collected data on 54 and 43 standing starts in the two directions
and 217 and 270 rolling starts in the two directions of travel during two days of
observations. Because of the strong grade along Polk St. ( about 4.5%) there was a
significant difference in the speeds of the rolling start bicyclists in the two directions.
The signal timing along Polk St. favored bicyclists rolling through on the green, and
relatively few bicyclists had to stop for the signal. The numbers of bicyclists in each
direction was too small to produce a good statistical distribution, but fortunately the
intersection itself is flat so there is no significant difference between the northbound and
southbound standing start bicycling, and it was possible to combine the data for both
directions to produce a single distribution. The standing start trajectories for the two
directions of travel are shown in Figure 3.1.
Figure 3.1 Standing- start bicyclist trajectories on Polk at Sutter, Northbound on left and
Southbound on right
13
The red and orange profiles superimposed on these trajectories represent the formulas
that were recommended by the City of Vacaville for adult bicyclists ( red) and child
bicyclists ( orange) respectively. Although the Vacaville formula for children would
serve most of the adult bicyclists at this site, the formula for adults would only serve the
fastest half of this bicycling population. Note the wide range of starting locations for
these bicyclists, who had to contend with vehicle traffic and parked vehicles on this
crowded street and could not always stop right at the stop line.
The contrasts in the rolling start results reflect the strong grade on Polk Street. Figure 3.2
shows the histograms of the rolling speeds in the two directions and Figure 3.3 shows the
cumulative distributions.
Figure 3.2 Histograms of rolling start speeds on Polk St., northbound on left and
southbound on right
Figure 3.3 Cumulative distributions of rolling start speeds on Polk St., northbound on
left and southbound on right
Because there were only a limited number of standing starts, and the direction of travel
did not appear to have a significant impact on bicyclist behavior, the data for northbound
14
and southbound standing start bicyclists were combined into a single dataset for analysis.
The histogram of standing start offset times is shown in Figure 3.4 and the cumulative
distribution is in Figure 3.5.
Figure 3.4 Histogram of standing start offset times at Polk at Sutter
Figure 3.5 Cumulative distribution of standing start offset times at Polk and Sutter
15
The final crossing speeds for the standing start bicyclists at Polk and Sutter are depicted
in the histogram of Figure 3.6 and the cumulative distribution of Figure 3.7. These show
that we found a few very fast, sporty bicyclists here, but they are far removed from the
large majority of the bicyclists. These speeds are comparable to the cruising speeds of
the uphill rolling start bicyclists at this intersection.
Figure 3.6 Histogram of final crossing speeds of standing start bicyclists at Polk and
Sutter
Figure 3.7 Cumulative distribution of final crossing speeds of standing start bicyclists at
Polk and Sutter
16
3.2 Data for Marina at Cervantes, San Francisco
The data at this intersection covered both directions of travel, eastbound and westbound,
in a location dominated by recreational bicyclists on a Saturday. This location had the
highest density of bicyclist traffic of any of the sampled locations, and in some cases the
density was so high that it was hard to distinguish individual bicyclists moving in
clusters. The pedestrian traffic at this location was so dense that in some cases it
impeded the motions of the bicyclists, so these data samples were not analyzed because
they are not relevant for determining the crossing times of bicyclists who need to actuate
green cycles through detection systems ( in these cases, pedestrian calls are going to
determine the selection of minimum green times).
The processed data for this intersection cover the speeds of the rolling start crossing
maneuvers ( 107 westbound and 64 eastbound), but not the standing starts. Unfortunately
the video imagery of the traffic signal status was not good enough to enable
determination of the phase changes, which made it impossible to identify the offset times
of the standing start bicyclists.
Figure 3.8 shows the histograms of the eastbound and westbound rolling start crossing
speeds at this intersection. The cumulative distributions of these speeds are shown in
Figure 3.9. Even though the shapes of the histograms look quite different from each
other at this level of aggregation, when we consider the full data set in the cumulative
distribution we can see that the key percentile values are really quite similar for both
directions of travel. At the median and lower percentiles, the speeds are very similar for
both directions. The upper tail of the westbound distribution shows higher speeds
because this included the bicyclists who rode in the curb lane of Marina Blvd in that
direction, not only the bicyclists who used the pedestrian crossing.
Figure 3.8 Histograms of crossing speeds for rolling start bicyclists at Marina and
Cervantes, eastbound and westbound directions respectively
17
Figure 3.9 Cumulative Distribution of Crossing Speeds for Rolling Start Bicyclists at
Marina and Cervantes
These bicyclist speeds are significantly slower than the rolling start speeds observed at
the other intersections, including the intersections with significant positive grades. This
shows the significance of the bicycling population and trip purpose for bicyclist speeds.
This location was the one location with a strong recreational flavor and with a higher
proportion of families and children among the bicyclist population, indicating that the
bicyclist signal timing needs to be adjusted based on factors such as these.
3.3 Data for Venice Blvd. at Beethoven, Los Angeles
At this intersection, we collected data for westbound bicyclists, primarily in the bicycle
lane on Venice Blvd., as they crossed Beethoven. Over two days of observation, we
captured usable data for 79 standing start and 171 rolling start bicyclists, with a very
diverse bicycling population including serious cyclists ( about 50%), commuters, tourists
and high school students ( about 10%). The high proportion of serious cyclists is
probably associated with the fact that this is a relatively intimidating bicycling
environment, with very fast vehicle traffic along Venice Blvd.
Westbound
Eastbound
18
The intersection is flat, with a width of 63 feet for the crossing of Beethoven, and the
bicyclists have very good visibility of the cross traffic, so they do not need to build in
extra margins for dealing with uncertainty about the cross traffic. The trajectories of the
standing start bicyclists at this intersection are shown in Figure 3.10.
Figure 3.10 Trajectories of standing start bicyclists on Venice crossing Beethoven
The speeds of the rolling start bicyclists are shown in the histogram of Figure 3.11 and
the cumulative distribution of Figure 3.12.
19
Figure 3.11 Histogram of rolling start crossing speeds on Venice at Beethoven
Figure 3.12 Cumulative distribution of rolling start crossing speeds on Venice at
Beethoven
The standing start bicyclist crossings are characterized by their offset times and final
crossing speeds. The offset time histogram is shown in Figure 3.13 and its cumulative
distribution is in Figure 3.14.
20
Figure 3.13 Histogram of standing start offset time for Venice at Beethoven
Figure 3.14 Cumulative distribution of standing start offset time for Venice at Beethoven
The final crossing speeds for the standing start bicyclists on Venice at Beethoven are
shown in the histogram of Figure 3.15 and the cumulative distribution of Figure 3.16.
21
Figure 3.15 Histogram of final speeds for standing start bicyclists on Venice Blvd.
crossing Beethoven
Figure 3.16 Cumulative distribution of final crossing speed for standing start bicyclists
on Venice at Beethoven
3.4 Data for Anderson at West 8th Street, Davis
22
This intersection, in a residential area of Davis, had very heavy bicyclist traffic.
Although we were hoping to observe many school children using their bicycles here, the
bicycling population was dominated by U. C. Davis students commuting to and from
classes. The volume of bicyclists was high enough and the flow was sufficiently
directional based on the start and end of the school day that it was possible to distinguish
differences between the morning and evening commute pattern bicycling trips. In two
days of observations, we recorded 426 southbound rolling start crossings and 266
southbound standing start crossings ( morning commute direction). In the northbound
direction, we added another 161 rolling start crossings but did not have enough standing
start crossings to do a separate analysis for this direction of travel.
Figure 3.17 Trajectories of southbound standing start bicyclists on Anderson Rd.
crossing West 8th Street
The histograms of the rolling start bicyclist speeds in both directions along Anderson at
West 8th Street are shown in Figure 3.18, and the cumulative distributions of these speeds
are shown in Figure 3.19. Although the population of bicyclists is largely the same
( university students) and the traffic conditions similar, the northbound speeds are
noticeably higher. The best explanation we can find for this is that the southbound trips
were morning rides toward the U. C. Davis campus and the northbound trips were
afternoon rides back home, when the riders were more eager to reach their destinations.
For the southbound standing start bicyclists, the histogram of starting offset times is
shown in Figure 3.20 and their cumulative distribution is in Figure 3.21. The final rolling
speeds for these bicyclists are characterized by the histogram of Figure 3.22 and the
cumulative distribution of Figure 3.23.
23
Figure 3.18 Histograms of Rolling start speeds on Anderson Rd., southbound ( morning)
on left and northbound ( afternoon) on right.
Figure 3.19 Cumulative distributions of rolling start speeds on Anderson Rd.,
southbound ( morning) on left and northbound ( afternoon) on right
Figure 3.20 Histogram of standing start offset, southbound, on Anderson Rd.
24
Figure 3.21 Cumulative distribution of standing start offset on Anderson Rd.,
southbound
Figure 3.22 Histogram of final crossing speeds for standing start bicyclists, southbound
on Anderson
25
Figure 3.23 Cumulative distribution of final crossing speeds for standing start bicyclists
on southbound Anderson Rd.
3.5 Data for Main at Marine, Santa Monica
The width of the crossing of Marine could be considered to range from 48 feet to 84 feet,
depending on whether the bicyclist starts at the stop line behind the pedestrian crossing or
at the curb line where the cross traffic passes. This is in a busy commercial area, two
blocks from the beach, with moderate cross traffic on Marine. The bicyclists include
tourists ( about 40%), serious cyclists ( about 40%), and commuters. The visibility of
cross traffic for bicyclists depends on the starting location. The signals along Main Street
seem well suited for bicyclists, generally keeping them moving smoothly. This means
we observed many more rolling bikes than standing start bikes at this intersection. We
also observed a lot of semi- rolling and early start bikes, anticipating the signal change. In
total, we recorded usable data on 79 standing start bikes and 240 rolling bikes in three
days of observations.
The trajectories of the standing start bikes are plotted in Figure 3.24, which shows the
wide range of starting positions of the bicyclists here. This diversity of starting positions
( and therefore of crossing width) made it impossible to characterize this intersection with
a single value of width for purposes of data summarization.
26
Figure 3.24 Standing start trajectories for southbound crossing of Marine on Main St. in
Santa Monica
The rolling start bicyclists are characterized by the histogram and cumulative distribution
plot of their cruising speeds, as shown in Figures 3.25 and 3.26.
Figure 3.25 Histogram of rolling speeds of bicyclists crossing Marine on Main
27
Figure 3.26 Cumulative Distribution of rolling start bicyclist speeds on Main at Marine
The standing starts are characterized by their offset times and final cruising speeds. The
offset time histogram is shown in Figure 3.27 and its cumulative distribution is in Figure
3.28. One bicyclist distracted by a conversation during a signal change accounted for the
single extremely long offset time sample.
Figure 3.27 Histogram of offset times for standing start bicyclists on Main at Marine
28
Figure 3.28 Cumulative distribution of offset times for standing start bicyclists on Main
at Marine
The final cruising speeds of the standing start bicyclists are shown in the histogram and
cumulative distribution of Figures 3.29 and 3.30.
Figure 3.29 Histogram of final crossing speed for standing start bicyclists on Main at
Marine
29
Figure 3.30 Cumulative distribution of final crossing speeds of standing start bicyclists
on Main at Marine.
3.6 Comparisons of data from all observed intersections
The relationships between bicycling behavior and the characteristics of the intersections
only become apparent when the data from the different intersections are plotted together,
so in this section we combine the cumulative distribution plots from all the intersections
that had full data sets. This begins with the cruising speed for the rolling starts, which is
the simplest parameter to compare, as plotted in Figure 3.31.
It is clear from Figure 3.31 that two of the three slowest cruising speeds are for the uphill
bicyclists in San Francisco and Berkeley and two of the three fastest cruising speeds are
for the downhill bicyclists at the same intersections, so the strong effect of grade is
obvious. The slowest cruising speeds of all, at the slow tail of the distribution, are for the
family recreational bicyclists using a pedestrian crossing along Marina Blvd. in San
Francisco, indicating the importance of accounting for the local bicycling population and
peculiarities of the crossing. In contrast, the other fast speed distribution is for the
vigorous young adults leaving the Stanford campus during the evening commute period.
The bicyclists at the flat intersections in Davis and the Los Angeles area were clustered in
the middle. The more recreationally oriented bicyclists in the heavier traffic of Santa
Monica were somewhat slower than the U. C. Davis students in their low- density
residential area, and as previously observed the Davis students going home in the evening
were somewhat faster than they were heading toward the campus in the morning.
30
Based on these data, it looks reasonable to assume a 50% ile cruising speed of about 12
mph at flat intersections, with a 20% ile of about 10 mph and a 10% ile of about 8 mph.
These values need to be reduced where there is a significant grade and where the
bicycling population is weighted toward recreational bicyclists and/ or families with
children, or where the bicyclists must use a pedestrian crossing.
Figure 3.31 Cumulative distributions of speed observations for rolling starts at the
observed intersections
The cumulative distributions of the offset times for the standing starts are plotted in
Figure 3.32. For the offset times, the critical parts of the distributions are the upper
percentiles, to ensure that signal timings can accommodate most of the population.
The offset time data for most of the intersections are relatively tightly clustered, with 80th
percentile values around 4 seconds and 90th percentile values around 5 seconds. The
outlier for offset times is Park Blvd. at El Camino Real in Palo Alto, where the offset
times are exceptionally long ( despite the youthful, vigorous population of bicyclists)
because of three factors – limited visibility of the cross traffic, extremely fast and
dangerous cross traffic requiring great caution on the part of the bicyclists, and a steep
crown on El Camino making the acceleration more difficult than at most intersections.
Eastbound Russell St. at Telegraph in Berkeley also had longer high percentile offset
times than most intersections, again because of a visibility issue. In this case, there is a
bus stop near the corner, so when a bus is stopped there it blocks the bicyclists’ view of
the approaching cross traffic and makes the start- up more difficult.
The third distribution of interest describes the final crossing speed for the standing- start
crossings, when the bicyclists have reached a constant speed after accelerating from a
31
stop, as shown in Figure 3.33. This plot shows a remarkably diverse set of results across
the sampled intersections.
Figure 3.32 Cumulative distributions of offset times for standing start intersection
crossings
Figure 3.33 Cumulative distributions of final speeds for standing start bicyclists.
Park Ave. at El Camino Real was again the outlier, but in this case on fast side rather than
the slow side. There are several reasons that the final speeds observed here were much
higher than at any of the other intersections:
32
- these bicyclists were vigorous young adults in a hurry to get home at the end of
the work day;
- they are crossing the widest street of any of the intersections for which we have
data, which allows more time to accelerate up to a higher cruising speed within
the observation range;
- the cross street has a strong crown profile, which means that after the bicyclists
reach the mid- point of the street they are on a negative slope, which helps them
accelerate to a higher speed. ( When the data were re- analyzed based on the
bicyclist speeds at the midpoint of their crossing of El Camino Real they were
much closer to the distributions for the other intersections.)
The intersection of Russell at Telegraph had the second- highest speeds across most the
cumulative distribution. It is no coincidence that this was the second- widest street where
we collected data, so the street width appears to be particularly significant to this
distribution. The intersections at Beethoven, Polk and Anderson were all in the range of
60 feet wide, while the intersection at Marine varied from 48 to 84 feet wide, depending
on where the bicyclists actually started their crossing.
The key percentiles of the observed bicyclist crossing behaviors observed at the selected
intersections were as tabulated in Table 3.1 below:
Table 3.1 Key Percentiles of Observed Bicyclist Crossing Behaviors
% ile accommodated Start- Up
Offset Time
Final Speed from
Standing Starts
Constant Rolling
Speed ( Roll thru)
90% Park Blvd., Palo Alto
90% Russell St., Berkeley
90% Anderson Rd., Davis
90% Polk St., S. F.
90% Venice Blvd., L. A.
90% Main St., Santa Monica
90% Marina Blvd., S. F.
8.1 s
6.0 s
5.6 s
4.8 s
4.5 s
5.2 s
--
11.5 mph ( 10% ile)
8.2 down, 7.6 up
6.4 mph
7.8 mph
6.1 mph
6.7 mph
--
10.0 mph ( 10% ile)
9.3 down, 6.8 up
8.1 ( AM) 8.9 ( PM)
11.5 down, 7.3 up
8.6 mph
7.1 mph
5.4 mph
80% Park Blvd., Palo Alto
80% Russell St., Berkeley
80% Anderson Rd., Davis
80% Polk St., S. F.
80% Venice Blvd., L. A.
80% Main St., Santa Monica
80% Marina Blvd., S. F.
7.0 s
5.0 s
4.9 s
4.35 s
3.7 s
4.2 s
--
12.3 mph ( 20% ile)
8.8 down, 8.4 up
7.0 mph
8.3 mph
6.8 mph
8.9 mph
--
10.6 mph ( 20% ile)
10.5 down, 7.8 up
9.3 ( AM) 10.4 ( PM)
13.4 down, 8.4 up
10.0 mph
8.3 mph
6.3 mph
50% Park Blvd., Palo Alto
50% Russell St., Berkeley
50% Anderson Rd., Davis
50% Polk St., S. F.
50% Venice Blvd., L. A.
50% Main St., Santa Monica
50% Marina Blvd., S. F.
5.5 s
3.7 s
3.8 s
3.5 s
3.1 s
2.8 s
--
14.2 mph
10.4 down, 9.8 up
8.2 mph
9.3 mph
8.0 mph
9.1 mph
--
14.1 mph
12.8 down, 10.0 up
11.6 ( AM) 13.2 ( PM)
16.3 down, 10.0 up
12.5 mph
10.5 mph
8.8 mph
33
4. Simulations to Show Traffic Impacts of Increased Minimum Green
In this chapter, we use an example corridor to study and discuss the impact of reflecting
bicycle green time requirements in signal timing. Since such impact has previously been
shown to be negligible under congested traffic conditions, the focus of this section is on
the impact under low to medium flow conditions.
4.1 Corridor and Scenario Description
The study corridor is Bouquet Canyon Road in Santa Clarita, California. The SYNCHRO
model files for this corridor are completely coded with road geometry and turning
volumes, as well as signal timing information. The 3- mile study section is between
Lowes and Plum Canyon Road along Bouquet Canyon Road with twelve signalized
intersections ( See Figure 4.1). The cycle length along the corridor is 120 seconds and
most intersections have a pedestrian phase available if called for.
The corridor uses different timing plans for morning peak, afternoon peak, and mid- day
traffic. For purposes of this study, we start with the mid- day traffic as the moderate flow
condition, which is significantly lower than the AM/ PM peak volumes; and for the low
flow condition we further reduce the mid- day volume significantly.
Figure 4.1: Study Corridor
34
SYNCHRO is used in this study to test the impact of minimum bicycle clearance time
requirement. Table 4.1 shows the minimum split requirements for the different scenarios
that were tested. Note that since it is typically bicycles traveling on a side street that
would require a longer than usual minimum green to cross the major arterial, Table 4.1
lists minimum green requirements only for the phases that serve the side streets.
Scenarios
In Scenario 1, the signal timing is chosen based entirely on serving the vehicle traffic
along this corridor, without regard to bicyclists or pedestrians. Thus, in Scenario 1,
Minimum split = Minimum initial + Yellow + All red
The signal timing splits and offsets are optimized and network wide MOEs are recorded
( Table 4.2).
Then, in Scenario 2, the minimum green is set to reflect the bicycle green time
requirement ( from Caltrans document TOPD 09- 06) based on the distance that a bicycle
needs to clear to cross the intersection safely. In this scenario,
Minimum split = Minimum initial for bicycle + Yellow + All red
Network MOEs for Scenario 2 are also recorded, where Scenario 2a reports MOEs under
un- optimized timing plans and Scenario 2b represents the network running re- optimized
signal timing plans.
Scenario 3 represents the corridor when there are significant pedestrian volumes and the
pedestrians are requesting pedestrian crossing phases, so that these become the minimum
split:
Minimum split = Pedestrian walk + Flash don’t walk + Yellow + All red
Scenario 4a adds bicycle minimum green requirement on top of Scenario 3 and Scenario
4b demonstrates the impact with a re- optimized timing plan.
Table 4.1. Minimum split requirement
Intersection No. of
Lanes
Auto
( seconds)
Auto+ bicycle
( seconds)
Auto+ ped
( seconds)
Auto+ ped+ bicycle
( seconds)
Lowes * 10 12 14.6 12 * 14.6
Newhall
Ranch
13 11 17.3 39 39
Best Buy * 10 12 14.6 12 * 14.6
Espuelle 10 8.5 14.6 39.5 39.5
35
Seco Canyon 7 15 11.9 35 35
Alamogordo 6 8.5 11.2 33.5 33.5
Central Park 6 8.5 11.2 34.5 34.5
Centurion 6 12 11.2 22 22
Haskell 9 9 13.9 33 33
Urbandale 8 8.5 13.2 33.5 33.5
Wellston 8 10.5 13.2 33.5 33.5
Plum Canyon 8 13 13.2 33 33
Note: The intersections noted with a * do not have a pedestrian phase accompanying the
side street phases.
4.2 Impact under moderate traffic flow conditions
As stated in Section 4.1, the moderate flow condition represents the mid- day network
demand, as shown in Figure 4.2 below.
Figure 4.2: Intersection Turning Volumes
36
With moderate demand, the network performance MOEs for each scenario are shown in
Table 4.2.
Table 4.2. Network MOE Comparison ( moderate flow condition) for different signal
timing scenarios
Scenario
1
Scenario
2a
Scenario
2b
Scenario
3
Scenario
4a
Scenario
4b
Total Delay ( hr) 124 124 124 168 169 169
Number of Stops 13153 13601 13807 14218 14322 14711
Average Speed ( mph) 28 28 28 25 25 25
Total Travel Time
( hr)
320 320 320 364 365 365
Distance Traveled
( mile)
9100 9100 9100 9100 9100 9100
Note:
Scenario 1 – Auto only;
Scenario 2a – Auto+ Bicycle;
Scenario 2b – Auto+ Bicycle, signal timing re- optimized;
Scenario 3 – Auto+ Pedestrian;
Scenario 4a – Auto+ Pedestrian+ Bicycle
Scenario 4b – Auto+ Pedestrian+ Bicycle, signal timing re- optimized.
As shown in Table 4.2, the network wide MOEs are very similar among Scenarios 1, 2a,
and 2b, and among Scenarios 3, 4a, and 4b. Adding the bicycle green requirement ( going
from Scenario 1 to 2a or 2b) has an imperceptible effect on total delay, average speed and
total travel time and causes only a 3.4%~ 4.9% increase in number of stops network- wide,
while adding the pedestrian green time requirement ( Scenarios 3 and 4) causes a much
bigger impact ( 26% increase in total delay, 11% decrease in average speed, 12% increase
in travel time, and 8.1% increase in stops). Table 4.3 shows green splits for through
traffic on the mainline, which provides an intersection- level comparison of the scenarios.
Table 4.3. Comparison of green splits under moderate flow condition ( seconds)
Cross Street
Name
Traffic
direction
Scenario 1 Scenario 2b
(% change 1)
Scenario 3
(% change 2)
Scenario 4b
(% change 3)
NE 97.0 96.3 (- 0.7) 97.0 # 1 Lowes ( 0.0) 96.3 (- 0.7)
SW 57.5 56.8 (- 1.2) 57.5 ( 0.0) 56.8 (- 1.2)
# 2 Newhall NE 52.0 54.0 ( 3.8) 46.0 (- 11.5) 46.0 ( 0.0)
Ranch SW 41.0 42.0 ( 2.4) 46.0 ( 12.2) 46.0 ( 0.0)
# 3 Best Buy NB 76.0 76.2 ( 0.3) 76.0 ( 0.0) 76.2 ( 0.3)
SB 72.0 72.4 ( 0.6) 72.0 ( 0.0) 72.4 ( 0.6)
# 4 Espuelle NB 70.1 70.1 ( 0.0) 59.0 (- 15.8) 59.0 ( 0.0)
SB 57.8 57.8 ( 0.0) 59.0 ( 2.1) 59.0 ( 0.0)
# 5 Seco EB 95.0 95.0 ( 0.0) 85.0 (- 10.5) 85.0 ( 0.0)
Canyon WB 55.0 55.0 ( 0.0) 50.0 (- 9.1) 50.0 ( 0.0)
37
EB 89.7 89.7 ( 0.0) 83.5 # 6 Alamogordo (- 6.9) 83.5 ( 0.0)
WB 65.7 65.4 (- 0.5) 62.5 (- 4.9) 64.5 ( 3.2)
# 7 Central EB 69.7 69.7 ( 0.0) 62.0 (- 11.0) 62.0 ( 0.0)
Park WB 92.2 92.2 ( 0.0) 77.5 (- 15.9) 77.5 ( 0.0)
# 8 Centurion NE 80.5 80.5 ( 0.0) 64.5 (- 19.9) 64.5 ( 0.0)
SW 53.5 53.5 ( 0.0) 48.5 (- 9.3) 48.5 ( 0.0)
# 9 Haskell EB 60.0 59.6 (- 0.7) 55.0 (- 8.3) 55.0 ( 0.0)
WB 53.0 52.6 (- 0.8) 50.0 (- 5.7) 50.0 ( 0.0)
# 10 Urbandale NE 68.1 68.1 ( 0.0) 63.0 (- 7.5) 63.0 ( 0.0)
SW 55.5 55.5 ( 0.0) 50.5 (- 9.0) 50.5 ( 0.0)
# 11 Wellston EB 37.0 36.8 (- 0.5) 26.5 (- 28.4) 26.5 ( 0.0)
( 60scycle) WB 37.0 36.8 (- 0.5) 26.5 (- 28.4) 26.5 ( 0.0)
# 12 Plum NE 25.8 25.8 ( 0.0) 39.3 ( 52.3) 41.7 ( 6.1)
Canyon SW 50.9 50.9 ( 0.0) 62.4 ( 22.6) 63.8 ( 2.2)
Note:
Scenario 1 – Auto only;
Scenario 2b – Auto+ Bicycle, signal timing re- optimized;
Scenario 3 – Auto+ Pedestrian;
Scenario 4b – Auto+ Pedestrian+ Bicycle, signal timing re- optimized.
1 Percentage change comparing with Auto only scenario
2 Percentage change comparing with Auto only scenario
3 Percentage change comparing with Auto+ Pedestrian scenario
As shown in Table 4.3, adding bicycle minimum green requirements ( Scenario 2b) has a
negligible impact on the green time provided to through movements along the corridor
( maximum 3.8% decrease, with most intersections unaffected). In comparison,
pedestrian green time requirements pose a much bigger impact ( Scenario 3). Figure 4.3
provides a more visual comparison of the green splits of the different scenarios.
4.3. Impact under low traffic flow condition
To study the bicycle green time requirement impact under low traffic flow conditions ( to
represent late night or early morning), the mid- day volumes used for analysis in Section
4.2 are further reduced by 80% to a low traffic flow level and the Scenarios defined in
Section 4.1 are compared under this flow condition in this section. Network performance
MOEs for each scenario under low traffic flow condition are shown in Table 4.4.
Similar to the results under moderate flow conditions, the network- wide MOEs are very
similar among Scenarios 1, 2a, and 2b, and among Scenarios 3, 4a, and 4b. The bicycle
green requirement has minimal impact on all reported network wide MOEs, especially
after the signal timing is re- optimized. Under low flow conditions, the pedestrian green
time requirements again show a bigger impact ( 12.5% increase in total delay, 3.0%
decrease in average speed, 3.5% increase in travel time, and 20.8% increase in stops).
Adding the bicycle green requirements on top of those for pedestrians, again no
additional impact is observed ( as shown in Table 4.4, Scenarios 3, 4a, and 4b). Using the
38
green splits for through movements along the corridor, Table 4.5 provides an
intersection- level comparison of the scenarios.
Figure 4.3. Mainline through movement green split comparison ( moderate flow
condition)
Table 4.4 Network MOE Comparison ( low flow condition)
Scenario
1
Scenario
2a
Scenario
2b
Scenario
3
Scenario
4a
Scenario
4b
Total Delay ( hr) 16 16 16 18 18 18
Number of Stops 1480 1507 1479 1789 1789 1788
Average Speed ( mph) 33 33 33 32 32 32
Total Travel Time
( hr)
55 55 55 57 57 57
Distance Traveled
( mile)
1820 1820 1820 1820 1820 1820
Note:
Scenario 1 – Auto only;
Scenario 2a – Auto+ Bicycle;
Scenario 2b – Auto+ Bicycle, signal timing re- optimized;
Scenario 3 – Auto+ Pedestrian;
Scenario 4a – Auto+ Pedestrian+ Bicycle
39
Scenario 4b – Auto+ Pedestrian+ Bicycle, signal timing re- optimized.
Table 4.5. Comparison of green splits under low flow condition ( seconds)
Traffic
direction
Scenario 1 Scenario 2b
(% change1)
Scenario 3
(% change2)
Scenario 4b
(% change3)
NE 82.0 82.0 ( 0.0) 84.0 # 1 Lowes ( 2.4) 82.4 (- 1.9)
SW 41.5 41.5 ( 0.0)
45.5 ( 9.6) 44.9 (- 1.3)
NE 34.0 30.7 (- 9.7) 46.0 ( 36.3) 46.0 ( 0.0)
# 2 Newhall
Ranch SW 34.0 30.7 (- 9.7) 46.0 ( 36.3) 46.0 ( 0.0)
# 3 Best Buy NB 61.0 60.4 (- 1.0) 67.0 ( 9.8) 65.4 (- 2.4)
SB 61.0 60.4 (- 1.0) 67.0 ( 9.8) 65.4 (- 2.4)
# 4 Espuelle NB 45.6 45.9 ( 0.7) 59.0 ( 29.4) 5 9.0 ( 0.0)
SB 43.2 43.5 ( 0.7) 59.0 ( 36.6)
59.0 ( 0.0)
# 5 Seco EB 79.0 79.0 ( 0.0) 71.0 (- 10.1) 71.0 ( 0.0)
Canyon WB 40.0 40.0 ( 0.0) 44.0 ( 10.0) 44.0 ( 0.0)
# 6 Alamogordo EB 82.1 82.1 ( 0.0) 68.5 (- 16.6) 68.5 ( 0.0)
WB 44.1 44.1 ( 0.0) 38.5 (- 12.7) 38.5 ( 0.0)
# 7 Central EB 46.0 46.0 ( 0.0) 42.0 (- 8.7) 42.0 ( 0.0)
Park WB 82.5 82.5 ( 0.0) 67.5 (- 18.2) 67.5 ( 0.0)
# 8 Centurion NE 61.5 61.5 ( 0.0) 51.5 (- 16.3) 51.5 ( 0.0)
SW 33.5 33.5 ( 0.0) 32.5 (- 3.0) 32.5 ( 0.0)
# 9 Haskell EB 43.0 42.1 (- 2.1) 43.0 ( 0.0) 43.0 ( 0.0)
WB 43.0 42.1 (- 2.1) 43.0 ( 0.0) 43.0 ( 0.0)
# 10 Urbandale NE 48.1 48.1 ( 0.0) 43.0 (- 10.6) 43.0 ( 0.0)
SW 43.6 43.6 ( 0.0) 38.5 (- 11.7) 38.5 ( 0.0)
# 11 Wellston EB 32.5 30.8 (- 5.2) 23.5 (- 27.7) 23.5 ( 0.0)
( 60scycle) WB 32.5 30.8 (- 5.2) 23.5 (- 27.7) 23.5 ( 0.0)
# 12 Plum NE 34.9 32.0 (- 8.3) 42.9 ( 22.9) 43.2 ( 0.7)
Canyon SW 62.5 58.6 (- 6.2)
61.5 (- 1.6) 59.8 (- 2.8)
Note:
Scenario 1 – Auto only;
Scenario 2b – Auto+ Bicycle, signal timing re- optimized;
Scenario 3 – Auto+ Pedestrian;
Scenario 4b – Auto+ Pedestrian+ Bicycle, signal timing re- optimized.
1 Percentage change comparing with Auto only scenario
2 Percentage change comparing with Auto only scenario
3 Percentage change comparing with Auto+ Pedestrian scenario
As shown in Table 4.5, adding bicycle minimum green requirements ( Scenario 2b) has a
very small impact on the green time provided to through movements along the corridor
( maximum 9.7% decrease, with many intersections unaffected). In comparison,
pedestrian green time requirements pose a substantially bigger impact ( Scenario 3).
Figure 4.4 provides a more visual comparison of the green splits of the different
scenarios.
40
Figure 4.4. Mainline through movement green split comparison ( low flow condition)
4.4 Conclusions
Using SYNCHRO as the signal timing simulation and optimization package, this chapter
shows the impact of bicycle minimum green requirements under moderate and low traffic
flow conditions. The results show that applying a reasonable bicycle minimum green
requirement has a small to negligible effect on the performance of the corridor, both
network- wide and at the intersection level, under both moderate and low volume traffic
conditions. Previous simulation work already showed that under high traffic volumes we
should expect enough vehicular traffic on the cross streets to actuate a green phase at
least as long as minimum needed for bicyclist crossings, so the increased minimum
bicyclist crossing time requirement has no practical effect on traffic.
41
5. Signal Timing Recommendations
The field data show significant diversity in the timing that bicyclists needed to cross
intersections throughout California. With a limited number of intersections and many
variables that could explain the variations, it was not possible to separate out all of the
effects directly to develop a comprehensive bicyclist signal timing handbook at this stage.
We have focused on intersection width as an obvious and measurable influence on the
time that bicyclists need to cross, and suggest formulas based on width to represent the
timing needed to accommodate the 80% ile and 90% ile bicyclists at each intersection.
However, width does not tell the whole story because the crossing times also depend on:
- bicyclist demographics ( age, bicycling experience, trip purpose and time of day)
- visibility that bicyclists have of cross traffic and the speed and density of that
cross traffic
- local intersection geometry ( grades, road surface crown).
The total crossing time distributions for standing start bicyclists should be used to select
the total time provided for clearing the intersection ( green plus yellow plus all- red
interval). Agencies often have their own specific rules for limiting the duration of yellow
and all- red intervals, but the selection of the yellow interval should at least be informed
by the distribution of bicyclist rolling start speeds so that bicyclists do not get caught in
the dilemma zone with undue frequency.
The total crossing time distributions as a function of crossing width W ( ft.) can be
summarized based on summation of the distributions for offset times and intersection
width divided by final crossing speed. In our previous report, we were able to show that
the offset times and final crossing speeds of individual bicyclists were not correlated, so
the distributions can be added without introducing bias. The combinations of offset times
and cruise speed crossing times produce equations for the 80th and 90th percentile total
crossing times of:
– T80 = 7.0 + 0.055 W ( Park Blvd., Palo Alto)
– T80 = 5.0 + 0.079 W ( Russell St., Berkeley)
– T80 = 4.9 + 0.097 W ( Anderson Rd., Davis)
– T80 = 4.35 + 0.082 W ( Polk St., S. F.)
– T80 = 3.7 + 0.10 W ( Venice Blvd., L. A.)
– T80 = 4.2 + 0.077 W ( Main St., Santa Monica)
– T90 = 8.1 + 0.059 W ( Park Blvd., Palo Alto)
– T90 = 6.0 + 0.086 W ( Russell St., Berkeley)
– T90 = 5.6 + 0.106 W ( Anderson Rd., Davis)
– T90 = 4.8 + 0.087 W ( Polk St., S. F.)
– T90 = 4.5 + 0.112 W ( Venice Blvd., L. A.)
– T90 = 5.2 + 0.102 W ( Main St., Santa Monica)
42
When these are plotted it is possible to see the diversity of these crossing behaviors
graphically, as shown in Figure 5.1. In this figure, the 80% ile crossing times are
indicated by dashed lines and the 90% ile crossing times are solid lines, representing the
six intersections for which we have substantial data, each of which is plotted in a
different color. Because the crossing distance at Main St. in Santa Monica varied
significantly among bicyclists, its data ( orange lines) show an exceptionally wide
variation between the 80% ile and 90% ile samples and are not assigned any specific value
of intersection width here.
Figure 5.1 Crossing Times as a Function of Street Width
Superimposed on top of the data is a black line representing the minimum bicycle timing
defined in Table 4D- 109 of Caltrans’ Traffic Operations Policy Directive 09- 06, issued
September 10, 2009. This line, which was defined based on a subset of the data reported
here, appears to represent a reasonable approximation to the 85% ile bicyclist needs.
Before these timing criteria are applied to a specific intersection, it would be advisable to
consider whether there are special conditions that could affect the bicyclist needs at that
intersection. The conditions that could require longer signal timing for bicyclists include:
- significant proportion of children or casual recreational bicyclists
- restricted visibility of cross traffic by bicyclists seeking to cross
- high- speed cross traffic ( posted speed above 30 mph) posing an increased threat
to bicyclists
- significant grades or road surface crowns making it more difficult for bicyclists to
accelerate to full speed.
0 1 2 3 4 5 6 7 8 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
0 20 40 60 80 100 120 140 160
Street Crossing Width ( ft)
Total Green + Yellow + All Red Time
Park
Blvd.
Russell
St.
Anderson
Rd.
Polk
St.
Venice
Blvd.
Caltrans
Table 4D- 109
( CA)
Yellow + All Red Clearance
43
On the other hand, if the bicycling population is exceptionally vigorous and physically fit
at an intersection, it may be possible to shorten the timing slightly from the values shown
here.
The observed rolling start bicyclist speeds can also be used to estimate yellow plus all- red
clearance intervals for bicyclists who are just entering the intersection at steady cruising
speed at the yellow onset. The observed 10% ile and 20% ile bicyclist speeds of 8 and 9
mph respectively would indicate yellow clearance intervals of:
Y80 = 0.076 W to accommodate 80% of bicyclists
Y90 = 0.085 W to accommodate 90% of bicyclists
These are indicated by the black lines shown in the lower part of Figure 5.1.
Unfortunately these values are so much larger than the values that would typically be
applied at the wider intersections that it is likely to be difficult to gain acceptance of these
values ( such as 10 seconds for the 125 foot width of El Camino Real at Park Blvd.).
44
References
Shladover, S. E., Z. Kim, M. Cao, A. Sharafsaleh and J.- Q. Li, “ Bicyclist Intersection
Crossing Times: Quantitative Measurements for Selecting Signal Timing”,
Transportation Research Record No. 2128, 2009, pp. 86 - 95.
S. E. Shladover, ZuWhan Kim, Meng Cao, Ashkan Sharafsaleh, JingQuan Li and Kai
Leung, “ Bicycle Detection and Operational Concept for Signalized Intersections”,
California PATH Research Report UC- ITS- PRR- 2009- 37.
California Department of Transportation, Traffic Operations Policy Directive TOPD No.
09- 06, “ Provide Bicycle and Motorcycle Detection on all new and modified
approaches to traffic- actuated signals in the state of California”, September 10,
2009.