ISSN 1055- 1425
September 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 6410
CALIFORNIA PATH PROGRAM
INSTITUTE OF TRANSPORTATION STUDIES
UNIVERSITY OF CALIFORNIA, BERKELEY
Assessment of the Applicability of Bus Rapid
Transit on Conventional Highways— Case Study
Feasibility Analyses Along the Lincoln Boulevard
Corridor
UCB- ITS- PRR- 2009- 38
California PATH Research Report
Alex Skabardonis, Mark A. Miller, Irene Yue Li, Robert Cervero,
Jin Murakami, Zhijun Zou, Neal Richman, Norman Wong
CALIFORNIA PARTNERS FOR ADVANCED TRANSIT AND HIGHWAYS
Assessment of the Applicability of Bus Rapid
Transit on Conventional Highways ─ Case
Study Feasibility Analyses Along the Lincoln
Boulevard Corridor
University of California at Berkeley ( California PATH
Program/ Institute of Transportation Studies and Department
of City and Regional Planning)
Tongji University ( Beijing, China)
University of California, Los Angeles ( Center for Neighborhood
Knowledge/ School of Public Affairs)
September 7, 2009
ii
ACKNOWLEDGEMENTS
This work was performed by the California PATH Program and the Department of City and
Regional Planning at the University of California at Berkeley, Tongji University in Beijing, China,
and the University of California at Los Angeles under the sponsorship of the State of California
Business, Transportation and Housing Agency, Department of Transportation ( Caltrans),
Division of Mass Transportation, Division of Research and Innovation ( DR& I) ( Interagency
Agreement # 65A0208). The contents of this paper 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.
The authors thank Elaine Houmani, Wendy Johnsen, Bradley Mizuno, Sebastian Oduni, and
Scott Sauer of Caltrans for their support during this research. The authors also want to thank
Paul Casey and Benjamin Steers of the Big Blue Bus, City of Santa Monica for their support of
this research. The authors would also like to thank Yunus Ghausi, Sin Kim, and Kim Gia Nauyen
of the Caltrans District 7 Office in Los Angeles. Finally, the authors would like to very much
thank Wei‐ Bin Zhang, our PATH colleague, for his contributions to this research.
Author List
University of California, Berkeley:
Alex Skabardonis ( Principal Investigator)
Mark A. Miller ( Project Manager)
Irene Yue Li
Robert Cervero
Jin Murakami
Tongji University:
Zhijun Zou ( Principal Investigator)
University of California, Los Angeles:
Neal Richman ( Principal Investigator)
Norman Wong
iii
ABSTRACT
This report presents the results of a performance assessment of the applicability of bus rapid
transit on conventional highways in the setting of a site‐ specific case study along the Lincoln
Boulevard corridor in Santa Monica, California. When bus rapid transit systems are
implemented on conventional highways, especially on arterials, there are numerous bus priority
treatments that can be applied and each has associated with it issues that need to be
investigated. In this study, we are investigating concurrent flow curb bus lanes based on the
removal of peak period parking along the Lincoln Boulevard corridor. We have focused on traffic
and ridership impacts associated with this type of bus rapid transit system implementation.
Key Words: bus rapid transit, bus‐ only lane, traffic impacts, ridership
iv
EXECUTIVE SUMMARY
This report presents the results of a performance assessment of the applicability of bus rapid
transit on conventional highways in the setting of a site‐ specific case study along the Lincoln
Boulevard corridor in Santa Monica, California. When bus rapid transit systems are
implemented on conventional highways, especially on arterials, there are numerous bus priority
treatments that can be applied and each has associated with it issues that need to be
investigated. In this study, we are investigating concurrent flow curb bus lanes based on the
removal of peak period parking in the context of the Lincoln Boulevard corridor. We have
focused on the traffic and ridership impacts associated with this type of bus rapid transit system
implementation.
For the traffic impacts study we have used the VISSIM package as the primary tool with which to
simulate the Lincoln Corridor in the context of converting during the morning and afternoon
peak periods the curbside parking lane to a bus‐ only lane over the course of two miles. Both the
curbside and adjacent travel lanes were simulated and traffic impacts were accumulated for
each of them. VISSIM represented detailed geometric settings, traffic conditions and control,
and bus operational characteristics. Initially, the models were calibrated using data collected
from the Lincoln Boulevard corridor; and the outputs from the “ before” model and the “ after”
model have been used to evaluate the lane conversion impacts. The outputs include Measures
of Effectiveness ( MOEs) for both traffic and bus operation status, such as delay, travel time,
speed, and queue length for general traffic and buses.
The findings from the simulation runs have been summarized to show which factors or
combination of factors would affect the lane conversion impact significantly. Note that the
summary will be based on simulation results for the case study site and thus the result is
site‐ specific, however, the factors/ combinations discovered to be important should be the ones
that need to be studied closely for a site that is considering the lane conversion strategy. The
simulation study’s objective was to test and compare different curb lane operational strategies
in a simulated environment. Five scenarios were defined:
v
 Scenario 1: Do Nothing
No change is made to the existing state, whereas the curb lane remains as a parking
lane during the peak periods. This provides a baseline reference scenario with which all
other scenarios can be compared.
 Scenario 2: Bus Only Lane
The curb lane operates as a bus only lane during peak periods. Scenario 2 consists of
both the Rapid 3 and Local 3 Lines being allowed to operate in the curb lane during
peak periods.
 Scenario 3: Mixed Traffic Lane
The curb lane operates as a mixed traffic or general purpose lane open to all types of
vehicles during peak periods.
 Scenario 4: Special Vehicle Lane for Buses, Taxis, and Charter Buses
The curb lane operates as a special vehicle lane only open to buses ( Rapid 3 and Local
3), taxis, and charter buses during peak periods.
 Scenario 5: Dynamic Dedicated Bus Rapid Transit ( BRT) Lane
The curb lane may dynamically convert from a mixed traffic lane to a bus only lane
when a bus appears, and convert from a bus only lane back to a mixed traffic lane when
not used by a bus.
The simulation study implemented these five scenarios in the simulation model and derived
measures of effectiveness ( MOEs) analysis results in terms of delay, travel time and speed for
both buses ( Rapid 3 and Local 3) and non‐ buses, and queue length. VISSIM produced these
MOEs down to the level of each link within the two‐ mile corridor and for each 30‐ minute time
period within each three‐ hour peak period: 7AM‐ 10AM and 4PM‐ 7PM. Over the entire corridor
during the peak periods, Tables ES‐ 1 through ES‐ 4 show the simulation findings on a corridor
basis across all scenarios so that comparisons relative to the Do Nothing ( Scenario 1) may be
made. Numbers in parentheses are the percentage change in a particular MOE relative to
Scenario 1.
vi
As could be observed from the data, with the curb lane converted into a travel lane, the MOEs
are all improved compared with the do‐ nothing scenario, that is, delays decrease across all
alternative scenarios, travel times decrease across all alternative scenarios, speeds increase
across all alternative scenarios, and queue lengths decrease across all alternative scenarios;
however, no single alternative scenario does better than all other alternative scenarios across all
MOEs.
Among all scenarios 2 through 5 on a corridor level basis, Scenario 2 has the lowest Rapid 3 and
Local 3 bus delay, lowest Rapid 3 and Local 3 travel time and highest Rapid 3 and Local 3 bus
speed, and Scenario 3 has the lowest non‐ bus delay, highest non‐ bus speed, and shortest queue
length. However, Scenarios 4 and 5 give values for delay, travel time, and speed for the Rapid 3
and Local 3 buses that are close to Scenario 2’ s values; moreover, they are not statistically
different from each other in most cases for different MOEs based on a set of statistical tests
performed on link level data for scenarios 2, 4, and 5. While Scenario 5, meanwhile, appears to
be a good compromise between exclusive bus use and entirely mixed traffic use, it also requires
a lot more technology and is considered an experimental scenario and needs more extensive
investigation if it is to be seriously considered to be implemented. Another observation is that
the travel time and speed gap between the Rapid 3 bus and non‐ buses decreases considerably,
especially for Scenarios 2 and 4.
A primary question we were tasked to answer in this study was to measure the impact on
non‐ bus traffic if a curbside lane with parking privileges were to be converted to a bus only lane
during the morning and afternoon peak periods. Among the study’s findings is the fact that
there is no negative impact on non‐ bus traffic for each of the bus‐ only lane scenarios ( 2, 4, and
5). In fact there are even benefits to non‐ bus traffic for these three scenarios, just not as large as
for Scenario 3, which makes available the curbside lane to all vehicles during the peak periods.
Alternatively, Scenario 3 generates benefits for buses – both the Rapid 3 and Local 3 – just
considerably smaller than available through Scenarios 2, 4, or 5.
vii
Table ES‐ 1 Average Vehicle Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
Non‐ Bus 2.7 2.4 (‐ 11.1%) 1.7 (‐ 37.0%) 2.3 (‐ 14.8%) 1.8 (‐ 33.3%)
Rapid 3 4.5 2.8 (‐ 37.8%) 3.9 (‐ 13.3%) 3.2 (‐ 28.9%) 3.0 (‐ 33.3%)
Local 3 10.8 8.3 (‐ 23.1%) 9.8 (‐ 9.3%) 8.4 (‐ 22.2%) 8.6 (‐ 20.4%)
Northbound
( AM Peak)
Non‐ Bus 1.8 1.7 (‐ 5.6%) 1.2 (‐ 33.3%) 1.7 (‐ 5.6%) 1.3 (‐ 27.8%)
Rapid 3 4.1 2.5 (‐ 39.0%) 3.3 (‐ 19.5%) 2.5 (‐ 39.0%) 2.6 (‐ 36.6%)
Local 3 7.5 5.3 (‐ 29.3%) 6.1 ( 18.7%) 5.4 (‐ 28.0%) 5.4 (‐ 28.0%)
Table ES‐ 2 Average Vehicle Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
Non‐ Bus 6.7 6.5 (‐ 3.0%) 5.8 (‐ 13.4%) 6.3 (‐ 6.0%) 5.9 (‐ 11.9%)
Rapid 3 9.5 7.8 (‐ 17.9%) 8.9 (‐ 6.3%) 8.2 (‐ 13.7%) 8.0 (‐ 15.8%)
Local 3 15.8 13.3 (‐ 15.8%) 14.8 (‐ 6.3%) 13.4 (‐ 15.2%) 13.5 (‐ 14.6%)
Northbound
( AM Peak)
Non‐ Bus 4.9 4.7 (‐ 4.1%) 4.2 (‐ 14.3%) 4.7 (‐ 4.1%) 4.3 (‐ 12.2%)
Rapid 3 7.9 6.3 (‐ 20.3%) 7.2 (‐ 8.9%) 6.3 (‐ 20.3%) 6.5 (‐ 17.7%)
Local 3 11.3 9.2 (‐ 18.6%) 9.9 (‐ 12.4%) 9.2 (‐ 18.6%) 9.2 (‐ 18.6%)
viii
Table ES‐ 3 Average Vehicle Corridor Speed ( mph) over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
Non‐ Bus 23.0 23.6 (+ 2.6%) 26.1 (+ 13.5%) 24.0 (+ 4.3%) 25.5 (+ 10.9%)
Rapid 3
17.2 20.1 (+ 16.9%) 17.6 (+ 2.3%)
19.0
(+ 10.5%) 19.3 (+ 12.2%)
Local 3
9.8 11.5 (+ 17.3%) 10.4 (+ 6.1%)
11.4
(+ 16.3%) 11.3 (+ 15.3%)
Northbound
( AM Peak)
Non‐ Bus 24.4 25.5 (+ 4.5%) 27.3 (+ 11.9%) 25.6 (+ 4.9%) 27.1 (+ 11.1%)
Rapid 3
16.5 18.9 (+ 14.5%) 17.3 (+ 4.8%)
19.0
(+ 15.2%) 18.6 (+ 12.7%)
Local 3
10.0 12.4 (+ 24.0%) 11.4 (+ 14.0%)
12.3
(+ 23.0%) 12.4 (+ 24.0%)
Table ES‐ 4 Average Corridor Queue Length ( feet) over Peak Periods
Direction
( Time Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic Dedicated
BRT)
Southbound
( PM Peak)
28.1 27.4 (‐ 2.5%) 10.0 (‐ 64.4%) 24.9 (‐ 11.4%) 12.2 (‐ 56.6%)
Northbound
( AM Peak)
19.7 18.2 (‐ 7.6%) 7.4 (‐ 62.4%) 18.3 (‐ 7.1%) 8.9 (‐ 54.8%))
For the ridership impact study, we used multiple regression models, referred to as Direct
Modeling, to estimate ridership as a function of station environments and transit service
features, which provides a fine‐ grain resolution suitable for studying relationships between
built environments, transit services, and ridership. The accessibility of station‐ area residents
to jobs and shops via transit versus auto are sometimes included in such models, thus in this
sense, performance attributes of competitive modes are imbedded in the analyses. Direct
ridership models generally have small sample sizes since observations consist of transit
stations or stops. Thus degree of freedom constraints often limit the number of variables that
ix
can be included as well as their specifications. It is because of these limitations that direct
models fall under the rubric of sketch‐ planning tools. They provide order‐ of‐ magnitude
insights for testing of various system designs and land‐ use scenarios. Collected data included
ridership from the Rapid 3 along Lincoln Boulevard together with other bus rapid transit lines
in Los Angeles County, e. g., numerous Metro Rapid Lines and the Metro Orange Line.
Findings from the model specification and ridership forecasting shows that substantial increases
in average daily boardings can be anticipated from the planned service enhancements on the
Rapid 3 Line. The adjusted model estimates that average daily boardings across the six Rapid 3
stops along the Lincoln Boulevard corridor will increase by between a factor of 3.5 and a factor
of 8.3. The average increase in boardings for the six stops on Rapid Blue Line 3 is estimated to
be more than 500%. Such large surges in ridership could be on the high side, again reflecting the
more transit‐ conducive environment of Metro Rapid services in denser, more congested Los
Angeles City ( that dominated the database). We note that the approximately five‐ fold average
increase in ridership relative to current counts on the Rapid 3 Line is not inconsistent with the
differentials in average boardings between the Metro Orange Line stops and other Metropolitan
Transportation Authority Metro Rapid stops. While no one has a crystal ball and can predict
with any precision what the future ridership will be on the Rapid 3 Line, experiences with
dedicated‐ lane services in Los Angeles County suggest that the impacts will be appreciable.
x
TABLE OF CONTENTS
SECTION PAGE
ACKNOWLEDGEMENTS ii
ABSTRACT iii
EXECUTIVE SUMMARY iv
LIST OF TABLES xii
LIST OF FIGURES xv
1.0 PROJECT OVERVIEW 1
1.1 Motivation 1
1.2 Objectives 2
1.3 Contents of the Report 3
2.0 BUS RAPID TRANSIT RUNNING WAYS: ARTERIAL‐ RELATED BUS PRIORITY
TREATMENTS 4
3.0 LINCOLN BOULEVARD CASE STUDY: TRAFFIC IMPACTS ASSESSMENT 7
3.1 The Simulation Site, Scenarios, and Data Collection 7
3.1.1 Simulation Site 7
3.1.2 Scenarios 10
3.1.3 Data Collection 10
3.2 Simulation Network Modeling 11
3.2.1 Network Geometry 11
3.2.2 Traffic Demand Coding 12
3.2.2.1 Non‐ Bus Traffic Demand 12
3.2.2.2 Bus Traffic Demand 15
3.2.3 Signal Controllers Coding 15
3.3 Implementation of Scenarios in the Simulation 17
3.3.1 Scenario 1 – Do Nothing ( Baseline) 17
3.3.2 Scenario 2 – Bus Only Lane 17
3.3.3 Scenario 3 – Mixed Flow Traffic Lane 19
3.3.4 Scenario 4 – Special Lane for Buses, Taxis, and Charter Buses 19
3.3.5 Scenario 5 – Dynamic Dedicated Bus Rapid Transit Lane 20
3.3.5.1 Dynamic Bus Rapid Transit Lane Operation Rules 20
3.3.5.2 Implementing Dynamic Bus Rapid Transit Lane Operation via VISSIM COM
Programming 22
3.4 Simulation Model Calibration 25
3.5 Measures of Effectiveness Analysis 26
3.5.1 Comparative Analysis Across Scenarios 28
3.5.1.1 Delay 28
3.5.1.2 Travel Time 38
xi
SECTION PAGE
3.5.1.3 Speed 47
3.5.1.4 Queue Length 55
3.5.2 Major Corridor‐ wide Findings 57
3.6 Recommendations and Conclusions 64
References 66
4.0 LINCOLN BOULEVARD CASE STUDY: RIDERSHIP IMPACTS ASSESSMENT 67
4.1 Modeling Approach and Sample 68
4.1.1 Sample Selection 69
4.1.2 Model Specification and Variables 70
4.2 Direct Model for Estimating Bus Rapid Transit Ridership 74
4.3 Prediction Accuracy of the Direct Ridership Model 78
4.4 Forecasted Daily Ridership for Six Rapid 3 Line Stops 81
4.5 Conclusions 86
References 87
xii
LIST OF TABLES PAGE
TABLE 3.1 Calibration Results: Comparison of Average Bus Travel Times 25
TABLE 3.2 Numbers and Definitions of Links for Evaluation 26
TABLE 3.3 Average Non‐ Bus Delay ( seconds) per Corridor Link over Peak Periods 29
TABLE 3.4 Average Non‐ Bus Corridor Delay ( minutes) over Peak Periods 31
TABLE 3.5 Total Non‐ Bus Corridor Delay ( hours) over Peak Periods 31
TABLE 3.6 Average Rapid 3 Bus Delay ( seconds) per Corridor Link over Peak
Periods 33
TABLE 3.7 Average Rapid 3 Bus Corridor Delay ( minutes) over Peak Periods 35
TABLE 3.8 Total Rapid 3 Bus Corridor Delay ( minutes) over Peak Periods 35
TABLE 3.9 Average Local 3 Bus Delay ( seconds) per Corridor Link over Peak
Periods 36
TABLE 3.10 Average Local 3 Bus Corridor Delay ( minutes) over Peak Periods 38
TABLE 3.11 Total Local 3 Bus Corridor Delay ( minutes) over Peak Periods 38
TABLE 3.12 Average Non‐ Bus Travel Time ( seconds) per Corridor Link over Peak
Periods 39
TABLE 3.13 Average Non‐ Bus Corridor Travel Time ( minutes) over Peak Periods 40
TABLE 3.14 Total Non‐ Bus Corridor Travel Time ( hours) over Peak Periods 41
TABLE 3.15 Average Rapid 3 Bus Travel Time ( seconds) per Corridor Link over Peak
Periods 42
TABLE 3.16 Average Rapid 3 Bus Corridor Travel Time ( minutes) over Peak Periods 44
TABLE 3.17 Total Rapid 3 Bus Corridor Travel Time ( minutes) over Peak Periods 44
TABLE 3.18 Average Local 3 Bus Travel Time ( seconds) per Corridor Link over Peak
Periods 45
TABLE 3.19 Average Local 3 Bus Corridor ( minutes) over Peak Periods 46
xiii
TABLE 3.20 Total Local 3 Bus Corridor Travel Time ( minutes) over Peak Periods 47
TABLE 3.21 Average Non‐ Bus Speed ( mph) per Corridor Link over Peak Periods 48
TABLE 3.22 Average Non‐ Bus Corridor Speed ( mph) over Peak Periods 49
TABLE 3.23 Average Rapid 3 Bus Speed ( mph) per Corridor Link over Peak Periods 50
TABLE 3.24 Average Rapid 3 Bus Corridor Speed ( mph) over Peak Periods 52
TABLE 3.25 Average Local 3 Bus Speed ( mph) per Corridor Link over Peak Periods 53
TABLE 3.26 Average Local 3 Bus Corridor Speed ( mph) over Peak Periods 54
TABLE 3.27 Average Queue Length per Corridor Link over Peak Periods 55
TABLE 3.28 Average Corridor Queue Length ( feet) over Peak Periods 57
TABLE 3.29 Average Vehicle Corridor Delay ( minutes) over Peak Periods 58
TABLE 3.30 Average Vehicle Corridor Travel Time ( minutes) over Peak Periods 60
TABLE 3.31 Average Vehicle Corridor Speed ( mph) over Peak Periods 61
TABLE 3.32 Total Corridor Delay over Peak Periods 62
TABLE 3.33 Total Corridor Travel Time over Peak Periods 63
TABLE 3.34 LOS Values Comparison Across Scenarios 65
TABLE 4.1 Comparison of Average Daily Ridership Among BRT Services in
Los Angeles County 75
TABLE 4.2 Descriptive Statistics for Dependent Variable and Independent
Variables that Enter the Direct Ridership Model 75
TABLE 4.3 Direct Ridership Model for BRT in Los Angeles County 77
TABLE 4.4 Comparison of Actual and Predicted Average Daily Boardings for
October 2008, Six Rapid 3 Line Stops 80
TABLE 4.5 Existing Conditions of Six Stops on Rapid 3 Line ( October 2008) 83
xiv
TABLE 4.6 Future BRT Scenario for Six Stops on Rapid Blue Line 3 83
TABLE 4.7 Forecasted Ridership for Six Stops on the Planned Dedicated‐ Lane
Rapid 3 Line 83
TABLE 4.8 Inputs for Existing 68 Buses that will operate as BRT Services for Six
Stops on Lincoln Boulevard 85
TABLE 4.9 Forecasted Ridership for Six Stops for Existing 68 Buses on Lincoln
Boulevard as well as Total Including New Rapid Blue Line 3 Services 86
xv
LIST OF FIGURES PAGE
FIGURE 3.1 Lincoln Boulevard Corridor between Wilshire Boulevard and
Rose Avenue 8
FIGURE 3.2 Lincoln Boulevard Corridor between Rose Avenue and
Washington Boulevard 9
FIGURE 3.3 Lincoln Boulevard Corridor between Washington and
Jefferson Boulevards 9
FIGURE 3.4 Typical Lane Channelization of Lincoln Boulevard 12
FIGURE 3.5 Background Map Cut from Google Earth 13
FIGURE 3.6 Geometric Layout and Lane Channelization Modeled in Simulation
Network 13
FIGURE 3.7 OD Zone Numbers and Intersection Numbers along Lincoln
Boulevard 14
FIGURE 3.8 Illustration of Bus Traffic Demand Coding of Rapid 3 Line 15
FIGURE 3.9 Illustration of Signal Controller Coding in VISSIM 16
FIGURE 3.10 Snapshot of the Simulation Animation for Scenario 1 17
FIGURE 3.11 Snapshot of the Simulation Animation for Scenario 2 18
FIGURE 3.12 Snapshot of the Simulation Animation for Scenario 3 19
FIGURE 3.13 Snapshot of the Simulation Animation for Scenario 4 20
FIGURE 3.14 Illustration of Related Settings for Dynamic BRT Lane Operation 21
FIGURE 3.15 Snapshot when the Curb Lane Opens to Mixed Traffic 23
FIGURE 3.16 Snapshot after a Bus has just passed the Bus Approaching Detector 23
FIGURE 3.17 Snapshot when the Curb Lane has Temporarily Converted to a
Bus‐ Only Lane 24
FIGURE 3.18 Snapshot when a Bus has just passed by the Bus Departing Detector 24
xvi
LIST OF FIGURES PAGE
FIGURE 3.19 Range in Percentage (%) Variation for Non‐ Bus Delay over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 30
FIGURE 3.20a Range in Percentage (%) Variation for Rapid 3 Bus Delay over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 34
FIGURE 3.20b Range in Percentage (%) Variation for Rapid 3 Bus Delay for all Links
except Links 6 and 16 of Alternative Scenarios ( 2, 3, 4, 5) Relative to
Scenario 1 34
FIGURE 3.21 Range in Percentage (%) Variation for Local 3 Bus Delay over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 37
FIGURE 3.22 Range in Percentage (%) Variation for Non‐ Bus Travel Time over all Links
of Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 40
FIGURE 3.23 Range in Percentage (%) Variation for Rapid 3 Bus Travel Time over all
Links of Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 43
FIGURE 3.24 Range in Percentage (%) Variation for Local 3 Bus Travel Time over all
Links of Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 46
FIGURE 3.25 Range in Percentage (%) Variation for Non‐ Bus Speed over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 49
FIGURE 3.26 Range in Percentage (%) Variation for Rapid 3 Bus Speed over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 51
FIGURE 3.27 Range in Percentage (%) Variation for Local 3 Bus Speed over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 54
FIGURE 3.28 Range in Percentage (%) Variation for Queue Length over all Links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 56
FIGURE 3.29 Southbound Corridor Delay Across Scenarios 59
FIGURE 3.30 Northbound Corridor Delay Across Scenarios 59
FIGURE 3.31 Southbound Corridor Travel Time Across Scenarios 60
FIGURE 3.32 Northbound Corridor Travel Time Across Scenarios 61
xvii
FIGURE 4.1 Locations of 69 BRT Bus Stop Observations used for Estimating Direct
Ridership Model 70
FIGURE 4.2 A Plot of Predicted Boardings ( Vertical Axis) and Actual Boardings
( Horizontal Axis) for 69 Metro Rapid Bus Stops 79
FIGURE 4.3 Plot of Actual and Predicted Average Daily Boardings for October 2008,
Six Rapid Blue Line 3 Stops 81
FIGURE 4.4 Comparison of Forecasted Ridership for Six Stops on the Planned
Dedicated‐ Lane Rapid 3 Line 84
1
1.0 PROJECT OVERVIEW
This report constitutes the final deliverable for PATH Project Task Order 6410  “ Assessing
Bus Rapid Transit Implementation on Conventional Highways”. The project has examined
opportunities for implementing bus rapid transit systems on conventional highways,
whether on freeways or arterials by performing a review of the literature of bus lanes and
bus rapid transit systems use of conventional highways together with a consideration of
California bus rapid transit systems practice, and performing a corridor‐ specific case study of
the Lincoln Boulevard Big Blue Bus ( Santa Monica) Rapid 3 Line currently running in mixed
flow traffic. The remainder of this section discusses the motivation for, objectives of, and a
summary of the contents for the remainder of this final report.
1.1 Motivation
Bus rapid transit ( BRT) systems are commonly viewed as an alternative travel mode to help
make bus transit more attractive by enhancing customer level of service with an ultimate
goal of increasing ridership that contributes to relieving traffic congestion. The elements that
comprise any rapid transit system consist of:
 Running Ways;
 Stations;
 Vehicles;
 Intelligent Transportation Systems;
 Fare Collection;
 Service Patterns; and,
 Identity and Branding.
Running ways are the key element of BRT systems around which the other components
revolve since running ways serve as the infrastructural foundation around which the other
elements function. Moreover, it is the running ways that should allow for rapid and reliable
movement of buses with minimum traffic interference to provide a clear sense of presence
and permanence. The types of running ways for BRT service can range between mixed flow
traffic operation and fully grade‐ separated busways ( Diaz, R. B., et al., 2004), ( Kittelson &
Associates, Inc., et al., 2007), and ( Levinson, H. S., et al., 2003).
2
An existing mixed flow lane on an arterial represents the most basic form of running way.
BRT vehicles can operate with no separation from other vehicle traffic on virtually any
arterial street or highway. Increasing levels of segregation begin with operations in mixed
arterial traffic, through exclusive arterial lanes ( curbside or median), contra‐ flow freeway bus
lanes, normal‐ flow freeway HOV lanes, grade‐ separated lanes or exclusive transitways on
separate rights‐ of‐ way and bus tunnels. Increasing levels of separation from other vehicle
traffic add increasing levels of travel time savings and reliability improvement for the
operation of BRT services. Fully grade‐ separated, segregated BRT transitways have the
highest cost and highest level of speed, safety and reliability of any BRT running way type.
Because of the incremental nature of bus rapid transit systems deployment, the ease and
relatively low cost with which BRT systems can initially be implemented in the setting of
mixed flow traffic, and the number of such deployments in the U. S. in general and in
California specifically, we are motivated by a desire to focus on the conversion of the running
way for a BRT system from mixed traffic flow to one of increasing levels of separation from
other vehicle traffic, in particular, to a bus‐ only lane at curb‐ side during peak periods.
1.2 Objectives
Of particular importance to consider when implementing bus rapid transit is its deployment
on conventional highways including arterials and freeways because of the need to integrate
BRT within an existing roadway infrastructure with specific land use patterns. Such
integration may require changes including removal of peak period parking to allow for a
bus‐ only travel lane, replacement of conventional traffic signal control systems with transit
signal priority systems, or removal of an existing curbside travel lane during peak periods to
allow for a bus‐ only travel lane. Moreover, such changes are likely to have impacts that need
to be examined. The overall objective of this project is to identify and assess such impacts
resulting from the removal of peak period parking in the context of a site‐ specific case study
along the Lincoln Boulevard corridor in the cities of Santa Monica and Los Angeles.
3
1.3 Contents of the Report
This is the first of four sections of the report. Section 2 provides a review of bus rapid transit
systems on conventional highways from the literature. Section 3 discusses the study of traffic
impacts we conducted using modeling and simulation methods; and Section 4 discusses the
study of ridership impacts we conducted using transportation planning and analysis
methods.
4
2.0 BUS RAPID TRANSIT RUNNING WAYS: ARTERIAL‐ RELATED BUS PRIORITY TREATMENTS
There are several types of arterial‐ related bus priority treatments for Bus Rapid Transit
running ways, as follows:
 Mixed traffic flow
 Concurrent flow curb bus lanes
 Concurrent flow inside curb bus lanes
 Contra‐ flow curb bus lanes
 Median bus lanes
 Bus‐ only streets
The running way setting for the Lincoln Boulevard bus rapid transit corridor is an arterial
street with current bus priority treatment for the Rapid 3 Line as mixed traffic flow. Bus rapid
transit systems generally operate in mixed traffic flow when physical and/ or traffic factors
preclude bus lanes or busways from being initially implemented. There are tradeoffs with
implementing BRT in mixed traffic flow; advantages include low costs and fast
implementation with a minimum of construction; however, mixed traffic flow operations can
limit bus speeds and service reliability due to the BRT vehicle having to travel in this
environment with other vehicles; system identity can also suffer without specific actions
taken to equip either or both the BRT vehicle and the BRT stop/ station with a single unified
BRT brand identity. In the Rapid 3 case, such actions have been taken to provide a brand
identity.
There are several examples of BRT systems implemented in California in addition to the
Lincoln Boulevard corridor Rapid 3 Line that currently operate in mixed traffic flow all of
which having a distinctively unique brand identity associated with their buses and bus stops,
as follows:
 Los Angeles County Metropolitan Transportation Authority’s Metro Rapid Lines with
the first two lines implemented in 2001 on Wilshire and Ventura Boulevards.
 AC Transit’s San Pablo Rapid traveling on State Route 123 ( San Pablo Avenue)
between San Pablo and Oakland
 Santa Clara Valley Transportation Authority’s Rapid Line 522 along the El
Camino/ Santa Clara Street/ Alum Rock Avenue corridor ( State Route 82), which
5
provides service along the east‐ west length of Santa Clara County between the
Eastridge Shopping Center in San Jose and the Palo Alto Transit Center.
 Sacramento County’s Regional Transit Line 50 E‐ Bus on the Stockton Boulevard
corridor
Buses will also benefit from customary street and traffic improvements that reduce overall
travel delay. The range of transit‐ related traffic improvements can include grade separations
to bypass points of delay; street expansions to improve traffic distribution or to provide bus
routing continuity; traffic signal improvements including signal coordination and bus transit
signal priority. Other transit‐ focused enhancements include turn controls that exempt buses,
bus stop lengthening, effective enforcement of parking restrictions, and bus stop design
improvements.
Of bus lane and bus street priority treatments, normal flow curb bus lanes are the most
common; they are generally considered when it is not practical to install other on‐ street bus
service options. They are appropriate for implementation under the following conditions:
 No parking or stopping along the curbs during the time periods that the bus lanes
would be in effect
 At least two other moving general traffic lanes in the same direction except in cases
on two‐ way four‐ lane streets where left turns are not permitted during peak period
traffic time periods.
 Curb access for other services to adjacent properties can be readily prohibited during
the time periods of bus lane operation; such services can include loading, unloading,
deliveries
They are the easiest to implement, have the lowest installation costs because they normally
involve only pavement markings and street signs, and have minimum impact on intersecting
driveways and street routings. Customarily, such bus lanes have been used to facilitate bus
movements in Central Business Districts by separating buses from other traffic; however,
such bus lanes are also used along outlying arterials.
Experience in the U. S., however, has shown that they are least effective in terms of travel
time saved, image and brand identity, ability to be enforced, and that they may impact curb
access requirements such as deliveries. Another disadvantage is that right‐ hand turns, when
6
allowed may conflict with bus flow; thus efforts should be made to either totally eliminate or
at least restrict right‐ turning movements that would impede BRT service.
Concurrent flow bus lanes can operate at all times or only during peak period times. On
one‐ way and two‐ way streets, an 11‐ to 13‐ foot bus lane should be provided along the curb.
When street width and circulation patterns permit and peak bus volumes exceed 90 to 100
buses per peak period hour, dual bus lanes should be considered. Figure 1 depicts four
typical concurrent flow bus lane designs for two‐ way streets. The four designs vary by
number of non‐ bus traffic lanes ( one or two) and whether left turns are allowed. For design
numbers 1 and 3, no left turns are allowed. Designs 1 and 2 each have a single non‐ bus
traffic lane; designs 3 and 4 each have two non‐ bus traffic lanes. The width ranges of the
right‐ of‐ way for each of the four designs are provided at the top of the figure adjacent to
each design. Right turns from the bus lane may be prohibited or permitted.
The primary example of a concurrent flow bus lane in California is in San Francisco under the
operation of the San Francisco Municipal Railway ( Muni) on various streets within the city
including:
 Sacramento and Clay Streets, which employ peak‐ hour curbside lanes that prohibit
parking during peak periods.
 Mission Street operates curbside lanes between 7am and 7pm that dedicate a traffic
lane to bus‐ only use, though convert to mixed flow use between 7pm and 7am.
 Third Street between Townsend and Market Streets operates a bus lane throughout
the day; taxis are also allowed to travel in the lanes with buses
7
3.0 LINCOLN BOULEVARD CASE STUDY: TRAFFIC IMPACTS ASSESSMENT
For the Lincoln Boulevard corridor case study, simulation methods were used to quantify the
impact, both on the bus‐ lane and adjacent traffic lane resulting from the lane conversion.
The research team built microscopic simulation models using VISSIM1 to represent detailed
geometric settings ( for instance, lane configuration), traffic conditions ( volume, capacity),
traffic control ( type, signal timing plan), as well as bus operational characteristics. The
models were initially calibrated using data collected from the Lincoln Boulevard case study
corridor; and the outputs from the “ before” model and the “ after” model have been used to
evaluate the lane conversion impacts. The outputs will include Measures of Effectiveness
( MOEs) for both traffic and bus operation status, such as delay, travel time, speed, and
queue length for general traffic and buses, both the Rapid 3 and the Local 3 buses. The
models have been used under different traffic settings, such as different volume, capacity,
lane configurations, to gain a better understanding of the impacts.
The findings from the simulation runs have been summarized to show which factors or
combination of factors would affect the lane conversion impact significantly. Note that the
summary will be based on simulation results for the case study site and thus the result is
site‐ specific, however, the factors/ combinations discovered to be important should be the
ones that need to be studied closely for a site that is considering the lane conversion
strategy.
3.1 The Simulation Site, Scenarios, and Data Collection
3.1.1 Simulation Site
The portion of Lincoln Blvd that that is under consideration for lane conversion is
approximately 4.1 kilometers ( 2.5 miles) in length. It extends between Washington and Pico
Boulevards. To capture the boundary conditions, as well as possible downstream movement
of potential “ choke points”, the section for the simulation study was extended on both north
and south ends of the corridor. More specifically, the total length of the study corridor is
1 VISSIM is a microscopic, behavior- based multi- purpose traffic simulation program
8
approximately eight kilometers ( 5 miles), and it includes the following three segments:
 North of Pico to Wilshire: approximately 1.5 kilometers ( 0.9 miles) with eight
signalized intersections
 Washington to Pico ( where the lane conversion is being considered): approximately
4.1 kilometers ( 2.5 miles) with 12 signalized intersections
 South of Washington to Jefferson: approximately 2.6 kilometers ( 1.6 miles) with six
signalized intersections
The simulation site belongs to two municipal jurisdictions: City of Los Angeles ( Jefferson to
Rose) and City of Santa Monica ( Rose to Wilshire). See Figures 3.1 through 3.3 for maps of
the corridor. Figure 3.1 shows the corridor between Wilshire and Pico Boulevards ( north of
the lane conversion site) and Pico Boulevard and Rose Ave ( part of the lane conversion site).
This entire part of the corridor lies within the City of Santa Monica. Figure 3.2 shows the
corridor between Rose Avenue and Washington Boulevard ( part of the conversion site) and
which belongs to the City of Los Angeles. Figure 3.3 shows the corridor between Washington
and Jefferson Boulevards ( south of the lane conversion site).
Figure 3.1 Lincoln Boulevard Corridor between Wilshire Boulevard and Rose Avenue
9
Figure 3.2 Lincoln Boulevard Corridor between Rose Avenue and Washington Boulevard
Figure 3.3 Lincoln Boulevard Corridor between Washington and Jefferson Boulevards
10
3.1.2 Scenarios
The objective of the simulation study was to test and compare different bus curb lane
operational strategies in a simulated environment. Six scenarios have been defined:
 Scenario 1: Do Nothing
No change is made to the existing state, whereas the curb lane remains as a parking
lane during the peak periods. This provides a baseline reference scenario with
which all other scenarios can be compared.
 Scenario 2: Bus Only Lane During Peak Periods
The curb lane operates as a bus only lane during peak periods. This scenario was
further subdivided into two sub‐ scenarios, labeled 2 and 2B. Scenario 2 consists of
both the Rapid 3 and Local 3 Lines being allowed to operate in the curb lane during
peak periods; scenario 2B consists of only the Rapid 3 Line being allowed to operate
in the curb lane during peak periods.
 Scenario 3: Mixed Traffic Lane During Peak Periods
The curb lane operates as a mixed traffic or general purpose lane open to all types
of vehicles during peak periods.
 Scenario 4: Special Vehicle Lane for Buses, Taxis, and Charter Buses During Peak
Periods
The curb lane operates as a special vehicle lane only open to buses ( Rapid 3 and
Local 3), taxis, and charter buses during peak periods.
 Scenario 5: Dynamic Dedicated Bus Rapid Transit ( BRT) Lane
The curb lane may dynamically convert from a mixed traffic lane to a bus only lane
when a Rapid 3 appears, and convert from a bus only lane back to a mixed traffic
lane when not used by a Rapid 3 bus.
The simulation study has implemented the above six scenarios in the simulation model,
and provided MOEs analysis results.
3.1.3 Data Collection
Four types of data have been collected for the simulation study:
 Geometric data such as lane and intersection geometry obtained through Google
maps
11
 Intersection turning volume data from the two cities for the intersections in their
respective jurisdictions
 Traffic signal data from the two cities, respectively, and
 Bus schedule and operations data from Big Blue Bus
In addition, drawings from Caltrans District 7 that illustrate parking restrictions were also
used in building the simulation model. The intersection turning volume data for different
intersections were collected in different years. The most recent are from 2007 ( intersections
in the City of Santa Monica), and the oldest are from 1997. To bring the volume data to a
comparable level, a growth factor of 3% per year is assumed. Furthermore, the turning
volume data could not be used directly in VISSIM, since the software only takes
origin‐ destination ( OD) demand as input for analysis of a stretch. Thus, a conversion from
turning volume data to OD demand data was performed.
3.2 Simulation Network Building
The simulation network building mainly involves the coding of network geometry, traffic
demand, and signal controllers.
3.2.1 Network Geometry
In VISSIM, links and connectors are used to model network geometry. Based on the
background image from Google Earth map, the geometry layout and lane channelization of
the studied stretch of Lincoln Boulevard were modeled into a simulation network.
As illustrated in Figure 3.4, the lane channelization of Scenario 1, also the existing state
of Lincoln Boulevard differs from that of other scenarios. In Scenario 1, each direction of the
cross section has two lanes and the curb lane is a parking lane not open to traffic during
peak periods. While in other scenarios, the curb lane converts to a lane open to particular
types of traffic, thus, each direction of the cross section has three lanes. As such, the curb
lane is set to be closed to all traffic for Scenario 1, but open to particular types of traffic for
all other scenarios.
12
Figure 3.4 Typical Lane Channelization of Lincoln Boulevard
Figure 3.5 is the background map cut from Google Earth. Figure 3.6 shows how the
geometric layout and lane channelization are modeled in the VISSIM simulation network.
3.2.2 Traffic Demand Coding
3.2.2.1 Non‐ Bus Traffic Demand
The data needs for coding of traffic demand is the Origin‐ Destination ( OD) matrix among
the inlets and outlets of the corridor. Figure 3.7 shows the OD zone numbers and
intersection numbers along Lincoln Boulevard.
13
Figure 3.5 Background Map Cut from Google Earth
( Intersection of Lincoln Blvd‐ Ocean Park Blvd)
Figure 3.6 Geometric Layout and Lane Channelization Modeled in Simulation Network
( Intersection of Lincoln Blvd‐ Ocean Park Blvd)
Curb lane closes to traffic for scenario 1,
but opens to traffic for other scenarios
14
Figure 3.7 OD Zone Numbers and Intersection Numbers along Lincoln Boulevard
15
3.2.2.2 Bus Traffic Demand
There are two bus routes along Lincoln Boulevard, which are the Rapid 3 and Local 3 Lines.
Each line has a departure rate of approximately 15 minutes during commute periods.
In VISSIM, there are three steps to code bus traffic demand. The first step is to place bus
stops along the corridor according to their locations. The second step is to define each bus
route and then add related bus stops to the route. The third step is to set the departure rate
of each route. Figure 3.8 is an illustration of bus traffic demand coding for the Rapid 3 line.
Figure 3.8 Illustration of Bus Traffic Demand Coding of Rapid 3 Line
3.2.3 Signal Controllers Coding
Timing plan( s), detectors deployment, and the positions of signal heads are the major
components for coding actuated signal controllers.
Bus stop belongs
to Rapid 3 Line
Bus stop not belongs
to Rapid 3 Line
Pico Blvd
Lincoln Blvd
16
VISSIM has a NEMA2 standard signal controller emulator module, which can simulate fully
actuated signal controllers as well as coordinate and semi‐ actuated coordinate signal
controllers. Through a transfer process, other signal controllers like Type 170, ASC/ 2070 can
also be emulated via the VISSIM NEMA module. Figure 3.9 is an illustration of signal
controller coding in VISSIM.
Figure 3.9 Illustration of Signal Controller Coding in VISSIM
2 NEMA = National Electrical Manufacturers Association
Signal Head
Signal controller configuration interface
Detector
17
3.3 Implementation of Scenarios in Simulation
3.3.1 Scenario 1 ‐ Do Nothing ( Baseline)
For the existing scenario, except for setting the curb lane closed to all traffic, no additional
configuration or setting is needed. Figure 3.10 is a snapshot of the simulation animation of
Scenario 1.
Figure 3.10 A Snapshot of the Simulation Animation of Scenario 1
It can be seen from Figure 3.10 that no vehicle including buses ( in green color) travels
on the curb lane, but right turn vehicles ( in red color) can move to the curb lane when
approaching very close to the intersection, and can thus make a needed right turn from the
curb lane.
3.3.2 Scenario 2 ‐ Bus Only Lane
For Scenario 2, the curb lane is set to be open to buses but still closed to non‐ bus vehicles.
18
Figure 3.11 is a snapshot of the simulation animation for Scenario 2. It can be seen from
Figure 3.11 that only buses ( in green color) are allowed to travel in the curb lane,
right‐ turning vehicles ( in red color) can change lanes to the curb lane when approaching very
close to the intersection, and can thus make the right turn from the curb lane.
Figure 3.11 Snapshot of the Simulation Animation for Scenario 2
19
Figure 3.12 Snapshot of the Simulation Animation for Scenario 3
3.3.3 Scenario 3: Mixed Traffic Lane
For Scenario 3, the curb lane is set to be open to all types of vehicles, that is, it is a general
purpose traffic lane. Figure 3.12 is a snapshot of the simulation animation of scenario 3. It
can be seen from Figure 3.12 that all types of vehicles are allowed on the curb lane.
3.3.4 Scenario 4: Special Lane for Buses, Taxis, and Charter Buses
For Scenario 4, a new vehicle class named Transit is defined, which includes the following
vehicle types: buses ( Rapid 3 plus Local 3 lines), taxis, and special‐ purpose charter buses.
The curb lane is then set to be open to Transit vehicles only. Figure 3.13 is a snapshot of the
simulation animation for Scenario 4.
It can also be seen from Figure 3.13 that only buses ( in green), taxis ( in blue) and charter
buses ( in blue) are allowed to travel on the curb lane of the upstream section.
20
Figure 3.13 Snapshot of the Simulation Animation of Scenario 4
3.3.5. Scenario 5: Dynamic Dedicated Bus Rapid Transit Lane
3.3.5.1. Dynamic BRT Lane Operation Rules
Basically, for a subject link between two intersections, when there is a bus approaching from
an upstream link, the curb lane of the subject link will convert from a mixed traffic lane to a
bus only lane. Then, the curb lane will convert back from a bus only lane to a mixed traffic
lane after the bus leaves the subject link. Figure 3.14 is an illustration of related settings for
dynamic BRT lane operation. Below are the rules of the dynamic BRT lane operation:
a. A link consists of an approach section and an upstream section. When the BRT lane
operation is triggered, the curb lane of the upstream section will be a bus only lane, the curb
lane of the approach section will be a right turn and bus only lane.
b. When a bus is detected by the Bus Approaching Detector ( BAD), the accumulated
counter number of BAD increases by 1. Then, if currently the curb lane of the subject link is a
mixed traffic lane, a new conversion to BRT lane will be triggered and indicator lights along
the curb of the subject link will turn on.
c. When a bus is detected by the Bus Departing Detector ( BDD), the accumulated
counter number of BAD increases by 1. If the accumulated counter number of BDD is equal
Charter Bus
Bus
Taxi
21
to that of BAD, which means there is no bus on the curb lane, then the indicator lights of the
subject link will turn off. If the accumulated counter number of BDD is less than that of BAD,
which means there are one or more buses still in the curb lane, then the indicator lights will
remain on. The Dynamic Dedicated BRT lane is very similar to what is referred to in the
literature as Intermittent Bus Lanes ( Viegas, 2007), ( Currie, 2008), and ( Eichler, 2005). Viegas
and Currie discuss implementations of intermittent bus lanes in Lisbon, Portugal and
Melbourne, Australia, respectively.
Figure 3.14 Illustration of Related Settings for Dynamic BRT Lane Operation
Bus & Right turn Only
Bus Only
Bus Only
Bus Only
Bus Approaching Detector
Bus Leaving Detector
Subject Link
Approach Section
Upstream Section
22
3.3.5.2 Implementing Dynamic BRT Lane Operation via VISSIM COM Programming
VISSIM provides a COM3 interface which can be used to realize some additional functions
not provided by the standard module. Through COM programming, we can implement
dynamic BRT lane operation during the simulation, which is not available in the standard
VISSIM module.
Figures 3.15 through 3.18 show major stages for the dynamic BRT lane operation during the
simulation, described as follows:
 Figure 3.15 shows the situation when the curb lane is open to all traffic. The
vehicles in red are right turning vehicles while those in black are through vehicles.
 Figure 3.16 shows the situation when a bus ( in green) has just passed by the Bus
Approaching Detector of the subject link, which triggered the curb lane of the
subject link converting from a mixed traffic lane to a bus only lane. Since it is the
initial period of the lane conversion, there are some non‐ bus vehicles already on the
approach section that may keep traveling on the curb lane.
 Figure 3.17 shows the situation when the curb lane has converted to a bus only lane
and all the non‐ bus vehicles have cleared off the curb lane. It can be seen from this
figure that only buses ( in green) are allowed on the curb lane and right turning
non‐ bus vehicles ( in red) can change to the curb lane of the approach section to
make the required turn.
 Figure 3.18 shows the situation when a bus has just passed by the Bus Departing
Detector, since there is no other bus on the curb lane of the subject link, thus the
curb lane has just converted from a bus only lane back to a mixed traffic lane. It can
be seen from the figure that non‐ bus vehicles have already changed lanes to travel
in the curb lane.
3 COM = communication
23
Figure 3.15 Snapshot when the Curb Lane Opens to Mixed Traffic
Figure 3.16 Snapshot after a Bus has just passed the Bus Approaching Detector
A bus just passed by the
Bus Approaching Detector
Bus Approaching Detector
Non- bus vehicle keep
traveling on the curb lane
24
Figure 3.17 Snapshot when the curb lane has Converted to a Bus Only Lane for a while
Figure 3.18 Snapshot when a Bus has just passed by the Bus Departing Detector
25
3.4 Simulation Model Calibration
Scenario 1 as previously defined represents the current traffic situation along the corridor
and is used for model calibration. Bus travel times, calculated from bus GPS data, were used
as the “ ground truth” to calibrate the model. More specifically, GPS data that was obtained
from Big Blue Bus included arrival and departure times at time points along the Local and
Rapid # 3 routes; however, because a minimum of two time points were required along the
segment coded in the simulation model to calculate travel time, only data from Local # 3
buses could be used as the Rapid # 3 has only one time point along the coded segment. The
two time points along the corridor are at Ocean Park and Washington Boulevards. Traffic
demand as well as simulation model parameters were calibrated so the travel times from the
simulation model match the travel times resulting from GPS data; the calibrated demand
and model parameters were then used in the other scenarios to evaluate the various
operating strategies. Calibration was performed for both northbound ( NB) and southbound
( SB) directions and for AM and PM peak periods and Table 3.1 compares the average travel
times from the two sources ( GPS data vs. simulation results). All simulated travel times fell
within 12% of GPS‐ observed data; however, more noteworthy is that for NB AM peak and SB
PM peak – the travel direction and time period combinations under consideration for lane
conversion – errors were within 5% of ground truth.
Table 3.1 Calibration Results: Comparison of Average Bus Travel Times
Direction
Time
Period
From GPS Data
( seconds)
From Simulation Result
( seconds) Percentage Difference
NB AM 644.02 671.21 4.2%
SB AM 498.60 557.72 11.9%
NB PM 536.28 564.76 5.3%
SB PM 797.10 760.10 ‐ 4.6%
NB = Northbound
SB = Southbound
26
3.5 Measures of Effectiveness Analysis
Several Measures of Effectiveness ( MOEs) were selected and used in determining the traffic
impacts – both for non‐ buses as well as buses – under the various scenarios. Such MOEs
consist of
 Delay4
 Travel time
 Speed
 Queue length5
The MOEs analysis is based on the links between Pico Blvd and Washington Blvd. Table 3.2
gives the numbers and definitions of the links used in the impact analysis evaluation.
Table 3.2 Numbers and Definitions of Links for Evaluation
Direction
Link
Number
Start Intersection End Intersection
Length
meters ( feet)
Southbound
1 Pico Ocean Park 903.4 ( 2,957.3)
2 Ocean Park Ashland 331.9 ( 1,086.5)
3 Ashland Marine/ Navy 302.2 ( 989.3)
4 Marine/ Navy Rose 350.7 ( 1,148.1)
5 Rose Brooks 541.3 ( 1,772.0)
6 Brooks California 151.1 ( 494.6)
7 California Superba 333.2 ( 1,090.8)
8 Superba Venice 488.4 ( 1,598.8)
9 Venice Washington 566.8 ( 1,855.5)
Northbound
10 Washington Venice 562.1 ( 1,840.1)
11 Venice Superba 464.6 ( 1,520.9)
12 Superba California 338.1 ( 1,106.8)
13 California Brooks 146.8 ( 480.6)
4 Vehicle delay on a link is defined as the difference between the vehicle’s actual travel time and its travel time
over this link under free flow conditions. The delay over a 15‐ minute time interval during a peak period across
a particular link is the average of such vehicle delays over all vehicles traveling on this link during this time
interval.
5 Queue length is defined per second; and the queue length on a link is calculated. In a 15‐ minute time interval,
there are 15* 60 queue lengths; the average queue is the sum of all these 15* 60 queue lengths divided by
15* 60.
27
Direction
Link
Number
Start Intersection End Intersection
Length
meters ( feet)
14 Brooks Rose 509.7 ( 1,668.6)
15 Rose Marine/ Navy 343.3 ( 1,123.8)
16 Marine/ Navy Ashland 281.5 ( 921.5)
17 Ashland Ocean Park 341.1 ( 1,116.6)
18 Ocean Park Pico 902.0 ( 2,952.8)
The total simulation time of each run is 3 hours consisting of 6 statistical time intervals each
of 30 minutes in length. For southbound, the OD demand during the 4‐ 7 PM peak period is
used; for northbound, the OD demand during the 7‐ 10 AM peak period is used.
Results for the various MOEs have been derived for the 6 time intervals ( 30 minute time
periods) per link for the AM and PM peak periods for all links6 across each of the scenarios.
To display all such results would require approximately 170 figures or tables, an amount
which could overwhelm the reader in detail without necessarily contributing to an
understanding of the general findings. Moreover, because the links are not uniformly
equivalent across characteristics such as length, network geometry, and others, there will be
variation from link to link.
To better represent the analysis results and to improve understanding of the findings, we
present below in Tables 3.3 through 3.12 the average value for each MOE during the two
three‐ hour peak periods ( delay, travel time, speed, and queue length) for non‐ buses, Rapid 3
buses, and Local 3 buses for each link across the six scenarios. For example, to show all
results for the delay MOE for non‐ buses would require 17 tables or figures ( one for each link)
to display the results for each 30‐ minute statistical time interval of the simulation for the
appropriate 3‐ hour peak period. Instead, for each link we use the value for the delay MOE
averaged over the 6 30‐ minute time periods for both the AM and PM peak periods and
6 There are 18 links however on Link 18 ( see Table 3.1), the Rapid 3 must leave the bus lane in order to prepare
to make a left turn on to Pico Blvd.; so MOEs were not captured for this link. While this affects the total
corridor‐ wide MOEs for the northbound direction ( see Table 3.1) in the AM peak, it does not change the
general behavior patterns for MOE values for Scenarios 2 through 5 relative to Scenario 1.
28
present the aggregated results in a single table. In this study, for scenario 4, we assume the
percentages of the traffic demand of taxis and charter buses are 2% and 1%, respectively.
3.5.1. Comparative Analysis Across Scenarios
In this section we present and compare the findings from the corridor simulation runs for
each measure of effectiveness across the various scenarios.
3.5.1.1 Delay
In this section we discuss the delay MOE for non‐ buses, the Rapid 3 bus and the Local 3 bus.
Non‐ Buses
Table 3.3 presents the results for each link across all scenarios. Recall that Links 1 through 9
represent the southbound direction of travel during the PM peak and Links 10 through 17
represent the northbound direction of travel during the AM peak. The numbers in
parentheses for a given scenario show the percentage change of the non‐ bus delay MOE for
that scenario relative to Scenario 1, the Do Nothing or Baseline Scenario. For example, from
Table 3.3 for Link 2, the non‐ bus delay for Scenarios 2, 3, 4, and 5, decrease relative to
Scenario 1 by, respectively, 36.8%, 54.4%, 38.6%, and 47.4%.
For more than 80% ( 14/ 17) of the links, non‐ bus delay decreases across Scenarios 2 through
5 relative to Scenario 1. There are, however, a few links with increases in delay for non‐ buses
for particular scenarios.
Typically, one would intuitively expect to observe that delay decreases from Scenario 1 to
Scenario 2, Scenario 4, Scenario 5, and finally to Scenario 3 and thus that Scenario 3 would
dominate – have the best value for ‐‐ the delay MOE; we expect this pattern because for
Scenarios 2 and 4 there is no change in the number of lanes for non‐ buses; however buses
are removed from the flow of traffic. In Scenarios 5 and 3 an extra lane for non‐ bus travel is
made available either all the time ( Scenario 3) or when Rapid 3 and Local 3 buses are not
present ( Scenario 5). This pattern is generally true on average on a link‐ by‐ link basis with a
few exceptions ( Table 3.3); it is definitely true on a corridor level basis ( Tables 3.4 and 3.5).
29
Figure 3.19 displays more of the full range of such percentage variation for non‐ bus delay
over all links of alternative Scenarios ( 2, 3, 4, and 5) relative to Scenario 1 ( Do Nothing). For
example, for Scenario 2 the average percentage change over all links relative to Scenario 1
for non‐ bus delay is a 16.5% decrease with a range between a 2.8% increase in non‐ bus
delay to a 71.4% decrease in non‐ bus delay. Analogous figures are provided for each of the
other MOEs discussed. We can see from Figure 3.19 how with rare exception across the
scenarios the delay MOE decreases relative to Scenario 1.
We make the following additional observations from Table 3.3 and Figure 3.19:
 For the corridor as a whole, adding a General Purpose lane ( Scenario 3) for use
during the peak periods shows the most improvement in delay of Scenarios 2
through 5 relative to the “ Do Nothing” baseline ( Scenario 1); however, at the
individual link level during the peak periods, link variability contributes to other
scenarios achieving the delay decreases relative to Scenario 1 surpassing that of
Scenario 3.
 Scenario 5 may be viewed as a hybrid scenario between Scenarios 2 and 3 because it
allows mixed flow traffic ( like Scenario 3) to operate in the curbside lane when no
buses ( Rapid 3 or Local 3) are present then excludes such traffic ( like Scenario 2)
when such buses are present. One would then expect simulation results for Scenario
5 to be between those of Scenarios 2 and 3, which is nearly the case across all links.
Certainly, this is the case for the overall corridor value relative to Scenarios 2 and 3.
Table 3.3 Average Non‐ Bus Delay ( seconds) per Corridor Link over Peak Periods
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5 ( Dynamic
Dedicated BRT)
Southbound ( PM Peak Period)
1 21.1 21.7 (+ 2.8%) 15.9 (‐ 24.6%) 21.6 (+ 2.4%) 18.1 (‐ 14.2%)
2 5.7 3.6 (‐ 36.8%) 2.6 (‐ 54.4%) 3.5 (‐ 38.6%) 3.0 (‐ 47.4%)
3 15.8 15.8 ( 0.0%) 11.6 (‐ 26.6%) 15.8 ( 0.0%) 11.9 (‐ 24.7%)
30
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5 ( Dynamic
Dedicated BRT)
4 12.4 11.6 (‐ 6.5%) 11.9 (‐ 4.0%) 11.4 (‐ 8.1%) 11.5 (‐ 7.3%)
5 15.5 15.7 (+ 1.3%) 14.9 (‐ 3.9%) 16.1 (+ 3.9%) 16.1 (+ 3.9%)
6 2.8 0.8 (‐ 71.4%) 1.1 (‐ 60.7%) 0.7 (‐ 75.0%) 1.0 (‐ 64.3%)
7 12.4 10.9 (‐ 12.1%) 5.9 (‐ 52.4%) 10.0 (‐ 19.4%) 6.9 (‐ 44.4%)
8 48.0 37.9 (‐ 21.0%) 22.5 (‐ 53.1%) 32.0 (‐ 33.3%) 22.6 (‐ 52.9%)
9 28.0 28.3 (+ 1.1%) 16.3 (‐ 41.8%) 25.9 (‐ 7.5%) 19.0 (‐ 32.1%)
Northbound ( AM Peak Period)
10 45.8 41.5 (‐ 9.4%) 23.0 (‐ 49.8%) 45.7 (‐ 0.2%) 28.7 (‐ 37.3%)
11 8.0 7.7 (‐ 3.8%) 7.9 (‐ 1.3%) 7.9 (‐ 1.3%) 7.0 (‐ 12.5%)
12 3.6 2.5 (‐ 30.6%) 2.5 (‐ 30.6%) 2.6 (‐ 27.8%) 2.1 (‐ 41.7%)
13 2.9 2.3 (‐ 20.7%) 2.0 (‐ 31.0%) 2.1 (‐ 27.6%) 2.0 (‐ 31.0%)
14 13.9 13.6 (‐ 2.2%) 11.5 (‐ 17.3%) 12.5 (‐ 10.1%) 10.8 (‐ 22.3%)
15 4.1 2.8 (‐ 31.7%) 1.6 (‐ 61.0%) 2.5 (‐ 39.0%) 1.9 (‐ 53.7%)
16 4.6 3.0 (‐ 34.8%) 2.7 (‐ 41.3%) 2.7 (‐ 41.3%) 2.7 (‐ 41.3%)
17 27.4 26.3 (‐ 4.0%) 18.8 (‐ 31.4%) 25.3 (‐ 7.7%) 20.2 (‐ 26.3%)
Figure 3.19 Range in Percentage (%) Variation for Non‐ Bus Delay over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
31
Table 3.4 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( non‐ bus) basis. Table 3.5 transforms the results from
Table 3.4 by accounting for the total number of vehicles ( non‐ buses) traveling along the
corridor during the two peak periods. As expected, we observe that Scenario 3 performs the
best of alternative scenarios; however, Scenario 5 performs the closest to Scenario 3 relative
to the change in non‐ bus delay for the corridor as a whole due to the fact that when the
Rapid 3 and Local 3 buses are not present, all non‐ bus vehicles are allowed in the bus lane.
Table 3.4 Average Non‐ Bus Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
2.7 2.4 (‐ 11.1%) 1.7 (‐ 37.0%) 2.3 (‐ 14.8%) 1.8 (‐ 33.3%)
Northbound
( AM Peak)
1.8 1.7 (‐ 5.6%) 1.2 (‐ 33.3%) 1.7 (‐ 5.6%) 1.3 (‐ 27.8%)
Table 3.5 Total Non‐ Bus Corridor Delay ( hours) over Peak Periods
Direction
( Time
Period)
Scenario
1 ( Do
Nothing)
Scenario
2 ( Rapid 3
+ Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
35.3 32.4
(‐ 8.2%)
23.1 (‐ 34.6%) 30.4 (‐ 13.9%) 24.7 (‐ 30.0%)
Northbound
( AM Peak)
23.6 21.2
(‐ 10.2%)
15.2 (‐ 35.6%) 21.4 (‐ 9.3%) 16.2 (‐ 31.4%)
Rapid 3 Buses
Rapid 3 bus delay is expected to decrease from Scenario 1 to Scenario 3, then to Scenario 4
32
and Scenario 5 and finally to Scenario 2. We expect this pattern because for Scenario 3 Rapid
3 buses continue to travel with all other vehicles as in Scenario 1 though with the availability
of an additional lane during the peak periods; however, this changes for Scenarios 4, 5, and 2
with Rapid 3 buses traveling only with Local 3 buses for all three scenarios and taxis and
chartered buses for Scenario 4. From the simulation data, we found that this is the case for
some links but not for all links ( Table 3.6); however, it is definitely true on a corridor level
basis ( Tables 3.7 and 3.8). However, high fluctuation results were also found in two links:
Links 6 and 16. What’s more surprising is that, for some links, the Rapid 3 bus delay even
increases from Scenario 1 to Scenario 3.
We then did further analysis of the simulation process and found that fluctuation of bus
arrival times at traffic signals is the main reason for this. Although the bus dispatch time
follows the same schedule in each scenario, there are still some variations of the Rapid 3 bus
arrival time at each signal. A bus arrives at green time in Scenario 1 may experience much
lower delay than in Scenario 3 if it happens to arrive in red time in Scenario 3. Since the
statistical time interval length is 30 minutes, there are only three bus arrivals on average
within each statistical interval. As such, the discrepancy delay of even one bus may
significantly influence the statistical result in that interval. Regardless, in higher congestion
level links such as Links 8 and 10, almost every bus may have to stop at the signal, thus other
factors, like the lane number and lane operation measures will be more influential to the bus
delay.
We make the following observations from Table 3.6 and Figure 3.20:
 Approximately 60% of all links across all scenarios show Rapid 3 bus delay decreases
 Scenarios 3 and 5, which allow non‐ buses to travel in the added curbside lane have
the most instances of Rapid 3 bus delay increases across all links
 Figure 3.20a and 3.20b show the percentage change for all scenarios relative to
Scenario 1 for all links and for all but Links 6 and 16, respectively. Figure 3.20b
removes the effect of the high fluctuation values in Links 6 and 16 and provides a
picture of the percentage change for all scenarios relative to Scenario 1 that is more
33
consistent with expectation. In particular, the average percentage changes for
Scenarios 2 through 5 indicate delay decreases relative to Scenario 1.
Table 3.6 Average Rapid 3 Bus Delay ( seconds) per Corridor Link over Peak Periods
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound ( PM Peak Period)
1 40.9 31.8 (‐ 22.2%) 43.1 (+ 5.4%) 35.7 (‐ 12.7%) 36.5 (‐ 10.8%)
2 35.3 26.3 (‐ 25.5%) 32.3 (‐ 8.5%) 31.4 (‐ 11.0%) 29.6 (‐ 16.1%)
3 2.7 3.3 (+ 22.2%) 4.9 (+ 81.5%) 1.4 (‐ 48.1%) 4.5 (+ 66.7%)
4 2.1 0.0 (‐ 100.0%) 1.9 (‐ 9.5%) 1.3 (‐ 38.1%) 0.8 (‐ 61.9%)
5 48.8 27.2 (‐ 44.3%) 38.8 (‐ 20.5%) 36.3 (‐ 25.6%) 29.2 (‐ 40.2%)
6 0.6 3.9 (+ 550.0%) 5.6 (+ 833.3%) 3.5 (+ 483.3%) 2.4 (+ 300.0%)
7 41.5 26.9 (‐ 35.2%) 37.3 (‐ 10.1%) 30.2 (‐ 27.2%) 30.0 (‐ 27.7%)
8 32.0 6.7 (‐ 79.1%) 19.6 (‐ 38.8%) 15.9 (‐ 50.3%) 14.5 (‐ 54.7%)
9 65.5 41.7 (‐ 36.3%) 50.6 (‐ 22.7%) 38.7 (‐ 40.9%) 33.7 (‐ 48.5%)
Northbound ( AM Peak Period)
10 78.2 35.5 (‐ 54.6%) 49.4 (‐ 36.8%) 35.6 (‐ 54.5%) 36.0 (‐ 54.0%)
11 32.4 25.7 (‐ 20.7%) 32.3 (‐ 0.3%) 24.6 (‐ 24.1%) 22.8 (‐ 29.6%)
12 4.1 5.1 (+ 24.4%) 2.5 (‐ 39.05) 3.1 (‐ 24.4%) 2.3 (‐ 43.9%)
13 30.3 24.6 (‐ 18.8%) 30.5 (+ 0.7%) 25.6 (‐ 15.5%) 23.6 (‐ 22.1%)
14 13.9 8.1 (‐ 41.7%) 17.6 (+ 26.6%) 9.3 (‐ 33.1%) 15.7 (+ 12.9%)
15 35.3 22.1 (‐ 37.4%) 27.1 (‐ 23.2%) 22.1 (‐ 37.4%) 22.8 (‐ 35.4%)
16 0.3 1.6 (+ 433.3%) 0.4 (+ 33.3%) 2.6 (+ 766.7%) 0.6 (+ 100.0%)
17 48.7 28.1 (‐ 42.3%) 40.1 (‐ 17.7%) 25.2 (‐ 48.3%) 33.9 (‐ 30.4%)
34
Figure 3.20a Range in Percentage (%) Variation for Rapid 3 Bus Delay over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Figure 3.20b Range in Percentage (%) Variation for Rapid 3 Bus Delay for all links except
Links 6 and 16 of Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing).
Table 3.7 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( Rapid 3 bus) basis. Table 3.8 transforms the results
from Table 3.7 by accounting for the total number of Rapid 3 buses traveling along the
corridor during the two peak periods. For the corridor as a whole for a Rapid 3 bus delay
decreases between approximately 29% and 38% for alternative scenarios 2, 4, and 5 in the
35
southbound direction and 37% ‐ 39% for northbound direction; accounting for the number
of Rapid 3 buses during each peak period shows a percentage delay reduction range
between 31% ‐ 39% for alternative scenarios 2, 4, and 5 southbound and 35% ‐ 40% for the
northbound direction, which are considerably more reduction than experienced by Scenario
3.
Table 3.7 Average Rapid 3 Bus Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
4.5 2.8 (‐ 37.8%) 3.9
(‐ 13.3%)
3.2 (‐ 28.9%) 3.0 (‐ 33.3%)
Northbound
( AM Peak)
4.1 2.5 (‐ 39.0%) 3.3
(‐ 19.5%)
2.5 (‐ 39.0%) 2.6 (‐ 36.6%)
Table 3.8 Total Rapid 3 Bus Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
9.0 5.5 (‐ 38.9%) 7.7
(‐ 14.4%)
6.2 (‐ 31.1%) 6.0 (‐ 33.3%)
Northbound
( AM Peak)
8.1 5.0 (‐ 38.3%) 6.7
(‐ 17.3%)
4.9 (‐ 39.5%) 5.3 (‐ 34.6%)
Local Buses
Local 3 bus delay is expected to follow the same behavior pattern as Rapid 3 bus delay, that
is, to decrease from Scenario 1 to Scenario 3, then to Scenario 4 and Scenario 5 and finally to
Scenario 2. From the simulation data, we found that again this is the case for some links
( Table 3.9); however, it is definitely true on a corridor level basis ( Tables 3.10 and 3.11).
36
We make the following additional observations from Table 3.9 and Figure 3.21:
 Nearly 90% of all links across all scenarios show Local 3 bus delay decreases
 Scenario 3 is the only scenario to have instances where the Local 3 bus delay
increases
 Figure 3.21 indicates very similar behavior among Scenarios 4, 5, and 2, which is to
be expected.
Table 3.9 Average Local 3 Bus Delay ( seconds) per Corridor Link over Peak Periods
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound ( PM Peak Period)
1 91.7 73.8 (‐ 19.5%) 91.8 (+ 0.1%) 80.2 (‐ 12.5%) 77.2 (‐ 15.8%)
2 36.1 27.1 (‐ 24.9%) 31.4 (‐ 13.0%) 28.9 (‐ 19.9%) 27.1 (‐ 24.9%)
3 39.5 28.9 (‐ 26.8%) 32.9 (‐ 16.7%) 29.7 (‐ 24.8%) 28.9 (‐ 26.8%)
4 52.0 39.0 (‐ 25.0%) 46.6 (‐ 10.4%) 37.6 (‐ 27.7%) 37.3 (‐ 28.3%)
5 118.4 97.5 (‐ 17.7%) 108.7 (‐ 8.2%) 99.4 (‐ 16.0%) 95.2 (‐ 19.6%)
6 42.9 32.1 (‐ 25.2%) 39.3 (‐ 8.4%) 30.4 (‐ 29.1%) 31.4 (‐ 26.8%)
7 75.5 60.3 (‐ 20.1%) 70.9 (‐ 6.1%) 60.1 (‐ 20.4%) 60.8 (‐ 19.5%)
8 102.7 66.2 (‐ 35.5%) 85.8 (‐ 16.5%) 66.8 (‐ 35.0%) 76.3 (‐ 25.7%)
9 86.9 74.8 (‐ 13.9%) 81.8 (‐ 5.9%) 68.5 (‐ 21.2%) 78.9 (‐ 9.2%)
Northbound ( AM Peak Period)
10 94.3 48.5 (‐ 48.6%) 62.1 (‐ 34.1%) 54.0 (‐ 42.7%) 51.6 (‐ 45.3%)
11 59.9 51.6 (‐ 13.9%) 55.2 (‐ 7.8%) 50.6 (‐ 15.5%) 49.4 (‐ 17.5%)
12 58.9 46.0 (‐ 21.9%) 48.8 (‐ 17.1%) 46.1 (‐ 21.7%) 46.6 (‐ 20.9%)
13 29.5 28.0 (‐ 5.1%) 29.4 (‐ 0.3%) 25.0 (‐ 15.3%) 23.8 (‐ 19.3%)
14 55.4 46.7 (‐ 15.7%) 59.3 (+ 7.0%) 48.9 (‐ 11.7%) 47.5 (‐ 14.3%)
37
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
15 43.4 22.5 (‐ 48.2%) 27.8 (‐ 35.9%) 22.5 (‐ 48.2%) 22.4 (‐ 48.4%)
16 38.5 25.4 (‐ 34.0%) 27.9 (‐ 27.5%) 26.2 (‐ 31.9%) 25.4 (‐ 34.0%)
17 67.9 51.8 (‐ 23.7%) 56.0 (‐ 17.5%) 51.4 (‐ 24.3%) 54.8 (‐ 19.3%)
Figure 3.21 Range in Percentage (%) Variation for Local 3 Bus Delay over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.10 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( Local 3 bus) basis. Table 3.11 transforms the results
from Table 3.10 by accounting for the total number of Local 3 buses traveling along the
corridor during each of the two peak periods. We observe again how alternative scenarios 2,
4, and 5 outperform Scenario 3 in terms of percentage delay reduction for both the
southbound and northbound directions.
38
Table 3.10 Average Local 3 Bus Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
10.8 8.3 (‐ 23.1%) 9.8 (‐ 9.3%) 8.4 (‐ 22.2%) 8.6 (‐ 20.4%)
Northbound
( AM Peak)
7.5 5.3 (‐ 29.3%) 6.1 ( 18.7%) 5.4 (‐ 28.0%) 5.4 (‐ 28.0%)
Table 3.11 Total Local 3 Bus Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
21.2 16.7 (‐ 21.2%) 19.6 (‐ 7.5%) 16.7 (‐ 21.2%) 17.1 (‐ 19.3%)
Northbound
( AM Peak)
14.8 10.7 (‐ 27.7%) 12.4
(‐ 16.2%)
11.0 (‐ 25.7%) 10.9 (‐ 26.4%)
3.5.1.2 Travel Time
In this section we present the travel time MOE for non‐ buses, Rapid 3 and Local 3 buses.
Non‐ Bus
Simulation results are shown in Tables 3.12, 3.13, and 3.14, and in Figure 3.22. The expected
behavior pattern for the travel time MOE for non‐ buses is the same as that for the delay
MOE, that is, to decrease from Scenario 1 to Scenario 2, Scenario 4, Scenario 5, and finally to
Scenario 3. While this pattern is not true for each of the links in Table 3.12, it is true on a
corridor level as shown in Tables 3.13 and 3.14.
We make the following observations from Table 3.12 and Figure 3.22:
 More than ¾ of all links ( 13/ 17) across all scenarios show non bus delay decreases
39
 Scenario 3 is the only scenario to record delay reductions for all links
 Figure 3.22 indicates very similar behavior among Scenarios 4, 5, and 2, which is to
be expected.
Table 3.12 Average Non Bus Travel Time ( seconds) per Corridor Link over Peak Periods
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound ( PM Peak Period)
1 75.5 76.9 (+ 1.9%) 71.1 (‐ 5.8%) 76.8 (+ 1.7%) 73.3 (‐ 2.9%)
2 26.4 23.9 (‐ 9.5%) 22.9 (‐ 13.3%) 23.8 (‐ 9.8%) 23.3 (‐ 11.7%)
3 33.8 34.2 (+ 1.2%) 30.1 (‐ 10.9%) 34.3 (+ 1.5%) 30.3 (‐ 10.4%)
4 34.2 33.5 (‐ 2.0%) 33.6 (‐ 1.8%) 33.2 (‐ 2.9%) 33.3 (‐ 2.6%)
5 48.1 48.7 (+ 1.2%) 47.9 (‐ 0.4%) 49.1 (+ 2.1%) 49.1 (+ 2.1%)
6 11.9 10.0 (‐ 16.0%) 10.3 (‐ 13.4%) 10.0 (‐ 16.0%) 10.3 (‐ 13.4%)
7 32.8 31.3 (‐ 4.6%) 26.3 (‐ 19.8%) 30.4 (‐ 7.3%) 27.3 (‐ 16.8%)
8 77.7 67.7 (‐ 12.9%) 52.3 (‐ 32.7%) 61.8 (‐ 20.5%) 52.4 (‐ 32.6%)
9 62.7 63.0 (+ 0.5%) 50.9 (‐ 18.8%) 60.5 (‐ 3.5%) 53.6 (‐ 14.5%)
Northbound ( AM Peak Period)
10 80.9 75.8 (‐ 6.3%) 57.3 (‐ 29.2%) 79.9 (‐ 1.2%) 63.0 (‐ 22.1%)
11 37.0 36.1 (‐ 2.4%) 36.3 (‐ 1.9%) 36.3 (‐ 1.9%) 35.3 (‐ 4.6%)
12 25.0 23.1 (‐ 7.6%) 23.2 (‐ 7.2%) 23.3 (‐ 6.8%) 22.8 (‐ 8.8%)
13 11.5 11.3 (‐ 1.7%) 11.0 (‐ 4.3%) 11.1 (‐ 3.5%) 11.0 (‐ 4.3%)
14 45.1 44.7 (‐ 0.9%) 42.6 (‐ 5.5%) 43.6 (‐ 3.3%) 41.9 (‐ 7.1%)
15 25.3 23.8 (‐ 5.9%) 22.6 (‐ 10.7%) 23.5 (‐ 7.1%) 22.9 (‐ 9.5%)
16 22.0 20.2 (‐ 8.2%) 19.8 (‐ 10.0%) 19.9 (‐ 9.5%) 19.8 (‐ 10.0%)
17 47.7 47.1 (‐ 1.3%) 39.6 (‐ 17.0%) 46.1 (‐ 3.4%) 41.0 (‐ 14.0%)
40
Figure 3.22 Range in Percentage (%) Variation for Non Bus Travel Time over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.13 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( non‐ bus) basis. Table 3.14 transforms the results from
Table 3.13 by accounting for the total number of non‐ buses traveling along the corridor
during the two peak periods. As expected, we observe that Scenario 3 performs the best of
alternative scenarios; however, Scenario 5 performs the closest to Scenario 3 relative to the
change in non‐ bus delay for the corridor as a whole due to the fact that when the Rapid 3
and Local 3 buses are not present, all non‐ bus vehicles are allowed in the bus lane.
Table 3.13 Average Non‐ Bus Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
6.7 6.5 (‐ 3.0%) 5.8 (‐ 13.4%) 6.3 (‐ 6.0%) 5.9 (‐ 11.9%)
Northbound
( AM Peak)
4.9 4.7 (‐ 4.1%) 4.2 (‐ 14.3%) 4.7 (‐ 4.1%) 4.3 (‐ 12.2%)
41
Table 3.14 Total Non‐ Bus Corridor Travel Time ( hours) over Peak Periods
Direction
( Time Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
89.2 87.5 (‐ 1.9%) 78.2 (‐ 12.3%) 85.5 (‐ 4.1%) 79.8 (‐ 10.5%)
Northbound
( AM Peak)
65.3 62.5 (‐ 4.3%) 56.5 (‐ 13.5%) 62.7 (‐ 4.0%) 57.5 (‐ 11.9%)
Rapid 3 Bus
Simulation results are shown in Tables 3.15, 3.16, and 3.17, and in Figure 3.23. The expected
behavior pattern for the travel time MOE for Rapid 3 buses is the same as that for the delay
MOE for Rapid 3 buses, that is, to decrease from Scenario 1 to Scenario 3, Scenario 4,
Scenario 5, and finally to Scenario 2. While this pattern is not true for each of the links in
Table 3.15, it is true on a corridor level basis as shown in Tables 3.16 and 3.17.
We make the following observations from Table 3.15 and Figure 3.23:
 Approximately 60% of all links ( 10/ 17) across all scenarios show Rapid 3 bus delay
decreases
 No scenario records travel time decreases for all links; however, Scenarios 2, 4, and 5
have the fewest instances of travel time increases compared to Scenario 3.
 Figure 3.23 indicates very similar behavior among Scenarios 2, 4, and 5, which is to
be expected. There is very little variation among the average percentage reduction in
travel time for these three scenarios: 12.6%, 11.0%, and 11.5% for Scenarios 2, 4, and
5, respectively
42
Table 3.15 Average Rapid Bus Travel Time ( seconds) per Corridor Link over Peak Periods
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound ( PM Peak Period)
1 108.6 99.4 (‐ 8.5%) 110.8 (+ 2.0%) 103.4 (‐ 4.8%) 104.2 (‐ 4.1%)
2 61.0 52.0 (‐ 14.8%) 58.1 (‐ 4.8%) 57.2 (‐ 6.2%) 55.4 (‐ 9.2%)
3 25.1 25.7 (+ 2.4%) 27.4 (+ 9.2%) 23.8 (‐ 5.2%) 26.9 (+ 7.2%)
4 28.7 26.6 (‐ 7.3%) 28.5 (‐ 0.7%) 27.9 (‐ 2.8%) 27.4 (‐ 4.5%)
5 89.5 67.9 (‐ 24.1%) 79.5 (‐ 11.2%) 77.0 (‐ 14.0%) 69.9 (‐ 21.9%)
6 12.0 15.3 (+ 27.5%) 17.0 (+ 41.7%) 14.9 (+ 24.2%) 13.9 (+ 15.8%)
7 67.0 52.3 (‐ 21.9%) 62.7 (‐ 6.4%) 55.6 (‐ 17.0%) 55.4 (‐ 17.3%)
8 68.9 43.6 (‐ 36.7%) 56.5 (‐ 18.0%) 52.8 (‐ 23.4%) 51.4 (‐ 25.4%)
9 108.4 84.6 (‐ 22.0%) 93.4 (‐ 13.8%) 81.6 (‐ 24.7%) 76.6 (‐ 29.3%)
Northbound ( AM Peak Period)
10 122.0 79.3 (‐ 35.0%) 93.2 (‐ 23.6%) 79.4 (‐ 34.9%) 79.8 (‐ 34.6%)
11 68.3 61.6 (‐ 9.8%) 68.2 (‐ 0.1%) 60.5 (‐ 11.4%) 58.7 (‐ 14.1%)
12 30.6 31.6 (+ 3.3%) 29.0 (‐ 5.2%) 29.5 (‐ 3.6%) 28.7 (‐ 6.2%)
13 41.2 35.5 (‐ 13.8%) 41.4 (+ 0.5%) 36.5 (‐ 11.4%) 34.5 (‐ 16.3%)
14 52.6 46.8 (‐ 11.0%) 56.3 (+ 7.0%) 48.0 (‐ 8.7%) 54.4 (+ 3.4%)
15 61.8 48.6 (‐ 21.4%) 53.6 (‐ 13.3%) 48.6 (‐ 21.4%) 49.4 (‐ 20.1%)
16 21.9 23.3 (+ 6.4%) 22.1 (+ 0.9%) 24.2 (+ 10.5%) 22.2 (+ 1.4%)
17 74.2 53.6 (‐ 27.8%) 65.4 (‐ 11.9%) 50.7 (‐ 31.7%) 59.4 (‐ 19.9%)
43
Figure 3.23 Range in Percentage (%) Variation for Rapid 3 Bus Travel Time over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.16 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( Rapid 3 bus) basis. Table 3.17 transforms the results
from Table 3.16 by accounting for the total number of Rapid 3 buses traveling along the
corridor during the two peak periods. For the corridor as a whole for a Rapid 3 bus, delay
decreases between approximately 14% and 18% for alternative scenarios 2, 4, and 5 in the
southbound direction and 18% ‐ 20% for northbound direction; accounting for the number
of Rapid 3 buses during each peak period shows a percentage delay reduction range
between 16% ‐ 19% for alternative scenarios 2, 4, and 5 southbound and 18% ‐ 20% for the
northbound direction, which are considerably greater percentage reduction than
experienced by Scenario 3.
44
Table 3.16 Average Rapid 3 Bus Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time Period)
Scenario 1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4 ( Bus
+ Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
9.5 7.8 (‐ 17.9%) 8.9 (‐ 6.3%) 8.2 (‐ 13.7%) 8.0 (‐ 15.8%)
Northbound
( AM Peak)
7.9 6.3 (‐ 20.3%) 7.2 (‐ 8.9%) 6.3 (‐ 20.3%) 6.5 (‐ 17.7%)
Table 3.17 Total Rapid 3 Bus Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time Period)
Scenario 1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4 ( Bus
+ Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
19.0 15.4 (‐ 18.9%) 17.5 (‐ 7.9%) 16.0 (‐ 15.8%) 16.0 (‐ 15.8%)
Northbound
( AM Peak)
15.8 12.7 (‐ 19.6%) 14.3 (‐ 9.5%) 12.6 (‐ 20.3%) 12.9 (‐ 18.4%)
Local 3 Bus
Local 3 bus travel time is expected to follow the same behavior pattern as Rapid 3 bus travel
time, that is, to decrease from Scenario 1 to Scenario 3, then to Scenario 4 and Scenario 5
and finally to Scenario 2. From the simulation data, we found that again this is the case for
some links ( Table 3.18); however, it is definitely true on a corridor level basis ( Tables 3.19
and 3.20).
We make the following additional observations from Table 3.18 and Figure 3.24:
 Nearly 90% of all links across all scenarios show Local 3 bus travel time decreases
 Scenario 3 is the only scenario to have instances where the Local 3 bus delay
increases
 Figure 3.24 indicates very similar behavior among Scenarios 2, 4, and 5, which is to
45
be expected. The average percentage reductions in travel time across all links for
these scenarios are, respectively, 16.6%, 16.6%, and 16.4%.
Table 3.18 Average Local Bus Travel Time ( seconds) per Corridor Link over Peak Periods
Link # Scenario 1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound ( PM Peak Period)
1 159.3 (%) 141.4 (‐ 11.2%) 159.4 (+ 0.1%) 147.8 (‐ 7.2%) 144.8 (‐ 9.1%)
2 61.9 (%) 52.9 (‐ 14.5%) 57.2 (‐ 7.6%) 54.7 (‐ 11.6%) 52.9 (‐ 14.5%)
3 61.9 (%) 51.3 (‐ 17.1%) 55.3 (‐ 10.7%) 52.2 (‐ 15.7%) 51.3 (‐ 17.1%)
4 78.6 (%) 65.5 (‐ 16.7%) 73.2 (‐ 6.9%) 64.1 (‐ 18.4%) 63.9 (‐ 18.7%)
5 159.1 (%) 138.2 (‐ 13.1%) 149.3 (‐ 6.2%) 140.1 (‐ 11.9%) 135.9 (‐ 14.6%)
6 54.3 (%) 43.5 (‐ 19.9%) 50.6 (‐ 6.8%) 41.8 (‐ 23.0%) 42.7 (‐ 21.4%)
7 100.9 (%) 85.8 (‐ 15.0%) 96.4 (‐ 4.5%) 85.6 (‐ 15.2%) 86.3 (‐ 14.5%)
8 139.6 (%) 103.1 (‐ 26.1%) 122.7 (‐ 12.1%) 103.7 (‐ 25.7%) 113.3 (‐ 18.8%)
9 129.7 (%) 117.7 (‐ 9.3%) 124.7 (‐ 3.9%) 111.3 (‐ 14.2%) 121.7 (‐ 6.2%)
Northbound ( AM Peak Period)
10 138.1 (%) 92.3 (‐ 33.2%) 105.8 (‐ 23.4%) 97.8 (‐ 29.2%) 95.4 (‐ 30.9%)
11 95.8 (%) 87.5 (‐ 8.7%) 91.1 (‐ 4.9%) 86.4 (‐ 9.8%) 85.3 (‐ 11.0%)
12 85.3 (%) 72.4 (‐ 15.1%) 75.2 (‐ 11.8%) 72.5 (‐ 15.0%) 73.1 (‐ 14.3%)
13 40.5 (%) 39.0 (‐ 3.7%) 40.4 (‐ 0.2%) 36.0 (‐ 11.1%) 34.7 (‐ 14.3%)
14 94.1 (%) 85.4 (‐ 9.2%) 98.1 (+ 4.3%) 87.6 (‐ 6.9%) 86.2 (‐ 8.4%)
15 69.9 (%) 49.0 (‐ 29.9%) 54.3 (‐ 22.3%) 49.0 (‐ 29.9%) 49.0 (‐ 29.9%)
16 60.2 (%) 47.1 (‐ 21.8%) 49.6 (‐ 17.6%) 47.9 (‐ 20.4%) 47.1 (‐ 21.8%)
17 93.3 (%) 77.3 (‐ 17.1%) 81.4 (‐ 12.8%) 76.9 (‐ 17.6%) 80.2 (‐ 14.0%)
46
Figure 3.24 Range in Percentage (%) Variation for Local 3 Bus Travel Time over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.19 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( Local 3 bus) basis. Table 3.20 transforms the results
from Table 3.19 by accounting for the total number of Local 3 buses traveling along the
corridor during the two peak periods. We observe again how alternative scenarios 2, 4, and
5 outperform Scenario 3 in terms of percentage travel time reduction for both the
southbound and northbound directions.
Table 3.19 Average Local 3 Bus Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time Period)
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
15.8 13.3 (‐ 15.8%) 14.8 (‐ 6.3%) 13.4 (‐ 15.2%) 13.5 (‐ 14.6%)
Northbound
( AM Peak)
11.3 9.2 (‐ 18.6%) 9.9 (‐ 12.4%) 9.2 (‐ 18.6%) 9.2 (‐ 18.6%)
47
Table 3.20 Total Local 3 Bus Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time Period)
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
31.1 26.6 (‐ 14.5%) 29.6 (‐ 4.8%) 26.7 (‐ 14.1%) 27.1 (‐ 12.9%)
Northbound
( AM Peak)
22.4 18.3 (‐ 18.3%) 20.2 (‐ 9.8%) 18.8 (‐ 16.1%) 18.7 (‐ 16.5%)
3.5.1.3 Speed
In this section we present the simulation results for the speed MOE for non‐ buses, Rapid 3
and Local 3 buses. Each scenario’s speed was calculated as a weighted average of the
link‐ level speeds for that scenario with each link’s weight equal to the proportion of that
link’s length over the entire corridor.
Non‐ Buses
Tables 3.21 and 3.22, and Figure 3.25 show the simulation findings for the speed MOE for
non‐ buses. Table 3.21 shows the value of non‐ bus speed for each link ( overall 30‐ minute
time periods during the peak periods) together with the percentage changes exhibited in
parentheses. It is expected that non‐ bus speeds would increase the most for Scenarios 3 and
5 because for each of these scenarios additional lane space is provided for non‐ buses and for
most links this is true as well as being true for the average corridor‐ wide speed ( Table 3.22).
We make the following additional observations:
 Approximately 76% of all links across all scenarios show non‐ bus speed increases
 At worst, non‐ bus speeds decrease at most 2.0% ( Link 5, Scenarios 4 and 5)
 Figure 3.25 shows for each scenario the percentage change in speed for non‐ buses
compared with Scenario 1. The figure depicts that the speed changes are
overwhelmingly increases and that the magnitude of speed decreases are very small.
48
Table 3.21 Average Non‐ Bus Speed ( mph) per Corridor Link over Peak Periods
Link # Scenario 1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound ( PM Peak Period)
1 26.7 26.2 (‐ 1.8%) 28.4 ( 6.2%) 26.3 (‐ 1.7%) 27.5 ( 3.0%)
2 28.1 31.0 ( 10.5%) 32.3 ( 15.3%) 31.1 ( 10.9%) 31.8 ( 13.3%)
3 20.0 19.7 (‐ 1.2%) 22.4 ( 12.3%) 19.7 (‐ 1.5%) 22.3 ( 11.6%)
4 22.9 23.4 ( 2.1%) 23.3 ( 1.8%) 23.6 ( 3.0%) 23.5 ( 2.7%)
5 25.1 24.8 (‐ 1.2%) 25.2 ( 0.4%) 24.6 (‐ 2.0%) 24.6 (‐ 2.0%)
6 28.3 33.7 ( 19.0%) 32.7 ( 15.5%) 33.7 ( 19.0%) 32.7 ( 15.5%)
7 22.7 23.8 ( 4.8%) 28.3 ( 24.7%) 24.5 ( 7.9%) 27.2 ( 20.1%)
8 14.0 16.1 ( 14.8%) 20.8 ( 48.6%) 17.6 ( 25.7%) 20.8 ( 48.3%)
9 20.2 20.1 (‐ 0.5%) 24.9 ( 23.2%) 20.9 ( 3.6%) 23.6 ( 17.0%)
Northbound ( AM Peak Period)
10 15.5 16.6 ( 6.7%) 21.9 ( 41.2%) 15.7 ( 1.3%) 19.9 ( 28.4%)
11 28.0 28.7 ( 2.5%) 28.6 ( 1.9%) 28.6 ( 1.9%) 29.4 ( 4.8%)
12 30.2 32.7 ( 8.2%) 32.5 ( 7.8%) 32.4 ( 7.3%) 33.1 ( 9.6%)
13 28.5 29.0 ( 1.8%) 29.8 ( 4.5%) 29.5 ( 3.6%) 29.8 ( 4.5%)
14 25.2 25.5 ( 0.9%) 26.7 ( 5.9%) 26.1 ( 3.4%) 27.2 ( 7.6%)
15 30.3 32.2 ( 6.3%) 33.9 ( 11.9%) 32.6 ( 7.7%) 33.5 ( 10.5%)
16 28.6 31.1 ( 8.9%) 31.7 ( 11.1%) 31.6 ( 10.6%) 31.7 ( 11.1%)
17 16.0 16.2 ( 1.3%) 19.2 ( 20.5%) 16.5 ( 3.5%) 18.6 ( 16.3%)
49
Figure 3.25 Range in Percentage (%) Variation for Non‐ Bus Speed over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.22 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( non‐ bus) basis. As expected, we observe that Scenario
3 performs the best of alternative scenarios; however, Scenario 5 performs the closest to
Scenario 3 relative to the increase in non‐ bus speed for the corridor as a whole since when
the Rapid 3 and Local 3 buses are not present, all non‐ bus vehicles are allowed access to the
bus lane.
Table 3.22 Average Non‐ Bus Corridor Speed ( mph) over Peak Periods
Direction
( Time Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
23.0 23.6 (+ 2.6%) 26.1 (+ 13.5%) 24.0 (+ 4.3%) 25.5 (+ 10.9%)
Northbound
( AM Peak)
24.4 25.5 (+ 4.5%) 27.3 (+ 11.9%) 25.6 (+ 4.9%) 27.1 (+ 11.1%)
50
Rapid 3 Bus
The next measure of effectiveness examined was speed with respect to Rapid 3 buses. Tables
3.23 and 3.24, and Figure 3.26 show the simulation findings for the speed MOE for Rapid 3
buses. As previously described, Table 3.23 shows the value of Rapid 3 bus speeds for each
link ( overall 30‐ minute time periods during the peak periods) together with the percentage
changes exhibited in parentheses. It is expected that Rapid 3 bus speeds would increase the
most for Scenarios 2, 4, and 5 because for each of these scenarios exclusive additional lane
space is provided for Rapid 3 buses and for most links this is true. Moreover, it is certainly
true for the average corridor‐ wide speeds for Rapid 3 buses ( Table 3.24). We make the
following additional observations:
 Approximately 60% of all links across all scenarios show bus speed increases
 At worst, Rapid 3 bus speeds decrease at most 29.4% ( Link 6, Scenario 3)
Table 3.23 Average Rapid 3 Bus Speed ( mph) per Corridor Link over Peak Periods
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4 ( Bus +
Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound ( PM Peak Period)
1 18.6 20.3 ( 9.3%) 18.2 (‐ 2.0%) 19.5 ( 5.0%) 19.4 ( 4.2%)
2 12.1 14.2 ( 17.3%) 12.8 ( 5.0%) 13.0 ( 6.6%) 13.4 ( 10.1%)
3 26.9 26.2 (‐ 2.3%) 24.6 (‐ 8.4%) 28.3 ( 5.5%) 25.1 (‐ 6.7%)
4 27.3 29.4 ( 7.9%) 27.5 ( 0.7%) 28.1 ( 2.9%) 28.6 ( 4.7%)
5 13.5 17.8 ( 31.8%) 15.2 ( 12.6%) 15.7 ( 16.2%) 17.3 ( 28.0%)
6 28.1 22.0 (‐ 21.6%) 19.8 (‐ 29.4%) 22.6 (‐ 19.5%) 24.3 (‐ 13.7%)
7 11.1 14.2 ( 28.1%) 11.9 ( 6.9%) 13.4 ( 20.5%) 13.4 ( 20.9%)
8 15.8 25.0 ( 58.0%) 19.3 ( 21.9%) 20.6 ( 30.5%) 21.2 ( 34.0%)
9 11.7 15.0 ( 28.1%) 13.5 ( 16.1%) 15.5 ( 32.8%) 16.5 ( 41.5%)
Northbound ( AM Peak Period)
10 10.3 15.8 ( 53.8%) 13.5 ( 30.9%) 15.8 ( 53.7%) 15.7 ( 52.9%)
11 15.2 16.8 ( 10.9%) 15.2 ( 0.1%) 17.1 ( 12.9%) 17.7 ( 16.4%)
12 24.7 23.9 (‐ 3.2%) 26.0 ( 5.5%) 25.6 ( 3.7%) 26.3 ( 6.6%)
51
Link # Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4 ( Bus +
Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
13 8.0 9.2 ( 16.1%) 7.9 (‐ 0.5%) 9.0 ( 12.9%) 9.5 ( 19.4%)
14 21.6 24.3 ( 12.4%) 20.2 (‐ 6.6%) 23.7 ( 9.6%) 20.9 (‐ 3.3%)
15 12.4 15.8 ( 27.2%) 14.3 ( 15.3%) 15.8 ( 27.2%) 15.5 ( 25.1%)
16 28.7 27.0 (‐ 6.0%) 28.4 (‐ 0.9%) 26.0 (‐ 9.5%) 28.3 (‐ 1.4%)
17 10.3 14.2 ( 38.4%) 11.6 ( 13.5%) 15.0 ( 46.4%) 12.8 ( 24.9%)
Figure 3.26 Range in Percentage (%) Variation for Rapid 3 Bus Speed over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.24 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( Rapid 3 bus) basis. For the corridor as a whole for a
Rapid 3 bus, speed increases between approximately 11% and 17% for alternative scenarios
2, 4, and 5 in the southbound direction and approximately 13% and 15% for the northbound
direction;
52
Table 3.24 Average Rapid 3 Bus Corridor Speed ( mph) over Peak Periods
Direction
( Time
Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4 ( Bus
+ Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak) 17.2 20.1 (+ 16.9%) 17.6 (+ 2.3%) 19.0 (+ 10.5%) 19.3 (+ 12.2%)
Northbound
( AM Peak) 16.5 18.9 (+ 14.5%) 17.3 (+ 4.8%) 19.0 (+ 15.2%) 18.6 (+ 12.7%)
Local 3 Buses
The next measure of effectiveness examined was speed for Local 3 buses. Tables 3.25 and
3.26, and Figure 3.27 show the simulation findings for the speed MOE for Local 3 buses. As
previously described, Table 3.25 shows the value of Local 3 bus speed for each link ( overall
30‐ minute time periods during the peak periods) together with the percentage changes
exhibited in parentheses. It is expected that Local 3 bus speeds would increase the most for
Scenarios 2, 4, and 5 because for each of these scenarios semi‐ exclusive lane space is
provided for Local 3 buses and for all links this is true. Moreover, it is certainly true for the
average corridor‐ wide speeds for Local 3 buses ( Table 3.26). We make the following
additional observations:
 Approximately 90% of all links across all scenarios show bus speed increases
 At worst, Local 3 bus speeds decrease at most 4.1% ( Link 14, Scenario 3)
 Figure 3.27 shows how rare, if ever, Local 3 bus speed decreases and this occurs for
Scenario 3 as expected.
53
Table 3.25 Average Local 3 Bus Speed ( mph) per Corridor Link over Peak Periods
Link # Scenario 1 ( Do
Nothing)
Scenario 2 ( Rapid
3 + Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound ( PM Peak Period)
1 12.7 14.3 ( 12.7%) 12.6 (‐ 0.1%) 13.6 ( 7.8%) 13.9 ( 10.0%)
2 12.0 14.0 ( 17.0%) 13.0 ( 8.2%) 13.5 ( 13.2%) 14.0 ( 17.0%)
3 10.9 13.1 ( 20.7%) 12.2 ( 11.9%) 12.9 ( 18.6%) 13.1 ( 20.7%)
4 10.0 12.0 ( 20.0%) 10.7 ( 7.4%) 12.2 ( 22.6%) 12.2 ( 23.0%)
5 7.6 8.7 ( 15.1%) 8.1 ( 6.6%) 8.6 ( 13.6%) 8.9 ( 17.1%)
6 6.2 7.8 ( 24.8%) 6.7 ( 7.3%) 8.1 ( 29.9%) 7.9 ( 27.2%)
7 7.4 8.7 ( 17.6%) 7.7 ( 4.7%) 8.7 ( 17.9%) 8.6 ( 16.9%)
8 7.8 10.6 ( 35.4%) 8.9 ( 13.8%) 10.5 ( 34.6%) 9.6 ( 23.2%)
9 9.8 10.7 ( 10.2%) 10.1 ( 4.0%) 11.4 ( 16.5%) 10.4 ( 6.6%)
Northbound ( AM Peak Period)
10 9.1 13.6 ( 49.6%) 11.9 ( 30.5%) 12.8 ( 41.2%) 13.2 ( 44.8%)
11 10.8 11.9 ( 9.5%) 11.4 ( 5.2%) 12.0 ( 10.9%) 12.2 ( 12.3%)
12 8.8 10.4 ( 17.8%) 10.0 ( 13.4%) 10.4 ( 17.7%) 10.3 ( 16.7%)
13 8.1 8.4 ( 3.8%) 8.1 ( 0.2%) 9.1 ( 12.5%) 9.4 ( 16.7%)
14 12.1 13.3 ( 10.2%) 11.6 (‐ 4.1%) 13.0 ( 7.4%) 13.2 ( 9.2%)
15 11.0 15.6 ( 42.7%) 14.1 ( 28.7%) 15.6 ( 42.7%) 15.6 ( 42.7%)
16 10.4 13.3 ( 27.8%) 12.7 ( 21.4%) 13.1 ( 25.7%) 13.3 ( 27.8%)
17 8.2 9.8 ( 20.7%) 9.4 ( 14.6%) 9.9 ( 21.3%) 9.5 ( 16.3%)
54
Figure 3.27 Range in Percentage (%) Variation for Local 3 Bus Speed over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.26 shows the average delay for the entire corridor length ( southbound and
northbound) on an individual vehicle ( Local 3 bus) basis for each of the two peak periods.
We observe again how alternative scenarios 2, 4, and 5 outperform Scenario 3 in terms of
percentage travel time reduction for both the southbound and northbound directions with a
range in average speed increases for scenarios 2, 4, and 5 of approximately 15% to 17%
( southbound) and 23% to 24% ( northbound). For Scenario 3, the average percentage speed
increase is approximately 6% ( southbound) and 14% ( northbound).
Table 3.26 Average Local 3 Bus Corridor Speed ( mph) over Peak Periods
Direction
( Time
Period)
Scenario 1
( Do
Nothing)
Scenario 2
( Rapid 3 + Local
3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated BRT)
Southbound
( PM Peak) 9.8 11.5 (+ 17.3%) 10.4 (+ 6.1%) 11.4 (+ 16.3%) 11.3 (+ 15.3%)
Northbound
( AM Peak) 10.0 12.4 (+ 24.0%) 11.4 (+ 14.0%) 12.3 (+ 23.0%) 12.4 (+ 24.0%)
55
3.5.1.4 Queue Length
The next measure of effectiveness examined was queue length. Tables 3.27 and 3.28, and
Figure 3.28 show the simulation findings for the queue length MOE. It is expected that
queue lengths would decrease the most for Scenarios 3 and 5 because for each of these
additional lane space is provided for non‐ buses and for most links this is true and is also true
on average over all links. We make the following observations from Table 3.27 and Figure
3.28:
 More than 50% of all links across all scenarios show queue length decreases
 Scenarios 3 and 5 have queue length decreases across all links
Table 3.27 Average Queue Length per Corridor Link over Peak Periods
Link # Scenario 1
( Do Nothing)
Scenario 2 ( Rapid
3 + Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound ( PM Peak Period)
1 28.2 31.3 (+ 11.0%) 13.7 (‐ 51.4%) 32.5 (+ 15.2%) 17.2 (‐ 39.0%)
2 4.9 3.8 (‐ 22.4%) 1.2 (‐ 75.5%) 3.8 (‐ 22.4%) 1.6 (‐ 67.3%)
3 27.6 30.1 (+ 9.1%) 11.7 (‐ 57.6%) 29.5 (+ 6.9%) 13.7 (‐ 50.4%)
4 16.8 17.5 (+ 4.2%) 11.2 (‐ 33.3%) 16.9 (+ 0.6%) 10.8 (‐ 35.7%)
5 22.1 31.7 (+ 43.4%) 15.3 (‐ 30.8%) 31.9 (+ 44.3%) 20.4 (‐ 7.7%)
6 3.2 0.2 (‐ 93.8%) 0.2 (‐ 93.8%) 0.1 (‐ 96.9%) 0.1 (‐ 96.9%)
7 16.5 16.0 (‐ 3.0%) 4.5 (‐ 72.7%) 14.5 (‐ 12.1%) 5.7 (‐ 65.5%)
8 90.9 70.2 (‐ 22.8%) 20.2 (‐ 77.8%) 56.1 (‐ 38.3%) 24.1 (‐ 73.5%)
9 43.1 46.1 (+ 7.0%) 12.0 (‐ 72.2%) 38.4 (‐ 10.9%) 16.0 (‐ 62.9%)
Northbound ( AM Peak Period)
10 67.7 60.8 (‐ 10.2%) 18.0 (‐ 73.4%) 68.7 (+ 1.5%) 27.7 (‐ 59.1%)
11 8.0 9.7 (+ 21.3%) 6.7 (‐ 16.3%) 9.7 (+ 21.3%) 5.8 (‐ 27.5%)
56
Link # Scenario 1
( Do Nothing)
Scenario 2 ( Rapid
3 + Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
12 1.5 0.8 (‐ 46.7%) 0.9 (‐ 40.0%) 1.5 ( 0.0%) 0.7 (‐ 53.3%)
13 2.9 2.6 (‐ 10.3%) 1.3 (‐ 55.2%) 2.2 (‐ 24.1%) 1.2 (‐ 58.6%)
14 18.2 21.4 (+ 17.6%) 10.7 (‐ 41.2%) 17.5 (‐ 3.8%) 10.5 (‐ 42.3%)
15 3.8 2.8 (‐ 26.3%) 0.5 (‐ 86.8%) 2.0 (‐ 47.4%) 0.8 (‐ 78.9%)
16 4.0 2.3 (‐ 42.5%) 1.2 (‐ 70.0%) 1.8 (‐ 55.0%) 1.2 (‐ 70.0%)
17 51.7 45.4 (‐ 12.2%) 19.6 (‐ 62.1%) 42.9 (‐ 17.0%) 23.4 (‐ 54.7%)
Figure 3.28 Range in Percentage (%) Variation for Queue Length over all links of
Alternative Scenarios ( 2, 3, 4, 5) Relative to Scenario 1 ( Do Nothing)
Table 3.28 shows the average queue length for the entire corridor length ( southbound and
northbound) for each of the two peak periods. We observe how alternative scenarios 3 and
5 outperform Scenarios 2 and 4 in terms of percentage queue length reduction for both the
southbound and northbound directions with a range in average queue length decreases for
scenarios 3 and 5 of approximately 57% to 64% ( southbound) and 55% to 62% ( northbound).
For Scenarios 2 and 4, the average percentage queue length decrease ranges between
57
approximately 3% to 11% ( southbound) and 7% to 8% ( northbound).
Table 3.28 Average Corridor Queue Length ( feet) over Peak Periods
Direction
( Time
Period)
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak) 28.1 27.4 (‐ 2.5%) 10.0 (‐ 64.4%) 24.9 (‐ 11.4%) 12.2 (‐ 56.6%)
Northbound
( AM Peak) 19.7 18.2 (‐ 7.6%) 7.4 (‐ 62.4%) 18.3 (‐ 7.1%) 8.9 (‐ 54.8%))
3.5.2 Major Corridor‐ wide Findings
Based on the simulation results, the performance of each scenario averaged over all the links
is summarized in Tables 3.28 through 3.31 for queue length, delay, travel time, and speed,
respectively. Tables 3.32 and 3.33 account for the number of vehicles ( non‐ buses and buses)
and show the total value for delay and travel time across all alternative scenarios. As could
be observed from the data, with the curb lane converted into a travel lane, the MOEs are all
improved compared with the do‐ nothing scenario, that is, delays decrease across all
alternative scenarios, travel times decrease across all alternative scenarios, speeds increase
across all alternative scenarios, and queue lengths decrease across all alternative scenarios;
however, no single alternative scenario does better than all other alternative scenarios
across all MOEs.
Among scenarios 2 through 5 on a corridor level basis, Scenario 2 has the lowest Rapid 3 and
Local 3 bus delay, lowest Rapid 3 and Local 3 travel time and highest Rapid 3 and Local 3 bus
speed, and Scenario 3 has the lowest non‐ bus delay, lowest non‐ bus travel time, highest
non‐ bus speed, and shortest queue length. These corridor‐ wide findings are shown in Table
3.30 and graphically in Figures 3.31 and 3.32 for the travel time MOE and in Table 3.29 and
graphically in Figures 3.29 and 3.30 for the delay MOE. However, Scenarios 4 and 5 give
58
values for delay, travel time, and speed for the Rapid 3 and Local 3 buses that are close to
Scenario 2’ s values. In fact they are not statistically different from each other in most cases
based on a set of non‐ parametric statistical tests7 that we conducted on data shown in
Tables 3.3, 3.6, 3.9, 3.12, 3.15, 3.18, 3.21, 3.23, and 3.25 for scenarios 2, 4, and 5.
Scenario 5, meanwhile, appears to be a good compromise between exclusive bus use and
entirely mixed traffic use, however, it also requires a lot more technology ( e. g. a warning
signal, either in the pavement or on the roadside, to communicate to vehicles that a bus is
approaching), and is considered an experimental scenario and needs more extensive
investigation if it is to be seriously considered to be implemented.
Table 3.29 Average Vehicle Corridor Delay ( minutes) over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
Non‐ Bus 2.7 2.4 (‐ 11.1%) 1.7 (‐ 37.0%) 2.3 (‐ 14.8%) 1.8 (‐ 33.3%)
Rapid 3 4.5 2.8 (‐ 37.8%) 3.9 (‐ 13.3%) 3.2 (‐ 28.9%) 3.0 (‐ 33.3%)
Local 3 10.8 8.3 (‐ 23.1%) 9.8 (‐ 9.3%) 8.4 (‐ 22.2%) 8.6 (‐ 20.4%)
Northbound
( AM Peak)
Non‐ Bus 1.8 1.7 (‐ 5.6%) 1.2 (‐ 33.3%) 1.7 (‐ 5.6%) 1.3 (‐ 27.8%)
Rapid 3 4.1 2.5 (‐ 39.0%) 3.3 (‐ 19.5%) 2.5 (‐ 39.0%) 2.6 (‐ 36.6%)
Local 3 7.5 5.3 (‐ 29.3%) 6.1 ( 18.7%) 5.4 (‐ 28.0%) 5.4 (‐ 28.0%)
7 We used the Mann‐ Whitney and Kruskal‐ Wallis tests, which are each nonparametric tests for the significance of the
difference between the distributions of independent samples; two such samples for the Mann‐ Whitney test and three or
more samples for the Kruskal‐ Wallis test.
59
Figure 3.29 Southbound Corridor Delay Across Scenarios
Figure 3.30 Northbound Corridor Delay Across Scenarios
60
Table 3.30 Average Vehicle Corridor Travel Time ( minutes) over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
Non‐ Bus 6.7 6.5 (‐ 3.0%) 5.8 (‐ 13.4%) 6.3 (‐ 6.0%) 5.9 (‐ 11.9%)
Rapid 3 9.5 7.8 (‐ 17.9%) 8.9 (‐ 6.3%) 8.2 (‐ 13.7%) 8.0 (‐ 15.8%)
Local 3 15.8 13.3 (‐ 15.8%) 14.8 (‐ 6.3%) 13.4 (‐ 15.2%) 13.5 (‐ 14.6%)
Northbound
( AM Peak)
Non‐ Bus 4.9 4.7 (‐ 4.1%) 4.2 (‐ 14.3%) 4.7 (‐ 4.1%) 4.3 (‐ 12.2%)
Rapid 3 7.9 6.3 (‐ 20.3%) 7.2 (‐ 8.9%) 6.3 (‐ 20.3%) 6.5 (‐ 17.7%)
Local 3 11.3 9.2 (‐ 18.6%) 9.9 (‐ 12.4%) 9.2 (‐ 18.6%) 9.2 (‐ 18.6%)
Figure 3.31 Southbound Corridor Travel Time Across Scenarios
61
Figure 3.32 Northbound Corridor Travel Time Across Scenarios
Table 3.31 Average Vehicle Corridor Speed ( mph) over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Scenario
1 ( Do
Nothing)
Scenario 2
( Rapid 3 +
Local 3)
Scenario 3
( General
Purpose)
Scenario 4
( Bus + Taxi)
Scenario 5
( Dynamic
Dedicated
BRT)
Southbound
( PM Peak)
Non‐ Bus 23.0 23.6 (+ 2.6%) 26.1 (+ 13.5%) 24.0 (+ 4.3%) 25.5 (+ 10.9%)
Rapid 3 17.2 20.1 (+ 16.9%) 17.6 (+ 2.3%) 19.0 (+ 10.5%) 19.3 (+ 12.2%)
Local 3 9.8 11.5 (+ 17.3%) 10.4 (+ 6.1%) 11.4 (+ 16.3%) 11.3 (+ 15.3%)
Northbound
( AM Peak)
Non‐ Bus 24.4 25.5 (+ 4.5%) 27.3 (+ 11.9%) 25.6 (+ 4.9%) 27.1 (+ 11.1%)
Rapid 3 16.5 18.9 (+ 14.5%) 17.3 (+ 4.8%) 19.0 (+ 15.2%) 18.6 (+ 12.7%)
Local 3 10.0 12.4 (+ 24.0%) 11.4 (+ 14.0%) 12.3 (+ 23.0%) 12.4 (+ 24.0%)
62
Table 3.32 Total Corridor Delay over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Units Scenario 1 ( Do
Nothing)
Scenario 2 ( Rapid 3
+ Local 3)
Scenario 3 ( General
Purpose)
Scenario 4 ( Bus +
Taxi)
Scenario 5 ( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
Non‐ Bus Hours 35.3 32.4 (‐ 8.2%) 23.1 (‐ 34.6%) 30.4 (‐ 13.9%) 24.7 (‐ 30.0%)
Rapid 3 Minutes 9.0 5.5 (‐ 38.9%) 7.7 (‐ 14.4%) 6.2 (‐ 31.1%) 6.0 (‐ 33.3%)
Local 3 Minutes 21.2 16.7 (‐ 21.2%) 19.6 (‐ 7.5%) 16.7 (‐ 21.2%) 17.1 (‐ 19.3%)
Northbound
( AM Peak)
Non‐ Bus Hours 23.6 21.2 (‐ 10.2%) 15.2 (‐ 35.6%) 21.4 (‐ 9.3%) 16.2 (‐ 31.4%)
Rapid 3 Minutes 8.1 5.0 (‐ 38.3%) 6.7 (‐ 17.3%) 4.9 (‐ 39.5%) 5.3 (‐ 34.6%)
Local 3 Minutes 14.8 10.7 (‐ 27.7%) 12.4 (‐ 16.2%) 11.0 (‐ 25.7%) 10.9 (‐ 26.4%)
63
Table 3.33 Total Corridor Travel Time over Peak Periods
Direction
( Time
Period)
Vehicle
Type
Units Scenario 1 ( Do
Nothing)
Scenario 2 ( Rapid 3
+ Local 3)
Scenario 3 ( General
Purpose)
Scenario 4 ( Bus +
Taxi)
Scenario 5 ( Dynamic
Dedicated BRT)
Southbound
( PM Peak)
Non‐ Bus Hours 89.2 87.5 (‐ 1.9%) 78.2 (‐ 12.3%) 85.5 (‐ 4.1%) 79.8 (‐ 10.5%)
Rapid 3 Minutes 19.0 15.4 (‐ 18.9%) 17.5 (‐ 7.9%) 16.0 (‐ 15.8%) 16.0 (‐ 15.8%)
Local 3 Minutes 31.1 26.6 (‐ 14.5%) 29.6 (‐ 4.8%) 26.7 (‐ 14.1%) 27.1 (‐ 12.9%)
Northbound
( AM Peak)
Non‐ Bus Hours 65.3 62.5 (‐ 4.3%) 56.5 (‐ 13.5%) 62.7 (‐ 4.0%) 57.5 (‐ 11.9%)
Rapid 3 Minutes 15.8 12.7 (‐ 19.6%) 14.3 (‐ 9.5%) 12.6 (‐ 20.3%) 12.9 (‐ 18.4%)
Local 3 Minutes 22.4 18.3 (‐ 18.3%) 20.2 (‐ 9.8%) 18.8 (‐ 16.1%) 18.7 (‐ 16.5%)
64
3.6 Recommendations and Conclusions
The decision to choose among the alternative scenarios
 Bus/ HOV ( Scenarios 2 or 4)
 General Purpose Lane ( Scenario 3)
 Hybrid ( Scenario 5)
mostly depends on how much the different values of MOEs really make a difference. To
determine that, a Level of Service ( LOS) comparison for both general traffic and transit
( instead of comparing the exact numbers) is made next.
The LOS for an arterial is based on its average travel speed ( Transportation Research Board,
1994). For Lincoln Blvd, which is an urban arterial, divided, with some parking at curbside,
the appropriate LOS values are the following:
Arterial LOS Values
LOS Speed ( mph)
A >= 30
B >= 24
C >= 18
D >= 14
E >= 10
F < 10
Based on these LOS values and findings from Table 3.31 ( and a weighted average of
southbound and northbound speeds), the LOS ratings resulting from the five scenarios are
shown in Table 3.34.
65
Table 3.34 LOS Values Comparison Across Scenarios
Measures of
Effectiveness
Scenario
1
Scenario
2
Scenario
3
Scenario
4
Scenario
5
Average Non‐ Bus
Speed ( mph)
23.7 24.6 26.7 24.8
26.3
Non‐ Bus LOS C B B B B
From Table 3.34, Scenarios 2, 4, and 5 provide the same LOS for non‐ bus traffic as Scenario 3
does, while providing better service ( in terms of reduced bus delay and increased bus speed)
to buses. Thus Scenarios 2, 4, and 5 appear to be better than Scenario 3; moreover, there is
another issue as it relates to Scenario 3 that needs to be considered:
 Possible impact of generated traffic, that is, diverted traffic ( trips shifted in time,
route and destination), and induced vehicle travel ( shifts from other modes, longer
trips and new vehicle trips). With the LOS improved along Lincoln Boulevard for
Scenario 3 relative to Scenario 1 more traffic will likely be attracted to the curbside
lane especially from the off‐ peak to the peak periods; and such growth in traffic
could result in deteriorated LOS again over time and thus would continue to favor the
alternative bus‐ only scenarios ( Litman, 2009), and ( Cervero, 2002). The amount of
traffic generated by a road project varies depending on site‐ specific conditions.
Generated traffic usually accumulates over several years and under typical urban
conditions, more than half of added capacity is filled within five years of project
completion by additional vehicle trips that would not otherwise occur, with
additional but slower growth in later years ( Litman, 2009).
Because Scenario 5 would require an investment of substantial technology and extensive
investigation if it is to be seriously considered for implementation, in the short term we
recommend that Scenarios 2 and 4 be pursued.
66
References
Viegas, J. M. et al., “ The Intermittent Bus Lane System: Demonstration in Lisbon”, 86th
Annual Meeting of the Transportation Research Board CD‐ ROM Compendium of Papers,
Transportation Research Board, Washington, D. C., January 2007.
Currie, G. and H. Lai, “ Intermittent and Dynamic Transit Lanes”, Transportation Research
Record: Journal of the Transportation Research Board, No. 2072, pp. 49‐ 56, Transportation
Research Board of the National Academies, Washington, D. C., 2008.
Eichler, M. D., “ Bus Lanes with Intermittent Priority: Assessment and Design”, Master’s
Thesis of City Planning, University of California, Berkeley, 2005.
Litman, T., Generated Traffic and Induced Travel – Implications for Transport Planning,
Victoria Transport Policy Institute, February 2009.
Cervero, R., “ Induced Demand: An Urban and Metropolitan Perspective”, Working Together
to Address Induced Demand: Proceedings of a Forum, Eno Transportation Foundation,
Washington, D. C., 2002.
Transportation Research Board, Highway Capacity Manual, Special Report 209, Washington,
D. C., 1994.
67
4.0 LINCOLN BOULEVARD CASE STUDY: RIDERSHIP IMPACTS ASSESSMENT
Direct modeling of transit ridership has emerged as an alternative to traditional four‐ step
travel‐ demand modeling for corridor and station‐ levels analyses ( Cervero, 2006). Direct
models estimate ridership as a function of station environments and transit service
features rather than using mode‐ choice results from large‐ scale models. This provides a
fine‐ grain resolution suitable for studying relationships between built environments,
transit services, and ridership. Moreover, the amount of resources needed to code a
network, set up a regional travel model, and then do a mode choice analysis favored a
sketch planning approach like a direct ridership model.
Because direct models predict demand for a specific node or location versus the
origin‐ destination attributes of a trip, some variables normally found in mode‐ choice
models, such as comparative travel times and prices of transit versus auto, are noticeably
absent. The comparative accessibility of station‐ area residents to jobs and shops via
transit versus auto are sometimes included in direct models, thus in this sense,
performance attributes of competitive modes are imbedded in the analyses.
Direct ridership models generally have small sample sizes since observations consist of
transit stations or stops. Thus degree of freedom constraints often limit the number of
variables that can be included as well as their specifications ( e. g., inclusion of interactive
terms). It is because of these limitations that direct models fall under the rubric of
sketch‐ planning tools. They provide order‐ of‐ magnitude insights for testing of various
system designs and land‐ use scenarios.
To date, direct modeling has been used to estimate station‐ and corridor‐ level ridership
for rail transit investments and expansion proposals in areas as diverse as
Charlotte‐ Mecklenburg County ( NC), St. Louis ( MO), the East Bay of the San Francisco Bay
Area, and Boise ( ID) ( Cervero, 1998; Cervero, 2004; Fehr and Peers, 2005). For a host of
reasons, including fiscal constraints and development densities that are too low for rail
investments, more and more U. S. cities and regions are turning to Bus Rapid Transit ( BRT)
68
as a cost‐ effective alternative to rail transit. As far as we know from the literature, no
direct ridership model has been estimated to date for a BRT proposal.
The remaining sections of this chapter present a Direct Ridership Model developed to
estimate ridership levels for a proposed dedicated bus‐ only lane Big Blue Bus BRT service
in Santa Monica, California along the Lincoln Boulevard corridor. The section is divided
into the following remaining sections. First, we discuss the sample frame used to conduct
the analysis as well as candidate variables that were considered for entry into the Direct
Ridership Model. This is followed by a presentation of a best‐ fitting regression model that
conforms with travel‐ demand theory, yields interpretable and statistically significant
results, and demonstrates a capacity to produce ridership estimates for existing Big Blue
Bus ( BBB) patronage that are reasonably accurate. The final section of the chapter uses
the validated model to estimate ridership for six BBB bus stops that are being considered
for a significant upgrade in BRT services – notably, the creation of a dedicated, bus‐ only
operating lane.
4.1 Modeling Approach and Sample
Limited real‐ world experiences with Bus Rapid Transit in the U. S. constrain the ability to
draw upon empirical experiences to inform ridership estimates. While foreign cities like
Curitiba, Brazil and Bogota, Colombia have accumulated considerable experience with
dedicated‐ lane BRT operations, vast cultural, socio‐ economic, and institutional differences
with the U. S. limit the use of empirical evidence from such settings.
Fortunately one of the most proactive regions of the U. S. in advancing BRT services has been
Southern California. The Metropolitan Transportation Authority ( MTA) phased in the Metro
Rapid Program between June 2000 and December 2000 with the goal of improving bus
speeds within urbanized Los Angeles County. Four pilot routes ‐‐ along Wilshire Boulevard
( 720), Broadway ( 745), Vermont Avenue ( 754) and Ventura Boulevard ( 750) – used Next Bus
technology at most stops to inform waiting customers of estimated bus arrival times. Metro
Rapid buses consist exclusively of low‐ floor buses and have their own distinctive color
69
scheme and markings. Other features include transit signal prioritization, frequent headways,
and comparatively long spacings between bus stops.
A new stage in BRT services was reached in October 2005 when MTA’s Metro Orange Line
opened. The Orange Line is one of the first “ full‐ service” BRT systems in the United States,
featuring a dedicated busway ( running on a disused rail corridor), high‐ capacity articulated
buses, “ rail‐ like” stations ( incorporating level boarding and off‐ board fare payment) and
headway‐ based schedules. The 14‐ mile route connects the western terminus of the Red Line
subway at North Hollywood with Warner Center, the third largest employment center in Los
Angeles County. As of late 2008, Southern California’s Metro Rapid Program consisted of 28
routes in total providing 450 directional miles of service. MTA operates all but two of the
routes. The Santa Monica Big Blue Bus operates Rapid 3 Line along Lincoln Boulevard and
Rapid 7 connecting downtown Santa Monica and the Rimpau Transit Center in Los Angeles
along Pico Boulevard. The Rapid 3 line is under consideration for conversion to higher end
BRT services with a dedicated bus lane, and is the subject of the ridership forecasts
presented in this section.
4.1.1 Sample Selection
In order to obtain a sample of sufficient size to draw statistically reliable inferences, 50 MTA
bus stop locations were sampled across 20 different Metro Rapid lines. Each location had a
stop on each side of a road, meaning ridership as well as service‐ level data were compiled
for both stops at each location. Additionally, in order to account for characteristics of BBB’s
own operating environment and to incorporate data for the BRT corridor of interest, data
were compiled for six bus stop locations of the Rapid 3 Line. Lastly, to reflect the
relationships between services and ridership for “ high end” BRT services, data for 13 Orange
Line stops were also compiled. Figure 4.1 shows the locations of the 69 total bus stop
locations that constituted the sample frame for our Direct Ridership modeling. Average daily
ridership data were obtained for each stop for October 2008. Accordingly, data for
explanatory variables were obtained for time periods as close as possible to the October
70
2008 date.
4.1.2 Model Specification and Variables
Direct Ridership models estimate boardings ( and/ or exits) at a stop or station for a defined
period of time ( e. g., daily) as a function of three key sets of variables related to each stop or
station:
Figure 4.1 Locations of 69 BRT bus stop observations used for estimating Direct Ridership
Model: 50 Metro Rapid stops, 13 Orange Lines stops, and 6 Rapid 3 stops
( 1) Service Attributes – e. g., frequency of buses ( headways, buses per hour),
operating speeds, feeder bus connections ( number of lines or buses), dedicated lane
71
( 0‐ 1 variable), vehicle brand/ marketing ( 0‐ 1 variable), etc.;
( 2) Location and Neighborhood Attributes – e. g., population and employment
densities, mixed land use measures ( 0‐ 1 scale), median household incomes and
vehicle ownership levels ( as proxies for levels of “ transit dependence”), distance to
nearest stop ( as a proxy for catchment size), accessibility levels ( e. g., number of jobs
that can be reached within 30 minutes over transit network in peak periods),
terminal station ( 0‐ 1 variable), street density ( e. g., directional miles of street divided
by land area), connectivity indices ( e. g., links/ nodes of street network), etc.; and
( 3) Bus Stop/ Site Attributes – e. g., bus shelters ( 0‐ 1), Next Bus passenger information
( 0‐ 1), bus benches ( 0‐ 1), far‐ side bus stops ( 0‐ 1), park‐ and‐ ride lots ( 0‐ 1, or number
of spaces), bus bulbs ( 0‐ 1), etc.
Often, service attributes like bus headways do not vary within bus lines though they can and
often do vary across lines. Travel‐ demand theory holds that transit riders, particularly choice
users, are more sensitive to service quality and operating features than other factors.
Accordingly we expected some measures of a bus stop’s service quality to enter the Direct
Ridership model. Other attributes of the operations, like fare levels, are usually so similar
across passengers who board buses at each stop that they are not of use for Direct Ridership
models. The one service‐ related variable that we felt would significantly enter the model
was whether a stop received an exclusive‐ lane service. No factor can begin to make
bus‐ transit more time‐ competitive with the private car than operating in a bus‐ only lane.
Accordingly, the 13 Orange Line bus stops were “ dummy‐ coded” ( binary 0‐ 1 variable) to
denote their qualitatively higher service levels than the other bus stops in the data base.
Location variables aim to capture attributes of the immediate operating environment, such
as nearby densities and distances to nearest stop. The farther a bus stop is from the next
nearest stop, for instance, typically the stop’s geographical catchment area increases in size.
Being a terminal station often boosts ridership even more since end‐ line stations also serve
72
big catchments8. If stops with large catchment average high population densities, boardings
at the stop should go up even more. And if nearby residents average relatively low incomes
and car ownership rates, then boarding can be expected to further rise. Factors like dense
street networks with high connectivity ( i. e., link‐ to‐ node ratios) can bump up ridership, at
the margin, by expediting pedestrian flows to stops. One issue pertains to the appropriate
size of the geographic buffer drawn around bus stops to capture neighborhood attributes.
In keeping with other research on the walkability to transit, we opted to create ½ mile
buffers around stops. Overlaying these buffers onto census tract polygons allowed variables
like population density within ½ mile of a stop to be estimated using GIS techniques.
Lastly, some of the bus‐ stop attribute variables – such as the presence of bus shelters or
far‐ side bus stops – are binary ( 0‐ 1) and thus are used in the models as dummy variables.
These variables largely represent passenger amenities and relative to variables that
traditional mode choice theory holds infl