ON-STREET MID-BLOCK OFF-LINE BUS Faheema... · 2020. 1. 20. · Publications Journal Articles...

MODELLING BUS STOP CAPACITY FOR

ON-STREET, MID-BLOCK, OFF-LINE BUS

STOPS

Faheema Hisham

B.Sc. (Mathematics)

Submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

School of Civil Engineering and Built Environment

Science and Engineering Faculty

Queensland University of Technology

2020

In memory of my loving father,

who always believed in me and taught me to believe in myself…

KEYWORDS

On-street bus stop, mid-block bus stop, off-line bus stop, bus service, capacity,

TCQSM, dwell time, traffic blockage, bus-bus interference, loading area, bus stop

operation, reliability, failure rate, operating margin, clearance time, re-entry delay,

yield-to-bus.

ABSTRACT

Arterial roads are a type of on-street bus (OSB) facility where buses and other forms

of traffic share the same lanes. The performance of an OSB facility is highly dependent

on the interaction between buses and other vehicles. When the adjacent lane carries a

high volume of general traffic, the interaction between buses and traffic will affect the

capacity and Quality of Service (QOS) of the bus facility (Hisham et al., 2019a). The

critical stop, which has the lowest capacity generally governs the bus facility capacity.

This is due to queue spillback upstream of the bus stop or a possible capacity

bottleneck downstream of the bus stop. It is essential to understand the operation of

any potentially critical bus stop in order to understand and manage such a facility.

The traditional methodology for capacity estimation is given in the third edition of

Transit Capacity and Quality of Service Manual (TCQSM (Kittelson and Associates,

2013a)). TCQSM provides the definition for capacity estimation as the product of

capacity of each of its loading areas, traffic blockage adjustment factor and number of

effective loading areas. However, TCQSM model lacks accounting for various

dimensions of an on-street bus stop operation, such as adjacent lane traffic volume,

degree of saturation and upstream wait time. This research has developed a reliability-

based bus capacity analysis methodology for on-street bus stops, particularly located

mid-block.

OSB operations are problematic in nature. This research through detailed analysis

identified elements and processes that influence capacity of an on-street bus stop.

Factors that influence the capacity were quantified as time components in the total

loading area processing time, in order to obtain detailed information about the time

contribution. To assist with the theoretical modelling and to visualize the operation of

an on-street bus stop, a microscopic simulation model was developed using AIMSUN.

Based on these models, a novel capacity analysis methodology was developed

comprising of new models, ‘Bus stop capacity with adjacent lane requirements’ and

‘Bus stop Maximum Working Capacity’. In addition to the traditional parameters, the

new models accounts for adjacent lane traffic flow rate, degree of saturation of the

adjacent lane and degree of saturation of the loading area, and upstream average

waiting time. The applicability of this new model was shown through demonstrative

case studies. Additionally, these case studies revealed the importance of incorporating

adjacent lane traffic volume and degree of saturation in the capacity estimation of on-

street, mid-block, off-line bus stops.

Publications

Journal Articles

Hisham, F., Bunker, J. M. & Bhaskar, A. 2019. Capacity estimation of on-street,

mid-block, off-line bus stops considering yield-to-bus rule. Transportation

Research Record: Journal of the Transportation Research Board.

Hisham, F., Bunker, J. M. & Bhaskar, A. 2019. Incorporating practical degree of

saturation in capacity estimation of on-street, mid-block, off-line bus stops .

Transportation Research Record: Journal of the Transportation Research Board.

(In-press)

Conference Proceedings

Hisham, F., Bunker, J. M. & Bhaskar, A. Development of a modified bus stop

capacity model. Transportation Research Board (TRB) 97th annual meeting, 7-

11 January 2018 Washington Dc. Transportation Research Board of the national

academies.

Hisham, F., Bunker, J. M. & Bhaskar, A. Improving capacity estimation of high

volume on-street bus facilities with yield-to-bus rule. Australasian Transport

Research Forum (ATRF), 40th, 2018 proceedings, 2018b Darwin Convention

Centre, Australia.

TABLE OF CONTENTS

Keywords ................................................................................................................................ iii

Abstract .....................................................................................................................................v

Publications ............................................................................................................................ vii

List of Figures ....................................................................................................................... xiii

List of Tables ....................................................................................................................... xvii

List of Abbreviations ............................................................................................................ xix

Statement of Original Authorship ......................................................................................... xxi

Acknowledgements ............................................................................................................. xxiii

Introduction ...................................................................................... 1

Overview ........................................................................................................................1

Background .....................................................................................................................1

Research Motivation .......................................................................................................4

Thesis Statement .............................................................................................................6

Research Questions .........................................................................................................6

Research Objectives........................................................................................................6

Scope ..............................................................................................................................7

Significance ....................................................................................................................8

Thesis Outline .................................................................................................................8

A Review of Measures, Modelling Approaches and Evaluation of

Bus Stop Capacity .................................................................................................... 11

Overview ......................................................................................................................11

Role of a Bus Stop ........................................................................................................11

Bus Stop Capacity.........................................................................................................15

Improving Bus Stop Capacity .......................................................................................31

Gaps in Knowledge ...................................................................................................... 34

Summary ...................................................................................................................... 36

Methodology .................................................................................... 37

Overview ...................................................................................................................... 37

Fundamental Appreciation of Operation of an On-street, Off-line, Mid-Block Bus Stop

37

Methodological Approach ............................................................................................ 39

Summary ...................................................................................................................... 43

Quantifying Bus Stop Capacity in terms of Processing Time .... 45

Overview ...................................................................................................................... 45

Problem Conceptualisation .......................................................................................... 45

Dwell Time Model ....................................................................................................... 46

Methodological Approach ............................................................................................ 48

Model Development ..................................................................................................... 50

Comparison between TCQSM model and MBSC model ............................................ 56

Examination of MBSC Model ..................................................................................... 62

Summary ...................................................................................................................... 63

Microscopic Simulation Modelling of an Off-Line, Mid-Block Bus

Stop 65

Overview ...................................................................................................................... 65

Microscopic Simulation Modelling Approach ............................................................. 65

Microscopic Simulation Model Development ............................................................. 66

Model Verification ....................................................................................................... 68

Microscopic Simulation Model Implementation.......................................................... 70

Summary ...................................................................................................................... 73

Theoretical Model of On-Street, Mid-Block, Off-Line Bus Stop

Capacity with Adjacent Lane Traffic ..................................................................... 75

Overview ...................................................................................................................... 75

Influence of Adjacent Lane General Traffic on Operation of an On-Street, Mid-Block,

Off-Line Bus Stop ...................................................................................................................75

Model Development .....................................................................................................77

BCAL Methodology to Estimate Capacity of an On-street, Mid-block, Off-line Bus Stop

81

Comparison between TCQSM and BCAL Models ......................................................82

Summary .......................................................................................................................83

Maximum Working Capacity of an On-Street, Mid-Block, Off-

Line Bus Stop 85

Overview ......................................................................................................................85

Refined Definition of Bus Stop Failure ........................................................................85

Degree of Saturation at a Bus Stop ...............................................................................87

BCAL Model Improvement Considering Maximum Working Capacity .....................88

Specification of Practical Degrees of Saturation at an OS-MID-OFF Bus Stop ..........93

BMWCA Routine to Estimate Maximum Working Capacity for On-street, Off-line,

Mid-Block Bus Stops ..............................................................................................................97

Comparison between TCQSM model and BMWCA Model ........................................98

Summary .....................................................................................................................101

Parametric Study of Bus Stop Maximum Working Capacity with

Adjacent Lane Traffic Model ................................................................................ 103

Overview ....................................................................................................................103

Influence of Dwell Time on Bus Stop Maximum Working Capacity ........................104

Influence of Upstream Average Waiting Time on Bus Stop Maximum Working

Capacity ................................................................................................................................106

Influence of Number of Loading Areas on Stop Capacity .........................................108

Yield-to-Bus Rule .......................................................................................................109

Summary .....................................................................................................................115

Conclusions ................................................................................... 117

Overview ....................................................................................................................117

Summary of the Thesis .............................................................................................. 117

Theoretical Contributions of the Research ................................................................. 121

Practical Contributions of the Thesis ......................................................................... 122

Recommendations and Future Research .................................................................... 123

Concluding Remarks .................................................................................................. 124

References ............................................................................................................... 127

LIST OF FIGURES

Figure 1-1: Bus Transit Facility Classification ............................................................ 1

Figure 1-2: Bus Facility Types ..................................................................................... 2

Figure 1-3: Structure of the thesis ................................................................................ 9

Figure 2-1: On-street bus stop types .......................................................................... 12

Figure 2-2: On-line bus stop ...................................................................................... 14

Figure 2-3: Off-line bus stop ...................................................................................... 15

Figure 2-4: Bus stop failure in bus stop ..................................................................... 24

Figure 3-1: Buses approaching the bus stop .............................................................. 38

Figure 3-2: Buses blocking each other at a bus stop .................................................. 39

Figure 3-3: Bus re-entering from the bus stop ........................................................... 39

Figure 3-4: Schematic diagram of the research methodology ................................... 40

Figure 4-1: Bus channel layout and passenger flow on commonest Brisbane

buses ............................................................................................................. 47

Figure 4-2: Steps followed to quantify the influencing capacity reduction factors

...................................................................................................................... 50

Figure 4-3: Processing time taken by a bus during a signal cycle ............................. 55

Figure 4-4: Comparison of time components of loading area total processing

time per bus between TCQSM model and MBSC Model ........................... 58

Figure 4-5: Influence on traffic blockage on loading area processing time per bus

and stop capacity .......................................................................................... 59

Figure 4-6: Influence of bus-bus interference on loading area processing time

per bus and bus stop capacity ....................................................................... 60

Figure 4-7: Influence on green time ratio on loading area processing time per bus

and bus stop capacity ................................................................................... 61

Figure 4-8: Loading area bus processing time comparison with technological

advancements ............................................................................................... 63

Figure 5-1: Layout of the simulation testbed of type bus stop of this research ......... 67

Figure 5-2: Testbed limit state bus stop capacity vs dwell time according to

simulation model and TCQSM model (Kittelson and Associates,

2013a). ......................................................................................................... 69

Figure 5-3: Limit state bus stop capacity determined using TCQSM and

simulation testbed vs adjacent lane flow rate (Kittelson and Associates,

2013a) ........................................................................................................... 71

Figure 6-1: Flow chart for OS-MID-OFF bus stop capacity estimation using

BCAL model ................................................................................................ 81

Figure 6-2: Comparison between TCQSM and BCAL models of OS-MID-OFF

bus stop capacity vs. adjacent lane traffic flow rate .................................... 82

Figure 7-1: Flow chart for bus stop maximum working capacity estimation using

BMWCA model ........................................................................................... 97

Figure 7-2: Comparison of bus stop maximum working capacity vs adjacent lane

flow rate ....................................................................................................... 99

Figure 8-1: Basic overview of parameters that influence the performance of an

OS-MID-OFF bus stop according to BMWCA model of Chapter 7 ......... 103

Figure 8-2: OS-MID-OFF bus stop maximum working as a function of average

dwell time (two loading areas, 10s upstream average waiting time)

according to BMWCA model .................................................................... 105

Figure 8-3: OS-MID-OFF bus stop maximum working as a function of upstream

average waiting time (two loading areas, 20s average dwell time) ........... 107

Figure 8-4 OS-MID-OFF bus stop maximum working as a function of number

of loading areas (10s upstream average waiting time, 20s average dwell

time) ........................................................................................................... 108

Figure 8-5: Bus stop maximum working capacities vs adjacent lane flow rate

with levels of YTB rule.............................................................................. 112

Figure 8-6: Re-entry delay based on YTB conditions ............................................. 113

Figure 8-7: Processing margin for YTB conditions ................................................. 114

LIST OF TABLES

Table 2-1: Comparative analysis of bus stop locations (Fitzpatrick et al., 1996) ...... 12

Table 2-2 : Individual passenger service times suggested by TCQSM (Kittelson

and Associates, 2013a)................................................................................. 20

Table 2-3: Failure Rates and corresponding ‘Z’ values (Kittelson and Associates,

2013a) .......................................................................................................... 25

Table 2-4: Effectiveness of loading areas for on-line and off-line bus stops

(Kittelson and Associates, 2013a)................................................................ 28

Table 4-1: TCQSM model and MBSC model comparison of bus stop capacity ....... 57

Table 8-1: Maximum working capacity comparison between level of YTB rule.... 112

LIST OF ABBREVIATIONS

AIMSUN Advanced Interactive Microscopic Simulator for Urban and

Non-Urban Networks

API Application Programming Interface

AVL Automatic Vehicle Location

BCAL Bus Capacity with Adjacent Lane Requirements

BMWCA Bus stop Maximum Working Capacity

BRT Bus Rapid Transit

CBD Central Business District

HCM Highway Capacity Manual

LA Loading Area

LHT Left Hand Travel

MBSC Modified Bus Stop Capacity

NZ New Zealand

OBF Off-Board Fare collection

OSB On-Street bus facility

OS-MID-OFF On-Street, mid-block, Off-line

PATH Partners for Advanced Transportation Technology

QOS Quality of Service

QUT Queensland University of Technology

RMSE Root Mean Square Error

SEB South Eastern Busway

SEQ South East Queensland

TCQSM Transit Capacity and Quality of Service Manual

TRB Transportation Research Board

YTB Yield-to-Bus

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis has not been previously submitted to meet

requirements for an award at this or any other higher education institution. To the best

of my knowledge and belief, the thesis contains no material previously published or

written by another person except where due reference is made.

Signature:

Date:

QUT Verified Signature

ACKNOWLEDGEMENTS

I would like first like to express my greatest gratitude to my principle supervisor

Associate Professor Jonathan M. Bunker, for all his motivation, support and patient

guidance I received throughout my PhD journey. His ideas and constructive

suggestions helped met to complete my thesis successfully. I would also like convey

my sincere thanks and appreciation to my associate supervisor Associate Professor

Ashish Bhaskar, who was always available to respond and guide me.

I am very thankful to Queensland University of Technology (QUT) for supporting me

with QUTPRA scholarship research facilities for my PhD. I would also like to thank

my colleagues in Brisbane who made me feel like home with their continuous

presence. Thank you for always making good times better and hard times easier.

I am very grateful for my father, Hisham who is not with me at this moment, but would

have loved to see where I stand today with his guidance and prayers. Thank you to my

strong mother who was always there for me giving her ultimate support throughout

her life. And my deep love goes to my precious brothers Imran, Fatheen and Aiyash

and my sister-in-law, Shafnas for always motivating and supporting me.

Finally, I would like to thank my beloved husband, Insaf for always being there for me

during hard times and being my biggest strength. My deepest love goes to my baby

Rayyan and my niece Maryam for bringing such joy into our lives.

Chapter 1: Introduction 1

Introduction

Overview

This chapter establishes the motivation behind this research and defines the thesis

statement and objectives. This is followed by a description of scope and relevance of

this research. This chapter then outlines the structure of the thesis.

Background

Bus stops are the first point of contact between the passenger and the bus transit

facility. The spacing, location, design and operation of bus stops significantly

influence bus transit facility performance.

Figure 1-1: Bus Transit Facility Classification

Figure 1-1 classifies the most common types of bus transit systems in use. As the

priority to buses increase the efficiency and the operational performance of the bus

facility increases.

Figure 1-2 shows the various examples of the bus facility types. In a mixed traffic

environment, buses share their lanes with other types of vehicles, such as cars, trucks

and bicycles. Buses are subjected to the same forms of traffic control as that of other

traffic, although buses may have some priority in certain locations. Buses are also

faced with delay that is caused by turning vehicles, pedestrian crossings, high traffic

volumes and roadside parking. Mixed traffic conditions are the most common type of

operation practiced around the world (American Public Transit Association, 2012).

Semi-exclusive Exclusive Grade -

separated

Priority for buses

Mixed Traffic

2 Chapter 1: Introduction

(a)- Mixed traffic (Source: https:// /www.abc.net.au)

(b)-Semi-exclusive

(Source:

https://www.infrastructure.sa.gov.au)

(c)- Exclusive (Source: https:// www.rms.nsw.gov.au)

(d)- Grade-separated (Source: https:// /www.abc.net.au)

Figure 1-2: Bus Facility Types

A semi-exclusive facility is partially designated for buses but is allowed to be used by

other vehicles and pedestrians during certain times of the day. The purpose of having

a semi-exclusive facility is to reduce or eliminate certain types of general traffic

interference that can delay the buses. Figure 1-2 shows a bus lane that operates between

7am to 7pm, whereas during other times the lane will be shared with other vehicles.

Exclusive facilities and grade-separated facilities are exclusively for buses. However,

grade-separated facilities give greater priority to buses than exclusive facilities because

they do not pass through signalised intersections that also need not to accommodate

general traffic, pedestrian crossings and speed restrictions.

On-street bus (OSB) facilities have buses operating with general traffic, and therefore

are the type of bus facility with most conflict. The performance of an OSB facility is

highly dependent upon the interaction between buses and other vehicles. When buses


operate in mixed traffic conditions, bus stops are generally located in the kerbside lane,

or adjacent to the kerbside lane in bus bays, which are also known as on-line bus stops

and off-line bus stops respectively. When a bus dwells in an on-line bus stop, the

stopped bus creates a temporary conflict between buses and other traffic. However,

off-line bus stops operate slightly differently than on-line bus stops. In an off-line bus

stop, a dwelling bus might not create a conflict between other traffic; however, a bus

seeking to re-enter into the adjacent traffic lane may be delayed due to gap acceptance

or may cause delay to the adjacent traffic due to forced merging. Furthermore, when a

bus is dwelling in the bus stop, there is a possibility that other buses may queue

upstream of the bus stop waiting to enter into the loading area, which may eventually

block the through traffic reducing the transit facility capacity.

Capacity of a transit facility is important for several reasons. It allows the analyst to

determine the ability of the facility to accommodate the number of buses and

passengers that wish to use the facility. At a more detailed level, it provides for the

estimation of the number of loading areas that are required to serve a particular bus

route or passenger flow along an arterial road. It also gives the ability to estimate how

bus speeds will decline as bus volumes approach capacity (St. Jacques and Levinson,

1997).

Transit facility capacity addresses the movements of both transit vehicles and

passengers in those vehicles. It is generally defined as the maximum number of

passengers who can travel through the facility during a given period of time (Jaiswal,

2010). This can be estimated by the product of the number of buses that can use the

facility during a given period of time and the number of passengers who can be

accommodated in each bus. However, the bus stop with the lowest bus capacity will

normally constrain the capacity of the whole facility. This stop, also known as the

critical bus stop, is usually the stop with longest dwell time, or heavily congested by

the general traffic.

Because of these special features in mixed traffic, bus stops are often considered as the

key bottlenecks of bus facilities and play an important role in facility capacity.

Therefore, to understand and manage an OSB facility, it is essential to understand the

operation of its critical bus stop.


Research Motivation

In the past several decades transit analysts, designers and scholars have paid

considerable attention to evaluating the performance of bus stops and its effects on

facility capacity. A prominent research achievement was the development of a

consistent set of guidelines for evaluating the quality of service and capacity of transit

services, facilities and systems, which was sponsored by Transit Cooperative Research

Program (TCRP) in the United States. The procedures that were developed have been

incorporated into the Transit Capacity and Quality of Service Manual (TCQSM),

which was the subject of TCRP Report A-15. The most updated set of guidelines and

methodologies for estimating bus facility capacity is given in its 3rd edition in

(Kittelson and Associates, 2013a) as TCRP Report 165.

The TCQSM model was developed to estimate the achievable facility capacity with

regard to the operation of its critical bus stop. Capacity of the critical stop is the product

of the capacity of each of its loading areas, the number of effective loading areas, and

a traffic blockage adjustment factor.

The TCQSM model includes an operating margin on dwell time to estimate the

maximum amount of time that a bus can dwell on a loading area without creating a

‘bus stop failure’. A failure is defined by the TCQSM as a situation that arises when a

bus arrives to use a loading area only to find another bus is still occupying (Kittelson

and Associates, 2013a). Failure rate is a combination of dwell time and dwell time

variability; by assuming that dwell times are normally distributed, the operating

margin on dwell time is calculated by assigning a standard normal variable

corresponding to a desired failure rate and multiplying it by mean dwell time and

estimated coefficient of variation of dwell time.

Under the TCQSM methodology, the addition of the operating margin on dwell time

to the mean dwell time achieves the design dwell time, which is then used in

determination of a loading area design capacity that reflects a desired level of

operational reliability. TCQSM recommends design failure rates between 7.5% and

15% for downtown areas and 2.5% for outside downtown (Kittelson and Associates,

2013a). However, it also mentions that design capacity is maximized when the failure


rate is set to 25%. This research study will pay particular attention to the definition of

bus stop failure and the accuracy of capacity estimation when it is used.

Traffic blockage is another factor that impacts the capacity, and it is a phenomenon

that affects the buses operating on an OSB facility. As mentioned before, on an OSB

facility buses entering and leaving the bus stop will share the adjacent general traffic

travel. In such a situation, the traffic uses some of the capacity of the travel lane that

would otherwise be available for buses in the immediate vicinity of the bus stop. The

reduction in capacity due to this traffic blockage is incorporated in the TCQSM (2013)

methodology for capacity estimation by way of a traffic blockage adjustment factor

for traffic in the lane used by the buses at the stop itself. The methodology allocates

this effect to the whole bus stop and only considers bus stops within the influence of

signalised intersections. It recommends not to apply the traffic blockage adjustment

factor for bus stops away from their influence. This research study will also pay

particular attention to whether it is reasonable to consider the potential effects due to

general traffic at mid-block bus stops that are away from the influence of signalised

intersections.

Degree of saturation is an important traffic performance measure of the maximum rate

of flow of traffic. It is used extensively in traffic control and design, and is defined as

the ratio of volume to capacity of a movement or lane (Akcelik, 1981). It is a

particularly important measure because it represents the sufficiency of a lane to

accommodate to the vehicular demand (Rodegerdts et al., 2004). A degree of

saturation of less than 0.90 at a signalised intersection generally indicates that

sufficient capacity is available for general traffic to travel without experiencing

unexpected delays or queues. As the degree of saturation approaches and exceeds 1.0,

traffic flow may become unstable and queues may grow (Hidalgo et al., 2013, Akçelik,

1980, Rodegerdts et al., 2004). It is noteworthy that the TCQSM (2013) methodology

does not incorporate degree of saturation in the estimation of bus stop capacity.

This research was designed to develop suitable procedures in estimating bus stop

capacity by addressing the abovementioned issues.


Thesis Statement

The interaction between buses and general traffic at an on-street, mid-block, off-line

(OS-MID-OFF) bus stop is a complex phenomenon that is not yet fully understood.

This thesis claims that performance of on-street, mid-block, off-line bus stops can be

analysed more effectively and accurately across the full range of bus and adjacent lane

general traffic flow rates, by developing a novel deterministic model for use in an

improved analytical methodology.

Research Questions

The abovementioned thesis statement has been guided by articulation of the following

four research questions:

1. How accurate is capacity estimation of an OS-MID-OFF bus stop using the

existing methodology of the Transit Capacity and Quality of Service Manual

(TCQSM)?

2. What other tools are available to understand and analyse performance of an

OS-MID-OFF bus stop when it is impractical to collect field data?

3. How can we improve upon the existing methodology of TCQSM, under

specified operating conditions?

4. How can we develop an improved methodology, to include control parameters

to reflect adequate levels of service?

Research Objectives

The following research objectives have been established in order to develop, support

and explain the abovementioned thesis statement in the context of the guiding research

questions:

1. Understand the operation of an OS-MID-OFF bus stop. (responds to RQ1 to

develop thesis statement)


2. Review the literature to understand the state of the art about performance

analysis of OS-MID-OFF bus stops and identify the parameters affecting bus

stop capacity. (responds to RQ1, RQ2 to develop thesis statement)

3. Develop a microscopic simulation model testbed for an OS-MID-OFF bus stop

using the methods identified in objective 1. Use the microscopic simulation

model to generate data that is reflective of a range of operational conditions for

varying adjacent lane general traffic flow rates. (responds to RQ2 to support

thesis statement)

4. Develop a deterministically based methodology to accurately estimate the stop

capacity for an OS-MID-OFF bus stop capacity considering operating

conditions and the influence of the adjacent lane traffic flow. (responds to RQ3

to support thesis statement)

5. Address the importance of degree of saturation of an OS-MID-OFF bus stop

and the adjacent general traffic lane and incorporate these parameters to further

improve the deterministic methodology. (responds to RQ4 to support thesis

statement)

6. Demonstrate the use of the developed deterministic methodology by means of

a test bed OS-MID-OFF bus stop that can be used to test the full range of bus

stop and adjacent lane general traffic flow rate scenarios. (responds to RQ3,

RQ4 to support thesis statement)

7. Provide recommendations for future OS-MID-OFF bus stop research, design,

operational analysis and management. (responds to all four research questions

to explain thesis statement)

Scope

While the motivation of this research is to better understand bus stop operation across

a full range of bus and adjacent lane general traffic flow rates, the scope of this research

is limited to on-street, mid-block bus stops, which are a prevalent bus stop type on


busy arterial roads. An exception is Chapter 4, which gives some consideration to

effects at bus stops located adjacent to signalised intersections. The model can be

modified to further consider the influence of signalised intersections, but it is beyond

scope of this research.

Significance

Limited understanding of any particular bus stop operation could lead to an inaccurate

analysis and design of the system and potentially overestimation of bus stop capacity

and therefore bus facility capacity. This research develops a novel, deterministic

methodology to analyse performance of an OS-MID-OFF bus stop which accounts a

full range of operating conditions reflected by influencing variables.

The main novel contribution from this research is a stepwise methodology to

accurately estimate stop capacity of an OS-MID-OFF bus stop, based on development

of valid relationships between these variables. This study helps to expand disciplinary

knowledge by clearly identifying influences of general traffic operation on on-street

bus stops.

Thesis Outline

Figure 1-3 illustrates the structure of the thesis. The thesis is divided into three main

sections.

The thesis statement is developed in Chapters 1 to 3 through introduction, literature

review and identification of gaps in knowledge, and establishment of the research

methodology.

The thesis statement is supported in Chapters 4 to 7, which present the approach

towards developing a novel deterministic model for use in an improved analytical

methodology.

The thesis statement is explained in Chapters 8 and 9, which present a parametric

analysis using the improved analytical methodology, followed by the overall

conclusion to the thesis.


Figure 1-3: Structure of the thesis

Chapter 1: Introduction

Chapter 2: Literature review

Chapter 3: Research methodology

Chapter 4: Quantification of capacity in terms of processing

i

Chapter 5: Simulation model development

Chapter 6: Theoretical model development with adjacent lane

traffic flow

Chapter 7: Bus stop capacity model development with degree of

saturation

Chapter 8: Parametric analysis

Chapter 9: Conclusions and future directions

Obj.-1, 2

Obj. - 2 Obj.-3

Obj.-4

Obj.-5

Obj.-6

Obj.-7

Thesis statement development

Thesis statement explanation

Obj.-1

Obj.-1

Thesis statement support

Chapter 2: A Review of Measures, Modelling Approaches and Evaluation of Bus Stop Capacity 11

A Review of Measures,

Modelling Approaches and Evaluation of

Bus Stop Capacity

Overview

This chapter presents a literature review on the role of bus stops and its importance.

The second part of the review includes bus stop capacity theories, in terms of measures,

modelling approaches and evaluations. The measures cover the state-of-art in theory

for estimating bus stop capacity. The third part of the chapter presents the operational

strategies and practices in a bus stop. Due to the limited literature on improving the

capacity of a bus stop, several similar studies are reviewed to inspire possible methods

for this study. This chapter fulfils research objectives 1 and 2.

Role of a Bus Stop

Bus stops are designated areas where buses stop to serve passengers wishing to alight

and passengers wishing to board the bus. Different types of bus stops exist depending

on the location, demand and operational purposes.

This study is concerned with on-street, mid-block, off-line (OS-MID-OFF) bus stops.

This study is restricted to bus stops that are directionally separated such that buses

cannot overtake across the oncoming side of the roadway. A bus stop consists of one,

or multiple linear loading areas. While we are concerned with this type of stop, it is

useful to describe the range of bus stops.

Types of Bus Stops Based on Location

There are three types of bus stops based on the location: nearside, far side and mid-

block as shown in Figure 2-1.

12 Chapter 2: A Review of Measures, Modelling Approaches and Evaluation of Bus Stop Capacity

Figure 2-1: On-street bus stop types

Near side bus stops are the stops located immediately prior to an intersection, while

far side bus stops are located immediately after an intersection. Mid-block bus stops

are located within the block. Table 2-1 shows a comparison between bus stop types.

While each has its advantages and disadvantages, location of any bus stop is subject

to local and wider network conditions. Mid-block bus stops are prevalent on arterial

roads in major cities and are the focus of this study.

Table 2-1: Comparative analysis of bus stop locations (Fitzpatrick et al., 1996)

Advantages Disadvantages

Near

side stop

Minimizes interferences when traffic is heavy on the far side of the intersection

Allows passengers to access buses closest to the pedestrian crossing

Eliminates the potential of stopping twice

Allows passengers to alight and board when the signal is red

Provides driver with the opportunity to look for oncoming traffic, including other buses with potential passengers

Increases conflicts with left turning vehicles

May result in stopped buses obscuring kerbside traffic control devices crossing pedestrians

May cause sight distance to be obscured for cross vehicles stopped to the left of the bus

May block the through lane during peak period with queuing buses

Increases sight distance problems for crossing pedestrians

Far side

stop

Minimizes conflicts between left turning vehicles and buses

May result in the intersection being blocked during peak periods by stopping buses


Provides additional left turn capacity by making kerb lane available for traffic

Minimizes sight distance problems on approaches to intersection

Encourages pedestrians to cross behind the bus

Creates shorter deceleration distance for buses since the bus can use the intersection to decelerate

Results in bus drivers being able to take advantage of the gaps in traffic flow that are created at signalised intersections

May obscure sight distance for crossing vehicles

May increase sight distance problems for crossing pedestrians

Can cause a bus to stop far side after stopping for a red light, which interferes with both bus operations and all other traffic

May increase number of rear-end accidents since drivers do not expect buses to stop again after a red light

Could result in traffic queued into intersection when a bus is stopped in travel lane

Mid-

block

stop

Minimizes sight distance problems for vehicles and pedestrians

May result in passenger waiting areas experiencing less pedestrian congestion

Encourage patrons to cross street at mid-block (jay walking)

Increases walking distance for patrons who cross at intersections

Types of Bus Stops Based on the Alignment

2.2.2.1 On-line Bus Stops

A linear, on-line bus stop does not have a separate bay for buses to stop and serve

passengers. Here buses stop in the travel lane to load and unload passengers. In a mixed

traffic environment, on-line stops allow buses to proceed again with the travel lane as

soon as the dwell time to serve passenger alighting and boarding are completed.

However, during the dwelling time of the bus, the general traffic may be obstructed

and form an upstream queue from the bus stop. Figure 2-2 shows the layout of an on-

line bus stop.


(a) Layout of an on-line bus stop with two loading areas

Figure 2-2: On-line bus stop

On-line bus stops are simple in design, and easy and inexpensive for the responsible

agency to install. They are also easy to relocate upon requirement. These bus stops

provide easy access for bus drivers and results in minimal delays to buses. However,

on-line bus stops can cause traffic to queue behind a stopped bus and cause congestion

in the vicinity. Furthermore, it may causes non-stopping buses to make unsafe

manoeuvres when changing lanes in order to avoid delay behind a stopped bus.

2.2.2.2 Off-Line Bus Stops

Off-line bus stops are separated from the traffic lanes in order to provide convenience

to alighting and boarding passengers (Fitzpatrick et al., 1996). At an off-line bus stop,

buses stop outside of the traffic lane to serve passengers. This allows the adjacent lane

general traffic to pass without obstruction while a bus is dwelling. They are suitable at

locations with high traffic volume, high speed roadways or sections with a high

number of alighting and boarding passengers and often result in higher vehicle

throughput along the facility. Figure 2-3 shows the layout of an off-line bus stop.

The standard layout of an off-line bus stop is shown in Figure 2-3 (b). An off-line us

stop has three zones: bus entry zone, loading areas zone (where buses dwell to serve

alighting and boarding passengers) and bus exit zone. These entry and exit zones

facilitate a bus to safely enter into the stop from the traffic lane and leave the stop to

merge back into the traffic lane.


(b) Layout of an off-line bus stop with three loading areas Figure 2-3: Off-line bus stop

Off-lines bus stops are advantageous for passengers in terms of safety because they

provide a protected area away from moving vehicles for both stopped bus and

passengers, and allow the passengers to alight and board out of the travel lane. Because

of their reduced interaction with other moving traffic, off-line bus stops minimizes the

delay to through traffic. However, these bus stops still have their disadvantages, being

more expensive to install and relocate. Problems may also occur to bus drivers when

trying to re-enter to the traffic lane, especially during periods of high traffic volumes.

Bus Stop Capacity

A bus stop plays an important role in the total travel time and the capacity of the

facility. This is because scheduling a bus to serve a bus stop inflates its travel time due

to the delay caused by the stop. According to St. Jacques and Levinson (1997), transit

facility capacity is solely dependent on the critical stop/ station along its corridor. With

respect to the bus stop, ideally buses arrive at the bus stop, serve passengers and leave

the bus stop, allowing the next bus to arrive according to its timetable. There are

several components that contribute to the time taken for a bus to be processed through

a stop.


Factors Affecting Bus-Stop Capacity

2.3.1.1 Dwell Time

The time spent by a bus at a stop to serve its passengers including the time taken to

open and close doors, is commonly known as dwell time. It is a significant parameter

which take up a significant portion of the total trip time and is key to system

performance and Quality and Service (QOS). It is also important to understand

variability in dwell time between buses. Dwell time can be divided into sub

components, comprising of delay associated with physical and mechanical properties

of a bus, delay associated with passenger transfers and delay associated with fare

collection methods.

2.3.1.2 Passenger Alighting and Boarding

For a specific bus, the door opening time and closing times are generally fixed and not

subject to the impact of passengers at the bus stop or in the bus; therefore, alighting

and boarding times at bus stops can be the most significant factors causing dwell time

variations.

According to Levinson (1983), dwell time for any bus is directly proportional to the

passenger demand. As one of the earliest studies in quantifying the bus dwell time,

Levinson’s study developed a model consisted of two primary contributing factors;

number of alighting and boarding passengers, and time taken to open and close doors.

According to the study dwell time was determined to be equal to 2.75s per passenger

(alighting and boarding) plus an additional 5s.

Similarly, another study was conducted by creating a link between dwell time

estimation and the fare collection system. The study highlighted that, where dwell time

increases, the service time per passenger decreases as the number of passengers at a

stop increased (Guenthner and Sinha, 1983). They developed a logarithmic model to

determine the dwell time. This equation resulted a maximum result when the numbers

of passengers are equal to 24. Therefore, the authors conclude that the dwell time will

be increased proportionally when there are more than 24 passengers and assigns a

value of 1.2s per passenger.


; Equation 2-1

; Equation 2-2

Where,

= Dwell time of a bus (s)

= Number of alighting and boarding passengers

Following these single variable models, research developed on multi-variable dwell

time models. Alighting and boarding passengers were considered as two independent

variables in the modified approach. Vuchic (2005b) suggested that the dwell time for

a bus where alighting and boarding take place in different doors, is the maximum of

total alighting and boarding time plus an additional constant reflecting lost time at the

station including time taken to open and close doors. A modified equation was

suggested for a bus where alighting and boarding take place in all doors according to

Equation 2-3.

Equation 2-3

Where,

= Dwell time of a bus (s)

and = Number of alighting and boarding passengers

and = Alighting and boarding time per passenger (s)

= Station lost time (s)

The research conducted by Rajbhandari et al. (2003) identified passenger alighting and

boarding as one of the most important parameter that affects the dwell time. The study

highlighted the importance by indicating that the dwell time reduction caused by

passenger demand could save more time than installing bus priority systems.

TCQSM 2003 (Kittelson and Associates, 2003) suggested an identical dwell time

model as above, which is represented below.

Equation 2-4

Where,

= Average dwell time (s)

and = Number of alighting and boarding passengers

and = Alighting and boarding time per passenger (s)


= Time taken to open and close bus doors (s)

This multi-variable model implies that alighting and boarding occur in series at the

busiest door. The dwell time model is defined to be the time taken to serve the

passengers through the busiest door, plus time taken to open and close doors. Any

passenger alighting through the rear door are neglected in the standard bus model since

their activities occur in parallel to the front door, which is implied to be the critical

door.

TCQSM 2013 (Kittelson and Associates, 2013a) includes a deterministic method to

estimate dwell time by relating bus channel layout and number of alighting and

boarding passengers. For buses with front and rear door, alighting and boarding occur

in different ways (front single, front double, all door). Thus passenger flow times are

calculated for each door channel.

Equation 2-5

Where,

= Passenger flow time of the door channel (s)

and = Number of alighting and boarding passengers through the door channel

and = Alighting and boarding per passenger in the door channel (s)

For a bus with number of door channels, the maximum flow time is estimated by;

Equation 2-6

Maximum passenger flow for a bus (s)

The longest passenger flow is then incorporated into find the dwell time along with the

boarding lost time.

2.3.1.3 Fare Collection Method

On-board fare collection systems are the most common in bus systems. The

complexity in the fare collection system can increase the passenger service time as the

number of passengers increase at a bus stop (Kraft, 1975). This is because some media

of fare collection require more time than the others.


Zografos and Levinson (1986) examined the dwell time for no-fare bus systems. They

developed a relationship between service times per boarding passenger with the

number of passengers already on the bus. They found that service time per passenger

was 2s per passenger for a no-fare bus system until passengers exceeded the seating

capacity.

Marshall et al. (1990) analysed dwell times and bus capacities for a specific area in the

US by collecting field data. They found that different fare structures, such as- coins

only, bills only, and combined payment methods, create an average passenger service

time of up to 8s. However, that study occurred before the widespread implementation

of smartcards and magnetic strip fare collection system.

Milkovits (2008) created a relationship between dwell time estimation and fare

collection system. They used data from automatic passenger counting, automatic fare

counting and automatic vehicle location systems installed on buses to estimate a dwell

time model and analysed the impact of fare collection system on dwell time. They

found that smart media fare cards are estimated to have a 1.5s faster transaction time

than magnetic strip tickets in uncrowded situations. The study also highlighted that

with 100% use of a smart media fare, the bus dwell could possibly to decrease by

22.8%.

Most of the studies found in the literature have focused on the impact of the fare

collection media on the time required to serve each boarding passenger. However,

some media such as smart cards may require the passenger to “tap-off” while alighting,

which might have an impact on the total passenger service time. In the absence of field

data TCQSM suggests values for passenger service times considering alighting and

boarding passengers.


Table 2-2 : Individual passenger service times suggested by TCQSM (Kittelson and Associates, 2013a)

Situation Average Passenger Service Time (s/p)

Observed Ranger Suggested

BOARDING

No fare payment 1.75-2.5 1.75

Visual inspection 1.6-2.6 2.0

Single ticket or token 2.9-5.1 3.0

Exact change into fare box 3.1-8.4 4.5

Mechanical ticket validator 3.5-4.0 4.0

Magnetic strip card 3.7-6.5 5.0

Smart card 2.5-3.2 2.75

ALIGHTING

Front door 1.4-3.6 2.5

Rear door 1.2-2.2 1.75

Rear door with smart card check-out 3.4-4.0 3.5

2.3.1.4 Vehicle Type and Size

Vehicle type plays an important role in the stop capacity measure. Standard buses can

operate at relatively high frequencies and therefore short headways. Articulated buses

can have capacities that are approximately 50% greater than those of standard buses,

but tend to operate at slightly lower frequencies and therefore longer headways. Such

high capacity buses are used on Bus Rapid Transit (BRT) systems in Bogota, San

Paulo and Curitiba.

Compared to conventional buses, low floor buses have an influence towards reducing

dwell times. Since its platform is designed to reach the level of the kerb, elderly and

passengers with disabilities find it easier and quicker to alight and board. Levine and

Torng (1994) conducted a research on the impacts of low floor buses on dwell time.

They used three data sources, one being a data set with wheelchair access. The study

concluded that low floor design have the potential to reduce passenger alighting and

boarding by 13-15%. Similarly, King (1998) pointed out that low floor busses offer

reduced dwell times through faster alighting and boarding. The second edition of

TCQSM (Kittelson and Associates, 2003) later suggested to reduce boarding time 0.5s


per passenger if the bus is low floor. Low floor buses are now standard in many

countries.

2.3.1.5 Other Factors

Jaiswal et al. (2010b) introduced boarding lost time as an important component

that affects the bus’s dwell time, which represents the time passengers take to walk to

the bus door from their waiting position on a BRT station platform. The study analysed

the differences between alighting and boarding times at three loading areas at one

station. Results showed that dwell time model which included boarding lost time gave

higher values, therefore increased loading area processing times for buses. This was

later incorporated into the dwell time model of TCQSM 3rd edition (Kittelson and

Associates, 2013a).

Equation 2-7

Where,


= Maximum passenger flow time (s)

= Time taken to open and close bus doors (s)

= Boarding lost time (s)

Some of the previous dwell time models included a variable to capture the effect of

crowding. Even though crowding is vague in its terminology, some researches related

the number of seats as the value above which passengers on-board begin to affect the

alighting and boarding process. Lin and Wilson (1992) developed models with

different linear and non-linear combinations of standees, alighting and boarding

passengers. They investigated the effect on dwell time with a linear crowding variable

and a non-linear crowding variable. According to their study the non-linear

relationship gave a better fit to the observational dwell time. They concluded that an

exponential crowding variable is more suitable because each additional standee or

alighting and boarding passenger above a critical value will take more time than the

previous average. Similarly, Milkovits (2008) incorporated passenger crowding in his

dwell time model. He mentioned that crowding is created by passengers obstructing

the alighting and boarding process and suggested that the number alighting and

boarding must be subtracted to calculate the standees that will affect crowding. The


crowding impact is then measured by the number of standees squared multiplied by

the total passenger activity. The model then compares crowding and non-crowding

scenarios with various fare collection methods to establish an effective operational

practice.

Clearance Time

Clearance time is that required for a bus to exit a loading area after the end of its dwell

time and to be replaced by a waiting bus. Geometric delay is a result of acceleration

and deceleration of the bus and could be within, or partially in addition to, clearance

time.

Start-up time is the time taken by a bus to start up from rest at the loading area and

travel its own length and the next bus to pull in, . Re-entry delay incurred by the

bus while waiting for a gap in the adjacent lane, (Kittelson and Associates, 2003).

Start-up time has a fixed value and corresponds to the mechanical properties of the

bus, while re-entry delay can vary depending on the stop attributes. Clearance time is

equal to the sum of start-up time and re-entry delay.

Meng and Qu (2013) developed a probabilistic methodology to estimate a bus’s

clearance time for an off-line bus stop. They assumed that the time headway of vehicles

on the adjacent lane and the inter-arrival time of passengers boarding on the bus stop

are two random variables that each follow an exponential distribution. The bus driver’s

decision-making process regarding whether or not to enter the adjacent lane was

modelled by Bernoulli trials. According to the study they found out that, the time taken

to enter into the adjacent lane followed a combined geometric-exponential distribution

for off-line bus stops. The study concluded that, service time at off-line bus stops has

a high degree of uncertainty because of the merging behavior of buses into the adjacent

lane.

TCQSM (2013) developed a deterministic methodology to estimate the clearance time.

When a bus stop is located away from an upstream signalised intersection and outside

of the influence of a downstream signalised intersection, traffic is assumed to arrive

randomly at the bus stop. Then buses would wait for a suitable gap to enter into the

traffic lane. In this case, the re-entry delay for the bus is the time taken for the bus to


find an acceptable gap. When a signalised intersection is present, the signal would

release the traffic as platoons. The remaining traffic will queue upstream of the

signalised intersection until the signal turns green. When a bus stop is located upstream

of the signalised intersection, the built-up queue will block the exiting bus. In such a

situation, first the buses will have to wait for the queues to clear, and second, wait for

a suitable gap to merge into the traffic lane. Therefore, re-entry is the addition of the

time taken by the bus waiting for the queues to clear and the time taken by the bus

waiting for an acceptable gap.

However, a mid-block bus stop is by definition away from the influence of a nearby

signalised intersection. Therefore, re-entry delay will be the delay caused by the gap

acceptance. According to TCQSM, gap in traffic delay is given by the following,

2.3.2.1 Gap acceptance:

Equation 2-8

Equation 2-9

Where:

= Re-entry delay (s)

= Capacity of the re-entry movement (veh/h)

= Number of actual loading areas

= Demand flow rate in the kerb lane (veh/h)

= Critical headway of the re-entry movement = 7s

= Follow-up time for the re-entry movement = 3.3s

This method to estimate produces an estimate of maximum average delay that

could occur in waiting to enter the adjacent traffic lane.

Failure Rate

A bus stop failure is defined by the TCQSM as a situation that arises when a bus arrives

to use a loading area, only to find that another bus is still occupying it (Kittelson and


Associates, 2013a). It may be reasoned that this definition strictly reflects the situation

that arises when two buses arrive to use the loading area consecutively at a headway

equal to the inverse of the specified capacity of the loading area, but the first bus dwells

on the loading area for a duration longer than the average dwell time plus a specified

operating margin, causing the second bus to wait until the first bus clears the loading

area.

Figure 2-4: Bus stop failure in bus stop

(Source: TCQSM 3rd edition (2013))

Many studies have determined facility bus capacity using a specific value for failure

rate. However, Gu et al. (2011) defined failure rate (FR) rather differently than any

other, for a bus stop with a single loading area. For uniform bus arrivals, they assumed

that bus service time follows an Erlang-k distribution. The ratio of bus inflow ( ), to

the loading area service time ( ) is set to be equal to ; where is the

coefficient of variation of the service time.

Levinson and St. Jacques (1998) modified the Highway Capacity Manual (HCM,

1985) formula for transit capacity estimation by using field studies and simulations.

The TCQSM model includes failure rate as a combination of dwell time and dwell

time variability by assuming that dwell times are normally distributed. This to account

for the maximum time that a bus can dwell on a loading area without creating a

‘failure’. The model is deterministic, although it assumes dwell time to be varied

normally. The operating margin is calculated by assigning a standard normal variable


corresponding to a desired failure rate and multiplying it by an estimated coefficient

of variation and mean dwell time.

Under the reasoning made above, this means that operating margin is the maximum

amount of time by which mean dwell time can be exceeded under a predefined bus

stop failure rate. This is calculated by selecting the standard normal variable for the

predefined design failure rate and applying it along with coefficient of variation of

dwell time ( and the average dwell time to estimate the operating margin. TCQSM

relates the design value applied to develop a design bus stop capacity that reflects a

desired level of operational reliability. Therefore it recommends design failure rates

between 7.5% and 15% for downtown areas and 2.5% for outside downtown areas

(Kittelson and Associates, 2013a).

Table 2-3: Failure Rates and corresponding ‘Z’ values (Kittelson and Associates, 2013a)

Failure Rate Z

1.0% 2.330

2.5% 1.960

5.0% 1.645

7.5% 1.440

10.0% 1.280

15.0% 1.040

20.0% 0.840

25.0% 0.675

50.0% 0.000

The failure rate is then used to estimate an operating margin. Mathematically,

Equation 2-10

Where,

= Operating margin on the dwell time (s)

= Coefficient of variation of dwell time assuming a dwell time is normally distributed


= Standard normal variate corresponding to the design failure rate

Capacity Theories

Line capacity of a transit system is defined as the maximum passenger volume that a

bus facility can transport during a given period under specified operating conditions.

Therefore it depends on number vehicles, operation of vehicles, passenger and traffic

volumes and operating policies of the transit agency (Levinson and St. Jacques, 1998).

Two types of capacity measures are considered to measure the capacity. They are

vehicle passenger carrying capacity and facility bus capacity. Vehicle passenger

capacity depends on number of seats and the floor area per standee (Vuchic, 2005b).

Facility bus capacity is the number of buses that the facility can accommodate. Bus

capacity can be estimated at three levels; loading area capacity, bus stop capacity, and

facility capacity.

TCQSM 2013 (Kittelson and Associates, 2013a) states that bus capacity is merely

dependent on the bus facility; the greater the exclusivity the higher the capacity. When

the facility offers different service types, the capacity is restricted by the capacity of

the critical bus stop; that is the stop with the lowest capacity (St. Jacques and Levinson,

1997)

Fernández et al. (2007) introduced the concept of divided bus stops. The objective of

the research was to increase capacity and reduce interference between buses by

dividing a bus stop into two subs stops. They experienced a capacity reduction in

upstream bus stop and concluded that weaving distance and weaving manoeuvres have

an impact on the stop capacity.

Gu et al. (2011) took account of the bus arrival pattern via variations in bus headway

to estimate capacity for isolated kerbside bus stops. According to the authors this

model can be used to predict the amount of variation in bus headway and bus service

time that can diminish stop capacity.

Hidalgo et al. (2013) observed from the theoretical models, that the theoretical capacity

is not realistic in practice, because of the stochasticity in vehicle arrivals and in dwell


time. Therefore, they proposed a practical stop capacity for a station with multiple sub

stops considering service policies, infrastructure and stochastic variations.

Equation 2-11

Where,

= Number of sub stops,

= Boarding/alighting time(s),

= Minimum interval between two successive buses,

= Percentage of buses that do not stop at the sub stop

= Accepted degree of saturation of the sub stop, which depends on the operation policy, infrastructure and the queuing capacity

This is the first article that has been found that specifically address the degree of

saturation of a sub bus stop and hence a stop. They used a saturation level of 0.6

considering three sub stops with a queueing capacity of two buses at each stop. The

study found that the practical capacity of the stop can be increased by increasing the

number of sub-stops, platforms and queuing capacity at stations; improving the

operational reliability and enhanced control strategies to allow higher saturation levels.

However this method was developed considering a bus stop with multiple sub stops,

with each sub stop having a single loading area.

TCQSM

TCQSM (2013) is an improvement of the empirical model introduced in the Highway

Capacity Manual (HCM). The most recent version of the model is present in TCQSM

2013 (Kittelson and Associates, 2013a) and provides a systematic methodology to

estimate bus facility capacity.

According to TCQSM, bus facility design capacity (bus/h) differs from the theoretical

bus facility capacity (bus/h). Design capacity represents an achievable flow rate under

restricted, safe working conditions resulting in maximum achievable frequency with

minimum headway. A failure rate was first introduced in 1999 through an operating

margin to account for the non-reliability in the dwell time. When the failure rate is

excluded, true theoretical capacity can be estimated; which represents a condition

when loading area failures occur continuously.


The design loading area capacity using the TCQSM 2013 (Kittelson and Associates,

2013a) method may be stated as,

Equation 2-12

When the bus stop is located adjacent to a signalised intersection, only a portion of the

time permits bus movement. This is during the green phase of the cycle and given by

the ratio so must be taken into consideration. Therefore the loading area bus

capacity is,

Equation 2-13

Where,

= Loading area bus capacity (bus/h)

= Green time ratio (effective green time to cycle time of any signalised intersection adjacent to the bus station either upstream or downstream)

= Clearance time (s)


= Operating margin on dwell time (s)

When there is more than one loading area present, the effective loading areas would

always be less than the total number of loading areas. This is because of temporary

occupation at the rear most loading areas creates the potential to prevent buses from

accessing the front most loading areas, which manifests as interference between buses.

The effectiveness of loading areas depend on whether the stop has on-line or off-line

loading areas.

Table 2-4: Effectiveness of loading areas for on-line and off-line bus stops (Kittelson and Associates, 2013a)

Loading Areas

On-line Off-line

Efficiency %

No of cumulative loading areas

Efficiency %

No of cumulative loading areas

1 100 1.00 100 1.00 2 75 1.75 85 1.85 3 70 2.45 75 2.60 4 20 2.65 65 3.25


Bus stop capacity is therefore the product of a single loading area bus capacity, traffic

blockage adjustment factor and the effective number loading areas.

According to TCQSM bus stop capacity is expressed as:

Equation 2-14

Where,

= Bus stop capacity (bus/h)

= Number of effective loading areas

= Traffic blockage adjustment factor

Microscopic Simulation in Capacity Estimation

Microscopic simulation can be used for situations where there is a need to represent

real world situations and reproduce its behaviour. In contrast to deterministic models,

microscopic simulation modelling provides a visual representation of each scenario.

Another key advantage of simulation is that it permits the user to test operation across

a complete range of bus volumes and adjacent lane traffic volumes on the testbed. In

most circumstances it would be infeasible to collect data across such a complete range

at a field bus stop.

Even though there are many traffic microsimulation packages available, only a few

can be used to model bus transit vehicles. Aimsun micro is one of them. It consists of

a collection of dynamic modelling tools. Aimsun also has sub-models such as car-

following, lane changing, gap-acceptance and overtaking models to effectively

represent traffic conditions. The simulation software has been applied extensively in

commercial projects across a wide range of environments, where its capability of

microsimulation is tested and verified (TSS, 2016, Widanapathiranage et al., 2015).

Car-Following and Lane Changing Models in Aimsun

Aimsun uses its car-following and lane changing models to realistically match the flow

of individual vehicles in the network. Both of these models evolve from the Gipps

model (Gipps, 1981, Gipps, 1986). Car-following behaviour describes how a given

pair of vehicles interact with each other. In Aimsun, the car-following model ensures


that a safe following distance is maintained, and the driver’s behavior is adapted to

always maintain it. The model assumes that the following vehicle chooses its speed

such that it can maintain a safe distance behind a lead vehicle by accelerating and

decelerating whenever needed.

Decision making for lane changing in Aimsun occurs in terms of possibility,

desirability and necessity. These are governed by the turning possibility, the distance

to the next turning position and local traffic conditions. For a roadway with a single

lane, this model is used when buses need to re-enter the traffic. For instance, when a

bus driver tries to re-enter into the adjacent traffic lane from the bus stop, Aimsun

recognizes it as a necessity to change lanes because the distance between the current

position and the next turn is very low. Lane changing during a necessity is different

from lane changing when it is possible or desirable, because vehicles are being forced

to reach their desired lane when there is a necessity. During this immediate action of

the lane changing bus, drivers of vehicles in the adjacent lane would modify their

behaviour in order to allow a gap large enough for the bus to merge into and make the

lane changing possible (Barceló and Casas, 2005). This is reflective of bus operation

with yield-to-bus (YTB) rule, where buses would re-enter the traffic with no delays

while vehicles on the traffic lane will slow down or even come to a complete stop

while giving way to the re-entering bus.

Public Transport Model in Aimsun

The required inputs in Aimsun for buses are, route of each line, stop location, departure

frequency (fixed or stochastic) and dwell time (fixed or stochastic) for each stop.

Capacity Estimation

Many studies in the literature have used microscopic simulation modelling to model

real world cases in order to estimate capacity. Siddique and Khan (2006) used

NETSIM, a stochastic microscopic simulation to evaluate facility capacity of the

system 20 years later. Their objective was to investigate whether the facility can handle

high volumes of buses with the increasing demand. For the chosen corridors, scenarios

incorporating horizon year 2021 traffic were analysed for congested conditions and

the results were compared with the condition representing year 2001. The results were


used to provide recommendations for planners and policy makers to improve

operations of the public transit facility.

Some studies have related stochasticity and randomness in facility bus capacity

estimation. Ortiz and Bocarejo (2014) estimated capacity of the Transmillenio Bogota

using a VISSIM microscopic simulation model. They quantified the difference in

capacities when randomness of bus system operations are included. Siddique and Khan

(2006) used NETSIM, a stochastic microscopic simulation to evaluate facility capacity

of the system 20 years later. With three scenarios presented, they compared the

estimated capacity with TCQSM deterministic model to show the importance of

incorporating stochasticity into the calculation.

Improving Bus Stop Capacity

Various design modifications in bus operations and bus facility infrastructure can be

made to increase the capacity of a bus facility. The design modifications are intended

to meet the passenger demand while carrying more passengers and utilise the bus stop

according to the demand. Many studies have examined operational measures to

increase bus facility capacity. Fernández (2010) described a way to model stop

operations by means of microscopic simulation. They evaluated the increased

performance of a divided bus stop than a regular, multi-berth bus stop. Gardner et al.

(1991) and Germani and Szasz (1980) found that dispatching buses in an ordered

manner could increase bus stop capacity. Gibson et al. (1989) and St. Jacques and

Levinson (1997) proposed reconfiguration of a stop’s geometry.

Vehicle modifications

If a corridor is liable to reach its capacity, one solution has been identified as increasing

bus size. It has been claimed that replacing a 12m bus with an 18m bus can result in a

significant increase in passenger carrying capacity. The bus rapid transit (BRT) system

of Curitiba is equipped with 24m bi-articulated buses each having a capacity of 270

passengers. This significantly expanded the carrying capacity of the system (Cervero,

2013). However, it is important to note that longer buses require longer loading areas

on bus stop and may cause interference between buses across loading areas. Where

platform length is limited, a trade-off between the number of effective loading areas


available and therefore bus stop capacity, and the vehicle passenger carrying capacity,

requires consideration.

Platform modification

Raised kerb and level platforms can increase capacity by facilitating faster passenger

service times, both alighting and boarding, are particularly beneficial to the elderly and

disabled persons. This is now commonplace.

Off-board fare collection

Off-board fare collection is most commonly found in Asia, Latin America, and France.

Keeping the passenger payment process away from the bus has been reported to reduce

passenger service times, dwell time and can therefore lead to an improvement in bus

stop capacity.

Skip-stop operation

The concept of spreading a bus stop into several sub stops, known as skip-stop

operation provides the ability to operate bus stations close to capacity (Kittelson and

Associates, 2013a, Wu et al., 2015) . With skip-stop operation bus routes are generally

grouped together by geographical area in order to provide a common stop group for

destinations that are served by multiple routes. For example, if the bus routes are

grouped in to two groups, namely A and B, the first bus stop along the corridor would

serve buses of group A. B grouped buses would skip the first stop and stop at the

second bus stop. Accordingly A and B follow an alternate stopping pattern along the

corridor. Advantages of operating a skip-stop pattern include; routes with high

demand can arrive directly to a particular station, and reduced passenger travel times

(Wu et al., 2015). A carefully scheduled operation is necessary in order to achieve the

objective of skip-stopping or else passengers may need to walk extra distances to

transfer between buses, which would result in longer travel times.

Platooning/ Convoy Operation

This operational protocol has been used in some cases where insufficient space exists

for passing lanes or other operational treatment has not improved bus capacity. Convoy


or platooning involves two or more buses running along the bus facility in close

formation operating similar to an extended set of rail cars. The bus consist

synchronises motion and moves along dedicated roads as a single unit (Gordon and

Lidberg, 2015). Experiments conducted by PATH (Partners for Advanced

Transportation Technology), California, reported that twice the road capacity can be

achieved by operating convoys than operating buses individually (Bergenhem et al.,

2012b).

Yield-to-Bus Rule

Not all of the above mentioned operational modifications are suitable for OSB

operations. Yielding to buses at the exit area of the bus stop at off-line bus stops could

possibly increase throughput across the bus stop. Meng and Qu (2013) conducted a

study to reduce re-entering delays at an off-line bus stop. They found that introducing

yield-to-bus (YTB) laws at the exit of the bus stop has the potential of reducing the

clearance time form 30% to 7%, which will ultimately increase the facility bus

capacity.

Yield to bus laws have been implemented in many parts of North America. King

(2003) investigated the use and experience of YTB laws implemented in British

Columbia, California, Florida, Oregon and Washington. Zhou et al. (2011) assessed

the impacts of YTB laws in Florida and highlighted that YTB behaviour depends on

location of the bus stop, hourly traffic volume, number of lanes, speed environment

and public attitude towards buses at a specific location. Hyde and Smith (2017)

quantified the economic and other benefits of YTB rules for bus services in New

Zealand. They established a relationship between number of cars and the delay to buses

for locations where YTB rule applies, by conducting video data analysis. They found

that average delay to buses exiting the bus stop equates to 5.69s and concluded that

changes to other road users due to YTB are marginal or negligible. The results also

showed that buses experienced re-entering delays at 25.45% of the stops. Among the

280 movements recorded, despite the fact that 14.3% benefitted from the YTB rules,

the conclusion is case dependent.

Hisham et al. (2018b) developed a microscopic simulation model using Aimsun to

study operations across a full range of bus and adjacent lane general traffic volumes.


They found out that bus stops with YTB rules resulted in achieving higher bus

capacities for mid-block and near side, off-line bus stops.

Gaps in Knowledge

Based on the literature review, important findings were made in estimating bus stop

capacity. Even though the TCQSM model can be applied across a range of bus

facilities, it does not sufficiently address adjacent lane general traffic movements and

their impacts on bus stop capacity. Especially in mid-block bus stops TCSQM

recommends not to apply the effect of adjacent lane traffic. The most important finding

made from the literature review is that the existing methodologies are not suitable to

effectively analyse on-street, mid-block, off-line bus stops.

The current theoretical methodology presented in the Transit Capacity and Quality of

Service Manual (TCQSM) (2013) defines bus stop failure as a situation that arises

when a bus arrives to use a loading area only to find another bus is still occupying it.

However, as reasoned earlier in this chapter, failure may be defined more accurately

as a situation that arises when two buses arrive to use the loading area consecutively

at a headway equal to the inverse of the specified capacity of the loading area, but the

first bus dwells on the loading area for a duration longer than the average dwell time

plus a specified operating margin, requiring the second bus to wait until the first bus

has cleared the loading area.

Importantly, from this more accurate definition it may be deduced that the current

TCQSM model presumes that any further failure is fully attenuated once the second

bus reaches the loading area. This presumption, however, is incorrect unless a third

bus arrives at a headway equal to or less than the inverse of the specified capacity of

the loading area minus the excessive dwell time of the first bus, and the second bus

had departed the loading area upon the third bus’s arrival. Consequently, the TCQSM

model does not allow for accumulation of delay due to successive buses arriving

unevenly.

TCQSM includes a table of failure rate as a percentage, representing the probability

that bus stop capacity is exceeded, and relates this to the desired level of service. This

was developed through a simulation under particular conditions. It also provides the


theoretical means of estimating operating margin. Mathematically, bus stop capacity

is greatest when failure rate is set to 50%, however this would represent a case with a

constant upstream queue. Therefore, by means of simulation it is suggested to use 25%

as the optimal failure rate in order to achieve a maximum design capacity. Following

the logic of the above clarification about failure, the application of operating margin

and associated failure rate can be feasible when bus arrivals are relatively evenly

spaced. However, as bus and adjacent lane general traffic flow rates increase, buses

are more susceptible to unevenly spaced arrivals, which would increase interference

between buses and would lead to increased failures (Hisham et al., 2019a). However,

as discussed above the TCQSM definition of failure rate is problematic, and the means

of prescribing failure rate has not been sufficiently studied.

Therefore, some researchers have developed novel approaches instead of the failure

rate approach in capacity estimation. Fernandez and Planzer (2002) identified that bus

stop degree of saturation is an important measure in bus stop capacity estimation. The

degree of saturation of the bus stop indicates how busy the bus stop is. This information

can be used in designing bus stops to estimate a suitable combination of bus flow and

passenger demand. However, the estimated degree of saturation does not relate in the

methodology of capacity estimation. Furthermore, TCQSM does not consider the

degree of saturation in its calculation, which means the degree of saturation of the bus

is taken to be 1.0. However, in reality bus stops operate with lesser degree of saturation

in order to maintain the operational reliability (Fernández, 2010).

From the literature review, it is evident that there is need to develop a generalised

methodology to estimate bus stop capacity where it is possible to control the DOS of

the bus stop by ensuring adequate service levels in the bus stop area. In this regard, an

OSB facility with an adjacent lane has not been found to have been considered in the

literature, so has been identified as an important gap in this research.

Therefore, based on this literature review, specific findings in relation to the TCQSM

model include:

1. Failure rate in the TCQSM model is assumed to have an impact on the dwell

time alone rather than the total time that the bus takes to be processed at the

loading area.


2. TCQSM (2013) provides the methodology for capacity without reference to

reliability and operational issues. The methodology consists of ‘failure rate’,

which reflects a design capacity that reflects a desired level of operational

reliability. However, the definition for failure rate is problematic because of its

insensitivity towards other factors such as cascading interference between

buses due to unevenly spaced headways or accumulated delays.

3. The TCQSM model does not account for the influence of the adjacent lane

traffic flow rate on an OS-MID-OFF bus stop. This means that the theory does

not address the relationship between bus stop capacity and the traffic volume

of the adjacent lane at this type of stop, especially near to, or at saturated

conditions.

Summary

This chapter presented the literature review that is conducted for this thesis in response

to the research questions 1 and 2 to develop the thesis statement. The review first

presented the background for this research to properly understand and evaluate the

operation of an-street, off-line, mid-block (OS-MID-OFF) bus stop. The fundamental

components of bus stop capacity estimation methodologies followed by bus stop

capacity theories are reviewed. On the basis of capacity estimation, the review

highlighted the limitations of existing methodologies and identified gaps in knowledge

concerning the operation of OS-MID-OFF bus stops.

Chapter 3: Methodology 37

Methodology

Overview

Chapter 2 identified the gaps in knowledge that have limited the suitability of

established methodologies for capacity estimation of on-street, mid-block, off-line

(OS-MID-OFF) bus stops. This first aim of this chapter is to identify the parameters

that influence the performance of an OS-MID-OFF bus stop in order to achieve the

research objectives that develop, support and explain the thesis statement. In order to

identify the influencing parameters, Section 3.2 will describe a fundamental

appreciation of the operation of an OS-MID-OFF bus stop with an adjacent lane across

a full range of bus and adjacent lane general traffic flow rates. The second aim of this

chapter is to establish the methodological approach towards the research, which is

presented in Section 3.3.This fulfils research objective 1.

Fundamental Appreciation of Operation of an

On-street, Off-line, Mid-Block Bus Stop

Arterial roads are a type of on-street bus (OSB) facility where buses and other forms

of traffic share the same travel lanes. The performance of an OSB facility is highly

dependent upon the interaction between buses and other vehicles. When the adjacent

lane at a bus stop carries a high volume of general traffic, the interaction between buses

and traffic is expected to affect the capacity and QOS of the bus facility (Hisham et

al., 2019a). It is essential to understand the operation of any potentially critical bus

stop in order to understand and manage such a facility.

For an OS-MID-OFF bus stop, operation can be complex because there is continuous

interaction between buses and general traffic. A fundamental way to consider the

functioning of this type of bus stop is that it is a server with a number of channels each

represented by a loading area. If a bus is occupying a particular loading area within the

stop, no other bus can enter into that loading area. Loading area capacity may be

38 Chapter 3: Methodology

defined as the number of buses that can be processed by the loading area within a given

period of time.

Figure 3-1: Buses approaching the bus stop

Figure 3-1 shows a bus approaching a bus stop with three loading areas. First the bus

will decelerate from its running speed and enter into front-most available loading area

that is accessible from the adjacent lane. As buses and other vehicles share the same

lane, while buses are approaching some of the lane capacity will be used by the general

traffic. During this process, additional processing time may often be needed due to

conflicts and disturbances due to general traffic. These conditions are highly dependent

upon the degree of saturation of the adjacent traffic lane.

Once the bus stops at a loading area, doors will be opened to allow passenger exchange.

The door opening and closing times mainly depend on the mechanical properties of

the bus. This usually contributes between 2s to 5s to the dwell time (Kittelson and

Associates, 2013a) and is independent on the number of passengers alighting and

boarding. Other factors that influence the dwell time were discussed in Chapter 2.

Underutilisation of certain loading areas can occur as a result of the dwelling of buses

in certain others. For example, as shown in Figure 3-2, the third loading area can be

occupied while the second loading area is available. The bus dwelling in the third

loading area is blocking the next arriving bus from accessing the second loading area.

This underutilisation reduces the effectiveness of loading areas and hence the

effectiveness of the entire bus stop.

Loading Area Travel direction

LA 1

LA 2

LA 3


Figure 3-2: Buses blocking each other at a bus stop

Furthermore, a passenger who travels in the fourth bus will experience added delay

depending on the processing time of the third bus. Should more buses arrive and queue

upstream of the bus stop, this delay will accumulate. Permitting a large upstream queue

means permitting a high rate of failure, which will result in a higher bus stop capacity

according the TCQSM theory. However, this needs to be studied in detail because

upstream queuing may obstruct the general traffic in the adjacent lane and cause

operational concerns.

After serving passengers in the loading area, the bus driver will seek a gap in the

adjacent travel lane, and re-enter upon finding a suitable gap. Figure 3-3 shows a re-

entering bus.

Figure 3-3: Bus re-entering from the bus stop

Methodological Approach

Based on this fundamental appreciation of the operation of an on-street, mid-block,

off-line (OS-MID-OFF) bus stop, Figure 3-4 outlines the methodological approach of


1

2

3 4

LA 1

LA 2

LA 3


Platform

LA 1

LA 2

LA 3


this research in order to meet the research objectives that develop, support and explain

the thesis statement of Section 1.4.

Develop a deterministic capacity analysis model for on-street, mid-block, off-line bus stops

1 Reflect stochasticity of bus operations 2 Reflect desired level of operational reliability 3 Provide insights on bus stop operations

Ch.2-Literature review 1 Explore actual bus operations 2 Understand the fundamentals of bus stop

capacity 3 Analyse existing capacity methodologies 4 Explore tools available to analyse

performance of a stop

PHASE 1

Ch.4- Quantify bus stop capacity in terms of processing time

1 Examine parameters that influence bus stop capacity

2 Revisit the TCQSM model 3 Investigate traffic blockage, bus-bus

interference and effect of adjacent signalised intersection

Ch.5-Model using Microscopic Simulation

1 Develop a microscopic simulation model

2 Cross-validate the model with the deterministic model

3 Investigate outcomes with adjacent lane traffic flow

PHASE 2

Ch.6-Incorporate adjacent lane traffic requirements

PHASE 3

Ch.8-Undertake parametric analysis 1 Implement model and case study 2 Analyse sensitivity of parameters

Ch.9-Conclusions Provide conclusions and future directions

Ch.7-Incorporate degree of saturation

Figure 3-4: Schematic diagram of the research methodology


Phase 1-Develop thesis statement

Phase 1 comprises of background study and literature review that is conducted for this

thesis in response to research questions 1 and 2 to develop the thesis statement. The

aim of this phase is to examine the existing performance and identify parameters that

influence operations of an OS-MID-OFF bus stop. Methodologies to analyse

performance of a bus stop are reviewed and presented in Chapter 2. Based on the

review, research gaps were identified. This phase identified that research objectives 1

and 2 of this thesis are met. The methodology for this research is subsequently

established upon the research gaps presented in Chapter 2.

Phase 2-Support thesis statement

The aim of this phase is to support the thesis statement by improving upon the existing

deterministic model for capacity estimation of OS-MID-OFF bus stops and by

developing a microscopic simulation model to cross-validate the deterministic model

and further investigate outcomes with adjacent lane traffic.

The standard procedure to estimate capacity that is reviewed in Chapter 2 identifies

primary and secondary influences on bus stop capacity. However, unlike primary

influences, secondary influences such as blockage of the lane used by buses to travel

by general traffic and interference between buses have not been quantified as time

components. Although the model provides deterministic outcomes for bus stop

capacity, the model does not provide an understanding of how each of the secondary

influences contribute towards the total loading area processing time as time

components. Therefore, to support the claim made in the thesis statement, and to

analyse the secondary influences more accurately, these influences will be quantified

in the first part of this phase and will fulfil research objective 2. A detailed analysis of

the quantification of these influences is given in Chapter 4 based upon the publication

(Hisham et al., 2018a). This will provide more detailed information to analyse bus stop

operations across traditional and non-traditional practices. This responds to research

questions 1 and 3.

The next part of phase 2 is to explore tools that are available to analyse bus stop

operations. This will fulfil research objective 3. It is not feasible to develop empirical


models from real data to estimate bus stop capacity for conditions where it is difficult

collect real data. In particular, it is very difficult to observe situations with high

adjacent lane traffic flow rates that are close to saturation. In contrast to empirical

modelling of real data alone, microscopic simulation modelling can effectively

represent real world situations and reproduce its behaviour under a controlled

environment, hence it has been extensively used in transport research (Fernández,

2010, Kittelson and Associates, 2013a, Hisham et al., 2018b). Microscopic simulation

provides opportunities for controlled experiments where detailed analysis of various

operating conditions on bus facilities can be performed, and also allows the user to

estimate bus stop capacity across a broad range of test conditions and testing scenarios.

In this research, Aimsun micro will be used as a simulation tool. This platform has

been used extensively for commercial and research, which has been validated in the

literature (Siddique and Khan, 2006, Widanapathiranage et al., 2015). Detailed

discussion of the microscopic simulation modelling is given in Chapter 5. The

developed Aimsun model is cross-validated with the deterministic model for on-street

operational characteristics to test conditions across the full range of bus and adjacent

lane general traffic flow rates. This respond to research questions 2 and 3.

In Chapter 6, we contend that at an on-street, mid-block, off-line bus stop, buses will

obstruct adjacent lane general traffic flow during a certain component of the loading

area processing time period, so additional time may be required to accommodate

adjacent lane traffic under saturated conditions. To fulfil research objective 4, the

influence of the adjacent lane traffic will be incorporated into the deterministic model.

A detailed explanation of the model development will be given in Chapter 6 based

upon the publications of Hisham et al. (2018b) and Hisham et al. (2019a). This

responds to the research questions 1, 2 and 3.

TCQSM does not consider the degree of saturation of either bus stop or adjacent lane

traffic in its methodology. It may be inferred that degree of saturation of the bus stop

is taken to be 1.0 in the case where there is no operating margin applied. However, in

reality bus stops operate with lesser degree of saturation in order to maintain the

operational reliability (Fernández, 2010). Based upon the in-press publication of

Hisham et al. (2019b) the deterministic model will be further improved in Chapter 7

by incorporating the degree of saturation of the OS-MID-OFF bus stop and the degree


of saturation of the adjacent lane by exploring operating margin. Chapter 7 will fulfil

the research objective 5. This responds to research questions 1, 2 and 4.

Overall, the principal contribution of this phase is the development of novel

deterministic model to effectively analyse the performance of OS-MID-OFF bus stops

across the full range of bus and adjacent lane flow rates. The new deterministic model

will reflect stochasticity of bus operations and also reflect the capacity for a desired

level of operational reliability. Furthermore, the new model will be used to provide

insights on on-street bus operations.

Phase 3-Explain thesis statement

The final part of this phase is to demonstrate the applicability of the developed model

to OS-MID-OFF bus stops using a case study and present in Chapter 8. The model is

used to test various non-traditional and advanced operating practices with the aid of

sensitivity analysis as a response to research objective 6. Contributions of this section

will be used to provide conclusion and future recommendations for this research and

will respond to the research objective 7.

Summary

This chapter successfully identified that time required to accommodate general traffic,

degree of saturation of the adjacent traffic lane and upstream waiting time for buses

are some of the parameters that influence the performance of an on-street, mid-block,

off-line (OS-MID-OFF) bus stop. A research methodology was developed to further

investigate these influencing parameters that develop, support and explain the thesis

statement.

This clearly will enhance knowledge by contributing a novel stepwise methodology to

accurately estimate stop capacity of an OS-MID-OFF bus stop, based on development

of valid relationships between these influencing parameters.

After establishing the research methodology, the next step is to study the secondary

capacity influencing factors in the TCQSM methodology that appear as capacity

reduction factors. Detailed analysis of this quantification is given in Chapter 4.

Chapter 4: Quantifying Bus Stop Capacity in terms of Processing Time 45

Quantifying Bus Stop Capacity

in terms of Processing Time

Overview

This chapter is based upon the paper “Development of a Modified Bus Stop Capacity

Model (Hisham et al., 2018a)” presented at the Transportation Research Board 97th

Annual Meeting. The chapter considers how the TCQSM bus stop capacity estimation

model (Kittelson and Associates, 2013a) uses capacity reduction factors and then

examines how a proposed alternative, deterministic Modified Bus Stop Capacity

(MBSC) model is able to represent the capacity reduction phenomena in terms of

components of loading area total processing time. While this overall thesis is

concerned with on-street, mid-block, off-line (OS-MID-OFF) bus stops away from

the influence of signalised intersections, this chapter takes a general approach in

developing an alternative deterministic model to the TCQSM for bus stop capacity

estimation.

The latter part of the chapter compares the TCQSM with the MBSC model, followed

by sensitivity analysis on the influencing parameters. Finally, a case example shows

how the new MBSC model can be applied and its usefulness in assessing advanced

operational approaches. This chapter fulfils research objective 2.

Problem Conceptualisation

Previous researchers have identified many factors that affect bus stop capacity. Dwell

time, dwell time variation and fare type are highly associated with the number of

boarding and alighting passengers, while clearance time is highly correlated with

traffic operation and buses’ mechanical performance. They identified these above

mentioned factors as primary influences to bus stop capacity. Secondary influences

include signalised intersection red time periods, blockage of the lane used by buses to

travel by general traffic, along with interference between buses at the stop. The current

methodology presented in the TCQSM to estimate bus stop capacity has incorporated

46 Chapter 4: Quantifying Bus Stop Capacity in terms of Processing Time

these primary and secondary influences (Kittelson and Associates, 2013a). It is

apparent from Equation 2-14 that, unlike the primary influences which are accounted

for as components of loading area processing time per bus, the model allocates

secondary influences using capacity reduction factors for the whole stop. Although

this model serves the purpose of capacity estimation, it does not provide an

understanding of how adjacent signalised intersection red time periods, traffic

blockage and interference between buses contribute as individual components of total

processing time on a loading area.

Traffic blockage is incorporated into the TCQSM model through a traffic blockage

adjustment factor and interference between buses is incorporate through bus-bus

interference factor . Green time ratio corresponds to the ratio that buses

can access the bus stop during the green time of the adjacent traffic signal. These

factors were considered in the capacity estimation considering the whole bus stop.

The TCQSM model does not provide more detailed information to estimate the

capacity under non-traditional operating practices. In order to compare and contrast

the capacity with different operational practices, a model that quantifies and

incorporates all of the relevant loading area bus processing time components due to all

influences will be necessary.

In this chapter the above-mentioned factors will be quantified as time components that

contribute towards the loading area processing time. Although the overall thesis

focuses on an on-street, mid-block bus stop, to broaden the applicability of the model

we consider a general bus stop with an adjacent traffic signal in this section.

Dwell Time Model

It was pointed out in the literature that passenger alighting and boarding, fare collection

methods, vehicle type and size and bus floor plan layout have a direct impact upon the

time imposition incurred at the bus stop. Considering Brisbane, Australia as the case

study bus system, operating equipment used in Brisbane includes 12.5m two axle rigid

buses, 14.5m three axle rigid buses, and 19.0m articulated buses. The commonest type

operating on the Brisbane bus system are 12.5m double-door, two axle buses and

14.5m double-door, three axle buses.


To understand the passenger flow during the dwelling period, the bus layout needs to

be considered carefully. Both bus types mentioned above have double channel front

doors.

Figure 4-1: Bus channel layout and passenger flow on commonest Brisbane buses

Figure 4-1 shows a typical rigid bus floor channel. The 12.5m bus has a single channel

rear door while the 14.5m bus has a double channel rear door. The number of channels

on the rear door does not have any impact on the boarding process, because on the

Brisbane system boarding is presently limited to the front door. However, it benefits

the alighting process, because in the double channel rear door case both channels can

be used by passengers to alight.

As the bus commences dwelling, passengers alight from either or both doors, after

which boarding passengers use the front door. TCQSM (Kittelson and Associates,

2013a) assumes that, with two doors available 25% of the passengers alight through

the front door and the remaining 75% alight through the rear door.

Considering the front door, each door channel performs differently. The right door

channel is used exclusively by passengers using the right-side smart card reader (Cubic

GoCard). The left door channel is used by a mixture of passengers using the left side

smart card reader and a minority of passengers, who require driver assistance, visual

inspection of a paper ticket or pass, or to purchase a ticket. This implies that the left

door channel serves a wide range of passengers, which accounts for longest passenger

flow, and is the busiest door channel amongst the available door channels. Therefore

passenger flow is estimated for the left door channel of the front door using Equation

2-5. Boarding passengers are accounted for in two components, including passengers

Driver’s cabin Driver’s cabin


using smart card and passengers require driver assistance. Therefore passenger flow

time for front door left floor channel can be obtained by modifying Equation 2-5:

Equation 4-1

here

= Passenger flow time for front door left door channel (s)

= Total passengers alighting bus at loading area

= Proportion of passengers alighting through front door

= Alighting time through front door (s)

= Total passengers boarding bus at loading area

= Proportion of passengers boarding through front door left channel requiring driver assistance

= Boarding time for front door left channel requiring driver service (s)

= Proportion of passengers boarding through front door left channel not requiring driver assistance

= Boarding time for front door left channel not requiring driver assistance

Using Equation 2-7, dwell time can be expressed as,

Equation 4-2

Where,


= Time taken to open and close the bus doors (s)

= Boarding lost time (s)

Methodological Approach

By definition, loading area bus capacity is equal to the number of buses that are able

to be served by a loading area during a given period of time. It is a function of bus

dwell time, clearance time and operating margin as defined previously (Equation

2-13). To address the matters identified earlier, this section aims to understand how

the effects of bus-bus interference, adjacent lane traffic blockage and an adjacent

signalised intersection may be considered as components of total processing time of a

bus at a loading area, rather than through factors applied to whole of bus stop capacity.


The principal reason for this task is to work towards an improved methodology of bus

stop capacity estimation.

The development of the model will be carried out in five steps as shown in Figure 4-

2. The model is referred as ‘Modified Bus Stop Capacity Model (MBSC)” throughout

this thesis.


Figure 4-2: Steps followed to quantify the influencing capacity reduction factors

Model Development

A bus stop can be configured in a variety of ways. While this thesis is overall focused

on mid-block stops, this section is more generally applicable to near side and far-side

Include the influence of the green time to the total loading area processing time

TCQSM loading area bus capacity ( and bus stop capacity

Adjust to suspend the influence of green time ratio on the loading area bus capacity

STEP 1

Include the influence of traffic blockage and bus-bus interference as factors on loading area total processing time

STEP 2

Extract the added time towards the loading area processing due to traffic blockage and interference between buses

STEP 3

Isolate the time taken toward traffic and interference between buses

STEP 4

STEP 5

MBSC Model


bus-stops; those where buses cannot overtake except on the adjacent passing lane. It

includes a linear kerbside (either with platform or waiting pad) in the direction of

interest, with one or more loading areas in series.

A loading area is defined as a section of the stop which is designated for a single bus

to stop and dwell to serve passengers. According to TCQSM (Kittelson and Associates,

2013a) the bus capacity of a loading area is equal to the number of buses that are able

to be served by a loading area during a given period of time.

TCQSM includes the green time ratio as a factor to reflect that buses cannot access a

loading area on a bus stop that is either immediately upstream or downstream of a

signalised intersection during red time (whose ratio can be represented in context of

equation 1 as ) for the movement carrying the buses. However, to reach

the goal of developing the MBSC model, a more convenient way of considering the

influence of the adjacent signalised intersection is to consider that buses are processed

throughout the entire hour, but that the red time periods on the movement carrying the

buses, when buses cannot access the loading area, contribute an additional red time

component towards the loading area total processing time per bus. This remains

conceptually consistent with the other time components denoted in the denominator of

Equation 2-13 that occur during the effective green time periods for the movement

carrying the buses.

The next consideration is that bus stop capacity is defined by the existing TCQSM

model as follows (Kittelson and Associates, 2013a):

Equation 4-3

Where:

= Bus stop capacity (bus/h)


= Traffic blockage adjustment factor

The temporary occupation of a rearward loading area by a bus can prevent access by

another bus or buses to any forward loading areas. Therefore, a bus stop having several

loading areas is characterised by underutilisation of the forward-most loading areas.

The TCQSM (2013) methodology reflects the reduction in capacity due to interference

between buses within the stop by way of loading area effectiveness (Kittelson and


Associates, 2013a). It substitutes the actual number of loading areas with the number

of effective loading areas, . The product of

the capacity of a single loading area and the number of effective loading areas give the

reduced bus stop capacity that is reflective of this bus-bus interference. Importantly,

the TCQSM (2013) methodology assigns this influence of bus-bus interference as a

capacity reduction factor to the whole bus stop. However, to reach the research the

goal of developing the MBSC model, a more convenient way of considering the

influence of bus-bus interference may be to consider that it contributes an additional

time component towards the loading area total processing time per bus.

Where a bus stop is located on-street, the buses approaching and departing the bus stop

will share the travel lane with general traffic. Where a bus stop is located along a bus

lane facility or a BRT corridor, the buses approaching and departing the bus stop may

share the travel lane with express buses and/or any vehicles that are permitted to use

that lane, such as service vehicles. In both cases this traffic uses some capacity of the

travel lane otherwise available for buses in the immediate vicinity of the bus stop. The

TCQSM model reflects the reduction in capacity due to this traffic blockage through

a traffic blockage adjustment factor, .

Importantly, the TCQSM model allocates the impact of adjacent lane traffic as a

capacity reduction factor for the whole stop. However, to reach the goal of developing

the MBSC model, a more convenient way of considering the influence of traffic

blockage is to consider that it contributes an additional time component towards the

loading area total processing time per bus.

As shown in Figure 4-2, Step 1 requires alteration of Equation 2-13 to form a base

model that suspends the influence of green time ratio as a factor on loading area bus

capacity as follows:

Equation 4-4

Where:

= Loading area capacity under base model (bus/h)


The reciprocal of Equation 4-4 multiplied by 3,600 equals the base loading area

processing time per bus, exclusive of the time components due to bus-bus interference,

traffic blockage and red time periods on the movement carrying the buses, as follows:

Equation 4-5

Equation 4-3 uses the number of effective loading areas to represent the influence of

bus-bus interference. In step 2 it is convenient to replace this influence using a factor,

being the quotient of the number of effective loading areas to the number of actual

loading areas, such that . Effects of both traffic blockage and bus-bus

interference are incorporated into a reduced loading area bus capacity model as

follows:

Equation 4-6

Where:

Loading area capacity model reduced to account for traffic blockage and

interference between buses (bus/h)

The reciprocal of Equation 4-6 multiplied by 3,600 equals the loading area processing

time per bus, inclusive of the time components due to bus-bus interference and traffic

blockage, but exclusive of red time periods, on the movement carrying the buses, as

follows:

Equation 4-7

In step 3, the difference between Equation 4-5 and Equation 4-7 is equal to the sum of

additional time components of loading area total processing time per bus due to traffic

blockage and interference between buses as follows:

Equation 4-8

Where:

= additional time component towards loading area total processing time per

bus due to bus-bus interference (s)


= additional time component towards loading area total processing time per

bus due to traffic blockage (s)

Equation 4-8 provides a relationship between . This can be distributed to

develop approximate relationships for each of and as follows:

Equation 4-9

Equation 4-10

The terms and correspond to possible errors that could have occurred during this

distribution. These error terms are associated with and , and are very small

values. Therefore, in step 4, by assuming , the processing time components due

to traffic blockage and bus-bus interference can be estimated as follows:

Equation 4-11

Equation 4-12

With traffic blockage and bus-bus interference effects expressed as processing time

components, the total time taken by a bus to process at a loading area ( can be

estimated by:

Equation 4-13

In Equation 4-4 through Equation 4-13, the influence of green time ratio on loading

area bus capacity was suspended.

In step 5, if there is no adjacent signalised intersection this step can be skipped. For

the case of a mid-block bus stop, this step is not required as will be the total

loading area processing time. However, under the logic that is implied by Equation

2-13 when there is an adjacent signalised intersection, the influence of the red time

periods must be considered.


Figure 4-3 illustrates the influence of this red time periods component towards loading

area total processing time per bus, which is denoted by , in proportion to the sum

of all other processing time components, which occur during effective green time

periods when buses can access the loading area in accordance with the assumptions of

the original TCQSM model.

Figure 4-3: Processing time taken by a bus during a signal cycle

The additional time component of loading area total processing time per bus due to red

time periods on the movement carrying the buses may be determined proportionally

as follows:

Equation 4-14

It is important to note that this model is consistent with the TCQSM model, in that a

proportion of the bus dwell time equal to is able to occur during the red

time periods component.

In step 6 the total average processing time per bus at a loading area under this new

model is then equal to the following summation:

Equation 4-15

The modified loading area bus capacity (bus/h) can then be stated as follows:

Equation 4-16

In Step 7, the MBSC model can be stated to estimate bus stop capacity (bus/h) as an

alternative form to Equation 2-14 and Equation 4-3 as follows:

Equation 4-17


Comparison between TCQSM model and MBSC

model

The principal difference between the TCQSM model of Equation 2-14 and the MBSC

model of Equation 4-17 is that the former assigns the influences of adjacent signalised

intersection, traffic blockage, and bus-bus interference as factors that can be

attributable at the stop level, while the latter assigns these influences to the processing

of buses at the loading area level. However, the bus stop capacities estimated by both

models are equal. In order to compare the benefit of the MBSC model with the

TCQSM model, both were applied to a practical scenario.

TCQSM allows the user to apply the model into a wide range of bus facility types. A

scenario of a bus stop on an on-street bus facility is considered in the following

example: The stop consists of three off-line loading areas with an adjacent lane for

general traffic. TCQSM (2013) states that the equivalent number of loading areas is

equal to 2.65 (Kittelson and Associates, 2013a). A signalised intersection with an

equivalent green time ratio of 0.75 and a cycle time of 90s is located immediately

downstream of the stop for side street. General traffic and occasional non-stopping

buses constitute an equivalent of 300veh/h in the adjacent lane. This results in a traffic

blockage factor of 0.85 on the facility at the bus stop. A start-up time between buses

on the loading area of 10s is assumed. Re-entry delay is calculated to be 1.9s based on

a 7.0s critical headway and a 3.3s follow-on time. Under the parameters listed above,

clearance time on the loading areas is equal to 11.9s. In order to perceive a maximum

achievable capacity, operating margin was not considered in this case.

The system is assumed to operate with an on-board, touch-on, touch-off fare smart

card collection system. Of the four alighting passengers per bus, 25 percent are

distributed to the busiest of the two front door channels. The five boarding passengers

per bus are distributed evenly between the two front door channels. TCQSM (2013)

states that passenger service time for smart card payment is equal to 2.75s per

passenger. It is assumed that the average number of alighting and boarding passengers

using the busiest front door channel is 3.6 p/bus. The assumed door opening and

closing time is 3.5s. The boarding lost time is assigned to be 4s based on (Jaiswal,

2010). These together yield an average dwell time of 17.5s.


Table 4-1: TCQSM model and MBSC model comparison of bus stop capacity

Parameter TCQSM

Model

MBSC

model

Change

Loading area total processing time per

bus (s/bus)

25.1 44.6 +78%

Loading area bus capacity (bus/h) 107.5 80.7 -25%

Green Time Ratio (Stop Level) 0.75 NA NA

Traffic Blockage Factor (Stop Level) 0.85 NA NA

Bus-bus Interference Factor (Stop Level) 0.883 NA NA

Number of Actual Loading Areas 3 3 0%

Bus stop capacity (bus/h) 242 242 0%

Table 4-1 compares the calculation of bus stop capacity between the TCQSM model

and the MBSC model. The final bus stop capacity is the same; however, the

intermediate calculations of loading area processing time, loading area bus capacity,

and the factors used differ. Figure 4-4 illustrates the differences in calculation of

loading area total processing times per bus between the models. The contributions of

traffic blockage time component, red time periods component, and bus-bus

interference time component are clear. The MBSC model provides a more complete

picture about the time components of the loading area total processing time per bus,

which is not explicit in the TCQSM model.


Figure 4-4: Comparison of time components of loading area total processing time per bus

between TCQSM model and MBSC Model

This section shows the MBSC model can be used to gain greater insight in bus stop

capacity consideration.

Sensitivity Analysis on Traffic Blockage Factor

In the TCQSM model, traffic blockage factor ( depends on the lane type, traffic

volume and capacity in the lane used by the buses at the stop, the configuration of the

bus stop, and the proximity of the stop to a signalised intersection. It is a particularly

useful parameter for on-street bus operations. A value near 0 would represent

congested traffic conditions on the lane shared by the buses at the bus stop, for instance

a left turning lane under Left Hand Travel (LHT) conditions. A value close to 1.0

would represent conditions such as restrictions on general traffic and exclusive travel

lane for transit vehicles.

As it also incorporates the influence traffic blockage, The MBSC model can be used

to study the impacts of traffic blockage in various aspects. For the purposes of this

sensitivity analysis, full flexibility of the MBSC model in accounting for traffic

blockage irrespective of proximity to a signalised intersection is assumed.

Figure 4-5 is based on the case study presented for Table 4-1, but with traffic blockage

adjustment factor varying between 0.5 and 1.0. The former reflects half of the capacity

of the lane used by buses at the bus stop being consumed by general traffic, and the

0

10

20

30

40

50

TCQSM MBSC

Load

ing

Area

Pro

cess

ing

Tim

e (s

/bus

)

td(g/C) tc ttb tbbi tred


latter reflects no general traffic in the lane used by the buses at the bus stop. The figure

demonstrates the considerable influence that traffic blockage can have on loading area

processing time per bus and therefore capacity of the bus stop.

The impact of traffic blockage factor on the interdependent time components of

loading area total processing time per bus, due to both bus-bus interference and red

time periods, is apparent. Meanwhile, there is no change to either dwell time or

operating margin as these are not parameters related to the traffic volume. In this case

changes to clearance time occur due to the increase in re-entry delay when a greater

amount of adjacent travel lane traffic is present.

Figure 4-5: Influence on traffic blockage on loading area processing time per bus and stop

capacity

Sensitivity Analysis on Bus-Bus Interference Factor

Bus-bus interference factor is equal to the ratio between number of effective loading

areas and number of actual loading areas at the bus stop. For a bus stop with three on-

line (First-In-First-Out) loading areas, under random arrivals the number of effective

loading areas is 2.45, while under platooned arrivals the number of effective loading

areas is 2.56 (Kittelson and Associates, 2013a). As in the previous case study, when


loading areas are off-line the number of effective loading areas is 2.60. These equate

to bus-bus interference factors of 0.82, 0.86, and 0.87 respectively.

Figure 4-6 is based on the case study presented for Table 4-1, but with bus-bus

interference factor varying according to the values above to reflect the different

loading area configurations. The figure demonstrates the marginal influence that

loading area configuration and hence bus-bus interference can have on loading area

processing time per bus and therefore capacity of the bus stop.

Figure 4-6: Influence of bus-bus interference on loading area processing time per bus and

bus stop capacity

The impact of bus-bus interference factor on interdependent time components of

loading area total processing time per bus due to red time period is noticeable, however

there is a slight unnoticeable impact on traffic blockage as well. There is no change to

clearance time, as that component reflects re-entry into the travel lane in bus turnover

on the loading area as a separate occurrence from manoeuvring under bus-bus

interference.

Sensitivity Analysis on Green Time Ratio

Green time ratio is equal to the ratio between effective green time of the movement

carrying the buses and signalised intersection cycle time. The case study example

above assumed an on-street bus facility with an adjacent signalised intersection with a


green time ratio of 0.75. Where two major facilities intersect a green time ratio of 0.5

is typical. A green time ratio of 1.0 reflects the absence of any adjacent signalised

intersection.

Figure 4-7 is based on the case study presented for Table 4-1, but with green time

ratios of 0.5, 0.75 and 1.0. The figure demonstrates the major influence that green time

ratio can have on loading area processing time per bus and therefore capacity of the

bus stop.

Figure 4-7: Influence on green time ratio on loading area processing time per bus and bus

stop capacity

The impact of green time ratio on interdependent time components of loading area

total processing time per bus due to both traffic blockage and bus-bus interference is

apparent. In this case, dwell time also changes due to the assumption that some of the

dwell time occurs during red time periods at the adjacent signalised intersection. For

this reason, dwell time is at a maximum for a green time ratio of 1.0, reflecting no

adjacent signalised intersection presence. There is a marginal change to clearance time

due to re-entry delay increasing with a reduction in green time ratio.


Examination of MBSC Model

A key motive of developing the MBSC model was to be able to better understand the

influence of traffic blockage, bus-bus interference and signalised intersection impacts

at the loading area level, which is particularly useful when examining alternative

operating practices. To demonstrate use of the MBSC model in this respect, a

hypothetical advanced operation is considered containing an off-board fare collection

system with all-door alighting and boarding. With an off-board fare collection system

passenger service time was suggested to be 1.75s per passenger (Kittelson and

Associates, 2013a). The four alighting and five boarding passengers per bus are

distributed evenly between the two front and two rear door channels. Other parameters

are kept unchanged in order to observe the impact on loading area bus processing time.

Figure 4-8 demonstrates how each component of processing time is affected by this

operating practice’s influence on dwell time. The first stacked bar represents the total

of all components of loading area processing time per bus from the case study of Figure

4-4 while second bar represents the MBSC model with Off-Board Fare (OBF)

collection system.

A rational explanation for the reduction in bus-bus interference time component is the

quicker turnover of buses on the loading areas under the OBF operation reducing the

time required by a given bus to reach an available loading area. A rational explanation

for the reduction in red time periods component is the quicker turnover of buses on the

loading areas under the OBF operation reducing the time required by a given bus to

progress through red periods to reach an available loading area. While a rational

explanation for the reduction in traffic blockage component is the quicker turnover of

buses on the loading areas under the OBF operation reducing the time required by a

given bus to progress through the travel lane traffic to reach an available loading area.

Hence the OBF system beneficially reduces all components of loading area processing

time per bus apart from clearance time. The resultant bus stop capacity is 254 bus/h,

which is an increase of 16 bus/h over that of the original system.


Figure 4-8: Loading area bus processing time comparison with technological advancements

Summary

The theoretical improvement presented in this chapter by way of development of the

Modified Bus Stop Capacity (MBSC) model enhances the TCQSM model (Kittelson

and Associates, 2013a) by quantifying the factors that influence bus stop capacity as

components of processing time at the loading area level. The model was developed

considering a general bus stop to expand the applicability of the model. Comparison

of the MBSC model with the TCQSM model indicated that the MBSC model assigns

the influences of adjacent signalised intersection, traffic blockage and bus-bus

interference as additional time components of the loading area total processing time

per bus, whereas the TCQSM model accounts for these influences by way of factors

at the stop level. This is an important contribution to bus stop capacity estimation

because the MBSC model provides detailed information to estimate capacity for both

traditional and non-traditional operating practices.

Bus stop capacities determined using both MBSC and TCQSM models gave similar

results; however, the intermediate estimations of loading area processing time, loading

area bus capacity and the impacts of influencing factors are new. The contributions of

adjacent signalised intersection, traffic blockage and bus-bus interference are clear

because they appear as time components in the total loading area processing time. As

there are no whole-of-stop capacity reduction factors involved, this is a particularly


important clarification that will better support the development of theory in later

chapters of this thesis.

The effectiveness of the MBSC model is demonstrated by conducting sensitivity

analyses for the quantified parameters. The MBSC model can be applied to study bus

stop operation for a wide range of bus facility types. Therefore, the model can be used

to better understand facility capacity under various operational practices. This will also

be helpful to transit analysts in bus route planning and design, particularly for peak

periods.

This chapter has partly answered research questions 1 and 3 by identifying the

accuracy of the TCQSM model for capacity estimation of an OS-MID-OFF bus stop.

The TCQSM model was improved by quantifying secondary influences such as traffic

blockage, bus-bus interference and adjacent signalised intersection to

comprehensively analyse their impacts on bus stop capacity.

Chapter 5: Microscopic Simulation Modelling of an Off-Line, Mid-Block Bus Stop 65

Microscopic Simulation

Modelling of an Off-Line, Mid-Block Bus

Stop

Overview

This chapter is based on the paper ‘Improving Capacity Estimation of High Volume

On-Street Bus Facilities with Yield-to-Bus Rule’, presented at the Australasian

Transport Research Forum (2018). The aim of this chapter is to present the

development of a microscopic simulation model of an on-street, mid-block, off-line

(OS-MID-OFF) bus stop in order to understand on-street bus operations with adjacent

lane general traffic flow.

This chapter first describes the microscopic simulation approach used in this research

to model the OS-MID-OFF bus stop. Section 5.3 presents the simulation model

development of the test bed bus stop. Section 5.4 demonstrates the model development

and the model cross validation with the deterministic bus stop capacity model. Model

outcomes are then analysed in section 5.5 to conclude the chapter. This fulfils research

objective 3.

Microscopic Simulation Modelling Approach

As was stated in Chapter 3, it is not feasible to develop empirical models from real

data to estimate bus stop capacity under all possible conditions for the following

reasons:

1. It is difficult to estimate the potential capacity of a given bus stop because most

bus stops operate below capacity due to the necessity of conservative

timetabling. Those which have been observed to reach potential capacity only

do so for a short period of time, thereby providing limited data.

66 Chapter 5: Microscopic Simulation Modelling of an Off-Line, Mid-Block Bus Stop

2. It is very difficult to collect real data for bus stop capacity for the full range of

general traffic flow rate in the adjacent lane because of its limited variation in

time. In particular, it is very difficult to observe situations with high adjacent

lane traffic flow rates that are close to saturation for long periods of time.

3. Overall, arterial roads and bus stops on them usually operate with degrees of

saturation across limited ranges.

Microscopic simulation is useful when there is a need to reproduce behaviour of real-

world systems. In contrast to theoretical models, microscopic simulation modelling

can provide a visual representation of the system. A key advantage of microscopic

simulation for this research is that it is able to test bus stop operation across a complete

range of bus and adjacent lane traffic flow rates.

This chapter describes the development and cross-validation of a simulation model

with the MBSC model from Chapter 4 in order to understand the feasibility of using

simulation across a wide range of bus and adjacent lane flow rates, in turn to develop

a widely accurate model for an OS-MID-OFF bus stop.

Even though there are many traffic microsimulation packages available, only a few

can model transit vehicles, including Aimsun (Advanced Interactive Microscopic

Simulator for Urban and Non-urban networks). The basic data required for model

development using this package includes; network geometry, characteristics of

vehicles, driver characteristics and driver behaviour, travel demand, traffic control

systems, and traffic flow models (car-following and lane changing). According to its

developer, the simulation software has been applied extensively in commercial

projects across a wide range of environments, where its capability of microsimulation

is tested and verified (TSS, 2016).

Microscopic Simulation Model Development

The microscopic simulation model developed in this research provides a realistic

representation and reproduction of a testbed OS-MID-OFF bus stop. In this study, the

testbed comprises of a linear off-line bus stop adjacent to a general traffic lane, which

is shown in the layout demonstrated in Figure 5-1. A public transport plan is included


with several public transport lines with preassigned dwell times and standard

deviations. The arrival pattern of the buses was assumed to follow a normal

distribution. Preassigned values such as bus dwell time and bus headway were adjusted

according to suit the simulation scenario. The buses are simulated according to a car-

following model. Gap acceptance logic is applied by Aimsun to model the bus merging

manoeuvre. Aimsun, in its standard manner, generates public transport vehicles

(buses) stochastically according to a normal distribution defined by mean headway and

standard deviation of headway.

Figure 5-1: Layout of the simulation testbed of type bus stop of this research

In order to monitor the behaviour of buses, detectors were placed along the testbed

section. The model considers stopping buses at bus stops, vehicles travelling in the

adjacent lane, and buses trying to re-enter into the adjacent lane. The goal is to

reproduce the fundamental operation of an OS-MID-OFF bus stop. Various

combinations of bus flows and adjacent lane general traffic will be used to establish

the relationship between bus stop capacity and adjacent lane general traffic flow rates.

The following are parameters used for the simulation experiments. Buses were

assumed to be standard 12.0m (40ft) rigid. Drivers’ reaction time, reaction time at a

stop, and reaction time at a traffic signal are some of the parameters that govern the

traffic flow models; car-following and lane changing models and will also affect the

performance of the entire network. Aimsun requires estimation of the driver’s

performance characteristics of reaction time. A simulation step of 0.20s was used in

order to ensure that the drivers’ behaviour is accurately modelled (TSS, 2016).

Widanapathiranage et al. (2015) identified that driver reaction times varies between

0.75s and 1.5s. For this study, driver reaction time for both buses and cars were

assigned to be 1.20s at a bus stop. This study also reflected a basic model of operation


in order to develop a fundamental relationship. The upstream section of the bus stop

was created with 10km length to avoid any virtual queue being created during the

simulation.

Potential capacity of buses was measured as outflow from the test bed using the

detector downstream of the bus stop (Figure 5-1). An Aimsun Application

Programming Interface (API) was used to obtain the timestamp for the detectors placed

on the testbed. Clearance times for each bus were measured from the resultant

timestamps from the detectors. Because Aimsun is a stochastic simulation model,

results differ with each replication. Each replication was carried out for one hour and

ten replications were performed to estimate an average for a reliable result.

Model Verification

We define the limit state bus capacity of a bus stop to be the maximum achievable

outflow of buses (Widanapathiranage et al., 2015). Using Equation 2-14, limit state

bus capacity was estimated using the TCQSM model but with no operating margin and

no adjacent signalised intersection. It can be quantified deterministically as follows.

Equation 5-1

Where,

= Limit state bus capacity (bus/h)

= Clearance time (s)

= Dwell time (s)


The number of effective loading areas suggested by the TCQSM for an off-line bus

stop with two loading areas equal to 1.85 was used. The limit state bus stop capacity

achievable under the simulation model was found by modelling conditions of

continuous upstream bus queuing with no adjacent lane general traffic. This was

attained by creating a saturated state such that inflow to the bus stop exceeded the

outflow of the bus stop. The detector placed downstream of the bus stop was used to

measure the exiting bus flow rate, being the limit state capacity. A range of conditions

of dwell times were modelled. In all cases, all of the buses stop at the bus stop using


one of the two loading areas such that there were no non-stopping buses in the passing

lane. Dwell times ranging from 5s to 90s were simulated. A 5s dwell time represents

the case where a bus arrives at the bus stop, opens and then closes doors, and departs

almost immediately. This was simulated to attain the highest capacity achievable at

the bus stop. Average dwell times range from 10s to 60s for a bus stop located at an

arterial road (Kittelson and Associates, 2013a). However, to obtain a lower range of

capacity, a dwell time of 90s was also simulated.

Figure 5-2 illustrates values determined for the bus capacity from the simulation across

the ranges of average dwell time. The bus capacity was also calculated using Equation

5-1 as a function of dwell time with no operating margin.

Figure 5-2: Testbed limit state bus stop capacity vs dwell time according to simulation model and TCQSM model (Kittelson and Associates, 2013a).

Ten replications for each dwell time were simulated. The average of the simulated

values was cross validated with the values obtained from the TCQSM theoretical

model. The models were compared by finding the Root Mean Square Error (RMSE).

RMSE was found to be 0.91 ( =0.84). This shows that the simulation model fits well

with Equation 5-1. Therefore, comparisons could then be made between the TCQSM

model and the simulation model in terms of their estimation of OS-MID-OFF bus stop

capacities with adjacent lane general traffic.

050

100150200250300350400450

0 10 20 30 40 50 60 70 80 90

Stop

bus

capa

city

(bus

/h)

Dwell time (s)

Simulation TCQSM


Microscopic Simulation Model Implementation

The limit state bus capacity of a bus stop was determined using two methods; the

TCQSM method of Equation 5-1, and the Aimsun simulation model testbed developed

in this study. General traffic shares the same lane with buses and is assumed to arrive

randomly upstream of the bus stop and pass the bus stop in the adjacent general traffic

lane. The traffic flow rate was varied from 0 veh/h to 1,800 veh/h, and for each input

traffic flow rate, maximum achievable outflow of buses and outflow of general traffic

were measured using the virtual detector placed downstream of the bus stop. The

maximum achievable outflows reflect the limit state capacities of buses and general

traffic respectively.

According to the TCQSM theory for an OS-MID-OFF bus stop, buses arrive, dwell in

the loading area to serve passengers, and re-enter the adjacent general traffic lane from

the bus stop. If there is a sufficient gap the bus would re-enter into the traffic lane

immediately, otherwise the bus would wait for an acceptable gap in the traffic lane. As

the adjacent lane general traffic flow rate is increased, longer re-entry delays are

expected at the bus stop accordingly. An average dwell time of 20s was used

throughout this study to reflect a typical bus stop operation (Widanapathiranage et al.,

2015) and with no operating margin. Start-up component of clearance time was

assigned as 10s for a standard bus (Levinson, 1997). Traffic blockage factor was

omitted, as the OS-MID-OFF bus stop was assumed to be away from the influence of

a signalised intersection and away from any queue generated by the traffic signal

(Kittelson and Associates, 2013a). Re-entry delay was estimated using the TCQSM

theory (Equation 2-7) and subsequently bus stop capacity was calculated using

Equation 5-1. Figure 5-3 illustrates the relationship between adjacent lane traffic flow

rate and achieved bus stop capacities using each of the two models.


Figure 5-3: Limit state bus stop capacity determined using TCQSM and simulation testbed

vs adjacent lane flow rate (Kittelson and Associates, 2013a)

Figure 5-3 illustrates stop bus capacities measured on the simulation testbed and the

TCQSM model as adjacent traffic volume varies. According to the TCQSM model,

bus stop capacity decreases gradually as the adjacent flow rate increases. This is due

to the increasing clearance time. According to the theory, clearance time comprises of

two components; the time for the bus to start up and clear its own length plus the time

taken to re-enter to the adjacent lane. In this study we assumed that the buses have the

same start up time. For a mid-block bus stop, away from the influence of any signalised

intersection, re-entry delay is the time taken for gap acceptance.

It is apparent from Figure 5-3 that capacity reduces gradually until the adjacent lane

general traffic flow rate reaches 900veh/h. Subsequently, the values drop in a concave

manner. This can be explained by the rapid increase in the re-entry delay. Because the

bus stop capacity and the clearance time are inversely proportional to each other with

the increase in the adjacent lane traffic, buses have to wait longer in the bus stop to re-

enter due to shorter headways in the traffic. The TCQSM model produces a one-way

effect; the adjacent lane traffic flow rate (X axis) affects the bus stop capacity (Y axis),

however the bus stop capacity does not affect the adjacent lane traffic capacity, as

adjacent lane traffic flow rate is purely an input to the deterministic model.

In the simulation testbed for each major stream flow rate, each point in the clusters of

10 reflects a 1h simulation run. Bus inflow was simulated with a 10s headway, which

0

40

80

120

160

200

240

0 300 600 900 1200 1500 1800

Bus S

top

Capc

ity (b

us/h

)

Adjacent lane flow rate(veh/h)

TCQSM Simulation testbed

Adjacent lane saturation effect


corresponds to an inflow of 360 bus/h. With no adjacent lane traffic, it was observed

that outflow was similar to the inflow. Similar to the TCQSM model, the simulation

testbed shows a reduction in theoretical bus stop capacity when as adjacent lane traffic

flow rate increases. The reduction in bus capacity as adjacent lane traffic flow rate

increases is more substantial under the simulation method. It is evident that the

simulation testbed produces a two-way effect; the adjacent lane general traffic flow

rate (X axis) affects the bus stop capacity (Y axis), while any particular measured bus

stop capacity is also reflective of the maximum volume of adjacent lane general traffic.

However, it will be useful to have a model a two-way model where the adjacent lane

traffic flow rate and the bus stop capacity will be interdependent. This will be

addressed in the future chapters.

Both curves show similar trends until the adjacent lane flow rate reaches 1,200veh/h.

For adjacent lane flow rates larger than 1,200veh/h the TCQSM model and simulation

testbed behave rather differently. The curve representing the TCQSM model shows a

gradual decrease in bus stop capacity until the adjacent lane becomes saturated.

Remarkably, the curve returns a non-zero value for the bus stop capacity even when

the adjacent lane is saturated. This implies that the TCQSM model assumes that while

the high traffic flow rate can reduce the capacity due to the traffic blockage, the bus

stop can still have a considerable rate of buses reaching the bus stop. However, this

may be impossible in practice, because once the lane has become saturated, there will

be no space to accommodate buses. Hence, no buses will arrive at the bus stop.

However, this is reflected in the simulation testbed curve. The reduction in capacity

due to adjacent lane saturation is marked in Figure 5-3 using grey arrows.

The simulation testbed revealed interesting behavioural patterns of the adjacent lane

traffic while re-entering, which is not reflected in deterministic models. In locations

where there is no yield-to-bus rule, one or both of the following can occur.

1- When the adjacent lane operates near capacity, the bus driver will find

it to be practically impossible to re-enter into the adjacent lane until they find a

suitable gap. However, if a second bus arrives at the bus stop before the first bus

has departed, then the second bus will have to wait until the first bus re-enters

into the traffic lane. Therefore, the stopped second bus will obstruct the adjacent


traffic lane, which will cause a temporary queue upstream of the bus stop until

the first bus has completely departed.

2- When a bus driver is trying to re-enter, general traffic in the adjacent

lane cannot pass the bus stop. This will cause the vehicles in the adjacent lane to

queue upstream of the bus stop. Therefore, the re-entering bus driver will see

compressed adjacent lane stream in the exit area of the bus stop. Once the bus

re-enters, the queue will be released as a platoon. This is a repetitive event.

These observations indicate that the simulation testbed does represent operation of this

bus stop type more accurately than the TCQSM model. Following the literature review

of Chapter 2, no research has been found to have addressed these effects in

deterministic models.

Summary

This chapter demonstrated that microscopic simulation modelling can be used to study

and analyse the operation of an on-street, mid-block, off-line (OS-MID-OFF) bus stop,

in order to determine limit state bus stop capacity.

The chapter concludes that the TCQSM model and the Aimsun microscopic simulation

testbed of the type bus stop developed in this study produce similar limit state bus

capacity when there is no general traffic present in the adjacent lane. However, as

adjacent lane general traffic increases, OS-MID-OFF bus stop capacity measured by

the microscopic simulation testbed does reduce more than the TCQSM model, because

some of the lane capacity is consumed by general traffic.

This research has found that when applied to an OS-MID-OFF bus stop, the TCQSM

model does not provide a clear understanding of the case where adjacent lane flow rate

reaches saturation flow rate. It assumes that, even though high traffic flow rate can

reduce capacity due to high re-entry delay, buses are still able to reach the bus stop. In

reality, this will not be the case. Once the adjacent travel lane reaches its general traffic

capacity, the lane will no longer be able to accommodate buses. This is, however,

reflected in the simulation output, which yields a bus stop capacity of zero under such

conditions.


A further advantage when using the simulation testbed output is that, unlike the

TCQSM model, its relationship provides an understanding of the maximum flow rate

of the adjacent lane traffic that can be supported at the OS-MID-OFF bus stop, for a

known bus stop capacity.

Future chapters of this thesis will further explore and address issues identified in this

chapter. In particular, a deterministic model will be developed beyond the existing

TCQSM model, which will account for adjacent lane general traffic that is more

reflective of the operation observed using the microscopic simulation model in this

chapter. It is also important to develop a two-way model, where the OS-MID-OFF bus

stop and the adjacent lane traffic flow rate can be considered interdependently. A two-

way model can benefit the transport analyst in optimizing bus facility operation during

peak periods when congestion is most likely to occur.

This chapter has answered research question 2 by developing microscopic simulation

model for a testbed OS-MID-OFF bus stop. The simulation model was then used to

test scenarios across a range of bus and adjacent lane flow rates to observe the

behaviour of buses and adjacent lane traffic. Observations made through the simulation

indicated that TCQSM model can be further improved by considering the adjacent lane

behaviour, which has partly answered research question 3.

Chapter 6: Theoretical Model of On-Street, Mid-Block, Off-Line Bus Stop Capacity with Adjacent Lane Traffic

75

Theoretical Model of On-Street,

Mid-Block, Off-Line Bus Stop Capacity with

Adjacent Lane Traffic

Overview

The contents of this chapter are based on the journal article, “Capacity Estimation of

On-Street, Mid-Block, Off-Line Bus Stops Considering Yield-to-Bus Rule (Hisham et

al., 2019a)” published in Transportation Research Record.

The aim of this chapter is to present the development of a deterministic methodology

to better understand the relationship between bus stop capacity and adjacent lane flow

rate for an on-street, mid-block, off-line (OS-MID-OFF) bus stop.

The first part of this chapter discusses the necessity of including the influence of

adjacent lane traffic on capacity estimation of an OS-MID-OFF bus stop. The next part

presents the methodological approach towards model development. The latter part of

the chapter then compares the TCQSM model with the model developed here and

analyses model outcomes. This fulfils research objective 4.

Influence of Adjacent Lane General Traffic on

Operation of an On-Street, Mid-Block, Off-Line Bus

Stop

The theoretical methodology for bus stop capacity presented in TCQSM (2013)

incorporates the influence of traffic flow surrounding a potentially critical bus stop

through a ‘traffic blockage adjustment factor’. This factor is a function of the volume

to capacity ratio of the kerb lane at the intersection and a location factor, which is

dependent upon lane type and bus stop location with respect to the signalized

intersection. Importantly the TCQSM (Kittelson and Associates, 2013a) states that “if

76 Chapter 6: Theoretical Model of On-Street, Mid-Block, Off-Line Bus Stop Capacity with Adjacent

Lane Traffic

the stop in question is located more than one-half block away from a traffic signal, and

outside the influence of a queue of stopped vehicles generated by the signal, this step

can be skipped”. At a mid-block bus stop away from the influence of a signalized

intersection, no traffic blockage adjustment applies. Therefore, under this case the

TCQSM does not address the interaction between bus stop operation and adjacent lane

traffic flow.

However, the previous chapter revealed that operation of buses obstructs adjacent lane

traffic flow, and conversely adjacent traffic flow obstructs the bus stop operation.

Observations from the previous chapter are as listed below.

Re-entering buses obstruct the adjacent lane traffic flow thereby creating a

compressed adjacent lane stream past the re-entry area of the bus stop.

Formation of the compressed adjacent lane flow results in general traffic

passing with shorter headways. This means more adjacent lane vehicles are

being accommodated within a more limited time-space at the expense of bus

stop re-entry capacity.

When the adjacent lane operates at or near capacity there will be no space to

accommodate buses. The saturated adjacent lane obstructs bus stop operation

by limiting the arrival rate of buses at the bus stop.

For a particular value of bus stop capacity, the simulation testbed provided an

understanding of the maximum flow rate of the adjacent lane that can be supported at

the bus stop. The inter-dependency between the buses and the adjacent lane traffic

flow has not found to be addressed in the literature for an OS-MID-OFF bus stop, so

it is necessary to address this by developing an improved deterministic model.

To incorporate the influence of the adjacent lane traffic flow rate, in this thesis we

contend that at an OS-MID-OFF bus stop, buses will obstruct adjacent lane general

traffic flow during certain components of the loading area processing time period, so

additional time may be required to accommodate adjacent lane traffic under saturated

conditions. The compressed general traffic stream will affect the re-entry delay of

buses due to gap acceptance. The next section develops a deterministic model, which


77

addresses the matters mentioned above to more accurately to represent operation of an

OS-MID-OFF bus stop.

Model Development

This section presents the development of an alternative deterministic model for

estimation of capacity of an OS-MID-OFF bus stop, which improves the TCQSM

model presented above, by ensuring that stop bus capacity operation does not allow

the adjacent lane general traffic to become over-saturated. This model requires the

quantification of average total processing time per bus, which can be considered as the

sum of fundamental components (Hisham et al., 2018a) as follows:

+ Equation 6-1

Where

= Average total processing time per bus (s)

= Start-up time (s)


= Dwell time (s)

= Additional time requirement for adjacent lane traffic (s)

= Operating margin on dwell time (s)

Hisham et al. (2019a) introduced the term as an additional time requirement

during the total loading area processing time per bus that may be needed in order to

accommodate adjacent lane general traffic such that it does not become over-saturated.

The start-up time is required as an input.

Hisham et al. (2018a) quantified the bus-bus interference factor as follows:

Equation 6-2

Where is the bus-bus interference factor, is the number of effective loading

areas according to the TCQSM (Kittelson and Associates, 2013a) definition and

is the number of actual loading areas.


Lane Traffic

The additional time component towards average total processing time per bus, due to

bus-bus interference, is modified from Equation 4-12 of Chapter 4 as follows:

Equation 6-3

Where

= Time taken due to bus-bus interference (s)

= Bus-bus interference factor

For a given average total loading area processing time per bus, the total time required

(s/bus) for adjacent lane traffic to proceed through the bus stop at saturation flow rate

in order to meet a given arrival flow rate is given by:

Equation 6-4

Where

= Average time taken by adjacent lane to proceed through the bus stop (s)

= Average total processing time per bus (s)

= Adjacent lane general traffic arrival flow rate (veh/h)

= Adjacent lane saturation flow rate (veh/h)

The time available for the adjacent lane to pass during the average total loading area

processing time is equal to the sum of the time components of average total loading

area processing time per bus, during which the bus does not obstruct the adjacent lane

and is given by:

+ Equation 6-5

Substituting Equation 6-1 and Equation 6-5 into Equation 6-4 will result in:

Equation 6-6

Thus, a relationship for can be derived. Therefore, we get,

Equation 6-7


79

When the adjacent lane is under-saturated (adjacent lane degree of saturation,

), will theoretically be less than zero. When the adjacent lane traffic flow rate

reaches the point of saturation ( ), will be zero. In these cases, adjacent

lane flow rate can be accommodated within the total loading area processing time

without requiring any additional time requirement. Therefore, by setting

when the adjacent lane is under-saturated, sufficient average total loading area

processing time exists to achieve the loading area bus capacity. However, for adjacent

lane traffic flow rates exceeding the value at the point of saturation, in order to

maintain and not exceed a degree of saturation, = 1.0, must increase from zero.

To accommodate all three cases, the additional time requirement to accommodate

adjacent lane traffic flow rate given by Hisham et al. (2019a) is as follows:

Equation 6-8

Therefore, the loading area bus capacity can be estimated using the existing and

derived variables. If a bus requires, on average, (s/bus) to process through a loading

area, the number of buses able to be processed through the same loading area during

one hour is by definition the loading area bus capacity, (bus/h/LA).

An important variation from the TCQSM (Kittelson and Associates, 2013a) model is

acknowledgement that general traffic in the lane adjacent to a given loading area has

a theoretical capacity, which can be stated (veh/h) as:

Equation 6-9

The degree of saturation of general traffic in the adjacent lane is then given by:

Equation 6-10

This model maintains the gap acceptance approach according to Equation 2-8 and

Equation 2-9. However, because adjacent lane traffic cannot pass during start-up time


Lane Traffic

and bus-bus interference time, we must consider that the re-entering bus driver will

see a compressed adjacent lane stream passing the bus stop during other times. For

purposes of estimating re-entry delay due to gap acceptance, the adjacent lane traffic

flow rate is adjusted according to:

Equation 6-11

Using Equations Equation 6-9 through Equation 6-11 the adjusted adjacent lane traffic

flow rate ( is given by:

Equation 6-12

The value of is then substituted into Equation 2-9 and Equation 2-8 to calculate re-

entry delay. Equation 6-12 also enables us to determine whether the adjacent lane is

operating at a saturated condition, according to its calculated degree of saturation.

However, adjacent lane degree of saturation, must be known in order to estimate

, and therefore in order , , , , , , , and hence itself must be

known. This requires recursive estimation using Equation 6-1 through Equation 6-12

until the adjacent lane adjusted flow rate settles, according to the following objective

function:

Equation 6-13

represents the starting value that is used to estimate the value of . For the

first iteration the trial value can be set to .

Bus stop capacity can then be estimated as follows:

Equation 6-14

This model will be referred as the Bus Capacity with Adjacent Lane Traffic

Requirements (BCAL) model throughout the remainder of this thesis.


81

BCAL Methodology to Estimate Capacity of an

On-street, Mid-block, Off-line Bus Stop

Based on the theoretical methodology presented in the previous section, the procedure

to estimate capacity for an OS-MID-OFF bus stop is presented below.

Once the recursive process settles the values obtained for , ,

, can be used to estimate and thus

Y

N

Input

Estimate

?

Assign

Figure 6-1: Flow chart for OS-MID-OFF bus stop capacity estimation using BCAL model


Lane Traffic

Comparison between TCQSM and BCAL

Models

For direct comparison, the theoretical bus capacity of an OS-MID-OFF bus stop was

determined both the TCQSM theory of Equation 2-14 and the BCAL model of

Equation 6-1 through Equation 6-14 under conditions where adjacent lane general

traffic has absolute priority over re-entering buses. An average dwell time of 20s was

used throughout this study to reflect a typical bus stop operation. The operating margin

was set to zero to ensure a limit state condition on the bus stop (Widanapathiranage et

al., 2015). The start-up component of clearance time was assigned to be 10s for a

standard bus (Levinson, 1997). Re-entry delay was estimated using a 7.0s critical gap

and 3.3s follow-up time as per TCQSM (Kittelson and Associates, 2013a). The number

of actual and effective loading areas were assigned to be 2 and 1.85 respectively. For

the TCQSM model, traffic blockage factor was omitted as the bus stop is assumed to

be away from the influence of a signalised intersection (Kittelson and Associates,

2013a). Figure 6-2 illustrates the relationship between adjacent lane traffic flow rate

and theoretical stop bus capacity using each of the two theories.

Figure 6-2: Comparison between TCQSM and BCAL models of OS-MID-OFF bus

stop capacity vs. adjacent lane traffic flow rate

0

25

50

75

100

125

150

175

200

225

250

0 300 600 900 1200 1500 1800

Stop

bus

capa

city

(bus

/h)

Adjacent lane volume (veh/h)

BCAL TCQSM

Compressed adjacent flow effect

Adjacent lane saturation effect


83

Figure 6-2 confirms that the BCAL model gives the same bus capacity as the TCQSM

model when no adjacent lane traffic is present. Both methods reflect a reduction in

theoretical bus stop capacity when as adjacent lane traffic flow rate increases.

However, as adjacent lane flow rate increases, under the BCAL model the adjusted

adjacent lane flow rate of Equation 6-13 results in a re-entry delay that is increasingly

greater than the TCQSM model. Therefore, the resultant bus capacity is increasingly

smaller and is shown using a grey arrow in Figure 6-2.

Further, once the adjacent lane traffic flow rate reaches approximately 1,400veh/h, bus

stop capacity declines more steeply under the BCAL model. This is due the adjacent

lane having reached the point of saturation, so the additional time requirement to

accommodate adjacent lane traffic flow rate according to Equation 6-8 increases from

zero. For a given loading area processing time period, this additional time requirement

proportionately reduces the amount of time available to process the bus. When

adjacent lane flow rate reaches the assigned saturation flow rate of 1,800veh/h, bus

stop capacity theoretically reaches zero under the BCAL model. The BCAL model

accurately represents the observations made using the simulation testbed.

Summary

The theoretical model developed in this study for an OS-MID-OFF bus stop with

adjacent lane traffic requirements (BCAL) enhances the TCQSM model by ensuring

that bus stop capacity conditions do not cause the adjacent lane general traffic to

become over-saturated. Comparison with TCQSM theory indicated that the BCAL

model estimates bus stop capacity properly once the adjacent lane operates at

saturation, whereas the TCQSM cannot accurately represent conditions over this

range. This is a significant improvement to bus stop capacity estimation on facilities

having OS-MID-OFF bus stops, where buses share the travel lane with high volumes

of general traffic, such as major urban arterial roads.

This chapter answered research questions 1, 2 and 3 by developing a deterministic

model to account for adjacent lane general traffic. The model accounts for obstruction

that can occur on the adjacent lane due to re-entering buses, and an additional time

required of the loading area processing time to accommodate adjacent lane traffic


Lane Traffic

under saturated conditions. The next chapter will improve the model developed in this

chapter to address the impact of degree of saturation of the adjacent lane and the bus

stop itself.

Chapter 7: Maximum Working Capacity of an On-Street, Mid-Block, Off-Line Bus Stop 85

Maximum Working Capacity of

an On-Street, Mid-Block, Off-Line Bus Stop

Overview

The contents of this chapter are based on the manuscript, “Incorporating practical

degree of saturation in maximum working capacity estimation of on-street, mid-block,

off-line bus stops (Hisham et al., 2019b)” submitted to Transportation Research

Record.

Previous chapters in this thesis provide a novel, deterministic limit state capacity

estimation model for an on-street, mid-block, off-line (OS-MID-OFF) bus stop.

However, due to the stochastic nature of bus stop operation, the maximum working

capacity of a bus stop at a given adjacent lane general traffic flow rate is necessarily

less than its limit state value. This chapter provides an improved understanding of OS-

MID-OFF bus stop operation with respect to its maximum working capacity. In order

to do so, the BCAL model is further improved to analyse the performance of an OS-

MID-OFF bus stop with respect to practical degrees of saturation of the bus stop and

the adjacent lane that ensure delays are moderated. This fulfils research objective 5.

Refined Definition of Bus Stop Failure

According to the TCQSM model of capacity estimation, an operating margin is added

to the loading area processing time to give the bus a reasonable time to accommodate

any irregularities in the dwell time. This is an additional time added to the dwell time,

which is also the maximum amount of time a bus can dwell on a loading area without

creating a ‘bus stop failure’. A failure is defined by the TCQSM as a situation that

arises when a bus arrives to use a loading area only to find another bus is still

occupying it (Kittelson and Associates, 2013a).

However, given the fundamental theory behind the TCQSM model, we define failure

more accurately as a situation that arises when two buses arrive to use the loading area

86 Chapter 7: Maximum Working Capacity of an On-Street, Mid-Block, Off-Line Bus Stop

consecutively at a headway equal to the inverse of the specified capacity of the loading

area, but the first bus dwells on the loading area for a duration longer than the average

dwell time plus a specified operating margin, requiring the second bus to wait until the

first bus has cleared the loading area. Importantly, from this more accurate definition

it may be deduced that the current TCQSM model presumes that any further failure is

fully attenuated once the second bus reaches the loading area. This presumption,

however, is incorrect unless a third bus arrives at a headway equal to or less than the

inverse of the specified capacity of the loading area minus the excessive dwell time of

the first bus. Consequently, the TCQSM model does not allow for accumulation of

delay due to successive buses arriving unevenly. This was addressed by Bunker

(2018), who developed a model to estimate average delay to buses upstream of a

loading area.

Because the TCQSM model includes failure rate as a combination of dwell time and

dwell time variability, by assuming that dwell times are distributed normally, the

operating margin on dwell time is calculated by assigning a standard normal variable

corresponding to a desired failure rate and multiplying it by mean dwell time and

estimated coefficient of variation of dwell time. Under the TCQSM model, addition of

the operating margin on dwell time to the mean dwell time achieves the design dwell

time, which is then used in determination of a loading area design capacity that reflects

a desired level of operational reliability.

TCQSM recommends design failure rates between 7.5% and 15% for downtown areas

and 2.5% for outside downtown areas (Kittelson and Associates, 2013a). However, it

also mentions that design capacity is maximized when the failure rate is set to 25%.

Following the logic of the clarification made above about failure, the application of

operating margin and associated failure rate can be feasible when bus arrivals are

relatively evenly spaced. However, as bus arrival flow rate and adjacent lane general

traffic flow rate increases, buses are more susceptible to unevenly spaced arrivals,

which would increase interference between buses and would lead to increased failures

(Hisham et al., 2019a). With respect to OSB operations, interference to general traffic

will also yield a higher sensitivity to loading area failures. Therefore ‘failure’ is an

attribute that could occur not only with the dwell time, but also with both interference

and clearance between buses with respect to the loading areas of a bus stop, which are

its servers. Therefore, to overcome this issue, this chapter will consider ‘failure’ with


respect to all contributing factors. In addition, this chapter will also improve the model

developed to estimate the maximum working bus stop capacity, which ensures that

sufficient capacity is available for adjacent lane traffic and that delays are kept to

acceptable levels.

Degree of Saturation at a Bus Stop

We defined degree of saturation of a bus stop as the quotient of the demand bus flow

rate and the theoretical, limit state capacity during a given period of time. Fernandez

and Planzer (2002) identified that bus stop degree of saturation is an important measure

in bus stop capacity estimation. The degree of saturation of the bus stop indicates how

busy it is. This information can be used in designing bus stops to estimate a suitable

combination of bus flow and passenger demand. However, their estimation was

particular to exclusive bus lanes. It would be useful to have a way of estimating bus

stop capacity to ensure that a bus stop can be designed to accommodate the required

number of buses. This number can be chosen such that it will ensure an adequate

service level in the stop area.

Degree of saturation of the adjacent lane also has an impact on bus stop capacity.

TCQSM allows for high traffic flow rate in the adjacent lane for re-entry delay in the

capacity estimation. As was discussed in Chapter 3, because it is a one way model that

outputs bus stop capacity only, it does not address the capacity of the adjacent lane for

a particular bus stop capacity.

According to Akcelik (1981), when degree of saturation exceeds 0.9 on an arterial

road, it is typically considered as suffering from congestion, of which excessive delay

is a manifestation. Thus, the application of this practical degree of saturation means

that the working capacity of a traffic lane is marginally less than its saturation flow

rate.

Degree of saturation of the bus stop directly affects approach delay of the buses and

therefore queuing. Therefore, in this chapter, we identify degree of saturation of the

adjacent lane and the degree of saturation of the bus stop as important parameters that

are direct measures of operating conditions experienced by buses (and their

passengers) using the stop and by the adjacent lane general traffic. The following


section will provide a methodical approach to quantify these degrees of saturation in

order to incorporate them into the deterministic model to determine maximum working

capacity of an OS-MID-OFF bus stop.

BCAL Model Improvement Considering

Maximum Working Capacity

This section presents the development of a model that modifies the ‘Bus Stop Capacity

with Adjacent Lane traffic requirements (BCAL)’ model developed by Hisham et al.

(2019a) by considering bus stop and adjacent general traffic lane degrees of saturation.

The model is called the ‘Bus Stop Maximum Working Capacity with Adjacent Lane

Traffic’ (BMWCA) model.

The BMWCA model considers loading area operation as being the fundamental

building block of stop operation. The loading area average total processing time per

bus, may be considered as the sum of fundamental components as follows:

Equation 7-1

Where

= Average loading area total processing time per bus (s)

= Start-up time (s)


= Dwell time (s)

= Time taken due to bus-bus interference (s)

= Processing margin on total loading area processing time (s)

The processing margin in Equation 7-1 differs from the operating margin according to

the TCQSM model, in that it applies to all of the four listed components of average

loading area processing time per bus, and not just the dwell time.

It follows that the loading area average total processing time per bus net of processing

margin is given by:


Equation 7-2

Where

= Net loading area total processing time (s)

The BCAL model of Hisham et al. (2019a) implies a maximum feasible degree of

saturation of the bus stop itself as 1.0, should a value of zero be assigned for the

operating margin on dwell time. This would also be the case for the TCQSM model.

Similarly, the theoretical model developed so far implies a maximum feasible degree

of saturation of the bus stop itself as 1.0, which would correspond to a margin on total

loading area processing time per bus of zero.

In the BMWCA we replace operating margin with the processing margin on loading

area average total processing time per bus, as follows:

Equation 7-3

Where

= Processing margin on total loading area processing time (s)

= Average loading area total processing time (s)

= Maximum working degree of saturation of the loading area

Equation 7-3 ensures that on average, the loading area remains idle for a portion of

total loading area processing time per bus, which is equal to the additive inverse of the

designated loading area maximum working degree of saturation, . From,

Equation 7-1 and Equation 7-3 the processing margin can therefore by restated as:

Equation 7-4

The time available for the adjacent lane to pass the bus stop during the average total

loading area processing time per bus is equal to the sum of the time components of

average total loading area processing time per bus, during which the bus does not

obstruct the adjacent lane, and is given by:

Equation 7-5

Where


= Average time taken by adjacent lane to proceed through the bus stop (s)

As with the BCAL model, an important difference between the BMWCA model and

the TCQSM (Kittelson and Associates, 2013a) model is that we acknowledge that

general traffic in the adjacent lane to a given loading area has a theoretical capacity

(veh/h), which was defined by Hisham et al. (2019a) as:

Equation 7-6

Where

= Theoretical capacity of the adjacent lane (veh/h)

= Saturation flow rate of the adjacent lane (veh/h)

The degree of saturation of general traffic in the adjacent lane is given by:

Equation 7-7

Where

= Degree of saturation of the adjacent lane (veh/h)

= General traffic flow rate of the adjacent lane (veh/h)

Equation 7-1 requires that bus re-entry delay be quantified. As with the BCAL model,

the BMWCA model maintains the gap acceptance approach according to Equation 2-8

and Equation 2-9. However, we acknowledge that the adjacent lane traffic is obstructed

during start-up time and bus-bus interference time. Therefore, the merging bus driver

will see a compressed traffic stream in the adjacent lane passing by the loading area

during other times. For purposes of estimating re-entry delay due to gap acceptance,

Hisham et al. (2019a) made the following adjustment to adjacent lane traffic flow rate

for bus re-entry gap acceptance:

Equation 7-8


From

Equation 7-6 and

Equation 7-7, the adjusted adjacent lane traffic flow rate is given by:

Equation 7-9

Under this model, this adjusted value is applied to Equation 2-8 and Equation 2-9 to

calculate re-entry delay, .

The interference between buses at a bus stop may be reflected by a bus-bus interference

factor as follows (Hisham et al., 2019a):

Equation 7-10

Where is the number of effective loading areas according to TCQSM (Kittelson

and Associates, 2013a) values and is the number of actual loading areas.

The additional time component towards average total processing time per bus due to

bus-bus interference (s/bus) can be estimated as a margin upon the sum of the time

components of loading area average processing time per bus, excluding processing

margin, from Equation 4-12 is as follows:

Equation 7-11

The system of Equation 7-1 through Equation 7-11 allows us to determine the loading

area average total processing time per bus, and the time available for the adjacent lane

to pass the bus stop during the average total loading area processing time per bus,

provided that all inputs are known. The start-up time and average dwell time are

typically inputs to analysis. However, according to Equation 7-9, knowledge is

required of the adjacent lane degree of saturation for determination of bus re-entry

delay. The additional time component towards average total processing time per bus

due to bus-bus interference (s/bus) can then be determined directly using Equation

7-11. Further, according to Equation 7-4, calculation of the processing margin requires

knowledge of the loading area maximum working degree of saturation. Therefore,


these two unknown degrees of saturation must be resolved in order to solve the whole

system of equations.

This must be undertaken in two steps. First, the optimal adjacent lane flow rate must

be determined. It is the highest adjacent lane flow rate at which the bus stop’s loading

areas (assuming equal utilization) operate at their common practical degree of

saturation, and the point at which the adjacent lane reaches its practical degree of

saturation, . We define practical degree of saturation as the greatest value that

yields an acceptable delay, and therefore uncongested operation.

It may be proven mathematically that optimal adjacent lane flow rate is determined

directly by:

Equation 7-12

Where re-entry delay, is calculated using adjusted adjacent lane flow rate,

. Here and are both specified directly, as is discussed in the next

section.

Second, the adjacent lane degree of saturation and loading area maximum working

degree of saturation are determined. Here, we need to separate into two regimes,

by comparing it with the optimal flow rate, Regime 1 occurs when

, Regime 2 occurs when .

Adjacent lane degree of saturation is then calculated under each regime as follows:

Equation 7-13

In Regime 1, needs to be estimated as a function of re-entry delay and average

total processing time per bus net of processing margin. To calculate these two

components of processing time, the adjusted adjacent lane flow rate is used, which is

recursively a function of adjacent lane degree of saturation. Therefore, it is necessary


to apply the following objective function to determine the adjacent lane degree of

saturation for the given flow rate:

Equation 7-14

Where we recommend an initial trial value for the objective function,

.

Loading area maximum working degree of saturation is then calculated dependent

upon adjacent lane flow rate regime as follows:

Equation 7-15

Once the loading area maximum working degree of saturation is determined for the

correct regime, the stop maximum working bus capacity may be determined as

follows:

Equation 7-16

Specification of Practical Degrees of Saturation

at an OS-MID-OFF Bus Stop

In the BMWCA model, Equation 7-12 to Equation 7-16, requires specification of

loading area practical degree of saturation, and adjacent lane practical degree of

saturation, in order to ensure that delays do not become unacceptably high. These

are required in order to calculate the following; optimal adjacent lane flow rate,

adjacent lane degree of saturation when adjacent lane flow rate is less than optimal,

and loading area maximum working degree of saturation when adjacent lane flow rate

is greater than optimal.


Adjacent lane practical degree of saturation is considered first. If we specify this value

to be constant across all adjacent lane flow rates, then according to the theory above,

when adjacent lane flow rate is less than optimal adjacent lane flow rate, adjacent lane

degree of saturation will be less than practical degree of saturation. When adjacent

lane flow rate is equal to or greater than optimal adjacent lane flow rate, adjacent lane

degree of saturation will be equal to practical degree of saturation.

For the common constraint point of a signalised intersection on an arterial road, the

typical default recommended practical degree of saturation is equal to 0.9 (Akçelik,

1980). We suggest that this value of practical degree of saturation is also applicable to

the case of adjacent lane traffic passing a mid-block, off-line bus stop on an arterial

road.

The loading area practical degree of saturation is considered second. According to the

theory above, in regime 1 where adjacent lane flow rate is less than or equal to optimal

adjacent lane flow rate, loading area maximum working degree of saturation is equal

to practical degree of saturation. In regime 2 where adjacent lane flow rate is greater

than optimal adjacent lane flow rate, loading area maximum working degree of

saturation will decline from practical degree of saturation, to zero at the point where

the adjacent lane becomes practically saturated, being at .

Along with its value, the assumption about whether loading area practical degree of

saturation should remain constant with adjacent lane flow rate in regimes 1 and 2

requires careful consideration, as follows.

Bunker (2018) discussed that the processing of buses through a loading area of a bus

stop has similar characteristics to operation of an unsignalised intersection. However,

the loading area as a server is subject to less fluctuation than the head of the queue on

a minor movement at an unsignalised intersection. The increase in upstream average

waiting time (delay) with degree of saturation were noted by Bunker (2018) as being

less pronounced. However, the waiting time upstream of a loading area at a bus stop

is more consequential, because of the effects of bus queuing upon bus stop and adjacent

lane operation. Bunker (2018) developed the following equation to estimate upstream

average waiting time:


Equation 7-17

where system time, .

Bunker (2018) discussed that Equation 7-17 is scalable. Where there are multiple

loading areas and assuming that all loading areas at the bus stop are equally utilized,

the estimate of upstream average waiting time applies to the whole bus stop. It is

important to note that in Equation 7-17, for a given adjacent lane flow rate, will

be constant regardless of loading area degree of saturation.

In regime 1, adjacent lane flow rate is less than or equal to optimal adjacent lane flow

rate. Equation 7-17 may be rearranged to determine an appropriate loading area

practical degree of saturation for a specified practical upstream average waiting time:

Equation 7-18

where is a specified practical upstream average waiting time and

corresponds to a given adjacent lane flow rate.

Loading area practical degree of saturation should not cause excessive upstream

average waiting time, particularly as adjacent lane flow rate approaches optimal, which

corresponds to adjacent lane practical degree of saturation. Bus drivers arriving at the

stop to access a loading area should be able to do so within a time associated with the

mechanical and geometric properties of the buses alone, and not components affected

by demand fluctuation including dwell time, re-entry delay, and bus-bus interference

time. Therefore, we recommend that to determine optimal adjacent lane flow rate and

associated loading area practical degree of saturation, practical upstream average

waiting time is limited to a value equal to the start-up time between buses.

Equation 7-13 enables loading area practical degree of saturation to be calculated in

regime 2. Equation 7-18 enables loading area upstream average waiting time to be


calculated. Mathematically, these values will both be less than the respective values at

optimal adjacent lane flow rate.

In regime 1, where adjacent lane flow rate is less or equal to the optimal value,

Equation 7-15, Equation 7-14, Equation 7-18 and Equation 7-15 may be solved

recursively to determine both loading area practical degree of saturation for a specified

upstream average waiting time and adjacent lane practical degree of saturation.

Mathematically, loading area practical degree of saturation will increase slightly as

adjacent lane flow rate increases.

Figure 7-1 illustrates the routine to estimate maximum working capacity bus stop of

an OS-MID-OFF bus stop by way of a flowchart. This covers the method that was

discussed in this section starting from Equation 7-1 through Equation 7-18. Based on

the theoretical methodology presented in the previous section, the routine to estimate

maximum working capacity of an OS-MID-OFF bus stop is presented below.


BMWCA Routine to Estimate Maximum

Working Capacity for On-street, Off-line, Mid-Block

Bus Stops

,

,

Y

Y

N

N

,

Input

Estimate

Assign

Figure 7-1: Flow chart for bus stop maximum working capacity estimation using BMWCA model


Comparison between TCQSM model and

BMWCA Model

To compare the two models, the maximum working capacity of a stylised on-street,

mid-block, off-line (OS-MID-OFF) bus stop is determined using each of the TCQSM

model based on Equation 2-14 and the BMWCA model of Equation 7-1 through

Equation 7-16, under conditions where adjacent lane general traffic has absolute

priority over re-entering buses. A mean dwell time of 20s is used throughout this study

to reflect typical, busy bus stop operation. The start-up component of clearance time

is assigned to be 10s for a standard bus (Levinson, 1997). Re-entry delay is estimated

using TCQSM default values of 7.0s for critical headway and 3.3s for follow-up

headway (Kittelson and Associates, 2013a). The bus stop is designated to contain two

actual loading areas. A value of 1.85 effective loading areas is assigned according to

the TCQSM model.

To estimate bus stop maximum working capacity and degree of saturation across a full

range of adjacent lane flow rates using the BMWCA model, optimal adjacent lane flow

rate must first be determined: The corresponding adjusted adjacent lane flow rate,

veh/h. Using Equation 2-8 and Equation 2-9 the corresponding

re-entry delay, s. Using Equation 7-11 the bus-bus interference time,

s. Using Equation 7-2 loading area average total processing time per bus net of

processing margin, s. Using Equation 7-18 with a specified practical

upstream average waiting time of 10s, loading area practical degree of saturation,

. Finally, using Equation 7-16 the optimal adjacent lane flow rate,

veh/h.

The routine illustrated in the flowchart of Figure 7-1 is then used to determine the bus

stop’s maximum working capacity and degree of saturation, as well as adjacent lane

degree of saturation, all of which satisfy their practical limits, across a range of

adjacent lane flow rate, , between 0 veh/h and 1,620veh/h. Figure 7-2 illustrates

the relationship between the stylized bus stop’s maximum working capacity and

adjacent lane flow rate using the BMWCA model along with the TCQSM model

results for comparison.


Figure 7-2: Comparison of bus stop maximum working capacity vs adjacent lane flow rate

The three marked curves in Figure 7-2 correspond to the BMWCA model; the red

curve corresponding to a scenario with a recommended upstream average waiting time

of 10s equal to the start-up time between buses. The two regimes discussed in the

above sections are apparent in each of the BMWCA curves, corresponding to each

practical upstream average waiting time. The left side of the curve represents the

regime where . When there is no adjacent lane flow, the upstream queue

will comprise of buses only, in which the greatest value of maximum working capacity

of the bus stop can be achieved. Depending on the assigned upstream average waiting

time, the loading area degree of saturation will be constrained. For instance, where

there is no traffic flow in the adjacent lane, to limit the maximum waiting time to 10s,

the loading area will be able to operate at a maximum working degree of saturation

equal to 0.32 while the adjacent lane will operate at a degree of saturation equal to 0.

As adjacent lane flow rate increases, both bus re-entry time and bus-bus interference

time increase, leading to an increase in the loading area average total processing time

per bus, net of processing margin. According to Equation 7-18 when the assigned

upstream average wait time is held constant, as adjacent lane flow rate increases, the


increase in the processing margin will result in a decrease in the loading area maximum

working degree of saturation that can be afforded. This in turn corresponds to an

increase in required processing margin relative to loading area average total processing

time per bus. As a consequence of these phenomena, loading area average total

processing time per bus gradually increases with adjacent lane flow rate, resulting in a

gradual reduction in bus stop maximum working capacity.

The right-hand side of the curves represents regime 2 where , such that the

adjacent lane is operating at practical degree of saturation. In Equation (22) the only

variable parameter in this regime is adjacent lane flow rate. According to this equation,

loading area maximum working degree of saturation reduces linearly to a value of zero

when adjacent lane flow rate reaches its theoretical maximum value of

. The amount of time available to accommodate the processing of any buses

on the loading area tends towards zero necessarily.

The marked blue dotted curve and the marked green dashed curve were plotted to

represent cases having average upstream waiting times equal to 20s and 30s

respectively. If the analyst considered an upstream wait time of 30s to be tolerable for

the bus stop configuration, almost twice the bus stop maximum working capacity could

be achieved than that with the recommended 10s average upstream waiting time.

Although higher average upstream waiting times can have high outputs in terms of bus

stop working capacity, due to increased travel time it can result in a worsening of

quality of service both for passengers within the bus and for passengers waiting to

board. High wait times can also lead to excessive queue lengths for OSB operation,

which can impact upon the adjacent lane general traffic through delay.

Failure rate of the TCQSM follows a cumulative normal distribution curve. Failure

rates relevant to areas outside downtown with off-line bus stops are recommended to

have values in the range between 2.5% and 7.5% (St. Jacques and Levinson, 1997).

This is to reflect the blocking that may arise in the traffic lane when a bus failure

occurs. Therefore, we considered cases with 5% and 2.5% failure rates, which are

shown by the orange dashed curve and the grey dotted curve respectively in Figure

7-2.


Interestingly, the BMWCA curve with 30s upstream waiting time aligns in-between

the TCQSM curve with 5% failure rate and 2.5% failure rate. Even though the values

may seem relatively similar when there is no adjacent lane traffic, they were estimated

using different assumptions. One of the differences is that the TCQSM model only

relates the stochasticity of the dwell time, by a setting a design value for its failure.

This sets a probability that variability in dwell times will not affect interference

between buses and upstream bus queue. By contrast, the BMWCA model allocates a

processing margin to the loading area processing time to encompass the stochasticity

amongst the entire bus stop process. For example, stochasticity of the bus arrival

patterns will affect the interference between buses. This may cause additional queuing

upstream of the bus stop, which will not be reflected if the failure rate approach was

used alone.

The differences in the trends between the BMWCA and TCQSM models are more

evident as adjacent lane flow rate increases. While adjacent lane flow rates on an urban

arterial road would not normally exceed 900veh/h, the BMWCA model yields

substantially lower bus stop maximum working capacity than the TCQSM model for

any combination of upstream average waiting time and failure rate shown. This is

directly a consequence of two inclusions in the BMWCA model. One is the setting of

a practical upstream average waiting time that limits the loading area practical degree

of saturation. The other is the two-way capacity effect incorporated into the BMWCA

model, which is not incorporated into the TCQSM model, which limits the loading

area maximum working degree of saturation, once the adjacent lane reaches practical

saturation. For adjacent lane flow rates in the generally feasible range below 900veh/h,

comparison between the BMWCA and TCQSM models still shows that if the analyst

were to limit upstream average waiting time to 10s, the bus stop maximum working

capacity would only be half that of the TCQSM capacity estimate. This example

demonstrates the strong need for setting policy about acceptable upstream average

waiting time.

Summary

This chapter presented a deterministic model to estimate maximum working capacity

of an OS-MID-OFF bus stop by incorporating measures of adjacent lane degree of


saturation and bus stop maximum working degree of saturation, and developing

criteria for their practical limits. This is a novel contribution to bus stop capacity

estimation and provides clearer insights about operation of this type of stop on an on-

street bus facility. In the development of the Bus Stop Maximum Working Capacity

with Adjacent Lane Traffic (BMWCA) model, characteristics of the operation of the

bus stop were considered to be similar to those of an unsignalised intersection.

Practical degrees of saturation were introduced for the adjacent lane, and for the

loading areas collectively, with the intent of maintaining upper bounds to delays at this

location along an on-street bus facility. In order to do so it was essential to introduce

the quantity of ‘processing margin’ for the bus stop, which ensures that sufficient slack

time is available on the loading areas to accommodate stochasticity within all

components of the processing of buses on the stop’s loading areas, in addition to arrival

headways.

The BMWCA model is more complex than the TCQSM model because it uses a

recursive algorithm to obtain a suitable value for the adjacent lane degree of saturation.

The results obtained by the BMWCA model show that upstream average waiting time

is a crucial parameter in capacity estimation and QOS because it directly affects

passengers both in the buses and waiting to board buses. While greater upstream

average waiting times result in high bus stop capacities, the analyst must make a policy

decision to specify a suitable value, such that the desired operational efficiency is

achieved. We recommended a value equivalent to the start-up time between buses on

a loading area.

This chapter successfully addressed research questions 1, 2 and 4 by developing a

deterministic model to properly understand the maximum working performance of an

OS-MID-OFF bus stop with consideration to practical limits of adjacent lane general

traffic and bus arrival flows. The next chapter will demonstrate the use of the

developed BMWCA model conducting a parametric study of all the parameters

affecting the performance of a bus stop.

Chapter 8: Parametric Study of Bus Stop Maximum Working Capacity with Adjacent Lane Traffic Model 103

Parametric Study of Bus Stop

Maximum Working Capacity with Adjacent

Lane Traffic Model

Overview

This chapter provides a study of the parameters of on-street, mid-block, off-line (OS-

MID-OFF) bus operation and their influence on bus stop capacity according to the Bus

Stop Maximum Working Capacity with Adjacent Lane Traffic (BMWCA) model

developed in the previous chapter of this thesis.

Figure 8-1 provides a basic outline of the parameters that influence the performance

of an OS-MID-OFF bus stop, which have been identified in this thesis. The

connections made between each parameter show their direction of influence. This

chapter will analyse these parameters in detail to examine and evaluate the operational

performance of an OS-MID-OFF bus stop. This chapter fulfils research objective 6.

Figure 8-1: Basic overview of parameters that influence the performance of an OS-MID-

OFF bus stop according to BMWCA model of Chapter 7

Number of loading areas

Loading area DOS

Upstream waiting

time

Loading area processing time

Processing margin

Bus-bus interference

time

Re-entry delay

Bus stop maximum working capacity

Dwell time

104 Chapter 8: Parametric Study of Bus Stop Maximum Working Capacity with Adjacent Lane Traffic

Model

Dwell time is a primary parameter that contributes towards to the loading area

processing time (Jaiswal et al., 2010a, Kittelson and Associates, 2013b). The influence

of dwell time on maximum working capacity with respect to the BMWCA model will

be discussed in Section 8.2. According the BMWCA model, upstream average waiting

time directly affects the loading area degree of saturation, which affects the processing

margin of the loading area. Detailed discussion on the influence of the upstream

average waiting time on bus stop capacity will be discussed in Section 8.3 along with

its significance in policy making. The influence of number of loading areas on

maximum working capacity will be discussed in section 8.4. Furthermore, the chapter

will demonstrate the applicability of the yield-to-bus rule through the BMWCA model

in section 8.5. The chapter concludes with a summary in section 8.6. Outcomes of this

analysis will be used to provide conclusions, future research and recommendations.

Influence of Dwell Time on Bus Stop Maximum

Working Capacity

Dwell time is the time spent by a bus at a stop to serve passengers. It is the summation

of door opening and closing time, passenger service time and the boarding lost time

(Kittelson and Associates, 2013a). It is normally considered as an average value during

the time period of interest. Detailed discussion on the theoretical estimation of average

dwell time was given in Section 2.3. In this section we discuss the impact of dwell

time on the maximum working capacity through the newly developed BMWCA

model.

To compare between various operational scenarios we consider a stylised OS-MID-

OFF bus stop with two loading areas similar to that of Section 7.7. The start-up

component of clearance time is assigned to be 10s for a standard bus (Levinson, 1997).

Re-entry delay is estimated using TCQSM default values of 7.0s for critical headway

and 3.3s for follow-up headway (Kittelson and Associates, 2013a).

Next, a suitable value for the upstream average waiting time must be prescribed.

Ideally, buses arriving at the bus stop should be able to access the loading area soon

after the previous bus pulls out of the loading area. As was discussed in Chapter 7,

waiting time should correspond to the time that the previous driver take to pull out of


the loading area alone, rather than other components such as interference with other

buses or the dwell time itself. Therefore, we set this value to reflect the geometric

properties of the bus which will approximately be equal to 10s.

Equation 7-1 through Equation 7-16 are used to estimate bus stop maximum working

capacity for dwell times varying within the feasible range for an on-street bus stop

between 10s and 60s, as presented in Figure 8-2.

Figure 8-2: OS-MID-OFF bus stop maximum working as a function of average dwell time

(two loading areas, 10s upstream average waiting time) according to BMWCA model

The smallest average dwell time specified was 10s. Based on Figure 8-2, average dwell

time does not have a linear effect on bus stop maximum working capacity under the

BMWCA model. Its effect on capacity reduction becomes less significant as average

dwell time increases. For a typical adjacent lane general traffic flow rate on an arterial

road of 600veh/h, a 10s average dwell time will yield a bus stop maximum working

capacity of 92bus/h, while a 20s average dwell time will yield a bus stop maximum

working capacity of 55bus/h (more than half), and a 60s average dwell time will yield

0.0,0.41

0.18,0.39

0.36,0.35

0.54,0.31

0.72,0.26

0.90,0.21

0.0,.320.18,0.30

0.35,0.28

0.53,0.56

0.70,0.22

0.90,0.17

0.0,0.260.17,0.25

0.35,0.24

0.52,0.22

0.69,0.19

0.90,0.16

0.0,0.22 0.17,0.210.34,0.20

0.52,0.180.69,1.17

0.90,0.14

0.0,0.19 0.17,0.19 0.35,0.18

0.0,0.17 0.17,0.17 0.35,0.16 0.53,0.15 0.70,0.140.90,0.12

0

20

40

60

80

100

120

140

160

0 300 600 900 1200 1500 1800

Bus s

top

max

imum

wor

king

cap

acity

(bus

/h)

Adjacent lane flow rate (veh/h)

td=10s td=20s td=30s td=40s td=50s td=60s


Model

a bus stop maximum working capacity of 17bus/h (more than one sixth). In the typical

range of adjacent lane general traffic flow rate on an arterial road of less than 900veh/h,

bus stop maximum working capacity is therefore most sensitive to fluctuation in

average dwell time for smaller average dwell times. These are the more common dwell

times for such a facility so care needs to be taken in their estimation when estimating

capacity.

According to the figure above, bus stops operating at small average dwell times are

highly sensitive to increasing degree of saturation of the adjacent lane. When an

adjacent lane operates approximately at degree of saturation of between 0.17 and 0.18,

a 10s average dwell time will yield 85% of the capacity of 0.0 degree of saturation, a

20s average dwell time will return 90% of the capacity of 0.0 degree of saturation and

a 60s average dwell time will yield 95% of the capacity of 0.0 degree of saturation.

Therefore increase in the adjacent lane traffic flow rate has less significance in capacity

reduction for bus stops with large average dwell times.

Influence of Upstream Average Waiting Time on

Bus Stop Maximum Working Capacity

It was identified in Chapter 7 that that upstream average waiting time is an important

parameter in capacity performance measure because it quantifies the delay experienced

by passengers even before the bus starts to process at the loading area. It also directly

affects queue length. Upstream average waiting time can be used to define bus stop

maximum working capacity and is also a measure of QOS.

Equation 7-18 in Chapter 7 provides the loading area practical degree of saturation as

a function of an assigned upstream average waiting time and loading area net

processing time per bus, for a system time of 1h. The threshold values can be set to

reflect the expected operational performance.

Equation 7-1 through Equation 7-16 are used to estimate bus stop maximum working

capacity for assigned upstream average waiting times of 10s, 20s and 30s for an

average dwell time of 20s and two loadings areas, as presented in Figure 8-2.


Figure 8-3: OS-MID-OFF bus stop maximum working as a function of upstream average waiting time (two loading areas, 20s average dwell time)

Based on Figure 8.3 upstream average waiting time does not have a linear effect on

bus stop maximum working capacity. The increment in bus stop capacity is less

significant as average upstream waiting time increases. For a typical general traffic

flow rate less than 900veh/h, adoption of a 20s upstream average waiting time will

result in about 50% more capacity than adoption of a 10s upstream average waiting

time, while adoption of a 30s average upstream waiting time will result in about 90%

more capacity than adoption of a 10s upstream average waiting time. It is apparent that

increase in the average wait time does not increase the bus stop maximum working

capacity proportionally.

According to Figure 8.3, loading area degree of saturation increases with increasing

upstream average waiting time. This is because, as the upstream waiting time

increases, the BMWSC model allows a higher bus inflow to the bus stop and result a

higher bus stop capacity. However, higher upstream average waiting times also result

in operational concerns. In these cases, the upstream queue may extend from the back

0.0,0.260.17,0.25

0.35,0.24

0.52,0.22

0.70,0.19

0.9,0.16

0.0,0.49019,0.47

0.37,0.44

0.54,0.41

0.72,0.36

0.90,0.30

0.0,0.59

0.19,0.58

0.38,0.55

0.57,0.51

0.74,0.45

0.90,0.39

0

20

40

60

80

100

120

140

160

0 300 600 900 1200 1500 1800

Bus s

top

max

imum

wor

king

capa

city

(bus

/h)

Adjacent lane flow rate (veh/h)tw=10s tw=20s tw=30s


Model

of the bus stop onto the adjacent general traffic lane, depending upon the length of the

off-line bus stop. Unlike an unsignalised intersection where queues can be

accommodated upstream of the stop line, at a bus stop there is limited space to store

buses. This may cause inconvenient delays to passengers on-board and also to

passengers waiting at the bus stop, and reduce general traffic capacity of the adjacent

lane. Interactions with upstream signalised intersection can also cause queue

spillbacks. This stresses the importance of setting policy regarding specification of a

limiting value of upstream average waiting time.

Influence of Number of Loading Areas on Stop

Capacity

A bus stop’s capacity depends on the capacities of its individual loading areas. It is

also dependent upon the traffic flow rate that interferes with a buses’ access to the

loading area. Generally, at a bus stop with multiple loading areas, underutilization of

front most loading area/s occurs. The diminishing effect on the bus stop capacity due

to the underutilization of the loading areas discussed below.

Figure 8-4 OS-MID-OFF bus stop maximum working as a function of number of loading

areas (10s upstream average waiting time, 20s average dwell time)

0.0,0.34 0.18,0.320.35,0.30

0.52,0.270.70,0.24

0.90,0.18

0.0,0.320.17,0.30

0.35,0.28

0.53,0.25

0.70,0.22

0.90,0.17

0.0,0.300.17,0.29

0.35,0.27

0.53,0.24

0.70,0.21

0.90,0.16

0

20

40

60

80

100

120

140

160

0 300 600 900 1200 1500 1800

Bus s

top

max

imum

wor

king

capa

city

(bus

/h)

Adajcent lane flow rate (veh/h) Nla=1 Nla=2 Nla=3


It is apparent from the above figure that higher bus stop maximum working capacities

can be achieved with increased number of loading areas. However, the effectiveness

of having multiple loading areas needs to be investigated further.

A bus stop with two loading areas produces bus capacities less than twice of the bus

stop having one loading area. Also, a bus stop with three loading area produces bus

capacities only little more than twice that of a bus stop with one loading area..

Therefore, by adding linear loading areas to an off-line stop does not proportionally

increase the expected capacity of a single loading area. For this reason, it is not

efficient to build bus stops with more than three loading areas (Fernández, 2010). One

way to alleviate this issue is to create divided bus stops with one or two loading areas,

which is related to skip-stop operation. Another method, where feasible, is to use an

upstream holding strategy, where buses would be grouped based on their routes. The

holding strategy can then release buses as platoons into the bus stop. If the bus stop

consists of three loading areas a platoon with three buses would be released into the

bus stop and occupy all three loading areas at the same time. This mechanism can

result in closer to 100% efficiencies for all three-loading area due to no upstream

waiting and interference between buses.

Yield-to-Bus Rule

Implementing yield-to-bus laws are considered as an operational advancement towards

an increased bus stop capacity (Zhao et al., 2018). For on-street bus (OSB) operations,

as adjacent lane traffic increases it becomes more difficult for the bus driver to find an

acceptable gap re-enter from the bus stop into the traffic lane Imposing yield-to-bus

rule (YTB) for off-line bus stops can minimize the impact of general traffic on bus

stops (King, 2003). To reduce bus delays and improve total travel time many countries

Europe and also United States, Canada and Australia already have enacted laws to

impose the YTB rule.

According to the TCQSM model, clearance time is the sum of the time taken by the

bus to travel its own length, the next bus to pull in and the re-entry delay (Kittelson

and Associates, 2013a). Re-entry delay component is the time taken for the adjacent


Model

lane general traffic queue formed by an adjacent traffic signal to clear plus the time

taken for the bus driver to find a suitable gap and re-enter into the adjacent lane. Start-

up time has a fixed value while re-entry can vary depending on the stop attributes. At

a mid-block bus stop, this re-entry delay reduces to gap acceptance delay only. The

method to estimate re-entry delay produces an estimate of maximum average delay

that could occur while waiting to enter the adjacent traffic lane and is estimated using

Equation 2-8 and Equation 2-9 which are re-stated below.

Equation 2-8

Equation 2-9

Where:


= Capacity of the re-entry movement (veh/h)

= Number of actual loading areas

= Demand flow rate in the kerb lane (veh/h)

= Critical headway of the re-entry movement = 7s

= Follow-up time for the re-entry movement = 3.3s

TCQSM (2013) suggests that in places that have adopted the YTB rule, the re-entry

delay can be minimized or eliminated depending on how well drivers comply with the

rule (Kittelson and Associates, 2013a). In this section we adjust the BMWCA model

to analyse the impact of YTB conditions.

The theory presented in Chapters 6 and 7 implied no YTB rule. In this parametric

study, to compare between YTB conditions, a scenario of an OS-MID-OFF bus stop

is considered similar to the case discussed in Section 7.7. The stop consists of two

loading areas, with adjacent lane general traffic. A mean dwell time of 20s was used

throughout this chapter to reflect a typical bus stop operation. The start-up time was

set to 10s (Levinson, 1997) similar to the previous chapter and the upstream average

wait time for buses was set to 10s.


In a universal sense, YTB behaviour exists on a spectrum. At the lower end of the

spectrum, with no YTB rule, bus drivers yield absolute priority to adjacent lane general

traffic. TCQSM (Kittelson and Associates, 2013a) prescribe values for the critical

headway, which is the minimum headway between vehicles in the adjacent lane that

buses can use to re-enter to the traffic, and the follow-up time which is the headway

between two successive re-entering buses, as 7s and 3.3s respectively for the default

case where no YTB rule applies. At the upper end of the spectrum, adjacent lane

drivers may yield to re-entering buses unless they are positioned within the follow-up

headway of the bus at the bus driver’s re-entry decision time. This condition with the

YTB rule fully effective may be modelled by applying a bus-entry critical gap equal

to the follow-up headway of 3.3s. At the middle of the spectrum, adjacent lane drivers

may have limited priority over re-entering buses. This condition with the YTB rule

partially effective may be modelled by applying a critical gap that equals the bus

follow-up headway plus the saturation headway of adjacent lane general traffic, the

latter reflecting that the bus driver would not expect a following adjacent lane driver

to slow down as the bus re-enters. For a typical adjacent lane, maximum feasible flow

rate is equal to the product of saturation flow rate of 1,800veh/h and practical degree

of saturation of 0.9, the corresponding bus re-entry critical gap would equal 5.5s.

Based on the above critical headways and the follow-on times, the maximum working

capacity of the OS-MID-OFF bus stop is illustrated in Figure 8-5. The orange, solid

curve represents the case with no YTB rule. The blue, soft-dotted curve represents the

case with the YTB rule fully effective, while the green, hard-dashed curve represents

YTB rule partially effective. The differences in theoretical stop bus capacity are most

significant at the point where the adjacent lane general traffic stream reaches its

practical degree of saturation. Table 8-1 compares the maximum working capacity

between no YTB, partial YTB and Full YTB cases.


Model

Figure 8-5: Bus stop maximum working capacities vs adjacent lane flow rate with levels of

YTB rule

Table 8-1: Maximum working capacity comparison between level of YTB rule

0.19,0.30

0.35,0.28

0.53,0.26

0.70,0.22

0.87,0.18

0.90,0.17

0.35,0.30

0.53,0.28

0.70,0.26

0.90,0.17

0.0, 0.32 0.18,0.310.27,0.310.44,0.31

0.62,0.300.80,0.29

0.90,0.17

0

10

20

30

40

50

60

70

80

0 300 600 900 1200 1500 1800

Bus S

top

Max

imum

Wor

king

Cap

acity

(bus

/h)

Adjacent Lane Flow Rate (veh/h)

No YTB Partial YTB Full YTB

Data labels-

Adjacent Lane Flow Rate

(veh/h)

Bus Stop Maximum Working Capacity (bus/h)

No YTB Partial YTB YTB

Partial YTB improvement over no YTB

YTB improvement over no YTB

0 71 71 71 - -

300 63 66 69 5% 10%

600 54 60 66 11% 22%

900 42 53 64 26% 52%

1200 30 45 60 50% 100%

1500 18 36 57 100% 217%


It is apparent from Table 8-1 that, for any adjacent lane flow rate, bus capacities with

any degree of YTB show improvements in bus stop capacities over the case with no

YTB rule.

The differences in the shape of decline in stop bus capacity with adjacent lane flow

rate in Figure 8-5 reflect the differences in efficiency of each level of YTB, which is a

direct effect of the critical headway of the adjacent lane. The full YTB curve shows

significantly higher values for bus stop capacities because buses are given priority

during the re-entering process. Bus drivers can re-enter even with a 3.3s of critical

headway and only headways less than 3.3s will be wasted during gap acceptance.

Under the partial YTB, buses accept all gaps in the adjacent lane except the gaps less

than 5.3s. Under no YTB case, all gaps in the adjacent lane less than 7s are wasted

during the gap acceptance process. Figure 8-6 shows the re-entry delay based on the

critical headway for each YTB condition.

Figure 8-6: Re-entry delay based on YTB conditions

Based on Figure 8-6, it is apparent that when the adjacent lane reaches practical degree

of saturation, re-entry delay can be as low as 4.3s for full YTB conditions or can be as

high as 37.3s for no YTB conditions. Depending on the degree of YTB rule

implemented near the bus stop, the re-entry delay can fluctuate between these values.

Xal=0.9

Xal=0.9

Xal=0.9

0

5

10

15

20

25

30

35

40

0 300 600 900 1200 1500 1800

Re-e

ntry

Del

ay (s

)

Adjacent Lane Flow Rate (veh/h)No YTB Partial YTB Full YTB


Model

Therefore, transport engineers and policy makers may choose the degree of YTB that

they wish to implement, such that the desired operational efficiency and reliability of

the bus system is achieved.

Once the adjacent lane reaches the optimal flow rate at 1,562veh/h corresponding to

the highest adjacent lane flow rate at which the bus stop’s loading areas operate at their

common practical degree of saturation and the point at which the adjacent lane reaches

its practical degree of saturation, the re-entry delay remains fixed according to Figure

8-6. However, the bus stop capacity drops more steeply after the adjacent lane having

reached the optimal flow rate (Figure 8-5). This can be explained using the ‘processing

margin’ parameter which was introduced in Chapter 7. Figure 8-7 shows the behaviour

of the processing margin with increasing adjacent lane flow rate.

Figure 8-7: Processing margin for YTB conditions

The processing margin ensures sufficient idle time exists at the loading area. In this

case the upstream average wait time was prescribed as 10s. Once the adjacent lane

reaches its optimal flow rate, the processing margin increases rapidly to ensure that

0

100

200

300

400

500

600

700

800

900

0 300 600 900 1200 1500 1800

Proc

essin

g M

argi

n(s)

Adjacent Lane Flow Rate (veh/h)No YTB Partial YTB Full YTB


sufficient space exist to accommodate the adjacent lane traffic. Hence the decrease in

the bus stop capacity for higher adjacent lane flow rates.

Summary

This chapter analysed the influence of average dwell time and average upstream

waiting times, and number of loading areas on the bus stop maximum working capacity

with the new BMWSC model. First section of the chapter investigated the importance

of dwell time in bus stop maximum working capacity. The analysis showed that bus

stop maximum working capacity is most sensitive to fluctuation for smaller average

dwell times. Therefore, accurate estimation of dwell time is very important in bus stop

capacity estimation.

Upstream average waiting time was identified to be one most important parameters in

analysing the operations of OS-MID-OFF bus stops and this chapter demonstrated its

influence on the bus stop maximum working capacity. The analysis showed that bus

stop maximum working capacity is most sensitive to fluctuation for smaller average

upstream average waiting times. On the other hand, bus stop capacity increased with

increasing upstream average waiting time. However, higher upstream average waiting

times could also result in operational concerns such as extended queue formation and

queue spill backs. Therefore this recommended a smaller value for the upstream

waiting time and demonstrated the importance of setting up a policy for a threshold

value for the upstream average waiting time.

Furthermore, this chapter analysed the number of loading areas and its influence on

the maximum working capacity. This section demonstrated that adding loading areas

to an OS-MID-OFF bus stop does not necessarily yield a fully proportional capacity

increase of a single loading area. This section also provided ways to improve this issue,

by creating multiple sub stops or by developing a holding strategy at the upstream

signalised intersection.

Finally, this chapter demonstrated that existing theories to estimate bus capacity at OS-

MID-OFF bus stops do not account for re-entry conditions under the yield-to-bus

(YTB) rule. This study demonstrated that, No YTB, partial YTB, and full YTB,

conditions may be modelled by adjusting re-entry critical gap between 7.0s, 5.5s, and


Model

3.3s respectively. It was found that reduction in re-entry delay can significantly

improve maximum working capacity of buses under a given adjacent lane traffic flow

rate. The analysis showed that, bus capacities with any degree of YTB showed

improvements in bus stop capacities over no YTB condition. However, whether the

YTB rule operates effectively in a given jurisdiction needs to be considered carefully,

including the impact from the perspective of general traffic.

This chapter demonstrated the applicability of the BMWCA through a parametric

analysis. However, the outcomes of the analysis relies on the case study characteristics.

This chapter successfully responded to research question 2 and 4 by demonstrating the

BMWCA model to understand and analyse the performance of an OS-MID-OFF bus

stop through a parametric study.

Chapter 9: Conclusions 117

Conclusions

Overview

This chapter concludes this thesis and summarizes the analyses, discussions and results

presented in past chapters. A brief thesis summary is provided in section 9.2, followed

by section 9.3, which discusses the contributions of this research to the existing

knowledge and practice. Section 9.4 discusses the implications of the research findings

for both theory and practice. Section 9.5 provides the conclusions of this research

work, while recommendations for future research is given in section 9.6. This fulfils

research objective 7.

Summary of the Thesis

This research presented the development of a deterministic model to analyse

performance of on-street, mid-block, off-line (OS-MID-OFF) bus stops more

effectively and accurately. This research has also identified various parameters that

affect bus stop capacity with respect to OSB operations. The thesis was divided into

three main phases, where the 1st phase was to develop the thesis statement, the 2nd

phase was to support the thesis statement and the 3rd phase was to explain the thesis

statement.

Develop Thesis Statement

Chapter 1 responded to research question 1. This chapter established the research

statement of this thesis and presented the research questions, research objectives and

the scope. An outline of the research was given towards the end of the chapter. This

chapter partially fulfilled research objective 1.

Chapter 2 responded to research questions 1 and 2. This chapter presented a

background for this research and discussed operational issues of on-street bus stops,

particularly located in the mid-block section. The chapter highlighted the limitations

of exiting methodologies with regards to estimating capacity and identified research

118 Chapter 9: Conclusions

gaps. TCQSM theory presented in the 3rd edition was identified to be the most

commonly used methodology in estimating bus stop capacity (Fernandez and Planzer,

2002). The review suggested that, existing methodologies for estimating capacity

cannot be used for OSB operations because of their complexity due to the general

traffic movements and was identified as the key research gap. In addition, this chapter

also pointed that, TCQSM definition for failure is problematic and the means of

prescribing failure rate has not been sufficiently studied. This chapter fulfilled research

objectives 1 and 2.

Chapter 3 formalized the research problem based on the gaps identified in chapter 2.

In particular, the chapter studied the operations of OS-MID-OFF bus stops and

identified various parameters that influence the performance of OS-MID-OFF bus

stops. When buses and general traffic share the same lane, additional time may be

required due to conflicts between general traffic. However, this condition is highly

dependent on the degree of saturation of the adjacent lane. Moreover, with high

adjacent lane traffic volume, it is expected to have upstream vehicular queues, which

will affect the reliability and QOS of the bus facility. This chapter identified that time

required to accommodate general traffic, degree of saturation of the adjacent traffic

lane and upstream waiting times for buses are some of the parameters that influence

the performance of OS-MID-OFF bus stops. A methodology was proposed and

presented to further investigate these influencing parameters to develop, support and

explain the thesis statement. This chapter fulfilled research objective 1.

Support Thesis Statement

Chapter 4 partially responded to research questions 1 and 3. This chapter considered

bus stops in general to develop an alternate deterministic model to the TCQSM model

for bus stop capacity estimation. In this chapter the TCQSM model was improved by

quantifying secondary influences such as traffic blockage, bus-bus interference and

adjacent signalised intersection to analyse their impacts on bus stop capacity. The

improved Modified Bus Stop Capacity (MBSC) model assigns the above mentioned

influences as additional time components of the loading area processing time per bus,

whereas the TCQSM model assigns these influences as capacity reduction factors at

the stop level. The effectiveness of the MBSC model was demonstrated through a case


study and showed that the model can be applied on its own to study bus stop operations

for a wide range of bus facility types. This chapter fulfilled research objective 2.

Chapter 5 responded to research question 2 and partially addressed research question

3. This chapter stated that it is difficult to collect field data for bus stop capacities with

a full range of general traffic flow rate in the adjacent lane because of its variation in

time. Also it is practically impossible to observe conditions close to saturation for a

long periods of time. Therefore, Aimsun microscopic simulation modelling was

chosen as a tool to affectively represent actually scenarios and reproduce their

behaviour under a controlled environment. A suitable microsimulation model was

developed for this research with adjacent lane requirements and cross-validated with

the MBSC model developed in Chapter 4. The simulation model was then used to test

scenarios across a range of bus and adjacent lane flow rates to observe the behaviour

of buses and adjacent lane traffic. Observations made through the simulation indicated

that TCQSM model does not provide a better understanding for higher adjacent lane

flow rates and the TCQSM model can be further improved by considering the adjacent

lane behaviour. This chapter fulfilled research objective 3.

Chapter 6 responded to research question 1, 2 and 3. Based on the observations made

through the simulation model, a theoretical model was developed for OS-MID-OFF

bus stops with adjacent lane requirements (BCAL). To incorporate the influence of the

adjacent flow rate, in this phase it was assumed that at an OS-MID-OFF bus stop,

buses will obstruct the adjacent lane traffic flow during a certain components of the

loading area processing time period, so additional time may be required to

accommodate adjacent lane traffic under saturated conditions. The BCAL model

enhanced the TCQCM model by ensuring that bus stop conditions do not cause the

adjacent lane general traffic to become over-saturated. The model accounts for

obstruction that can occur on the adjacent lane due to re-entering buses, and an

additional time required of the loading area processing time to accommodate adjacent

lane traffic under saturated conditions. The effectiveness of the BCAL model was

demonstrated through a case study application, and the study indicated that BCAL

model estimates bus stop capacity more accurately once the adjacent lane reached

saturation, whereas the TCQSM model cannot properly represent conditions over this

range. This chapter fulfilled research objective 4.


Chapter 7 responded to research questions 1, 2 and 4. This chapter presented novel

deterministic model to estimate maximum working capacity of an OS-MID-OFF bus

stop by incorporating measures of adjacent lane degree of saturation, upstream average

working time and bus stop maximum working degree of saturation, and developed

criteria for their practical limits. In the development of the Bus Stop Maximum

Working Capacity with Adjacent Lane Traffic (BMWCA) model, practical degrees of

saturation were introduced for the adjacent lane and the loading areas collectively, with

the intent of maintaining a desired level of operational reliability at this location along

an on-street bus facility. This chapter identified that TCQSM definition for ‘failure’

only attributed with the variability in dwell times. However, considering the stochastic

nature of the buses and increasing adjacent lane flow rates, failure can occur not only

with dwell time, but also with interference between buses which could be caused due

to unevenly spaced arrivals. With respect to OSB operations, interference to general

traffic will also yield a higher sensitivity to loading area failures. Therefore, in this

thesis ‘failure’ was defined with respect to all contributing factors. A quantity of

‘processing margin’ was introduced, to ensure sufficient slack time is available on the

loading areas to accommodate disruptive events within the process, such as bus-bus

interference, bus bunching and uneven upstream queuing. This chapter fulfilled


Explain Thesis Statement

Chapter 8 responded to research question 2 and 4. This chapter analysed the bus stop

maximum working capacity using the MBWCA model for various parameters. For

each case, a case example was used to demonstrate the application of BMWCA model

and its outcome.

The chapter demonstrated the influences of dwell time, upstream average waiting time,

number of loading areas through a parametric analysis. The analysis revealed that bus

stop maximum working capacity is highly sensitive to small dwell time values and

care should be taken in estimating dwell times in capacity estimation. Furthermore

upstream average waiting time is considered to be an important parameter in capacity

estimation. The study showed that assigning high values for upstream waiting time

returns a high bus stop capacity, however it could result in operational concerns at the

bus stop and adjacent traffic lane. Therefore, a low value for upstream average waiting


time was recommended. Moreover this chapter also demonstrated that adding loading

areas does not necessarily return proportionally higher bus stop capacities. Final

section of this chapter demonstrated the application of yield-to-bus (YTB) law through

the BMWCA model. The study suggested that, no YTB, partial YTB, and full YTB,

conditions may be modelled by adjusting critical gap in the re-entry estimation. It was

observed in the analysis that, any degree of YTB rule implemented near the bus stop,

showed improvements in bus stop working capacities over no YTB rule. However,

further investigations are recommended to determine whether the forced re-entry of

buses impact on general traffic movements in the adjacent lane. This chapter fulfilled


Theoretical Contributions of the Research

This research provided a detailed understanding of bus stop performance of OS-MID-

OFF bus stops and determined the influence of adjacent lane traffic flow rate on bus

stop maximum working capacity. The major contributions towards state-of-art of the

knowledge are listed below.

This research is primarily focused upon improving the TCQSM (Kittelson and

Associates, 2013a) methodology. This research first focused on a general approach in

developing an alternative deterministic model to the TCQSM model for bus stop

capacity estimation. The theoretical improvement carried out enhanced the TCQSM

model by quantifying the influence of traffic blockage, bus-bus interference and

adjacent signalised intersection. The improved Modified Bus Stop Capacity (MBSC)

model, quantifies the influence of traffic blockage, bus-bus interference and adjacent

signalized intersection as additional time components of the loading area total

processing time per bus, whereas the TCQSM model accounts for these influences by

way of factors at the stop level. (Hisham et al., 2018a). This is a significant contribution

because the improved model can be used to study a wide range of operations such as

a traditional on street operation or a bus rapid transit operation with improved

operational protocols, in terms of capacity (Chapter 4).

It was identified in this thesis that the TCQSM model does not provide accurate

understanding for on-street, mid-block, off-line (OS-MID-OFF) bus stops with

adjacent lane general traffic flow, especially where the adjacent lane reaches the point


of saturation (Hisham et al., 2018b). Furthermore, it was not possible to develop an

empirical model from real data because it is challenging to collect data for bus stop

capacity with varying adjacent lane traffic flow rates. Additionally, it is difficult to

observe situations with high adjacent lane traffic flow rates close to saturation.

Therefore, to observe the behaviour of the bus stop in terms of capacity, with varying

adjacent lane flow rates, an Aimsun microscopic simulation testbed was developed.

The key contribution from this research is the BMWCA model, a deterministic model

to analyse the performance of OS-MID-OFF bus stops incorporating adjacent lane

traffic requirements (Hisham et al., 2019a), degrees of saturation of the adjacent lane

and practical degree of saturation on the loading area due to specification of limit upon

upstream average waiting time of buses. The upstream average wait time was

incorporated into the model to maintain an acceptable operational reliability and was

identified as the one of the most important parameters because it is closely related to

passengers and hence QOS. The model results a maximum working bus capacity

which corresponds to the input upstream wait time and assigned degree of saturation

of the adjacent lane. This allows to maintain upstream average general traffic queue

within the region by limiting the upstream waiting time of the buses.

Analyses showed that maximum working capacity of BMWCA is relatively low than

that of the TCQSM bus capacity. This is because BMWCA considers the operational

reliability by limiting the upstream average waiting time. Moreover the model

considers the stochasticity at the bus stop by and allocating a ‘processing margin’.

These collectively increase the loading area processing time and results in a lower bus

stop capacity (Chapter 7 and Chapter 8).

Practical Contributions of the Thesis

The BMWCA model can be used to determine bus stop capacity based on the setting

of an upstream average waiting time that suits policy considerations. Upstream average

waiting time affects the QOS of the system greatly because it directly affects the

passengers both in the bus or waiting to board onto a bus, along with adjacent lane

general traffic should upstream bus queueing spill into the adjacent lane. Analyses

showed that increased waiting times can be used to achieve higher bus stop capacities.

However, this could in result operational concerns such as extreme queue formation


and queue spillback upstream of the bus stop. Therefore, a low value for the upstream

waiting time is recommended in this thesis. Nevertheless, this can be set by the transit

analyst, by determining a suitable value for the upstream average waiting time such

that it reflects the desired operational efficiency.

The BMWCA model categorizes the adjacent lane flow rates into two regimes based

on the optimal adjacent lane flow rate . Analyses showed that, in regime 2,

which is the range when the adjacent lane flow rate is equal or greater than the optimal

adjacent lane flow rate, bus stop operations becomes highly volatile. Generally, we

would not expect to see such high adjacent lane flow rates on urban arterial roads

because they would correspond to fully saturated conditions. However, BMWCA

model can be used to determine to avoid conditions in regime 2.

The BMWCA model can be used to test advanced operations such as, yield-to-bus

rule, advanced fare collection systems, etc., as described in the earlier chapters in this

thesis. BMWCA model assigns the influences including bus-bus interference and

adjacent lane general traffic as additional time components of the loading area

processing time per bus. This is a significant contribution to the practice because it

provides detailed information about all the components that affects the bus stop

capacity in terms of loading area processing time for traditional and non-traditional

practices.

The BMWCA model can be extended for application to capacity estimation of Bus

Rapid Transit (BRT) systems where non-stopping buses pass the subject station.

The BMWCA model can help the transit analyst in improving scheduling and in turn

could be used to address travel time reliability. This model can also be used to perform

robust capacity analysis for future bus stops or provide predictions for existing highly

congested bus facilities.

Recommendations and Future Research

The research has identified numerous future research directions as follow:

1. This research assumed bus dwell times were normally distributed. However it

is evident that some bus stops follow a log normal distribution


(Widanapathiranage, 2015). This study can be used to further investigate with

other types of bus arrival distributions.

2. This research can be further developed to analyse the performance of BRT

stations where stopping and non-stopping buses pass the critical station.

3. This research was conducted particularly for mid-block, offline bus stops.

Future research can be conducted on off-line bus stops located at signalized

intersections, mainly because these locations are common locations of critical

stops for OSB operations.

Concluding Remarks

This thesis found that understanding the parameters governing the bus stops is

absolutely essential to fully understand and analyse OS-MID-OFF bus operation. The

traditional approach to estimate bus stop capacity is found to be less reliable when an

adjacent lane operates with general traffic. Prior to this research no methodology was

known to exist to properly analyse on-street bus operations in terms of capacity.

A microscopic simulation model was developed using Aimsun and cross-validated

against TCQSM model, in order to observe the behaviour of an on-street bus operation

at saturated conditions. Results obtained from the Aimsun model was used to develop

‘Bus Stop Maximum Working Capacity with Adjacent Lane Traffic (BMWCA)’

model was developed to analyse the performance an on-street, off-line, mid-block bus

stops. The BMWCA model can be used in a broader spectrum. It can be used to test

various advanced operational practices and their impacts on bus stop capacity. Since

it provides a detailed analysis on the influencing factors, it is up to the planning

engineer to determine suitable values for these parameters such that the desired

operational efficiency of a bus operation is achieved.

References 127

References

AKCELIK, R. 1981. Traffic Signals: Capacity and Timing Analysis, ARRB Group.

AKÇELIK, R. 1980. Time-Dependent Expressions for Delay, Stop Rate and Queue Length at Traffic Signals.

ARRINGTON, G. & CERVERO, R. 2008. TCRP Report 128: Effects of TOD on Housing, Parking, and Travel. Transportation Research Board of the National Academies, Washington, DC, 3.

BARCELÓ, J. & CASAS, J. 2005. Dynamic Network Simulation with AIMSUN. In: KITAMURA, R. & KUWAHARA, M. (eds.) Simulation Approaches in Transportation Analysis: Recent Advances and Challenges. Boston, MA: Springer US.

BARTELS, A., RUCHATZ, T. & BROSIG, S. 2014. Intelligence in the Automobile of the Future. Smart Mobile In-Vehicle Systems. Springer.

BERGENHEM, C., HEDIN, E. & SKARIN, D. 2012a. Vehicle-to-Vehicle Communication for a Platooning System. Procedia-Social and Behavioral Sciences, 48, 1222-1233.

BERGENHEM, C., SHLADOVER, S., COELINGH, E., ENGLUND, C. & TSUGAWA, S. Overview of Platooning Systems. Proceedings of the 19th ITS World Congress, Oct 22-26, Vienna, Austria (2012), 2012b.

BIMBRAW, K. Autonomous Cars: Past, Present and Future a Review of the Developments in the Last Century, The Present Scenario and the Expected Future of Autonomous Vehicle Technology. Informatics in Control, Automation and Robotics (ICINCO), 2015 12th International Conference, 2015. IEEE, 191-198.

BRT, F. E. 2016. Segregated Busways or Bus-Only Roadways [Online]. Available: http://www.brtdata.net/ [Accessed].

BUNKER, J. M. 2018. High Volume Bus Stop Upstream Average Waiting Time for Working Capacity and Quality of Service. Public Transport, 10, 311-333.

CERVERO, R. 2013. Bus Rapid Transit (BRT): An Efficient and Competitive Mode of Public Transport. Institute of Urban and Regional Development, Calofornia, USA.

CHAUDHARY, N., ABBAS, M. & CHARARA, H. 2006. Development and Field Testing of Platoon Identification and Accommodation System. Transportation Research Record: Journal of the Transportation Research Board, 141-148.

128 References

CURRIE, G. 2006. Bus Rapid Transit in Australasia: Performance, Lessons Learned and Futures. Journal of Public Transportation, 9, 1.

DIAZ, R. B. A. D. H. 2009. Characteristics of Bus Rapid Transit for Decision Making. Federal Transit Administration, Washinton, D. C.

FERNANDES, P. & NUNES, U. Platooning of Autonomous Vehicles with Intervehicle Communications in SUMO Traffic Simulator. Intelligent Transportation Systems (ITSC), 2010 13th International IEEE Conference 2010. 1313-1318.

FERNÁNDEZ, R. 2010. Modelling Public Transport Stops By Microscopic Simulation. Transportation Research Part C: Emerging Technologies, 18, 856-868.

FERNÁNDEZ, R., BURGOS, V. & CORTÉS, C. E. 2007. Results of the Microscopic Modelling of Traffic Interactions at Stops, Junctions and Roads for the Design of Bus Rapid Transit Facilities. Association for European Transport and Contributors.

FERNANDEZ, R. & PLANZER, R. 2002. On the Capacity of Bus Transit Systems. Transport Reviews, 22, 267-293.

FINN, B., HEDDEBAUT, O. & RABUEL, S. Bus with a High Level of Service (BHLS): The European BRT Concept. TRB 2010 Annual Meeting CD-ROM, 2010.

FITZPATRICK, K., HALL, K., PERKINSON, D. & NOWLIN, L. 1997. Location and Design of Bus Stops. Institute of Transportation Engineers. ITE Journal, 67, 36.

FITZPATRICK, K., HALL, K., PERKINSON, D., NOWLIN, L. & KOPPA, R. 1996. TCRP Report 19: Guidelines for the Location and Design of Bus Stops. Transportation Research Board of the National Academies, Washington, DC.

GARDNER, G., CORNWELL, P. & CRACKNELL, J. 1991. The Performance of Busway Transit in Developing Cities, Citeseer.

GBD. 2017. Global BRT Data [Online]. Available: http://brtdata.org/ [Accessed 03.03.2017].

GERMANI, E. & SZASZ, P. A. COMONOR-A Bus Convoy System. Vehicular Technology Conference, 1980. 30th IEEE, 1980. IEEE, 413-417.

GIBSON, J., BAEZA, I. & WILLUMSEN, L. 1989. Bus-Stops, Congestion and Congested Bus-Stops. Traffic Engineering & Control, 30.

GIPPS, P. 1986. Multsim: a model for simulating vehicular traffic on multi-lane arterial roads. Mathematics and Computers in Simulation, 28, 291-295.

GIPPS, P. G. 1981. A behavioural car-following model for computer simulation. Transportation Research Part B: Methodological, 15, 105-111.

References 129

GOLOTTA, K. & HENSHER, D. A. 2008. Why Is the Brisbane Bus Rapid Transit System Deemed a Success? Road & Transport Research: A Journal of Australian and New Zealand Research and Practice, 17, 3.

GORDON, T. J. & LIDBERG, M. 2015. Automated driving and autonomous functions on road vehicles. Vehicle System Dynamics, 53, 958-994.

GU, W., GAYAH, V. V., CASSIDY, M. J. & SAADE, N. 2014. On the Impacts of Bus Stops Near Signalized Intersections: Models of Car and Bus Delays. Transportation Research Part B: Methodological, 68, 123-140.

GU, W., LI, Y., CASSIDY, M. J. & GRISWOLD, J. B. 2011. On the Capacity of Isolated, Curbside Bus Stops. Transportation Research Part B: Methodological, 45, 714-723.

GUENTHNER, R. P. & SINHA, K. C. 1983. Modeling Bus Delays due to Passenger Boardings and Alightings. Transport Research Record 7-13.

HARDY, M. & PROPER, S. 2006. Analyzing the Impacts of Vehicle Assist and Automation Systems on BRT. Journal of Public Transportation, 9, 3.

HIDALGO, D., LLERAS, G. & HERNÁNDEZ, E. 2013. Methodology for Calculating Passenger Capacity in Bus Rapid Transit Systems: Application to the Transmilenio System in Bogotá, Colombia. Research in Transportation Economics, 39, 139-142.

HIDALGO, D. & MUÑOZ, J. C. 2014. A Review of Technological Improvements in Bus Rapid Transit (BRT) and Buses with High Level of Service (BHLS). Public Transport, 6, 185-213.

HIDAS, P., AITKEN, S., SHARMA, S. & XU, M. Evaluation of Bus Operations by Microsimulation in Sydney CBD Corridor. Australasian Transport Research Forum (ATRF), 32nd, 2009. The Growth Engine: Interconnecting Transport Performance, the Economy and the Environment, 2009.

HISHAM, F., BUNKER, J. M. & BHASKAR, A. Development of a Modified Bus Stop Capacity Model. Transportation Research Board (TRB) 97th Annual Meeting, 7-11 January 2018a Washington DC. Transportation Research Board of the National Academies.

HISHAM, F., BUNKER, J. M. & BHASKAR, A. Improving Capacity Estimation of High Volume On-Street Bus Facilities with Yield-to-Bus Rule. Australasian Transport Research Forum (ATRF), 40th, 2018 Proceedings, 2018b Darwin Convention Centre, Australia.

HISHAM, F., BUNKER, J. M. & BHASKAR, A. 2019a. Capacity Estimation of On-Street, Mid-Block, Off-Line Bus Stops Considering Yield-to-Bus Rule. Transportation Research Record: Journal of the Transportation Research Board.

HISHAM, F., BUNKER, J. M. & BHASKAR, A. 2019b. Incorporating Practical Degree of Saturation in Capacity Estimation of On-Street, Mid-Block, Off-

130 References

Line Bus Stops. Transportation Research Record: Journal of the Transportation Research Board.

HOFFMAN, A. 2008. Advanced Network Planning For Bus Rapid Transit: The “Quickway” Model as a Modal Alternative to “Light Rail Lite”.

HYDE, R. & SMITH, D. 2017. Quantifying the Economic and Other Benefits of Enabling Priority Bus Egress from Bus Stops.

IBARRA-ROJAS, O., DELGADO, F., GIESEN, R. & MUÑOZ, J. 2015. Planning, Operation and Control of Bus Transport Systems: A Literature Review. Transportation Research Part B: Methodological, 77, 38-75.

IBEAS, Á., DELL’OLIO, L., ALONSO, B. & SAINZ, O. 2010. Optimizing Bus Stop Spacing in Urban Areas. Transportation Research Part E: Logistics and Transportation Review, 46, 446-458.

JAISWAL, S., BUNKER, J. M. & FERREIRA, L. 2007. Operating Characteristics and Performance of a Busway Transit Station.

JAISWAL, S., BUNKER, J. M. & FERREIRA, L. 2010a. Influence of Platform Walking on BRT Station Bus Dwell Time Estimation: Australian Analysis. Journal of Transportation Engineering, 136, 1173-1179.

JAISWAL, S., BUNKER, J. M. & FERREIRA, L. Modelling Bus Lost Time: An Additional Parameter Influencing Bus Dwell Time and Station Platform Capacity at a BRT Station Platform. TRB 89th Annual Meeting Compendium of Papers, 2010b. Transportation Research Board of the National Academies.

JAISWAL, S. K. 2010. Busway Platform Bus Capacity Analysis. Queensland University of Technology.

KING, R. D. 1998. New Designs and Operating Experiences with Low-Floor Buses, Transportation Research Board.

KING, R. D. 2003. Yield to Bus: State of the Practice, Transportation Research Board.

KITTELSON AND ASSOCIATES, B. P., KFH GROUP, TEXAS A&M TRANSPORTATION INSTITUTE, ARUP 2013a. Transit Capacity and Quality of Service Manual, 3rd Edition. Transportation Research Board of the National Academics, Washinton D.C. .

KITTELSON AND ASSOCIATES, K. G., PARSONS BRINCKERCHOFF QUADE AND DOUGLAS, DR. KATHERINE HUNTER-ZAWORSKI 2003. TCRP Report 100 : Transit Capacity and Quality of Service Manual, 2nd Edition. Transportation Research Board of the National Academics, Washinton D.C.

KITTELSON AND ASSOCIATES, P. B., KFH GROUP, TEXAS A&M TRANSPORTATION INSTITUTE, ARUP 2013b. Transit Capacity and Quality of Service Manual, 3rd Edition. Transportation Research Board of

References 131

the National Academics, Washinton D.C. .

KRAFT, W., H. 1975. An Analysis of the Passenger Vehicle Interference of Street Transit Systems with Applications to Design Optimization. New Jersey Institute of Technology, Newark, New Jersey.

KULYK, W. & HARDY, M. 2003. ITS Enhanced Bus Rapid Transit Systems. Transportation Research Record: Journal of the Transportation Research Board, 1-43.

LEVINE, J. C. & TORNG, G.-W. 1994. Dwell-Time Effects of Low-Floor Bus Design. Journal of Transportation Engineering, 120, 914-929.

LEVINSON, H. & ST. JACQUES, K. 1998. Bus Lane Capacity Revisited. Transportation Research Record: Journal of the Transportation Research Board, 189-199.

LEVINSON, H., ZIMMERMAN, S., CLINGER, J., RUTHERFORD, S., SMITH, R. L., CRACKNELL, J. & SOBERMAN, R. 2003. Bus Rapid Transit -TCRP-90. Transport Research Board.

LEVINSON, H. S. 1983. Analyzing Transit Travel Time Performance. Transport Research Record, 1-6.

LEVINSON, H. S. 1997. Operational Analysis of Bus Lanes on Arterials, Transportation Research Board.

LIN, F.-B., TSENG, P.-Y. & CHANG, C.-W. Capacities of Exclusive Bus Lanes with On-Line Linear Bus Stops on Urban Arterials. Proceedings of the Eastern Asia Society for Transportation Studies, 2011. Eastern Asia Society for Transportation Studies, 311-311.

LIN, T.-M. & WILSON, N. H. 1992. Dwell Time Relationships for Light Rail Systems. Transportation Research Record.

MARSHALL, L. F., LEVINSON, H. S., LENNON, L. C. & CHENG, J. 1990. Bus Service Times and Capacities in Manhattan. Transportation Research Record.

MENG, Q. & QU, X. 2013. Bus Dwell Time Estimation at Bus Bays: A Probabilistic Approach. Transportation Research Part C: Emerging Technologies, 36, 61-71.

MICHAEL, J. B., GODBOLE, D. N., LYGEROS, J. & SENGUPTA, R. 1998. Capacity Analysis of Traffic Flow Over a Single-Lane Automated Highway System . Journal of Intelligent Transportation System, 4, 49-80.

MILKOVITS, M. 2008. Modelling the Factors Affecting Bus Stop Dwell Time: Use of Automatic Passenger Counting, Automatic Fare Counting, and Automatic Vehicle Location Data. Transportation Research Record: Journal of the Transportation Research Board, 125-130.

ORTIZ, M. Á. & BOCAREJO, J. P. Transmilenio Brt Capacity Determination using

132 References

a Microsimulation Model in Vissim. 93rd Annual Meeting of Transportation Research Board, National Academics, Washington, DC, 2014.

PARKINSON, T. & FISHER, I. 1996. Rail Transit Capacity, Transportation Research Board.

RAJBHANDARI, R., CHIEN, S. & DANIEL, J. 2003. Estimation of Bus Dwell Times with Automatic Passenger Counter Information. Transportation Research Record: Journal of the Transportation Research Board, 120-127.

RATHWELL, S. & SCHIJNS, S. 2002. Ottawa and Brisbane: Comparing a Mature Busway System with its State-of-the-Art Progeny. Journal of public Transportation, 5, 8.

RODEGERDTS, L. A., NEVERS, B., ROBINSON, B., RINGERT, J., KOONCE, P., BANSEN, J., NGUYEN, T., MCGILL, J., STEWART, D. & SUGGETT, J. 2004. Signalized Intersections: Informational Guide.

SAM LOTT, J. & GETTMAN, D. 2016. Preparing for Automated Roadway Technology Impacts on Guideway Transit. Automated People Movers and Automated Transit Systems 2016.

SHLADOVER, S. 2012. Recent International Activity in Cooperative Vehicle–Highway Automation Systems.

SHLADOVER, S. E. 2007. Lane Assist Systems for Bus Rapid Transit, Volume I: Technology Assessment. Publication RTA 65A0160, US Department of Transportation, Washington, DC.

SIDDIQUE, A. J. & KHAN, A. M. 2006. Microscopic Simulation Approach to Capacity Analysis of Bus Rapid Transit Corridors. Journal of Public Transportation, 9, 10.

SISLAK, K. G. Bus Rapid Transit as a Substitute for Light Rail Transit; A Tale of Two Cities. Light Rail: Investment for the Future. 8th Joint Conference on Light Rail Transit, 2000.

ST. JACQUES, K. R. S. & LEVINSON, H. S. 1997. Operational Analysis of Bus Lanes on Arterials: Application and Refinement, Transportation Research Board, National Research Council.

STF. 2016. 'A Better Solution?' [Online]. Available: https://smarttransitfuture.org/tag/autonomous-vehicles/ [Accessed 27.05.2016].

SZETO, W. Y. & WU, Y. 2011. A Simultaneous Bus Route Design and Frequency Setting Problem for Tin Shui Wai, Hong Kong. European Journal of Operational Research, 209, 141-155.

TAN, H.-S., BU, F., JOHNSTON, S., BOUGLER, B., ZHANG, W.-B. & SUN, S. 2009. Field Demonstration and Tests of Lane Assist/Guidance and Precision Docking Technology. California PATH Program, Institution of transport

References 133

studies, University of California, Berkeley.

TAN, H. S. & HUANG, J. The Design and Implementation of an Automated Bus in Revenue Service on Bus Rapid Transit Line. 2014 American Control Conference, 4-6 June 2014 2014. 5288-5293.

TRB 1985. Highway Capacity Manual. Transportation Research Board, First Edition.

TSS. 2016. Available: https://www.aimsun.com/aimsun/ [Accessed 25.05.2016].

VIS, H. & BOUWMAN, R. The Phileas - Integral Safety Approach for an Electronically Guided Vehicle. Intelligent Vehicles Symposium, 2008 IEEE, 4-6 June 2008 2008. 416-421.

VUCHIC, V. R. 1981. Urban Public Transportation Systems And Technology, Prentice-Hall Inc.

VUCHIC, V. R. 2005a. Light Rail and BRT: Competitive or Complementart? : Public Transport International 5.

VUCHIC, V. R. 2005b. Urban Transit: Operations, Planning, And Economics.

VUCHIC, V. R. 2007. Urban Transit Systems and Technology, John Wiley & Sons.

WEI, J., SNIDER, J. M., KIM, J., DOLAN, J. M., RAJKUMAR, R. & LITKOUHI, B. Towards a Viable Autonomous Driving Research Platform. Intelligent Vehicles Symposium (IV), 2013 IEEE, 23-26 June 2013 2013. 763-770.

WIDANA PATHIRANAGE, R., BUNKER, J. M. & BHASKAR, A. 2013. A Microscopic Simulation Model to Estimate Bus Rapid Transit (BRT) Station Service Capacity with Mixed Stopping and Non-Stopping Bus Operation. OPTIMUM 2013: International Symposium on Recent Advances in Transport Modelling. Mantra on Salt Beach Resort, Kingscliff, NSW Australia.

WIDANA PATHIRANAGE, R., BUNKER, J. M. & BHASKAR, A. 2014. Case Study: South East Busway (SEB), Brisbane, Australia.

WIDANAPATHIRANAGE, R. 2015. Modelling Busway Operation with Mixed Stopping and Non-Stopping Buses. Queensland University of Technology.

WIDANAPATHIRANAGE, R., BUNKER, J. M. & BHASKAR, A. A Microscopic Simulation Model to Estimate Bus Rapid Transit Station Bus Capacity. Australasian Transport Research Forum (ATRF), 36th, 2013 Proceedings, 2013 Queensland University of technology, Australia.

WIDANAPATHIRANAGE, R., BUNKER, J. M. & BHASKAR, A. 2014. Modelling the BRT station capacity and queuing for all stopping busway operation. Public Transport, 7, 21-38.

WIDANAPATHIRANAGE, R., BUNKER, J. M. & BHASKAR, A. 2015. Modelling the BRT Station Capacity and Queuing for All Stopping Busway Operation.

134 References

Public Transport, 7, 21-38.

WU, C. & MURRAY, A. T. 2005. Optimizing Public Transit Quality and System Access: The Multiple-Route, Maximal Covering/Shortest-Path Problem. Environment and Planning B: Planning and Design, 32, 163-178.

WU, J., SONG, R., WANG, Y., CHEN, F. & LI, S. 2015. Modeling the Coordinated Operation between Bus Rapid Transit and Bus. Mathematical Problems in Engineering, 2015, 7.

YOSHIOKA, L. R., MARTE, C. L., FONTANA, C. F. & CARDOSO, J. R. Intelligent Transportation Systems with Autonomous Guidance–an application to the improvement of efficiency for median capacity Urban Transportation Systems. Recent researches in telecommunications, informatics, electronics and signal processing, 2013 Baltimore, MD, USA.

YOSHIOKA, L. R., MARTE, C. L., MICOSKI, M., COSTA, R. D., FONTANA, C., SAKURAI, C. A. & CARDOSO, J. R. 2014. Bus Corridor Operational Improvement with Intelligent Transportation System based on Autonomous Guidance and Precision Docking. international journal of systems applications, engineering & development, v. 8. International journal of systems applications, engineering & development, 8, 116-123.

ZHAO, J., CHEN, K., WANG, T. & MALENJE, J. O. 2018. Modeling Loading Area Effectiveness at Off-Line Bus Stops with No Clear-Cut Separation of Berths. Transportmetrica A: Transport Science, 1-21.

ZHAO, X. M., GAO, Z. Y. & JIA, B. 2007. The Capacity Drop Caused by the Combined Effect of the Intersection and the Bus Stop in a CA Model. Physica A: Statistical Mechanics and its Applications, 385, 645-658.

ZHOU, H., BROMFIELD, S. & LIN, P.-S. 2011. An Overview of Yield-to-Bus Programs in Florida. Journal of Public Transportation, 14, 8.

ZOGRAFOS, K. & LEVINSON, H. 1986. Passenger Service Time in a no Fare Bus System. Transportation Research Record, 1051, 42-48.

References 135

136 References

ON-STREET MID-BLOCK OFF-LINE BUS Faheema... · 2020. 1. 20. · Publications Journal Articles...

Documents

Transcript of ON-STREET MID-BLOCK OFF-LINE BUS Faheema... · 2020. 1. 20. · Publications Journal Articles...