Delft University of Technology Evaluating the Implementation Of

Delft University of Technology

Evaluating the implementation of Transit Signal Priority on a bus network, with a focus on equity MSc Transport Infrastructure & Logistics

Final 13th April 2015 |

Student: Rachel Ivers Student Number: 4248538

Committee: Prof. Dr. Ir. Bert van Wee Delft University of Technology TPM Transport and Logistics Dr. Jan Anne Anemma Delft University of Technology TPM Transport and Logistics Dr. Ir. Niels van Oort Delft University of Technology CiTG Transport and Planning Jose Izquierdo Arup

Preface

This study is the final componet in my Masters in Transport Infrastructure and Logistics in Delft University of Technology. With this study I was able to combine my passion between the interaction of public transport and society which I gained through the last two years in Delft and my BSc in Dublin. By combining my knowledge acquired through living in these two countries I feel I have a unique perspective on how transport is provided and has developed.

This thesis was written with Arup. Firstly I would like to thank my colleagues at Arup for their time and the opportunity they gave me to work in such a friendly and experienced company. Working in a company with offices around the world gave me not only access to a network of transport planners but a number of opinions and perspectives. Secondly I would like to thank my committee Prof. Bert van Wee, Dr. Jan Anne Annema, Dr. Niels van Oort and Jose Izquierdo for the patence and time they provided that enabled me to finish my thesis. Next I would like to thank the National Transport Authority in Dublin for providing data and time without which I would not be able to do this topic. Dublin Bus, Dublin City Council, Cyclist.ie and AA roadwatch also helped me by providing information on this topic. Finally I would like to thank my family and friends; without their support I would not have had the opportunity to come to the Netherlands.

Rachel Ivers

April 2015

vii

Louth

Location of the Greater Dublin Area in Ireland and its administrative boundaries.

vii

Summary

This study evaluates the implementation of the introduction of a transit signal priority system on a bus network and uses a new transport appraisal method that focuses on equity. The main research question to be answered is ‘What are the equity and efficiency impacts of introducing transit signal priority in a city network?’ The method used to answer this question is divided into three phases. Phase One describes the components necessary to implement Transit Signal Priority (TSP), outlines some available transport appraisal methods and identifies how they do not deal with the issue of equality within society. Phase Two explains the importance of equality in transport provision, the components necessary for, and introduction of, an equitable transport appraisal method. Once this research is conducted Phase Three applies the information gained in the previous phases and identifies two contrasting zones (socially advantaged and socially disadvantaged) in the chosen study area, County Dublin, Ireland, by mapping socio-economic Census data using ArcGIS. From that, transit signal priority is modelled on the two identified routes using the National Transport Authorities ‘Greater Dublin Area Transport Model’. Next the change in travel times and modal split between the Do Nothing (base case) and Do Something (TSP case) is evaluated using an equitable transport appraisal method identified in Phase Two which uses the Lorenz curve and Gini index and a more common financial based transport appraisal method, a Cost Benefit Analysis. There are two steps to the results in this study. The first are the results from implementing TSP on the two bus routes. These bus routes have differing socio-economic demographic. The North Clondalkin corridor links a socially disadvantaged zone to employment and the Stillorgan corridor links a socially advantaged zone to the same employment zone. The purpose of the contrasting zones is to determine the equity effects on both zones and to test the equity based transport appraisal method. The results are below. They show a 10 minute reduction in travel time by bus on the North Clondalkin corridor and a 7 minute reduction in travel time on the Stillorgan corridor. Both the car travel times and the cycle travel times see no significant change (less than 1 minute). The second sets of results are the equitable and financial results of the transport appraisal methods. The equitable transport appraisal methods using the Lorenz curve and Gini index show that there is a 57% change in equality from the North Clondalkin zone and a 36% change in equality from the Stillorgan zone. Both zones increase the level of equality within the zones and balance the level of equality between the zones when TSP is implemented. The cost benefit reveals a net present value benefit of € 285,000.00, generating a cost benefit ratio of 1.8. The use of the Lorenz curve and Gini index enables different transport proposals to be tested to see the effects it has on different users and helps policy makers decide a proposal that generates the most equitable result. This approach has the ability to see the effects a transport proposal has at a disaggregate level rather than the aggregate level of a cost benefit analysis. Focusing on the aggregated level has the potential to hide negative results in one area with large positive results in another area; therefore increasing the level of inequality, this issue is tackled with the proposed equitable transport appraisal method.

vii

Contents

Preface ...... vii Summary ...... vii Contents ...... viii Figures...... xi Tables ...... xii Equations...... xiii Abbreviations ...... xiv

1 Introduction ...... 1 1.1 Problem description ...... 1 1.2 Problem statement ...... 3 1.3 Hypothesis ...... 3 1.4 Research question ...... 3 1.5 Scope ...... 4 1.6 Social and societal relevance ...... 5 1.7 Who is this report aimed at ...... 5 1.8 Methodology ...... 5 1.9 Report structure ...... 6

2 Transit Signal Priority System ...... 8 2.1 Introduction ...... 8 2.2 TSP system architecture ...... 8 2.3 TSP requirements for the study area ...... 9 2.4 Conclusion ...... 12

3 Transport Appraisal Methods ...... 13 3.1 Cost Benefit Analysis ...... 13 3.2 Multi Criteria Analysis ...... 13 3.3 Efficiency Analysis ...... 13 3.4 Accessibility Analysis ...... 14

viii

3.5 Conclusion ...... 14 4 Importance of equitable transport ...... 16 4.1 Urban development ...... 16 4.2 Definitions ...... 16 4.3 Discussion ...... 17 4.4 Conclusion ...... 18

5 Components necessary for an equitable Transport Appraisal Method ...... 19 5.1 Categories of transport equity ...... 19 5.2 Accessibility based equity analysis -two ethical principles ...... 20 5.3 Socially Relevant Accessibility Impacts ...... 21 5.4 Conclusion ...... 23

6 Introducing an equitable Transport Appraisal Method ...... 24 6.1 Lorenz curve and Gini index ...... 24 6.2 Common fields of application ...... 25

7 Identifying socially (dis)advantaged zones in study area ...... 27 7.1 Introduction ...... 27 7.2 Individual component ...... 28 7.3 Land use component ...... 31 7.4 Individual and Land Use Conclusion ...... 31 7.5 Transport component ...... 32 7.6 Temporal component ...... 33 7.7 Scope to introduce TSP ...... 34 7.8 Conclusion ...... 37

8 Modelling Transit Signal Priority ...... 38 8.1 Summary of model ...... 38 8.2 Limitations and considerations ...... 39 8.3 Strengths ...... 40 8.4 Final plan ...... 41 8.5 Determining zone numbers ...... 41

8.6 Implementing TSP in GDA Model ...... 42 8.7 Results & analysis ...... 46 8.8 Conclusion ...... 51 8.9 Discussion ...... 53

9 Lorenz Curve and Gini Index ...... 55 9.1 Lorenz curve ...... 55 9.2 Gini index ...... 57 9.3 Conclusion ...... 58 9.4 Discussion ...... 59

10 Cost Benefit Analysis ...... 61 10.1 Components ...... 61 10.2 Results ...... 65 10.3 Conclusion ...... 65 10.4 Discussion ...... 66

11 Conclusion and Recommendations ...... 67 11.1 Conclusion ...... 67 11.2 Recommendations on improvements of the methodology ...... 71 11.3 Recommendations for further research ...... 71

12 References ...... 73

13 Appendix ...... 78

Figures

Figure 1 Study area boundary ...... 4 Figure 2 Report Structure ...... 7 Figure 3 Location of AVL points...... 9 Figure 4 Map based presentation of the bus positions ...... 10 Figure 5 Virtual detectors ...... 10 Figure 6 Dublin County Councils ...... 11 Figure 7 Relationship between components of accessibility ...... 21 Figure 8 Lorenz curve and Gini index ...... 24 Figure 9 Professional workers and Unskilled workers ...... 28 Figure 10 Students and Unemployed having lost or given up previous job ...... 29 Figure 11 Households with no cars ...... 29 Figure 12 Means of travel to work, school and college by bus, car passenger and rail respectively ...... 30 Figure 13 Employment and retail locations ...... 31 Figure 14 Origin & Destination in study area ...... 31 Figure 15 Dublin Frequent Transit Map and Road network ...... 32 Figure 16 Stillorgan QBC dwell time analysis ...... 34 Figure 17 North Clondalkin Route 40 18 hr period Kphs ...... 36 Figure 18 Stillorgan Route 145 18 hr period Kphs ...... 36 Figure 19 CUBE and ArcExplorer screen shots ...... 41 Figure 20 Example junction information in SATURN ...... 43 Figure 21 GDA Model process screen shots...... 44 Figure 22 Addition of extra step in PT Assignment ...... 45 Figure 23 North Clondalkin Lorenz curve ...... 56 Figure 24 Stillorgan Lorenz curve ...... 57 Figure 25 Lorenz curve and Gini index example ...... 57 Figure A.26 Data classification field in ArcGIS ...... 79 Figure A.27 Population density ...... 80 Figure A.28 Means of travel by motorised vehicles ...... 81 Figure A.29 Modelled Area and Model Zoning ...... 85 Figure A.30 Section of the Road Network as coded in the GDA Transport Model ...... 86 Figure A.31 Section of Public Transport Network – coded in GDA Transport Model ...... 86 Figure A.32 GDA model structure...... 87 Figure A.33 Sample trip matrix ...... 88 Figure A.34 Structure of Mode choice / Hour of travel choice and Route choice feedback loop ...... 90 Figure A.35 Software packages used per model step ...... 94 Figure A.36 Stillorgan SM route ...... 98 Figure A.37 North Clondalkin SM route ...... 98 Figure A.38 Stillorgan HW image and cross section ...... 99 Figure A.39 Donnybrook HW image and cross section ...... 99 Figure A.40 Ballyfermot HW image and cross section ...... 100

Figure A.41 Sarsfield HW image and cross section ...... 100

Tables

Table 1 Perspectives on accessibility and components ...... 22 Table 2 AM and PM comparative bus and car journey times (mins) ...... 33 Table 3 Number of bus route on the North Clondalkin and Stillorgan QBC’s ...... 33 Table 4 Number of signalised junctions and length of studied bus route ...... 34 Table 5 List of unscheduled stops ...... 35 Table 6 SATURN Record type 1 ...... 43 Table 7 SATURN Record type 3 ...... 43 Table 8 North Clondalkin public transport travel time ...... 46 Table 9 Stillorgan public transport travel time ...... 47 Table 10 North Clondalkin & Stillorgan bus route distance & study area distance ...... 47 Table 11 North Clondalkin highway travel time ...... 47 Table 12 Stillorgan highway travel time ...... 48 Table 13 GDA highway network travel time ...... 48 Table 14 North Clondalkin and Stillorgan soft mode travel times ...... 48 Table 15 Modal split percentages ...... 49 Table 16 Modal split total user increase above base ...... 49 Table 17 Sub-modal split increase above base ...... 50 Table 18 PT ordered from largest passenger km travelled to smallest ...... 51 Table 19 North Clondalkin summary results ...... 52 Table 20 Stillorgan summary results ...... 52 Table 21 Modal split total user increase above base ...... 52 Table 22 North Clondalkin input values ...... 55 Table 23 Stillorgan input values ...... 55 Table 24 Gini Index results ...... 58 Table 25 CBA results ...... 65 Table A.26 North Clondakin QBC unscheduled stops ...... 82 Table A.27 Origin and Destination zone numbers ...... 95 Table A.28 North Clondalkin SATURN run results ...... 97 Table A.29 Stillorgan SATURN run results ...... 97

xii

Equations

Equation 1 Gini index ...... 24 Equation 2 Gini index from Lorenz curve and line of equality ...... 24 Equation A.3 Mode choice ...... 91 Equation A.4 Route choice HW ...... 92 Equation A.5 Trip Assignment PT ...... 93

xiii

Abbreviations

AA Automobile Association AVL Automatic Vehicle Location ED Electoral District CBA Cost Benefit Analysis CIÉ Coras Iompair Éireann CSO Central Statistics Office DEA Data Envelopment Analysis DMUs Decision Making Units DPTIM Dublin Public Transport Interface Model DUTC Dublin United Tramways Company GDA Greater Dublin Area GIS Geographic Information Systems GPS Global Positioning System GTFS General Transit Feed Specification HW Highway ITS Intelligent Transport System LCY Total Cycle Time MCA Multi Criteria Analysis NITA National Irish Taxi Association NTA National Transport Authority POWSCAR Place of Work, School and College. Census of Anonymised Records Pax Passengers PT Public Transport QBC Quality Bus Corridor RTPI Real Time Passenger Information SA Small Area SAPs Small Area Population Statistics SCATS Sydney Coordinated Adaptive Traffic System SIRI Service Interface for Real time Information SM Soft Mode SRAI Socially Relevant Accessibility Impacts TAM Transport Appraisal Method TT Travel Time TSP Transit Signal Priority VoT Value of Time

xiv

1 Introduction

1.1 Problem description Dublin tramways started operating in 1872, the largest operator being the Dublin United Tramways Company (DUTC). At the turn of the 20th century Dublin had 66 miles of tram lines (National Transport Museum, n.d.), with some lines operating at the impressive frequency of four minutes (Mullen, 1989). The DUTC opened its first bus route in 1925 and with the last bus replacing trams in July 1949. The reason for the decline in trams is said to be due to a number of factors. Firstly buses and trams ran on the same route and buses were able to overtake trams and ‘grab’ passengers (Clontarf Online, 2012); secondly buses were able to service Dublin’s expanding suburbs faster and at less cost than trams and finally trams were (considered) outdated and old technology (Stewart, 1955). There was a belief that buses were cheaper to run then trams and that the tram system was in a poor state of repair. Many buses to this day still run on the same tram lines and use the same numbers. Following the Transport Act 1944, control of the DUTC was vested in the newly formed Coras Iompair Éireann (CIÉ) (Office of the Attorney General, 1944).

The 1960s saw a focus toward the car with the publication of the Traffic in Town (1963), or the Buchanan Report as it’s more widely known, which was advanced by the Roads in Urban Areas in 1966. Both reports were published in the UK but heavily influenced Irelands approach to road design (Dept of Environment, Community and Local Government, 2014). One of the most expansive examples of this influence can be seen in the Dublin Transportation Strategy (1971) which sought to reshape inner Dublin into a functional system of one-way street systems, ring roads and motorways in order to relieve congestion (although this was never realised).

Today, the need for increased public transport provision, a higher level of accessibility and an increased awareness of the role public transport plays in an inclusive society has grown. Lucas (2012) looks at accessibility not as reducing social exclusion but as enhancing the quality of life. Kenyon et al., (2002) suggests there is a strong correlation between a lack of access to adequate mobility and a lack of access to opportunities, social networks, goods and services. Finally Jones and Lucas (2012) seek to highlight the importance of understanding the social impacts and consequences, as well as the distributional effects, of transport decision-making. As a result there is a new focus towards increasing the provision of public transport in urbanised areas and making it more advantageous.

There are many ways in which public transport can provide increased accessibility to society for example by increasing the number of available opportunities, reducing travel time, cost savings, providing to the needs and abilities of users and mitigating any time restrictions (Geurs & van Wee, 2004). One possible way in which cities are making public transport more advantageous is by improving bus service reliability. Traffic, weather and driver behaviour can all reduce public transports reliability (van Oort & van Nes, 2007). However

there are many ways in which the effects of these events on the service reliability can be reduced. Van Oort and van Nes (2007) outline at three different levels how this can be achieved. At the strategic, tactical, and operational level, service reliability can be reduced through the prevention of bunching, overcrowding and uncertainty. These can be controlled through a number of measures like detours, slowing down, speeding up and the use of software and technology such as Traffic Signal Priority (TSP).

This study looks at evaluating the introduction of TSP to a network. Traffic Signal Priority is a traffic light control strategy that favours an approaching bus over the rest of the traffic on the road by introducing a little extra green time or a little less red time at traffic signals to reduce the time buses are slowed down by traffic signals [p. 3 (Smith, et al., 2005)].

When implementing a new public transport innovation, the most common way of determining its feasibility is through an economically focused analysis most commonly a Cost Benefit Analysis (CBA). This is achieved by defining the project and any alternatives; then by identifying, measuring, and valuing the benefits and costs of each, or through the use of a Multi Criteria Analysis (MCA); by evaluating alternatives on a set of criteria reflecting the decision-maker’s objectives (De Brucker, et al., 2011). The MCA differs from a CBA in that the scores achieved with an MCA do not need to be conveyed in monetary terms, but can simply be expressed in physical units or in qualitative terms (De Brucker, et al., 2011). These two methods determine the benefits gained from implementing such an innovation. However there is little research conducted on the ethics of such innovations and the effect it has on creating an equitable society. For instance, an ethically based analysis for a public transport innovation should aim to reduce the societal gap between those who will gain and those who will not, and the disparity and inequality that can result when different innovations are implemented (van Wee & Lucas, in press). This study aims to contribute to that knowledge by evaluating the introduction of TSP on a network with a focus on the impacts of equity.

The equity analysis used in this study derives from a paper by van Wee and Lucas (in press) stemming from two ethical principles of egalitarianism and sufficientarianism and using the Lorenz curve and Gini index. This is explained in greater detail in chapter 5.

In order to test the equality impact of TSP on a society there is the need to use a city as a case study. The cities position in this study is purely to facilitate the application of TSP and to test the equitable transport appraisal method. The city chosen is Dublin. Dublin City Council has just developed software to implement TSP. The city is currently applying TSP on certain isolated junctions which are identified by Dublin Bus (bus service operator) as locations in which serious delays occur. However Dublin Bus and Dublin City Council (traffic managers) have not implemented TSP on a route bases (but aspires to), they have also not modelled the effects on the network nor have they conducted an equality based appraisal. This study aims to model the effects of TSP on two bus routes as well as conduct an equality based appraisal of the implementation.

1.2 Problem statement Many cities around the world such as in America and the Netherlands have implemented TSP (ITS America, 2005) (Furth & Muller, 2000) (Feng, 2014). Dublin is heavily dependent on its bus network; in 2013 114.4 million trips were made by bus (Dublin Bus, 2014), 70% of those trips carried passengers on high frequency routes on the Quality Bus Corridors (QBCs). Other public transport modes available in Dublin include Luas and DART, both rail based systems, which carried 5.41% and 12.94% respectively. The mode share of the morning peak traffic into Dublin is 46% by public transport, 37% by car, 9% by walking, and 4% by bicycle (NTA, 2013). Congestion in cities is on the rise and the need for a more attractive and reliable public transport service is required. In Dublin there have been many plans for new public transport infrastructure (Transport 211) but it is unlikely to be realised in the near future due to a lack of government resources. Transport surveys demonstrate that it is usually the poorest and most socially disadvantaged within society who experience transport disadvantage (Lucas, 2012); therefore they are likely to be affected the most by a lack of transport development. However transport appraisal methods do not (yet) assess the level to which transport innovations reduce this disadvantage. This study aims to determine if TSP will reduce inequalities in a society and if it is financially feasible.

1.3 Hypothesis The testing hypothesis is that introducing transit signal priority will result in a reduced travel time for bus passengers, have minimum negative impact on other road users and finally, balance the societal equity among differing socioeconomic areas.

1.4 Research question The objective of this study is to establish if implementing transit signal priority achieves the aim of improving the travel times of the bus in the study area, and to balance social equality, as stated in the hypothesis.

Therefore the aim of the research and the research questions is to determine:

What are the equity and efficiency impacts of introducing transit signal priority in a city network?

In order to determine the research question the following sub categories of questions need to be answered which are divided into three phases..

1 Transport 21 was an infrastructure plan announced 2005. It aimed to greatly expand Ireland's road network, along with investment in public transport in the form of buses and rail. Due to the economic decline, the Transport 21 project was cancelled in May 2011. Replaced with the Draft Transport Strategy for the Greater Dublin Area 2011-2030.

Phase One

1. What are the technical requirements for implementing TSP? 2. What methods are currently used for appraising transport projects?

Phase Two

3. Why is transport equality important? 4. What components are necessary for an equitable transport appraisal method? 5. Formulate an equitable transport appraisal method

Phase Three

6. What are the travel time and resulting equity impacts from implementing TSP?

1.5 Scope

1.5.1 Geographical In order to answer the research question a city needs to be chosen to apply TSP. The cities position in this study is purely to facilitate the application of TSP and to test the equitable transport appraisal method. The city chosen does not have to be limited to a defined level of bus provision, the level of equality or the socio- economic demographic of the city. Therefore any city can be chosen. The city used in this study is Dublin, Ireland which is bounded to the East by the sea and to the North, South and West by the M50 motorway. These are the geographical boundaries of the study area. The bus routes included in this case study includes all Dublin Bus routes that run on one of the 16 dedicated bus only lanes called Quality Bus Corridors (QBCs). Figure 1 Study area boundary

1.5.2 Scientific This study contributes to scientific research as it is the first to test a transport proposal in an equitable manner. This study takes the concept developed by van Wee and Lucas’ (in press) paper ‘Developing a method to evaluate accessibility: combining ethical theories and accessibility-based approaches’ in its approach to assessing transport from an ethical perspective and builds upon it by applying and testing this new equity approach to a case study through the introduction of a transit signal priority system on a bus network. The scope limits itself to the travel time of the bus in AM peak using 2006 base case data, assessing TSP from an ethical perspective, and comparing it against a traditional financially based analysis.

1.6 Social and societal relevance The social and societal relevance of this this study is looking at changing a transport network from a societal perspective, through an ethical evaluation and not through the more common economic and political evaluations. The aim is to implement transit signal priority in order to achieve a more balanced society by enabling bus users to achieve the same level of accessibility to key destinations in the same travel time as car users.

1.7 Who is this report aimed at This study is aimed at a number of actors related to transport namely the Dublin County Councils who are in charge of road maintenance and control, the bus operator, and the policy and investment makers. It is important to aim this report at these actors to highlight this new appraisal method and show the importance of including an equitable appraisal method in future transport investment projects. This study benefits the field of social, transport and policy science as it provides a new method to evaluate transport that specifically looks at the equality affects it has within a society.

1.8 Methodology There are a number of steps to be taken in order to achieve a successful evaluation of transit signal priority with a focus on equity.

1. First the components of transit signal priority are outlined and what apparatus are required in the study area, this is conducted mainly through a desktop analysis and is further explained in chapter 2, 2. Next transport appraisal methods are described in chapter 3 in which the aim is to show that common transport appraisal methods do not have the provision to assess equality between zones in a defined study area, 3. Following this the importance of having equitable transport in cities is expressed in chapter 4, 4. A proposed equitable transport appraisal method using the Lorenz curve and Gini index is outlined in chapters 5 and 6, 5. Next current fields of application of the Lorenz curve and Gini index area introduced both within and outside the sphere of transport seen in section 6.2.

In order to apply this new equitable method two bus routes are taken from which to introduce transit signal priority and determine the effects.

6. Two contrasting suburban residential zones that have different socio-economic characteristics are found in the study area from which to apply transit signal priority, which is explained in greater detail in chapter Error! Reference source not found.7 7. An initial check of these routes is conducted using Automatic Vehicle Location data (Nov 2013) to see if the bus routes have contrasting service characteristics regarding junction dwell time, travel time, speed and modal split, seen in section 7.7,

8. Once confirmed that these bus routes have different social and service characteristics the National Transport Authorities Greater Dublin Area (GDA) transport model is used to model the effects of transit signal priority in chapter 8. Strengths, limitations and assumptions to the model are also made. 9. The model results are put into the Lorenz curve and a Gini index is generated to see if there is a greater level of equality on those routes in chapter 9. 10. In chapter 10 the results of the model are used to conduct a Cost Benefit Analysis in order to compare the results from the Lorenz curve and Gini Index. 11. Finally, conclusions are drawn in chapter 11 as to whether using the Lorenz curve and Gini index is a good equality appraisal method for evaluating the implementation of TSP on a bus network. Final recommendations are then made for further study.

The first two steps in the methodology answers Phase One of the research questions on page four, steps three, four and five answer Phase Two of the research questions, and steps six to ten answer Phase Three. Finally step eleven draws a conclusion on the process and proposes recommendations for further study.

1.9 Report structure The report structure is shown below with the methodology above.

Phase One describes the components necessary to implement Transit Signal Priority and available transport appraisal methods. Phase Two explains the importance of equality in transport provision, the components necessary for, and introduction of, an equitable transport appraisal method. Once this research is conducted Phase Three applies the information gained and identifies two contrasting routes in the study area and then models transit signal priority using the National Transport Authorities ‘Greater Dublin Area Transport Model’, and then evaluates the results from an equitable and financial manner. Finally conclusions and recommendations are drawn.

Research Question Chapter One

Componements of TSP

Chapter Two Phase One Phase Transport Appraisal Methods Chapter Three

Importance of equality in transport provision, Components necessary for and introduction of an equitable transport appraisal method

Chapters Four, Five, Six Phase Two Phase

Identifying two contrasting study areas from which to apply TSP

Chapter Seven

Modelling TSP

Chapter Eight Phase Three Phase

Evaluating the modelling results in an equiable and financial manner Chapters Nine, Ten

Conclusion and Recommendations Chapter Eleven

Figure 2 Report Structure

2 Transit Signal Priority System

The aim of this section is to outline the components necessary to introduce a Transit Signal Priority (TSP) system which answers the first research sub-question in Phase One: What are the technical requirements for implementing transit signal priority?

2.1 Introduction Giving priority to transit vehicles at signalized intersections is one of the most powerful strategies for improving a transit service that operates in separated or mixed traffic (Furth & Muller, 2000). Furthermore eliminating unplanned stops e.g. at traffic lights is the best way to improve reliability (van Oort, 2011). Advances in technology continue to create new and less costly means of providing priority, while increasing traffic congestion and concerns about air quality, liveability and the cost of rail constructions make the need for transit priority ever more pressing (Furth & Muller, 2000). Early focus of TSP applications was to do with improving speed, reducing operating cost and passenger ride time. Now it has progressed on to mitigating a major source of randomness in operations at intersection delays and improving service reliability. Before TSP, regimes to control the travel time and improve service reliability were through holding early vehicles at control points. However this only controls early vehicles, it is unable increase the performance of late vehicles. With TSP it is now possible to influence the performance of late vehicles.

2.2 TSP system architecture TSP priority requires a number of elements in order for it to work. The TSP system architecture is as follows, taken and summarised from ITS America (2005).

At the forefront of the TSP structure, a vehicle detection system is in place to deliver vehicle data (location, arrival time, approach, etc.) and to activate the TSP. This is one of the most important components of TSP as it is necessary to distinguish transit vehicles from the rest of the road traffic. The accuracy of the detection is also an important factor. It is worth noting that usually the type of vehicle detection method is made in tandem with the selection of TSP software. As a result there are a number of detection techniques such as; Hard-Wired Loop Detection, Light-Based (infrared) Detection, Sound-Based Detection, Radio-Based Detection and Satellite (GPS)-Based Detection with Automatic Vehicle Location (AVL).

A communication system is used for the vehicle detection / priority request information from the vehicle to the intersection or the management centre. This system should be capable of capturing data for later analysis. There are two types of communications 1) Vehicle detection communications and 2) Communications between intersections and management centre. Some applications only require one system but usually two are required. Typically intersection controllers and the management centre are physically connected with fibre optics, however newer systems use wireless technology which eliminates those physical costs.

The traffic signal control system is responsible for acting on the priority request. This could be taken locally by the traffic signal controller or through a centralized traffic signal control 8

system. Depending on the predefined parameters the traffic signal control system may or may not make actual changes to the signal indications. A limitation of using an AVL system for vehicle detection and then prompting the traffic signal control system for a priority request is that AVL is polled on average every 20 seconds which is too long for TSP. This requires a separate model to be interfaced with the control system.

2.3 TSP requirements for the study area

2.3.1 Vehicle detection system At the forefront of the TSP system vehicles are detected through an AVL system. All buses in the study area (Dublin Bus fleet) have an AVL system fitted with a SIRI (service interface for real time information) standard interface for connection to the SCATS2 system with a polling rate of 30 seconds (in practice it can be the nearest to 20 seconds) (Dublin Bus, 2011). Figure 3 shows the location of the AVL points over the whole bus network.

Figure 3 Location of AVL points

2 SCATS (Sydney Coordinated Adaptive Traffic System) 9

The polling rate for vehicle detection of 30 seconds through the AVL is slow and means that it is not possible to use this method for individual junction priority. SCATS has an intelligent transport system (ITS) port that allows a separate model to be interfaced e.g. for TSP (O'Brien & O'Donnell, n.d.). Therefore a Dublin specific interface module has been designed allowing it to process data and interface with SCATS to provide a feed of public transport data to influence network decisions. This system is called Dublin Public Transport Interface Module for SCATS; otherwise abbreviated to DPTIMS. In order to visualise this geo-spatial data a map based graphical user interface was developed as seen in Figure 3. Other transport systems such as SCOOTS (Split Cycle Offset Optimization Technique) and UTOPIA / SORT incorporate a level of public transport priority capability (Ireland i, 2010).

In service bus

In service bus at a stop

Bus in congestion

Bus within virtual detector and has the correct routing information

Figure 4 Map based presentation of the bus positions

Different layers of the map are presented in Figure 4 such as the traffic signal junctions operated by SCATS, bus routes highlighted in different colours, bus garages and virtual detectors. Each bus location is represented by a shaded dot. In-service buses are represented in yellow, with a red boundary when at a stop. Buses that are in congestion are represented in pink and a blue one means that the location of the bus is within a virtual detector and has the correct routing information. The different bus routes and their journey patterns are presented on the road network (O'Brien & O'Donnell, n.d.).

Figure 5 Virtual detectors

Figure 5 presents the start and end journey time for virtual detectors for a section of the road network. At the start the virtual detector is highlighted orange to indicate that there is a bus within its geo-spatial area and it is travelling in the right direction. The end virtual detector is presented in grey as it contains no buses of interest yet (Kinane & O'Donnell, 2013).

Location of bus stops Bus stops are located on the far side of the traffic signals as much as possible, in the Quality Bus Corridor’s (QBCs) this is especially evident. This helps ensure that when a call is made there are no delays in availing of the priority change at the junction.

2.3.2 Communication system 3 A high speed fibre optic network link between the Córas Iompair Éireann CIE computer centre and Dublin City Council (DCC) traffic management offices has been installed and is now operational.

2.3.3 Traffic signal control system Traffic signal control system SCATS is used by Dublin City Council and gathers data on traffic flows in real-time at each intersection. This data is fed via the traffic control signal box to a central computer. The computer makes incremental adjustments to traffic light timings based on minute by minute changes in traffic flow at each intersection. SCATS performs a vehicle count at each stop line and also measures the gap between vehicles as they pass through each junction. As the gap between vehicles increases the lights are wasting green time, and SCATS seeks to reallocate green time to where demand is greatest. The SCATS system is used in many urban areas including Hong Kong, Sydney (where it originated), Melbourne, and Oakland County (MI) and Dublin (Ireland i, 2010).

There are four county councils in the study area, Dublin City Council, Fingal County Council, Dun Laoghaire / Rathdown County Council, and South Dublin City Council. They do not all use the same SCATS system as Dublin City Council uses (who developed DPTIMS). SCATS system is used in Dublin City Council, Fingal County Council and Dun Laoghaire / Rathdown County Council while SCOOT is used in South Dublin City Council (Dept. of Transport, Tourism and Sport, 2011). This means that Dublin Bus routes in South Dublin will be unable to have TSP on their entire route, unless they develop something similar to DPTIMS for their traffic control system.

Figure 6 Dublin County Councils

3 Córas Iompair Éireann (CIÉ) is Ireland's national public transport provider. It is the holding company for the National Train Service - Iarnród Éireann/Irish Rail, the National Bus & Coach Service - Bus Éireann, Dublin’s Bus Service - Dublin Bus, and Coach Tours Service – CIE Tours International. 11

2.4 Conclusion It is concluded that introducing TSP at route level is possible as the technical requirements, vehicle detection, communication system and traffic signal control system have all been developed. It is assumed for this study that Fingal has the ability to implement TSP the same way as the rest of the Dublin county councils. Dublin has started testing TSP on isolated junctions in the summer of 2014 (Doherty, 2014).

3 Transport Appraisal Methods

The purpose of this section is to determine the best method to appraise the benefits and costs of implementing TSP now that the technical requirements have been determined. As this study is evaluating the implementation of TSP with a focus on equity the aim of this section is to find a transport appraisal method already used and that deals sufficiently with appraising the equality outcome on society of transport proposals. This section provides a selection of some of the different types of transport appraisal methods available, explains how the methods work and their benefits and pitfalls. This section answers sub question two of the research questions ‘What are the methods currently used for appraising transport projects?’

3.1 Cost Benefit Analysis A Cost Benefit Analysis (CBA) is an overview of all the pros (benefits) and cons (costs) of a project or policy option. These costs and benefits are as much as possible quantified and expressed in monetary terms. Costs and benefits that occur in different years are discounted and presented as so called net present values. Final results are often presented in summarizing indicators, such as the difference between costs and benefits, the return on investment, and the benefit-cost ratio (van Wee, 2012). Cost and benefits can include investment costs, maintenance costs, operational costs, and revenue costs / benefits, while social savings include travel time savings, safety increase, environmental benefits, reliability savings and operational savings.

3.2 Multi Criteria Analysis For Multi Criteria Analysis (MCA) alternatives are evaluated on a set of criteria reflecting the decision-maker’s objectives, and ranked on the basis of an aggregation procedure. Scores achieved do not necessarily need to be conveyed in monetary terms (as with a CBA), but can simply be expressed in physical units or in qualitative terms (De Brucker, et al., 2011).

3.3 Efficiency Analysis Data envelopment analysis (DEA) is a linear programming based technique that provides an objective assessment of the relative efficiency of similar organizational units (Sarica & Or, 2007). These organizational units are known as Decision Making Units (DMUs) in DEA analysis i.e. the DMUs are the different transport alternatives (Cooper et al.,2000), cited in (Caulfield, et al., 2013).

Sensitivity analysis is required to evaluate the robustness of the results obtained. Four terms, robustly efficient, marginally efficient, marginally inefficient and significantly inefficient are used to describe how efficient the DEA model is when some of the DMUs are removed. Finally, DEA could also be used in conjunction with a CBA to compare different investments and mode options on one route.

In Dublin, Ireland, DEA was employed to identify the most efficient public transport solution, a city centre to airport route and to establish the reasons for inefficiency. It aimed to determine the best transport system to invest in the given cost constraints.

3.4 Accessibility Analysis Spatial Network Analysis of Public Transport Accessibility (SNAPTA) is a GIS-based accessibility model that has been developed to measure the accessibility by public transport to different urban services and activities. Karou and Hull (2014) define accessibility as ‘whether or not people can get to services and activities at a reasonable cost, in reasonable time and with reasonable ease’.

SNAPTA relies on three accessibility measures to improve accessibility by public transport to six types of activity opportunities (Karou & Hull, 2014).

Accessibility measures:  Access time to city centre  A contour measure  A potential accessibility measure

Six types of activities opportunities selected:  Central Business District  Health opportunities  Employment  Leisure and recreational  Retail opportunities opportunities  Education opportunities

The model is said to respond to several limitations in other existing accessibility models in planning practice4. It offers an alternative and practical tool to help planners and decision makers examine the strengths and weaknesses of land use – transport integration.

The model developed in this study is not intended to provide the complete picture of transport accessibility but it attempts to cover adequately the required aspects of accessibility measurement.

3.5 Conclusion This chapter shows that many appraisal methods are available, some of which can be seen to tackle equality in transport provision through efficiency and accessibility appraisals however none deal with the level of equity within a society, they all deal with society as a whole.

While there are only a few examples of transport appraisal methods given above the CBA is a very popular ex-ante evaluation method in many countries with MCA being an alternative for

4 Models can be inflexible, non-user friendly, extra requirement for an external function to be integrated or being restrictive to only one transport mode. 14

CBA as well as combining MCA and CBA (van Wee, 2012). Consequently the pitfalls of these two are looked at in more detail, especially relating to how they deal with equality.

Caulfield et al. (2013) say that studies (Browne and Ryan, 2011; Damart and Roy, 2009; Tudelaetal., 2006) have indicated that it is very difficult to monetise all the impacts of transport projects (i.e. CBA and MCA), in particular benefits or costs that do not have constant economic values. Take travel time as an example, which factors should be taken into account when putting a monetary value on time saved i.e. geographic zone, profession, social status, or travel purposes?

Van Wee and Lucas (in press) state that a CBA generally ignores the distributional and equity effects and other ethically important implications of choice options, e.g. the area of social exclusion. The authors go on to state that this is an important oversight for contemporary policy analysis, as it is now widely recognised that ‘sound’ policies should meet three criteria: i) effectiveness, ii) efficiency and iii) equity (e.g. Young and Tilley, 2006) and so a CBA is incomplete if equity or fairness is not considered (van Wee, 2012). This can also be seen through the other methods of efficiency and accessibility as they look at these criteria in isolation and don’t combine them with other criteria.

A disadvantage noted by van Wee and Lucas (in press) with a MCA is the risk of double counting. The authors give the example that, a new bus service may reduce travel times as well as the level of social exclusion, but social exclusion is reduced as a result of the reduction in travel times, so there is overlap between the impacts of the two indicators. MCA also does not inherently identify the spatial or social distribution of impacts across different population groups and this needs to be specifically identified as a separate necessary step if there is concern to protect or enhance the accessibility of particular ‘at risk’ groups.

To conclude further research needs to be conducted to formulate a transport appraisal method that sufficiently deals with the question of equity. None of the methods above deal sufficiently with reducing the inequality seen within a network; they deal with reducing travel time or increasing accessibility for a single mode but not the effect that has on other mode users or users of the same mode but who do not receive the same benefit. The following three chapters deal with this issue.

4 Importance of equitable transport

This chapter answers the third research sub question which is the start of Phase Two ‘Why is transport equity important?’ So far this study has outlined some different transport appraisal methods and highlighted the most common one but the study has not yet discussed why equality, ethics or accessibility are important when evaluating transport. The next section will discuss this.

4.1 Urban development It is the goal of many cities around the world for vulnerable members of society to have access to transport, jobs and education (Rosler & Mc Donald, 2011) (Welsh, 2013). The reason for this goal is to reduce the level of social inequality and ensure equal opportunities for everyone.

However due to past urban planning policies some cities moved vulnerable members of society from city centre locations to social housing units on outer suburban rings. The urban slum clearances in the 1960’s (in Ireland) was also at the time when there was a shift towards car based transport. As a result these social housing areas had inadequate public transport services while the residents who previously lived in city centre locations had no use for a car and could not afford to buy one.

With the growth of cities in the 1960s, the availability of land and the desire for households to have space between their neighbours resulted in low density housing being developed. Consequently this urban development typology was difficult to serve by public transport, this was also the ‘time of the car’ so adequate public transport provision was not seen as a necessity. The result was the potential for households to become isolated both from jobs, education and from interacting with other communities causing social exclusion.

Therefore today some of the aims to providing public transport is to reduce the level of isolation, increase accessibility and the level of social inclusion. The aim is to reduce the level of inequality seen between areas with regard to transport.

Today there is a greater understanding that society, public transport and equality are interlinked, that spatial development forms transport provision and vice versa. This chapter aims to understand this relationship better. To start off a few definitions of accessibility, social exclusion and equity are provided.

4.2 Definitions At its simplest level, accessibility is the ease of reaching opportunities or the ease of being reached (Jones, 1981) cited in (Halden, 2011). In order to understand why accessibility is important and the role accessibility plays in society and its relationship with social exclusion and poverty it is important to distinguish between these two terms before continuing. Poverty and social exclusion are sometimes considered the same, however they are not. Poverty should be seen as only one dimension of social exclusion. A weakness of the understanding

of poverty is that it assumes there is always a fixed level of basic needs and income which is insufficient to provide the needs which defines the poverty line [Foley, 1999, p. 3; cited from (Kenyon, et al., 2002)]. The concept of relative poverty allows for considerations of the multiple deprivations arising from an inadequate income. Today lack of access to a car is considered a characteristic of poverty (Kenyon, et al., 2002).

Mobility related social exclusion is defined by Kenyon et al., (2002) as ‘the process by which people are prevented from participating in the economic, political, and social life of the community because of reduced accessibility to opportunities, services and social networks, due in whole or in part to insufficient mobility in a society and environment built around the assumption of high mobility’.

Determining equity is difficult because there is no standard definition of distributional equity for transportation benefits (Martens, et al., 2012). However, equity, as stated by Santos (2008) refers to the fairness and justice of the distribution of the impacts (benefits and costs) of an action on two or more units. Depending on the available data and the chosen equity (or inequality) measure, units can stand for individuals or groups. For the definition of groups, one can use collective units, such as households, disabled people, non-drivers, land-use type, or regions, and characteristics, such as income, travel cost, population, or age.

4.3 Discussion The concept of social exclusion in its different dimensions has progressively become an important element in social policy discourse, often limiting its scope to the field of economic poverty and income-disadvantages. However, social exclusion represents a complex notion that includes several dimensions, including economic, that considers spatial, political, societal, personal, and temporal disadvantages, among others, which can be exacerbated by poverty (Kenyon, et al., 2002).

Bernardo Secchi (2013) looks at social classes and makes a comparison between them and the necessity to have a balanced transport system between the rich and the poor which he relates to the justice and injustice of transport. He speaks about society wanting justice but never acknowledging the injustice that is out there and the injustice on the poor. Stating that policies are in general discriminatory on the basis that a project can have an excellent environmental / financial report and therefore get funding but it would still miss the social requirements to make it successful. It is important to appreciate that project funding will help improve the stated catchment area but in turn it can cause a greater gap between the rich and poor, by excluding people outside the catchment. This can be similar for all projects including mobility projects. The point Secchi is making, and relating it to this study, is that everybody should have access to public transport and that any improvements that are made should be made system wide and not isolated to a number of routes as this will create injustice in transport and society.

Kenyon, et al., (2002) state that it is assumed that people with access to a car have an increased level of mobility then those with no or minimal access. Therefore inequality is present in society regarding the level of accessibility people have depending on their mode of transport. As a result a lack of accessibility is a key component of social exclusion, influencing its many dimensions. Transport difficulties can be a key barrier to employment opportunities; education and training opportunities; leisure facilities; hospitals and social services. Lack of mobility plays a social function as it can increase isolation and separation by not allowing people to participate in society. Consequently it can be seen that public transport plays a crucial role in the equality seen in society.

It is assumed in today’s society that most people own a car, or at least have access to one. However there are people in society that do not have such access. As a result they have to rely on other modes of transport such as a bicycle, foot, public transport, or a combination of the three in order to be able to go to the same locations as people who use a car. Consequently different levels of accessibility arise for certain locations and activities depending on the mode of transport that is available. If it is determined that out of all possible locations people can visit, car users have a greater level of accessibility then people that do not have a car then they could potentially experience social exclusion (Kenyon, et al., 2002). This means that for example bus users could have a lower level of accessibility purely because it is the only mode available to them.

A counter argument to the statement ‘car users have a greater level of accessibility’ is that the independence gained from having a car also creates a related dependency on the car, thus creating a paradox. The moment when traffic problems become too great, too expensive or environmentally damaging to drive a car the paradox will become clear and the car will be seen as a problem that one will attempt to be free from (Jensen, 1999).

4.4 Conclusion The purpose of this section was to outline why transport equality is important. It was identified that low density urban development from the 1960s has made providing public transport difficult which caused a reduced level of accessibility and social inclusion. It was also found that are many dimensions to equality. Social exclusion, poverty and accessibility have all been discussed in literature which aims to give thresholds on individual areas but equality aims to go further and looks at the difference between these levels and tries to reduce this by achieving a higher and fairer level of transport provision. It can be concluded that transport equality is important to reduce the level of social exclusion, balance accessibility and to ensure equal access to opportunities.

However the majority of appraisals for public transport proposals do not look at the equity benefits to society. They mainly focus on financial based appraisals as found in the previous chapter. Therefore the reason used to improve an area is the one factor they do not actively and precisely measure. The next chapter determines the components necessary to formulate an equitable transport appraisal method.

5 Components necessary for an equitable Transport Appraisal Method

Now that it is understood why equitable transport is important and that there does not yet exist a working equitable transport appraisal method the purpose of this section is to understand and define the components necessary in order to measure the equity of transport proposals. These components are discussed in this chapter. This answers the forth research sub question ‘What components are necessary for an equitable transport appraisal method?’ In order to successfully answer this question first it is necessary to determine the category of transport equity to work with, which theory to apply to increase transport provision, then determine accessibility measures and ensure that the accessibility measures meet the policy agenda identified through the chosen theory.

5.1 Categories of transport equity Determining equity is difficult because there is no standard definition of distributional equity for transportation benefits (Martens, et al., 2012).There are a number of equity types and principles categorised by Delbosc & Currie (2011), Thomopoulos, et al., (2009) and van Wee & Lucas (in press) including horizontal, vertical equity and territorial equity, territorial cohesion, transport users should pay their way, individuals that are negatively affected by policies need to be compensated, egalitarianism, solidarity and finally spatial and social equity.

In general, most transport infrastructure projects address spatial, horizontal, vertical, environmental, social, intergenerational and regional equity (Thomopoulos, et al., 2009). Delbosc & Currie (2011), Foth, et al., (2013) and van Wee and Lucas (in press) identify two general categories of transport equity horizontal and vertical equity. Consequently these two equity forms are looked at in more detail.

5.1.1 Horizontal equity Horizontal equity (fairness) is concerned with providing equal resources to individuals or groups considered equal in ability (Delbosc & Currie, 2011). A horizontal equity perspective emphasizes the importance of treating people in equal circumstances equally. For example, a planner assuming a horizontal equity philosophy would suggest that all people living within a community deserve equal access to public transportation. When selecting projects, this planner would try to distribute projects evenly throughout the community without any group receiving a disproportionate amount of the burdens or benefits of transit projects. One of the major problems with the horizontal equity approach is that it fails to address or even consider existing inequalities (Bertolaccini, 2013).

5.1.2 Vertical equity Vertical equity (social justice, social inclusion) is concerned with distributing resources between individuals of different abilities and needs (Delbosc & Currie, 2011). A planner

following a vertical equity philosophy would believe that disadvantaged populations, such as lower income families or ethnic minorities should receive priority consideration in public transportation projects (Bertolaccini, 2013).

These two types of equity usually conflict; if vulnerable groups are being prioritized then everyone is not being treated equally.

5.2 Accessibility based equity analysis -two ethical principles Khisty (1996) (cited in (Thomopoulos, et al., 2009) recommends that the six equity principles form a starting point:

 |Egalitarian,  Equal shares distribution (benefits equally distributed to all groups),  Rawls’ principle (benefiting more the lowest income group),  Maximising the average net benefit with a minimum range of X units,  Maximising the average net benefit with a minimum floor of X units,  Maximising the benefit for the community as a whole (utilitarian approach).

Where any of the principles or objectives are not relevant to the particular project, they will be assigned low (or zero) weights by stakeholders and then not considered further in the analysis. van Wee and Lucas (in press) propose adopting an approach that brings together accessibility based equity analysis underpinned by two ethical principles of egalitarianism and sufficientarianism. The reasons being, that they are seen to be more suited to redistributing transport resources toward currently disadvantaged population groups and areas. In order to do this the authors went one step further and identified ‘socially relevant accessibility impacts’ (SRAI), which take the form of accessibility measures and indicators for egalitarian or sufficientarian theories. Consequently these two theories are looked at in more detail.

5.2.1 Egalitarian Theory This theory is useful to legitimate the evaluation of the (in) equality of accessibility, so directly accessing unemployed to employment areas rather than over all journey time reduction on whole population and therefore (improved) accessibility to basic destinations (Van Wee & Geurs, 2011) (van Wee & Lucas, in press). Secondly it encourages a focus on the relative level of accessibility between different social groups (van Wee & Lucas, in press).

5.2.2 Sufficientarian Theory While Egalitarian theories focus on the difference between people, sufficientarianism assumes that everybody should be well off up to a certain minimum threshold. The lower their welfare the more important the policy priority is. This concept provides an ethical justification to determine the minimum threshold levels of accessibility to key destinations – below which people are considered to be socially excluded (Lucas, 2012).

By understanding these two theories it is possible to determine the best way to increase public transport provision in a city, e.g. by substantially improving the journey time on a small number of lines, improving the base line travel time on all lines, or by decreasing the travel time to the transport locations and making the journey by foot / bike more accessible.

5.3 Socially Relevant Accessibility Impacts Since socially disadvantaged groups should receive some special attention in transportation planning, defining these groups is a very important step. A common way to define these groups is by using a social indicator. Social indicators identify underprivileged groups lacking access to goods and resources compared to the rest of the society at large (Townsend et al., 1988, cited in Forth, et al., 2013). Van Wee and Lucas (in press) developed an evaluation framework for socially relevant accessibility impact (SRAI) based upon the principles of egalitarianism and sufficientarianism.

5.3.1 Accessibility measures Van Wee and Lucas (in press) state that an appropriate list of accessibility measures need to be determined to insure that social inclusion is secured. They state that four components of accessibility exist that are important for policy makers to consider when evaluating a transport system. They are ordered below to emphasize the importance of the components for discussion on social exclusion:

 Individual component (What do people need?)  Land use component (Where are locations of activities needed?)  Transport component (How to get there?)  Temporal component (When to do there?)

They are further illustrated in the figure below.

Figure 7 Relationship between components of accessibility

Geurs and van Wee (2004) state that these four accessibility components need to be looked at from four different measures; infrastructure, location, person and utility based measures. These measures can be used as social indicators if they show the availability of social and economic opportunities for individuals and social equity impacts. Together they form the matrix:

Table 1 Perspectives on accessibility and components

It’s only by assessing these issues in society that increasing the level of accessibility and social exclusion is possible.

5.3.2 Indicators

Egalitarianism and accessibility measures Accessibility should be measured for the activities with the most relevance to this policy agenda, including employment, education, health and welfare services, etc.

Indicators A way to measure journey time is for all people living in a certain area to a (selection of) destination(s) that are assumed to be most relevant from a SRAI perspective (van Wee & Lucas, in press).

Sufficentarianism and accessibility measures Threshold values for accessibility need to be defined below which it can be assumed individuals will be socially excluded. This is difficult as social exclusion is related to the norms of society. Related to this is whether minimum standards should be based on the opportunity to participate in activities or revealed levels of participation.

Indicators By defining a threshold value for accessibility from which it can be stated that everyone below this threshold is socially excluded. Van Wee and Lucas (in press) state that this in itself is a challenge as the condition of being socially excluded is relational.

5.4 Conclusion Going forward the principles of vertical equity using the egalitarian theory and corresponding SRAIs will be used in this study. The reason for this is that vertical equity is concerned with distributing resources based on the different needs and abilities of individuals while horizontal equity is concerned with distributing projects evenly throughout society therefore failing to address existing inequalities. As this study is concerned with the equity effects of introducing TSP on society vertical equity is considered more relevant as the different needs and abilities of individuals can be assessed and TSP can be implemented in socially disadvantaged areas with the aim of reducing the gap between socially advantaged and disadvantaged areas. Using the egalitarian theory these socially disadvantaged areas can be assessed in their accessibility to employment areas and improve their accessibility to basic destinations through the introduction of TSP. Sufficientarianism fails to look at the differences between groups but says that everyone should be well off up to a certain minimum threshold and therefore is not used in this study.

In order to test these theories a socially advantaged area and socially disadvantaged area will be found to test TSP.

6 Introducing an equitable Transport Appraisal Method

Chapter 5 looked at the theory and components required for an equitable transport appraisal method. Vertical equity, the egalitarian theory and corresponding SRAIs were found. The purpose of this chapter is to introduce a (statistical) method that incorporates these components in order to measure the level of equality in socially disadvantaged (and advantaged) areas, explain how the method works, and finally outline current fields of application of this method. This chapter answers the fifth sub question which is to formulate an equitable transport appraisal method.

6.1 Lorenz curve and Gini index Having once identified a list of accessibility measures and related theory the next step is to identify the distribution of accessibility between different social groups. The Gini coefficient or index (Gini, 1936, cited in (van Wee & Lucas, in press)) is a widely accepted statistical technique with which to express the distribution of an item over a group of people, the dominant application being the distribution of income over the population of a country (e.g. (Weymark, 1981). The Lorenz curve shows the distribution of income over the population. The Gini index is the area between the line of equal distribution and the Lorenz curve, divided by the triangle covering the x-axe, the y-axe and the line of equal distribution. The Lorenz curve [Gini 1912 pg. 7 cited in (Bertolaccini, 2013)] plots the cumulative proportion of the population, ordered from lowest to highest income, against the cumulative proportion of income earned.

푛 퐺1 = 1 − ∑푘=1(푋푘 − 푋푘−1)(푌푘 − 푌푘−1) Equation 1 Gini index

퐴 − 퐴 퐺푖푛푖 = 푒푞푢푎푙 퐿표푟푒푛푧 퐴푒푞푢푎푙

Equation 2 Gini index from Lorenz curve and line of equality

Figure 8 Lorenz curve and Gini index The mathematical calculation of the Gini index is complex (Equation 1), but can be approximated using the following formula where Xk is the cumulated proportion of the population variable, for k = 0, . . . , n, with X0 = 0, Xn = 1 and Yk is the cumulated proportion of the public transport service variable, for k = 0, . . . , n, with Y0 = 0, Yn = 1 (Delbosc & Currie, 2011).

Equation 2 is used to calculate the Gini index from the Lorenz curve and line of equality.

Where ALorenz is the area underneath the Lorenz curve and Aequal is the area underneath the line of equality. Because the axes are scaled from 0 to 1, Aequal will always equal 0.5. A perfectly even distribution of supply would result in a Gini index of 0 while a perfectly unequal distribution of supply would result in a coefficient of 1.

The above methodology describes the process visualizing egalitarianism through the Lorenz curve. The Gini index is then an indication for the level of (in)equality of the accessibility indicator (journey time). The process for visualizing sufficientarianism seen in Figure 8, where the vertical line x represents the threshold value. The right side of this value represents people who are above the threshold and to the left the people who are below the threshold and therefore socially excluded. The corresponding x axis value indicates the percentage of the population which are socially excluded (below the threshold).

6.2 Common fields of application

This method was chosen as it has many previous applications both in and outside the field of transport which are summarised below. The most common and well known application of the Lorenz curve and the Gini index is in showing the income distribution of a country.

 Health - Gonzalez et al (2008) determined a relationship between social exclusion and health inequality using the Gini Coefficient in Mexico.  Education - a study conducted by Halffman and Leydesdorff (2010) applied the Gini index to university rankings in order to assess whether universities are becoming more unequal, at the level of both the world and individual nations.  Port activities – Gonzalez-Cancelas et al (2013) used the Gini Index and Lorenz curve to show the Spanish Port System by type of goods from 1960 to the year 2010 for business units.

Using Lorenz curve and Gini index as a transport appraisal method:

 Road network design – Santos et al. (2008) selected three different equity measures (one of which was the Gini index) and incorporated them into an accessibility- maximization road network design model.  Ramp control schemes – Tian et al. (2012) investigated the efficiency and equity of morning peak ramp control schemes in a freeway corridor with limited capacity in china.  Roadway Tolls – Franklin (2005) examined the distributional effects of transport policies, using a bridge toll as a case study.  Public Transport Services – Delbosc and Currie (2011) assessed the overall social exclusion effects of the public transport service in Melbourne, Australia.  Transit Supply - MSc student Bertolaccini (2013) assessed the equity of transit supply distribution in metropolitan areas and the effects of using different scales, levels of

data resolution, and various demand measures when calculating Gini scores for interregional comparisons in the USA.  Transport access among affordable housing units – Welsh (2013) develops a comprehensive method to quantify the quality of service and accessibility at each transit node in a network, combined with an index to measure the inequity (concentration of quality service) at the micro scale. These measures are applied to the distribution of all residential housing units to determine if the subsidized housing programs are achieving major policy objectives of providing equitable transit access to vulnerable groups (USA).

Any equity analysis that has been conducted as been done so after the proposal or policy has been implemented. No research could be found that deals specifically with the equality of (public) transport proposals on socially advantaged and disadvantaged areas and different mobility users, which is why the researcher is focusing on it for this study.

7 Identifying socially (dis)advantaged zones in study area

Up until now this study answered the first two phases which comprise of the first 5 research sub questions. What Phase Three aims to achieve is to appraise the introduction of TSP to the study area using the previously outlined method and theories. This and the next three chapters are concerned with answering the final research sub question ‘what are the travel time and resulting equity impacts from implementing TSP?’ In order to do this a recap of the equity categories, theories and accessibility components is made relating them specifically to the study area to identify two transport coriddors that have a contrasting demographic to be able to appropriately assess the benefits of TSP. Finally the extent to which TSP might benefit the study area is shown through an initial analysis. A detailed methodology can be found in appendix A.

7.1 Introduction Below the components in chapter 5 are explained and related to the study area. This ensures there is a clear focus on equity when assessing the results of introducing TSP.

 Vertical equities key point is ‘the distribution of resources between individuals of different abilities and needs’. This is applied to the case study by defining areas of differing socio-economic profiles and ensuring equal travel time between social classes. This study is looking at vertical equity by identifying two areas with different social classes and applying the same level of TSP provision. This provides a comparison analysis when assessing the equity benefits of TSP using the Lorenz curve and Gini index.  Egalitarian theory requires accessibility to basic destinations. This is applied to the study area through access to main employment and / or public transport transfer nodes. This theory is met by ensuring that the areas of differing socio-economic profiles have access to the same employment locations and / or public transport transfer nodes. o Indicator: travel time for all people in the contrasting social areas.

Next it is necessary to identify how the four accessibility components can be applied to the study area. This was achieved through the use of 2011 Census data, annually published Quality Bus Corridor monitoring reports, Dublin Bus AVL data and desktop analysis.

 Individual – vehicle ownership, means of travel to work, school, and college, principle economic status, and social class area all considered. Preferably the two locations will differ in the majority of these classifications.  Land use – two residential areas of demand with different socio-economic characteristics, restrictions to geographical area include inside the M50 bounding the North, South and West of the study area, to the East is the Irish Sea. For land use supply – a high concentration of employment, retail, schools, and colleges are required.

 Transport – both demand and supply land use locations must have a direct bus route using a QBC link.  Temporal – transport must have a frequency of 15 minutes or more and must run throughout the day. Shops and services at the destination location must accommodate travellers throughout the day.

Having identified the above components it is possible to get a clear outcome of the actual equity results of introducing TSP seen later on in the study.

7.2 Individual component

Population by social class

Figure 9 Professional workers and Unskilled workers

There is a high density of Professional Workers raditating to the South-East, East and North East of the city, indicated in dark green (Figure 9 left), while the Unskilled Workers, seen in Figure 9 to the right, are located to the West and North. For this a clear devision of classes is visible.

Population aged 15 years and over by principal economic status

Figure 10 Students and Unemployed having lost or given up previous job

It can be seen that Students are located to the South / South-East (Figure 10 left), while the Unemployed having Lost or Given Up Previous Job are located from the West to the North (Figure 10 right) with distinctive zones in dark green comprising of the largest unemployment areas.

Number of households with cars

Figure 11 Households with no cars

Along the South and South-East of the study area there is a clear boundary visible where there are households with more then zero cars, indicated in light green. This corresponds to the location of students and professional workers (in the previous figure). This is in contrast with the rest of the study area where the West sees a higher number of households with zero

cars (dark green) corresponding to Unskilled workers and Unemployed having lost or given up previous job.

Population aged 5 years and over by means of travel to work, school or college

Figure 12 Means of travel to work, school and college by bus, car passenger and rail respectively

Including Figure 11 with Figure 12, electoral districts (ED’s) to the West have a high number of ‘households with no cars’ and high bus use (left image); the South-East has a low number of ‘households with no cars’ but still a high bus use. Both the West and South-East of Dublin have similar bus use, with the West slightly higher, the South-East has a higher level of car access which relates to the higher level of car ownership (middle image).

Rail based transport (Luas - pink line, DART-blue line along the coast) have strong catchment areas (right image). However due to the limited number of rail based lines and the minimal overlapping catchments with bus based modes rail based transport is not considered any further. It is worth noting that these lines have a highly concentrated use by the surrounding population. The mean of travel ‘car drivers’ is not shown above as there is an even distribution over the whole study area so not much differentiation can be determined5.

5 Can be seen in the appendix B 30

7.3 Land use component The highest employment areas are in the city centre and in isolated ED’s on the circumferance of Dublin. Highlighted employment areas on the circumference are industrial estates to the North, South and West accommodating a mixture of industries.

Figure 13 Employment and retail locations

7.4 Individual and Land Use Conclusion This section has identified two areas that meet the requirements set by the accessibility measures outlined at the start of the chapter. Two areas with different land uses (to the West and South East) have been identified that have different socio economic profiles. These areas have differing (individual component) social classes, economic status, number of households with no cars, and means of travel to work, with the South West having a higher socio-economic demographic then the West. With regard to the Land Use component the highest area of employment and retail locations is the city centre. Next it will be determined if these areas have the required transport and temporal components to link the origin and destination zones. To conclude the North Clondalkin zone to the West is the socially 1 3 disadvantaged area and the Stillorgan zone to the South East is the socially advantaged area. These two areas will be used in the following chapters as a comparison 2 between the provision of transport, travel times and modal split of different socio-economic zones within a study area and therefore to determine the level of equity within and between these zones. Figure 14 Origin & Destination in study area

1: North Clondalkin zone (origin) 2: Stillorgan zone (origin) 3: Dublin City Centre zone (destination)

7.5 Transport component

Transport provision in study area To understand the layout of the Dublin public transport system, below is the Frequent Transport Map of Dublin indicating Dublin Bus, Luas and DART routes that operate every 15 minutes or more often at peak travel periods.

Figure 15 Dublin Frequent Transit Map and Road network Source: (Broderick, 2014) and Ordnance Survey Ireland

The majority of Dublin Buses network consists of a combination of mixed traffic and exclusive bus lanes called Quality Bus Corridor’s (QBC). There are 16 QBC’s in Dublin radiating out from the city centre. Dublin has a clear radial structure, the majority of which is built up of 110 radial, cross-city and peripheral routes provided by Dublin Bus as well as two Luas (light rail) lines and one DART (heavy rail) line. This can be seen in the left figure above.

A similar radial structure is present in the road network Figure 15 (right figure). Upon crossing the two canals North and South of the city with a total of approximately 17 bridges (which effectively makes the city an island) there are differing traffic restrictions in the city including one way streets, counter-flow bus lanes, pedestrian only streets, and speed restrictions down to 30kph. Located in the city are a number of multi storey car parks as well as on street parking all of which are paid.

The two areas identified in the previous section have a QBC to the city centre Stillorgan QBC to the South East and North Clondalkin QBC to the West, both rail and light rail are outside the catchment6 area of the QBCs.

6 Defined as 600metres 32

Travel time Table 2 AM and PM comparative bus and car journey times (mins)

Dist. Bus Ave. Car Ave. Time Corridor Section measured Am/Pm % Diff (km)7 Journey time journey time difference Coldcut Rd. to 7.4 AM 37 23 14 -39% North Cornmarket Clondalkin Cornmarket to PM 47 31 16 -35% Corncut Rd. Mount Merrion 5.5 AM 19 21 2 9% Ave. – Lesson St. Stillorgan Leeson St. – PM 17 26 9 50% Stillorgan Rd

Upon analysis of Table 2 it is seen that when comparing bus travel times with car travel times the bus on the Stillorgan corridor is faster than the car, however for the North Clondalkin corridor the bus is slower than the car both for the AM and PM peak (QBC monitoring report, 2011).

7.6 Temporal component

Initial bus route analysis in chosen study areas Table 3 Number of bus route on the North Clondalkin and Stillorgan QBC’s

Corridor Dublin Bus route numbers North Clondalkin 18 40 13 68 123 76/a 79a 46a 84x 7d 7b 67x 66x 51x Stillorgan 39 32x 25x 145 118 116 11 46e 41x 39a 77x 47 75

Table 3 above shows the number of bus routes on each corridor. While a number of bus routes run on each corridor there are only three that run the entire length from the origins to the destination, that run all day, and are considered high frequency (operating every 15 minutes or more often at peak travel periods) they are route 40 on the North Clondalkin corridor and both route 46a and route 145 on the Stillorgan corridor. Dublin Bus and the NTA use the 40 bus route on the North Clondalkin corridor and the 145 route on the Stillorgan corridor when conducting studies and reports. Consequently this study is using the same routes for comparison purposes.

7 Measured through maps.google.com and geodistance.com 33

7.7 Scope to introduce TSP The purpose of this section is to see if TSP can be introduced on these routes. A desktop analysis was conducted as well as information gathered from NTAs GDA model (described later on in the study) to find the following information.

Table 4 Number of signalised junctions and length of studied bus route

No. signalized junctions Length of QBC North Clondalkin QBC 25 9,7 km Stillorgan QBC 39 15,7 km

There are a number of signalised junctions on both routes which can be seen in the table above as well as the length of both QBCs. The signalised junctions on both routes comprise of vehicle traffic signals and pedestrian signals with four pedestrian signals on the Stillorgan QBC and five on the North Clondalkin QBC.

Junction dwell time

100% 17 90% 25 21 20 23 25 23 80% 33 10 16 70% 8 16 13 11 13 60% 15 50% 40% 69 30% 67 64 64 67 64 64 52 20% 10% 0% Thurs Fri Mon Tues Wed Thurs Fri Average

Running time % Total junction dwell % Total boarding and alighting dwell %

Figure 16 Stillorgan QBC dwell time analysis

A study conducted by O’Connor and Kavanagh (2014) summarised in Figure 16 shows that junction dwell time accounts for 13% of all journey time in a survey that was undertaken on the Stillorgan QBC. There were 32 signalised junctions along the Stillorgan QBC in the study ranging from signalised junctions to pedestrian crossing (their study is marginally shorter than then the being undertaken in this study). With the average dwell time per junction of 7 seconds this equates to an average junction dwell per journey of 3,3 minutes. No studies were conducted for any other QBCs.

Referring to Table 5 below it is seen that there are delays at junctions on the North Clondalkin QBC which could be mitigated with the introduction of TSP.

Location of unscheduled stops Table 5 represents delays encountered on North Clondalkin’s QBC. Delays are registered when the doors are closed, the bus is not moving, and the bus is not at a bus stop. The delay analysis was conducted during the AM peak 0700-1000 in November 2013. The plus or minus after the stop name indicates if the delay was registered before or after the stop. The highlighted rows have a traffic light that could be causing the delay; this was determined using desktop analysis (maps.google.com and geodistance.com).

Table 5 List of unscheduled stops

No. No. Ave. Stop time Rank Stop Name8 Stop No. Pct % Ave. Distance Stops Trips (sec) 1 Emmet Road South Circular Road (+) 01992 226 562 40 67 47 2 Ballyfermot Rd Drumfinn Road (+) 02696 263 928 28 46 128 3 Ballyfermot Rd Blackditch Drive (+) 02689 143 669 21 49 73 4 Emmet Road Camac Close (-) 01989 127 623 20 52 -135 5 Ballyfermot Rd Drumfinn Road (-) 02696 122 928 13 52 -171 6 … … … … … … …

There are 21 unscheduled stops along the North Clondalkin QBC, Table 5 is an extract from this data set. The purpose of showing this information is to show that delays are present on this corridor.

8 For the full list of unscheduled stops for the North Clondalkin route see the appendix C 35

Automatic Vehicle Location analysis The purpose of this section is to show that there is a variance in travel times throughout the day as well as between each AVL segment. This shows that there is the scope to introduce TSP to reduce the travel times and increase reliability.

North Clondalkin Route 40 30 25

15 Kph 10 5 0

Inbound Outbound Threshold

Figure 17 North Clondalkin Route 40 18 hr period Kphs

Stillorgan Route 145 30

15 Kph 10

Inbound Outbound Threshold

Figure 18 Stillorgan Route 145 18 hr period Kphs

Figure 17 and Figure 18 above shows the inbound and outbound travel time for 18 hours of scheduled operation of the two bus routes. Hour periods are on the X axis and speed is depicted on the Y axis. The AVL locations for route 40 are from Cherry Orchard Hospital to Emmet Road - South Circular Road and for the 145 route from Westminister Road to

Morehampton Road - Morehampton Terrace. This study area is slightly smaller than the one being conducted in this study.

During the AM peak inbound, North Clondalkin’s Dublin Bus route 40 has a speed of 12,6kph while Stillorgan’s Dublin Bus route 145 has an average speed of 14,4kph, these two QBCs are ranked 6th and 9th slowest out of 17 respectively. Outbound the North Clondalkin’s Route 40 has an average speed of 13,7kph while Stillorgan’s route 145 has a speed of 18,7kph, these two QBCs are ranked 6th and 13th slowest out of 17 respectively. It can be concluded that both routes do not meet the average speed threshold set by the National Transport Authority of ≥22kph.

7.8 Conclusion There is a balanced population density throughout the city, but an unbalanced distribution of social class, mode of transport in travel to work, school, college, and car ownership. All of these increase the societal gap between residential zones.

Both the socially advantaged and disadvantaged corridors have access to public and private transport modes however the travel times for these modes on the corridors differ. The North Condalkin corridor has a faster car travel time then the bus while the Stillorgan corridor has a slower car travel time then the bus. It can be seen that there is an imbalance of bus travel time over the entire day on and between the two routes and both routes do not meet the required travel speed of 22kph on the QBC set by the NTA. Therefore the implementation of TSP would aim to improve this situation by taking a step to towards reducing the travel time differences between the car and bus on the North Clondalkin corridor and increasing the average route speed on the two bus routes.

It can be concluded that the two zones have different land use, transport, temporal, and individual component characteristics. These socially contrasting zones have different levels of equality and therefore it is possible to model the equity affects that introducing TSP has on these zones. The next chapter will model TSP on these two corridors and determine the results. These results will be equitability appraised using Lorenz curve and Gini index and finally it will be determined if introducing TSP is financially feasible.

8 Modelling Transit Signal Priority

The last chapter found the socially advantaged and socially disadvantaged origins, an employment and education destination and corridors linking the origins to the destination. The aim of this section is to generate travel times for base case and TSP case for highway (HW), public transport (PT) and active soft modes (SM) as well as their corresponding modal splits. These results are then input for the next chapter which determines if these changes are equitable, as well as for the CBA appraisal method for the following chapter.

8.1 Summary of model9 The NTA’s Greater Dublin Area (GDA) transport model is a strategic multi-modal, network based transport model covering the Greater Dublin Area (i.e. the counties of Dublin, Meath, Kildare, Wicklow and Louth). The GDA transport model is owned by the NTA, who are the authority responsible for its maintenance and use.

Zoning: All 666 GDA zones are used covering the entire modelled area however this study is only concerned with the zones within Dublin County.

Model Periods: Only the AM-peak period covering the three-hour period from 0700-1000 is used as this is the most concentrated travel period, the other period is the inter- peak 1400-1500.

Base and Forecast Years: The base year for the current peak model is 2006, while the main forecast year is 2030. The 2006 Base year is used.

Modelled Networks: The model contains coded networks for all mechanised modes of travel – including car, goods vehicles, bus, heavy rail, LUAS and Metro. The car and public transport networks are considered. Goods vehicles are not considered; while they could be effected they are not used for work (commuting) trips and they have an alternative routes for crossing the city as they are not allowed in the city centre (HGV restrictions 0700-1900, 7 days a week) (Dublin City Council, 2007).

Travel Demand: In the case of both the AM-peak and Inter-peak models, travel demand is broken down by six journey purposes – i.e. Work (commuting), Education, Employer’s Business, Shopping, Other and Non Home Based. Travel demand is further segmented by two person types – i.e. those with a car available for their trip (Car Available), and those without a car available for their trip (Car Not Available). For AM peak only Work trips are modelled.

9 The components of the GDA model are explained further in the appendix D

8.2 Limitations and considerations

8.2.1 Time limitations The model was only available for seven and a half days, in which it was necessary to understand both how the model worked and how to use the software. Another time limitation was that it took, on average two working days to run the model if it was unable to be set up to run overnight. Time was also a limitation in terms of the opening hours for the office, 11 ½ hours a day. That meant that the model had to be set up and started before closing otherwise the next day was wasted running the model. The number of iterations in the mode choice, hour of travel choice and route choice / PT assignment was limited to two iterations due to these time constraints.

8.2.2 Data limitations Regarding gaining access to data there were time delays due to the NTA’s own commitments. However once access was granted there were no limitations to data. It was the case that there was so much data available that the time spent using the model required a clear timetable and output requirements which had to be lucid, concise and vigilantly adhered to.

8.2.3 Model considerations QBC link segment sees the speed 8, 15, 22 or 30 kph, or Max {QBC speed, prevailing car speed}. The QBC speed depends, in part, on the location of the QBC in the GDA. In the outer suburbs the speed might be 30kph and the city centre might be 8kph. Whole QBC corridors might not have consistent exclusive lanes due to limitations on road space.

 One option was to increase the travel time on the defined corridor to Max {22pkh, prevailing car speed} regardless if there is a QBC corridor on a link or not. This means that changes made to favour the bus also favours the car.

The second model consideration is how the traffic light sequences are provided for in the model. As mentioned in Chapter 2 within County Dublin there are four county councils each responsible for traffic management. There are two traffic management systems used (SCATS, SCOOTS) by the four County Councils. SCATS is used by Dublin City Council, Fingal County Council and Dun Laoghaire / Rathdown County Council. SCOOT is used by South Dublin City Council. It is assumed in this study that only one system is used, SCATS. Within the SCATS system there are approximately eight different sequences available in the traffic system to implement depending on the time of day, congestion of the links, and policy aims of that County Council. The GDA model runs the AM peak (0700-1000) period and cannot provide a dynamic traffic light sequencing system. This results in the average green and red time for a junction to be coded as input into the model.

 A second option to test TSP is to increase the green time on the junctions on the corridors to simulate what might happen if TSP was introduced. This affects all traffic on the road including cars, cyclists, buses, goods vehicles and pedestrians.

The third consideration is that this model covers the whole GDA, a population of 1.6 million people. This has consequences for the results as any modal split or travel time gains generated from TSP will be spread and therefore diluted over the whole population and not on the two corridors identified in this study. This has the potential to produce results that show no changes to modal split.

8.2.4 Active soft mode considerations Soft modes of transport comprise of cyclists and pedestrians. The majority of the soft mode users are pedestrians; cyclists only represent approximately 4% of the GDA modal split. This needs to be taken into consideration when analysing the results further on in the study as the origin zones are 9,7kph and 15,7kph away the destination zones and assumed too far for pedestrians to walk for commuting purposes.

8.3 Strengths Strengths of the model are outlined below.

 The highway assignment stage takes full account of junction delays caused by congestion, and thereby produces a realistic representation of car and bus journey times on the road network,  The model includes a high level of segmentation of trip makers and their journey purposes,  Travel behaviour as represented in the model is based on comprehensive and detailed travel surveys and travel datasets,  The model covers the full GDA, and takes full account of travel within, into and out of the modelled area,  As the model is also used as the basis for scheme evaluation, the transport networks represented contain a level of detail beyond that which would be required for its use as a strategic transport planning tool,  To enhance its functionality, the GDA transport model includes an additional stage (“hour of travel choice”) in the modelling process. This additional stage is used to represent the phenomenon of peak spreading as a response to congestion and is not captured in many strategic models of this kind.

Another strength of the model is the level of interaction between the public transport and highway assignment. The bus lanes and QBC speeds are modelled fairly well. The QBC lanes taken have been removed ahead of time from the network, from the car point of view. For straight ahead movements on a QBC the bus sees the speed 8, 15, 22 or 30 kph. The car sees one less lane and junction delay. If green time increases (with the introduction of TSP), and there is a bus lane then the bus sees Max {QBC speed, prevailing speed}, and creates a delay file which includes junction delay.

 A third option to test TSP is to remove the delays seen by the two PT routes on the corridors. This involves manually modifying the model while it is running by

stopping the model at the point where the delay files are created for the three PT assignments time periods, identifying the links and manually changing the delay values to see no delay.

8.4 Final plan It was decided, through consultation with the NTA that applying all three options would not generate the best results especially due to the time constraints. Due to the strengths and limitations identified it was decided that a two-step approach was necessary to model TSP. From the highway side, modifying the traffic light sequences, and from the public transport side modifying the delay file. Overall the final plan is a simple intervention to mimic the effects of TSP on the network.

8.5 Determining zone numbers Having identified the start of the North Clondalkin and Stillorgan corridors as two contrasting areas both terminating in Dublin city centre, the next step is to determine their zone numbers in the GDA model. The catchment area of these corridors is not the whole corridor itself but rather two areas at the beginning of these corridors taking into consideration the accessibility components previous outlined. This enables a clear comparison of the contrasting areas. Origin zones are identified as starting at Cherry Orchard Hospital (North Clondalkin corridor) and Loughinstown Hospital (Stillorgan corridor) and 1,000metre inbound. There is a 600metre buffer either side of this area to represent the maximum walking distance of a bus user. The 1,000x600metre zone ensures a large enough population catchment. On the North Clondalkin Corridor if an ED was within the 600 metre radius but was on the far side of the M50 (to the west), the rail line (to the south) or the Chapelizod Bypass (to the north) it was not taken due to the inability to for pedestrians to cross these physical barriers.

The destination zones are identified starting where the two routes meet at O’Connell Bridge, an 800 metre buffer radius taken from this point. The buffer at the destination is slightly larger to take in the majority of the city centre which comprises of employment, retail and third level educational opportunities, and to account for users who get off at a previous or later bus stop.

Both CUBE and ArcExplorer software’s were used to insure the accuracy of the zones. The two system screens are shown below:

Figure 19 CUBE and ArcExplorer screen shots

By using CUBE, on the left (with route 145 shown), to follow the PT routes, while checking and confirming the zone numbers with ArcExplorer, on the right, it was possible to ensure that the correct zones were identified to define each corridor. A list of the 16 origin and 66 destination zones were found10.

8.6 Implementing TSP in GDA Model In order to model the effects of TSP on the network a number of changes had to be made at the route choice / public transport assignment stage in the GDA model. They are outlined below. In order to compare the Do Nothing (2006 Base Case) with the Do Something (2006 Base case with TSP modifications) first the GDA model was run in the Do Nothing state. The changes made to the modal to mimic TSP are outlined below. The trip demand was not changed in this study.

8.6.1 Highway changes In the route choice / trip assignment step of the GDA model SATURN software undertakes a detailed simulation of traffic flow through junctions on the selected routes, by calculating travel times explicitly including the travel time of delays caused by traffic congestion at junctions.

One mechanism for implementing TSP on the highway is changing the stage duration (increase or decrease) at junctions to reduce the travel time of the bus. Changing these stage durations at junctions is one way to model the effects of TSP. It is important to note that the LCY (total cycle time for the node) is not changed.

The GDA model contains coded networks for all mechanised modes of travel. In the Dublin County area full junction details are included for all major junctions (the simulation network). The simulation network is encoded in a large text file containing details for all 2,000 + junctions which are simulated. It is possible to change the signal green time to mimic TSP.

Each record holds many columns and descriptions, below are the ones relevant for this study.

10 Listed in the appendix E 42

Table 6 SATURN Record type 1

COLS. NAME DESCRIPTION 1-5 NODE Node number 11-15 JTYPE* Node type 0 for EXTERNAL NODES 1 for PRIORITY JUNCTION 2 for ROUNDABOUTS 3 for TRAFFIC SIGNALS 4 for DUMMY NODE 5 for ROUNDABOUT with U-turns 16-20 JCIR Time to circle roundabout (in seconds) (roundabout only) NSTAG OR Figure 20 Example junction information Number of stages – traffic signals only in SATURN 26-30 LCY Cycle time for this node (+) 31-35 NUC Number of time units per cycle (+)

Table 7 SATURN Record type 3

COLS. NAME DESCRIPTION 11-15 STAGL Stage duration 16-20 INTG Duration of following inter-green

In the above table’s Record type 1, identified by node number, was only considered if there was a node type 3 for Traffic Signals. Then Record type 1 would be used again to identify the cycle time for the node from which Record type 3 could distinguish the stage duration and the inter-green time, subsequently the stage duration and the inter-green time could be modified to accommodate the introduction of TSP.

SATURN test runs A number of tests runs were conducted using only the SATURN network to see the effects that changing the stage durations at the junctions would have on the delay and demand flow on the identified routes and on the whole network. By running multiple different stage duration lengths the best increase stage duration and inter-green time was determined11.

It was finalised that each node received an increase of 5 or 10 seconds with only a few nodes receiving no increases. These values were taken after consultation with the NTA, research and the level of green time already provided at the junctions. If the remaining arms on the junction had 10 seconds of green time or 4 seconds of inter-green time left no increases could be made to the junction arm on the route as 10 seconds are required for a bus to clear a junction (Embarq India, n.d.). Once this file was re-written it was saved. This file is used for each time period and iteration in the Mode choice, Hour of travel choice and Route choice feedback loop until convergence was reached. Two junction files now exist, the original Do Nothing file and the Do Something file.

11 Table of changes can be seen in appendix F 43

Public Transport changes On the public transport side the highway link travel times are passed from the SATURN assignment and used in the calculation of bus travel times; therefore the change in junction times is passed on to the buses. This means that implementing TSP will affect bus and car travel times equally. However this study aims to give the advantage of TSP to the bus over the car. This requires one more step to be completed on the public transport side by intervening and manually removing any delay seen by the bus at a junction within the two corridors. This can be done with the use of the file ‘satdelay.dat’ which is created during the PT assignment time periods 1, 2, and 3 (0700-1000). The detailed explanation of these two methods is below.

GDA Model The Main Model loop PT Assignment Time Period 1

Figure 21 GDA Model process screen shots In order to make these changes the process follows:

1. Each step in the GDA model is run individually until the Main Loop Model step, 2. Each step in the Main Loop model is run individually until PT Assignment Time Period 1 is reached, 3. Each step within the PT Assignment Time Period 1 is run individually until the creation of satdelay1.dat file, 4. Once this file is created it is opened.

This file is structured by node sequence per bus route. The 145 and 40 bus routes are located in the file, there are two tables for each bus route the first one has the junction nodes and the second one has the corresponding delay values for each junction node,

5. Identify the beginning node and end node that bound the study area in the table, 6. Identify the corresponding delay times (seconds) in the table below, 7. Change the delay times to represent zero delay at the junction.

Once the satdelay1.dat file was re-written it is saved. A copy was made of this named satdelay_iter1.dat. This process was repeated for PT Assignment TP2 and TP 3. In the end three files, satdelay1.dat, satdelay2.dat, satdelay3.dat, were created, as were their copies satdelay_iter1.dat, satdelay_iter2.dat, satdelay_iter3.dat. Once the third file was created the model was allowed to run the rest of the Assignment period, as well as the Main Model loop

uninterrupted. This is the first iteration. This first iteration was saved and results and analysis sections were run (described in more detail further on).

For the second iteration two changes were made to the Main Model loop configuration. Firstly the loop was conditioned to stop after the first iteration. Secondly in the PT Assignment Time Periods 1, 2 and 3 an extra step was added after the creation of the satdelay.dat files in order to save time running each step manually. In the second iteration the process was changed to follow: In the original PT Assignment the steps exist to create the satdelay.dat files which is then the input file name to the next step. However the satdelay_iter files was modified during the first iteration to include zero delays seen by PT between specific AB junction nodes, so an extra step in the PT Assignment is made to override the satdelay file that will be created in second iteration with the satdelay_iter file that was created in the first iteration. Both the model interface and the background code had to be modified for this to happen for each time period 1, 2, and 3. The modal interface addition can be seen in Figure 22.

Figure 22 Addition of extra step in PT Assignment

8.7 Results & analysis Once the modal had completed its two iterations, the results and analysis were run. This was broken down by mode per time period. This study was only concerned with analysis files that contained the travel time per mode and modal split.

8.7.1 Public transport results The results provided below for time period 1, 2, and 3 are in decimal fraction of minutes, while the final time period 1 – 3 (TP13) are in minutes.

Table 8 North Clondalkin public transport travel time

Study Area travel time Entire bus route travel time Ave Speed Kph

TSP 16,7 42,3 36,2 TP1 Base 22,9 48,6 31,5 Increase above base -6,2 -6,3 4,7 TSP 20,9 49,6 30,9 TP2 Base 33,0 62,2 24,7 Increase above base -12,0 -12,5 6,2 TSP 21,4 50,4 30,5 TP3 Base 34,4 64,0 24,0 Increase above base -13,1 -13,6 6,5 TSP 19,7 47,5 32,5 TP13 Base 30,1 58,2 26,7 Ave. Increase above base -10,4 -10,8 5,8 TSP 19,4 17,3 32,3 TSP13 Base 30,1 58,1 26,4 min.ss Increase above base -10,3 -10,5 +5,5

Each time period 1, 2 and 3 are broken down above. The average value for TP13 was found. Dublin Bus route 40 reveals an average travel time saving for TP13 of 10,3 minutes from the study area which covers 38% of route 40’s service distance. Taking route 40’s total service distance into account there is an average overall travel time saving of 10,5 minutes and route speed increase of 5,8kph.

Table 9 Stillorgan public transport travel time

Study Area travel time Entire bus route travel time Ave Speed Kph

TSP 25,4 44,8 38,3 TP1 Base 30,7 50,5 34,0 Increase above base -5,3 -5,7 4,3 TSP 34,1 62,3 27,6 TP2 Base 42,6 71,7 23,9 Increase above base -8,5 -9,4 3,6 TSP 33,9 60,4 28,4 TP3 Base 43,8 70,9 24,2 Increase above base -9,9 -10,5 4,2 TSP 31,1 55,8 31,4 TP13 Base 39,0 64,4 27,4 Ave, Increase above base -7,9 -8,5 4,1 TSP 31,1 55,5 31,3 TSP13 Base 39,0 64,2 27,2 min,ss Increase above base -7,5 -8,3 +4,0

For Dublin Bus route 145 the analysis reveals an average travel time saving of -7,5 minutes in the study area which covers 55% of route 145’s service distance. Taking route 145 total service distance into account there is an overall average travel time saving of 8,3 minutes and an average overall route speed increase of 4kph.

Table 10 North Clondalkin & Stillorgan bus route distance & study area distance

Route Distance (km) Study area distance (km) Percentage (%) 40 25,6 9,7 38% 145 28,6 15,7 54,8%

8.7.2 Highway results Using a skim time matrix (666x666) for each time period the average highway travel times for the study areas can be found. These tables are too large to present and so a summary is provided below. The results provided below for TP1, 2, and 3 are in decimal fraction of minutes, while the final TP13 are in minutes.

Table 11 North Clondalkin highway travel time

Cum, Travel time TP1 TP2 TP3 TP13 Ave. TP13 min,ss TSP 992,4 1155,4 1175,9 1107,9 18,3 Base 979,7 1137,4 1129,8 1082,3 18,0 Increase above base +12,7 +17,9 +46,1 +25,5 +00,3

Table 12 Stillorgan highway travel time

Cum, Travel time TP1 TP2 TP3 TP13 Ave, TP13 min,ss TSP 1566,5 1915,8 1891,4 1791,2 29,5 Base 1594,7 1936,0 1907,1 1812,6 30,1 Increase above base -28,2 -20,2 -15,7 -21,4 -00,2

The results show that there is an increase in travel time along North Clondalkin highway of +30 seconds, but a reduction in travel time along Stillorgan highway of -20 seconds.

Table 13 GDA highway network travel time

TP1 TP2 TP3 TP13 Ave. TP13 min,ss

TSP 1923,6 2627,4 2718,6 2423,2 40,2 Base 1922,3 2623,6 2709,9 2418,6 40,2 Increase above base 1,2 3,8 8,7 4,6 +00,1

The overall travel time on the GDA network was also looked at, with a GDA highway travel time increase of +10 seconds.

8.7.3 Active soft modes results In the model active soft modes are not modelled per time period as the notion of peak spreading for these modes is not considered an issue. This mode does not take part in the route choice stage of the model either, and their cost of travel is assumed to be a simple combination of travel distance and time.

Consequently their travel time is taken from Dublin cycle planner provided by Transport for Ireland12. These values are used as the Base case. For the TSP case the travel time saving percentage seen by the highway is applied to the cyclists as they (for the most part) share the same route, road provision and have the same level of priority at junctions, assuming that they obey the red light. The results are provided in min,ss format.

Table 14 North Clondalkin and Stillorgan soft mode travel times

North Clondalkin Stillorgan

TSP 35,5 61,2 Base 35,0 62,0 Increase above base +0,5 -0,40 Route length Kph 8,9 16,1

The results are similar to the highway network with North Clondalkin route gaining 50 seconds while Stillorgan route loses 40 seconds.

12 Screen shots can be seen in the appendix G 48

8.7.4 Modal split results The level of modal split was taken at three levels, origin level, destination level and GDA level. The reason for this was to ensure a thorough analysis of modal split was made as the GDA model is conservative in its model split estimation.

Table 15 Modal split percentages

HW % PT % SM %

TSP 41,83 24,25 33,92 North Clondalkin Base 41,84 24,21 33,95 (origin) Increase above base -0,01 0,05 -0,03 TSP 54,82 21,47 23,71 Stillorgan Base 54,83 21,47 23,70 (origin) Increase above base -0,02 0,01 0,01 TSP 31,8 45,12 23,1 City Centre Base 31,82 45,10 23,10 (destination) Increase above base -0,012 0,017 -0,006 HW: Highway TSP 51,76 22,15 26,09 PT: Public Transport GDA Base 51,76 22,14 26,10 SM: Soft Mode TSP: Transit Signal Priority Diff above base 0,00 0,01 -0,01 GDA: Greater Dublin Area

The next table converts the percentage modal split into actual number of passengers. The percentage change above equates the following number of people for destination and GDA only.

Table 16 Modal split total user increase above base

TP13 PT Orig PT Dest HW Orig HW Dest SM Orig SM Dest North Clondalkin +7 - -1 - -4 - (origin) Stilorgan +1 - -1 - +3 - (origin) City Centre - +43 - -16 - -6 (destination) GDA +58 - +72 - -12 -

Origin results North Clondalkin and Stillorgan origin modal split (approx. 8EDs each) seen in Table 16 represents people leaving the origin study areas only. For North Clondalkin there was a -0,01, +0,05, -0,03% change for HW, PT and SM respectively. For Stillorgan there was a -0,02, +0,01, +0,01 change for HW, PT and SM respectively. These changes are negligible.

Destination results Looking at the destination (approx. 66 EDs) there are an extra +43 PT passengers, -16 HW users and -6 SM users (Table 16). This equates to a +0,02% increase for PT, and a reduction

of -,008% for HW use and -0,003% for SM use. This results in an extra 21 people travelling during the AM peak on the two routes with the introduction of TSP. The city centre destination (66 EDs) represents people entering the destination study area. These values are different from the origin values as they take into account the effect of TSP being introduced at corridor level. These can be seen in Table 15.

GDA results Minor modal split percentage changes, +0,01% increase in PT and -0,01% decrease in SM, with no change for HW. Value wise there are +72 more people travelling by PT, -58 less people travelling by HW, -12 less by SM. These can be seen in Table 15 Table 16.

For the two areas there are more people leaving by PT and fewer leaving by HW. In North Clondalkin there are fewer people leaving by SM but in Stillorgan there are more people leaving by SM.

For the destination zones in the city centre, it is seen that more people arrive by PT and fewer by HW and SM. The larger values then the origin zones include benefits seen in the entire study area.

8.7.5 Sub PT modal split results Results above showed that there were only minor changes in the modal split between HW, PT and slow modes. This was a possibility as the mode shares for the whole GDA would have hugely diluted the results considered on a corridor basis. The model is also very conservative in the probability of switching from HW to PT.

The table below (TP13) shows a change in sub-mode choice within PT with an increased passenger distance being seen in the bus sub-mode and a decreased passenger distance being seen on the other PT sub-modes. This suggests that there are more transfers happening for example from regional bus to city bus within Dublin County. Passenger distance is the main attribute used to compare sub-modes as this shows that passengers are either travelling further on the mode and / or more passengers are using that sub-mode and can be directly related to an increase in farebox revenue.

Table 17 Sub-modal split increase above base

Pax, Pax, Boarding’s / Pax, Time Trip Length Pax, Km Pax, Time Mode Name Distance Transfers (mins) (km) %Change %Change (km) DART -265 -3.457 -7.074 13 -1,6% -1,6% SuburbRAIL -123 -2.668 -4.017 22 -0,5% -0,6% City Bus 1.149 11.756 -9.259 10 +1,2% -0,4% Regional Bus 30 428 -2.480 14 +0,1% -0,3% LUAS -597 -3.766 -10.284 6 -2,7% -2,9% Total PT 193 2.293 -33.115 12 +0,1% -0,7%

It can be seen that there is a clear increase in boarding onto the city centre bus service with an average increased trip length of (11,756/1,149) 10.2km. There are a total of 193 new PT users with reductions in users on LUAS, DART, and suburban rail.

Table 18 PT ordered from largest passenger km travelled to smallest

Pax. Boardings / Pax. Distance Pax. Time Rank Line No. Line Name Transfers (kms) (mins) 1 1451 Kilmac-Heuston 193 2.225 2.677 2 401 LVSC-OCB-Finglas 286 2.114 1.040 3 4112 SlieveB Tulla-Bel 28 1.706 1.929 4 253 LucanEskrMerrnSq 44 472 928 5 1232 WalknstownMarino 61 318 1.045 6 155 Stocking-Benson 34 272 1.192 7 4111 SILVERDAWN Port-B 5 243 288 8 132 GrngCstleHarrstwn 39 187 725 9 343 SkerresDPT-SSG 6 182 281 10 283 HarristownEdenQy 25 161 290

10 793 Parkwest - Aston -35 -263 -720 9 3189 KearnsRochfortDub -5 -377 -400 8 7703 LUAS_BrdGln-StnGrn -53 -385 -847 7 255 LucanFoxbMerrnSq -37 -386 -859 6 3203 AthKinnHeustDub -5 -398 -367 5 9671 Carlow >> Heuston -5 -402 -357 4 259 FoxbrghMerrnSq -52 -438 -958 3 9106 Grstnes >> Mlhide -32 -502 -919 2 9228 Arklow >> Connolly -23 -529 -790 1 7905 LUAS_Tallght-Point -119 -590 -2.008

The table shows PT ordered from largest to smallest passenger Km’s by line for TP2 08.00- 08.59, the busiest hour. Both routes which have TSP implemented have seen the largest passenger distance. This table shows two things, first that the changes to mimic TSP and increase priority on only two routes were successful. This was to be expected considering changes were made to give preference to these routes over all other routes. Secondly some routes received a large negative effect. This indicates that the results in the table should be considered with caution as some routes received a negative effect that are far away from the modified routes and outside the study area catchments.

8.8 Conclusion The study has made an intelligent intervention to mimic and model the average effect of TSP for buses and it has found quite a considerable positive benefit, both within the corridors and the whole GDA. However it should be noted that results from transport models should be

considered with caution. They are a representation of what might happen when different policy measures are introduced. However it does not mean that they will happen that way.

Table 19 North Clondalkin summary results

Modal Split % PT SM HW TSP 24,25 33,95 41,83 Base 24,21 33,95 41,84 Increase above base +0,04 0 -0,01

Travel Time (mins) PT SM HW TSP 19,4 35,5 18,3 Base 30,1 35,0 18,0 Increase above base -10,3 +0,5 +0,3

Table 20 Stillorgan summary results

Modal Split % PT SM HW TSP 21,47 23,71 54,82 Base 21,47 23,70 54,83 Increase above base 0 +0,01 -0,01

Travel Time (mins) PT SM HW TSP 31,1 61,2 29,5 Base 39,0 62,0 30,1 Increase above base -7,5 -0,4 -0,2

As mentioned in the limitations and assumptions section previously SMs comprise of pedestrians and cyclists. The majority of slow mode users are pedestrians, not cyclists. The 2006 Census revealed a GDA modal split for cyclists to be 4% (Central Statistics Office, 2006). As it is not possible to determine the exact cyclist percentages that travel from origin to destination for both corridors the GDA average is used and the remainder of the travellers in the slow mode category are no longer considered.

Table 21 Modal split total user increase above base

HW PT SM

City Centre -16 +43 -6 (destination) GDA -58 +72 -12

Sub-mode There is a clear net growth in passenger Km’s of +2293km (Table 17) which represents a net growth of +0,1%. This would translate directly into a 0,1% growth in fare revenue.

The results generated in this section are inputs for the next section where the implementation of TSP will be appraised from an ethical perspective using the Lorenz curve and Gini index.

8.9 Discussion There are a number of results from the previous section that are worthy of further discussion.

8.9.1 Public transport and highway results It is interesting that the public transport saved on average 9 minutes with the introduction of TSP and the highway and soft mode network have not been negatively affected by any great degree. Secondly the highway users on the North Clondalkin route gain travel time while the highway users on the Stillorgan route receive a reduction in travel time. This difference can be explained as follows. The route that highway users take along the Stillorgan corridor is the same as the public transport route for the majority of the trip. Only when entering the city centre does the public transport and highway route deviate. This means that the highway and public transport routes see the benefits of TSP. For North Clondalkin the highway and public transport route is not the same. Near the start of the corridor the highway and public transport route diverge with the highway route taking a shorter path to the city centre destination. As a result the highway route does not see the benefits of TSP that has been created for the public transport route but rather the negative results from a reduction in green time on side roads that have resulted in increased delays.

Another consideration to be taken into account for the difference in travel times is the quality of provision in QBCs and road space along the two routes. The Stillorgan corridor has a dedicated unbroken QBC route from origin to destination (both inbound and outbound) for the majority of the route, apart from the period passing through Donnybrook Village just before the city centre. The road provision up until this point is also consistent with a dual carriageway. For the North Clondalkin corridor there is not the same level of provision with segmented sections along the route providing a QBC, and one lane for car traffic. Illustrations of this can be seen in the appendix H. This could account for the differences in results.

8.9.2 Soft mode results Soft mode results consist of walking and cycling. It is known from gathering census data that the soft mode percentage is overestimated. This value consists of both pedestrians and cyclist and there is no way to further separate these modes in the model, so the census results are being used instead. The percentage of the population who cycled to work, school or college in the GDA is 4% (Central Statistics Office, 2006). The majority of the soft mode movements are pedestrian movements, not cyclist movements. As these values are not separated in the model it was necessary to apply the GDA average for cyclists as commuters to the two routes. While these values are the same for the two routes it is necessary to remember that

due to the different route lengths and socio-economic profiles these routes may have different values.

Bike lane provision is scattered along the two routes from having shared space with the highway, QBC or having a cycle lane on level with the pedestrian path, but rarely a separate cycle track. With this in mind it is considered appropriate that the travel time results from the HW be used to calculate the cyclist results.

However along the Stillorgan corridor there is a greater provision of a continuous cycle lane then the North Clondalkin corridor with a consistent separate cycle lane for the most of the corridor. While cyclists obey the same rules as vehicle traffic along the corridor assuming they stop at the traffic lights the presence of a dedicate cycle lane has the ability to reduce the travel time.

8.9.3 Modal Split Results The difference in modal split for both areas is a result of a combination of the different socio- economic demographic outlined in the previous chapter as well as the differing distances to the city centre.

It is possible that due to the fact that only two iterations were made the model did not run a full modal split and an error could have occurred.

8.9.4 PT Assignment changes in the model Using the results from the first iteration for the second iteration in the PT Assignment had to be considered carefully. There were a few downsides to this approach as the delays experienced by all other routes are the same as the first iteration, as well as slight variations that might have been present in the two routes. However through consultation with the NTA it was considered that any delay changes would not be that great between iterations. This approach was also taken due to access constraints with the GDA model. In a sense, for pragmatic reason, this seemed to be the best approach to model TSP with minimal intervention and relative simplicity.

8.9.5 Affects for the operator The operational effect of introducing TSP with a 7 and 10 minute savings on the two routes enables the bus operator to get significant savings. A 10 minute saving on a route enables the NTA and Dublin Bus to re assess the operation of that route. The minor savings at individual junctions can lead to a real saving for the operator. Dublin Bus’ aim is to achieve a 10% travel time saving over the entire network (Doherty, 2014), something that could be possible with the introduction of TSP.

The increase in passenger distance of +0,1% equates to a direct increase in revenue for the operator of +0,1%.

9 Lorenz Curve and Gini Index

The aim of this chapter is to determine if the results found in chapter 8 create a greater level of equity between the two zones, and if so to what extent. In order to do this the Lorenz Curve and Gini Index are used which was outlined in chapter 6.

9.1 Lorenz curve

9.1.1 Method The inputs used (found in the previous section) are the travel time of the three modes and their corresponding modal split. With regard to soft mode pedestrians are taken out and only cyclists are represented as previously discussed.

Table 22 North Clondalkin input values

Mode Modal split % Travel time (mins,ss) TSP SM 4 35,5 PT 24 19,4 HW 42 18,3 Base SM 4 35,0 PT 24 30,1 HW 42 18,0

Table 23 Stillorgan input values

Mode Modal split % Travel time (mins,ss) TSP SM 4 61,2 PT 21 31,1 HW 55 29,5 Base SM 4 62,0 PT 21 39,0 HW 55 30,1

Executed using Microsoft Excel first the number of individuals in the data set were ordered starting with the slowest travel times to the fastest travel times. Then the travel times were summed to get the total travel time. Next the percentage of population for each individual was generated.

Subsequently the percent of travel time for each individual is calculated in two steps;

1. (Travel time of individual / Travel time of total travel time)*100 2. Then calculate the cumulative travel time. The last individual should see 100%.

Finally the Lorenz curve is plotted with the population percentage on the x axis and the travel time accumulation on the y axis.

The most common use of the Lorenz curve is determining equality in income. This means allocating the percentage of population with the lowest income to the largest on the graph, which coincides with the smallest number to the largest number (e.g. 10,000€ to 100,000€). This results in the graph seen previously in Figure 8. However when applying travel time the percentage of population who have the longest travel time to the shortest travel time is used (e.g. 60 minutes to 10 minutes). This means the travel time needs to be divided by 1 (1/TT).

9.1.2 Results Figure 23 North Clondalkin Lorenz curve

North Clondalkin Lorenz Curve 100

90 80 70 60 50 Equality 40 TSP 30 Base 20

Cumulative Travel Time % Time Travel Cumulative 10 0 0 20 40 60 80 100 Cumulative Population %

Figure 23 shows the result that the introduction of TSP (red line) reduces the equality imbalance by moving closer to the line of Equality.

The first 4% in the above Lorenz curve represent the cyclists with the longest travel time, this travel time increased slightly with the introduction of TSP, this can be seen in the graph. The increase in travel time for the highway cannot be seen due to its representation in the graph by being the fastest travel time and therefore being at the top of the graph (60-100%).

Figure 24 Stillorgan Lorenz curve

Stillorgan Lorenz Curve 100

90 80 70 60 50 Equality 40 TSP 30 Base 20

Cumulaitve Travel Time % Time Travel Cumulaitve 10 0 0 20 40 60 80 100 Cumulative Population %

The result for this route is similar to the previous route. The result shows that the introduction of TSP (red line) reduces the equality imbalance by moving closer to the line of Equality. However it can be seen that the Stillorgan corridor is closer to the line of equality before TSP is introduced.

9.2 Gini index

9.2.1 Method

The Gini index is the Area A bounded by the area of equality and the Lorenz curve, and divided by the right triangle (0,5) seen in the Figure 23, Figure 24, and Figure 25 above.

The differences between the theory and real data is that real data does not have a continuous function (or curve), but instead line segments of data. By taking either the lower bound or the higher bound values would either underestimate or overestimate Area A. To overcome this, the average of the lower bound and the higher bound is found, and this is the area below the Lorenz curve.

Figure 25 Lorenz curve and Gini index example

Area under the first rectangle in the graph is found by (Highbound + lowbound)/2 which is the mean of the first rectangle, multiplied by the width of the rectangle (e.g. if there are 100 individuals, 1/100 = 0,01), (Highbound+lowbound)/2*0,01 for each rectangle.

Do this for each rectangle and then sum the results. This sum is the area under the Lorenz curve. But this study requires the Area A, the area between the line of Equality and the Lorenz curve. To get this: 0,5 – area under Lorenz curve.

Finally the Gini index is found by dividing Area A / 0,.5. The Gini index is between 0-1 with 0 being total equality and 1 being total inequality.

9.2.2 Results The results are below:

Table 24 Gini Index results

North Clondalkin Stillorgan TSP 0,06 0,06 Base 0,14 0,09 Increase above base -0,08 -0,03 % change 57% 36%

It can be seen that both routes reduce the inequality by -0,08 for the North Clondalkin area and -0,03 for the Stillorgan area. It can also be seen that introducing TSP on both routes balances the level of equality from 0,14 for North Clondalkin and 0,09 for Stillorgan to 0,06 and 0,06 respectively. The zones with lower equity value, the North Clondalkin zone sees a higher change in the Gini index of 57% rather than 36% for the Stillorgan zones.

9.3 Conclusion It can be seen that by using vertical equity and identifying accessibility components through the egalitarian theory distributing TSP between individuals of different abilities and needs it can be concluded that introducing TSP to an area with a lower individual and transport components increases the equity over an area then for an area with a higher individual and transport components.

Both visually through the Lorenz curve and numerically through the Gini index introducing TSP in both study areas regardless of the difference in accessibility measure components is positive and balances the level of equality. The difference in modal split and travel times for both study areas is reflected in the shapes of the Lorenz curve. The smaller Gini index change for the Stillorgan study area can be considered a result of the smaller modal split towards PT and the lower level of travel time reduction.

9.4 Discussion A perfectly even distribution of transit supply does not imply that demand for transit service is being perfectly met. A transit system can receive a “perfect” score of zero regardless of how well the system is able to meet the total demand (Bertolaccini, 2013). It is not only travel time fairness that is important but also modal split balance which relates to mode availability that is important for equitable transport.

For there to be equitable transport every mode needs to have a similar travel time. However not only travel time is important; secondly there needs to be a more balanced modal split, for example every mode highway, public transport and cyclist should aim to represent 33% modal split each and have equal travel times, such as 30 minutes. A third element affecting equitable transport is the distance to destinations. This can be seen through the Stillorgan zone (socially advantaged zone) that has a longer distance of 15km and has a cycle travel time of 61,2 minutes and a highway and public transport of approximately 30 minutes. Creating a public transport and highway network having a similar travel time to Cyclists would not be advantageous and it would require increasing in the current public transport and highway travel time; however it would create a more equitable transport. Therefore having a transport policy that aims to increase the number of Cyclists in this area could decrease transport equality. What is required in this location is a greater modal split towards public transport and highway in order to compensate for the large Cycle travel time.

Regarding the North Clondalkin zone (socially disadvantaged zone) a similar conclusion can be drawn from the Stillorgan zone due to the larger travel time for Cyclists then public transport and highway. However when considering the travel time difference between North Clondalkin and Stillorgan zones the North Clondalkin Cycle travel time and the Stillorgan public transport and highway travel time are similar approximately 30 minutes. Due to the shorter distance of 9,7km in the North Clondalkin zone a transport policy that aims to increase the number of cyclists in this area could increase transport equality between the two study areas.

A policy change to increase modal split towards cycling and implement TSP may overall improve the equality within the North Clondalkin zone and between the two zones. While the highway may be left untouched for both zones, implementing TSP for the Stillorgan zone would also balance equity.

It should be remembered that the North Clondalkin corridor has a more challenging urban landscape and does not have enough space to improve all three modes of transport, so choices may have to be made. This is unlike the Stillorgan corridor where there is enough road space on the majority of the route to make changes to all three modes.

The Lorenz curve and Gini index is a good analysis and visualisation tool to see the effects of different transport policies in different areas. It can be clearly seen in this example that while a blanket implementation of TSP for the whole QBC network may balance equity it may not provide the best results in isolation. A mixture of measures is required for different areas regarding socio-economic distribution, distance to destinations, current modal split and mode

availability, and current travel times. So far this study shows that the Lorenz curve and Gini index will be a good aid for transport policy and analysis.

10 Cost Benefit Analysis

It was found in chapter 9, that introducing TSP in the study area had increased equality. The aim of this chapter is to determine if introducing TSP on the study area generates similar positive results using another transport appraisal method, a cost benefit analysis.

10.1 Components There are two scenarios for the CBA, Do Nothing (base case) and Do Something (with TSP), the change between these two values are used in the CBA, in other words the net present value. The parts included are listed below. The appraisal period ranges from 2014 to 2030, 16 years to reflect the time span of other transport proposals and investment plans in Dublin. There are a number of actors responsible for implementing TSP:

 Dublin City Council – Traffic management  Dublin Bus – Operator  National Transport Authority (NTA) – Funding allocation and policy maker

10.1.1 Financial costs

Investment costs Regarding Dublin City Council (DCC) costs, there was a complex bespoke software module developed (DPTIMS) by a contractor to allow for the Transit Signal Priority to work in SCATS. Also, software was required to get the AVL data into required formats and fed into DCC from Dublin Bus servers. There is also some hardware involved such as a server to deal with the AVL data and CCTV monitor / remote controller units to view key junctions from staff workstations. The total cost for DCC for this setup is approximately €1.1million, funded by the NTA.

In addition, the NTA fund three DCC staff members to work full time on the AVL Bus Priority Project – the NTA work together with this team and Dublin Bus to identify the areas to target (using the latest AVL supplied delay data), and then the DCC team work on signal interventions and monitor their effects; approximately €180.000 a year.

The costs above relate to the introduction of TSP on the whole GDA network, however this study is concerned with only two out of 16 QBCs. Therefore the above costs are reduced to an eighth.

The purchase of buses with AVL capabilities has been implemented in line with the natural upgrading process of the Dublin Bus fleet that was completed in 2011. The buses are purchased by the NTA, and it is considered that there are no costs for this upgrade.

Maintenance and operational costs In the longer term the staff requirement is likely to reduce as all of the interventions get programmed into the system, but there will always be a need to monitor/tweak/react to new

bus delay elements required by DCC staff. For the purpose of this CBA it is considered that three staff will be required until 2020 then reduced to two until 2025 and then one from 2026 onwards.

Again the costs above relate to the introduction of TSP on the whole GDA network, however this study is concerned with only two out of 16 QBCs. Therefore the above costs are reduced to an eighth.

Revenue The fare box revenue earned by the operator is directly comparable to the increase in passenger distance of 10,2km, + 0,1% as seen in Table 17. While there are benefits through increased revenue with the introduction of TSP this value is not included in the CBA.

10.1.2 Social costs

External costs The travel time benefits and costs are taken from the GDA model. The travel time (TT) is an average of the travel time on the two routes. The number of passengers affected is taken from the increase above base of passengers (Pax)13 entering the destination zones when TSP introduced. The Value of Time (VoT) is known and taken from the GDA model in order to monetise the travel time results. However the model is based on 2006 data. In order to project the TT to the CBA study period (2014 - 2030) some extra analysis needed to be conducted. To bring the Pax value to 2030 first the 2011 Census Means of travel to school, college or work modal split percentages were used to identify any changes in modal split from the known 2006 and 2011 values. These percentage changes were applied to GDA modal results. The Central Statistics Office conducted a population projection from 2016 to 2031 with a GDA annual growth of 0,9%14. Taking the 2011 Pax. values, applying an annual 0,9% increase and extrapolating until 2030 it was possible to determine that approximately 18.000 people will be affected by the introduction of TSP by 2030. This assumes the same network characteristics in 2006 and 2030 (except for the introduction of TSP) there should be no modal split difference between the years.

There are other costs to consider when conducting a CBA such as environmental costs, safety, and reliability. These costs are not quantified in this study but rather qualified below.

Reliability Reliability can be measured in many ways such as connectivity reliability, travel time, headway and passenger wait time reliability or schedule reliability. Travel time reliability is defined by (Liu, 2007 cited in (van Oort, 2011) as ‘to what extent the specific actual time matched the scheduled time’ while service reliability is ‘the matching degree of the promised

13 Note the number number of passengers used in the Lorenz curve and Gini index was taken from the origin zones to represent the differences between the two routes. This is not possible with a CBA as so the destination zones were taken. 14 This was determined by the Central Statistics Office from the 2011 Census 62

and actual public transport services and its impacts on passengers’, which are mainly travel time related.

Summarised from Van Oort (2011), concerning service reliability, both the supply side (i.e. TSP) and the demand side of public transport are important, since unreliability is caused by the interaction of both sides. The supply side consists of the service provided by the operator, being trips in time and space. The demand side is defined as the passenger side including their behaviour and experiences. In an ideal situation, vehicles depart on time from the terminal and drive perfectly according to the schedule. Therefore, deviations and variations of the supply side may be categorized into two source types:

 Terminal departure time variability: is the distribution of schedule deviations (early or late) of the vehicle trip departure at the terminal,  Vehicle trip time variability: is the distribution of the trip times along the route.

Concerning vehicle trip time, two sub elements may be distinguished:  Driving times,  Dwell times.

Driving times consist of actual driving times from stop to stop and unplanned stopping times between stops. All these time elements may vary over time and these elements together affect the variability of total trip time, expressed in a trip time distribution. Unplanned stopping is the stopping of vehicles at a location where no boarding and alighting is enabled, for instance at traffic lights. Eliminating this source completely is the best way to improve public transport reliability. Unfortunately, in urban public transport (bus, tram and light rail), unplanned stopping occurs and results in both delays and service variability. The main reason is other traffic. Unlike a metro system which has exclusive right of way, these systems share parts of the infrastructure, lanes and junctions with car, bicycle and even pedestrian traffic. The other part of vehicle trip time is dwell time, which is the time used for boarding and alighting at a stop.

In Dublin QBCs provide a dedicated bus only lane, however this has greater provision on the Stillorgan corridor than the North Clondalkin corridor. Removing unplanned stops, through eliminating stops at traffic lights, is the best way to improve public transport reliability.

It is considered that introducing TSP improves the reliability of the bus service in the study area, and would therefore have a positive value in the CBA. For this study reliability is not included in the CBA but could be monetised by: ‘number of delays per month*12*Av delay(min)/60*no. of passengers affected*CPI’. Total travel time savings are then the sum of reliability savings plus travel time savings calculated in the previous external costs section.

With regard to the reliability of trip times along bus routes, Dublin Bus has Real Time Passenger Information (RTPI) on when a bus is due to arrive at the bus stop. RTPI information is calculated through historic AVL data that is recycled every two weeks. This is

available visually at the majority of bus stops, or online and through the Dublin Bus App. The reliability of the RTPI information is found to be operating at average 95% accuracy within 3 minutes and 91% within 60 seconds through a survey of 990 buses in July 2012 (O’Mullane, 2013). However this has no correlation to passenger timetables as no bus stop specific timetables exist on Dublin Bus route only terminal departure times. Therefore there is no way of telling (from a passenger perspective) if the bus is keeping to the timetable. On the Dublin Bus side there are desired location times along the route where the bus should be, displayed on their monitoring system and from that it is possible to tell if a bus is behind schedule or not. However these internal location times is an area where Dublin Bus are working to redefine as they now have a bank of AVL data that can be analysed to improve their timetable. In conclusion it is difficult to determine the resulting change in reliability that would occur with the introduction of TSP but is assumed that there would be an increase in reliability.

Environmental costs Environmental costs include costs generated from changes in noise, air quality and green- house gasses. A change in modal split would generate different values for these components most notably green-house gasses. Considering that there is only a minor modal split, it is not believed that there will not be any reduction in environmental costs. However it is not thought that there would be any increase either. In conclusion no change in environmental costs is seen in this study.

For HW users there may be a slight increase in idle time on cross streets due to the increase in red time, however there is likely to be a decrease in costs for the bus operator and other with flow traffic due to a reduction in acceleration and deceleration required at junctions as a result it is assumed in this CBA that the costs and benefits cancel each other out.

Safety It is assumed that there is no change in the safety of road users with the introduction of TSP. However in an interview with the Chairman of Cyclist.ie (Killen, 2014) there are concerns with the concept of buses being allowed to go through junctions without previously stopping while other traffic is at a standstill, which Killian (2014) says has the potential to cause safety concerns to on street cyclists. However for the purpose of this study it is assumed that no change in safety is seen for this study.

Sensitivity analysis A sensitivity analysis is undertaken in a CBA for the purpose of testing the robustness of the results in the presence of uncertainty. It is assumed for this study that the scenario presented by the modal is correct and changes are seen once implemented. However it is possible in CBAs that other scenarios are tested ranging from weak to robust.

Questions that can be asked in a sensitivity analysis are:

 Are the values generated from the transport modal reliable?

 Is the discount rate correct?  What if the operational and maintenance costs were higher or a reduction in staff is not seen?

By altering these values in the CBA it can be determined if the proposal still results in a positive net present value or not under different scenarios.

10.2 Results Represented in € per year summed for the study period.

Table 25 CBA results

Cost-Benefit analysis (euro) TOTAL ( Net Present Values) 2014 - 2030 Costs Benefits Costs Investment costs (year 0) -137.500,00 Personnel costs ( year 0) -22.500,00 Operational and maintenance costs (year 1 – 16) -196.262.12

Transport travel time benefit (Totals) HW (year 1 – 16) -603,60 PT (year 1 – 16) 642.356,97 Cycle (year 1 – 16) -424,86

TOTAL -357.290,57 642.356,97 Benefits minus costs Net Present Value € 285.066,40

Cost Benefit Ratio 1,8

In Table 25 it can be seen that with the introduction of TSP on the two corridors generates a total net present value benefit of € 285.066,40 euro and a cost benefit ratio of 1.8. A cost benefit ratio over 1 is positive.

10.3 Conclusion A simple CBA was conducted to find the cost of implementing TSP through a combination of investment, maintenance and operational costs and travel time savings. The rest of the CBS components were not considered for this study but were represented qualitatively.

The limited input in the CBA is considered sufficient as all other components are not (negatively) affected and it is the interaction between travel time and investment, maintenance and operational costs that are the most significant. It was found that there was a total net present value benefit of +€285.066,40 euro and a cost benefit ratio of 1.8.

10.4 Discussion Both methods CBA and Lorenz curve and Gini index generate a positive result when TSP is implemented in the study area. However one clear difference between these methods is the scale at which they gather their input values. For the Lorenz curve and Gini index it was possible to see the effect that TSP had on both routes individually and how much that affect the study areas. For the CBA this is not possible and only the overall result is possible. This has the potential to hide negative effects of introducing such transport changes.

The Lorenz curve and Gini index has the potential to be used in conjunction with a CBA to highlight the distribution of social benefits and costs with study area. Through the step by step methodology outlined in identifying the necessary components the results of the CBA can be used to conclude that the equitable transport innovation is financially feasible.

11 Conclusion and Recommendations

This chapter first formulates the main conclusions to the research questions set out in the first chapter. Secondly recommendations are made to improve the methodology and finally recommendations for further research are mentioned.

11.1 Conclusion This study aimed to answer the research question:

What are the equity and efficiency impacts of introducing transit signal priority in a city network?

The hypothesis also proposed that ‘introducing TSP will result in a reduced travel time for bus passengers, have minimum negative impact on other road users and finally, balance the societal equity among differing socio – economic areas’.

In order to answer the research question a number of research questions needed to be answered. The sub questions and answers are explained below cumulating in the answer to the main research question.

Phase One included chapters 2 and 3 and answered following sub questions: 1. What are the technical requirements for implementing TSP? 2. What methods are currently used for appraising transport projects?

 Conclusions drawn from chapter 1 is that three main components are required to implement TSP; vehicle detection systems, communication systems and traffic control systems. It was discovered that the study area has acquired all three components and started testing the TSP system on a number of isolated junctions in the summer of 2014.  Secondly a number of transport appraisal methods were identified in chapter 2, all methods lacked the ability to appraise transport proposals in an equitable manner. The most common method is the cost benefit analysis (CBA) which lacked sufficient equity analysis as it was said that it generally ignores the distributional and equity effects and other ethically important implications of choice options, e.g. the area of social exclusion.

Phase Two included chapters 4, 5 and 6 and answered the following sub questions: 3. Why is transport equality important? 4. What components are necessary for an equitable transport appraisal method? 5. Formulate an equitable transport appraisal method.

 Chapter 4 found that are many dimensions to equality. Social exclusion, poverty and accessibility have all been discussed in literature which aims to give thresholds on individual areas but equality aims to go further and looks at the difference between these levels and tries to reduce this by achieving a higher and fairer level of transport provision. It can be concluded that transport equality is important to reduce the level of social exclusion, balance accessibility and to ensure equal access to opportunities.  Following from this is was necessary to understand what components were required to assess transport from an equitable perspective. It was concluded that vertical equity in combination with egalitarian theory was the most relevant. This required identifying four accessibility components, namely the individual, land use, transport and temporal components. In conjunction with this an egalitarian related indicator had to be created, this was defined as ‘measuring journey time is for all people living in a certain area to a (selection of) destination(s) that are assumed to be most relevant from a SRAI perspective (van Wee & Lucas, in press)’.  The final section outlined a method to assess equity using the Lorenz curve and Gini index: the Lorenz curve shows the distribution of travel time over the population and the Gini index, being a widely accepted statistical technique with which to express the distribution of an item over a group of people, the dominant application being the distribution of income over the population of a country e.g. (Weymark, 1981).

Phase Three included chapters 7, 8, 9 and 10 and answered the following sub question: 6. What are the travel time and resulting equity impacts from implementing TSP?

 The travel time results from implementing TSP showed an approximate 7 minute decrease in travel time for the North Clondalkin public transport route in which was identified as the socially disadvantaged zone in chapter 7 and an approximate 10 minute reduction in travel time for Stillorgan public transport route in the socially advantaged zone. Both the highway and cyclist users on the North Clondalkin corridor received minimal increase in travel time of approximately 1 minute, while the Stillorgan highway and cyclist users received a slight decrease of approximately 1 minute. It was concluded that the increase in highway and cycle travel time on the North Clondalkin corridor was a result of differing route choice in comparison to the public transport route while the Stillorgan corridor used the same route for public transport, highway and cyclists.  The testing hypothesis is that introducing transit signal priority will result in a reduced travel time for bus passengers, have minimum negative impact on other road users and finally, balance the societal equity among differing socioeconomic areas. It can be concluded that the first part of the hypothesis is verified as TSP did reduce travel time for the bus and have minimum negative impact on other road users.  When testing the equality of these results using the Lorenz curve and Gini Index the results showed 57% increase in equality for the socially disadvantaged zone (North Clondalkin) and a 36% increase for the socially advantaged zone (Stillorgan). Overall the resulting equality between the two zones reduced to a more balanced level of 0,06 Gini index for both zone when TSP was introduced. The second part of the hypothesis was also verified in that introducing TSP did balance the equity among contrasting socially advantaged areas.  Finally when testing the results using a common financial based transport appraisal method, CBA, the introduction of TSP on the two corridors generated a total net present value of € 285.066,40 euro, and a cost benefit ratio of 1.8.

Conclusions drawn about the use of Lorenz curve and Gini index with the CBA  By conducting the equity analysis it was possible to show that one area gained greater advantage over the other, the CBA was unable to see this. This could have a number of consequences, firstly that if only the CBA is conducted and the benefit is not sufficient overall then the project might not be realised. However with the Lorenz curve and Gini Index an extra analysis step was possible and showed that perhaps the benefits gained in one area are significant enough to override the lower level of benefits seen over the entire network. Secondly it has the potential to prioritise the location of certain investments through the understanding of the level of equity that is gained in and between each area.

Conclusions drawn about the Lorenz curve  Regarding the Lorenz curve both corridors for the base case already had a high level of equity as seen in Figure 23 and Figure 24 where the curve is not overtly pronounced. When TSP is introduced the routes become very close to 0 or the line of equity. This could be due to a number of reasons, one of which is that the chosen zones may not have been that distinctive in the first place due to the fact that they both had QBCs15.  This method has the ability to influence policy, network and local level.  At policy level the use of the Lorenz curve and Gini index enables different transport proposals to be tested to see the effects it has on different users and helps policy makers decide a proposal that generates the most equitable result. This high level approach has the ability to directly impact the lower level users in the most positive way.  At network level it can determine which routes gain the best results and which routes do not benefit. This is important for two reasons. First that transport is there to improve society and so the benefits can be seen. Secondly areas where the proposal generates the greatest benefit can be identified. Therefore if the introduction of TSP is required in stages then the 16 QBCs can be prioritised in order of who benefits the most from it.  At local level this method identifies the current equality level in transport provision and enables policy makers to pin point what is required to achieve a more equitable transport. It delivers real benefits to areas by focusing on specific modes per area that increase accessibility and reduce travel time to key destinations. It prioritises areas in terms of investment in order to reduce the level of inequality both within areas and between areas by implementing proposals that reduce the Gini index.  This approach has the ability to see the effects a transport proposal has at a disaggregate level rather than the aggregate level of a cost benefit analysis which has the ability to hide negative results in one area with large positive results in another area therefore increasing the level of inequality.  What the Lorenz curve and Gini index is unable to do is deal with the health or environment benefits of using a particular mode. Taking the example of the Stillorgan Corridor with a cycle travel time of 60 minutes and bus and car travel time of approximately 30 minutes, in order to make this corridor completely equitable car and bus travel times must increase to that of the bike or cyclists must be removed as a mode as their speed is unlikely to reach that of motorised vehicles. Both options have the potential to cause a negative effect on both the health of citizens and the environment.

15 Discussed more in recommendations 70

11.2 Recommendations on improvements of the methodology This chapter outlines areas where the model and methodology could be improved for further research.

11.2.1 Definition of accessibility components By changing the transport component in section 7.6 from limiting it to QBCs only would open up more isolated zones in the study area. One reason for the similar Lorenz curve results could be due to both public transport corridors running on a QBC. This could also account for the high level of equality on both routes, and the minimal curve seen.

11.2.2 Origins and destinations used Only two corridors and one destination were considered, while these two routes differ with regard to length and socio-economic characteristics it would be interesting to model all 16 QBCs into the city centre destination with TSP to see the results.

11.2.3 Junction changes 5 – 10 seconds was taken as an appropriate value to increase the green time at junctions, this value was taken after consultation with the NTA, research and the level of green time already provided at the junctions. If the remaining arms on the junction had 10 seconds of green time or 4 seconds of inter-green time left no increases could be made to the junction arm on the route. (10 seconds for a bus to clear a junction (Embarq India, n.d.). More detail into the effects of these signal changes at local level could be done.

11.2.4 PT changes Only two iterations in the mode choice, hour of travel choice and route choice / PT assignment were made. This gives the possibility that the convergence values would be different then after only two iterations, however due to time restrictions this was not possible to do. The first iteration had much larger values suggesting that the model was working and that two iterations, while not ideal have given sufficient results.

Using the delay file from the first iteration for the second iteration was also a compromise in order to save time and manual input. It is unknown what the delay values were, however it is considered by the NTA that the values would have been similar. While it is considered that the change between the second iteration and convergence is minimal, this cannot be known for certainty. Therefore allowing the model to run to convergence could be considered.

11.3 Recommendations for further research Further research into applying the Gini index increase above base results into the CBA would fully incorporate the two methods and not leave them as comparison and supplementary transport appraisal methods as they now currently stand.

 This study took bus routes with a similar level of exclusive lane provision (QBCs), further studies could look at taking bus routes with and without exclusive lane provision. It was stated in problem statement in section 1.2 that 70% of Dublin Bus trips are conducted on QBCs. This could result in the remainder of the bus routes experiencing a reduced level of equity because they do not have QBC provision. This has the potential for greater differences in the Gini index.  The changes made to the transport model were intended to mimic the effects of implementing transit signal priority. A more comprehensive model could be run which could include letting the model run further then two iterations, and modifying until a model split is seen which would be more likely to happen given the travel time savings generated.  The Lorenz curve and Gini index does not include the cost of travel outside of the costs included in the transport model that generated the travel times. Further analysis could include the up-front cost of travel by each mode. This is an important factor in an equitable society as not everyone has access to the same level of income and therefore may not have access to the same modes.  Safety: Further study into the effects of safety on road cyclists with TSP could be conducted.  To explore the possibility of being able to weight the travel time in the CBA with the Gini index results. This would enable greater level of interaction between the values used from the Gini index into financially based appraisal methods.

12 References

Bertolaccini, K., 2013. Msc: Assessing the equity of transit supply distribution in metropolitan areas using Lorenz curves and Gini coefficients, s.l.: University of Connecticut.

Broderick, C., 2014. Colin Broderick. [Online] Available at: http://cbroderick.me/dublin/ [Accessed 18 03 2015].

Caulfield, B., Bailey, D. & Mullarkey, S., 2013. Using data envelopment analysis as a public transport project appraisal tool. Transport Policy, Volume 29, pp. 74-85.

Central Statistics Office, 2006. Census 2006 reports. [Online] Available at: http://www.cso.ie/px/pxeirestat/Statire/SelectVarVal/Define.asp?maintable=C1203&Plangua ge=0 [Accessed 06 03 2015].

Central Statistics Office, n.d. Census 2011 boundary files. [Online] Available at: http://www.cso.ie/en/census/census2011boundaryfiles/ [Accessed 10 09 2014].

Clontarf Online, 2012. Public transport in Clontarf 1880. [Online] Available at: http://www.clontarfonline.com/transport1880.htm [Accessed 18 03 2015].

De Brucker, K., Macharis, C. & Verbeke, A., 2011. Multi-criteria analysis in transport project evaluation: an institutional approach. Eurpoean Transport, Volume 47, pp. 3-24.

Delbosc, A. & Currie, G., 2011. Using Lorenz curves to assess public transport equity. Journal for Transport Geography, Volume 19, pp. 1252-1259.

Dept of Environment, Community and Local Government, 2014. Design Manual for Urban Roads and Streets, Dublin: Dept of Environment, Community and Local Government.

Dept. of Transport, Tourism and Sport, 2011. Article 17 (1) of Directive 2010/40/EU National Report for Ireland. s.l.:Government Ministy.

Doherty, P., 2014. Interview with Dublin Bus CEO [Interview] (12 8 2014).

Dublin Bus, 2011. Annual Report & Financial Statements, Dublin: s.n.

Dublin Bus, 2014. Dublin Bus annual report 2013, s.l.: Dublin Bus.

Dublin City Council, 2007. HGV management strategy. [Online] Available at: http://www.dublincity.ie/hgv [Accessed 29 02 2015].

Embarq India, n.d. The Hub. [Online] Available at: http://www.embarqindiahub.org/online-publications/51-its-and-applications- city-bus-operations [Accessed 31 Janurary 2015].

Fellendorf, M., n.d. Public transport priority within SCATS - A simulation case study Dublin, s.l.: s.n.

Feng, W., 2014. Analyses of bus travel time reliability and transit signal priority at the stop- to-stop segment level, s.l.: Portland State Univeristy.

Foth, N., Manaugh, K. & El-Geneidy, A. M., 2013. Towards equitable transit: examining transit accessibility and social needs in Toronto, Canada. Journal of Transport Geography, Volume 29, pp. 1-10.

Franklin, J. P., 2005. The distributional impacts of transportation policies: A research design for the case of roadway tolls. Kansas city, MO, USA, ACSP 46th Annual Conference.

Furth, P. G. & Muller, T. H., 2000. Conditional bus priority at signalised intersections: better service quality with less traffic disruption. Transportation Research Record, Volume 1731, pp. 23-30.

Geurs, K. & van Wee, B., 2004. Accessibility evaluation of land-use and transport strategies: review and research directions. Transport Geography, Volume 12, pp. 127 - 140.

Gonzalez-Cancelas, N., Palomino-Monzon, M. C., Soler-Flores, F. & Almazan-Garate, J. L., 2013. Gini coefficient, dissimilarity index and Lorenz curve for the Spanish port system by type of goods. s.l., Scientific Conference.

Gonzalez-Perez, G. J. et al., 2008. A socio-spatial analysis of social exlusion and inequity in health in Mexico. REVISTA DE SALUD PÚBLICAv, 10(1), pp. 15-28.

Halden, D., 2011. The use and abuse of accessibility measures in UK passenger transport planning. Research in Transportation Business & Management, Issue 2, pp. 12-19.

Halffman, W. & Leydesdorff, L., 2010. Is inequality among universities increasing? Gini coefficients and the elusive rise of elite universities. Minerva, Issue 48, pp. 55-72.

Headicar, P., 2009. Transport policy and planning in Great Britain. first ed. Oxon: Routledge.

Ireland i, 2010. Adaptive road traffic control systems in use in Ireland. [Online] Available at: http://www.iol.ie/~discover/traffic.htm [Accessed 17 09 2014].

ITS America, 2005. Transit Signal Priority (TSP): Planning and Implementation Handbook, s.l.: s.n.

Jensen, M., 1999. Passion and heart in transport - a sociological analysis on transport behaviour. Transport Policy, Volume 6, pp. 19-33.

Jones, P. & Lucas, K., 2012. The social consequencess of transport decision-making: clarifying concepts, synthesising knowledge and assessing implications. Journal of Transport Geography, Issue 21, pp. 4-16.

Karou, S. & Hull, A., 2014. Accessibility modelling: predicting the impact of planned transport infrastructure on accesibility patterns in Edinburgh, UK. Journal of Transport Geography, Issue 35, pp. 1-11.

Kenyon, S., Lyons, G. & Rafferty, J., 2002. Transport and social exclusion: investigating the possibility of promoting inclusin through virtual mobility. Transport Geography, Volume 10, pp. 207 -219.

Killen, M. M., 2014. Interview with Chairman of Cyclist.ie [Interview] (13 08 2014).

Kimpel, T. J. & Strathman, J. G., 2004. Improving scheduling through performance monitoring using AVL/APC data, s.l.: Potland State University.

Kinane, D. & O'Donnell, M., 2013. Dublin public transport interface model: A pilot study. Dublin, ITRN.

Lucas, K., 2012. Transport and social exclusion: Where are we now?. Transport Policy, Volume 20, pp. 105 - 113.

Martens, K., Golub, A. & Robinson, G., 2012. A justice-theoretic approach to the distribution of transportation benefits: Implications for transportation planning practice in the United States. Transportation Research Part A, Volume 46, pp. 684-695.

Mullen, B., 1989. Dalkey Homepage. [Online] Available at: http://www.dalkeyhomepage.ie/trams9.html [Accessed 18 03 2015].

National Transport Museum, n.d. Passenger Vehicles - Dublin United Tramways and tram No. 253. [Online] Available at: http://www.nationaltransportmuseum.org/b005.html [Accessed 18 03 2015].

NTA, 2013. Report on trends in mode share of people crossing the Canal Cordon 2006 - 2012, Dublin: National Transport Authority.

O’Mullane, A., 2013. ITS for public transport management. Thessaloniki, s.n.

O'Brien, B. & O'Donnell, M., n.d. Centralised public transport priority in SCATS. s.l.:Dublin City Council.

O'Connor, D. & Kavanagh, P., 2014. Stillorgan QBC dwell time analysis. Dublin, Irish Transpsort Research Network.

Office of the Attorney General, 1944. Irish Statute Book. [Online] Available at: http://www.irishstatutebook.ie/1944/en/act/pub/0021/index.html [Accessed 18 03 2015].

Rosler , K. & Mc Donald, M., 2011. The relationship between transport and disadvantage in Australia, s.l.: Australian Government.

RPA, 2013. Record year for Luas in 2013. [Online] Available at: http://www.rpa.ie/en/news/Pages/RecordyearforLuasin2013.aspx [Accessed 18 10 1204].

Santos, B., Antunes, A. & Miller, E. J., 2008. Integrating equality objectives in a road network design model. Washington D.C., 87th Annual Meeting of Transportation Research Board.

Sarica, K. & Or, I., 2007. Efficiency assessment of Turkish power plants using data envelopment analysis. Elsevier, Issue 32, pp. 1484-1499.

Secchi, B., 2013. A new urban question. Cascais, s.n.

Smith, H. R., Hemily, B., Ivanoic, M. & Flemming Inc, G., 2005. Transit Signal Priority (TSP): A planning and implementation handbook. s.l.:United States Department of Transport.

Stewart, D., 1955. Dublin City passenger transport services. Journal of the statistical and social inquiry society of Ireland, Volume 108, pp. 136-155.

Strathman, J. G. & Kimpel, T. J., 2003. headway deviation effects on bus passenger loads: Analysis of Tr-Met's archived AVL/APC data, s.l.: Potland State University.

Thomopoulos, N., Grant-Muller, S. & Tight, M., 2009. Incorporating equity considerations in transport infrastructure evaluation: Current practice and a proposed methodology. Evaluation and Program Planning, Volume 32, pp. 351-359.

Tian, Q., Huang, H.-J., Yang, H. & Gao, Z., 2012. Efficiency and equity of ramp control and capacity allocation mechanism in a freeway corridor. Transportation Reseach Part C, Issue 20, pp. 126-143.

Transport for Ireland, 2015. Cycle Planner. [Online] Available at: http://www.journeyplanner.transportforireland.ie/cp/XSLT_TRIP_REQUEST2 [Accessed 03 02 2015].

Transportation Economics, 2014. Transportation benefit-cost analysis. [Online] Available at: http://bca.transportationeconomics.org/ [Accessed 04 09 2014]. van Oort, N., 2011. Service reliability and urban public transport design, s.l.: TRIAL Reseach School.

van Oort, N., Sparing, D., Brands, T. & Goverde, R. M., 2013. Optimizing public transport planning and operations using automatic vehicle location data: the Dutch example. Dresden, Germany, International conference on models and technologies for intelligent transport systems, MT-ITS. van Oort, N. & van Nes, R., 2007. Improving Reliability in urban public transport in strategic and tactical design. s.l., 87th Annual meeting of the Transportation Research Board 2008. van Wee, B., 2012. How suitable is CBA for the ex-ante evalutaion of transport projects and policies? A discussion from the perpecctive of ethics. Transport Policy, Issue 19, pp. 1-7.

Van Wee, B. & Geurs, K. T., 2011. Discussing Equity and Social Exclusion in Accessibility Evaluations. EJTIIR, 4(11), pp. 350-367. van Wee, B. & Lucas, K., in press. Developing a method to evaluate equitable accessibility: combining ethical theories and accessibility-based approached. Transportation.

Vigar, G., 2002. The politics of transport: transport, the environment and public policy. First ed. London: Spon Press.

Welsh, T. F., 2013. Equity in transport: The distribution of transit access and connectivity among affordable housing units. Transport Policy, Volume 30, pp. 283-293.

Weymark, J., 1981. Generalised Gini inequality indices. Mathematical Social Sciences, 1(4), pp. 409-430.

13 Appendix

A Socio-economic analysis A socio-economic analysis is undertaken of the Dublin area using the 2011 Census data aquired through the National Transport Authority (NTA) and analysied using ArcGIS software. The census data can be categorised under a number of boundaries, the one relevant for this study is Election Districts (ED’s) of which there are 3,440 in the country (Central Statistics Office, n.d.).

There are a number of characteristations that need to be determined in order to evaluate the demographic of Dublin. Such characteristations include:

1. Population density 2. Population by social class 3. Population aged 15 years and over by principal economic status 4. Employment location 5. Number of households with cars 6. Population aged 5 years and over by means of travel to work, school or college

Data Collection In order to undertake this analysis data had to be collected from a number of sources in order to build the ArcGIS file, namely:

Central Statistics Office (CSO) website:  ED ArcGIS file (.shp, .shx, .dbf)

‘Transport for Ireland’ website:  The General Transit Feed Specification (GTFS) for Dublin Bus (.txt)

NTA:  All public transport routes for the country inlcuding their shape_id and route_id (.xlsx, .shp.xml)  Employment data in POWSCAR_CSOED (.xlsx)  Industry Group = Wholesale, Retail Trade, Transportation and Storage, Accommodation and Food Service Activities and the count in POWSCAR_CSOED (.xlsx)  SAPS data on the themes, joined at ED level (.xlsx) o Prinicpal status o Social class and socio-econoimc group o Commuting o PC and internet access (number of households with cars)

Method In order to achieve the required information, the above data had to be built in ArcGIS.

 The first step in this process is to upload the ED’s for the whole country into ArcGIS which was laid over a Irish basemap. It is important to ensure that the same gepgraphical reference is used in joining the longatudes and latitudes.  Next the GTFS file and PT file which included Dublin Bus routes and the Luas and heavy rail lines were layed on top.  Next the POWSCAR and SAPS data is uploaded and joined by ED code.  At this stage it is possible to reduce the file from country level to the study area level by selelcting the ED’s within the geographical scope that hold all the Dublin Bus routes and consequently any other PT routes within this catchment,  From here it is posisble to obtain the information for the questions previously outlined.

The data is represented by classifying numerical fields for graduated symbology. When classifying the data in ARcGIS there are many standard classification methods to choose from. It was determined that the ‘defined interval’ classification method was preferential and gave intervals between 3 and 7 classes. This option determines the number of classes based on the interval size and the range of all field values. An example of this is given below. It is important to test the different classifications as some have the potential produce maps that could be misleading.

Figure A.26 Data classification field in ArcGIS

Population density The population is evenly distributed over the city, with expected lower population density in the suburbs compared to the city centre.

Figure A.27 Population density

For AVL analysis The information for the following sections was obtained after a meeting with Dublin Bus CEO Paddy Doherty and Declan O’Malloy who is responsible for DPTIMS and AVL in Dublin Bus. Three documents were provided with the following information:

1. Each of the 16 QBCs were broken down into AVL segments with the distance (metres), sample size, average travel time in seconds for each hour period throughout the day, as well as the overall standard deviation for the day. 2. The second piece of information given is similar to the information above but it’s the average QBC speed for the 16 QBCs throughout the whole day not broken down into AVL segments. On a second excel sheet the inbound and outbound QBC speed (kph) peaks are represented 08.00-08.59 and 17.00-17.59 respectively, with the QBCs ordered from slowest to fastest. 3. The final piece of information provided are five QBCs which were analysed for unscheduled stops. The North Clondalkin QBC is one of these five, there are a number of data columns per QBC in this file; Stop Name, Stop No., number of Stops, number of Trips, Percentage of stops in trips, Average Stop time, Average Distance (before or after bus stop), and Impact. Stillorgan QBC is not represented in this data.

B Population aged 5 years and over by means of travel to work, school or college. Sum of motorcycle or scooter, car driver, car passenger and van

Figure A.28 Means of travel by motorised vehicles

C Table North Clondakin QBC unscheduled stops Table A.26 North Clondakin QBC unscheduled stops

Month Timeband Rank Stop Name Stop No. Stops Trips Percentage% Average Stop time Average Impact Distance Nov'13 7-10 1 Emmet Road South Circular Road (+) 01992 226 562 40 67 47 15032 Nov'13 7-10 2 Ballyfermot Rd Drumfinn Road (+) 02696 263 928 28 46 128 12034 Nov'13 7-10 3 Ballyfermot Rd Blackditch Drive (+) 02689 143 669 21 49 73 7026 Nov'13 7-10 4 Emmet Road Camac Close (-) 01989 127 623 20 52 -135 6635 Nov'13 7-10 5 Ballyfermot Rd Drumfinn Road (-) 02696 122 928 13 52 -171 6338 Nov'13 7-10 6 Ballyfermot Rd Ballyfermot Parade (+) 02697 97 924 10 61 80 5949 Nov'13 7-10 7 Sarsfield Rd St. Mary's Avenue (+) 02718 56 587 10 53 363 2993 Nov'13 7-10 8 Ballyfermot Rd Blackditch Drive (-) 02689 46 669 7 51 -63 2346 Nov'13 7-10 9 Sarsfield Rd Woodfield Place (-) 02719 40 482 8 58 -115 2317 Nov'13 7-10 10 Sarsfield Rd Woodfield Place (+) 02719 37 482 8 49 109 1798 Nov'13 7-10 11 Ballyfermot Rd Ballyfermot Parade (-) 02697 15 924 2 63 -114 941 Nov'13 7-10 12 Sarsfield Rd St. Mary's Avenue (-) 02718 18 587 3 41 -90 737 Nov'13 7-10 13 Ballyfermot Rd Cleggan Park (+) 02688 10 667 1 64 90 643 Nov'13 7-10 14 Ballyfermot Rd Kylemore Road (-) 02713 11 322 3 53 -167 586 Nov'13 7-10 15 Emmet Road Myra Cottages (+) 01990 6 563 1 48 53 290 Nov'13 7-10 16 Emmet Road Myra Cottages (-) 01990 5 563 1 54 -77 270 Nov'13 7-10 17 Ballyfermot Rd Longmeadows Park (+) 02716 2 323 1 70 62 140 Nov'13 7-10 18 Ballyfermot Rd Cleggan Park (-) 02688 1 667 0 67 -25 67 Nov'13 7-10 19 Ballyfermot Rd Longmeadows Park (-) 02716 1 323 0 67 -78 67 Nov'13 7-10 20 Ballyfermot Rd O'Hogan Road (-) 02715 1 325 0 47 -126 47 Nov'13 7-10 21 Ballyfermot Rd Cherry Orchard I.E. (-) 02687 1 668 0 34 -75 34

D GDA Model

Introduction The NTA’s GDA transport model is a strategic multi-modal, network based transport model covering the Greater Dublin Area (i.e. the counties of Dublin, Meath, Kildare, Wicklow and Louth). The model includes the main surface modes of travel (including travel by car, bus, rail, goods vehicles, walking and cycling). The model currently comprises of a morning peak model covering the three hour period between 07:00 and 10:00 and an afternoon inter-peak model covering the single hour between 14:00 and 15:00. The GDA transport model is vested in by the NTA (but owned by the State), who are the authority responsible for its maintenance and use.

The last update of the GDA transport model was completed in 2009 and incorporated the 2006 Census travel to work data and data from the GDA travel to education and household travel surveys (both undertaken in 2006). The model was re-calibrated to observed 2006 travel behaviour and conditions and the Trip Attractions & Generation and Trip Distribution Models were re-developed to incorporate the 2006 land uses and travel data sets. The afternoon inter-peak model was developed at this stage to have a similar structure and functionality to the morning peak model and calibrated to the observed 2006 off-peak travel conditions.

Model characteristics Listed below is a summary of the main characteristics of the NTA’s transport model for the GDA in terms of the area covered, zoning system used, time periods modelled, model base and forecast years, transport networks modelled and the classification of travel demand.

Zoning: The GDA Transport Model covers the full Greater Dublin Area (GDA) and county Louth. The current model has 657 internal fine zones covering the modelled area and 9 external point zones representing travel between the modelled area and the rest of Ireland 666 zones in total). In the metropolitan area, some zones are subsets of the District Electoral Divisions (DED’s) used to compile Census data16. In the hinterland area, zones are much larger and are an amalgamation of DED’s.

In order to represent travel patterns at a more aggregate level, the model has the facility to amalgamate the 657 fine zones to 75 strategic zones (called sectors), or to 21 coarse zones. In addition, travel demand can be analysed in terms of travel within and between 87 District Centres defined for the GDA. The modelled area and fine zoning system is shown in Figure A.29 below.

16 The majority of the NTA zones coincide with DEDs. 83

Model Periods: Two separate periods of the day are modelled. The am-peak model covers the three-hour period from 07:00 to 10:00, while the inter-peak model covers the single hour from 14:00 to 15:00.

Base and Forecast Years: The base year for the current peak and off-peak models is 2006, while the main forecast year is 2030 – the target year for the GDA Transport Strategy.

Modelled Networks: The model contains coded networks for all mechanised modes of travel – including car, goods vehicles, bus, heavy rail, LUAS and Metro. The highway network has two distinct regions. In the Dublin County area, full junctions details are included for all major junctions (the simulation network), while outside Dublin County, junctions details are not included (the buffer network). The model has a background goods vehicle demand which was calibrated approximately against available count data.

The bus network contains details on Dublin Bus, Bus Eireann and private operator bus services operating within, into and out of the GDA. Quality Bus Corridors and bus priority measures are included as part of the highway network, and in the simulation area their impact on junction capacity is coded17.

The rail network contains details on Iarnrod Eireann (intercity rail) services operating in and out of the GDA. Existing and future LUAS and Metro lines and services are coded in the model as part of the rail network. Sections of the road, and public transport networks as coded in the model are shown in Figure A.30 and Figure A.31 below.

Travel Demand: In the case of both the am-peak and off-peak models, travel demand is broken down by six journey purposes – i.e. Work (commuting), Education, Employer’s Business, Shopping, Other and Non Home Based. Travel demand is further segmented by two person types – i.e. those with a car available for their trip (Car Available), and those without a car available for their trip (Car Not Available).

17 To what extent? – the average bus speed in a bus lane was taken from data gathered in the QBC monitoring reports. In locations where road surface space was taken for QBC (e.g. Rock Road) this has been coded as one lane removed, and unavailable for road traffic. This is an approximate exercise at best, because of set-backs for turning movements etc. 84

Louth

Figure A.29 Modelled Area and Model Zoning

Figure A.30 Section of the Road Network as coded in the GDA Transport Model

Figure A.31 Section of Public Transport Network – coded in GDA Transport Model

Model structure The AM peak model includes an additional stage from the afternoon inter-peak model18 called the Hour of travel choice. This enables the modelling of peak spreading where people decide to depart at an earlier or later time to avoid congestion or crowding.

Car Ownership / Availability Trip Generation Model Pre Process

Trip Distribution Iteration loop between Mode Choice, Hour of Mode Choice Travel Choice, and Route Choice until equilibrium across Hour of Travel Choice modes, time periods and route choice is achieved. Models peak Route Choice / Trip Assignment spreading (AM-peak model only).

Figure A.32 GDA model structure Each model component is run in sequence as shown above. There is an iterative loop between mode choice, hour of travel choice and route choice stages. Iterations proceed until equilibrium is achieved across travel modes, hour of travel and route choice.

Model components / stages This section explains the model components shown in the figure above.

Trip Generation This stage of the model refers to the estimation and prediction of the number of trips that will be generated by people in each model zone and the number of trips attracted to each model zone in any given modelled time period – i.e. the trip generation stage refers to both trip generation and trip attraction.

18 Only applies to Home-Work trips. Education trips etc are generally constrained to arrive by 9am, i.e. there is no choice, as such. 87

The Trip Attraction & Generation Model (TAGM) is a sub model of the NTA’s transport model that has been developed within the OmniTrans modelling software package. Datasets used are as follows:

 2006 Census travel to work data (POWCAR),  2006 GDA travel to education survey,  2006 GDA household survey,  2006 CSO small area population statistics (SAPS).

Trip Distribution The purpose of the Trip Distribution Model (TDM) is to determine the pattern of trips between the sets of trip generations and trip attractions (called trip ends) produced by the TAGM. The TDM is a sub model of the GDA transport model and is also implemented using the OmniTrans software package. Its function is to determine to what zones the trips generated at any particular origin will travel. Given that the GDA transport model has 666 zones, the output of the TDM (for each pair of Attractions and Generations) is in the form of a matrix of trip patterns of dimension 666 x 666. A sample section of such a matrix output is illustrated in Figure A.33 below.

Figure A.33 Sample trip matrix (first 10 origins and destinations only shown)

In addition to using the forecast year trip ends as output from the TAGM, the TDM also uses known calibrated trip distributions from the base year. These distributions are based on observed travel patterns from the GDA household and education surveys and the POWCAR 88

travel to work dataset. The travel patterns represented by the base year matrices are also tested and calibrated in the GDA transport model to ensure that when assigned to the transport networks, the model outputs match closely to observed network characteristics (in terms of traffic flows and journey times). This process is known as base year model calibration.

The TDM uses the following inputs:

 All day forecast year Trip Generations and Attractions from the TAGM,  Base year trip distribution matrices for the AM-peak and inter-peak periods, covering a 12 hour period from 7 AM to 7 PM.

Each of these inputs are broken down by six journey purposes and segmented by car available and car not available persons. Mode choice / hour of travel choice and route choice feedback loop The final three stages of the model – mode choice, hour of travel choice and route choice are interlinked. They are run together iteratively in a feedback loop until an equilibrium is achieved. The mode choice and hour of travel choice are only applied to Home-Work. The hour of travel choice stage is only executed in the case of the AM-peak model19.

The feedback loop shown in Figure A.34 begins with the mode choice stage whose function is to split trips into the different modes of travel - i.e. Car, Public Transport and Soft Modes (walk & cycle20). Following this, the hour of travel choice stage further splits trips into the three modelled hours. In the final stage of the feedback loop, car and public transport (PT) trips are assigned to their respective transport networks. In this final stage, trips are assigned to specific routes (on either the highway or PT network) between their origin and destination. At this route choice stage, the costs of travel by car and public transport are calculated and then fed back up to the mode choice stage of the model to begin the loop again.

It should be noted that trips by soft modes (walk & cycle) do not take part in the route choice stage of the model, and their cost of travel is assumed to be a simple combination of travel distance and time.

19 The inter-peak is only a one hour period 20 Initially based on the base year mode shares estimated for each zone and journey purpose and time of day 89

Trip Attraction / Generation Model and Trip Distribution Model - Trips by all modes 0700-1000

Car Available Car Not Available Trips Trips (CA) (NA)

CA SM Trips CA Mech Trips NA SM Trips NA Mech Trips

Car Trips 0700 - 1000 PT Trips 0700 - 1000

Car Trips 7-9 Car Trips 9-10 PT Trips 7-9 PT Trips 9-10

Car Trips 7-8 Car Trips 8-9 PT Trips 7-8 PT Trips 8-9

SATURN - HW Assignment / Route Choice TRIPS - PT Assignment / Route Choice

Abbreviations CA Car Available NA Car Not Available SM Soft Modes Mech Mechanised Modes PT Public Transport

Figure A.34 Structure of Mode choice / Hour of travel choice and Route choice feedback loop21 The feedback loop between mode choice, hour of travel choice and route choice continues until an equilibrium position is achieved. This equilibrium is based on the premise that all trip makers wish to minimise their cost of travel, through the most cost efficient choice of mode, hour of travel and route. The point of equilibrium corresponds to a scenario where no trip maker can lower his / her cost of travel by switching either mode, hour of travel or route.

Mode choice The purpose of the mode choice stage of the model is to divide the trip matrices as output by the TDM into the different modes of travel. The mode choice stage is replicated for each of the six journey purposes. The model is hierarchical - with the top level choice between travel by soft modes (walk & cycle) and mechanised modes (car and public transport). This is followed by the split of mechanised mode trips into trips by car and trips by public transport – this lower level split is only carried out in the case of car available trips. Car not available trips have a single level split between trips by soft modes and trips by public transport.

Each level of the mode choice model represents a choice between two competing modes of travel (say mode 1 and mode 2) for trips between any origin and destination. The purpose of the model is to predict the percentage of trips that will opt to use each mode, given the

21 Note that other structures are possible 90

relative costs of travel by each. To do this, the model uses a so called “logit” formulation of the form:

1 푃 = 1 (1 + 푒−λ∆퐶)

Equation A.3 Mode choice Where P1 = the percentage of trips that will travel by mode 1 ΔC = the difference in cost of travel by mode 2 and mode 1 = C2 –C1 λ = a measure of people’s sensitivity to changes in travel costs.

Hour of travel choice The AM-peak model includes an additional stage to the traditional 4-stage transport model. This stage, known as “hour of travel choice”, is executed following the mode choice stage and captures the phenomenon of peak spreading as a response to congestion. This stage is not included in the afternoon inter-peak model, where peak spreading is not considered to be an issue.

In the AM-peak, the trip matrices that are the main input into the mode choice, hour of travel choice and route choice stages of the model incorporate trips by all modes for the three hour period 7 to 10. As explained above, the mode choice stage splits these matrices by mode of travel (Car, Public Transport and Soft Modes). Following the mode choice stage, the purpose of the “hour of travel choice” model is to split these trips down further by hour of travel – i.e. 7 to 8, 8 to 9 or 9 to 10.

It should be noted that trips by Soft Modes (Walk & Cycle) do not participate in this stage of the model – i.e. the notion of peak spreading for these modes is not considered an issue. It should also be noted that “hour of travel choice” is only applied to trips to Work (i.e. commuter trips). In the case of all other journey purposes, the base year proportion of trips travelling in each of the three am-peak hours is assumed to remain constant in the forecast year.

At the end of the hour of travel choice model, the am-peak model has six trip matrices for each journey purpose – i.e. a trip matrix for each of the three modelled hours divided into trips by Car and trips by Public Transport. These trip matrices are then assigned to the respective transport networks (highway and public transport) in the route choice stage of the model. This is the final stage of the model, and is described below.

Route choice / Trip assignment The final stage of the modelling process is the “route choice” or “trip assignment” stage. In this stage, the trip matrices that have been broken down by mode and hour of travel are assigned to the respective transport networks.

In the case of the AM-peak model, separate assignments of trips for each of the three modelled hours are undertaken for both Car and Public Transport trips (i.e. six assignments in total). These assignments are undertaken in sequence, starting with trips for the hour 7 to 8 and finishing with the hour 9 to 10. In the case of the inter-peak model, Car and Public transport trips are assigned for a single modelled hour 2PM to 3PM. Car trips are assigned using the SATURN software, while public transport trips are assigned to the combined bus and rail network using the TRIPS / CUBE software.

Highway Assignment using SATURN In the case of SATURN, the goal of the assignment process is to select a route or routes on the highway network for trips from each origin to each destination that will minimise the trip makers’ cost of travel. For car trips, the cost of travel is a combination of travel time, travel distance and fixed costs (such as road tolls, parking charges etc.):

퐶푐푎푟 = 훼. 푡 + 훽. 푑 + 푃

Equation A.4 Route choice HW Where Ccar = Cost of travel by car on a given road link t = Travel time on the link d = Travel distance on the link P = Fixed penalties / costs (road tolls, parking charges etc) α = Weightings to be applied to travel time β = Weighting on travel distance – i.e. car running costs per km (fuel, maintenance, tax & insurance, etc.)

In urban road networks (as in the GDA), where traffic congestion has a significant impact on people’s perceived cost of travel, travel time tends to have a much higher weighting than vehicle running costs (i.e. α will be much greater than β). This would not be the case in rural / uncongested road networks where travel distances tend to be longer and vehicle running costs are a significant proportion of the overall cost of travel, and also affect rural trip changing and trip frequency.

Within the transport model, travel costs are represented in time units (minutes or seconds). Hence, monetary costs (e.g. running costs, tolls or parking charges) must be converted to units of time by dividing by the perceived value of time.

SATURN undertakes a number of route assignments in an iterative fashion with the goal of minimising travel costs for all trips on the highway network. Following each trip assignment iteration SATURN undertakes a detailed simulation of traffic flow through junctions on the selected route / routes – where these junction details are coded in the simulation area. This means that in calculating travel costs, SATURN explicitly includes the travel time cost of delays caused by traffic congestion at junctions and on links – the flow delay curves also contribute to delay. In the parts of the network that are in Buffer (outside Dublin County), SATURN relies exclusively on the use of flow-delay curves that relate travel costs to link characteristics and the volume of traffic on each link.

Public Transport Assignment using TRIPS / CUBE The assignment of public transport trips is undertaken using the TRIPS / CUBE software. However, highway link travel times are passed from the SATURN assignment and used in the calculation of bus travel costs. In undertaking public transport route choice, the TRIPS / CUBE software uses a logit type choice formulation to calculate the proportion of trips that will choose competing routes based on their relative cost of travel. As trips are assigned to a combined bus and rail network, trip makers can interchange freely22 between bus, and the different rail modes (Heavy Rail, DART, LUAS, Metro etc). The generalised cost of travel by public transport has a number of components that can be represented as follows:

퐶푃푇 = 푎1. 푡푤푘 + 푎2. 푡푤푡 + 푎3. 푡푖푣 + 푎4. 푡푖푐 + 푎5. 퐹 + 푎6 Equation A.5 Trip Assignment PT Where CPT = the generalised cost of travel by public transport twk = walking time to access public transport node (i.e. bus or rail station) twt = waiting time at public transport node (= half the PT service headway) tiv = in vehicle time tic = interchange penalty F = fare (s) charged a1…a5 = weightings to be applied to the different elements of PT travel costs a6 = mode constant penalty that reflects the perceived disutility of using public transport relative to car (i.e. in terms of comfort, convenience, personal space etc.)

As all public transport costs in TRIPS / CUBE are represented in time units (i.e. minutes), the weighting a5 in the above formula is the inverse of the value of time and is used to convert the public transport fare in euros into minutes23. All other weightings were determined during the model calibration process.

22 But with a transfer penalty and a fare for each leg of the journey 23 The value of time for PT route choice is €8.10 per hour. 93

Other model functionalities In addition to the main model components the GDA transport model incorporates other functionalities that allow it to simulate travel behaviour and responses to different interventions. 1. Handling other traffic modes including Goods Vehicles (HGV’s) and Park and Ride (P&R) 2. Capturing the impact of travel demand management measures including traffic tolling, cordon charges, area charge, road user charge, parking restrictions, destination charge.

However, none of these are considered in this study. Summary of software packages used

OMNITRANS TAGM and TDM

Mode Choice, Hour of Travel Choice, TRIPS / CUBE PT Assignment / Route Choice

Highway Assigment SATURN / Route Choice

Prepare model ArcView / GIS, inputs / Analyse Accession, Office more outputs / Tools Reporting

Figure A.35 Software packages used per model step The figure above summarises the software packages used in the GDA model. This study focuses on two software packages, CUBE and SATURN, for the Mode choice / hour of travel choice and route choice feedback loop between the PT assignment and HW assignment.

E Origin and Destination zone numbers Table A.27 Origin and Destination zone numbers

City Centre North Clondalkin Stillorgan (Destination) (Origin) (Origin) 13101 20101 20152 28205 50459 13102 20102 20153 28217 50507 13103 20103 20154 28295 50518 13111 20104 20156 28296 50519 13112 20105 20161 28298 50521 13121 20106 20162 43322 50558 13122 20111 20164 43419 51561 13123 20112 20165 43444 51563 13124 20113 20166 13125 20114 25201 13126 20115 25202 13131 20116 25211 13132 20121 25212 13133 20122 25232 13141 20123 25271 13142 20124 25272 13143 20125 25273 13151 20126 25281 13152 20127 14161 20131 14162 20132 14171 20133 14172 20134 14173 20151

F SATURN runs

SATURN.111 file - Junction details file example

Do Min

Do Something

SATURN run results

Table A.28 North Clondalkin SATURN run results

Run No. Description

time

Base Base

Flow

(secs) (secs) (secs)

Delay

base

(total)

Demand

Junction

Diff above Diff above Diff above Diff

greentime Cycle

1 Base 1521 - 3000 3015 - 45984 -

2 All AB nodes on 145 and 40 routes 1521 0 3000 2928 -87 45668 -316 increased to 22kph 3 AB traffic signals increased + 10 seconds 1777 256 3000 2815 -200 33051 -6265 (where possible) 4 AB traffic signals increased +20 seconds 1947 426 3000 2876 -58 34834 -5865 (where possible) 5 AB traffic signals increased by +5, +10 1746 225 3000 2746 -269 30963 15021 seconds (where possible)

Table A.29 Stillorgan SATURN run results

Run Description

No. 25

Base Base

Flow

(secs) (secs) (secs)

Delay

base

(total)

Demand

Junction

Diff above Diff above Diff above Diff

greentime Cycletime

1 Base 2699 - 4430 3549 - 96063 -

2 All AB nodes on 145 and 40 routes 2699 0 4430 3391 -158 96941 878 increased to 22kph 3 AB traffic signals increased + 10 seconds 3110 411 4430 3415 -134 67958 -4447 (where possible) 4 AB traffic signals increased +20 seconds 3275 576 4430 3460 -89 66519 -2150 (where possible) 5 AB traffic signals increased by +5, +10 3034 335 4430 3308 -241 81691 -14372 seconds (where possible)

The description section refers to the increase in green time given to the bus route. It is important to note that the LCY (total cycle time given to the junction) is fixed meaning that the increase in green time on one arm resulted in a reduction in green time on one or more of the other arms.

In order to get the model to run the NUC values had to be changed. The NUC ‘Number of time units per cycle (+)’ has a maximum value and anything over the value creates an error. When the junction times were changed this increased the NUC value. In order to reduce the NUC value junction had to be taken far away from the study area and their NUC values reduced. This is a known issue with the GDA Model.

24 Of known values 25 Of known values 97

G Screen shots of Transport for Ireland cycle travel times

Figure A.36 Stillorgan SM route Loughlinstown Hospital – O’Connell Street

Figure A.37 North Clondalkin SM route Cherry Orchard Hospital – O’Connell Street

Source: (Transport for Ireland, 2015)

H Corridor QBC and car provision

Stillorgan Corridor

Figure A.38 Stillorgan HW image and cross section

Figure A.39 Donnybrook HW image and cross section

North Clondalkin Corridor The North Clondalkin route does not have as much segregation for the bus, this is in spite of some areas having enough road width to do so which can be seen below.

Figure A.40 Ballyfermot HW image and cross section

Figure A.41 Sarsfield HW image and cross section It can be seen between the two corridors that road space is not consistent throughout the corridor nor is it consistent between corridors.

Sources: Google.Maps and Streetmix.net

The images were captured in Google Maps with an image capture data of 2009 and 2011

100