CRC for Rail Innovation

Transcription

1 CRC for Rail Innovation Corridor Capacity Analysis

2 DOCUMENT CONTROL SHEET CRC for Rail Innovation Floor 23, HSBC Building Brisbane Qld 4000 GPO Box 1422 Brisbane Qld 4001 Tel: Fax: Document: Title: Corridor Capacity Analysis Project Leader: Authors: Wardrop Project No.: R3.104 Peter Pudney Peter Pudney, Phil Howlett, Scott Mackenzie, Derek Harris, Alex Project Name: Corridor Capacity Analysis Synopsis: This project has undertaken a literature review and surveyed industry experts on key issues associated with assessing network performance. This report does not aim to be an exhaustive discussion and analysis of all of the issues. Rather, it aims to accessible summary of the issues to help inform other Rail CRC projects. The main contributions of the report are to: identify performance measures that should be considered when analysing network performance identify methods and tools that can be used to assess capacity and network performance identify and discuss potential methods for increasing network capacity discuss other requirements that need to be considered to achieve a doubling of the rail task. REVISION/CHECKING HISTORY REVISION DATE NUMBER 0 [insert date] ACADEMIC REVIEW (PROGRAM LEADER) INDUSTRY REVIEW (PROJECT CHAIR) APPROVAL (RESEARCH DIRECTOR) DISTRIBUTION DESTINATION Industry Participant for Review REVISION x Established and supported under the Australian Government s Cooperative Research Centres Program Copyright 2009 This work is copyright. Apart from any use permitted under the Copyright Act 1968, no part may be reproduced by any process, nor may any other exclusive right be exercised, without the permission of University of South Australia. CRC for Rail Innovation 2 February 2010 Page i

3 Table of Contents 1. Current and future rail transport in Australia Freight Passenger Modal shift Perspectives on network performance and capacity Train and network performance indicators Capacity and utilisation Punctuality Transit time Dwell time Reliability Safety Energy and emissions Cost Customer complaints Further research Analysing capacity Literature review Methods used by Australian operators Tools Is capacity a useful measure? Conclusion Techniques for increasing network performance Faster trains Increased payload Better track Better operations Better schedules Key findings CRC for Rail Innovation 2 February 2010 Page ii

4 CRC for Rail Innovation 2 February 2010 Page iii

5 Preface This is the final report of project R3.104 of the CRC for Rail Innovation. This study was conceived as an enabling project to provide a context for other Rail CRC projects focussing on more technical issues associated with rail operations and engineering. The key objectives of the project were to: identify and analyse methods for assessing capacity and capacity utilisation on Australian rail networks identify key factors that impact on the capacity of a rail network identify and develop methods for assessing and ranking the impact of projects designed to improve the capacity of a rail network assess the improvement in capacity required to support a doubling of the rail task. Some of the key issues to be addressed were: How do you measure the practical capacity of a rail network? What is the current capacity of the major bulk freight, inter-modal and urban passenger networks in Australia? Identify the changes required to achieve a 20 per cent increase in capacity. Extrapolate these changes to give a projection of the changes needed to double the freight task. This will consider: o What are the factors that limit the capacity of a rail network? o What are the most cost-effective ways of increasing the capacity of a rail network? o What else is required in order to double the rail task? To this end, the project has undertaken a literature review and surveyed industry experts on key issues associated with assessing network performance. This report does not aim to be an exhaustive discussion and analysis of all of the issues. Rather, it aims to be an accessible summary of the issues to help inform other Rail CRC projects. The main contributions of the report are to: identify performance measures that should be considered when analysing network performance identify methods and tools that can be used to assess capacity and network performance identify and discuss potential methods for increasing network capacity discuss other requirements that need to be considered to achieve a doubling of the rail task. During the course of the project it became apparent that it may not be meaningful to attempt to determine the capacity and capacity utilisation of specific rail corridors, and that it was, at any rate, a task beyond the resources available to this project. This element of the scope has not, therefore, been pursued. The project has focussed on the core task of setting out a conceptual framework that can provide a point of reference for other Rail CRC projects to assess their potential contribution to increasing the performance of the rail sector. CRC for Rail Innovation 2 February 2010 Page 4

6 Chapter 1 Current and future rail transport in Australia 1. Current and future rail transport in Australia Rail transport in Australia is a highly diverse activity. The main rail networks and corridors are shown in the map below. Rail services in Australia include: suburban passenger services in Sydney, Melbourne, Brisbane, Perth and Adelaide rural and interstate passenger services domestic and export coal in Queensland and New South Wales mineral ores in Western Australia sugar cane in Queensland grain in Queensland, New South Wales, Victoria, South Australia and Western Australia bauxite in Queensland and Western Australia intermodal and general freight, primarily between capital cities. Over a long period of time there has been a gradual divergence between the passenger and freight businesses. Rail networks and operators, which were once highly integrated, have increasingly specialised in one or other task. Looking to the future it is therefore useful to consider the two tasks separately Freight CRC for Rail Innovation 2 February 2010 Page 1

7 Chapter 1 Current and future rail transport in Australia Bureau of Infrastructure, Transport and Regional Economics (BITRE) estimates of a doubling of the freight task over 20 years have been seized on and widely quoted through the media and industry. It is important to appreciate though that this doubling of the task is essentially an extrapolation of historical growth rates. The graphs below use data from BITRE (2009) to illustrate the growth of bulk and non-bulk freight in Australia. To place a potential doubling of the rail freight task in context it is important to appreciate the nature of current rail activities and where the sources of growth lie. BITRE has recently released an updated analysis of the Australian freight task (BITRE, 2009). This analysis highlights the respective strengths of the road, rail and sea sectors and considers where volume growth may occur into the future. Importantly, it highlights the dominance of bulk traffic in the make-up of the rail freight task and the equally significant dominance of non-bulk freight to the road task. CRC for Rail Innovation 2 February 2010 Page 2

8 Chapter 1 Current and future rail transport in Australia The report notes that, for a range of technical and commercial reasons, this fundamental pattern of traffic is unlikely to change significantly in the future. Rail is expected to continue to dominate the bulk freight market and road the non-bulk market. This is not to say that there is not scope for modal competition in the freight sector. This clearly exists in the key intercapital general freight markets. However, it is important to appreciate that in total volume terms the potential size of any shift of volume between modes is considerably smaller than the volume growth that will come from the underlying expansion of the task within each sector s core market Passenger Urban passenger services dominate the rail passenger task in Australia. This is likely to continue or even strengthen into the future. Governments are increasingly focussing on metrofication of their networks with increasing separation of routes and increasing train frequencies on core lines. The trend is toward integrated land-use planning that puts public transport at the centre of urban design Modal shift In terms of modal shift, the transport industry is clearly entering a phase where the external environment is increasingly favouring rail. Likely increases in the oil price, carbon pricing, labour market trends and the pressure for reform of road pricing and institutional structures are all likely to increase the attractiveness of rail. In response, there are a range of initiatives that the rail sector can either take of its own initiative or support in a policy context to support this opportunity for modal shift both the freight and passenger sectors. These include: working together coherently as an industry promoting fair pricing of road and rail infrastructure (whatever fair means) promoting fair regulation of road and rail track enhancing reliability introducing more regular services introducing better routes offering a more appealing travel environment improving rolling-stock design achieving compatibility with international rollingstock providing better customer information educating customers, the public and policy makers. CRC for Rail Innovation 2 February 2010 Page 3

9 Chapter 2 Perspectives on network performance and capacity 2. Perspectives on network performance and capacity The study took a two-pronged approach to researching the issues around network performance and capacity. The first was an extensive literature review to understand past research and analysis of the issues in both the Australian and international contexts. The second was consultation with practitioners in the Australian industry. In November 2008, we asked Australian industry experts to comment via questionnaire on issues of rail network performance, capacity, capacity analysis, and methods for improving capacity. The specific questions asked were: 1. Train and network performance indicators What do you consider to be the Key Performance Indicators (KPIs) for train and network operations (e.g. capacity, transit time, reliability, safety)? How do you define and measure these KPIs? 2. Analysing capacity What methods and tools do you use to measure capacity and to determine how well a corridor or network will cope with different traffic scenarios? What are the strengths and weaknesses of these methods and tools? What tools would you like to have? 3. Techniques for increasing capacity Some possible methods for increasing capacity are listed below: Faster trains o increased power/mass ratio o improved braking performance (e.g. ECP brakes) Increased payload o ensuring that current trains are fully loaded o heavier axle loads o lower-mass rolling stock o longer trains o double stacking Better track o reducing temporary speed restrictions o curve and gradient easing o additional loops, passing lanes and track duplication o increased capacity at terminals Better operations o more efficient safeworking systems o uniform train speeds o improved maintenance scheduling o improved policies for recovering from unexpected delays o improved communication Better schedules o different pricing for peak and off-peak times (congestion pricing) o fleeting o greater focus on reliability o fixed timetables versus headway running CRC for Rail Innovation 2 February 2010 Page 4

10 Chapter 2 Perspectives on network performance and capacity o greater train frequency (urban passenger) Integrated land use and transport planning Which of these or other methods for improving capacity are likely to have a high benefit/cost ratio? What methods are not effective, and why? Several industry experts provided comments by way of questionnaire. In addition, general discussions on network performance and capacity were held with a second group of industry experts. The remainder of this report summarises the findings from the literature review and responses to the questionnaire, adopting the same topic structure as the questionnaire. It concludes with a set of key findings. CRC for Rail Innovation 2 February 2010 Page 5

11 Chapter 3 Train and network performance indicators 3. Train and network performance indicators What are the Key Performance Indicators (KPIs) for train and network operations (e.g. capacity, transit time, reliability, safety)? How should these KPIs be defined and measured? The purpose of this section of the analysis was to understand the strategic context of corridor capacity, how it relates to overall rail network performance, and what the important factors are in determining the success of rail operations. In response to the questionnaire, one industry expert made three important points regarding KPIs: Appropriate KPIs depend on the user they are intended to inform. Service providers, customers, policy makers and regulators have different views of how well a rail network is performing, and so will have different KPIs. It seems that a problem common to most of the published indicators is that they were not originally designed to measure service quality as perceived by customers, but rather to provide feedback to railway managers (BTCE 1999, p. 13). Where possible, KPIs should be based on data that is collected in the course of normal operations, otherwise the effort of collecting data can be too onerous. KPIs must be unambiguous, so they are not misinterpreted or misused. As with almost any performance data, the distribution of performance may be more important than the average performance. The KPIs identified through the literature review and by the questionnaire respondents are described in the following sections. However, many of the key KPIs are not well defined. This section should be read and considered in this context. This issue is discussed further in Section Capacity and utilisation High capacity utilisation is desirable, particularly to policy makers and regulators, because it indicates economies of density. Unfortunately, capacity and capacity utilisation are difficult to define and difficult to measure. Capacity and utilisation also have high dependence on the type of railway and on the types and mix of trains, which may vary with the time of day. The theoretical capacity of a track section can be found by dividing some nominal time interval by the headway of the trains operating during that interval. For example, if trains can operate with a six-minute headway then the theoretical capacity is 10 trains per hour. However, this definition of capacity does not allow for any delays, and so is not useful except as an upper limit. The operational capacity of a track section must take into account random delays due to signalling systems, station dwells, train performance and track speed limits, and so is much more difficult to define and to calculate. Furthermore, any statement of operational capacity should be qualified by a corresponding reliability measure. Service providers, policy makers and regulators are generally interested in the mass or volume of freight or the number of passengers that can be transported, whereas train planners are more interested in the number of train paths that can be provided to meet demand. These two measures are related by the capacity of the rollingstock. CRC for Rail Innovation 2 February 2010 Page 6

12 Chapter 3 Train and network performance indicators The capacity of a rail system also depends on fleet size and availability. Issues associated with defining and measuring capacity and capacity utilisation will be addressed in more detail in Section Punctuality Assessing punctuality is not straightforward: Current rail industry practice [in the freight market] is to publish punctuality indicators based on train arrival times, but customers are more interested in cargo availability times. One of the insights from monitoring was that train punctuality and cargo availability statistics can give different measures of service quality (BTCE 1997, p. 40). Comparing actual train departure and arrival times to scheduled departure and arrival times gives a reliability measure for only part of the transport process. Rail transport users are interested in the reliability of the entire delivery process: pick up, transport to the terminal, loading, rail transport, unloading, and delivery. The rail freight transport portion of the process loading, rail transport and unloading involves several organisations including terminal operators at the origin and destination, train operators and rail network managers. For suburban passenger networks, punctuality is usually indicated by the proportion of trains arriving on time. This measure of on-time reliability needs to be considered in the context of train frequency, train cancellations and trip durations, since there is considerable scope for making trade-offs between these indicators Transit time Transit time (or average speed) is a useful KPI for some freight customers, and is a good measure of how rail freight transport compares with road transport for that portion of the market for which transit time is important. In this context, however, the door-to-door time, or some other measure, may be more appropriate than just the train travel time. For passengers, transit time is usually defined to be the time it takes to travel from the origin station to the destination station, including interchange durations Dwell time Dwell time indicates how long a train is stationary. Planned dwell can be determined from working timetables with little additional effort. There can be many reasons for dwell, including crew changes, crew breaks (particularly with driver-only operation), loading or unloading, refuelling, and waiting for track segments to become available. Dwell time needs to be interpreted carefully, since not all dwells indicate capacity issues Reliability Reliability can be indicated by: the proportion of trains that arrive on time the number of rolling-stock failures, or the time lost due to rolling-stock failures the proportion of the fleet that is available delays due to track failures, speed restrictions and possession over-runs the proportion of scheduled services that are actually run. CRC for Rail Innovation 2 February 2010 Page 7

13 Chapter 3 Train and network performance indicators 3.6. Safety Safety data is collated and reported by the Australian Transport Safety Bureau. The data collected includes Signals Passed at Danger (SPADs), derailments, injuries, load shifts, and other reportable incidents Energy and emissions 3.8. Cost Climate change and dwindling oil reserves mean that energy efficiency is becoming an increasingly important performance indicator, of particular interest to policy makers and regulators. For freight, energy efficiency should be measured in net tonne-km per megajoule, not gross tonne-km per megajoule. For passenger transport, energy efficiency should be measured in passenger-km per megajoule. To compete with other transport modes, rail has to operate at a lower cost or else provide a superior quality of service Customer complaints One passenger service provider use the number of customer complaints per million journeys as a KPI Further research Many of the KPIs used by the rail industry do not have precise definitions that are used consistently throughout the rail industry. In November 2008, the CRC for Rail Innovation started a new project, Train Planning Assessment Tools, which aims to develop a set of precisely-defined KPIs that can be used by network operators, train operators and rail customers to assess train plans and rail network performance. The project summary identifies some additional KPIs that might be important: What makes a good train plan? Traditionally, train plans are assessed in terms of capacity utilisation and transit times. However, these terms are neither well defined nor consistently applied. In any case there may be more important criteria for network and train operators. Better criteria might include: Feasibility: the knowledge that a train plan can be implemented without exceeding the capabilities of the track, safe working systems or train controllers. Flexibility: the ability to change a train plan in response to changed requirements from one train operator without adversely affecting other trains or train operators. Flexibility is also related to the ability to add new trains without adversely affecting existing trains. Robustness: the ability of trains to meet key arrival times when subjected to unplanned disturbances. Typical disturbances include late entry of trains onto the network, slow running, and train and track failures. Train planning processes assume that trains will run to schedule but it would be desirable if schedules were designed in such a way that unplanned disruptions could be easily managed. CRC for Rail Innovation 2 February 2010 Page 8

14 Chapter 3 Train and network performance indicators Timeliness: the degree to which trains arrive at key destinations to meet critical deadlines. Key destinations include passenger stops, locations where trains are loaded or unloaded, and locations where crews are changed. For many operators, meeting a designated arrival time at a key location may be more important than minimising transit time. One of the key requirements of a rail network is reliability. Reliability is typically measured by the proportion of services for which the train meets key departure and arrival times. Robustness is a characteristic of a train plan that allows reliable operation. Methods for calculating robustness will be a key outcome of the new project. CRC for Rail Innovation 2 February 2010 Page 9

15 Chapter 4 Analysing capacity 4. Analysing capacity What methods and tools can be used to measure capacity and to determine how well a corridor or network will cope with different traffic scenarios? What are the strengths and weaknesses of these methods and tools? What tools are required? The capacity of a rail network is a measure of how much can be transported on the network during some time interval. Capacity is usually thought of in terms of the number of trains that can be run during some interval. This definition may be useful for network managers, but is not particularly helpful for transport planning, where the number of passengers or the mass or volume of freight are more appropriate measures. There are many different ways to define capacity. Furthermore, many of the standard definitions of capacity and capacity utilisation are not particularly useful. Spare capacity is of no use if it is at times when there is no demand, if it is so fragmented that additional trains cannot be run, or if adding new trains reduces other key performance measures such as reliability. Capacity can also be difficult to measure Literature review BTE (1995) list the following as indicators that a railway line is approaching maximum throughput: decreasing ability to meet existing service specifications diminishing ability to handle new traffic offered increased time required to recover from disruptions difficulty in maintaining the condition of the track structure due to the traffic diminishing ability to increase daily production. However, quantifying the number of trains that can run reliably on a network is not easy. Abril et al. (2007) give a good overview of the problem. Calculating the number of trains that can be run is straightforward if you assume that all trains are identical, services are evenly spaced throughout the day, and train movements are not subject to disruptions. In practice none of these assumptions are true, and so practical capacity will be significantly lower than theoretical capacity. Furthermore, there will generally be a trade-off between capacity, reliability and other key performance indicators. Abril et al. (2007) identify infrastructure, traffic and operations parameters that influence capacity. These include: the number of tracks the definitions of lines and routes junctions speed limits safeworking systems the mix of train types train speeds train priorities concentration of traffic at particular times timetable requirements track maintenance quality of service requirements. CRC for Rail Innovation 2 February 2010 Page 10

16 Chapter 4 Analysing capacity Ferreira (1997) discusses the relationship between railway infrastructure and benefits including transit time, capacity and reliability. He also gives a good overview of research into optimisation of train schedules, reliability of train schedules, classifying delays, train dispatching, rolling stock scheduling, infrastructure planning and track maintenance planning. Kieran (2001) discusses methods and practices in pricing railway track access, and addresses the issue of how much capacity each train uses. He identifies the important factors as: speed limits, which are determined by track geometry, track quality and other external factors the distribution of train speeds and priorities loop spacing on single-line track, and the distribution of loop lengths relative to train lengths the proportion of multiple tracks the spacing of crossovers on multiple tracks signal block spacing train power, mass and length traffic peaking and directional imbalance disruptions. He also identifies several methods that could be used to define the capacity of a track: the number of trains that will cause the system to lock up the number of trains that can be run before the slowest train becomes unacceptably slow the number of trains that maximises production the maximum number of trains that can be run without excessive queuing delays the maximum number of trains that can be run and still allow recovery from major service disruptions within an acceptable time limit. One of Kieran's conclusions is: Almost all real situations require someone to build a model of the operation and establish quantitative objectives (the model could be a mental picture, some numbers on the back of an envelope, a spreadsheet, or an elaborate computerized simulation). Once the model is in place, the challenge becomes finding the common ground for communication. Gibson (2003) also discusses the allocation of capacity, and identifies the following key features of rail capacity: different trains can require train paths with significantly different speeds and stopping patterns the value of a train path to an operator depends on the overall pattern of services of that operator even minor reallocation of capacity can require significant adjustments to the train plan, affecting many operators some services are required to meet social obligations, and this impacts on the market value of capacity it is difficult and costly to produce train plans, since automation of the process is not yet fully developed. CRC for Rail Innovation 2 February 2010 Page 11

17 Chapter 4 Analysing capacity The costs of increasing capacity can be high, and so it is often more practical, though still difficult, to reallocate existing capacity. Gibson goes on to discuss three mechanisms for allocation capacity: administered, cost-based, and market based. Brewer and Plott (1996) describe a possible mechanism for auctioning train paths that addresses the issues associated with market-based allocation of capacity, including interdependence of train paths. The International Union of Railways (UIC) (2004) has developed a standard method, called UIC 406, for assessing the capacity of a rail network, and defines capacity as the total number of possible paths in a defined time window, considering the actual path mix. The method recognises that capacity depends on the mix of traffic and on the train plan, and calculates capacity utilisation by compressing an existing timetable and measuring the proportion of time used on each segment. Abril et al. (2007), Landex et al. (2006) and Landex (2007) give overviews of the UIC method, and describe different ways it may be used. As an illustration of the UIC method, the train graph below shows the ARTC 2008 timetable for Dry Creek to Crystal Brook. The horizontal axis represents one week of operation; the vertical axis is location, from Dry Creek at the bottom to Crystal Brook at the top. The UIC method considers each loop-to-loop segment separately. The trains are placed on each segment in the order specified by the timetable with the minimum allowable spacing between them. On segments that can be occupied by only one train at a time, the segment utilisation is the sum of the segment running times plus the sum of the minimum inter-train spacings, divided by the duration of the scheduling interval. If we assume a section clearance time of five minutes then the results for Dry Creek to Crystal Brook are: 19% Snowtown - Redhill 18% Nantawarra - Snowtown 17% Long Plains - Bowmans 17% Redhill Rocky River 15% Bowmans Nantawarra 14% Dry Creek Bolivar 14% Mallala Long Plains 12% Two Wells Mallala 12% Bolivar Two Wells 9% Rocky River Crystal Brook A minor shortcoming of the compression method is that it does not consider the effects of changing the sequencing of trains, which may allow even further compression. A more significant shortcoming of compression methods is that they do not indicate how much usable capacity remains, or the relationship between capacity utilisation and reliability. The following train graph shows the Dry Creek to Crystal Brook trains spaced as closely as possible. CRC for Rail Innovation 2 February 2010 Page 12

18 Chapter 4 Analysing capacity The duration of this timetable is about 20 per cent of the original timetable, and the resulting capacity utilisation is close to 100 per cent, but the timetable is clearly not practical a delay to any train will cause cascading delays to later trains. The UIC considers the practical limit for daily capacity utilisation to be about per cent, and the practical limit for peak capacity utilisation to be about per cent (UIC, 2004). Burdett and Kozan (2006) define absolute capacity as a theoretical value (overestimation) of capacity that is realised only when critical section(s) are continuously occupied. They then describe methods for calculating absolute capacity on lines and networks. As with the UIC method, capacity depends on the mix of trains. The simplest case is a bottleneck section of a single line with no intermediate signals. In this case, absolute capacity is determined from the proportion of each train type and direction, and the section running times for each train type and direction. For more complicated scenarios, involving sections with intermediate signals or dwells, or involving networks of simple lines, calculating absolute capacity may require solution of an optimisation problem. Because absolute capacity is an upper bound on capacity, if the method shows that there is insufficient capacity then there is no point in trying to develop a schedule. However, even if the method shows that there is sufficient capacity, it is no guarantee that a practical schedule can be found. Cambridge Systematics (2007) has analysed the capacity of the United States rail networks for the Association of American Railroads (AAR), and produced maps that show capacity utilisation on each of the primary corridors. However, the methods used to determine the capacity of each corridor in this exercise are overly simplistic. They use three dominant factors to estimate the capacity of each primary corridor: CRC for Rail Innovation 2 February 2010 Page 13

19 Chapter 4 Analysing capacity the number of parallel tracks the type of safeworking system automatic block signalling (ABS), centralised traffic control (CTC) or train order working (TO) the mix of train types and train speeds. They use the following table, based on expert opinion, to determine the capacity of each corridor in trains per day: Tracks Safeworking system Multiple train types Single train type 1 TO ABS CTC TO ABS CTC This data is not particularly convincing. For example, the ratio (trains per day with a single train type) / (trains per day with multiple train types) varies from 1.2 to 1.6. Furthermore, the Hunter Valley in Australia has is a mixed train flow of almost 100 trains per day, which is much greater than the numbers suggested by Cambridge Systematics (Wardrop, personal communication). Mattsson (2004) considers interactions between capacity and reliability. The paper: describes the UIC methods for determining capacity utilisation, and an alternative method used by the Swedish National Rail Administration to calculate the proportion of time each track section is occupied gives a simple example that illustrates the dependence of capacity on the number of tracks, the speed of the traffic, the presence of an extra station, and the heterogeneity of train speeds shows that having a mixture of fast and slow trains has a significant detrimental impact on capacity shows how the fundamental law of traffic flow = speed x density can be used to calculate approximate capacity presents analytical methods of delay analysis, including a method for analysing probable delays when you know the distribution of train speeds but don't have a timetable gives definitions of delays presents an example where the capacity of a line varies from five trains per hour with a reliability of 70 per cent, down to two trains per hour with a reliability of 99 per cent Mattsson concludes: microsimulation is the only reasonable way to model in any detail the very complex processes by which different trains interact with each other and with the infrastructure, [ but requires ] detailed data about the infrastructure, the performance of the trains and, perhaps most importantly, about the timetable. Luscombe and Lee (2005) describe a method for identifying the best place to add a crossing loop to a rail corridor. They use the Schedulemiser train planning system to generate many different train plans for a corridor, then analyse these plans to determine where the corridor is most congested. In their example, they consider five candidate track sections to receive a new loop, and three possible CRC for Rail Innovation 2 February 2010 Page 14

20 Chapter 4 Analysing capacity loop positions for each track section. For each of the 15 possible loop positions, Schedulemiser software was used to determine the new average delay on the corridor. In some cases, adding a loop shifted the congestion location and increased the average delay. The best location reduced average delay by 7 per cent. Finally, Parkinson and Fisher (1996) describe factors that influence passenger rail transit capacity and procedures for estimating practical rail transit capacity in North America. They define achievable capacity as the maximum number of passengers that can be carried in an hour in one direction on a single track allowing for the diversity of demand and use the formula C = L x T x D where C is the achievable capacity in passengers per hour, L is the line capacity in trains per hour, T is the train capacity in passengers per train and D is a loading diversity factor. Line capacity is defined as L = 3600/(t s + t d + t m ) where t s is the minimum train separation supported by the control system, t d is the maximum station dwell and t m is an operating margin, all measured in seconds. The steps for estimating capacity are: 1. Determine the weak link. The following steps will be calculated for the corridor bottleneck. 2. Calculate signalling system throughput. The report gives typical separation times for three aspect, cab control and moving block systems. Speed restrictions may increase the separation. 3. Estimate station dwell time. 4. Select an operating margin. The operation margin is used to absorb any unplanned delays. Higher margins give greater reliability, but less throughput. 5. Estimate a loading level in passengers per metre of car or passenger per car. 6. Select a loading diversity factor. Uneven loading of passengers into trains and cars reduces the overall capacity. Typical factors are 0.8 for heavy rail, 0.75 for light rail, and 0.6 for electric commuter trains 7. Calculate the achievable capacity using the formula above. The report cautions that: The best method to estimate capacity is with a complete system simulation involving models of the signalling system, power supply system and train performance. The methodology involves simplifications and approximations. Correctly applied with reasonable input values, it should estimate capacity with ±10%. Parkinson and Fisher conclude that the best ways to increase transit system capacity are through advanced train control systems and shorter station dwells Methods used by Australian operators The respondents to our questionnaire identified three classes of methods used to assess capacity and capacity utilisation: CRC for Rail Innovation 2 February 2010 Page 15

21 Chapter 4 Analysing capacity Static analysis. By examining a train plan, the proportion of time each track section is occupied during some time interval can be determined. The occupation time for each section should include the time required for the rear of each train to clear the section, and any delays associated with signalling or route setting. This is essentially the compression method for calculating capacity utilisation that is recommended by the UIC (2004). Acceptable levels of capacity utilisation depend on the track configuration, on the mixture of traffic, and on how well the network is managed. Utilisation greater than about 70 per cent is generally unsustainable. The main advantage of static analysis is that it is easy to perform, and can give a quick indication of whether or not a proposed train plan is likely to be sustainable. A key disadvantage of static analysis is that it does not consider the impact of random delays, and so relies on prior experience with similar scenarios to determine whether the calculated utilisation is likely to be sustainable in practice. Another disadvantage of static analysis is that it does not consider alternative train plans that may better utilise the available capacity, or may be more reliable. Dynamic simulation. The aim of dynamic simulation techniques is to incorporate the impact of random variations into the assessment of train plans, and to predict the reliability of train plans. The sequence of train movements is based on an input train plan, but the movement times and durations are subjected to random delays with given distributions. Alternatively, operational dispatch rules can be used to change the sequence of train movements on a section if some trains are delayed significantly. These models can be calibrated by adjusting the delays so that the output of the simulation matches historical operations. This type of modelling can be particularly effective if the models are built, run and interpreted by a team that has skills in both modelling and in rail operations. The disadvantage of dynamic simulations models is the time, effort and data required to build, calibrate, run and interpret the models. Automated train planning. One of the main reasons for measuring capacity and capacity utilisation is to determine whether or not a network can support a proposed level of service. An obvious answer to this question is that if you can generate a (robust) timetable for the proposed services, the network has enough capacity. But most timetables are generated using essentially manual processes, and it can take many weeks to generate a single feasible timetable. Recently, however, ARTC, Alex Wardrop and others have been using automated train planning tools to generate hundreds of different timetables for a given set of train requirements (Pudney and Wardrop, 2008). These timetables can be further analysed to determine congestion patterns and capacity constraints that are independent of precise timetables or even independent of train departure times (Luscombe & Lee 2005, ARTC 2007a, ARTC 2007b, ARTC 2008). One of the disadvantages of these systems is that they are currently configured to search for train plans that minimise delays. As a consequence, the resulting timetables may not be particularly robust. Further work is required to incorporate robustness into the search objective. With any of these methods, but in particular with the simulation and optimisation methods, it is almost impossible to capture all of the nuances of train operations, and so the opinions of drivers, controllers and signallers are immensely valuable. The Australian Rail Track Corporation (ARTC) has used three methods to analyse capacity on their East West, North South, and Hunter Valley networks (ARTC 2007a, ARTC 2008, ARTC 2007b): The ARTC theoretical capacity model calculates the daily capacity of a link by dividing the duration of the day by the section running time on the longest section plus an allowance for safeworking. The practical capacity is assumed to be 65 per cent of the theoretical CRC for Rail Innovation 2 February 2010 Page 16

22 Chapter 4 Analysing capacity 4.3. Tools capacity on single-line track, and 70 per cent of the theoretical capacity on double-line track. The ARTC theoretical delay model calculates delays to trains by assuming that each cross within a network sector causes a fixed delay, based on the average section running time for the sector plus an allowance for safeworking. The ARTC Sketch model uses Sketch scheduling software to generate timetables for a given set of trains, and then analyses the delays to each train. The Sketch software uses the scheduling method developed by UniSA and TTG Transportation Technology to generate many alternative schedules, from which the best can be selected (Pudney and Wardrop 2008). Sketch has also been used by a mining company to determine how many more trains could use the network. In this case, additional pre-processing software was developed to determine the best sequence of trips to each of the mines; Sketch was then used to calculate detailed train plans that minimised the total delay experienced by the trains on the network. The number of trains was then increased until delays and congestion became too severe (Pudney and Wardrop, 2008). In their analysis of the Hunter Valley, ARTC (2007b) conclude that the best way to determine whether or not a network has sufficient capacity to support proposed train services is to generate a fully-resolved timetable. Manual planning methods take too long, but with automated scheduling tools it is possible to generate and evaluate hundreds or even thousands of alternative, feasible timetables. ARTC use Sketch software to generate timetables for given sets of train requirements (departure times, train lengths, section running times), with custom pre-processing software to generate different train requirement scenarios, and custom post-processing software to analyse the results. The Hunter Valley Coal Chain Logistics Team is a joint venture between two track owners, two train operators, a port owner and a port operator. By taking a cooperative approach to logistics planning, the Logistics Team has been able to achieve a substantial increase in throughput, from 69 million tonnes in 2002 to 80 million tonnes in The aim of the Logistics Team is to develop coal chain plans that maximise the rate at which coal is delivered from mines to vessels. The Logistics Team has developed tools to help build and coordinate vessel loading plans, cargo assembly at the port, and the scheduling of train trips from mines to the port. However, the planning process is essentially manual. It takes about six hours to develop a single feasible plan for the next three days. Plans are continually refined as circumstances change. Falko software is being used by an Australian operator. Falko is a timetable construction and rail network simulation tool developed by Siemens. OpenTrack is being used to perform randomised network simulations to model network performance at current and projected traffic levels. The main aim of this work is to determine the reliability of arrival times. There are some simple methods that can be used to visualise network operations and identify potential problems: the number of weekly events at key locations the distribution of event times at key locations congestion maps the average speed of trains, compared to the maximum possible speed. CRC for Rail Innovation 2 February 2010 Page 17

23 Chapter 4 Analysing capacity Beyond these simple methods, there are three classes of tools that can be used to help evaluate plans for increasing railway capacity: train simulation network simulation timetable optimisation. These tools are also discussed by Hollins et al. (2004) and by Wardrop (2007b) Weekly events The following diagram shows the number of events occurring at each location in the ARTC Master Train Plan (5 August 2007). This plan is for the ARTC interstate network from Kalgoorlie to Broken Hill and Albury. An event is an arrival or departure at a location; however, arrivals and departures that coincide are counted as a single event. The radius of each circle is proportional to the number of events. The table below shows the number of arrivals and departures at key locations on the East West corridor, and the number of distinct events. Location Arrivals Departures Events Parkeston Tarcoola Spencer Junction Port Augusta Crystal Brook Coonamia Dry Creek Islington Keswick Laverton CRC for Rail Innovation 2 February 2010 Page 18

24 Chapter 4 Analysing capacity Tottenham South Dynon Wodonga Platform Kanandah Event distributions The number of weekly events occurring at a location is not necessarily a good indicator of congestion, since it does not tell you anything about the spacing of events. The following diagram shows the number of arrivals and departures per week at major locations on the network. We have not used locations on the edge of the network because we are missing train data from adjacent networks. The horizontal axis of each graph represents time, from midnight Sunday to midnight a week later. Each red vertical bar represents an arrival (top half) and departure (bottom half); the length of the bar indicates the rate at which events are occurring. You may have to magnify the graphs to see the detail. Parkeston Tarcoola Spencer Junction Port Augusta Coonamia Crystal Brook Dry Creek Islington Keswick Laverton Tottenham CRC for Rail Innovation 2 February 2010 Page 19

25 Chapter 4 Analysing capacity South Dynon Wodonga Platform Kanandah Train speeds Plotting the cumulative number of arrivals against time, along with a line indicating the average arrival rate, is perhaps a better way of visualising the variability of train spacing at each location. The mean and standard deviation of inter-arrival times may also be good measures of activity. These types of visualisation produce congestion maps that show when and where congestion occurs. It is important to realise, however, that the locations and times of congestion are highly dependant on the timetable even small changes to the timetable can cause significant changes to congestion locations and times. By analysing congestion patterns for many different timetables, you can get a better idea of which locations are causing congestion (Luscombe & Lee 2005, ARTC 2007a, ARTC 2007b, ARTC 2008). Dividing the average speed of a train by its free speed that is, the speed it would achieve without delays due to interactions with other trains also gives a good indication of the level of congestion on the network. Free speed can be determined using simulation Train simulation The motion of a train along a track can be simulated to answer questions such as: how long will it take to travel between points on the journey? how fast will the train be travelling at particular points on the track? The simulation must take into account: traction and braking forces available at various train speeds the length, mass and mass distribution of the train resistive forces due to rolling resistance and aerodynamic drag forces on the train due to track gradient and curvature track and train speed limits. The simulation may also allow different driving modes or even optimal driving. Some single-train simulators used in Australia include: CRC for Rail Innovation 2 February 2010 Page 20

26 Chapter 4 Analysing capacity Energymiser (TTG Transportation Technology), which is based on the journey optimisation system developed by TTG Transportation and the University of South Australia. This system can calculate the most energy-efficient driving strategy for given journey times, and is the engine inside the Freightmiser driver advice system. OpenTrack Rail\\Train (TTG Transportation Technology) simulates the motion of a single train along a route. It is also able to simulate a fleet of trains travelling along a corridor, and the interaction of these trains with the signalling system. Rail\\Train is based on the classic MTRAIN simulator. It is typically used to calculate minimum-time journeys, but other control strategies can be specified. SimTrain (Samron Pty Ltd) Network simulation RailSim Train Performance Calculator (SYSTRA) TSR Calculator. In 2006, the University of South Australia developed a simple train simulator specifically designed to calculate the impact of temporary speed restrictions on journey time (Rail CRC Project 154). For each temporary speed restriction, the tool calculates the time that will be lost at that speed restriction by a standard train, and the time that will be regained if the restriction is removed. (These two times are not necessarily the same, as the speed profiles for adjacent speed restrictions may overlap). This tool is described further in Section A network simulation tool is used to simulate the motion of many trains on a rail network to answer questions such as: is the timetable feasible? where and when do trains interact? how much delay is experienced by each train? where and when is the network congested? The information required by the network simulation tool includes: a description of the rail network a description of safeworking systems trip descriptions, which specify the sequence of locations to be visited by a particular class of train and the running time between adjacent locations departure times and trip types for each train. A network simulator will generally use simple rules, such as order of arrival, or priority, to determine which train should go first when two or more trains need to use the same track segment. (Tools which optimise such decisions are described in the next section.) Some network simulators used in Australia include: Falko (Siemens) CRC for Rail Innovation 2 February 2010 Page 21