Trade-offs summary document Trade-offs summary document Trade-offs summary document Background and Aims Context The work was initiated to provide an enhanced understanding of the relationship and trade-offs between outputs and the impact of the trade-offs on cost, revenue, wider economic benefits and customer satisfaction in order to make best use of available capacity. The National Task Force and the industry forum, Planning Oversight Group (POG) established a working group, led by Network Rail, to develop the strategic thinking and supporting analysis to progress the industry’s approach to managing the trade-offs between performance, capacity, journey time and cost to inform discussions with government, support the development of the Strategic Business Plan (SBP) and the specification of future franchises. Background and Hypothesis The overall objectives for this work can be summarised as looking to understand changes in operational aspects which have the potential to influence performance including timetable, rolling stock, train crew and incident management. With strong anecdotal evidence to support the assertions that the more trains you run the more reactionary delay will occur, the greater the speed mix the more reactionary delay will occur, the less recovery time you have the more lateness you will experience. A timetable can be measured through: Train kms travelled Trains per hour (through a particular section) Trains per hour per track Maximum white space per hour Total in service time (i.e. the sum of all journey times within the timetable) Average speed Average recovery (or slack) time per train Average differential time per train It should be noted that none of these measures is an indicator for the quality of the timetable, and the quality of the timetable plays a significant role in delivery of good performance. An attempt to compress the linkage between the timetable and infrastructure, into one simple measure of how full the network is, was expressed through CUI (Capacity Utilisation Index). CUI is an established measure supported by the UIC but tends to favour simple stretches of track rather than junctions. Network Rail Rolling stock characteristics that can impact on performance include acceleration and deceleration profile, dwell time requirements (door cycle, number of doors), length of train and maintenance cycles. More importantly though is the link between rolling stock availability and performance including number of trains in service / diagram, average turn round time, complexity of train diagrams (interworking of routes). As timetable train miles increase we would expect performance to drop if not accompanied by a corresponding increase in available rolling stock. Similarly, train crew rostering has a significant impact on flexibility when the railway operation is perturbed and the possibility that the delay will migrate onto other services through the diagram. Trade-off Outcomes The work on trade-offs informs the development of a change control mechanism through CP5 to reflect the changing requirements that may be demanded by the industry (for example between capacity, journey times and performance). The outputs will help to inform CP5 performance risks and to work towards development of a framework/toolkit to examine tradeoffs. This framework and toolkit could support the development change control mechanism to assess performance impacts of specific changes through CP5. Study Aims and Approach An independent study was commissioned with the technical work being managed by Steer Davies Gleave, whilst the case study analysis was undertaken by a joint consultancy team comprising Steer Davies Gleave and Arup. Network Rail provided the base data and has undertaken bespoke analysis to support the study. The aim of the study was to improve understanding of the trade-offs between key ‘drivers’ (inputs) of performance and performance outputs (in particular PPM). It achieved this by starting to quantify the relationships and use this to support understanding of CP5 performance risks. It has taken us some way to understanding what might be required to develop a framework/toolkit to predict impact of making trade-offs, potentially for use as part of a change control mechanism. The study was conducted between June and October 2012 by ARUP, SDG and Network Rail with support from East Coast Trains and Southeastern Trains. 1 Trade Offs summary document Case studies Trade-offs summary document Case Studies Phase Technical Analysis Outputs Phase A Findings From the Southeastern study, the following conclusions were drawn: Phase A (June – August) Phase B (August – September) Phase C (Potential work) ‘Deep Dive’ case studies – East Coast & Southeastern Evaluation of performance and trade-offs made during CP4 Detailed understanding of specific case study performance issues/relationships Strong pointers on nature of relationships, potential ‘tipping points’ and supporting metrics Identify promising avenues of further research in Phase B ‘Hot-spot’ analysis Generalising analysis of tradeoffs and relationships, primarily at Service Group level Review of CP5 schemes and other proposed changes Substantial further understanding of performance drivers and relationships High-level assessment of key trade-off related performance risks and issues in CP5 Clearer idea of what a framework (set of tools and models) to predict performance could comprise Targeted further work on understanding relationships Development of tools Framework and toolkit Phase A Approach Phase A looked in detail at East Coast and Southeastern services and sought to Identify changes to key ‘inputs’ or drivers, through development of a ‘timeline’ of key changes to timetables, management and incidents and a ‘deep dive’ analysis of the trade-offs implicit in these changes. These changes were then reviewed for changes to ‘outputs’ through a review of performance data (such as PPM and delay minutes) by period, time of day (peak and off peak) and line of route. The analysis enabled some conclusions to be drawn about the relationship between inputs and outputs. From the East Coast study some additional conclusions were made: No simple relationship: usually a number of factors acting on performance at the same time and not always immediately obvious what they are Essential to document all factors and the assumptions being made The importance of timetable change is confirmed – but not just number of trains Case studies provide clear pointers to factors affecting performance and risk areas Understanding of performance risk areas, and where proposed change is likely to have a material impact of performance could trigger change control mechanism if outside Control Period planning assumptions. In overall terms, this phase of the study concluded that: Network Rail It is demonstrable that extra trains do not always mean worse performance: timetable structure changes to reduce complexity can compensate and deliver an improved performance outcome Putting extra time into trains does not necessarily mean better PPM Additional congestion appears to be a factor at Cannon St. but further work required to understand root cause Development of high level metrics will be complex – a range of factors are interacting The Delay Minutes/PPM relationship changed after the Dec 09 timetable change. Increasing the number of train services will generally worsen performance Journey time/speed is critical – shortening journey times can result in worsening performance Complexity is a significant factor in performance – Speed mix (fast and slow trains) – Number of crossing movements at junctions Performance can differ markedly by direction – Complexity is greater towards major termini as services converge – ‘Imported delay’ is greater at the end of route (absolute delay increases with journey length) Worse performance on longer distance routes partly explained by the distance travelled (more opportunity to incur delay and greater interaction) Increasing termini capacity utilisation can worsen performance The quality of the timetable is a key driver of secondary delay. 3 Trade-offs summary document Phase A Outputs In order to understand the relationship between trade-offs and performance a number of hypotheses were tested, including: The number of trains impacts delay and PPM A change in average speed impacts PPM Timetable complexity impacts PPM Complexity is going to increase with freight growth Terminal Capacity – Time Taken Vs Number of Trains (Cannon St.) The number of Trains impacts delay and PPM Comparison of Number of Trains and Average Delay The graph compares average delay on particular routes by number of trains for the London – Severn Tunnel section of the GWML and the London – Newark North Gate section of the East Coast route. This shows the average delay per train mile along the route against the number of trains in each hour. Delays in stations are excluded. The number of trains is only a proxy as it includes all trains on all running lines. Analysis suggests a relationship in both directions and for both routes increasing busyness higher delay per mile. It shows a similar relationship for both routes, albeit at different levels. For both routes, delays are higher as services approach London As pressure for increased frequency continues in CP5, there is an implied risk to performance. 0.10 R² = 0.3877 R² = 0.3594 0.08 Average Delay per train mile 0.12 R² = 0.3319 GWML Paddington ‐ Severn Tunnel 0.06 GWML Severn Tunnel ‐ Paddington ECML Kings Cross ‐ Newark NG R² = 0.4197 ECML Newark NG ‐ Kings Cross 0.04 0.02 0.00 0 20 40 60 80 100 120 140 Number of weekday trains along route per hour Network Rail 4 Trade-offs summary document Change in average speed impacts PPM Commentary & Interpretation There is a relationship between change in delay and change in PPM, but there are outliers B Absolute Change in PPM (P01-P04) Area with high increase in PPM compared with delay change Group B PPM increases, higher than expected compared to decrease in Delay per Train On 11 out of 14 routes speed has decreased Group A A Decrease in PPM, Decrease in Total delay per Train On 9 out of11 routes where PPM has fallen and total delay per train has fallen, speed has increased. CP5 Risk: Pressure to reduce journey times likely to result in worsening performance PPM decreases, Total Delay per Train decreases Note: Data used was for Periods 01-04, 2008 & 2013. N.B. Bubbles represent size of a Service Group compared to one another Network Rail 5 Trade-offs summary document Timetable complexity impacts PPM Findings Timetable complexity is measured as % of service group train miles that interacts with other service groups. Greater complexity results in increasing risk to PPM. Long distance service groups generally have the most interaction. CP5 risk complexity being added (e.g. Thameslink) Note: Data used was for Periods 01-04, 2008 & 2013. Network Rail Increase in Interaction 6 Trade-offs summary document Complexity is going to increase with freight growth Findings This chart shows those passenger service groups that are forecast to see an increased interaction with freight in CP5. CP5 freight growth: 23% in CP5 – mostly long distance, much of it deep sea boxes Potentially complex interactions which may impact on freight: LSE Servcie Groups on routes from ports Long distance service groups Network Rail 7 Trade-offs summary document Terminal Capacity – Time Taken vs Number of Trains (Cannon St.) Time taken = GBTT at Cannon Street – GBTT at London Bridge + Lateness at Cannon Street – Lateness at London Bridge Between 09:00 and 09:59 higher time taken to get over congestion of high peak hour (08:00 to 08:59) The impact of moving from 7 trains per hour (tph) in Dec 06 TT and Dec 07 TT to 9 tph is an extra 1 minute time taken between London Bridge and Cannon Street In the Dec 09 timetable (TT), the extra trains between 07:00 and 07:59 gives an increase of 1 extra minute between 08:00 and 08:59. Findings Time taken between London Bridge and Cannon Street used to show factors relating to congestion on the route and at the terminal – indicates time waiting outside station related to turnaround issues at terminus. Relationship exists between trains per hour and time taken. Evidence that congestion from an increase in services in Dec 09 TT (2 tph 07:00 and 07:59) has affected the time taken in the following hour. There have been changes to the routing of outbound Cannon Street trains as a result of the Thameslink Programme works. Network Rail 8 Trade-offs summary document Phase B The focus of Phase B was to utilise the understanding, metrics and approaches developed during Phase A to forecast the potential impacts of changes in CP5 (whether originating from HLOS or elsewhere) on performance. Analysis was undertaken in these areas: Documentation of current performance of baseline hotspots Review of changes which will impact performance in HLOS or expected in CP5 Establishing basis for predictions – agreeing assumptions Development of metrics to permit quantified performance forecasts to be made Development of storyboard to support presentation of CP5 assumptions and principles Phase B Findings The “hot spot” analysis showed that there are key nodes in the network that are likely to cause impact, based on current levels of primary and reactionary delays. The top four locations were then reviewed against predicted growth in CP5 and the effect of major planned works as follows: Reading [Predicted 12% Growth in CP5]: There has been significant work through CP4 continuing into early CP5 to transform Reading. Extra platforms, the main line flyover and the separation of freight services from the trains heading to the South West will act to mitigate performance risks by reducing congestion and network complexity in the area. London Bridge [Predicted 5% Growth in CP5]: Major risk from Thameslink work. London Victoria [Predicted 6% Growth in CP5] and London Cannon Street [Predicted 3% Growth in CP5]. Phase A of the analysis already showed the negative impact of extra peak trains at these locations on performance. Major risk from Thameslink work. Manchester Piccadilly [Predicted 6% Growth in CP5] and Leeds [Predicted 19% Growth in CP5]: Risk from Northern Hub work. Network Rail 9 Trade-offs summary document Additionally, the highest and lowest PPM service groups were analysed and some conclusions could be drawn. For the majority of Service Groups, the expected relationship between a decrease in delay and an improvement in PPM and, and vice versa, was shown to hold true. There are two groups where this relationship is either overly sensitive compared to normal values, or not true. Journey time seems to have played a role in both of these groups: For Service Groups in which both PPM and Delay per Train have fallen, an increase in speed on those lines appears to be a partial cause in the majority of these. For Service Groups with a large increase in PPM and a large decrease in Delay per Train (above normal levels), speed has decreased on the majority of these. First Great Western form a large part of this group The analysis then went on to look at the interaction of service groups on lines of route and the conclusions were: PPM is generally lower for Service Groups which share their network with other Service Groups (same TOC or different). The more a Service Group shares the network, the more varied PPM becomes. Long Distance Services share their networks with other Service Groups the most The analysis leads to the conclusion that Long Distance services have, collectively, the worst performing PPM. Freight is going to grow substantially on lines used by Long Distance Service Groups. Long Distance Service Groups will therefore become even less self contained, which could have a negative impact on their PPM for the remainder of CP4 and in CP5. The lessons learned from the trade-offs analysis were applied to CP5 performance risk assessment in the areas of “hotspots”, CP5 changes and CP5 risks. The large enhancement projects shown here are considered to have the largest potential risk to CP5 performance: % All Growth Trains PPM Train Miles % All Trains (2013 P1-4) (2013 P1-4) (2013-2019) (2019) Thameslink Direct 18.10% 92.20% 2.40% Service Groups 17.60% (EG01) Bedford Mainline, (EG02) Brighton Mainline, (EG03) South London, (EG04) Northern Inners, (EG05) Northern Outers, (EG06) FCC : Blackfriars - Kentish Town (Joint), (HU01) Kent Mainline Off-Peak, (HU02) Kent Metro Off-Peak, (HU03) Kent Rural, (HU04) Kent Mainline Peak, (HU05) Kent Metro Peak, (HW04) South London Lines (Off Peak), (HW05) South London Lines (Peak) Thameslink Indirect 4.10% 86.60% 1.50% 4.00% (EM04) East Midlands Inter City, (HB01) Anglo-Scottish, (HB02) West Yorkshire, (HB04) West Yorkshire - Kings X - Bradford / Hull, (HB05) Anglo - Scottish (Aberdeen / Inverness), (HW02) London - Sussex Coast (Peak), (HW03) London - Sussex Coast (Off Peak), (HW07) London Victoria - Gatwick Airport Crossrail Direct 5.50% 92.80% 4.10% 5.40% (EB01) Great Eastern Inners, (EF05) Outer Thames Valley - London, (EF06) Inner Thames Valley - London, (EF07) Reading & Oxford Suburban, (EF08) Thames Valley Branches, (HM01) Paddington to Heathrow Express Crossrail Indirect 4.10% 91.60% 8.70% 4.20% (EB02) Southend & Southminster, (EB03) Great Eastern Outers, (EB04) Anglia Inter City, (EF01) London - Bristol, (EF02) London - South Wales, (EF03) London - Cotswolds, (EF04) London - West Of England GW Direct 4.30% 89.90% 11.50% 4.60% (EF01) London - Bristol, (EF02) London - South Wales, (EF03) London - Cotswolds, (EF04) London - West Of England, (EF05) Outer Thames Valley - London, (EF06) Inner Thames Valley - London, (EF07) Reading & Oxford Suburban, (EF08) Thames Valley Branches Northern Hub 6.90% 92.70% 7.80% 7.00% (EA01) North Trans Peninne, (EA02) South Trans Pennine, (EA03) North West, (ED02) Lancashire & Cumbria Local, (ED08) North Manchester, (ED09) Merseyrail City Lines, (ED10) South Manchester Network Rail 10 Trade-offs summary document Applying trade-off findings to identification of performance risks for CP5 CP5 Change Forecast Likely Performance impact Decision Basis Thameslink Reduction Specific Thameslink analysis of Thameslink stage works in CP5 and 24 tph timetable showed considerable negative risk Refranchising Reduction (Long Distance) No information about new franchise specifications, but if the new bids for the WC franchise prove a good indicator of future TOC intentions than there is likely to be an increase in services and decrease in journey times Freight Reduction More freight miles forecast Great Western Uncertain Will infrastructure enhancement be adequate to compensate for increased service frequency/changed service pattern? Northern Hub Reduction 6 tph Northern Trans Pennine from Dec 16. New works to take place during CP5 with minimal impact, other new services to begin in CP6 Crossrail No change during CP5 New works mainly confined to new tunnel and full Crossrail service will not commence until CP6 Electrification No change Construction likely to have minimal impact on performance as likely to occur at weekends and Bank Holidays. Reliability and rolling stock improvement offset by journey time reduction and rolling stock bedding in period IEP No change during CP5 Some reliability improvement but offset in CP5 at least by rolling stock bedding in In addition to the risks from enhancement projects, two further risks for CP5 were presented. Timetable Changes in the timetable have been shown to have an impact on performance through CP4, particularly where they are designed for faster journey times or change the level of timetable complexity. A significant number of timetable changes are due to take place in CP5 through HLOS and the franchising process. Timetable re-casts are a key method of improving performance within a given timetable specification as long as the interactions and trade-oiffs are fully understood. At the very least, Thameslink, West Coast, East Coast, Great Western and Edinburgh-Glasgow timetables will all be re-cast during CP5 with unclear impact at this stage due to incomplete specification and modeling. Freight growth CP5 freight growth is forecast to be around 16% by the end of CP5 and this at a time when passenger services are also changing and increasing. There will be an inevitable increase in freight / passenger interaction and the impact will depend on the type of goods (heavy load or containers), location / routes and actual growth rates. Risks were identified but no quantification has been undertaken to date. There is a a possible future phase to look at some of the interactions and risks in more detail and the output of this work will feed into the Delivery Planning for CP5. Many of these timetable changes will be in December 18 with minimal impact on end-CP5 PPM but modeling will take place during CP5. Network Rail 11 Trade-offs summary document Next Steps The next stage will be to use the evidence to understand the various inputs to performance so as to develop a potential framework and toolkit that will help the industry understand the impacts of changes during CP5. Through understanding trade-offs we can develop a set of tools and models to predict performance in such a way that the industry and stakeholders recognise the legitimacy of the modeling outputs and therefore be more accepting of associated decisions. This could support strategic planning by ensuring that performance is understood in the wider context of a rapidly developing industry. It absolutely will be required to under-pin a change control mechanism for service changes prior to detailed timetable specification and modeling and will support operational planning in the production of robust and resilient timetables. Prospects for Future Development The trade-offs analysis could be used and extended to provide a comprehensive understanding of actual performance: ‘Hotspots’, service group, category of delay The impact of incidents – self contained or major ‘ripple effects’ The distribution of delay – routine problems or only on worst x% days Relationship with performance outputs e.g. PPM, overall delay Where performance may be approaching a ‘tipping point’ – where actual performance is often ok, but where indicators suggest instability will grow if drivers of performance changed adversely Extension of trade-offs analysis to establish more general relationships that could underpin development of models to predict performance Would be of significant value to the industry Evidence from this study suggests that relationships do exist between input drivers and performance The number of interacting drivers makes development of a single all-inclusive model challenging but potentially deliverable Conclusion Better understanding of trade-offs, together with associated Change Control is seen as a fundamental part of delivering outputs in CP5. Without this mechanism the concern is that the industry will be at risk of making the wrong decision around trade off requirements for capacity, journey time and cost owing to the pressure to protect committed performance targets. Network Rail 12