AN INTRODUCTION TO RAILWAY PERFORMANCE MEASURES

CONFUSED BY WHAT ALL THE DATA MEANS?

Well you are not alone; I’ve even come across senior railway managers who don’t understand it. So, don’t feel alone, there are plenty of others out there! Partly this confusion arises between the rail industry performance regimes and the public performance measure. The former is generally known as “Schedule 8” and the latter as “PPM”.

This Blog, and subsequent posts are designed to introduce interested observers to the archane word of railway performance stats.

SCHEDULE 8

Between Network Rail (NR) and each Train Operator (TOC) is a contract which governs access for trains to the national rail network for which the operator pays a set of track access charges.  The main contract document is likely to be less than 40 pages. The bulk of the document (frequently runs to more than 300 pages) is a set of supporting schedules. The 8^th Schedule deals with the operation of a performance regime. The basis is that in return for paying track access charges, the train operator has a reasonable expectation that NR to provide and maintain a network on which those services can run. Where a train operator can run the vast majority of trains on time then NR becomes entitled to what might be described as bonus payments. These are designed to reflect that when a train service is both reliable and regularly runs to time then patronage rises; know in the rail industry as the “Fare Box Effect”. The counter to that is also true, where trains are seen as being unreliable and suffer frequent cancellations, patronage falls. In these cases, where NR are at fault, then the TOC receives additional payments to top-up the adverse effect on their “Fare Box”. Conversely when it is the TOC that is the cause of delay, possibly because they fail to maintain their rolling stock or employ sufficient numbers of staff to operate the service, then NR becomes entitled to receive compensatory payments. In the main these are designed to off-set any payments that NR becomes liable to other train operators, whose trains become delayed and over-crowed as a result of the other operator’s poor performance.

To make all this vast sea of data manageable, performance is calculated on a daily basis. Each day’s calculations are then combined into a single reckoning every 28 days. In railway jargon, these are “Railway Accounting Periods”. Thus, it is quite possible for a train operator to experience two or three days of very poor performance from Network Rail, yet at the end of the period they are still required to make a bonus payment to Network Rail.

DELAY ATTRIBUTION

Running alongside the Schedule 8 regime, is a system known as TRUST DELAY ATTRIBUTION (TRUST DA). TRUST (Train Running Under System TOPS) is a database of every train when is planned to operate across the national rail network. TRUST DA then takes this information (of planned services) and using data from the signalling system compares how each individual train has performed at almost every station is stops at and at a number of key junctions. TRUST DA makes a comparison between planned and actual and where the difference is 2 mins or less then TRUST DA automatically records the reason for the delay as “”ZZ – Unknown other”. Where the delay is 3 mins or more, then TRUST DA flags the train up for investigation. At that point a NR employee makes what he believes to be an attribution to a main cause. This will be to NR or to the TOC and will either add this to an existing delay cause (for example a problem with the signalling system) or create a new cause (perhaps “Unruly customers). At this point the Train Operator can contest the cause of the delay if, for example they believe that the delay was caused by Network Rail rather than themselves. It is alleged that across the whole UK rail network 300+ people are engaged in running the TRUST DA process of attributing delays to specific causes.

Below is an example of a train schedule and what data gets recorded by NR. (Data sourced from www.realtimetrains.co.uk). It shows the journey of 2L74 the 06:13 from East Grinstead to London Bridge on Friday 8^th December 2017.

On this day, a major power failure was suffered by NR following a fault on the national grid supply at Ashburton (South Norwood). Having Left Lingfield 1 minute late, it was then held at Hurst Green until power was restored, departing there 42 minutes late. Having regained a minute, a minute was lost at Riddlesdown, as presumably the train was close to full and bursting making it difficult for waiting passengers to board. Further delay occurred at East Croydon (48 mins late). A fast run to London Bridge saw it arrive there 45 mins late.

So, what delays did the train experience? The 1-minute delays at Lingfield and Riddlesdown would automatically have been treated as two events attributed as “ZZ”. 41 minutes at Hurst Green would have been attributed to the NR Power outage along with the 6 minutes at East Croydon. Its quite possible that some of the delay was down to the numbers of passengers wanting to leave the train to transfer to a Victoria service whilst a platform full of passengers were waiting to join. In some cases, this delay may be attributed to the TOC, however, following such a major disruptive event, what might best be described as ‘reactionary delays’ will also get attributed.

So how late was the train? As can be seen although 2L74 is due into London Bridge at 07:11, the time advertised to passengers is 07:14. So having left East Croydon 48 mins late it is recorded as arriving into London Bridge 45 mins late for internal railway purposes, but for passengers it was just 42 mins late.

This is the first area where we start to get controversy. The train departed East Grinstead on-time and arrived into London Bridge 45 mins late. During the journey it accrued two 1-minute delays, a 41-minute delay and a 6-minute delay. On some longer journeys it is quite possible for a train to depart on-time and arrive at its destination 15 minutes late, but as all the delays along the route were in multiples of 1 and 2 minutes such that there was no attributable delay.

So that’s the first conundrum; how late a train is? Also the causes for its total delay are not the same. Trains may frequently experience more delays on-route, than how late they are at destination. The later because the TOC has given itself a small additional time advantage between when the train is planned to arrive and when they advertise its arrival time to its passengers.

Cancellation and significant lateness (CaSL)

To provide TOCs with a degree of protection from on-going serious disruptive events attributable to NR, a separate calculation is used when one of the following events are recorded:

·         It is cancelled at origin

·         It is cancelled en route

·         The originating station is changed

·         It fails to make a scheduled stop at a station

·         It is significantly late (ie it arrives at its terminating station 30 minutes or more late).

PUBLIC PERFORMANCE MEASURE (PPM)

This a the most commonly quoted performance statistic. It is simple a ‘yes’/’no’ test. Did the train arrive at its destination at the time advertised to the pubic (as opposed to internal working time) or within 4 mins and 59 seconds of that time. If it did it registers as being ‘on-time’. If it arrives 5 minutes later than advertised, then it is described as a ‘PPM Failure’.

The problem with this statistic is that it gives every train an equal weighting; to both delays or cancellations, the value of a lightly used train in the middle of the day scores just as much as if it were a loaded commuter train. In part this can influence behaviour, should a train be turned round short of its destination and sent back to London on-time or should we allow it to run late? Also, there is no indication of who is to blame. Both NR and a TOCs own delays are captured as a single figure

This graph, (taken from http://trains.im) gives the PPM figure over the past 10 weeks for Govia Thameslink Railway (GTR); highlighted is the figure for Friday 8^th December.

On Friday morning there was a catastrophic failure of the National Grid Supply at Ashburton (South Norwood) which rendered the signalling system from Three Bridges to Streatham and Sydenham useless. Against this cause there were a total of 696 PPM failures. But because the disruption was so widespread a further 600 PPM failures were attributed to NR against ‘Network Operations’. Coupled with a point failure at Hove early that morning NR, in the South East, were responsible for a total of 1,315 PPM failures. Later than day a problem with the 15:50 London Bridge to Tattenham Corner caused 66 PPM failures. A range of other technical problems caused 26 PPM failures, problems with platforming trains at London Bridge caused 21 PPM failures and problem with Train Crew resulted in a further 46 PPM failures.

So, across the whole of the South East, NR were responsible for 89% of PPM faiures leaving GTR to pick up the remaining 11%. In terms of time lost, the NR delays accounted for 379 Hrs 16 mins of delay and GTR were responsible for 34 hrs and 26 mins. This pushes the NR total up to 92% of all delay.

However, it was a bad day for GTR themselves on the Midland Main Line were technical problems with their trains caused 37 PPM failures and resulted in 5 hrs and 53 mins of delay. A major points failure at Cambridge resulted in NR Anglia causing a further 153 PPM failures resulting in 37 Hours and 34 mins of delay.

WHICH TIMETABLE?

The timetable, which NR has agreed with the TOC at 19:00 hours the previous day is the timetable which is measured. Thus, if a planned disruptive event, typically at weekends when NR needs to undertake renewals and maintenance work, has been included in the public timetable for that day, then it is against this amended timetable, that all performance decisions are made.

  

COMING UP

A more detailed explanation as to how Schedule 8 operates. For which the two important factors are lateness and delays