Return to Archive -by date - by topic.
Captain Deltic revisits fleet reliability and finds that more-rigorous monitoring has revealed that even the best are not as good as we thought.
In January 2004, Informed Sources published a fleet-by-fleet analysis of passenger train reliability based on figures collated by the Association of Train Operating Companies under the National Fleet Reliability Improvement Programme (NFRIP – say it enfrip) A furore ensued.
Much as I would like a furore every month, they are fickle things. But this was a classic because lots of different people were upset by the revelations that not all trains are equal.
There were stroppy letters from ATOC and train manufacturers. TOC MDs who has probably thought that NFRIP was some obscure techie thing, saw their poor performance exposed and heaped opprobrium on their engineering teams.
Readers loved it, however, so here is the second annual Informed Sources Fleet Reliability Analysis. For TOC MD's I should point out that Informed Sources provides features you won't find in the NFRIP figures which your engineers either slap confidently on your desk covered in magic marker, or try to hide when you go into their office.
To maintain continuity, the latest figures are for Period 7 of the current year. The basic measure is miles per casualty (MPC), with a ‘ casualty' defined as a technical or maintenance defect on a train causing a delay of 5 or more minutes.
One reason why manufacturers were touchy is because NFRIP doesn't go in for sophistry or carefully manipulated ‘contractual failures'. For example NFRIP counts cancellations and part cancellations. And since Period 2 of 2003/04 ‘casualties' have also included all technical cases of 'No Fault Found' (NFF).
Nor should we forget the recent impact of TPWS. On its own the train mounted equipment has a failure rate around 60,000 MPC. New solid state receivers should improve this.
Technical faults on empty stock moves also count, ditto delays exacerbated by operational error and adverse weather conditions. However faults due to vandalism, by proven infrastructure defects and by other external factors are excluded.
So, a really tough regime. Note that the mileages are for units. So an IC225 is one unit, but two Class 465s running in multiple still count as two units.
In each table, fleets are ranked in descending order of MAA reliability. There are three MPCs: for Period 7 2004-05, the Moving Annual Average (MAA) MPC, either for the preceding 13 periods or since the start of a replacement franchise, and Period 7 2003-04.
Last year's ranking are also given and I have calculated the percentage change in MAA over the past year. Colour coding identifies the year on year change.
At first sight it looks as though reliability has got dramatically worse. This is, generally, not the case.
What has happened is that ATOC has audited the preparation of NFRIP data by every maintenance depot. And when ATOC's Engineering Director Rebeka Sellick audits a depot, they don't get a letter with a tick the box questionnaire; whoever is responsible for collating the data gets interviewed in person and their procedures checked.
For example, following the audit in July at Hornsey Depot, which maintains my local trains, the number of technical casualty figures from Period 6 2003/04 to Period 4 2004/05 were revised upwards. The new technical casualty figures for each of these periods are reflected in the latest MAA.
Comparing the MAA before and after the revision, the MPCs have fallen by 58% for Class 313, 59% for Class 317 and 41% for Class 365. And as you can see, the WAGN Class 365 fleet has dropped back from the best to merely good yet still ranks second in the ‘Three phase drive EMU' category.
But there is a further factor. The South Eastern Trains Class 365 fleet, which was recording half the MPC of its sisters on WAGN, has now joined them and it will be some time before the team at Hornsey get the MPC up to their standards.
So, treat the year on year comparisons with caution. It is not so much that things have got worse, as that the NFRIP data is much more accurate and consistent across operators. The failures were there in the past, but not all were being reported
This is confirmed by Table 1 which compares total train performance now and then. Note that both train and unit miles have increased. If this is factored in, the impact minutes have increased by around 5000. This could well be explained by withdrawal of the ultra-reliable Mk 1 stock.
|
Train miles (millions) |
Impact minutes |
Unit miles (millions) |
Casualties |
P7 2004-05 |
20.524 |
342,907 |
26.166 |
3719 |
P7 2003-04 |
20.135 |
331,136 |
25.538 |
3004 |
If a fleet has improved its MAA it is shown in green. Given that the NFF ruling and the depot audits have seen more failures recorded, any fleet within 10% of last year's MAA is counted as no change and colour coded yellow. Finally, the red colour coding identifies a fall in reliability of more than 10%. .
Pacers are unpopular. Last year's survey partly explained this perception. This year's figures show that a lot of work has been done by the maintenance engineers concerned.
A special mention must go to Arriva Trains Wales for the Class 143 fleet. While it is still bottom of the table, note that Period 7 MPC of 6,781. Last year's champagne challenge for an MAA of 5000 MPC from the Welsh Pacers remains open.
With no fleet under 3000 MPC this year, it looks to me as though an MAA of 5000 MPC should be the aim. After all, these are simple pieces of kit and the fragile bits should be well understood.
Red Ranking: 33%
Aspiration; 5000 MPC
After last year's review there was needle all round. Between GNER and the Virgins and between the HST operators.
This year, GNER retains No 1 spot with its mid-life re-engineered IC225 fleet showing a 4% improvement in MAA. This may seem modest, but given the changes already mentioned is quite an achievement.
Among the HSTs, First Great Western have added 100 miles to their MAA to take third spot overall, while GNER have lost 1,200, but retain fourth. Midland Main Line's MAA reflects the long term impact of the rather scruffy Rio fleet taken over from Virgin. But while the MAA is up slightly, note MML's ‘spot' MPC for Period 7.
Anglia 's loco hauled fleet has fallen from grace, but now, with everyone's favourite Class 90 in the stud, will be expected to improve. Virgin West Coast's locomotive hauled sets are on the way out, but even so the spot MPC suggests that everyone in preoccupied with the Pendolini, which are slowly improving.
Virgin Cross Country's Voyagers are classified with the modern DMUs. However their MAA's would rank the Class 220 second and the 221s third in the InterCity table.
Red Ranking 42%
HST only
Minimum MPC – 5000
Expected MPC – 7500
Aspiration* - 10,000
This is where things start to get interesting. Not least because some of the fleets (identified with a #) are maintained by the manufacturers. My prejudice in favour of manufacturers maintaining their products is under review.
Top of the list is c2c's Bombardier maintained Class 357 fleet. Note the consistent set of numbers. The MAA is slightly down, but the MPC is above the latest MAA, so the engineers have a grip on reliability.
One reason for this is Bombardier's excellent diagnostics. Not only does this help solve real problems quickly; if an NFF is down to finger trouble by the driver the engineers know it wasn't their train.
Even so, the contractual MPC is 100,000km or 62,000 miles – so the Class 357s are still only halfway there.
For reasons already explained the figures for last year's top kit, the WAGN Class 365s are in transition. As a user, I expect them to be significantly better next year.
And then, what's this? It's Almost (anag.) taking third and fourth. In the case of Gatwick Express you could argue that eight trains in a small depot is cheating, but the Class 460s are modern EMUs full of electronics running on DC lines, which is doing it the hard way.
Modern trains are a mass of microprocessors – a Pendolino has over 250. When a system stops working, rebooting the train (the equivalent of ctrl.alt.del on a computer) may get it working again, but it takes time for all those microprocessors to count their fingers and toes and report all systems go to the Train Management System (TMS).
Typically a re-boot takes five minutes, which means you have a casualty. Alternatively the driver can use the TMS to trouble shoot and get the train moving. My experience is that Gatwick Express drivers have got to grips with the TMS better than most TOCs.
Also performing well is the manufacturer maintained South Eastern Trains Class 375s. Actually maintenance is the responsibly of SET Maintenance Limited, 60% owned by Bombardier.
So why is my prejudice wavering? Look at the lower end of the table. A year after winning a champagne challenge for not (quite) being a year late with the Class 350 Desiros for SWT, Nemesis (anag.) is still looking at an MAA for these trains of around 6,000 MPC compared with the contractual figure of 50,000.
As regular readers well know I have no faith in the ability of test tracks to grow reliability, so this performance feeds another prejudice. Where a test track does help is in proving modifications, as with Pendolino at Old Dalby, but you can't simulate real journeys with real passengers at real stations going round and round in circles.
My chums at SWT reckon that coming Periods will see a rapid and substantial improvement – at least for the Class 350s. But with the Class 444s, and the longest running Desiros of all, the Class 360s, ranking below even the Class 458 you would need a heart of stone not to smile, given the all the hype from SWT and its supplier.
Last year I reckoned that a minimum MPC for these new EMUs was 10,000 with 20,000 expected. Only three classes beat the minimum which suggests that a massive improvement is long overdue.
Red ranking: 53%
Minimum MPC – 10,000
Expected MPC –20,000
Aspiration* - 50,000
Talking of hype, you have to realise that we have an extremely, probably uniquely, tough operating regime in Britain , which can lead to achievements being underestimated. Foreign railways regularly exaggerate train reliability and other factors in the national interest. Naïve old-railwaymen in Britain assume everyone is playing to the same rules and get out the sackcloth and ashes.
On the stand at Railtex I was admiring the shiny new MTU engine to be evaluated in a pair of Angel HST power cars. When I stressed the extreme operating conditions created by the binary driving of HSTs, the MTU man simply couldn't take it in. He seemed happy that the engine had passed the International Union of Railways tests and thus HST would be no problem.
We shall see. Sitting at one end of the stand was the baby MTU engine used in the Turbostars.
Now we have more hype about the Hitachi Channel Tunnel Rail Link Domestic Stock EMUs, with Transport Secretary Alistair Darling going on about the wonderful new technology. So here is Captain Deltic's Sake challenge. If the MAA at the end of the first year of operation is over 10,000, he'll be at St Pancras with a bottle.
End of digression.
With the British Rail EMUs, we tend to get much more coherent data. It looks as though the chart topping Silverlink Class 321 fleet was affected by Depot Audit Syndrome, but London lines' ace engineering team is heading onwards and upwards again.
A special mention for SWT with the Class 455, with camshaft control and commutators on the, admittedly iconic EE507, traction motors showing most of the newfangled three phase drive trains what reliability really means. This is a coherent set of figures, so why is Southern getting half the MPC from its Class 455/8 fleet?
Well, for one thing, SWT runs eight car trains all day, which has a financial cost while Southern reforms trains en route, and uncoupling and coupling has a reliability cost.
Sadly, the Captain's favourite high speed cruiser, the camshaft controlled son of 4-REP in a Mk 3 bodyshell with power doors, has, according to Informed Sources, suffered from a badly specified and executed C4 overhaul programme that could take up to a year to put right. Even so, the Class 442 still makes fourth place.
Red Rating: 64%
Minimum MPC – 10,000
Expected MPC –20,000
Aspiration MPC - 40,000
As Rebeka Sellick pointed out in a follow up letter to last year's review, different service patterns and duty cycles affect on MPC. We see this in Hull Trains chart topping performance. Hull Trains is about as undemanding as its gets for a Class 170, purring happily along the East Coast Main Line with few station stops to annoy the doors.
Below the class leader, the next three places are filled with Networker Turbos. Note the significant drops in the year-on-year MAA compared with Chiltern's Class 168/1 which has climbed from thirteenth to fifth on improved performance.
Voyager performance is a disappointment and another blow to my manufacturer/maintainer prejudice. A year on and the MAA is flat, when they were expected to be over 10,000 MPC by now.
Further doubt grows with the Alstom Coradias. The Class 180 has given my engineering chums at FGW a few headaches, but is obviously getting better. In contrast, its little sister, the Alstom-maintained Class 175, is not only bottom of the new DMU chart but has the lowest MAA of any established fleet and is about a 1000 miles worse than the worst Pacer. Go figure, as they say.
Red Rating 65%
Minimum MPC – 7,000
Expected MPC – 12,500
Aspiration* - 20,000
In this category, there are two lessons. As remarked last year Metro-Cammell's Class 156 continues to demonstrate the long term benefits of quality. The second is the number of Class 158 fleets apparently showing their age.
Last year I suggested that anything under 5000 MPC for these nicely run in, well understood and largely electronics-free DMUs was unacceptable. NFRIP's recent work confirms this view.
One NRFIP initiatives is the Class 158 Users Group, which brings together Depots Engineers from the various operators. The Group has analysed the reliability of systems and components for each Class 158 fleet.
It's a sort of fault tree analysis applied to reliability. Each fleet's MPC is broken down to individual parts of the train - doors, engine, transmission and so on. The sum of these ‘unreliabilities' gives the train's MPC.
And the results of this study highlights another downside of privatisation – the fragmentation of engineering. The best Class 158 has an MPC of just under 6,000. But if you created a Class 158 with the most reliable components across all the fleets, this ACME Class 158 would have an MPC of 50,000.
More on this another time.
Red Rating: 51%
Minimum MPC – 6,000
Expected MPC – 10,000
Aspiration* - 15,000
Last year, SWT's Class 421/5 fleet took overall top spot with a six figure MAA. This year SWT retains the crown with a remarkably consistent performance by the fast dwindling Class 421/8s.
As new fleets come in and the Mk 1s are withdrawn, a fall-off in reliability is to be expected. Even so only Hull Trains Class 170 stops the SWT Class 421s filling the podium when it comes to Victor Ludorum. And I am indebted to Mr Miles for pointing out that the eight remaining Class 421/8 units ran faultlessly in Period 7 to give an MPC of infinity.
These paragons of functional design yet again demonstrate the challenge facing the manufacturers of the Mk 1 replacement stock. SET still have to double or treble Class 375 reliability to match the Class 423s. Southern have to double Class 375 performance to equal some of the least reliable Mk 1 units. As for SWT, if we are generous and set the Class 450s' Period 7 figure of 10,000 MPC against the worst Mk 1 MAA, a doubling is necessary. I'll let readers do the sums for Class 444 vs 442.
Red rating: 78%
Top TOC is again SWT with its Class 159s and the Mk 1 fleet. However Top Franchisee is National Express with Silverlink and c2c heading two EMU categories and Wessex the Pacers.
On the manufacturing front, there is no change. Derby Litchurch Lane is top mechanical parts manufacturer, with Preston the most successful electric traction equipment factory.
Similarly, Porterbrook is once again the top ROSCO.
Finally, what are the prospects for train reliability? I expect to see a lot more yellow and green in the January 2006 review charts. Mandatory fitting of On Train Monitoring and Recording (OTMR) will bring modern diagnostics to existing stock.
But the maintenance depots are not immune to the cost cutting regime which currently dominates the railway. And Mrs Sellick is clear that engineers can either make trains more reliable or they can cut the cost of maintenance. ‘If we have enough trains and enough people and enough time in the depot reliability will get better'. Whether that happens is a political decision.