Return to Archive -by date - by topic.

INFORMED SOURCES March 2004

Another measure of reliability

Miles per casualty tell you how often a train fails – but the passenger is more concerned about what happens next.

Since everyone greatly enjoyed the January review of train reliability, well nearly everyone – see this month's letters column – here is another type of reliability analysis based on delays resulting from failures. It is not new.

Back in the days when the class 91s were being teethed, InterCity engineers used the concept of Impact Minutes to prioritise remedial work. What this did was monitor how many minutes' delay each failure cost. Development could then be concentrated on those faults which upset the passenger most.

Thus a frequent sporadic minor fault which tripped a circuit breaker which could be re-set immediately by the driver was less urgent than something which went ftang twice a week and brought the East Coast Main line to a halt for a couple of hours while a fitter was called in.

Today, of course, we are much more sophisticated. We have delay minutes for which, under Schedule 8 of the Access contract, someone has to pay compensation. In other words impact minutes with pound signs attached.

So if engineers monitor delay minutes as well as failure rates, they not only have a better handle on what needs to be done, they can make a financial case for doing it. This assumes, of course, that you can find a pattern in delay minutes.

Here are two graphs showing delay minutes for the last year. Putting lots of operators on a single graph creates an interesting abstract impressionist aesthetic but not much illumination. Comparing similar operators can be informative – or pointless.

Chart 1 shows the fleet delay minutes for the three big south of the Thames TOCs and it looks to me as though it comes into the ‘why bother' category. South Central are the steadiest performers, but you would expect this since they have the simplest Network and the minimum scope for causing mayhem in the peak.

What you will notice is that over the period covered, the changes in delays for all three TOCs are pretty well synchronous, down in the spring and up in the autumn. This suggests that the Mk 1 fleets might well be suffering the aches and pains which come with age and are especially annoying in the cold and wet.

InterCity

Now look at Chart 2, which comes into the informative category. Best performer is GNER which rolls along at around 3000 delay minutes per period – that's 50 hours. But don't forget, delay minutes include time lost by all trains affected. So, well done GNER.

Now, you could argue that Christopher Garnett and his team have it easy on a relatively simple railway with lots of four track. But note that Anglia , running with a smaller fleet in among some busy commuter trains generates more delay minutes than the Blue Team with 400 miles runs.

Well, Mk 2 stock and Class 86s are going to be less reliable than IC125s and 225s. But more to the point, when something does go wrong between Stratford and Liverpool Street , say, you can delay an awful lot of commuter trains in the peaks, piling on delay minutes.

What about the others? Well First Great Western is fairly solid, except for that spike in Period 5. And here we see a message emerging. Everybody gets worse in Period 5. But whereas GNER coughs, FGW get the flu and MML has double pneumonia.

Overall, the lesson of both graphs is that there is little rhyme or reason for changes in fleet delay minutes and little sign of continuous improvement. With once exception.

Note that while Virgin Cross Country is pretty volatile, the overall trend is clearly downward. This must reflect Bombardier's excellent work on improving Voyager reliability, plus the ability to recover lost time with nigh on a quarter of a Deltic under the floor, plus the new Regulation agreements.

So that is probably a better performance than GNER in relative terms.

Of course, total delay minutes doesn't is a relative, rather than and absolute comparator. To determine how well TOCs are doing in absolute terms we need to normalize the statistics and the Table shows Delay minutes per 1000 fleet miles.

I leave you to draw your own conclusions while looking forward to receiving readers' explanations of that Period 5 blip. And more NFRIP shocks next month.

Countdown to Pendolino

213 days to go

Return to Archive -by date - by topic.