Return to Archive -by date - by topic - 2001 Archive.
Human fallibility means it's fingers crossed until TPWS is in and working
Here's a graph to study, courtesy Her Majesty's Railway Inspectorate. It shows the 12month rolling totals of Signals Passed At Danger effectively since Railtrack became a separate entity. It provides, I hope, a useful corrective to the media hysteria we now enjoy when the monthly SPAD figures are published.
What really concerns us are the Category 3-8 severity SPADs. Just to remind you, the severities are as follows:
3) Overrun greater than overlap plus all overruns greater than 200 yards and no damage, injuries or deaths.
4) Track damage only with no casualties.
5) Derailment with no collision and no casualties.
6) Collision (with or without derailment) and no casualties.
7) Injuries to staff or passengers with no fatalities.
8) Fatalities to staff or passengers
The table shows the proportion of total SPADs within each of these categories. And, like me, no doubt the first thing you noticed was that row of 0%' for Category 8 SPADs ie those which kill people.
So here's another corrective. As the HMRI explains, the 0% for some severity category 7, and all category 8 events, is caused by rounding down when converting to percentages. There has in fact been one Category 7 incident this year and one Category 8 incident in each of the years 94/95, 96/97, 97/98, and 99/00. Which means that 1 in 200 SPADs have been fatal in four out of the last five years.
Now back to the graph. Where you will note that, despite the massive investment in SPAD reduction and mitigation measures, the number of Severity 3-8s has remained the same, or sensibly constant, as we used to say in physics experiments, since 1995.
Of course, rolling annual totals are a pretty crude measure and both the HMRI and Railway safety are applying some fancy analytical techniques to see whether things are getting better or worse. Which is not always apparent from the raw data, because there are so few SPADs to analyse.
For example, as a rule of thumb, counts of data such as SPADS will vary in the range plus/minus twice the square root of their mean. In other words if the underlying level is 16 serious SPADs per month, the variation observed in the actual figures is likely to be plus or minus twice the square root of 16, equals eight. So, in 19 months out of 20, simply by chance the number of SPADs could range from 8 to 24. In one month out of 20 chance alone can be expected to produce an even more extreme variation.
The official viewThe latest twelve month rolling total for serious SPADs is 198, 16 higher than the corresponding figure up to July 2000 (182), which also happened to be the best ever twelve month record for serious SPADs. The difference is not statistically significant, and could be entirely ascribable to random variation. The total for the last three months (49) is very close to that for the corresponding figure last year (50). Since late 1999 the twelve month rolling total has been in the range 182 to 205, but over the last 12 months has shown a fairly steady movement from the bottom towards the top of this range. The small numbers involved mean that one cannot safely conclude that there is a worsening trend in underlying performance (particularly since much of this period saw disrupted operations due to the Hatfield derailment). What is clear is that there has been no evidence for improvement since the middle of last year. HMRI July 2001 Monthly SPAD Report |
Now what I suspected no one was brave enough to say is that the present level of SPADs is probably close to the irreducible minimum within the hard wired propensity for fallibility in the Mk 1 human brain. But at a recent HMRI briefing Chief Inspecting Officer Vic Coleman agreed that we probably are nearing the limits of what we can achieve by means of process improvements [and this] provides the justification for pressing ahead with more physical train protection systems.'
Or to put it another way, the present unprecedented focus on SPADs will not reduce the numbers significantly, nor should we expect them to. Which is not the same as saying that it is wasted effort, because it is the many and varied SPADRAM initiatives which keep the railway as safe as it is.
So for the next 27 months, until the Train Protection & Warning System is fully operational, safety rests, as ever, in the hands of drivers and signallers. And for my money, the best way to support their amazing record until then is through human factors and management not threats of one SPAD and you're sacked'.
Finally, let us not forget that TPWS will not stop SPADs, as the media believes. But it will stop trains in the overlap, that often overlooked signal engineers' recognition of human frailty.