Prof. John McDermid of the University of York asked me if I had documentation for the suggestion in my post on the Buizingen collision that the number of fatalities to trackside workers expected in installing ATP universally on rail tracks might be larger than the number of passenger lives expected to be saved by ATP. I was thinking specifically of British railways. Prof. Dr. Jens Braband of Siemens Rail Automation, who has also taught a course on risk at the Technical University of Braunschweig for many years (if you read German you might like to check out his textbook on it) and who currently does a lot of work for and with the European Rail Authority, took a look at ERA statistics and didn’t find support for the assertion.
This morning I surveyed the annual Safety Performance Reports from the U.K. Rail Safety and Standards Board, from 2000 to 2009. All of them include details of individual fatalities. I only considered fatalities, and not serious injuries, so this is not an assessment of total personal risk. Details follow. There are three columns: Year, Trackside Employee Fatalities, specifically restricted to those struck by moving trains, and Passenger Fatalities, specifically restricted to those occurring in collisions between two moving trains. Note that these figures must be extracted specifically from the descriptions of individual accidents.
Year: Trackside Employee Fatalities: Passenger Fatalities
2009: 1: 0
2008: 3: 0
2007: 2: 0
2006: 2: 0
2005: 5: 0
2004: 8: 0
2003: 0: 0
2002: 1: 0
2001: 4: 0
2000: 2: 0
1999: ?: 31 (Ladbroke Grove)
Note that there are significant passenger-fatal train accidents in 2000, when 4 people were killed in the Hatfield derailment and 10 people, including the driver of the automobile, two train drivers, two other train staff, and five passengers, at Great Heck. In 2002, 6 passengers were killed in the derailment at Potters Bar.
In summary, it is not clear what one may infer from these figures. 31 passengers died at Ladbroke Grove in 1999 as a consequence of a train collision following from a SPAD. In the ten years since then, no passenger has died as a consequence of a train collision following a SPAD. 28 trackside workers have died.
Can we go further back, maybe, to obtain some more figures. Not really – the environment has changed significantly. Because of systematic attention paid to them over the last decade since Ladbroke Grove, the number of SPADs in the last few years has been under 15% of the 2001 baseline level. Pre-1999 statistics come from a different safety environment.
Conclusions? This is a typical case in which the risk of a very rare event is compared with the risk of rare but regularly-occuring events. One big accident, such as Ladbroke Grove, changes the figures completely. It seems that ATP installation risk versus passenger lives saved is a difficult balance to assess.