Learning from accidents: Hatfield 15 years on

Quntinshill, Lewisham, Ladbroke Grove – an accident’s location alone can stir the memory and chill the blood. Yet as time passes, such names grow symbolic too of recommendations made and actions taken. On 17 October 2000, another was added to the list: Hatfield, where a derailment at speed killed four and injured many more. Why did it happen, what came next – and what kind of railway came out of it?

Shattered rail, shattered lives 

The day began ordinarily enough. The crew had worked ECS to Skipton, back to King’s Cross and then took a short break. Their next train was 1D23 – the 12:10 King’s Cross – Leeds, which left the capital ‘right time’ and snaked over the points at the station throat, before entering the dark cavern of Gasworks Tunnel. The speed increased as it passed Alexandra Palace, Oakleigh Park and Hadley Wood. By 12.23, it was rounding the Welham curve at about 115mph. After passing under the Oxlease Avenue overbridge, the driving crew noticed that the brake pipe pressure had fallen to zero, initiating a full brake application. The reason soon became apparent – derailment. Though the locomotive and first two carriages stayed on the track, the remaining eight had come off. Some had turned on their sides and one – the buffet car – struck an OLE stanchion, bringing it down. It was here that the four fatalities occurred. More than 70 people were also injured, four of them seriously (including two members of staff).

Investigators found that the high rail of the canted curve at this point had broken into more than 300 pieces. Beyond this, the rail was intact – though displaced – for about 44 metres, but there was a further fragmented length of 54 metres beyond that. Hatfield was the first major train accident involving a broken rail since Hither Green in 1967, though the underlying causes were very different – not that many people really appreciated how different at the time. The problem was that the shattering had been caused by gauge corner cracking, a form of rolling contact fatigue that could be difficult to detect. In the aftermath, concern within Railtrack over the rail condition at other sites led to the blanket imposition of severe speed restrictions until all the facts could be ascertained.

The irony of change

Sometimes when we’re striving to improve, bitter irony will come in to play. The SPAD and multi-train collision at Harrow & Wealdstone in October 1952, for example, occurred as BR was about to start trials of the Automatic Warning System, a system that would have saved 112 lives there alone. It was a similar case with Hatfield, for the HSE and the (then) Rail Regulator had already commissioned the Transport Technology Centre to investigate broken rails and Railtrack’s management of them. The report was published that November, one month after the accident.  It (and of course Hatfield itself) led to the establishment of a taskforce to research metallurgy, wheel–rail interaction, brake design, suspension design, and ultrasonic rail flaw detection – the fruits of which were borne in subsequent years.

There were also three inquiries – by RSSB’s predecessor Railway Safety, the HSE and the ORR – which, together with the Transport Technology Centre report, found (inter alia) serious flaws in the training given to patrolmen, the way contractors were monitored and the way asset records were kept. But there were other factors that may have contributed, such as the post-privatisation increase in passenger and freight traffic, which put great strains on a ‘stretched, ageing and fragile’ infrastructure. Furthermore, the RSSB/Railway Safety report noted that those reputation-damaging, passenger number-damaging speed restrictions came in not only because of Railtrack’s lack of knowledge of its own assets, but because the gauge corner cracking at the root took it – and everyone else – by surprise. A fair point…except it wasn’t quite true – people did know about it and were talking about it.

In fact, the phenomenon was first noticed in the early days of diesel traction, when heavily-laden wheelsets revolving at high speeds were found to ‘flake’ the railheads. Instances were few at first, but started to become more prevalent on the West Coast Main Line in the 1980’s. This led to a series of reports in the 90’s that showed ‘GCC’ to be playing an increasing role in defects and broken rails. There was increasing awareness too that high curve rail positions (such as Welham) were more vulnerable to damage than others. When you add in an increase in the manufacturing hardness of rails and changes to the wheel profile that meant the same section of rail was being used all the time, then perhaps Railtrack should have seen what was coming, should have increased rail grinding, should have upped renewals and replacements.

This ‘retrospectoscope’ approach is all very well, but I’m reminded of the situation in the 1960’s when the use of new rigid-framed wagons at the speeds possible with new diesel traction led to hunting, which led to derailments which led – one July day in 1967 – to a wagon foul of the adjacent line at Thirsk. It was struck by a passenger train with fatal results. Hunting, that strange, alarming oscillation – perhaps like gauge corner cracking – came out of a concatenation of events that was hard to predict. For BR, accidents like Thirsk spurred research into the wheel-rail interface that not only made us safer, but also gave us the Mark III coach, capable of 125-mph travel, and still in front-line service some 40 years after introduction.

So what was done after Hatfield?

In 2001, the government refused to go on helping Railtrack with its spiralling repair bill and, on the order of the (then) Transport Secretary (Stephen Byers), the company went into administration. The High Court allowed it to be taken out again in 2002 so that Network Rail could take over. In September 2005, both Balfour Beatty and Network Rail (as Railtrack’s successor) were found guilty of breaching health and safety legislation. The former was fined ￡3.5 million, the latter ￡10 million. Both were ordered to pay ￡300,000 in costs. This was the same year, however, that Network Rail took maintenance contracts in-house, which led to a more strategic approach to the way the infrastructure is managed. There were also some changes that were coming anyway – including changes to legislation.

ROGS and other animals  

At the end of 2000, the Safety Case Regulations came into force, replacing the earlier 1994 version. Under their aegis, any infrastructure controller, train operator or station operator was required to prepare a safety case, a key element of which was a safety management system. Railtrack had to have its safety case accepted by the HSE as a condition of using the railway; train operators were required to have theirs accepted by Railtrack. Then – in 2006 – came ROGS. Obviously ROGS – the Railways and Other Guided Transport Systems (Safety) Regulations – is all about the European Rail Safety Directive. However, one important change it brought about was the safety regulator’s ‘permissioning’ of railway operations, meaning that infrastructure managers and train operators had to apply to the ORR for a safety certificate (TO’s), safety authorisation (IM’s) or, where appropriate, both.

By this time, CIRAS and the RAIB had been established – the latter closing a recommendation from the Cullen Inquiry and setting up the mechanism we know today for undertaking independent investigations and improving our learning capabilities. And of course RSSB came into being. Not only did this lead to improvements in safety performance reporting and risk analysis, it also led to the establishment of System Interface Committees – getting people round the table with the right expertise to focus on specific links between different aspects of railway assets. One ‘SIC’ deals with the vehicle/track interface, which includes rolling contact fatigue and adhesion; issues that were particularly relevant to the Hatfield derailment. This process – and the research it spawned – meant that things got better. Indeed, there hasn’t been a passenger fatality in a train accident of any kind since Grayrigg in 2007. More specifically, the figures show there to have been a notable reduction in the number of broken rails over the first half of the last decade, due to the substantial improvements that Network Rail made in track management. This is built on a much better evidence base than was available at the time of Hatfield, and is thanks to world-leading research by industry through RSSB, and by academia through groups like the Institute of Railway Research at Huddersfield University.

Monitoring the situation overseas

But if you’ve read any of my other articles, you’ll know of my belief in the need to avoid complacency, and that one of the ways of doing so is by monitoring the situation overseas. This became more pertinent in July 2013, when – in the space of 22 days – we had the runaway and explosion in Quebec (6), a derailment in Paris (12), a high-speed derailment in Spain (25) and finally a dispatch error in Switzerland (29), which led to a collision that killed the driver.  The Paris incident occurred at 90 mph near a station and resulted from defective track on or near a set of points. Specifically, a broken fishplate had got lodged in the ‘V’ of a double slip. There was a parallel here with a derailment at Southall East Junction on 24 November 2002. RSSB sent SNCF the report on that incident, as it’s vital that we share learning in an industry like ours. But we learned too. In Britain, fishplates are reportedly used at around 20 sites on 100mph lines to secure track to switches and crossovers and allow the rail to expand and contract under prevalent temperature conditions. Fishplate breaks increased by 30 per cent between 2008 and 2011, and though the trend reversed in 2012-13, the Paris accident spurred Network Rail to review bolted crossing risk and brief its inspection teams to highlight the issues that came out of it.

With a railway as safe as ours has become in the 15 years since Hatfield, we’ve started to look, then, not only at Close Calls – incidents lower down in Bird’s triangle – but also accidents from overseas, which we use to keep a check on any gaps in our processes, rules and methods. And we’ll go on doing that until the future means we no longer need to.

The future though, as someone once said, is slower than you think…