Sometimes an accident warrants examining because it is extraordinary. The Girona report needs study because it is ordinary
If Professor James Reason was looking for the ideal real-life example to demonstrate his famous "Swiss cheeses" model of what enables accidents to happen in a basically safe system, the just-published report on the September 1999 Boeing 757 accident at Girona in Spain has it all.
Now the investigators reveal a web of interacting forces, circumstances and influences, even including what secular insurers still call an act of God. Gamblers would call it a wild card.
It is worth using an event like Girona to examine whether, just sometimes, things like this are bound to happen and we have to accept it, or whether something could - or should - have been done that might have blocked the chain of events. Although this 757, with 245 people on board, careered off the runway at high speed with almost all its controls disabled or malfunctioning, only one person died as a result. Even the low toll in human life could be considered a matter of luck.
This aircraft set off from the UK for Girona, knowing its destination and all its alternates were affected by a band of stormy, frontal weather - but it was the type that might delay a landing until a storm cell passed rather than prevent it. The captain loaded an additional 15min of fuel above standard company diversionary minima to allow for this. On calling Girona it was clear that a storm cell was close to the airfield, but it was dark and its precise location was not communicated. The wind - not strong - had shifted from southerly to northerly, so the captain decided on a runway 02 non-precision approach rather than a precision approach on to 20 with a tailwind. The trouble with 20 is that it has a strong downslope - just above the International Civil Aviation Organisation recommended maximum. And the runway - adequate but not generously long at 2,400m (7,900ft) - was going to be wet, so the captain opted for the risk of a non-precision approach rather than the alternative risk of a tailwind landing on runway 20. He also took over as the pilot flying at that point. But the VOR/DME approach did not go well, the tower/approach controller advised during it that the storm cell was now over the airfield, and the captain carried out a go-around. Meanwhile, the wind was shifting again to southerly, making an ILS to runway 20 plausible for the second attempt. Then the aircraft flight management system advised the crew they were approaching company fuel minima, so the captain was under pressure to make a decision whether to divert or not. It was dark, turbulent and the rain over the airfield was "torrential", but had they diverted to any of the alternates it might have been the same. As they established on the ILS for runway 20, the "must land" mindset would have been a tempting one to adopt. The approach was turbulent and not stable relative to the glideslope, but the runway lights were visible before decision height and, despite a sink-rate warning from the ground proximity warning system on short final, the captain clearly thought the landing could be safe.
Then fate played the wild card. The runway lights went out for 11s just as the captain needed them to judge the final descent and flare. The report says a contributory factor in the very hard, nose-down landing that followed was "the effect of shock or mental incapacitation on the pilot flying at the failure of the runway lights, which may have inhibited him from making a decision to go-around". Most pilots have experienced an unexpected loss of visual contact with runway lights at the last moment - usually due to a patch of fog caused by a local micro-climate effect - but by the time the pilot has registered the loss of contact the lights usually reappear again.
A pilot is "the system's goalkeeper", and this one got past the 757's crew. So where else did the defences fail? Girona, with its steeply sloping runway, is scarcely perfect, but that was a known part of the risk management calculation. With hindsight, 15min of extra fuel was not enough. A recommendation that go-around manoeuvres below decision height should be a mandatory part of recurrent training seems useful because it would help pilots override a "must land" mindset. Another recommendation - that more precise real-time weather information should be immediately available to controllers - would have helped the captain with decision making. There's plenty else - the report is a gift for nitpickers.
The Girona report should be required reading for airline and airport safety committees, because there is no "silver bullet" solution for this one. It was an ordinary situation that got out of hand, and all operators can face challenges like these at any time.
Source: Flight International