Fuel leaks, like many systems failures, are rare events, but pilots have to cope with them. And they need help from manufacturers and trainers

Now that the full story of the Air Transat "glider" incident in 2001 has been released, the Portuguese investigators have given the aviation world plenty to think about. It is not the main points of the story that make the difference here - those were known soon after the Airbus A330's impressive gravity-powered arrival at Lajes air force base, Azores in August 2001. It is the minutiae of how it all worked that provides food for thought.

The aircraft actually worked exactly as it was designed to do from the time the maintenance-caused leak began - insidiously, but quickly - to deprive the A330 of fuel. But the question the inquiry quite reasonably asked was: why was this process insidious? Why did the aircraft's sophisticated and - by current standards - intuitive warning system not spell out to the crew that there was a fuel leak? The system provided plenty of clues that all was not well, but no warnings as such. The process of tracking the clues back to the cause remains, in many state-of-the-art cockpits, a matter of detective work of the type that flight engineers used to be best qualified to practise. But since flight engineers have been replaced by computers, the computers had better start doing what the flight engineers did best - talking to the pilots in their own language to explain what seems to be wrong.

It is clear that it would be difficult, or impossible, to design a total aircraft fuel system that would detect fuel escaping from any point within it, so the answer has to lie in an extension to the existing system that detects fuel usage/quantity anomalies. Airbus has already reacted to this with a modification that means the electronic centralised aircraft monitor now provides an alert to the pilots that a discrepancy is developing between the fuel on board and the rate of fuel use by the engines. The most probable cause for the rapid development of such a discrepancy is a fuel leak. The problem then - and here the crew would still have to revert to detective work - is to find out where the leak is, so they can isolate it. That is when the crew should, according to standard procedure, go to the quick-reference checklist (QRC). This is what the Air Transat crew failed to do when the ECAM warned them there was a fuel imbalance, with far less fuel in the right wing tanks than the left. They carried out the fuel crossfeed procedure without reference to the QRC, and so missed the "caution note" that if a fuel leak is suspected, they should go straight to the page for fuel leaks. In the fuel leaks procedure, if the location of the leak is not known, the crossfeed valve should not be opened.

The trouble is, the crew had their attention diverted for nearly an hour before the fuel imbalance advisory was presented because they were preoccupied with unusual, but not critical, engine oil readings. So the ECAM raw fuel readings and the rapidly dropping fuel-at-destination values available via the multi-purpose control display unit had received less attention than they normally would.

Crew preoccupation with a non-critical issue while a serious threat develops unnoticed has brought down many aircraft. Accident reports often turn to pilot training for the solution in their recommendations, and the report on the Air Transat glider is no exception. The Portuguese investigators recommend that flightcrew operating manuals and checklists should "contain adequate information related to fuel-leak situations". That is fair enough, but they add that airlines should "review flightcrew training programmes to ensure they adequately prepare crews to diagnose and take appropriate actions to mitigate the consequences of fuel-leak events".

Fuel leaks are rare. Pilot recurrent training - logically - tends to focus on the more likely events, especially those that are critical and require instant action. Incorporating unlikely occurrences in training could divert attention from the higher risks. The solution is well-managed line-oriented flightcrew training (LOFT). This is not a new idea, but neither is it a part of all airlines' training programmes. In LOFT, crews "fly" long, "normal" sectors in a simulator or advanced flight-training device, and are trained in systems failures, including those that are rarer and more difficult to solve. "Trained" is the operative word. If crews see LOFT as a test, it has failed. If a systems failure is rare and solving it complex, crews will need retraining in it, and as systems become more reliable and real failures rarer, LOFT will become the main method for pilots to reinforce their theoretical systems learning and practise dealing with the problems that may never happen, but with which they have to be familiar.

Source: Flight International