The crash of the National Air Cargo 747-400F at Bagram, Afghanistan, was a real out-of-control accident in the sense that the aircraft was clearly outside its flight envelope well before it hit the ground, and pilot attempts to recover were likely to have been futile. Although no formal information has been made available, the suspicion is that the freight shifted aft during the take-off run or after rotate, because video footage of the early climb clearly shows that pitch-control was lost.
Among the other jet crashes, there have been some more traditional accident scenarios, particularly the SCAT Bombardier CRJ200 crash in January 2013 on an instrument landing system (ILS) approach in fog at Almaty, Kazakhstan. This appears to have been a classic poor visibility accident in which the crew either failed to monitor glidepath and speed or deliberately ignored the glidepath guidance in hope of making visual contact with the runway lights.
The same appears to be true of the Lion Air 737-800 approach at Denpasar, except it was on a non-precision approach using VOR/DME guidance. The crew clearly elected to continue the approach through their decision height without having sight of the runway, and eventually paid the price for it with a broken aeroplane. Similarly, the UPS A300F crew at Birmingham, Alabama, was carrying out a non-precision localiser/DME approach at night when the aircraft hit rising ground on short final approach. It did not help that the runway had only edge lighting and no approach lights.
The Asiana accident – the first-ever fatal accident for a 777 – was in a category that is becoming more common: the pilots’ failure to monitor or control the aircraft’s airspeed and rate of descent on approach. It was disclosed at the routine public hearing on the accident that the National Transportation Safety Board’s interview with the pilots revealed another factor – the pilot’s admitted unhappiness with carrying out an approach with the ILS glideslope inactive, even in daylight and good visibility.
It became clear in the interview that the pilot flying, a new captain under instruction by the examiner in the right-hand seat, did not completely understand the autopilot and autothrottle modes, but was also reluctant to trip them out.
A manual selection of the throttle levers to idle because the aircraft was too high on the early approach did not disconnect the autothrottle, but put it in “hold” mode, so the engines remained at idle and the aircraft gradually dipped below the approach path, simultaneously slowing the 777 to an airspeed far less than the commanded 136kt (252km/h).
The pilots did not know about the “hold” mode and assumed that the autothrottle would work as normal. This is not the first time the autothrottle mode has confused pilots: it confused EASA test pilots when they were testing the 777 and 787, but they were monitoring the aircraft’s behaviour so no ill came of it. The Asiana pilots failed to recognise that the power levers did not move again and did not monitor the aircraft’s speed or descent profile until it was too late.
It has taken around 1,300 deaths from LOC-I accidents in the last 20 years for the industry to recognise, formally, that highly automated cockpits are de-skilling pilots in subtle but dangerous ways, and that they are not taught properly how to make best use of the sophisticated flight management systems. Finally, however, the US Federal Aviation Administration has -published its long-awaited study by the Commercial Aviation Safety Team (CAST) Flight Deck Automation Working Group (dubbed FltDAWG) entitled The operational use of flight path management systems.
This FAA-led work is a seminal report and it is the first attempt by an aviation authority anywhere in the world to define the problem rather than treat the symptoms.
Yet the FltDAWG has not proposed solutions, it has just defined what needs to be solved. The FAA has handed the task of deciding what should be done over to the Air Carrier Training Steering Group. When solutions have been agreed, airlines would then be under pressure to adopt the recommendations voluntarily. The agency cannot impose regulatory solutions because, with safety in the USA as good as it is, new regulation would fail the cost-benefit analysis test even if it would, one day, save lives.