Asiana: one of several similar events

The Asiana Airlines Boeing 777 accident at San Francisco is one of several events that emphasise the need to review, in precise detail, what the effects of high levels of aircraft automation are, not only on the way pilots carry out their task, but also how they perceive their role in the control of the aircraft. 

Evidence, although circumstantial at present, suggests this perception is changing and they are beginning to feel functionally peripheral.

Some work on specific areas has begun, but more is needed. Following the 2007 Thomsonfly 737-300 incident at Bournemouth, UK, in which the approach airspeed decayed 20kt below the reference figure before being noticed, the airline arranged for pilot eye movement tracking to be carried out. The carrier found that instrument scans were often un-methodical and incomplete. 

Since pilots are still provided with the same basic flying training syllabus that has been used for many decades, it is important to understand whether this ineffective scanning habit is widespread, whether a disciplined instrument scan was never learned, or whether this represents a gradual degradation from a previously adequate level. Finally we must examine whether excessive trust in automated systems has anything to do with that degradation.

The Thomsonfly incident and the 2009 Turkish Airlines 737-800 crash on approach to Amsterdam Schiphol share with the Asiana accident a failure by the crew to recognise airspeed degradation until the situation was dangerous. In the Turkish case, recognition came so late the aircraft stalled fully. 

According to National Transportation Safety Board investigation so far, the difference of these two from the Asiana case is that Thomsonfly entailed an autothrottle failure, Turkish involved a malfunctioning radio altimeter affecting a serviceable autothrottle, whereas – so far – no systems problem has been detected in the 777. Autopilot mode selection is therefore under examination, as well as further checks on serviceablility, to determine why the throttles remained at the idle position.

But back to scanning. On a visual approach like that of Asiana flight 214, aircraft attitude can be determined pretty accurately by reference to the real horizon, and the approach path ascertained from the runway perspective and precision approach path indicators. But the other factors critical to the performance of a safe approach can only be determined by looking at the instruments. Frequent reference to the airspeed indicator, altimeter and engine power indications is vital, because if the automatic systems deviate from their expected performance, divergence from the safe approach parameters is very rapid.

The airspeed indicator in particular needs to be scanned at least every 5s on long final, and every 2-3s on short final approach. So there is plenty for the eyes to do, both outside and in. 

But the evidence shows this is not happening. 

Why? We need to know. 

, , , , , , , , , , , , , ,

6 Responses to Asiana: one of several similar events

  1. Robert Taylor 17 July, 2013 at 1:07 pm #

    Experience on type also tells you (instinctively) where the thrust levers should be (arm length etc) on a “normal” approach (need to monitor airspeed as ever though).

    Once the flaps are out of UP during approach a hand should be on the thrust levers at all times whether or not the autothrottle is in use.

  2. Mike 17 July, 2013 at 1:07 pm #

    I am an interested layman, so please ignore this if appropriate.

    My cousins wife is a senior capatain flying B747-400. His airline also flies Airbus 330. He told me something that stuck in my memory. When the autothrottle is engaged on a Boeing, the throtlle levers move driven by the computer, but on Airbus the throttle levers stay at flight idle irrespective of what the engines are doing. I note that the pilot of the Asiana a/c had come from Airbus 320. I wonder if he subliminally noted that the levers weren’t moving, and ignored it, because that was what he was used to and ‘knew’ autothrottle was engaged. Possible?

    • Ron Boyle 9 August, 2013 at 4:07 am #

      In reply to Mike, your information on the Airbus Autothrottle/Lever position is incorrect. On a normal Autothrottle coupled approach, the Levers stay in the Climb/Cruise position. The lever position defines the maximum power that the autothrottle may use during this phase of flight under normal conditions. The max power available for that throttle position is displayed on the engine instrument as a blue circle. On approach the autothrottle then modulates the power to maintain the approach speed that has been bugged.
      If flying a manual throttle approach, the throttles are used and moved just like in any other aircraft.
      You are correct in saying that the levers move when the A/T is engaged on a Boeing. The bottom line with both systems is that power/airspeed needs to be constantly monitored, especially at critical stages such as on approach. For what it’s worth I have spent considerable time using both systems, and are aware of several failures in the Boeing system (including one leading to a stalled 400 in a major airline, and several close shaves), but have not heard of such a failure in an Airbus. This doesn’t mean that it hasn’t happened in an Airbus, but I just haven’t heard of one.

  3. David Learmount 17 July, 2013 at 1:07 pm #

    Everything is possible, but the proposal has an air of pointlessness about it because even the pilot himself will never know for sure if that is true!

    But the crucial cues to whether the throttles are doing what you want them to do is not what the throttle lever angle is, it’s what power the engine is developing (see engine instruments) and what the aircraft’s airspeed is (see airspeed indicator).

    You do not judge performance in a car by looking at (or feeling) where the accelerator pedal is. You judge it by the car’s speed (speedometer and how fast things are going past the window). Speed is the product you are demanding by using the accelerator, so you adjust your accelerator according to the speedo, not the other way around.

    Boeing pilots who haven’t flown Airbuses have a habit of saying that the sidestick/throttle mode in the Airbus would be a disaster, because in a Boeing you always know where you are. Apparently this is not always true: all three incidents mentioned in which airspeed was ignored happened to be in Boeings.

    There’s a certain mystery and fascination about aviation accidents, so people also have a habit of looking for the abstruse when the truth is staring them in the face. The problem is that pilots are still not trained properly for operating in the highly automated environment, so they manage it very badly, whether in Airbuses or Boeings.

  4. Anthony 17 July, 2013 at 1:07 pm #

    David, you write of possible “excessive trust in automated systems”; is the implication that (some) systems should not be trusted or was the intention to suggest of excessive ‘reliance on’ automated systems? I’m sure you’ll state the answer is that pilots need always assume that no system is infallible and therefore needs a watchful eye over it!

  5. David Learmount 17 July, 2013 at 1:07 pm #

    If the systems were infallible we wouldn’t need pilots. But they are fallible so we do need pilots – sceptical ones!

Leave a Reply