AF447: After two long years, six short minutes

France’s Bureau d’Enquetes et d’Analyses has released details of the final few minutes of flight AF447, following the two-year recovery effort to retrieve the Airbus A330′s flight-data and cockpit-voice recorders. Presented here are the main sections of the BEA’s statement (in bold) and an explanation of the significance of each point:



The controller asked the crew to maintain FL350 and to give their estimated time at TASIL.

Flight AF447 is told to keep cruising at 35,000ft and inform Brazilian air traffic control when it expects to reach the waypoint ‘TASIL’ – the entry point into Senegalese airspace, where responsibility for watching over the flight will transfer to the control centre in Dakar.


The Captain woke the second co-pilot and said “[...] he’s going to take my place”.

On a long flight it is normal practice to carry an extra pilot to allow one of the crew – the captain in this case – to rest. Of the two pilots left in charge, one had 4,479h on A330/340s, and was the most experienced of the three on the A330 in terms of flight time, although all of the captain’s time on the type had been in command.



The Captain attended the briefing between the two co-pilots, during which the PF said, in particular “the little bit of turbulence that you just saw [...] we should find the same ahead [...] we’re in the cloud layer unfortunately we can’t climb much for the moment because the temperature is falling more slowly than forecast” and that “the logon with Dakar failed”.

The pilot flying the A330 highlights the relatively high air temperature outside the aircraft. Higher temperatures mean the air is less dense and reduces the ability of the wing to generate lift, meaning the aircraft would be unable to sustain flight at a higher altitude.

The pilot also mentions that the aircraft has been unable to connect with Dakar air traffic control ahead of crossing the airspace boundary between Brazil and Senegal. This absence of contact, close to the transfer point, may be significant in the context of the subsequent failure to realise that AF447 had gone missing.



The airplane began a slight turn to the left, the change in relation to the initial route being about 12 degrees. The level of turbulence increased slightly and the crew decided to reduce the speed to about Mach 0.8.

Probably to avoid an area of heavier turbulence, about which they were aware, the crew deviated from the flightpath. The A330 nevertheless encountered turbulence and the crew chose to slow the aircraft from Mach 0.82 to Mach 0.8. This is normal practice when penetrating storm clouds, in order to prevent the aircraft’s speed from slipping outside of design limits in rough air.



The autopilot then auto-thrust disengaged and the PF said “I have the controls”.

Investigators have not confirmed why the autopilot and autothrust disengaged, although this is a normal response of the A330 if its computers detect inconsistency in the airspeed data received from the pitot tubes on the nose.

The airplane began to roll to the right and the PF made a left nose-up input. The stall warning sounded twice in a row.

While the pilot appears to have rolled the aircraft to the left, to counter a roll to the right, he also pulled the aircraft’s nose upwards – for reasons yet to be explained. This would have put the A330 into a climb, but the aircraft already had little flexibility to climb higher, because of the temperature, and the slower speed would have further reduced the lift available from the wings. The stall warning was the first indication that the A330 was struggling to maintain lift.

The recorded parameters show a sharp fall from about 275 kt to 60 kt in the speed displayed on the left primary flight display (PFD), then a few moments later in the speed displayed on the integrated standby instrument system (ISIS).

On the captain’s side of the aircraft – his seat occupied by a relief pilot – the airspeed on the cockpit display suddenly appeared to fall. This also occurred to the backup, or standby, display. These readings did not necessarily mean the aircraft had actually slowed to 60kt; the displays are fed information from the pitot tubes on the nose, and a blockage in the tubes – from ice, for example – could have generated useless data in the airspeed computers.



The PNF said “so, we’ve lost the speeds” then “alternate law [...]“.

The non-flying pilot realises that the airspeed information is unreliable and that the aircraft has switched, as designed, to following less rigid flight control laws. Critically these laws no longer automatically protect the aircraft from reaching high angles of attack, at which it risks stalling.

The airplane’s pitch attitude increased progressively beyond 10 degrees and the plane started to climb. The PF made nose-down control inputs and alternately left and right roll inputs. The vertical speed, which had reached 7,000 ft/min, dropped to 700 ft/min and the roll varied between 12 degrees right and 10 degrees left. The speed displayed on the left side increased sharply to 215 kt (Mach 0.68). The airplane was then at an altitude of about 37,500 ft and the recorded angle of attack was around 4 degrees.

Positioned increasingly nose-up, the A330 climbs rapidly by 2,500ft – the investigators have not clarified whether the climb was intentional – but the pilot then reduces the climb by pushing the control sidestick nose-down.

The captain’s displayed airspeed suddenly leaps to 215kt. It is unclear whether this figure is reliable; if it is, the aircraft has slowed dramatically.



The PNF tried several times to call the Captain back.

The non-flying pilot appears to consider the situation serious enough to warrant recalling the captain.



The stall warning was triggered again. The thrust levers were positioned in the TO/GA detent and the PF maintained nose-up inputs. The recorded angle of attack, of around 6 degrees at the triggering of the stall warning, continued to increase.

The A330′s angle of attack is too high, and the aircraft is losing the battle to sustain lift, as demonstrated by the stall alarm, yet the pilot is still keeping the nose pointing upwards – in apparent contradiction to a basic principle of flight: escaping a stall requires the nose to be pushed down, in order to regain a smooth, fast airflow over the wings.

Even though the crew pushes the engines to full power, the high altitude probably renders this ineffective. The resulting thrust is still not enough for the heavy aircraft to generate enough lift in the thinner air.

At the time Air France’s procedures for an A330 stall alarm required pilots to increase thrust immediately, to take-off/go-around power, and reduce pitch attitude. Airbus and other airframers have since drawn up different procedures to prioritise reducing angle of attack rather than powering out of a stall.

The trimmable horizontal stabilizer (THS) passed from 3 to 13 degrees nose-up in about 1 minute and remained in the latter position until the end of the flight. Around fifteen seconds later, the speed displayed on the ISIS increased sharply towards 185 kt; it was then consistent with the other recorded speed. The PF continued to make nose-up inputs. The airplane’s altitude reached its maximum of about 38,000 ft, its pitch attitude and angle of attack being 16 degrees.

The aircraft automatically trims itself to correspond with the nose-up attitude. In the meantime the standby airspeed indicator’s reading suddenly rises, bringing it into line with the reading on the captain’s display; the two speeds have been inconsistent for less than a minute.

Despite the stall condition the flying pilot still holds the nose of the aircraft upwards.



The Captain re-entered the cockpit. During the following seconds, all of the recorded speeds became invalid and the stall warning stopped.

The altitude was then about 35,000 ft, the angle of attack exceeded 40 degrees and the vertical speed was about -10,000 ft/min. The airplane’s pitch attitude did not exceed 15 degrees and the engines’ N1′s were close to 100%. The airplane was subject to roll oscillations that sometimes reached 40 degrees. The PF made an input on the sidestick to the left and nose-up stops, which lasted about 30 seconds.

All of its lift having been sapped, the aircraft virtually free-falls from its peak altitude of 38,000ft. With its nose still pitched up, and the pilot still giving nose-up commands, it drops at 180km/h on a steep trajectory downwards. Although the A330′s engines are operating at high power, the stall attitude prevents restoration of smooth airflow over the wings. The stall warning had only stopped because the system ignored the extreme readings being generated by the aircraft’s predicament.



The PF said “I don’t have any more indications”, and the PNF said “we have no valid indications”. At that moment, the thrust levers were in the IDLE detent and the engines’ N1′s were at 55%. Around fifteen seconds later, the PF made pitch-down inputs. In the following moments, the angle of attack decreased, the speeds became valid again and the stall warning sounded again.

At this point, shortly after the captain re-entered the cockpit, comes the first indication of an attempt to unstall the aircraft, as the pilot starts pushing the nose down. The engines are throttled back to idle power, possibly to aid the rescue by reducing the tendency for thrust at lower altitude to pitch the aircraft up. The decrease in angle of attack and the return of valid speeds – which in turn bring back the stall warning – suggest the beginning of a recovery.



The PF said “we’re going to arrive at level one hundred”. About fifteen seconds later, simultaneous inputs by both pilots on the sidesticks were recorded and the PF said “go ahead you have the controls”. The angle of attack, when it was valid, always remained above 35 degrees.

The A330 descends towards 10,000ft. The pilot surrenders control to another member of the crew – not necessarily the captain - but the angle of attack is still excessive and the aircraft is still falling rapidly.



AF447 runs out of altitude and time. Having failed to recover from the stall, it pancakes onto the ocean surface at 200km/h and disintegrates.

, , , , , ,

10 Responses to AF447: After two long years, six short minutes

  1. Nik 1 June, 2011 at 9:39 am #

    We now have to learn from this accident.

    Misjudged situation caused other crash such as probe icing on take off that led to an indicated airspeed rising rapidly and the crew failing to recognise the discrepancy of the indicator pulled the plane to stall. This accident is used for icing danger awareness training.

    The issue are for the airline training:

    - Why a fully trained crew failed to understand the stalled situation and how to prepare crews to face abnormal attitude with partial instrument panel at high altitude.

    - Could any additional technical solution (incidence indicator) have helped the crew?

    And issue for the manufacturer:

    - Did the PNF see what the PF stick inputs were (I doubt it)? Is sidestick without feedback a relevant option for future aircraft?

    Finally and above anything to my point of view:

    There have been similar airspeed indication incident on the same type in other company within the preceding few months and every time the crew reported difficulty to maintain flightpath. Did the manufacturer give adequate feedback to airlines, and if so, did the airlines take any action to inform the crew so they can get prepared with this kind of situation ahead (that’s finally the main purpose of the pilot job: get prepared for dramatic situation that will certainly never happen).

  2. Anton Beeckman 3 June, 2011 at 5:52 pm #

    The BEA report, released on May 27 2011, states: “The trimmable
    horizontal stabilizer (THS) passed from 3 to 13 degrees nose-up in about 1 minute and
    remained in the latter position until the end of the flight.”

    With regard to these THS data, can you answer the following questions:

    1. Is it correct that this would always happen automatically, or are there factors not mentioned, that would imply the THS was set manually?

    2. If the THS setting did happen automatically, was the pilot’s manual nose-up stick input, the only factor leading to the (nose-up) auto-trim action? Or, put differently: were there no other factors to be taken into account with regard to the THS position ?

    3. Suppose the THS got stuck (not necessarily mechanically speaking, but rather “software-stuck”) in the nose-up 13 deg. position, what chances would the crew have had, if they had pushed the sticks in the full nose-down position ? In other words, if the aircraft found itself in a fully stalled position, would it have been possible to recover from the stall, with a THS in a 13 deg nose-up position ?

    Thanks in advance for your answers.

  3. David Kaminski-Morrow 3 June, 2011 at 7:09 pm #

    1. Is it correct that this would always happen automatically, or are there factors not mentioned, that would imply the THS was set manually?

    While there is no formal confirmation over the THS behaviour, auto-trim would have been functioning and manual trim is rare. Probability overwhelmingly favours a scenario in which the auto-trim brought the THS into the 13deg position automatically – in the same way that there was a full-up auto-trim of the THS in the Perpignan A320 accident on 27 Nov 2008:

    It’s worth reading this report to see the similarities – and differences – to AF447. Manual trim is hardly ever required and there is no indication that the THS was trimmed manually.

    2. If the THS setting did happen automatically, was the pilot’s manual nose-up stick input, the only factor leading to the (nose-up) auto-trim action? Or, put differently: were there no other factors to be taken into account with regard to the THS position ?

    My understanding is that the auto-trim responds to the pilot’s pitch input.

    3. Suppose the THS got stuck (not necessarily mechanically speaking, but rather “software-stuck”) in the nose-up 13 deg. position, what chances would the crew have had, if they had pushed the sticks in the full nose-down position ? In other words, if the aircraft found itself in a fully stalled position, would it have been possible to recover from the stall, with a THS in a 13 deg nose-up position ?

    Again, I’d refer you to the Perpignan A320 crash analysis by the BEA. The THS in that case would have responded if the elevators had been driven past their neutral position – but that did not happen.

    “The nose down commands applied by the Captain on the sidestick brought the elevators, due to the load factor, to the neutral position, without however pushing them to the stops. Consequently, the trimmable stabilizer did not move even though the flight control law was normal.” (p49)

    “The elevators must go beyond the neutral position before the auto trim function adjusts the position of the stabilizer.” (p49)

  4. Alex Erochenko 4 June, 2011 at 2:21 am #

    First of all, blessings for all lives lost in that tragic accident. I can immagine those 4 minutes seemed like 4 hours for the pilots.

    It is evident that they had a clear case of “unreliable airspeed” and followed by a stall down to the ocean. You have to do extensive reseach to judge in the right manner what happened, everything indicates that the pilots did not recognize in time what went on in order to apply immediately the Unreliable Airspeed procedure which at that altitude and heavy G weight does not leave room for error plus if you add turbulence to it (report says had none).

    THS operates automatically on the ‘bus, you pull back on the sidestick and there it trims to attain the new attitude since this is automatic.

    The fly-by-wire protections were lost, so the case is the plane can trim itself to the stall if you keep putting back presure.

    Anyways, what it really suspicious to me is that all 3 ADR (air data ref) units failed to give the right indications the pitot heaters work with 110V heating elements that heat-up the probes to the point in which the metal color turns kind of violet showing how hot they can get at altitude.

    So, ice on all the probes… impossible you have like 4 different ones: 1 for ADR#1, 1 for ADR#2, 1 for ADR#3, the last one for the ISIS (Standby attitude indicator altimeter & airspeed) one sistem could eventually fail.

    All 4, hard to believe.. by the way.. all of these systems are digitally controlled the only old fashioned one is the ISIS, I believe (mechanical, like normal airspeed indicator in a small plane).

    There have been cases in which a similar situation happened and pressing the “probe-window heat button” saved the day.

    Probe window heat is automatic, you never mess with that button unless the window and pitots are iced up on a cold morning, first flight of the day…

    Otherwise, when the PF sets the throttles to TOGA or FLEX for takeoff the electrical anti-icing kicks on for the remainder of the flight. So, this system was automatically turned on but for some strange reason many times when you have strange speed indication, if you turn this one switch to ‘on’ it tells something to the computers controlling the anti-ice for the probes and window that normalizes airpeed indications (even though they are already being heated from takeoff).

    I am not trying to say that I know what happened, this is just an input from exprience of other pilots have had.
    I strongly believe the onboard computers threw something that was not really happening, the system believed it and that’s why the autopilot renders the plane to the pilots.

  5. Anton Beeckman 4 June, 2011 at 8:24 am #

    Many thanks for your clear answers !

  6. AirBoss 15 June, 2011 at 2:56 am #

    What was the Centre of Gravity (% MAC) at time of the incident, was the tail trim tank in use, and was there any attempt to pump it dry and in so doing move the C of G forward?

  7. David Kaminski-Morrow 15 June, 2011 at 9:56 am #

    There has been no final figure on the centre of gravity.

    But the interim report into the accident gave a balance of 23.3% of the MAC, for a forward limit of 22.7% and an aft limit of 36.2% at take-off.

    After take-off the fuel control system automatically adjusts the centre of gravity using the trim tank – there has been no suggestion that this did not occur on AF447.

    Investigators have previously estimated that, shortly before the accident, the balance would have been 37.3-37.8%, which was “within the limits of the operating envelope”.

    There has been no indication of any abnormal action involving pumping fuel from the trim tank.

  8. AirBoss 15 June, 2011 at 12:52 pm #

    Thanks, David. 38% MAC being the limit, that’s close, if “within the limits…”.

  9. H. Bennett 17 June, 2011 at 10:17 pm #

    With the wing stalled, and the tailplane at +13 degrees relative to datum, would it not be even more deeply stalled, and the elevator completely ineffective? The classic condition for a “deep stall” (tailplane within “shadow” of stalled wing) perhaps need not necessarily be present for an irrecoverable situation to exist. Or am I just hopelessly out of date?

  10. Private Pilot 20 June, 2011 at 3:01 am #

    The initial crash report draws no conclusions, but it’s fairly clear that pilot error was a fundamental cause, due to improper reaction to the stall warning horn. The pilot pulled the nose up repeatedly and according to the report, maintained the nose up attitude (due to control inputs) for the entire 4 minute duration of the rapid descent.

    This sounds like the same pilot error that was made by the Continental Connection pilot landing in Buffalo …. he overpowered the stick shaker to pull the nose up when the stall warning sounded, causing an aggravated stall at just 1,000 feet.

    It’s hard to believe that 3 experienced pilots could make such a basic mistake and not correct it over a 4 minute period, yet that’s what the report implies.

Leave a Reply