Menu lock


Jul 6, 2012

AF447 final report could drive critical safety reforms

There is substantially more inside the final report of the French safety agency’s inquiry into the crash of Air France flight AF447 on 1 June 2009 than came out in the press conferenc

Ben Sandilands — Editor of Plane Talking

Ben Sandilands

Editor of Plane Talking

There is substantially more inside the final report of the French safety agency’s inquiry into the crash of Air France flight AF447 on 1 June 2009 than came out in the press conference in Paris overnight.

And that is notwithstanding some very serious issues that were highlighted in that media briefing on the full report into the crash that killed all 228 people on board an Airbus A330-200 when it slammed down onto the mid Atlantic Ocean more than several hours into a flight from Rio de Janeiro to Paris.

As previously established in the second interim report by the agency, the BEA, the disaster sequence began when external measuring devices called pitots iced up causing the auto-pilot to disconnect, after which two junior co-pilots lost control of the jet, in a four minutes and 23 seconds sequence of events in which the captain, summonsed from his rest break, was also unable or unwilling to effectively intervene to diagnose that the jet had been stalled.

The icing and loss of speed data that triggered the lethal series of events was transitory, and has been dealt with by the pilots of other Airbus wide-bodied aircraft without loss of control or of life, yet in the few minutes that it took before the speed measuring devices cleared and came back on line, the jet had been mishandled to the point where the BEA says without an exceptional and knowledgeable input by the pilots a crash had become inevitable.

The autopilot disconnection, which took the two co-pilots left to manage the jet by surprise, began at 2 hours 10 minutes and 4 seconds into the flight, and at 2 hours 14 minutes and 27 seconds, with all of its mechanical and electronic systems fully functional, it struck the sea at a vertical speed of 10,912 feet per minute ( 102 feet per second), at high engine output, nose high, with the synthetic voice in the cockpit saying  ‘pull up, pull up’.

At 2 hours 13 minutes and 32 seconds the pilot flying said “we are going to arrive at level 100” which is 10,000 feet.

At  2 hours 14 minutes and 6 seconds the pilot flying says “I’m pitching up” and second co-pilot not flying says “ Well we need to we are at four thousand feet”.

At 2 hours 14 minutes and 9 seconds the co-pilot not flying says “Let’s go, pull up, pull up, pull up” which is what the annunciator is also saying and has been since it ceased saying ‘sink rate’ three second earlier, some seconds after it ceased saying ‘stall, stall’.

At 2 hours 14 minutes and 23 seconds the co-pilot not flying says “(expletive) We’re going to crash”.  A second later he says “I can’t believe this”.

A second after that the co-pilot flying says or makes a noise or observation not transcribed.

This is followed in the final second by the co-pilot not flying saying “But what’s happening” and the last words, spoken by the captain are “ (ten) degrees pitch attitude”.

The final report says that for most of the control crisis the jet was flown at an angle of attack of up to 40 degrees and seldom below 35 degrees when a valid parameter for this was being supplied to its flight management system.

It says that when there was not a valid pitch measurement recorded the stall warning, which at one stage is heard continuously for 54 seconds on the cockpit voice recorder, ceases, and that intermittent cessation of the stall warning has already caused Air France to criticise Airbus for a ‘fault’ which contributed to the crash by misleading its pilots.

At the press conference the chief investigator, Alan Bouillard said “The crew never realised that the plane had stalled”.

He said Air France had not properly trained its pilots to deal with ‘surprises’ in the cockpit nor to handle high altitude stalls (in fact Air France had not been explicitly required to do this under the A330 operating rules) and said the report called for significant revisions and improvements to pilot training, and to redress limitations that the BEA had identified in the safety case that had been applied to Air France’s, and by implication, other airline A330 operations.

The report says that the investigation had uncovered  a “profound loss of understanding” in the cockpit in a moment of surprise, when the aircraft went into a stall and lost lift.

The pilots lacked training for stall scenarios, and the safety investigator  recommended that flight simulation training be reviewed.

The BEA report points to ‘indications’ of shortcomings in how information is displayed to pilots on the A330, in particular the so-called flight director that pilots rely on to fly the aircraft. It says that after the crew were surprised by the autopilot disengaging the flight director displays at times disappeared and there was also aural overload from alarms that were triggered by the compromised state of the aircraft. While this lead the pilots of AF447 to make incorrect or inappropriate control inputs, the BEA report details 13 instances where pilots were on top of the situation and reacted properly when flying A330s and A340s.

The flight-recorder readings revealed that Chief Pilot Marc Dubois, 58, had been on a routine break when the autopilot disengaged, and that he at no point until the crash took back control of the jet. Instead, the two junior co-pilots, aged 37 and 32, shared the task of stabilizing the plane, with the senior pilot giving occasional commands from the background.

The report says  that the failure in the context of flight in cruise completely surprised the crew.

“The apparent difficulties in handling the aeroplane in turbulence in high altitude resulted in over- handling in roll and a sharp nose-up input” by the co-pilot, the report said.

“The startle effect played a major role in the destabilisation of the flight path and in the two pilots understanding of the situation.”

However it also said other things more explicitly on a full reading, and the English version can be found at this page, where you will need to also download various annexes including the voice recorder transcript using separate links not just the one marked full report.

It says of the conduct of the flight that:

The two co-pilots left in the cabin after the departure of the captain for a rest break  were left with an uncertain strategy for the rest of the flight.

The captain had not formally declared the pilot flying, the less experience of the two co-pilots, to be in charge, and did not answer his concerns before his departure for the rest break about handling the turbulence that was anticipated while crossing the inter tropical convergence zone [a storm belt that stretches across the mid-Atlantic from South America to Africa].

In terms of experience on the type and total experience the pilot designated as flying was significantly less experienced than the other co-pilot.

The latter also held a managerial position with Air France at its operational centre and was therefore considered to be an expert by his peers.

This raises questions about the rationality of (informally) designating this co-pilot as the relief captain. The difference in experience of the two co-pilots naturally resulted in the more experienced of them taking over from the informally designated relief captain, and “without generating any conflict” this takeover lead  rapidly after the auto-pilot disconnection to the inversion of the normal hierarchical structure in the cockpit.

The leadership role switched to the pilot not flying without the command structure formally and explicitly (being) transferred.

The operators training program doesn’t give co-pilots the opportunity to systematically develop the mind-set necessary to perform the role of relief captain aboard flights with augmented crews.

It draws attention to the more experienced co-pilot diverting much of his attention to efforts to bring the captain back to the cockpit from his rest break than perceiving the seriousness of stall warning sounding in the background of the cockpit voice recording and notes that there is no evidence that either co-pilot actually recognised the warning for what it was.

The more experienced co-pilot, who was not flying the jet, tells the captain once he had arrived that they had tried everything but did not understand the situation and had lost control of the aircraft.

The report says:

Only an extremely purposeful crew with a good comprehension of the situation could have carried out an manoeuvre that would have made it possible to recover control of the aircraft. The crew had almost completely lost control of the situation. Until impact there was no valid angle of attack of less than 35 degrees.

In the first minute after the disconnection of the autopilot neither crew member had the clarity of thought necessary to take the corrective actions.


The report emphasises the failure of knowledge and appropriate responses by the crew several times, and uses language which is similar to that employed by psychologists in discussing clinical hysteria, rather than histrionics.

The crew progressively becoming destructured,  and likely  never  understood they were faced with a simple loss of three sources of air speed information.

The loss of coordination and the willing but chaotic cooperation in managing  the surprise generated by the autopilot disconnection  led quickly to loss of cognitive control of the situation and subsequently to loss of physical control of the aircraft.

As foreshadowed in an earlier article, AF447 may become something of a lightning rod for concerns that undue reliance on automation in any modern Airbus or Boeing poses increased safety risks.

This is something the major airline manufacturers have already raised out of the public gaze at safety and technical forums.  There has been a disconnection between airline managements and flight or safety standards as executives increasingly come from business schools that may not understand nor value the ‘excessive’ costs of engineering or piloting excellence.

Just as there has been a disconnection between some in those professions with the economic realities of airlines in the 21st century.

However while this divide can on one hand contribute to airlines going broke slowly, it can on the other hand result in the quick and unexpected death of hundreds of people in an avoidable crash, destroying brand and shareholder value even faster the ferocity of airline competition.

AF447 shouldn’t be a debate about how airliners are designed so much as one about how they are operated by appropriately trained professionals in an effective regulatory environment.

We recommend

From around the web

Powered by Taboola


Leave a comment

10 thoughts on “AF447 final report could drive critical safety reforms

  1. anonflightattendant

    “There has been a disconnection between airline managements and flight or safety standards as executives increasingly come from business schools that may not understand nor value the ‘excessive’ costs of engineering or piloting excellence.”

    Well said Ben. You could easily apply that sentence to the kind of “marketing creep” that has taken safety information out of safety demonstration videos. Air New Zealand is a good case study – that company’s marketing department has slowly prioritised viral clickability over passenger preparedness for an emergency. The preflight safety video (which once featured a safety demonstration) …

    … has slowly been replaced with inane advertorials promoting celebrities, sports teams and others …

    I’m not aware of any safety data that shows passenger survivability increases after watching celebrities talk about themselves.

    Air New Zealand is a good example of how a marketing department can erode the value of some safety assets to the point where those assets have no potential for a return on safety investment. To quote a commenter on youtube:

    “Why don’t you just show a 4 minute montage of plane crash footage, overlaid with text saying “Read your safety card”

    Perhaps that will be the next one.

  2. Treeguy

    The discussion on automation dependency and training in the world of modern aircraft has been building for some time in the industry, maybe this will be the catalyst for public understanding to drive action rather than discussion.

    This vid is a great example

    Say what you like about the big “legacy” carriers in the USA, but one of the legacies they have does seem to be an appreciation of aviation professionalism.

  3. Alex

    A fascinating and insightful read, thank you, Ben.

  4. caf

    It seems like it should be axiomatic to any airman that increasing angle of attack will not arrest a stall, and if you think that’s what the aircraft is telling you then you or it is mistaken.

    Maybe it should be required for commercial pilots to take regular flights in a little bugsmasher just to keep the intuition about the basic physics of flight?

  5. Geoff

    This is an extraordinary report considering that it comes from a continent that takes upon itself the role of assessing non-European Airlines and banning them from it’s airspace if they do not meet European standards.

    It would appear that Air France is guilty of ignoring whatever safety process it has to ensure that relief captains are capable of handling any upset that can occur in the cruise. It would also seem that there is lack of oversight by the French regulator.

    Previous indications from France indicate that a lot of people are going to find themselves on trial.

  6. fractious

    This sort of post is perhaps the major reason I read this blog. I have no experience of flying and don’t work in the industry, but it is evident to me from the report and your comments Ben that there is something seriously wrong in the state of Air France. I was always of the belief that those up the pointy end had undergone extensive and disciplined training precisely so that they knew how to act and react when the $hit hit the fan, and being a majuor European carrier I had also assumed that training would be state-of-the-art.

    Seems that’s not necessarily so. That the chief pilot hadn’t properly handed off is bad enough. That the two co-pilots either didn’t know what to do or did the opposite of what they should have when stall warnings started blaring is dumbfounding. Like caf above (and from watching episodes of air crash investigators and various youtube vids – yes I know that’s not ideal and I know that don’t make me an expert) I imagined that when certain automated systems start doing unexpected things you go back to basic instruments – altitude, air speed, turn/ bank and attitude – to check and cross reference what the plane is actually doing. I don’t want to pile on these crew members (especially since they would have been a lot more distressed than anyone else) and I don’t want to imagine the pressure they were under, but jeepers… not “perceiving the seriousness of stall warning”? There “…is no evidence that either co-pilot actually recognised the warning for what it was”??

    That’s just scary.

    I know sfa about automatic systems on aircraft. But I know automation is almost everywhere these days – computers, cars, supermarkets, trucks, traffic lights and so on. Done right and in many circumstances it takes a lot of the dull repetetive workload off the operator and so prevents boredom and tirdedness and so is a good thing. But it is not and should never be a replacement for skill and knowing how to act and react when things don’t go according to plan. And skill comes from aptitude and experience and – in the first place – proper training. Proper training is expensive, because proper training involves quality and that doesn’t (and shouldn’t) come cheap. I accept that not all expensive training is good training, but that doesn’t mean that all expensive training should be cut to the bone, as this tragic incident demonstrates. I just hope that the bean counters at Jetstar (to name one) get the wake up call they really do deserve.

  7. Smith Trevor

    The BEA report is a whitewash of the fact that the junior co-pilot flying panicked on hearing the stall horn and pulled the joystick fully back until it crashed at terminal velocity (124 mph), thinking he was in a low-altitude take-off and go-around. The crew had never been trained to handle high-altitude stalls; the fatal stall was pilot-induced!

    Unlike other flights that night, these Bozos hadn’t re-routed to avoid the storms and the weather radar had been set on the wrong range until it was too late. Captain’s fault. The captain then goes for a nap without designating command or replying to the junior pilot’s question about the inter-tropical convergence zone. Only the co-pilot not flying notices the weather radar discrepancy later and advises junior ‘Go left a bit’.

    Stinginess by Air France meant this plane’s pitot air-speed tubes had not been upgraded, so all 3 iced-up in the storm, causing dis-engagement of the autopilot. Junior pulls the plane up to its ceiling of 38,000 feet, then it plummets like a stone with its nose up in the rarified air.

    Chaos reigned: turbulence, popping ears, loss of data screens, warning horns. The pitot tubes came back online quite quickly, so they could have re-engaged the autopilot but the junior co-pilot flying was pulling the nose up for all he was worth. Even if he had released the joystick the situation would have stabilised rapidly.

    Captain eventually returns but doesn’t take the controls, or even notice junior is pulling the joystick back like a maniac as the stall warning horn sounds 75 times.

    Senior co-pilot fails to press over-ride button to disable junior’s joystick; at times they are both making inputs, and the A330 accepts both!

    Under control of The Three Stooges, a crash is inevitable. Lessons have been learned but Beoing’s dual-yoke system would have saved the day. Airbus is unlikely to relinquish its video-game joysticks as blame ricochets between themselves and Air France.

  8. drpixie

    While not one to always agree with Ben, this article is a great example of straight-down-the-line analysis – well done.

    Fractious and Trevor – don’t be too quick to blame a specific airline or aircrew. The root of the problem is that airlines no longer choose aircrew of the old-style … cool-as-a-cucumber with masses of varied experience. Instead airlines now prefer to employ young crew with little “real” experience but just sufficient tick-the-box training (preferably self-funded). Management like that situation – because no accountant wants to authorise $1 more right now, and no MBA CEO wants to feel an any way inferior to their staff.

  9. fractious

    @ drpixie – as I said “I don’t want to pile on these crew members…” especially given the pressure they would have been under and that they can’t answer for themselves. Perhaps I didn’t make it especially clear but I also am not singling out AF – as Ben and others have been pointing out, many “reputable” airlines are as likely as AF to be part of the “world’s best practice training” (aka lowest common denominator) cabal, as witness the number of indicents involving Jetstar. Your point about the calibre of aircrew is well made and – regrettably – it is a product of the same mindset that permeates a lot of skilled professions these days. The aircrew who are “old-style … cool-as-a-cucumber with masses of varied experience” as you put it are more often than not wholly incompatible with the current “managerialism” style that most businesses ascribe to and whether the business we are talking about is running a pizza parlour or an airline, we are all the worse off for it.

  10. NeoTheFatCat

    As someone with all of 8 hours in a C152, I know zero about what those pilots were facing. However, as a volunteer firefighter I know a fair bit about how people react under pressure. I have experienced the tunnel vision that comes from adrenalin (my wife constantly reminds me how I get into the ‘zone’ when my pager goes off).

    Is this the missing ingredient? Should pilots be trained to learn how they respond to adrenalin, and how to manage that in an emergency?

    When I think back over the accident reports I have read, it seems that many pilots enter the tunnel where there is only one response, and all other possibilities are ignored or wished away.

    But as I say, I have zero experience to draw on…