Do we do nothing if AF 447 remains a mystery?

It is still possible that the AF 447 recorders will be found.

But if they never are, can the industry afford not to explore what is likely to have happened, rather than what is merely possible?

The list of theoretical possibilities is, at present, so long that an assumption by the industry that any of them might have occurred would lead to a failure to take any action at all.

So what is “likely”? Let’s not get too philosophical here or we’ll end up nowhere useful.

The Airbus A330 was under control in the cruise when some airspeed indication problems occurred and the autopilot and autothrottle tripped out. This happened a couple of hours after midnight when the crew would have been at their deepest circadian low.

The flight control law also changed from normal into alternate, but the latter does not alter the way in which the crew would act to exercise control of the aircraft, nor how the aircraft feels to fly.

The next significant known and understood fact is that the aircraft hit the sea.

There are only two alternative generic scenarios that describe what happened between cruising altitude and the sea: either the aircraft, for an unknown reason, became actually uncontrollable; or it was controllable but the crew was unable to control it.

A study of airline accident history – both recent and going back decades – would suggest the latter is the more likely.

Take two recent nighttime accidents that also happened over the sea: pilot disorientation caused loss of control in  the Flash Airlines accident (2004), and in the Adam Air case (2007) the cause was pilot fixation on troubleshooting a fault followed by pilot disorientation and loss of control. This cannot be ruled out as a cause in the AF 447 case.

But the question is, do we rule it in?

Yes, if that would move the industry to take action prevent such events in future.

That’s my opinion anyway, and here’s why. 

Flash and Adam Air actually happened and the accident investigators revealed the cause, but practically nothing has been done since then about this phenomenon. There is increased interest in upset recovery training, but no agreement on how it could be done.

The fact that AF 447 might have been caused by the same phenomenon just adds additional urgency to the argument for action.

Still not convinced?

Flash and Adam Air were not the only ones. There was also the Gulf Air (2000) and Armavia (2006) A320 crashes. They, also, were caused by pilot disorientation at night over the sea. Nothing was wrong with the aeroplanes in either case.

And a week or so ago we had the Yemenia A310 accident at Moroni in the Comoros Islands. It went into the sea at night too, at about 02:00 local time, and the crew had not reported any problems with the aircraft.   

 Here are some linked truths:

  • Loss of control accidents are becoming proportionately more common as a serious accident category;
  • operating highly automated aircraft give pilots less practice at physical aircraft manipulation, and deprives them of practice in operating and thinking with raw data.
  • pilot training is a soft target for cost cutting because there is no instant perceptible effect from reducing it, just an unquantifiable increase in risk.
  • simulators are essential to affordable training, but they are best for teaching systems knowledge and management, and standard operating procedures.
  • simulators are at their least good when training pilots in manipulative flying skills because their greatest shortcoming is their motion system. It cannot – and will never be able to - replicate reality.
  • Handling training in simulators “does not transfer” to the real aeroplane (US DoT Volpe Center).

But what if simulator motion systems were able to improve, so that “flying” the box feels much closer to flying the real thing? This could enable pilots to be refreshed more effectively in the stick and rudder skills that they lose as a result of flying highly automated aircraft. 

Well, there is such a system, and the industry – including the major simulator manufacturers – should give it a chance. Because if it were proven to deliver transferable training in pure handling skills – like landings in crosswind, for example, or recovery from extreme attitudes – it might have the potential to save lives and aluminium while adding only minimally to training costs.

It’s called Lm², I’ve written about it before, and it’s the brainchild of Filip Van Biervliet of Sabena Flight Academy – Development (SFA-D). It needs to be taken seriously because it delivers. Here’s my description of what I thought when I first “flew” it.

And now I’ve just flown it again (see below) on a Boeing 737-800 full flight simulator at CAE’s Hoofddorp training centre near Amsterdam Schiphol, and have no reason to change my opinion.


  Now what is needed is for a research agency like NASA or the Volpe Center to run trials to see if, with Lm²,  manual flying skills training in a simulator really does transfer to the actual aircraft for the first time.

Because if it does, it is impossible to overstate the importance of this for aviation safety. One thing’s for sure, the industry cannot afford to revert to training in real aeroplanes, so the least it can do is to use the next best training solution.

Is Lm² the next best solution? There’s no excuse for not finding out. 


17 Responses to Do we do nothing if AF 447 remains a mystery?

  1. David Nicholas 10 July, 2009 at 10:23 am #

    Jet upset accidents are not new, but even in the 1960s (when they first became prominent) their were connections made between pilot training and this phenomenon. Of course, pilots have been losing control of aircraft since the advent of manned flight, and handling skills backed up by theoretical knowledge overcame much of the risk so that provided sufficient altitude remained, and adequate airspeed, most pilots could recover from most situations in an intact aircraft. High altitude flight was the domain of the military, and both aircraft and crews were more capable of withstanding the physical forces associated with loss of control and recovery. Once civil aircraft started routinely flying in the thin air of the upper atmosphere, their occupants in shirt sleeves and with loose objects around the cockpit, an upset was much more challenging to the crew. Aircraft (often with podded engines and swept wings) were optimised for high speed level flight and could be structurally intolerant of an upset. Clear Air Turbulence was the probable cause of many of these accidents, the aircraft typically shedding outer wings, empennage and engines as they fell.
    So what has changed in almost 50 years? The aircraft are aerodynamically similar (if much more refined in engineering and systems) and the atmosphere in which they fly is the same. The pilots, however, are less likely to have had a traditional “hands on” training in a military environment, and certainly will have little practical experience of hand-flying the aircraft in anything other than a normal condition. Simulators have achieved standards of fidelity undreamt of by the designers of the fixed-base analogue systems in which the 707 and DC8 pilots practiced their skills, and yet are still simulators – unable to recreate the boundaries of the envelope in a way that pilots can use to practice recovery from unusual or extreme attitudes. Lm2 seems like an advance in some ways, but how would this offer the experience of upset recovery (something that has to be got right first time to protect the aircraft structure) in a synthetic environment?

  2. Lawrence 11 July, 2009 at 3:11 am #

    In my opinion the only way in which this accident can be unfolded through another accident. If in the future a similar accident repeats itself (let’s hope not) data from these accidents can be shared. Assuming that data from the latter accident can be collected. I believe that IF there is a flaw there is a certain probability that a similar accident would repeat itself.

    I don’t see another way how such an accident can be solved without the recorders.

  3. Usmani, Iftikhar J. 11 July, 2009 at 6:16 pm #

    Here I’d agree with Mr Learmont that new automated jets do make pilots over dependant on the automation. Any eventuality of a manual flight gives them jitters, not because of pilots’ incapability but because they are mostly out of practice for a manual flight. Not that they cannot fly it but these jets are designed to be flown automated right after takeoff to just short of touchdown. All the pilot has to do, is to monitor systems and, make an input where required. This automation actually has taken the charm out of manual flight and has made the pilots lazy.

    Automation in aircraft is no doubt a blessing in today’s crowded airspace, flight duty cycles and weather – which is becoming more intense owing to global warming. How many and how often do pilots actually hand-fly aircraft would probably give the analysts a fair idea of how the pilots have become dependant on automation. The pilots of new generation automated aircraft usually hand fly during line and/or recurrent simulator training sessions, but when released online what is practiced is company policy and Captain’s discretion who more often prefers relying on automation than other pilot’s flying skills. This analysis of handflight versus automated flights might give an insight into the boon and bane of automation.

  4. Iain 12 July, 2009 at 10:14 pm #

    Don’t forget the Kenyan Airways 737NG crash in 2007 where no one really know what happened.

  5. Oliie361 13 July, 2009 at 12:09 pm #

    David, as usual, your valuation of the AF447 event is dead centre.
    I shared with you this same opinion for quite sometime and expressed my view that excessive automation would lead to deterioration of professional skills.
    Now I am worried that the Airbus Automation philosophy, for what concerns A380, might lead to similar if not more spectacular disasters, unless flying in uncomfortable environment situations (moderate to heavy turbulence) is stricly forbidden.
    OUr fraternity now is divided between the “goodness” of the Airbus philosophy and Boeing’s.
    I found out that experienced professional pilots favour Boeing’s.
    OTOH, trying and developing the Lm² simulator system is definitely a “MUST” for now!
    If it fails to deliver I think more down to earth training (Stalls & Spirals, Unusual Attitude Recovery Training on traditional a/c trainers should become mandatory.
    This was and remains my opinion.

  6. Andrew 14 July, 2009 at 4:25 am #

    I remember many moons ago, QANTAS who have a pretty impressive safety record used to operate 2 HS125 business jets configured in the cockpit to resemble the 707.

    A lot of their serious training took place in those aircraft.

    Perhaps this is what is needed now.

    It is hard to see for instance how pilots flying a 14 hour leg probably only twice a month with as little a 10-15 minutes of manual flying per flight can possibly stay as sharp as would be desirable through simulation.

    Unfortunately precision airmanship is not like riding a bike. It takes regular practice.

  7. VAMVAKOS 14 July, 2009 at 8:53 am #

    If they suspect the Pitot tubes do not give at times correct indication of speed, then why they do not take advantage of the GPS-aids for measuring speed?

  8. Oliie361 14 July, 2009 at 11:28 am #

    Endeavour Shuttle Launch postponed for the 5th time yesterday.
    Will the A380 suffer the same postponements on takeoff when turbulence and lightning within 20 miles from TO position are reported?
    My sarcasm/doubt derives from my conviction that excessive automation will lead to similar decisions, especially for what concerns the Airbus wide-bodies. Decisions not taken by pilots but by future safety regulations that will be imposed from the constructors for obvious reasons.
    On the Endeavour the pilots are cargo from lift-off to orbit. I hope they will be more involved on the A380 take-off!

  9. oceanrider 14 July, 2009 at 3:25 pm #

    Isn’t this scenario more likely:

    You think airspeed-problems came first.

    What if running into bad weather came first – the pilots struggeling to regain control – and crashing into the ocean on the belly.

  10. Paul 14 July, 2009 at 10:28 pm #

    There is no such thing as a mystery. We simply do not have the right tools since we have some data. May be little data but enough to find out. The team responsible to find out what happenned will eventually compute probabilities and analyse even further with the debris available the more probable course of event. And again, i am confident, Airbus and Boeing will provide us with the best machines humain kind has ever produced.
    Keep up the good work David you are on the top of my list.

  11. David Nicholas 15 July, 2009 at 11:14 am #

    Oceanrider, a deep stall such as that which afflicted the TU154 is a condition peculiar to high-tailed, rear engined aircraft. As such, the A330 would have been in a different aerodynamic situation. That said, there is the stated evidence that much of the fuselage (at least) descended with a low forward speed but, of course, a high vertical speed. From early examples of jet upsets (for example the BOAC B707 accident in Japan in 1964) the extremities (tail, engine pods, outer wings and perhaps the nose) tend to separate leaving the centre section of the wing and fuselage to descend vertically (like a falling leaf), more or less horizontal but with zero forward speed. That would be consistent with the loss of control hypothesis in this case, initiated, I would suggest, by the turbulence-induced tripping out of the autopilot leaving the crew with an aircraft doing things that they had never experienced, in the dark, and perhaps in cloud. This situation does not allow time to experiment – recovery (even if theoretically possible) would have to be carried out correctly and immediately before aerodynamic forces overcame the integrity of the airframe.

  12. Pitchlock 16 July, 2009 at 12:08 am #

    It makes sense for Air France to “cop a plea” and accept that the pitot tubes malfunctioned / iced over which caused erroneous airspeed intications in the cockpit and at the flight data recorder input. That would exonerate AIr France – with the exeption that possibly a low level maintenence director could take the blame.

    But, it has evaded almost every commentators thought process that you penetrate severe weather using a fixed power setting for turbulence penetration airspeed, and disregard the ASI.

    Next you fly attitude, keep the wings level and the nose on the artificial horizon.

    If the pilot can’t hand-fly the Airbus with the side stick controller, maintaining attitude and power only, than the problem is with the automated flight control system of the Airbus A330.

    I’ve done this for hours at a time in a Douglas C-133A, even flying through a typhoon which took us off course 600 miles, but we survived.

    Can Airbus do that? Maybe not!

  13. Taras 18 July, 2009 at 10:20 am #

    Tail of airbus made out of composite material, means plastic. It is not durable as aluminum, simple crack may cause it to broke off. It happen once before in New York. Air France try to hide it now, they occupy all families of victims with their lawyers (, means families will get nothing and no investigation will be made other then Air France one. It’s sad that other people now in great danger of flying those airplanes from Airbus.

  14. Gerrit NIJS 13 August, 2009 at 9:42 am #

    Dear Mr. Learmount,

    I read with interest your article « Do we do nothing if AF 447 remains a mystery ? ».
    The lessons to be learned in your article base themselves on the assumption that the AF 447 flight was controllable but that its pilots were unable to control it.
    If this would be the case then it would only highlight, as you mention, the urgent need to address Upset Recovery Training for pilots.
    As I’m an airline captain on B747-400 airplanes and also aerobatic instructor, giving aerobatic instruction and Upset Recovery Training on a top of the line aerobatic airplane, a Walter Extra 300L (, I can give you some more insight and corrections on the matter.
    It is true that the highly automated airplanes we fly today leave the pilots with little manual flying.
    This phenomenon is recognized by several airlines and they have, over the years, adopted their operation manuals in a way that pilots are encouraged to fly manually when conditions permit.
    Unfortunately, some other airlines did not yet reach this elevated level of good practice in their airline operation.
    Your suggestion that the Lm2 programm of Mr. Van Biervliet of SABENA Flight Academy -Development improves the reality of simulator flying is correct, however, it does not provide a solution to the issue of Upset Recovery Training.
    As is stated by test pilots from Boeing and Airbus in the Upset Recovery Training Aid, simulators are lacking some basic elements in order to be used as tools to provide Upset Recovery Training ; first of all, there is no data available for flight beyond the normal flight envelope, and secondly, the necessary sensory feedback is not available (G-forces).
    They even advise airlines not to use simulators for Upset Recovery Training as it can constitute a negative training (pilots flying the numbers as no, or wrong sensory input available, wrong performance data as data for flight beyond normal flight envelope not available and simply extrapolated from normal flight envelope data, etc.).
    There is simply no alternative to Upset Recovery Training in a real aerobatic airplane.
    Loss of control accounts nowadays for a higher number of airplane accidents than Controlled Flight Into Terrain (CFIT).
    As with CFIT, it took the industry years to tackle the problem, but they eventually did as they couldn’t ignore it anymore.
    It seems to me that the industry is following the same path, let’s just hope that the Air France accident will speed up the process.
    Unfortunately, the authorities have gone the opposite way in recent years by eliminating for instance basic aerobatic training, spins, steep turns as a requirement to obtain a CPL or ATPL pilot’s license.
    JAA has regressed training standards in this matter with the introduction of the JAA pilot’s license, and EASA has not corrected it yet.
    Strangely enough, the pilots that do follow my Upset Recovery Training course are business jet pilots and not airline pilots.
    Does it mean that owners of business jets value their own life more than managers value their passengers life ?
    The airline industry has to deal with the Loss of control issue, as they did with CFIT, let’s hope it will be rather soon than late, as more loss of control accidents will happen, and more people will die because of it.
    Best regards,

    Captain Gerrit NIJS

  15. Jaz 15 September, 2009 at 3:39 pm #

    Not an expert on aircrafts but one keeps on thinking that the “Tail Fin” broke up in the air causing the crash.
    This crash is very similar to the “AA” crash of New York some time back, there also the “Tail Fin” was found in (what looks to) as per this Air France Crash.
    Any further Info on this.

  16. buffalo 28 January, 2010 at 11:04 pm #

    Finding the vertical stabilizer in one part doesn’t mean it caused the crash. The AF A330 hasn’t anything in common with the AA A300 that crashed in NY except it’s fuselage diameter. It’s simple physics that massive parts like a stabilizer don’t go up in dust especially when a airplane disintegrates at high altitude. At very high altitudes the minimum speed (needed for lift in the thin air) and maximum speed (the mechanical load the airplane can handle) are very close. Getting into very turbulent air at high altitude a single heavy gust can disintegrate a airplane.

    Usually nobody would ever fly into a bad weather region like AF-447 did. But today all these MBAs in the airlines executive ranks urge pilots to go beyond the once common sense safety rules.

    Just my 2¢

  17. David Learmount 29 January, 2010 at 10:25 am #

    Buffalo, I agree there are good reasons why a composite tailplane doesn’t “go up in dust”, as you put it. The main one is that, in its manufactured form, it is effectively a monolithic structure, and very strong and rigid.

    The other (more contextual) points are these: the investigators, the BEA, have said that the wreckage they have recovered indicates that the aircraft was “substantially intact” on impact; and as for proceeding through weather, all aeroplanes that go from southern hemisphere to northern or vice-versa have to fly through the inter-tropical convergence zone (ITCZ), which is always littered with cumulo-nimbus build-ups that have to be avoided.

    Navigating the ITCZ regularly is just part of being a long-haul airline pilot. Again, the BEA says that there was nothing special about the weather in the ITCZ that day, and there were other airliners, including another AF A330 ploughing through that same piece of sky at the same time, and none of them came to grief.