Pilots: who needs them?

“Unless pilots are trained to cope when automation fails, logic dictates that manufacturers might as well design them out altogether.”

That’s the logic. But will it survive the argument?

————————

The greatest service the Dutch National Safety Board could render to aviation in its investigation of the Turkish Airlines Boeing 737-800 crash at Schiphol is to use the study to dig deep into the psychology of pilots working in the modern aviation environment. And the psychology of their employers.

The same is true of the French investigators working on the XL Airways Airbus A320 crash during its post-maintenance test flight from Perpignan last November.

Of course, the Dutch will have to look into the system design issues that led the autothrottle to retard the power to idle at about 2,000ft during the instrument landing system approach, but that is the easy bit. Human factors is where it gets complicated.

What is unsettling about these two accidents is that the flightcrew appear to have thrown common sense out the DV window, along with some of the fundamentals of safe piloting. Both crews died, but since they clearly didn’t intend to let their aircraft get into the unrecoverable situations that killed them, the world needs to understand what happened. The Turkish crew forgot to monitor power and indicated airspeed on final approach; the XL crew carried out a flight test that took the aircraft to the very edge of its flight envelope at a suicidally low level while approaching an aerodrome.

The inquiries need to compare these accidents with others like them and try to find out what the common human factors are. In every case, the training the pilots received needs to be put under the spotlight, as does the selection process.

Highly automated aeroplanes have brought higher levels of safety – meaning fewer accidents – than their predecessors.

The downside of high automation appears to have two main facets: the first is that the complexity of fourth-generation systems and their logic means that, although technical faults or anomalies are more rare, unique and unforseen combinations of circumstances with the potential to confuse pilots are more likely; second, although the type of training given for these aircraft ought to have changed appropriately since the days of Lockheed Constellations, it has not changed much, despite industry acknowledgement that traditional pilot skills and awareness can be eroded by automation.

Regulators and the airlines have a lot to answer for: the regulators for their tendency to cling to recurrent training tradition like a drowning man to a piece of driftwood; and the airlines for assuming the latest aeroplanes should deliver not just improved reliability, but lower crew training costs. In fact increased systems complexity, combined with a lack of line flying practice at manual procedures, demands a different pilot training approach.

Unless pilots are trained to cope when automation fails, logic dictates that manufacturers might as well design them out altogether.

But since the uproarious outcry at the loss of the world’s first pilotless passenger aircraft is totally predictable, it will not happen for a very long time. Meantime, the industry had better learn to prepare its pilots properly, because it clearly isn’t doing it now.

18 Responses to Pilots: who needs them?

  1. John Buchanan 5 March, 2009 at 11:46 pm #

    Do pilots no longer monitor ILS indications whilst on autopilot coupled approach?

  2. David Learmount 6 March, 2009 at 9:58 am #

    It appears, from the rather sparse preliminary report, that the autopilot held the aircraft on the glideslope and localiser until the stickshaker operated. If that is confirmed, monitoring the ILS was not the issue because the stickshaker woke the pilots up to what was happening. Failing to monitor the airspeed and power was the core problem.

  3. SDW 6 March, 2009 at 11:11 am #

    As usual a very interesting article by DAVID who I had the pleasure to meet on many occasions and enjoy very interesting conversations.

    An impression that I have with the present PILOT TRAINING is that we are no more paying attention to the real art of flying an AIRCRAFT, as we were once lucky enough in our initial training to have done by basic experiences like GLIDING ( or is it SOARING)?

    I have the impression we are creating PILOTS flying a COMPUTERBOX and no more an AIRCRAFT.

    Very soon they wil be UAV operators instead of PILOTS

    I expect a lot of reactions ???

  4. David Nicholas 6 March, 2009 at 12:43 pm #

    These loss-of-control at low level events seem to be becoming almost endemic, having been relatively uncommon (certainly when compared with other causes, for example CFIT) in recent memory and yet they largely appear to be rooted in poor airmanship. In recent years the very term “airmanship” has almost left the vocabulary of pilots and safety organisations (interestingly, the marine equivalent, seamanship, is still instilled in sailors). Perhaps pilots who have been trained from the start in automated glass-cockpit aircraft have a different perception of aviation and a shallower understanding of aerodynamics. As you suggest, whatever possessed the A320 crew to undertake a slow-flight test on what was effectively the downwind leg of a circuit, especially as they considered other options (do it enroute to Frankfurt or skip it altogether)? The potential hazard of this seems evident to a PPL holder, let alone three commercial pilots who all went along with it without demur, until the aircraft stopped flying.

    The differences between a large aircraft and a small one may in terms of equipment, systems and performance be greater than ever; however, the law of gravity and the forces of lift, weight, thrust and drag apply across the spectrum. Take any aircraft to the edge of the envelope and it is much more likely to bite you. Cross the edge at low altitude and it will kill you.

    In stark contrast to the Hudson River ditching (almost a definition of good airmanship by a pilot with very wide flying experience), recent high profile accidents resulting from loss of control at low altitude perhaps involve human factors concerning one or more of the following – sloppy cockpit procedures, poor application of SOPs, crew fixation on something unusual to the exclusion of watching what the aircraft is doing, poor or non-existent instrument scans – in other words, poor airmanship. That is not to seek to prejudge the outcome of investigations but is based upon what has already come into the public domain.

  5. leson 6 March, 2009 at 8:47 pm #

    Pilots-who needs them?
    Check Hudson river…

  6. Anonymous 7 March, 2009 at 8:57 am #

    To draw the above conclusions from two recent unfortunate accidents is both facile and ignorant. I speak as a pilot of both old and new technology aircraft so I feel I am qualified enough to state that automation incidents,which occur on a daily basis around the world, are generally well handled by professional crews who still rely on common sense sense and airmanship to ensure a routine outcome.
    SDW, I’ve enjoyed dual FMC failures,EFIS failures,instrument,autopilot and flight director malfunctions and dealt with them as I’ve done since I first learnt to fly in single engine piston aircraft i.e. fly the aircraft,know the MSA and remember that
    attitude + power = performance.
    I would also like to quote an instructor of mine,a Vietnam veteran,who ,when a fellow pilot was pontificating about the performance of a recently deceased colleague simply stated-”there but for the grace of God go I”.

  7. David Learmount 7 March, 2009 at 10:48 am #

    Which conclusion is facile and ignorant?

    There is only one conclusion in this piece and that is that pilot training needs to be fit for purpose, and the industry – by and large – has still not adjusted its training to acknowledge the different training needs of pilots who fly highly automated aircraft.

    I may only cite two recent examples in this piece, but in my job I study and report on many like them where pilots depart from SOPs for no good reason and pay the price.

    There will always be pilots like Sully, thank God, and that’s why flying is so safe. But that does not mean the industry can rest on its laurels while accidents continue to happen that could easily have been avoided. Did you listen the pasting Sully gave to the airlines at the congressional hearing?

  8. Doy 8 March, 2009 at 5:17 pm #

    Good piece of analysis as usual David.

    We are now faced with a long string of human factors related accidents – aren’t they all except sabotage? Lesson one – day one of multi crew operations – SOMEONE NEEDS TO FLY THE PLANE! That is even written at teh top of our QRC…. Some of these failures are highly insiduous to spot, and no amount of training can cover it all. I fear for the MPL products sitting next to me at some point….experience will count for much. My passengers frequently stick their heads in prior to departure “looking for a bit of grey hair”…unfortunately they see quite a bit these days……

    I am surprised you made no mention of another recent fatal accident, since your words about the THY accident, and I quote

    “What is unsettling about these two accidents is that the flightcrew appear to have thrown common sense out the DV window, along with some of the fundamentals of safe piloting. Both crews died, but since they clearly didn’t intend to let their aircraft get into the unrecoverable situations that killed them, the world needs to understand what happened. The Turkish crew forgot to monitor power and indicated airspeed on final approach;…..”

    can fully be ascribed to the Buffalo accident….that accident will eventually be found to reek of inadequate training, low experience, fatigue issues for commuting regional pilots who make peanuts and cannot afford to live in base and therefore jumpseat in to cover an assignment.

    I enjoy your writing David. Thank you for your years of highly readable reading in Flight.

  9. Hanan 9 March, 2009 at 1:59 pm #

    I fully agree with David’s analysis. I feel that the term “airmanship” must be updated today, to include the understanding of the in and outs of the new avionics. It’s not just “fly the aircraft” today. It’s also “monitor the automation and react”. I’m sure that the crew of the Turkish Airlines 737-800 saw that somenthing went wrong there but they have not reacted, probably because they didn’t understand what is happening.
    Also, it is very strange to me how do regulators allow a flight critical automatic system to have a single point failure mode. I come from the military side of aviation and there is no such thing in the newer machines (F-16 and up)

  10. alloycowboy 9 March, 2009 at 7:24 pm #

    It’s rather apperant that todays pilots are now more control room operators watching LCD’s rather then the traditional stick and rudder men flying on a wing and a prayer.

  11. Peter Bore 10 March, 2009 at 12:03 am #

    There seems to be agreement that the THY 737 crash, though initiated by an altimeter fault, could have been avoided if the crew had been monitoring that most precious of aviation commodities – airspeed. There are striking similarities with the BA 777 Heathrow crash. Though the initiating event was very different and the BA crew had less time and no power whatever they did, like the THY crew, they left the aircraft on autopilot coupled to the glide slope and only intervened when that aircraft had slowed by 35 knots and the stick shaker activated. In both cases the pilot’s inaction made a stall inevitable.

    To gain my PPL I had to demonstrate that in the event of power loss I was able to bring the aircraft into contact with whatever terrain was available whist it still had flying speed and was still under control. Airline pilots have done this in the past (Alidair Viscout 700 approaching Exeter and the recent Hudson A320). Is it unreasonable to expect that they should still be able to demonstrate those basic skills?

  12. David Nicholas 10 March, 2009 at 10:54 am #

    Peter,
    the THY crew recognised belatedly that their loss of airspeed was the outcome of the autothrottle reducing power. One attempt (after the stickshaker operated) to manually open the throttles was only briefly effective as once the hand was removed the autothrottles once again reduced the power to idle. By this time the captain had taken control from the autopilot. By the time full power was applied again the speed had decayed below the stall and the aircraft fell out of the sky.
    There seems to be a substantive difference here between this event and the BA 777 last year: in the BA case their was no suggestion (that I was aware of) of the crew failing to monitor their airspeed. The marginal but successful attempt to stretch the glide was almost certainly inspired by the ground obstructions beneath the aircraft. As a split-second decision it was justified by the outcome. Had the power loss (or more accurately the refusal of the engines to accelerate when required) occurred slightly earlier then the outcome could have been significant loss of life and damage to property.
    In contrast, the THY crew (with a third pilot on board) seemed oblivious to the declining airspeed until they had run out of the altitude necessary to regain it.
    You are absolutely right about basic skills – it seems almost unbelievable that professional pilots can lose control of their aircraft through apparent weakness in basic airmanship.

  13. Luca Chittaro 10 March, 2009 at 1:32 pm #

    To counteract the “logic”, we could consider Lisanne Bainbridge’s paradox (“The more advanced the automation, the more crucial may be the contribution of the human operator”).

    Her 1983 paper about the “Ironies of Automation” (http://www.bainbrdg.demon.co.uk/Papers/Ironies.html) seems well suited to today’s dilemmas.

  14. pete 10 March, 2009 at 6:52 pm #

    I understand what some says about experience and basic skills (totally agree with), but nowadays that is not enough. It’s not only about flight and managing systems, but it’s abount overseeing complex digital systems that are flying the plane with some type of high level decision making authority that was not there before… not surprisingly all problems we had in the recent past are not on badly maintained planes / old planes / poorly trained crew / small regional airlines / but on the latest equipment with generally top experienced pilots and multiple crews (more than 2 pilots)etc etc… we better get working on this fast as to me is a sign that there is something missing in the training and interaction with advanced systems. so far we have been quite lucky… and new planes with new systems, more automations, new materials, are here… a380, a350, b787… are we fit to fly/control those machines?

  15. Peter Bore 11 March, 2009 at 1:10 am #

    David,

    The suggestion that the crew of BA 777 G-YMMM may not have managed their airspeed optimally is based on the following.

    1. Unless an aircraft’s wheels are on the ground or several thousand feet from the ground, flying an aircraft down to stalling speed is usually helpful.

    2. The AAIB Interim Report indicates that G-YMMM remained on autopilot until the stick-shaker operated at a height of 170 feet and an airspeed of about 110 knots. From that situation, even with power available, a high G impact with the ground would have been probable. Without power it was inevitable.

    3. To maximise gliding range you need to fly at the best L/D ratio. I do not have access to 777 performance data in the landing configuration but as a generalisation L/D ratio is sub-optimal at speeds close to the stall.

    Whether or not the crew’s performance was adequate in the circumstances remains to be seen, Certainly they had very little time and we have not yet heard their side of the story. But there are some grounds for suspecting that, perhaps with different training, it was a performance which could have been bettered.

    Peter

  16. Steve 12 March, 2009 at 1:57 am #

    The FMS section of my company’s Boeing 767 systems manual is 267 pages long. The next longest section, flight instruments, is only 100 pages long. The other day I spent an entire afternoon configuring a new, and very basic, cell phone. There is a pattern here.

    I know of no one who has actually read the entire FMS section of the manual. Rather, there are a set of normal programming and operating tasks that one becomes very familiar with. The manual exists as a reference. The FMS is actually capable of much, much more than we use it for. I suspect that the FMS can do a lot of things simply because it was very easy to add additional lines of code to its software. This leads into the same problem my new cell phone has. We are so enthralled with manipulating microchips that we do so simply because we can, not because we need to. In doing so, we lay traps.

    We actually use about as much of the FMS capability as I do my cell phone’s capability…probably less than twenty percent. That does ninety five percent of what today’s needs require. However, every once in a while, we take a trip down a rabbit trail that undoubtedly leads to somewhere within those 267 pages, but with which we have almost no experience. Then we get in trouble, particularly when the thing starts making attitude and power changes we hadn’t planned on.

    The errors made at Amsterdam, Buffalo, Perpignan, and others are all likely to fall well within the scope of James Reason’s threat and error model. It seems probable that they all involve periods of low residual attention (line training, low time in type, maintenance/acceptance testing), combined with tasks that were vulnerable to automation traps. Low residual attention (Stanley Roscoe’s term) easily leads to slips and lapses, since there simply isn’t enough attention left over to review and update. Misinterpretation of data can lead to selection of the wrong rule for the situation, or selecting the wrong desired outcome. Worst of all, low residual attention can lead to utter confusion when faced with atypical indications…and the inter-relationships of modern automation easily present atypical, counterintuitive indications. None of these errors is “basic”; there is no such thing. None are solved by good old-fashioned airmanship.

    We build standard operating procedures with the objective of preventing excursions down these rabbit trails. SOPs include error traps and mitigations developed through safety analysis and service history. They can vastly improve the situation by expanding the available residual attention. Compliance is essential, but we are often guilty of a failure to educate on the safety system architecture. Compliance can become a challenge to the pilot’s inner Yeagerism, just as it can become a challenge to management’s desire to do so much with so little. If we are going to change our approach to training, it needs to be towards a more comprehensive and professional understanding of threat and error management, risk mitigation, and system safety. Flight crews need to understand the hardware, the software, but in particular, the rational for procedure design, their own vulnerabilities to error, and how procedural compliance provides a defense. The kind of discipline that we are aiming for is not the same kind as is found on the parade ground. This may be disappointing to many armchair investigators. It is, rather, a discipline based on sound understanding of human performance and personal limitations.

    Because, in the end, you simply can’t think of a thought until you think of it. It follows that you will never, ever see the one coming that gets you. The accident is all over with before you really comprehend what has happened. And that can happen to any pilot.

  17. Anonymous 12 March, 2009 at 2:08 am #

    I’m afraid that the Peter Bore comments are typical of someone who speculates based on limited & misunderstood knowledge of an aviation accident (limited because the official report has yet to be published!)

    To suggest that the YMMM crew were ‘unaware’ or did nothing until the stickshaker is fascile and grossly defamatory.

    The accident was most certainly not caused by over-reliance upon automatics – indeed allowing the autopilot to maintain control of the flightpath would have given some much needed spare capacity to the crew who were trying to diagnose/resolve a loss of thrust at a late stage of the approach… the thrust levers were already against the stops!

    Disconnecting the autopilot in this case and maintaining a speed above the stall by increasing rate of descent (the only viable option to achieve that) would have led to the aircraft landing short of the airfield perimeter, most likely onto the Bath Road which crosses the undershoot of 27L resulting almost certainly in a large loss of life.

  18. peter bore 16 March, 2009 at 11:26 pm #

    The comments of Anonymous (March 12) are so wide of the mark that they hardly deserve comment. However the controller of this blog deemed it appropriate to publish them so I think that a response is in order.

    1. The Philosophy of Safety.
    Steve (March12) hits the nail on the head when he says “You cannot think of a thought until you think of it”. It is a significant disadvantage of the human condition. However speculation does allow us to think about what might have happened or what might happen again and thus to think about preventing similar events in the future. The effort which has gone into predicting problems, and therefore preventing them, is the probably the most important of the many contributions to aviation safety. We may never know for sure why the engines on G-YMMM failed to accelerate but the speculation that, having flown through some very cold air that night, fuel icing may have occurred allowed the AAIB to think about the problem and then do some experiments which demonstrate that fuel icing was a probable cause. (AAIB Second Interim report re G-YMMM) “Perhaps the most important distinguishing feature of high reliability organizations is their collective preoccupation with the possibility of failure” (James Reason). To introduce the word “defamatory” into attempts to understand why the crew did what they did would significantly impede the progress of aviation.

    Speculation should not be confused with judgement or with apportioning blame.

    2. Consistency.
    Whilst castigating me for speculating on the basis of limited evidence, Anonymous is prepared to make assertions about the cause which involve the phrase “most certainly”.

    3. Accuracy.
    (a). An official report does in fact exist (AAIB Interim report re G-YMMM), and it contains considerable amounts of data. The final report is yet to come.

    (b). I did not, as anonymous claims, say that the pilots were “unaware” of their airspeed. That is a possibility but there are other possibilities. Did they deliberately fly the aircraft down to stick-shaker speed at 170 feet? That would seem to be unlikely but we have not yet heard from the pilots. Remember that these pilots were only two miles from touchdown after a very long flight and then read again the fourth and the final paragraphs of Steve’s March 12 contribution. Alternatively was there some other factor which prevented them from increasing their airspeed? We must await the final AAIB report.

    4. Aerodynamics.
    Anonymous makes no reference to the difference between rate of descent (i.e. vertical speed as a function of time) and angle of descent (i.e. vertical speed as a function of horizontal speed). It is the latter which is most important in achieving the maximum gliding distance and it is determined by the L/D ratio i.e. the ability to convert kinetic energy (airspeed) and potential energy (height) into lift as efficiently as possible. I do not know what airspeed corresponds to the maximum L/D ratio for a 777 in landing configuration (Perhaps there is someone out there who can supply that information.) but the graphs of pitch, rate of descent and airspeed in the AAIB report suggest that over the last minute of G-YMMM’s flying career, the angle of attack varied from around 5 degrees to almost 20 degrees (plus, of course, any difference between the pitch of the fuselage and the pitch of the wing). It cannot have been at the angle which produced the optimum L/D ratio for all of that time. (The ‘optimum’ L/D ratio might well have been less than the maximum L/D ratio because the limited amount of potential energy (height) available to accelerate the aircraft.)

    Moreover G-YMMM descended at an almost constant 700-800 feet per minute over the two minutes shown on the FDR readout. That seems to be about the right number for an aircraft on a 300 feet per mile glideslope and travelling at two and a half miles a minute. The constant slope of the graph of altitude against time indicates that the autopilot did not reduce the rate of decent, but it did reduce airspeed. If the rate of descent remained the same but horizontal speed decreased then the angle of descent must have increased. Basic PPL aerodynamics teaches that increasing pitch-up at low airspeeds will increase the angle of attack and the drag to an extent that will often outweigh any gain in the coefficient of lift. Basic PPL teaching is that on the descent, airspeed is controlled by elevator trim and descent rate by power. Since most people come to aviation after driving a car, that is counter-intuitive. But aeroplanes are different. Vertical fight-path (particularly in slow flight when one is below the speed for maximum L/D ratio and there is no kinetic energy to spare) is a function of thrust and drag. Speed is controlled by the elevators because, for any given state of trim, most aircraft are stable with respect to speed as a result of the movement of the centre of lift with changing airspeed. These outcomes are elegantly demonstrated in the AAIB FDR data which shows that rate of descent remained constant as the aircraft pitched up from 2 degrees to 14 degrees but airspeed fell. How much further G-YMMM would have travelled had it been closer to the ideal L/D ratio I cannot calculate but had it arrived at ground level whilst it still had flying speed it might have avoided a 2.9G impact.

    The pilots may have performed as well as could be expected in the circumstances. We are not yet in a position to make judgements. But we can look at the available data. That data gives no support to the view that gliding distance was increased by allowing the autopilot to remain coupled to the glideslope. We can, and should, speculate that pilots with different training might, in future events, perform even better. I am sure that the holder of an ATPL knows that it is preferable to bring an aircraft into contact with the ground whilst it is still under control and I am similarly sure that they would have the flying skills to do that. The question is what prevented them from achieving that outcome in this particular situation. The pilots involved will probably remember G-YMMM on every approach they fly in the future. I would expect that they would welcome any assistance that could help them to do it better if it happens again. Of course it is easy for me having had hours to ponder over this data. For the crew, who had to make a decision within a few seconds knowing that their own lives as well as those of their passengers were at stake, it was much more difficult. Though I have been in situations where a very swift decision was needed to save life, the life in question was never my own.

    As Steve points out, when things happen quickly it is well nigh impossible to cope with them optimally unless it is a situation that has been predicted, is expected and has been trained for. If we do not ask the question “Could the pilots performance have been better?” we will never seek answers as to how their performance, and that of pilots in general, could be improved in future events.

    This is not a simple problem and I am not an aviation professional. I have no access to 777 performance data. Thus the above is based on general aerodynamic principles. I accept that there may be errors in my analysis. If so, I hope someone will tell me but I do expect any correction to (a) explain where and why I was wrong and (b) to be consistent with the data from the G-YMMM flight data recorder.

    P.S. There is a typo in my March 11 contribution. “Helpful” in line 4 should of course read “unhelpful”

    The comments of Anonymous (March 12) are so wide of the mark that they hardly deserve comment. However the controller of this blog deemed it appropriate to publish them so I think that a response is in order.

    1. The Philosophy of Safety.
    Steve (March12) hits the nail on the head when he says “You cannot think of a thought until you think of it”. It is a significant disadvantage of the human condition. However speculation does allow us to think about what might have happened or what might happen again and thus to think about preventing similar events in the future. The effort which has gone into predicting problems, and therefore preventing them, is the probably the most important of the many contributions to aviation safety. We may never know for sure why the engines on G-YMMM failed to accelerate but the speculation that, having flown through some very cold air that night, fuel icing may have occurred allowed the AAIB to think about the problem and then do some experiments which demonstrate that fuel icing was a probable cause. (AAIB Second Interim report re G-YMMM) “Perhaps the most important distinguishing feature of high reliability organizations is their collective preoccupation with the possibility of failure” (James Reason). To introduce the word “defamatory” into attempts to understand why the crew did what they did would significantly impede the progress of aviation.

    Speculation should not be confused with judgement or with apportioning blame.

    2. Consistency.
    Whilst castigating me for speculating on the basis of limited evidence, Anonymous is prepared to make assertions about the cause which involve the phrase “most certainly”.

    3. Accuracy.
    (a). An official report does in fact exist (AAIB Interim report re G-YMMM), and it contains considerable amounts of data. The final report is yet to come.

    (b). I did not, as anonymous claims, say that the pilots were “unaware” of their airspeed. That is a possibility but there are other possibilities. Did they deliberately fly the aircraft down to stick-shaker speed at 170 feet? That would seem to be unlikely but we have not yet heard from the pilots. Remember that these pilots were only two miles from touchdown after a very long flight and then read again the fourth and the final paragraphs of Steve’s March 12 contribution. Alternatively was there some other factor which prevented them from increasing their airspeed? We must await the final AAIB report.

    4. Aerodynamics.
    Anonymous makes no reference to the difference between rate of descent (i.e. vertical speed as a function of time) and angle of descent (i.e. vertical speed as a function of horizontal speed). It is the latter which is most important in achieving the maximum gliding distance and it is determined by the L/D ratio i.e. the ability to convert kinetic energy (airspeed) and potential energy (height) into lift as efficiently as possible. I do not know what airspeed corresponds to the maximum L/D ratio for a 777 in landing configuration (Perhaps there is someone out there who can supply that information.) but the graphs of pitch, rate of descent and airspeed in the AAIB report suggest that over the last minute of G-YMMM’s flying career, the angle of attack varied from around 5 degrees to almost 20 degrees (plus, of course, any difference between the pitch of the fuselage and the pitch of the wing). It cannot have been at the angle which produced the optimum L/D ratio for all of that time. (The ‘optimum’ L/D ratio might well have been less than the maximum L/D ratio because the limited amount of potential energy (height) available to accelerate the aircraft.)

    Moreover G-YMMM descended at an almost constant 700-800 feet per minute over the two minutes shown on the FDR readout. That seems to be about the right number for an aircraft on a 300 feet per mile glideslope and travelling at two and a half miles a minute. The constant slope of the graph of altitude against time indicates that the autopilot did not reduce the rate of decent, but it did reduce airspeed. If the rate of descent remained the same but horizontal speed decreased then the angle of descent must have increased. Basic PPL aerodynamics teaches that increasing pitch-up at low airspeeds will increase the angle of attack and the drag to an extent that will often outweigh any gain in the coefficient of lift. Basic PPL teaching is that on the descent, airspeed is controlled by elevator trim and descent rate by power. Since most people come to aviation after driving a car, that is counter-intuitive. But aeroplanes are different. Vertical fight-path (particularly in slow flight when one is below the speed for maximum L/D ratio and there is no kinetic energy to spare) is a function of thrust and drag. Speed is controlled by the elevators because, for any given state of trim, most aircraft are stable with respect to speed as a result of the movement of the centre of lift with changing airspeed. These outcomes are elegantly demonstrated in the AAIB FDR data which shows that rate of descent remained constant as the aircraft pitched up from 2 degrees to 14 degrees but airspeed fell. How much further G-YMMM would have travelled had it been closer to the ideal L/D ratio I cannot calculate but had it arrived at ground level whilst it still had flying speed it might have avoided a 2.9G impact.

    The pilots may have performed as well as could be expected in the circumstances. We are not yet in a position to make judgements. But we can look at the available data. That data gives no support to the view that gliding distance was increased by allowing the autopilot to remain coupled to the glideslope. We can, and should, speculate that pilots with different training might, in future events, perform even better. I am sure that the holder of an ATPL knows that it is preferable to bring an aircraft into contact with the ground whilst it is still under control and I am similarly sure that they would have the flying skills to do that. The question is what prevented them from achieving that outcome in this particular situation. The pilots involved will probably remember G-YMMM on every approach they fly in the future. I would expect that they would welcome any assistance that could help them to do it better if it happens again. Of course it is easy for me having had hours to ponder over this data. For the crew, who had to make a decision within a few seconds knowing that their own lives as well as those of their passengers were at stake, it was much more difficult. Though I have been in situations where a very swift decision was needed to save life, the life in question was never my own.

    As Steve points out, when things happen quickly it is well nigh impossible to cope with them optimally unless it is a situation that has been predicted, is expected and has been trained for. If we do not ask the question “Could the pilots performance have been better?” we will never seek answers as to how their performance, and that of pilots in general, could be improved in future events.

    This is not a simple problem and I am not an aviation professional. I have no access to 777 performance data. Thus the above is based on general aerodynamic principles. I accept that there may be errors in my analysis. If so, I hope someone will tell me but I do expect any correction to (a) explain where and why I was wrong and (b) to be consistent with the data from the G-YMMM flight data recorder.

    P.S. There is a typo in my March 11 contribution. “Helpful” in line 4 should of course read “unhelpful”