G-YMMM: could the crew have done any more?

After an accident, pilots always discuss whether a crew could have done more even if they appear to have done a good job. It’s not usually vicious or critical, it normally feels more like pilots experimenting with ideas about how they might handle a situation like it if they were to meet one themselves.


The crew of flight BA 038, a Boeing 777-200ER on short final approach to Heathrow’s runway 27L after a flight direct from Beijing, were suddenly faced with a technical problem that has never happened before. Not in the effects it had anyway. So any suggestion that they should have been better prepared for it is somewhat wide of the mark.


The presence of some ice in fuel is a known issue, but systems are designed to cope with it. On this day, however, the ice behaved in a unique way, a function, investigators believe, of the particular combination of a long flight in lower than usual outside air temperatures and low average power settings throughout.


The problem showed itself when – on short final approach - the autothrottle system demanded power following many hours at low power settings. As the fuel flow tried to increase, “sticky” ice that had accumulated on the fuel pipe walls was suddenly dislodged and washed downstream, accumulating in a mushy mass on the front face of the fuel/oil heat exchangers which are just upstream of the engines. This restricted fuel flow, denying first the autothrottle and then the pilots the power they were demanding. The engines settled at a setting not far above idle power.


Here’s an account of what happened from that point. Heights are above ground level, times are seconds to go to G-YMMM’s actual touchdown short of the runway, and autopilot and autothrottle are both engaged. The problems begin at 720ft AGL according to the flight data recorder, but it is a few seconds more before they become apparent to the crew:


  • 800ft: Copilot takes control according to BA standard operating procedures;
  • 720ft agl/57sec: Right engine fails to respond, then rolls back power to 1.03 EPR;
  • 50sec: Left engine retards to 1.02EPR, but both engines remain above idle;
  • 600ft/48sec: Copilot becomes aware of a split in power lever positions and, distracted by this, he fails to trip out the autopilot as he had briefed he would do at 600ft;
  • 34sec: Crew recognise low speed and that there is a problem maintaining glideslope. They apply manual power but there is no engine response. Airspeed drops further below reference speed and aircraft descends below the ILS glideslop, but the autopilot continues to raise the nose;
  • 240ft: captain reduces flap from 30deg to 25deg, which AAIB says stretched the glide by about 50m
  • 200ft: Stickshaker operates, copilot pushes stick forward to prevent stall. AIRSPEED LOW master caution warning operates;
  • 150ft: Autopilot is disconnected by copilot’s action in pushing on the control column;
  • 3sec: Capt makes Mayday call;
  • 0sec: Aircraft hits ground at a 1,400ft/min (25m/s) rate of descent, 330m short of runway 27L, and slid 372m before coming to rest. Peak touchdown vertical acceleration 2.9g.

 So what more could the pilots have done?


The only thing the investigators reveal, from their tests reproducing fuel system ice behaviour, is that if the crew had reduced the low power settings actually to idle, the ice blockage would have dissipated. The trouble is that reducing power at a time when your problem is not having enough thrust is totally counter-intuitive.


So forget that option.


They could have pulled the gear up to reduce drag. Yes, but when the gear broke off at touchdown because of the very high rate of descent, it absorbed some of that vertical energy. Without it, the hull impact would have been harder, the tail would have impacted well before the front fuselage, raising the risk of a fuselage breach. And the aircraft still wouldn’t have made it to the runway anyway.


What about the captain’s decision to raise the flap from 30deg to 25deg? The AAIB says it stretched the glide by 51m. Raising it further would risk a stall. Not changing the setting would still have seen the aircraft make it “over the hedge” into Heathrow, and maybe just about clearing the ILS localiser aerial array as well. So doing nothing was an option, but what the captain chose to do definitely reduced the risk.


I couldn’t have done better, and I would be pretty pleased if I could have done as well as Capt Peter Burkill and SFO John Coward did on that day in January 2008.










14 Responses to G-YMMM: could the crew have done any more?

  1. Orlando Giacich 11 February, 2010 at 11:32 am #

    David my friend, you’re too kind.
    My idea of gear up would have obtained safer results and less short final headahes and apprehension.
    Low speed an stick shaker at 200ft, as you discribed, would not have materialized if the flaps were kept at 30°, trust me, and the plane’s attitude would not have been such to cause a hard landing on the tail. My opinion of course, but belive me I
    saw and experienced myself flights (C119 propeller overspeed, ferry flight) with gear down and consequently acquired this fixation that in an emergency situation with gear dow and low speed,retracting the gear up would have been be my first option.
    My best friend crash-died in Congo because of this landing gear down with propeller overspeed at takeoff. Unfortunately he could’nt retract the landing gear because the crew, in the haste to expedite the return to base after the flight test of a new engine, neglected to pull out the landing gear safety pins before taxiing. And yet, he was a brilliant pilot, he had manged to keep the aircraft up for almost five minutes, trying to land it back to the field.
    FYI, my friend flew one engine for two and a half hours to Luluabourg a couple of days before and that was the reason of the flight test.
    Regards oliie361

  2. Eric Hallberg 14 February, 2010 at 12:21 am #

    I feel that it is very harsh to criticize the decision to leave the gear down, it would mean taking a chance since it is unclear what the actual effects would be (unless the investigation says otherwise) keep in mind that a aircraft of the 777 size has landing gear doors the size of a barn door and the first thing that happens is that they open with the result of more rather then less drag. With the engines still turning over the hydraulic power shouldn’t be affected but it still takes some time. My point however is not wether it would produce more or less drag but that unless you have tried it or at least had a good think through about the effects on that specific aircraft type it is very hard to justify such an action.
    Further I would like to object to the statement regarding “doing nothing” even without any configuration change it would still demand some accurate handling. The most impressive bit is that they didn’t freeze or panic, they got on with flying the aircraft.

  3. David Learmount 14 February, 2010 at 10:33 am #

    Thanks for your comment, Eric.

    I’m not sure where you detected the criticism. My personal verdict, which I thought was clear, is that they could not have done any better than they did even if they had tried the alternatives, and they might have done worse.

    But over the two years since the accidents I have heard many pilots getting quite passionate – as pilots sometimes do – about possible alternative courses of action, like retracting the gear.

    I thought this would give people a chance to argue their case, now that the AAIB has provided the missing technical detail about what really happened.

  4. Darren 15 February, 2010 at 12:19 am #

    Good article, David. I am not a pilot, but I have a reasonable understanding of aircraft. I think the crew did a great job with this landing. Reducing the flap angle at such a low airspeed was very interesting. Thankfully, the FO lowered the nose to reduce the AoA, which in turn conserved IAS, and lengthened the plane’s “trajectory”. Please correct me if I am wrong (it wouldn’t be the first time).

    Amazingly, some of the PAX didn’t think the plane had crashed, even though the plane impacted the ground at 23 ft/sec. The undercarriage certainly did its job. One thing is for sure: 777s are built like brick “outhouses”. Well done, Boeing.

  5. Peter Bore 4 March, 2010 at 11:24 pm #

    G-YMMM: could the crew have done any more?
    David Learmount’s statement that the G-YMMM crew encountered a hitherto unexperienced technical problem is true but the effect it had, almost total power loss, is far from unique. Moreover it has happened to two further aircraft since the G-YMMM incident.
    I think the question David poses needs splitting into at least two separate questions. (a) Was it reasonable to expect the pilots of G-YMMM to have done more? (b) Is doing more in future similar events theoretically possible?

    To the first question – These pilots were only two miles from touchdown at their home base after a very long flight. Engine failure came as a big surprise. From the start of the problem to impact was less than one minute. One might reasonably argue that to do better in those circumstances would have been a formidable task. As Steve commented on this topic (March 12 2009) the problem which gets you is the one you have not anticipated.

    To the second question – Gliding distance 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. The graphs of pitch, rate of descent and airspeed in the AAIB report indicate that over the last minute of G-YMMM’s flying career, the pitch angle varied by some 15 degrees whilst rate of descent remain constant (until the last few seconds) and airspeed fell. Thus angle of descent must have steepened and angle of attack increased by more than the increase in pitch angle. Thus G-YMMM cannot have been at the angle of attack which produced the optimum L/D ratio for all of that time. The period from 30 -50 seconds into the graph on page 50, when pitch angle was close to zero, flaps 30 and airspeed close to the target approach speed might indicate that the target approach speed was a fair approximation to the best gliding speed for this configuration. If the aircraft had maintained 135 knots and the same vertical descent rate over the last 40 seconds during which time its speed fell from by 20 knots then it could have traveled on average 10 knots faster for 40 seconds before reaching zero altitude. In 40 seconds at a speed of 10 knots it could have flown another 200 metres and still had speed to flare. It would still have landed short but had it arrived at ground level whilst it still had flying speed it could have avoided a 2.9G impact.

    There now emerges a third question. Is that theoretical improvement a reasonable goal for the future? I would suggest it is.

    1. A 15 hour student pilot embarking on his first solo, flying a piston single, has to be prepared for total loss of power and is expected to be ably to fly at the speed which maximizes gliding distance and to arrive at the ground still with flying speed and still under control. Thus the demands on pilot knowledge and technical skill are not inordinate.

    2. When computing take-off performance it is assumed that if an engine fails just below V1 the pilots will recognize the problem and initiate a rejected take-off within a few seconds. They succeed because until they hear V1 called, they are mentally primed to initiate a rejected take-off and often they have mentally rehearsed the RTO procedure shortly before commencing take off. Thus, in these circumstances, recognizing and responding to a problem within seconds is considered acceptable.

    The most important question now is “ If doing more was possible, what needs to be done to ensure that in the future, a crew in similar circumstances will do better?” The answer I suspect is that sometime during the descent the crew should be reminded that there is no guarantee that thrust will be available when they open the throttles and thus they need to consider how they will respond if there is loss of power.
    (Minimise drag if feasible – fly as close as possible to speed which gives the best L/D ratio for the configuration – arrive at the ground still with flying speed and still under control.)

    I would not be so bold as to disagree with David and say the pilots should have been better prepared but, on the basis of the above, I think that they could have been better prepared. The next time total power loss occurs on the approach an extra 200 meters glide may avert a catastrophe.

    Remember Steve’s comment. “The problem which gets you is the one you have not anticipated.”

    Peter Bore

  6. David Learmount 5 March, 2010 at 10:36 am #

    Excellent analysis, Peter, and well-argued.

    I agree pilots could be better prepared in the future.

    I cannot honestly remember if, in my regular C-130 flying days, I carried in my head – ready for instant retrieval if required – the IAS for an optimum L./D ratio taking the weight into account at all stages of the trip, but I think not. The manuals are always there, but you need plenty of height to have the luxury of consulting those.

    But as you say, you knew the best L/D speed when you were flying your first solo, so why not promote a more general awareness among professional transport pilots? I bet that Cessna Caravan pilots and those in charge of other single-engine transport machines carry that awareness with them.

    I now await a flood of responses telling me that the best L/D IAS is always in the forefront of the mind of numerous pilots of four-engined aircraft. If you can honestly say that you’ll win my sneaking admiration, but no-one loves a smartass.

    Sully has said quite a lot about the considerations running through his mind on his power-free finals for the Hudson, but mostly I recall they were about stability, and achieving a sensible speed at touchdown with wings absolutely level. But apart from avoiding bridges, he had a luxuriously big “runway”, so stretching his glide was not the primary consideration.

  7. Peter Bore 9 March, 2010 at 3:51 am #

    Thanks David,

    The problem of remembering critical speeds for best L/D ratios may not be too demanding. The ideal would be an angle of attack (AOA) meter on the instrument panel and one would simply fly the speed which gave the optimum AOA for the aircraft in question. AOA can be sensed by aerodynamic vanes (as in some stall warning devices) or could be computed from airspeed, vertical speed and pitch angle – allowing of course for the “rigging angle “ of the wings. The situation gets a bit complicated when flaps are extended. In general flaps increase both lift and drag but not both at the same rate. Whilst flaps do not increase the stalling AOA, leading edge devices do and because, at some extensions, flaps effectively change the rigging angle of the wing (because the angle of the chord changes), the pitch angle of the fuselage may be decreased at the stall. I have not succeeded in finding any information as to whether or not the AOA corresponding to L/D max changes with flap setting. It probably does but in a manner which will be aircraft specific. However all of these variables could be incorporated into an AOA meter designed for use on a particular aircraft.

    If you cannot afford the luxury of an aircraft with an AOA meter (I think Concorde was the only civil aircraft to have one) then some easily memorable approximations may be available. I emphasis that I do not possess either the aerodynamic expertise nor the aircraft performance data to do anything more than offer some slightly informed speculation but I will do that at least in the hope of promoting a discussion. My professional backgound is in surgery. My only aviation qualification is a little-used PPL.

    After take-off
    At take off you want the aircraft to use as much of its thrust as possible to climb, and as little as possible to be expended on drag. These are not the only considerations but they suggest that climb-out speeds will be a reasonable guide to the best L/D ratio at the take off weight and configuration being used. I would guess that a guide such as V2 plus so many knots or V2 x 1 point something (the ‘so many knots’ or the ‘1 point something’ will probably be aircraft specific) might be acceptable. V3 or V4 might be used in place of V2. I presume that V1,V2,V3 and V4 are all ‘bugged’ on the ASI.

    Most aircraft have a defined long range cruise speed and most operators these days are very keen to have their aircraft cruise at the best L/D ratio in order to save fuel so a reasonable guide if you loose all power in the cruise is to keep doing the speed you are already doing. But as David points out, in the cruise you have the luxury of time to consult manuals.

    As I suggested previously, late on the approach with large amounts of flap extended the target approach speed will be a useful guide but it too would need to be expressed as TAS x 1 point something if the TAS requires an AOA greater than the AOA for L/Dmax.

    Late in the climb or early in the approach when only small amounts of flap are in use one would have the ability clean up and use the long range cruise IAS but on the other hand small amounts of flap contribute more lift than drag and it may be just as good to maintain the existing configuration and IAS in the subsequent glide. (Interesting thought – can you improve gliding angle by using a small amount of flap as opposed to clean?)

    Dare I suggest that in many, though not all, situations, if you disconnect the autopilot and take your hands off everything, a well-trimmed speed-stable aircraft will, at least for starters, make some sensible decisions by itself.

    As I said earlier, the above are no more than minimally informed guesses but I hope that I might have persuaded someone with more aerodynamic expertise and more access to aircraft performance data than I possess, to ratify, refute or refine these suggestions.

    I look forward to the discussion. Even if I am totally wrong, I may learn something.

    Peter Bore

  8. David Connolly 10 March, 2010 at 1:13 pm #

    Peter Bore represents the optimal of surgical SARPS= standards and recommedded practices, namely that the only stupid question, is the question not asked. And always a listening reader of a second opinion. Any pilot under his scalpel should be confident of an optimal arrival.
    Peter is informatively correct in observation of V1-REF inclusive. Of course all V Speeds are deferential to the referential of a stall from V2, which is VS X 1.2. That is why V2 is the first speed appearing in the A/P’s MCP window and in magenta at the top of the PFD’s speed tape. Indeed, as said, pilots are always primed to reject at V1 and it is normal SOP for V1 to be called 5 kts before so that the Captain’s hand is off the throttles at V1.
    VREF is the naked stalling speed margin for a landing flap setting of VS X 1.3. To that is added a prevailing wind additive of 0.5 of the steady state wind plus the full gust value to obtain the target speed, max addition being 20 kts. On my type, the B-744, at 250 tonnes AUW with flaps 30 set for landing, the VREF would be 149 kts. Assuming the tower-reported winds are 8 kts, gusting to 16 kts, the target speed is 161 kts (149 + 4 + 8). The threshold speed is VREF+Gust being 157 kts in this case, bleeding back towards VREF in the flare at 50 ft RADALT. Therefore, in still or calm conditions, the minimum target threshold speed is VREF + 5kts. Think of it as an “Assumed Gust Additive”. The threshold target speed is referred to as VREFAS=VREF Final Approach Speed. Flaps 25 instead of 30 is minus 6 kts from the flaps 30 VREF and a good reflex to remember if one is in a BA-038 situation of puffing instead of thrusting on short final.

    So transposing a BA-038 scenario to a B-744, I don’t think I would do anything differently, concentrate, maintain humour, perspective and discipline. Drag is best left to transvestites,in extremis. Being given no choice is very liberating, almost to the point of Zen. To ratify, refute or refine ? as Peter says, Well, every day is a school day-and as in surgery, we aim for perfection without quite achieving it, for if we did, in Nirvana theory, we would stagnate and regress. Though on occasion, ambient conditions conform with a MET perfect ISA world, however fleetingly. And rembember, the Boeing or Airbus airship travels in True/AS but feels in Indicated/AS. And if not deferential to the referential from VR-VREF, it’s feelings can be very badly hurt and dented with very expensive crunching and scraping sounds.

  9. Peter Bore 17 March, 2010 at 10:49 am #

    David Connolly is, I suspect, very good at the clipped, clear, precise and unambiguous mode of communication which rightly is the hallmark of an aircraft cockpit. It seem that he sometimes tires of it and needs a complete change. It is then that he writes to Flight.

    When I can understand them I find his contributions to be pertinent, thoughtful, densely technical in their language, (something to which doctors are not immune) and a touch mystical. Into this mixture is added a dash of humour, philosophy, irony and occasionally a hint of an almost fundamentalist “Boeingophilia.”

    I have managed to translate much of his G-YMMM contribution but I cannot reliably discern whether my comments have generated a hint of approval or just irony. And, given my incessant references to AOA and L/Dmax, I am puzzled by there being no reference to them. However he does point out that around take-off and landing all of the V numbers are derived from Vs. Given that Vs and L/Dmax are both related to AOA then I am encouraged to persist with the idea that speeds derived from the V numbers might provide some easily memorable guidance at critical times.

    Any pilot imagining himself in the G-YMMM scenario must surely wish that there would be something to remind him that he had been distracted from the task of disconnecting the autopilot and, that that reminder would have occurred before the aircraft was at stick-shaker speed at 200 feet. A clear idea of the desired airspeed just might have led to an earlier disconnection of the autopilot. To me it seems that the villain in this story (the second villain perhaps since the first was the FOHE) is the autopilot which was prepared to fly the aircraft almost to the stall at low altitude whilst pursuing an unobtainable goal. The autopilot would have had the data, but apparently not the ‘intelligence,’ to recognise that staying on the glideslope was not possible. Perhaps, to be provocative, an Airbus autopilot would have recognised this and either disconnected or simply flown airspeed and attitude. Perhaps not, but the scenario is a good illustration of the point John Farley refers to (in another current Learmount blog) about the way autopilots respond when faced with data they cannot cope with.

    I wonder if this demi-god from 30,000 feet (a position which William Langeweische and Antoine de St. Exupery would insist gives a new and valuable perspective on the world) can supply a translation for this ground based mortal who can only get to 30,000 feet through the good offices of the David Connollys of this world.

  10. David Learmount 17 March, 2010 at 10:57 am #


    Connolly knows a thing or two, but the explanation for his prose is that he has kissed the blarney stone.

    And – only his friends know this – he is really a leprechaun.

    I dare you to deny it, Connolly!

  11. David Connolly 23 March, 2010 at 3:56 pm #

    Well ,be Gosh and be Gorra, I wish I was in Gomorra with a crock of whores instead of a crock of fool’s gold under an ephemeral rainbow, I’d get more lucky. Fool’s Gold is Gordon Brown’s speciality. Being a black-Beamish-Murphys-Guinness-sheep Corkonian Leprechaun a few nautical miles East of Blarney town in Cork City, I expect nothing less. And be Gob its no coincidence that ye do be both writing to me on the InterWeb on St. Paddy’s Day ? Clearly the advantage I possess, is that of an ex-communicated Irish Roman Catholic-Roman-Katholikos=Greek for Universal=Local Oxymoronic. Like GDP in Greek-Hellenic accounting, which translates as Greek Decimal Point, a Protestant is merely a skeptical Catholic and has a lot to protest about. Being born into a Catholic Taliban state, ICTS, think GDR via religion, it simply gave me so much to recover from in particular and Ireland in general. Think how the Pontifax Maximus, a.k.a. ze Papenfuher, alias, Bennedict-Benny Joe Ratzo, runs ze Planet Vatican and execute a reciprocal heading PDQ ! Papa Ratzo is his cuddly, non-Rottweiler alias, more rot and less weiller or wolf in sheep’s drag in and to demonstrated effect, QED ! In particular the latest weasel words from the moral high ground to moral low road from eternal Roma. As Oscar Wilde said, “morals are the standards we use to judge those we despise”. Lesson for aviation ?…in one, for me, duty of care is not contingent upon rank, by any metric.

    I do not deny David’s comments of Blarney/Cork in general. It is in particular that me, being a Corkonian-Paddy, that both of your greatly back-handed complementary comments were transmitted on St. Paddy’s Day of March 17th. I dare say, coincidence or serendipity ?,it was transparent perhaps to you, the dual transmitters, in that chronological particular. Thanks for your general consideration, much appreciated !

    To the particulars. I-unlike most surgeons and consultant surgeons-am not “a demi-god from FL390(FL300=LRC these days)” that bury their mistakes, CFIT excepted. My “good office” mind is open, intellectually at least. Like a parachute, it is best used when opened. My father was a tradesman-joiner/carpenter and is living ,fairly well, through variable cardiac care, angioplasty, non-bypass. He pharma-rattles somewhat when he walks. My wife was forecast to be a mastectomy patient in 2007. This was greater than our oncologist led us to believe as a final surgical solution. Therefore, I demanded a 2nd opinion, from a 2nd surgeon at a 2nd hospital. My “clipped” communication to the surgeon was CVR-like thus :”Considering the patient’s de-facto pathologically complete response would you execute a mastectomy ?” . “No !, I would not !, I will write to convey my opinion to Dr.X” was his, slightly redacted response. Fee=Eighteen/18 Euro. Thereafter, I was somewhat shocked to have two leading cancer surgeons come to surgical solutions at each end of the operational spectrum, operationally speaking, in Learspeak. A third opinion was out in case it agreed with the first. In my view it was the best TOGA-Go Around call I ever made, bar none. This was Jan 3rd 2007 in the Christmas/New Year holiday period. Without a second opinion, my other half would have gone with the halved option without a balancing second opinion and the attendant reconstructive potential of being an asymmetric surgical patient for life thereafter. In CAT III autoland speak, this is CAT II- “fail passive” rather than CAT III- “fail operational” .The original surgeon performed the now relatively minor lumpectomy with a surgical transit time of 180 minutes, as opposed to mastectomy’s 90 minutes. I only questioned his judgement, not his scalpel skill. The de-jure lumpectomy-defacto biopsy was a de-jure pathologically complete response, thus far, de-facto in hope. A new twist on ETOPless, so to speak. The patient wife, very patient to put up with technocratic me obviously, is now on a Pre-Tamoplex-Post-Arimidex 3 out of 5 year pharama-transition exit and gyms and swims daily, like myself, when opportune. No time for fitness=more time for illness. And she periodically bungee jumps with me in Scheveningen-Den Haag/The Hague, 90 minutes drive from Brussels. Formula simple, motivation difficult, diet-exercise with a dash of adrenalin, stitched with innate humour. So, I am very glad of real demi-gods like Peter Bores, luckily for me and mine. I most certainly am not and have never presumed to have seen myself as such. A stout-pint drinking-aerosexual-womanizing Leprechaun shall suffice upon my epithet, in between Michael O’ Leary and Willie Walsh in temperament. My faith is not absolute though. The only stupid question is the one not asked, particularly of hierarchy and orthodoxy of any metric. I have been under the surgical scalpel myself, a few times, since childhood for minor issues from hernias to circumcision, and I’m not even Jewish, more is the pity. I recommend “The Checklist Manifesto” by Dr. Atul Gawande, that actually brings a “clipped” checklist philosophy to the operating theatre, that may save more lives in aggregate than any pilot’s checklist, ironically in isolation. The best surgeon knows when not to operate and most surgical fatalities are the result of the anaesthetist/goalkeeper’s failure, I also hold a historic 1998 F/E licence, so know how the side saddle pilot F/E can kill the forward jockeys at the first hurdle. I prefer economy of words to prevent inflation, like the noun of hero, devalued, in my view, to the point of hyper-inflation. How about thorough professional instead ?. Flight’s 250 word letter limit imposes ECON Prose, I have been hyper-economical ever since, which is more optimal than hyper-inflation, by any metric. I am though, through training and philosophy an advocate of “Boeingophilia”, without apology.
    I am no fundamentalist either, per se. But I am fun loving, like Norn Iron’s First Mistress, Iris Robinson is. I have flown Airbus simulators, A-340 in Toulouse and A-320 in Northwest’s NATCO in Minneapolis and while I see the overall benefits, from a handling perspective, apart from taxing, I felt somewhat peripheral, but that is just my perspective. The taxing I found to be very crisp and far superior to the B-737 version in fairness. It is perhaps more equal than the B-744/777. I’m being somewhat rebellious here as conformity can be a prelude to collapsing chaos and while hindsight is retrospective foresight, it prevents overreaction in the future from a present mindset. Think environmentalism, think clearly through it’s fog, more mental than enviro. Like psychiatry, it makes organized religion appear scientific. Organised crime and religion in Ireland have proved to be synonymous via the immutable and infallible law of gravity, ask Benny Joe Ratzo ! Needless to say, gravity is mutable via ze infallible Papenfuhrer. QED !
    Boeing builds a plane for the worst pilot. That is, muppet pilot-me, I am my own most strident critic. I generally agree with John Farley, regarding automating handling, but as being an ex-Roman Catholic, not the craven lapsed variety, never trust anything absolutely, so a little bit of Stick and Rudder Princely handling is really important to keep one’s hand active, so to speak to note Peter’s quoted imminent and pertinent authors.
    I refer below to the A320 of 1992 and A-300-600 of 1994 and the A-330 of 1994, for Airbus and autopilots reference.
    An Air Inter A-320 F-WWDP crashed on Jan 20 1992. While on approach to Strasbourg the aircraft impacted into the side of a mountain. The cause of the crash was found to be a faulty design in an autopilot mode selector switch which led the flight crew to inadvertently select a 3,300 V/S Vertical Speed foot per minute descent rate on the approach instead of the desired 3.3 FPAー Flight Path Angle. This decimal disaster design fault has been corrected.
    On April 26 1994, a China Airlines A-300-600 in Nagoya stalled on autopilot when the stabilizer trimmed against the F/O’s pushing on the yoke, the stabilizer won and the plane was lost. This brilliantly novel feature was designed to prevent an autoland being destabilized. Somewhat plane stupid in my view. As BA-038 demonstrated, the autopilot is sensitive enough to give the plane back to the pilot when it is being pushed around, it is more democratic in effect. If you want to try and fly an ILS with a thrust deficit, it will do that, to the best of it’s ability to a stability deficit over widely variable stalling speeds over a static stalling AOA and yawing over an equally variable VMCA. On the B-744 for CAT III “fail operational” compliance, active control of the rudder,other than the full time “defacto’ autoYAWpilot-yaw damper, is only active after pressing the MCP’s APP switch which arms the other 2 autopilots for engagement at 1500 RADALT, making an engine out approach and autolanding, a relative non-event. The autopilot is not telepathic though-yet.
    The 1994 A330 test flight crash occurred on 30 June 1994 when an Airbus A330-300, registration F-WWKH, crashed at Toulouse-Blagnac Airport while undergoing performance tests. It was the first fatal accident involving an Airbus A330 and the first in which an Airbus 330 was lost. It remained the only fatal accident involving an A330 until Air France Flight 447 crashed in the South Atlantic en-route from Rio to Paris on 1 June 2009 with 228 fatalities. Apparently the submarine technology capable of locating AF-447 exists and the US DOD is willing to locate it-if formally requested. French hubris stands as an obstacle, as the US DOD will not lease sensitive technology to France due to a trust deficit.
    The objective of the flight was to test the performance of the aircraft with the centre of gravity near its aft limit. Tons of water were carried in bladders in the rear of the aircraft’s cabin to move the center of gravity to the desired position. The particular test that led to the crash was a simulated engine failure after takeoff, which meant shutting down one of the aircraft’s Pratt & Whitney PW4000 engines and switching off a hydraulic circuit. During the test, the aircraft’s autopilot would also be set to fly the plane to an altitude of 2000 feet.
    Onboard were three crew and four passengers. Two of the passengers were Airbus employees and two were Alitalia employees selected to observe the A330, which the company was considering purchasing.
    • Nicholas Warner (born June 7, 1943 in Colchester England), chief test pilot and captain. 7,713 flying hours experience.
    • Michel Cais (born November 4, 1940 in Paris France), co-pilot. 9,558 flying hours experience.
    • Jean Pierre Petit (born August 23, 1943 in Boulogne-sur-Mer, France), engineer. 6,225 flying hours experience.
    • Alberto Nassetti, Alitalia employee (pilot).
    • Pier Paolo Racchetti, Alitalia employee (pilot).
    • Philippe Tournoux, Airbus employee.
    • Keith Hulse, Airbus employee.
    The aircraft had just successfully completed a landing in extreme weather conditions and was in the process of completing a takeoff in the same conditions. The aircraft was flown by the co-pilot while the actions to shut off the #1 engine and hydraulic circuit, and engage the autopilot, were to be carried out by the captain. The takeoff was completed successfully and the captain shut off the engine and hydraulic circuit. However there was a problem with the autopilot and it took several attempts to engage it. When the autopilot finally engaged, the aircraft started to ascend to 2000 ft. However, the aircraft rose too sharply and began losing speed. The speed decreased to 100 knots, but the minimum speed for controlling the aircraft in yaw is 118 knots VMCA. The aircraft started to roll so the crew reduced power to the operating #2 engine to reduce the thrust asymmetry; however, this made the problem worse and the aircraft rolled over 90 degrees left and pitched down by 15 degrees and shortly afterwards crashed into the ground.
    Reasons for crash
    There were many reasons for the crash. One is pilot error: specifically, the captain had too many tasks to execute in a short amount of time. The test was carried out at an altitude too low for such a potentially-dangerous test. In addition the autopilot may not have been working correctly, or the crew may not have been fully aware of how to use it. It is also claimed that the autopilot was using experimental software which was being tested for Category III approaches. The autopilot was also in control of speed and angle of attack and it is believed that this vital part was malfunctioning through a software design error. Unlike a naked V/S mode, the integrated VNAV mode assumed it had the same vertical speed schedule before the engine was shut down as after. In short, it pitched up to attain an altitude it did not have the thrust to acquire from initial autopilot engagement. Or as Peter says, “whilst pursuing an unobtainable goal”
    So, answering Peter’s “incessant” L/D Max A/P AOA pertinent points, an Airbus autopilot has a greater “demonstrated” bundled or bungled ? potential to pursue unobtainable goals, from it’s record from Strasbourg to Toulouse via Nagoya. The Airbus autopilot philosophy is Alpha Floor-L/D max and max thrust to get on the front side of the drag curve. Airbus autopilot is therefore pitch stability biased. Boeing’s is speed stability biased. So on the B-777, when even in manual flight operating the twin(power and direction) trim switches on the bull wheel yoke do not move the stabilizer directly in flight, but send an IAS speed demand to the PFC-Primary Flight Computer or through the 3 AFDC-Autopilot Flight Director Computers to PFC-Primary Flight Computer and from there to the Actuator Control Electronics converting the digital PFC to analog ACE to the hydraulic PCU-Power Control Unit, to move the elevator, hold it 3.5+/- seconds and trim out the load to fair the elevator to neutral.
    Peter is entirely correct in advocating a greater awareness of pilots of the right V numbers at the right time. This would have avoided the near disaster of Singapore Airlines B-744 tail drag VMU-demo in Auckland on March 12 2003 and Air Canada’s A-340-300 on June 14 2002, that’s the one powered by 4 hairdryers and only appears to climb because of the earth’s curvature.
    Regarding the Autopilot flying the plane to a stall and preventing it, that is what the stickshaker is for and it does provide a margin, for the Good, it is not God, singular or plural. It provides a compelling audio and tactile wakeup call, conspicuous by it’s absence on the A-318-380 series. On the B-744 and all current production Boeings, the autopilot has a Vertical Speed switch called V/S. It also stands for Very Special. That is because it offers no horizontal IAS speed protection. Why ?, because of the need to be able to unbundle autopilot modes into basic constituent default modes for flexibility and safety, ironic as it may sound. Lets assume VNAV is inop, so we are climbing out using FLCH. Flight Level Change is only seeking to close the gap between the MCP and PFD in 2 minutes using variable thrust up to reference without advertising it on the PFD. Think of it as a pitch and thrust blended mode. So THR is PFD annunciated. Then imagine a TCAS Climb RA, then one winds in a 6000 FPM V/S impossible demand. Well ,the A/T Mode will change to SPD and it will pitch up to meet that demand all the way to a stall. Normally a TCAS TA from above would have one tweaking the V/S wheel to wind off the normal VNAV or FLCH climb rate. In the CRZ the BA-038 flight used V/S to slope rather than step climb, to minimize throttle movement associated with VNAV or FLCH climbing. Good practice, notwithstanding the Rolls Royce FOHE latent design fault.
    Lastly Peter’s diagnosis of my condition, is, in this patient’s view, 100% correct. My written word is the long shadow of my face spoken word. I avoid phones mostly, even from family, I keep it generally textual. And if one loses one’s sense of humour one should qualify immediately for a disabled parking permit. I shall not be seeking a second opinion from any other doctor apart from my Brussels EBBR-Zaventem FAA 1st Class medical examiner, luckily, I drink less than her-I think. And yes, she is a ravishing minx that sets my pulse racing, while maintaining 120 over 80 optimal, demi-gods willing… And David, sorry, but I never kissed the Blarney stone, that is the prerogative of tourists like yourself, Slainte…

  12. old timer 25 May, 2010 at 8:46 pm #

    Good argument by all, however for someone like me who does not have any experience on flying i find it worrying to accept the fact nothing could have been done to recover from this accident
    i do not like the blame and train theory however being defensive by saying nothing could have been done by the pilot does not give me the comfort as PAX

    So please concentrate on what could have been done rather than what can not be done
    We need to see the rate of the captain error percentage reduced

  13. David Learmount 26 May, 2010 at 6:20 pm #

    Old Timer, you’re missing the point. The question the blog poses (could the crew have done any better?) does not presume they made any errors. We know they did not make any errors. The verdict is that they made the best of a bad situation. The engines failed to deliver power, and there was absolutely nothing the crew could have done to get the power back.

  14. Peter Bore 3 June, 2010 at 3:36 am #

    David. The question ‘Could the G-YMMM crew have done more?’ is no longer the question which needs an answer. The question now is can the next crew who encounters a similar scenario to G-YMMM do more. I suspect that that was the question ‘old timer’ intended. (Perhaps you should start a new blog asking that question.) The problem has already happened at least three times since G-YMMM; the Hudson A320, the Amsterdam 737 and the Biggin Hill Citation – different initiating causes but a final common path of loss of thrust at low altitude.

    In January 2008 the G-YMMM crew could not have done more given that they had so little time to respond to an unexpected event. The sequence of events from the AAIB report was as follows;
    60 second to impact – first one engine, followed 7 seconds later by the other engine, rolled back.
    45 seconds to impact – the crew became aware of asymmetric thrust lever positions and were distracted from disconnecting the autopilot.
    30 seconds before impact – the crew became aware that the speed was 10-15kt below the TAS and cycled the thrust levers.
    20 seconds before impact – the crew are now fully aware that they a flying a glider and raised flap from 30 – 25.
    10 seconds before impact – stick shaker operates 4kts above the stalling speed.

    With the knowledge of hind-sight can anything be done to assist a future crew to do more?
    Undoubtedly the non-disconnection of the autopilot had significantly unhelpful consequences. The autopilot’s attempts to stay on the glide slope allowed the speed to decay to 108 knots. That did not maximize the gliding distance and recovery from stick shake speed in an aircraft at 170 feet and without power made a heavy landing (2.9g) inevitable. Thus, the aerodynamics say that a better outcome was possible but it would have required that the crew had an early and unequivocal indication of thrust loss and had been able to react promptly to an event which they had anticipated and rehearsed. The AAIB report notes that this was an unexpected and unrehearsed situation but made no recommendations on that aspect of the problem. The Rejected Take Off close to V1 scenario indicates that crews can react with a co-ordinate sequence of responses within a few seconds of an anticipated problem emerging.

    Sitting in an economy class seat one hears the engines spool up but probably not in the cockpit. Since the thrust increase was to maintain airspeed, not increase it, no sensation of acceleration would be expected. Is it not possible to devise a method of advising the crew of a discrepancy between commanded thrust and actual thrust? Such a device might have given the crew an extra 20 seconds of realization that they were in a glider ie 40 seconds before impact not 20.

    At 40 seconds before impact G-YMMM was at just departing its TAS at 500 feet with 2500 metres to go to the threshold. At 135 kt 2500 metres would have taken 37 seconds – three seconds less than it did stay in the air. It would have needed to achieve a rate of descent no greater than 800 feet/min. In the 124 seconds from the start of the record in fig 20 the aircraft descended 1350 feet a rate of decent of 650 feet/min.(There are some uncertainties since I do not know if the small amounts of thrust made significant differences to this number) But it seems that with that in these circumstances G-YMMM might just have made it to the threshold and with enough speed to flare. But the crew needed both an early warning of the loss of thrust and a rehearsed mental check list that reads 1. In the absence of better guidance, maintain speed at TAS. 2. Reduce drag if possible.

    I am told that all jet transports can achieve a glide angle of better than (i.e. less than) 3 degrees in the clean configuration. Thus if you are on the glide slope you can, in the absence of thrust, clean up and accelerate and still reach the runway providing you are at a point sufficiently far out that you can afford the short-term loss of altitude involved in the clean-up/acceleration. Perhaps power should always be applied before reaching that point. As G-YMMM demonstrated, leaving it until 600 feet means you are considerably constrained by the shortage of both time and height. No doubt there are some disadvantages (noise and fuel consumption would be obvious ones and David Connolly will no doubt tell me the others) but the principle of confirming that thrust is available whilst you still have viable options in the event that thrust is not available, is attractive.

    I would be interested in answers to the following;

    1. There must be some pilots, who since January 2008, have flown the G-YMMM scenario on a simulator. What do they have to say?

    2. Are any airlines modifying their descent procedures so that failure of the engines to spool up on final approach is planned-for event?

    I am reminded again of the words of Steve “The problem which gets you is the one you have not anticipated.”