New account of Flight 447 disaster published

Lack of manual flying experience contributed to the crash of a fully functional commercial airliner two years ago, killing all 228 people aboard. Air France Flight 447 crashed while flying through an Atlantic storm in July 2009, the worst ever French aviation accident. The black box recorders were not recovered for almost two …

[Steve the Cynic]

slang

Yes, indeed, the transcript avoids giving the accurate literal translation of 'putain' ('whore'), but in that situation, he isn't saying that. It's just an expression of alarm / frustration / etc., sort of like saying "damn" or "bloody hell". And it has an excellent mouth-feel for expressing this, as living in France for any length of time will tell you. (I'm racing towards three years, and yes, I say it too, sometimes. It occasionally is heard in the enhanced version, "putain de merde".)

Similarly, "bordel" isn't translated at all. The observant will note its similarity to "bordello", and it does indeed mean nothing more and nothing less than "brothel" when used in a technical sense. When used in casual speech, however, it means something similar to "tangled mess" - "dans ce bordel de merde" (lit: "in this brothel of shit") is sometimes heard.

[Robin]

re: slang

"It occasionally is heard in the enhanced version, "putain de merde".)"

Works in a similar way in Spanish, "puta mierda".

[Chris Miller]

Eventuellement = possibly, not eventually

It's a notorious 'false friend' (words in two languages that sound similar but have different meanings), so:

02:08:03 (Robert) Tu peux éventuellement le tirer un peu à gauche. = You might need to go left a bit.

Apart from that, it's all very sad. The icing on a couple of airspeed probes caused experienced pilots to drop an otherwise perfectly servicable plane into the drink with the loss of over 200 lives. It raises issues with the automation software and the pilot training - which in this case combined to cause an accident.

[Anonymous Coward]

Not sure you can really blame the software, it detected an anomaly from the sensors which it couldn't handle, so did what it was programmed to do - yelled for help, i.e. turned control over to the supposedly-experienced pilots, who then ignored the information the plane was giving them while *they* screamed for help. When the captain returned to the cockpit and eventually realised what the situation was, it was beyond recovery. All in all, not unlike the inexperienced driver who brakes on snow, feels the car start to slide, and desperately stands on the brakes even harder trying to make the car stop, but sadly with far more serious consequences. One can only hope that, despite their public attitude, Air France do learn something about pilot training from this.

[Chris Miller]

You don't just 'blame the pilots' or 'blame the software'. Accidents in modern aircraft rarely have a single cause. And to say 'the pilot was insane' doesn't help much in preventing reoccurrences, which is the object of the exercise.

There were problems with (at least the UI of) the software. One particular instance was the fact that the stall warning was inhibited once the aircraft was fully stalled (essentially zero airspeed) - this is done to prevent erroneous warnings. From that point on, even when the Pitots unfroze and were working correctly, when the pilots tried to do the right thing (lower the nose) the airspeed exceeded the inhibition limit (but still far below stall speed) and the stall warning returned - a horrible trap.

[Aaron Em]

"A horrible trap"

Sure, but only if you've so badly stalled the aircraft to begin with that the pitot tubes aren't getting enough airflow to provide reliable AOA indication -- I gather that's what shut off the stall horn, rather than the low airspeed. Hard to call this a UI failure, I think, or solely a UI failure at least, when it didn't happen until the pilot had got the aircraft's attitude incredibly screwed up to begin with.

[Tom 13]

While it is true that you can't blame the software

for essentially the same reason you can't blame the girder that failed and collapsed the building, you can and ought to blame the software in the sense that if the pilot isn't getting regular practice and isn't aware of the circumstance causing the emergency, he isn't going to be able to pull out of it. We've got the same problem over here in the states with passenger trains that are mostly run on autopilot. When they dump control back to the fleshy in a red alert emergency situation, the fleshy has no idea what the correct procedure is and lots of lives can be lost.

[Mondo the Magnificent]

Fly by wire?

The Popular Mechanics article makes for interesting reading indeed, especially how the two "joysticks" on the Airbus are not stynchroised, thus allowing the two co-piots at the controls to totally contradict each others actions

As sad as the event was and still is, the information gained from the flight data recovery will ensure this doesn't happen again.

[miknik]

All the data recorder reveals

is a school boy error.

Stalling is exercise 10b of the JAR syllabus for a private pilot's licence and covers symptoms of the approaching stall, the stall itself and standard stall recovery (which is the same in any plane - full power, lower the nose)

Looking through my logbook I did this exercise after 6 hours of ever flying a plane. The fact one of the first officers has not only gained his PPL, but also then a CPL and then an ATPL and then type approval for the A330 yet managed to forget standard stall recovery is a worrying one. Even flying without a functioning ASI is something any competent pilot should be able to manage, I'm surprised there wasn't a checklist to follow in such a situation.

I'd rather walk than fly Air France

[Gordon 10]

Really?

You don't know enough about what the pilots were experiencing to make the call you are making. You are massively oversimplifying the situation.

The fly by wire in alternate mode, the stall warning shutting off, the averaging of stick inputs, the complete disorientation the pilots were experiencing all added to the tragedy.

There was a significant chain of causality.

You are cheapening what really happened and I think you need to STFU.

[miknik]

@Gordon 10

You don't know enough about aviation. You shouldn't fly into storms (other pilots in the area at the time avoided this one). There will be standard procedures for instrument failure wirh a specific set of instructions to follow in such an eventuality. Loss of the ASI shouldn't result in crashing into the sea. You still have ground speed from the gps and an attitude indicator so you know if you are nose high or not, you have wind information from forecasts and from during the flight before the ASI failed so coupled with ground speed you have a reasonable idea what your airspeed is. The stall warning is independent of the ASI so no reason to ignore it when it sounds 75 times. If the nose is elevated 18 degrees, the stall warner is blaring and your groundspeed is around 100 knots in an A330 and you don't reduce the angle of attack then bad things are going to happen.

Go to your local flying club and ask anyone with more than a few hours flight experience and they will all tell you the junior co pilot basically screwed up and killed everyone. That's certainly the consensus at mine.

[Gordon 10]

I didn't say that it wasn't pilot error yes Bonin was the single main cause however there were significant contributing factors and blaming it on the training only was a cheap shot and you know it.

And as for your friends the only ones worth listening to would be those who have type certs on the a330/40.

Until then I'll rely on my own opinion based on 12 years aviation experience including working directly with the engineers who take 330's to pieces on a regular basis and an Airline who run some of the highest duty cycle 330's in the world since the model was brand new.

[Tchou]

miknik..

The pilot made the error in combination with other factors.

The certification is not in cause.

If you were to rely only on companies that NEVER had crashes, you'll sure not fly often.

And certainly not on American ones.

[Glenn 6]

Scary fail for France's licensing standards?

Active stall recovery, much like with you, was done frequently and very early on when I was training for my license. In Canada where I'm from, so are recovering from spiral dives. Its been a while since flight school however I seem to recall my instructor telling me that the actively (meaning really doing it while flying a real airplane) recovering from stalls and spiral dives is NOT manditory in every country. Some countries, such as the US, just cover it in the classroom.

So either this Bonin guy (whom in my opinion single-handedly crashed that aircraft and killed all those people) was a complete moron who somehow ended up as a copilot on a commercial aircraft, or France's pilot training standards are severly lacking.

[Chris Miller]

@Glenn

Swept-wing jet airliners are not like the Cessna on which you probably learned to fly. If you stall them (which there are multiple safeguards to prevent happening) the flight path can soon become irrecoverable. It's quite likely that, once AF447 was deep in the stall regime the pilots were merely front-seat observers of the unfolding accident. We'll probably never know whether the world's greatest pilot could have recovered the plane - no-one's ever going to recreate the conditions for real, and they are way outside the limits of what any simulator is certified to replicate.

Recommended reading: Handling the Big Jets: An Explanation of the Significant Difference in Flying Qualities Between Jet Transport Aeroplanes and Piston Engined Transport Aeroplanes

(David P. Davies, 1973)

[John 62]

@Chris Miller

You're much better read than me, but the Popular Mechanics article said one of the expert pilots they interviewed for the story had subsequently flown that situation in a simulator and had got out of it.

[Steve the Cynic]

Exiting stall...

Actually, some planes don't respond easily to that method of exiting a stall.

The phenomenon is called "deep stall", and mainly affects two types of planes. The first type is anything with a T-tail, where the horizontal stabilisers end up in the "dirty" airflow downstream of the stalled wings, and therefore you lose pitch authority. The second type is the F-16, whose flight computers are notorious for obstructing pitch-down input during a stall.

None of that, of course, excuses the pilots of that plane.

[Anonymous Coward]

"especially how the two "joysticks" on the Airbus are not stynchroised, thus allowing the two co-piots at the controls to totally contradict each others actions"

I literally felt sick when I read that. I knew some of the guys that wrote the A320 software and have tried to avoid the aircraft ever since as a result, but someone should be in the dock for that piece of shit design.

[ColinP]

"Computer rashly let meatsacks take over"

Is this really the sort of story for an attempt at humourous Reg-isms?

[Aaron Em]

Is there any other sort of story?

Or are we going to get a bunch of whining about respect for the dead, &c.? I hope not; one of the reasons I read .co.uk rather than .com is because so few of my countrymen do likewise, and there's really nothing more American than the kind of complaint it sounds like you're making.

[asdf]

please

Remember Bill Hicks and George Carlin were both American were both as irreverent as any Ricky Gervais(sp?). Don't let our very loud very small minority of bible thumping right wingers give you the wrong impression.

[asdf]

300+ million people in USA

but the opinionated limey knows us all.

[Aaron Em]

"The opinionated limey"

was born and raised in Mississippi, thank you, and my experience tells me that progressives are at least as susceptible as traditionalists to this kind of sanctimonious moralizing.

[asdf]

fyi Jackwaggon

Colin is much more likely a UK name than an American one.

[Anonymous Coward]

ll

If you say "Colin" to a Merkin in my corner of Merkinland they'll spell it with two l's.

[asdf]

nothing more European

than killing 200 people due to the European love of bureaucracy. Nobody could be bothered to take ownership of preventing such an obviously preventable tragedy.

[Anonymous Coward]

@ColinP

To be fair, an *actual* sack of meat may have done a better job in the cockpit than this hapless trio of cheese-eating surrender monkeys!

[Anonymous Coward]

Bad stall training in the airline industry

The issue that is grabbing most attention is that pilots are being taught to pull the stick back in a stall in order to retain height. This seems to be fundamentally wrong and a far bigger cause of the accident that this article implies.

[Aaron Em]

Not according to Popular Mechanics

The linked article clearly states that pilots are trained to drop the nose in the event of a stall, and indeed anyone who's spent more than fifteen minutes screwing around in Microsoft Flight Simulator, or even just looking at the Wikipedia page on aerodynamic stalls, can figure that one out for himself.

For that matter, unless France's pilot certification requirements are a joke -- which I doubt -- everyone on that flight deck would have had to demonstrate ability to recover a stall before being allowed anywhere near a revenue flight.

[Dirk Vandenheuvel]

True...

It is very weird they continued to pull up. You natural instinct in a stall is even to go down and increase speed, your body can usually feel a stall. All very strange. Then again I have never experienced a stall in a big plane like that, maybe you don't feel it.

[rjmx]

Loss of height

I've read that some of that is due to some (FAA?) requirement that stall recovery should involve a specified maximum loss of height -- to the point where some pilots undergoing checks got reluctant to push the stick too far forward, lest they exceed that maximum. True, stall recovery should involve losing as little height as possible, but specifying a maximum height loss gets counter-productive.

[Peter2]

It probably depends what your flying

It your flying a single engine prop powered light aircraft, if you stall then you probably do need to nose down as you don't have the power/weight ratio to simply power your way out of a stall.

I don't know, but I suspect that 4 jet turbofans at full throttle probably produces so much thrust that the benefit of going nose down is marginal, since I suspect that a packed airliner has a gliding profile similar to a brick. If that is the case, then diving wouldn't help much in the first place, and would put you nose down when 4 jet turbofans start pushing you along the line of flight which might take you into the ground in the time it took the plane to recover and get nose up if on approach.

Since I doubt those people writing the airline safety documentation are idiots, you have to suspect that someone has sat down and calculated that it's simply safer to keep nose up and go full throttle to recover.

Still, i'm sure your better qualified to comment than me or the experts!

[Jim Morrow]

> I suspect that a packed airliner has a gliding profile similar to a brick

wrong. the typical jet performs remarkably well as a glider considering it has 2 or 4 big engines dangling from its wings. which mean the aerodynamics can't be anywhere near as good as they'd be on something that was actually designed to be a glider. despite this, jets of today tend to have glide ratios of 18:1 or theresabouts. ie for each foot of altitude lost, the plane goes forward 18 feet. even a shitty 50 year old one-prop cessna has a glide ratio of 12 or 14 to one.

the idea that any plane immediately falls from the sky when the power is taken off is just silly. remember the airbus that ditched in the hudson after it lost both engines at take off?

btw take a wild guess what the engines of a jet are doing in the last 20 minutes or so of a flight when the plane's descending to land.

[Michael Jennings]

Gimli Glider

The "Gilmli Glider" incident in Canada in 1983 is also worth mentioning. A Boeing 767 ran out of fuel mid-flight (an imperial/metric mixup combined with a faulty fuel gauge) and the pilots managed to successfully land the plane with only minor damage. The minor damage was due to the nose wheel not being fully down due to the power loss and due the the brakes having to bit hit very hard due to coming down a bit too late, rather than the plane doing anything brick-like. Everyone on board the plane was completely okay and the aircraft was back in service soon afterwards.

http://en.wikipedia.org/wiki/Gimli_Glider

There are not many incidents like this, because multiple engine failures are incredibly rare. You have to run out of fuel, have simultaneous bird strikes, fly through a cloud of volcanic ash, or something like that. Independent mechanic failures due to faulty engines simply do not happen, as modern engines are incredibly reliable.

[AndrewG]

"I don't know, but I suspect that 4 jet turbofans at full throttle probably produces so much thrust that the benefit of going nose down is marginal"

Errr you'd be wrong in your suspicion I'm afraid. The biggest problems with any jet engine is they have to spool up, gravity's instantly responsive. Also the aim of a stall recovery is to get the air flowing over the wings the right way, a stall means your descending so pointing the wings in the direction your travelling makes the recovery one hell of a lot quicker.

And according to the transcript, it looks like the people driving sort of forgot to pump the throttles for a while and then when they did they tried to go for a nose up max climb attitude amore appropriate to sea level.

Theres so many basic mistakes and areas for improvement in that reconstruction its scary. Add that to the later QANTAS A380 engine debacle where the crew had to spend 30 mins clearing alerts to understand what was wrong and you've got to wonder if there are some seriously wrong UI decisions in Airbus

Who said this didn't have an IT angle?

[laird cummings]

@AndrewG

"And according to the transcript, it looks like the people driving sort of forgot to pump the throttles for a while and then when they did they tried to go for a nose up max climb attitude amore appropriate to sea level."

Just so.

Basically, a behind-the-power-curve maneuver. IOW - Not enough thrust available at that point to do what they wanted. That late into the stall, even with throttles to the firewall, the engines would not have been able to power the aircraft out of the 'stall' portion of the envelope without a healthy assist from gravity.

My father used to do incident investigation, and his opinions on the cockpit discipline and coordination in this case is unprintably profane.

[Tom 38]

putain

While the literal translation is 'whore', your typical frenchy will use it almost as an all purpose swear word - think of it as a french version of 'bloody', and about as naughty.

[Uncle Slacky]

Poor old Vladimir

Because of the similarity, French newsreaders take care to refer to him as "Poutine" (pronounced "poot-een") rather than what it really sounds like.

[I_am_Chris]

Not quite.'putain' is much stronger than 'bloody'; it's much nearer to fuck or shit. There's no way you'd see anyone on prime time tv use that word unlike 'bloody' over here.

[Steve the Cynic]

Wrong.

French TV does have a "watershed" time, but it's not immediately clear to me (a) what time it is and (b) what's allowed before versus after. Certainly I've heard both 'putain' and 'merde' on 101%, a thirty-minute magazine show that airs daily at 7pm on Nolife.

That said, the French have odd standards on age ratings. /Bring Me The Head Of Alfredo Garcia/ is rated 18 in the UK, but the French DVD of it I just received has a 12+ rating (not allowed for under-12s).

[Arthur Dent]

@I_am_Chris

I think you are wrong - putain is not very strong. I don't know about prime time TV, but I remember hearing two of Brassen's songs that used it (it occurs in the chorus of "Putain de toi" and in the last verse of "La complainte des filles de joie") about half a century ago on prime time radio in France (while his "Fernande" was banned from radio because the chorus contained "je bande" and "la bandaison papa ça n'se commande pas" - that use of bander/bandaison was thought to be a bit too much). Best translation I can think of for putain in a phrase like "putain de toi" is "you tramp", or for "putain" on its own "oh damn". The dictionary I use on the rare occassions when I need a French disctionary gives it as "putain: exclamation exprimant la surprise; (grossièrement) prostituée" so I guess its use as an exclamation is not regarded as grossière.

[Michael Nidd]

I understand that one stick doesn't move when the other does, like in a Boeing, so they don't feel the actions of the other pilot; but why isn't there some sort of alarm when both are being moved at the same time, so they at least know about the conflict?

[Voland's right hand]

Cockpit is no place for democracy

"Averaging" between inputs is a terminally bad idea. Better to have priority in cases like this and the captain being able to flip it if need be from the back seat.

[TeeCee]

That's why every pilot knows (or ruddy well should) that "You've got her" (or similar) means you are to take control and that "I've got her" is the correct response, indicating he should now keep his paws off. In the event of an emergency where a pilot thinks he should take control unilaterally, he should announce the fact and ensure that an appropriate response is received (assuming his counterpart is conscious).

If both pilots are attempting to fly the same plane at the same time, there's been a catastrophic failure of basic procedures somewhere along the line. This sort of thing is usually referred to as "pilot error" in the inevitable post mortem.

There is one exception here and that is when for some reason (e.g. hydraulic failure) the pilot flying the plane is unable to move the control surfaces, at which time he may call for the bloke in the second seat to assist. Applying extra "grunt" in this manner is utterly pointless on a fly-by-wire aircraft and should never be used in such.

Thus if lack of synchronisation becomes an issue, the lads at the controls have fucked up. Big time. Fortunately the aviation industry isn't above telling pilots what they should damned well already know, so I'd expect such an alarm to be amongst the recommendations of the inquiry.

[JimC]

From what I can read elsewhere

that summary goes a lot further than the official enquiry has done so far. It seems plausible, but I wouldn't guarantee that the official report will be identical.

[itzman]

That is probably something that will be considered REALLY carefully from now on.

Its a classic 'we never thought of that one' situation ..

[Intractable Potsherd]

Multiple stick input on FbW planes ...

... is surely redundant, and therefore shouldn't be allowed to happen. Is there any reason why there shouldn't be only one stick live at a time (subject to the relevant override controls being suitably distributed)?

[Phil O'Sophical]

Wonder if this will get reported in France?

It's interesting (if that's the right word) that as soon as the first indication of pilot error came out, Air France responded "couldn't be, not OUR pilots", and the French mainstream press stopped reporting anything about it.

[Destroy All Monsters]

Well, *some* parts of the US press where into "Airbus plastic material failure / fly-by-wire-too-much-automatics failure, couldn't happen to Boeing oh no"

I was disgust.

[Anonymous Coward]

Airbus Crazy

I'm disturbed to know that Airbus has two modes "normal" and "alternate" law. And that pilots honestly believe they can't exceed the safe envelope in "normal law".

One thing is for certain: both of these pilots were unaware of "alternate law". I wonder if airbus cockpits had a clear indication of which mode was in effect?

And dual input controls? What the hell? One pilot must be in charge at any one time. The idea that the system "averages out" both controls is ludicruous! And resulted in death in this situation.

The report makes for terrible reading both for Air France pilots but more disturbingly Airbus!

Blame the French!

[peter 45]

No training in alternate

I am gobsmacked that they did not have any training in this alternate law control. it is beyond my ken that there was nothing in the extensive training package that these guys have to go through, that there was no simulated instrument failure that switched to alternate control.

The 'averaging' joystick is beyond weird. Who on earth thought it was a good idea*? At the very least an 'I have control' button should should have been mandated.

Probably the same person who wrote the software for a well known British Military Aircraft to display position data from a dual Inertial Navigation System (two gyros) and decided to just average the two. Seemed to work fine even when one IN drifted its heading by one degree west, and the other one degree east.........until the pilot turned exactly north and the system averaged the heading to exactly 180 degrees!

[Gary B.]

Actually, there are three modes: "normal", "alternate", and "direct". I forget the specifics of how the displays work, but there is an indication when the aircraft is no longer is "normal law", which yes, does prevent the pilots from exceeding the normal flight envelope. Alternate and direct law provide increasingly less envelope protection, with "direct" providing nearly zero protection.

The averaging of the two controls is dumb, though. Pilot A: Left! Pilot B: No, right! Computer: Ok, straight!

[Chris Miller]

It's worse than that, Jim

There's at least one more: 'direct' law. In normal law (which is all you'd ever expect to see, unless there had been multiple failures in redundant systems - pilots only encounter 'alternate' law in the sim) it is (almost) impossible to exceed the safe envelope. So heaving back on the stick (while bad practice and utterly alien to anyone who'd learnt to fly on, say, a Cessna) is normally quite safe - but in this case it was fatal. The change from normal to alternate law is shown on the flight computer, but there were a lot of simultaneous announcements when the problem occurred - before the recovery of the black boxes, the only data was from the in flight transmission of engineering data which showed about 80 (IIRC) warnings in a few seconds.

You're quite right that only one pilot should be handling the aircraft at any one time - which is why the behaviour of the sidestick controls is not normally a problem. But this does highlight the lack of tactile feedback for the non-handling pilot, which those brought up on Boeings tend to dislike.

Despite all this, it should have been a non-event. All pilots are supposed to have memorised the appropriate settings of power and aircraft pitch to be used in the event of failure of the airspeed indication, which would have allowed AF447 to simply fly through the cloud until the pitots unfroze and all returned to normal. No-one has come up with a very satisfactory explanation of why the handling pilot reacted as he did.

[Aaron Em]

Pilots *can't* exceed the flight envelope under the normal law; that's what it is for. (The autopilot can't exceed the envelope either; if it tries under normal law, it's automatically disengaged and a warning horn sounds.)

Alternate law exists for cases in which the ship's computer system isn't getting enough information to be certain it can fly the aircraft safely -- as, for example, in the case of pitot tube icing, which was what happened here. It's the computer's way of telling the pilot "look, I'm not sure I know what's going on correctly, so you'd better take over."

I'm not sure whether there's a clear indication in the cockpit of when the flight control law changes, but it would very much surprise me if there weren't; after all, switching off part or all of the fly-by-wire system is going to significantly change the aircraft's behavior, wouldn't you say?

But I *am* sure that Airbus doesn't deserve blame for the crash, not least because Boeing's fly-by-wire aircraft use a very similar arrangement for graceful degradation of that system in case of malfunction or untrustworthy data inputs. Judging by the transcript, the fault clearly belongs to Bonin, who either wasn't properly trained or who failed to keep his head in the situation, at the cost of over two hundred lives including his own.

[SkippyBing]

It actually has more than two modes, depending on what systems are available. In 'normal' the computers shouldn't allow you to conduct manoeuvres that will depart from controlled flight, e.g. it won't let you stall, if you pull fully back on the stick the computer will only let the airframe achieve the AoA for maximum lift no more. Effectively the pilots are allowed freedom of operation inside the safe flight envelope, the computers prevent the aircraft going beyond that.

In 'alternate' mode, recognising that some systems have failed and that it's unable to provide full flight envelope protection the aircraft reacts in a more traditional way, obviously it's important that the pilots recognise this.

Dual control inputs, yes, otherwise it's a bit pointless having two pilots, I did have the whole thing explained to me a while back, I can't remember the exact details but I'm fairly sure there's an over ride so one of the controls can have primacy.

[itzman]

That is in fact unfair.

Of COURSE a plane out of autopilot is in a different regime and I am sure they were aware of what not being in normal law meant - theoretically. Its probably in the manual somewhere.

The fact is they were not awarer - or one of them was not aware - that it applied at this point, and they seem to have displayed no sense of being trained in its use.

there is a little of blame attached to airbus..the sticks not being tied is one..but its mostly a terrible lack of training and a terrible casualness when faced with a situation outside the normal..as if they relied on the plane to fly itself out of trouble when it had told them it couldn't.

And that's fair and square in the court of ALL modern pilots who simply don't fly seat of the pants planes enough - if at all - to understand what the instruments put together are telling them.

Or at least fly enough emergencies in the simulators to effectively give them that experience.

[Jim Hague]

Re: Airbus Crazy

Based on a play last night on an official A320 sim (full motion an' all, owned by Czech Airlines), I can tell you that yes, you can move both input controls at the same time.

And if you do, the warning voice bawls 'dual input' at you. Repeatedly. Which I submit is a respectful request to the overlord meatsacks to bloody sort themselves out.

We'll have to wait for a real Airbus jockey to come by and tell us if the mode is apparent. Based on the response to an engine fire on approach, I suspect that yes, it does let you know. And probably advises on what to do about it.

As an aside, the instructor took us for a most entertaining low-level pass by the Eiffel Tower. Said warning voice was fully occupied yelling 'Terrain', 'Terrain, 'Pull up', 'Terrain'. The sim was colossal fun. I want one.

[William Boyle]

This is why...

This is why pilot-less commercial aircraft will never "fly". When the computer and instruments are snafu, the only thing keeping people alive are the skills of a human controller (pilot). Unfortunately in this case, the pilots did not have the requisite skills and experience to deal with the situation, unlike the pilot that landed that plane on the Hudson River a few years ago, saving all the passengers and many more on the ground after a serious bird hit. See http://en.wikipedia.org/wiki/US_Airways_Flight_1549 for details about that miraculous event.

[Tony Humphreys]

or the captail who got a knobbled 777 inside the Heathrow boundries, and actually onto the airfield. He was rewarded for his skill in saving a couple of hundred onboard, and countless on the busy London streets by being turned down from jobs because of an accident on his record.

Thats the state of pilot training these days.

[Chris Miller]

@Tony

The good news is that Capt Burkill got his job back at BA and is back flying 777s. Highly recommend his book "Thirty Seconds to Impact".

[RIBrsiq]

The autopilot gives control of the plane to its pilots because it realizes it might be in over its head. Those pilots then proceed to literally fly the plane into the sea. And you conclude the right course of action is to *retain* the human pilots...?

[Lars]

@William Boyle

True, but then again the number of accidents due to the human error is incredible large, as large as technical errors or even bigger.

Trying to aid pilots bye introducing automation is good and inevitable. Just look at what is introduced into modern cars.

"the pilots did not have the requisite skills and experience to deal with the situation"

Right, then again nor did the computer system.

Funny, is not a funny word, but still how can a "plane" be so completely unaware of its altitude, after all, how much does a reverse beep on a car cost.

[Anonymous Coward]

Uhh come again?

Humans took over a plane and crashed it, and that's your reason why humans should be able to manually take control of plans?

I'm not disagreeing but this article dosen't seem to help your case.

[ZweiBlumen]

nice Fairytale...

but not the truth. It was basically the avionic on this Airbus that landed the plane safely. Which is not to discredit the pilot for acting appropriately, but this landing was as good an advert for fly by wire as there will ever be. It's just that people need a real hero, like a pilot, not some boring, faceless, aerospace engineering team.

[Anonymous Coward]

But

I read in another source if the "Pilots" had switch the Autopilot back on when the pitot tubes started working the computer would have sorted it out and had the plane flying true in seconds.

Airbus needs to work on their Autopilots work- there was another crash in Russia of an Airbus when some pilot let his kid fly the jet (commercial) and the kid "partially" disengaged the Autopilot and there was no warning. Plane impacted the ground at a high rate of speed. Ground won.

I wonder how good the Autopilot is on the A380 .................

[Martin Gregorie]

and also....

... the LOT pilot who did a superb job of putting a 767 down at Warsaw when its undercarriage failed to come down.

http://airpigz.com/blog/2011/11/2/2-videos-lot-767-lands-wheels-up-in-warsaw-poland-11-1-11.html

The second video, taken from outside the airfield, catches the touchdown perfectly. The pilot put it right on the threshold with the exact amount of flare to grease it on. Its hard to imagine how it could have been done better.

I hear that, like Sully, he is a glider pilot...

[Anonymous Coward]

Not miraculous, just down to damn good flying.

[Jim Morrow]

the airbus's avionics had as much to do with the successful outcome of us1549 as the undoubted skill of the pilots. the avionics "flew" the plane -- keeping the optimal angle of attack until the last few seconds -- while the pilots figured out where and how they'd get the plane down as well as run through the emergency ditch and shutdown procedures in much less time than these procedures need. if the pilots had to control the plane all that time as well, there was a good chance something would have been overlooked in the tiny amount of time they had to do anything. so we'd have been talking about pilot error being a contributing factor in the loss of life from that one-in-a-million freak event.

[Anonymous Coward]

Fear of flying

Good job I dont have a fear of flying.

Or at least I didnt until now.

[Yet Another Anonymous coward]

@This is why...

That would be the pilot of the airbus whose computers automatically kept the aircraft on course at a proper descent rate and speed - even when the pilot let the speed fall below the limit which would have caused a stall.

The NTSB report details the role of the avionics system and credits the captain with starting the APU which allowed more of the system to continue functioning that minimum manual controls available from the RAT.

[Anonymous Coward]

The transcription of the flight...

...is one of the most disturbing things I've ever read.

[Destroy All Monsters]

Then you haven't read the details on KAL 007.

Due to epic pilot error (which includes somehow not groking that if the sun doesn't come up at the expect time you are probably NOT at the position you think you are), leisure fly over Soviet territory where stimmed youth in fast planes then shoot you down.

That was before GPS though.

[Stratman]

I know it's not really helpful here, but way back when I was gaining my PPL, it was ingrained into us to "Believe the instruments". Maybe they themselves were 'going on autopilot' and their training was kicking in.

[JeffyPooh]

"...the crash of a fully functional commercial airliner..."

That's S.O.P for Airbus. Airbus aircraft are typically in perfect aerodynamic, mechnical, and electrical* condition one millisecond before impact. [* Software and UI? - No comment...]

Boeing aircraft are typically fatally damaged long before they hit the ground.

There are exceptions of course, but this is (IMHO) a valid observation.

[Destroy All Monsters]

Except for that Iranian Airbus which had a bad encounter with a US missile?

[Michael Chester]

I think that comes under "impact"

[JeffyPooh]

"...the crash of a fully functional commercial airliner..." Part 2

Okay then, let me be even more emphatic about the distinction between Airbus and Boeing:

Observation 1: *Most* (significantly more than 50%) Airbus major crashes involve aircraft that are essentially in good mechanical condition in the millisecond before impact.

Observation 2: *Most* (significantly more than 50%) Boeing major crashes involve aircraft that are already significantly damaged, mechanically, long before they hit the ground.

Based on everything I've seen, I strongly believe that the two observations above are in fact true. I'm not saying one cause of disaster is better than the other. But the root causes appear to have an obvious divergence. Not "all", just 'most'. Downvote me all you want. Facts is facts.

Flight 447 is just the latest in a long sequence of famous Airbus CFIT incidents.

[Anonymous Coward]

I don't see any facts

Citations please? Rather than your "gut".

[Ragarath]

Logging in the cockpit

We can all see from the transcript what went wrong and that inexperience caused the crash. It is very sad and had the captain realised earlier what was happening then it may have been different.

my question is, is there any sort of logging going on in the cockpit, on a screen that an absent member can then briefly cast their eyes over?

Yes the instrumentation is there to tell them what is happening now but if there had been a simple visual system that told the returning captain what had been happening then it could have been averted.

For example:

1) Autopilot OFF

2) Stall warning issued

3) response from the pilot e.g. Stick pulled back X Degrees for X amount of time.

4) Speed and angles etc. from working instruments.

All of that could be in a visual interface so that it takes seconds to look over. The captain would have been able then to tell the co-pilots what to as soon as he arrived.

Does anyone know if such a system exists on these aeroplanes?

It's all well and good logging into a black box for detailed examination after the problem but it does not help those that had the problem.

Sad face for the sad circumstances.

[Anonymous Coward]

HAL says ...

"This sort of thing has cropped up before, and it has always been attributable to human error ..."

[rurwin]

This is plainly and self-evidently not the fault of the pilots.

If a co-pilot could panic to such a degree that he forgot everything he was taught, and disregarded everything the plane was telling him then he should not have been on the flight-deck. That would have been a fault of Air-France.

If the plane did not tell the pilots it was in "Alternate Law", they should not be blamed for not knowing it was.

If that truely is the way the jopysticks work, then whoever designed the control system is guilty of gross negligence. If the pilots command opposite control movements, it is a clear indication of pilot failure. At the very least it should have triggered an alarm. In fact there should be a switch to select the active joystick. Or is the Airbus designed to be flown by committee?

If the plane had failed safe and given adequate indication, then one panicing co-pilot would have been quickly and effectively cut out of the control loop and everyone would have survived.

[Intractable Potsherd]

Pilots ...

... shouldn't be panicking. A panicking pilot is as much use as a panicking surgeon, and any tendency towards it should be checked for, vigorously. Any pilot with that tendency should not be allowed anywhere near the pointy-end of a passenger flight (at the very least). Therefore, this is a pilot failure, and an airline failure, exacerbated by some very difficult-to-understand aspects of the control system, most of which you identify.

As a fan of Airbus and AirFrance, I am extremely concerned about this, and will be considering very carefully which airlines I fly on for the near future, until I hear that both organisations have taken proper ownership of the clear failings, and taken appropriate steps to minimise the chances of them happening again.

[Glesga Snapper]

Stick priority.

If both side sticks are moved at the same time an aural warning "dual input" is given by the aircraft.

In addition to this, there is also a visual warning light on the glare shield, just below the pilots' eye line, that illuminates to indicate that the other pilot is making control inputs.

Pressing the red button on the side stick gives that stick priority and cancels out the inputs from the other stick.

[asdf]

no expert but

You would think will the mass lay offs and tight economy the pilots left in the industry would be the best of the best perhaps even with combat experience. Alas once again like in most other industries it seems sociopathic middle managment instead keeps the cheapest talent not the best around.

[Destroy All Monsters]

So how is Captain America on the joystick going to help here?

And if you think european commercial airline pilots are cheap, well...

[laird cummings]

@Destroy All Monsters

Combat experience seperates the goats from the sheep - You find out how will panic, and who will not. Plus, having been shot at by people who mean for you to die horribly tends to put other emergencies in perspective, thus improving response in other, lesser emergencies.

[itzman]

well..

the usual compound cockup with exactly the wrong thing done at every decision point that marks most avoidable crashes.

And an indictment of pilots who actually are more bus drivers these days than seat of the pants pilots with an understanding of aerodynamics.

I reckon even I would have known what -2000 fpm, a nose up angle of 30 degrees and an airspeed of 90 meant..

and shoved the stick forward.

[Anonymous Coward]

Don't know much about flying, but...

Wouldn't the GPS have given them speed and approximate altitude sensors to show the plane was climbing and slowing even without the air speed indicators and then the other co-pilot could've realised the problem.

Secondly why didn't the pilots (or the plane offer) to go back to auto-pilot/normal envelope when the speed sensors de-iced and returned to normal?

[b166er]

OK,

I've read that now.

It dumbfounds me that this can even happen.

The aircraft has a GPS system doesn't it? And in the flying game, altitude is what it's all about, right?

Why its the altitude not in BIG FUCKING NUMBERS on a VERY VISIBLE display in the cockpit?

Why, when midflight, does the alarm wait until 2000 feet before sounding?

Finally, why is such a noob allowed anywhere near the controls during a storm?

[Chris Miller]

@b166er

Wow, you should be an aircraft systems designer - no-one ever thought of an altitude display, at least not until 1904. They knew their height and they knew they were descending, what they were (apparently) unable to recognise was that the plane was stalled and no amount of pulling back on the stick and firewalling the engines was going to change that.

[paulc]

First Principles...

They forgot them... if in doubt, fly the plane by pitch and power... ie. set the nose at the correct angle and then set the power so that the aircraft altitude remains steady.

[Anonymous Coward]

I can't believe that both joysticks can give conflicting instructions!

That would seem pretty elementary to me, that there is one person who actually controls where the aircraft is going at any given time.

Also, another good primer on why to avoid thunderstorms, even if it adds time and fuel use.

[Sean Baggaley 1]

@"Airbus Crazy":

Whenever I read lines like "better training required", I mentally translate it into "the user interface sucks".

In an emergency situation, and regardless of the mode of transport involved, the *last* thing users will be able to do is mentally flick through their memories to try and remember some random nugget of information that they've never had to remember until now.

The cockpit's user interface design is at fault here. Fly-by-wire does have many advantages, but it still needs to interface with us fallible humans, so there are limits on what can be removed.

E.g. synchronising joysticks should not be necessary as only one pilot is ever needed on the yoke at a time, but it *should* be a requirement that BOTH pilots are involved in transferring control from one position to the other: it should be impossible for either one to just seize control of the plane without the other pilot even knowing about it. THAT is poor interface design.

Furthermore, there clearly needs to be a more obvious "Plane in Alternate Law Mode" warning, as none of the pilots appeared to be aware of it. Had they been aware of the status, it's possible they'd have paid more attention to that "STALL! WHOOOP!! STALL! WHOOOP!!" warning that was apparently blaring out throughout the crisis period.

This is a design issue. Humans are fallible and it's about damned time engineers stopped assuming everyone who uses their increasingly complex machinery has not only read their 1000-page manuals, but has also memorised them perfectly. Short of requiring pilots all have eidetic memories, there's no way to guarantee that.

Nevertheless, the notion that humans are magical beasts, capable of superhuman feats of physics-defying piloting is bullshit too. Humans require input, just as computers do. Feed us garbage data and we'll shit out garbage reactions to it, just as a computer would. This incident is a clear demonstration of that.

[Steven Roper]

As a web developer

the issue of user interface design comes up for me on a daily basis.

A major problem that any engineer (software, plane design or otherwise) faces, amounts to a direct conflict between customer requirements and ease of use.

That is, a customer tells me, "I want it to do this and this and this and this, and I want it to store that info and that and that and that". Then they complain about complexity and training costs if I present them with a Web form that has 50 input boxes and controls on it for them to set all the parameters and input all the information that they asked for.

There are many ways of solving this problem, but they all have their flaws. I could simplify the form by making assumptions about default settings for certain items and then masking those controls from the user, but this introduces the issue of lack of control if those defaults need to change in a specific case. So then I can create an "easy mode" and "complex mode" to cover this (in the same way Airbus have with their "normal law" and "alternate law" etc modes), but then this introduces the problem of insufficient training of staff to handle the "complex mode" side - since the object of doing this was to reduce training costs and time in the first place.

In the end, the more complex the task you as the end user want the system to accomplish, the exponentially harder it becomes to simplify the inputs required to accomplish that task. Engineers of all stripes are constantly working every day to try to find ways around this problem but we're far from being able to achieve perfection, if ever.

It really does come down to a balance between "How much do want this system to do?" versus "How easy do you want it to be to use?" The two issues are counterpoints by definition, and this finds expression in the famous Murphy's Law corollary, "Build a system that even an idiot can use, and only an idiot will be able to use it."

[Anonymous Coward]

"E.g. synchronising joysticks should not be necessary as only one pilot is ever needed on the yoke at a time, but it *should* be a requirement that BOTH pilots are involved in transferring control from one position to the other: it should be impossible for either one to just seize control of the plane without the other pilot even knowing about it. THAT is poor interface design."

Did you think about this before writing it? Just wondering how your excellent system will cope when the plane is in a crisis situation and the person who was flying the damned thing has passed out/had a heart attack/is otherwise incapacitated and unable to nicely pass on control to his co-pilot? Would you be bleating about the stupid requirement for BOTH pilots being involved in transferring control then? The current design is fine in principle (though not so sure about the averaging of inputs) - it just requires that 2 people talk to each other (which, according to other posters, is standard practice) in order to work out that control is being passed on and to whom.

As for the rest of the rant, do you really think that anyone should be able to just walk in and fly a plane? That it should be so easy that a child could walk in and go "yup, got it!"? Pilots are quite well paid - and rightly so - because they have (or are supposed to have) specialist skills in, and a lot of experience of, how to fly planes, including how to handle emergency scenarios. In fact, they are paid to read and understand those manuals you complain about - it's called "knowing what you're doing" and is quite important in most jobs.

It seems that these days the pilot's job is not much to do with the normal running of the plane - they seem, mostly, to do that pretty well themselves - but to take over when something unexpected happens. The pilots involved here obviously weren't up to that job.

[Denarius]

GIGO

computers are only as good as the input data, and usually worse. Similar pitot failures on Airbusses avoided disaster because the pilots were taught to identify when the input to the computers was wrong and fly the plane by attitude. eg, if airframe undamaged, power == cruise and angle of attack == +4 degrees (guessing) then airspeed _must_ be OK, even if indicated airspeed is 550 knots and rate of climb is shown as 5000 feet/min. The pilots were flying in bad weather so this would have required them to follow the artificial horizon which should be a gyro based device feeding the glass cockpit. At night in cloud windows are useless.

Even in recreational flying a similar over-dependance on computers is becoming evident at the simple level of navigation. GPS makes it so easy, until the battery goes flat. Oldtimers just revert to compass and map, but the younger pilots.... {s}

[Anonymous Coward]

Airbuii have a number of modes of operation: normal, alternate and direct are the major modes. In normal and alternate modes the computers keep the aircraft within certain bounds and in direct mode it flies like any other aircraft. Normal mode has ALL the protections. The modes are commonly referred to as 320,737 and DC-9 respectively and humorously.

Regarding the control sticks - in front of each pilot is a big light stating who has control. In situations where both pilots apply input they are "added" together. One pilot CAN override the other by pressing a switch.

Now here's the killer:

Pilots flying Airbus have training to understand the modes of operation, to find out that the pilots were unaware of "alternate law" is f****** stupid on Air France's part. The equivalent in a car migth be that after a crash you had no understanding of what brakes or steering were. I could understand this is Air France's training was based on a few hours of flying MS Flight Sim but an A330 pilot should have quite a few years of training and experience in other aircraft, the basics of flight etc.

Regarding the sticks, the pilots are also trained in cockpit procedures for this, hence the phrase "I have control" and the corresponding challenge. BASIC absolutely BASIC training.

"Blame the French"...do you read the Daily Mail?

[Pascal Monett]

Interesting

One thing that I take from this article is that Air France apparently finds it normal to put two inexperienced pilots in the driver's seats, with no experienced pilot able to intervene when a mistake is made.

I always thought that airline companies had a Captain with experience at the helm, accompanied by a co-pilot that was learning the ropes, so to speak.

Seems that I was wrong, and budget cuts have led to the inevitable loss of experienced (and more expensive) pilots to tutor the new guys, so just bash two newbies in and pray for the best.

Can't say that inspires confidence in the future of commercial flight.

[Anonymous Coward]

Pitot tubes: voting between 3 tubes of 2 dissimilar designs.

I've deliberately not read the whole PM article but I have searched for the word Pitot.

The reason I did that is that one of the first consequences of AF447 was an airworthiness directive that resulted in the replacement of some Pitot rubes on some aircraft. Read it at

http://edocket.access.gpo.gov/2009/E9-21368.htm

I was under the impression that this had been done because AF447 provided some evidence of a previously believed "impossible" scenario - 2 tubes of identical design misbehaving exactly the same way at exactly the same time (and thus outvoting the one, differently designed, working one).

How did design failure followed by regulatory failure ("the chances of two identical failures at the same time are negligible") magically vanish from the article, and turn into pilot error?

[Aaron Em]

Because

pitot tube icing wasn't what killed the aircraft and everyone aboard -- as another poster pointed out, if they'd just switched the autopilot back on once the iced tubes had cleared, they'd have been home and dry. Go back and actually read the PM transcript this time, and you'll see that they thought of this objection before you did and took a moment to explain why it doesn't relieve the pilots of responsibility for keeping the aircraft in the air.

[david wilson]

@AC

>>"How did design failure followed by regulatory failure ("the chances of two identical failures at the same time are negligible") magically vanish from the article, and turn into pilot error?"

Well, initially, people were understandably looking to see if there were any possible explanations which didn't go for the lazy route and simply assume serious mistakes in the cockpit, explanations which might suggest other aircraft were at risk and where remedial actions might prevent or reduce the risk of future problems.

Then, much later, the flight recorders were found and people found out what had actually happened - temporary loss of speed information, but not something that should have caused a disaster given appropriate crew actions.

It's not at all unlikely that if people are motivated to scratch their heads over a mystery disappearance of an aircraft (or a spacecraft, or a ship or whatever) where data is limited, they might well come up with one or more possible causes, and whether any particular one was partly or wholly to blame for the loss or not makes little difference to whether it's worth talking about all the possible problems, estimating how likely they are and trying to engineer ways round them.

[Anonymous Coward]

so..

the instruments are buggered giving false readings. The controls are not giving the same feedback to different crew members. But it's the pilots' fault that the instruments would not verify ascent or descent? and where actual airspeed was unknown or the readings untrustworthy? How would a pilot verify this during an incident, stick a hand out the window to gauge airspeed?

It's yet another Airbus incident where too many lines of code are between pilot and aircraft. If the aircraft does not respond consistently, it is practically impossible to react "properly" without benefit of hindsight. Especially when it seems any Airbus can react differently for the same external situation, and differently for each pilot. When the actual proper response for one situation can be totally devastating for another, it is imperative for a pilot to know what is actually going on. Any aircraft design that filters or modifies that in variable ways cannot be managed safely when conditions are adverse. Regardless of what 20/20 hindsight, armchair quarterbacking and EU Consortium politics try to sell you.

[Anonymous Coward]

Nope

Other pilots (including AF pilots) have managed to safely get out of "pitot-failure" situations. Pitot icing is a rare situation and requires well-trained pilots to handle it. This is done by using the still-functioning instruments such as the artifical horizon and the altimeter. Standard Operating Procedure in this case calls for maximum thrust and 4% angle of attack, which will bring the A330 in a safe mode of flight by virtue of the aerodynamic design, not by virue of any software.

Airline pilots must be skilled and highly trained professionals; not monkeys pushing buttons. Other pilots did recognize this type of situation and handled it correctly in several cases at AF alone. This is a case of a badly trained pilot team, period.

[LarsG]

Not all flight crew are equal...

Publishes book on the technical aspects of air disasters.

As technology advances it takes responsibility away from the pilots, the computer flies 98% of the route because it is more fuel efficient and cost effective. As a result pilot skills decline.

There are some advocates of pilotless airliners. Computers cannot think outside the box.

Remember, some pilots are naturals and others have to work to their limits to fly, they are not all equals.

I prefer to fly with the naturals, they make it look easy, not the adequate.

A pilot

[Anonymous Coward]

I'd rather have my pilots thinking inside the box...

Thinking outside the box by a copilot (or rather, thinking outside all his training and fundamentals of the theory of flight) was what caused this crash. If they had done -absolutely nothing- once the autopilot and autothrottle disengaged, the accident would not have occurred. If the copilot had relinquished control to the other copilot to allow him to drop the nose and recover from the stall (a warning of which was making a bloody great racket every couple of seconds in the background) it would not have occurred. To try to blame software, or the computers, in this case is just plain wrong. Thinking inside the box is exactly what I'd want from a commercial airline pilot flying me anywhere...

[b166er]

I don't have a fear of flying either, I do have a fear of suddenly not flying though!

[Anonymous Coward]

Law suit sequence

So the aircraft instruments fail (Airbus' fault) and then a couple of assholes that didn't know what do to kill the crew and passengers because they did not know how to fly straight. And the captain was somewhere else at the time (Air France's fault x2).

In US that would be a multi-billion dollar law suit that at least would partially comfort the victims families with a few million dollars each.

If I was the French president I would ask all planes from the company to be grounded until every single pilot and co-pilot could prove they can at least recover a commercial plane from a stall (Aviation 101 - Class 1 - Introduction).

I, too, blame the french.

[xj25vm]

"Even while the airline was plunging to earth at 10,000 feet per minute, the pilots were not certain whether it was climbing or falling."

I confess to not knowing much about these things, but if the pilots had no idea if they were going up or down, one begs the question if they ever stood a chance - no matter how much skill or training they would have had? Isn't it like flying completely blind folded?

Just wondering

[Aaron Em]

No

You still have the artificial horizon, which tells you the aircraft's attitude with respect to the surface, and the altimeter, which tells you how far away from the surface you are. The latter instrument by itself is enough to tell you whether you're going up, down, or neither, regardless of what the vertical speed indicator's telling you; granted those instruments may not agree at a given point, but it sounds from all I've been able to find like the VSI was going all over the place and the altimeter wasn't, which suggests to me at least that the VSI was the less trustworthy of the two.

Besides which, the condition which caused the instrument disagreement, flight control law change, autopilot disengage et cetera, namely the iced pitot tubes, cleared up long before the aircraft hit the water -- the PM transcript identifies the exact point at which the instruments ceased to be untrustworthy in any way. Beyond that point, the only thing wrong with the aircraft was going on between the ears of the people on the flight deck, specifically the one who kept the aircraft in a stall pretty much all the way down.

[Intractable Potsherd]

They should have known.

There are other instruments, such as the artificial horizon, which would have been giving them the information by telling them whether the plane was pointing towards the ground (sea in this case), or towards the stars.

[Anonymous Coward]

We are at an inflection point.

We could concentrate on giving people training in exactly what they need most when they're needed: When it gets hard. Even starting and landing an aeroplane is a solved problem, at least in fair weather, so there's hardly a reason why you'd have three pilots on board just for that. We might even have been better off with just two in this case: The one co-pilot might've called the captain back much sooner. But that is highly speculative, of course.

What killed everyone aboard was, as I read the transcript, two juniors getting stuck in a panic over a situation that a more experienced pilot mightn't have batted an eye over, and nothing that might've snapped them out of it. They've basically stalled themselves into the sea, and didn't notice--apparently this is possible in a big aeroplane. The simplest fix might even be to make the stick generate feedback both about the other stick and about the rest of the machine. You know, force feedback like you find in your favourite game console.

It is perhaps a testament to how good the technology is that it goes wrong so little, but this transcript does show that the tech has left those in charge of keeping the passengers alive short handed when the technology does drop the ball. Both on feedback, and on experience.

We could also build systems that never give up, not even when it gets hard, and drop the pilot from the equation. We simply do not allow the thing to drop the ball, ever. How? I don't know, I don't design aircraft for a living.

We cannot now mostly because of legal reasons --need a human to blame, rightly so-- but technically, well, those drones already take off and land themselves, routinely. They then manage to get themselves caught, but anyway. And even if we changed the law and we could somehow show that we'd be better off without human pilots, I would probably personally still pass on flying with such a thing, full well knowing it would be on frankly irrational grounds. That, however, doesn't change the argument:

It is clear that solving the easy stuff has had a side effect that the hard stuff will now be left to humans with irrelevant experience. This, too, we need to solve, this way or that way.

Sad? Yes, of course. It is a tragedy, a disaster. Yet the only way to learn is to study the whole thing regardless, draw hopefully correct conclusions, learn. That is something that the flying boys thankfully haven't forgotten.

[SkippyBing]

The extra pilot's onboard so the one landing it has a chance to get some rest, it actually takes quite a long time to fly from Brazil to France and the regulatory authorities don't like people who've been up for 15 hours trying to land at international airports.

You'll also find, surprisingly, that airline training already spends an inordinate amount of its time concentrating on emergencies because they're the things you need to get right the first time despite hardly ever actually having to do them. The question is whether Air France's training needs re-examining, bearing in mind the crew's initial reactions do appear to have been in accordance with the then current procedures.

[Anonymous Coward]

Bonin, in the right seat

had never seen St. Elmo's fire. That's inexperience. I'd say I saw that in the first year. When it got going, flickering across the windscreen like lightnning, and filling the cockpit with the smell of an underground/subway/metro, we would call up the new flight attendants to have a look.

The Popular Mechanics article inferred that the radar had not yet been "calibrated" or "setup" or "turned on". Approaching the Intertropical Front without looking for a hole on radar is not good.

Bonin's control stick is located on the right side and would probably required direct eye contact to determine that he was pulling back, continually. Robert received no feed back on his stick. It may have required physical contact to remove Bonin's grip assuming Robert could identify what Bonin was doing in all the confusion. If Robert's stick had been in the full back position that would have helped sort out the problem.

Another post stated that the pitot tube icing was a previous issue.

It's not usually one thing that kills you. Anyway, I managed to survive ... 21,000 hrs ... and at the end...

takeoffs.count == landings.count .

[Yet Another Anonymous coward]

Not co-pilots

The copilots are NOT learning the ropes.

They are exactly as trained as the pilot in command, typically the captain and first officer will swap piloting - one doing the take off and one the landing for example.

At almost all airlines the captain gets to be in the left hand seat simply because he has worked for that airline for longer. Quite often in the US the first officer is older or has more hours on the type than the captain - but happened to be laid off by another airline and has to start at the bottom again with the new one.

ps They hate being called co-pilots.

[Uplink]

Asynchronous joysticks

Reading the transcript, it looks like having asynchronous joysticks "helped" crash this plane, because the co-pilot Robert didn't have plane feedback to what Bonin was doing, and assumed Bonin was doing as agreed, even though he was doing something else.

If anything, Airbus should look into synchronizing the joysticks and retrofit all their planes free of charge.

Also... put those back between the pilots legs... not out of sight on the side... It's a friggin' plane, not an Xbox.

[Anonymous Coward]

An Automated System Could Have Prevented This

It becomes clear from this report that the aircraft (including its software) operated exactly as promised by the designers. The flight control software detected bad airspeed measurements and disengaged itself largely. It told the pilots that the pitots had iced and that it had reverted to directly channeling pilot's control input to the control surfaces. Even an inexperienced pilot should now know that the anti-stall feature is turned off in that situation. That why they have extensive training. But Bonin did something completely crazy; he pulls the aircraft into steep ascend.

He must have been asleep in theory training, because everybody is told that the speed envelope at this height is quite small. Fly too fast or too slow and you loose lift and fall down...

The A330 even yelled at them "STALL" for 75 times.

The proper operating procedure would be to look at the artifical horizon, pull up the plane to 4 degrees and set the engines to maximum thrust. Other pilots did this safely in the past dozens of times. Why this team was so inexperienced is beyond me. Air France's training program must be put under the microscope, it appears.

Also, it would have been quite easy for an automated system to perform the described procedure after detecting pitot failure. It is only because Airbus is indeed committed to human control that it did not happen. Contrary to what we can read here, humans are indeed in control on all Airbus models and it was humans who failed. More automation would have saved this plane.

[trafalgar]

What happens when a plane hits the water like this, does it break apart? Does it stay in one piece and sink with people trapped alive inside?

[Aaron Em]

Ever do a belly flop?

Imagine that, times a few million. An aircraft that ditches in water shortly after takeoff or shortly before landing, or at the end of a long and careful descent, will stay in one piece; an aircraft that falls out of the sky and slams into the ocean, the way this one did, won't.

FWIW, though, the passengers aren't any more likely to survive such an impact than the aircraft itself is; even leaving aside the effect of ten thousand pounds of airframe collapsing all around, a human body isn't built to survive a hundreds-of-miles-per-hour collision with anything, really.

Basically, if the airframe survives, most or all of the passengers will get off, absent some other problem like fire in the cabin et cetera; if the airframe doesn't survive, as happened here (did you see the news pic of the shattered vertical stabilizer being recovered from the seabed by a Brazilian frigate?) then neither will anyone aboard.

[Anonymous Coward]

At that Speed

Water behaves more like a concrete wall than a liquid. Aircrafts normally respond by spectacular break-up if they meet concrete walls...

[Dodgy Geezer]

Umm...

It was descending almost vertically at 10,000 ft/min. That's over 100 mph.

The deceleration from the impact alone would probably kill many of the people inside immediately by causing disruption of their internal organs and multiple fracturing of bones. Many necks would be broken.

Those passengers who were flat on reclined seats may have survived that initial impact. But the impact would have fragmented the entire underside of the fuselage and probably broken it into several pieces depending on the precise wave pattern that it hit. The top half of the fuselage would then disintegrate downwards in several large pieces, while the floor would similarly break upwards. Large pieces of debris and high speed jets of water would enter the gaps, striking people and tearing bodies apart. Violent accelerations would continue to cause injury.

Finally, the pieces of the aircraft would be plunged many feet below the surface, and most would then continue on to the bottom. A few bodies may have survived the break-up if they were strapped in their chairs, and died of drowning on the way down. But I doubt that anyone was conscious of this.

Is this what you were asking?

[Anonymous Coward]

"too many lines of code are between pilot and aircraft"

Nothing ever goes wrong with safety critical software does it?

Especially with compilers. I mean once the source has all been through AdaTest or whatever, tested exhaustively as people always do (it's required by 178A, right?), what could possibly go wrong? There's no way the compiler could generate the wrong code from the right source, is there? [Hint: check the gcc buglist for code generation bugs, and then look which compiler is being used for safety critical software at companies like Airbus, Boieng, Thales Aerospace and many others]

It's about as likely for something to go wrong at the compilation stage (and not be detected by testing the executable) as it is for two of the available three Pitot tubes to fail identically at the same time with a common-mode design fault, thus outvoting the one correct one. And for the airframe to have no correlation of the Pitot readings with the aircraft's other available airspeed measures as a sanity check, because "there's no need for it".

As another commenter already said, pilots are taught "trust the instruments". Sadly that may not have been the right thing to do in this picture.

But no, actually let's blame this tragedy on inexperienced pilots, because that hides the design failures and regulatory failures that have been sweeping stuff like common-mode failures and inadequacy of source-level code analysis under the carpets for years, because the risks were considered sufficiently low as to be "acceptable levels of risk".

After AF447, are these still "acceptable levels of risk"?

How much extra are you prepared to pay Air France, Easyjet, etc, to cover the additional technology, the additional training, etc?

[Aaron Em]

"Acceptable levels of risk"

depend on how many incidents have actually been the fault of the FBW system you're so upset about. How many can you cite? And, in this case, how do you justify the pilots' actions after the icing condition had cleared?

[Anonymous Coward]

Nonsense

All the data we have points to

A) iced pitot tubes and a correctly reacting flight control software

B) incompetent pilots

So what is the point of your post ? Scaremongering ?

Airbus has an excellent safety record, but their aircrafts are not MuppetProtectedTM. The worst safety risk of an Airbus aircraft has been and will be the humans sitting in row number 0.

[SkippyBing]

Interestingly they thought of this problem a few decades ago and the code for each flight control computer (there are three) is written by a different team which minimises the likelihood of each computer having the same error. And then they test everything a lot.

I'm also interested to know what other methods you think there are of measuring airspeed? Hint GPS doesn't give you airspeed, nor do radio navaids or Doppler nav. To measure the airspeed you need to measure the dynamic and static air pressure and that takes a pitot tube and a static vent.

[Anonymous Coward]

Surrogate Airspeed Indicator

Aircraft designers could deploy the RAT turbine to perform airspeed measurements.

[Anonymous Coward]

"the code for each flight control computer (there are three) is written by a different team "

OK, here's a simple question for you.

How many safety critical systems on an Airbus or Boeing or whatever are developed using independent design/development teams as you describe? For bonus points, how many are developed using dissimilar hardware so a common mode hardware fault is harder to provoke? Let's keep it simple: is it

a) None

b) Some

c) All

d) Don't Know.

I know what the answer is. Do you?

[b166er]

Chris Miller, Wrong!

'02:12:15 (Captain) Alors, là, je ne sais pas!

Well, I don't know!

As the stall warning continues to blare, the three pilots discuss the situation with no hint of understanding the nature of their problem. No one mentions the word "stall." As the plane is buffeted by turbulence, the captain urges Bonin to level the wings—advice that does nothing to address their main problem. The men briefly discuss, incredibly, whether they are in fact climbing or descending'

-------------------------------------------------------------------------------

'02:13:43 (Robert) Alors descends... '

'(Descend, then...)'

45.4 seconds later they hit the sea

[Anonymous Coward]

A bit of disinfo flying about

In my opinion, the cause of this crash is a combination of complacency and inadequate training and another fine example of profit over people.

To clear up some confusion:

GPS is not much use when correct airspeed information is required (the speed of air moving across the wing, affected by things such as wind speed, direction, etc)

A change to alternate law is clearly indicated, requiring immediate acknowledgement, along with a warning that protections have been lost.

When both pilots are using the sidestick there are clear indications.

The stall alarm went on and off intermittently and was no doubt a source of confusion (assuming the crew were even aware of it given the stress they were under).

Their fates were sealed the moment the stall warning discontinued, shortly after the Captain reappeared on the flightdeck. It discontinued because by that point the angle of the aeroplane was so acute that the stream of air required to feed the sensors was not entering the pitot tubes in great enough quantity to be measured, and as a result airspeed dropped off to under 60kts, and 60kts is the minimum airspeed the flight computers will accept before deciding that the information is incorrect and disregard it, thereby shutting off the stall warning.

This is one of those situations that could not be foreseen by the designers of the aircraft, and the reason why (properly trained) humans will always be required (until the advent of true AI).

[CeeOfGee]

So very easily prevented

They forgot that they were flying an aircraft and set about trying to troubleshoot the messages instead of acting on the basic flight information. All pilots are taught how to stall and recover from the stall to the point where you can recognise the signals and instinctively know how to recover from a stall. I was taught this within 10 hours of starting flying and I still practice it today.

If in doubt about the airspeed indications, most SOPs on the Bus call for 90% N1 and 5% Nose up - it may not be easy or elegant or smooth - but you will maintain a rough altitude in a stable attitude. You will survive.

As Korean did years ago - Air France have a serious training issue on their hands - an issue that needs to be addressed very quickly!

Air Chance......second only to Aeroflot now in terms of passenger deaths per miles flown.

[Jamie Kitson]

SatNav

It sounds like when people use SatNav and end up not knowing where they hell they are.

[Sir Lancelot]

Everyone is a pilot ... and an accident investigator

After reading some of the comments I have to admit I was never aware the Register reader corps included so many qualified pilots type-rated on the Airbus A330 series aircraft!

For the arm chair pilots: back to your MS Flight Sims please before you embarras yourselves even more by posting additional non-sense. Perhaps you should start by reading the DGAC reports released so far. The final report is not out yet i.e. the Truth is still out there..

[maccy]

The pilots were partly responsible but it was the computer that really killed them all.

This is the killer: lack of feedback. "Even while the airline was plunging to earth at 10,000 feet per minute, the pilots were not certain whether it was climbing or falling"

For some reason the pilots were unable to figure out the direction of the most important vector a plane has. Most of the blame here has to go to the avionics and the UI. Pilots can be trained to overcome these issues, but once stress kicks in the extra intellectual effort needed to decipher a poor display simply disappears.

[Magnus_Pym]

In all fairness...

... the storm was the primary big nasty one.

My guess is the pilots had very little information to go on due to the possibility of failed instruments, The huge storm throwing the plane about like a toy and the lack of good visibility. They made a judgement call on very little information and it turned out to be a wrong one.

[silver fox]

Only one mention of the effects of the storm....

Can't believe no one's taking the effects of a big violent storm into account. You're all talking as if the plane was in the dark or in cloud but basically flying smoothly (in whatever direction!). A big t-storm would be throwing the biggest of jets around like a paper plane.

If you thnk that you've just been thrown to the left when you've actually just been thrown to the right you're going to have a very hard time believing any instrumentation telling you otherwise. Also, if you know you have some systems/instrumentation failure how do you know that anything any instrument says is true?

And that's all without the psychological effects of increasing stress.

I think the consensus has come round from blaming the pilots to realising how goddam awful some of the design was. The pilots biggest error was not seeing a potential danger and just avoiding it. Once they became embroiled in it, it was only ever going to be a bad day in the office.

[Anonymous Coward]

Alarm faulty?

Ok, the black box says that the stall warning is blaring but do we know for sure that they could actually hear it? Apparently there is no mention of it at all from the pilots, is that really because they ignored it? Still, they should have the experience and knowledge to know it had stalled without the plane telling them.

[Anonymous Coward]

That's the cockpit voice recorder.

It records the sounds heard by a (few?) microphones in the cockpit.

If it's recording the alarm going off, then it really was sounding - and it's also really simple to work out if it's sounding loud enough over everything else going on to be heard.

They either ignored it completely or never commented on it.

Going by their actions, it seems like the copilots mostly ignored it.

This quote is what terrifies me most about it:

"02:11:47 (Robert) On a totalement perdu le contrôle de l'avion... On comprend rien... On a tout tenté...

We've totally lost control of the plane. We don't understand at all... We've tried everything."

For the last two minutes the plane has repeatedly told you it was probably going to stall, and you're surprised and don't understand when you've lost control? Doesn't that mean you probably have stalled?

The real question is how the **** they managed to kill more than two hundred people and themselves before Air France realised they were less than competent.

[James 132]

An AOA gauge and training in including it in the scan may have helped them. It would at least have told them what the wing was doing, but for some reason this isn't seen as necessary, although they do turn up in some 737 options.

For whatever reason, he encountered (and it was a bumpy night) a situation that overwhelmed him and his reaction to it killed everyone on board.

[Dan Paul]

Similar to flight 3407???????? Airlines just don't learn from their mistakes

Living in the Buffalo, NY USA area, we are regularly reminded of the crash of Colgan Air (Continental) Flight 3407 Dash-8 Q400.

Same issues, both pilots inexperienced in flying planes under manual control, bad user interface, pilot fatigue, adverse flying conditions (icing conditions in Buffalo during the winter, who'd have thought?) and resultant stalling and terrible crash and loss of life.

Here are my suggestions:

Icing - Automatic temperature sensing of icing conditions to operate all wing, flap, instrument heaters. Redundant fail over heaters. NO MANUAL OVERRIDE to turn them off, only one to turn them on. Icing Problem (mostly) Solved.

Training - Pilots who will fly in icing conditions should be required to train in ACTUAL ICING CONDITIONS UNDER MANUAL CONTROL and be CERTIFIED to fly in regions that have inclement weather. I would rather trust a flight out of Buffalo with a local pilot than one INTO Buffalo with a warm weather pilot from somewhere else.

Regulation - All new aircraft flown commercially shall be manufactured with automatic de-icing systems (as above) and all existing commercial aircraft shall be retrofitted with same or the ENTIRE Airline will be decertified. Same goes for the pilot training.

NOW if something goes wrong, pilot error will more obviously be the fault instead of the equipment.

Right now, many Pilots are nothing more than glorified Taxi Drivers who rarely if ever fly planes in manual.

[Richard 12]

Those anti-icing systems you talk about? They are already fitted.

The pitots etc have very powerful heaters in them, and in this case they did in fact clear the blockage on one of them roughly two minutes after they first iced up.

Sometimes the heating can't cope, sometimes it fails, that's pretty rare and why they do have several independent units of each instrument.

The thing I find the most upsetting?

It really sounds like if the pilots done *nothing at all*, just re-engage the autopilot once the instrument deiced, the plane would have been ok and everyone would have survived.

Maybe it really is time to take the pilots out of the front seat.

[b166er]

Sir Lancelot: What the fuck indeed, is that just a blatant troll?

You accuse people of posting nonsense with nothing whatsoever in your spew to demostrate that.

I think most intelligent readers can determine from the flight recorder cockpit transcript roughly what happened to those 228 people without being pilots.

By the same logic, are you unable to grasp concepts that don't fall squarely within your area of expertise?

If you ARE a qualified pilot, perhaps you could answer an earlier question: Why is it that the low altitude alarm doesn't sound until 2000 feet during midflight?

I would expect that 2000 feet is seriously well below expected altitude in that stage of the flight.

Meantime, I'll get back to Microsoft Flight (in Beta shortly) if you get back to your Mail :)

I think it's also wise to remember, that solutions often come from left of field and from persons who aren't tightly focussed on problems particular to their subject matter.

[Sir Lancelot]

B166er:

Why don't you read this first concerning the 'accuracy' of the Popular Mechanics article:

http://www.pprune.org/rumours-news/466259-af447-final-crew-conversation-36.html#post6891763

"You accuse people of posting nonsense with nothing whatsoever in your spew to demostrate that."

One example but there are many more: using GPS-derived ground speed in an analysis of the aerodynamical state of a stalled airplane descending in a tight right-hand turn!

"I think most intelligent readers can determine from the flight recorder cockpit transcript roughly what happened to those 228 people without being pilots.".

Of course they can (the plane crashed and the people died, didn't they?) but determining what the probable causes and contributing factors of the accident are, is an entirely different thing, isn't it?

"By the same logic, are you unable to grasp concepts that don't fall squarely within your area of expertise?"

I really think you need a bit more than grasping a few concepts about flying to be an accident investigator...

"If you ARE a qualified pilot, perhaps you could answer an earlier question: Why is it that the low altitude alarm doesn't sound until 2000 feet during midflight?"

A simple one to answer: it was not designed to measure heights larger than 2500 ft above sea/ ground level (the range of the radio altimeter). You do know the RADALT is part of the Groud Proximity Warning System, don't you? This system was designed to prevent CFIT (controlled flight into terrain) accidents. The chances of encountering terrain are rather slim at flight level 350 don't you think? The RADALT is only used by the GPWS and by autolanding systems. Pilots are not allowed to use it as a primary altitude measurement system. Your primary altitude measurement system is a barometric instrument with a selectable reference height/altitude/pressure.

On top of that: in IMC pilots are expected to scan a number of essential flight parameters including altitude and vertical speed. From the interim report released by the BEA (did you read it yet?) it is clear that all AF447pilots where perfectly aware of the plane's altitude and vertical speed during the decent. You do know that altimeters and VSIs use the static pneumatic ports which were not iced up, don't you? In others words: pilots should not be dependent upon altitude alerts to know what altitude the airplane is flying at! Scanning your instruments is one of the primary pilot tasks!

Better stick to simming...

[Anonymous Coward]

Boeing patented synchronous joysticks?

Boeing patented synchronous joysticks? Or is there some other reason the Airbus is "asynchronous".

[Johan Bastiaansen]

A blind trust in technology

A blind trust in technology without understanding the mechanism behind it, can you get killed.

Any pilot that pulls the stick back in a stall, should be executed in front of his family. Even when you've only dabbled in M$ Flight simulator you know that.

What about the captain leaving the cockpit when he knows the plane is heading towards heavy weather? What about leaving the less experienced co-pilot in charge.

Why didn't he take the controls when he finally, after being asked twice, came back to the cockpit?

These are bad decisions, but they were made under stress. Who can claim he would never make such mistakes?

I don’t agree with the optimistic note at the end: “From now on, every airline pilot will no doubt think immediately of AF447 the instant a stall-warning alarm sounds at cruise altitude.”

What? We need a plane plunging into the ocean with the stall warning blaring, just to remind us that when the stall warning is given, the plane might actually be stalling? And is stall warning something we should look into only when flying over the ocean? Or also over land? Geez, how old are these pilots? I wouldn’t let them ride a bicycle.

But the biggest mistake is in the design of the plane. Who figured it would be a good idea to average out the input from both sticks? Find out who it was and invite his family to witness his execution. Was it decided in committee? Good, there’s plenty of room in the mass grave.

This in the only way to really learn from this mess. Because the managers who managed it all wrong will take control of this “learning process” and manage to manage it all wrong again. Don’t believe they have learned a thing. If they don’t literally FEEL the consequences of their stupidity, they’ll never learn. They’ll only learn to cover their arse with more new speak.

Kill them. It’s self defence. If we don’t, they’ll kill us.

You think this is harsh? Don’t blame me, blame Darwin. Survival of the fittest, not survival of the fattest.

[Anonymous Coward]

Of course it's easy

"Any pilot that pulls the stick back in a stall, should be executed in front of his family. Even when you've only dabbled in M$ Flight simulator you know that."

Try it at night with no visual horizon, travelling at 450 knots at 37,000 feet while flying through a violent storm with no airspeed indicators. A little more challenging, don't you think?

[Dodgy Geezer]

Umm...

I am not a qualified pilot, but I can't help thinking that most of the comments above don't really address the major issue.

A lot of them say that Bonin should not have held the stick back and so Air France training must be faulty. I really can't believe that you can get onto the flight deck of any commercial passenger airline without a pretty clear idea of what to do in a stall. So I'm pretty sure that Bonin knew.

Though not a pilot, I am a well-qualified human being, with many years of living experience, and I think I recognise the major problem that affected Bonin. It's called 'digging a hole for yourself'. All humans experience it on occasions. What happens is that all your higher brain functions get put on hold, and you keep pursuing the one course that you are following. Usually with disastrous results, particularly if it happens during an interview (which is a very common place for me to experience it).

I think Bonin, spooked by his first experience of running into a tropical storm, reverted to a well-understood method of avoiding ground impact, pulling back on the stick. That was quite safe to do in 'normal law'. And there is no point talking about better UI or cleverer displays or better training at this point, because he was locked into this action. I suspect you could have slapped him and shouted in his face to let go of the stick, and he still wouldn't have. I know that when I'm ballsing up an interview like this I often know I'm going wrong, but it doesn't stop me carrying on regardless.

The only chance at that point was for someone to recognise what Bonin was doing. There was a possibility the Captain might have done - he was just starting to realise what was going on when they hit. So it is probably worth stopping having independent control movements with no signal between the control positions. But if you want to get to the root of the control problem here you will have to redesign the human brain....

[get off]

Hideous and horrible. The whole thing. I went on to read the writer's blog...

'What did the passengers feel'

Plus, the writer does some very wacky things himself.

"Today we fire machine guns"... for a laugh!

http://jeffwise.net/

It's all a bit boy's own and alarmingly close to Top Gear territory, for my own sensibilities (I don't watch it) but interesting all the same.

[BatCat]

Mistake number 1...

... was not re-routing round the storm. The report says other aircraft routed round it so why didn't this one? Maybe they didn't have enough fuel contingency? A few years ago, when I was involved in reporting an incident to the CAA, I spent a bit of time reading around the causes of incidents in UK airspace. Apparently, one of the major reasons for requesting emergency / priority landings at UK airports was lack of fuel. Airlines were cutting corners on fuel contingency because of the cost of hauling spare fuel around the place. I wouldn't be surprised if cost-cutting in this way was the primary cause of the incident.

I'm not a pilot, real or virtual, but I do know that when I had a car with traction control I got into the habit of accelerating far too hard and early exiting corners - when it was in for a service, I had a lower spec car on loan, I nearly crashed it on the first wet roundabout - sounds like a similar thing going on here with dependency on fly by wire.

[Hairy Airey]

Once the pitot tubes were frozen their chances of survival were minimal

I was once told by a pilot on a 757 flightdeck (back when they allowed passengers in the cockpit) that there are always three manual instruments, airspeed, altimeter and gyroscope. To lose the readings from one of those makes flying difficult indeed. Flying into a storm you wouldn't necessarily get an accurate altimeter reading since these rely on atmospheric pressure and the plane could be thrown up and down as well as atmospheric pressure changing. So that leaves one reliable instrument.

They would have needed to descend to an altitude at which the pitot tubes would defrost without ending up in the ocean, virtually impossible in those conditions. A real tragedy.

[Dodgy Geezer]

Once the pitot tubes were frozen their chances of survival were minimal..

...so once the tubes were un-frozen their chances of survival were assured?

The tubes were frozen for a very short while - a minute or so. This confused the pilots, particularly Bonin, who was anticipating major problems flying heavily laden through a tropical storm.

When he flew the aircraft into a stall and held it there, he confused Richard, who started to believe that the controls were faulty..

When they got the captain up to the office, he saw what the problem was in a minute or two, particularly once Bonin explained what he had done. Too bad there was only 1.5 seconds left....

[Anonymous Coward]

Where was the pilot?

What I don't understand is why the cockpit stall warning was not also played into the crew "resting" (I assume this means sleeping) area? Any time you have a major alarm I'd want to get every hand on deck. If the stall warning was played into the captains bunk he would have got his arse back to the cockpit a little faster. And if it *was* played to his bunk and he rolled over and ignored it like the two co-pilots...

Sure, no need to wake everybody up for trivial alerts, but for major alarms (like stall, terrain, etc) you want to make sure you have all the available help. Who knows, maybe both co-pilots have passed out and there is nobody awake at the stick.

The other part I don't understand is why the plane didn't revert to "Normal Law" once the instruments came good? I can see an argument for not automatically reverting to "Normal Law", but I would expect the plane to prompt saying it's a really good idea to revert.

[bep]

Grim reading

It appears that instrument flying, with no reliable visual reference, is as hard as it ever was.

As some posters have alluded to, the Airbus approach does seem to somewhat isolate the pilots from the key indicators while suddenly overwhelming them with input. It's not surprising the most inexperienced pilot panicked, but he should have been a spectator by that point. I don't know what to say about the averaging of the flight controls though, I just can't see how anyone would think this was a good idea.

I also think we're all entitled to an opinion on this, any one of us could have been on board that aeroplane.

[blackjack205]

what the hell are you guys saying?

how many people who posted very poor assumptions here are actual airline pilots? (atpl with a jet rating, at least 2000 hours flying time..) we are required to do a "recurrent" training at least every six months inside a simulator to train for these scenarios (airspeed unreliable/recovery from a stall, etc). but I tell you, no matter how good you are at flying, simulated or actual, nothing can really prepare you for the real thing. ABSOLUTELY NOTHING. bottomline is, ask any airline pilot and if that pilot says "that won't ever happen to me" then he has no respect for the situation and he has no idea of the dangers.

[Anonymous Coward]

independent design teams, dissimilar hardware

@SkippyBing 21:20 11 Dec

There are safety critical systems on pretty much every commercial airline that do not use independent design teams and/or do not use dissimilar hardware for resilience.

This actually isn't a great concern, what's more a concern is the people round here who think these techniques are in common use and therefore that all will be OK.

Everything probably will be OK, but the use of independent design teams and dissimilar hardware is not necessarily a reason why.