Air France 447 - On topic only!

[Gabriel]

I can tell you with confidence that if I had been one of these pilots on that night I would very likely make the same mistakes. Why? Because of what I didn't know and what I wasn't prepared for. FIrstly, I had no training in high altitude manual flight, no training in procedure for loss of autoflight at cruise, no training for unreliable air speed, no training for stall/upset/unusual attitudes recovery and no training for CRM when flying with another F/O.

Think about that. What is so hard to understand?

What?

Look, they do hand-flown take-offs and landings day after day. They know, and have a practiced motor-skill, that pulling back on the stick makes the plane pitch up.

I can understand lack of training in what relates to not realizing about the UAS situation, or about not knowing the UAS procedure, or at being inefficient and imprecise while trying to hold a target attitude (lack of practice hand-flying at high altitude).

I CANNOT understand lack of training in what relates to make bold, sustained and frequent pull-up inputs that put a plane cruising in what they knew was already the limit of the envelope in a sustained violent climb that reached 12° nose-up and 7000 fpm climb, and keep it in that attitude for several seconds, and when the stall warning triggers (before an actual stall) they pull up hard actively stalling the plane, and kept pulling up with the stall warning shouting in clear English STALL STALL STAL uninterruptedly for 54 seconds.

Sorry, I will NEVER understand how lack of UAS training, high altitude upset recovery, and high altitude manual flight, relates to the above. The pilot inputs would have stalled a non-UAS, non-fly-by-wire DC-8 flying at low altitude, or a Cessna 152 for the matter.

I am NOT saying that the appropiate training would have not prevented the accident. On the contrary, it would have likely done that. Just that the lack of training ALONE doesn't explain the extreme, large, sustained, repeated, irrational inputs.

All of the above said, I do see how Airbus made things more complicated (and I know that, in many cases, a Boeing plane would have made no difference).
- The computer knew it had disconnected the AP and AT due to an unreliable airspeed, but didn't tell the pilots so.
- The flight directors might have presented confusing info, but apparently did not command the level of pull-up made by the pilots, just a slight climb. Since it is required that the pilots switch the FD off whenever the AP disengages for itself, the computer actively disengaging the AP could have killed the FD too.
- The sidestick lacks force feedback. In Boeing FBW planes the yoke has force feedback.
- The plane even in alternate law lacks longitudinal stability. In fact, even more in alternate law. In normal law at least the AoA protection acts like a longitudinal stability at high AoA. And in direct law at least the stick inputs are proportional to the elevator deflection, so the plane will have longitudinal speed and AoA stability. Boeing FBW planes have positive longitudinal speed and AoA stability in all flight control laws.
- There is a lack of throttle feedback too: the position of the throttles doesn't change when the thrust change in auto-thrust, and also when the autothrust self-disengages. The throttles were left at climb with the actual thrust being quite less than that.
- The sidesticks don't move together and the sidestick in use is hard to see for the pilot not-flying. Boeing planes have twin connected yokes that are big, bold, placed just in-front of the pilot and move together. it is likely that the non-flying pilot, and later the captain, didn't knew that the flying pilot was mostly pulling up until he said, 5 seconds before the crash "But I've been pulling up the whole time!" In a Boeing, it would be hard for the non-flying pilot to miss the yoke against his stomach.
- The stall warning, after sounding continuously for 54 seconds, started to behave stupidly. The captain must have noticed the high deck-angle as he "climbed" the aisle toward the cockpit, must have felt the stall buffet, and BEA said that he had to hear the stall warning just before entering the cockpit, time at which the stall warning stopped. The reason why it stopped is because at very slow speeds the AoA values are rendered invalid, and the reason for those very low speed readings was NOT the UAS (it had been already reverted by then), but that the very high AoA (of some 40°) left the pitot tubes severely misaligned with the air stream, rendering them ineffective. At one point the NFP took control and lowered the nose. By doing that he reduced the AoA, which diminished the large misalignment of the pitot tubes that started to show a faster airspeed, rendering the AoA values valid (and of course still too large) which triggered the stall warning. Then there were new pull-up inputs, the speed showed too low again, and the AoA became invalid again, and the stall warning stopped. So the illusion was that pushing down triggered the stall warning and pulling up stopped it. Let's be clear on this: This was AFTER some 40 seconds of irrational pulling up which ultimately triggered the stall warning to begin with, PLUS 54 seconds of continuous stall warning that were addressed with more pull-up inputs. I don't have any faith that, had the stall warning kept working properly, they would have solved the problem that they created and failed to solve for about one minute and a half. But at minimum it didn't help, especially not to the captain that got into the cockpit just as the stall warning stopped. Despite of that, the stall warning triggered (and stopped) dozens of times later during the fall. So the captain must have known that AoA was an issue, but perhaps the intermittent nature of the warning made him think that they were just at a marginal AoA case, which of course was inconsistent with the nose of the airplane pointing high up with the plane falling like a rock, which is like shouting DEEP STALL ASSHOLE!

So, while the pilot inputs would have stalled any plane at any altitude and any phase of flight, maybe, just maybe, this accident would not have happened in a Boeing because of the AoA stability, force feedback, and the NFP noting early in the sequence the stupid nose-up inputs.

[Black Ram]

obviously it will take a while to get through the whole thing, but start with this fault: "However, no explicit indication that could allow a rapid
and accurate disgnosis [sic] was presented to the crew."

strike 1 airbus

this will make the boeing guys happy:

"He may therefore have embraced the common belief that the aeroplane could not stall, and in this context a stall warning was inconsistent."

i have to say that this has to be a belief among some pretty uneducated folks. that or airbus trainers are really screwing up.

"No failure message is provided that identifies the origin of these other failures: in particular, the rejection of the ADR’s and of the speed measurements."

strike 2?

Here's the thing. The idea the crew were confused about what was causing the automation failures and error messages, overwhelmed with trying to diagnose the root cause and figure out what exactly was going on with "HAL", and thus forgetting to fly the plane may seem compelling, but is something that was believed prior to the extraction of the CVR. What the CVR revealed in the short publication from spring 2011 is that they knew RIGHT AWAY they were in an UAS situation and one of the pilots did call out "UAS...then Alternate Law". So no confusion over error messages, no confusion over non-moving thrust levers when on AT. They had all the information to diagnose the problem right away and they did so, as recorded by the CVR.

Offside: TeeVee?

What follows then is a complete mismanagement of the problem. They failed to follow the UAS procedure, which guarantees safe flight. So did the other crews who experiences this, but as many others have pointed out, in the case of AF 447 the pilots embarked upon a very bizarre course of action, with decision-making and control inputs that don't make sense in general. This was the only crew to do this, thankfully. Also, you can't say the aircraft was not functioning properly. It was behaving and performing as designed and everything was working fine, other than the speed indicators. They still had the Attitude Indicator, the Altitude, the VS...they had a stall warning for almost a minute. Any attempt to properly analyze the situation should have quickly given them an idea of what was going on. And of course, they climbed to 38000ft, when just two minutes before they had declared 37000ft was out of reach.

[Myndee]

Interesting that this thread (V2.0) is alive and well.

I think it's summed up in the last recorded comment on the CVR (which Air France did NOT want released): "...but what is happening?"

I can't help but wonder if AF had become too complacent with their pilots. These copilots were fairly young men. Are the new generation of pilots flying FBW aircraft getting the benefit of the same training as the generations before them? Are pilots being taught to let the computer do its thing, and if it fails, don't worry because backup systems A, B, and C will fix it?

I hate to see pilots vilified, especially if their airline didn't provide all the training they needed. Methinks AF is getting out of this a little too easy...

[Spectator]

Here's the thing. The idea the crew were confused about what was causing the automation failures and error messages, overwhelmed with trying to diagnose the root cause and figure out what exactly was going on with "HAL", and thus forgetting to fly the plane may seem compelling, but is something that was believed prior to the extraction of the CVR. What the CVR revealed in the short publication from spring 2011 is that they knew RIGHT AWAY they were in an UAS situation and one of the pilots did call out "UAS...then Alternate Law". So no confusion over error messages, no confusion over non-moving thrust levers when on AT. They had all the information to diagnose the problem right away and they did so, as recorded by the CVR.

Please don't spoil our fun with a few facts... UAS was called out, the procedure wasn't followed. We shouldn't even be discussing the irrational pulling up and crazy pitch angles. The fact is they should have been flying on the memory items and check list. No doubt clearer messaging, instrumentation and warning sequences may have just maybe given them a clue as to what they had got themselves into, but from the comfort of my reply-with-quote-box the only person who SHOULD have wanted better instrumentation/warning sequences would have been the captain who walked in half way through the episode.

That said, I think the fault handling of the glass cockpit could be improved, but it is in no wise a "ten minute" upgrade! There is a lot of complex logic in there, and the ramifications of even the smallest change are big. There'll be testing and review and regulatory approval and and and and before it reaches a 'plane that one of us will ever board.

[Evan]

Sorry, I will NEVER understand how lack of UAS training, high altitude upset recovery, and high altitude manual flight, relates to the above. The pilot inputs would have stalled a non-UAS, non-fly-by-wire DC-8 flying at low altitude, or a Cessna 152 for the matter.

Gabriel, have you read the conclusions and findings? There is a bit there about how disorientation due to the 'startle factor' has repeatedly caused pilots to lose their cognitive bearings and do things that would seem contrary to reason. This is what I believe actually caused the crash. But this can happen on any aircraft currently flying, even a rusty old DC-9. The industry has known about this for a very long time and has made provisions to deal with it.

They are called procedures, and the process is called CRM. If airmanship alone could reliably step in during these situations, we wouldn't have developed these procedures and CRM. But it can't. These procedures become even more important as planes have become more complex, where functions are affected by situations in ways the pilots would not consider. For instance:

I am NOT saying that the appropiate training would have not prevented the accident. On the contrary, it would have likely done that. Just that the lack of training ALONE doesn't explain the extreme, large, sustained, repeated, irrational inputs

But following the flight director guidance, combined with over-control might begin to explain it. This is why both Boeing and Airbus have procedure that tells them very simply to turn the flight directors off. Your situational awareness might be flawed. Take a few moments to sort it out; but do these things in the meantime. Because we know from ages of experience that you might not be thinking clearly at this very moment, and these are the correct things to do to stabilize the situation, to buy you time to get your situational awareness in order and to work the problem methodically.

black and white when it is required, and creative when required. Remember the manuals have not been written in a day, and the default must always be adherence

And ensuring that pilots use these memory items and QRH checklists is a matter of training and practice alone. Going to these procedures must be their first instinct. Until that is true, these things will keep occurring and we will continue to wonder what the pilot could have been thinking at that moment.

It's training training training. These guys did have basic airmanship skills (though not practiced stall avoidance technique). You see what good that does you when your mind is disoriented. They couldn't access them.

What I was hoping would come from this is an industrywide wake-up moment, followed by a humility on the part of pilots who believe their superior airmanship would prevent them from ever doing something like this. It won't if they can't think clearly (due to information design flaws in the human mind) and so they need to get very familiar with the procedures and CRM discipline.

Training, practice and procedural discipline are the only ways to combat these flaws of the mind.

- The computer knew it had disconnected the AP and AT due to an unreliable airspeed, but didn't tell the pilots so.

Yes it did (and they recognized it). The sequence of messages, including the ones that remain on the ECAM as the others scroll off it, were indication enough for adequately trained pilots. There was also an obvious inconsistency with the airspeed indicators. A 767 pilot does not have an ECAM, but he still recognizes UAS. The ECAM shouldn't have to spell it out for them, but in light of this accident, I would prefer that it did. It's not the solution though, and it's not the road we want to go down.

I don't understand why the report speaks so much about a lack of clear indication of UAS, since this clearly did not play a part in the accident. MY findings and conclusions are that this report was politically influenced to avoid placing blame squarely on Air France, the BEA's biggest 'customer'.

[Evan]

CNN seems to have gotten it right.

Absolutely. CNN's Richard Quest is the first major mainstream media reporter to get it right and explain it clearly. All props to him.

[MCM]

OK,

So did all three pilots not understand what was going on?

Did both pilots comprehend the fact, at the start, that it was unreliable airspeed?

Or did ONE pilot not become in the loop and mishandle the situation, while the other pilot thought the handling pilot was trying to do the right thing, unaware he wasn't?

[TeeVee]

Here's the thing. The idea the crew were confused about what was causing the automation failures and error messages, overwhelmed with trying to diagnose the root cause and figure out what exactly was going on with "HAL", and thus forgetting to fly the plane may seem compelling, but is something that was believed prior to the extraction of the CVR. What the CVR revealed in the short publication from spring 2011 is that they knew RIGHT AWAY they were in an UAS situation and one of the pilots did call out "UAS...then Alternate Law". So no confusion over error messages, no confusion over non-moving thrust levers when on AT. They had all the information to diagnose the problem right away and they did so, as recorded by the CVR.

Offside: TeeVee?

What follows then is a complete mismanagement of the problem. They failed to follow the UAS procedure, which guarantees safe flight. So did the other crews who experiences this, but as many others have pointed out, in the case of AF 447 the pilots embarked upon a very bizarre course of action, with decision-making and control inputs that don't make sense in general. This was the only crew to do this, thankfully. Also, you can't say the aircraft was not functioning properly. It was behaving and performing as designed and everything was working fine, other than the speed indicators. They still had the Attitude Indicator, the Altitude, the VS...they had a stall warning for almost a minute. Any attempt to properly analyze the situation should have quickly given them an idea of what was going on. And of course, they climbed to 38000ft, when just two minutes before they had declared 37000ft was out of reach.

my gather from the report is that the investigators were NOT convinced the crew were in agreement over UAS. while they both mentioned "speeds" they never actually stated "UAS" or anything like this.

"The impression of an accumulation of failures created as a result probably did not incite the crew to link the anomaly with a particular procedure, in this case the “Vol avec IAS douteuse” procedure"

"The symptoms perceived may therefore have been considered by the crew as anomalies to add to the anomaly of the airspeed indication, and thus indicative of a much more complex overall problem than simply the loss of airspeed information."

"The PNF detected the climb based on observation and reasoning (“according to all three you’re climbing”), which indicates the beginning of a loss of confidence in the instrument readings."

I could go on quoting for a day or so, but the idea is that these guys were lost. Scared shitless and lost. The captain returned in the midst of absolute mayhem and he too became lost.

All i'm sayin is that the aircraft KNEW EXACTLY WHAT WAS WRONG WITH IT FROM MOMENT ZERO and IT FAILED TO INFORM ITS CREW IN PLAIN LANGUAGE: UAS. i mean CHRIST! the computers are so sophisticated and monitor every system to a ridiculously detailed degree and are programmed to disconnect or shut down certain systems in the event of UAS but are not programmed to flash a simple 3 fuckin letter code on the ecam: UAS???????

yes, two trained pilots should be able to say, "hey my airspeed indicator is funky." "mine too!" and then in unison, "UAS!" well, for whatever reason, and the report explores the possibilities, they did not. perhaps a simple UAS indication on the ecam would have removed the doubt from their mind.

this is a horrible chink in airbus' logic and programming. a chink, that in my worthless opinion, was a contributing factor in this crash. One that I'm fairly certain some lawyers are going to point out at trial.

we can continue to point out how miserably the crew failed, how miserably AF failed, how miserably the caa failed. but at the same time, we should be pointing out how miserably airbus failed.

[Evan]

OK,

So did all three pilots not understand what was going on?

Did both pilots comprehend the fact, at the start, that it was unreliable airspeed?

Or did ONE pilot not become in the loop and mishandle the situation, while the other pilot thought the handling pilot was trying to do the right thing, unaware he wasn't?

12 seconds into the event the PM says:

"We’ve lost the the the speeds so… engine thrust A T H R engine lever thrust"

At the same time the PF says:

"We haven't got a good display of speed

One minute and 36 seconds into the event the PF says:

"I have the impression (we have) the speed"

[TeeVee]

oh, and evan, coming from the same guy that railed on about the lack of coordinated water rescue facilities in a poor island nation and has bitched about other lack of something or other regarding safety, i'm very surprised--well maybe not--that you would not want airbus and perhaps boeing as well to require this simple function.

[TeeVee]

12 seconds into the event the PM says:

At the same time the PF says:


One minute and 36 seconds into the event the PF says:

and 45 seconds later says he thought they had some crazy speed, indicating he may have believed they were in an overspeed situation.

[3WE]

Whoops....why has no one said this (at least not recently)?:

If you freeze your pitot tube and climb, the indicated airspeed increases.

So you pull up some to slow down, but then your speed increases more, so you pull up more.

I know, it unthawed later, and yeah, I don't like pulling up hard, when the nose is already up and warnings and stuff are happening, but other planes with frozen pitots have been stalled out while the pilot worked to slow down since indications were that he was going too fast.

[kris]

All of the above said, I do see how Airbus made things more complicated

Also the switching to alternate law "for the remaining of flight". Why isn't the normal law returning with speeds? That (with protections back) would save them too.
The need to switch off air data units, not very helpful.
Of course a basic stall understanding (of at least one flight crew member) would fix this and other problems even better.

[kris]

Whoops....why has no one said this (at least not recently)?:

If you freeze your pitot tube and climb, the indicated airspeed increases.

Why exactly?

[Gabriel]

Whoops....why has no one said this (at least not recently)?:

If you freeze your pitot tube and climb, the indicated airspeed increases.

So you pull up some to slow down, but then your speed increases more, so you pull up more.

I know, it unthawed later, and yeah, I don't like pulling up hard, when the nose is already up and warnings and stuff are happening, but other planes with frozen pitots have been stalled out while the pilot worked to slow down since indications were that he was going too fast.

That is in theory (a fully air-tight blocked pitot) and not the case here, where the false speeds were very low.