Air France Flight 447 report

[LondonJohn]

Hence the settings of the engines to Take Off Go Around. He used the training for a stall near ground level at an altitude getting close to the maximum for that aircraft. If he had done what the Pilot Not Flying had said, to "do as little as possible with the controls", everything would have been OK.

TOGA isn't really an anti-stall manoeuvre. It's simply the technique for gaining maximum vertical speed at optimum thrust, so as to safely gain altitude and airspeed when close to the ground.

It was a massive, massive error for the pilots to think that TOGA parameters were in any way applicable to the situation they found themselves in. They were flying at high altitude, and their altimeter/VS indicator and AI were working fine at all times. Quite apart, therefore, from their seeming ignorance of the radically different aerodynamic characteristics of flying at high altitude vs ground level, it's also totally unforgivable that they didn't look at their major instruments and notice that 1) the aircraft was in a nose-up attitude (as much as 10-12 degrees at times), yet 2) the aircraft was losing altitude. Even with a total absence of airspeed indication (which in any case was only a transient problem), these two indicators should have told the pilots unequivocally that the aircraft was aerodynamically unstable, and that the wings were not generating lift. When you add in the unmistakable stall warning, it's almost beyond belief that the pilots weren't taking correct remedial action well before they got anywhere near 20,000 feet.

 

[LondonJohn]

I'm aghast. It seems as though the less-experienced pilot (I can't bring myself to call anyone with an ATR inexperienced) pilot somehow got focused on the idea of needing to climb, to the exclusion of all else. I only fly gliders, but my take would be that if you're not sure what's going on, retreating to a position of stabilized straight-and-level flight while you figure it out wouldn't be such a bad thing.

Yes: it appears that he became fixated on trying to fly over the bad weather. And even with the temporary loss of indicated airspeed, the pilots had no reason to think that the altimeter or AI were in any way malfunctioning. So, as you say, the basic response to bad airspeed readings in their situation would be to 1) check the altimeter and VS indicator to verify that you were at good height and were flying straight and level; 2) check the AI to ensure the aircraft had a good level attitude - maybe even nudging the nose down a little to assist airflow over the wings, especially since they had a huge amount of safe altitude to play with; 3) set throttle at around 60-70%, to ensure adequate thrust (better to have slightly too much thrust in such a situation than too little); 4) Only then, have one pilot continue to fly the aircraft manually in this configuration - with constant instrument monitoring - while the other pilot runs through the checklist for lost airspeed indication.

The other very illuminating thing about this crash (and it was the same for the Buffalo crash that someone else referenced earlier) was that neither pilot made any radio communication of being in difficulty. This appears to me to be due to a psychological flaw that needs to be addressed far more comprehensively in training: a reluctance for the pilots to make any sort of admittance of confusion or inability to control the aircraft. I believe that in both these crashes (which had certain similarities), the pilots had the mentality of "ah, we'll sort this out in no time", and were thus reluctant to share their problem with anyone outside the cockpit. I think that in the pilots' minds, they made the subconscious choice that had they radioed their difficulties, and had then recovered the situation, they might either be ridiculed for their poor airmanship or disciplined over it. Better then - in their minds - to "keep it a secret" while they sorted it out, and then with luck they might be able to pretend it didn't happen. After all, the FDR and CVR would be overwritten pretty soon, and provided the aircraft didn't have to pull any high-g manoevres or climb/descent at great speed, the passengers would probably never be any the wiser.

 

[LondonJohn]

Another slight side factor behind this accident has been the predictable attempts by the BEA (French accident investigation department) to minimise the role of pilot error, and to almost seek to protect the pilots and the airline where there is any hint of ambiguity. The BEA has plenty of form in this area. It seems extraordinary that some of the BEA investigators are serving pilots in Air France, and it's also unhelpful that the French government remains a major (20%) shareholder in the enlarged Air France-KLM group.

The BEA has behaved in less-than-transparent ways before in high-profile investigations, most notably the Concorde CDG accident and the Airbus Mulhouse crash. By contrast, the AAIB (UK), BFU (Germany), TSB (Canada) and NTSB (USA) are all universally regarded as fair, objective investigation bodies, who have an unfettered determination to seek the truth behind any accident, regardless of the political or economic consequences.

 

[TjW]

Yes: it appears that he became fixated on trying to fly over the bad weather. And even with the temporary loss of indicated airspeed, the pilots had no reason to think that the altimeter or AI were in any way malfunctioning. So, as you say, the basic response to bad airspeed readings in their situation would be to 1) check the altimeter and VS indicator to verify that you were at good height and were flying straight and level; 2) check the AI to ensure the aircraft had a good level attitude - maybe even nudging the nose down a little to assist airflow over the wings, especially since they had a huge amount of safe altitude to play with; 3) set throttle at around 60-70%, to ensure adequate thrust (better to have slightly too much thrust in such a situation than too little); 4) Only then, have one pilot continue to fly the aircraft manually in this configuration - with constant instrument monitoring - while the other pilot runs through the checklist for lost airspeed indication.

The other very illuminating thing about this crash (and it was the same for the Buffalo crash that someone else referenced earlier) was that neither pilot made any radio communication of being in difficulty. This appears to me to be due to a psychological flaw that needs to be addressed far more comprehensively in training: a reluctance for the pilots to make any sort of admittance of confusion or inability to control the aircraft. I believe that in both these crashes (which had certain similarities), the pilots had the mentality of "ah, we'll sort this out in no time", and were thus reluctant to share their problem with anyone outside the cockpit. I think that in the pilots' minds, they made the subconscious choice that had they radioed their difficulties, and had then recovered the situation, they might either be ridiculed for their poor airmanship or disciplined over it. Better then - in their minds - to "keep it a secret" while they sorted it out, and then with luck they might be able to pretend it didn't happen. After all, the FDR and CVR would be overwritten pretty soon, and provided the aircraft didn't have to pull any high-g manoevres or climb/descent at great speed, the passengers would probably never be any the wiser.

It's interesting to compare and contrast this to UA232 -- the DC-10 that lost all hydraulics some years back. I've always felt sorry for Captain Haynes and his crew -- not so much that they had the accident; someone was going to be flying that plane when the fan disk failed -- but that they weren't able to save everyone.

There's a transcript of a talk Captain Haynes gave about the Iowa crash. One of the first things he says in that talk is:

I think there are five factors that contribute to the degree of success that we had at Sioux City: that is, luck, communications, preparation, execution, and cooperation.

He goes on to say:

When we declared an emergency, which Bill did, the copilot, Bill ?? declared an emergency, everything stopped on the ground. Everything went to us. They cleared the frequency for us, they gave us all the help they could get. I don't know how many light airplane pilots you have here today, but I've talked to several groups of several pilots, and this one you can pass on Mary, they're afraid to say anything, they don't like to declare an emergency, they're afraid they're going to cause some problems or something like that. And they said to me, well, you have all these resources of United Airlines at your disposal, and the center, and all this. So do you Three words: I'm declaring an emergency, and you've got it. All the help you want. You've got American Airline's maintenance facility, United maintenance facility, if you stay in the air long enough, they'll patch you through to them, and you can talk to them. So you've got all kinds of help, if you just do that. Now, if you do that, and then land, and nothing happens, you've got a lot of reports to fill out, and you're going to have a lot of airline pilots that are upset at you, especially at a place like Chicago, because they're out holding, while they're getting you on the ground. And you'll probably hear about it. But in truth, and when I've said this, the pilots, yeah, they shake their heads, yeah, they're going to complain, but they're really glad you made it, and they're very happy you're able to use the services available to you. So communicate with the ground, tell them your problem, and they'll help you. They really will.

 

[a_unique_person]

TOGA isn't really an anti-stall manoeuvre. It's simply the technique for gaining maximum vertical speed at optimum thrust, so as to safely gain altitude and airspeed when close to the ground.

It was a massive, massive error for the pilots to think that TOGA parameters were in any way applicable to the situation they found themselves in. They were flying at high altitude, and their altimeter/VS indicator and AI were working fine at all times. Quite apart, therefore, from their seeming ignorance of the radically different aerodynamic characteristics of flying at high altitude vs ground level, it's also totally unforgivable that they didn't look at their major instruments and notice that 1) the aircraft was in a nose-up attitude (as much as 10-12 degrees at times), yet 2) the aircraft was losing altitude. Even with a total absence of airspeed indication (which in any case was only a transient problem), these two indicators should have told the pilots unequivocally that the aircraft was aerodynamically unstable, and that the wings were not generating lift. When you add in the unmistakable stall warning, it's almost beyond belief that the pilots weren't taking correct remedial action well before they got anywhere near 20,000 feet.

Pilot, not pilots. The Pilot Not Flying tells Bonin exactly what to do, Bonin indicates he is complying.

02:10:27 Robert Faites attention à ta vitesse. Faites attention à ta vitesse. Pay attention to your speed. Pay attention to your speed.
2:10:28 Bonin OK, OK, je redescends. Okay, okay, I'm descending.
02:10:30 Robert Tu stabilises... Stabilize…
02:10:31 Bonin Ouais. Yeah.
02:10:31 Robert Tu redescends... On est en train de monter selon lui… Selon lui, tu montes, donc tu redescends. Descend... It says we're going up... It says we're going up, so descend.
02:10:35 Bonin D'accord. Okay. As the de-icing system takes effect, one of the pitot tubes starts working again, and the cockpit displays again show correct speed information
02:10:36 Robert Redescends! Descend!
02:10:37 Bonin C'est parti, on redescend. Here we go, we're descending.
02:10:38 Robert Doucement! Gently! Bonin decreases his backward pressure on the stick, and stall warnings stop as the airspeed increases
02:10:41 Bonin On est en… ouais, on est en "climb." We're... yeah, we're in a climb. Bonin does not push the nose down. Robert presses a button to call the captain.
02:10:49 Robert Putain, il est où... euh? Damn it, where is he?
02:10:55 Robert Putain! Damn it! The other pitot tube starts working again; all aircraft systems are once again functional, and the pilots simply have to push the nose down to recover the aircraft from its stall.
02:11:03 Bonin Je suis en TOGA, hein? I'm in TOGA, huh? TOGA means take-off, go round. Bonin is attempting to increase speed and climb, but he is at 37,500ft where this maneuver is inappropriate. 02:11:06 Robert Putain, il vient ou il vient pas? Damn it, is he coming or not?

Robert gets distracted because he is trying to get the Captain to come back from his break, and misses that Bonin is doing exactly the opposite of what he is being told to do.


Bonin's wife was on the flight, which could help explain the confused and panicked response.

 

[gumboot]

Another slight side factor behind this accident has been the predictable attempts by the BEA (French accident investigation department) to minimise the role of pilot error, and to almost seek to protect the pilots and the airline where there is any hint of ambiguity. The BEA has plenty of form in this area. It seems extraordinary that some of the BEA investigators are serving pilots in Air France, and it's also unhelpful that the French government remains a major (20%) shareholder in the enlarged Air France-KLM group.

The BEA has behaved in less-than-transparent ways before in high-profile investigations, most notably the Concorde CDG accident and the Airbus Mulhouse crash. By contrast, the AAIB (UK), BFU (Germany), TSB (Canada) and NTSB (USA) are all universally regarded as fair, objective investigation bodies, who have an unfettered determination to seek the truth behind any accident, regardless of the political or economic consequences.

Wouldn't surprise me. When the BEA investigated an Air NZ hand-over test flight (a German airline had been leasing the aircraft and the German pilots were at the controls with Air NZ crew observing) the report pretty much focused entirely on issues with the crew, despite the fact that, according to their own report, the cause of the crash was damage to AOA sensors when French maintenance crew failed to protect the sensors during a hose-down.

The report basically totally ignored the French company that caused the accident and instead fixated on the German and NZ crew.

 

[a_unique_person]

I don't think anyone can doubt the BEA this time. The easiest thing would have been to leave the aircraft where it was. Instead now we know exactly what happened.

 

[LondonJohn]

I don't think anyone can doubt the BEA this time. The easiest thing would have been to leave the aircraft where it was. Instead now we know exactly what happened.

It's really not that simple. Had the aircraft been left where it was, there would have been a massive - and justified - uproar from victims' families and across the aerospace community. The rough location of the crash had been identified by other aircraft which had spotted floating wreckage. The initial searches were in fact carried out by the Brazilian Navy, on account of the number of Brazilian victims. Similarly, the French Government (not the BEA) funded and coordinated the French rescue efforts, on account of the number of French victims and the fact that the aircraft was operated by the French national flag carrier.

As it happens, the final search that located the CVR and FDR was primarily funded by Airbus. The manufacturer had a huge commercial interest in finding out the cause of the accident: Airbus wanted to dispel rumours that it may have been caused by either mechanical failure or design faults. It's not hard to see that even rumours of such things being causes of the crash would have had a negative effect on Airbus' image and potential future sales.

So the BEA could not have stymied the search and recovery effort, even if it had wanted to. And in any case I was never asserting that the BEA would have been motivated against a full search and recovery effort. What I did assert was that the BEA has a habit of interpreting the available evidence in a manner which tends to cast the best possible light on Air France pilots, mechanics and other employees. That's what the BEA appeared to do at a number of points throughout this investigation: they even tried to claim the final words on the CVR were inaccurately reported (under the partial guise of "distress", but probably chiefly because it indicated that the pilots were panicked and out of control of the aircraft).

 

[LondonJohn]

By the way, the archetypal instance of a national regulator seeking to protect the national flag carrier and its personnel was the investigation into the Air Egypt 990 crash. All the data show with near-total certainty that the cause of the crash was the deliberate actions of the relief First Officer, who flew the aircraft into the Atlantic Ocean in an act of suicide and murder. Yet the Egyptian air accident investigators seemed unable and/or unwilling to believe this cause, and instead insisted that there had been a mechanical failure (even though no possible scenario that involved mechanical failure could be reasonably married with the data). In this instance, the Egyptian investigators made the BEA look like models of probity and objectivity.

 

[LondonJohn]

Pilot, not pilots. The Pilot Not Flying tells Bonin exactly what to do, Bonin indicates he is complying.

Robert gets distracted because he is trying to get the Captain to come back from his break, and misses that Bonin is doing exactly the opposite of what he is being told to do.


Bonin's wife was on the flight, which could help explain the confused and panicked response.

But to me, Robert is equally culpable here. He was the senior of the two co-pilots, and should have therefore had no problem pulling rank over his junior colleague (had it been the other way around, the failure of the more junior pilot to intervene may have been more understandable, even though modern CRM training instructs junior pilots to countermand even the captain if they feel that the aircraft is being dangerously flown).

In this instance, Robert seemed to have only marginally more insight than Bonin into the situation. Yes, Robert was seemingly aware that climbing was not the correct course of action. But he seemingly remained unaware of the spatial/aerodynamic issues that were so critical to this crash, and appeared to be ignorant of the disconnect between the aircraft's attitude (which he could easily have divined with no more than a glance at the large AI right in front of his nose) and its vertical speed/altitude.

What's more, Robert actually took control briefly (at 02:11:32), yet kept pulling back on the stick (the FDR records individual stick inputs, so we can be sure that it was Robert who was trying to push the nose upward). If Robert was any kind of decent pilot, he would have had no trouble adding up the various sensory inputs (unmissable stall warning with stick shake, even a cursory glance at the plunging altimeter, even a cursory glance at the AI to determine the nose-up attitude), and would have applied the correct remedy - which, as I've said before, is the most basic airmanship knowledge. And Robert also advocated climbing later on, once the captain had returned to the cockpit (02:13:40). So he was also guilty of massive errors of judgement and knowledge at critical times leading up to the impact.

The one thing which very likely muddied the water was the way in which the pilots may have placed blind faith in the Airbus' FBW systems acting to keep the aircraft inside its flight envelopes. And there's a distinct possibility that they were insufficiently versed in the differences between normal law and alternate law flying. But I can't help coming back to the most basic conclusion: it's almost beyond belief that the two pilots (whether between them, or because one pilot took control from the other) failed to take the most basic remedial action to correct a situation that was explicitly playing out before their very eyes on their instrument panels.

Make no mistake, this crash was almost entirely due to pilot error on behalf of both co-pilots, and the captain should probably also bear a small part of the responsibility for his failure to take decisive command when he could see that the two co-pilots were unable to fly the aircraft. To my mind, after the captain asked his initial questions on returning to the cockpit, he should have immediately booted Robert out of the left hand seat and resumed control. As it was, the captain spent two and a half minutes before the crash without touching the controls. And, what's more, he even professes that he doesn't know what to do! Had Dubois taken control within 30 seconds of returning to the cockpit, it's highly likely that he could have recovered the aircraft in time by looking at the instruments with a fresh eye, and simply putting the aircraft into a slight nose-down attitude at maximum thrust. As captain, he could and should have taken responsibility at such a critical time.

 

[a_unique_person]

But to me, Robert is equally culpable here. He was the senior of the two co-pilots, and should have therefore had no problem pulling rank over his junior colleague (had it been the other way around, the failure of the more junior pilot to intervene may have been more understandable, even though modern CRM training instructs junior pilots to countermand even the captain if they feel that the aircraft is being dangerously flown).

In this instance, Robert seemed to have only marginally more insight than Bonin into the situation. Yes, Robert was seemingly aware that climbing was not the correct course of action. But he seemingly remained unaware of the spatial/aerodynamic issues that were so critical to this crash, and appeared to be ignorant of the disconnect between the aircraft's attitude (which he could easily have divined with no more than a glance at the large AI right in front of his nose) and its vertical speed/altitude.

What's more, Robert actually took control briefly (at 02:11:32), yet kept pulling back on the stick (the FDR records individual stick inputs, so we can be sure that it was Robert who was trying to push the nose upward). If Robert was any kind of decent pilot, he would have had no trouble adding up the various sensory inputs (unmissable stall warning with stick shake, even a cursory glance at the plunging altimeter, even a cursory glance at the AI to determine the nose-up attitude), and would have applied the correct remedy - which, as I've said before, is the most basic airmanship knowledge. And Robert also advocated climbing later on, once the captain had returned to the cockpit (02:13:40). So he was also guilty of massive errors of judgement and knowledge at critical times leading up to the impact.

At 2:11:32, according to what I can see, Robert took the controls, and Bonin took them straight back again. Robert didn't get to make any inputs. Robert said several times, more or less, to just keep the plane flying straight and level, but he was distracted with trying to get the captain.

 

[LondonJohn]

At 2:11:32, according to what I can see, Robert took the controls, and Bonin took them straight back again. Robert didn't get to make any inputs. Robert said several times, more or less, to just keep the plane flying straight and level, but he was distracted with trying to get the captain.

Ah, I should have written 2:11:37. Here's the excerpt from the article (my bolding):

02:11:37 (Robert) Commandes à gauche!
Left seat taking control!

At last, the more senior of the pilots (and the one who seems to have a somewhat better grasp of the situation) now takes control of the airplane. Unfortunately, he, too, seems unaware of the fact that the plane is now stalled, and pulls back on the stick as well. Although the plane's nose is pitched up, it is descending at a 40-degree angle. The stall warning continues to sound. At any rate, Bonin soon after takes back the controls.

 

[Pulvinar]

Was there any indication that Bonin was suicidal? Other than his actions here, that is. It might have been an opportunistic "way out" of a marital problem.

 

[Daald]

Was there any indication that Bonin was suicidal? Other than his actions here, that is. It might have been an opportunistic "way out" of a marital problem.

From the last thing he says on that transcript, I would say no. He was clueless to the end.

02:14:23 (Robert) Putain, on va taper... C'est pas vrai!
Damn it, we're going to crash... This can't be happening!

02:14:25 (Bonin) Mais qu'est-ce que se passe?
But what's happening?

02:14:27 (Captain) 10 degrès d'assiette...
Ten degrees of pitch...

Exactly 1.4 seconds later, the cockpit voice recorder stops

 

[Skwinty]

Before disrespecting the pilots please read the following link.



http://www.wired.com/autopia/2011/07/an-airbus-captains-take-on-the-air-france-disaster/



"
The key ingredient most everyone seems to be overlooking: The flight control laws of an Airbus.


And here’s where the pundits really don’t get it. A so-called Transport Category airplane like an Airbus is not required to be recoverable from deep stalls like the one that may have occurred with AF447. They’re supposed to be difficult to stall, and pilots are trained to avoid stalls. In fact, we train to recover from “approaches to stalls,” not “full stalls” or “deep stalls.” So even if the pilots did everything they were supposed to do to recover from a stall, the airplane still may not have been recoverable once it entered a deep stall and exceeded the critical angle of attack.


That’s my best take. In any case, let’s not throw blame at the pilots before the engineers and investigators can sort out what really happened."

 

[Sideroxylon]

That may be, but was it not a serious pilot error that got the plane into the stall situation?


ETA: I am not sure that "disrespect" is a good description of the criticisms have been made so far. To err is human.

 

[Daald]

Before disrespecting the pilots please read the following link.



http://www.wired.com/autopia/2011/07/an-airbus-captains-take-on-the-air-france-disaster/



"
The key ingredient most everyone seems to be overlooking: The flight control laws of an Airbus.


And here’s where the pundits really don’t get it. A so-called Transport Category airplane like an Airbus is not required to be recoverable from deep stalls like the one that may have occurred with AF447. They’re supposed to be difficult to stall, and pilots are trained to avoid stalls. In fact, we train to recover from “approaches to stalls,” not “full stalls” or “deep stalls.” So even if the pilots did everything they were supposed to do to recover from a stall, the airplane still may not have been recoverable once it entered a deep stall and exceeded the critical angle of attack.


That’s my best take. In any case, let’s not throw blame at the pilots before the engineers and investigators can sort out what really happened."

That link is older than the new evidence.

He misses quite a few things.

1. Alternate law.
2. Bonin was pulling the stick back the whole time.
3. The instruments recovered and they had plenty of altitude.

 

[Skwinty]

That link is older than the new evidence.

He misses quite a few things.

1. Alternate law.
2. Bonin was pulling the stick back the whole time.
3. The instruments recovered and they had plenty of altitude.

Link to new evidence ?

Did you read the link about the Airbus flight laws?

 

[LondonJohn]

Before disrespecting the pilots please read the following link.



http://www.wired.com/autopia/2011/07/an-airbus-captains-take-on-the-air-france-disaster/



"
The key ingredient most everyone seems to be overlooking: The flight control laws of an Airbus.


And here’s where the pundits really don’t get it. A so-called Transport Category airplane like an Airbus is not required to be recoverable from deep stalls like the one that may have occurred with AF447. They’re supposed to be difficult to stall, and pilots are trained to avoid stalls. In fact, we train to recover from “approaches to stalls,” not “full stalls” or “deep stalls.” So even if the pilots did everything they were supposed to do to recover from a stall, the airplane still may not have been recoverable once it entered a deep stall and exceeded the critical angle of attack.


That’s my best take. In any case, let’s not throw blame at the pilots before the engineers and investigators can sort out what really happened."

I disagree. Firstly, the pilots didn't even make an intelligent effort to recover from the stall - something that even the most minimally qualified pilot (or even trainee pilot) should have known to try. And secondly, the flight data show pretty conclusively that up until perhaps the final 30 seconds of flight, the aircraft was easily recoverable. It was not in a "deep stall", and nor was it spiralling in a stall turn. Literally all that was needed was a slight push forward on the stick and sufficient thrust. Even removing all control inputs would probably have self-corrected the aircraft.

The post you've quoted is of course correct in that aircraft like the A330-200 a) are not designed to go into stalls, and b) should, in normal law, be incapable of stalling. However, that's somewhat immaterial when it can now be proven that the pilots were presented with absolutely unmistakeable evidence of a stall. So whether or not they thought a stall was impossible, they should have initiated a stall recovery, and worried about the theory-vs-practice later on. I'm sorry, but regardless of the (brief) period when the pilots had no accurate air speed data, the combination of the...ermm.... STALL WARNING!!, plus a glance at the altimeter and AI (which were both reliable throughout, were both right in front of both pilots, and which showed the aircraft in a nose-up attitude yet rapidly losing altitude), should have immediately told the pilots what was happening.

The combination of the CVR and FDR show conclusively that neither pilot had sufficient wherewithal to fully recognise what was going on, and what the simple remedy should be. We know for certain what the pilots were saying to each other, and what their precise instrument inputs were. And together they add up to the inescapable conclusion that both pilots (and, to a lesser extent, the captain on his late arrival on the scene) were grossly negligent and were the primary cause of the accident. The post you quoted might possibly have a point if the data showed that the pilots had tried to take proper stall recovery actions at the appropriate time, but had failed to recover in time (although again the data suggest that a simple stall recovery procedure would easily have remedied the situation). But the data showed instead that the pilots did pretty much the opposite of what was required in the situation. Unpalatable as it often is to ascribe human error to a multiple-fatality accident, there's no way to escape that conclusion in this particular incident.

 

[Skwinty]

I disagree. Firstly, the pilots didn't even make an intelligent effort to recover from the stall
*snip*

So all the air crash investigators can go home now ?