It’s certainly easier to stall an aircraft at low speed. However, a stall condition is based on angle of attack, not airspeed. If they dismissed it because of their thoughts about airspeed, then this was an additional mistake made by everyone in the cockpit.
They actually had a discussion about whether they were climbing or descending, despite what the instruments were telling them. This was another error made by everyone in the cockpit.
If he’s unaware of what his copilot is doing, then they aren’t communicating like a cockpit crew should; that’s a very important CRM failure for all of them. Presumably their training on the Airbus A330 includes developing an understanding of the control modes (and an awareness of the mode indicator as you describes upthread). Wondering why the plane isn’t nosing over when I push my stick forward? Wait a minute, this is the A330, with two independent sticks, the guy next to me might be providing input. Robert should have asked Bonin what he was up to, or told him to take his hand off, or declared “I have the plane.”
A check of the AoA meter would have shown that their AoA was insanely high (as much as 40 degrees!), and that the plane was indeed pitching forward in response to Robert’s forward stick input (when Bonin wasn’t pulling back on his).
For a professional pilot, no, it’s not reasonable. NTSB records are rife with incidents in which pilots (and their passengers) died because the chose to disregard/disbelieve their instruments and fly by the seat of their pants in IFR conditions.
Another failure on their part. The control settings do not determine the totality of their situation; as professionally trained pilots, they should have known that full power and a nose-high attitude can correspond to either a rapid ascent, or a stall condition.
Except that account is complete fiction. AF 447 suffered a conventional stall, stayed fully intact and hit the ocean on its belly, with its nose pointing upwards; ie. almost the exact opposite of that account.
What I don’t understand is, why don’t they use motorized control levers that provide feedback on the setting? It seems to me, if the controls on the two seats aren’t mechanically connected, they can (and should) be made to move together using servo motors or something.
It doesn’t make sense to me, but Airbus procedure was to pull the nose up and climb if airspeed indicators became unreliable. Was the copilot who kept pulling back just following his training?
Pilots are not perfect, and they’re not superhuman. Expecting them to reach perfectly accurate conclusions about the state of the aircraft in 4 minutes, under adverse conditions (a turbulent, electrical storm at night, out of contact with ATC, with one pilot intentionally and continuously doing the wrong thing) is expecting too much.
My point was just that, if I am flying along at 0.85 Mach and my airspeed indicator becomes unreliable, and subsequently I receive a stall warning, I might be inclined to dismiss the stall warning as a consequence of the invalid airspeed indicator.
Note that Air France did not provide training for airspeed indicator failure in high-altitude cruise.
A discussion about which instruments have failed and which can be trusted is not an error, it’s a perfectly rational discussion to have when one instrument has failed already and the airplane is not doing what you expect it to. Ultimately, they did correctly determine they were descending.
I agree that poor CRM played a significant role. However, my understanding is the plane did nose over (since the two control inputs effectively cancelled), but that the stall alarm sounded, and Robert instinctively removed his nose-down input. I continue to think that this confusing stall alarm behavior played a major role in the accident. If the stall alarm had not sounded when Robert provided the correct input, he might have then observed the plane accelerate and determined that Bonin was screwing up. As it was, the airplane effectively punished him for doing the right thing, and I suspect this is what prevented him (and anyone else) from figuring out what was actually happening.
The A330 does not have an AoA indicator readily available to the pilots.
When an instrument has just failed, a cross-check of the remaining instruments is not just reasonable but required, especially when your control inputs are not doing what you expect and the airplane systems are doing other weird things.
Yes, but:
The stall alarm was not on.
When Robert attempted to recover from a suspected stall, the alarm turned on.
I continue to think that it is unreasonable to expect the other pilots to have figured out what was happening in the 3 minutes it took for them to crash after the stall began, given Bonin’s inexplicable incorrect inputs, the turbulence, and the counter-intuitive design of the air data system and stall alarm.
Yes, it is theoretically possible for them to have figured out what was happening. But people are not good at logical thinking under pressure. In fact, it’s almost impossible. That’s why pilots train heavily and rely on checklists - so that they know instinctively what to do in an emergency, and don’t have to figure it out on the fly. Unfortunately, they were not trained for this event (high-altitude airspeed indicator failure), nor were they trained for handling it when their copilot other pilot handles it improperly.
Blaming the other pilots because they did not figure out, in 3 minutes under adverse conditions, that Bonin was screwing them all over and the stall warning system was not working as they expected, is not reasonable, IMO.
Can’t the pilots see what input the other pilot is providing just by turning their head and looking at what they are doing on their stick? Eg shouldn’t it have been obvious to Robert that Bonin was pulling back?
Note that, in this photo, the pilots have their seats in the far aft position. When the seats are adjusted further forward so that the pilots can reach the rudder pedals, as would be the case in flight, each pilot’s body mostly blocks the other pilot’s view of his sidestick.
You don’t want one pilot to have to overpower the servo motors if he wants to make an emergency correction to what the other pilot is doing.
Additionally, servos can malfunction just like anything else. You wouldn’t want a short in the feedback/interlock system you are describing to result in the stick going full deflection in one direction and having the plane fly out of control until the pilots can manage to hit the “control stick feedback servo disable” button.
The system Airbus came up with is just fine. The two sticks inputs combine, which makes intuitive sense, and there is a warning displayed when both pilots are providing input, as well as a way for each pilot to disable the other’s sidestick.
Robert could have regained control of the airplane - the problem was not that Bonin’s inputs were preventing him from doing so. He likely would have succeeded in regaining control of the airplane, had the stall warning not gone off when he tried. In the configuration the airplane was in, going to neutral elevator (the result of a combined full up + full down input) would have reduced their attitude and allowed recovery from the stall, no matter what Bonin did (the stall was only maintained because of his full-up input). Robert was essentially prevented from doing so because of the counterintuitive stall warning system.
I read a book a few years ago about how to design machine interfaces. The author was critical of the Airbus sidestick compared to traditional, mechanically-linked yokes. The yoke is the biggest control, reflecting its importance. There’s no question about what inputs the plane is receiving. You don’t need to look over at the other pilot to see what he’s doing, you can feel it. In this accident, both pilots were trying to control the plane. I remember hearing of other cases where each pilot thought the other was doing the flying.
Maybe some of that was FUD spread by Boeing, and Airbuses have a good safety record. Yes, there are mechanisms for one pilot to override the other, but during an emergency I think the important things should be as intuitive as possible. People shouldn’t have to adapt themselves to the needs of a machine; the machine should be designed to work with people’s natural habits and perceptions.
It is a committee airplane.
Any way to find out which committee designed the flight control system and how much high time pilot input there was and how much of that they ignored.
No cites but hanger flying from some airline pilots tell of other things that airbus did with are so counter to universal pilot training and all other large airplane operating parameters that it SHOULD be criminal IMO.
Lets just go with the side stick connection for now.
Lets use only opinions from 20,000 hr plus airline pilots on how good & well planned it is and what they think of using it in an emergency …
These thigs are big & really complicated and even Sully missed several things because the manual is just too big, to counter intuitive and there was not time.
They do not put young low time pilots in these things, or they should not. Lots of lives there. But they want a pilot who all his flying career he was taught that this was what to do and this is what is safe to do and then throw them in a plane that what was safe is now unsafe.
Then the emergency and they sometimes use their oldest and strongest and safest training for all those hours & years and the fact that on page 330 in fine print at the bottom in one place only they say, “Oh you do that & the airplane will break unlike every other airliner in the world.”
If one pilot can provide an input that the other will not notice in an emergency situation that will kill them all, then there is bad design.
Crew management mistakes is just another link that will compound the problems …
Forgive a non-aviator for asking what may be a silly question, but having read this thread, the transcript linked in the OP, and some other commentary, what I cannot understand is why no-one saw the altimeter spinning its way to zero throughout the whole sequence? Surely that would have been a strong indication that they were in a serious stall? Or did that in fact happen, and the problem was that no-one noticed Bonin had the stick back? Amazing that 3 qualified pilots could get things so wrong.
They did see it, and they did correctly attempt to recover from a stall even though the stall warning horn was not going off. However, when they tried to do so, the stall warning horn turned on, and so they stopped, at which point the horn turned back off.
IMHO, this probably confused them enough to prevent them from recognizing the actual problem in time. Instead, they probably thought that something was seriously wrong with the airplane, and did not want to risk making it even worse by “entering” a stall.
I have not seen anything that suggests that they received any training about the operation of the airspeed indicators and stall warning system at very low airspeeds / in deep stalls. If they had had more time, they might have been able to figure out that they were in fact flying so slow that the stall warning system was inoperative - but they only had between 2 and 3 minutes.
The more I think about this, the more I conclude that the most egregious error here was the design of the stall warning system. When the pilots did the right thing, provided the correct control input and actually improved their situation (by lowering the nose and increasing their airspeed), the airplane systems told them they had made it worse (because the airspeed was now high enough for the stall warning system to function, and it told them they had just entered a stall). As a result, they stopped doing the right thing - and died.
It is true that the whole incident could have been avoided if Bonin had been properly trained and/or hadn’t panicked, or if the crew had practiced better CRM, etc. But in the end, the airplane’s avionics lied to them about what was happening. It told them they were not in a stall when they actually were, and when they actually decreased the severity of the stall and began to recover, it told them they had made things worse.
Exactly. I’m just a non-pilot layman, but the Airbus A330 cockpit design seems idiotic to me. The control stick, which should be the most important input device in the cockpit, has been relegated off to the side. If there had been two control yokes mechanically linked together, it would have been obvious to Robert that Bonin was pulling back, because his own yoke would also be pulled back.
Also, this idea of the plane’s fly-by-wire system averaging conflicting pilot inputs seems stupid. Suppose the same type of system was in a car, and the intent was to avoid a pothole in the road. If one pilot swerved left, and the other swerved right (either of which would result in going around the pothole), the averaging of inputs would cause the vehicle to drive right into it.
Finally, the placement of the sidesticks (to the left of the pilot, and to the right of the copilot) seems tailor-made to discourage the development of “muscle memory” if a pilot/copilot were to switch seats, not to mention being forced to use one’s non-dominant hand to fly the plane in such a design.
But wouldn’t any pilot of an aircraft the size of an A330 understand the sort of time necessary for the plane to gain some useful speed, once the nose is lowered?
I fully agree that “Stall warning comes on as airspeed increases” is a highly dubious scheme, and very plausibly contributed to the problem here. But it seems you also need some notable lack of understanding by the pilots for that to have blocked a proper stall recovery. A significant point is that proper airspeed indications were restored almost 4 minutes before they hit the water.
But then the airspeed indications became invalid again, not because of icing this time, but because the plane was flying too slowly for the sensors to register. They were not sure of how fast (or slow) they were flying.
They certainly suspected they were in a stall, but did not know how the stall warning system operated. So, when they tried to recover from a stall by lowering the nose, and only then the stall warning horn turned on, they thought “Shit, I guess we’re not in a stall after all, what the fuck is going on?”
The book that I mentioned used a very trivial example to make a point; doors. If you design a door correctly, you should never need to put the words “push” or “pull” on it. Make a big, flat panel and people will instinctively push on it. If there’s a big, curved bar, everybody knows that they should pull on it. A properly designed machine takes advantage of those natural perceptions.
They have a very good safety record, and no one will rest until it’s perfect.
Aircraft safety is an amazing field. The lengths that we go to to investigate accidents is extraordinary. Planes today are as good as they are because thousands of people have busted their asses to find out why previous planes crashed. These recorders were found almost two years after the crash, under 12,000 feet of water. Airbus planes are very, very safe. It’s now their job to learn from this accident and make their next one better.
