News French plane lost over Atlantic

[kamaz]

Wait, wait.

So you're saying that the correct response to the stall alarm in normal law is to pull the stick back?

 

[garyw]

Wait, wait.

So you're saying that the correct response to the stall alarm in normal law is to pull the stick back?

What am I waiting for here?

And READ my reply. Actually, I'll bold what I actually said.

He didn't though, He pulled the stick BACK and expected the Airbus fly by wire to save the situation. In Normal Law you can do this and the plane will go nose up and slow down but it won't fall out of the sky.

In fact, let's hear Bruce Dickinson say it as well.

http://youtu.be/IKBABNL-DDM?t=4m10s

 

[kamaz]

Let me rephrase my question:

I understand that in normal law, the plane CAN be recovered from stall by pulling the stick back.

My question is whether this is the CORRECT way of doing this, i.e. are the Airbus pilots really taught to pull back on the stick if they hear the stall alarm?

 

[garyw]

My question is whether this is the CORRECT way of doing this, i.e. are the Airbus pilots really taught to pull back on the stick if they hear the stall alarm?

No. They are taught to lower the nose and increase the speed but they are also taught that in Normal law you can't stall the plane.

Have a read of this:

02:13:43 (Robert) Alors descends... Alors, donne-moi les commandes... À moi les commandes!
Descend, then... Give me the controls... Give me the controls!

Bonin yields the controls, and Robert finally puts the nose down. The plane begins to regain speed. But it is still descending at a precipitous angle. As they near 2000 feet, the aircraft's sensors detect the fast-approaching surface and trigger a new alarm. There is no time left to build up speed by pushing the plane's nose forward into a dive. At any rate, without warning his colleagues, Bonin once again takes back the controls and pulls his side stick all the way back.

Source: http://www.popularmechanics.com/tec...ally-happened-aboard-air-france-447-6611877-2

Robert, the second in command did the right thing. Bonin, the most junior of the crew did the wrong thing. I don't know Bonin's training history but I suspect that he had very little experience of non-commercial aviation beyond the basics required to get his pilots licence and that first foot on the ladder of commercial aviation.

 

[ADSWNJ]

Let me rephrase my question:

I understand that in normal law, the plane CAN be recovered from stall by pulling the stick back.

My question is whether this is the CORRECT way of doing this, i.e. are the Airbus pilots really taught to pull back on the stick if they hear the stall alarm?

This is all wrong! In normal law, the Airbus has full protections enabled, meaning it keeps the plane in the safe flight envelope and you will not stall it.

In normal law, the flight stick works in what seems to me like an arcade mode. If you want to turn left, you move the stick to the left to command the roll angle, then let go of it. The plane then flies a perfect circle, correcting for altitude and yaw, until you do the opposite command. If you want to go up, you pull back to command the rate of altitude increase, and then let go and it climbs. (Contrast a little plane, where you know to increase throttle to increase altitude, as pitching up just slows you down, and pitching up too much stall you).

In the case of this AF447 flight, the plane lost all three pitot tubes (all three off and on, enough not to know what was going on), leading to the avionics not being able to calculate its airspeed (or mach number). At this instant, the plane drops from arcade mode into real world mode (read: Alternate Law), where depending on the exact mode (not shown to the pilot), the stick does different things. In this mode, you are back to being able to stall at any time, and you have to fight any wind gusts yourself.

Also consider this: when in the cruise at close to max altitude, the difference between optimal performance and stalling is as little as 1.5 degrees of angle of attack. So you are suddenly thrown from the comfort of all this automation into a pandemonium of alarms and ECAS warnings, as the plane is being violently jostled in the super-cell storm, in pitch darkness so no horizon for reference, and probably in cloud too. You *must* know the quick-response procedure at this point (CLB thrust / 5 degree pitch), or you will get in trouble very quickly.

So why does the pilot pull back at this point? Because they spend thousands of hours in Normal Law, where the response to decreasing altitude is to pull back on the stick and let the automation correct itself. In Alt Law, this is a potential death sentence.

---------- Post added at 11:20 PM ---------- Previous post was at 11:14 PM ----------

Trouble started at 37 000 feet, which was likely above the weather. As for darkness, that's what the artificial horizon and stars are for...

In this case, they hit the top of super-cell convection over the tropics (ICTZ), where the volume of ice crystals was enough to overwhelm the pitot heaters. Beyond that, they ran straight into the top of the cell (no stars, and a flight deck bouncing all aver the place)

Right, I have no idea why they reduced power and throttled back up several times. Were they experiencing engine problems and hoping that would fix it? But I still don't understand why they didn't pitch nose down. For the most part, they're trying to pull nose up.

With low-slung big engines, full throttle can result in a pitch-up moment that can be hard to control. Throttling back (e.g. to climb setting or lower), can break this pattern. It's therefore OK to do this if you are trying to figure things out. AFAIK, the engines were 100% operational to the moment of impact.

 

[Andy44]

This over-reliance on automation over basic stick and rudder flying seems to be a recurring theme in accidents and incidents. IIRC the KAL 777 that crash-landed at SFO also involved pilots confused at what to do when automation fails to behave nominally.

 

[Urwumpe]

This over-reliance on automation over basic stick and rudder flying seems to be a recurring theme in accidents and incidents. IIRC the KAL 777 that crash-landed at SFO also involved pilots confused at what to do when automation fails to behave nominally.

Actually - it did behave nominally. The pilot just used it for something that it was not designed for and did not realize the results of the nominal behavior until it was too late.

 

[kamaz]

This is all wrong!

Please bear with me, I'm just trying to understand this.

In normal law, the Airbus has full protections enabled, meaning it keeps the plane in the safe flight envelope and you will not stall it.

OK. Note that in Bruce Dickinson's flight Gary posted, he sets throttles to idle and pulls back the stick -- and there is no stall warning. The computer just increases throttles silently.

In AF447 there was a stall alarm, which means that the plane was already outside the envelope (which implies it was already in the alternate law). I posit that a properly trained pilot should have had this situation rehearsed (i.e. he should know how to manually recover from stall in the alternate law).

At this instant, the plane drops from arcade mode into real world mode (read: Alternate Law), where depending on the exact mode (not shown to the pilot), the stick does different things. In this mode, you are back to being able to stall at any time, and you have to fight any wind gusts yourself.

So what you're saying is that in normal law, the stick sets the desired pitch angle, while in alternate law the stick sets the pitch speed (i.e. controls the ailerons), and the pilot is not told which mode he is in?

To use an analogy: you have a car, in which position of the gas pedal normally controls the speed. In alternate mode, position of the pedal controls the acceleration. The car switches from one mode to another without telling you. You keep the pedal 75% depressed to travel at 75% top speed. The car suddenly switches to alt mode, and now you start accelerating. In panic you release the pedal hoping that the car will stop. The computer interprets this as the order to apply zero acceleration, i.e. maintain current speed. You need to use the brake pedal to stop, but you can't remember where it is, because you have never used it.

Brilliant design! /s

So why does the pilot pull back at this point? Because they spend thousands of hours in Normal Law, where the response to decreasing altitude is to pull back on the stick and let the automation correct itself. In Alt Law, this is a potential death sentence.

Translation: the pilots don't know how to fly the plane (because that's what the controls do in Alt mode), they only know how to tell the computer to go up/down/left/right (because that's what the controls do in normal mode).

 

[boogabooga]

It is all easy in Orbiter, when you can rely surface MFD to display the truth all the time. But what if Surface MFD could "lie" to you? We are all used to take Orbit MFD with a grain of salt. But could you imagine Orbit MFD to display complete bollocks?

Sounds like an add-on idea to me...

:tiphat:

 

[Andy44]

Actually - it did behave nominally. The pilot just used it for something that it was not designed for and did not realize the results of the nominal behavior until it was too late.

Oh, okay, but the result is still confused pilots in each case.

The problem with automation is that in addition to knowing how to pilot the plane, you also have to know how to "fly" all the automatic systems, and in addition you also have to know how to tell the difference between the two, how to tell which mode you're in, etc.

Automation is great as it reduces fatigue, both physical and mental. But does it also have a nasty downside?

 

[Urwumpe]

Automation is great as it reduces fatigue, both physical and mental. But does it also have a nasty downside?

You can't have it all. Not automatisation and its advantages and still the same old physical demanding unautomated piloting of the past. You can only make sure that everybody read the manual before attempting to fly a modern aircraft.

Trying to make aircraft idiot-proof will only result in nature developing better idiots. Same with the normal law vs alt law for Airbus: You can't make the warning loud enough that you are in alternative law, that nobody will fail to realize it. Somebody simply will, maybe just for this one single second that really decides it all.

 

[Andy44]

Trying to make <insert noun here> idiot-proof will only result in nature developing better idiots.

That should be on a billboard somewhere.

 

[ADSWNJ]

Please bear with me, I'm just trying to understand this.

Sure - all in the spirit of a good discussion :thumbup:.

OK. Note that in Bruce Dickinson's flight Gary posted, he sets throttles to idle and pulls back the stick -- and there is no stall warning. The computer just increases throttles silently.

In AF447 there was a stall alarm, which means that the plane was already outside the envelope (which implies it was already in the alternate law). I posit that a properly trained pilot should have had this situation rehearsed (i.e. he should know how to manually recover from stall in the alternate law).

If there's a stall alarm, then for sure the plane will be in Alt 1, Alt 2 or Direct Law, as it will prevent this is Normal Law. So yes, as soon as the airspeeds started to disagree (as the pitot froze over one by one), the AP dropped out (2:16 in the reconstruction video), and the plane dropped into Alt Law (2:19)

At this point, very close to "coffin corner" (upper edge of the flight envelope), the plane was manually commanded to fly higher and well outside the envelope where the plane could maintain level flight. At 3:06 in the reconstruction, the stall buffet was first detected by the systems, at which the pilots needed to take immediate action to recover (i.e. drop nose and let the wings install). Tragically, they behaved as if they were still in Normal Law, right to the end.

So what you're saying is that in normal law, the stick sets the desired pitch angle, while in alternate law the stick sets the pitch speed (i.e. controls the ailerons), and the pilot is not told which mode he is in?

Not quite, but close enough. In normal law, the stick controls the bank angle, or the rate of altitude change. In alternate law, it works in different ways depending on the sub-mode (Alt 1, Alt 2, Alt 2a, etc, or Direct mode). Each mode doing similar by different things. **However** the only indication to the pilot is "F/CTL ALTN LAW" on the ECAS, and yellow X symbols onto he primary flight instrument. No indication of the sub-mode of Alternate Law.

To use an analogy: you have a car, in which position of the gas pedal normally controls the speed. In alternate mode, position of the pedal controls the acceleration. The car switches from one mode to another without telling you. You keep the pedal 75% depressed to travel at 75% top speed. The car suddenly switches to alt mode, and now you start accelerating. In panic you release the pedal hoping that the car will stop. The computer interprets this as the order to apply zero acceleration, i.e. maintain current speed. You need to use the brake pedal to stop, but you can't remember where it is, because you have never used it.

Yep - this is a close enough analogy. And imagine this is at night in pitch black, in a storm, with G-forces in the cabin, panic, alerts everywhere ... and the imminent threat to life, and you can only imagine the stress that they were under.

Translation: the pilots don't know how to fly the plane (because that's what the controls do in Alt mode), they only know how to tell the computer to go up/down/left/right (because that's what the controls do in normal mode).

Precisely! You have arrived at that same shakes-head moment that I did a few years back when the black boxes were recovered.

SO ... what to do about it? More automation or less? Or is it a fundamental human-computer interaction rethink?

 

[Andy44]

SO ... what to do about it? More automation or less? Or is it a fundamental human-computer interaction rethink?

Maybe make "alt" mode the new "normal" mode, and define "normal" as old-fashioned stick and rudder?

Kind of a "yes means yes" for pilots, if that makes sense.

 

[Urwumpe]

Maybe make "alt" mode the new "normal" mode, and define "normal" as old-fashioned stick and rudder?

Kind of a "yes means yes" for pilots, if that makes sense.

So back to the point where travelling 15,000 km with just 4 pilots is impossible.

What you all forget: In 99.9% of all cases, Normal law works perfectly and saves lives in quite a few situations which would be extremely demanding in direct mode, that you suggest there.

Also on your "yes means yes": if flashing red lights and an audio warning that the autopilot dropped out and alternative law has been activated is not catching the attention of a pilot, the guy must be Helen Keller. Same with "dual input" warnings - these are also not sounding in a cockpit for testing the speakers.

Is pulling the stick back standard procedure on Airbus aircraft? Absolutely not - if you look at the emergency procedures ("Stall recovery"), you can see that its not different to other aircraft: It calls for reducing pitch and wings level, while putting the throttles to TOGA. (And if you are not in a clean configuration: Prepare for pitch up moment by the increasing thrust)

 

[kamaz]

Is pulling the stick back standard procedure on Airbus aircraft? Absolutely not - if you look at the emergency procedures ("Stall recovery"), you can see that its not different to other aircraft: It calls for reducing pitch and wings level, while putting the throttles to TOGA. (And if you are not in a clean configuration: Prepare for pitch up moment by the increasing thrust)

So -- are we going to pin that one on "did not read the manual", a.k.a. lack of training?

---------- Post added at 12:30 PM ---------- Previous post was at 12:16 PM ----------

Also on your "yes means yes": if flashing red lights and an audio warning that the autopilot dropped out and alternative law has been activated is not catching the attention of a pilot, the guy must be Helen Keller. Same with "dual input" warnings - these are also not sounding in a cockpit for testing the speakers.

If one pilot did not understand the warnings, then you could attribute that to the pilot being dumb.

But here, two pilots failed to understand what is happening. If two people cannot understand the indicators (and a third one does, but it takes him too long), this points to a serious HMI design problem.

 

[garyw]

So -- are we going to pin that one on "did not read the manual", a.k.a. lack of training?

a lack of understanding of the automation and more importantly, what happens what that automation goes for a coffee break.

If one pilot did not understand the warnings, then you could attribute that to the pilot being dumb.

But here, two pilots failed to understand what is happening. If two people cannot understand the indicators (and a third one does, but it takes him too long), this points to a serious HMI design problem.

Or it points to the wrong training, i.e. too much reliance on the automation.

This accident happened at the worst point of a flight, a few hours in, the crew are relaxed, there are storms ahead but it's no big deal then a few minutes later they are bouncing around the sky, warnings are going off, things are going wrong, there is no visual reference and it's LOUD on that flight deck.

And they didn't communicate. If Bonin had said "I'm pulling back to climb, the stall can be ignored because of the flight protections" then I think that the accident would have been averted.

This flight crew simply did not communicate.

 

[Urwumpe]

So -- are we going to pin that one on "did not read the manual", a.k.a. lack of training?

No. I rather think that its a "simpit pilots" view that the other guy simply didn't follow the right procedures. Its easy to come to that conclusion if you are used to sit alone in your simpit, and never in a real big airliner cockpit. All of the active crew didn't follow CRM standards - that made the situation go wrong. In an absolutely non-standard situation human communication simply failed.

But aside of that: You have to ask yourself the question "what did each pilot perceive?" Have they both decided with the same information available?

The tricky part of every airliner is the fact, that left and right side indicators are not getting their data from the same source. So, it is very common that the different pilots see different information. They need to communicate their information to the other side, especially when anomalies occur. Only the left side instruments and the stand-by instrument of the Airbus are recorded by the FDR, but these already disagreed - possibly all three instruments showed different numbers and both agreed to use the wrong one of the standby instrument in the center.

Robert and Bonin had both been thinking that they have to pull up - you can see that in their communication (Robert: "Remonte… remonte… remonte… remonte…"). A few seconds before the first warning sounded, the two pilots decided to fly manually around the thunderstorms and Bonin took control (Bonin: "J'ai les commandes."). Both pilots ignored the seventy-five stall warnings that sounded (continuously for 54 seconds).


Both pilots have not received any training for manual flight in such altitudes - this special lesson was introduced as consequence of the event. The plane banked by 8.4° in a short instance after the A/P failed. Also, Robert was more experienced in the A330 than the captain, but also failed to realize that the aircraft is in deep stall and not recovering (35° AOA) - the coffin corner had caught him as well.

About 20 seconds after the anomaly started, Roberts called Bonin to reduce pitch because they are climbing - Bonin did so, but not enough to avoid the stall, the sink rate increased as they expected, but this was the result of the stalling aircraft.

One minute before the crash, both pilots gave the captain the information, that they are pitching up.

This all happened in 3 minutes and 20 seconds - a very short time, when you are confused.

So, did they both simply not read the manual? Maybe. If the manual contains a chapter "What should you do if you are getting lost in a tropical thunderstorm with possibly all pitot tubes frozen". Both came to the correct conclusion that they have lost control, but both failed to realize that this was really a deep stall and not just a plain sensor failure anymore. Of course there is a manual chapter for that air data situation and both did it wrong - they likely both panicked as the confusing situation endured.

Thus blaming the right seat pilot alone there is sure wrong - as the CVR proves. Yes, just pushing the stick forward with resolve and trusting the artificial horizon would have been enough. But even the captain needed 1 minute to realize what was wrong - despite him likely already feeling the very nose high attitude when he entered the cabin.

IMHO, this isn't an issue about Normal or alternative or WW1-flight law. It is an issue how the situational awareness degraded to the point, that two pilots, including one experienced, got completely confused and left behind in panic. A simple instrument drill would have been enough to reassure what information the crew can trust and what information is possibly unreliable.

Even more: All the key events that led to the accident could also have happened in a Boeing - the captain would only have had an easier time spotting that the crew is pitching the aircraft up. But its doubtful he would have had the chance to react after the aircraft was already in deepest stall at 35° AOA - possibly the horizontal stabilizer was not even having any effect on the attitude anymore at such a high AOA.

 

[ex-orbinaut]

:hmm: I thought the rekindling of this thread might have been the result of some extrapolation based on a very recent Emergency AD (dated December 10, 2014), with precedent in 2012, all Airbus A320 family operators received, regarding actions in the case of two out of three AoA probe failures. Not the same type or situation, granted, but a common component to the A330, too. But no, I see why it was revived, now. Never mind.

Regarding comments on the "ice crystals"
[*] overwhelming the pitot system, I stop to think that the rest of the aircraft also flew through the same icing conditions.


[*]My understanding is that it was more likely supercooled droplets, which form glaze ice freezing. It forms instantly and takes a long time to melt once formed. It also forms beyond wing anti-ice protection areas, streaking back and creating an irregular surface which in cases of serious formation disrupts laminar flow, like in a high AoA condition, even at low AoA. In any case, it certainly would have reduced the available "window", if you will, for regaining control. Far from this being any attempt at exoneration the crews' culpability in this case (clearly something extremely amiss there), the question remains for me; if the icing was that bad and instantaneous, how flyable was that aircraft anyway after entering the icing conditions? I know all about that Airbus quickly demonstrated in a sim that correct unreliable airspeed procedures would have saved the day. I still only think "might" have saved the day. Sims are not infallible representations of reality, either.

:2cents:

 

[Urwumpe]

It is not even sure if the problem was really such an extreme weather condition (which could happen to all aircraft the same) or a special problem related to special pitot tubes by a special manufacturer, which had under-performing heaters that had also been prone to fail without the crew having any chance to detect the failed heaters.

Practically: If you are caught into supercooled droplets over a thunderstorm, that could ruin your aerodynamics in less than 30 seconds, you would also notice other systems fail first - especially the engines, which have also only limited de-icing.

But I doubt that such weather conditions are really able to harm intact airliners. The side mirror heaters of my car need 2 minutes to get from "covered in 3 mm ice caused by supercooled rain" via "dense local fog" to "clean and dry" - and these have hardly the power input of the de-icing systems of an aircraft.