Sounds good....

We discuss here a pilot who demonstrably has lost the "plot", and his partner, who evidently was easily pleased that things were acceptable, in the midst of a wild ascent.

Are you willing to suspend your strict definition of "pilot"? In the interest of entertaining a possibility? Once the climb established, and control was lost (i define upset as three seconds after loss of a/p and at first SW), it becomes difficult to retain textbook definitions, yes?

What manner of 'control' involves the stick venturing to and fro, a constant (net) NU, and Stall Warn bracketing a climb of three thousand feet in restricted airspace?

Can we agree that loss of SA may have at least potentially involved spatial disorientation?

rgds

Hi Lyman,

How did he manage to control roll if he was not looking at the attitude indicator? How did he determine that he had lost (lateral) control at 02:11:31? How did he manage to maintain 15° NU until he also lost longitudinal control?

Because he had confused the two axes, as the aircraft had taught him to, in the thirty seconds it took to kill them all? Have you never fixated on one cue, ignored another? Believed one, and not another? Always trusted dials and completely eliminated sensory? I think he fixed on Roll.

Isn't it clear that whatever he saw v/v Pitch, he pulled anyway? Aren't we looking for reasons to explain this lapse?

This is my point. In AL2b the aircraft behaves schizophrenically, by definition. It required a skill that this pilot had not only not practiced, but may not even have been aware of. Machinbird has explained this, and his comments are instructive. I think Roll had PF totally consumed, so by inferring that he saw AI and "should have" mastered AOA, is presumptive. This flight was lost in the first fifteen seconds, and simply by degrading into a Roll mode that demanded the complete attention of both pilots, such that they lost the awareness of flight path that may have saved them.

No one needs 'twitchy', not in those circumstances....

Hi Lyman,

You don't address my third question: How did he manage to maintain 15° NU until he also lost longitudinal control?

'Maintain' 15 degrees suggests he could suss attitude, and that 15 was what he wanted. That also happened post Roll issues, no? Given that he may have selected TOGA, is it possible he was comfortable with a) 15 is acceptable, the a/c is Stall protected, or b) he was flying "g pants".

Maintain is the key word, for it suggests that however he sussed the flight path, his Pitch was 'appropriate', in his opinion. He also may not have been aware of the value.

Hi HazelNuts39,
Probably by following these erroneous FD commands.

Time FD MODE
2 h 10 min 08 ALT CRZ HDG
2 h 10 min 08 - 17
2 h 10 min 17 - 21 ALT CRZ * HDG
2 h 10 min 21 - 26
2 h 10 min 26 - 36 V/S +6000 HDG
2 h 10 min 36 - 42
2 h 10 min 42 - 43 V/S +1400 HDG
2 h 10 min 43 - 47
2 h 10 min 47 - 2 h 11 min 40 V/S +1400 HDG
2 h 11 min 40 - 2 h 12 min 52
2 h 12 min 52 - 2 h 12 min 58 V/S +1400 HDG
2 h 12 min 58 - 2 h 13 min 57
2 h 13 min 57 - 2 h 13 min 58 Not recorded- period of associated
parameter sampling insufficient
2 h 13 min 58 s - end of flight
See Page 95 Final report.

Page 198 Conclusions:
"The Flight Directors did not disconnect.
The crossbars disappeared and then re-appeared on several occasions, changing mode several times."

The previous few posts have highlighted IMO the single minded focus the PF had in over-controlling the roll (PIO) at the expense of paying any attention to pitch attitude, vertical speed or altitude. At the back of his mind was his initial introduction to all the good things associated with Airbus FBW, i.e.

In Normal Law it is a protected aircraft with three thresholds incorporated in the protection:

• Alpha Prot(ection), which is the maximum attainable stick-free AOA. The auto-trim stops there because there is no valid reason to fly at such a low speed for a lengthy period of time; the speed brakes, if extended, retract automatically.
• Alpha floor, which is the AOA where engine thrust increases to TOGA even with auto-thrust selected off.
• Alpha max, which is the maximum attainable AOA with the side stick held fully back.

It took 18 odd seconds for the PNF to announce (I suspect rather quietly), "ALT LAW, protections lost". During that time the PF had just assumed that provided he kept the SS back, the longitudinal stuff would look after itself. It didn't, and with less than 50 seconds of the flight remaining, his hallelujah announcement, "But I’ve been at maxi nose-up for a while", kind of says it all.

None of the above excuses the lack of CRM or lack of reference to the QRH, when it was patently obvious that they knew they "had lost the speeds".

Just another question that occurred to me, to long-range wide-body pilots in particular:

Shortly before A/P disconnect, the CVR recorded this exchange:

2 h 07 min 01 (PF): See, we’re really on the edge of the layer (and under the squall)
2 h 07 min 05 (PF): I mean I’m sure that with three six zero no standard would be good, we’d be ok eh (j’veux dire je suis sûr qu’avec un trois six zéro non standard on serait bon, on serait pas mal hein)
2 h 08 min 03 (PNF): Don’t you maybe want to go to the left a bit?

At about 2 h 10 min 26 the airplane was climbing through FL360.

Could the PF have been looking outside and, aquiring some visual references, attempted to fly by those?

mm43, hiya....

I think that "50 seconds of the flight remaining, his hallelujah announcement, "But I’ve been at maxi nose-up for a while", kind of says it all."

Was a directed response to a request (urgent?) to climb. As such, PF tips his mind set, and it is damning. He is unwittingly saying that keeping the stick full back has not resulted in....a climb. Which means that he thinks it should, though obviously it has NOT. It also (parenthetically) shows us that all three, AT THIS POINT, do not know what is happening. Sorry, they know WHAT, they do not know HOW.

HazelNuts39

Are there any widebody Captains still here? Because I am tempted to ask who in the world would fly visual here? Aren't they discussing Radar returns?

Lyman,

Good analysis!:ok:

HN39; Lyman;

I agree with that "good analysis".

It doesn't quite explain why keeping the SS back on approach to the stall with SW echoing in his ears, seems to have worked for him.:\

A lesson quickly learned and acted upon again when the SW returned - but in NCD mode.:eek:

Simulating this in simulator ( PPC additional training)
 

As part of my airline's recurrent simulator training, we have been asked to simulate the Air France incident. In order to practice the 'Startle Effect' (and in order to avoid the pilots knowing what is going to happen next)it has been left to the discretion of the instructor on how to simulate this in the sim.

I have some ideas in mind ( Volcanic ash encounter and blocked pitos)however would appreciate any feedback from fellow TRI's.

mm43

"It doesn't quite explain why keeping the SS back on approach to the stall with SW echoing in his ears, seems to have worked for him."

And it cannot, because he is unaware of his altitude. He cannot have been, given the discussion just prior re: temps. It is ironic how the conclusions are in, and present the fantastic as normal, and reality as bizarre..... It is in the nature of directed thought and propaganda, however....

May we hear the CVR?

Thought not. Although one hopes the complete data will be provided to NTSB and FAA. If it is, it will be demanded via FOIA....

Patience.

Technically speaking, the TRE in our sim experiments simply failed 2 ADRs in turbulent conditions and triggered a brief thrust asymmetry to simulate the initial roll. The narrative aspect will likely be at the TRE's discretion.

CVR
 

Bear,

May we hear the CVR?

:mad:

@Andy24
Is it possible to get a little more about your experience?
Which UAS procedure have you used, How much initial pitch did you use, Did you simulate the LOC? Which airline? Which Sim?
Thanks

HN39;

Re, "Could the PF have been looking outside and, aquiring some visual references, attempted to fly by those?",

Almost certainly no, he's not flying or attempting to fly clear of cloud by means of visual references. Reasons: a) instruments, not visual cues would be the primary guidance; b) too dark, (no moon) to see the tops, c) even with a moon it is very difficult to judge height of cloud in front...it would be by chance that cloud would be avoided, d) it's just never done, (as in, it is not considered a viable, legitimate manoeuvre nor is it trained); e) change of altitude or track is never done without an ATC clearance. The pilot-in-command of an aircraft an deviate from an ATC clearance in an emergency where the safety of the flight is at immediate and clear risk. As you know, it is my opinion that the loss of airspeed information and the degradation to manual flight is not an emergency.

Lyman;

"Aren't they discussing Radar returns?"

Yes, almost certainly that is what is being discussed. The conversation I "hear" is just what I would have heard in my own cockpit...very ordinary, very operational...normal.

It is extremely rare that the actual CVR is released. Not even the FOIA will do this.

Usually it is notions like "the public interest" that would cause a consideration to do so and even then it would be heavily restricted and "in-camera" in the court so ruling with only the participants absolutely required by the process.

Hello to all of you, I need one info.

Does anybody have the AF447 book by J.P.Otelli?

If yes, can you tell me is there an english edition or just in French?

Thank you.

BEA report is better in any case

Lyman:
Maybe. It depends on what you mean by spatial disorientation.

You can have spatial disorientation without any "feel induced" disorientation playing a part, a common affliction in a non-motion simulators. It isn't hard to wind up massively spatially disoriented (where you are in the air and what you are doing versus where you think you are in the air and what you are doing) by having a lousy instrument scan and a bit of a ham fist. I offer myself and my first few instrument navigation sims during flight training as Exhibit A. I think the term "utterly fargin' lost" (or words to that effect) crossed my sim instructors lips at least once. Weak/poor instrument scan and possibly rough hands look to have been part of this crew's problem.

Back to your question.

I suggest that between time zero and time five seconds of this event (pitot data deemed bad and all three rejected by aircraft systems) SA itself was partial to begin with.

Spatial disorientation loss, or at least impairment, need not have been in the "feel" domain at all. Over the past years on this topic, it has been my estimation that an initial "scan impairment" was exacerbated by the mental effort being expended upon "what's it doing now," which takes you back to some fundamental systems knowldege and training issues, CRM and systems trouble shooting methods. Won't repeat them here.

Based on what can be put together after the fact by looking at the record of outcomes (pilot actions with controls and aircraft performance), the degree of SA retained remained at best partial
-- one cue being as you mentioned, rate of descent and the attendant cue of decreasing numbers on the altimeter seem not to have registered --
and I'll even assert that SA decreased as the time moved forward from zero to impact.

In shorter terms, the crew started behind the aircraft, and got further behind as the event progressed.

This in turn leads me back to both upset and "out of control flight." Out of control flight is more or less the situation in which you make certain inputs to the controls, expecting response A, but you get response B (or perhaps no response at all).

You don't have to be stalled to be out of control.

In one fixed wing aircraft I am familiar with, you can be in a spiral (high roll rate/rotation descent) with symptoms similar to but not the same as a spin (which begins in a stall and gyrates as the plane falls).

If you make control input A, anti-spin control inputs, what you won't get is a recovery from a spin (nor the actual spiral) since you aren't stalled in the first place. What you'll get is response B which is that the plane stays in the spiral until you either hit the ground or you figure out what is actually happening and put in control input B, which would be recovery from a spiral and then fly away.

Until you recoginze what the plane is doing (Situational Awareness at its most basic) your inputs may not yield the outcome you expect, nor the one you desire.

Because the flying pilot believed that the plane was still flying (SW noise and other inputs notwithstanding), he made contol input A, kept making control input A even after a few responses to his crew mate on him going up, and he kept getting response B.

The "aha moment" of "we are stalled" never arrived.

SA was never achieved in any meaningful way beyong the pilot not flying
-- observing that the protections had been lost
-- offering corrections to PF in re "stop going up" that are taken form the CVR.

Because this was a crew, I have to assess the SA of the CREW, as well as individual SA of each pilot who made up the crew.

The sum of their SA was most certainly NOT greater than the two parts. :( The outputs don't suggest either got vertigo, but "disorientation" in terms not "seeing" what their primary flight instruments were showing them is a rather obvious post mortem finding.

Respectfully asserted to those involved in the discussion at this point:
IMO, somatogravatic illusion is a red herring in this particular case.

Lonewolf_50;
I agree.

Of more importance, and to Lyman's observation of the crew's comment regarding having been "nose-up 'for a while' ", the inability on the part of all three crew members to assess as a stalled condition the high rate of descent which could not be arrested by pitching the nose up is, (as well as the absence of effective CRM and the lack of SOPs), set the pathway to this accident.

Gotta go with PJ2. Wolf and Lyman and others......

I have previously praised the aero characteristics of plane that made a stall so benign. Many planes would have been rocking/shuddering or even having uncontrollable roll. Then there's the attitude that " I can't stall this plane", not appreciating the reversion laws and such. Hence the comment about having kept the nose up for such a long time to no effect.

Besides that, what bothers me most is the initial reaction and initial control inputs. Why not just "hold what you have'? So the flight directors may have provided bad cues and the crew blindly followed them.

I, too, would appreciate a post from Andy as to the sim ride using the AF447 initial conditions. Was it set up after the loss of the A/P? Was it set up immediately prior to loss of the/ A/P? How much had Andy read about the whole profile? In other words, did the sim IP not say that this would duplicate AF447 but simply failed some air data inputs to get the A/P to disconnect? and the beat goes on....

@ alogobotur: I do.
I have no clue if it's been published in english, but here (and also here) are some comments about it I wrote at the time.

I concur with roulishollandais: the final report is better (but was released later). The book is an analysis based on what looks like (smells like, feels like) a leaded early draft of the CVR transcript.

@ Lonewolf_50 re: "IMO, somatogravatic illusion is a red herring in this particular case." ==> :D:D

From what I've been told and given to understand - while the stall characteristics are relatively benign, there would be some noticeable buffeting in the stall regime. The problem here is that in that situation it takes a clear head to recognise the difference between the buffeting caused by stall versus the effects of turbulence the crew had been experiencing and were expecting to continue. The obvious way to differentiate would be the presence of the stall warning at the outset and continuing for about a minute, but for some reason this wasn't heeded.

Correct me if I'm misremembering, but wasn't there a school of thought that had the PF misdiagnosing the buffeting - particularly the unusual airflow noise - as an indication of overspeed (or "crazy speed" as he put it)?

Correct, Doze, about the reference to a mach/high speed buffet. I can't read minds, especially those of ghosts, so the comment about "crazy speed" is still puzzling.

I discount confusion about the turbulence, as seems most of use can easily tell the diference between "chop" and an airframe buffet ( high or low speed). My own experience with approach to stall in the bent wing birds is that the buffet/burble/shaking is a lot different than turbulence in a storm or near a storm or even "light chop" you see in clear air when near the tropo boundary or the edge of the jet stream. Departure in the VooDoo was prefaced with noticeable buffet and wing rock. The SLUF also shook and rocked. The Deuce was as smooth as silk and the only thing you felt was a buzz ( could hear it, also). Next thing you knew you had the VVI pegged going down until you released back pressure and moved the stick forward.

and so the saga continues, heh heh

Dozy, I recall a few threads ago some detailed discussion on what the PF may have heard, and what may have fed him a false input into his SA via the audio channel. (If that sound registered, why not the cricket chirp of the Stall Warning, one wonders? )

With that considered, the A330 was bopping along at something near to Mach 0.8, which doubtless has a sound signature to it if you are in the cockpit. As the initial climb was underway, might there have been a subtle decrease in that ambient noise? As I have no time in the front seat of an A330 at cruise, I have no idea how noticeable that auditory cue would be for a given pilot, nor how subtle or obvious a change in noise related to the change in speed as the aircraft slows down before the stall.

Would the air flow, once stalled, of the plane falling at somewhere around 10-14,000 feet per minute have the same audible signature? I don't know. I can't recall if anyone had a good answer for that.

But it's somewhat irrelevant, other than as a distraction to the primary task of instrument flying at altitude. To establish SA, the noise is a secondary cue related to the information on the primary flight instruments. (FPV addressed by gums up a few posts). So even though this issue of auditory data input has been raised, we are back to disorientation in the visual domain, in terms of a good (or bad) instrument scan and seeing, as well as grasping, (or not) the information displayed on the primary flight instruments.

I am aware of how attuned most pilots are in the auditory channel. Our ears often tell us of a state change, or alerts us to start looking for indication of a change due to "that doesn't sound right." So I won't poo poo the idea that the audio domain may have contributed to disorientation in this case.

In my experience, a change in the sound of something (often engines, or for the rotary winged bretheren the sound of the rotors changing as well as engines) gets your attention and gets you busy until you either fix what's wrong, or at least satisfy yourself that you know "what doesn't sound right."

The CVR doesn't give us much in terms of any discussion they had along those lines, but that doesn't mean neither pilot was dealing with that. It will remain an unknown.

Initial pitch-up
 

Quote from gums:
"...what bothers me most is the initial reaction and initial control inputs. Why not just "hold what you have'? So the flight directors may have provided bad cues and the crew blindly followed them."

This is the crux of the matter. What I think we can say beyond reasonable doubt is that the PF cannot have made a conscious decision to climb a significant amount above FL350. That he allowed a sustained climb seems inexplicable.

But one thing still bugs me: that sudden INDICATED loss of 300ft caused by the sudden drop of INDICATED IAS/CAS (my deliberate tautology). Pilots are generally unhappy to find they have lost (or gained) that much height in the cruise. The initial reaction may be to recover it asap, however unnecessary that haste may seem. Could this explain the PF's initial stick movement, if not the magnitude of it? If so, his subsequent preoccupation with roll control in roll-direct, and the startle factor causing him to tense up, may go some way to explaining his apparent inability to recognise his inappropriate pitch-up input, and correct it.

Hello Chris;
Yes, I think there is something to that in examining the data in the initial 12 to 14 seconds of the initiating (UAS) event. However, the altitude indication returned to FL350 within that time frame the pitch achieved was 10deg NU and rapidly increasing. The fact that no one who is experienced at flying these transports would maintain or even continue such attitudes and/or control inputs is sufficient I think to indicate a rapid loss of SA in terms of what transport aircraft are like to fly at cruise altitudes/high Mach numbers. A rapid return to nominal pitch attitudes in the first 20, possibly 30 seconds would have begun the stabilization process.

After the loss of energy and the increase of the AoA to about 6deg, the nose would have had to have been lowered say, to 5deg ND or even lower, for an effective initial attempt at recovery.

The air at cruise altitudes is, as we all are aware, very thin and the damping effects much reduced compared to the thicker air say, at FL200 so a long time would be needed to stabilize the airplane at that point.

At that point, the only awareness that would have saved the flight was an awareness that the wing was stalled. For a pilot who never hand-flew and who was raised and trained on auto-flight systems and who would never have seen a pitch attitude of say, -10 or -15deg, pointing the airplane down that far with an-already dramatically high VS would have been extremely challenging, but it was the only way to save the airplane.

It is gratifying to hear of so many changes being quietly implemented in training regimes and re-arranged priorities as a result of this tragic accident. That said, knowing how to fly a jet transport airplane using pitch and power in all flight regimes, normal and abnormal, is absolutely, fundamentally paramount in this business, period. Computers and autoflight are huge safety tools but are high-speed idiots that are dumber than a bag o' hammers when it comes to actually flying an airplane.

Hi PJ2,

That's a cogent analysis of the first half minute; a longer time frame than I had in mind. I think the loss of situational awareness becomes indisputable round about FL352 climbing, with the absence of any attempt to arrest the increasing pitch-attitude, let alone reverse it.

Quote:
"...knowing how to fly a jet transport airplane using pitch and power in all flight regimes, normal and abnormal, is absolutely, fundamentally paramount in this business, period."

Amen to that. Those of us who have hand-flown jets in the cruise for long periods know only too well that a single degree of pitch represents nearly 1000 ft/min (rather more on Concorde...). Practising it without the luxury of the FD, with or without IAS display, needs at least to be encouraged. That would best include a step-climb at CLB thrust with IAS, and a step-descent.

Lonewolf_50,

I reckon that, as you suggest, the audio signature would have changed considerably as the aircraft slowed down: engine noise gradually becoming more apparent as the air noise diminished. But, whereas subtle changes in sound can attract attention when you are relaxed, I suspect that they might not when the adrenalin is running?

Hi Chris;

Re the longer period - the reduction in indicated altitude lasted about 12 seconds with its lowest point about 2-3 seconds after the loss of speed indication and A/P-A/T disconnection.

So in that sense, even though there was no negative 'g' loading to speak of associated with the indication of height loss, a reaction to correct the loss, about 340' indicated could be construed as reasonable although in my mind unusual.

This pull on the PF's stick occurred within a half-second of the indicated airspeed & altitude loss, reaching 10deg backstick in about 2 seconds, (of 16deg full backstick available or about 65% backstick), then back almost to neutral then 11deg NU again within 6 seconds. The AoA went from about 2deg to 6 about 4 seconds after the loss, and the pitch went from and average of around 2deg to 11deg in the same time period or at a rate of about 1deg/sec with a max 'g' of about 1.5+g. The stall warning blipped at 6deg AoA about 7 seconds after the UAS event.

All this occurred within the first 12 seconds after the initial UAS event.

Those who have flown these aircraft will understand the "inertia" of such a dynamic change in pitch and would know intuitively that it would take some gentle, but quick handling to get the nose back down without significant negative 'g' for those in the back and loss of energy for the airplane.

After the initial twelve seconds, the side stick position varied back and forth but almost all the time within the NU area of stick movement.

" We have been NU for a while"
If the Captain had been told that this was for over 2 minutes... By then it might well have been too late.
PF might not have known. It is difficult to estimate time PAST. Time from NOW is easier.
("When DID I put the egg in to boil ?" Or "I must take the egg out in 4 minutes time.")

When navigating, one was trained to note the TIME of an observation. ( "Was that fix 30 OR was it 35 minutes ago ?" meant that it was somewhat dubious.)

A change of sounds on the F/D may cause one to look around to try and see that " All is Well"... Nothing noticed ?
(The THS was following the PF's order to go more NU than usual ? Altimeters were moving too rapidly to read ? Was it only possible to NOTICE the "10,000ft" because of the change if the first digit?)

Linktrained
Quite possible - analogue instrumentation also uses a different cognitive channel so you can react to visual patterns while reading other numbers. The crew were showing all the signs of cognitive overload: fixation on roll not pitch and flight path. (cognitive tunneling) - disregard of auditory cues - such as the stall warning.

An altimeter needle 'winding down' may have been more of an attention getter for them - and definitely for the captain.

Remember the digits change colour to emphasise abnormal rates...

Are the colors indexed to a particular LAW?

Is there a QCRH? (Quick Reference Color Handbook)

Just the ticket, additional sensory load....

My intuition from the outset was mechanical issue. After nearly four years, it boils down to a change in TYPE, on the fly....

Interface is a species of Mx... Loss of control by definition happened at the first Stall Warn.... None of the three pilots had a good get on the aircraft after that.

Read the recommendations, and then we can talk about PE.

Manual trim is not available with G+Y hydraulic failure, only elevator is being held in last position by the integrator it is like trim through the elevator and after gear down even that is not available. THS cannot be moved without G and Y.

Dozy
And once changed they stay that way. The aircraft was dropping at 11,000fpm or so... An analogue altimeter winding down at that rate would wake up the most tunnel visioned pilot. :eek:

Generally if your altimeter digits are in amber, you have an issue that needs correcting. If they're red, then that issue is seriously dangerous. The "tape" display is also an indication of vertical speed.

When we did the sim experiment, the digits were still amber when we began the recovery procedure - but it was very noticeable. Unwinding "steam gauge" style is definitely a secondary indicator that can be more easily picked up than read, but so is the colour change and the "tape" display.

@Lyman - no, the altimeter digit colour and tape display are not dependent on control law.

Red button
Crash de l?Airbus Rio-Paris d?Air France: la dernière expertise - Challenges
In english (approximate .. :) )
Google*Traduction

Very approximate, I'd say...

:}

We've already covered the ins-and-outs of the "red button" on these threads several times though, so it's not really worth going into it again - because no-one on either side of the argument is likely to change their minds.

Dozy;

As a refresher, the altimeter scale is displayed in white, the digital numbers in green, all against a gray background.

The analog vertical speed needle is green and the scale is white on a gray background. The digital display normally "attaches" itself to the analog needle as shown by the first image.

http://www.smugmug.com/photos/i-6MN9...-6MN9wbF-L.jpg

As shown in the 2nd image, at vertical speeds greater than 6000fpm the analog needle stays at the end of the scale and it is both the digital display and analog needle which turn amber. The digital rate of descent is displayed at the bottom or top of the VS display as appropriate.


http://www.smugmug.com/photos/i-Vc8R...-Vc8RFmq-L.jpg


In our sim exercise (A330) the analog standby altimeter needle was going around the dial once every 3 seconds...THAT gets one's attention.