I realise I am rather late at the 'They followed the FD' dining table, but why does this possibility not feature more in the BEA report? Is it not a horrendously poisonous cocktail to present to a confused crew - an apparently 'computer generated so must be right' driven VS demand, totally seductive, when all else has fallen around your ears? Why not a simple 'Select Vertical mode' inhibit?

BOAC

The FD on the bus, even on the A300, is just one of many traps.

Fortunately the airline I used to work for put some emphasys during training on that issue.

@BOAC

May be because PF was really concerned with "Overspeed"...

Procedure AF for unreliable speed:
Select FD OFF
AF447 crew don't followed the procedure ... sad ...

YES, but the bars already disapeared when ASI went bust....and the AP kicked off.

A trap!

One question I have that I think actually determines my layperson opinion on this is why, given the duration of the incident, there appears to have been no vigorous attempt to acknowledge that the response wasn't having the desired effect and to at least articulate alternative strategies. Odd to persevere with failure and not fight harder.

Did the autopilot stall the airplane on bad airdata before it clicked off and handed over to the pilots, or did the PF immediately imply an overspeed, and stall it with the long period of stick back ?

.

Interesting!
Is that behaviour known and clear to an average AB Crew?
Technically I can see the logic behind but on the other hand that could indeed be somewhat misleading if the behaviour is not properly understood by the Crew. And indeed it appears the behaviour was not clear to them because after the intial rapid climb one could conclude they followed the 1400fpm of the FP.
It might at least be a good idea to automatically disconnect the FD in the same way as the AP in case of UAS so that it deliberately has to be reengaged. This might help a crew that failed to properly follow the procedure for whatever reason.

Altogether I do not really share the feeling that the report put too little emphasis on the FD. These reports are always very indirect in language and considering that I felt they highlighted the FD rather prominently. To me the report implies most of the committee also had the impression that the FD played an important role especially during the extermely relevant first minute of the incident.
After that things were pear shaped anyway and recovery was getging more and more unlikely. Once they were at 40° AoA things were set firmly for a bad ending. (Anyone trained in doing a 45° Nose dive at 30kFt in a 200t airliner at night in IMC with uncertainty about validity of instrument readings?)

Audible Alarms/Noise
 

What about the affect of the audible alarms. Doesn't the noise make it more difficult to think clearly? Is it really very helpfull to have the alarms keep repeating? I know that when I am in a stressfull situation i just want everyone to shut up for a bit so I can think. Has there been research done on that subject?

Sorry if this has been discussed before. I have read many of the threads, but alas, not all of them.:O

I would certainly expect so, it's part of knowing one's airplane. But it is one of those transient characteristics that is somewhat "invisible" because normally we're on the way to something else in terms of a selection. On the engine-out on takeoff, we would level off the A320 by selecting "0" on the V/S but when initially rotating the knob it would synch with the current VS, from which an adjustment would be made.

The UAS memorized items require that the AP/AT/FDs are all turned off.

Automatically, (meaning the system does it, not the pilot), turning the FDs off with an abnormality such as an NAV ADR Fault may or may not be a good thing. Solving one specific issue always has a "dozen" other perhaps-unanticipated outcomes so it would have to be studied, just like the THS continued trimming after a stall warning, and the loss of the stall warning (NCD - no computed data) below 60kts. Perhaps it should latch?...but that can't be determined until a full analysis of all possible/knowable outcomes are tested.
Well, there has been input, (by Owain Glyndwr) that recovery may have been possible at various altitudes. Owain indicates that a reasonable "unloading" of the wing/reduction in AoA could be done with a steadily-held 10deg ND and in fact lower down, (thicker air), the recovery would be slightly quicker.

aircarver
 

The PF flew a perfectly flyable, reasonably safe and fairly stable plane into a stall.

AF procedures for "unreliable air speed" were not followed at any time.

In the press meeting the BEA had to answer the same question (recovery possible ?)
Their answer was they don't know .. as this was never tested in real life on a A330
But like they stated it .. it's seems that for the BEA that recovery was impossible .. unless a miracle ......

jcjeant; On reflection, I believe the BEA have developed the Report on the basis that their mandate was to determine the actions and/or events that lead to the aircraft departing the Normal Flight Envelope. The report then covers factual information concerning the LOC through to impact.

Consequently, they are very vague on possible alternative outcomes for which no validated data is available, and it is up to the airframer and regulator to establish procedures that can be used in a LOC situation where the envelope has been "extended".

Why no AOA indicators ? The report indicates the data is available buried several pages down, but it isn't an indicator you can fly to.

Seems pitch was being referenced as AOA when they were wildly different.

The airplane was being nervelessly flown in 'coffin corner' by the autopilot with good air data, to save fuel. Once the pitot inputs were lost, the autopilot threw the chore back to the humans, who were even less equipped to do the task without reliable speed data.

Yet the PF kept trying to keep the altitude that it was almost impossible to hand fly at. Why not a procedure requiring dropping down to a less demanding (altitude) in the flight envelope when the autopilot goes on strike. Screw saving fuel until the other problems are sorted out.

.

Kinda disappointed by PJ's reference to setting FD modes and such.

Fer chrissakes, attitude and power and an understanding of the aero for your jet should dictate what control inputs you make, and not the flight director or its modes.

Again, I do not advocate the crew reverting via a switch to "direct" law.

The final report makes a case for concern by the crew with overspeed. I can understand that. One thing that happens when the mach gets too high is aileron reversal, and this could have been a concern. Nevertheless, the report clearly mentions over and over the inapropriate application of back stick.

A shameless plug for the primitive FBW system and control laws I flew for 4 years - our AoA was related to the gee command. So at the max allowable AoA we could not command more than one steekeeng gee! The nose could not be commanded for a further rise, and the tail surfaces would be commanding for nose down. So we sat there with full back stick and 27 degrees AoA and one gee until we pushed forward on the stick. Falling like a rock, but we could command less than one gee and gain energy and the fight was still on.

Gotta go....

Aerodynamically I could well imagine that recovery was possible until quite late in the sequence. Maybe even down to 10 kFt. After all it is a conventional tail airliner and they were not in a spin. So the Tail was in free airflow. With full ND elevator and maybe neutral or at least less extreme NU trim it is liekely that the Nose would have dropped sufficiently to at some point re- attach the airflow over the wing. Might have taken 30s sustained ND or even more but personally I don't share the pesssimism that it wasn't technically possible.

It is more the practical probability which I would consider exteremely low. At no time the crew appeared in a mental constitution to unambiguously diagnose what was going on and to act accordingly and sustain that action until it yields the results (which could have taken rather long and decided action, much much longer and decidedly than they seemed to be willing to pursue a scheme). It appeared they tried something and if it didn't yield within seconds they reverted the action.

@gums:

In my mind, PJ2 has always been one of the staunchest and unswerving advocates of attitude & power, flight path stabilization, and 'knowing' your aircraft, both aero & systems-wise.

It appears to me that he is only explaining how the system works & what they were dealing with. But I don't want to speak for someone else.

@ The stall:

I flew 3 T-tail air transport aircraft which did not 'deep stall' and one military conventional tail aircraft that did.

The bottom line on this recovery and associated techniques is that none were ever attempted because there was no recognition that they were ever required.

If the term 'deep stall' is not meaningful enough or does not fit the formal definition, or offensive to anyone, call it a 'pterodactyl' or a 'lemon drop'. That won't change it one bit.

Just don't call me late for dinner. :)

gums;

Well, if true, and the Report does spend some time on this, yes, it is disappointing that the pilots in this case actually "obeyed" the FDs, but quite frankly I have never heard of any pilot I know actually blindly following the flight directors and ignoring the instruments.

Guys I flew with, regularly turned the FDs off if they weren't following them - it's just something one does on the Airbus without even thinking about it, so drilled into Airbus pilots is this requirement.

Ever since the A320 accident at Madras, Airbus pilots have been made VERY aware that if the flight directors aren't going to be followed, turn them off, (the original problem concerning IDLE-OPEN DESCENT with one FD on has long since been fixed).

Now if you're following them, you're not going to turn them off, but if you're going to follow them you had better be damn certain in your own aviator's mind where these bars are taking YOUR airplane.

When a few first suggested that the PF might actually have followed the FD I dismissed it - no transport pilot, esp. an Airbus one, would blindly follow FDs without always questioning what the FD is leading one towards - the computers are high-speed idiots, nothing more. We regularly talked of "looking through" the FDs during momentary changes in flight dynamics and we'd wait for the FDs to catch up to the airplane while temporarily ignoring them. That just comes from a lot of hand-flying but you'll never get it from turning the knobs.

OK465 - thank you - you're quite correct - I'm just explaining the system.

It's an ordinary, very good FD that deserves the same respect and treatment any computer system does - "check six", "trust, but verify", hand-fly raw data - it works because it's an airplane and a beautiful one at that; it is not a computer platform or an iPlane.

Henra - we're on the same page - yes, it could have been recovered. No, I wouldn't have expected this crew to try given how cockpit discipline was conducted.

Thanks, PJ, it's what I thot, but some newbies may not realize the fine points. Blindly following the FD's is not always recommended when the steam gauges are telling you that something is awry.

For OK, the jet was not in a "deep stall", but rather "deeply stalled". A nose down stick and maybe even power could have recovered the thing, given 35,000 friggin' feet to do so.

It comes down to realizing what the jet is doing and what you can do about it with the tools you are given.

The complicated reversion laws and pilots thinking they have this and that "protections" is not healthy, IMHO. Ya gotta have a baseline that you can count on without a manual switch to "direct" law, which I do not recommend unless you are Chuck Yeager. And then we have the THS that follows the pilot back stick to achieve the commanded gee, NOT THE AOA!!! So to recover, the pilot had to roll the THS trim down and also push forward on the stick in order to provide the nose down moment required. I was surprised that BEA discussed the longitudinal static stability. Considering that the jet control laws disregard AoA other than warnings and some limits in "normal law", would seem to me that we could look at that aspect a bit more.

gums;

What "de-confuses" the airplane is flying it like any other airplane. I don't think I was unusual in the way I thought about the A320/A330/A340...I was never "aware" that there were "protections"...it was flown like a DC8, or a Lockheed or a B767 - just never took anything for granted. It is astonishing that this should be permitted to change as it is so fundamental to staying alive and keeping others who are with you alive.

Demonstrating one's knowledge of the autoflight should be one of many exercises instead of the way the entire sim exercise is conducted. These days I suspect one risks drawing criticism if one insists on hand-flying some exercises. But the instructors have to sign the sheets saying such stuff was covered, so the sheets have to change.

Raw data, non-FD, manually-flown (including autothrust OFF), ILS and non-precision approaches ought to be regular exercises in recurrent sims.

The beancounters and even ops managers may resist that thinking but indications are, here in the Report and in the US particularly after Colgan, (as mentioned in the Report) that it is time to re-prioritize skills, teaching/training, standards and checking.

This now appears to be a question of adjective versus adverb...:p

Like most of us, you never had to apply the relevant memory items in real life, but for the simulated life you have been asked to get the maximum and just rely on the protections. Anything else and you didn't follow the Published Emergency Procedure ...

Point taken OK, and I had posted a graph of the pitch moment for the Viper a year or so ago. Flew at least two jets that could be "deeply stalled", but pushing forward and keeping wings level with rudder would enable a prompt recovery. In the Viper, we had about a ten deg AoA range that did not allow a nose down effect using full elevator. The 'bus folks claimed that the cee gee for AF447 was well forward of the allowable value, so the jet had a positive nose down pitch moment through out the drill.

I go with PJ a 100% on a few drills flying the jet without all the FD's and otto help. There's no need to be macho and "prove" your manhood/womanhood. Just demonstrate you can fly the damned jet without all the gizmos.

Finally, I re-iterate my instense dislike for the term "protection" versus "limit". For some reason, I found it easier to remember the jet's "limits" than how it was supposed to "protect" me.

As the A330 was never tested in real life for stall and recovery ... the AF447 pilots were (against their will) promoted test pilots when the aircraft stalled
Unfortunately they do not look like they were qualified for the job ...

jcjeant;If they didn't know how not to stall it, recovery was then an even bigger ask.:sad:

Not knowing you were stalled, made the outcome inevitable.:eek:

A3330 stall test
 

JCJEANT...are you sure they never tested A330 stall recovery? Is that possible?

- firstly, this is not really relevant to this accident since the crew did not have the first notion that they were 'stalled', but way back in the dark annals of the multiple threads on this I posited that 20k would be my guess at the absolute lowest recovery altitude and below that they were in yet another 'coffin corner'.

Posters talking about '30 secs sustained ND or even more' need to remember that

a) This a/c was descending at around 10,000 fpm ie from 10k, one minute to impact
b) A pitch change of around 30-40 degrees nose down would have been required to initiate unstall, which would probably have raised the r o d to around 20k fpm
c) Now pull out at xxx g?

BOAC - again not to resurrect things but your numbers were about what my sim exercise produced from FL350 - one was just above FL200 - the others around FL 250 or so IIRC. Descent rates were as high as 17,000fpm. I did not try it lower and now wish I had.

Owain Glyndwr's numbers show a 10deg ND steady pitch attitude, for recoveries at FL350, FL200 and FL60 at 3, 2 & 1 deg/sec AoA recovery rates.

The FL350 scenario recovers between FL250 and FL220 which coincided with most but not all our exercises; the FL200 scenario recovers between FL145 and FL125 and the last scenario recovers about a thousand feet above the sea.

I agree it is not really important in the given case,
just out of curiosity I had again a rough look at the physics behind.

We are probably talking about somewhere between 10 and 20kFt.

I appologise upfront for re- doing the calc.
[Assuming stall speed of ~180kts @200t, no flaps, we need 250kts IAS for a 2g recovery. At 10 kft that corresponds to ~300kts IAS. At an angle of 45 ° this would produce ~22kfpm (300kts x sin(45°)). If we assume linear acceleration (which I admit is not 100% true) that would give us an average 16kfpm.
In 30s that would mean a drop of 8k.
Edit : At 30° - 35° RoD would be 16kfpm. Simplified average 13kfpm.
This would give us 6500ft drop.

Now the 2 g recovery:
R = v^2/a: with a = 9,81 m/s (Out of the 2g maneuver capability @300kts TAS (=250kts IAS) one g is required for Sir Isaac Newton). That gives us a radius of
2430m. At an angle of 45° the corresponding lost altitude would be (1- cos(45°)) x Radius. That would gives us an additional 710m (2370ft)].
Here I err on the conservative side, as recovery in the given example would have been below 3kft and thus 250kts IAS wouldn't have meant 300kts TAS but rather 260kts.
(Edit: 1500ft at 35°)

So we have lost somewhere between 6,5 and 10kft for acceleration to a speed where recovery could be executed and between 1,5 kft and 2,5kft for the recovery itself. However it would be easy to start to early with the recovery and run into the next stall.

However, from a purely theoretical PoV somewhere close to 10.000ft might have been possible.
Apologies again for beating this poor dead horse again.

But I fully agree this has no practical meaning in the given case.

- I admit I cannot understand these figures - with an AoA of 35-40 degrees, and a pitch attitude of ?13-18 degrees? an instantaneous change to 10 degrees nose down still leaves you with an (admittedly improving) AoA (ball park) of between 12 and 17 degrees. ie no 'instant' unstall. How long do you wait? Are we not confusing attitude with Aoa here? Go back to ?mm's? early AoA pics and overlay a 10 degrees nose down pitch.

I am also uncertain how representative a sim exercise would be with AoAs of that value - can they even be achieved and is the software up to the job?

I was hoping to stay lurking until I had time to read the whole of the final report, but since several people have referred to my theoretical study I thought I'd better chip in.
Yeah I agree the process is not intuitive. I gave my best shot at an explanation in Thread 8 post # 175. Not an 'instant' unstall though - it takes quite a while to get the AoA down to sensible levels. I also fully agree with the statement that this represents theoretical possibilities and that the psychological pressures on the pilot might well inhibit maintaining the necessary ND attitude for long enough. And no, I think I understand the difference between attitude and AoA well enough.
As for the BEA response - I think they said they didn't know because nobody had done the sums or (certainly) any relevant flight tests. They didn't (AFAIK) say it was impossible.

More errors on BEA website.
 

Another (minor) error on BEA website 5 July 2012 press briefing

In the English version of the Summary, Paris time is stated to be 5 hours ahead of UTC. French, German and Brazilian (Portuguese) versions appear to be correct.

They have also transposed the links to DE and BR summaries with http://www.bea.aero/fr/enquetes/vol....let2012.br.pdf under the German flag and http://www.bea.aero/fr/enquetes/vol....let2012.de.pdf under the Brazilian flag.

Hi DozyWannabe,

Your comment on 19 April http://www.pprune.org/tech-log/48235...ml#post7143575 post #94
was amazingly accurate.
It explains (to me) the pitch attitudes PF attempted to follow and his errors.

No - in fact it's for that very reason that the values are constantly compared and the AP will disengage if there's a discrepancy. The aircraft did not stall until it reached the apogee of the zoom climb, at which point it had been under manual control for almost a minute.

That's the big question, and judging by the content of the final report there's not enough clear-cut information to provide a definitive answer there. Instead what we have is a series of possibilities, each of which has to be eliminated. Those giving the report short shrift because it does not provide a definitive answer are missing the point - it's not the fault of the investigators, it's that there was insufficient information available to provide a simple answer.

For one thing, I don't think it's a coincidence that a push for flight-deck CCTV monitoring began around the time the final report was being compiled.

Why thank you. I feel compelled to admit I wasn't the only one pointing that out, however. :)

That said, I don't think the FDs alone were the cause of the PF's pitch-up commands*, just as I don't think fear of overspeed or expectation of being able to rely protections would do it in isolation. I suspect it was a combination of any or all of those factors and possibly more.

This is what irks me about some of the criticisms of the report in this thread - because if the accusations of the BEA trying to protect Airbus and AF by pinning it on the crew were true, they could easily have taken a look at the probable FD behaviour, stated that the crew contravened procedure, stalled and crashed the aircraft and left it there.

Instead what we have is a fairly exhaustive review of all the possible factors that led to the accident, inclusive of shortcomings on the part of the manufacturer, operator and the industry as a whole - there's even a whole section devoted t othe handling of the recovery operation and how to rectify mistakes made there. While there's a lot of material dedicated to how the crew mishandled the situation, the findings and conclusions only relate that factually. The implicit reprimands seem to be largely directed at the industry as a whole.

[* - The reason for this is that we don't know precisely when they re-appeared, how long they re-appeared for, we only have a theory on what they might have displayed based on systems behaviour - and probably most importantly, if the PF even saw them when they did re-appear on each occasion that they did. On its own it's an interesting rather than compelling theory, but in concert with everything else working against the crew it's definitely something that needs to be eliminated in future.]

Hi DozyWannabe,
the times and modes are here
http://www.bea.aero/docspa/2009/f-cp...nexe.02.en.pdf
I hope Airbus changes the FD logic. If the FDs are withdrawn automatically, then they should remain withdrawn until the pilots reselect them.
To have them reappear automatically in a different mode to that which the pilots originally engaged them, can be very confusing.

Agreed. That behaviour is far from optimal and should be altered regardless.

That said, while the times and modes are displayed in the annex, that still only provides an estimate of what might have been displayed. Taking a look at the timeline, the FDs reappeared twice during the climb phase of the accident sequence - first in ALT CRZ* mode, and subsequently in VS/HDG mode. They then reappeared during the descent phase several times.

Here I need a bit of clarification - in this mode, will the FDs lock to the current vertical speed or the selected vertical speed?

Hi DW,
The reference to "vertical speed selected" is what the FDs automatically acquired on automatic re-engagement.
The reference to "vertical speed" is what the aircraft is actually doing.

Level D simulators generally have a touch screen instructor station display. On the systems/flight diagnostic page there is an icon labeled "INPUT GUIDANCE".

Selecting this page allows you to choose a specific internal software parameter, such as ALPHA, and have it 'artificially' displayed on the ND where VOR DME would normally be. This function allows two different parameters to be selected at any given time, i.e. VOR 1 DME & VOR 2 DME could be set to display ALPHA & BETA. (A poor man's AOA :))

The ALPHA values at any point are very consistent with those from the FDR, a credit to the Sim manufacturers and the value of the 'predicted' data provided with the software 'flight package'. (BETA is another story)

I believe the mathematical analyses of recovery altitudes assumed a constant average nose down pitch rate (based on actual pitch rates generated for short periods from the FDR data) and a specific constant 'G' applied at a given speed on the pull-out. Correct me if I'm wrong.

This is all well and good if the FCS is 'agreeable' to providing those pitch rates over the extended period of time encompassing the full exercise. In other words, the analysis is, at least partially, independent of the specific flight control system as the SS is held forward for longer and longer periods, the 'assumed' long term pitch rate being extrapolated from short periods.

It may be entirely representative...:confused:

This is correct for the entry and exit manoeuvres, but for by far the longest time the aircraft was assumed to be held at constant pitch attitude - in most cases 10 deg ND but other attitudes were looked at. The thinking was that attitude was about the only thing he could rely on and hold easily. 10 deg was a "for instance" value which might not have been too extreme to be believable, and was held for a long time just to see if it would have been effective. So your next point is OK - the sums did not depend on the ability to hold specific pitch rates over an extended period and weren't really any sort of function of FCS except insofar as the entry rate of pitch was concerned, and even there one could show that the entry rate did not significantly affect the outcome.