MountainBear, Jando:

I think you both have good points. I've questioned the stall warning logic before.

Irrespective of whether or not it would have made any difference to the outcome of this accident, it seems bad and wrong to me.

I don't think anyone could reasonably have foreseen that it might be necessary or helpful to warn the crew of a civil aircraft that they were so far outside the envelope the AOA sensors were no longer giving useful data.

Until this accident.

MountainBear, with the benefit of hindsight the aural progression should perhaps be SILENCE -> STALL STALL STALL -> FLY THE F**KING AIRCRAFT

It's not something that was considered before, but if you're below 60kts and in the air, those computers should be screaming at you, under (Asimov) Laws 1, 2, and 3.

Ozone smell probably came from the Cb overshoot they were flying near. I believe the top of the Cb was estimated by Meteo France as 52000 feet, and by Vasquez as 56000 feet. Ozone is present in significant amounts near the tropopause and the Cb overshoot reached into the ITCZ tropopause by 6,000 feet according to Vasquez.

The PNF seems to have recognized what it was.

@ Jando - was the real aircraft airspeed < 60kts or was this an erroneous reading due to icing? It is a pity if it is the case that when airspeed is unreliable, or all 3 ADC disagree or are out of range, the stall warning computer decideds that the aircraft is not in a "real" airborne situation and decides not to play ball when the situation actually may be recoverable.

It would appear the absence and presence of the stall warning may have added to some of the confusion on the flight deck, though on some aircraft and dependent on pilot technique, the stall warning or stick shaker can be activated during recovery from the stall. But maybe these guys had never experienced this and of course its always easier from the armchair.

In system design, probability is regularly used when deciding how safe a system needs to be or how complex the logic should be. It may be that loss of all air data was of a suitably low probability to not be considered in the stall warning inhibit logic. As someone else said earlier maybe this is one of those 1 in 10000000 events.

My apologies, I've oversimplified it into "untrue" category. I've only described what it looked like to me when I'was flying the beast. It isn't pitch hold, rather it's flight path hold yet with no speed excursions or no vertical windshear, it will result in maintaining constant pitch.

I've flown 320/19 with MSNs in low 1000s and never spotted the tendency to pitch up while handflying high. As DozyWannabe mentioned: this quirk was ironed out of TA FBW Airbuses. Difference between gravity on the Earth's surface and at FL390 is about 0.4%, between the equator and poles is 0.5%, I don't think that having constant 1G value of 9.81 ms^-1 in your FBW was good idea, if it ever was used. Could be miscalibration of vertical accelerometer and that there are some Airbi that pitch down slowly, instead of up. We'll have to turn Mythbusters' attention to the issue.

Because aeroplane is so out of flight envelope that AoA, VS, speed, mach, flightpath angle constantly switch themselves off and on. Could be flight/ground logic that gets them but my french is not that good to understand the whole report. However, baro altitude is constantly recorded from top of descent to end. No spikes or surprises there.

It was real. Pitot 1 unblocks at 2:10:35, pitot 3 at 2:11:08. After that they agree until AoA goes above 30° and everything becomes mess.

Fortunately, it's not true. Stall warning worked just fine after all airspeed was lost. It has thrown in the towel after AoA exceeded 30°. That's area where no flight test was done before - for a good reason.

Despite FWC shouting "STALL STALL STALL". 54 seconds.

If we ever get the answer to that, it won't be very palatable.

Well Airbus manuals would say that wouldn't they.(about pitch attitude & height)
However if you take the guidance out (Autopilot) at 40,000ft on a 332 and wait.. The a/c TENDS to climb. Tried it a few times on different days as was interested after I knocked the disengage button on the side stick button with our ungainly hard plastic tech log.. Noticed the a/c was STARTING a gentle climb.
Whatever, I still believe that an attitude indicated indicating 16 deg NU. At height would look weird to any heavy jet pilot. If that 1.6 g spike is true that probably explains the INITIAL climb.
Again think too much blame pointing at the pilots..(all dead men who can not defend their corner)

Really? I read it as the airline taking the most stick - their training and procedures were either not followed or were clearly a dreadful mess at the time.

The question here is not " at which point didn't "the software" give them what they asked for" (Airbus - and BEA - were very quick to answer that).

The problem here is that, FBW Airbuses are flown differently from conventional airliners. They purely don't give you any feed-back on the side-sticks (and on ATS). And when confronted with a situation of multiple failures, pilots tend to fly by their pants...that is why the PF Co-Pilot voiced his impression about going with a crazy speed.
They were in a turbulent area (no sense of posture), they had no correct speed informations (no complete visual clues) and no feed-back on the side-sticks.

jcjeant shows it all...
Remember when you went to your first flying lessons? What were the first clues our instructors told us about identifying an approach to stall? The reduction on the forces needed to move the stick, the lack of reaction and amplitude required from the controls, the buffeting, etc...
The lack of feed-back (and interconnection between) the side-sticks was IMO a major factor, on the inability of the pilots to properly identify the stalling conditions.
One can always argue about the stall stall synthetic voice warning, but we must remember also, the number of gear-up landings that have been made in the past with the horn ON, all the way down to the tarmac...

Now put that voice coming on when you put the nose down and turn that madness situation into an accident...

Yes, the systems behaved exactly as designed.
Yes the situation was poorly handled by the whole crew.
But when we take away from the man/machine interface such an important clue (feed-back) we are forgetting that those aircrafts are ultimately flown by men and that men revert to basics when in face of an outstanding dangerous situation.

No offence, but your statement is unqualified.

Computer pilots are handed an unfimiliar situation when the AP drops out (remember Roselawn ?).

With less hands-on training for the "men", what do you expect ?

You obviously haven't been following these threads.

According to many prior posts here, the Stall Warning shuts off below 60 kt IAS. Why they have IAS input to SW is beyond me. The DC-10, for example, uses only AOA for SW.

Regardless of IAS, the plane was moving through the air fast enough to keep the AOA vane aligned with the relative wind, even as they were falling like a brick.

@aguadalte - Tactile feedback will come into it, yes - but it's worth remembering that the same thing happened to the Birgenair 757 which had interconnected yokes, and the psychological impact and tragic result was much the same.

@Graybeard - As I said, I suspect that Airbus will be revisiting their logic over this one. However at the same time I think Ranger one was right when he said:

Remember that the stall warning was sounding for 57 seconds before the values went out of usable range and it shut off. How in the hell did that happen?

Unfortunately, that points to the pilots.

Mike X, welcome.

Rhetorical question, my apologies.

Dozy, we're following this on two threads.

I believe that no-one wishes to put forth an opinion that may bury them in the future.

Off for some shut eye. Later.

Some posts conclude with aspects which tend to ‘blame’ the crew.
I suggest that before coming to any conclusion a wide range of ‘human’ aspects should be considered.
The crew were faced with a sudden failure of an airspeed display: This is not obviously an unreliable airspeed – there is no display; thus, no demanding thought of using an UAS checklist. Remedial action might focus on reinstating the display or using an alternative source.
With all airspeed displays failing, again there is no obvious link to anything being ‘unreliable’. The priority action has to be to fly the aircraft, but without any specific checklist, basic aero / control concepts have to be recalled from memory and used in an aircraft where the control characteristics might not be best suited for a ‘novice’ pilot.
Add a stall warning to this scenario, which initially and logically, might have been disregarded as false: – Technically, an inappropriate association of stall with airspeed, but ‘stall’ is shown on the airspeed scale and everyone talks about stall speed, thus without any speed display why believe a stall (speed) warning. This was a mindset which unfortunately was not reversed later even with a valid warning, but there were no other stronger cues to trigger a change of view, particularly in a rapidly developing and very confusing situation.
These humans – the crew, may not have deduced that the aircraft had actually stalled until very late in the event, if at all.

Why did the aircraft climb: The loss of airspeed affects the ADC in a way in which the indicated altitude suddenly reads low (see FDR and previous events). In this instance, the crew focus on altitude recovery – avoidance of an altitude bust – they are ‘the juniors’ responsible to the absent captain and thus have responsibility for a safe and accurate flight. Did they deduce turbulence or another affect of adjacent Cbs. Add this to the developing scenario as above, then apparently inattentive control may not be surprising.
Possibly both crew focused too much of their ‘cognitive resource’ on deducing the ‘technical’ situation and problem solving, flying the aircraft was no longer #1 priority. This might not have been a result of a conscious choice, but due to the surprise and confusion of an unfolding complex event in a timescale incompatible with an ability to think, understand, and act.

The question is what were he control surfaces doing while the PF was practicing his signature?

Hi,

Page 115 french report N°3
http://i.imgur.com/176xQ.jpg

IMHO, and FWIW, I lay the blame for the crash into the sea on the shoulders of the pilot, for not doing the correct stall recovery drill, i.e. unload or drop the nose to recover the airspeed and at 30'000 ft, hang the altitude loss. The fact that the aircraft was descending rapidly is not a valid reason to pull back on the side stick.
The reason for the stall I lay squarely on the shoulders of Airbus/Air France for not fixing the pitot problem years earlier.

"According to many prior posts here, the Stall Warning shuts off below 60 kt IAS. Why they have IAS input to SW is beyond me. The DC-10, for example, uses only AOA for SW.

Regardless of IAS, the plane was moving through the air fast enough to keep the AOA vane aligned with the relative wind, even as they were falling like a brick."

I completely agree. As long as the plane is in the air there is always 60 kts of relative wind, even if the angle of attack is so high that the pitot measure wrong speeds. The fact that they certified the logic behind this is for me a huge mistake. As long as the AOA is not insane it doesn't matter. But if the aircraft is capable of 45 degrees AOA, they should do a logic that also works in this case (irrespective of the issue of displaying the AOA info to pilots).

Also, what matters is not the fact that it sounded for 57 seconds straight. What matters is the fact that pilots push the stick forward and then the stall warning sound. If you were piloting this plane with all the other faulty sensors, wouldn't you start to think that the stall warning is faulty too?

This kind of behavior of the stall warning is only caused by the faulty logic, and the BEA does not seem to want to address this...

No, the speed of less than 60 kts wasn't real. The relative wind speed was always above 60 kts, just by looking at the vario and ground speed it is actually above 100 kts.
So the AOA data shouldn't have been "rejected" at all. The system worked as designed, as stressed by the BEA. But the design on this point is bad, and adds unnecessary confusion during emergencies.

I don't see why we should blame the pilots for discarding a warning system that is behaving erratically and against your intuition (nose up= stall, nose down=no stall).
Agreed it sounded 54 seconds straight and they should have done something way earlier. But they were about to fix the stall problem multiple times, when they pushed the stick forward (and the stall warning sounded again at that time... preventing them from continuing in this direction).

damning report
 

Having read report 3 and the CVR transcript I am no longer surprised by the ineptness of the handling crew and the crash, more so the fact that they actually managed to get the plane airborne in the first place.
they overlooked just about every lesson that one learns within the first few hours of primary training, simply a horrific testament to modern training and modern electronically flown a/c where manual flying is not so much a requrement but actually discouraged. I hope the whole flying community learns from this crash.

what faulty logic? the only faulty logic was the pf who looked at the vsi which was pinned to the bottom, and he couldnt deduce that they were fully stalled out. a very elementary mistake

WX Radar
 

My question!
What was the WX-Radar showing? How big was the cell there going thru? Green, yellow or red? Is there anywhere a reconstruction of the picture?

Actually if you look at the variometer data from the BEA report, it is not pinned to the bottom, but also flickers too.

Conclusion:
- failure from the pilots to realize that the variometer data was almost right even if it flickered
- failure from the pilots to realize that the lack of stall warnings was wrong because they flickered

that's it! pilot failure. case closed.

The only faulty sensors were the pitots, and then only briefly, so to say "all the other faulty sensors" is a bit disingenuous. At this point I believe a competent pilot should disregard all other annunciations other than the thrust settings and the pitch as displayed on the ADI (double-checking the ADI with the standby instrument if you're unsure). Once you've got those in the right ballpark you should know you're going to be OK.

If they didn't want to address it, they wouldn't have mentioned it in the report, but the important point (which goes back to your first paragraph) is this - simply teaching stall warning and avoidance, as had been done prior to this accident, simply isn't enough. Stall recognition and recovery procedures should be a mandatory part of recurrent training, which you'll notice has become something of a hot topic in the industry over the last couple of years.

Again, only the press are saying that the report is "blaming" the pilots. If you read the report itself, it becomes pretty clear that the BEA considers this an industry-wide issue that should be clarified and adopted by manufacturers and airlines alike.

I'm beginning to wonder why the press are unilaterally oversimplifying the interim findings this way - you only have to look at the report itself to see that it is a very dry document that does not apportion blame as such, but does criticise the apparent lack of CRM and flying knowledge exhibited in the flight deck. It isn't blaming the pilots, it's saying there's likely a systemic problem that needs to be addressed - unfortunately that doesn't make for a "sexy" byline.

Flying high is a new experince. High altitudes stalling is not very frecuent To practice.Has anyone here done it¿ I wonder how low density of the air and the inercia of the stall due to the long time stalling affects the recovery tecnique.

STALL WARNING MUST ALWAYS WORK. Its unveliable that on the Airbus with that pitot probe on icing condicions doesnt.

Its funny for me how people blame the pilots because the stall warning sound for 57sg and they didint recognize the stall. Dont you understand that when the stall warning stops means NOT STALL, but the aircraft wad actual stalling.

This is what should be:
Sound of the stall warning should mean STALL
Not sound of stall warning should mean NOT STALL


If you relay on stall warning when the sounds stops should be that you are out of stall. I think they pilots of the Airfrance understood that the stall warning was not working right, but I think they probably felt overspeed due To de strong turbulence and icing they were in.

Apologies if this article is already familiar in this domain, but
The Case For An Angle Of Attack Indicator | Air Safety Week | Find Articles at BNET

seems to summarise quite well the conditions under which an AOA indicator available to the pilots is useful. Good to see it the the BEA recommendations.

DozyW: Was there a UAS procedure training in force at that time? (I honestly can't remember) As I said before, my experience with Boeing ended long time ago in the early 90's and I can't remember training UAS at that time.

Timpara, I agree with you 100%. That is exactly what I was saying in an earlier post. Attitude and speed are secondary to the AOA in relation to the performance of the wing.

From the BEA report:

I can hardly blame the pilots. If something you do silences a warning there's a natural temptation to keep on doing it.

Horrible systems design I say.

One of the problems highlighted by the investigation was inadequate pilot training.
Why don’t the various National Pilot Unions alone or better still in concert devise their own Standards of Simulator Training for different categories of commercial flying; then use this as a benchmark to judge the quality of training given by companies. In this era of transparency publish the results on the web, with a section for the company to reply to any criticism if they wish, and give the Public / Journalists more information to make their Carrier decisions.
Perhaps this is already done; perhaps not. It would at least open the area for more debate and scrutiny.
Just a thought

Anyone know if an English translation of the 3rd interim report in French (the 117 page pdf) is going to appear?

Difficult to grasp the details when my technical French is not up to the job.

Sim training will only take you so far. I refer again to Davis' impassioned plea to get ordinary airline pilots some extraordinary training - in real aircraft.

My original edition does not include this - has anyone got one of the later editions they could copy a quote from? (I read an F/O's in about 1999)


While flying in a 'retirement job' I found myself in a twin Turboprop with a full load of passengers, hopping from Aldergrove to City, at night, in rough and rainy conditions.

The stall warning went off, with shaker and siren.

It's hard to describe how that 7 mins or so of that beaten-into-me-brain pair of warnings affected me, even after thousands of hours in command on Turboprops, and really steeling myself against what had become an almost certain false warning. For a very long 60 seconds, it was almost impossible to believe the instruments.

Apart from the fact we were near lumpy ground, the 447 crew's sensory inputs and scrolling data were I would think, much more confusing. I feel deeply sorry for them, but do feel that they should have nailed attitude and power within a band appropriate to that aircraft as an absolutely fundamental part of aircraft handling.

Stall AOA is ~7deg when high&fast and ~16deg when low&slow.
Besides, AOA vanes are designed and calibrated for situations when air comes from generally front direction, not side or bottom. AOA was severe enough to cause Pitot readings off, variometer indications got flaky, etc.

Nope. It's perfectly possible, as evidenced by previous accidents, to get airspeed below 60kts in a large plane (some poor guys happened to get negative airspeed before they crashed). But I agree that the real airspeed was above 60kts in this particular case.

Certification does not bring much attention to situations so far away from flight envelope. Many airframe designs are considered totally unrecoverable even by heroic measurements when stalled so severely, yet they get certified.

The stall warning was on for long enough to take proper action. It's debatable if the airplane could be recovered if they initiated proper action after stall warning continued for the whole minute.

You, and many posters here concentrate on what happened after the plane has severely stalled. But from a flight safety PoV things were already critically bad. Dealing with problems during "falling out of the sky" phase won't improve safety much (if at all).

What is really important is how it entered the stall and how there was no immediate proper attempt at unstalling it. Aural stall warning was active then, attitude indication was good, vario was good, even speed was good at least on PM's display and backup (ISIS) display.

So, if even, there is some human interface problem pertaining to that phase of flight ain't aural stall warning silencing itself when attitude gets so far away from flight envelope that pitot probes stop working.

Yes, because there had been several UAS incidents with the Thales AA pitots. There was a specific bulletin on the subject if I recall correctly.

@ Feathers, it's due later this week.

What exactly is negative airspeed? Plane flying backwards? :confused:

Did the pitots stop working because of the snow or being away from envelope?

What does it say?
 

Unfortunately the BEA's report isn't available in English yet.

But from the discussions here it seems to still be silent on many things.

Did the pilots say nothing during the last three minutes? Because I have not seen anything mentioned.

The BEA did, apparently, say the plane was fully and completely responsive throughout, so how did they explain the THS being apparently stuck up, ignoring the
down instructions (of which there were at least a few).

Hi ross_M,
At first, because of ice crystals. This issue #1 lasted a bit less than 1 minute and it was cleared before they were actually stalled (but stall warnings were not affected by this fault).

Later in the sequence, alpha went so high (above 35°) that both pitot and alpha probes were out of boundaries (hence, Stall warnings stopped). issue #2 happened 40-50 seconds after the full stall developed.

The main confusion is due to #2 starting too closely after #1, the crew being not able to deal with any of them (neither UAS nor Stall).

it does seem weird that the PF in the RHS was not wearing any kind of seat belt,having warned the cabin crew of turbulence ahead.. i understand Airbus provide a nifty table where the yoke sits on most other aircraft

is it a possible scenario that he had the seat pushed back and was tucking into his dinner? so that when the Auto Pilot/Auto thrust clicked out he had to reach out further than usual to grab the side stick ...plus maybe with a tray full of food in the way ... apologies if this has been done to death before.. just a thought .

at least the Airbus engineers now know an AoA of 35 degrees positive is possible in commercial flight, and it shouldnt pose a condition for the silencing of a stall alarm......:ugh:

I bet there are lot of other limiting software conditions that pose create additional risk at the very edges of the improbable flight envelope , where one needs the info the most.

CVR last minutes of flight
 

I was watching French TV 5 Monde last week, and they aired the actual conversation of the last few minutes of the flight, narrated by an announcer with text superimposed over a graphic of the cockpit. All three crew members were present in the cockpit. It was clear that they were confused and, in fact, could not believe what was actually happening. One crewmember (maybe the captain, but I'm not sure) repeated several times "pull up" and "pull". My French isn't very good, so I was concentrating more on the speech rather than which crewmember said it. Maybe some of our friends on the East side of the Pond have some contacts at TV 5 Monde and can get the video. I can't find it on their website.