MurphyWasRight, sensor_validation's words led me to think "he" referred to the captain rather than PNF. But, I note further on in sensor_validation's remarks he says PNF was unhappy with what PF was doing, implying he was more alert than the PF.

I won't let you guys have it both ways.

LYKA:

Not clear if this applies to all operators. How much training is typical in an airline environment? Given the importance of avoiding CBs, substantive formal training would seem to be warranted.

Has anyone developed a radar simulator? By comparison with a full flight simulator, it would appear be quite cheap and simple to develop one.

Hi.

All of this is consistent with loss of speed information to the computers (and the flight deck instruments) due to iced-up pitot tubes, and it would appear that the PNF correctly diagnosed this in a little over 10 seconds. The plethora of warnings came after whatever caused the pitch-up and stall condition.

Then why didn't the Birgenair Captain in his 757 do just that in the same situation? This is where the psychological/human factors aspect comes in to play.

And the point I'm making is that the aircraft is, by design, doing exactly as the pilot commands. At this point there is no "interface" as such - the computers are simply doing everything they can to pass the pilot's commands along to the control surfaces.

Anyway - I *really must* leave it there!

DJ77:
The zoom climb and resulting loss of a/s and increase in AoA was the apparent cause of the stall, which the crew either didn't recognize or was unable to fly out of. So, after the loss of reliable a/s, it was essentially the proximate cause of the accident. And the report gives no real hints as to why the PF commanded nose-up. As the FDR didn't record the a/s displayed on the PF's display, it may, in the end, be unresolvable to a certainty.

Although it appears that this accident was a result of incorrect crew response to a UAS, that hardly absolves anyone else of responsibility. Airbus built a flight control system around critical flight data coming from instrumentation that proved, in the end, insufficiently redundant. Thales designed a pitot tube that, although meeting Airbus specs, proved far more prone to icing than did its competitors. AF did not appear to be sufficiently alarmed by previous instances of pitot icing to act to fix the problem with sufficient urgency. And, of course, the training provided to aircrews to deal with the situation proved inadequate.

Lots of blame to go around, based just on what's known so far. I do find it telling, though, that AF seems better at lining up the holes in the cheese than its competitors. That's not to say AF is unsafe; no major carrier is "unsafe" in comparison to competing modes of transport. But no other carrier has managed to have the first instance of hull loss in revenue service and/or loss of passenger life in four different aircraft types (Concorde, A320, A330, and A340.)

The French have managed superlative safety records in comparable high-tech enterprises (nuclear and high-speed rail in particular). AF appears an outlier in several ways.

Why no clarification of ACARS in BEA paper?
 

Hi!
I have been following these discussions on several forums for some years now.and learnt a lot. However, there are some unexplained, and unexpected, messages in the ACARS stream. Some of these seems to be ignored in the current discussion. Why? E.g. 2:11:55-.1/FLR/FR0906010210 27933406EFCS1 X2,EFCS2X,,,,,,FCPC2 (2CE2)/WRG:ADIRU1 BUS ADR1-2 TO FCPC2,HARD.
If WRG means wiring problems or signal problems it is a strange message from a healthy aircraft system.

Wouldn't a "left nose up" command without any change in thrust cause airspeed to decrease?
For a stall warning to sound wouldn't the airspeed signal still need to be valid?
If with no increase in thrust and a nose up input after a stall warning sound along with a climb rate of up to 7000 ft/m cause the airspeed to rapidly fall?

I believe the correct position for the horse is in front of the cart not behind it.

JD-EE
JD-EE and others.
Most of you may not recognize what sensor_validation is doing with his analysis, but if anyone is familiar with the communication analysis techniques of Sidney Dekker in (for example) "The Field Guide to Understanding Human Error" (a highly recommended read) they will recognize the process.

Sensor_validation gets a big attaboy:ok: for those initial efforts.
There is a lot that can be learned from this type of analysis process.

What we are up against here, is that the BEA has provided a brief description of what happened, but not why. In that context the ACARS message you have mentioned is a "maintenance" message that is indicating a fault that occurred at 0210. It may be indicating something that is complicit in what was happening, or a by product. Svarin has spent some time trying to analyse what it was trying to tell us, but until the BEA have completed their investigation, the relationship to this accident and the above ACARS message will remain a mystery.

Having said that, I did mention it a few pages back in relation to what caused the initial upset.

JD-EE:
The only "fact" is that the PNF (recently asleep) was calling the Captain, whether he was seriously unhappy or just following protocol for any unusual situation is not mentioned in the report.

I would suspect it well could be both, however the point remains that he may not have been at the top of his game when things went wrong and thus less likely to be able and or willing to offer suggestions or intervene.

If I understand correctly he could have taken control at any point but did not utill much later.

----

As to Captain taking over the controls, not likely since his body was found, hence not strapped in.

DozyWannabe, all that's correct about loss of airspeed. However, I am an engineer and apparently so are the BEA people. They specifically noted the RHS is not recorded. So they never directly said all airspeed was lost. They could not sit down and directly prove it without "circumstantial evidence".

Yes, but not by that much. We're not told how much deflection went into the "left nose up" input, nor are we told how long it was held.

No, it's as much a function of AoA as anything (while the A330 doesn't have a specific indicator, it *does* have a vane sensor). Stall *protection* requires valid airspeed (which is why the degradation to Alt 2 law).

Not by that much and not that quickly.

I believe that the BEA and Airbus considered the ACARS messages they received consistent with pitot icing. Is your supposition that this was not the case, and if so, what do you suppose it was?

JD-EE - see above.

(Damn, caught posting *again*!)

They were dealing with the FDR and CVR, and it may well be that data will be available from the QAR. I doubt if we will know that until the next Interim Report.

MurphyWasRight - it looked like sensor_validation was drawing conclusions or making inferences beyond the bare facts that I didn't think were particularly valid for the very procedural and ritualized aspects of being in a cockpit - or even being a waiter going on break briefing his replacement.

Dozy
I think you are missing a big aspect of the crew's control problem that night.

If the aircraft is right wing heavy, the aircraft will only fly straight with continuing left stick. How easy is it to separate the roll chanel from the pitch channel in this case? This aircraft was flying in Direct Law in roll.

Think about what this means to the lateral trim situation.!!!
NO Lateral Trim.
NO tendency to return to wings level. You must fly the roll angle full time.
Yes you have enough roll authority to level the wings.
You cannot let go of the stick-It will roll.
Do you see how a guy settling into this situation might apply nose up at the same time???

If not, do this experiment:
Place right arm on table in front of you. Point thumb up.
Put a stick in hand. Tilt stick to left.
With opposite hand, resist the attempt to tilt stick. Do you accidentally tilt stick toward you?? If not, apply more force until you have to cup stick in hand instead of with finger tips.
Get the idea?

Quoth DozyWannabe:
OK, let's look at energy. Potential energy is mass times gravitational force times height, mgh. Kinetic energy is 1/2 the mass times the speed squared.

So we should be able to calculate speed lost from the height gained. "m" or mass is the same in both equations. So we can leave it out. The change in potential energy was m * 32 feet per second times 1500 feet. The change in kinetic energy must equal this. So 1/2 m dV^2 (delta-V squared) is 32*1500. Or dV^2 = 32*1500 or about 219 feet per second which is about 130 knots REAL speed. That needs to be converted to a change in kCAS and I'm not sure of that conversion. But it puts the plane's velocity change within the ballpark of what was shown.

Thanks for tweaking me into doing the rather simple math. They note no change in thrust took place. PF may have expected, for whatever reason, CLB thrust to be asserted. This might be because he moved the stick before the stall warning was played.

The initial storm climb
 

Although not an airline pilot, I have read every post on this thread. There are a couple of things as yet unexplained:
1) there seems to be an assumption that the first (2) stall warnings (just after 2h 10m 05) were spurious, possibly related to the IAS error. Firstly, there is nothing in the BEA statement to suggest they were spurious; and secondly, as the stall warner is fed from the AOA indicators and not the pitots, the IAS error would surely not cause a stall warning
2) the "storm climb" at 7000fpm for 2500 feet is not explained. The PF made "a left nose-up input" not, as he is later stated to be making, "sustained" nose-up input. And there is nothing in the text to suggest that he pitched the aircraft up to the extent needed to climb at this rate on cruise power settings - surely that would have required 20-30 degree pitch angles that would surely have been mentioned? There is also a mention of nose-down inputs in this phase

On the other hand, a big updraft would both lift the aircraft without big pitch-up, and cause the AOA to increase, perhaps setting off the stall warner?

Where this leaves the explanation of the crash I don't know, but we do not yet seem to have satisfactorily explained this first 45 seconds.

Machinbird . . .
 

Excellent example with the stick! That clarified much of the difficulty I was having in understanding the continual trim up. That, combined with the sudden mixed rules of operation in different stick axes . . . in a crisis . . . what a recipe for disaster.

Back in post #260 I made some assumptions about the THS and the AoA. The basic post is repeated below with a small addendum and should be self explanatory.
My interpretation is that the aircraft was in a well established stall (deep stall?) and passing through FL250 with RoD about 10,000 feet/min, and even though both engines N1's were 55%, the pitch-down inputs decreased the AoA and the airspeed data became valid. However, that IAS was low and the stall warning activated again but ceased with further pitch-up commands as the data again became invalid.

Without AoA information, neither pilot had any idea in what part of the stall regime they were, and seemed to react as if a return of the SW was indicating Vmo, hence the continuing pitch-up command. So failure at this time to persist with the pitch-down command effectively left them doomed, as even then with the time taken for the THS to readjust (if it was going to**) and for the wings to start flying, meant that the chance of reducing the RoD to zero before FL0 was marginal.

It strikes me that no-one on the flight deck realized that the AoA data was only valid when IAS was greater than 60KTS. Even though they went looking for the FPV, it seems it was only after the SW stopped at 2:11:40 plus a few seconds, and therefore to no avail - see ACARS FPV messages timed 0212z. Their final chance to grab the FPV page was missed shortly after 2:12:02 when the stall warning sounded again - meaning valid airspeed!

** The THS question?? As IAS had gone below 60KTS, the aircraft was no longer in Alternative 2 Law, but was now in Abnormal Law which means that the THS must be manually trimmed. Even so, in the description given above, I believe that with the low airspeed the elevator was still capable of making an attitude change, provided it received a nose-down command.

This note from the report links the vane angle signals to the validity of the airspeed.

Later when the nose is lowered and airspeed increases back to above 60kts the stall warning sounds again. The pilot responds by pulling and the warning stops as the airspeed drops below the lower limit.

So while we cannot say what caused the airspeed to drop we do know that it happened after the nose up command and after the stall warning.

We also know that the pitch attitude increased to 10° and the aircraft began to climb.

We also know that after the first mention of nose down input the VS dropped from 7000f/m to 700f/m and the airspeed increased sharply to 215kts by which time the aircraft had climbed 2500ft

The first mention of thrust increase is after this initial climb after a further stall warning.

If I bunt at the top of a zoom climb but fail to increase thrust will this cause a stall warning as the g reloads after the bunt?

I'm not discounting anything but I'm not adding information that isn't described in the record of events described in the report.

All fair points.

Firstly, was the aircraft right-wing heavy?

Secondly, I can see how it would be a handful, but the force we're talking about here is not an accidental tweak in the wrong direction. To get the THS up to 13 degrees requires full back pressure for several seconds. This is the problem that is not going to be solved until at least an interim report is out, and anything we suggest is ultimately pointless.

Thirdly, I couldn't get your experiment to work - would the fact that I am left-handed be a problem there?

Try the mirror image experiment.
The idea is that to exert force toward the body's center, you are probably going to use the wrist joint to do this if the thumb is pointed up.

The THS movement was in response to the sustained nose up input as described here:

Bigdave599

The initial storm climb
Although not an airline pilot, I have read every post on this thread. There are a couple of things as yet unexplained:
1) there seems to be an assumption that the first (2) stall warnings (just after 2h 10m 05) were spurious, possibly related to the IAS error. Firstly, there is nothing in the BEA statement to suggest they were spurious; and secondly, as the stall warner is fed from the AOA indicators and not the pitots, the IAS error would surely not cause a stall warning
2) the "storm climb" at 7000fpm for 2500 feet is not explained. The PF made "a left nose-up input" not, as he is later stated to be making, "sustained" nose-up input. And there is nothing in the text to suggest that he pitched the aircraft up to the extent needed to climb at this rate on cruise power settings - surely that would have required 20-30 degree pitch angles that would surely have been mentioned? There is also a mention of nose-down inputs in this phase

On the other hand, a big updraft would both lift the aircraft without big pitch-up, and cause the AOA to increase, perhaps setting off the stall warner?

Where this leaves the explanation of the crash I don't know, but we do not yet seem to have satisfactorily explained this first 45 seconds.

The time frame from 15 seconds prior to, and 30 seconds after a/p drop are the focus. The rest is marbles and chalk. In your post is a gob of important thought.

Short some one who is bilingual and conversant with Physics and aero, the majority of 'aha' is here. Energy( including ambient), Flight Path, and Input by both a/c and PF.

I'm not 'anti airbus' - I've never flown one. But...

It strikes me that the bus flies quite unlike other aircraft even when everythings working as it should, which is practically always. I'm talking about the way it processes roll and pitch inputs, with the a/c staying rolled or pitched when pilot input ceases.

But when the autos have gone off-line and instruments have stopped indicating normally, its behaviour is going to change more than a non-airbus would. ie it changes back into what a non-airbus would be like with the same failures.

Someone mentioned having to roll the wing level by hand. The pilot of a non-airbus would not be surprised at suddenly having to do this. An airbus pilot, with probably very small actual 'stick time' (never mind stick time in IMC on partial panel) would probably be very surprised to fiind that his 'bus had stopped helping him in this way.
Also, unless given the chance to practice in this degraded flight condition, is it really surprising the pilots struggled?

The trouble with these autos is that the better they get, the more confused you are when they fail.

All I'm saying is that a highly automated Airbus makes a pilot highly dependant on its autoflight system and 'easy handling'.

Maybe a bit too dependant.

Gotta go and read that ref now - but we are just given a glimpse into the mindset of the the BEA investigators who can't say what they are not certain about, or AF / AB could object to - but sure they will by now have more potentially controversial 'facts' and theories. I hope they can recreate what speed indication was on the RHS display.

On consulting mon interprète there's little doubt "réveille" means "wake", but perhaps there's nothing in the captain "assiste" rather than just "attend" the handover, and the PNF "éventuellement" translated to "maybe" could have been "possibly" and not be passive in context - but why did BEA highlight that but not whose decision to reduce speed. Maybe they wanted to make the point he was fresh and alert after "power napping"?

By the way - if you study the A340 airprox Appendix B zoom climb it appears (as has been pointed out before in this thread) pitch-down commands are plotted as positive which is somewhat counter-intuitive isn't it?

Circumstantial evidence only at this point. All turns were made to the right following the initial problem. We will have to wait for BEA to confirm.

Pointless? Not necessarily. Once we have an idea of how this might have happened, we can explore other implications. The number of pathways we have to explore are far fewer than before AF447 was found.
It takes a while for the light to come on with some regarding what the crew faced. The more lights that come on, the easier it is to discuss other factors. At the moment, it seems no one on the thread is volunteering any of their experiences with the ALT2 flight control law. Or maybe they only did it with the aircraft in lateral trim (due to a kind simulator instructor) Maybe no one trained for it with lateral imbalance. That would be significant.

AoA would decrease.

L337. Other poster's correct. Increase.

Folks

May I point you to this Spiegel online article with the following

I have no idea about the veracity of this quote but I would tend to say that this is generally a serious publication. If confirmed this would obviously paint a very different picture.

I distinctly remember when doing my AEF in a Chipmunk that moving the stick to the left would put me in a left turn and returning it to neutral would hold me at that angle of bank. How is a FBW Airbus in Alt2 or Direct Law any different? You're not positioned as precisely as you would be in Normal Law, and the computers can't hold you there if you're being buffeted - but that's a pretty basic flying skill surely? Isn't that what recurrent training is for?

L337
 

How's that then ?

little known facts
 

Well, well, well.

As someone asked the 'bus drivers some posts ago, how many realized that AoA was considered invalid if airspeed below 60 knots? Oh!!! Sheesh, the airspeed is deemed invalid. What now, kemo sabe? Where's my AoA "protection"? Oh! There it is again, but I am stalling and maybe getting a speed warning. I am confused.

So "attaboy" to mm43 and Hoppy.

I then ask all to look at the "reversion" sequences in the manual and all the "protections" lost at each stage. Having trouble with capturing the page, so check back later or look at previous ref by 'bird, et al.

To Doze:

We luv ya, man.

I want you to code my next hypersonic space vehicle's flight control system and guarantee the system will work exactly as SPECIFIED, not DESIGNED. Jez kidding, and I feel your pain ( not really, but we're in the stag bar now and I can be honest, heh heh).

Jocularity aside.

The concern over weather seems overdone. From the BEA report, it appears climbs and dives were not done with fairly level flight attitudes as we would see in a severe updraft/downdraft or mountain wave or microburst. They seem to be directly related to pitch attitude and later, a flight path vector in a fully-developed stall. Sorry, but that's my story, and I'm stickin' to it.

Secondly, unlike several here who wish to have a "conventional" control mode, I can't see it. Even in the older jets there are dampers and limiters on the control deflections and deflection rates. Some are based on dynamic pressure, some on body rates. Last jet I flew with zero of those was in 1971. So the best I can see is a FBW backup mode that emulates the older jets we are used to as the Boeings and original Airbus.

And we don't have it!!

My biggest problem with the laws for the FBW 'bus is they try too hard to "protect" basic attitude and roll angle without emphasizing AoA. Then there are all those connections with the autopilot. I really like some of the law changes in "TOGA", wow, couldn't go around without all those, huh?

On the good side, the gee command for pitch is really neat ( versus AoA command like a normal plane), and allows for the so-called autotrim. Actually, all it does is maintain 1 gee, even in a bank up to certain limits. My jet didn't do that, so we needed to click back a notch of trim in a turn. Big deal.

On the bad side, a reliable backup law for the 'bus might consider primarily the gee and AoA for limits, while still applying body rates and maybe "standby gains' for dynamic pressure, and allow airspeed to take a lesser role ( especially unreliable airspeed).

I see far too many "notes' in the manual and combinations of failure modes that are not clearly explained unless you read all the fine print.

L337 - perhaps I need some clarification. If the air flow vector at the front of the plane changes from horizontal towards the tail into something with a vertical upwards component towards the tail and upwards would that not look to the plane as if it had just had its nose planted upwards at a higher angle of attack?

Are you sure? The BEA Update says after 2:10:51: I believe to have read somewhere that full back pressure demands 1.5 g. I further believe that the 7000 ft/min was reached after approx. 20 - 25 seconds, which would correspond to 1.2 g or less, i.e. 40% of stick travel. Stand to be corrected, of course.

When stationary the AoA vane, which is pivoted, will be at an angle to the ground. The vane requires airflow over it to give accurate readings, I would therefore humbly suggest that the actual airspeed required for an accurate reading is 60kts.

HN39: The PF maintained nose-up inputs. So the question is whether a nose-up input not to the wall will activate autotrim or not. I suspect it will simply because the ride for the SLF would be kind of rough if autotrim only responded to a full forward or backwards stick motion.

In that case maintaining the nose-up stick input will bring THS upwards to its limits over time.

An updraft would initially increase the angle of attack, and the airplane, being statically stable (one hopes) would react by pitching down in order to regain its trimmed angle of attack. What happens next depends on whether the pilot (or autopilot) attempts to hold altitude against the updraft. It is not correct, however, that an updraft ipso facto would decrease the airplane's angle of attack.

WRT the Spiegel article, the leak regarding the pilot's statement is extremely interesting, and it is hard to see why BEA would have withheld it from the interim report. Some statements in the article are incorrect, however, for instance the claim that the pax were held in their seats only by their seatbelts. At a steady rate of descent, the pax would experience 1 G.

It is unlikely that a recovery could have been made without manually trimming the THS. At high alpha the elevator is not very powerful compared with the stabilizer. That this was not done suggests that the crew were not aware of the -13 degree THS setting. Doesn't it seem unlikely that if they were aware of it, and had identified the stall, they would have been deterred from using it by their simulator training? They were about to die. Or did I overlook something in the report?

JD-EE : I have it on good authority that it does. If the limit of elevator authority is reached and the pilot is still commanding full back-stick, then autotrim will move the THS to comply with what is being asked.

..
 

As a professional translator, assiste in Portuguese means to watch, as in watch TV, watch a football game etc.
I am almost 99% sure it is the same in French
It does not mean help or actually assist in any way

Just a little more of the CVR would help ! Presumably the CVR will show what the intent was and for some reason, legal or otherwise, someone is uncomfortable with divulging too much information. Doesn't it rather look like the initial climb was intentional: perhaps to get out of an bad icing level (with the storm below being unwelcoming) ?