lomapaseo

I take these kind of breaking news reports with a grain of salt.

I have played arround with some manufacurer's simulators trying to recreate scenarios and some interesting things happened. After some anaysis of the surprise it was mostly the limitation or programmng of the simulator that was causing the reaction. So ... I'm not ready to accept the reported facts in the Der Spiegel report as useful until the BEA completes their analysis.

Page after page of this discussion convinces me that somewhere one of you is going to be lucky and guess at some smoking gun contributor but until the total picture is put in perspective by the BEA I'm not about to scream for a corrective definied action.

Meanwhile carry on with helping us understand how the thing is supposed to work

rudderrudderrat

Hi bearfoil

Alternate Law.
"LATERAL CONTROL

When the aircraft flying in pitch alternate law, lateral control follows the roll direct law associated with yaw alternate or mechanical."

So a bit of Left Rudder would have levelled the wings.

Lonewolf_50

Bear, I am questioning what seems to be your substituting attitude instrumentation (artificial horizon) for AoA instrumentation (Angle of Attack indication). Apologies if you already understand what follows.

If I fly with my nose at (for example) 2 degrees below the horizon, and vary the airspeed, AoA will vary. If I keep my airspeed and attitudes within the normal operational range, my AoA stays comfortably away from critical, and I don't stall.

To say the pilots did not have attitude indication (artificial horizon) is not correct. Three displays of the aircraft's attitude are in the cockpit. One is in front of each pilot, and one back up in the ISIS instrument cluster.

That most airliners apparently don't have separate AoA gages does not stop pilots flying day to day airline routes from having an attitude reference: attitude (artificial horizon) is the primary reference instrument in instrument flying.

While AoA is related to attitude ... and AoB ... and airspeed ... and power ... and g load ... and air density ... etecetera), it is not correct to derive from that an AoA gage being the equivalent of an artificial horizon, which is an aircraft attitude (pitch and roll) instrument.

A33Zab

Bearfoil:

Abnormal law is a sub-law available in alternate 1 or 2 when e.g. alpha exceeds +30 or -10 or any other predetermined limitation exceedance.

Abnormal law:
- in roll: the yaw alternate law.
- in pitch: an adapted Nz law, without autotrim.
After A/C recovery, and until landing, the available laws become:
- in roll: the yaw alternate law.
- in pitch: the Nz law, with recoverd autotrim.

Touch the trimwheel -AND- and after A/C recovery it will stay in abnormal law with auto pitch trim.

See previous post: http://www.pprune.org/tech-log/45283...ml#post6486875

Manual Pitch trim movement will not initiate a law-change and has always priority!

t54

"From 2 h 10 min 05 , the autopilot then auto-thrust disengaged and the PF said "I have the controls". The airplane began to roll to the right and the PF made a left nose-up input. The stall warning sounded twice in a row."

...and because of this the PF followed standard (low altitude) stall procedure, setting power to full and raising the nose somewhat?

There is no timeline quoted her. The Stall warning might have been the first indication that the PF had that something was amiss.

bearfoil

Lonewolf50

Yes, I get. In Unreliable a/s, the a/c is in Alternate Law, out of autopilot and autothrottle. So some (ill-defined?) combination of Manual and residual control is necessary. To maintain cruise, (that's the idea?), until A/S is recaptured and auto pilot can be reselected, the idea is Pitch and Power. At the beginning of the thread, my assumption is, well, Pitch and Power, then. Some combination of NU and %N1, yes? Table, PNF, memory, Bob's yer oncle?

I may have missed a key piece of info in the plethora of posting around maintaining aero flight. I may need to re-assess; Did our boys have access to means of pulling the fat back from the fire, and screwed the Poodle?

Within the realm of cruise in a commercial airliner, and a knife fight with duelling Vipers or Phantoms, AoA? Am I wrong in taking for granted that in Commercial flight at the "edge", Pitch and N1 are insufficient to keep the flight safe?

If Pitch and Power are the fallback, de jure, someone needs to teach the Airbus pilot to fly AoA? Okay, Fine?

bear

t54

"From 2 h 10 min 05 , the autopilot then auto-thrust disengaged and the PF said "I have the controls". The airplane began to roll to the right and the PF made a left nose-up input. The stall warning sounded twice in a row."

This is reliant on the undivulged time line, for BEA state only "From....."
Anyways, after the 2:10:05 id. "Twice in a row" means to me, two short alarms, and my guess is that these were related to AoA rate transients, not a Stall per se. "chirp, chirp." In any case, the Stall warning happened at the same time (between a/p drop, and PF's left,NU input). This is in legal jargon, exculpatory.

"...and because of this the PF followed standard (low altitude) stall procedure, setting power to full and raising the nose somewhat?"

Not "raising the nose" in recovery from approach to Stall, but "Maintaining Altitude", two very different things. Since the Nose was dropping at handover, PF's flying was by the book, no matter the Training Syllabus.

Airbus instituted its "MODIFICATION OF STALL RECOVERY" AFTER 447 went in.

LW

"Put another way, if your aircraft stalls at 6, or 8, or 10 or 12 units/degrees AoA, and you are at 30, you are well behind the aircraft. A design assumption seems to be "if you get this far into stall, the computer may be a problem contributing to the situation, get it out of there so you can get this bird out of a stall!" "

So some means of "get" and control movement needs some tweaking, eh?

I may be letting some AB philosophy in, finally. "That does not absolve the airframe builder of responsibility here. Selling someone some hardware and not training it exhaustively is a chasm of Arrogance."

This will no doubt be at least one of Plaintiff's claims (theories).

Lonewolf_50

This bolded part is where a stalled aircraft could require trim wheel input to help move the THS to a position where it helps recover from the stall, since if that law kicks in your sidestick/elevators may not get the THS moving. (Based on the control authority points made previously regarding elevators and THS)

If one does not practice stalling that far
(some would call that bleeding practice, since the idea of stall recovery is typically "unstall as soon as you can, don't wait for it to get worse")
one might not recall that change -- using a secondary flight control, a trim wheel -- when one is playing catch up to the aircraft.

Put another way, if your aircraft stalls at 6, or 8, or 10 or 12 units/degrees AoA, and you are at 30, you are well behind the aircraft. A design assumption seems to be "if you get this far into stall, the computer may be a problem contributing to the situation, get it out of there so you can get this bird out of a stall!"

The more I think of how that law is set up, the more it makes sense.

If you are an unusual attitude, you don't want the computer interfering with your attempts to fly out of it. This law means that the aircraft not only allows you to fly (pitch, anyway) manually without computer interference, but requires you to fly manually without computer assistance.

But if you don't train to do it ... will you remember to fly it that way when you need it?

OK465

Academic point only (assuming you're talking A330):

If you could not cancel the stall warning with the EMER CANC button in your sim, you may want to get your simulator checked/fixed. Unless it varies by model, I believe only gear related warnings & ground prox stuff can’t be cancelled with EMER CANC. However even these continuous audio warnings can be temporarily cancelled by holding down EMER CANC. They will resume when it is released.

It’s interesting that no one has commented on the fact that the initial stall warning termination (though assumed due to unreliable data) also coincided with the Captain’s arrival and attempt to provide verbal info. If I were up front I might instinctively hit the EMER CANC so I could fully hear what he was saying. (Maybe someone has brought this up and I missed it here in the 3 volumes of “War & Peace”.)

BOAC

The problem with the AB system appears to be that no-one actually understands it so heaven help pilots caught in a software maze! Each AB 'expert' comes here and we get different shades of the event. Even Flight Global appears confused:

The abnormal attitude law is a subset of alternate law on the aircraft andis triggered when the angle of attack exceeds 30° or when certain other inertial parameters - pitch and roll - become greater than threshold levels.
Alternate law allowed AF447's horizontal stabiliser to trim automatically 13° nose-up as the aircraft initially climbed above its assigned cruising altitude of 35,000ft.
The stabiliser remained in this nose-up trim position for the remainder of the flight, meaning that the aircraft would have had a tendency to pitch up under high engine thrust.
Crucially the abnormal attitude law - if adopted - would have inhibited the auto-trim function, requiring the crew to re-trim the aircraft manually.
After stalling, the A330's angle of attack stayed above 35°. But while this exceeded the threshold for the abnormal attitude law, the flight control computers had already rejected all three air data reference units and all air data parameters owing to discrepancy in the airspeed measurements.
Abnormal law could only have been triggered by an inertial upset, such as a 50° pitch-up or bank angle of more than 125°. "That never occurred," says French accident investigation agency Bureau d'Enquetes et d'Analyses."

Can anyone give me a clear, unambiguous explanation of why 'Abnorml Law' did not engage and how the two underlined bold bits go together??

bearfoil

For one thing, as a "subset", ABNORMAL LAW IS technically "ALTERNATE LAW", and until one sees the actual data, perhaps BEA is parsing the Camel? So let's be broad minded, give the benefit of the doubt to BEA and entertain that the a/c went DIRECTLY into Abnormal Law.

For discussion only, the slim pickens are well short of any definitive answers, by design, I would say. Not even bear is pushing a theory at this point.

"Calling for conclusions not based on facts in evidence". .....Perry Mason 101.


"Any change in available controls, operation, limits, or results shall constitute a LAW change." Don't change the rules and call it something it ain't.

t54

bearfoil
"Not "raising the nose" in recovery from approach to Stall, but "Maintaining Altitude", two very different things."

Point taken.

I read somewhere that the initial two stall warnings were a valid response to the invalid 60kt pitot reading. Maybe the stall warning comes first in the relevant part of the computer program , before the validation of the data upon which the stall warning is based.?
Anyway, could the assumption by the PF that the initial short stall warnings were valid explain the climb?

PuraVidaTransport

In the Flight Global article, they are making the distinction that the aircraft never went into the abnormal attitude law because the flight computer had rejected the information coming in from the ADRs before those parameters were met. The aircraft was already in alternate law from the loss of airspeed. The last underlined part is saying that the flight computers would have to see 'acceptable' data from the ADRs to trigger this law like the examples given.

Lonewolf_50

For BOAC: we both seem to have seen something amiss in two different threads, and that article, from a different entering point.
I share your question. Does it or doesn't it (in re AoA at 30?)

For Bear:

The idea of teaching Air France pilots to fly AoA would require an AoA gage in the cockpit (not likely any time soon) and probably isn't as important a training issue for Air France as:

"How do we confirm basic fly pitch and power handling, and basic instrument flying skills, are up to the standards we expect and assume?"

The answer to that ain't trivial. What if the average AF pilot is a perfectly good, or even above average, instrument pilot. What if the line pilots are found to be good to very good through a sampling that shows skills and knowledge up to standards.

Then what do you do? What if that particular person/event match up was anomalous? Would you tamper with a system (back to Deming's caution against that) when most of your data show it operating well?

This is where the pilots in Air France doubtless have a crucial role, and ought to be listened to carefully, in any attempt to change something.

For OK:

Sounds like a PNF action.

Welsh Wingman

I have been reading PPrune posts for many years (always appreciate from afar what PJ2 has to say) before finally feeling compelled to register and throw in my thruppence worth (for what it is worth, having long since retired from B747 classics, from my retirement cottage...).

This reminds me somewhat of the Erebus/TE901 crash back in 79, where an air crash investigator (used to small aircraft investigations) put all the blame on the DC-10 flight crew. It took Justice Mahon to look at the systemic failings at Air NZ, and point out that programming the plane and the pilots of a low flying sightseeing plane near to high ground to fly to different locations in Antarctica was unlikely to result in a happy ending (particularly last minute amendments to the flight plan co-ordinates without notifying the flight crew). There is seldom a single failure leading to a plane crash, but rather an entire chain of unfortunate events that could have been, but were not, interrupted at any stage....

BEA have got to be very careful about what they release, including from the CVR/FDR, and will dare not release anything at this stage that they are not absolutely certain they can verify. To put it mildly, they don't want the lawyers of Air France and/or Airbus pulling anything they say/imply apart and we will have to be patient. I have confidence that BEA will get there, but they have huge domestic commercial sensitivities (albeit the last thing that they want is an international observer, e.g. from the AAIB, disavowing their final report) to factor in (rightly or wrongly) to avoid needlessly damaging on the road there and we must accept that.

The one thing that we can be certain about is that the BEA are, as we write to each other, going through all areas and particularly the interface between machine and pilots, and this will take time. There will be agonising/soul searching over Airbus design/philosophy issues and Airbus/Air France pilot SOPs - high altitude A/P and autothrust disengagement and the flight envelope/crew response implications, the lack of redundancy with the loss of air speed data (and why the Thales pitot tubes were not changed sooner, given this obvious redundancy issue and criticality to safe A/P operations), the PF's stick movements (backwards in particular) and the PNF's (& Captain's from FLT 350 down) ability to see these movements, that climb up to FLT 380 and the THS going to 13 for the duration of the alternate law flight, weather radar training and the possible weather deviation limitations for this plane on this particular route (unless the flight crew want to land in Bordeaux and endear themselves to flight ops...), why the pilots did not appear to recognise the high altitude stall at all (including no AoA indication or BUSS installation on this aircraft) when they had FLT 380 to get out of it (troubling to an old salt like me trained by the Fleet Air Arm in the early-60s to lower nose/increase power!) and the ability of flight crews to immediately correctly identify the problem and correctly respond if possible/permitted by the remaining systems (in the face of such an urgent problem, used to so much automation and never touching the manual trim wheel).

This all being said, keep up the good work and keep your thoughts/ideas coming. I bet somebody from the BEA will be taking a look at some of the more informed comments. Just be patient with the BEA. There will be a lot more, on the CVR in particular. For what it's worth, I don't expect and can't see (save for any "smoking gun" in relation to unexpected system performance outside of normal law) any party (Airbus/Air France/the flight crew) being totally "exonerated" on this one.......

bearfoil

I am weary of folks accepting counterintuitive and plain wrong assumptions, including me.

If the a/s was 60 knots, and the a/p was in, we need to talk. If the Stall warning was on, and the a/c was not in ABNORMAL LAW, likewise. Monsieur takata swears that a/p drop was UAS driven, in spite of a time line provided by BEA.

Screw ACARS, all right? Old habits, etc. If the a/p was in, and the PF had a Stalling a/c in his mitts, he had a correct response.

Cruise flight is not easy to parse into BITS, it is a dynamic Dance, and these guys had an "Unfolding" (Unwinding) Flight Path. May we start there? Back to read only.

BOAC

- there you go - we are first told Stall warnings are AoA driven now we 'read' they are IAS driven.

Those first two certainly puzzle me, whatever triggers them..

jcjeant

Hi,

Sorry if I'm dumb .. but .. someone can make a technical comment (explanation) about the text in bold .. ?

From the 27 May BEA report:

Thank's in advance.

rudderrudderrat

Hi jcjeant,

The only way I can imagine that is if they "bunted over the top" so the wing loading was much less than 1g.

Lonewolf_50

t54, I am confused again.
Stall warnings, and stall α, are shown in HazelNuts39's graph
They are based on AoA, the magnitude of which is influenced by airspeed/mach number.

I don't understand how you reach the idea of a stall warning based on an invalid (or even a valid) airspeed reading, since stall warnings are (should be?) based upon AoA which is not the same as airspeed. It is a different parameter, influenced by airspeed, attitude, gross weight, angle of bank, G, air density ... etc

From earlier posts: what the 60 knots threshold seems to have triggered is a disabling of a stall warning, which is itself enabled by an AoA reading as shown in HazelNuts39's graph.

Does that make sense?
Do you mean due to an increase in power as a response to stall warning?

Not an Airbus driver, but my assumption on stall response is decrease AoA via attitude change, and increasing power with the intent of increasing speed and thus decreasing, for the same attitude, AoA further away from stall margin. (EDIT: if you don't recognize or know you are stalled, or believe you are not stalled, you might not make that response).

For bear:
Based on PPRuNe posts regarding this crash, posts made by actual airline pilots who fly heavies at those altitudes, posts since about 02 June 2009, the answer to your question is that pitch and power are indeed, and should be, sufficient to maintain flight.

The estimates back then frequently led to "didn't know to fly pitch and power? If they didn't why didn't they?"

Info to date released by BEA seems to confirm that setting pitch and power for that altitude and desired performance isn't what happened ... but the why remains elusive at present.
Not really. If pitch and power are the standard fall back, (and it appears that numerous pilots and airlines have procedures that are precisely just that for UAS conditions, and these work,) then "flying AoA" is a subsequent skill set suitable for other flying applications.

Referring to AoA as a crosscheck if Airspeed becomes unreliable has been suggested (to confirm "what is my wing doing?") as a suitable improvement to the pilot's tool kit. (A lesson learned, if you wish).

DozyWannabe

The simulator is not the aircraft - given that it seems that they were in a situation that had not even been encountered in test flying, so no data with which to program the simulator. There are no details as to the specific actions Prof. Huettig performed, or how they correlate to the actual actions as they were performed in the flight deck.

I find it quite ironic that you of all people, who considers software so unreliable and so many of the people who create software to be... What was it? I believe the word was "dolts" (no offence taken, sir...) would be willing to put his whole weight behind a "smoking gun" derived from a computer simulation that was based on incomplete data.

It's simple really - past the dry engineering language it boils down to the fact that Alt1 has protections, Alt 2 (effectively) does not. If I were a pilot, I would be very wary of relying on any of the remaining protections in Alt Law to back me up, doubly so in a situation when there's an obvious data capture issue (or in aviation terms, unreliable instrument readings).

Correct.

AoA information and functioning AH/AI/ADI (whichever you want to call it) are not one and the same thing, bear. On modern aircraft (talking about anything post 757/767), attitude information is provided from a redundant pair of electronic gyro boxes known as an AHRS. Attitude information does not rely on either AoA vanes or Pitot/Static sensors.

I refer you again to the Birgenair accident where the aircraft (a 757) was effectively in the same situation, though it was a case of the captain's blocked pitot tube being hooked up to the FMS (autopilot), thus inadvertently creating a single point of failure rather than all pitots being blocked. In fact by programming the FCU to detect multiple pitot failure and kick the autopilot out upon detection, Airbus actually created a safer design. The Birgenair 757's FMS actually tried to fly the aircraft on the bad information, pitching the aircraft up to an extreme AoA, and it was only that extreme AoA that caused the FMS to switch out. If I recall correctly (and please, anyone, correct me if I'm wrong) Boeing later retrofitted the 757 and 767 fleet to enable selection of pitot/static data to the FMS, so that a failure on one side would not be catastrophic.

In that case, the all the ADIs (powered by AHRS) were functioning perfectly, and the relief F/O repeatedly called out "ADI!" on the way down, pointing out that the aircraft was in an extreme nose-high attitude (the other F/O, who had a working ASI repeatedly called out that they were stalling) - all the way down to the ocean. The pilot in that case, as I've said before - no low-hour newbie, but an experienced ex-Air Force jockey appeared to be so overwhelmed that he failed to check the ADI in front of him that was telling him he was nose-high, even as his F/Os were emphatically telling him what to look at.

Bear, you've got to be careful how you phrase this stuff. A newcomer to this thread would think that what you're presenting is a known fact rather than a theory posited by a single poster that is based on no evidence whatsoever.

BOAC, If the context wasn't so serious, I'd be willing to wager with you that what comes out of this will prove all this talk of software and laws to be something of a red herring. We've seen two 757s go down due to pitot/static failure and their controls were not software-driven.

That said, if what some posters are saying about inadequate training in use of the manual trim wheel is true, then it's a major problem, but it's not a software problem. Ultimately, even if you're in Normal Law and you don't like what the trim is doing you can grab that wheel and set it manually.

I know that the increased presence of computers in the flight deck is an emotive issue for pilots, but I can assure you that from the perspective of this software engineer, and, I'd be willing to wager even more - every software engineer who worked on the FBW aircraft that are flying today - we are on your side. We are trying to make your job easier. The systems we built were specified and designed from requirements put forward in the main by pilots. We are not intending to replace you. If your employers are saying that the technology effectively reduces you to systems monitors - and that hand-flying is discouraged - then they are abusing what we gave them, and as PJ2 says, you *must* fight them on it for the sake of every person who boards your aircraft. That was not the intent, and I hope never will be in my lifetime.