@PJ2

You forgot the first rule of Pprune:

Never blame the pilot. :=

Blame the pilot
 

Given PJ2's background, he's totally entitled to do so ... and he is doing it in a very honest and respectful manner.
Airbus' designers are not treated as well by some other contributors ...

Stall warnings
 

Stall Warning

In normal law, given the protections available, the stall warning was designed to avoid its activation before those angle of attack protections and not to provide an additional protection. So, this warning is triggered in normal law as soon as the value of the angle of attack of one of the three sensors reaches 23 degrees in landing configuration, higher than the angle of attack associated with Vs1g

N.B.: For a given configuration, Vs1g is the lowest speed at which the aeroplane can develop lift equal to the weight of the aeroplane

In alternate anddirect law,this threshold is modied in order to allow the stall warning to be set off at a lower angle of attack than that associated with Vs1g

(from Air NZ - Airbus A320-232 ,page 30)

Do these tighter AoA thresholds (in alternate law, along with turbulences) explain why stall alarms (going hand in hand with the revertion to the alternate law) were frequently observed: in 9 cases over the 13 past Pitot incidents ? (analysed in the 2nd BEA interim report).
In the Air Caraïbe case, as far as I can remember, the stall warning was triggered for an AoA around 4.3°.
So we get:
-a variable meaning/sensitivity (thresholding) for the stall warnings (depending on the law/mode you are in),
-when the airspeeds become unreliable and when the ADR disagree (triggering the alternate 2 law), this tighter stall warning threshold seems to become a function of the airspeed (not in normal law),
-variable ways to handle the stall warnings (depending whether you read the ECAM messaging "risk of undue stall warnings" or the unreliable IAS C/L urging to consider seriously the stall alarms since they are based on the AoA)
-a stall warning supposedly independant from the airspeeds but inhibited when the IAS are no longer valid
:ugh:

FE Hoppy;

I wasn't aware that there were "no-go" zones in this discussion. Are we under an informal rule or do we stay true to our craft?

The notion of "blame" is not in play in any discussion regarding the accident.

If people choose to read a discussion this way, that is beyond the control of the contributor who chooses to discuss all aspects of this accident.

Blame has no place in the discussion but examining the evidence available and positing notions and otherwise going where data and evidence and knowledge of the airplane etc lead, is a primary investigative principle, one which, I will add, is not honoured (despite their membership in ICAO), in many countries and in a number of accidents which have been discussed on PPRuNe.

The entire picture of course does not snap into focus - that's what "blame" does...attempts to snap things simply and leave it at that. Instead, bits and pieces are examined, placed aside for later consideration or placed on the table for closer examination or placed in what would be an extensive discussion of factors working in parallel.

That is the kind of discussion I would expect from those who want to "find out" rather than hand it off and "blame the ____".

As a pilot I am keenly aware of the dangers of "blame". I am a flight safety person, not some lawyer with an interest.

FWIW:

If the airspeed indicator isn't working, and pilot is on instruments, a possible answer to that is "yes?" (Or, if the AS indicator is working, but the pilot thinks it isn't, and is on instruments ...)
You ask a good question: does that make sense as a design case?

But even if one accepts that as a possible reason to display AoA ... one is still left with an instrument scan, and all of the other cues, none of which are seat of the pants.

What is Attitude Indicator telling you? (Pitch & Roll, turn & slip)
What is Altimeter telling you?
What is VSI telling you?
What is the Thrust/Engine indication?
What is are your configuration display telling you (flaps, gear, etc)

I am still wondering, and will probably not get the answer to:

What did the PF see? Where was his scan taking him?

Hopefully, info from the CVR will eventually reveal what PNF's response was to the climb from FL 350 toward FL 380?

What did he see, what did he say?

EDIT: to make sure I understand correctly

Stall Warning is intended to sound/go off at an AoA less than stall AoA.

Point being (like stick shakers or rudder shakers on various aircraft) to cue pilot to act to prevent the stall. Stall warning would remain on until AoA goes below that value, to indicate either "you are still close to stall" or "you are stalled" either of which is a cue to do something to unstall the aircraft, or get away from the stall threshold. A simplistic way of translating Stall Warning is "change your AoA, it's too high" as noted in various posts in this and other threads.

(Yes, pitch and power remain the basics).

Without question, the role of bias is part of every investigation. It is a part of every human being who is not perfect. Better it should be established, and weighed (also with bias), than denied, whether naively or slyly.

Commercial carriage is a capital venture subject to the whims of market, but mostly to the realities of competition.

An example, an excellent example, is the Thales Probes. I will apologize for calling them "Junk", but only insofar as it may have offended a friend. Here, there is no argument in evidence that the research and performance were faulty, that these devices should be accorded a benefit of doubt, or other accommodation. They did not perform according to expectation, an expectation that was made with warranty, and involved the possibility of killing passengers. This is not proven. Instead, mitigation in the training of operators was devised, and approved.

How many are then guilty of at least poor judgment, if not criminal culpability? In hind sight, the proof appears stronger than than the "Defense".

"Blame" and "responsibility" are cousins. "Fault finding and Fact finding are brothers.

How distracting does it become that some would excuse the "discipline" by saying "we don't do this"? No one is excused, and if feelings are ruffled, that is a good thing.

People are dead, and thus far, it is becoming clear that no "Act of God" is responsible, and some may be responsible in some way for these terrible losses. In my opinion, it is a good argument that preserves the emotions in play here.

My bias is a matter of record. The worst thing that could happen is for this accident to fade. In its loss of currency and care would be the continuation of a way of "doing business" that is repugnant to those who care about people before profit.

Does this happen? Can we change it? I do not know. But if we give in to the pressure to keep the current methods in play, these souls will never forgive us.

Wolf

you ask.... ".....What did the PF see? Where was his scan taking him..."

BEA..... "...The PITCH progressed beyond ten degrees, and the a/c started to climb......" (emphasis mine)

I haven't seen data re: AoA just prior and just after a/p loss. Pitch and power, if accomplished would have established a Pitch of far less than Ten degrees, (more like 3.5). So Pitch was not, and did not become, compliant with UAS procedure.

That is it. BEA has the rest. The fifteen seconds on either side of loss of a/p are the meat, the rest is potato. Clearly, the danger to flying passengers has been determined to be within limits, if not the actual data of this accident?

Plenty of time for a measured and patient response from the authority, then?

As an answer to all of the posters insisting on the CVR content....

What would you have said ???

Unlike some well-known tennis players, yelling/shrieking at every stroke, I doubt the average pilot in such a situation would have punctuated his every action with yells and shrieks and expletives (except for maybe the occasional "putain" and "merde".....).

CRM does not always equate to being able to give an intelligble running commentary for your colleagues and the CVR... esopecially if you're confused yourself.
It's not tennis.....

LW_50;

Regarding what the PF saw..., CONF iture asked the question a while back about whether the QAR would have the CAS parameter displayed on the F/O's PFD.

There is no easy way to answer the question for, as discussed in the post on data frames, QARs are different animals and are subject to more variations in what they record and how they record it (in terms of rates). The parameter is certainly available and, optimistically, I would think it would be on the QAR, which of course is primarily used for FOQA/FDM analysis by the airline's safety department. A lot of other stuff will be there as well and being solid state it may have survived even in water.

We can't say for sure whether the three airspeeds failed in the same way because we don't have the data. But I think they did fail the same way. I think it was gums posted an A300 manual's schematic page which showed very clearly what happens downstream and over a short period of time to airspeed indications when the pitot head or drain holes are plugged with ice or water. It is more difficult to imagine differences than it is similarities although until the data is available, it is possible.

What would cause the ADRs to read a higher speed and send it to the DMUs which display same on the F/O's PFD, that permits at the same time, the loss of airspeed on the Captain's display and the ISIS, and the uniform increase in AoA parameters to 16deg and beyond?

The questions concerning the attitude, sideslip, altimeter and IVSI are good ones, and we need to understand what else the PNF saw and perhaps said. None of us here expect that this release was not closely vetted.

As a basic watch this or die type indication, something calibrated to the specific aircraft in the form of the old F4 indexer lights could be an answer to the crew getting swamped in a high stress situation.

http://aviation.watergeek.eu/images/...oa_indexer.jpg Not the dual bulb shoot light on the right

OR

http://www.dcswiki.com/images/c/cf/A...ackIndexer.png

OR

http://prideaircraft.com/images/AOA-01.jpg

Something simple and mounted away away from a confusing/failed/corrupted lcd display.

just my 10p worth, nomex applied.

gladrag, isn't that AoA indexer only activated when the gear is down? (Memory faulty on that one ... maybe that depends on aircraft model ... )

Bear:
bear, I didn't ask what the FDR recorded, I asked what the pilot saw. Not necessarily one and the same. I am asking what had his attention. A scan can turn into a boresight under stress.

ChristiaanJ
Hard to say. Were I a copilot, and I saw the pilot climbing, and we were at a phase of flight meant to be straight and level, and he hadn't told me he was going to climb, I suspect I'd have said: "You are climbing." Maybe "Do you mean to be climbing?"

Even more likely, had I seen the climb, (and assuming AF SOP is similar to an old outfit of mine), I might have said, "Two hundred feet high and climbing" since assigned alt was FL 350 ...

But that's a guess.

I don't know what PNF was looking at. I've no idea what cue first caught his attention of something in the flight profile being "not quite right."
I was thinking more along the lines of some standard CRM and copilot functions, not a Hollywood style set of emotive statements. Nor a grunt. :p (PS, pet peeve of mine, all that tennis grunting ...)
I know it isn't tennis, having flown in mutiplace aircaft. I am not asking for running commentary. I am keenly aware of how little one says when one is catching up to an aircraft. (Been there) Maybe he said almost nothing.

I am interested in what cues he did, or didn't, respond to in executing copilot duties in support of his pilot.

What he said (as recorded on the CVR) may give a clue at which things in the cockpit he was looking at, and which things he brought to the attention of the flying pilot.

Or not. (You make a very good point! :ok: )

PJ2
Yes. My question includes "of all the things in front of him, what most caught his attention, or what group of things did he pay attention to?" He may or may not have communicated to his copilot what he did or didn't see, or what was his main concern, beyond "alternate law" and comments on the speeds that have been released ...

I have myself, while on instruments, had the whole instrument panel in front of me but had my world collapse into a form of tunnel vision where all I could "see" was attitude indicator, airspeed, VSI and Altitude ... barely heading ... I had to force the scan around ... as I fought vertigo. You could have asked me power setting and I'd not have been able to tell you, as I wasn't seeing it, it was where I had left it. You could have asked me AoA and I'd have had no idea.

Aye.

Hi,

PJ2

This does not seem to be the case so I just re-read the note from the BEA.
And my doubts were founded.
When the autopilot disengaged (for any reason whatsoever) the aircraft banked to the right.
What happens if the pilot does not correct this inclination?
The aircraft was stable before the autopilot disengaged:
Sure .. the aircraft was stable .. but why ?
Because the autopilot was making the adjustments necessary to maintain stable flight of the aircraft (his duty)
When the autopilot disengaged .. it was the pilot duty to maintain a stable flight of the aircraft ...
The aircraft banked to the right .. so ........ hands off ?

My inclination is to say that the a/c was likely not well trimmed, and was only "seemingly" in stable flight, perhaps "hunting" a good deal. Whatever the cause of a/p quitting, UAS or turbulence, the a/c itself made the decision it was not "stable", nor could it be stabilised (by the a/c). This makes a comment like:

"...the airplane was stable before the event and, doing nothing while the airspeeds return will keep the airplane stable. Move it, and you've moved the one "known" in the equation that you had and you've taken yourself into no-man's land with no way back, without a great piece of luck and skill.."

quite patronising, and dismissive of a very likely reason for not "sitting". Also, perhaps a bit biased in favor of the "Platform".

Yes, would probably have been the better choice....

It is hardly conceivable that this roll momentum would have been so violent that they would have ended upside down within seconds.

And then a more careful reaction (with less potential for over-reaction) would have been to trim the rudder to compensate for the roll. Especially since a rudder mistrim is a premier candidate for being reponsible for the roll in the first place. A strong fuel imbalance would be really strange. That normally does not happen. At least not without a prior warning of a pump failure or similar. And that you would have found on ACARS. So I would consider that highly unlikely.

Knowing that it is really hard to fight the urgency to do 'something' if something unexpected occurs.

Edit:
@PJ2: Thank you very much for your rational and well balanced and informed posts !!! :ok:

If one wants to learn something from this tragedy it is vitally important to keep an open mind. Even if it would mean that the Pilots in this case made Errors that mainly caused this accident.
And if it was that way it must be allowed to say this here otherwise it is not worth the discussion.
As PJ2 and others have said: This is to improve safety and not to put blame.

(From PJ) Design point to consider with this in mind: if airspeed is unreliable, or if AoA is on the fritz, other warnings (engines, for example, if icing is the condition driving airspeed awry) would be useful and even necessary. Establishing a heirarchy makes sense if you are to avoid sensory overload in the cockpit.

Wolf

I am not persuaded that the A330 needs additional design work. This accident is I believe, more basic than a hugely anomalous glitch might infer.

I don't think BUSS would have made much difference, nor the addition of steam gauges as "back-up".

Credit where due, this a/c did climb with X energy, and reached X-y somewhere near the apogee of a long curve to the Ocean.

I truly believe that the answers are few, easily connectable, and straightforward.

I honestly don't believe BEA parsed the data to "set-up" the pilots.
There is sufficient responsibility available for all, and eventually, it will out, to a very fine accuracy.

I think what we have here are Blunders, not cheese holes. If, in identifying these shortcomings, a new and improved way of driving the Bus (and other a/c), is proposed and accepted, fine.

I also believe that underneath the Blunders will be some of the less attractive aspects of Human behaviour, not limited to Flight crew, by any means.

The Bus was built by humans, and flown by humans. So, not necessarily PE, but H-E.

I would not want to inhabit the moccasins of those responsible for making this horrible crash likely, rather than remote.

Laws, functions, protections, limiters
 

Sorry if I appear to harp on "too many" laws. It comes from reading and examining all the Airbus manuals I have gotten courtesy of this forum. And I thank you all.

I am gonna take a pic of the FLCS ( flight control system) I flew with back in 1979 and you will see many "laws". Will post it soon. The "laws" were those small graphs of impact pressure versus control deflection, body rate limits, AoA limits, etc.

I am also gonna show my problem with all the "modes" and "laws" and "protection". Let's face it, I flew a very simple jet and it had vastly different operational requirements.

That being said, we never used the word "protections". We used "limits". Maybe it was psychological to provide we dweebs some reason that our piloting skills were relevant, but it worked. Heh heh.

Before I take the pics and upload them, I am compelled to comment upon the THS mechanization. I know, one mo' time Gums, when will you let go? Nevertheless I feel it is more important than the pilot action to correct a roll after the AP disconnected.

With unreliable airspeed (think impact pressure, dynamic pressure, CAS/IAS), I look at some of the alternate laws and still see clues that the confusers are trimming the THS. So without airspeed, the jet is still trimming to maintain one gee. It does not care about the pilot aft stick input other than to keep the stick deflection about the same angle regardless of speed. How does it do this? It uses the accelerometers that are the major players for the basic control concept of a gee command versus an AoA command system. You don't need airspeed for the accelerometers to work. Same for body rates. Saw this on one of our accidents when a pelican wiped out pitot tubes, hemispherical air data gizmo and AoA vanes.

So a constant, even a small aft stick, that is commanding maybe 1.1 gee could result in the confusers increasing the THS angle to maintain the same stick angle for the commanded gee. No air data required. This might explain the THS increasing angle.

REMEMBER: The pilot is not commanding an AoA or pitch rate or attitude. The final command to the control surfaces utilize some of those inputs, but the primary output is related to gee. Clever addition of AoA, speed, body rates and such can make the jet "appear" like the last one the pilot flew. But down deep, the control laws are very different.

Bear
1. If you are referring to my post on why you'd clip an AoA input, it was a defense of clipping AoA signal, as currently done, based on the audio warning heirarchy that PJ2 pointed out. This is a different tack than my earlier "why is AoA clipped below a given speed?" query. It made no sense to me to clip AoA signal below a given airspeed. (I suspect that discussions on that point were non trivial in the design process). The warning heirarchy provides some of the reasoning behind that design decision. There was doubtless other reasoning involved.

2. BUSS seems an interesting approach to a problem. No skin in that game.

3. Would an AoA gauge (steam or digital cockpit display) have been incorporated into the scan in this particular instance (and when would that have been practiced in the training syllabus?) ... my guess is that at least one scan got narrow pretty quickly ... but there's something more basic here.

If you don't train for a stall recovery on instruments (for reasons previously explained, to include sims that can't replicate it), what does a pilot finding himself in a stall have in his training background and experience to draw on in order to redirect his scan. His task chain requirement at that point, when stall is identified, is to change his instrument scan from
"flying and getting this bird to the airspeed and altitude I want" scan
to
"recover from stall in the goo" scan
to
"recover from unusual attitude (unstalled)" scan
to
"get back on course, speed, and altitude" scan.

The first scan, not trained, has to be made up on the fly.

Some of the folks here have for two years pointed out that you train to prevent a stall. There are some good reasons for that philosophy, given training limitations. The pitch and power refrain has been sung, like the Hey Jude chorus, since about June 2, 2009.

Considering what training can't be done, see sim limits, that song is number one on the hit parade of the remedy concert to be performed in due course ... but it won't be the only song to be sung.

Still chewing over gums' point on using g to direct control surface positions.

I'll show you mine, you show me yours
 

Sorry for poor pic, but I don't have a scanner and didn't wanna drive down to the UPS store.

The graphs at top are the "laws", or functions. lower right are stick pressures versus gee or roll rate.

I can do it better if anyone cares, but I would really like to see a similar graphic for the 'bus in any of the "modes" or "laws".

Note that in the upper middle we have the "standby gains" for when air data was deemed unreliable. So those values were applied to many functions for control surface movement and rates. I see no such concept/implementation for the Airbus. Why is that?

The left side of the page I used has much more detail, but I need to get a good PDF scan to show how all the functions are added/divided/integrated etc.

We didn't have all this in our operating manuals. I only have it as I was the Chief of Academics for the first training squadron and had to answer questions from the dweeb students, heh heh.

enjoy,

http://www.sluf.org/misc_pages/laws.jpg

That's why I'd NOT turn off the stall warning of there is no weight on the wheels. I'd add an altitude test if there's a reliability problem with only using weight on wheels.

Then it may be wise to enter the inertial reference's output into the equations at least for short term stall warnings. Then maybe train in a 10 second or 15 second wait before acting on stall warnings at cruise altitudes.

bearfoil

sadly, you are correct I think. IMHO this accident will go down, correctly, to that much malgned term 'pilot error'.

I've been in a MoFo CB and it's not pretty. But hey, although my instruments werre working, I couldn't read them!

Sitting here, my brane 'says, pole the plane on its attitude and leave the throttles alone'- suck it up!

WTF would you climb? Fear?

Caution About AoA Indexer
 

Glad Rag, again, I need to caution everyone here who has not routinely flown AoA, AoA indications at approach speeds assist in quite precise pitch control, but not at cruise speeds because AoA changes very little with airspeed at the higher speeds. It's a bit like balancing on a bowling ball. It's possible but not easy.

LW50: Yes, the indexer ordinarily is activated only with the gear down, but that could be changed, of course. On the other hand, the typical AoA indicator would certainly give the pilot an indication of AoA which was near stall.

Would I like to see an AoA indicator installed and used in transport aircraft? YES! Proper use of an AoA indicator might have prevented this accident. Of course, so could some other things.

Concur.

The C* Law in the A320/A330/A340 is gee driven. Alternate Law 1/2 are variations/protections bolted onto the C* Law. In Direct Law, the stick is a B737 control column and its WYSIWIG including trim.
Lonewolf_50;
I think this is one thing that is going to come out of this accident - proper stall recognition and training and not mere entry-and-recovery (where minimal loss of altitude has been not only stressed but is an actual IFR-ride failure if more than nnn feet are lost in the recovery!! - clearly designed for low-altitude stalls, (and even that didn't work for THY).

To do this, both simulators and curriculae have to change and that is a very tall order given the comments on sim performance beyond loss of control regimes.

Given the past ten years' data on stall accidents however, the industry is showing that it needs to do something. Along the way, the discussion about automation and the man-machine interface will take place in earnest just like we wanted it to when we checked out on the A320 in 1992.

Smilin'....
Because the L/D curve is as shallow as a dinner plate and one is trying to keep a marble in the middle... ;-) !

Smilin_Ed,PJ2
 

Cheers guys just a shot across the bows so to speak.

PJ2 made a comment I hoped to elicit, "DO NOTHING was the correct response and that is the action which would have "prevented" the loss of control."

I also note that something like this had happened before to a fair number of planes that sailed through it with fine aplomb.

What might have been different with AF447 other than a different cast of characters? In my mind I don't picture pilots acting on whim or instant panic. (And the BEA highly edited release seems to indicate rationality all the way down.) Was there something the other planes had which saved them, like a bright Moon or another extra form of reference that reassured them? Or was something missing with AF447? Or did training simply break down?

(I seem to remember talk here about those rules being standard at the time of this accident. Wasn't some aspect of it changed to "always believe a stall warning?" Has a future problem been setup?)

As a note his roll input doesn't bother me. A strong enough pull on the stick to cause the plane to zoom 2000' like a marble rolled on the tabletop towards an incline (up and slower is the result) bothers me. So please let me ask, "if the controls really are sensitive enough that the "natural" bend of the wrist to correct a roll would also induce a climb?" I believe somebody remarked taht the sticks are mounted at a modest angle to vertical to accommodate human the restraints of the human skeleton. So that supposition of accidental input seems wrong. So I reask, "What made the PF decide he needed to climb? Might it be to avoid something and he forgot throttle was frozen? Might it be he thought there was an over speed condition and he needed to slow down immediately?"

Your UAS list comment is one we should enter formally into the record and discuss. It does appear to at least partially answer the above. 2000', however, seems to cost a tremendous amount of speed. I'd think 500' or 1000' would be about the limit of what you'd want to lose in a pull up action.

If this series of inputs and reactions still does not make any sense let's put a "post it flag" on that time and look at the other intervals in the release. I'm looking for actions or even stimuli that could be modified slightly to have eliminated or recovered from the bad situation that led to the crash.

JD-EE 'Rules' are always valuable.

But they should never replace what your instructor taught you on your PPL.

While this is surely true in retrospect, it seems contrary to most training.
The pilot assumed control as the airplane banked.

How does he know that the bank is self limiting?
Is not the drill to level the wings?

Hi etudiant,
It's not self limiting.

"LATERAL CONTROL
When the aircraft flying in pitch alternate law, lateral control follows the roll direct law associated with yaw alternate or mechanical.
ROLL DIRECT LAW
BANK ANGLE PROTECTION
Not provided."

Shallow L/D curve and AoA and speed
 

Well put, Smilin' and PJ. OTOH, I wouldn't call it like balancing a marble on a dinner plate.

The Airbus manuals have a diagram depicting "stall protection", and that graphic looks like that of a Cessna. Trust me ( and Smilin'), the graphic does not show the relatively gentle curve that most all swept wing planes have. Those suckers do not have a sharp "break", then a stall. They tend to buffet a lot, maybe have some roll oscillations, and finally fall off on a wing or "mush". If the jet has super directional stability, it may stay fairly wings level and not yaw too much. Look at the F-18 demo's and think about the jet's AoA and how easily it passes the reviewing stand without a lotta wing rock or yaw.

The other thing is that the curve is also fairly gentle with respect to IAS/CAS. So 10 or 20 or 30 knots is no big deal.

Doubt if anyone here flew a delta or the Concorde, but who knows. My limited time in the F-102 was "educational"( 80 hours or so). No real stall you could see or feel. Thing just descended with the VVI pegged at 10,000 feet per minute. Mild buffet, more of a buzz. Viper was about the same due to the variable camber courtesy of the leading edge flaps.

later, and I shall try to have some of those archaic FBW control diagrams in a great format for history.

Symphony of Flight
 

Pardon if this comparison is lacking; my musical skill is limited to playing the radio.

An airliner on departure, climb, cruise, descent, approach and landing is like a symphony orchestra, with all the components working in concert.

The conductor can be the FMS, Flight Management System, or the aircraft commander.

It's all seamless and symphonic until a component hits a sour note. Then you really see the contribution of the other components and what is driving them.

It's basic to conventional airplane dynamics, as taught in Langewiesche's "Stick and Rudder" and hundreds of other instructional manuals:

Pitch controls airspeed.
Power controls altitude.

The A330 Flight test accident in 1994 uncovered the flaw in Airbus logic. The plane was on autopilot in Altitude Capture mode when the pilot pulled the power on one engine. The AP responded to lack of power by pulling the pitch up to over 30 degrees. Instead, it should have kept the pitch at desired speed, and told the pilot, "We can't do that."

The TK951 B737 approach accident at AMS in 2007 revealed the same flaw. An undetected erroneous input to the A/T caused it to go to Flight Idle. The autopilot tried to stay on Glideslope by pulling the nose up and trimming to lower speed to make up for lack of power. Instead, it should have kept the desired speed, and let the plane sink below glidepath. If the pilot hadn't caught that, he would have heard, "TOO LOW! GLIDESLOPE."

Firewalling the throttles would have stopped the sink, and not have caused such a pitchup, as the elevator would still have been trimmed for the correct speed. The pilot could have intentionally traded some airspeed for altitude, and without fighting full aft trim.

How does this apply to 447? We'll see. Trying to control altitude with pitch is a loser, regardless.

Bearfoil, is it a total blunder if you have a UAS condition to apply UAS procedures? Did he pull the stick before the stall warning or after? It appears to be "before" in the rude imprecise chronology of the released data.

And simply for my own information I have some related questions.

* If he should have waited to apply UAS procedures then how long should he have waited?

* What is the intended purpose of the UAS procedure. It seems to be to slow down the plane to get it into a safer margins condition between stall and overspeed, which was already happening.

* Might the pilot have asserted the "slow me down quickly" planning to lose maybe 50 knots by climbing 500' to 1000' and then level off only to have the stall warning derail his thought processes?

* Is everyone happy with the UAS scenario PJ2 suggests. It seems to partially fit to my untutored eyes. If it was not UAS, why was it not? If it was not UAS, then what the heck else might the pilot have been doing presuming he is rational with no deathwish? Too many here are willing to accept conflicting data in such a manner as to suggest the pilot simply messed up. I'm ornery. WHY might he have messed up? Absent a good reason I'm not willing to give the "pilot messed up" crowd an ear, including the French Transport Minister.

We need i dots and t crosses applied to the appropriate letters here.

The trouble with the scarebus is, when the autos roll over, what you're left with is a Boeing.

And as you are a scarebus pilot, you don't know how to fly a boeing.

Gums, I was starting to read PJ2's posting with its quote from your 1698 when a coin dropped in my mind.

I got this far, "So a constant, even a small aft stick, that is commanding maybe 1.1 gee...."

1.1 G according to which measurement facility, inertial, measured stresses on airframe, or what else? I just want to get on the same page as you since this is important.

Also important is your remark about vocabulary. The words "protections" and "limits" are seen quite differently by most people. The former gives a sense of safety and the latter gives a sense of urgency.

JD-EE
I'm not sure Do Nothing was a valid course of action either. As I understand it, there is nothing automatic to keep the machine right side up in roll while in Alt2 Law. The pilot has to fly the wing roll attitude.

If it was in lateral trim, probably no big deal to fly it in roll direct law, just be gentle but firm.

Out of lateral trim, it would be a complete handful-until you balanced it out with rudder trim.

Thrust
 

PJ2,

In the most recent discussions regarding pilot actions, stall warnings, etc., the one thing that isn't mentioned much are thrust settings. Recall I had asked about the N1% differences between normal cruise at Mach 0.83 and reduced turbulence N1% for Mach 0.80. The questions I had behind this has to do with how this may have affected subsequent events. With a reduced N1% at cruise at 35K, for some reason (probably pitot tube blockage due to ice crystals), the AP/AT suddenly dropped out in a probable turbulent atmosphere. I am assuming when this happened, the engine thrust level remained at that for turbulence penetration, whether or not thrust lock occurred, as the pilot had prior reduced the thrust setting manually. If so, then the first input from the PF was to correct a roll situation while inputting a nose up command with the side stick. As the turbulence thrust setting was for level flight, would the aircraft given weight, COG and reduced speed be more vulnerable to a potential stall situation more rapidly? We know that speed was deteriorating rapidly after the nose up side stick command, but perhaps the pilots didn't know this. However, when the stall warning activated, my thoughts are that it was genuine, the aircraft was saying, "Keep doing what you are doing and we are going to stall". From what data has been released by the BEA, the throttle levers were never moved manually until such time later in the sequence when the throttles were moved to TOGA.

What are your thoughts on this?

JD-EE;
I made that comment shortly after the airspeed-ACARS discussion began in July, 2009 and have done so several times since to get that aspect of the discussion going. "Do nothing" not only makes complete sense to a pilot because the airplane was stable and doesn't need an airspeed indicator to fly, but doing something takes one into unknown territory because one is no longer at the original pitch and power setting and there are no breadcrumbs back and no airspeed. I later posited that the pull-up was in response to one of the memorized items in the UAS drill. Maybe, maybe not, but the airplane didn't pitch up all by itself.

They can be, if one is ham-fisted but that applies to every transport I've flown. Yes, a climb can be induced accidentally, perhaps under stress or turbulence (discussed previously) but I know of no pilot, especially of this experience, who, after having unintentionally applying such movement, would permit the result to develop into what it did without stopping it. An immediate, tiny, gentle correction would be all that was needed to return the airplane to the cruise altitude, while correcting the roll. Hardly a change in 'g' would be needed.

The airplane hand-flies beautifully at cruise altitudes. As always with any transport aircraft, you have to be gentle with it, making tiny control inputs. It will respond aggressively if you make it by using a handful of stick movement. I would not expect that of someone with as much time on the airplane as the F/O had.

Without the data we cannot say.
Yes...so the question is, why did a pitch-up maneuver make sense? Did it have to? We cannot say without the data.

Since the BEA's note, we've all been at it, working overtime to fill in the frustrating holes and blanks, trying to make sense of something that does not lend itself easily to such a task and is difficult to come to terms with. This is a rational enterprise in a thoroughly rational industry so that is how we work it to answer the questions we are all so familiar with and talk about. The notion is discussable as one idea among many. The data's the thing and cranes, not skyhooks must do the heavy lifting where there is a disconnect in this or that theory. But I have yet to see an explanation involving the AFS which fits within the known facts about the aircraft and what we know happened, without the use of skyhooks.

JD-EE

Bearfoil, is it a total blunder if you have a UAS condition to apply UAS procedures?

Let's turn that around. Why initiate UAS c/l if UAS has not been identified?
It is not until 2:10:16 that PNF notes "lost speeds" then "Alternate Law". The a/p is reported to have been lost at 2:10:05 ("I have controls"). From the release, the pilot is said to have made control inputs after a/p loss, and prior to "Lost speeds". If he was flying in ALT LAW at a/p loss, it would seem he was unaware (if his NU was intended). That is unlikely, so he may have made inputs in what was likely Normal Law. All protections, and fly the a/c by hand, in comfort, but without auto pilot. No big.

Did he pull the stick before the stall warning or after? It appears to be "before" in the rude imprecise chronology of the released data.

I agree, and nothing is noted after the "input", said to be left, NU. At the Stall warning, one would think he would relent on the NU (if he indeed knew he was inputting NU) but continue levelling the wings, already underway.

And simply for my own information I have some related questions.

* If he should have waited to apply UAS procedures then how long should he have waited?

Once again, it is not known (though it is strongly suggested) that the a/c had dropped into ALAW, and his wing levelling (and Pitch correction, if intended) was appropriate. At this juncture, it is not known whether the PF continued with his initial corrections, as BEA Say:"...PF made one input...."

* What is the intended purpose of the UAS procedure. It seems to be to slow down the plane to get it into a safer margins condition between stall and overspeed, which was already happening.

One thing, afaik, Stability. Each of the logged incidents has a somewhat different and "ad lib" recovery. Here, the argument against BUSS loses credibility, "It eliminates Anemometric input until landing." Well, speeds are gone, and if they come back, fine no need for BUSS? But if they do not, the need is for speed data for a long safe home, and BUSS might be just the thing.

* Might the pilot have asserted the "slow me down quickly" planning to lose maybe 50 knots by climbing 500' to 1000' and then level off only to have the stall warning derail his thought processes?

I doubt it. The consensus at brief was climbing was ill advised, due warm temps. But that was for a discretionary climb, and as you say, the need for urgent loss of velocity is not eliminated.

* Is everyone happy with the UAS scenario PJ2 suggests?

I am. It was PJ2 who let me know that a vanilla loss of autopilot does not degrade Law, the ship stays in Normal. Now "A/S reads unreliable" was noted eleven seconds after a/p drop. If uas was the deal, a/p would have been dropped at AD disagree, and that would have occurred, of course, prior to loss of autopilot. "I have controls" (after loss of a/p) is first, "Alternate Law" 11 seconds after. It may be merely an anomaly of CVR. I have not seen it explained yet, however.

One does not know, without DFDR data, the accelerations on and around the airframe. A/P disconnect is the first item on UAS c/l. Airbus does it automatically. The AutoPilot also drops out when it cannot keep up with its own manuevering limits.

Machinbird;
By "do nothing", I didn't mean literally sit there, hands-off doing nothing :) , I meant that the airplane had to be hand-flown like a real airplane, maintaining, (not changing) pitch and power, and kept that way in stable flight until things settled down, and the checklists were called for, etc.

JD-EE, take a look at the BEA Reports, (first one I think) for the UAS drill and checklist.

At the MSA [Minimum Safe Altitude around the departure airport], and the aircraft is well away, either in the climb or in cruise and above FL100, one doesn't "wait" for a while to do the drill...one does not do the drill at all. One "levels off", and begins troubleshooting, which means get out the checklists for pitch and power to stabilize the airplane in cruise. There is not sufficient data to determine if this was done or not - we just have the back-stick and the climb.

Turbine D;

I've slowed down to turbulence penetration speed many times. The power change is minimal, the speed change from M0.83 to M0.80 is, (on the road & going from memory) about 9 or 10 kts. Cruise CAS was around 273kts and the stall speed IIRC, was around 196 kts - Bump speed was around 265 kts or so, so both power and speed were a long way (in level cruise flight) from permitting a possible "rapid" degradation of flight conditions, even in heavy turbulence, (which they weren't in). However, a pitch-up of ten degrees at that altitude is unthinkable as the result would be known intuitively.

I hope that helps...

FCOM quotes for supplementary thoughts ...
 

FCOM3 - supplementary techniques - flight controls