AF 447 Search to resume

[UNCTUOUS]

Hopefully, some clarifications.....

GreatBear Said: "UNCTUOUS. I'm still having trouble picturing the auto throttle creep necessary to your "favoured" Mach tuck scenario when, by measuring ground speed until the upset (A/P disconnect, ten seconds after the last position report), the A/C seemed to be traveling at a constant cruise speed... see diagram. Perhaps creep towards the upper right of the coffin corner (+20kts? +40knts? +more? to a flick-roll) occurred in a very short time-span? Short enough not to influence the 0200 to 0210 distance traveled segment?

BBC 178
MARTIN ALDER (A330 pilot). Our speed range is quite limited. Typically it could be ten knots either side of the cruise speed.

If, over a 6 minute period (say), the average of the undetected CAS speed increase was 5 knots, it would only make a half-mile extra distance gained along track. As the pitot became increasingly blocked, the rate of the displayed speed decrease trend (requiring auto-thrust increments to retain a system-stable displayed speed) would accelerate - so logically the last half of that 10+kt undetected speed-gain would occur over the last few minutes. i.e. not a uniform acceleration.

ACARS messages are queued for transmission and can be sent out of sequence. Positions relayed with messages are probably related to msg formatting times.... rather than Tx times. Pls correct me if I'm wrong there. However precision in the reported positions might be an unsupportable assumption anyway.

BOAC said: "Pilots neglected to add power (i.e. left throttles in CLB) after dial-selecting a slowdown to turbulence penetration speed"

BOAC's query: "excuse a dumb q from a non-AB chap - isn't the A/T supposed to do that?

The BBC assertion seems to be (sequentially) that:
a. a lower speed (turb penetration of M0.82) was dialled in due to radar-perceived down-track cell activity

b. at some time before they reached that speed, the "pure precipitation instant freeze-over" of some/all pitots occurred, causing the autopilot and autothrust disconnect.

c. in order to preclude a stall, the pilots would have then had to set 85% power and a 5 deg nose-up attitude (says Alder). Because of all the distracting alarms and reversion to ALT1, they possibly neglected to set the power, lost speed (unbeknownst to them) and stalled. i.e. they'd have been very acclimatized to autothrottles taking care of power to sustain selected speeds (another automation trap). A selection of max power at FL350 DURING a high AoA stall would stand a good chance of compressor stalling one engine (at least). That would introduce a severe asymmetric component and increase the chance of an autorotative "departure".

BBC208 - In 10 previous incidents of airspeed failure, the crews failed to increase thrust immediately.

TONY CABLE. In quite a number of them, it’s clear that the crews were very slow to get onto manual throttle operation.

BOAC said: Unctuous- I am now totally confused as to what you are presenting regarding BBC/ACARS. First you 'present' "An ACARS message from the airplane recorded the sudden onset of critical Mach" with no comment and then you tell GB "Well think it through. Why would there be in any airspeed pitot compromised scenario? It's not a missing message, it's a disabled capability."

Sorry, didn't mean to mince words/semantics, but I did say "recorded" rather than "reported" (in both the original and the response to the query above). I was trying to get across the timing of the triggering event (causing immediate autopilot and autothrust drop-outs), rather than infer that there was a specific Mach crit report.... which of course there couldn't have been - as (at that time) there was no airspeed data from which to derive it.
.
Ref henra's comment about the time taken to accelerate a 210tonne A330 to Mcrit, I've really no idea. The whole leg could've involved a quite slow acceleration (over 30 minutes say) or it may have been <10 mins. All such timing depends upon the rate of ice crystal accretion within the pitots, requiring autothrottle offset for a sustained display speed.
.

[bearfoil]

Fixation seems contagious. Dark, bumpy, somewhat complacent, most likely the Captain not flying, and a/p A/T quit, followed by a workload that doesn't stand a chance while surprised pilots try to get current with 'what is she doing now?'.

Then a slowly rotating a/c hits the sea vertically in "Line of Flight" (Which means what under the circumstances?). Trying to pin a time on the upset, I would suggest it happened quickly, followed by loss of power and hopes of regaining controlled flight.

Speed limit alert and Alternate Law controls. Maybe someone could advance a possibility given the cockpit environment (imagined). Throw in a powerful and unseen airmass with some very strange noises emanating from engines and airframe. With upset comes the need for immediate and precise controls on a good day, once established, how could recovery not be impossible.?

Stall Recovery is not taught here, it is deemed virtually impossible to happen. Yet the a/c can Stall. Therein lies the hubris and the politics for GreatBear. Colgan doesn't train Stall recovery either. At least not back in February, 09. For the record, a recovery was likely impossible with 447 in any case.

bear

Perhaps the defensiveness came from the original heated debates re: FBW.

[Hyperveloce]

delay before exiting the flight enveloppe

it seems that the A/C could have exited its flight enveloppe underspeeding or overspeeding once the A/P & A/THR gone OFF. but I also feel that a large excursion in the high mach numbers (overspeed) take some time to become critical (see http://www.pprune.org/5102068-post4106.html for a basic/not validated attempt to get an idea), even descending at 1 or 2° with N1=95 or 100%. In past cases, 5 crews reacted to stall alarms by a nose down and a descent, -3500 ft max. but if there were increased turbulences close to the Cb cell and/or a turn to avoid a weather hazard, maybe the aerodynamic margins were further narrowed ? I thought that the overspeeding possibility was more probable / explainable given the known conditions/procedures/past cases, but if the plane is really south of LKP (so far from its route) close to the region where the pingers were detected by the French Navy & Thales, then the high altitude upset (in alt2, no alpha prot, no low speed stability) must have been more sudden/immediate than an overspeeding event ? (at 02:10 or early 02:11, not 02:12 or over). Then the other possibility, the low speed stall (BEA: low index vs the pich remaining locked)(possibly accompanied with a large roll excursion complicating the recovery) would appears more probable if the flight ended around 02:15Z in the pinger area ? in one past case of Pitot freezing the A/THR remained engaged and the N1 was varied between 48 and 100% (airchasing in turbulences ?). And as it was said, a low speed stall at high altitude can turn into an overspeed. Can a low speed stall due to a low N1 (50, 60%) locked occur rapidly from A/P A/THR OFF ? (say less than 20-30s)
Jeff

[ChristiaanJ]

Question from an ancient stupid here....

What's wrong with an "overspeed"?
Increased drag, increased buffet, possibly some decrease in controllabily. Oh, and in the books you're not supposed to go there.
But otherwise?

The DC-9 was deliberately flown to over Mach 1 during flight testing.
At least one 747 exceeded Mach 1, even if that was not deliberate, and returned to tell the tale.
The RAF VC-10s were over-powered, and some, towards the end of certain flights, were taken way beyond the formal Mcrit.... and they returned.
I'm sure there are others.

Now, I don't know zip about the A330. So maybe it DOES fall apart the moment you exceed Mcrit by two percentage points. I doubt it.

On the other hand, when an aircraft stalls, it's like any other aircraft.... it no longer flies.
And I mean when it "physically" stalls (not with various FBW protections having kicked in, and the aircraft is mushing down, but with still some kind of control) but with loss of airflow over the wings, and settled in a nice flat spin.

My uninformed conclusion is that they got in the "stall" corner of the coffin corner, not the Mcrit one.

Christian

[BOAC]

they got in the "stall" corner of the coffin corner, not the Mcrit one.

- if you understood 'coffin corner'..........................................

[CONF iture]

Originally Posted by ChristiaanJ

What's wrong with an "overspeed"?

I've no clue either but they build some protections for the 330 and sisters ... there must be a reason, a way to allow some structural weight saving maybe, or is it just for the fun to take control away from pilots ...

[ChristiaanJ]

Originally Posted by CONF iture

I've no clue either but they build some protections for the 330 and sisters ... there must be a reason, a way to allow some structural weight saving maybe, or is it just for the fun to take control away from pilots ...

CONF,
As I said, I'm an ancient...... Concorde.... remember?

We didn't call them "protections" at the time.
MAX CRUISE was simply an autopilot mode, that kept the aircraft within those three basic limitations: Mach 2.04, Tt 127° C, and IAS 530 kts.

Mostly there to make sure we could use the aircraft again next time.... or more seriously, to make sure it would would stay within the certification limits, to meet the certified supersonic cycles requirements.
And yes, we went up to Mach 2.23, and we did not use that aircraft (Delta Golf) again for commercial flight afterwards, even if it was thought about seriously at one time.

I've seen M 0.8 or suchlike mentioned for the A330. Was she certified to M 0.9 and beyond (as she should have been)? Or DID they "save some structural weight" somewhere?

CJ

[HarryMann]

What's wrong with an "overspeed"?

In smooth air when you know what's what and have been there before....

or when it's rough as hell and you don't really know what's what?

[CONF iture]

FCTM for reference :

HIGH SPEED PROTECTION

When flying beyond maximum design speeds VD/MD (which are greater that VMO/MMO), there is an increased potential for aircraft control difficulties and structural concerns, due to high air loads. Therefore, the margin between
VMO/MMO and VD/MD must be such that any possible overshoot of the normal
flight envelope should not cause any major difficulty.
High speed protection adds a positive nose-up G demand to a sidestick order, in order to protect the aircraft, in the event of a dive or vertical upset. As a result, this enables a reduction in the margin betwen VMO/MMO and VD/MD.

Therefore, in a dive situation:
. If there is no sidestick input on the sidestick, the aircraft will slightly overshoot VMO/MMO and fly back towards the envelope.
. If the sidestick is maintained full forward, the aircraft will significantly overshoot VMO/MMO without reaching VD/MD. At approximately VMO + 16 / MMO + 0.04, the pitch nose-down authority smoothly reduces to zero (which does not mean that the aircraft stabilizes at that speed).

MMO = M.86
MD obviously above M.90 ?

[OVERTALK]

At what point is the High Speed Protection nullified?

My interpretation of the intended thrust of CONFiture's post on High-speed Protection follows.

As Normal Law would've been in play when (and if) the A330 hit Mcrit, for these protections to have been then overcome (and an upset occur), what particular anomaly could've been involved at the same point or as an immediate preliminary (i.e. to cause a transitioning to a degraded law - and kill the protections)?

It's becoming a bit "chicken and eggish".

but IMHO, it's looking more and more like it would've been the developing airspeed splits that started the downgrade in FC Laws and caused autopilot and autothrust to jump out. That would have instantly placed the pilot in the cross-hairs for an upset, for all the usual reasons:

a. surprise,

b. unfamiliarity,

c. loss of airspeed read-out,

d. engine parameter screen not being displayed,

e. a THS out-of-trim pitch-down into a high mach (possibly compounded by a misinterpretation of the situation as a stall - and max power and a lower nose attitude being selected),

f. or alternatively, mere distraction due to the aural and visual alarms allowing an unusual attitude to develop. Aircraft at height under manual control in ALT1 would be inherently unstable, particularly if manual thrust was changed significantly to affect the trim-state.

This scenario tends to accord more with what's known of the prior instances of pitot icing at high levels.

[Machinbird]

What's wrong with an "overspeed"?

In an aircraft designed for subsonic flight, there is a lot of reason not to get too close to M 1.0. The primary reason is the center of pressure shift which occurs rather abruptly.
The aerodynamic center of the wing shifts from the quarter chord to the 50% chord (Actually the whole airframe generates lift based on the Angle of Attack of the individual components).
Since your lift is being generated further aft, the nose wants to tuck. The only way to get it back up will probably be THS trim since the elevator will be ineffective, but that has hazards all of its own, particularly when you decelerate below the transition point.
Then you might well break something when all that nose up trim takes hold.
If you haven't seen it before, take a look at the A380 flutter test video. YouTube - Airbus A380 Flutter Test
In addition to Mach tuck and flutter when you exceed the demonstrated envelope, subsonic airliners buffet and occasionally lose bits off the aircraft.
It was impressive to see how much nose up stabilator (stabilator itself was nose down) was required to fly level in a supersonic F-4.

[CONF iture]

Originally Posted by OVERTALK

As Normal Law would've been in play when (and if) the A330 hit Mcrit, for these protections to have been then overcome (and an upset occur), what particular anomaly could've been involved at the same point or as an immediate preliminary (i.e. to cause a transitioning to a degraded law - and kill the protections)?

If all airspeed indicators under read at a similar level, the high speed protection won't trigger even if the real MMO is already passed. At this point the airplane is still in Normal Law as no malfunctions have been identified. The message for the Pitot tubes malfunction can appear only later on when finally the split between airspeeds reach the required threshold ... only then appear the ACARS messages as we know them ...

Very unlikely does not mean impossible ...

[jcjeant]

Hi,

Sorry for return again on the "tail problem" but I read something interesting there ... some comparison of crashs outcome about the tail ....

AF 447 : questions de dérives : Les dossiers noirs du transport aérien

From up to bottom
Pic N°1 is AF447 tail (lost on sea "BEA" or in air ?????? )
Pic N°2 is raw drawing of attachement system (A330 and A300 same system of attachement)
Pic N°3 is tail at Tripoli crash on ground (stay attached)
Pic N°4 is tail Monterey crash A300 on ground (stay attached)
Pic N°5 is tail NY crash A300 (lost in flight)

[bearfoil]

jcjeant

Thank you. Food for thought. It comes to mind that the tail (complete) is built as a mating of cantilevered stabilizers and articulating control surfaces. When confronted with an impact into the ground, or water, it fails as a unit, due to overwhelming and reasonably conforming and directional stress. Failure of individual surfaces and panels suggests a mix of stressors, rather than one massive arrestment. It is intuitive. An impact with the ground or water would then have a tendency to remove much of the tail at once. This suggests to me (not conclusively, but certainly as a result of patent failures), that 447's V/S came off as a result of stress that was not sufficient to remove panels and surfaces that were not attendant to lateral failure, eg: HS or dorsal skin. I consider it odd that the VS was found alone, likewise the elevator. This isn't probitive, but it meets the test of possibilities. The elevator has its trailing edge fractured from the main panel which has removed from the THS, but the TE remains attached to the elevator. This just doesn't square with a single and overwhelming failure of all the parts. Likewise, the splinter at the TE of the elevator should have been explosively removed, and here's the thing: Separating from the THS infers a separate stress from that which (is proposed per BEA); the description of impact by BEA would have the reader think that one impact was suffered by an all intact tail, there would have been no time separation for a sequential, and partial failure of any part, it had to come off the tail as a unit, like the V/S Rudder.

Any failure which served to have distributed parts in the SEA, would have thus been caused by the airstream, and prior to impact.

bear

[HarryMann]

Any failure which served to have distributed parts in the SEA, would have thus been caused by the airstream, and prior to impact.

Could we have at least some guide to your qualifications to make that statement please?

[bearfoil]

Harry Mann

The statement stands; I am sure my qualifications would not satisfy you, and you are welcome to rebut or invite another to do so. If the mods allow it, I propose that you do as well, and consider the statement on its own "merits", subject of course to your critique.

[MCScott]

UNCTUOUS comments on BBC scenario

All,

I am not a pilot, nor an aerospace engineer per se. But I do have an MSME degree, have spent most of my life in computer programming, and have an elementary grasp of statistics and probability. I've also followed these AF447 threads with great interest, since the beginning a year ago. I have some comments regarding UNCTUOUS' very interesting theory, as expressed in comments to the BBC script.

- The BBC obviously did not do much research on previous incidents of IAS inconsistencies. As you said, most of these seem to have occurred in contexts that suggest ice crystals, rather than supercooled water, as the primary cause.

- I had understood the possibility, before, that engine thrust at AT disconnect might have been different than the pilots may have intuitively expected, due to intervention by the flight management system in response to turbulence settings or other phenomena. I had not known, though, that selection of the "fuel monitoring" screen would further obscure the true current engine thrust settings.

- Without doing any in-depth analysis, your theory of relatively gradual degradation in speed readings leading to mach-effect overspeed phenomena, while plausible, does not seem very likely. The pitot tube is a very simple analog device, whose basic design predates the dawn of aviation. It operates primarily on pressure difference, rather than flow, per se - very little air actually flows through the tube (some air does exit through drain holes, etc, but that is by design a small amount, or the tube won't serve its intended purpose). This in turn means that the tube will need to be almost fully blocked before significant pressure and airspeed indication loss is observed. And that, in turn, means that airspeed indication loss due to icing will very, very likely be 1). rapid, 2). unstable, and 3). inconsistent between redundant tubes. This is consistent with the summaries of known IAS anomalies previous to AF447. While I have no knowledge of the airspeed validation algorithms in the Airbus flight management systems, it seems likely that they would be programmed to reject airspeed data inconsistent with the physical capabilities of the aircraft and known limits of windspeed variation, which would further limit the kind of airspeed variation which could lead to your scenario. It is certainly possible that high-altitude upset could have occurred due to overspeed because of pilot input in a confusing environment with no accurate feedback, but I don't see that overspeed due to AT correction is especially likely.

- The issue of training for high-altitude upset (stall, mach-tuck, etc) is interesting. The apparent inability of most commercial simulators to simulate stall in combinations of pitch and roll is something I'd not understood. Beyond that, there is the fact that humans are not machines, and will react differently when primed to expect surprises than when they are lulled into complacence by thousands of hours of uneventful automated flight at cruise altitude. The point made by the military pilot - that military craft are almost always manually flown, and thus the pilot is always in that "zone of awareness", is significant. And then there's the question: with all of the air miles flown today, and the fragmentation of air traveler miles into more and more flights on smaller and smaller planes, can we possibly expect all of the burgeoning ranks of pilots to have the degree of manual flight skill needed to recover from the hypothetical AF447 situations?

- It is easy to identify accidents related to failures of automated flight systems, or the misuse thereof. It is impossible to track accidents avoided by such systems. The overall statistics, which seem to show fewer and fewer accidents per air mile traveled, suggest that if automation is either a net safety benefit, or it is being overwhelmed by other safety improvements. I have no data to say one way or another.

[CONF iture]

What about this one ?



Amazing, it didn't really happen in the middle of the Atlantic, but that's the only photo we have seen so far ... Some know how to control the information.

[Machinbird]

What about this one ?

Yes, but we know it flew into the water nose down at fair speed.
I think what we are seeing is that the Airbus VS structure is fairly robust and will survive a fair beating relatively intact. The attach lugs tell the story of how they came off the rest of the airframe.

[HazelNuts39]

Buffet margins

(DingDong) Ladies and gentlemen, your captain speaking:
We have just arrived at our cruise altitude of 35,000 feet, popularly known as "the coffin corner"'. Enjoy your flight with XYZ Airlines ...

EDIT:: Just to put Martin Alder's remark in the BBC documentary into perspective: in this case as shown in the graphic, the 'coffin corner' is actually 102 kt wide: from M=0,58/192 ktCAS to M=0,86/294 ktCAS.

HN39