The blocked pitot explanation certainly makes the possibility look feasible, but surely it would be an umpteen-million to one shot that all of them got blocked at the same time.

How many does an A330 have? A left, a right and a backup?

Four I think, but they all point into the same slush.

Three actually, and that slush could effectively block all of them at the same rate, so as nobody disagree : NO warning NO caution ...

Yes, the spec is a bit broad. I was thinking that for an R/C oscillator entombed in parafin wax within a very solid steel/titanium case, the external temperature would eventually have an effect on its frequency and that a xtal/piezo filter may be in the R/C loop. Thinking in terms of manufacturing to meet the nominal frequency +/- 50Hz, I'd move away from an R/C circuit and use something less affected by temperature and voltage. The stuff I can find just quotes the tolerances without any schematics to back it up.

As there are three timing issues involved, 37.5kHz [output frequency], 1Hz [period] and 375/37500Hz - 10m/s [duration]; simple xtal divider logic would seem more practical.

With regard to the Doppler shift, my feeling is that differentiating that from multi-path would be a nightmare. Perhaps careful reference to the local bathymetry may help sort it out. Best of luck to them!

mm43

And here they all are, on the actual aircraft in question photographed a day or two before it crashed.

http://i49.tinypic.com/zlooht.jpg

mm43

The 2 black thingies inboard and aft of the TAT probes might be “Ice Detection Probes”.

I am only new to this so I may be mistaken !

SPARGO Craic

You are correct - the two items labelled "LOC Antenna" are the ice detection probes. :ok:

Wristwatches typically use 32.768 kHz quartz tuning fork oscillators. They are also very low power and survive modest shocks. I'm not sure they'd survive the shock and vibration specified for the recorders. Going the divider route with a higher frequency crystal uses more power for no practical benefit. The crystal is still vulnerable.

You can, however, find nice stable resistors that hold 0.1% or less over temperature. You can also find nice "NPO" ceramic capacitors that hold within a few dozen ppm/degree C. That's where the temperature variation would come from. Manufacturing tolerances of the capacitors would have to be in the 2.5% range or else trimmed with a parallel resistor for the R in the RC that is calculated and soldered in during manufacture. For reasons of cheap getting it within 1kHz or about 2.5% is "good enough."

After weeks of cold soaking there would be no further temperature inspired variance in the oscillator frequency. And there are oscillator circuits that are remarkably insensitive to battery voltage that could be used. So figuring out that they heard two distinct signals should not be all that difficult.

{^_^}

Again see old thread for more detail discussion last year

Not slush at -40 degC, probably micro-fine ice crystals, which would be harder to melt out, could also insulate the TAT probe, and cause problems with engines...

Icing not expected at cruise altitude, so computer fault? Clearly used to be an issue:-

http://www.iag-inc.com/premium/airbu...ablespeeds.pdf

Subtle Difference in the Blockage Process
 

CONF iture said: Slush? Is not the problem an ingestion of ice crystals over time within Cirro-Stratus/CirroCumulus cloud rather than a "heater OFF" style freezing of moisture at a drain hole (followed by a blockage, partial or complete of the pitot tube itself)? i.e. those high-level cloud types are just composed of ice crystals. That's the important difference.

Concentrate upon the possible difference in the two blockage processes and it may become clearer as to what actual effect it had upon automation.... and why the Thales tubes' heaters were more prone to being overcome. i.e. they were never designed to combat being hit by already frozen ice particles. Think in terms of calories/BTU's in and out - and it's not hard to figure why the pitot heaters finally were overcome. That's the design deficiency.

It would have to be something unusual yet simultaneously common to all three pitot heads (as CONF iture said:"no alarm-raising disagree" to provoke a triple redundancy-based comparative alarm).

........a. a given concentration of ice particles (i.e. fairly thick and continuous cloud at quite a low temperature)
........
........b. a TAT low enough that the accumulation rate of ice particles would ultimately overpower the heating capability of the pitot heaters.
........
........c. If the accumulation/compaction rate was sufficiently subtle (i.e. slow) then the effect would be insidious. Thinking here that the speed loss seen by the system would not be actual but would be acted upon by autothrottle incremental increases, thereby putting the constant airspeed/mach (but actually accelerating) aircraft ever closer to the corner of the coffin corner envelope at that height.
........
When, in turbulence perhaps, the aircraft actually hit that critical mach (that you must avoid), what happened then?
........
Did the autopilot disconnect? Did the pilots assume a coffin corner aerodynamic stall (because they were seeing a low IAS), and then take stall recovery action (lower the nose/cob the throttles?).
........
What would the effect be of doing that? Last time I did "alt and comp" (and had a mach-induced pitch-up), I was intentionally inverted (because I had a very inquiring mind in those days). It was an interesting (and quite aerobatic) ride and I had much greater respect for MCrit after that.
........
Can you ever (and have you ever?) ran this scenario in a simulator? If so, does it faithfully replicate what would happen in the airplane? Or is this subtle and insidious pitot blockage scenario a totally unforeseen circumstance, despite all the prior instances of Thales pitot induced upsets?
I have the gut feeling that to eventually determine with some certainty what happened to cause AF447's loss of control, we must first completely understand the precipitating event.
........

Is it possible for probes to be damaged or removed by hail, has anybody heard of this happening?

My thoughts exactly; nicely put. I might add that it's likely the design of these pingers hasn't changed much in decades (maybe replacing mica caps with NPO ceramic?). Once a design is certified, they probably follow the principle: "if it's not broken, don't fix it."

Hi,
The main problem with such scenario is that it just doesn't fit with ACARS reported:
1. Pitot Probes freezing was actually detected by the systems;
2. Auto Pilot and Auto Throttle also disengaged automatically as a consequence of 1 (as designed, and it was not a manual disconnection).

How the pilots reacted is certainly open to speculation but nothing is telling us at this point that an insidious freezing was an issue for the system to change any of the flight parameters. So an upset due to speed issues is very less likely to have happened by itself if it was the only factor in cause.

S~
Olivier

Without regard to 447, I wonder if there hasn't been an overlooked accumulation of ash affecting the bleed holes in pitot probes of planes in the past two months? It would show up as increased speed error. By how much would depend on the design of the probe.

GB

Blockage could have been simultaneous, but unblockage was not ... so appear the ACARS messages, but at this time the overall situation is already highly degraded ...

Hi,
Well, why not if you want it this way... Nonetheless, this sequence is very unlikely: In one hand, simultaneous blockage of 3 probes located at different spot alongside the fuselage when flightpath is notoriously always a bit assymetrical would be a probability seriously close to zero while, on the other hand, unfreezing is due to simultaneous imput of heaters with variations very limited in time. Consequently, all those related "probes" faults would probably not be displayed at all in the first place.
S~
Olivier

Takata and others

The ACARS msg content is that ADIRU:s reported air-speed values disagree. Pitot problem is only one of many possibilities. The error can occur anywhere between the pressure on the pressure transducers (dynamic and static) and the computed speed value produced by the corresponding ADIRU. It has apparently happeded that ADIRUs have malfunctioned (their IR-part) and the error not properly detected/saved by the BIT system (on a Boing, however). Don't look at only one possibility, please.

The Flaps were stowed at impact. (BEA). The spoilers are part of the upper wing surface. The recovered Spoiler shows evidence of collapse in a deployed position.

The spoiler shows evidence of no water impact from underneath, its damage is consistent with overspeed, aerodynamic collapse and removal by the airstream. The Radome, though recovered in pieces, shows much the same sort of failure.

The Vertical Stabilizer (leading edge) shows collapse as though "pushed back". Its damage is symmetrical, intimating that the leading edge was impacted by a force that was distributed while the VS was a part of the airframe. If water impact caused this symmetric failure, the VS would have to have entered the water edge on; though possible, this is highly unlikely. The damage to the vertical take up arm (the missing "end"), cannot have occurred without damage to the Hinge axis, unless it was shaken loose by vibration, something that in itself would not have disrupted the integrity of the Rudder Hinges.

A/P. The auto system uncouples per the following parameters: an inability to keep the a/c from rolling greater than 45 degrees, pitching up more than 15 degrees, pitching down more than nine degrees. If it is assumed that the uncoupling occurred due to one or more of the limits having been attained, it must also be assumed that these rather extensive excursions were abrupt; a pilot would not allow an a/c to slowly degrade its ride to these limits. The ACARS are maintenance hallmarks only, and are sequenced plus or minus one minute? SATCOM has some antenna requirements, so during the cascade of these messages, the a/c must have been in antenna posture aspect that limits conjecture about the end relative to upset.. The Spoiler speaks of a need to slow, other evidence points to overspeed. How fast?

The answer I think is in the debris location. Where were the VS, radome bits, and bodies? This means the first body, laptop and VS establish the first of a series of trajectories (airborne loss of integrity), or, the only one. The Ocean is a mystery to me, so I'll keep reading and staring at the work of others.

bear

Independently, they could have various meaning. But this serie's meaning is quite obvious. Pitot problem is clearly identified with an ACARS telling that each pitot probe is reporting a different value! Then, ADR disagree (unrelated to IR).
You should have another look at BEA's reports as this part is explained in details.

Or, as BEA mentioned also, it possibly separated due to water pressure when wings sank while being (fully/partially) deployed due to (vertical) forces at impact.

... and this aircraft then crashed 5 minutes later, level, with low horizontal and strong vertical acceleration? ... A/P & A/T going off were the first ACARS to be sent and the reason why they disconnected doesn't look obviously documented by the 20+ following ACARS?...

But nobody can find any submerged wreckage around this place!

If there was any measurable evidence of overspeed by studying the wreckage, it should have been reported by investigation in last december's report like they did for other parts showing some clues about what happened. If it was not reported, it means that, either they disagree about it, either that no evidence at all of overspeed was actually found. Otherwise, it would have been a huge conspiration in order to hide something so obvious for you.

Then, look at the daily debris fields recovered during searches (BEA annexes). But as floatting stuff is moving, this is not the best way to draw any conclusion from that, even if the tailfin has been actually recovered in the middle of bodies and other parts of the airframe. Only what is still at the bottom of the ocean would be enlighting, starting with the wreck's location compared to flight plan.
S~
Olivier

takata
Please take a look at the systems schematic. This has been shown multiple times in the old thread. Any signal had to pass a digitizer then to the ADIRU until a speed is computed and the ADR value becomes available.

bearfoil;
I cannot claim to be particularly knowledgeable about structures, but some things would seem to me to be fairly obvious by just looking at the drawings so kindly provided by BEA. Take a look at the eight hinge brackets. Viewed from above, each is a triangular structure, the base of which is the rear spar of the V/S, and the top is the rudder hinge pivot. That structure is shaped to take loads perpendicular to the hinge line, and is very rigid with respect to these loads.
Now look at these brackets in side view, and you see just slender rods that are quite flexible with respect to loads acting along the hinge line, normal to the triangular planes. They can take a relatively large displacement in the direction of the hinge line without permanent deformation or failure. They are not intended to take any loads in that direction, because that function is exclusively assigned to the vertical load pick-up arm '36g', which is rigid in that direction and dimensioned to resist a load of 36 times the weight of the rudder. That arm fractured only locally but otherwise stayed in place, i.e. the overall geometry of the structure was not impaired, and the hinge brackets were therefore not subjected to significant deformation perpendicular to their plane. It is therefore only logical that these brackets show no sign of failure.

HN39

So long since I messed with oscillators, that the KISS principle deserted me! A fundamental frequency oscillator combined with a simple divider chain for other timings, and all well within specs, is as you say "good enough".http://images.ibsrv.net/ibsrv/res/sr...ies/thumbs.gif

mm43

HazelNuts39

I am away from those photos and drawings. From memory, there are two conditions that are puzzling from a structural standpoint. The vertical arm in the VS with its end missing, and the "rods" for lateral stiffening on the aft tail decking, common to an eyebrow rail mount between the right and left VS mounting saddles (three conditions). In both cases it shows that these "added" structural members will not fail (or even come into play) until after the failure of the structure "benefitting" from their installation. Note in particular the bead of sealant at the shoulder of the through bolt.at 36g. This bead is unaffected by any slip that may have caused the arm to lose its tip. This through bolt cannot be loose in its hole, it must be solidly inserted, perhaps driven in at assembly. I would enjoy inspecting a disassembled bolt/arm joint for fit, any looseness would aggravate its purpose; So the tip went missing without vertical movement, it has to be something other than a design condition.

At the VS/fuselage remnant a good picture of the lateral snubbing rods shows failure in the lateral plane (imo). Once again, the VS is not installed loosely, wear on the saddle brackets and eye bolts cannot be allowed. Since the VS cannot move relative to the tail decking, any "movement" that needs protection can only occur when or as the VS undergoes removal by catastrophic forces. I also have the suspicion that as the VS "rocks" back and forth in failure, the rods act as jacks, aggravating the strain that causes the loss of this Fin. In other words, to me, there seems no allowance for a day to day need for these assemblies, only in failure does their use seem to satisfy their design. Your thoughts?

bear

lateral load pick-ups
 

bearfoil,

After arm ‘36g’, now to the lateral load pick-ups in the vertical stabilizer attachments. It took me a little while to understand how they failed. Please forgive me a somewhat lengthy ‘approach’ before I ‘land’ on them. According to BEA’s analysis, the V/S departed in a forward motion relative to the fuselage, under inertia forces as the forward motion of the airplane was decelerated by water forces acting on the lower parts of the airplane.

The inertia force on the V/S acts through the center of gravity of V/S plus rudder, in a direction opposite to the deceleration of the airplane, i.e. forward and essentially horizontal. It is opposed at the level of the three V/S attachments by rearward forces. The forward and rearward forces together result a pitch-down moment on the V/S, which is opposed by an upward or push-force at the front attachment, and a downward or pull-force at the aft attachment.

The forces exerted by the V/S on the fuselage are opposite to those just described: forward at the three attachments, downward (push) at the front attachment, and upward (pull) at the aft attachment. The forward attachment is thus pushed down on the rear pressure bulkhead beneath it, while the hoop-shaped fuselage frames under the aft attachment are less resistant to the pull forces exerted on it, and fail. The center attachment then gets twice that pullforce (because the moment arm is halved) and fails also. At the forward attachment, the back-up structure of the V/S front spar is weaker than the fuselage, so it fails. The V/S is now free to move forward, taking parts of the fuselage frames below the center and aft attachments with it which, held back at their fracture point, swing backwards.

The lateral load pick-ups at the center and aft attachments are placed in the extended plane of the respective V/S spars. It should be noted that the extension of the center spar passes in front of the pivot of the center attachment, whereas the extension of the aft spar, and thus the lateral load pick-ups attached to it, is aft of the pivot of its attachment. Therefore, when the frames together with the lower part of the attachments between them swing backwards, both pick-ups at the center attachment fail symmetrically in tension, while those at the aft attachment fail in compression.

Could the failure of the lateral load pick-ups be the result of lateral loads? Their function is to prevent or alleviate lateral loads on the male and female lugs of the six attachments. I cannot see a way in which they would fail in left/right symmetry, without lateral failure of the main lugs or a failure in the structure between left and right attachments.

Does this remove some of your doubts regarding BEA’s description of the tail separation?

HN39

Unlikely and offers a rather pessimistic view on structural engineers' comprehension...

Iced Pitots
 

There seems to have been wide agreement here for a day or two that − in level flight, right at the beginning of this event − whichever ASIs were showing erroneous readings are more likely to have been over-reading the actual IAS/CAS, than under-reading it. Thus, if the crew were misled by the false readings, they might have slowed the aeroplane into a stall near the cruise altitude. This would be likely to produce a far steeper descent profile, fitting Machinbird's and mm43's suggestion that the search should not ignore areas close to the LKP. One or two of you have even described the well-known effect of a blocked pitot in the climb (increasing over-reading). Remember: this event started in level flight.

My argument, originally indicated in a PS to my "Au Sud?" post #1175, 48 hours ago (and again in post #1180) , is that iced pitot probes are more likely to result in UNDER-readings. If these (falsely) showed the IAS as being dangerously low, this might have led the crew instinctively to opt for lowering the nose rapidly, leaving the cruise altitude (FL350) into an accelerating descent.

So what does the BEA think? They studied the 13 best-documented A330/340 events involving ASI (ADR) anomalies. Their results, detailed in Interim Report #2 (December), include the following. (My annotations are in square brackets, and under-scorings are all mine.)

"The speed anomalies can be characterised by two distinct signatures:
- intermittent falls (peaks);[sic]
- Fall followed by levelling off (continuous period).
They were accompanied by an instant increase in displayed static temperature (and total when recorded), and with a 'drop' in altitude on airplanes equipped with altimetric correction (A330-200)."
[These phenomena imply under-reading of CAS – ref. 1.6.11.6 of the same report.]
"In both cases, the lower speed limits recorded were below 100 knots.
The maximum continuous duration of invalid reported speeds was three minutes and twenty seconds."

"...Nine cases of triggering of the stall warning were observed...
...(this) triggers when the angle of attack passes a variable threshold value. All of these warnings are explicable by the fact that the aircraft is in Alternate law at cruise Mach, and in turbulent zones. Only one... was caused by clear inputs on the controls."
[This seems to imply that the warning is genuine, but apparently triggered by a high angle-of-attack (or rate of increase??) in turbulence.]

So, with a possible exception where they have ambiguously used the term "peaks" (apparently when describing intermittent falls, so something has been lost in the English translation), they are describing false, often alarming, falls in displayed IAS/CAS. These are often combined with (phase-advanced??) stall warnings caused by rapid AoA fluctuations, presumably in what many pilots would call "severe" turbulence.

One last quote:
"The pilot flying gives priority to piloting the airplane and to the flight path, by maintaining a cruise altitude or by performing a descent to increase the margins for evolution within the flight envelope..." [To get out of 'coffin-corner'.]
"The descent can also be decided following the triggering of a stall warning."

In the descent, with rising static pressure, the (possibly) trapped pitot pressure would lead the ASI to under-read even more.

Chris

Back in Post #1179 in which I posted the ACARS list of received messages, I mentioned that the message timing was to the "integer of the minute". On further reference to the BEA Interim Report No.1, I am bound to say that the wording should actually be "rounded to the nearest minute", as per:-
That being the case, then the #0210 AOC position could be between 02:09:31 and 02:10:30, but as it takes precedence, it was most likely timed and transmitted at 02:10:30 and received 4 seconds later.

Hope that doesn't confuse the issues, and the original post has been edited.

mm43

iced pitots
 

Chris;

Not me. I feel it is not impossible but somewhat unlikely. The UAS procedure mentions the possibility of an overspeed warning, but I think that is primarily to cover the case of an airplane taking with probes and/or ports blanked off. On page 60 Sensor Validation points to the possibility of pitot overpressure due to drain blockage, but adds that he doesn’t know by which percentage. Until someone comes up with that information, I’m inclined to think that the effect is quite small.

Where you had a difficulty with the translation, the french version reads:
To my knowledge, stall warnings are not phase advanced (stickpushers sometimes are). Based on the cL-AoA data obtained from the QF72 event (the whole manoeuvre was between Mach=0,808 and 0,818), at a typical cruise turbulence penetration condition of Mach=0,8; FL 350; 200t the margin to the onset of stall warning at AoA=4,2 degrees permits a loadfactor of 1,35 g.

HN39

see image here: https://docs.google.com/leaf?id=0B0C...NjRhZGMw&hl=fr

Before the post gets so far back there that it is hard to find, would some of the Airbus jockeys please look at page 60 of this thread, post #1184 http://www.pprune.org/5716260-post1184.html, at the part of the posting entitled "Out of the Loop". The question I have is, should Mmo/Vmo protection be tripped in Normal law by a simultaneous bogus pressure build up in each pitot system, how would you recover?
From Pilotaydin's original post last year, it appears that this scenario is sufficient to put a cruising aircraft promptly into a stall.
If you assume the AF447 aircrew deslected autothrottles in an attempt to get control of the situation early on in the ACARs sequence of events, then it seems to me that this scenario would likely put the aircraft in the water in a deep stall in the appropriate time frame.
Don't worry about how this hypothetical pressure build up could happen just yet. The key questions I am hoping to answer are,
(1) Would this scenario likely proceed to a sad conclusion?
(2) Is Pilotaydin's sim experience technically flawed in its description?
(3) If the sim experience is technically valid, how could you break the sequence of events?
I'm just an old steam gauge vintage retired military guy and these questions are better answered by someone on the front lines.
Thanks for your thoughts.
Machinbird

Hi,
Well, an elementary methodology is at least to start with correct settings before making this kind of assumption:
This proposition is invalid as AF447 Autothrottles (as well as autopilot) were OFF from the start of the sequence (disengagement by the system - not pilot action), due to three pitot probes reporting different readings, causing ADR DISAGREE. From that point also, most of the flight enveloppe protections were OFF due to the same pitot issues causing the selection of flight mode ALTERNATE 2 by the system.
In short, at 02.10, the system identified probes issues and auto-protected the aircraft by disconnecting all the systems that could be contaminated by these corrupted data. After this point, AF447 was flown in manual ALT2 mode.

Now, I'm feeling that we are back to the first days following the crash because some people just want to push their "system upset theory" (or agenda) disregarding any available information which doesn't fit with their own preconcieved theory.

Having spent a lot of time studying this issue and having participated myself in this "old thread", my schematics and ACARS/fault study doesn't point to this conclusion at all. Triple PROBE fault is the origin of ADR DISAGREE as shown by ATA codes reported. This fault sequence is quite coherent while your proposition (multiple and simultaneous module failure) disregarding reported PROBE faults is certainly not.
S~
Olivier

auv-ee
"If you can still get the famn dool parts."

(I'm enough of an antique that I know from hard experience this is a pesky detail. I wonder if anybody could build the GPS frequency synthesizer and distribution unit today to the schematics generated from my design back in the mid 70s. It used already completely obsolate TTL parts because they were rad hard and the newer parts were not hard enough.)

{^_-}

Thank you Takata,
I'm not "invested" in any particular theory on AF447 other than that airspeed indications can rise as well as fall with an iced pitot tube.

The comment relative to autothrottles was made in an attempt to synchronize a Vmo/Mmo pitchup in Normal Law with the observed information (That is, it would have already happened before the first ACARS event) but I have to agree, it is improbable a crew would put the throttles to idle and the airspeed mismatch is quite sufficient to disconnect the autothrottles.
To summarize your reply though, Alt 2 flight law would remove the Vmo/Mmo protections and restore control.
(FWIW lost most of my references in a recent computer crash and haven't had time to rebuild)
Thank you for taking the time to respond.
MB

takata
"Triple probe fail". This is not necessary at all to generate the ACARS msg ADR disagree. Two bad and different speed values are enough, because the system can't decide which of the three values is correct.
A sobering example from my own safety research field. We had a near catastrophic event in the Forsmark nuclear power station a couple of years ago when two out of four almost identical parallell safey systems failed at the same time (always assumed impossible). The two failing ones together managed to shut down the instrumentation in the control room. The two surviving ones kept the reactor going. These systems had higher safety requirement rating than most things on any aircraft.
I can even imagine some kind of mechanical fault in the rack containing the ADIRU:s from the inertial forces when going through bad weather causing bad and erratic signals. Even some loose object may have hit them. Both of these possibilities have been subject to safety directives, although for other aircraft types.

cL vs AoA
 

A few weeks ago I posted a link to a graph of liftcoefficient vs angle of attack, derived from DFDR data in the ATSB Interim Report on an uncommanded pitch-down event during flight QF72. That graph has been updated. The main change is that a line approximately representing the stall warning threshold has been added, and the extrapolations of the various cL-AoA lines above that threshold have been suppressed. Anybody who has downloaded the old graph is recommended to discard it and to replace it with this one:
https://docs.google.com/leaf?id=0B0C...MzMyNDlk&hl=fr

I too am in a business where the products have longer production runs than their components, so I certainly agree about substitutions. I'd be willing to bet that the topology of the circuit has not changed much over the years.

Regarding mm43's conjecture that dividers are used, I bet not. The spec sheet (Dukane model DK120 and DK100 Underwater acoustic locating beacons) shows a different tolerance for the frequency (1kHz/37.5kHz =2.7%) than for the pulse width and repetition rate, each at 10% (each listed as 9ms and 0.9sec from a likely, but unstated, nominal of 10ms and 1 sec, respectively). This means that they are not derived from the same time base. Dukane pingers have been around since 1969 (Dukane Audio Visual, Ultrasonics, and Seacom Divisions), and although the first CMOS ICs appeared in 1968 (CMOS - Development history), the pingers probably were not originally designed with digital dividers. I expect that the design is based on independent 10% tolerance one-shot timers for the pulse length and repetition rate.

Hi,
We surely can imagine a lot of things, including a space alien mothership intervention. My point is not about ADRs racks but the specific PROBE-PITOT 1+2/2+3/1+3 (9DA) fault you are just ignoring. I'll stick with my proposition if you don't mind.
S~
Olivier

HazelNuts39

Thank you for your response. I must say that on short notice, your description of events in defense of forward separation at water impact is most elegant. It is scholarly and compelling.

Before I respond, I wish to thank you for sharing your experience here. This device is something new to democracy, and will at least keep interest in important matters alive, not necessarily subject to the mundane lurching from screed to screed that is a legacy from poor reporting in the past.

BEA explains 447's water entry aspect as "En Ligne de Vol", with a port yaw. This is a challenge to logic; the description goes on to describe a left wing low. A left bank, slight Pitch up, but 'In line of flight'? I won't have it. The description is meant to convey a Point of View, but framed as a Finding of fact.

You have relied on evidence, and bravo. However, the evidence you submit is equally persuasive when applied to another "Point of View". BEA's description describes virtual "controlled flight". If you don't think so, will you at least admit that it can be taken so? They then proceed to say "low forward speed". High vertical acceleration is an attack on the posture of the a/c as described, so shall we excuse the report as simply a misunderstanding in transit from the French to English? No.
English is the language of International flight, by agreement.

There is a gap in ACARS/Aircraft radio connection. Even though BEA would have us envision an a/c that is complete and on the verge of recovery from jet upset as it unluckily hits the Water first, this visual is an artifice. My evidence? ACARS itself.
Over what is believed to be the time frame of this flight's demise, it infers some controllability on the part of the flight crew. If the link (connection) to the satellite demands line of sight, we can accept an upset in Pitch, with Roll excursions limited to orientation that does not include the inverted. Except perhaps once.

Back to the Vertical Stabilizer. The folding of the structure beneath the VS mounts you use to imply a conclusion of partition by vertical acceleration that is interrupted twice. It is possible certainly. Wait, let's instead assume that the VS didn't separate from the Aircraft, but the a/c separated from the VS. At high speed, even thin air can be a machete to crew desperately trying to raise the nose to slow; it may demand energetic control inputs, such that the integrity of the airframe is compromised. Could the elevators and HS have corrupted the tail mountings of the VS? Inertia. Try to ignore BEA's suggestion of a tame descent, though desired for many reasons, I think it unlikely. An overloaded HS/Elevator structure? Imagine.
The HS swivels, and demands a structure that has the ability to resist the load of a moving structure. In a sense, the HS is the more petulant of the two basic structures.

In an uncontrolled descent, well above Vne (a fluctuating value, to be sure), control surface loads can easily exceed their critical limit. Flutter becomes a lethal noise, a devil that has haunted Aeronautics since its beginning. If in this descent the elevators were commanded to a sweep that imposed these (demonstrable) loads, the HS could have broken its blocks in the tail, and transferred its negative loads to the structures beneath the VS imposing these two wrinkles while in the air.

It is possible the a/c entered the Sea in four reasonably distinct pieces, Tail, aft section, wings (with center section) and mid fuselage, and the nose. The BEA's suggestion that the a/c was intact cannot escape the thought that a complete a/c is sleek, and wants speed, belying the conclusion of slow horizontal entry. As separate
sections, speed at entry would be limited by a draggy shape.

The Galley? the crew module? The damage they share could quite easily have been the result of aerodynamic loads sufficient to disrupt attachments and sign their presence with a bent mount or two.

The failures you describe so eloquently also wordlessly question the design of the tail, indeed, they ask why plant the Tail on the Fuselage instead of building the entire a/c as a unit. Because the Tail was built thousands of miles away? What was the cost of shipping? The pictures supplied also show the historic problems with combining Composites with metal into separate structures.

My hope is that the possibility I describe is wrong. Can you help dismiss it?

bill

Hi again!
I am not ignoring the fact that three different speed values were reported by the three ADIRU:s. However, we don´t know if one of them was correct. I fully agree with you and others that pitot failure is a probable cause - but not necessarily the only one. I don´t think the ACARS system measures any analogue signals, e.g. the pressures from the ports.

If you read the BEA 2:nd report carefully you will find that not all ACARS events have been fully understood/explained.

Personally I think that temporary hardware or software errors perhaps also may have interfered with some operations of the data busses. There is at least one incident report where one faulty computer locked up a bus thus preventing the other healthy computer access. When I was involved in the design of very large automated multi-computer controlled systems years ago (probably more complex than those on the Airbusses) it was often forbidden for safety reasons to use bus-type signal connections because these were sensitive to single electronic faults anywhere along the bus. However, I understand the need for bus-type connections on an aircraft due to weight limitations etc.

Your conclusion, not BEA which mention only a possibility of.

Of course not, it is left to ADMs (air data modules) which transmit to ADIRUs. Then, what is the chance that simultaneously damaged ADIRUs (for whatever reason, including aliens) monitors the same fault at the same time from different systems?