AF447

[vittorio66]

Incapacitation

The BEA says:
last radio exchange between the crew and Brazilian ATC occurred at 1 h 35 min 15 s. The airplane arrived at the edge of radar range of the Brazilian control centres, at 2 h 01, the crew tried, without success for the third time, to connect to the Dakar ATC ADS-C system,
No ATC messages were received or transmitted by F-GZCP. Only three attempts were made to connect up to the Dakar centre ADS-C system and were recorded on 1st June at 1 h 33,
1 h 35 and 2 h 01.

What was the last known crew action? At 2h01z or at 1h35Z? It seems that during the emergency initiated at 2h10z the crew took no action, what leads to the possibility of crew total incapacitation (depressurization, or any gas like CO???). Its a possibility to be investigated? This could explain why the flight did not any evasive action to the weather conditions and also the way it impacted the water. This is also coerent with no preparation to any emergency.

[DJ77]

SaturnV, Thank you for your detailed explanations. Actually, I didn't looked closely at the part of BEA's report regarding ATC communications, but I understand BEA's concerns about the apparent delay to scramble emergency researches.

[augustusjeremy]

FPC2 only

o FCPC2(2CE2)/WRG:ADIRU1 BUS ADR1-2 TO FCPC2 (2 h 10)
ATA: 279334
Source: *EFCS1
Identifiers: *EFCS2
Class 2, HARD
This message indicates that FCPC 2 no longer considers as valid the information that is
delivered to it by ADR 1 (via bus 2). The ATA code beginning with 27 indicates that the fault
was not detected by any other FCPC during the three seconds that followed (otherwise this
message would have been classified ATA 34). This message has not been fully explained at
this stage of the investigation.

Faulty ADR1 data was detected only by FCPC2 (PRIM 2 ?).

Regarding correlation of messages maybe a given system reporting an external fault (class 2) probably predicts but doesn't command an specific class 1 message.

Which means that a class 2 message may be followed by a class 1 message giving a different diagnostic to the first message.

Once again it is important to notice that apparently only PRIM2 reported the ADR1 fault - Would it be enough to rule out ADR1 to the other PRIMS ?

[vovachan]

The ill-fated Pulkovo 612 crew did not even realize they were spinning until they fell from the clowds and saw what was going on. Before that moment they seemed to think their instruments had gone crazy.

So the forces generated can't be too severe.

[takata]

vittorio66:
This could explain why the flight did not any evasive action to the weather conditions and also the way it impacted the water. This is also coerent with no preparation to any emergency.

ACARS are only maintenance messages. So, they won't tell us a lot about what the crew did or did not. Possibly some messages were triggered by crew troubleshooting after 0210. But the aicraft is also fully monitored. Then, until the SATCOM failed, we should be aware that no decompression happened and no poisonous gas was released in the cabin.

"No preparation for an emergency" is not exactly what the BEA mentioned from what was recovered so far. The point was that no life jacket was discovered "inflated" (implying no crash survivor), and this point was particularly weak as nobody would inflate its life jacket until clearing off from the aircraft. It seems way too early to extrapolate the situation of the aircraft passengers and crew before it crashed without knowing a lot more about the forensic evidences or recovering the CVR/FDR.

S~
Olivier

[funfly]

takata, thanks for your last answer. It brings me on to a more relevent point.

There is obviously going to be political mileage in the pitot tube situation.

Consider, the aircraft was on a stable cruising stage at a pressure setting (i.e.not altitude). The pitot tubes may or may not have been iced up but whatever the case this can not have been the catalyst of the disaster as there would have been no obvious indication of this to crew or instruments as the difference pitot/static would have remained the same.

If the pitot heads were iced then they wouldn't have had any adverse effect until speed or FL changed.

In fact the information transmitted about differing pressures between l & r pitots give a very good indication that both were operational.

[threemiles]

AMDAR report

(I am guessing that EU is LAN Ecuador.)

EU denotes a European aircraft for AMDAR purposes. Coding does not allow to identify operator to avoid spotting. Altitude report is most likely height over MSL, thus 325 may be something different in FL (FL310,330).

[einhverfr]

A few thoughts after reading the BEA report.....

First, the last ACARS positional message (around the beginning of the rapid degradation of the situation) seems to occur inside a region of strong convection within the larger meso-scale convective system. At this point, I think it is reasonable to ask why pilots flew into "that weather." Does AF require sufficent WX radar training? Are proper procedures (and training) in place to help pilots avoid at least the worst of the weather? From the available data, it does not appear that any weather avoidance was taken.

Secondly, could strong updrafts be a major factor, up to and including the final descent and crash? As I understand it, aircraft stalls are a function of angle of attack, and pitot tubes don't function well at high AoA values. A 50kt updraft at a speed of 447 kts (mach 0.78 at fl350) would add about 6.4 degrees to the angle of attack. (Showing my work: Inverse sine of 50/(447^2 + 50^2)^0.5... is roughly 6.4 degrees) Even with a 3 degree angle of attack, this puts the plane close to an angle where the pitot tubes become unreliable even without icing. Depending on the automated response to the angle of attack, this could get worse very quickly. The main point is though that any pitch up may have rendered the pitot tubes unreliable even without blockages. Oddly this gets back to what Tim Vasquez wrote about turbulance being the primary factor on his list shortly after the incident. In short could a strong updraft have caused a lack of reliable airspeed indications, degraded controls, etc?

Thirdly, what woud the autopilot have done in a strong updraft before disconnecting?

Fourth, assuming AoA was reduced by pitching the nose downward (possibly restoring airspeed data), what would have happened when the airplane went from a strong updraft to a strong downdraft? Could this have also caused a stall of some sort via too low of an AoA?

So this leads to a couple of other questions that are now lurking in my mind:

1) Can we shelve the idea that pitot icing was a factor at this point and focus more on AoA issues?

2) Could the Cabin Vertical Speed advisory have referred to a falling descent? Could this be some sort of stall?

[vittorio66]

Thank you Takata for some explanation, still I have few doubts, referring to the l/jacktes there is no evidence so far that the bodies recovered by the Frigate Ventose had the l/jackets doned , they stating "Clothed and relatively well preserved" no mention what so ever to L/J.
I beleive even if one had the L/J on would tell the investigators that pax where informed or some problem on the A/C.
This what I'm referring as an emergency on board.
About ACARS in the report state that "F-GZCP was programmed to automatically transmit its position approx. every ten minutes"....
Air France received total 26 messages 2 belonging to the previous flight about toilet and 24 received June 1st between 2:10:10 and 2:14:26.
They shouldn't have received also the position more frequently than only one time since the airplane was set to an automatic transmission ?
Also other aircrafts same type had similar situation encountering turbolence but the pilots were able to handle the emergency , the other aircraft did also sent ACARS or were programmed to send them ? Would not beneficial for the investicgators to compare "if possible" ACARS from other planes.
By now I guess you know I'm not a Pilot , I work in transportation field as well but in a different element Water not Air but this unfortunate event opened my eyes about things that I took for granted and now I'm finding out are not.

[safetypee]

IMHO the speculation on a spin, flat or otherwise, is misplaced. There is little evidence to support this; all we have is that the aircraft was most likely in a level attitude moving forward as it hit the sea – nothing of what happened in-between, time scale – when, how or why.

I would like to see referenced evidence of any large aircraft in a stable spin; incipient, yes I would accept. I do not know of any such event and this is not for the want of being in conditions conducive to a spin in commercial aircraft.

Aircraft are built and certificated to withstand inadvertent Cb encounters; they do not automatically stall, nor is a spin an automatic consequence of a stall. Large aircraft tend to depart controlled flight relatively gracefully in comparison small aircraft or high performance fighters. The engines do not flame out in Cbs without other contributions, and all aircraft, with their natural stability can transit the relatively short duration of these events in a reasonable stable manner when the controls are used to maintain attitude.

funfly re #3010, the pitot tubes incorporate a leak path – drain hole. See previous posts for likely icing mechanisms. Drain hole blocked = over pressure - higher airspeed. Pitot blocked (before the drain) = ambient pressure – lower airspeed. All other combinations – fluctuating airspeed.

[SaturnV]

threemiles, thank you.

Is the flight number 0046 also masked; i.e., 0046 is not the actual flight number?

This flight would have preceded LH507 by about ten minutes, and LH507, flying at FL350, deviated west of the track by 10 NM after passing ORARO.

[surplus1]

Originally Posted by ”PJ2”

That's exactly what the BEA is trying to convey: level aircraft attitude, vertical trajectory with possible slight forward and perhaps even aft, trajectory, (possible explanation of the condition of the one spoiler found - torn out by reversed airflow, as someone posited a thousand posts ago) - not a straight line down, in other words.

As usual your post is relevant and well thought out. I don’t argue with your hypotheses but let’s take this a little farther if I may. For starters, I will assume that the BEA report is accurate, i.e., the aircraft struck the ocean in a ‘level attitude” with a near vertical trajectory. Why?

I would suggest that a ‘near vertical trajectory’ (descent) – in a level or flat attitude - is not possible unless both wings are fully stalled – the aircraft isn’t flying, it is falling and there is very little ‘forward motion’. I would postulate further that in such a prolonged “stall” condition a spin of some type, would almost certainly develop. If there was a “spin” there would also be rotation. If there was a conventional spin, the nose would be down – unless the ‘spin’ was a “flat spin” – in which the tail could be down and the nose up with reference to the horizon. The BEA made no mention of “rotation” on impact.

I do not understand how an aircraft can descend with a vertical trajectory if it is not spinning. I do not understand how it can “spin” without rotation. Even if the stick was held full nose up after the stall and throughout the descent, I do not see what wound keep one wing from dropping off at some point and producing a conventional nosed down spin. However, a rearward cg or some other factor could induce a "flat spin" - and that could fit the scenario implied by the BEA.

There has been considerable debate as to whether an airliner can enter a ‘flat spin”. The answer is: YES. How do I know this - Historical record from prior accident(s). [See BOAC 911 and Vladivostok Air 352]

We all know that a “spin” by definition, requires at least two elements: 1) the wing must stall [and remain stalled]; 2) there must be rotation (about the aircraft’s cg.), [and a near vertical descent will follow].

According to NASA, there are four (4) categories (modes) of “spin” classification:

NASA Spin Mode Classification

Spin Mode .................. Angle-of-attack range, degrees

Steep .................................20 – 30
Moderately steep ................. 30 – 45
Moderately flat .................... 45 – 65
Flat .....................................65 - 90

By definition, a “Flat Spin” = A spin in which the tail of the aeroplane drops and the machine rotates in an almost horizontal pane. The nose may be slightly below or above the horizon. A rearward CG will augment the tendency to flat spin. The following video illustrates an unrecoverable flat spin in a fighter aircraft.

http://www.youtube.com/watch?v=z7VS9_Ce0sg

I am not aware of any transport category airplane that can recover from a fully developed spin - especially a flat spin, or retain its structural integrity within one, given the forces that are experienced.

I do not see how, as inferred by the BEA report, a near vertical trajectory of descent, implying alpha of 65 to 90 degrees, could occur in anything other than a flat spin. Perhaps you or someone else could enlighten me if that is wrong.

In the BOAC 911 accident (a 707-463), it was determined that the aircraft entered an area of extreme (clear air) turbulence. The vertical fin failed and was torn from the aircraft taking other parts of the empennage with it. Subsequently the severe yaw tore the engines from their mounts and the remainder of the structure entered a flat spin until impact. [You have previously posted a photo of that aircraft in descent]. A military fighter sent to search for the wreckage also experienced extreme turbulence – with recorded 9g positive and 4g negative - and temporary loss of control. Its pilot was able to recover but that was a fighter aircraft. IMHO, Airliners cannot survive such forces intact.

In the Vladivostok Air accident (TU-154), the stall was induced by pilot error during the approach and the aircraft entered a flat spin, in which it remained for 22 turns until impact. Unlike the Boeing, this is a “T-tail” aircraft with engines mounted on the rear fuselage (which would likely negate their separation), plus T-tails are susceptible to deep stalls. Also, that aircraft was much closer to the surface when the stall occurred.

IF AF447 entered extreme turbulence, it would have experienced very high g forces, both positive and negative, in rapid succession and most probably stalled. It is likely there would have been severe skidding and slipping as well – all induced by the turbulence. Its pilots would not have been able to prevent this – once the area of turbulence had already been entered. A spin could have resulted – which might have caused the VS to separate – thus creating a “flat spin”. The fin didn’t have to separate by itself – the entire aft tail structure could have separated – somewhere behind the pressure bulkhead. The centrifugal forces would most probably immobilize the crew at least initially.

If there was a flat spin, it is probable that not only the VS could have separated but, the engines would not only have flamed out (as you correctly surmise) but might likely (one or both) have been torn from their mounts – by extreme yaw when the fin/tail separated, or by centrifugal force of the flat spin. [In the AA587 accident, if I recall correctly, at least one engine separated from its mounts due to the yaw.]

Just as extreme turbulence or a complete stall could disrupt airflow to the engines, could it not also disrupt airflow to the pitot tubes and thus generate the loss of dynamic pressure and the associated errors/warnings (unreliable airspeed, etc.,) [as opposed to icing]? Could not extreme turbulence independently exceed the limitations of the Autopilot and cause it to disconnect - absent other factors? Other components, such as the spoiler to which you refer, may also have been partially deployed and separated.

If either pilot was holding the side stick when the extreme turbulence occurred – could it not have raised his arm violently (negative g) and induced an unintended control input? Anyone who has had an inadvertent encounter with severe turbulence should be able to readily imagine what might happen in extreme turbulence.

At FL350 a CB with tops at FL500 or 550 is quite likely to be capable of producing extreme turbulence. In severe turbulence you might have a change; in extreme turbulence that is improbable.

Do you as an experienced Airbus captain believe that the built in computer ‘protections’ could prevent the aircraft from exceeding either end of its envelope in extreme turbulence? I don’t have the knowledge to guess but I don't believe so. Those things just aren't designed for conditions that are never expected to occur - neither is any other part of the aircraft.

This [flat stall/spin] could then permit the remainder of the structure (fuselage and wings) to descend ‘in a near vertical trajectory’ and strike the ocean in a right side up ‘level aircraft attitude’ – mostly intact – thus explaining the observed ‘compression’ of some recovered parts.

For this to happen, it would seem that the aircraft would also have to exit the area of turbulence – otherwise there is only a 50/50 chance of the falling structure to have remained ‘upright’ during the descent.

The plethora of ACARS messages may have been related to effects triggered by the turbulence – not to pitot icing - and not causal of the incident. I respectfully submit that there is a difference between severe turbulence and extreme turbulence. In the latter the aircraft is not controllable – with or without computers IMO.

We do not know if the aircraft encountered either severe or extreme turbulence, but it is just as probable as any of the other hypotheses. There was certainly weather in the vicinity that could have produced either. IF this did occur, we could well have a logical (though not factual) explanation of everything else.

This is NOT intended to disparage the crew. We do not know if they deviated, if they never detected the wx, or why. There could have been a variety of reasons not related to any failure on their part. I am not willing to judge them without certain knowledge. I am also convinced that they did not intentionally fly into a CB - no matter who was on the flight deck at the time. They were not that inexperienced.

If the ultimate loss of the hull did occur as a consequence of an encounter with extreme turbulence, that in no way reflects negatively on the design or structural integrity of the A-330. No airliner is designed to operate in extreme turbulence. That is why the possibility of such an encounter must be avoided at all cost.

The BEA’s assumption that compression damage observed in the wreckage indicates that the aircraft was “intact on impact” may only be partially correct. Some major parts of the fuselage may well have been intact and the wings still attached at impact. That however, does not preclude separation of other parts in flight. To bluntly state that there was “no break-up in flight” appears to be premature at best. The terminology is highly ambiguous.

Obviously, this is conjecture – but so is everything else that we have explored to this point, including the Preliminary Report

[IO540]

Secondly, could strong updrafts be a major factor, up to and including the final descent and crash? As I understand it, aircraft stalls are a function of angle of attack, and pitot tubes don't function well at high AoA values. A 50kt updraft at a speed of 447 kts (mach 0.78 at fl350) would add about 6.4 degrees to the angle of attack.

I don't think so. If you fly, on autopilot holding a constant pressure altitude, into an updraught, the AP pitches the aircraft down, to maintain the pressure altitude. Your airspeed increases - possibly a lot.

Thirdly, what woud the autopilot have done in a strong updraft before disconnecting?

It will pitch the nose down, and the airspeed will increase rapidly.

Whether the AP will disconnect just because of the updraught, I have no idea but I doubt it.

[surplus1]

Whether the AP will disconnect just because of the updraught, I have no idea but I doubt it.

Like everything else, the AP has limitations. If those limitations are exceeded, whatever the reason, it will disconnect.

[Squawk_ident]

Estimated time TASIL 0220

Not at Mach 0.82

Here is a recap from AFR447 actual or estimated. I have put the distance between them.

FEMUR: ACTUAL 0113 (COM at 01.14.31) DIST to INTOL 154NM: 20 Minutes
INTOL: ACTUAL 0133 ( COM at 01.35.15) Estimating SALPU at 0148 and ORARO at 0200
INTOL-SALPU is 122NM 15 minutes and SALPU-ORARO is 122NM 12 Minutes.

ORARO-TASIL IS 119NM and it should be about 11 minutes flight time, the ETO TASIL would have been at/around 0211 not 0220.


Knowing that the last radio contact with AFR744 was just after the SELCAL check at 01.35.43 and that no ETO TASIL was provided to Atlantico, we have to think that Atlantico has given an estimated time based on its own calculation to Dakar.
Another thing is that AFR447 would not have called over SALPU and ORARO but it is compulsory reporting points. Either contact was kept after INTOL with the AF447 and something is missing in the Radiocommunication transcript published in the Report, or contact was lost just after the SELCAL check and Atlantico would have tried to contact him repeatedly and it would be said in the Report.

If my calculation is correct It would mean that crew would have drastically reduced its speed after INTOL.

[AirRabbit]

I’ll try this one more time…

Originally Posted by Graybeard

The ventral fins (strakes) were not new on the MD-80. They were first fitted on the DC9-50, for stability at cruise. All DC-9 are subject to deep stall. The DC9-80 was fitted with a stick pusher after it became uncontrollable in a deep stall in flight test, and they had to deploy the parachute to recover.

1) The “strakes” mounted on the DC-9-50 were not mounted for stability at cruise. They were added to the DC-9-50 to improve airflow up and over the fuselage and around the vertical stabilizer to improve directional control during approaches. Up to the addition of the strakes, the DC-9 had a “rudder limiter” system installed that dropped a bracket around the rudder actuator in the tail that prevented the rudder from being displaced more than plus/minus 8 degrees either side of center (as I recall) when the flaps were retracted. On the other end, when flaps were extended, this “rudder limiter” was retracted out of the way, allowing the rudder surface to be displaced plus/minus 30 degrees either side of center (again, as I recall). The pilot annunciator panel had a “blue” advisory light that illuminated saying “Rudder Unrestricted” that had to be illuminated prior to initiating the approach, or the approach had to be flown at substantially higher approach speeds because of the potential limitation of the rudder. This system is still on the DC-9s - 10, 20, 30, and 40 series.

2) While the DC-9 was initially prone to the same malady that befell the French Caravelle – that of deep stall blanking out the tail sufficiently that elevator authority was lost. Therefore, McD widened the horizontal stabilizer, including the ends of the elevators, allowing both to extend out into the airflow that did exist beyond the “blanked out” areas in a deep aerodynamic stall. They also added a 3000 psi assistance to place the elevators into a full nose down position when either flight crew member pushed the control column full forward. Not only was this system successful, it was so successful that the first few pilots landing on contaminated runways and desiring firm nosewheel contact with the surface and pushed full forward on the control column to get it, found themselves “wheelbarrowing” with only the nosewheel on the runway – that is, they actually had so much lift on the tail they actually lifted the main gear off the runway surface!

3) Any “stick pusher” that may have been installed would have been for flight testing only, and I'm not at all sure that ever happened. I am not aware that there was ever a production DC-9 delivered with a stick pusher – they relied on the wider horizontal stabilizer and elevator and the 3000 psi hydraulic system to ensure the elevators were properly positioned to “nose down.” You can still see flight crews testing this system today when taxiing out behind any DC-9 series airplane.

[ttcse]

safetypee

I would like to see referenced evidence of any large aircraft in a stable spin;

Lookup Pulkovo 612, a TU-154 I believe.

Apparently stalled then spun from FL390 while trying to fly higher over convective activity but simply entered at a higher level. Very interesting transcript of cockpit conversation can be found.

[safetypee]

surplus1, in recognizing your conjecture I believe that some of the emphasis and the correlations with previous accidents are misplaced; I covered some of these points in #3014.

You discount a stable stall condition where the wings remain relatively level. I proposed an hypothesis leading to this scenario in #2825 http://www.pprune.org/rumours-news/3...ml#post5037329
Note that the A330 is reported to have good roll centering in ALTN law, which even with a roll oscillation could result in a ‘level’ attitude and a high descent rate. Hitting the sea at 100kts in a stalled condition, where pitch control might be non existent for a flare is still very fast.

The stalled condition above is not the same as the spin conditions which you and others relate too – the example, and ideas elsewhere invoke the loss of the fin which so far has been shown to be unlikely. Similarly, the example of a T tail stall is not applicable here, but I will check ttcse's reference.

I do not discount control mishandling which could have resulted in a stall / spin, but without evidence, this line of thought cannot be progressed. Also, whilst attesting from experience of Cb test flights, the levels of turbulence can create many hazards, they do not in themselves cause the type of upset or failure which you discuss.
In general, the certification requirements provide for a range of extremes associated with inadvertent Cb encounter, which are based upon extensive aviation history (and accidents). As yet, we only have evidence of the aircraft being close to Cbs and not actually in one.

Many big IFs lead to a range of alternative situations, each viewed by many people with their own views, and unfortunately at times, with bias (but not apparently in your post).
IMHO your analytical logic fails with the assumption that the pitot systems would see different air masses on each side of the aircraft. From experience of tests and often due to the architecture of the pitot systems (cross balance pipes) this occurrence is most unlikely. Also, IF the comparator software detected a cross system failure, then a subsequent comparator failure might then be comparing ‘same side’ systems for the next reported ‘failure’.

[Litebulbs]

Safetypee

That post was like a breath of fresh air. It pulls you back to the FACT that aircraft are fundamentally well designed and safe. Problems may have occurred in the past with T designs, but experience through accidents has allowed procedures to be established, to correct any latent faults in that design.

Large twins have never suffered fatal aerodynamic problems whilst in the cruise, no matter what the weather was doing.

[Rananim]

If either pilot was holding the side stick when the extreme turbulence occurred – could it not have raised his arm violently (negative g) and induced an unintended control input?

I think this may be central and am surprised that it hasnt be discussed more.Sidestick issues(sensitivity,size,position)in turbulence deserve greater focus.