AF 447 Thread No. 5

[auraflyer]

Gums, hope this helps.

Some points from my reading of the French (not a native speaker). The first two are my thoughts, the rest is my best attempt to translate the salient parts of the report; thoughts are theirs, not mine.

I'm sure I've made mistakes, will correct them as people point them out to me.


1. The pitots seemed to fail at 3 different points and differently. See p 31. The ISIS speed dropped, rose then dropped again. Not certain about this, but given the statement at 2:10:17 "we've lost the the the speeds", the third (unrecorded) speed might not to have fallen as far at that moment, so there was one high(er) (unrecorded speed) and two that had dropped to 73kt and coincided at that value.

2. The aircraft apparently left its known flight envelope at 2:10:54:

p43: "La validité du modèle se limite au domaine de vol connu d’après les essais en soufflerie et les essais en vol. Ainsi, la simulation a pu être menée sur la période de 2 h 10 min 00 jusqu’à 2 h 10 min 54."

As I read it, Airbus conducted a simulation limited to the flight envelope as known from wind tunnel/test flights. The ending of the simulation at 2:10:54 suggests that that was the limit of the envelope.

"Phase 1"

3. Background noise changed, sounds characteristic of ice crystals encountered at 2:09:46 (p75)

"Phase 2"

4. Indicated airspeeds first dropped at 2:10:05 (p76). The sharp ("brutale") drop caused disconnection of AP and autothrust. Turbulence caused a right roll of up to 8 degrees (p76)

5. PF's sidestick inputs were quick and large, almost from stop to stop (native speaker pilot might assist - "butee" (Lemurian?)). They caused a pitch increase of 11 degree in 10 seconds, and increased the AoA. The stall warning activated twice. [As I read it, activated briefly or intermittently - not certain about "de manière furtive", as I understand it to mean "almost silent/hidden"] This is probably what caused PNF to say "what's that?" (qu’est-ce que c’est que ça?). [2:10:10]

6. Although they identified unreliable airspeeds, neither of the two performed the associated procedure. They had only been trained in emergency procedures at low altitude. (There is a passage observing that a note had been circulated following earlier UAS encounters in the A330/340 fleet.)

7. Between 2:10:18 and 2:10:25 the PNF read the ECAM messages out of order (I read it as uncoordinated, ie not usual). He notes loss of auto thrust and alt law, and calls for wing anti icing.

8. PNF draws the PF's attention to speed. The two known (ie in the data recorders) speeds are less than 100 kts. Vertical speed has reached a maximum of 7000 ft/min. PNF notes the climb, tells PF several times to descend. PF does take action but the rate of climb is nonetheless still excessive. (p76)

9. Near 2:10:34, the left displayed speed becomes valid and increases to 215kt. The ISIS speed is still wrong. The aircraft has already lost about 60kt from the time the AP disconnected, which is consistent with climbing about 2000ft.

10. At 2:10:47, thrust is moved back to about 2/3 (N1 85%). 2 seconds later, pitch returns to a little under 6 degrees, roll is controlled, AoA is under 5 degrees, THS is 3 degrees up. Vertical speeds is still above 1100 ft min. The PNF calls for the captain.

11. Up to this point, neither of the two pilots has mentioned the values of speed, pitch, vertical speed or altitude. The flight directors had not disengaged but the "bars" have disappeared and reappeared several times. They can't say at this stage how this might have affected what the PF was doing.

12. At this point, the PF seems to have controlled the aircraft's trajectory. The initial increase in pitch and the resulting vertical speed were excessive for that altitude and should have been mentioned to the PNF. There is a passage about the lack of CRM and the lack of clear division of resposibilities by the captain. (p77)

"Phase 3"

13. At 2:10:51, the stall warning sounded again. AoA is aobut 6 degrees, which is the theoretical threshold for stall warning activation at M0.68. PF mostly continues to apply nose-up: pitch increases from 6 to 13 degrees, AoA from 6 to 10 degrees. The recorded values for vertical acceleration reflect a rapid onset of vibrations that could be buffet. Five seconds later, probably as a result of the stall warning, PF applies TOGA. "That is about the moment the aircraft left its flight envelope."

14. Despite some small contrary actions [I think this is what they mean by "piquer"] the PF generally maintains nose up inputs. Pitch is between 11 and 13 degrees and AoA between 11 and 23. The THS starts moving consistently with the PF's inputs and reaches a value of about 13 degrees about a minute later. Note that in alt law, auto trim is still active. On the other hand ,it is difficult for the crew to know the trim position and there is no alarm that it is moving [check: "deroule" - I usually understand as "unroll"].

15. At 2:11:06, after several attempts to call the captain, the PNF again notes his absence. "This concern probably increases the PNF's stress, who has found himself in a situation he does not understand."

16. A second later, the ISIS speed becomes valid again. ADR 3 is selected on the right hand PFD, and the PF's speed also becomes valid. It is then 183 kt and the three speeds are consistent. This is not the subject of any comment by the crew.

17. In the 30 seconds after the stall warning started, airspeed fell from 205 to about 160 kts. Vertical velocity fell to zero, then strongly negative (-4000 ft/min). Maximum altitude of 38,000 ft was at 2:11:10

18. A little after 2:11:30, PF announced that he had lost control of the plane. This might be explained by difficulty in maintaining roll control: it was rolling to the right even though he had the stick at full left stop (again, "butee").

19. At 2:11:37, PNF said "command to the left" [I read it potentially as "I have control"?], and made a short input to the left stop. PF took control again almost immediately without saying anything and continued to fly ("piloter").

20. About 2:11:42, the captain re-entered the cockpit, not long before the stall warning stopped. Altitude was 35,800 ft, vertical speed was -9,100 ft/min, IAS [I think - "vitesse conventionnelle"] was 100kts and dropping, pitch at 12 degrees and N1 at 102%. Neither of the pilots knew exactly what was going on ("what problems they had encountered or the actions undertaken"), other than that they had lost control of the aircraft and they had tried everything. In response, the Capt several times said "look at that"; he was undoubtedly pointing to the FPV. The recorders show the stall warning had stopped because the 3 AoA values had become invalid.

After 2:11:45, the speeds were no longer continuously displayed on the PFD.

21. At 2:12:04, PF said he thought they were in an overspeed situation, possibly because of the high aerodynamic noise inside the cockpit. Neither of the other two pilots analysed this hypothesis even though it was inconsistent with the pitch and the high negative vertical speed.

22. Until the end of the flight, the AoA values became successively valid and invalid. Each time that at least one value becomes valid, the stall alarm starts; when the values become invalid again, the alarm stops. Several small inputs are made that reduce pitch, and each time they make the AoA values valid again and cause the stall alarm to sound. It seems that the pilots react by making nose up inputs, which cause the AoA to increase, the measured speeds to decrease and hence cause the stall warning to stop. Until the end of hte flight, the AoA never drops below 35 degrees.

23. Neither of the two copilots formally identified the stall. They did not recognise it from the stall warning, by recognition of the buffet, or by the high value of the vertical speed and the pitch. Not quite certain of this: "It is notable that the buffet is the only indication of the onset [or approach] to stall at high altitude on other aircraft where the stall warning does not depend on Mach".

24. Not having been given relevant information by the copilots, the only information the Capt had, from the screens, was not enough to give him rapid situational awareness. He did not then ask questions that could have helped him understand the sequence of events.

25. The stall alarm sounded continuously for 54 seconds but neither of the copilots referred to it. It is likely that the Capt heard it briefly before he returned to the cockpit but it is equally probably that the multiple start/stops added to the confusion and affected his understanding of the situation.

Despite several references to the decreasing altitude, none of the three pilots seemed to know which information to trust: the pitch, roll and thrust values could have seemed inconsistent to them with the vertical speed values.