Another companion case, the AF 908 (FGNIH) between Paris and Antanarivo, very similar to the two Air Caraïbes events, and to the AF 447 seen through the ACARS, available at:
Eurocockpit - Archives
Quick translation:
FL370, on the AWY UB612 (OFFSET 1R), between the OBD and MLK waypoints, in contact with Khartoum, above the cloud layer, light turbulence, crepuscular lighting.
-at 15h10TU: Pitot 1 & 2, 2 & 3 et 1 & 3 FLT
AUTO FLIGHT AP OFF, REAC W/S DET FAULT, IAS DISCREPENCY , NAV ADR DISAGREE, ALTN LAW PROT LOST
it is noted that the turbulence strength had increased just before (at 15:09) this event when they entered the cloud layer, with a burnt smell in the cockpit, speed was reduced to Mach 0.80 (just above green dot), cabin crew warned, no ICE DETECTION alarm.
a few sec. later, the speed indication on the copilot PFD plunged from 280 Kts to 100 Kts in the red band during numerous sec. In the same time, on the CPT PFD, the airspeed rolled back 15 kts under the green dot (& displayed a speed trend of -50kts).
immediately followed by a Stall, Stall, Stall alarm... without the cricket sound, triggering a TOGA LK indication.
-at 15H11 : amber alarm F/CTL RUD TRV LIM FAULT
Since the speed trend was still of -50kts, the hand flying CPT initiated a slight descent and turn (to depart from the AWY route). A MAYDAY is issued by the copilot. The airspeed recovered at FL340, the A/THR was disengaged (to exit the TOGA LK state). The altitude being stabilized, the "unreliable airspeed" procedure was implemented. Anti ICE ENG & WING, PACKS FLOW set on HIGH and cross-checking between the airspeeds / altitude and the GPS ground speed / altitude + the winds data from OCTAVE. AP1 & A/THR reengaged. Descent at FL330 and MAYDAY cancelled. The crew joined the maintenance by satcom for a more throughout failure analysis and a PRIM/SEC reset was decided without results. An incoherence is found about the flap configuration between the QRH and the status (?).
Like in the other cases, the Pitot freezing & corrupted airspeeds event lasted between 3 and 5 mn. The crew set the weather radar on max sensibility for the rest of the flight (it was on CAL and did not detect anything suspect before the Pitot freezing).
Jeff
PS) -One of the differences between the AF 447 and all these cases with Pitot problems is the A/THR settings when the fault sequence began (engaged in the AF 447 case, off and N1 fixed for the turbulence penetration in the other cases). The visual clues: the Air Caraïbe case was in daylight, the AF 447 was a night flight (all in poor met conditions). I don't see any pilot refering the pre-stall buffeting indication in their in flight/post flight analyze of the stall alarm event in the safety reports.
The reasons why the Air Caraïbe crew is convinced that the stall aalrms are not justified are not explicitely stated in the report. Maybe it is linked to a question asked by a pilot to other pilots here.
-The speed trend on which the pilot based his decision about the stall alarm is completely artefactual/spurious: why this speed trend is still computed & displayed on the PFD during a corrupted airspeed event ? (Pitots FLR+ADR disagree). He probably also had in mind that they were just above the green dot just before the sudden sequence of FLRs.
-And in the same manner, why is the stall alarm still generated during a corrupted airspeed event ? Is it mandatory for the manufacturers that their planes have a functionnal stall/overspeed alarms in all circumstances ? (Airbus knows the risks and displays it on the ECAM "risks of undue stall warnings). An alarm plagued by false alarms is no more a valuable alarm, it is a danger. This danger can be further increased by the applicable procedures.
- we may have slightly different sequences of Pitot-static gradual failures, different timings for the different phases (before the fault isolation/detection by the automation, revertion to degraded modes, phase before the stall/overspeed alarms, after). Some will enable the crew to implement a part of the "unreliable airspeeds" proc. before the stall alarms may occur, other will not (AF 908). According to the sequence, the crew attention may be allocated differently on the failure analysis, procedures/check lists implementation, hand flight, cabin crew warning, etc...). I feel it would very useful to have a short-list of the indicators that may enable a crew to discriminate between a false and a justified stall alarms. How much time the cross-checking implemented by the AF 908 crew using OCTAVE can take ? Are there faster ways ? (when the fligh security is impacted). On the contrary, being able to know the "false friends" that you must avoid to rely on in such a case of unreliable airspeeds, seems equally interesting.