AF447

[DJ77]

In US terminology, incidence is defined as the angle between the reference chord line of the wing and the longitudinal axis of the fuselage. As far as I know, this angle generally lies in the range 0 to 3 deg. With this definition, we have in level flight:
Pitch attitude = Angle of Attack - incidence.

The A330 "Unreliable Speed Indic. / ADR Check Proc." contains a table of Pitch / Thrust values. In this table, for parameters corresponding to AF447 known data at the time of the events, ie:

clean configuration,
cruise phase,
gross weight > 190 t,
FL250 – FL 370,

you can read:

Speed = 260 kt,
Pitch attitude = 3.5 deg,
N1 = 90%.

The pitch attitudes values in the table are rounded to the closest 0.5 deg.

The report on the ACA incident says (page 11) that the "STALL" audio alarm is triggered when the AoA exceeds a predetermined value that depends on

Flaps or Slats position,
Speed or Mach (???),
Active control law (normal, alternate or direct).

Then it states that two STALL audio alarms occured during the incident for AoAs of 4.48 deg and 4.31 deg respectively. Those values were obviously obtained from FDR data. A note adds that the STALL alarm is triggered at a threshold AoA of 4.2 deg. I believe that this value only applies in the configuration of the incident and was obtained from Airbus personnel.

It may look stange that AoA depends on airspeed but probe litterature generally speak of "local AoA" and "corrected AoA". An important design stage for AoA probe installation is to find a place on the fuselage where the correction is quasi linear relative to airspeed. This is done experimentally in wind tunnel. One would expect that if the associated Air Data Computer detects unreliable airspeed, it uses a default conservative correction factor. Alternatively, the warning computer could use increased margins, which would account for the threshold AoA dependancy upon the active control law.

Since the configurations of the ACA flight and AF 447 were very close (cruise, clean, about same weight, same level), It seems very likely that AF 447 experienced STALL alarms. Moreover, if the 4.2 deg AoA threshold value is true, the formula:
Pitch attitude = Angle of Attack - incidence
3.5 deg = 4.2 deg - incidence
shows that, at this weight and flight level, even taking into account rounding and using zero incidence, the a/c is very close (less than one deg) to the stall warning alarm threshold. A little pitch up move of the sidestick or a small updraft could trigger it.

Posted by HazelNuts39: Pilots MUST respect stall warning

That is fairly true. This alarm is in the category of "time critical warnings", like GPWS or windshear alarms, that require immediate crew awareness and corrective action. Crew Resource Management teaches that when such an alarm occurs, it's not the time to start wondering what caused it, but to act.
The PF of the ACA flight fortunately did not perform the stall recovery procedure because he was convinced the plane was not stalled. This conviction was certainly based, among other cues, on the fact that he had been controlling airspeed manually for some time when the airspeed rollback event occurred. In fact, FDR data show they exited the event at the same airspeed they entered it. They did respect the stall warning in an appropriate way since they voluntarily ignored it.

Contrary to ACA, the A/T was controlling airspeed aboard AF447, until it disconnected. The crew was not at the same level of awareness about the airspeed before the indicators went astray. If a STALL alarm sounded, which I think is very likely, they logically respected the alarm the standard way: nose down, full thrust then ... loss of control ?

The main cause of the accident could be ... failure of the crew to disregard a false alarm.