@Clandestino: Thanks for the clarification.
Originally Posted by Clandestino
Number 32 was often quoted in threads referring to AF447.
Indeed - I was just quoting your mention of that number, without detailing the variations around it.
Originally Posted by Clandestino
In those cases that could be further analyzed, crew pushed forward when faced with stall warning, CM2 of AF447 pulled and kept pulling. My entirely subjective criterion of exception is thus met.
But that only addresses one reaction - to the stall warning. Other crews made
different errors, which the crew of AF447 did
not make (e.g. re-engaging the A/P during the UAS event). So depending on what specific criterion is considered,
other crews were exceptions and
not AF447.
That's exactly why I agree with you about the subjectivity - thanks again for your clarification.
@xcitation:
Originally Posted by xcitation
Indeed, a major puzzle why ignore the stall warnings for 3 or 4 minutes?
Absolutely - I said something very similar earlier.
The lack of apparent recognition of the stall warnings, but also the lack of explicit verbal reasons for ignoring them, is a puzzle.
How are we ever going to know the reason(s)? I can only think that the BEA Human Factors folks will eventually give us their best guess, which will be based on more info than we have e.g. intonation used on the CVR recording.
@Old Carthusian: Thanks for the summary.
Originally Posted by Old Carthusian
But this accident is about professionalism, training, culture, CRM and SOPs all mixed togther and above all knowing your machine.
Agreed - many factors indeed. I also believe that there is a large dose of the "Ironies of Automation" [Bainbridge] in there too - humans monitoring machines (any) don't do very well when they have to take over completely and suddenly, with no prior warning. That seems to be a human "failing", and not something that we can (should?) expect to "train out of people".
That's why I've been following with interest the comments here about the advantages of designing some kind of "ATT Hold" autopilot in the event of UAS - something to gradually
ease the transition from A/P to manual control, and lessen that "shock factor" (which is my currently favoured reason, from everything I've read so far, for the NU inputs from the PF - just muscular tension from the shock of having the plane "dumped in his lap". That's why he didn't know he was causing the climb, as it wasn't a conscious input).
I've learned here some of the difficulties of implementing something like that, especially if (as in this case), the plane is in some turbulance when it is engaged, and we may not want the
current attitude (which could be momentarily significantly NU or ND in turbulance) to be the one which is held by the ATT Hold system. That requirement for the PF to control the plane in the turbulance immediately, prevented PJ2's lovely phrase of "don't just do something, sit there" (IIRC) after the A/P disconnect.
If the onboard systems had given the crew time to think & get up to speed, instead of immediately forcing them to react, in a state of shock from their "cruise state", I wonder if we would have had a better outcome. The Human Factors analysis of the final accident report will make facinating reading, I expect.