Originally Posted by
chrisN
Dozy, it was one of the experts who I thought said they got the initial roll under control – in a part of a thread about PIO or whatever it is called nowadays.
I'm no expert, but judging by the DFDR output, every single roll input prior to stall was in excess of what was required to correct it. I'm of the opinion (as a non-expert
) that the initial roll was turbulence-induced, and that the initial overcorrection by the PF led to PIO in roll. The reason I'm fairly confident about the latter is that we know for certain that neither of the F/Os in the flight deck had ever done any high-altitude manual handling training, and the low-level manual handling they had done requires considerably greater stick deflection than required or desired at cruise level.
But it is not particularly relevant to the main point, which is that PF’s initial pitch inputs seemed to many to be incidental, even accidental, to his roll over-corrections. The full stick back came later.
I'm not so sure (but remember - non-expert!). Part of the reason I leapt at the chance to try this in the sim was to find out exactly how the real Airbus sidestick behaved. Looking at the DFDR, the initial pitch input varied between about 10-50%. From my subjective perspective based on the sim scenario, 50% deflection requires significant positive movement to overcome the spring of the centring mechanism - very different from a computer joystick, where that kind of deflection is fairly easy to achieve unintentionally.
Everything about this looks to me (as a relatively well-read layman) like a textbook startle response, followed by breakdown of instrument scan and intra-cockpit communication. It may be fair to censure the crew for not following UAS procedure, but it is not fair to censure the crew for attempting to improvise and fumbling the recovery, because the institution to which they belonged had never provided the training required to do so effectively.
@Lyman - I know a few disagree with me, but I don't think the roll was induced by positive autoflight intervention or mechanical failure. Everything I see is consistent with the autopilot disconnecting in the middle of a turbulence event, from which it was trying to correct. Rudderrudderrat's mention of rudder trim is an interesting one because if autoflight had been using it, it may have left them with a slight roll tendency until corrected, but even then nothing that would require inputs as aggressive as those made by the PF.
@gums - See above. Nothing in the traces indicates aileron damage. I'd need to get the books out, but I think the outer ailerons are centred and locked when manual control is taken at high altitude, which is why the outer aileron traces centre at A/P disconnect. The take-away from the aileron traces for me is comparing the significant difference in magnitude between the corrections made by autoflight (indicating the required amount of deflection for positive correction at high altitude) and those made by the PF (considerably larger and way in excess of that required). OK, so with outer ailerons locked the inner ailerons would have to work a little harder - but not that much harder!