Originally Posted by
Machinbird
Since AP had dropped out together with the autothrottles when the air data went south, what would have prevented the aircraft from rolling over onto its back, particularly since it had a significant tendency to roll to the right in its ALT2B mode?
Hmm... Based on earlier discussions I think the idea that the aircraft had developed a right-roll tendency [EDIT: in and of itself] is at best debatable and in my opinion (for all that's worth) actually fairly unlikely. OK, so the data shows the flightpath post-disconnect and stall as being a descent biased to the right - but there are several factors that need to be borne in mind.
Firstly, we know that the AP disconnect occurred roughly simultaneously with a "bump" of turbulence that left the aircraft with a pitch attitude of slightly below zero and a slight bank to the right. From then on the PF took manual control and was making inputs virtually constantly from onset to impact, thus making it impossible to ascertain what the aircraft's response to zero control input would have been. Even in Alternate Law, the aircraft should have remained stable in pitch and bank until an input was made.
Secondly, we also know that the PF was making significant lateral inputs on top of the pitch commands - during the zoom climb this seems consistent with overcorrection, if not PIO. Looking at the DFDR data you can see that the right bank in this phase correlates with the input from the PF's stick - if he were correcting a right roll tendency the roll would occur even with the stick laterally neutral in the traces with an occasional correction to the left.
Finally, once the aircraft was into the stall regime any bank commands attempted via aileron would have been ineffective and inconsistent, as the wing needs to be flying in order for the ailerons to function correctly. The only way to effectively control bank past that point would be to recover from the stall and subsequently level the wings, or in an absolute worst-case scenario use the rudder very gingerly to get some degree of stability during the stall recovery sequence.
Additionally (and perhaps tangentially), there is no mention on the CVR of the aircraft exhibiting a right-roll tendency [EDIT: prior to stall]. Circumstantial I know - but in every case I know of where a specific and consistent control difficulty has come up, it's always mentioned by the crew.