Surplus1 said
If a stall is approached in that mode, when the AP disconnects and power is applied there will be a severe pitch up that the elevator may not have the authority to overcome in a timely manner unless (and until) the trim is removed by the pilot.
Undesirable flight states; stall appears to be prevented in Alternate law (ALTN 1) by the low speed protection, IAS replaces the AOA function, but in this accident it appears that there was no valid IAS.
AFAIK this gives ALTN 2 law – no speed protections;
I accept what surplus1 says about a maintained stable stall being unlikely, however his arguments don’t apply to a spin scenario. Once entered and stabilized (for whatever reason), some aircraft will remain in a sustained stick-free spin until impact. It’s the nature of the beast and a function of the momentum built up by the couples in autorotation (hence the “auto”). The most recent precedent is the TU154 crash in Donetsk. It was a very similar situation except that the pilots had attempted to outclimb the storm clouds – rather than suffered an autopilot disconnect and pitch-up. They rode it down, attempting recovery, and even managed to punch out a distress call or two.
However I don’t believe that the aerodynamic forces in play at spin entry (or during a spin) would’ve been anywhere near that required to generate fin failure/detachment. The four minutes of ACARS messages stemming from that initial failure would have started at autopilot “kick-out”.
So if we now apply Occam's Razor and then summarize:
Can't rule out pilot incapacitation yet but it's more likely that the two copilots were minding the shop and were caught out when the autopilot suddenly kicked out and they faced a severely out-of-trim aircraft pitching up into a stall/spin situation (thanks to the faulty pitot tubes and auto-trim and a confounded ADIRS).
If the auto-trim progressively wound in lots of back-trim as a result of the pitot supplying increasingly false air-data, the A330 may not have been recoverable in Alternate Law (which they were suddenly in). It's one of the unattractive aspects of a trimmable horizontal stabilizer and an autopilot that's capable of holding large out-of-trim loads.
One big fallacy is that you cannot stall a FBW Airbus. That law is trumped by a dynamic pitch-up into the tenuous aerodynamics of coffin corner flight. i.e. to say that no further pro-spin flight control inputs are required. Thanks to the high nose attitude and the couples, the stall/spin scenario is the only logical aerodynamic outcome (with or without an engine hiccup). That's what can happen in a thin air regime that pilots very very rarely fly manually in. And you can reflect upon the likelihood that such corners of the operating envelope would never have been pursued by Airbus test-pilots. Engines that are suddenly exposed to a high AoA at high power in coffin corner are also likely to stall (due intake blanking). One engine stalling would have provided the asymmetry for rapid instantaneous entry into a flat spin.
Why could it have remained a flat spin? As the aircraft descended and the air thickened, the engine would have unstalled itself (that's what happens with N over root T stalls), provided high symmetric thrust and kept the spin flattened. Distracted pilots may not have noted that the auto-thrust had also dropped out and that the throttle positions were now dictating thrust directly (after all, during a developed spin there's lots of centrifugal force and the pilots were probably not able to focus on the ECAM). They'd have been trying stick and rudder inputs for recovery. The nature of a flat spin is that recovery via unsustained flight control inputs is highly unlikely and, because of that high centrifugal force, anything not tied down in the cabin or holds would've migrated aft, further exacerbating the aft CofG complication for a flat spin.
The nature of spins generically is that the correct flight control anti-spin inputs must be made AND HELD to be ultimately effective in recovery. CRM just doesn't operate in a terminal spin - and in a FBW Airbus there's got to be a formal hand-over/take-over and a pressing of side-stick buttons, or control inputs are either negated or reinforced. The galvanic nature of a spin is not conducive to two pilots caught by surprise, agreeing on the correct throttle and stick/rudder positions and holding them static - at least not long enough for their inputs to overcome the spin's considerable inertia. Coherent analysis, communication and trouble-shooting is just not possible in a full-blown spin. AF447 has likely been one of those "points of no return" developments that lurk beyond aviation's rarely visited thresholds.
A sudden lunge into the deep recesses of coffin corner, courtesy of the triple ADIRUs finally rejecting erroneous pitot-derived data - and reacting and rejecting- is to swiftly and unexpectedly cross one of those thresholds. When examining the hazards of a known deficiency, and the urgency of any need to correct it, both EADS and the FAA should be looking at the possible complications arising at (or just outside) the edge of the operating envelope - and extrapolate from there.
.