@ DozyW concerning your experiment in an A320
It´s about time to rethink the value of the your tests and the result it presented. If it would have been that simple then BEA would have adopted that kind of cheap and easy testflight and would have presented the results out of it.
But as we can see in the following excerpts from the final report BEA has completely missed that possibility (which i doubt) or BEA came to the result, that such a test would be far off from reality. They even could have used an A330 simulator instead an A320 like you did, but again, they didn´t.
So we should accept your A320 test as an experiment with some discussion value in the period before the final report came out, but without reference or firm basis for answering questions, even BEA didn´t dare to.
Let´s focus instead on the report, there is enough potential for discussion and rethinking of former own positions and asumptions in it.
BEA FR 1.16.4.1
The validity of the model is limited to the known flight envelope based on flight tests. Consequently, it was possible to conduct the simulation on the period from 2 h 10 min 00 s to 2 h 10 min 54 s.
BEA FR 1.18.4.2
ˆValidation (proof-of-match) tests compare the behaviour of the simulator with that of the aircraft. A set of technical data (the data package) compiled during flight tests and aircraft certification serve as reference data for this objective comparison. The data only covers the aircraft’s known flight envelope;
In a developed stall situation the aircraft has left its known flight envelope. The data package does not contain any data relevant to this situation. The simulator is not representative of the aircraft in a developed stall situation; it does not reproduce the deterrent buffet effect.
1.18.4.5 Information reported by the manufacturer and the operator:
The information provided by Airbus and Air France managers highlighted the following:
ˆˆThe data currently available in the data packages prevents the simulator’s flight envelope from being extended, since the data in this package is limited to the aircraft’s known flight envelope;
ˆˆSimulators do not indicate to the pilots and instructors that the simulator has been taken outside the envelope validated by the data packages.
Improvements to this situation would appear, however, to be possible;
1.18.4.6 Work currently underway on simulator fidelity and training:
Following the accident to the DHC-8-400 operated by Colgan Air(17) on 12 February 2009, one of the NTSB’s recommendations (A-10-24) was that operators (notably of public transport aircraft) should define and codify minimum simulator model fidelity requirements to support the training of pilots in how to recover from stalls, including stalls that are fully developed. These simulator fidelity requirements should address areas such as angle of attack and sideslip, motion cueing, proof-of-match with post-stall flight test data, and warnings to indicate when the simulator flight envelope has been exceeded.
ˆSimulator fidelity must be improved to avoid the risks of negative training;
2.1.3.5 End of the flight:
At about 2 h 12, descending though FL 315, the aeroplane’s angle of attack was established around an average value of about 40 degrees. Only an extremely purposeful crew with a good comprehension of the situation could have carried out a manoeuvre that would have made it possible to perhaps recover control of the aeroplane. In fact, the crew had almost completely lost control of the situation.
BEA 2.2.5
However, positive longitudinal static stability on an aeroplane can be useful since it allows the pilot to have a sensory return (via the position of the stick) on the situation of his aeroplane in terms of speed in relation to its point of equilibrium (trim) at constant thrust. Specifically, the approach to stall on a classic aeroplane is always associated with a more or less pronounced nose-up input. This is not the case on the A330 in alternate law. The specific consequence is that in this control law the aeroplane, placed in a configuration where the thrust is not sufficient to maintain speed on the flight path, would end up by stalling without any inputs on the sidestick.
€EASA modify the basis of the regulations in order to ensure better fidelity for simulators in reproducing realistic scenarios of abnormal situations; [Recommendation FRAN‑2012‑045]
I´m not saying that recovery would not have been possible at all, but to base such saying on your experiment after BEA report is out and BEA did not perform such a test or did not consider it as valid enough to include it in their report is far fetched.