JD-EE
It is, at the very least, "unkind" of the AP to "dump" control of the aircraft back in the hands of the pilot under such highly adverse circumstances - perhaps beyond the skill of
any human to deal with. I find the work being done by the folks at Bielefeld University (
AG RVS - Welcome to Networks and Distributed Systems) on the human/machine interface encouraging in this regard. Hopefully they, or others engaged in such work, will soon find a better mode of the AP relinquishing control when circumstances are beyond its capabilities. They have some additional comments on the AAIB report here:
AG RVS - Computer-Related Incidents with Commercial Aircraft. The AAIB recommended that overtaking aircraft indeed adopt a lateral separation approach but its implementation is complicated by rules in differing flight juristictions. Bielefeld also has an ongoing analysis effort related to
"Computer-Related Incidents with Commercial Aircraft" found here:
AG RVS - Computer-Related Incidents with Commercial Aircraft.
HN39
Unfortunately, I have not been able to find Appendix B of the AAIB report but I will search further later today. I had hoped the Bielefeld link I mentioned above would lead to the Appendix B but it did not. The AAIB, IMHO, "punted" on the autoflight aspects of the incident stating "
The fact that this law was not triggered until 10 seconds after the autopilot disconnected was a random event driven by the severity of the turbulence. Had the turbulence been more severe at the first encounter and coincident with the overspeed warning, reversion to AoA law could have been triggered as soon as the overspeed condition disconnected the autopilot. Nevertheless, it should be noted that had the autopilot remained engaged, the AoA law would not have been invoked because it is
inactive except in manual control." Note the italics and bold are mine for emphasis of that sentence. Perhaps AF 447 also experienced such a "random event".