I tend to disagree when you suggest the hands off calculation is highly speculative.
The excellent match between DFDR and the hands on calculation shows that the wind/gust/control combination was accurately measured, and from there to a hands off state is a simple subtraction of the response to the control applied.
If AI can't calculate that accurately then you won't have any approved simulators.
Hi Owain,
I am not commenting on the accuracy of the measurements implied. I am commenting on using those measurements in a situation where you are computing the effects on a theoretical aircraft whose flight path diverges from the known flight path.
If the the aircraft averages ~10 degrees wing down for the duration of the simulation, then it will be flying through a different part of a turbulent air mass, and the assumptions used to develop the effect on the aircraft will diverge from what would have happened in reality as time increases.
I think everyone would be better served by a Monte Carlo simulation of the possibilities so that we could better understand the aircraft's natural tendencies and the factors required to make it diverge substantially from those natural tendencies.