Originally Posted by
safetypee
PPRuNe is a valuable forum for sharing ideas and improving safety
It might be so, yet being anonymous internet forum it (as expected) suffers from quite low wheat/chaff ratio, mostly brought on by good folks who get their kicks from proudly displaying their aeronautical ignorance here. I find it especially funny when convoluted and basically wrong theories are supported by reference to author's self-professed experience & expertise.
I'll try to answer a few questions I think might be worth answering, to the best of my ability. As usual, feel free to correct me.
1. What happened here?
Pilot induced upset, followed by pilot caused and pilot sustained stall that was broken only by disintegration of the aircraft upon impact with the sea surface. G trace shows no turbulence signature, aeroplane behaved i.a.w. flight control orders made, as long as it was aerodynamically possible to do so.
2. How often does it happen?
AFAIK, there is no proper study of it but there are fairly useful indications that it happens very, very seldom but when it does, it tends to be fatal.
First, in connection with AF447 investigation BEA analyzed frequency of multiple pitot blockages and found 40-odd of them on 330 and 340. All of them resulted in degradation of flight control law to alternate and AP disconnect. Some of them even went unreported as the crew didn't think much about having to handfly in altn mode, at high altitude. Some received stall warning. AF447 was unique not just because full back stick was sustained after stall warning, it was the only one where reaction to stall warning was to pull! So much for the "uncoupled sticks are gonna kill ya and we need stickshakers because overloaded crew won't hear anythıng"
Second, do you really believe that in this day and age of enlightened investigation authorities, flight data monitoring, Flightradar24, AvHerald and social media it is possible for some airliner busting couple of levels in extreme attitude and RoC unsustainable even at SL, empty and with maximum thrust and then stalling but somehow recovering and landing safely to go unnoticed? Yeah, neither do I. There were some lucky escapes like Dynasty 006 back in 1985. where diving aeroplane managed to get into VMC so the crew finally figured out their AHs were not toppled and recoverd or Flagship 3701, which was a real shame that the crew finally managed to kill themselves as the post-accident interview, revealing what in the world were they thinking as they forced aeroplane into stall and how did they manage to gather their wits (I suspect the distinct lack of them in the first place made the task easier), to bring it into more-or-less controlled glide would be precious.
3. What do we do to stop it from happening again?
I don't know. It's really just the matter of psychology and from my (quite limited, I admit) perspective, I don't see any particular effort, coming from any aviation psychologist, in trying to make useful theory what the
happens in severe pilot-induced-upsets. AF447 report was very weak on it, QZ8501 is utter disgrace.
However, I can tell you a few ways that some believe might reduce chance of the QZ8501-like event reoccurring but won't do good at all.
You can't prevent it by fitting connected controls; there are far more cases of conventional controls aeroplanes being pulled into stall or spiral dive till impact than FBW Airbi and having interconnected yokes did not help at all. For all the cries of "I need to see what my effoh is doing to his stick!", no FCOM or FCTM reference to "As PNF, observe your PF control input procedure" was ever brought forward.
You can't preventing it by installing AoA gauge. Pilot who forgets about maintaining proper attitude, which is the very basic of flying from the first second of his very first flight, stands no chance of checking alpha.
You can't prevent it by installing pusher, for it was shut off at Staines and overridden at Buffalo and Jefferson City.
You can't prevent it by increasing upset recovery training. Aerobatic aeroplanes quite differ in characteristics from transport ones. It is not true that airliners' high altitude stall characteristics are unknown; they are tested but recovery is effected immediately past alpha max and that's the data that is fed to sim manufacturers so sim can realistically reproduce it but no test pilot is suicidal enough to pull to alphas above forty, so that is the area of quite some conjecture. Anyway, the biggest objection to more training is that everyone knows it is training and one is prepared and knows what to do. In real accidents, crews were presented with flyable aeroplane and they only had to minimal corrections (or nothing at all) to keep it flying. It's not that AF447 or QZ8501 crews did not do the UA or stall recovery properly, they never initiated it after creating upset themselves in the first place.
You can't prevent it by having more CRM or more experience. A pilot who has a panic attack and reverts to atavistic notion that aeroplane is trying to kill him by diving and only way to prevent it by pulling the stick was very well described by Wolfgang Langewiesche in his deathless tract "Stick and rudder", back in 1944. Conventional wisdom has it that better training of airline pilots and multi-crew concept should eradicate this kind of accidents in airline environment. Alas, this is only partly true. While better training really seems to reduce the number of occurrences, in the accidents where one pilot goes brains off and gets into upset, the other is very inefficient in figuring out what's going on, no matter what his experience might be. Capt Irıyanto was a former fighter pilot so he for sure knew a lot about unusual attitudes and stalls, yet his attempts at recovery were weak and ineffective, showing that he dıd not understand the gravity of the situation. AF447 CM2 was former glider pilot, so he had to know about energy management. Captain of Swiftair MD-83 who kept pulling all through the spin into the spiral dive was freakin' TRE! What more experience and skill do you want?
Issue is what made these pilots forget the very basics of flying a couple of minutes before they perished. For the time being, it's either we don't have enough data to answer this question or our analysis tools are inadequate. Perchance both.