I’m sure you’re probably referring to Air France 447 and/or Colgan 3407, both of which were stalled.
I think a lot of it has to do with the fact that most airline pilots haven’t actually stalled a plane in a really long time. Usually, pilots learn to stall an aircraft early in their training, and then it really isn’t required any more.
Well, I should put that in the past tense. Since those accidents, a lot more emphasis has been placed on both stall awareness as well as reducing automation dependence and hand-flying the airplane more often.
In those particular instances – In Colgan, there was an extraneous conversation about icing happening in the cockpit when they should have been focused on the approach (violated the sterile cockpit rule). The pilots reacted in a manner that would have been much more appropriate for tailplane icing, not realizing it was a plain old stall.
On Air France, the start of the problem was some erroneous indications due to a problem with the pitot tube, but the FOs did a terrible job of hand-flying it. In addition, a lot of the Airbus control and automation design theory made things more confusing. In particular:
- The Airbus does not prevent the pilots from providing different inputs, it merely averages them and provides a “Dual Input” warning which can easily be missed in the heat of a situation such as AF447.
- There are Angle of Attack probes, but the AoA is not displayed to the pilot at all.
- The plane flies differently in an emergency than it does every day. There are three separate sets of logic: Normal Law, Alternate Law, and Direct Law.
- The warning logic ignores certain inputs if it believes them to be incorrect.
Because of #2 and #4, the pilots didn’t really know whether they were in a stall. Eliminate #2 by showing the pilots the AoA and it’s obvious that they’re in a stall. Eliminate #4 and they may have realized that they were in a stall and they probably would have successfully recovered rather than getting close to recovery and un-doing the recovery because the warnings went off when they tried to recover.
Because of #1, the left seater’s inputs were ineffective because they were being averaged with the right seater’s inputs (stick all the way aft). There should NEVER be more than one person flying the plane, so I cannot understand why they would average the inputs, nor why they don’t set off all kinds of alarms when the plane is receiving multiple inputs.
Because of #3, the wrong habits are formed. The right seater’s input, holding the stick all the way back, is the correct input *when the plane is in Normal Law*, it will automatically hold the maximum non-stalling Angle of Attack, providing the most possible lift. However, the airplane was no longer in Normal Law and that envelope protection was gone. As a result, the input that’s right under normal circumstances (and is trained for) was not the correct input. Humans are creatures of habit, which is why training is so effective most of the time – Unfortunately, when the airplane doesn’t always behave the same, those habits no longer work.
Because of #4 and the fact that they were in a deep stall (~40º AoA), the computer decided that the AoA must be incorrect and shut off the stall warnings. When the pilots actually attempted a correct recover, the computer changed its mind and all the warnings went off, leading the crew to decide that the action they had taken was incorrect, whereupon they pulled it back into a deep stall.
It’s easy to blame the pilots in hindsight, but when you have only a few minutes from flying along “fat, dumb and happy” on autopilot at cruise altitude to being literally dead in the water, and the airplane seemingly working against you the whole time, it’s hard to piece everything together in time to save the day. RIP to all aboard.