The big problem was that the pilot-in-command simply forgot (or did not know) how to fly the plane. He overcorrected in reaction to turbulence and the rest is history. Normal law does not allow the pilot to set an angle of attack that would lead to a stall. In the alternate law modes this protection is not there and so when the PIC didn't adhere to the rule of "don't pull up while in a stall" there was no safety net to prevent him from exacerbating the situation. In fact he kept pulling back when the stall alarm went off.
A better UI would not have solved the self-inflicted problems. From the findings sections in the wiki entry:
- The pilots apparently did not notice that the aircraft had reached its maximum permissible altitude
- The pilots did not read out the available data (vertical velocity, altitude, etc.)
- the crew made inappropriate control inputs that destabilized the flight path;
- the crew failed to follow appropriate procedure for loss of displayed airspeed information;
- the crew lacked understanding of the approach to stall
- the crew lacked practical training in manually handling the aircraft both at high altitude and in the event of anomalies of speed indication
- the two co-pilots' task sharing was weakened both by incomprehension of the situation at the time of autopilot disconnection, and by poor management of the "startle effect", leaving them in an emotionally charged situation
- the crew did not respond to the stall warning
A better cockpit may have given the pilots a better chance of surviving but definitely wouldn't have guaranteed a better outcome. One only needs to look at Asiana to see that even Boeing, with their contrasting approach to UX, can't save pilots who lack basic airmanship. My understanding is that most long-haul pilots will fly a fairly small number of segments each month, and that most of those segments are going to involve a huge amount of time spent in the middle of nowhere with autopilot on. This leads to atrophying of skills needed to manually fly a plane.
Conversely, one only need to look at Qantas' QF32 incident to see that good airmanship can absolutely make catastrophic failure survivable. Yes, there is a lot of information to process in an emergency... but the Qantas crew took that information and delegated appropriately. The Air France pilots lost their cool and fucked up.
The pilot flying wasn't the pilot-in-command. Half the problem was the PIC was out of the cockpit at the time and the lack of any clear command structure between the two first officers in the cockpit. You can also argue that it may have been recoverable if the PIC took over as PF when he returned to the cockpit (had he acted rationally, etc.), instead of being confused by the situation they were in and unable to get out of, likely without knowing what the PF was doing.
> My understanding is that most long-haul pilots will fly a fairly small number of segments each month, and that most of those segments are going to involve a huge amount of time spent in the middle of nowhere with autopilot on. This leads to atrophying of skills needed to manually fly a plane.
Much flying nowadays is through RVSM (reduced vertical separation minima) airspace, and you're not allowed to hand-fly the aircraft there (in general, modern autopilots can maintain altitude better than a pilot can).
There were a Captain and two co-pilots aboard AF445. The junior co-pilot was PIC for much of the incident.
See also <https://aviation.stackexchange.com/questions/33414/who-is-th....
As in the AF447 case, the PIC will delegate to an acting PIC while they are on their rest break, but ultimate responsibility still remains with them.
PF Pilot Flying
PNF Pilot Not Flying
https://www.bea.aero/docspa/2009/f-cp090601.en/pdf/f-cp09060...
But where does your viewpoint meet up with someone like Chesley Sullenberger who says the pilots would have probably not crashed if the plane was a Boeing rather than an Airbus, because it would have had an angle-of-attack indicator?
The initially-designated co-pilot also was virtually fully occupied simply clearing alerts from the aircraft's control and diagnostics systems -- for about 45 minutes as I recall.
That's a near-deadly case of TMI, I suspect.
For an industry that is so safety conscious, I believe this is a huge flaw in disaster control. To put it in context that most readers here can understand - lets say your everyday computer uses OS X as its operating system, but one day, with deadlines to meet, you have a huge crash, and the OS reverts to Windows, and you have to find the problem and remember the appropriate commands under intense pressure...