> The Captain of the Air France flight had logged 346 hours of flying over the past six months. But within those six months, there were only about four hours in which he was actually in control of an airplane—just the take-offs and landings. The rest of the time, auto-pilot was flying the plane. Langewiesche believes this lack of experience left the pilots unprepared to do their jobs.
They should really be talking about the First officer. The captain knew exactly what to do but wasn't at the controls of the plane until after it had already been stalled.
Not sure how flying a plane while it's essentially doing nothing is helpful to anyone.
If anything airline pilots should be taught to troubleshoot a plane more and fly it less. Maybe if the First Officer understood the correct way to exit a high altitude stall and that applying inputs to his control cancelled out the other pilot's inputs the plane likely would not have crashed.
A big plane in a normal condition doesn't require much attention and can fly itself. Humans are best at exceptional circumstances (in theory) and computers best at doing the boring and tedious work.
Flight hours in this context are pretty meaningless, they should be judged by percentage of abnormal incidents survived in a simulator instead.
Maybe the whole cockpit should be disconnected after takeoff and turned into a simulator where the pilots can just practice disaster scenarios for the duration of the flight. Worst case they get a chance to ace a real disaster in the mix.
I like the idea of what'd happen if there were some runaway misaligned incentives and it ends up that long haul pilots are forced to vividly relive airline crashes for their entire 14 hour flight to Australia. Years later everyone can agree that it's beneficial ("It's always been done that way") but they can't remember why they started doing it.
Projected Headline: "According to investigators, pilots mistakenly believed their readouts and controls were still in simulation mode until seconds before the aircraft struck the mountain, killing everyone aboard."
I mean, yeah, practice for when the automation breaks is good, but during the flight? That's just creating more mode-switching error cases.
> Not sure how flying a plane while it's essentially doing nothing is helpful to anyone.
When we train engineers for oncall, we take a look at incidents they saw while shadowing. Just X days shadowing doesn't really cut it, they have to be meaningful.
I see the comparison to flying pretty clearly.
Chaos engineering and game days are excellent ways to get engineers more experience in a system. Does the aviation industry have something similar?
Commercial pilots perform frequent recurring training in highly accurate flight simulators which can recreate just about any kind of system failure or emergency situation. It is normal in this training that they are given multiple simultaneous failures, along with confusing or unhelpful ATC instructions.
Some professional pilots choose to engage in leisure activities that improve their aviation skills. Sullenberger was a glider pilot. I know a few ATPs that fly aerobatic aircraft on the weekend.
Many people who love flying really love flying. I once met an extremely attractive woman who was a pilot. The problem was that flying is all she ever wanted to talk about. (I should say that that's just a problem for me. If I loved flight too then I'd be head over heels.)
> If anything airline pilots should be taught to troubleshoot a plane more and fly it less.
Air France 447 is generally viewed as showing the opposite need.
The disaster happened because the pilots were not at all involved in the flying process and suddenly thrown into a problem situation without that situational awareness.
If anything training is already very focused on troubleshooting. And yes systems knowledge could be better. But there's many cases where there's just no time for it and that human awareness is no important. But they are kinda zoned out, similar to when you're a passenger in a car and have less geographical awareness due to not being involved.
AF447 is often misunderstood, there simply is no easy root cause for that particular accident. Or any other aerospace accident, it always is a combination of multiple factors.
That the instruments readings were contradictory is indisputable, but when the anti-stall alarms are blaring in the cockpit, and the plane is literally falling from the sky for 3.5 minutes, than pulling the stick full back is not the way to go. In fact, pulling the stick back, even when applying full power, never has been, and never will be the way to get out of a stall.
The Thales pitot tubes were clearly faulty. The flight system output was absolutely contradictory. But understanding which instruments and which readings a pilot could potentially rely on in such a situation goes down to fundamental, basic airmanship. And for that, the blame lays solely on Air France and on their pilot training.
Pulling back to avoid overspeed is important, but stalling is certain death. And if the VMO and flutter flight-test videos are anything to go by the plane will definitely let the aircrew know they are getting there.
Agreed, the crew was confused and stuck in bubble where they believed pulling up would get them out of a stall, it didn't. Until the very end, they didn't come to the correct conclusion.
Again, training, alarm logic combined with a, in itself trivial equipment failure, led to the disaster. Picking only one of those factors, and all those I didn't mention, is just ignorant.
> In fact, pulling the stick back, even when applying full power, never has been, and never will be the way to get out of a stall.
It works perfectly well in an Airbus when operating in normal law AIUI because of envelope protection. The problem is that it doesn't work in alternate law which was the active mode at the time. It suddenly became necessary to control the aircraft like a "real" aircraft without a computer altering the inputs, and the pilots weren't prepared for that.
447 occurred from a design flaw in Airbus aircraft that have independent joysticks instead of the Yoke available on Boeing fly-by-wire.
The first officer who had more experience than the pilot flying the plane was unaware that the pilot was pulling back on the joystick. If the plane were properly designed as Boeing fly-by-wire craft were designed, eg, 777 at the time, then the accident wouldn't have happened because the first officer would have realized the pilot was pulling back on the yoke for the stall.
A principle in safety is that you want things encoded in the "hardware", eg, you can't put your car in reverse without putting your foot on the brake, even though you are taught this in driver's ed.
The Airbus designers forgot this major principle by using independent joysticks, a lesson not forgotten by the Boeing fly-by-wire craft.
Sure it was a design flaw but again the captain wasn't at the controls.
The first officer and the relief first officer were the ones who were at the controls. The relief first officer (who was the most experienced with that airplane of the pilots) wasn't aware the first officer (the most junior of the pilots) was pulling up on the stick because of the independent inputs.
You can absolutely try switching to reverse at 200km/h in a manual transmission. car at cost of your tranmission and potentially your bones in you hand. If you do it while rolling forward it will make funny sounds.
Apart from that the pilot flying the aircraft was totally confused and pulled the nose up while having stall warning and actually stalling. That is quite some remarkable screw up not understanding to stall and thinking having your nose at an high attack angle while having 100% engine thrust will solve anything in this situation. Sure with non-independent joysticks the other might have called the non nonsensical maneuvers earlier and prevented it. But the lack of awareness in instrument flight of that pilot is quite remarkable.
Can and should be able to are different things. I would imagine on a manual transmission they want things to be simpler so they have to make trade-offs and can't encode certain things without adding hardware.
But it's still not something anyone except the "I like direct control of the machine" types actually want, average users would probably prefer the safeguards.
Thing is, pulling the nose up silenced the stall warning. In the end, the flying crew was, wastly over-simplyfied, convinced to get out of a stall by pulling up. That alert behavior was changed by Airbus.
Blaming the pilots, or flatout stating that one design pholosophy is better than the other, is just ignorant. That's worse than all those soccer coaches knowing everything better on subdays, or whatever the US equivalent is, because at sport no lives depend on it.
And regardless of the latest Boeing fuck ups, everyone in aerospace knows that. And the people designing those planes and systems are fully aware of that, and know what they do.
Edit: The Vanity Fair piece also focuses on the stick behaviour and fly-by-wire systems. I cannot emphasize it enough, that pullong up silenced the stall alarm. The issue was that flight crew never realized that they were still in a stall. Pilot training aroubd that particular edga case, including simulator training, was ammended. As was the stall alarm behaviour. Fly-by-wire and stick behaviour had not much to do with it.
On a different note, for everyone blaming the pilots for being clueless amateurs (I exagerate, but I do get the impression): they died too on that flight. Inclusing the family of one of those pilots. Flight crew had as much skin in the game as possible. And before people start crying for remote controlled planes, how much risk does a pilot sitting in cubicle hubdreds of miles away from the plane actually take?
"Blaming the pilots, or flatout stating that one design pholosophy is better than the other, is just ignorant."
I work with safety, I study it.
The Airbus was poorly designed. You want to have forcing functions in the hardware, not depend of operator training, the "software" in safety terminology.
Airbus should have had the proper design philosophy so that the forcing function was in the "hardware", eg, the yoke of Boeing.
The poor choice of having independent joysticks brought the plane down.
Not certain why the FAA, which does know better, approved of the Airbus design.
That alone is worthy of a study. Was it politics?
In summary, safety guidelines mandate putting safety in the "hardware" of real forcing functions and not "software" -- training. That is the reality. 447 crash occurred because of poor Airbus design and the FAA for approving this design that goes against safety principles.
On physically connected controls this can happen as well.
It is why some flight instructors actually carried a hefty stick into cockpit - sometimes the best solution was to physically hit the trainee pilot to get them out of nervous hold on controls that you couldn't always overpower.
> you can't put your car in reverse without putting your foot on the brake
I rarely put a car in reverse while putting a foot on the brake. Handbrake on sure, but my foot is almost always on the accelerator when reverse engages, same as with forward gear
Certainly it was a design flaw, but also a major fault in training. Any properly trained pilot should have been able to detect a stall situation and pushed stick forward.
Aside: Magenta the color is interesting - the color code is #ff00ff. When I used to use the BBC Micro you had 8 colors by combination of red, blue, green on and off. Magenta is red and blue on, green off. So the color is a result of lack of memory for more colors. Probably the same reason for planes.
I started this particular subthread, and I actually do know how to spell fuchsia.
In college once I made a reference to "my red notebook over there" and the entire room (all women except for me) erupted in "that's not red; that's fuchsia" with one lagging voice adding, "or hot pink."
So yeah, I can spell it, if only because I had never heard of that color before then and I looked it up.
The funny thing is that real Fuchsias display such a diversity of colours as to make naming the colour rather pointless. It is not substantially more sensible than calling a colour 'rose'. I'll think I will stick to the hex codes ;)
I feel like this is true with the new crop of personal 'evtol' aircraft. There's a lot of marketing being said about being so "easy" to fly that even your "concierge" could do it. I do wonder how much aviating you could do in those things and if they even have a "manual" mode so to speak.
Vortex ring state is an extremely real danger with these craft, the computer is what stands between the occupant and a fiery death. Hence the concierge treatment.
There's no manual mode per se with something like a quadrotor, even a simple one requires really unintuitive applications of thrust to affect things like yaw. You could fly one in 'rate' mode though - where the stick deflection commands pitch/roll/yaw rates directly.
Unlike a 172, it's pretty unstable flown like that, and most people would turn such a craft into a smoking crater shortly.
I have crashed an experimental heavy lift multirotor (80kg payload on board) due to VRS. Was maybe 30 feet off the deck, no time or altitude to recover. That was a fun one to explain.
I'm still a bit unclear here; I'll have to give examples.
I have "several" quadcopters. I don't actually know how many I have at the moment, but in addition to enough components to fill a couple of five-gallon buckets, the three I use most regularly are:
* an Autel Evo 2 Pro
* a 5" FPV quad
* a 3" FPV quad
The Autel is a dedicated photo/video quad. It's stupidly easy to fly, to the point that I can put the controller down and come back 20 minutes later to find that it hasn't moved.
I use the 5" for fun, mostly freestyle stuff. The 3" has enclosed props and is intended for videography. Neither of these have GPS. While they do have flight controllers, they are optimized for minimal input latency and do not attempt to automatically keep themselves upright whatsoever. Flying them is much like balancing a dinner plate on top of a broomstick; it requires constant input adjustments. That's difficult at first, but after ~20 hours of practice it starts to become unconscious. I've been doing it for a few years now and really don't even think about it anymore.
Prior to the availability of ARM STM32-based flight controllers, I believe there were some fully analogue solutions out there that allowed similar flight characteristics.
At the most detailed level, all of them are sending a PWM signal to each individual motor. Controller input is translated from 4 axes (pitch, roll, yaw, throttle) to the appropriate changes in the PWM signal for each motor.
I _think_ I could probably set up a fully manual quadcopter and get it to fly predictably without a flight controller, by using the radio receiver on the quad to output PWM directly to the individual motors. I'd have to set up the controller so that input on each axis is "mixed" appropriately to the respective per-engine output channels. The result should be (barely) controllable. I bet wind and the dynamics of lift would be much more difficult to deal with, but I think it'd be doable.
10 years from now nobody will be able to back an old car out of a parking space because old cars don't have backup cameras.
This doesn't make backup cameras bad but it does require that new drivers be trained for what happens when they fail. This idea is fairly new for driving instruction but it used to be mandatory for pilot training.
"Children of the magenta" should be required reading/listening for anyone interested in aviation.
https://99percentinvisible.org/episode/children-of-the-magen...