26

I'm trying to track down a story about an incident involving an Airbus airplane my father told me about 25 years ago, when he was working at a Boeing supplier.

Reportedly, what happened was the airplane's autopilot was set to bring them in to a certain airport, and the pilots noticed that it was descending too early. So, they took over the controls, lined up manually, and at about a hundred feet up, the autopilot wouldn't let them descend further. So, they did a go-around, tried the approach again, and the same thing happened. At this point, they diverted to another airport and landed without incident. Turned out that the airplane's database had an incorrect position for their destination: the plane thought they were overshooting the runway, and wouldn't let them perform what it saw was an off-airport landing.

Was this based on an actual incident, or just one of the "Airbus airplanes don't let the pilot fly" stories circulating among Airbus competitors?

Mark
  • 5,552
  • 1
  • 31
  • 54
  • 3
    never heard of such accident nor such capability of the airbus autopilot. I put my chips on the second hypothesis. – Federico Jun 17 '14 at 10:38
  • 7
    If you find a notable reference for this story, it being retold on a website or in a book, then its a good fit for http://skeptics.stackexchange.com – Jamiec Jun 17 '14 at 12:38
  • 1
    I'd say this must be an urban legend. Not sure if it was like this everywhere, but 25 years ago, there existed a rule for safety critical systems, that any kind of automation was only allowed if it could be shut down completely by a human. Note that it's not more like this nowadays. – Zane Jun 17 '14 at 13:41
  • @Zane, not anymore? got any links/documents to support the claim? – Federico Jun 17 '14 at 14:22
  • 1
    I don't know about not letting the aircraft land, but there was an incident where the flight control system wanted to make sure they landed. – voretaq7 Jun 17 '14 at 15:33
  • 9
    @voretaq7: No, it didn't. The aircraft didn't start to climb because it didn't have sufficient speed for it. The computer did prevent pulling up further, but if it didn't the aircraft would have stalled with the same overall result. They were simply too late applying power (which the computer would have done for them much earlier if they didn't disable it!!) and nothing could have saved them at that moment. – Jan Hudec Jun 17 '14 at 16:55
  • @Federico: Might in fact be that this is still a requirement in avionics, I don't know. It seems it's not a requirement for automotive, where software support emerged much later. – Zane Jun 17 '14 at 18:40
  • @Zane I find that extremely hard to believe, such requirement would create only problems and no benefits. – Federico Jun 17 '14 at 18:41
  • 5
    @Federico: which requirement? That a human can shut down the automation? I'd say this is a very natural requirement that causes no problem at all. – Zane Jun 17 '14 at 18:46
  • @Zane, that it got removed as you say in any kind of automation was only allowed if it could be shut down completely by a human. **Note that it's not more like this nowadays** – Federico Jun 17 '14 at 18:50
  • @Federico: I remember this from the safety critical systems lecture. I agree that this is dangerous. However, I think this is really true nowadays for automotive, maybe because it's less dangerous there. Might very well be that this is still a requirement in avionics and power plants. – Zane Jun 17 '14 at 19:25
  • @Zane we were speaking about aricraft here, not automotive. And you claimed that such a requirement was no more in place, that's why I asked for sources. Also, the subject had been address here before – Federico Jun 17 '14 at 19:36
  • 1
    @Zane: I don't think there is, or ever was, any such requirement for the kind of automation we are talking about here, i.e. something that translates direct human commands via non-trivial controller. In Airbus you sure can disable the FBW, but you have to use the circuit breakers and you would be left with elevator trim and rudder only (yes, the aircraft was demonstrated controllable with just that). – Jan Hudec Jun 17 '14 at 21:00
  • @Fedrico: sorry for being imprecise and puzzling you. No I don't have a source for avionics. – Zane Jun 17 '14 at 21:01
  • The details don't match up, but you may be interested in Air France Flight 296. – 200_success Jun 18 '14 at 07:29
  • 3
    There is no such requirement that pilots must be able to override any automation in an aircraft. Simple example that this is the is FADEC systems for engine control. The FADEC computers know more about the engine at any given moment than any pilot (or two) can process in their heads. Gone are the days of having to monitor temps to make sure you don't burn up your very expensive turbofan engine. And yes, FADEC is avionics -- aviation electronics. – Juan Jimenez Apr 12 '17 at 19:48

3 Answers3

31

I believe this is a hoax1. I've read through several documents describing the protections provided by Airbus flight laws (this is not autopilot; autopilot is a separate layer on top of it) and have never seen any mention of any kind of ground proximity protection. Only standard (E)GPWS which yells "terrain, pull up".

Update: Of course now there is EGPWS that gets aircraft position from GPS and can yell "terrain, pull up" when there is no airport ahead, but 25 years ago there was only GPWS based on radar altimeter which didn't know anything about position and would switch to simply read out2 the radar altimeter value whenever gear (and maybe also flaps; not sure here) was extended.

1Might have been even attempt at FUD against the then novel technology.

2That is say "one thousand", "five hundred", "two hundred", "one hundred", "fifty", "forty", "thirty", "twenty", "retard"3 (and "ten" or "five" when it stops saying "retard" and the wheels are not on the ground yet). This is useful during approach so the pilot can concentrate less on the instruments and more on outside reference and in the final phase to help judge when flare should be initiated. It will also warn "sink rate" when the sink rate is excessive (more than -1000 ft/min, the stabilized approach criteria).

3At 20 or so feet, if the thrust levers are not in idle gate, the Airbus system starts announcing "retard" and keeps announcing it until the thrust levers are placed in the idle gate (or reverse). This is specific to Airbus. I believe it can also announce "minimums" when approach minima are configured in the flight management system.

Jan Hudec
  • 56,247
  • 12
  • 155
  • 268
2

I would categorise this as an urban legend, as there is no this kind of function or protection in Airbuses. Any protection will not kick in until you are trying to exit aerodynamical envelope of the aircraft.

But many of these stories are based on true events, as early Airbus pilots were mostly experienced second-generation-jet pilots, who just couldn’t wrap their head around with new technologies A320 offered. As the pilots could not understand what the flight management system was doing, they disconnected it with various level of success and told the stories as they perceived it.

It took decades for the industry to actually understand the complex interaction between the pilot and the machine and to develop levels of training required to operate these modern aircraft. Paradoxically, now that pilots understand better all the modes of the system, they have less hesitation to take over manually when required.

busdriver
  • 3,340
  • 7
  • 17
1

This urban legend has been going around for ages and one flight (not sure if it was the start or if there were others) which started the myth that Airbus pilots are just along for the ride was the LH2904 accident in WAW.

In this case, the spoilers and reversors did not deploy as the conditions for deploying were not met resulting in aircraft going off the runway at highish speed.

The cause of the accident was pilot error with contributory factor of aircraft system/lack of guidance in the AOM. link to accident report here

Anilv
  • 4,566
  • 17
  • 16