Hacker News Reader

Comment by glbrew | Hacker News Reader

A child in the cockpit once accidentally disengaged some controls which eventually resulted in a crash that killed all 75 people on board.

https://en.wikipedia.org/wiki/Aeroflot_Flight_593

View on HN

This is actually an odd case, as it was mainly the result of improper training and different configurations between two planes. In a Boeing plane, autopilot will never disengage without specifically being disengaged. In an Airbus plane, giving a 'harsh' enough input will disengage the autopilot. The pilot here mainly had experience on Boeing, so was not aware of this minor but very important difference.

The Air France flight that crashed into the Ocean by Brazil was also partly the fault of different designs by the two major aircraft manufacturers. In Boeing planes, the input is synced between the pilot and copilot (if one pulls up, the other physically moves as well). In Airbus planes however, the input is merged together instead. If one pilot pulls up, and one is pushing down, nothing really happens.

This ended up partly leading to the crash, as one pilot heard the stall warning and pulled up (causing more speed loss), while the other pilot correctly pushed down. If the inputs were synced, the other pilot likely would have noticed his copilot was making the wrong input.

> In Airbus planes however, the input is merged together instead. If one pilot pulls up, and one is pushing down, nothing really happens.

Which is just crazy! I wonder who the product manager who thought "what should happen if one pilot is pushing on the stick and the other is pulling? I know, just average them and fly straight" was. There are various things you can do in that situation that would kind of make sense (the Boeing design being the best), and that is not one of them.

They might have thought "there is no sensible answer in that case, so just have the behavior be the natural operation of the existing systems as designed" which turned out to be adding the inputs.

Except that there is a sensible answer: you physically force the two controls to synchronize.

Once you settle on a design where the two controls move independently of each other, then yes, there's no sensible answer. But, don't do that.

Even then, anything else would make more sense, for example "the left stick always has priority and the right stick flashes a red 'disabled' light", or anything else.

Agreed. No solution will be good, but there are definitely less-awful ways than averaging the inputs.

I suppose what the Parent was getting at, is that it doesn't make any sense to have two sets of completely disconnected flight controls in the cockpit. That's a lot like having two different mice plugged into your computer at once...

I can't think of a situation that would warrant one flight control being pushed and the other pulled...

Most aircraft have both sets of flight controls physically connected (either via direct links, or actuators driven by flight computers). So if one pilot pulls back, the other set of controls comes back as if it's being pulled on. One of the advantages of this design is both the pilot and copilot can be aware of what the control input is at any given moment.

> The Air France flight

More information for those interested. It's a long read.

Or, you know, the captain could have told the co-pilot "I have control" as they are typically trained to do in handling emergency situations!

> This is actually an odd case, as it was mainly the result of improper training and different configurations between two planes.

That might be the case, but I would rather not be on a plane where children are allowed inside the cockpit. Parents can visit an aviation museum if they want to.

> One of the children had unknowingly disengaged the A310 autopilot's control over the aircraft's ailerons while seated at the controls. [...] With the autopilot active, Kudrinsky, against regulations, let the children sit at the controls

I remember visiting the cockpit pre-9/11 and it did not include being invited to sit at the controls. Or, in other words, children visiting the cockpit was probably more the proximate cause than the root cause.

The child sitting at the controls was the pilots child, that's the reason for the increased (but certainly unacceptable) access.

There is a difference between letting a child in the cockpit and having him sit in the pilot seat, as it was the case in the Aeroflot Flight 593 crash.

That transcript is harrowing.

Seems kind of similar to the plot of the Grisham novel Airframe.

s/Grisham/Crichton

Reply on news.ycombinator.com