2
$\begingroup$

In line with this great question, what's the possibility of a commercial airplane crashing due to a computer error? Or are there already known cases of this? As a passenger, should you be worried about this possibility happening to you someday in the future?

$\endgroup$
  • $\begingroup$ "should you be worried about this possibility happening to you someday in the future?" Here is some advice: "Don't worry about the future. Or worry, but know that worrying is as effective as trying to solve an algebra equation by chewing bubble gum." $\endgroup$ – DeltaLima Oct 7 '17 at 18:55
  • $\begingroup$ Air France Flight 296 had allegations of improper response by the computerized flight control system during the crash, but the official report stated pilot error. The computer did not allow more nose-up (alpha-protection engaged) to prevent going into a stall, however the captain said that it was an error in the fly-by-wire controls that did not give him elevator authority or increase the speed of the engines. $\endgroup$ – Ron Beyer Oct 7 '17 at 20:56
  • $\begingroup$ Do you really mean a computer error (e.g. the computer announces an error state, or produces incorrect output) or a general computer fault (something goes wrong with a computer system)? $\endgroup$ – Daniele Procida Oct 7 '17 at 22:07
7
$\begingroup$

First; let's be honest. No one really knows what the exact probability is.

But we do have a design certification system that requires developers to apply various development assurance processes to meet specific safety standards for the design. The primary requirement is covered by §25.1309 Equipment, systems, and installations.

That has a lot of words about identifying potential faults and mitigating the associated hazards. But there are no numbers. As with most rules, the FAA conveniently provides an Advisory Circular to help you show compliance with the rule. In this case, its 25.1309-1A - System Design and Analysis.

The acceptable standard for critical faults (those with a potential catastrophic result) is found at the bottom of para 10.

(3) Extremely Improbable failure conditions are those having a probability on the order of 1 X 10^-9 or less.

So, if the developers do their job to the standard and the standard (and the methods used to evaluate it) are accurate the answer to the question is:

The probability is less than once in every 1 billion hours of operation.

As a point of statistics, that number applies to the cumulative hours of all aircraft of that design.

| improve this answer | |
$\endgroup$
  • 2
    $\begingroup$ +1 for coming up with an actual quantitative regulation for that :) $\endgroup$ – TomMcW Oct 8 '17 at 1:41
  • 1
    $\begingroup$ What about periods of strong solar activity? VLSI chips are particularly sensitive. $\endgroup$ – mins Oct 8 '17 at 14:29
  • 3
    $\begingroup$ @mins Avionics design must take into account the possibility of neutron single event upsets (NSEU) and include appropriate mitigation techniques. OEMs such as Boeing specify expected flux densities (the airframe does affect the value). Non-critical systems often just use watchdogs and reset if needed. Critical systems have more complex protections including ECC and mirrored memory and modified software to detect and correct errors due to NSEU. I've had circuits tested at Los Alamos Labs where they are bombarded with neutron beams. And to your point, it gets harder as the parts get smaller. $\endgroup$ – Gerry Oct 8 '17 at 17:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.