In the movie Sully we have this section of dialogue

He wasn't following proper procedure at all. And I know because I had the QRH in my hands. He switched on the APU immediately after engine blowback. According to the Airbus that's the 15th thing on the list to do. 15th! If he had followed the damn rules, we'd all be dead.

Now, another question discusses the importance of the APU to the crash, but my question is, why is it so far down the list? I suspect the above is dramatized somewhat, but it does seem odd that it's so low. What does Airbus consider more important in their handbook?

  • 2
    $\begingroup$ Starting the APU is actually the 11th item on the checklist for dual engine failure, and quite a few items can be skipped (like the relight attempt, oxygen masks, notifying ATC, etc). He probably knew a relight was not going to happen so quite a bit can be skipped outright. $\endgroup$
    – Ron Beyer
    Commented Dec 30, 2017 at 5:23
  • 2
    $\begingroup$ Here are the excerpts from the QRH for the A320, dual engine failure. $\endgroup$
    – Ron Beyer
    Commented Dec 30, 2017 at 5:31
  • 2
    $\begingroup$ It's not particularly an error on Airbus' part, if it is before the oxygen mask stuff, and that failure happened at high altitude, then there's a risk of hypoxia. This is where memory items and training and common sense come in. Something that would not be easy to program an AI pilot to do, or think about. (A mere comment.) $\endgroup$
    – user14897
    Commented Dec 30, 2017 at 11:33
  • $\begingroup$ @MichaelKjörling Sully seemed very accurate on the terminology aspect. It wasn't Star Trek, where the names are made up, and the points don't matter. $\endgroup$
    – Machavity
    Commented Dec 30, 2017 at 16:47
  • $\begingroup$ @ymb1 It would probably be fairly easy to make separate checklists for, say, dual engine failure below 10,000 ft (where supplemental oxygen is unnecessary even for extended periods) and dual engine failure above 10,000 ft (where supplemental oxygen starts to become needed, depending on duration of operations, but you're able to glide farther). I guess I personally hope that Airbus will respond by making such checklists, plus one specifically for dual engine failure during early climbout; possibly by renaming the existing one "dual engine failure above 10,000 ft" or some such. $\endgroup$
    – user
    Commented Dec 30, 2017 at 16:59

1 Answer 1


The assumption was that a multiple engine failure happens at altitude, for instance in cruise while flying through volcanic ashes. This happened to British Airways Flight 9 and to KLM flight 876, with all four engines switching off due to the ash cloud. Total engine failure at altitude buys time, and if time is available first priority is to attempt to re-light an engine. APU will probably be off during cruise, and given the choice to re-light an engine or to light the APU it is clearly the engine that is the most beneficial to start.

All-engine-failure in airliners is very very unlikely. It is a function of number of engines - in a four-engined aeroplane, the chance of all four engines failing is assumed to be super extremely incredibly unlikely. With a statistical failure rate of 10$^{-5}$ per flying hour per engine, a four-engined aeroplane would experience all engine failure due to statistical causes once in $\frac{10^{20}}{24}$ flying hours1: a billion times a billion times four. So very uncommon that the only practical consideration would be a common mode failure from external causes, such as the volcanic ash cloud.

For a twin-engine like the A320, chance of all engine failure due to random causes is 2 * 10$^{-10}$, still considered an Extremely Unlikely event. Procedures are written for the least unlikely events, and updated according to events that actually happened. Now that a dual engine failure during take-off has actually happened, updating the procedures will be the result of failure analyses: how likely would it be to happen again, or what other procedures can be put in place to avoid this situation.

1: The chance that any of four engines fails is 4 * 10$^{-5}$ per flying hour. Then 3 * 10$^{-5}$ per flying hour for the remaining engines etc.

  • $\begingroup$ In the volcanic-ash scenario, you describe in your answer, would immediately starting the APU cause the APU to ingest volcanic ash, too? $\endgroup$
    – Dohn Joe
    Commented Feb 20, 2019 at 13:38
  • $\begingroup$ Yes it would, and it would cause loss of valuable time in getting the engines restarted. $\endgroup$
    – Koyovis
    Commented Mar 1, 2019 at 0:55
  • $\begingroup$ I think the 10^20 is not correct because the multiplication of the faillure rates assumes that all engine failures are independent statistical events. This does not account for an event that affects all engines at the same time (like the big flock of birds in case of the Hudson!) $\endgroup$
    – ROIMaison
    Commented Aug 23, 2023 at 11:42
  • $\begingroup$ @ROIMaison indeed, a common mode failure, like stated in the answer. $\endgroup$
    – Koyovis
    Commented Aug 24, 2023 at 20:38

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .