I wonder if there is an incident that pilots saved the plane or minimized the damage by bending or breaking the rules like Sully? (As it's known Sullenberger started the APU at first which was the 15th thing in QRH.)

Thanks in advance.

  • 59
    $\begingroup$ Pilots are, in fact allowed by law in 91.3(b) to deviate from any "rule" in an emergency to do whatever they see is the best course of action. It must happen often enough that the law specifically allows it... $\endgroup$
    – Ron Beyer
    Commented Feb 17, 2020 at 20:09
  • 10
    $\begingroup$ I wouldn’t call what Sully did bending or breaking any rules per se. Capt Sullenberger just had a very good understanding of the aircraft and it’s systems and what he was going to be needing in an emergency. $\endgroup$ Commented Feb 18, 2020 at 4:06
  • 8
    $\begingroup$ What Sully did was not exactly breaking a rule per se. $\endgroup$ Commented Feb 18, 2020 at 5:08
  • 14
    $\begingroup$ There is very, very little reporting on "prevented accidents," simply because nothing interesting happened on that flight. If the application of common sense to some unusual situation averted some bad outcome, you often don't have clear proof that "if we hadn't done ___ we could have crashed". You just know that things turned out okay. How bad they might have been is speculation. $\endgroup$
    – Ralph J
    Commented Feb 18, 2020 at 17:33
  • 6
    $\begingroup$ My soaring club requires you to make radio calls, but I once skipped those radio calls because I had one hand on the airbrakes and the other hand on the stick. If I'd attempted to make a radio call, that could have caused an accident. But something tells me that's not exactly what you were asking for. :D $\endgroup$ Commented Feb 18, 2020 at 21:52

6 Answers 6


First I have to admit that I have no idea what FAA regulations said about passengers taking part in controlling an airplane in 1989, but I guess it was at least frowned upon, even if the passenger happened to be a pilot.

The United Airlines flight 232 made a stunningly successful "landing" at Sioux City airport, much due to the fact that they took an extra crew member from the cabin to control the throttles of their severely disabled aircraft.

Unfortunately 112 people lost their lives as the plane cartwheeled after a very rough touchdown, and one might ask what is so successful about that... well: 184 people survived a situation that was in subsequent simulations deemed to be impossible to manage.

The throttle operator was a UA training check airman Dennis Edward Fitch, riding as a passenger on this ill-fated flight. It has been stated in many sources again and again, that without his presence in the cockpit, the outcome would have been much worse, even the worst possible.

It just so happened, that Fitch had been practicing a similar scenario in a simulator after a fatal crash of Japan Airlines flight 123 in 1985. The reason of that crash was total loss of hydraulics, Fitch wanted to find out if an airliner could be controlled with throttles only. Luckily, it was somewhat possible.

  • 6
    $\begingroup$ “but I guess it was at least frowned upon”—no, it was lauded as a great and proper application of cockpit resource management! $\endgroup$
    – Jan Hudec
    Commented Feb 17, 2020 at 21:43
  • 9
    $\begingroup$ Yes, after the accident, of course. This case has been displayed as a fine example of CRM, which not not a standard thing in the days. What I meant was before UA 232 took place. $\endgroup$
    – Jpe61
    Commented Feb 17, 2020 at 22:00
  • 1
    $\begingroup$ CRM was introduced as a concept in -79, yet as late as in -92 FAA published a handbook where it describe "recent studies" finding CRM to be very beneficial for flight safety. In late eighties it was used for sure, but it was still developing. It still is, but now it's an "industty standard". Despite that, we still see accidents where sub par CRM plays a fatal role. $\endgroup$
    – Jpe61
    Commented Feb 18, 2020 at 6:26
  • 11
    $\begingroup$ The passengers on flight 93 did not prevent a crash and loss of all on board, but they may have prevented an even worse outcome. Passengers who were using their cell phones were breaking the rules. But that is how they learned of the other hijackings. $\endgroup$ Commented Feb 18, 2020 at 19:23
  • 1
    $\begingroup$ There are two reasons I don't consider it an answer. The passengers aren't pilots, and the question was about pilots. And they didn't prevent an accident. A crash in a field is better than burning down the Capitol, but no one on board was saved. $\endgroup$ Commented Feb 19, 2020 at 21:00

The rules say to do whatever necessary to ensure safety of flight.

91.3 Responsibility and authority of the pilot in command.
(a) The pilot in command of an aircraft is directly responsible for, and is the final authority as to, the operation of that aircraft.
(b) In an in-flight emergency requiring immediate action, the pilot in command may deviate from any rule of this part to the extent required to meet that emergency.
(c) Each pilot in command who deviates from a rule under paragraph (b) of this section shall, upon the request of the Administrator, send a written report of that deviation to the Administrator.

  • 9
    $\begingroup$ I think OP meant to ask about specific instances of 91.3 (b) uses where the plane was saved and wouldn't have been otherwise. $\endgroup$
    – Jeffrey
    Commented Feb 17, 2020 at 20:41
  • 3
    $\begingroup$ I bet a lot of pilots have anecdotal stories of violating procedures for the sake of safety. It may be harder to find a documented incident. $\endgroup$
    – Dean F.
    Commented Feb 17, 2020 at 20:59
  • 1
    $\begingroup$ Doesn't 91.3(b) imply that no action that saved the plane could have violated the rules? So the answer to the question is "No" $\endgroup$
    – Bill K
    Commented Feb 20, 2020 at 17:23
  • 1
    $\begingroup$ I think you're right @BillK . Given 91.3(b) it seems impossible to prevent an accident by bending or breaking the rules. $\endgroup$
    – Dannie
    Commented Feb 20, 2020 at 19:49

As pointed out above pilots may deviate from any regulation in the event of an emergency per §91.3(b).

One of the more common events is a civilian aircraft making an emergency diversion and landing at a nearby military airbase, such as this 777 flight which diverted to Erickson AB, Shemaya, AK.

The OP brings up the case of Cactus 1549. But what Sully and Stiles did in powering up the APU first wasn’t a violation of the rules per se, though I’m sure there are plenty of odious airline sim check pilots that cut themselves on their overstarched undershirts who would disagree with this.

This even brings up an interesting discussion I once had with a flight instructor examiner I knew who had been a Air Safety adviser to Korean Airlines. He told me he once conducted a systems discussion with a group of KAL 747 pilots in regards to understanding their systems in an emergency. The KAL pilot corps at the time apparently handled their pilots in a very regimented and militaristic fashion. When the chief pilot walked in, they all jumped to their feet and stood at attention, etc.

So this examiner guy walks up to the front of the group of the group and asks them: “Ok, so what is the emergency procedure for an engine fire on a 747-400?”

The entire group barks out the QRH entry for this contingency at the top of their lungs perfectly and in unison.

“Okay.......so what two things do you do in the event of an engine fire if you only have time to do two things?”

Dead silence. The entire room stares blankly back at him like a second grader at a spelling bee asked to spell ‘photosynthesis’.

OEM approved procedures, standardized general operating practices and legal regulations for aviation are great guideline for safe and reliable operations but they can’t always cover every contingency perfectly. A fundamental knowledge of your aircraft and it’s systems from a correlation level can come in real handy sometimes if you encounter something serious that’s not been documented. It’s something test pilots know all too well, as it can allow them to make very judicious choices on flight test plans and handle unforeseen problems once in the air.

  • 9
    $\begingroup$ The company culture you described is a very effective inhibitor of proper CRM. $\endgroup$
    – Jpe61
    Commented Feb 18, 2020 at 23:37
  • 3
    $\begingroup$ Yup. And it happens out there. $\endgroup$ Commented Feb 18, 2020 at 23:51
  • 1
    $\begingroup$ This was posted as an answer, but it does not attempt to answer the question. etc etc $\endgroup$ Commented Feb 20, 2020 at 12:45

In 2017 an MD-83 aborted takeoff above V1. The pilot was widely criticized for that, which was against a lot of rules and conventional wisdom. The NTSB report determined that aborting above V1 was the most correct thing to do in this case.


  • $\begingroup$ This is probably the best ever example of how braking the rules may sometimes save lives! I'm embarrased for not remembering this one... $\endgroup$
    – Jpe61
    Commented Feb 20, 2020 at 22:47

There have been cases with incorrectly wired controls, with the aircraft doing the opposite action than such a wrongly wired control commands. Then the narrow rules about how to use the control to achieve intended action must be completely reversed, moving the stick in the opposite direction.

Doing so does not break the wider rules that cover such malfunctions. But I imagine how difficult is to realize them happening at the first place. This is the main reason I would never like to have just a robot in the cockpit.

  • 4
    $\begingroup$ @Jpe61 Actually, it is very hard to program an algorithm to correctly reverse a control if needed, but only then. Not very easy to create a simple rule that says if things get worse then try doing less of what you are supposed to do! $\endgroup$
    – MikeB
    Commented Feb 18, 2020 at 11:23
  • 1
    $\begingroup$ But it definitely is doable, IIRC even the venerable F-16 has such algorithms. Once the algorithm is set up properly, it will (with contemporary computing power) figure out very quickly how to adapt to different reactions from airframe. $\endgroup$
    – Jpe61
    Commented Feb 18, 2020 at 12:08
  • 3
    $\begingroup$ With the judicious application of (shoot me for saying this) machine learning, the computer can and would fairly quickly figure out that applying left stick was causing the plane to roll right and begin compensating for it. Once in this "alternate mode", it would, most likely, fly the plane more correctly and more consistently than a human pilot who would have to fight years of training and muscle memory with every control input. Writing, testing and certifying such a system would be an absolute nightmare, but I believe it's possible. $\endgroup$
    – FreeMan
    Commented Feb 18, 2020 at 14:47
  • 5
    $\begingroup$ Avionics systems are programmed very simply with old-fashioned and reliable feedback control techniques. Their behavior is specified in big documents written by aerospace engineers and handed to programmers. I don't think anything flying today uses "machine learning" or thinks about what it's doing. $\endgroup$ Commented Feb 18, 2020 at 22:09
  • 1
    $\begingroup$ @FreeMan very judicious. The current popular flavour of machine learning is running the training on huge datasets in datacenter, which creates a model, which you then run on the device. (Robot plane, in this case.) Your proposed system would have something akin to model training while in "live mode". Not impossible, but not what most people use machine learning for now. $\endgroup$ Commented Feb 19, 2020 at 9:59

I will note that in maritime law there is the General Prudential Rule which states that avoiding a collision takes precedence over strict adherence to other rules and regulations.

I would have to believe that there are similar provisions in aviation rules.

  • $\begingroup$ I believe that the motor law where I learned to drive has the same clause. At least, in my driving class many years ago, I was told that the law said that avoiding a collision trumped all other rules. $\endgroup$ Commented Feb 20, 2020 at 16:07

You must log in to answer this question.

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