Alaska Airlines flight 1282 NTSB preliminary report [pdf] (opens in new tab)

Very likely that number is meaningless. I suspect this is the kind of environment that incentivises hiding non-conformances whenever possible.

For example, better quality control usually results in an increase of number of defects, at least temporarily. But that just because large portion of these defects were undetected before.

So... you are looking at a number that you have nothing to compare to that also depends on how closely the process is monitored and also depends a lot on the definition of what is non-conformance.

It is like trying to give an answer to "what is the length of Britain's coastline?" Everybody knows that you can get whatever answer you want depending on how long the ruler is.

Given that if a worker doing the work raised this problem or took initiative to resolve it, they'ld probably be punished; I completely agree.

Reminds me so much of how, once-upon-a-time, there seemed to be actual engineering management in a cooperatively adversarial relationship with business managers but not anymore. Now any sort of engineering in business seems to be completely business managed and business minded. I'm sure it's great for profits while it lasts but I haven't observed engineering becoming better and I suspect business is suffering by overextending itself too, I just don't have any solid observations.

(Well, maybe one, my brother does warehouse / logistics management and says, despite there being every reason in the world, he has never seen the accounting software and the inventory software successfully and productively linked. So, big opportunity there for a serious player but maybe not the profitable compared to the issue?)

Apparently, there are two ticketing systems (one for "history of plane," Boeing internal, and one for "day-to-day onsite work," visible by contractors and Boeing management). The work to fix the rivets was logged in the day-to-day, but management and the onsite staff managed to convince themselves that merely opening the plug to fix the vacuum-seal trim did not constitute "removing" the plug, and since there was only an entry in the history-of-plane log for removing, not opening, they didn't log it there (when the intent was "there's no entry for 'just opening' because there's no such thing as 'just opening', breaching the pressure vessel at all constitutes 'removal of plug'").

The final inspection that should have caught the error would have been triggered by the update in the history-of-plane ticketing queue.

(And as for 'how many non-conformances,' the same source claims that Spirit is one of the few subcontractors with on-site staff at the factory because their parent company delivers such consistently shoddy out-of-compliance product that they are continuously doing final warrenty-work onsite. So maybe "fire that vendor" should be on the docket too).

If management in the aerospace industry works like management in the software industry, then I guess they are pushing for results as agressively as possible without much concern about safety or anything else.

> As a result, this check job that should find minimal defects has in the past 365 calendar days recorded 392 nonconforming findings on 737 mid fuselage door installations (so both actual doors for the high density configs, and plugs like the one that blew out). That is a hideously high and very alarming number, and if our quality system on 737 was healthy, it would have stopped the line and driven the issue back to supplier after the first few instances.

Source:

https://leehamnews.com/2024/01/15/unplanned-removal-installa...

Well, the report says "During the build process, one quality notification (QN NW0002407062) was noted indicating the seal flushness was out of tolerance by 0.01 inches.

So I'd say they've had about 2,407,062 quality issues :)

The planes they worked on did not share an assembly line with the 737 but another Boeing model…

Lol, they said the door plug "departed" instead of "blew the f* off"

So this all just serves to confirm that report, and what people suspected for a while; the bolts were just missing, removed for removing the door plug for the rivet rework and never reinstalled.

There must be a middle ground here- the paradox is that Google, Apple, etc have this ability to generate user friendly software and hardware at scale. But they aren't considered "battle proven". The expensive proprietary systems that are used instead tend to be hard to use and brittle, so what's the middle ground?

I mean, there is already a ton of documentation and process surrounding the construction of an airplane. Adding more process doesn't safety make. Having a safety culture without the fear of retaliation, on the other hand, makes a world of difference.

https://www.airlinepilotforums.com/safety/146074-boeing-inte...

The process was there so that the people would know there was work being done on the doors despite not being there for it. If you see an unfinished work from a previous shift, it does not mean you can start messing with it - there might be context you do not know.

Which is why such things are supposed to be noted in appropriate ways. Similarly why aviation has so many procedures everywhere - because we know and understand that sometimes you miss things. For any human reason, not just mismanagement. The process is a way to have reliable place to double check with.

This is different from over reliance on automation, which is arguably less of an issue of automation itself (it's just more visible in such areas) as much as getting out of training because you do not encounter certain things so often. 96 people died because in a stream of many deviations, among other things, the crew never trained how to do IFR landing without ILS, autopilot or no autopilot.

The process is the part that says "yeah, I haven't done this in a long time, I need to train, here is documentation that provides we need to do it and can't delay".

Similarly CMES is supposed to track "work was done on this part of the ticket, now different work needs to be done, do not assume it will be done by other teams"

https://www.federalregister.gov/documents/2023/12/04/2023-26...

  > Given how cheap flash is these days

Flash data retention at high temps is TERRIBLE (and gets worse for MLC/TLC/etc), see any flash datasheet. It is NOT nearly as simple a problem as you might think.

Yes, it is a solvable problem, but please do not dismiss it so outright as "trivial"

Irritatingly, they didn't even pick the top-of-the-line machine from the vendor at that time. They picked a middling one. And then put an LTS OS version on it that didn't fully support the motherboard chipset. I spent way, way too much time an energy trying to get the software to run on the sort of timescales necessary. It took me months to get anyone to let me talk to the vendor in order to sort out the fact that the storage was being run in legacy PATA mode, reducing our IO throughput by an order of magnitude and the application throughput by about a third.

Ten minutes on the phone and I got them to agree to give us a patch that aliased the chipset to one it was backward compatible with, that was actually supported by the OS. But they really wanted us to take the never version of the OS that didn't have this problem.

That's not even the most hard-ware crippled I'd ever been, but it was top three.

Also how do you cause a rapid decompression without a gun of some kind?

OXYGEN MASKS - DON

OXYGEN REGULATORS - Set to 100%

CREW COMMUNICATIONS - ESTABLISH

PRESSURIZATION MODE SELECTOR - MAN AC/MAN

OUTFLOW VALVE SWITCH - CLOSE

Hold in CLOSE until outflow Valve indicates fully closed

If Pressurization is Not Controllable

PASSENGER SIGNS - ON

PASSENGER OXYGEN SWITCH - ON

EMERGENCY DESCENT - ANNOUNCE

The pilot flying will advise the cabin crew, on PA system, of impending rapid descent. The pilot monitoring will advise ATC and obtain area altimeter setting.

PASSENGERS SIGN - ON

DESCENT - INITIATE

I do giggle a little at the thought of a door flying off, the air rushing out of the cabin, and the pilots responding by switching the seatbelt light on.

The plane was only at 16,000 feet when it lost its door and according to [2] you've got 20-30 minutes of 'useful consciousness' at such an altitude, even without your oxygen mask on. So there was no need for an abrupt dive.

[1] https://www.theairlinepilots.com/forumarchive/b737/b737memor... [2] https://skybrary.aero/articles/time-useful-consciousness

Edit: Re-reading, it was more like 16K feet when it popped, 10K is what ATC assigned them when requested. Still low enough not to be a critical emergency. Some people absolutely will get altitude sickness at that level, but it's likely to be mild. Many people climb mountains much taller.

Better to climb for a bit more as you get your oxygen mask on than to try to descend immediately and make some problem worse.

We know it was a door plug blowing out, but in the past it has been entire major sections of the airframe ripping off, in which case sudden extra stresses are not what you want.

Pilots can't look at the rear view mirror, and see the whole plane. Accident reports on engine malfunctions routinely mention that someone had to check their appearance through the passenger window, and relay that to the pilots. In case of a blast so severe that the door flies off, it is safer to assume the worst. Say, that part of the plane disintegrated because of sudden collision. In such conditions, indications on what works and what doesn't are probably really messy and unreliable, and there can be not enough means to control the plane properly. Lower the nose too much, and you might not be able to pull it up any more.

Pilots probably did checklists with one eye on the instruments to check that they were not losing speed, that angles were correct, autopilot inputs resulted in stable flying, and so on, and deduced that everything still worked. By that time, they were probably informed that the plane was seemingly intact, although with a hole in its side.

Not really. The cockpit door was blown open, and the pilot's headsets were blown off. It was a pretty chaotic event, and when you are flying an airplane, you definitely don't want to figuratively "jerk the wheel" - you remain calm and start running checklists.

Looking at the track, they descend to 10000' until they start their downwind to base turn. Once they start that turn, they get a lower altitude (looks like 7000') until they are established on final and can fly an approach.

Or just throwing the rebar in at random maybe? [0]

[0] https://www.gr-us.com/%E2%80%9Chorror-at-the-harmon%E2%80%9D...

There’s also bandwidth and satellite coverage not being magic of course.

That said, there’s a standard and reliable 25-hour flight voice recorder that solves this problem. But it’s only used outside the US. That’s a regulatory inertia situation and I suspect this incident will speed changes in this area.

However, finally, and particularly in relation to your proposal of streaming cockpit voice recordings to some cloud server. There is some resistance to this (and to longer recordings in general) from air crew on privacy grounds. The privacy issue is less about how much personal info is revealed in a crash situation and more about how easy it would be for a bad actor in management —or whatever operations group runs the audio storage—to listen in on conversations. And you can be sure this would happen if something like your system were implemented without the appropriate regulatory controls (and tbh even with them it would probably still happen).