Skip to content


Two major airlines have discovered loose bolts on multiple Boeing 737 MAX 9 aircraft, raising new concerns over the jet model.

US regulators grounded some 171 Max 9s on Saturday after a panel blew off an Alaska Airlines plane mid-flight from Portland, Oregon, forcing pilots to scramble to land the plane safely.

United Airlines – the biggest operator of the model, with 79 Max 9s in its fleet – announced on Monday night that its assessments of the jets had revealed loose bolts in panels that are similar to the part that came off during the Alaska Airlines flight.

And the claim at least is that this was a problem stemming from the factory…..

36 thoughts on “Ooops”

  1. In Walter Isaacson’s biography of Elon Musk, Boeing is shown in a poor light in several respects, one of which is that an executive there said that they would prefer people who stick around than those who are the best engineers. And so lost a stellar engineer to SpaceX.

  2. Bloke in the Fourth Reich

    We at Boeing believe it is very important that aircraft are assembled in diverse ways, and that we are consciously inclusive of the different ways in which aircraft express themselves while operating.

    We acknowledge the need to do more to ensure equality, and are therefore instituting a new committee to look at how we ensure all of our aircraft have equal opporunities to deconstruct themselves mid-flight.

    We commit ourselves to bringing equality of these diverse experiences to a wider range of passengers in the near future, and ensuring that all passengers can be included in the experience of an aircraft overriding white male pilot trim control inputs, causing the aircraft to fly into the ground at 450 knots.

  3. As usual my mind contemplates my hero Crassus. And the way he used to line the roads with the corpses of the crucified.

    But of course I’ve just flown back from Perth. And I’m thinking of flying to Sydney in March.

  4. Bloke in the Fourth Reich

    Fortunately the ancient, cramped, 737 is extremely rare in European skies. In any variant. At least in the service of any airline you would consider flying with.

  5. No need for Loctite – as designed, the nuts on the door bolts were supposed to be of the castellated kind and properly secured by a cotter pin/split pin. They have now retrieved the ex-door from someone’s garden, so we shall soon see if that was the case.

  6. Boeing has had issues with quality control for a decade: see the USAF tanker program for examples. It’s also been ranting about diversity for about the same time, as well as moving it’s HQ from Seattle where they do the majority of their manufacturing, to Washington DC, where they manufacture something else. Strange!

  7. Look, there’s some part of this story that isn’t making sense.

    Semi-permanent door plugs like the one in question are made to be larger than the hole in which they are fitted, and to be installed from inside – they are held in place and sealed by the pressure differential between the inside and the outside of the aircraft in flight. The securing bolts – there are only 4 in this case – are only there to stop the plug from literally falling out when the aircraft is not pressurised. The securing bolts are not designed or arranged to carry any of the loads on the door plug when the aircraft is pressurised, and, in fact, once the aircraft is pressurised, the securing bolts could be removed and the plug would be completely-secure. The bolts are like the hinges on a safe door – once the door is closed and bolted, they carry no loads. There’s something quite different that happened here, and it has nothing to do with the securing bolts. In a sense, it may be worse in that it suggests a failure of the structure of the aircraft or the design of the door plug.



  8. Boing merged with McDonnell Douglas, and the latter one out in engineering and manufacturing competence.
    They have a long history of doors blowing off.
    Nothing new here.

    Though interesting that the door fault seems to be “we didn’t bother to do up the bolts” just weeks after the 737 MAX problem with the rudder “we didn’t bother to do up the bolts”. I’m beginning to spot a common factor…

    Do Boing need a supply of spanners?

  9. Just as a completion – after reviewing the specific design of the plug door in question (which is designed just as I described) the only way the door plug could simply fall out (which is what seems to have occurred) is if all 4 securing bolts were entirely missing. The door plug is removed by lifting it upwards, against its own weight – the securing bolts prevent this movement. Chatter about these bolts being ‘loose’ is meaningless – the bolt functions only as a stop pin, and as long as it is present, it doesn’t matter if the securing nut is ‘loose’, as the nut itself is secured with a cotter pin and can’t fall off.

    The only scenario that seems to fit the facts is that all 4 securing bolts were missing entirely, the plug became misplaced upwards at some point (it would have had to move upwards by about 6 inches) to the point where it was no longer engaged with the outline of the plug opening, and then it blew out when the aircraft was pressurised in flight. Not sure how an aircraft can leave the ramp with the door plug so-obviously misplaced, but I bet there’s CCTV. Any road up, it’s an appalling black eye for someone.



  10. The BBC News channel played a short video of NTSB officials examining the plane from inside & out, which clearly showed a number of undamaged securing lugs round the edge of the fuselage opening – no bits of broken bolts or anything. There had to be more than 4 in total, judging by those I could see. Given that the door has now been found (in one piece), this suggests that the “plug” could only be attached from the outside!

  11. It’s also been revealed the aircraft had been suffering “Pressurisation” issues for some time, and Alaska Airlines had banned it from flying long over-water sectors. In my mind this further backs up the “Plug fitted from the outside” theory…

  12. Not too long ago United made a big announcement that their pilots would meet diversity quotas (competency be damned). Wonder if other jobs have been compromised?

  13. @ Dave Ward – the large number of securing lugs you see around the periphery of the door opening are supports only – they are seats for matching lugs around the periphery of the door plug. There are no securing bolts through these lugs. When the plug is slid into place, the lugs on the door plug bear on the inside faces of the lugs on the door frame, creating the ‘plug’ function whereby the plug cannot fit through the door opening and is forced against it by differential pressure. The sliding/guiding action that engages all these lugs is performed by 4x guide roller tracks, one at each corner of the door plug, which guide the plug into place. Once in place, the 4 securing bolts of which so much has been made are installed across the roller tracks, to prevent the plug from sliding back out, and secured using castellated nuts and split cotter pins. These bolts are not required to be axially “tight” for any load bearing function, but only to prevent rattling and wear.

    The plug is indeed offered up to the aircraft for initial assembly from the outside, but is initially misplaced vertically to allow the securing lugs on the door plug and the door frame to pass each other laterally. The plug is then slid vertically to engage the outside faces of the lugs on the plug with the inside faces of the lugs on the frame. The lugs have matching conical faces and conical recesses to align them horizontally and vertically. So in that sense, the plug is inserted from the inside of the aircraft, and, when correctly aligned, is too large to pass through the door frame.



  14. Additional reporting this morning says that the 4x roller-track securing bolts described on the incident aircraft are missing, although investigators are not speculating about how they came to be missing.

    Multiple reports of ‘loose’ bolts in door assemblies on other aircraft all refer to various mounting bolts in other parts of the assembly, including in the convenience hinges which allow the plug door to be partly-opened for maintenance, in spring-strut assemblies which assist in lifting/sliding the door into place, and in ground-bonding straps. None of the ‘loose’ bolts reported have any function in actually holding the door in place in service.



  15. And this, dear children, is why we use LockTite™ and a torque wrench in places….

    Just make sure you use the correct one, for the sake of all that is holy (and Maintenance staff’s sanity)

    Nothing worse that coming to undo a bolt that’s meant to be held with red or blue loctite to prevent it rattling loose and finding that some [redacted due to language only describable as a crime against humanity] has used green Loctite.

  16. Actually, this is exactly why we DON’T use ‘Locktite” on serviceable or separable bolted joints in most parts of aircraft structures – precisely because you can’t check and verify correct torque of the bolted fastener after the threadlocker has cured. If, for example, you spun a nut onto a fixed stud no more than finger-tight, but applied a ‘Locktite’-type threadlocker, once the threadlocker has cured, you cannot verify whether the nut is correctly tightened in the assembly to the required specification. At this point, if you apply a torque-measuring tool to the nut, you are only measuring the torque between the nut and the stud, not between the nut and the bolted face, and a finger-tight nut might well show a torque measurement that meets specification.

    Add to that the complete inability to control the ‘Locktite’ process in the field, and you have a result with just-too-many uncertainties for critical applications. That’s why aircraft are assembled with prevailing-torque locknuts, safety lockwire, tab washers, cotter pins – all highly-reliable, positive locking methods that can be accurately checked after assembly and are much-less susceptible to variation and human error.

    Threadlockers work great in controlled and specified environments, with rigorous process controls, and in applications which are not designed for regular or unplanned service disassembly. They are used extensively in areas which are serviced at the unit level, such as engines and actuators.

    Mind you, I suppose no locking method can do much good if the bolts are simply omitted, which now seems to be more-likely-than-not what occurred here.



  17. You can check the tightness of threadlocked bolts, if the bolt is going straight into the item to be bolted, not a through hole with nut.

    No doubt those fasteners are used in aircraft, but the vast majority of fixings are rivets as far as I can tell. Though my experience is limited to one area of the plane.

    Threadlockers are great anywhere, but really are most useful where vibration is an issue. You don’t want a tightened bolt coming loose over time and having the machine shake itself to bits. The trick is using the correct type. Blue loctite is best all round, with a good combination of retaining strength but not being hard to break if needed without heat treatment. Purple is ok, if weaker. Green is pretty much permanent without a blowtorch. As with anything, select for use correctly.
    And not slathering it on is important, a small drop is enough.

  18. In reply to @llamas – from a person I know at Boeing who has relevant knowledge

    The plug is not “Larger than the hole”. It does have a dozen lugs that line up with a frame the plug is in, but it IS designed to be opened for maintenance. In fact, the hinges at the bottom are spring loaded the inch or two (not 6″) the door needs to travel upward to disengage them. But the top also rides in a guide track. The door plug did not rise all the way out of that track, in fact, the track cracked under the pressure load. He IS right in that all of the bolts would have to be out to allow that to happen. The plug is normally installed by Spirit Aerosystems (in Witchita) but apparently they sometimes are opened for bringing materials inside the plane. So it may be difficult to assign blame. If it were me, I’d have those suckers safety wired, rather than just torqued.

    Also the exact same design has been in use on the 737-900 ER for years.

  19. @ FrancisT – the description ‘larger than the hole’ is perhaps simplistic, but describes how the door works. When in the service position, the lugs on the plug door engage on the inside of the frame, and the door cannot fall out. The plug door must move inwards before it can be moved upwards to disengage it from the opening. Unless the entire structure fails completely, the door in the service position cannot move outwards through the opening, because (put in simple terms) it’s larger than the hole it would have to pass through.

    Your friend has not described the door function correctly. The roller tracks are on the plug door, not the frame, the door frame has the roller pins that slide in the tracks.

    The upward motion of the plug door has two stages. To partly open the door for maintenance (using the convenience hinges and limiting cables) takes an upward motion of about 2 inches. To completely remove the door, the total vertical movement is more like the 6 inches I described, in order to completely clear the hinges, roller tracks, sliders and locking lugs. See the image I linked. Since the door obviously departed the aircraft entirely, the upward motion was more than the 2 inches required to open it on its hinges into the maintenance position, which is about 15 degrees of rotation on the hinges from the closed position. The latest published images show that the convenience hinges and lower sliders are still attached to the aircraft, so the amount of vertical motion required to release the door may be somewhere between the two.

    At least one upper door track has been reported as cracked, but no cause for the crack has yet been assigned. Bearing in mind that the recovered door fell 16,000 feet, it might be too early to say. It might well be (as has been suggested) pre-existing damage from an incorrect installation.

    The 4x locking bolts are supposed to be secured using castellated nuts and split cotter pins passing through radial holes in the bolts, which are functionally equivalent to safety lockwire.

    Regarding checking the tightness of threadlocked bolts going ‘straight into the item’, no, you cannot accurately confirm tightening torque once the threadlocker has cured. Again, consider a bolt inserted no more than finger-tight, with threadlocker of a suitable grade. Once the threadlocker has cured, any torque measurement will reflect only the torque between the bolt and the fixed thread, but will tell you nothing about the torque between the bolt and the mating part – so you can’t know the clamp force. Threadlocker only secures the bolt at whatever torque you had applied to it when the threadlocker cured, which could be anything, and you can’t measure it accurately, and you can’t re-tighten the bolt to the required torque because now the threadlocker is adding its own torque to the system. That’s what I mean by carefully-controlled processes when using threadlockers – torques must be set and checked before the threadlocker cures, and cannot be checked or adjusted afterwards. Too many people make the assumption that threadlocker somehow tightens a bolted joint – it doesn’t. It just locks the joint at whatever torque it was at when the threadlocker went off, and it doesn’t care if the torque was right or wrong.



  20. Bloke in the Fourth Reich

    I am confused about what actually blew out.

    Was it a normal panel? Or an emergency exit? Or something that on older 737 fuselages is an emergency exit but in the max is “disused”, and just built in in a similar way (without the handles, slides, etc) as a corner-cutting measure?

  21. It’s almost inconceivable but there’s rather a good graphic in today’s ‘Daily Mail’ that accurately shows the sequence of assembly and securing of the door plug in question

    – although the description of step 1 is incorrect. The vertical movement arrestor bolt shown being assembled in step 1 is actually assembled in step 3, at the same time as the loocking bolts in the upper tracks. This diagram also shows the two stages of vertical motion of the door – about 2 inches upwards (the DM says 40mm) to disengage the stop pads and sliders and allow the door to pivot outwards on the hinges, and the much-larger movement (about 6 inches) required to remove the door from its hinges entirely.

    The door in question is not a normal entry/exit door. It is an emergency-exit door location that is only required to be functional as an exit by FAA regulations for certain seating configurations of the 737 (most models, not just the 737 Max) based on number of doors vs number of seats. This aircraft does not require it to be a functional exit, but could easily be reconfigured to have enough seats that it would be required, so the opening is there in all models. When not required, the opening is plugged with a removable panel (the “plug door”) which operates in exactly the same manner as a functional exit would operate, except that the mechanism is secured in the service position by the 4 locking bolts of which there has been so much talk. When properly secured, the plug door cannot be removed from either inside or outside, and is trimmed over on the inside. For convenience in maintenance, the “plug door” is fitted with a set of hinges and cable limiters, which allow it to be partly-opened to allow ventilation – to open it to the maintenance position, the 4x locking bolts, nuts and cotters must be removed, and replaced once the door is placed back in the service condition. The hinges incorporate helper springs to support some of the weight of the door and help it to move upwards into the open condition(s) – these make the opening and closing of the plug door a one-person operation, as the weight of the door without them would make this a 2-person operation.

    If required to be a functional exit, the “plug door” is removed and replaced with an actual exit door, which is secured into the service position by the exact same system of locking lugs and tracks, but which has a cam-handle mechanism on the inside which allows it to be opened and jettisoned from the aircraft with a single motion. This mechanism is interlocked to prevent accidental operation (“doors to manual”). The exit door does not use the convenience hinges, helper springs and cable limiters, which are removed in this configuration.



  22. There are different possible layouts. One of which requires an extra emergency exit, others which don’t. So, if you don’t need the extra, there’s still that hole in the fuselage where it would go. So, you don’t need it, plug it instead. One of them.

  23. @ llamas – Thanks to your explanations and link to the DM I now have a much better understanding of the situation. But given that under normal circumstances it should be impossible for the plug to come loose – due to pressure difference holding the stop pads in contact – it strikes me that to “blow out” it would had to have moved downwards before take-off when no such pressure difference applied. In which case I imagine the noise of cabin air escaping would have been apparent.

    The pictures of the recovered plug don’t appear to show any of the stop pads (7 each side, according to the DM article), so is also possible these might have failed with the plug in the correct position? There is obviously considerable force acting on these when at cruising altitude, and one would hope that a lot of redundancy is built into the design, but if a manufacturing defect existed in one (or) more of them, there is always the risk of a progressive failure.

  24. There was a photo of the recovered door, but from the outside position, which appeared unblemished. They carefully didn’t show the reverse side.
    Anyone know how the exterior panel sheet is fixed to the lugs, bolts & framing framing? Epoxy?

    Anyway, Matt scores a perfect 10 in today’s cartoon.

  25. @Dave Ward – I guess it it possible that the 12x securing lugs and 4x roller/slider tracks are what failed and caused the door to “blow out”, but I think if there was any evidence of that, the NTSB would have had something to say about it. Since the identical door design is used on many models of the 737, over many years, an unexpected catastrophic failure of the design in a brand-new aircaft would have led them, I think, to ground all 737’s, not just this one model. In the second of two NTSB press conferences I have seen, they again specifically referred to the absent locking bolts and the question of whether they were present when the door departed the aircraft. I think they were telegraphing their punches – I think they’re pretty-sure that the bolts were missing, they just want to complete i) forensics on the door and frame and ii) review of the service records. They want to be able to say ‘the bolts were left out, and here’s who did it.’

    As you say, it’s not really possible (absent a gross structural failure) for a door that’s in the correct service position to “blow out” once there is significant pressure differential between the inside and outside of the aircraft. At operating altitude, the force on the entire door, pressing it against the securing lugs in the frame, could be up to 8000 pounds or 4 tons – you couldn’t move it if you tried. However – if the 4x securing bolts are missing, and the aircraft is not pressurised, then the force required to lift the door up and out of the service position and into a position where the securing lugs are no longer engaged is relatively slight – it amounts to the effective weight of the door, plus friction in the securing lugs, roller tracks and so forth. I could well see where a rough taxi or take-off run could jar the door upwards to where the securing lugs are no longer engaged. In this condition, a relatively-minor pressure differential could easily blow the door out.

    The airplane pressurization is at around 12 psia, which is equivalent to an altitude of about 7,000 feet. In climbing to that altitude, there would be little or no pressure differential across the door, the internal pressure at take-off is slowly bled off the match the outside pressure. But beyond that altitude, as the internal pressure is stablized at about 12 psi, the outside pressure continues to fall. Approaching 16,000 feet, the outside pressure is now about 8 psia, which gives a 4 psia pressure differential across the door. This equates to a force on the door of around 6000 pounds, or 3 tons. I think that might be enough to blow out a door that is no longer aligned with its securing lugs.

    The main structure of the plug door is made up of formed details of aluminum-alloy sheet material, rivetted and bonded together. The various forged and machined details of the roller tracks and securing lugs are attached using a variety of nuts and bolts, nut plates, threaded inserts and the like. I’m not familiar with this specific door assembly but I suspect the outer skin is primarily bonded to the inner structure.



  26. Bloke in the Fourth Reich

    Who knew we had an expert in the construction of aircraft doors here. This site really is one of the seven wonders of the internet.

    And I’ve no idea what the other six are.

  27. Not an expert by any means, but it just so happens that I worked on a programme which required interface with the emergency exits on a Boeing aircraft, not the 737 but the design is very similar.

    I watched the third NTSB press conference, which provided a mass of detail extracted from the aircraft recorders. After take-off, the aircaft passed through a band of noticeable turbulence as it turned south-east between 8,000 and 12,000 feet. The decompression occurred as the aircraft was climbing through 13,000 feet, at which point the internal and external air pressures were about 14 psia and 9 psia, respectively. The aircraft continued climbing to just-over 16,000 feet before executing an emergency descent to 10,000 feet under ATC.

    It was reported that both upper roller tracks on the door plug have fractures consistent with the door being forced outwards. They referred again to the missing securing bolts (as in, they haven’t found them yet) and the possibility that they were not present when the door departed the aircraft.

    I think it’s now pretty-much a certainty. The securing bolts were missing when the plane took off – if any one of them had been present, they would have left some signs on the roller tracks or hinge sliders when the door was wrenched free. The door was lost in two stages. During taxi, take-off or climb, the door plug was displaced upwards enough to disengage the 12x securing lugs, but not enough to disengage the upper door frame rollers from the roller tracks in the door. This is consistent with the operation of the helper springs, which assist the door upwards quite strongly when it is in the fully-down service position, but press it upwards less and less as the door rises upwards.The only thing now holding the door in place was the hinge sliders and some measure of engagement between the door frame rollers and the roller tracks in the door – neither of which is designed or capable of carrying even a tiny part of the loads of pressurization. As the pressure differential across the door increased as the aircaft climbed, they eventually let go and the whole door left the aircraft. It’s quite fortunate that it did not strike the horizontal stabilizer on the way by.

    Explicitly not a failure of design, but a basic, fundamental human error – leaving parts out, and then not checking the work when it was done. See Shak Lin, Constantine, et al.



  28. @llamas
    Gosh, your book reference took me back. Wrote a lot of good sense, did Mr Norway.
    Not least on the importance of dangerous hobbies to engineers.

    I got a copy of that book for everyone in my team when I ran a fault-tolerant system development stream, and also quoted the advice given to me when signing-off an aircraft as airworthy “Remember, you may be required to justify your decision at an inquest”.
    It concentrates the mind wonderfully, and makes people think. They are still susceptible to management bullying though.

  29. They referred again to the missing securing bolts (as in, they haven’t found them yet) and the possibility that they were not present when the door departed the aircraft

    If the bolts were present, but not secured, wouldn’t they have been sucked out along with all the other lose items?

  30. @ Pcar – well, anything’s possible.

    The securing bolts in question do not perform the task we often associate with nuts and bolts, namely, to clamp parts together. Instead, they act as pins. The body of the upper securing bolts pass cross the upper roller tracks to capture rollers in the closed ends of the tracks. The body of the lower securing bolts passes through mating holes in two telescoping cylindrical parts (part of the lower hinge assemblies) and locks the telescoping parts in a specific axial position. The diagram at the end of this report is quite good;

    The nuts on the ends of the bolts perform no compression function, and (within reason) it doesn’t matter how tight or loose they are – all they do is prevent the bolts from falling out. The nuts are prevented from unscrewing from the ends of the bolts by a split cotter pin that passes through both nut and bolt and is bent over in a precribed manner to secure it in place. So you see that it doesn’t matter whether these bolts are ‘tight’ or not, or whether they are ‘secured’ or not – but they do have to be there, assembled as required. I suppose it’s possible they were present, but simply laying on the floor, but in that case, they might as well have been in a drawer under my workbench in Michigan for all the bearing they would have on the incident.

    All the breathless reporting you have seen about ‘loose’ bolts discovered in other aircraft refers to other bolts found to be loose in other, adjacent parts of the plug door assembly – not these securing bolts. Not a good thing, of course, but none of these ‘loose’ bolts have any bearing on the door blowing out. It’s early in the morning here, and I haven’t scanned the news, but I have not seen any report to date that any other of the specific securing bolts in question have been found missing or ‘loose’.



Leave a Reply

Your email address will not be published. Required fields are marked *