Around 5:15 PM Tuesday afternoon, Nov. 4, 2025, an MD-11 cargo plane operated by UPS began rolling down the runway of Muhammed Ali International Airport in Louisville, Kentucky. Flight 2976 was on a routine flight bound for Hawaii, but as the pilot rotated the plane to get it into the air, the left engine separated from the fuselage. The resulting crash and fire killed all three crew members and twelve people on the ground.
Since then, National Transportation Safety Board (NTSB) personnel have been investigating the crash. After two days of hearings last May, the NTSB released a set of documents last Wednesday that reveal new details about why the engine fell off.
The structure that connects the engine to the underside of the wing is called a pylon, an elongated assembly that attaches to the wing at two points, one near the structure's center and one on its rear end. The rear attachment point hangs from a wing structure called a clevis. The thing that actually connects the wing's clevis to the pylon is an assembly called a spherical bearing. It's rather like a ball-and-socket joint in the human hip, and allows restricted angular movements that the dynamics of flight make necessary. But it also carries a large portion of both the static weight of the engine and the dynamic loads when the plane takes off.
Surrounding the ball is the socket part called a race. The race is held to the pylon with a couple of brackets called lugs, a forward and an aft lug. The lugs are shaped like my grandmother's old mantle clock: a circular middle part tapering to a flat surface on either side at the bottom, where the lugs attach to the pylon.
Back in 2011, Boeing discovered that metal fatigue exacerbated by a "design recess groove" in the spherical bearing had led to the bearing race cracking in two, like the two layers of a layer cake coming apart. This had happened on several airplanes already, and while Boeing issued a service letter noting that this was a problem and telling aircraft maintenance personnel how to check for it, there was nothing in the letter implying that such a failure could cause the engine to separate from the wing.
In the May hearings, both Boeing and the FAA admitted that they didn't realize the seriousness of the race cracking, which put extra stress on the lugs that they were not designed to carry. When the lugs surrounding the bearing let go, probably during the maximum stress of takeoff that would exert extreme downward tension on the mount, the engine was left hanging by only the center mount, which also eventually gave way, possibly because of the fire caused by ruptured fuel lines. Once the engine left the plane, the situation was hopeless.
Why wasn't the MD-11 inspected for this problem in time to catch it and fix it? An AP report says that several years earlier, Boeing asked the FAA for permission to relax a schedule of major required inspections from every 19,900 cycles of takeoffs and landings to every 29,260 cycles. The MD-11 that crashed had 21,043 cycles, so it would have been inspected under the old more rigorous schedule. Inspecting the spherical bearing is not an easy task, and usually the engine has to be removed to gain access to it. So the problem couldn't have been caught with more frequent minor inspections.
And even if it had, neither Boeing nor the FAA believed the bearing-race failure would lead to lug failure and loss of the engine. Maintenance on the aircraft was performed by STE San Antonio Aerospace, which followed the maintenance procedures required by the FAA-approved schedule. Unfortunately, that schedule wasn't sufficient to catch the problem in time.
Any time a fatal accident occurs and engineering is involved, the public wants to know the cause. And in rare cases, such a single-point failure that no one expected can be found. But in most cases, the chain of events leading up to the major accident is complicated. And usually precursors to it can be found: warning signs that, while not causing serious harm in themselves, can serve to alert those who are paying attention to a potential major problem that needs to be addressed before it gets out of hand.
That is what seems to have happened in the case of UPS Flight 2976. A faulty design including the recess groove (1) led to bearing-race failures (2) which were discovered but not taken seriously enough by either Boeing (3) or the FAA (4). After the service note alerted maintenance organizations to the problem, three more planes were found with cracked bearings, although none of them crashed. This fact probably encouraged a degree of complacency which in retrospect was unwarranted.
Following the crash, all MD-11s were grounded until they could have their spherical bearings inspected. They are now required to be checked after every 4,000 cycles of takeoffs and landings, but UPS has announced that they are retiring their fleet of MD-11s early in any case.
The NTSB is not finished with its investigation, which might take until the end of the year to complete. It must be a rather thankless task to pick through the huge pile of wreckage reassembled in some hangar in order to figure out what happened. But such jobs are necessary as long as lives depend on the smooth functioning of the incredibly complicated systems that we rely on for transportation and communication.
Engineering requires tradeoffs: tradeoffs between economy and increased frequency of maintenance schedules, tradeoffs about inexpensive designs that might not last as long as more costly ones, and on ad infinitum. The responsibility for missing the critical inspections and repairs that might have averted the crash last November appears to be spread among the four parties of Boeing, the FAA, UPS, and STE San Antonio Aerospace. Any one of these could have gone the extra distance of doing extra inspections to avert the crash, but no one seemed to know how serious the bearing problem was.
Unfortunately, experience is sometimes the best teacher in engineering as in other fields. And the results of this investigation will probably lead to increased inspections of the kinds of engine mounts that failed. And that will be a small victory in the never-ending battle to keep minor problems from turning into major ones.
Sources: I referred to an article in the Fresno Business Journal at https://thebusinessjournal.com/ups-cargo-plane-crash-investigation-maintenance-failures/, a more technical article at
https://theaircurrent.com/feed/dispatches/ups-boeing-md-11-ntsb-update-engine-mount/, and the NTSB preliminary report at https://www.ntsb.gov/investigations/Documents/DCA26MA024%20Investigative%20Update.pdf, as well as the Wikipedia article "UPS Airlines Flight 2976."
