Close

Flagship Disasters

The 200-Man-Hour Shortcut — Flight 191's Engine Change

Concept visualization for The 200-Man-Hour Shortcut — Flight 191's Engine Change

Taking the engine and pylon off as one piece on a forklift saved about 200 man-hours per airplane. It also cracked a bulkhead flange nobody saw. Eight weeks later, at rotation out of O'Hare, the engine tore off the wing — 273 people died.

Warning condition

29–30 March 1979, engine-and-pylon change, American Airlines maintenance facility, Tulsa

56 DAYS

If uncorrected

25 May 1979, takeoff from runway 32R, Chicago-O'Hare

Read the official report: NTSB NTSB-AAR-79-17
ATA 54engine pylon aft bulkheadNormsPressureLack Of CommunicationLack Of Awareness
What happened

During a C check at Tulsa on 29–30 March 1979, N110AA's No. 1 engine and pylon came off the wing to replace the pylon's spherical bearings per McDonnell Douglas Service Bulletins 54-48 and 54-59 — but not by the McDonnell Douglas sequence. The bulletins assumed the engine comes off the pylon first, then the pylon off the wing; American's engineering change order took engine and pylon down as a single unit cradled on a forklift, cutting the disconnects from 79 to 27 and saving about 200 man-hours per airplane. A McDonnell Douglas field representative reported that Douglas "would not encourage this procedure due to the element of risk" in remating the combined unit to the wing — and the manufacturer had no authority to forbid it. The clearances at the attach points were minimal: the Board found a minor mistake by the forklift operator, or imperceptible fork movement from a hydraulic pressure leak, could bring the aft bulkhead against the wing clevis with fracture-producing force. The work on N110AA spanned a shift change with the assembly hanging mid-procedure, and day-shift mechanics found the aft bulkhead's upper lug in contact with the clevis attach bolts. Out of Tulsa the bulkhead's upper flange carried a 10-inch overload crack, closed up and invisible behind the mated fitting. It held for eight weeks. At rotation on 25 May the flange failed, and the engine and pylon tore up and over the left wing.

Why it matters

The Board's probable cause names the separation as "resulting from maintenance-induced damage" through "improper maintenance procedures," and the contributing factors read like an industry indictment: pylon attach points vulnerable to maintenance damage, FAA surveillance and reporting systems that never detected the carrier-devised procedure, and operators, manufacturer and FAA failing to share what they knew — Continental Airlines had fractured two aft bulkhead flanges the same way, in December 1978 and February 1979, repaired them as routine events, and the knowledge changed nothing at American (NTSB AAR-79-17 Synopsis, pp. 19–20). The mandate this card carries: a manufacturer's disassembly sequence is engineering, an "equivalent" in-house method needs engineering proof, and maintenance damage anywhere in a fleet is intelligence owed to everyone.

Control sequence
  1. Remove major assemblies in the manufacturer's sequence — engine off the pylon, then pylon off the wing. A combined lift is a new procedure requiring engineering evaluation and manufacturer consultation, not just a man-hours case.
  2. Prove the lifting equipment can hold position to the tolerance the fitting demands. A forklift's forks can creep imperceptibly as hydraulic pressure bleeds — precision rigging is part of the procedure, not an accessory.
  3. Never leave a heavy assembly partially attached across a shift change or engine-off pause; if the work must stop, the load path must be made safe first.
  4. Any contact or forced misalignment during mating is a stop-and-inspect event for the fitting — before close-up hides the evidence.
  5. Report structural damage found in maintenance and share it: two identical flange fractures at another carrier were the warning that never arrived.
How often

The pattern was already in the fleet before the accident: Continental fractured aft bulkhead flanges twice with the same forklift method — December 1978 and February 1979 — and post-accident inspections brought the count to nine DC-10s with fractured flanges, plus a United DC-10 with a cracked upper spar web and 26 damaged fasteners (NTSB AAR-79-17 p. 20).

Consequences

The No. 1 engine and pylon separated at takeoff rotation and went up and over the wing, taking about 3 feet of the left wing's leading edge; the severed hydraulic lines let the outboard slats retract, raising the left wing's stall speed to 159 knots. Climbing at the commanded attitude, the crew — with the stall warning and slat disagreement indication lost with the engine — decelerated toward their prescribed climb speed, 6 knots below that stall speed. The left wing stalled, and Flight 191 rolled past vertical and struck an open field about 4,600 feet northwest of the runway's departure end. All 271 on board were killed, along with 2 people on the ground; 2 more on the ground were injured, and a hangar, several automobiles and a mobile home were destroyed (NTSB AAR-79-17 Synopsis).

Evidence gate: publish only after verification

  • The single-unit forklift method cut the engine-change disconnects from 79 to 27 and saved about 200 man-hours per aircraft. (NTSB AAR-79-17 p. 26) (NTSB AAR-79-17 p. 26)cited
  • McDonnell Douglas' field representative reported that Douglas "would not encourage this procedure due to the element of risk" in remating — and McDonnell Douglas had no authority to approve or disapprove its customers' maintenance procedures. (NTSB AAR-79-17 pp. 26–27) (NTSB AAR-79-17 pp. 26–27)cited
  • The pylon aft bulkhead's upper flange left the Tulsa engine change with a 10-inch overload fracture; the fracture surface carried a crescent-shaped deformation matching the shape of the lower end of the wing clevis. (NTSB AAR-79-17 pp. 19–20) (NTSB AAR-79-17 pp. 19–20)cited

ExperienceLearn it, live it

Coming soon

EON Life Simulation

Walk into the accident and practise freely, no score. The physics decides, unscripted.

Checking your score

Assessment & certificateProven under pressure