Tropical Helicopters McDonnell Douglas 369D accident

The helicopter crashed onto a rugged hardened lava flow following a loss of engine power during cruise. The pilots of other helicopters heard the pilot make a mayday call with the clear "engine out" audio warning in the background. A post crash fire consumed the majority of the wreckage. An airframe examination revealed no evidence of a preimpact malfunction in what remained of the airframe. An engine inspection and a metallurgical examination revealed that the compressor-coupling adapter, a Parts Manufacturer Approval (PMA) part made by a company other than Rolls-Royce, had failed above the shear point due to fatigue cracking initiated by fretting on the pilot diameter, which had disconnected the compressor from the turbine section. The metallurgist noted that the damage was consistent with the rotation of the aft piece of the fractured compressor-coupling adapter within the stub shaft after it had separated. Fretting damage to the impeller corresponded to the fretting damage on the compressor-coupling adapter, and indicated relative movement between the parts. Rolls-Royce had records of 12 other compressor-coupling adapters that had fractured and failed, with all instances of the failure/fracture occurring in this specific new coupling design, which was significantly different than the previous design. The nature of the fretting and fractures indicated that the newly designed couplings have a small amount of longitudinal movement that is occurring between the outer diameter of the compressor-coupling adapter and the inner diameter of the impeller, which was not a factor in the previous design. If the spur adapter gear, compressor-coupling adapter, and compressor impeller (all coaxial spline joints) were in alignment, there would be no significant longitudinal movement of the pilot diameter. However, any axial misalignment of these components during engine buildup, for example, could induce a misalignment that would result in relative motion between the components with each engine rotation. The misalignment would also induce bending stresses into the compressor-coupling adapter in addition to torsion stresses that could result in fatigue. The combination of fretting damage and cyclic bending stresses are most likely the main factor in the failures of these couplings. One month prior to the accident, company maintenance personnel had changed out the engine gearbox due to finding a loose stud on the gearbox to turbine section mounting. Changing the gearbox would involve disturbing the coaxial spline joints between the spur adapter gear, compressor-coupling adapter, and compressor impeller.

Verbatim NTSB analysis and probable cause from the NTSB dataset