On 4 July 1999, a Fokker F100 aircraft, on a direct service from Brisbane to Norfolk Island with 43 persons on board, experienced a severe vibration from the left main landing gear as the brakes were applied during the landing roll. The crew were able to bring the aircraft safely to a halt on the runway and then conducted an inspection of the aircraft. This revealed that the left main outboard wheel was missing. There was no other damage to the aircraft and no injuries to any of the occupants. The crew reported there was approximately 15 knots of crosswind for the landing.
The wheel was later located on the flight strip, where it was examined and then forwarded to the Australian Transport Safety Bureau (ATSB) for specialist fracture analysis.
Examination of the maintenance documentation showed the wheel had completed 99.8 hours and 77 cycles since it was last overhauled. During overhaul, the wheel had also undergone an approved repair to remove scoring from the hub caused by rubbing contact with the brake heat shield during service. The repair involved reducing the hub diameter by 0.02 inches by machining. The repair instructions specified that after the material was removed the repaired region was to be shot-peened and the shot-peening parameters were to be adjusted to produce a specific surface quality.
The specialist fracture analysis report concluded that the wheel failed because of a fatigue crack, starting at the surface of the metal, in the repaired region of the axle-hub to wheel web transition. It found that no single stress point concentrator had started the cracking. It had begun at numerous closely spaced points around the circumference of the hub, known as ratchet marks. This was consistent with sideways flexure of the wheel web, and with crack growth from the repaired surface of the hub. There was no indication the growth had started at any crack that had been present prior to the repair.
The manner in which the fatigue crack spread was also consistent with sideways flexing of the wheel web. Such flexing would occur when a turning moment (torque) was applied to the main landing gear while the wheels were rotating, such as during ground turning or crosswind landings. The crew reported that high crosswind components are regularly experienced during take-off and landing at Norfolk Island.
A comparison of the surface of the repaired region with the original surface of the wheel hub revealed that the intensity of shot peening was lower than that applied during manufacture. This variation would be expected to lower the resistance of the wheel to fatigue cracking. The lower level of compressive residual stress associated with the less intense shot-peening process applied to the repaired region would also increase the likelihood of fatigue failure under normal loading conditions. As part of the investigation into this incident, the operator's maintenance facility found that a part of the left main landing gear shimmy damper had been incorrectly re-assembled during last overhaul. Analysis of the wheel fracture surface evidence determined that this would have had minimal if any influence on the fatigue crack starting or spreading.
On 9 October 1999, this aircraft had a similar incident involving the failure of the left main landing gear upper torque links while landing at Norfolk Island, ATSB occurrence number 199904802. The ATSB specialist fracture analysis report found the torque link failure had started in similar circumstances to those for the wheel hub failure.
CONCLUSIONS
The left main landing gear wheel failed as a result of fatigue cracking. Although the cracking started at the site of the repaired area of the wheel hub, the cracking did not start as a result of any surface changes, such as scoring.
The surface treatment of the repaired region was significantly different to the "as manufactured" condition of the hub. The repaired region had evidently been shot peened by a lower intensity process.
A reduction in the intensity of shot peening, with an accompanying reduction in the magnitude of residual compressive stress, had made fatigue cracking more likely under normal loading conditions.
The wheel finally failed during the early stage of landing at Norfolk Island Airport while torque was being transmitted through the landing gear assembly. These significant landing gear torque loads were probably associated with crosswind take-off and landings.
RECOMMENDATION R20000310
The ATSB recommends that the UK Civil Aviation Authority review the repair and overhaul processes for the failed wheel and also for the failed torque links identified in occurrence 199904802, to ensure they conform to the appropriate airworthiness requirements.