What happened
On 3 September 2010, a Korean Air B737-800, registration HL7786, was cruising at 33,000 feet while operating flight KE2984 from Irkutsk, Russia, to Incheon, South Korea. While flying approximately 320 NM northwest of Beijing, the flight crew observed an "ENG FAIL 1" message on the lower display unit.
Initial checks showed the high-pressure compressor rotation RPM (N2 RPM) fluctuating between 88% and 94%, after which the error message temporarily cleared. However, the fluctuations persisted, leading the crew to declare an emergency. The aircraft descended to 29,000 feet and performed an emergency landing at Beijing Capital International Airport. There were no injuries among the 36 passengers and 8 crew members on board.
The investigation
The ARAIB examined the engine's accessory gear box (AGB) and the hand cranking shaft drive square. The investigation focused on the history of borescope inspections (BSI) and the condition of the magnetic chip detectors (MCD).
Investigators analyzed the damage to the AGB line 4 and performed a teardown of the accessory gear box. The examination of the engine components revealed that metallic debris, specifically an aluminum alloy, was lodged within the drive shaft square. The investigation also reviewed the maintenance history, noting that the airline had been performing MCD checks every 250 hours, significantly more frequent than the manufacturer's recommended 500-hour interval.
Findings
- The primary cause of the engine shutdown was an axial overload applied during a borescope inspection. To ensure a tool was properly engaged into the drive shaft square, excessive force was used without the inspector realizing that foreign metallic material was present in the square.
- This force created brinelling marks on the outer race of the ball bearing in the AGB line 4.
- During high-speed operation, friction from these marks caused severe damage to the bearing, leading to the fragmentation of the bearing cage and inner ring.
- The resulting gear shaft misalignment caused the N2 RPM to fluctuate, triggering the Electronic Engine Control (EEC) to automatically shut down the engine.
- The metal chips generated by this failure were trapped within the magnetic chip detector and were not detected during the last inspection, which had occurred 138 hours prior to the incident.