History of Flight
On 3 September 2004, a Piper PA-28-161 Warrior II (registration G-BTSJ) was conducting a training flight at Plymouth City Airport, Devon. During takeoff from Runway 31, the pilot heard a thud as the aircraft became airborne. The pilot continued the climb to 300 feet and contacted Plymouth Tower regarding the noise. The tower informed the pilot that part of the left landing gear had departed the aircraft and that the inner cylinder with the wheel attached, as well as debris, had been recovered from the runway.
The aircraft remained in the circuit and performed a fly-past of the tower to confirm the left landing gear was missing. While the runway was inspected for further debris, the pilot flew an orbit pattern. During the approach for landing, the pilot shut down the engine and isolated the electrical systems. To minimise damage, the pilot landed on the right landing gear and held the left wing up as long as possible. After the aircraft stopped, the pilot and student evacuated unhurt. Damage was limited to the left wing, left flap, and left landing gear.
Engineering Examination
Examination of the left landing gear revealed that the lower torsion link attachment bolt had fractured. This allowed the inner cylinder to become detached from the torsion link and fall from the outer cylinder when airborne. The torsion links prevent rotation between the inner and outer cylinders and retain the inner cylinder in flight.
The lower torsion link attachment bolt on the PA-28-161 can be one of two designs. The simplest is an AN 5-22 hexagonal head bolt with castellated nut and split pin. On aircraft with wheel spats, the bolt has a small threaded section, a hexagonal spacer, the bolt shank, and a final threaded portion. The small threaded portion attaches to the wheel spat; the spacer provides clearance; the shank passes through a sleeve in the torsion link and inner cylinder. G-BTSJ was not fitted with wheel spats but had the latter bolt configuration, indicating it had previously been fitted with spats.
Metallurgic examination showed the bolt failed due to rotational reverse bending fatigue with multiple crack initiation points. The shank had signs of corrosion and polishing from rotation inside the sleeve, indicating the bolt had become loose. The castellated nut and split pin were recovered from the runway with the split pin still inserted, but they were mislaid and not available for examination.
The last maintenance was seven hours prior to the accident, but there was no requirement to inspect the torsion link attachment bolts. The only Piper-defined maintenance for these bolts was lubrication every 100 hours. It was not known how long the bolt had been fitted or when it was last disturbed, and it was not possible to determine why the bolt became loose.
Previous Incident
A similar occurrence involving a Piper PA-28-161 (G-BTBC) was previously investigated by the AAIB (EW/G2003/10/19, published in September 2004 Bulletin). That accident also resulted from a failure of the lower torsion link attachment bolt, causing the landing gear to collapse on landing. The failure mode was slightly different, with a fracture in the bolt shank possibly due to over-torque. As a result, safety recommendations were made to the FAA and EASA calling for improved inspection procedures of torque link attachment bolts (Safety Recommendations 2004-20 and 2004-14). At the time of writing, the FAA were considering their response, and no response had been received from EASA. Since this incident, the maintenance organisation responsible for G-BTSJ undertook to replace all torque link attachment bolts during the next maintenance input on any Piper PA-28-161 aircraft they maintain.