What happened
On 21 October 2021, an Aeronca 65C, registration G-BTRG, departed from Birchwood Airfield in North Yorkshire for a private flight. The aircraft, which had recently undergone annual maintenance and a Permit-to-Fly renewal, was carrying one pilot and one passenger.
During the initial climb, passing through approximately 450 feet, the pilot attempted to reduce engine power. At this moment, the engine suddenly experienced a significant overspeed. The pilot immediately closed the throttle and maintained a best gliding speed of 65 to 70 mph. After evaluating available landing sites, the pilot opted to return to the airfield. During the approach to Runway 03, the aircraft's speed was too high for a safe stop, prompting a realignment to Runway 08. The aircraft landed on its main wheels but continued along the runway, eventually striking a hedge at the far end. The aircraft sustained extensive damage, and the propeller was found to be missing from the engine.
The investigation
The investigation focused on the mechanical failure of the propeller attachment. The aircraft was powered by a Verner Scarlett 7-cylinder radial engine. The propeller was secured to the engine flange using six M8 x 110 mm steel bolts.
Metallurgical examination of the failed bolts using a Scanning Electron Microscope revealed that all six bolts had failed due to high-frequency, reverse-bending fatigue. Analysis of the bolts showed evidence of grinding marks, indicating they had been shortened to ensure clearance for the alternator windings. Investigators found that the bolts had 'shanked' during installation. This occurred because the bolt shank was too long for the thickness of the propeller, washer, and crush plate assembly, meaning that while the correct torque was applied, the bolts did not actually stretch to create the necessary axial tension, or pre-load.
Findings
- The propeller departed the aircraft because all six attachment bolts failed due to fatigue caused by a lack of pre-load.
- The bolts had been shortened to prevent interference with the alternator, but this resulted in a shank length that was approximately 3 mm too long for the component stack.
- When the bolts were tightened to the required torque, they reached the end of the unthreaded shank before any tension could be applied to the threaded portion.
- It is possible that the selection of the 110 mm bolt length resulted from a misinterpretation of the engine manufacturer's installation requirements.
- A 100 mm bolt would have provided sufficient thread engagement and prevented the shanking issue.