Disk Crack Detection and Prognosis Using Non Contact Time of Arrival Sensors

Abstract

Early detection of cracks in engine rotors can prevent uncontained engine fractures. A detection system may also reduce maintenance costs by increasing the time between inspections and reducing the number of spares required. Currently, the only method to monitor engine rotor degradation is through periodic removal of the engine from service, disassembly, and inspection of each rotor. The objective of this study was to propagate a crack in a cyclic engine test and evaluate the ability of an eddy current sensor and Reasoner software system to isolate a crack and predict its remaining useful life. The engine used for these tests was a Spey RB168 Mk 101 engine with titanium blades. Prior to the engine test, low cycle fatigue (LCF) spin-pit testing was carried out on a pre-flawed disk to validate understanding of the disk crack growth and reduce the risk of fracture during engine testing. Time of arrival data was collected using the QinetiQ eddy current sensor based tip timing system, to investigate the blade movement caused by crack growth in the disk. A greater understanding of the different types of faults that might exist in a rotor assembly and the ability to differentiate between them using the tip timing system is essential if the true nature of a fault is to be diagnosed and reported. A software Reasoner, using physics-based structural transfer functions to relate blade tip timing measurements to the damage state of the disk, was validated during the spin-pit test and was subsequently used on the cracked disk test at MoD Shoeburyness. The Reasoner provided real-time monitoring of the crack growth and remaining useful life of the component.