Fatigue Crack Detection at Gearbox Spline Component using Acoustic Emission Method

Didem Ozevin, Justin Cox, William Hardman, Seth Kessler, and Alan Timmons
Submission Type: 
Full Paper
Supporting Agencies (optional): 
phmc_14_014.pdf1.33 MBAugust 29, 2014 - 6:57am

The spline section of helicopter gearbox structure is susceptible to fatigue crack, and non-redundant characteristic leads to the need for early flaw detection strategies. Acoustic Emission (AE) method relies on propagating elastic waves due to release of energy from active flaws. The initiation of damage is identified using the features of AE waveforms such as energy, amplitude and frequency centroid. The characteristics of the AE features are influenced by sensor type, sensor location and gearbox operational conditions. In this study, the AE data was collected from a helicopter gearbox with a notched spline section and realistic operational conditions using two different AE sensors located at two different positions. The data collection was conducted over one year under various operational conditions. The change in the crack state was collected periodically when the gearbox was not operational. The AE features were extracted from long duration waveforms (100 milliseconds) at every pre-defined time step (every 5 seconds). The frequency domain features of frequency centroid and energy distribution in various frequency bands were compared with gearbox operational conditions such as forward load, and temperature. The influences of sensor location, sensor type and operational conditions on the AE features are presented in order to decouple their influences from the AE features due to damage. The comparison between the predicted crack growth time using the AE data and the observed crack initiation shows that the AE method using frequency domain features of streamed waveforms has great potential to identify the crack initiation when the sensor type and location are preserved.

Publication Year: 
Publication Volume: 
Publication Control Number: 
Submission Keywords: 
acoustic emission
Crack Growth
frequency domain
Submission Topic Areas: 
Data-driven methods for fault detection, diagnosis, and prognosis
Submitted by: 

follow us

PHM Society on Facebook Follow PHM Society on Twitter PHM Society on LinkedIn PHM Society RSS News Feed