In-situ fatigue crack detection and monitoring in A high-frequency resonance-dwell fatigue test using digital image correlation

Abstract

Resonance-dwell fatigue tests represent a rapid way to accumulate fatigue damage in a specimen. These tests utilize the structural resonances of an assembly to generate large strains at high frequencies using relatively modest input forces. Historically, damage in resonance-dwell fatigue tests has been assessed using metrics based on the natural frequency of the assembly. As damage accumulates, the natural frequency decreases due to a loss in stiffness. This approach to crack detection is somewhat problematic, however, as it can only be reliably used to determine whether damage exists; not the extent of the damage. In this work, digital image correlation is used in lieu of frequency-based metrics to provide a more comprehensive view of how damage evolves in a resonance-dwell, high-cycle fatigue test. Towards this end, a multiply-cracked beam theory model is developed to describe how crack formation impacts the out-of-plane displacement of a continuous-radii specimen. The beam theory model provides a set of basis functions that can be put in the form of a separable nonlinear least squares regression problem, thereby reducing it to a minimization problem in two variables for each crack in the specimen. Using this model, crack depth is estimated by application of the regression model to full-field displacement measurements taken during fatigue tests. Finally, the crack depth estimates resulting from the regression model are validated using post-mortem computed tomography scans of the specimens.