Improved fracture predictions using Acoustic Emissions data

The mechanical health of any composite specimen can be monitored by recording the acoustic emissions (AE) from the surface and hence can be used to predict mechanical fracturing of the material. The non-destructive and non-invasive technique of AE is an established method to monitor the mechanical health of a structure. The AE captured from a specimen undergoing mechanical stress appears to display power law like distribution. While empirical data almost never follow power law behaviour completely, which, in turn makes the estimation of its exponent (b-value) to be highly erroneous. Literature shows that attempts have been made by the researchers to filter out the non-power law parts of the dataset, however, issues such as choice of upper and lower bounds, and fixed values of these bounds for different types of datasets remain a challenge for efficient and robust estimation of b-value. In the present work, the author has proposed a novel technique to rectify the inherent shortcomings of b-value and improved b-value (ib-value) estimation methodology. The proposed method, which is computationally light to implement, has been verified on synthetic, experimental datasets and reputed datasets from literature. The results show measurable quantitative improvements in the estimation of b-value parameter compared to previous reported methods in the literature.