Filtering approaches of AE data in damage severity assessment of highly emissive concrete structures under loading

Abstract

The application of the AE technique in the field of civil engineering requires overcoming various issues related to several factors such as structure complexity, material nonhomogeneity, signal attenuation, environmental noise. In this concern, multivariate statistical analysis was successfully applied to manage AE data and discriminate relevant features related to main damage mechanisms in concrete. Principal component analysis (PCA) and artificial neural networks (ANN) were applied with promising results. However, a fundamental limitation of these numerical methods is that they are not user-friendly and their application requires suitable scientific expertise. In such a context, the development of an easy de-noising protocol, able to simplify the AE data analysis for damage structure assessment and failure prediction, is a fundamental improvement toward the applicability of this technology on real-scale concrete structures. In the present paper, the analysis of AE data collected during loading and unloading cycles up to the failure on a real scale post-tensioned concrete beam is reported. Different denoising approaches were adopted to remove a large incoherent AE population originating from the friction of hydraulic actuator steel plate on concrete surface and friction on beam supports during loading/unloading steps. Filtered data were then synchronized with the beam deformation and crack width opening. De-noising algorithms have then been validated by structure damage severity assessment using statistical indexes such as calm ratio, load ratio, severity, and historical index. The procedure was tested with interesting results on PT concrete beams characterized by different pre-existing damages on steel tendons. EWGAE 35, Ljubljana, Slovenia, 13th – 16th Sep. www.ewg