Localized corrosion induced damage monitoring of large-scale RC piles using acoustic emission technique in the marine environment

• Published in: Construction & building materials 2020-05, Vol.243, p.118270, Article 118270
• Main Authors: Zheng, Yonglai, Zhou, Yujue, Zhou, Yubao, Pan, Tanbo, Sun, Limin, Liu, Don
• Format: Article
• Language: English
• Subjects: Acoustic emission (AE); Localized corrosion; Machine learning; Marine environment; Piles
• ISSN: 0950-0618; 1879-0526
• DOI: 10.1016/j.conbuildmat.2020.118270

[Display omitted] •6 large-scale piles were subjected to localized accelerated corrosion in a simulated marine environment.•ADPF filter provides an alternative viable method to the traditional filtering approach.•The AE monitoring can enable an early identification of damage induced by localized corrosion.•MLP models provide an effective method to quantify the corrosion degree of reinforcement in RC piles. An experimental study was conducted to apply the acoustic emission (AE) technique to monitor the corrosion process and cracking behavior in large-scale reinforced concrete (RC) pile specimens. In this study, the underwater and tidal zones of six RC piles were exposed to accelerated, localized corrosion in a simulated marine environment to reach 5%, 10% and 20% steel mass loss. The two piles with 20% steel mass loss together with another reference pile without corrosion were continuously monitored during the test via three attached AE sensors. Tidal action had a significant impact on the AE signals; accordingly, a novel amplitude-duration-peak frequency-based AE signal filter (ADPF) was proposed, which achieved better performance than the previous amplitude-duration-based filters. Additionally, the conjoint analysis of AE signals and the fractal dimension of cover cracks throughout the corrosion period enabled global detection of localized corrosion-induced damage of piles, regardless of sensor location. This study also presents an integrated corrosion-induced damage detection framework and four models based on multi-layer perception (MLP) networks to predict the corrosion level of the underwater and tidal zones of piles in the marine environment.