Detection of the onset of delamination in a post-tensioned curved concrete structure using hidden Markov modeling of acoustic emissions

Abstract

This paper presents acoustic emission (AE) monitoring of a large-scale curved, post-tensioned concrete wall under monotonically increasing prestressing loads. This structural system, which is commonly used in water storage tanks, silos, bins, and nuclear containment structures, is subject to hidden delamination defeats that may develop during posttensioning and lead to a premature brittle failure. To detect the onset of such defects, this study uses a network of AE sensors mounted on the outer surface of the wall and identifies common patterns in AE signals. Specifically, AE signals are clustered using k-mean clustering, and their sequence is modeled with a hidden Markov model. The comparison of the results with accurate through-thickness expansion measurements of the wall shows that certain patterns in AE signals are correlated with the onset of delamination, and thus can be used to detect it. This early detection of such delamination defects provides decision makers with sufficient time to take remedial and preventive action.