PSI - Issue 66

Vivek Vishwakarma et al. / Procedia Structural Integrity 66 (2024) 381–387 Vishwakarma and Ray/ Structural Integrity Procedia 00 (2025) 000–000

382

2

preferred for crack monitoring in concrete structures. However, it often falls short in situations where structural elements are not visible, and due to limited out-of-plane movement, stereo DIC is also unable to predict 3D crack profiles accurately. These limitations underscore the need for alternative methods. Acoustic Emission techniques offer a promising alternative for monitoring crack initiation and propagation within concrete. AE sensors capture stress waves generated by crack formation and growth, providing valuable insight into the damage evolution within the material. Numerous studies have explored AE for crack characterization and localization in concrete. Md Nor et al. (2021) employed AE to diagnose fatigue damage severity in reinforced concrete beams. Sagar et al. (2012) performed a probabilistic analysis of AE events and energy release during crack formation in cementitious materials under compression. The use of advanced AE parameters for structural health evaluation is discussed by Ma & Du, (2020), while Muñoz-Ibáñez et al. (2021) investigates AE monitoring of mode I fracture toughness tests. Clustering algorithms and validity indices relevant to AE analysis are reviewed by Maulik & Bandyopadhyay, (2002). However, accurately predicting the crack path, particularly in reinforced concrete, remains challenging due to the stochastic nature of crack propagation and the complex interaction of multiple cracks. Existing methods often struggle to capture the three dimensional nature of crack growth and the influence of reinforcement. This paper presents a novel probability-based approach for predicting crack propagation in lightly reinforced concrete beams using AE parameters. In this study, lightly reinforced beams are used as they fail in flexure with a single major crack at the centre, thus reducing the complexity of the problem. The proposed methodology combines Gaussian Mixture Model (GMM) clustering with a spatial binning strategy to simulate the crack path and plane. GMM clustering categorizes AE events based on AE parameters. After clustering, the cluster associated with the tensile mode failure was identified, and only those events with a high probability of being in that cluster were selected in the further analysis. The filtered AE data is then spatially binned to simulate the evolving crack path. Further, a temporal binning strategy is utilised to predict the crack length. Experiments were conducted on notched, lightly reinforced concrete beams under flexural loading, with AE data continuously recorded. The predicted crack paths were then validated against results obtained from DIC. This research contributes to structural health monitoring by offering a robust and real-time crack path prediction tool for reinforced concrete structures.

2. Experimental program 2.1. Material properties

A 43-grade ordinary Portland cement was utilized to cast concrete beam specimens. The mix design was carried out in accordance with IS 10262:2009. Locally sourced river sand with a specific gravity of 2.59 and a maximum coarse aggregate size of 12 mm were used. The mix proportion details and properties of the various ingredients used in the concrete preparation are summarized in Table 1. The average 28-day cube compressive strength of the standard sized specimens was 38 N/mm 2 .

Table 1. Concrete mix and material properties

w/c ratio

0.45

Mix proportion

1:1.74:2.2 38 N/mm 2 3.7 N/mm 2

f c (Comressive strength) f t (Tensile strength) ν (Poisson’s ratio) E (Young’s modulus

0.2

33980 N/mm 2 Deformed bars of grade Fe 550SD were used as reinforcement in the beam specimens, with an average yield strength of 653 N/mm 2 . The beams were cast with dimensions of 1000 mm X 200 mm X 120 mm and cured for 28