Identification of damages in steel structures using guided wave method

Abstract

Ultrasonic guided wave (GW) has shown clear superiority and strong potential for identification and real-time monitoring of service-induced damage in structures which need on-demand interrogation, by virtue of its unique capability for forward or inverse analysis. The main challenges in GW-based structural health monitoring (SHM) and nondestructive evaluation (NDE) derive from two aspects of complex wave analysis: first, in applications in real structures, the signals collected are complex due to the large number of overlapping reflections obtained in time-traces because of the multiple boundaries of the structures; second, in application with real damage, the effects of closing and opening of a fatigue crack make the acquired wave signals more complex than in instances of artificial notch or crack. However, relatively little research has been devoted to investigation of the inspection of fatigue cracks, especially in complex structures. In this thesis, a single piezoelectric lead zirconate titanate (PZT) actuator-sensor pair for locating a through-thickness crack within welded zone in thick steel plate is initially investigated through both finite element analysis and experiments to demonstrate GW-based techniques in plates with simple geometries which can be used to construct more complex structures for practical applications. The feature extraction method is applied in a welded tubular steel structure (WTSS). A probability-based damage imaging approach is developed. As validation, the approach is employed to predict the presences and locations of multiple slot-like damages in the welding zones of a WTSS. It can be concluded that the identification results using the extracted signal features are comparable, and accuracy when more damage-impaired sensing paths are involved. An energy-based damage imaging approach is evaluated by identifying a fatigue crack in a thick steel plate. The propagation of GWs in the plate-like structure is complicated by thick geometry, wave dispersion, boundary reflection, and the existing boundary notch used to initiate the fatigue crack, resulting in diverse forms of interference with fatigue crack identification. Hence, signal features are extracted from the wave energy distribution. Simultaneously, the proposed method is demonstrated by FEM and good agreement is obtained between the numerical and experimental results using a new developed fatigue crack model. The image-based approach is evaluated experimentally by detection and monitoring of a fatigue crack using time reversal method (TRM). Results indicate that several damage-sensitive features extracted in the normalized captured signals and different pattern recognition techniques are effective for monitoring of fatigue crack propagation in the steel plate, such as TRM, transmission coefficient and principal component analysis (PCA). Some of the experimental results are verified by FEM results. PCA is validated by monitoring of the propagation of a surface fatigue crack in a welded steel angle structure (WSAS) using GWs generated by a PZT sensor network which is surface-mounted to classify and distinguish different structural conditions due to fatigue crack initiation and propagation. Instead of directly comparing the changes between a series of specific signal segments, signal statistical parameters extracted from the frequency domain are demonstrated to have the capability of monitoring fatigue crack in welded steel structures. In summary, application of GW-based damage detection techniques using structurally integrated PZT transducers for SHM is still in its formative years, and one of the main challenges is use in complex real-world structures. Different approaches are validated systematically in this thesis via simulations and experiments. Results for typical cases indicate that the proposed methods are applicable and effective for detection and real-time monitoring of non-fatigue damage and fatigue cracks in engineering structures.