Ultrasonic Techniques for Baseline-Free Damage Detection in Structures

This research presents ultrasonic techniques for baseline-free damage detection in structures in the context of structural health monitoring (SHM). Conventional SHM methods compare signals obtained from the pristine condition of a structure (baseline signals) with those from the current state, and relate certain changes in the signal characteristics to damage. While this approach has been successful in the laboratory, there are certain drawbacks of depending on baseline signals in real field applications. Data from the pristine condition are not available for most existing structures. Even if they are available, operational and environmental variations tend to mask the effect of damage on the signal characteristics. Most important, baseline measurements may become meaningless while assessing the condition of a structure after an extreme event such as an earthquake or a hurricane. Such events may destroy the sensors themselves and require installation of new sensors at different locations on the structure. Baselinefree structural damage detection can broaden the scope of SHM in the scenarios described above. A detailed discussion on the philosophy of baseline-free damage detection is provided in Chapter 1. Following this discussion, the research questions are formulated. The organization of this document and the major contributions of this research are also listed in this chapter. Chapter 2 describes a fully automated baseline-free technique for notch and crack detection in plates using a collocated pair of piezoelectric wafer transducers for measuring ultrasonic signals. Signal component corresponding to the damage induced mode-converted Lamb waves is extracted by processing the originally measured ultrasonic signals. The damage index is computed as a function of this mode-converted Lamb wave signal component. An over-determined system of Lamb wave measurements is used to find a least-square estimate of the measurement errors. This error estimate serves as the damage threshold and prevents the occurrences of false alarms resulting from imperfections and noise in the measurement system. The threshold computation from only the measured signals is they key behind baseline-free damage detection in plates. Chapters 3 and 4 are concerned with nonlinear ultrasonic techniques for crack detection in metallic structures. Chapter 3 describes a nonlinear guided wave technique based on the principle of super-harmonic production due to crack induced nonlinearity. A semi-analytical method is formulated to investigate the behavior of a bilinear crack model. Upon comparing the behavior with experimental observations, it is inferred that a bilinear model can only partially capture the signal characteristics arising from a fatigue crack. A correlation between the extents of nonlinear behavior of a breathing crack with the different stages of the fatigue crack growth is also made in Chapter 3. In Chapter 4, a nonlinear system identification method through coherence measurement is proposed. A popular electro-magnetic impedance circuit was used to detect acoustic nonlinearity produced by a crack. Chapters 5 and 6 comprise the final part of this thesis where wavefield images from a scanning laser vibrometer are digitally processed to detect defects in composite structures. Once processed, the defect in the scanned surface stands out as an outlier in the background of the undamaged area. An outlier analysis algorithm is then implemented to detect and localize the damage automatically. In Chapter 5, exploratory groundwork on wavefield imaging is done by obtaining wave propagation images from specimens made of different materials and with different geometries. In Chapter 6, a hitherto unnoted phenomenon of standing wave formation in delaminated composite plates is observed and explained. Novel signal and image processing techniques are also proposed in this chapter, of which the isolation of standing waves using wavenumber-frequency domain manipulation and the use of Laplacian image filtering technique deserve special mention.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.