Damage Localization in Complex Structures by Built-in Ultrasonic Transducers
Niri, Ehsan Dehghan
MetadataShow full item record
It is known that civil infrastructures are subjected to deterioration due to aging, increased load, and natural hazards. To minimize the maintenance costs and to increase the operation lifetime, researchers and practitioners are increasingly interested in improving current nondestructive evaluation (NDE) technologies or building advanced structural health monitoring (SHM) strategies. Structural health monitoring (SHM) based on guided ultrasonic waves (GUWs) is a field that has received significant interest in the past few years, which has led to the development of a variety of systems and signal processing techniques for damage detection in complex structures. The most important aspect of a SHM system is the ability to "localize" the damage. Although source localization algorithms have been developed, unanswered questions have been posed regarding their reliability and accuracy. The inherent uncertainty in sensor measurements, caused not only by the sensor impreciseness and noise, but also from the ambiguities and inconsistencies present within the environment, and from an inability to distinguish between them, may hamper their reliability in terms of automatic damage localization. Recent advances in probability theory are used to characterize uncertainty in sensor data, and ultimately develop robust signal processing algorithms for source localization in complex structural systems such as plate-like and cylindrical structures. In plate-like structures, traditional damage or impact location is based on time-of-flight (TOF) triangulation of wave measurements taken at multiple receiving points. An alternative probabilistic approach, based on two nonlinear Kalman filtering methods, Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF), is proposed for AE source localization and wave velocity identification in isotropic and anisotropic plate-like structures. This approach, in which TOF and wave velocity are considered as Gaussian random variables, consists of two main stages. During the first stage, TOF measurements of Lamb waves are carried out by different signal processing methods such as a Continuous Wavelet Transform (CWT) accounting for systematic error and random noise; and the second stage uses the nonlinear Kalman filtering to adaptively fuse the information provided by sensors in the first stage and iteratively estimate the location of the AE source and the wave velocity. Several experiments on isotropic and anisotropic panels are validated the efficiency and accuracy of the probabilistic algorithm. In this dissertation a new damage source localization approach is proposed for SHM of cylindrical structures. The approach uses an array of permanently installed low profile piezoelectric transducers to generate and receive helical guided waves around the cylindrical structure. This approach is able to detect and locate active damage such as leak, cracks, pre-existing cracks and corrosion by toggling between two modes: 1. "passive" AE monitoring and, 2. "active" GUWs test. In the first mode, an advanced uncertainty quantification method based on Unscented Transformation (UT) is proposed to take into account uncertainty in leak localization. In the second mode, a multi-helical ultrasonic tomography (MHUT) approach is developed for localization of existing defects, such as a pipe wall thickness loss due to corrosion. Experimental tests are carried out on a steel pipe instrumented with six permanently attached piezoelectric transducers to validate the proposed approach. As the sensor network hardware such as wireless technology and software technology evolves the concept of autonomous independent embedded sensor nodes in SHM systems is becoming both economically and technologically feasible. However, the concept of "embedded" sensing cannot be fully realized if the SHM systems will require cables to access to traditional power sources or if batteries have to be periodically replaced. In the last part of this dissertation a novel tunable vibration-based piezoelectric energy harvester that can potentially supply electricity to a SHM node is proposed. Additional stiffness and axial load are introduced to the tip of a traditional cantilevered piezoelectric energy harvester, such that the resonance frequency of the device is now a function of both the stiffness and axial load. For the first time, the analytical solution of an axially loaded cantilevered piezoelectric energy harvester with tip stiffness, using Euler-Bernoulli beam assumptions, is presented. Analytical results show that the device is capable to be passively tuned to a wide range of frequencies. Furthermore, the device can be tuned to very low frequencies without the requirement of having a large proof mass.