Detection Of Microstructural Damages And Defects Using Electromagnetic Testing And Ultrasonic Guided Waves

Engineering components, widely used in aeronautic, petrochemical and power-generation industry, often contain defects and damages caused by manufacturing processes and long-time service. The growth of defects and damages can reach a critical size during service, and lead to a catastrophic failure of components. During the last few decades, electromagnetic and ultrasonic non-destructive testing (NDT) and structure health monitoring (SHM) methods were thus developed in order to achieve the safe operation of the components. However, doubts persist on the reliability due to the lack of an accurate evaluation of the damage by using obtained signals. In this work, a detailed study of the correlation between microstructure evolution and magnetic properties is carried out. Models for predicting magnetic properties considering the damage of the high temperature components are developed, and the effectiveness of those are verified by the experiments. In order to precisely detect the defects in the structure, the interaction of ultrasonic waves (Rayleigh and Lamb) with defects in the components is investigated, and quantified in terms of the scattering coefficients by using modal decomposition method, finite element analysis and experiments. The details of the investigation and some concluding remarks are as follows:(1) The effect of microstructure evolution on magnetic domain in commonly used heat-resistant materials is studied. Curie temperature and the magnetic properties characterized by magnetic hysteresis loop under 500-600℃ of 10CrMo910, P91 and 23CrMoNiWV88 are measured both by JLD-III set-up and vibrating sample magnetometer (VSM). An intelligent optimization algorithm (QB-PSO) is used to determine five hysteresis parameters of Jiles-Atherton (J-A) model. The results indicate that the size and distribution of magnetic domain are strongly influenced by the grain size, grain boundary, precipitations and dislocation density. It is also shown that magnetic properties of three materials remain detectable at their service temperature, therefore the hysteresis-based NDT can potentially be improved as an in-situ monitoring method at high temperature.(2) The pinning and effect-field scaling parameters in J-A model are modified by taking the effect of microstructure evolution at early creep stage on domain wall motion into account. Magnetic properties and microstructure of creep specimens are measured and observed by using VSM, OM, SEM and TEM respectively. Quantitative metallographic analysis is carried out to determine the grain size, dislocation density and amount of precipitations of each creep specimen. The results of hysteresis parameters in the modified J-A model are compared with those experimentally characterized from hysteresis loops, revealing good agreement between them. Note that without modifying the model, the calculated parameters can be 50 times higher than the experimental ones. The assessment of early creep damage by using minor hysteresis loop method is also conducted, results show that minor hysteresis loop coefficients are decreasing monotonically with increasing creep life fraction, and the coefficients can potentially predict the presence of the hidden defects in the material.(3) The effect of the localized creep damage on the hysteretic magnetic properties in a CT creep specimen is studied by using SQUID-VSM. The metallographic and three-dimensional finite element analyses are performed to correlate creep damage indicators with magnetic properties. It is found that the presence of creep cavities and high stress triaxiality significantly reduce the coercivity and remanence of the specimen. J-A model is modified by taking the effect of time-dependent creep-cavity growth on the magnetic domain wall motion into account. Creep damage rate reflecting the physics of the cavity growth controlled by power-law and diffusion coupled mechanism is incorporated into the model for predicting the magnetic properties of serviced components. The comparison between experimentally measured and numerically calculated magnetic properties confirms the effectiveness of the modified model in predicting the variation of magnetic property of serviced components.(4) The analytical approach to determine the scattering coefficients for Rayleigh and Lamb wave conversion at a crack follows the basic ideas of the modal decomposition method. The system of linear equations reflecting displacement and traction of reflected and transmitted waves is given based on continuity condition at the crack interface, and solved numerically in a least-square sense for determining energy-based scattering coefficients. In order to verify the results from the analytical framework, transient finite element (FE) simulations are conducted. The scattering coefficients of the propagating waves are determined by applying filtering techniques (time window and 2D-FFT) to the displacements from simulation. The analytical solutions are compared with results from numerical simulations. For both incident Rayleigh and Lamb wave, excellent agreement is found. Therefore, combing the scattering coefficients both at leading and trailing of the crack it is possible to quantify the interaction of ultrasonic waves with the damage in the structure. Moreover, the effective and reliable detection can be achieved if the aforementioned algorithm is integrated in the testing technique and also choose specific excited frequency (300kHz-500kHz) at which high order anti-symmetric modes can be an excellent indicator for the presence of crack in the structures.