The usage of ultrasound in imaging is widely used for material characterisation, alongside other defect-oriented tasks such as defect detection and localisation. In metallic materials factors such as grain size, orientation and shape are different useful properties to identify but come with challenges to estimate. For example, to measure the mean grain size from the backscattering intensity, it is usually assumed that the scattering is weak, so the interaction of ultrasonic waves with the microstructure may be described by the single scattering approximation. Additionally, conventional ultrasonic inversion methodologies require knowledge of elastic constants of the crystallites of the tested material. In this case measurements can be predicted by well-established models. However, experimental verification of the single scattering assumption is not possible in the traditional single probe measurement framework, and material properties of engineering alloys may not always be known. To address these limitations, in this paper a method of estimating the average grain size using full-matrix capture (FMC) data from a 1D ultrasonic array is developed. The method is based on an ultrasonic array grain scattering model and allows to estimate the grain scattering coefficient (Figure-of-Merit, FoM) at a range of frequencies directly from ultrasonic total focusing method (TFM) images. The technique is applied experimentally on several copper samples with different grain sizes and good agreement with optical microscopy results is found.

The tomography of a polycrystalline material’s internal microstructure is critical for modern non-destructive testing and the development of additive manufacturing (AM) materials, yet accurate in-situ characterisation of locally anisotropic metallic materials remains challenging. This study considers the propagation of 2D elastic body waves (P and SV) in a locally anisotropic medium and demonstrates that the spatially varying crystalline orientation angles can be accurately reconstructed using full-waveform inversion (FWI), with the reconstruction formulated as a PDE-constrained nonlinear optimisation problem. The FWI framework consists of forward wavefield modelling, which solves both the forward and adjoint equations using 2D spectral-element methods, and an inverse optimisation process that iteratively updates model parameters based on gradients computed via the adjoint method. We provide an explicit formulation of the adjoint method in FWI for efficiently computing the gradient of the misfit function (local sensitivity kernel) with respect to model parameters by introducing a specific inner product for the function space of vector fields. Furthermore, we introduce a novel sensitivity kernel with respect to the orientation angles, which substantially reduces the dimensionality of the parameter space and eliminates the parameter-coupling issues in most existing FWI studies. Within an appropriate frequency band, the results demonstrate that FWI can deliver high-resolution imaging of the internal grain structure in the tilted transversely isotropic medium considered in this study.

In the ultrasonic testing for Cr coating of zirconium-based cladding in nuclear reactor cores, the challenges arise from the coating's thin-multilayered structure and columnar crystalline microstructure. These challenges include complex sound beam propagation, overlapping echoes from multiple interfaces, and distortion of micro-defect signals, leading to significant errors in the ultrasonic quantitative characterization of coating defects, like debonding and blistering. On the basis of constructing an acoustic model considering the anisotropy of columnar crystals to analyze the mechanism of ultrasonic propagation, this study has developed a multi-resolution signal processing technique for high frequency water immersion focusing probe, incorporating with multi-domain feature fusion. The main contribution of this new technique is to enhance the accuracy of ultrasonic quantitative characterization for Cr-coating defects at the tens to hundreds of micrometer level on zirconium-based cladding. Specifically, the Electron Backscatter Diffraction (EBSD) technique is employed to analyze the characteristics of the columnar crystal structure of Cr coatings, and an elastic anisotropic ultrasonic finite element model incorporating coating crystal orientation is developed. The ultrasonic simulation signals of multiple models with varying columnar crystal characteristics are compared and analyzed. The propagation of high-frequency focused beams within Cr coatings is investigated, and optimal ultrasonic testing parameters for effective acoustic field focusing in the columnar crystal structure of Cr coatings are identified. Based on the high frequency ultrasonic signals collected by the columnar crystal anisotropy simulation model and the Scanning Acoustic Microscope (SAM) experimental system, Synthetic Aperture Focusing algorithm based on the acoustic field characteristics of the water immersion focusing probe is applied to delay and sum the signals. The synthetic signals with high signal-to-noise ratio are decomposed into multi-resolution sub-signals with different frequency components by means of an adaptive variable bandwidth Split-Spectrum Processing (SSP) algorithm. Furthermore, the multi-domain features such as time domain, frequency domain and time-frequency domain are extracted and fused to construct sensitive features for the micro-defect in Cr coatings. Using the sensitive features as quantitative indicators, the size of coating defects is quantified, combing with the half-wave height method. The results demonstrate that, when using a probe with a center frequency of 51.8MHz, compared with traditional ultrasonic C-scan imaging, the proposed method reduces the ultrasonic quantification error of debonding defects with optical microscopic dimensions of 36-176 μm, from 5.7%-208.3% to 2.8%-55.6%. Additionally, the imaging contrast is improved by over 50%.