Seminar on Advancing High-Resolution Ultrasonic Characterization: From Exploring Polycrystalline Microstructures to Imaging Complex Composites

This research seminar delves into the realm of ultrasonic wave propagation and scattering in complex media, aiming to enhance quantitative non-destructive evaluation (NDE) capabilities for engineering structures. The study commences by addressing the NDE challenges posed by polycrystalline microstructures, especially in critical components like welded joints widely used in aerospace, marine, automotive, and energy industries. The mechanical properties of welds are significantly influenced by the distribution of grain sizes within different subregions. To address this, a promising quantitative ultrasonic NDE tool is developed at the grain size scale using two-dimensional and three-dimensional finite element (FE) modeling, providing a more accurate description of elastic wave propagation and scattering within polycrystals. The proposed framework is validated by classical second-order approximation (SOA) modeling results.

In the forward problem, the influence of grain size on the ultrasonic attenuation and phase velocity dispersion is studied, reflecting statistically averaged scattering characteristics. The backscattering coefficient is then extracted, encapsulating temporal and spatial information as propagating through the weld. In the inverse problem, a convolutional neural network constructed via Long Short-Term Memory (LSTM) is employed for processing the sequential backscattering signals. This approach enables the identification of boundaries between different welding subregions and facilitates the assessment of their averaged grain sizes.

The motivation for high-resolution characterization is also driven by the imaging of complex-shaped carbon fiber reinforced polymer (CFRP) laminated components. These components are extensively used for critical structures in aircraft and rocket engines, and any design or manufacturing defects can compromise their in-service performance and reliability. While ultrasonic array techniques are commonly employed for CFRP component characterization, the curved geometry (usually with ply-drops), strong anisotropy, and inhomogeneity of these structures can severely distort the ultrasonic beam and reduce inspection accuracy.

To address these challenges, the research proposes an innovative ultrasonic imaging method based on Dijkstra's algorithm (DA) and ray-tracing techniques. This method evaluates side-drilled hole (SDH) scatterers, resin-rich regions, and interlaminar interfaces in CFRP laminated components. Unlike conventional methods assuming straight-line propagation of ultrasound, the proposed method combines DA with high-fidelity finite element (FE) modeling to retrieve realistic ray paths in complex media. By superimposing full matrix captured (FMC) ultrasonic data according to DA predictions, the accuracy of the delay law and hence the structural features imaging is improved. Imaging parameters can be adjusted, such as filtering frequency and aperture angle limit, to achieve optimal imaging for various features located deep within the structure. These advanced characterization results will play a crucial role in supporting reliable and lightweight design for advanced composites in the aerospace industry.

Join us for this insightful research seminar, where we explore cutting-edge techniques to push the boundaries of high-resolution ultrasonic characterization, benefiting various engineering applications and paving the way for more efficient and reliable inspection practices.