Ultrasonic internal multiple migration: An imaging method for defect characterization in complex-structure welds

Ultrasonic array imaging is widely used for defect characterization in the non-destructive evaluation of welds. Recent ultrasonic multi-view imaging extracts additional defect information from the dataset by accounting for wave reflections at structural boundaries. However, defect characterization in complex-structure welds remains challenging due to: (1) the difficulty in identifying geometrical artifacts, even when the weld structure is known; and (2) the challenge of selecting the optimal view and the intensively computational cost of calculating all possible views. Based on the principle of reverse time migration of multiples, this paper introduces ultrasonic internal multiple migration (UIMM) to addresses these challenges. The input velocity distribution is defined based on the known boundaries of the welded structure to eliminate geometrical artifacts and facilitate accurate defect characterization. Elastic wave mode decomposition and directional wave separation are conducted in the wavenumber domain to efficiently distinguish echoes in different modes. The decomposed wavefields are then correlated to generate multiple views and further reduce artifacts. Experiments on nozzle weld mockups with artificial reflectors showed that four mode combinations were sufficient for imaging cracks at a wide range of orientations. The results demonstrate that defects were imaged without interference from artifacts. Cracks of various sizes and orientations were precisely characterized, with positional and size accuracies within 0.70 mm and 0.82 mm, respectively, and an angular deviation of less than $2.2^{\circ }$.