Guided wave tomography of pipe bends for wall thinning detection using full waveform inversion

Pipe bends are critical areas prone to wall thinning due to flow-accelerated corrosion. Traditional monitoring methods rely on localized ultrasonic measurements, which are slow and costly. Guided wave tomography (GWT) offers an efficient alternative by utilizing permanently installed transducers to monitor an entire bend. However, an accurate forward model is crucial for GWT, especially for complex geometries like pipe bends. In this study, we propose a GWT approach based on full waveform inversion (FWI) for high-resolution thickness reconstruction of steel pipe bends. To improve the integration of the forward model with the FWI algorithm, we employ Thomsen parameters in a two-dimensional (2-D) domain to simulate the anisotropic nature of wave propagation in the bend. Additionally, we introduce a helical path separation technique, based on geodesic equations, to compute the propagation distance in the pipe bend, preserving wavefront information while accounting for the cyclic nature of the geometry. This method enables accurate separation of wavefronts, including focusing and scattered wave effects. The results were validated experimentally on a pipe with a 220-mm diameter and 1.5-d bend radius, where an artificial defect was introduced. Our findings demonstrate that FWI, combined with the helical path separation, effectively reconstructs the thickness map of defects, particularly those located near the extrados, confirming the feasibility of this approach for large-area, real-time assessments of structural integrity.