Interface voids in concrete-filled steel tube (CFST) members weaken steel–concrete composite action and are difficult to detect using conventional inspection. This study proposes a baseline-differenced, total-variation-regularized, and edge-enhanced simultaneous iterative reconstruction technique (BTVE-SIRT) for low-redundancy ultrasonic tomography of CFST interface voids. The method combines intact-baseline differencing, interface-prior weighting, total-variation constraints, and edge enhancement to suppress structural background effects and improve the recovery of near-interface low-velocity anomalies. A measurement-point scanning strategy was first used to balance detection reliability, false-alarm control, interface coverage, and acquisition efficiency. A 15-point circumferential layout was selected as the lowest-redundancy feasible scheme, providing 150 directed travel-time records for reconstruction. Tests on intact, annular-void, and spherical-cap-void specimens showed that BTVE-SIRT produced no stable false positives under the adopted criteria. For annular voids, the average IoU and Dice coefficient were 0.682 and 0.808, with arc-angle and thickness errors of 5.42° and 0.93 mm. For spherical-cap voids, the corresponding values reached 0.836 and 0.909, with a centroid error of 3.99 mm and a height error of 2.96 mm. These results demonstrate improved reconstruction stability and geometric recovery relative to the designed reference geometry.
Guided-wave-based Structural Health Monitoring (SHM) requires thorough assessment of damage sensitivity across a range of defect sizes and locations. This process typically involves manufacturing numerous mock-up samples with similar properties, which can be difficult, costly, and time-consuming. Removable artificial defects provide a practical means of reproducing selected characteristics of real damage while streamlining SHM validation.Soft materials in the form of colloidal suspensions can serve as suitable representations of real defects due to their significant influence on propagating Lamb waves. However, their viscoelastic properties must be accurately characterized to enable their effective use. In this work, we investigate the influence of colloidal suspension layers on the propagation of Lamb waves in thin-walled structures. We present a procedure for quantifying their effect through local frequency-dependent wavenumber and attenuation characteristics for different Lamb-wave modes.Furthermore, we identify the viscoelastic material properties of colloidal suspensions using an inversion procedure based on experimentally estimated attenuation and wavenumber values. The resulting material model is validated by comparing theoretical complex dispersion curves of Lamb waves propagating in plates with an added colloidal layer against experimentally obtained dispersion curves for plates made of three different materials: aluminum, PMMA, and CFRP. Finally, we demonstrate that colloidal layers can reproduce selected features of embedded Teflon-insert delaminations in CFRP plates, indicating their potential as artificial defects for SHM studies of such structures.
The performance of ultrasonic array imaging for defects in large-thickness laminated composites is affected by ultrasonic attenuation. Moreover, as the number of plies increases, combined with strong anisotropy, severe background noise, image distortion, and artifacts appear in array ultrasonic imaging of large-thickness laminated composites. To solve these problems, this paper proposes an array ultrasonic total focusing imaging method based on ultrasonic attenuation compensation with the propagation path and an adaptive pixel summation strategy. The novelty of the proposed method lies in calculating the travel time field via the anisotropic Eikonal equation, combining with the second-order focal adaptive weighting strategy and an attenuation compensation module based on the ultrasonic propagation path for pixel summation. In simulations and experiments, the imaging performance for a 2 mm delamination at 24 mm depth in a 30 mm thick laminated composite was tested. The proposed method was further applied to image debonding at the interface between a laminated composite and a metal layer. A comparison with advanced imaging methods demonstrates the advantages of the proposed method. Quantitative results also show that the proposed method significantly improves the signal-to-noise ratio (SNR) of the imaging results.
