Virtual Time-Reversal and Local Defect Resonance Analysis of Circumferential Interfacial Debonding in Concrete-Filled Steel Tubes

Circumferential interfacial debonding in concrete-filled steel tubes (CFSTs) weakens steelconcrete shear transfer, while the resulting first-arrival difference can remain small. This study combines a three-dimensional explicit finite-element model, virtual time-reversal (TR) reconstruction, geometric and interface-representation sensitivity analyses, and piezoelectric-transducer (PZT) guided-wave measurements to examine localized numerical responses, their local-defect-resonance (LDR) interpretation, and externally measurable residuals. The baseline defect is represented by a displacement-continuous, low-stiffness equivalent interphase; separation-permitted hard-contact variants with either frictionless tangential behavior or a friction coefficient of μ = 0.4 are also examined for the representative 90° case. For the baseline geometry, the early direct-reflection amplitude ratio at 135° is approximately 0.86%, whereas the late-time far-field residual coefficient (RC) increases from 1.16% at 45° to 4.52% at 135°. For the adopted virtual-TR reconstruction, the defect-center resultant-displacement peak and the far-field circumferentially averaged tangential residual peak differ by approximately 80.8 dB. The defect-center spectrum has a local peak near 17.2 kHz, above the 15.0 kHz excitation center, and a curved-shellfoundation estimate of 16.60–17.88 kHz provides a first-order interpretation based on local shell bending, cylindrical curvature, and concrete-core restraint. Response-concentration and circumferential-harmonic analyses show that the reconstructed response is spatially localized and contains mixed circumferential components. The near-end residual coefficient (RC N ) varies with wall thickness, axial defect length, tube size, and interface representation. Across 15 repeated PZT records for each specimen, the mean residual damage index (RDI) values are 5.77 ± 1.89%, 9.23 ± 1.65%, and 15.03 ± 1.50% at 45°, 90°, and 135°, respectively. After separate normalization, numerical RC N and experimental RDI follow the same angle-dependent ordering. These results support residual-based comparative assessment under the tested configuration when an intact reference is available.