A numerical study on vibration-based interface debonding detection of CFST columns using an effective wavelet-based feature extraction technique

This paper aims to investigate the impact of interfacial debonding on modal dynamic properties such as frequencies and vibration mode shapes. Furthermore, it seeks to identify the specific locations of debonding in rectangular concrete-filled steel tubular (CFST) columns during the subsequent stage of the study. In this study, debonding is defined as a reduction in the elasticity modulus of concrete by a depth of 3 mm at the connection point with the steel tube. Debonding leads to a lack of correlation between primary and secondary shapes of vibration modes and causes a reduction in the natural frequency in all modes. However, directly comparing changes in vibration responses does not allow for the identification of debonding locations. In this study, a novel irregularity detection index (IDI) is proposed based on modal signal processing via the 2D wavelet transform. The suggested index effectively reveals relative irregularity peaks in the form of elevations at the debonding locations. As the severity of damage increases at a specific debonding location, the relative irregularity peaks would increase only at that specific point; in other words, the detection or non-detection of a debonding location using IDI has minimal effects on the identification of other debonding locations.