Fig. 7 shows the time-history varia

Fig. 7 shows the time-history variations in AE hits and AE energy in correlation with external load, respectively. At the early stage, the load increases with a relatively low slope, which indicates that the original micropores in the steel–concrete interface are gradually compacted. The AE signals detected may be mainly from the closure of original cracks or pores under low stress. A short duration time is a significant characteristic of these AE waves. Anyway, the initial loading with a slow rate is not so evident for the HS-1 specimen. This is because a preloading with 20% of the ultimate bond capacity was firstly applied on this specimen. As a result, imperfections of microstructure in the steel–concrete interface were already eliminated before AE measurement. As the specimen is furtherly loaded, an obvious increase in the number of AE hits with longer duration is found. According to Mpalskas et al. [37], acoustic waves with a long duration time are usually related to damages from shear friction, while acoustic waves with a short duration time are associated to microcracks from tensile stress. Therefore, AE hits with longer durations are accompanied with occurrence of the slip between steel and concrete, and can be used to indicate the development of friction in the steel–concrete interface. In this paper, it is found that the AE signals with a duration time over 8 × 104 μs are highly related to an overall slippage of CFST interface. The time-history variation of cumulative AE energy has a simultaneous ascending trend with the load as shown in Fig. 7(b). It should be noted that the time-history curve of cumulative AE energy rises sharply at about 80% of ultimate bond capacity. After that, there is an inflection point of the curve, which is capable of slowing down its increasing rate. It is clear that AE energy emitted is able to reflect the extent of damage inside an object. Very little AE energy can be detected when tensional cracks of the original micropores occur inside the concrete. It indicates that most of the AE energy is released by the bond failure in steel­ –concrete interface. The transient release of AE energy exhibits a discrete pattern during the loading process. A rapid increase of AE energy is found when the external load reaches to about 50% of ultimate bond capacity. Once the bonding is broken, there will be a corresponding slip in the steel–concrete interface, and then a new bonding interaction is formed in the further depth of load transfer. With the load increasing, AE energy from interface bond failure is released again. There are several cycles of transient AE energy released during the bond stress redistribution. The transient AE energy tends to be stable after an overall slippage. 4.2. AE frequency distribution of bond failure The AE peak frequency is referred to the maximum value on the power spectrum of an acoustic wave [23]. This parameter can be used to distinguish the signal sources from various fracture behaviors. Fig. 8 shows the ranges of AE peak frequency from acoustic sources of bond failure at steel–concrete interface. These AE events with different peak frequencies may be generated from concrete cracking, chemical adhesion failure, steel–concrete debonding and steel breaking. According to Du et al. [31], AE sources from the failure of CFST were mainly located in two frequency bands, including a low-frequency band of 0–100 kHz and a high -frequency band of 150–400 kHz. Gutkin et al. [27] found that AE signals with low frequency usually implied tiny cracking of concrete, while AE signals with high frequency corresponded to largerscale damage of reinforcement in composite structures. In this paper, two frequency bands are also found from AE signals in the push-out test of CFST. These are a low frequency band with the average value of 100 kHz and a high frequency band with the average value of 310 kHz, respectively. It can be inferred that the bond-slip behavior at steel­ –concrete contact surface is mainly derived from two failure mechanisms. One is the steel–concrete debonding related to shear stress released through the steel panel, the other is the concrete microcracks related to tensile stress released through the infilled concrete. The AE peak frequency from the steel panel should be higher than that from infilled concrete. In this test, CFST specimen was reversely pushed out in another direction, and the corresponding AE peak frequencies were also obtained, as shown in Fig. 8(b). This operation is able to exclude the influence of chemical adhesion on distribution of peak frequencies. Anyway, there is no obvious difference on frequency range of AE signals between the two loading processes.