Keywords

1 Introduction

Concrete-filled steel tube (CFST) structures make full use of advanced mechanical properties of both concrete and steel, and have been widely used in large-scale civil engineering structures including super high-rise buildings, underground works, and long-span bridges in recent years [1, 2]. Due to the poor construction quality control methodologies and unavoidable shrinkage of mess concrete, various types of defects, such as void and interface debonding may occur under inner horizontal diaphragms of CFST components. Interface debonding and void defects will seriously affect the mechanical properties and service life of the CFST structure, which leads to risk in structural safety and serviceability [3,4,5]. Therefore, the development of defects detection technology for CFST structures is an emergent concern in civil engineering.

Zhangjinggao Yangtze River Bridge, with a main span of 2,300 m, is the world's largest main span suspension bridge under construction. The bridge has a steel box-CFST composite tower with a height of 350 m, and the diameter of each circular CFST column employed in the bridge tower is 3.6 m as shown in Fig. 1. Horizontal diaphragms, two vertical partition plates mutually perpendicular to each other and shear studs are employed to ensure the composite effect between the inner wall of the steel tube and the concrete of each CFST component. During the construction of the steel box-CFST composite tower, void defects may occur under the inner horizontal diaphragm due to complex internal structure details and insufficient vibration during concrete pouring. This pouring uses the high drop method, although the method is conducive to the self-compaction of concrete, but the internal design of this experimental component has inner horizontal diaphragm, and the discontinuity of concrete construction is prone to produce the situation of uncompacted pouring. There is a critical need to develop efficient methods for visualizing potential void in the full-scale CFST column of composite tower.

Fig. 1.
figure 1

The schematic diagram of the standard section of steel box-CFST composite tower

Full size image

To detect structural defects, several non-destructive testing (NDT) techniques have been developed [6], e.g., hammer impact method [7, 8], ultrasonic guided waves [9], impact-echo method [10], and infrared thermal imaging [11]. Unfortunately, the performance of some traditional approaches is dependent on the experience of operators. Ultrasonic guided wave methods have been used to detect concrete cracks, interlayer debonding of composite laminates, but are not capable of the defect detection for large-scale composite and hybrid engineering structures. Infrared thermal imaging method to determine the scope of defects through the heat conduction theory, but still needs to conduct in-depth research on the clarity of the infrared thermal image. The steel tube of CFST members as a metal medium with electromagnetic shielding, traditional electromagnetic method does not work for internal void defect detection of CFST members also. In contrast to the NDT method described above, electromechanical impedance (EMI) measurement shows its advantages in defect detection of engineering structures [12, 13]. Lukesh et al. [12] first described the advantages and disadvantages of piezoelectric impedance technology and the current engineering application in detail, followed by a discussion of the fabrication techniques of new sensors and data processing methods applied to the EMI method. The development prospects of the EMI method are analysed by focusing on the influence of sensor materials and external environmental on the detection accuracy. The new directions of EMI technology in the field of detection are proposed from various aspects, such as intelligent measurement technology, diversification of acquisition parameters and depth of data mining. Nguyen et al. [13] conducted a relevant experimental study and numerical simulation analysis after verifying the correlation between the admittance signal and anchor force using theoretical derivation in order to achieve dynamic monitoring of anchor force of prestressed anchorage using EMI technology, respectively. The change of anchor force was successfully simulated by changing the contact stiffness of the numerical model, and the detection of anchor force was tested on this basis. Li et al. [14] found that smart corrosion coupon (SCC) methods can accurately determine the corrosion of metal plates through experimental and numerical simulation studies based on peak frequency shift and peak variation of the conductance signal of PZT pasted on the metal plates. Luo et al. [15] combined magnetic materials with piezoelectric ceramic materials to design a PZT sensor with a magnetic base. The thickness of the magnetic base can not only affect the magnetic properties of the base, but also correlate linearly with the impedance resonance frequency of the magnetic sensor. To test the detection capability of the magnetic sensor, an experimental and numerical simulation study of bolt loosening monitoring was conducted. The experimental and numerical simulation results math well, and the location of the bolt loosening can be located using EMI measurements with the SMT. Wen et al. [16] proposed the EMI method to detect core concrete cavity defects of CFST columns. Considering the electromechanical coupling between PZT and the composite structure, an analysis model is established by ANSYS to analyse the impedance frequency curve of the composite structure with cavity defect and explore the mechanism of the EMI method for defect detection, which verifies the effectiveness and practicality of the method.

In this paper, an experimental study on the feasibility of EMI measurement-based void defects detection and visualization method for a large-scale CFST member in a steel box-CFST composite tower structure of the Zhangjinggao Yangtze River Bridge was conducted. Void defects with different sizes under the inner horizontal diaphragm of the specimen are mimicked and EMI measurement at selected frequency band of surface-mounted PZT sensors are performed. Regions of void defects in CFST specimens were visualized using an interpolation algorithm based on RMSD results of EMI measurements from number of PZT sensors.

2 Void Defect Detection Principle Based on EMI Method Using Surface-Mounted PZT Sensors

The one-dimensional coupling system composed of PZT sensors and detected structure was proposed by Liang et al. for EMI analysis as shown in Fig. 2 [17]. The model explains the correspondence between the electrical impedance of PZT and the structural impedance of tested structure, and evaluates the damage condition of the tested structure by the coupling properties between the PZT and the tested structure. The EMI method uses the electrical impedance response of PZT to reflect the local damage of the tested structure. The damage causes change in the local characteristics (mass, stiffness) of the tested structure, which leads to change in electrical impedance for the purpose of defect detection.

Fig. 2.
figure 2

EMI measurement system

Full size image

The RMSD value of the impedance signals of each sensor is calculated for visualizing the location of the mimicked void defects in CFST specimen.

$$ RMSD_i = \sqrt {{\frac{{\sum_{j = 1}^N (Z_{i,j}^1 - Z_{i,j}^0 )^2 }}{{\sum_{j = 1}^N (Z_{i,j}^0 )^2 }}}} \times 100{\text{\% }} $$
(1)

where, \(Z_{i,j}^1\) represents the impedance test data of the i-th sensor at the j-th frequency, N represents the total sampling points of the impedance analyzer in the test frequency band, and \(Z_{i,j}^0\) is the average of the impedance of PZT sensors at healthy positions at the j-th frequency.

The damage indexes adapted to the EMI are established using the RMSD, which can effectively reflect the data differences between measurement points at different locations in different states to obtain the damage of the structure. On this basis, the defect range is visualized according to the RMSD at different locations. Equation (1) expresses that if there are defects in the detection location, the greater the impedance difference with the healthy location, the greater the RMSD will be. If the detection location is in good condition, the impedance match with the healthy location and the RMSD difference is not large.

3 Full-Size CFST Specimen with Mimicked Void Defects and PZT Sensors Arrangement for EMI Measurements

3.1 Full-Size CFST Specimen

The diameter and height of the full-scale CFST specimen are 3600 mm and 3000 mm, respectively. The thickness of the steel tube is 30 mm. The thickness of horizontal diaphragm and vertical plates are 20 mm. To secure the bond between concrete and steel tube, shear studs are welded in the inner wall of the steel tube, which will effectively reduce the debonding phenomenon between concrete and steel tube wall caused by concrete shrinkage. C60 self-compacting concrete was poured into the steel tube using the high drop method of pouring concrete. Full-scale CFST member on site and the cross section are shown in Figs. 3 and 4. The impedance analyser is used to excite the sensor on the surface of the structure and obtain the electrical impedance data.

Fig. 3.
figure 3

Full-scale CFST specimen

Full size image
Fig. 4.
figure 4

Cross section diagram

Full size image

3.2 Artificially Mimicked of Void Defects

In order to mimic void defects between the horizontal diaphragm and concrete core, number of empty wooden boxes were pasted with epoxy resin on the lower surface of inner horizontal diaphragm at designed locations before concrete pouring. The empty wooden boxes were made of 3 mm thick wooden board material, and then filled with epoxy resin glue to fill the gap of the board to avoid the flow of concrete into the empty box. The dimensions of the mimicked void defects and the labels are displayed in Table 1. Each void defect corresponds to a healthy location under the same boundary conditions with the numbers KH1, KH2, KH3, KH4, and KH5, respectively. Some examples of the mimicked empty boxes installed under the horizontal diaphragm are shown Fig. 5. Concrete pouring is carried out after the installation of mimicked void defects is completed.

Table 1. Dimensions of mimicked void defects.
Full size table

To detect the region of five void defects KD1, KD2, KD3, KD4, and KD5, five surface-mounted PZT sensors are installed, one PZT sensor is installed at the centre of the mimicked defect and other four PZT sensors are located at 2 cm outside of the boundaries of the void defect. Taking the void defect KD4 as an example, sensor installed at the canter of the defect is named KD4, and additional four PZT sensors, named as KD4-1, KD4-2, KD4-3 and KD4-4, are located at 2 cm outside of the four boundaries of the mimicked defect as shown in Fig. 6. For comparison, surface-mounted PZT sensor labelled as KH are placed at healthy locations where no void defect is mimicked and are represented by cross lines as shown in Fig. 7. After the PZTs were welded with the wire, the negative electrode of the PZT is fixed on the surface of the structure by epoxy resin.

Fig. 5.
figure 5

Mimicked void defects in CFST specimen

Full size image
Fig. 6.
figure 6

PZT sensors installed at defect KD4

Full size image
Fig. 7.
figure 7

Locations of mimicked void defects and the PZT sensors at healthy regions of the expended view of the steel tube (unit: mm)

Full size image

4 EMI Test Results and Void Defect Detection and Visualization

4.1 Impedance Curve Analysis

According to previous studies, the appropriate frequency range is chosen to measure the impedance of the sensor [16]. Using the test system based on the impedance method, the sensors at different locations are tested separately and the impedance curves of each measurement point are obtained. EMI measurements of the sensor installed at the center of the artificially mimicked void defect were compared with the average value of all five sensors KH1, KH2, KH3, KH4 and KH5 at the healthy location, as shown in Fig. 8. The impedance curve of the average of the five PZT sensors at healthy conditions is smooth. Nevertheless, the impedance curves of the five PZT sensors KD1, KD2, KD3, KD4 and KD5 installed at the center of five void defects have obvious fluctuations and new peaks. The void defect can be detected successfully based on the EMI measurements. The EMI measurements collected from all the PZT sensors were used for RMSD calculations and results show that the RMSD of the sensors installed at the center of void defects are significantly higher than those at the healthy locations as shown in Fig. 9. The defect state at the tested location can be verified clearly by the impedance Z and the corresponding RMSD.

Fig. 8.
figure 8

Impedance of PZT sensors

Full size image
Fig. 9.
figure 9

RMSD values.

Full size image

4.2 Void Defect Visualization

The RMSD of the five PZT sensors around the mimicked defect KD4 shown in Fig. 10. The RMSD corresponding to the sensor KD4 installed at the center of mimicked void defect is significantly higher than the RMSD values of the other four PZT sensors located at 2 cm outside of the void defect boundaries. Based on the calculated RMSD values of the five PZT sensors, interpolation was used to calculate the RMSD values of the region covered by the five PZT sensors around each void defect. The position where the test piece is not arranged with sensors for measurement is by default the healthy position. The matrix of RMSD values at different locations was mapped to a color matrix using the colormap function in MATLAB software. The location and region of the mimicked void defects and the actual location shown in solid line in the expended view of the steel tube are shown in Fig. 11. Figure 11 depicts the expansion diagram of the whole specimen with the circumference of the specimen in horizontal coordinates and the height of the specimen in vertical coordinates. The locations and sizes of the five void defects were imaged by the RMSD results corresponding to all the sensors arranged around the artificially mimicked void defects. The mimicked void defect can be identified and visualized successfully.

Fig. 10.
figure 10

RMSD values of sensors around void KD4.

Full size image
Fig. 11.
figure 11

Imaging of detected void defects.

Full size image

5 Conclusion

In this paper, a void defect detection and visualization methods for a large-scale CFST specimen using EMI technology using surface-mounted PZT sensors is proposed and the validity is verified with a full-scale CFST specimen.

  1. (1)

    The existence of void defect makes the impedance curve of the surface-mounted PZT sensor located at the center of the defect fluctuate significantly, and the corresponding RMSD value is obviously greater than the RMSD value of PZT sensors at healthy location. By comparing the EMI measurement results of the surface-mounted PZT sensors, the existence of void defects can be detected efficaciously.

  2. (2)

    Using an interpolation algorithm based on the EMI measurements around a mimicked void defect, the location and region of the void defects can be visualized and the imagination results are consistent with the artificially mimicked void defects in the CFST specimen.

    The effectiveness of using EMI technique for the detection of concealed internal cavity defects is verified, and the EMI method has significant development prospect in the field of defect detection in large-section CFST structures. The quantitative analysis of cavity defects is still challenging research, and new data mining techniques combined with machine learning techniques are used in future research to perform deeper analysis of obtained impedance data.