Acoustic evaluation of concrete delaminations using ball-chain impact excitation

Corrosion induced delaminations are common defects in concrete bridge decks. Chain drag is the most popular method to detect shallow delaminations by dragging a chain over it and listening the “hollow” sound generated by delaminated concrete vibration. A major drawback of chain drag is that it relies on subjective interpretation of the inspector. In addition, this method does not provide achievable data, and is affected by traffic noise. It is also time-consuming to map the delamination areas after test. Henderson et al. (1999) and Costley et al. (2003) reported an automated chain dragging system that includes chains, a microphone, and signal acquisition and processing components. Acoustic signals were recorded continuously and an odometer recorded the position of test. If working properly, the automated system will be more efficient and consistent than the manual chain drag test. The Impact-echo (IE) test is another common NDT method which measures surface vibration using a contact sensor and gives the depth information of delaminations. Zhu and Popovics (2007) proposed air-coupled IE test by using a microphone to replace the contact sensor. Although the air-coupled sensing method significantly improves the IE test speed, the contact impact source remains a challenge for rapid scanning. Researchers have attempted to develop automated IE system to increase the efficiency and reduce cost (Popovics, 2010; Mazzeo et al., 2016). However, these systems need complicated electrical and mechanical controls to give consistent impacts.

Chain drag is a simple acoustic source for continuous excitation on concrete surface. However, when analyzing the acoustic signals generated by chains on concrete, the authors found that the chains generate broadband acoustic noises that may lead to misjudgment. In this study, the authors combined the advantages of impact-echo and chain drag, and developed a new ball-chain impact source to replace the traditional steel link chains. This letter presents a near-continuous acoustic method for detecting shallow delaminations in concrete using the new impact source. Experimental studies demonstrated that the ball-chain generate acoustic signals similar to the impact-echo signals, and therefore gives higher signal-to-noise ratio (S/N) than traditional link chains. It is found that the ball-chain jumps and impacts on concrete during dragging, while the link-chain mainly slides on concrete surface.

To compare the conventional link chains used in the chain drag test and the newly developed ball-chain impactor, we acquired acoustic signals using a steel link-chain and a ball-chain on concrete surface in solid and delamination areas, respectively. The link-chain is a galvanic steel chain with diameter of 6.35 mm (1/4 in.), length of 45 mm, and weight of 28 g per link, while the ball-chain consists of multiple 12.7 mm (1/2-in.) and 15.9 mm (5/8-inch) diameter brass balls with a 10 mm spacing. Using different sizes of balls will be able to detect delaminations with different frequencies. A larger ball is more effective for delamination with lower frequency. In each test, signals of 1.0 s duration were collected using a MEMs microphone (Adafruit, SPW2430) and a PicoScope 4824 oscilloscope. Each signal was processed by short-time Fourier Transform (STFT).

Figure 1 shows the STFT spectrograms of signal by dragging a link-chain and a ball-chain on solid and delamination concrete. On solid concrete surface, the link-chain signal [Fig. 1(a)] shows a dominant 15 kHz noises which comes from the clapping between chain links, while the ball-chain gives a very clean signal on solid concrete surface [Fig. 1(b)]. According to the semi-analytical equations of resonance frequencies for square delaminations by Kee and Gucunski (2016), the frequency range 0.5 to 5 kHz covers flexural responses of delaminations with a depth of 0.02 m ∼ 0.08 m and width of 0.2 m ∼ 1 m. When dragging a link-chain and a ball-chain on delamination area, STFT spectrograms of the signals are shown for the frequency range of 0.5 to 5 kHz, in Figs. 1(c) and 1(d). In order to compare test results with delamination locations, a laser distance sensor was used to record the real-time test positions. In Figs. 1(c) and 1(d), the horizontal axes are shown as the corresponding test distance, and the delamination boundaries are marked as two straight lines. Over the delamination (0.46 × 0.46 m², 38 mm depth), both the link-chain and ball-chain excited the resonant frequencies of the delamination. The link-chain signal shows higher amplitude than the ball-chain signal, which is probably caused by larger mass of the link-chain. However, the ball-chain's image clearly shows three resonance frequencies, while the link-chain's image has high level of noises. These noises may mask the indication of delamination or give false positive indication of delamination.

Slow motion videos were used to investigate behaviors of the steel link-chain and ball-chain when they were dragged on concrete surface. A GoPro camera was used to capture footages of dragging both chains on concrete surface at a frame rate of 240 frames per second (fps), then videos were played at a frame rate of 10 fps (1/24 original speed). In the slow-motion videos, it can be found that the link-chain mainly slid on the concrete surface (Mm. 1), while the ball-chain jumped and impacted on the concrete surface periodically (Mm. 2).

By comparing the ball-chain's video and its time domain signal, we confirmed that each ball impact generated a signal similar to a single impact-echo signal Fig. 2. In the STFT spectrogram, these impacts give multiple vertical highlight strips [see Fig. 1(b)]. The sliding friction between a link-chain and concrete leads to high noise level that may affect signal interpretation.

Moving speed will affect testing results on two aspects: spatial resolution and impact energy level. Spatial resolution is defined by the distance between two adjacent impacts. Increasing ball-chain dragging speed will increase the ball impact energy on concrete surface and amplitude of acoustic signals, but it may lead to poor spatial resolution due to large spacing between two impacts. In this study, a 12.7 mm diameter brass ball was dragged on the concrete surface at four different speeds (0.25, 0.50, 0.75, and 1.0 m/s). The spatial resolution is calculated as the total dragging length divided by the number of impacts. The number of impacts per meter are 73, 48, 31, and 14 for dragging speeds of 0.25, 0.50, 0.75, and 1.0 m/s, respectively, which correspond to spatial resolutions of 1.3, 2.1, 3.2, and 7.1 cm for the four dragging speeds. The peak voltage of each impact signal was measured and converted to sound pressure level (SPL) using the microphone's sensitivity (12.9 mv/Pa). The average SPLs at the four dragging speeds are 83.6, 87.4, 91.4, and 97.5 dB. Therefore, there is a tradeoff among testing speed, spatial resolution, and signal amplitude. On the basis of our experience, a speed of 0.6 m/s is a good balance of spatial resolution and impact energy.

In order to improve testing speed and signal amplitude without scarifying the spatial resolution, we connected multiple balls in parallel as the excitation source. Each ball will impact the concrete surface randomly and independently. This design increases the total number of impacts per unit length. On the basis of experimental data, at a speed of 0.86 m/s, the spatial resolutions were 4.3, 2.8, 2.3, and 1.8 cm when using 1, 2, 3, and 4 balls, respectively. It is found that the spatial resolution does not increase linearly with the number of balls. Since the ball size affects frequency content of impact, we combined two small size balls and two large size balls (such as two 12.7 mm balls and two 15.9 mm balls) to cover delaminations with different fundamental frequencies.

Both link-chain and ball-chain were used to scan a concrete specimen with four artificial delaminations. The dimensions and depth of delaminations are marked in Fig. 3. Resonance frequencies of the delaminations range from 1.4 to 2.9 kHz. A multichannel scanning frame was built to install six chains and six MEMs microphones at 200 mm spacing. In order to improve lateral resolution, the chain/sensor locations were shifted by 50 mm in lateral direction in each scan, and three shifted scans were performed. In total 24 channels of signals were collected. Each time domain signal was first processed by STFT to generate a spectrogram for each channel, as shown in Fig. 1. Then STFT amplitudes in the frequency range from 0.5 to 5 kHz were summed up to form one dimensional data set vs scanning distance for each channel, and then data sets from all channels were stacked together to generate a two-dimensional map, with two axes representing dimensions in longitudinal (scanning) and transverse directions.

Figure 3 shows the scanning images using the link-chain [Fig. 3(a)] and the ball-chain [Fig. 3(b)]. Both types of chains detected shallow delaminations (25 mm depth) #1 and #2 clearly. In Fig. 3(a), many false positives present on the solid area near delamination #3 (37.5 mm depth), due to very rough surface in this region. The ball-chain image shows cleaner image and more accurate localization for delamination #3. The link-chain failed to detect delamination #4 (50 mm depth), which was shown difficult for manual chain drag test too. Detailed analysis indicates the red spot in delamination #4 [Fig. 3(a)] is likely caused by rough surface noise instead of delamination response. In the ball-chain scanning image [Fig. 3(b)], delamination #4 can still be identified although the delamination response is weaker than other defects. The ball-chain gives higher S/N and is less sensitive to the surface roughness than the link-chain. It should be noted that these artificial delaminations appear more rigid than actual delaminations in bridge decks. Experience indicates that actual delamination with similar dimensions as #4 (50 × 50 × 5 cm³) can be detected by a manual chain drag text.

In this letter, a near-continuous impact source is reported for acoustic scanning of concrete structures. Comparison of testing results from a link-chain and a ball-chain shows that the ball-chain signals have higher S/N and similar to the impact-echo signals. Slow motion videos reveal that the ball-chain gives periodical impacts during dragging, while a link-chain mainly slides on surface. The dragging speed affects the spatial resolution and impact energy. Using multiple balls connected in parallel can increase test speed without sacrificing spatial resolution or signal amplitude. A concrete specimen with artificial delaminations was scanned by a multichannel acoustic system using both link-chains and ball-chains. The results indicate that the ball-chain is an effective acoustic source which gives high S/N and is not affected by surface roughness.

This study was supported by the Nebraska Department of Roads (NDOR).