ABSTRACT

In this study, Impact Acoustics parameters obtained from received sound generated by impact on concrete surface were investigated to develop a evaluation system of defects in concrete. As Impact Acoustics parameters, frequency distribution was employed. In addition to the experiment, 3 dimensional FEM analysis was carried out to understand the theoretical background of parameters. The results of FEM analysis showed good agreement with experimental ones. From analytical and experimental results, it was likely possible to estimate defect sizes using the relation between the resonance frequencies of impact sounds and defects diameter.

KEYWORDS
Impact Acoustics Methods, frequency distribution, FEM analysis, flexural resonance

INTRODUCTION

Tapping methods that receive sounds with human ears have been used in practice to inspect the concrete linings of railway tunnels [1]. However, such methods completely depend on the detection of "unclear sounds" to identify internal defects. Evaluation by human ears is greatly affected by the experiences and subjectivity of inspectors.

In order to solve these problems, methods using acoustic device such as microphone to receive sounds and to analyze the waveforms have been studied. The name "Impact Acoustics Methods" is introduced as a definition of such type of methods in a committee report [2] by Japan Concrete Institute.

This study investigated Impact Acoustics Methods that can quantitatively identify the sizes of internal defects and their depths from the surface using the frequency distributions. Experiments were conducted using concrete specimens with an artificial defect inside, which represented voids or delaminations inside concrete. Numerical analyses of the three-dimensional finite element method (FEM) were also conducted to analyze the relationship between frequency distribution and defect information. The frequency distributions were compared between the without defects and with defects concrete sections. The displacements of the concrete surface were monitored using an accelerometer to investigate the phenomena involved in impact acoustics.

OUTLINE OF EXPERIMENTS

Specimens
To evaluate frequency distributions, concrete slab specimens (length: 400 cm, width: 200 cm, thickness: 20 cm) that contained of a disk shaped artificial defects (styrene, thickness: 0.5 cm) were prepared. Photo. 1 shows the example of the specimen used in this study. The mixture proportions and mechanical properties of concrete are shown in Table 1. This size was likely to be large enough to not be affected by elastic wave reflection from the sides. The diameters of artificial defects were 5, 10, 15, 20, 30 and 50 cm. The depths of the defects were 3,5,7 and 10cm from the surface. Defects were arranged as shown in Photo. 2 and Fig. 1. The surfaces of the specimens other than the defect embedded portion are hereinafter referred to as "the sound portions".

Fig 1: Location of defects inside the concrete specimen (Ex. defect depth: 3cm and 5cm).	Photo 1: Concrete slab specimen.
	Photo 2: Arrangement of defects.

Impact method
Elastic wave was induced by impacting the surface of concrete using steel ball drop. Drop height was set at 10cm from the concrete surface. The diameter of steel ball was 19.05mm. Impact point was at the center of the concrete upper side of the embedded defects as shown in Photo. 3.

Photo 3: Test setup.

Elastic wave measurement
A condenser microphone (which has a flat sensitivity at 20 Hz to 30 kHz) was used to measure the radiating sounds during impact application. Displacement of the concrete surface was monitored using an accelerometer (frequency range: 0.1 to 45 kHz). The accelerometer was attached 5cm away from the impact point and microphone was positioned 10 cm above the point where the accelerometer was attached. The waveforms received by the microphone and the accelerometer were transmitted through an amplifier and A/D converter to a personal computer and were recorded. The frequency distributions were then derived by fast Fourier transform (FFT). Test setup for the measurement is shown in Photo. 3.

OUTLINE OF FEM ANALYSIS

Authors conducted analytical investigation using a general purpose program code (LS-DYNA). Fig. 2 shows the analytical model used in this study. Concrete part upper side of defect was modeled as a disc shaped plate. Concrete was assumed to be an elastic body (density: 2.3 x 10³ kg/m³, elastic modulus: 42 GPa, Poisson's ratio: 0.2), and all degree of freedom of the disc plate base was constrained in the analysis. Load was input as a wave shown in Fig. 3 to the center of the model. The duration of input Tc was derived using the following equation [3]:

(1)

where, D is the diameter of the steel ball (m). From Equation (1), the duration of impact to be used for the analysis was determined to be 80µs. The load F_max applied to concrete by dropping a steel ball was calculated using the following equation [4]:

(2)

where, m is the mass (kg) of the steel ball, h is the height (m) from which the ball was dropped, and g is the gravitational acceleration (m/s²). In this analysis, F_max was 0.88 kN.

Fig 2: Model for analysis.

Fig 3: Input waveform.

RESULTS AND DISCUSSIONS

Waveforms
The waveforms measured by microphone, accelerometer and calculated by FEM are shown in Fig.4 (diameter of defect: 20cm, depth: 3cm) and Fig.5 (diameter of defect: 20cm, depth: 7cm) respectively. Waveforms monitored in sound portion are also shown in Fig.6 for the comparison.

Fig 4: Waveforms (Diameter: 20cm, Depth: 3cm).

Fig 5: Waveforms (Diameter: 20cm, Depth: 7cm).

Fig 6: Waveforms (Sound portion).

From Fig.4 and Fig.5, it is obvious that the waveform obtained from sound portion and defects portion show quite different characteristics. In sound portion, duration of the waveform was shorter than that of defects ones. On the other hand, waveforms of defects portion show periodic history curves. With defects case, both defects depth is 3cm and 7cm, waveforms obtained from microphone and accelerometer show same tendency in terms of periodic history and attenuation characteristics. Tendency of waveforms obtained from FEM analysis show good agreement with experimental ones. In both experimental and analytical results, the greater the diameter of defect, the shorter the period.

Frequency distribution
Fig. 7 and Fig. 8 show the frequency distributions of impact sounds, surface displacement and analytical results at a depth of 3 and 7cm for defects of 20 cm in diameter. For comparison, frequency distribution of sound portion was also shown in Fig. 9 (only in experimental data). The figures show elastic wave frequency distributions that have clear single peaks and shapes that are completely different from those in the sound portions (Fig. 9). Intensities of each spectrum of defect case were much larger than that of sound one, which is thought to be the characteristics of defects part. However, in Fig. 9 (b), sharp and single peak is also appeared. This is not flexural resonance but longitudinal resonance corresponding to thickness frequency calculated by the following equation.

Fig 7: Frequency distribution (Diameter: 20cm Depth: 3cm).

Fig 8: Frequency distribution (Diameter: 20cm Depth: 7cm).

Fig 9: Frequency distribution (Sound portion).

(3)

Where f_t : thickness frequency(Hz), V: elastic wave velocity(m/s) and T: concrete thickness(m). In this study, V was 4500m/s, T was 0.2m. By using above values, f_t = 11.25kHz is obtained. This is close to experimental value. From Fig.7 and Fig.8, spectral peak obtained from microphone and accelerometer showed same value. This implies that the phenomena of sound generated by impact and surface displacement was equivalent. The larger the depth of the defects, the higher the frequency at the peak. This characteristic seems to show that the defect induced resonance inside the specimen, which caused a spectrum peak of impact sounds to appear. There are good agreements between the experimental and analytical results.

In vibrating disks, the flexural resonance frequency is known to increase as the depth of the disk increases if the disk area is constant [5]. The shift of the spectrum peak described in the above paragraph suggests that the peak component of the spectrum was due to the flexural resonance of the concrete above the defect.

Relation between peak frequency and defects diameters
Fig. 10 and Fig. 11 show the relation between resonance frequency and defects diameter obtained from experiment and FEM analysis respectively.

Fig 10: Peak frequency and defects diameter (Experiment) .

Fig 11: Peak frequency and defects diameter (FEM analysis) .

These figures show lower peak frequencies for larger defects and also show higher peak frequency for deeper defects. This characteristic can be observed both experiment and analysis. On the other hand, in the analytical results, resonance frequency of 10cm depth (equal to plate thickness is 10cm) show lower value than that of 7cm case except for diameter of defects are 10, 30, 50cm. This seems to be attribute to the difference between analytical assumption and actual condition of defects inside the concrete. In analysis, displacement of side in the model was not constrained. However, in experiment, when subjected to impact, the vibration of concrete part upper side of defects was not separated to around concrete. As a result, flexural resonance was not appeared in the analysis.

From Fig. 10 and Fig. 11, under same diameter of defects, the difference of resonance frequency caused by the increase of defects depth seems to be small. Taking into consideration of this relation, resonance frequency is thought to be effective index to evaluate the lateral expansion of the defects inside concrete.

CONCLUSIONS

The following conclusions were obtained from this study:

1. Waveforms obtained from experiment and calculated by FEM analysis showed good agreement. So model used in this study was proper.
2. Under the experimental conditions used, the resonance frequencies of the impact sounds and surface displacement mutually agreed in the defect portions, and they are likely to be equivalent.
3. Frequencies obtained from FEM are also good agreement with experimental ones.
4. To detect defects, it is effective to compare the frequency distributions of impact sounds for the defected and sound portions.
5. In this study, it is possible to estimate the size of a defect using a correlation diagram of resonance frequency and defect diameter.

REFERENCES

1. Nikkei Business Publications, Inc., NIKKEI CONSTRUCTION, No.240, pp.44-45, 1999.9
2. Japan Concrete Institute.TC 994: Technical Committee, Committee Report on Nondestructive Evaluation for Diagnosis of Concrete Structures, pp.72-80, 2001
3. M. J. Sansalone and W. B. Streett, Impact Echo,Bullbrier Press,Ithaca,N. Y. 1997
4. Shiratori M., Higai T. and Okamura Y.,"Analytical Study on the Response of Concrete Members for Light Impact Load", Proceedings of Japan Concrete Institute, Vol.14, No.1, pp.679-684, 1992
5. M.F. Ashby, Materials Selection in Mechanical Design, Butterworth-Heinemann, Oxford, 1992