According to past research, the decrease of the propagation speed of elastic waves show a proportional relationship with the existence of fragment in the rock mass [5] . Hence, the change of the elastic wave velocity under the different fragment condition was investigated with a plaster model in order to evaluate the influence of fragment in the rock mass on the P wave velocity. As shown in Figure 1, simulated rock with fragment was prepared by arranging a fixed number of gypsum artificial speciments in series, and P wave velocity passing through them was measured. Table 1 shows physical properties of gypsum samples used for this tests.

Figure 2 shows the relationship between the number of fragments and P wave velocity. From this figure, a good correlation is found between the number of fragments and the P wave velocity, and the elastic wave velocity decreases as the number of fragments increases. From this result, the impact of fragment existence in the measurement section could be quantitatively evaluated by investigating the P wave velocity and its attenuation behavior.

Blast-induced ground vibration is affected by the topography and geological features of a mine site as well as charge weight and distance from the blast source [6] [7] . Hence, it is very difficult to develop a common blasting prediction equations or methods for different locations and/or geological conditions. Thus, it is necessary to comprehend the propagation behavior of blast-induced ground vibration and develop the specific prediction formula in each mine site based on the results of field measurements.

A series of field tests was conducted at I Mine in Kagoshima prefecture, Japan. In these experiments, ammonium nitrate fuel oil (ANFO) was used as an explosive. Blast-induced ground vibration was monitored with a TEAC 707z three axial piezoelectric accelerometer (max acceleration 15 g, frequency response 3 to 5000 Hz, sensitivity 100 mV/g, g = 9.8 m/s²). In addition, a TEAC SA-611 amplifier and a LX-100 recording unit were also used. This digital recording unit can record low frequency signals and analyze waveforms with A/D converter [8] . One of the example layout of the field experiment is shown in Figure 3. In I Mine, hole spacing and its diameter are 2 (m) and 76 (mm), respectively. Burden is usually 2.5 (m). In addition, bench height of I Mine is 10 (m) and it angle is 80˚. Although the powder factor is depended upon operation situation, it is usually 170 (g/t). In this mine, the delay is 25 (ms).Typical blasting standard of I Mine is shown in Table 2 [9] .

The vibration sensor was installed along a traverse line set perpendicular to the row of blast holes. In these experiments, the Cartesian coordinate system is defined by the X axis parallel to the bench plane, the Y axis perpendicular to the bench plane, and the Z axis perpendicular to the ground. In addition, the piezoelectric element type pickup used in this experiment has advantages in high frequency and wide dynamic ranges measurement of vibration over the velocity type geophones or other seismometers for high sensitivity vibration since it is effective especially when examining the damping effect of vibration near the explosion source. By converting the data obtained as voltage (V) into vibration acceleration and integrating the acceleration waveform, it is possible to obtain the vibration velocity waveform and the peak particle velocity (PPV) [2] . Three to

five vibration measurement points were set from the center of the row of blast holes toward the back of the bench. Figure 3 is one of the example of a layout of sensor points for the ground vibration measurement. In order to discuss the damping tendensy of elastic wave, each sensor was arranged as far as possible (the distance between each sensor was at least 15 m).

P wave velocity propagating in rock mass was calculated based on the time difference of arrival of the blast-induced ground vibration wave between vibration sensors in the field test. Figure 4(a) shows the relationship between the fragment density and the ratio (v₁/V₁), where v₁ is P wave velocity of the sample containing the fragments obtained in the laboratory test and V₁ is the that of the intact rock sample. Moreover, Figure 4(b) shows that the relationship between the fragment density, measured by the number of cracks per unit length on the blasting face in the field test [9] , and the ratio (v₂/V₂), where v₂ is P wave velocity of the rock mass measured by accelerometer and V₂ is P wave velocity of the intact rock sample. Althogh the value is largely defferent due to difference of the length of measurement section and material, or effect of refelction wave in the indoor tests, these figures show similar tendency and it can be seen that fracture density increases with decreasing the value of the ratio. That is to say, fracture conditions in the rock mass can be estimated by evaluating the ratio of P wave velocity of rock mass and intact rock.

From the results of blast-induced ground vibration measurement at laboratory and operation site tests, it became clear that the elastic wave velocity decreases due to the existence of fragmentation intervening in the rock mass. Therefore, for more detailed discussion, we simulated the propagation behavior of the blast-induced ground vibration by means of two-dimensional finite element analysis code LS-DYNA [10] in order to investigate the influence of the fragment in the rock mass. Figure 5 shows the analysis model, and Table 3 shows the input physical property values for elastic material models.

Generally, the stress wave around the explosion source is expressed by the pressure function of the Equation (1) [11] .

P ( t ) = P 0 ξ { ( A t ) − exp ( − B t ) } (1)

Here, P(t) is the pressure (MPa), P₀ is the maximum pressure (MPa), and ξ, A, B are constants. At that time, assuming that 10% of the detonating pressure contributed to the vibration, and the velocity of detonation of the ANFO explosive was estimated as 3000 (m/s) in consideration of the energy consumed by such a rock breaking. Based on the detonation velocity of ANFO and length of explosive charge, the time of reaching peak pressure is inferred 2 (ms), and A = 375 B = 675 are determined. Figure 6 shows the pressure variation with time at blasting hole.

Moreover, the attenuation of blast-induced ground vibration is determined by its minimum frequency in this study. Therefore, Fast Fourier Transform (FFT) analysis was performed on the velocity waveform obtained from on-site test measurement, and the velocity-frequency spectra of each direction component at observation point was obtained. Figure 7 shows one of the result of the FFT analysis. Based on the results, the minimum vibration frequency of the vibration waveform was determined as 12 Hz, and the attenuation factor corresponding to it was set.

In this study, the Young’s modulus of the rock with excellent fragment was calculated using Equation (2) [5] in order to compare and comprehend the propagation behavior of stress wave under different rock mass conditions.

E C = E S ( v 2 V 2 ) 2 (2)

Here, E_c is the Young’s modulus of the rock mass, v is the P wave velocity measured by accelerometers in the field test, E_s is the Young’s modulus of the

intact rock sample.

Figure 8(a) and Figure 8(b) show pressure distribution diagrams of stress waves at 24 (ms) after detonation under the intact rock and rock mass condition, respectively. From both figures, it became clear that the velocity of stress wave becomes small under low Young’s modulus condition. This tendency has good accordance with the result of laboratory and field tests. In other words, Young’s modulus of rock mass become smaller than that of intact rock because of the existence of fragments.

As a next step, Figure 9 and Figure 10 show the relationship between scaled distance and PPV obtained from field experiment and numerical simulation,

respectively. Comparing the slope of the approximate linear, which means the damping coefficient of the blast-induced ground vibration, the damping coefficient obtained from numerical simulation considering the existence of fragments, −2.01, has good accordance with that obtained from field experiment measurement, −2.12. Based on this result, it can be concluded that the propagation behavior and PPV can be more accurately simulated and predicted by considering fragment condition in the rock mass. In other words, the PPV might be efficiently predicted by conducting test blasting, leading to economical mining operation. Although further test blasting in other mines or quarries, this prediction method is P wave based prediction method and this technique might be applied every mine or quarry.