*3.1. Characteristics of AE Time Series Evolution of Coal under Deep Mining Conditions*

Of the AE data obtained by simultaneous real-time monitoring, the cumulative ring count and ring count rate reflect the frequency of AE occurrence, and the cumulative energy and energy rate reflect the intensity of AE occurrence. Figure 6 shows the variation in the AE ring-down count rate (a1,b1,c1), cumulative ring-down count (a2,b2,c2), energy rate (a3,b3,c3) and cumulative energy (a4,b4,c4) over time under the simulated deep mining condition at a depth of 1000 m. In Figure 6a–c correspond to protective seam mining, top-coal caving and nonpillar mining, respectively.

**Figure 6.** *Cont*.

(c) Nonpillar mining

**Figure 6.** Evolution of the AE ring-down count rate, cumulative energy, energy rate and cumulative ring-down count over time.

For rocks with prefabricated cracks, the crack initiation point and propagation process are the usual targets of research [37,38]. For cylinder specimens without macroscopic cracks, Cai et al. [27] concluded that rock experiences four stages: crack initiation, microcrack formation, crack coalescence and macrocrack formation under compression conditions. Coal has a complex structure with viscoelastic characteristics such as instantaneous deformation, elastic hysteresis and irreversible plastic deformation. Hence, its mechanical behavior is also complex, and cannot be simply described by the five-stage constitutive rock model. Studying the characteristics of the AE time series evolution curve can make the research results of the crack initiation point and propagation process more convincing. Figure 6 shows that the overall AE activity exhibited three characteristic stages over time. During the first stage of AE activity, the overall AE activity was at a low level. In the second stage, the amplitudes of the AE ring-down count rate and energy rate showed a significant increase compared to those in the previous stage on average, and began to exhibit a cluster characteristic with time, but had a small range of fluctuation (see Figure 6 (a1,a3,b1,b3,c1,c3)). The AE cumulative energy and cumulative ring-down count began to increase slowly (see Figure 6 (a2,a4,b2,b4,c2,c4)). In the third stage, the amplitudes of the AE ring-down count rate and the energy ratio changed abruptly compared with those in the previous two stages on average, and exhibited an increasingly strong clustering characteristic with time. In this stage, the cumulative energy curve began to show a relatively high step, increasing by an order of magnitude in a few seconds. It can be concluded that the three variation stages of AE activity correspond to the three processes (i.e., microcrack initiation, stable propagation and unstable propagation) inside the coal.

Statistical analyses were performed on the characteristic points of the three AE activity stages of coal under three mining layouts, as shown in Table 1. The starting point for each of the three AE activity stages corresponding to the three mining layouts (i.e., nonpillar, top-coal caving and protective seam) is associated with an increasing stress level, allowing the occurrence timing of rupture initiation and main rupture activity under the above mining layouts to be ranked in ascending order.

Based on the three stages of AE activities quantitively divided previously, the stress-strain curves can also be divided into three parts. Figure 7a–c shows the relationship between the released AE energy and the axial strain at the three characteristic stages under each of the three mining layouts plotted in the same coordinate system. Combined with Figure 7d, the following can be observed. In the first stage of AE activity, the axial stress and strain under each of the three mining layouts were in an elastic range; the magnitude of released AE energy was on the order of 9.31 <sup>×</sup> <sup>10</sup>−<sup>18</sup> to 10−<sup>20</sup> J. In the second stage, the coal deformation progressed from the elastic stage to the elastoplastic deformation stage; the magnitude of released AE energy increased to the order of 9.3 <sup>×</sup> <sup>10</sup>−<sup>15</sup> to 10−<sup>17</sup> J. In the third stage of AE activity, the stress-strain relationship curve showed an increase in curvature and exhibited a stress plateau, indicating plastic deformation of coal in this stage; the released AE energy was in the range of 10−<sup>15</sup> to 10−<sup>17</sup> J. Therefore, it can be seen that, for each of the three mining layouts, the AE energy was mainly released during the second and third stages of AE activity. Thus, these two stages can be considered the main stages of coal rupture.

**Figure 7.** AE events, ring-down count, and energy under different mining layouts.


**Table 1.** Statistics of stress and time for typical stages of AE activity under different mining layouts.

#### *3.2. Spatial Evolution and Fractal Characteristics of AE of Coal under Deep Mining Conditions*

According to the defined standard for the three-stage evolution of the AE time series, Figure 8 shows the spatial relationship between the AE event and the strain under different mining layouts. In the first stage, AE events were distributed inside the sample, as shown in Figure 8a. At the end of the second stage, the coal material began to yield, and a main AE cluster area formed, as seen in Figure 8b,e. In the subsequent evolution process, the generated AE was concentrated in the main cluster area, as seen in Figure 8c,f, corresponding to the main coalescing area of coal cracks, as shown in Figure 8d. After the completion of the entire experiment, the cumulative morphological distribution of AE spatial localization had a consistent corresponding relationship with the macrocracks in the ruptured coal mass experiencing the complete mining process. The coal failure mode under each of the three mining layouts can be preliminarily determined from Figure 8d as follows: The sample corresponding to the protective coal-seam mining failed under compression, and the coal corresponding to top-coal caving mining and nonpillar mining failed under tension.

Fractal geometry provides a quantitative method for testing spatial morphology. Hirata et al. and Zhang et al. [39,40] used AE localization tests to verify that the spatial distribution of AE events has fractal characteristics. The evolution process of the AE spatial distribution under the influence of coal mining can be viewed as a transition from a disorderly to an orderly state, and the fractal dimension provides us with an accurate order parameter describing the AE activity during such a transition.

The general formula for solving fractal dimension *D* is

$$
\log \text{N} = \log a - D \log \delta,\tag{1}
$$

Boxes of different sizes δ were used to cover the study objects, and the total number of boxes covering the study objects was denoted as *N*. A set of (δ, *N*) data obtained in the covering process is presented in a log-log plot, and the slope is the fractal dimension *D*.

The cylinder covering method [41] was used for the fractal measurement of AE spatial localization. The mass center of the cylindrical sample was used as a base point, and a small cylinder (with a radius of *r* and a height of *h*) proportional to the sample with respect to the height-diameter ratio was taken. By simultaneously increasing *r* and *h*, the number of AE localization points covered by each cylinder was counted, and then Equation (1) was used to obtain the cluster dimension of AE spatial location points.

Figure 9 shows the relationship between the AE spatial fractal dimension and the peak stress at the coal peak stress location plotted in the same coordinate system. As the peak stress increased, the AE spatial fractal dimension decreased successively under the protective seam, top-coal caving and nonpillar mining layouts.

In a deep mining process, under the influence of confining pressure unloading (Figure 9), the spatial microcracks were clustered into two-dimensional (2D) variants, resulting in a relatively small fractal dimension corresponding to the peak stress. At the peak stress location, nonpillar mining corresponds to the lowest fractal dimension, which is closest to the 2D stress state, indicating that the nonpillar mining layout resulted in a higher degree of unstable coal mass rupture in front of the working face. The higher degree of coal mass rupture indicates a lower energy accumulation, and implies that the possibility of coal burst risk tends to be low, which coincides with views, based upon practical applications, that nonpillar mining is a relatively safe mining scheme to reduce rock burst likelihood. In nonpillar mining, the working face outside the seam line of the pillar is not reserved, and the roof strata will collapse to the goaf, generating roadside support under the influence of the gangue retaining system behind the support. That process also releases the accumulative energy in the surrounding rock, thereby reducing the risk.

(a)The first stage of AE activity (time) (b) The second stage of AE activity (time)

time)

(d) Corresponding relationship between surface cracks and the spatial cumulative distribution of AE after sample failure

(f) The third stage of AE activity (period of time)

**Figure 8.** Spatial distribution of AE under different mining layouts.

**Figure 9.** Relationship between AE spatial fractal dimension and stress under different mining conditions.

#### **4. Conclusions**

Based on three typical mining layouts, i.e., protective coal-seam, top-coal caving and nonpillar mining, the coal mining process at a depth of 1000 m was simulated through mechanical tests on a laboratory scale, and simultaneous AE localization of this process was carried out to study the AE characteristics under deep coal mining conditions. Based upon this, the following conclusions were drawn:


**Author Contributions:** Y.Y., T.A. and Z.Z. contributed to conducting the experiments and writing the manuscript. R.Z. and L.R. modified the writing of the manuscript. J.X. and Z.Z. assisted in the analysis of AE data. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Department of Science and Technology of Sichuan Province (CN) (Grant No. 2017TD0007) and the National Natural Science Foundation of China (Grant No. 51622402).

**Acknowledgments:** The authors thank anonymous colleagues for their kind efforts and valuable comments, which have improved this work.

**Conflicts of Interest:** The authors declare no conflict of interest.
