The surrounding rock of the ore body is mainly granite. We conducted conventional mechanical tests on the granite to assess its tendency for rockburst, in order to scientifically control the potential risk of rockburst. The granite samples were taken from the 1,300–1,480 level of the Ruihai gold mine, and more than 30 granite blocks were obtained. In the laboratory, diamond drill bits were used to drill rock cores, which were then processed into the following specimens using a cutting and grinding machine: the sample size for uniaxial compression strength (UCS) test was 50 mm × 100 mm, and the size of for Brazilian tensile strength (BTS) test was 50 mm × 25 mm. The ends and sides of the specimens were carefully ground, and the non-parallelism and non-perpendicularity were both less than 0.02 mm. The accuracy of the specimens met the experimental specifications recommended by the International Society for Rock Mechanics (ISRM). The rock specimens are shown in Figure 1.

During the processing, it was found that the granite specimens showed three different colors, which may be three different types of granite, as shown in Figure 1. For the time being, we named them Granite A, Granite B, and Granite C, in order to further analyze the differences in rock types, X-ray diffraction (XRD) analysis was performed on the three colors of granite, and the analysis results are shown in Table 1.

From the analysis results in Table 1, it can be seen that the rock composition of the three colors of granite differs significantly, and it can be determined that they are three different types of granite. Therefore, when conducting rock mechanics experiments, they need to be classified for testing. A total of 28 standard rock specimens for uniaxial compression strength test were processed, including 9 Granite A, 5 Granite B, and 14 Granite C specimens. There were also 5 Brazilian tensile strength test specimens for Granite A and Granite C, respectively.

The equipment used in this experiment was the ZTRE-210 microcomputer-controlled rock triaxial testing system produced by Changchun Zhantuo Experimental Instrument Co., Ltd., as shown in Figure 2. Displacement control was used to apply loading to the rock samples for uniaxial compression test and Brazilian tensile test, with a loading speed of 0.0002 mm/s. The purpose of the uniaxial compression test was mainly to obtain the basic mechanical parameters of the rock and to determine its rockburst tendency, while the purpose of the Brazilian tensile test was mainly to obtain the tensile strength of the rock. Five standard specimens were selected for uniaxial compression strength tests for each type of granite, and five Brazilian disc specimens were selected for Granite A and Granite C for Brazilian tensile strength tests (there is no Brazilian tensile test samples for Granite B).

During the experimental process, varying degrees of rock powder and debris were observed to be ejected from all three types of granite, accompanied by sound and large rock fragments being ejected when sample failure approached. Overall, the above conditions were more pronounced in Granite C, followed by Granite A, and finally Granite B. The failure modes of the three types of granite are shown in Figures 3A, 4A, 5A.

Figures 3B, 4B, 5B show the typical full stress-strain curves of the three types of granite. It can be seen from the figures that the ultimate strain of the three types of granite is less than 3‰, with Granite B and C having an ultimate strain of less than 2‰, which is typical of brittle rocks. This can also be seen from the characteristics of rock failure in Figures 3A, 4A, 5A, where the rocks exhibit tensile and brittle failure. Due to the high strength and strong brittleness of the rocks, although the control of the loading speed at 0.0002 mm/s, only a small amount of post-peak deformation was observed in some of the rock samples. The basic mechanical parameters of Granite A, Granite B, and Granite C are shown in Table 2.

Using typical rockburst tendency criterions such as rock brittleness criteria (Xu and Wang, 1999), deformation brittleness coefficient (Zhang et al., 2017), and Barton criteria (Barton et al., 1974), a prediction of rockburst risk was made for the Ruihai gold mine. The expressions and discrimination intervals of the criterions are shown in Table 3.

According to the experimental parameters of rock mechanics in section 3.3, the rock brittleness criterion B and the deformation brittleness coefficient K _U were applied to determine the rockburst tendency of three types of granite. The determination results of granite A, B, and C are shown in Tables 4–6. It can be seen from Tables 4–6 that the rockburst risk of granite C is moderate to intense, the rockburst risk of granite A is slight to intense, and the rockburst tendency of granite B is slight to moderate. The rockburst tendency of granite C is greater than that of granite A, and the rockburst tendency of granite A is greater than that of granite B.

According to the in situ stress test results of the 2# drill hole of the Ruihai gold mine, the stress field in the exploration area is mainly dominated by horizontal stress, and the ratio of maximum horizontal principal stress S _H to vertical principal stress S _v is 1.06–1.41, with an average of 1.24. The ratio of the two horizontal principal stresses is 1.19–1.54, with an average of 1.37 (Peng and Sun, 2016). The maximum value of the horizontal principal stress is 12.61–39.74 MPa, and the minimum value of the horizontal principal stress is 10.64–28.89 MPa.

From Figure 6, it can be observed that the two horizontal principal stresses and the vertical principal stress increase with the increase of the overlying rock layer thickness or burial depth, and they show a basically linear relationship. The linear regression equations for the maximum (S _H) and minimum (S _h) horizontal principal stresses of each measuring point in drill hole 2 of the gold mine in the northern waters of Sanshan Island are as follows:

S _H = 0.74+0.033D

S _h = 1.08+0.022D

S _v = 0.00+0.027D

Where D is the depth of the drill hole (positive downwards), m; S _H is the maximum horizontal principal stress, MPa; S _h is the minimum horizontal principal stress, MPa; S _v is vertical principal stress, MPa.

Based on the fitting relationship between the stress and depth, it can be inferred that the maximum principal stresses in the −1,480 and −1,300 levels are 49.6 and 43.6 MPa, respectively. By applying the rock mechanics parameters and stress conditions to the Barton criterion, the rockburst risks in the two sections can be predicted, as shown in Tables 7, 8. From the tables, it can be seen that in the −1,480 level, Granite A has a moderate to intense rockburst risk, Granite B and Granite C have a moderate rockburst risk; in the −1,300 section, Granite A and Granite B have a moderate rockburst risk, and Granite C has a slight to moderate rockburst risk.

From the deformation brittleness coefficient, rock brittleness criterion, and Barton criterion for the rockburst tendency determined results of Granite A, Granite B, and Granite C, it can be seen that all three criterions indicate that there is a risk of rockburst occurrence in these three types of granite. However, it can also be observed that different criterions may lead to inconsistent or even different rockburst risk assessments for the same type of granite. For example, when using the deformation brittleness coefficient K _U criterion, granite C has the highest rockburst risk among the three types of granite, with a moderate to strong rockburst risk. However, when using the Barton criterion, granite C has the lowest rockburst risk among the three types of granite, with a slight to moderate rockburst risk at −1,300 level. Different criteria seem to lead to some contradictions in the rockburst risk levels assigned to the same rock. In fact, despite a series of explorations and studies by domestic and foreign scholars on rockburst, the understanding of the mechanism of rockburst and the warning of rockburst has been deepening. However, because rockbursts are an extremely complex dynamic phenomenon, there is still no unified understanding of the mechanism of rockburst.

Although there are many rockburst criterions at present, these criterions are mostly single factor criterion that are considered one-sided and do not consider the necessary and sufficient conditions for rockburst occurrence, namely, the external driving force of rockburst occurrence, i.e., stress conditions, and the internal conditions for rockburst occurrence, i.e., the inherent energy storage and consumption characteristics of the rock. For example, criterions such as the Barton criterion, Russense criterion, and Hoek criterion criterion only consider stress conditions, without considering whether the rock itself has a tendency to rockburst. Energy and rock brittleness criteria, such as rock brittleness criteria, deformation brittleness coefficient, impact index K _E, and strain energy index W _et, only consider whether the rock mass itself has a tendency to rockburst, without considering whether the external driving force of stress conditions for rockburst occurrence has been reached. Both of the above criterions may lead to misjudgments of the occurrence level of rockburst. For example, in the first type of situation, the external driving force may reach the conditions for rockburst occurrence, but if the rock does not have a tendency to rockburst (such as soft rock), then the rockburst will not occur. In the second type of situation, although the rock itself is judged to be prone to rockburst, if the external driving force does not reach the rockburst threshold, then the rockburst will not occur, either. The above stress factors (external conditions) and the rock’s own rockburst tendency (internal conditions) are both necessary conditions rather than sufficient conditions for rockburst. Rockburst can only occur if the above two conditions are met at the same time. Moreover, the above conclusion is only obtained at the experimental scale. Due to the complexity of engineering rock masses, such as local stress changes, variable rock structures, and geological conditions, the above indicators have certain limitations in predicting and warning rockburst. However, rockburst criterions still have important significance, as they can provide guidance for mine design, roughly delineate key areas of concern for rockburst, and develop rockburst prevention and control strategies.

Based on the above sections, it is evident that the Ruihai granite has a tendency for rockburst under all three criterions, and some even assess it as having a high level risk of rockburst. Therefore, it is necessary to carry out on-site monitoring to conduct daily risk management and control of rockburst.

Based on previous engineering experience and in combination with the microseismic monitoring methods recommended by the ISRM (Xiao et al., 2016), a new generation of high-precision intelligent microseismic monitoring system for rock fracture produced by Hubei Seaquake Technique Co., Ltd was selected. This system can continuously monitor rockburst disasters in rock engineering for 24 h, achieving real-time automatic data acquisition, analysis, and positioning of microseismic data. It can obtain micro-fracture information of the entire process of rockburst disaster incubation, providing a basis for the evaluation of rock mass stability and disaster warning and prediction, and providing technical support for engineering management and rockburst prevention and control. The hardware of the microseismic monitoring system mainly includes sensors, data acquisition devices, communication devices, time synchronization servers, and data servers. Some hardware is shown in Figure 7. The sensor converts the ground motion (ground velocity or acceleration) into a measurable electrical signal. The data acquisition device is responsible for converting the analog electrical signal from the sensor into a digital signal, and then transmitting the data to the data server for analysis and processing through the communication equipment.

The software of the microseismic monitoring system mainly consists of system configuration management software, microseismic waveform data processing software, microseismic event visualization and interpretation software, and real-time display software for microseismic events, as shown in Figure 8.

The design of the development roadway excavation adopts the drilling and blasting method. Considering the characteristics of linear engineering of the development roadway and the optimized layout of the sensor array, combined with previous successful sensor layout schemes, the microseismic monitoring plan is as follows: one microseismic acquisition device and six microseismic sensors will be installed in the development roadway, including five uniaxial sensors and one triaxial sensor. Each sensor has a natural frequency of 10 Hz, a response ranging from 10 to 2,000 Hz, and a sensitivity of 100 V/m/s. They will be arranged in two cross-sections, with three sensors in each cross-section. The installation of the I-I monitoring section is selected at a distance of about 50–60 m from the working face (the specific distance depends on the on-site construction conditions), with the No. 1 sensor being a triaxial sensor and the other two being uniaxial sensors. The borehole depth for sensor installation is 2–3 m. The tunnel is excavated 40 m forward, which means that the I-I monitoring section is about 90–100 m away from the tunnel face. A second monitoring Section II-II is arranged at a distance of about 50–60 m from the tunnel face (the specific distance depends on the on-site construction conditions). Sensor 3 is a triaxial sensor, and the other two are uniaxial sensors. As the tunnel continues to excavate another 40 m, the I-I section sensor is about 130–140 m away from the face. The I-I section sensor is recovered and arranged at a distance of about 50–60 m from the face. This cycle continues, and every 40 m of excavation, the last row of sensors is recovered and arranged at a distance of about 50–60 m from the face until the tunnel excavation is completed. To improve monitoring performance, it is planned to adopt an interlaced arrangement of sensors within the section, that is, the adjacent sensors within the section are spaced apart by 2–4 m. At the same time, when the sensors move, the two sensors on the left wall (or right wall) are replaced by the two sensors on the right wall (or left wall), so that the sensors are arranged in a spiral alternating pattern. The specific layout plan is shown in Figures 9, 10.

After the installation of the microseismic monitoring system, the rockburst risk of the roadway can be quantitatively predicted by rockburst warning formula based on the microseismic information recommended by the Technical Specification for Rock Burst Risk Assessment of Hydropower Engineering. This warning method can provide different levels of rockburst risk probability (Chen et al., 2015; Hydropower et al., 2019). When the potential rockburst risk is high, the basic excavation strategy of “short footage and weak blasting” should be followed. At the same time, measures such as destress boreholes and destress blasting can be taken, and strengthened support measures such as “Combining active and passive support, using rigid-flexible support together, and combining multiple support methods” can be used. When there is an extremely high level of rockburst risk, measures such as evacuation and suspension of construction should be taken.

The first mining area is located in the −1,480 level, which is deeply buried. According to empirical rockburst criteria, the level of rockburst tendency is relatively high, and the volume of the mined rock mass in the mining area is large. The stress adjustment of the surrounding rock is bound to be severe, and the maximum tangential stress can be several times higher than the original rock stress. Therefore, the occurrence probability of rockburst with high intensity is relatively high.

Taking into account factors such as the distribution of the ore body and key monitoring areas (ore body enrichment area), mining methods and processes, tunnel space layout, coverage of microseismic sensors, and distribution arrays of sensors, a sensor encryption and more triaxial sensor layout plan will be implemented in the key monitoring areas. The specific sensor layout plan is as follows: Firstly, a triaxial sensor is installed on the middle roof of the first ort roadway. Then, two uniaxial sensors are placed on the roof near the intersection area of first ort roadway and the two tramway transport entries, respectively, and another two sensors are placed on the roof near the intersection area of the second ort roadway and the two tramway transport entries and a microseismic monitoring network is preliminarily formed. As the mining progresses, a uniaxial sensor is installed on the middle roof of the second ort roadway, and two uniaxial sensors are placed on the roof near the intersection area of third ort roadway and the two tramway transport entries. As the mining continues, a triaxial sensor will be installed in the middle of the roof of the third ort roadway. This process will be repeated until the sixth ort roadway, covering the key monitoring area. The specific arrangement of microseismic sensors can be seen in Figure 11, and the monitoring layout plan for other areas can be seen in Figure 11. A total of 28 microseismic sensors are arranged in the middle section of 1,480, including 6 triaxial sensors and 22 uniaxial sensors. 40 microseismic channels are required, and 5 high-precision microseismic acquisition instruments produced by Hubei Seaquake Technique Co., Ltd are selected. Considering factors such as the layout of microseismic sensor cables and external construction disturbances, the microseismic acquisition instrument is arranged in one side of the transportation roadway, and the microseismic monitoring sever should be located in chambers in case been disturbed and damaged.

During microseismic monitoring in the mining process, close attention should be paid to the area where microseismic events are clustered in the surrounding rock of the mining area. The more clustered the microseismic events are, the more active the micro-fracturing of the rock mass in that area, the more intense the stress adjustment, and the higher the energy accumulated in the surrounding rock. The probability of micro-fractures penetrating each other to produce large fractures is higher, and the risk of rockburst is also higher. Therefore, when the number of microseismic events and the released energy suddenly increase in a certain area over a period of time, appropriate measures should be taken to prevent and control rockburst.

In the area where microseismic events are clustered, measures such as destress blasting or destress boreholes during mining, reducing the single charge amount of mining to reduce disturbance to the surrounding rock, and strengthening the support of the mining area should be taken to mitigate or even control the risk of rockburst. The mining rate of the mining area should be reduced to slow down the stress adjustment. The number of personnel working in the area should be reduced. When the risk of rockburst is extremely high, personnel should be evacuated from the working face.