Introduction

Faults, as common geological structures in coalmines, are a geological factor that cannot be ignored in underground mining, and they represent a significant threat to safe coal production (Wang et al. 2018). In Chinese coalmines, fault structures are mainly simple, with small or medium-sized fall heights (Jiang et al. 2017). Field practices show that fault structures directly or indirectly control inducing rock bursts. Moreover, according to the statistics, more than 70% of accidents occur near faults. For instance, approximately 90% of rock burst accidents every year occur near faults in Laohutai coalmine. Further, in 50 rock burst accidents in Longfeng coalmine, a total of 36 times were statistically found to be directly related to faults. Hence, as mining depth and intensity increase, the frequency and intensity of rock bursts affected by faults rise; this results in ejection of rocks and coal, damage of supports, titling of devices, and even fatalities (Wang et al. 2016; Jiang et al. 2017). Figure 1 shows several field images captured after these accidents.

Fig. 1
figure 1

Field images captured after accidents

Full size image

Commonly used study methods to solve the scientific problems of underground engineering are laboratory tests, field monitoring, theoretical analysis, numerical simulations, and physical simulations. For instance, a true triaxial test was carried out to analyze the two kinds of potential fault plane models (Gong et al. 2015). Microseismic monitoring in the field was used to study the relationship between rock bursts and in situ stress (Xia et al. 2016). The voussoir beam model (through theoretical analysis) was established to study the instability of fault coal pillars (Li et al. 2016); the in situ stress distribution affected by faults is studied by using ADINA software and UDEC software (He et al. 2010; Jiang et al. 2018). The fault characteristic parameters and roof physical-mechanical parameters were studied by FLAC2D software and FLAC3D software (Li et al. 2008; Sainoki and Mitri 2015). Roof stability of tunnels or roadways in underground mines was studied with 3DEC software (Shreedharan and Kulatilake 2016; Sherizadeh and Kulatilake 2016; Xing et al. 2018a, b); the laws of stratum behavior and fault slipping were studied using physical simulation tests (Lai et al. 2016; Wang et al. 2014a).

In this paper, the physical simulation test method, which can intuitively recreate and artificially change the testing conditions, is chosen and used to simulate a progressive mining operation. This is because the evolution of deformation and movement and stress of overlying strata are often hidden and inaccessible. This method has many advantages, including being an easy to operate, clear, intuitive, and repeatable test; moreover, it also can ameliorate the disadvantages of field observation and theoretical analysis, such as high cost and time consumption.

Considerable studies regarding physical simulation tests within a fault occurrence have been conducted. For instance, Wang et al. (2017), Peng et al. (2013), and Liu et al. (2011) studied the failure and stress distribution of overlying strata affected by a fault. They obtained the roof caving height and concentrated stress of the coal body (which were large), with the footwall face advancing toward the fault. Dou et al. (2017) tested roadway deformation with and without tectonic stress, finding that the tectonic stress field had a significant effect on roadway deformation failure, which can provide a technical reference for the optimization of roadway layouts. Wang et al. (2014b), Jiang et al. (2013), and Bornyakov et al. (2016) studied the evolution of fault plane stress during the fault slipping process, and deduced that the fault activates more easily when the working face approaches the fault. Wu et al. (2007) studied the mining effect of fault zone rocks and obtained the different response characteristics between the two fault walls. Zuo et al. (2009) monitored roof horizontal displacement through theodolite, and confirmed that slip dislocation occurs near faults based on the monitoring results of the digital speckle correlation and discontinuous deformation methods. Zhang et al. (2015, 2017) developed a new similar solid–fluid coupling material and conducted a feasibility analysis of the simulated materials to study the expansion activation of concealed faults.

Previous studies were either confined to the laws of mine pressure with fault occurrences or only to the development of similar materials. However, the relationship among the movement of overlying strata, fault activation slipping, and rock bursts near faults has rarely been reported. Hence, in this paper, a physical model test with a working face passing through the fault is carried out; then, the mining effects on fault activation slipping, and the responses of abutment stress of coal body and fault plane stress during the process of fault slipping, are studied. Consequently, the relationship among mining operation, fault activation slipping, and rock bursts near faults is analyzed. Results can serve as a reference for predicting rock bursts, and their classification into hazardous areas under similar conditions.

Scheme of physical simulation test

By using existing equipment in a laboratory and study targets, a two-dimensional model frame with dimensions of 3000 (length) × 400 (width) × 1800 mm (height) was chosen, and a physical model was established based on the similarity theory (Wang et al. 2015a, b; Xuan and Xu 2017; Xia et al. 2017). Then, the similarity ratios (geometry, density, and strength) between the physical model and the prototype were determined as 1/200, 1/1.5, and 1/300, respectively, by considering the study scheme and referring to previous studies (Lai et al. 2016; Xie et al. 2016; Dai et al. 2010).

To create the physical model, similar materials (river sand, calcium carbonate, and gypsum) were applied (Wang et al. 2017). The simulation of different strata in a prototype should be distinguished by different proportions of the materials; this is termed the material ratio, and it is conventionally determined by several uniaxial compressive tests of rock specimens, with each group of specimens being repeatedly tested and compared. Finally, the material ratio was determined by referring to the literature (Li 1988) and previous test experiences. In addition, mica powder was used to simulate the stratification of rock layers by evenly paving between strata. The prototype strata were derived from the Yangliu coalmine in district 104. Properties of the prototype and physical models, and the material ratios, are listed in Table 1.

Table 1 Lithology and physical mechanical parameters of roof–floor strata (Wang et al. 2017)
Full size table

The physical model can be deployed by using the similar materials, and the procedure can be divided into six steps: material weighing, material mixing, blending evenly, adding water and mixing evenly, paving and ramming, and removing redundant material. Finally, total height of 1460 mm and roof height of 1100 mm were achieved. However, the physical model cannot fully simulate all strata of the prototype (mining depth of 270 m), and the remaining roof strata in the prototype were compensated by 500 kg iron weights on the physical model. Considering the weak characteristics of a fault, mica powder with a thickness of 10 mm was used in the simulation (Huang et al. 2009). Moreover, the thickness of the simulated coal seam was 8 m, the dip angle of the fault was 50°, and the fall height of the fault was 4 m, which was categorized as a small fall height.

After 7 days of drying, the model could be viewed, and then the plans of displacement and stress monitoring were designed. Among them, 16 monitoring lines were set up, and the 9th and 16th lines were marked as D9 and D16 and chosen to study the displacement variation by using Electronic Total Station; meanwhile, 35 stress sensors were applied to study stress variations by connecting to a DH3815N stress–strain testing system. Figures 2 and 3 show the overview and monitoring plans of physical model.

Fig. 2
figure 2

Physical model and test system (Wang et al. 2017)

Full size image
Fig. 3
figure 3

Plans of displacement and stress monitoring

Full size image

Results of physical simulation test

Mining process

Before mining, a coal pillar with a width of 150 mm was left at the open-off cut side to eliminate the boundary effect. Subsequently, a 50 mm step increment was synchronously mined from the front and rear of the model, and the interval time between two steps was set to 2 h. Figure 4 depicts the typical states of overlying strata during the process of model mining. Note that all the length parameters are converted into an in situ scale.

Fig. 4
figure 4

Movement and structure of overlying strata with a fault occurrence: a L = 40 m; b L = 30 m; c L = 20 m; d L = 10 m; e L = 0 m; f L = − 30 m; g L = − 70 m; h L = − 110 m

Full size image

When the distance between the working face and the fault exceeded 40 m, it could be defined as L ≥ 40 m. A typical saddle-shaped fracture pattern of overlying strata formed with L = 40 m, as shown in Fig. 4a. At present, the fault is not affected by mining operations, and the overlying structure evolves conventionally.

In Fig. 4b, the fault is beginning to be activated because the high-located strata have rotated, and the fault plane opened due to fault cutting; the maximum and minimum heights of the fault opening fracture (FOF) were 130 and 83 m, respectively. Moreover, the maximum height of the FOF was higher than the maximum height of bed separation (122 m), which further indicates the fault can result in more severe failure. Then, the fault activation was enhanced (Figs. 4c and d), resulting in FOF expansion; maximum and minimum heights were 174 m and 38 m, respectively, and the fracture height of the footwall strata increased to 165 m with L = 10 m. With the face continuing to advance, the fault opened completely, as shown in Fig. 4e, because of the sudden roof subsidence; the hanging wall strata rotated and slipped along the fault, which resulted in the high-located FOF reduction. Referring to Fig. 4f, when the face passed through the fault with L = − 30 m, the cantilevered main roof rotated and was cut down, which resulted in the fracture height in the hanging wall increasing to 69 m. Then, the hanging wall strata slipped, driven by fault cutting and overburden pressure (as shown in Fig. 4g), and the FOF tended to be close; the fracture height footwall strata increased to 178 m due to the hanging wall slipping.

As shown in Fig. 4h, the caved rocks in the hanging wall formed a support structure for overlying strata, which resulted in little effect of the fault on strata movement. The footwall strata moved fully, and the FOF basically closed.

Movement characteristics

Figure 5 depicts the movement characteristics of overlying strata at different layers to explain the mining effect affected by a fault quantificationally.

Fig. 5
figure 5

Displacement of overlying strata at different layers: a D9 monitoring line; b D16 monitoring line

Full size image

When L ≥ 30 m, the footwall high-located strata initially rotated to the fault, resulting in the displacement of D16 increasing, with a maximum value of 2.8 m; however, the low-located stratum movement was not affected by the fault, and the maximum displacement value was 5.8 m. When 30 > L ≥ 10 m, the high-location fault activation aggravated, and the strata slipped prominently, which resulted in the displacement of D16 sharply increasing to 5 m; meanwhile, the low-location fault started to activate, and the displacement of D9 slightly increased to 6.2 m. When 10 > L ≥ − 30 m, the hanging wall high-located strata rotated to the fault, and its maximum value of D16 increased to 1.2 m; however, the displacement of D9 in the hanging wall increased sharply because of the main roof cutting down with L = − 30 m. When − 30 > L ≥ − 70 m, the subsidence range expanded, and displacements of D9 and D16 increased continuously; meanwhile, the displacement of D16 in the hanging wall notably increased to 3 m, while that of footwall increased slightly to 5.4 m, due to the rotation and slipping of hanging wall strata.

Analysis of fault slipping

The displacement of two monitoring lines on two fault walls parallel to the fault plane was monitored. Then, the curve variations of relative displacement (which can represent the fault slipping) were depicted as shown in Fig. 6.

Fig. 6
figure 6

Variation of relative displacement of two fault walls: a relative displacement of two monitoring lines; b displacement of two monitoring points

Full size image

In Fig. 6a, when L ≥ 10 m, the mining effect on the fault aggravated gradually, and the footwall high-located strata rotated toward the fault, resulting in displacement of the footwall monitoring line being reduced downwards. When L = 0 m, the relative displacement increased sharply at the low location because of the sudden strata subsidence. When 0 > L ≥ − 70 m, it reduced notably at the low location and aggravated gradually, because of fault activation and hanging wall lower-located strata cutting-down.

Figure 6b shows the displacement variations of two points in the D9 monitoring line of two fault walls. It can be seen that when L ≥ 40 m, the mining effect on fault was slight, and fault slipping was small. Then, as the fault activation was enhanced, the displacement of footwall strata notably increased to 5.2 m with 40 > L ≥ 0 m, while that of the hanging wall only increased to 1.8 m, resulting in the relative displacement sharply increasing, with the peak value of 3.4 m. When 0 > L ≥ − 30 m, mining operations had a large effect on the hanging wall strata, and the displacement notably increased to 5.8 m, while it had little effect on the footwall strata and the displacement there displayed no significant change; hence, the relative displacement clearly dropped to 0.4 m with L = − 30 m. As the face continued to advance, the mining effect on the fault weakened, and the change of fault slipping was not clear.

Stress evolution of fault plane

According to the monitoring scheme of the physical test, the stress data of sensors marked as S23 (L2), S27 (L4) and S31 (L6) were analyzed. Then, the variation curves of normal stress for the fault plane were obtained, as shown in Fig. 7. In the model test, all stress sensors had to be reset to 0 before coal mining. Hence, the positive stress readings from the sensor represent the stress concentration, and this range can be defined as the stress-increased zone, while the negative stress sensor readings depict the stress relief, and this range is defined as the stress-relaxed zone.

Fig. 7
figure 7

Variation curves of normal stress for fault plane

Full size image

As shown in Fig. 7, the normal stress of the fault plane at different layers had different variation characteristics. At the initial stages of mining, the fault is not affected by mining activation; hence, the normal stress of the fault plane remained unchanged. As the face advanced toward the fault, only S27 and S31 at the high position were affected by mining disturbance, and both normal stresses were − 0.01 MPa. Then, the effect of fault activation was enhanced and the fault plane opened, resulting in the stress of S27 dropping to − 0.04 MPa; however, the stress of S31 recovered and increased, gradually affected by compaction of the hanging wall strata. Moreover, the stress of S23 increased to 0.06 MPa and then dropped sharply because of fault plane opening. Finally, the increased stress of S27 was clearly due to hanging wall strata rotation, which then tended to stabilize.

Abutment stress evolution of coal body

Six sensors, marked as S2, S5, S7, S9, S11, and S14 at the floor strata, were chosen to study the effect of fault activation slipping on abutment stress of the coal body. Among them, S2 and S5 were at the footwall and far from the fault, at 84 m and 30 m, respectively; S7 was near the fault and far from the fault, at 10 m in footwall; and S9, S11, and S14 were at the hanging wall and far from the fault at 20 m, 48 m, and 106 m, respectively.

As shown in Fig. 8, when the face approached the monitoring points, the abutment stress concentrated in S2 and S5, and gradually increased to a peak value. For instance, the peak stress of S5 was 0.08 MPa, and then dropped sharply; after that, it recovered as the caving rocks compacted, but it was still smaller than that of in situ stress. When the face approached the fault, the abutment stress at S7 increased notably with a peak value of 0.18 MPa, because of fault coal pillar reduction and fault activation. This value was 2.25 times the peak stress of S5, and greater than that of the non-affected zone. When the face just passed through the fault, the fault still activated, and the hanging wall strata rotated and slipped, which resulted in the abutment stress of S9 and S11 increasing. With the face continuing to advance, a stable structure was formed (as shown in Fig. 4); a high abutment stress was displayed at S14 with a peak value of 0.23 MPa, which was 2.88 times the peak stress of S5.

Fig. 8
figure 8

Variation curves of abutment stress of coal body

Full size image

Hence, it can be concluded that prior to passing the fault, the fault is activated from high location to low location, and it aggravates gradually, which leads to increasing abutment stress of the footwall coal. After passing the fault, the cutting-down strata are compacted and slipped, which can result in the abutment stress of the hanging wall coal increasing; however, the hanging wall strata move slightly, and the displacement is smaller compared to that of the footwall. The fault is activated and slips in the range of − 70 to 30 m, in which rock bursts are easily induced.

Analysis of field observation

Geological conditions

In this section, panel 14310 of Dongtan coalmine in Shandong province (as shown in Fig. 9) is taken as the study background of field observation. The mining depth and thickness of coal seam are 600 m and 8.2 m, and the length of the face 262.9 m. The thicknesses and lithologies of the panel strata are listed in Table 2 according to the borehole histogram of No.49. Based on the field report, a normal fault marked as NF6 is located in the middle of face, which has a notable effect on mining of panel 14310.

Fig. 9
figure 9

Plan of panel 14310

Full size image
Table 2 Lithologies and thicknesses of the coal and rock mass (Chen et al. 2016)
Full size table

Analysis of field observation

To study the mining effect in the field, microseismic events were monitored by a seismological observation system (SOS), which was installed in May 2010 in Dongtan coalmine (Bischoff et al. 2010; Sainoki and Mitri 2014). Figure 10 shows the distributions of microseismic activities during the period from July 26, 2010, to August 25, 2010; herein, the shapes represent different energy of microseismic events, and these positions represent the locations of microseismic events. Figure 11 depicts variation of total counts and total energy for the microseismic events per day during the period from July 11, 2010, to August 28, 2010.

Fig. 10
figure 10

Sectional distribution of microseismic events (Chen et al. 2016): a July 26; b August 1; c August 6; d August 11; e August 17; f August 25

Full size image
Fig. 11
figure 11

Energy and counts of microseismic events between July 11, 2010, and August 28, 2010

Full size image

During face mining before July 24, it was far away from the fault, exceeding 62 m; the fault was slightly affected by mining, and the microseismic events were mainly distributed in front of the coal wall and near the goaf. The total energy and total counts of microseismic events were very low, as shown in Fig. 11 for stage I, where the pressure appearance from the field observation is similar to the simulation results with L ≥ 40 m in section “Mining process.”

During the period from July 25 to August 2, the face was far away from the fault, at approximately 40 m; the counts of microseismic events increased and concentrated near the fault, as shown in Fig. 10a and b. This indicates that the fault begins to be affected by mining and aggravates gradually. However, the total energy did not change considerably, with mainly small energies (less than 5000 J), as shown in Fig. 11 with stage II. At this stage, the pressure appearance of field observations is similar to the simulation results with 40 > L ≥ 10 m in section “Mining process.”

In Fig. 10c and d, during the period from August 3 to August 14, the face passed through the fault approximately 24 m, and the counts of microseismic events increased notably due to violent fault activation and slipping, resulting in the total counts and total energy increasing sharply and rising to a peak, as shown in Fig. 11 for stage III. Moreover, several microseismic events with large energy (10,000–50,000 J) occurred near the fault. Hence, the pressure appearance of field observation is similar to the simulation results with 10 > L ≥ − 30 m in section “Mining process.”

During the period from August 15 to August 22, the face passed through the fault by approximately 68 m, and the distribution of microseismic events in the footwall did not change, while in the hanging wall they moved forward, as shown in Fig. 10e. Moreover, the total counts and total energy notably dropped and aggravated gradually, as shown in Fig. 11 for stage IV. This indicates that the mining effect on faults weakens, and the fault activation effect decreases, which is similar to the simulation results with − 30 > L ≥ − 70 m in section “Mining process.”

When the face passes through the fault exceeding 68 m after August 22, the microseismic events are no longer affected by the fault, and they redistribute in front of the coal wall and near the goaf, as shown in Fig. 10f. Moreover, the large-energy microseismic events also no longer occur, and the total counts and total energy drop and retain this low level, as shown in Fig. 11 for stage V. The pressure appearance of field observation at this stage is similar to the simulation results with L < − 70 m in section “Mining process.”

Hence, regarding mining effects caused by a fault tectonic, comparing the results of field observations and physical simulations, the results between these two methods display good agreement, which verifies the correctness and feasibility of the physical model test and its simulation results. Moreover, this also reflects the advantages of physical simulation, such as the ability to reconstruct phenomena that cannot be directly observed in the field.

Conclusions

In this study, a physical simulation test simulating a progressive mining operation with a fault occurrence was performed. The mining effect on fault activation slipping, the responses of abutment stress of coal body, and fault plane stress during the fault slipping process were studied and analyzed. We can obtain several conclusions as follows.

  • Prior to passing the fault, it is activated from high location to low location, and it aggravates gradually, which leads to the abutment stress of footwall coal increasing; moreover, the footwall strata move fully, and the displacement is large.

  • After just passing the fault, the cutting-down strata are compacted and slip along fault plane, which can exert load on the hanging wall coal and lead to increasing abutment stress of the hanging wall coal; however, the hanging wall strata move slightly, and the displacement is small compared to the footwall.

  • The fault is activated and slips in the range of − 70 to 30 m in this paper, and the slipping mutation of two fault walls occurs successively, which can easily induce rock bursts near the fault (the symbol “−” means the face has passed through the fault).

Finally, a field case regarding microseismic monitoring was used to verify the simulation results.

Study results can serve as a reference for predicting rock bursts and their classification into hazardous areas under similar conditions.