This object of study is the Gaohe Coal Mine, located in Changzhi City, Shanxi Province, China (Figure 2A); the current main coal seam is No. 3, with a thickness of 6.0 m and an average overburden depth of 600 m, and is mined using the consecutive single-wing mining method. The double-roadway layout means are adopted for the working face in Mining Area I in the western wing. The main research objects of this article are the W1308 headgate and W1309 tailgate. The double-roadway layout system involves driving along the roof strata of the coal seam, with a 7 m-wide coal pillar left in the middle. The cross-section of the W1308 headgate and W1309 tailgate is 5.0 m wide and 4.0 m in length. The three working faces of W1308, W1309, and W1310 are ceased and waiting for layout currently in the Mining Area W1 in the western wing, of which the working width of the working face W1308 is 180 m, with a length of 1,600 m (Figure 2B). The roof is composed of fine sandstone, mudstone, and medium sandstone in sequence. The floor plate is composed of mudstone, fine sandstone, and medium sandstone in sequence (Figure 2C).

The width of the coal pillar between the W1308 headgate and W1309 tailgate is 7.0 m and is driven along the roof. Figure 3 illustrates the roadways and coal pillar support parameters. Left-handed, non-longitudinal threaded steel bolts with a 22 mm diameter and a length of 2,400 mm were adopted for roof rock bolt support, with bolt spacing of 900 mm and each row of bolts spaced 800 mm apart, and six rock bolts per row. The roof rock bolts were perpendicular to the roof of the roadway, and were matched with square pallets and spherical anti-friction washers. The length, width, and height of the bolt plates were 150, 150, and 10 mm respectively, and the central aperture was 1.5–2.0 mm larger than the diameter of the bolt rod. CKb2335 and a K2360 resin anchoring agent were adopted for anchoring, with CKb2335 in the front, and K2360 at the rear. Mine anchor cables of a diameter and length of 17.8 and 6,300 mm respectively were used for roof anchor cable support. The spacing and row spacing of anchor cables were 1,600 mm and 800 mm respectively, with three anchor cables in each row, matched with locks and square plate. The length, width, and height of the anchor cable plates were respectively 300, 300, and 16 mm. One CKb2335 and two K2360 resin anchoring agents were adopted for anchoring the anchor cable, with CKb2335 in the front and K2360 in the back. The roof was laid with reinforced ladder beams made of 10# square mesh and 16# round steel. The length and width of square mesh were 5,200 mm and 1,100 mm respectively. Meshes must be overlapped with no less than 100 mm and must be tied firmly with barbed wire. The reinforced ladder beams used for the roof rock bolts were 4,700 mm long and 60 mm wide, while the roof anchor cables were 3,400 mm long and 60 mm wide.

Two ribs were supported using left-handed non-longitudinal rebar rock bolts with a size and length of 22 and 2,200 mm, respectively. The row spacing was 800 mm, and the bolt spacing was 800 mm. Four rock bolts were installed in each row, and the rock bolts on ribs were perpendicular to the rib, while the rock bolts at the rib corners were set at a 15° incline. Bolts were equipped with a square plate anti-friction washer, and a spherical washer. The length, width, and thickness of the rock bolt plates were 150, 150, and 10 mm respectively. The rock bolts were anchored with a CKb2335 and a K2360 resin anchoring agent, with CKb2335 in front and K2360 at back. The two ribs were laid with reinforced ladder beams made of 10# square mesh and 14# round steel. The square mesh was 4,200 mm long and 1,000 mm wide, while the two ribs of the reinforced ladder beam were 3,400 and 60 mm long and wide, respectively.

Gao and Stead (2014) introduced the “UDEC Trigon” logic to represent brittle rock fracture. The rock mass is described as a set of triangular blocks linked by internal contact to imitate brittleness material in the triangle procedure. These zones cannot fail if each triangular block is made of an elastic material and is split into triangular finite-difference areas. Shear or tensile stress damage can only occur across the contact surface, and the strength of the contact surface defines the effectiveness of the model. The stress-displacement connection is considered to be linear and is governed by stiffness k n (Inc., 2014) in the direction of vertical contacts. Δ σ n = − k n Δ u n (1) Where Δ σ n represents the effective normal stress increase and Δ u n represents the normal displacement increment.

Toward normal contact, there is a maximum tensile strength, τ s max , if the tensile properties are reached, Δ σ n = 0.

The reaction in the shearing direction is governed by a constant shear stiffness, τ s . A composite of contacting micro properties, cohesion, and friction, determines shear stress. As a result of: | τ s | ≤ c + σ n ⁡ tan ⁡ φ = τ s max (2)

Then Δ τ s = − k s Δ u s e (3) or else, if | τ s | ≥ τ s max (4)

Then | τ s | = sign ( Δ u s ) τ max (5) Where c is the cohesion, φ is the friction, Δ u s e is the elastoplastic component of increasing shear displacement, and Δ u s is the sum of the incremental shear displacements.

The two-dimensional UDEC model is used to describe the failure characteristics of the coal pillar and its supporting structure under the disturbance of repeated mining, as well as to develop different coal pillar and roadway support parameters (Figure 4). The model is 180 m wide and 70 m high. To enhance the computation efficiency of the model, only the area of interest, namely the W1308 headgate and W1309 tailgate, and the coal pillar range between them, are discretized using “UDEC Trigon” logic. The average size of the triangular blocks in the region of interest is 0.2 m, and a longer triangular block with a size of 0.5 m is used to represent the roof-to-floor rocks in the region of the two roadways and the coal pillar, and a narrow and long rectangular block is used to represent the coal seam and rock layers at the triangular block boundary.

Fixed displacements are conducted in the vertical and horizontal directions at the bottom and two sides of the model. According to the field stress test, the vertical stress is 15.0 MPa, the maximum principal stress is 18.2 MPa perpendicular to the axis of the roadway, and the minimum principal stress is 12.4 MPa parallel to the axis of the roadway. This original rock stress state is applied to the model, and a vertical stress of 15.0 MPa is applied to the upper boundary of the model to simulate the pressure of the overlying rock.

In general, the simulation calculation of the stability of the coal pillar and its supporting structure under the disturbance of repeated mining of the double-roadway layout system is mainly divided into four steps, as shown in Figure 5:

Step 1: Initial ground stress conditions are applied in the global model, to calculate the balance, and save it as a numerical model of the original rock stress balance.

Step 2: Excavation of the two roadways of the double roadway layout system, excavate the W1308 headgate and W1309 tailgate, and carry out bolt and cable support, calculate the balance, and save it as a numerical model after the roadways excavation.

Step 3: Mining simulation of the W1308 working face. It is a process of gradual mining by the coal mining machine for the excavation of the W1308 working face in the field. Therefore, segmented deletion of partial blocks is used for the mining simulation of the W1308 working face in the model to avoid discrepancy between the large dynamic loads and the site resulting in excessive errors in the simulation results. After the stepwise excavation, the balance is calculated and saved as a numerical model of the coal pillar disturbed by the mining at different excavation stages.

Step 4: W1309 working face mining simulation. The W1309 working face mining simulation method is the same as W1308 working face mining, the balance is calculated after distributed excavation, and it is saved as a numerical model of repeated mining disturbances in different excavation stages.

For the W1308 and W1309 working face mining simulation proposals: The excavation of the panels was recreated using a progressive excavation. Each step requires a 10 m advancing distance, 40,000 time steps were executed at each level to answer dynamic stress. (Gao et al., 2015).

Laboratory-obtained rock characteristics should be modified into rock mass parameters. The RQD method is frequently used to estimate the deformation modulus of a rock mass. Supported by a large amount of on-site monitoring data, Zhang and Einstein (Zhang and Einstein, 2004) proposed the relationship Eq. 6 between RQD and E m / E r , which was then utilized to adjust rock mass parameters. In this study, we used a borehole camera to measure the RQD value of rock mass: E m / E r = 10 0.0186 R Q D − 1.91 (6) Where E m , E r are the deformation modulus of rock mass and intact rock respectively. Unconfined compressive strength (UCS) can be modified based on the ratio of E m / E r (Singh and Seshagiri Rao, 2005): σ m / σ r = ( E m / E r ) n (7)

σ m , σ r are the strength of rock mass and intact rock respectively. In addition, the index n for splitting, shearing, sliding, and rotating modes is 0.56, 0.56, 0.66, and 0.72, respectively. In this research, the index n is 0.63—considering the failure process of a roof is so complex, there are several failure types. We also assumed that the ratio of rock mass tensile strength ( T m ) to intact rock tensile strength ( T r ) follows the same transfer function (Wu et al., 2019b), The parameters of the rock mass can be calculated, and the results are presented in Table 1.

These data’s rock mass properties parameters cannot be applied to the model directly, and the mechanic parameters of the contact surface and triangle utilized to represent the rock mass properties must be acquired by numerical calibration. Hence, UDEC Trigon logic calibration models were developed, which are the UCS test model with 2.0 m and a height of 4.0 m, and the Brazilian shear test model with a diameter of 2.0 m. The trial-and-error approach is used to calibrate the input parameters of the block and contact surface to fit the rock mass characteristics shown in Table 1; Figure 6 presents the calibration results for rock and coal (The term “Damage” shown in Figure 6 is introduced in Section Coal Pillar Failure Study Under the Original Support Scheme), as well as the micro-parameters in Table 1. Table 1 shows that the error between the UCS and E obtained by numerical simulation model and the results acquired by laboratory test is a little less than 10%. These parameters exactly match the rock mass properties in this research, indicating that the micro-mechanical parameters of coal and rock mass obtained in Table 1 are acceptable.

In this section, the W1308 headgate and W1309 tailgate are excavated with a width and height of 5.0 and 4.0 m, respectively. The numerical simulation procedure simulates the excavation of the roadways by deleting blocks. Conversely, rapid excavation may result in an imbalanced model response, resulting in a dynamic stress path around the deleting border. This type of dynamic stress generally results in a more extensive damage region than desired around the excavation (Cai, 2008). The roadways are progressively and constantly mined by the heading machine in the field excavation, and the static stress path is developed by the roadway’s boundaries (Hazzard and Young, 2004; Yang and Fall, 2021). To reflect this more realistic mining impact, the material softening approach is used to simulate the mechanical excavation of the roadways. In this procedure, the Young’s modulus of the element in the excavation region deteriorates to zero in stages to represent the impact of progressive excavation. This procedure reduces the impact of transients on material failure and provides a more static resolution.

In the numerical simulation, a rock bolt element is made up of two structural nodes and one structural component. The structural components are elastic and plastic materials, as shown in Figure 7, which yield under tension and compression and can resist bending moments. The rock bolt element can realistically realize the normal shear and axial tension of the bolt. The components can be disconnected and separated at the nodes. Fracture of the rock bolt element is realized depending on the user-specified tensile failure strain thresholds (tfstrain). The following criteria were used to determine the failure of the rock bolt element: ε p l ≥ t f s t r a i n , where tfstrain should be specifically set; ε p l is the sum tensile yield strain of every unit of the anchoring rod, and the standard UDEC module recognizes the following (Inc., 2014): ε p l = ∑ ε p l a x + d 2 θ p l L (8)

If the sum tensile yield strain of every unit of the anchoring rod exceeds the artificially set tensile failure strain limit (tfstrain), the stress and bending moment of the anchoring rod member will instantly decrease to zero, and it can be considered that the rupture occurs at the point of the anchoring rod. The built-in “cable” component represents the anchor cables, while the built-in “liner” component represents the roof and two reinforcing ladder beams (Bai et al., 2016; Bai et al., 2017). Figure 7 shows the schematic diagram of the composition of the anchoring rock bolts, the anchor cables, and the structural elements (Inc., 2014). The details of the support unit are provided in Table 2 of this study.

The Fish function has been used to measure the sum length of the cracks in the coal pillar, as well as the lengths of shear and tensile cracks induced by stress disturbances during excavation, first working face mining, then repeated mining, to obtain the failure characteristics of the coal pillar of the double-roadway layout system under the original support scheme. The damage parameters are proposed according to the standard of Gao and Stead (2014): D = L s + L t L s u m × 100 % (9) Where, L s is the sum length of shear cracks, L t is the sum length of tensile cracks, L s u m is the sum contact length.

The changes in coal pillar damage parameters during double roadway tunneling (Stage 1), first working face mining (Stage 2), and repeated mining disturbance (Stage 3) are shown in Figure 10A. During double roadway driving (Stage 1), the damage parameter in the coal pillar increased slowly from 0 to about 20%, the damage parameter in the coal pillar increased from 20% to about 55% during the first working face mining (Stage 2), and the damage parameter in the coal pillar increased from 55% to around 90% during the repeated mining disturbance period (Stage 3).

It can be seen from the analysis in section Coal Pillar Failure Study Under the Original Support Scheme that under the original support conditions, when the first working face is mining at a long distance from the heading face, the extent of damage in the coal pillar does not change much compared with the tunneling period; but when completely experiencing the impact of first working face mining, the damage of the coal pillar increased to 55% under the original support scheme. When affected by the repeat mining disturbance, the final coal pillar was almost in a broken state, and the W1309 tailgate was seriously deformed, which could not meet the requirements for safe use. Therefore, a reinforced support plan is proposed in this part, and the main reinforced support parameters are as follows:

In the original support plan, the length of the bolts of the coal pillar sidewall was increased from 2,200 to 2,400 mm, and the pre-tightening force of the bolts was increased from 180 to 300 Nm. At the same time, two bolts with a length of 7,300 mm and a diameter of 7,300 mm were installed in the middle of the coal pillar. The tension pre-tightening force of the anchor cable is 200 kN. The model is recalculated according to the calculation process of the numerical model in 3.1, and the failure features of the coal pillar under the condition of reinforced support are obtained as shown in Figure 10B.

It can be seen from Figure 10B that after the reinforcement support conditions are adopted, the extent of damage in the coal pillar has not changed much compared with the original support scheme in stage 1. During stage 2, namely the first working face mining, with the gradual increase of the degree of mining influence, the damage in the coal pillar gradually became serious, but compared with the original support scheme, the final damage degree is reduced by about 15%; in the reinforcement support scheme for the stage 3, the repeated mining disturbance stage, the damage degree in the coal pillar is significantly reduced, and the reduction even reaches an astonishing 27%. Currently, the coal pillar is in the plastic bearing state and still has the features of isolation bearing.

From the previous analysis, we can see that there is a significant difference between the strength of reinforced support and conventional support in controlling the damage degree of a coal pillar, especially in the disturbance stage of repeated mining; the degree of damage to the coal pillar is significantly lower, and the supporting intensity is significantly reduced. The internal mechanism is mainly because the reinforcement support during the double roadway excavation can effectively limit the expansion and extension of the initial fissures inside the coal pillar under the disturbance of a dynamic load. It can be seen from Figure 11 that after the reinforcement support scheme is adopted, the damage extent in the coal pillar is not high after the first working face mining is completed, and there is a partially complete core bearing area inside the coal pillar; while during the disturbance period of repeated mining, the slipping, stretch, and shear occur to the internal fissures of the coal pillar, but under the restriction of high pre-tightening force bolts and high pre-tightening forces on the anchor cables. In the end, the large-scale destruction of the internal crack of the coal pillar did not occur, and in the middle of the coal pillar there existed a region defined as “Lower damage”, which is expected from the literature (Gao et al., 2015; Wu et al., 2019b; Wu et al., 2019c; Fan et al., 2019; Li et al., 2021). The damage is calculated by an Eq. 9, and “Lower damage” is defined as a region where the value is below 35%. Therefore, it shows that the initial support of the coal pillar and roadway has a significant effect in the double roadway layout system, which ensures the primary stability of the coal pillar and effectively limits the damage caused by disturbance of the repeated mining upon the coal pillar.

Based on the above research, we proposed a double-roadway layout coal pillar stability control strategy under the disturbance of repeated mining. As shown in Figure 12A, it is mainly divided into two aspects: one is to improve the overall plastic bearing capacity of the coal pillar, and the other is to limit the expansion of the surface and internal cracks of the coal pillar.

The proposed control strategy has been used in the double roadway layout system of the newly excavated W1309 headgate and W1310 tailgate. The main support parameters of W1310 tailgate control are shown in Figures 12B,C.

Figure 13A shows the roadway deformation features of the W1310 tailgate during excavation, mining one side of the working face, and repeated mining disturbance. The monitoring results show that the deformation of the W1310 tailgate has been effectively controlled. During the excavation, the maximum value of the roof-to-floor convergence is 180 mm, and the maximum value of convergence of the rib-to-rib is 200 mm. During mining the first working face, the maximum value of roof-to-floor convergence is 420 mm, and the maximum value of convergence of the two sides is 500 mm. During the disturbance of repeated mining, the maximum value of roof-to-floor convergence is 620 mm, and the maximum value of rib-to-rib is 730 mm. Figure 13B shows the loads of rock bolts and anchor cables in the coal pillar during the repeated mining, within the influence range of the pressure, the loads of the anchor cables and rock bolts begin to increase. When the monitoring station is about 40 m away from the working face, the load of the anchor is increased to 140 kN, and the load of the anchor cable is increased to 195 kN. It shows that the load of the anchor cables and rock bolts increases rapidly and can bear the load in time, so the effect of controlling the deformation of the surrounding rock is better. Meanwhile, during the repeated mining disturbance, the imaging results of the borehole inside the coal pillar 10 m ahead of the working face are shown in Figure 13C. It can be seen from Figure 13C that at the position where the repeated mining disturbance is the most severe, there is still a certain complete area in the central part of the coal pillar, verifying the rationality of the control measures through the combination with the above deformation results.

The influence of the single hydraulic reinforcement support on the crack propagation and damage of the coal pillar and supporting structure should be considered, and the parameters of the rock bolts and anchor cables in the simulation model should be matched with the field data in detail in the future research.

The interaction and stability of the “large structure” of the “arc triangular block” formed by the lateral roof after panel mining as well as the “small structures” of the coal pillar and supporting structure are the focus of our future research.