Longwall mining is a primary method for underground mining. Longwall mining system with coal pillar left between adjacent mining panels has the problems of low coal recovery rate, high drivage ratio. Moreover, it may cause geological disaster such as coal burst due to the concentrated stress on the coal pillar (Wang et al. 2021; Lou et al. 2021; Wei et al. 2022). The automatic roadway formation method by roof cutting, proposed in 2009, is an effective method to address these problems (He et al. 2017a; Gao et al. 2018; Wang et al. 2018). In this method, roof cutting is performed in the roadway in advance, and the roadway is then automatically formed and reused for the next mining panel. Because the stress transmission from the gob roof is offloaded by roof cutting, the formed roadway is in a low stress area. Practice suggests that the automatic roadway formation method can effectively reduce the roadway drivage amount and improve mining safety (Gao et al. 2017; He et al. 2017b; Wang et al. 2020).

Considerable research has been conducted on investigations of the key technique, design procedure and practical application of the novel non-pillar mining method under different geological conditions (Ma et al. 2018; Sun et al. 2014). He et al. (2015) built a theoretical cantilever beam model and proposed three key technologies for the method. Zhang et al. (2016) compared the moving behavior and stabilizing process of the roadway surroundings in two kinds of non-pillar mining method. It was demonstrated that the automatic roadway formation method was also applicable to special geological conditions such as a caving zone that contains thick and hard sandstone stratum. Tao et al. (2017) used a finite difference simulation method to explore the distribution laws of the vertical stress, periodic pressure and shear stress of surrounding rocks affected by roof cutting. They revealed that roof cutting can promote gob roof caving and transfer the concentrated stress center to the gob. He et al. (2018) carried out a field test under thick coal seam and fast-extraction mining conditions. An active stability control approach for the roadway surroundings was introduced by building a steady combination structure using the upper strata of the main roof, bulking gangues at the gob and cutting cantilever of the roadway roof. Based on the short cantilever beam theory, Yang et al. (2019) investigated the applicability of the automatic roadway formation method under different geological conditions.

Roof cutting is crucial for the automatic roadway formation method. Roof cutting affects the caving of the gob-side gangues and the stability of the roadway surroundings. Deep hole blasting and hydraulic fracturing are two commonly used techniques to address roof related problems in coal mine. In terms of deep hole blasting, Wang et al. (2013) investigated the deep-hole pre-split blasting mechanism based on cylindrical cavity expansion theory and explicit dynamic analysis. The deep hole blasting technique was applied in shallow depth coal seams to control roof caving. Petr Konicek et al. (2013) proposed a destress blasting (long-hole drilling and blasting) technique to prevent rockbursts. Besides, many studies were conducted to optimize the deep hole blasting parameters (Chen et al. 2021; Zou et al. 2022). For the hydraulic fracturing technique, most studies are focused on investigating the fracturing mechanism and the propagation laws of the fractures (Ye et al. 2017; Yang et al. 2023; Shen et al. 2016). However, the damage scope of these two techniques is hard to be controlled and inappropriate operation may easily disrupt the stability of the formed roadway. There are also other rock fracturing techniques, such as liquid CO₂ blasting (Li et al. 2022), plasma blasting (Kuznetsova et al. 2022), instantaneous expansion (Zhang et al. 2020), but the unsatisfactory fracturing effects, low efficiency and high cost restrict their large-scale application in roof addressing in mining. Therefore, there is a well-justified need for an improved technique to perform roof cutting and a systematic evaluation of its effectiveness. In our study, a directionally single cracking (DSC) technique is introduced to perform roof cutting. Mechanical and numerical models are established to study the cracking mechanism and crack propagation laws in the roof rock using the DSC technique. Subsequently, the roof cutting parameters on the influence of roadway surroundings are systematically evaluated by discrete element simulations and experiments. The optimized DSC technique is finally tested and verified in the field application.

In the automatic roadway formation method, after the first mining cycle, the subsequent working panel only needs one excavated roadway in advance, the other one is automatically formed during the next mining cycle, and no coal pillar is left in the mining area. Roof cutting is an important stage of the automatic roadway formation method. Currently, blasting and hydraulic fracturing are two commonly used methods to conduct rock cracking or cutting (Juárez-Ferreras et al. 2008; Sobhaniaragh et al. 2016; Wang et al. 2013; Gao et al. 2021; Guo et al. 2020). Because the blast holes are located along the edge of the formed roadway, the impact of blast cutting should be considered. To that end, a DSC technique is introduced.

As shown in Fig. 1a, in the conventional blasting, a crushed zone is formed under the action of stress waves. The crushed zone is arbitrarily distributed around the original hole. The implementation of DSC technology mainly relies on a specially designed device, i.e., an energy-accumulated tube (Fig. 1b). The device surface has two rows of controlled grooves. The explosives are installed in the energy-accumulated tube. During blasting, a high energy flow is generated and concentrated in the set direction. Perpendicular to the energy-accumulated direction, a reflection tensile stress initiates the roof cutting crack and the detonation-generated gases further expand the cracks. The roof rock in the energy accumulated direction is under a tensile blasting stress, while the other directions of the blasting hole are mainly under a compressive blasting stress. In the energy non-accumulated direction, compressive shear cracks may be generated. As the explosion stress wave spreads, the strength of the stress wave gradually decreases. In a decoupling charge mode of rock blasting, the radial compressive stress in different positions can be expressed as follows (Wang 1984)

where, \(r\) is the distance between the rock mass point and blast hole; \(r_{\text{b}}\) is the blast hole radius; \(\alpha\) is the attenuation index; \(P_{{1}}\) is the peak compressive stress.

Schematic diagram of the DSC technique

When the dynamic compressive strength is less than the radial compressive peak stress, compressive damage occurs. The critical condition can be written as follows:

where \(\tau_{\text{c}}\) is the dynamic compressive strength.

In the energy accumulated direction, the rock damage is mainly caused by tensile stress. The hoop tensile stress in different energy accumulated positions can be expressed as

where, \(b\) is the proportionality coefficient; \(\xi\) is the energy accumulated coefficient, which is related to the charging structure; \(P_{2}\) is the peak tensile stress.

When the dynamic tensile strength is less than the hoop tensile peak stress, tensile damage occurs. The critical condition can be expressed as

where \(\tau_{\text{t}}\) is the dynamic tensile strength.

In fact, because the tube body can prevent transmitting a portion of the stress wave, the actual stress in the roof rock is small in the energy non-accumulated direction. In addition, the compressive strength of the roof rock is relatively high. Therefore, the critical condition of Eq. (2) is hard to achieve and the roadway roof is integrated with the energy non-accumulated direction. Due to the energy-gathered effects of the DSC technology, the peak stress in the energy accumulated direction is much larger than that in the energy non-accumulated direction. Additionally, the compressive strength of the roof rock is appropriately 10 times its tensile strength, which means that the rock mass is good at resisting compression; however, it fails under tension. Thus, the critical condition of Eq. (4) is easily achieved, and therefore, it is ideal to make a tensile crack in the energy accumulated direction.

The crack propagation laws and damage processes for the conventional blasting and the DSC technique are compared using an explicit finite element simulation method. LS-DYNA is an explicit and nonlinear finite element program. The program has obvious advantages in simulating impact, penetration, blasting and other obvious nonlinear problems (Li et al. 2017; Ma and An 2008; Xie et al. 2016). In this study, the DSC was simulated using LS-DYNA to explore the mechanical response and stress transmission in the roof rock.

3.1 Control equations

During the blast simulation, a high energy combustion model, i.e., JWL model, is used to describe the chemical reaction process. Previous studies show that the model can effectively reflect the blasting process of explosives and blasting effects on surrounding rocks.

1. (1)

   Control equations for explosive material.

The control equations for the blasting wave surface can be written as

where, \(\rho_{D}\), \(u_{D}\), \(C_{D}\), and \(p_{D}\) are the density, particle velocity, acoustic velocity and pressure of the detonation product, respectively; \(k\) is the multi-index; \(D\) represents the detonation velocity of the explosive; and \(\rho_{\text{e}}\) is the density of the explosive.

The mass equation can be written as

The energy equation can be written as

The JWL EOS equation for the simulation can be expressed as (Xie et al. 2017)

where, \(P\) is the pressure; \(V\) is the relative volume; \(E\) is the parameter regarding the internal energy; and \(A\), \(B\), \(R_{1}\), \(R_{2}\), and \(\omega\) are constants.

1. (2)

   Failure criteria of the rock material.

The propagation of the stress waves can cause compression-shear and tension damage in the roof rock. In the simulation, when the maximum tensile stress of the rock unit reaches the ultimate stress value of the fracture failure of the material in uniaxial stretching (Nguyen et al. 2016), the rock begins to yield or rupture, which can be expressed as:

where \(\sigma_{\text{t}}\) is the tensile stress under blast loading and \(\sigma_{\text{td}}\) is the dynamic uniaxial tensile strength of the roof rock.

The compression damage in a rock mass is judged by the von Mises effective stress failure criterion, which can be expressed as

where \(\sigma_{\text{cd}}\) is the dynamic uniaxial compressive strength and \(\sigma_{\text{VM}}\) is the von Mises effective stress that is calculated by

In addition, the erosion algorithm is introduced in the numerical simulation. When the stress or strain states of the rock element reach their damage criteria, the rock element fails and this process is irreversible. Therefore, as the rock element fails, it will be removed for rock cracking.

3.2 Modelling parameters

During the blasting, the explosion generates shock waves that interact with the blast hole wall. In the LS-DYNA program, there are three algorithms, i.e., the Lagrange formulation, the Eulerian formulation and the Arbitrary Lagrangian–Eulerian (ALE) formulation (Elmarakbi et al. 2009; JO 2006). Considering the deformation behavior of simulation materials, the ALE formulation is used for explosive materials and the Lagrange algorithm is used for the rock materials.

In the current study, a mine-used emulsion explosive is used to simulate the explosion. The parameters of the explosive and JWL EOS equation are listed in Table 1.

Two scenarios were performed in the simulation to demonstrate the effectiveness of the DSC technique, as illustrated in Fig. 2. In the two scenarios, normal restraint boundary conditions are set up on the vertical and right sides. A crustal stress of 2.5 MPa (\(\sigma_{x} { = }\sigma_{y}\)) was applied to the left and top sides of the model. The nonreflecting boundary condition is applied on the outer rounded surface. In scenario one, the explosives are placed in the hole center and the blasting is not controlled. In scenario two, the DSC technique was used. The mechanical parameters of the rock and energy-accumulated tube are listed in Table 2.

Model setup for the conventional and controlled blasting

3.3 Evaluation laws of the crack and effective stress

1. (1)

   Evaluation laws of the crack.

The evolution laws of crack propagation at different time steps in the conventional blasting mode are shown in Fig. 3a. When there is no energy-accumulated device in the blast hole, the crack initiates along the circumference of the blast hole. The cracks propagate into the deep rock perpendicular to the peripheral direction. At the time of Step_70, the longest crack has basically reached the length of the blast hole diameter. Finally, there are seven main cracks that have formed along the hole surface. In contrast, when the DSC technique is used, the evolution laws of crack damage are more regular, as shown in Fig. 3b. During the numerical simulation, the solid tube of the energy-accumulated device blocks and absorbs a portion of the stress wave. At the initial stage of the blasting, the damage crack grows in the energy-accumulated direction. As the explosion continues, the crack propagates approximately in a line rather than in a random direction; however, due to the anisotropy of the roof rock, the crack dose not strictly propagate along the center line.

Crack evaluation process in the conventional and DSC blasting modes

Because the roadway is retained for the next mining panel, the roadway roof should be stable. A descriptive index of D_m is introduced to evaluate the damage degree of the roadway roof. The D_m refers to the maximum distance of the roof cutting line to the outermost crack. As shown in Fig. 4, the values of the D_m under different time steps were recorded along the roof cutting line. For a same position along the roof cutting line, with the increase of the calculation steps, the D_m increases. Under the two blast scenarios, the damage to the roadway roof is quite different for the same step. Under conventional blasting condition, the maximum value of the D_m of 122 mm can be seen in the 250 mm position along the model length. From 150 to 340 mm position along the model length in Step_110, the values of the D_m are all over 30 mm in the first scenario. Comparatively, after using the DSC technique, the crack propagation is in a single desired direction. The values of the D_m in the second scenario are all less than 10 mm. Therefore, the integrity of the roadway roof is well protected after using the DSC technique.

Statistics of the D_m under different calculation steps

1. (2)

   Variation of the effective stress.

During the numerical simulation process, it was found that besides the difference in the crack evaluation laws, the effective stress of the rock element under different blasting conditions is different. Eight measuring points were arranged around the blast hole in the two blasting scenarios to explore the time-varying laws of the effective stress. Point A, B, C, and D are arranged in the conventional blasting scenario and Points E, F, G, and H are arranged in the directional blasting scenario. The distance between the eight measuring points and the center point of the blast hole is the same, as illustrated in Fig. 3.

The monitored results of the effective stress in the conventional blasting scenario are shown in Fig. 5a and b. When the explosion is not interfered, the variation process and trend of the effective stress around the hole are quite similar. The peak effective stress at Point A is only 2.6% higher than that at Point C. The peak effective stresses at Point B and D are 281  and 302 MPa, respectively. The monitored results under the free blasting condition indicate that the stress wave around the blast hole uniformly spread out to the periphery, thereby causing the damage crack to propagate into the roadway roof, which influences the stability of the roadway roof.

Variation of the effective stress using the conventional and DSC technique

In sharp contrast, when the explosion is controlled by the energy-accumulated device, the effective stress distribution around the hole is quite different. As shown in Fig. 5c, Point E and G are in the energy accumulated and non-accumulated directions, respectively. During the blasting process, the maximum effective stress of Point E is 567 MPa, whereas the peak effective stress of Point G is reduced by appropriately 58% to 236 MPa. The effective stresses in Point F (energy accumulated direction) and H (energy non-accumulated direction) have similar variation laws, as shown in Fig. 5d. The stress in the energy accumulated direction is more concentrated but the stress in the protected direction is lower. Thus, the DSC technique can effectively promote the crack to propagate in the desired direction and restrain the crack from growing in the roof rock that should be protected.

It can be seen from the above study that the effective stress and cracks can be controlled, and the roadway roof and gob roof can be separated by the DSC technique. When the holes are connected by the DSC technique, a roof cutting line or surface is generated. The roof cutting has a great influence on the stability of the formed roadway. Previous studies show that the stress distribution (especially the vertical stress) and deformation of the surrounding rock of the roadway are the main factors that reflect the stability of the roadway (Ghasemi and Shahriar 2012; Kukutsch et al. 2015; Tesarik et al. 2009). In this study, the influence of roof cutting parameters (roof cutting height, roof cutting degree and roof cutting angle) was comprehensively studied and evaluated by discrete element simulation and field experimental methods.

4.1 Roof cutting schemes

Field application indicated that the most important parameters that affect the behavior of the roadway surroundings are roof cutting height, roof cutting degree and roof cutting angle. Reasonable roof cutting parameters are favorable for the stability of the formed roadway. The roof cutting schemes are designed based on theoretical analysis and practical experience.

4.1.1 Roof cutting height scheme

The roof cutting height exerts certain effect on the bulking of the roof rock. The theoretical roof cutting height can be calculated by (Groccia et al. 2016):

where \(h_{\text{r}}\) is the roof cutting height, \(h_{\text{c}}\) is the mining height and \(b\) is the bulking coefficient.

The mining height of the coal seam in the study site is 4 m and the bulking coefficient of the roof rock is 1.38. Based on the theoretical calculation result and field condition, three scenarios for roof cutting heights (7.5, 9.0 and 10.5 m) were performed in the simulation and experiment. In scheme types H1, H2 and H3, the roof cutting degree and angle were used as invariants. The details of the roof cutting height scheme are shown in Table 3.

The roof cutting height can affect the stress transfer among the gob roof and roadway roof, and thus exert an influence on the deformation, stress distribution and stabilization of the roadway surroundings. To comparatively analyze the effects of the roof cutting height, the structural deformation of the retained roadway was investigated by numerical simulation, and the loads of the hydraulic support and stabilizing process of the gangues were monitored in the field experiment.

4.1.2 Roof cutting degree scheme

The roof cutting degree is defined as the contact degree of the gob roof and roadway roof after adopting the DSC technique. In the simulation, a roof cutting interface was built between the gob and roadway roof. By changing the interface parameters, the effects of roof cutting degree can be achieved in the simulation. In the field experiment, the charging amount can directly affect the roof cutting degree. The evaluation of cutting degree in the field experiment was realized by changing the charging structure. Three schemes for roof cutting degree were designed in the field. The charging amount for each hole of the three schemes were 1200, 1600 and 2800 g, corresponding to roof cutting degrees of weak, medium and well. In scheme types D1, D2 and D3, the roof cutting height and angle were used as invariants. The details of the roof cutting degree scheme are shown in Table 4.

The roof cutting degree can affect the frictional force between the roadway and gob roof during the caving and compacting process of the gob roof, and thus influence the structural deformation of the retained roadway. Besides, in the blasting process for roof cutting, the amounts of explosives cause the fragment size of the gangues to be different. For this reason, the caving span of the main roof for different roof cutting degree varies. In this study, to evaluate the effects of roof cutting degree, the structural deformation of the retained roadway was investigated by numerical simulation and the caving span of the main roof was recorded by field monitoring.

4.1.3 Roof cutting angle scheme

To explore the influence of the roof cutting direction on the stability of the retained roadway, three schemes were designed in the simulation and field experiment. The roof cutting angles are 0°, 10° and 20°, respectively. In scheme types A1, A2 and A3, the roof cutting height and degree were used as invariants. The roof cutting angle schemes are shown in Table 5.

The roof cutting angle affects the caving and movement of the gob roof. For a different roof cutting angle, the caving speed of the gob roof is different. In this study, the stability of the retained roadway was investigated by the DEM simulation. The caving speed of the gob roof behind the hydraulic support was monitored in the field considering the inconsistency of the support moving and roof caving. To quantificationally describe the caving speed of the gob roof in different roof cutting angle schemes, evaluation indicators of \(\delta_{\text{M}}\) and \(\delta_{\text{A}}\) were established.

where, \(\delta_{\text{M}}\) and \(\delta_{\text{A}}\) are the maximum and average areas of the unfilled room, m²; L₁ and L₂ are lengths of the topline and baseline of the unfilled area; h_r is the height of the retained roadway, as shown in Fig. 6.

Indication of the caving speed behind the hydraulic support

4.2 Research methodology

The states and stability of the roadway can be reflected by many aspects such as the strata behavior, surrounding rock deformation and stress distribution. Numerical simulation based on discrete element method (DEM) and field experiment were complementarily used in our study to evaluate the roof cutting parameters.

4.2.1 DEM numerical simulation

The discrete element method is used to simulate roof caving and deformation problems. During the simulation process, the noncontinuous media is represented by discrete blocks, and the discontinuous surface is the boundary surface of the blocks. The blocks or elements can shift, rotate or separate along the discontinuous surface.

1. (1)

   Constitutive model.

The Mohr–Coulomb elastoplastic constitutive model that considers the tensile strength is used for the deformable block material. The failure criterion is represented in the plane (\(\sigma_{1}\), \(\sigma_{3}\)), as illustrated in Fig. 7. From point A to B, the failure envelope is defined by the Mohr–Coulomb yield function (Group 2011):

where \(f^{\text{s}}\) is the shear yield function, \(\sigma_{1}\) and \(\sigma_{3}\) are the maximum and minimum principal stress, respectively, and \(N_{\phi }\) is a function related with the internal friction angle.

Mohr–Coulomb failure criterion used in the simulation

From point B to C, the failure envelope is defined by the tensile yield function:

where, \(f^{\text{t}}\) is the shear yield function; and \(\sigma_{\text{t}}\) is the tensile strength of the roof rock.

Various joint constitutive models are provided in UDEC. Among them, the Coulomb slip joint model with surfaces in contact, which fully considers the mechanical properties of the joint elastic stiffness, friction, cohesion, tensile strength and dilatancy properties, can provide a good linear description of the joint stiffness and yield limit. This joint model is especially suitable for an underground excavation simulation (Xia et al. 2013). Therefore, in this study, the Coulomb slip joint model with surfaces in contact was adopted to describe the joint behavior.

1. (2)

   Numerical model.

To explore the effects of roof cutting, numerical calculation models were established. The simulation was conducted according to the project background in the Ningtiaota coal mine. Details on specific geological conditions will be introduced later. The dimension of the model was 300 m long and 80 m high based on the geological column, as shown in Fig. 8. The calibrated mechanical parameters of the rock stratum in the DEM model are listed in Table 6. The horizontal displacement of the model boundary was restricted. It is worth mentioning that a subsidiary transport roadway was excavated adjacent to the formed roadway. The experiment was to explore the effects of roof cutting. The roof cutting was conducted along the side of the formed roadway. The effects of roof cutting were studied by changing the roof cutting parameters, including roof cutting height, degree and angle.

DEM numerical calculation model

4.2.2 Field experiment

The formed roadway is surrounded by the roof rock, floor rock, coal rib and gangue rib. The first three are formed during the early excavation stage, while the gangue rib is newly formed after mining. The formation and stabilized process of the gangue rib is an important indicator to reflect the stability of the formed roadway. Many behaviors in terms of roof strata pressure and displacement can reflect the formation and stabilized process of the gangue rib, such as the roof pressure in the gob, the compacting rate of the gangues and the caving speed of the gob roof. These behaviors are hard to be authentically demonstrated in the simulation. To better explore the influence of roof cutting, a section of the roadway was chosen to conduct the roof cutting experiment in the field. First, the rock cores of the roof were extracted using the core drilling rig. The rock samples were tested in the laboratory and the lithology of the roof could be obtained. Considering the roof lithology, the roof cutting and splitting schemes were then designed. Subsequently, a drilling machine was used to drill holes in the roadway roof. The depth and direction of the holes were determined according to the scheme. After the holes were formed, the explosives with DSC device were installed in the holes and the yellow mud was used to seal the holes. Finally, the explosion was conducted and the cracking effects were detected using the optical peephole device.The experimental process is shown in Fig. 9.

Experimental process of roof cutting in the field experiment

4.3 Results and analysis

4.3.1 Evaluation of roof cutting height effects

1. (1)

   Structural characteristics and deformation of the roadway surroundings.

The structural characteristics and deformation of the roadway surroundings under different roof cutting heights were obtained by the DEM simulation, as shown in Fig. 10. Generally, we can observe from the simulation results that the precrack effectively cuts the structural transfer of roadway and gob roof. The gob roof caves along the roof cutting line and a new roadway rib is formed by the bulking gangues. There are small dislocation deformations in the main roof, but the main roof above the roadway and gob is still a whole structure. Under different roof cutting conditions, the structural state of the caving rock strata and its bearing capacity to roof are different, thus causing differences in the roof deformations and roadway stabilities.

Structural characteristics and deformation of the roadway surroundings for different roof cutting heights

To explore the influence of roof cutting height, the roof cutting angles are all 10° tilting to the gob area and the parameters of the contact surface between the roadway roof and gob roof are the same. When the simulation is balanced, the structure and vertical displacement of the roadway surroundings were obtained for different roof cutting heights. To clearly show the caving and structural forms of the rock, the surroundings of the retained roadway were magnified.

When the roof cutting height is 7.5 m, the caved roof rock fills most of the mining room and becomes one side of the retained roadway. However, there is an obvious gap between the caved gangues and the upper main rock layer. The existence of the unfilled room causes the upper roof to sink and rebalance. In the process of reforming a balanced structure, the short cantilever beam structure is inevitably affected. Under this scheme, the overall deformation of the roof is the largest, reaching 725 mm. In addition, an obvious separation appears above the gob roof, and the roof separation reaches 2315 mm. In the second scheme, the rock mass in the gob is more bulked. The unfilled room between the caved gangues and the upper main roof is significantly reduced. The bulked gangues provide an oblique support for the roof structure. In this scheme, the rotational deformation of the overburden strata is significantly reduced compared with that in the first scheme. In addition, the overall deformation of the roadway roof is small, and the maximum sagging amount of the roof is appropriately 214 mm, which is 70% lower than that in scheme one. Thus, the stability of the roadway is improved after increasing the height. Furthermore, when the roof cutting height is 10.5 m, there is almost no unfilled room left between the caved gangues and upper main roof. The gob roof is fully bulked. The maximum deformation of the roadway roof is 232 mm, which is slightly larger than that in the second scheme. The short cantilever beam structure formed by roof cutting is closely related to its surrounding rocks. When the roof cutting height is too large, the weight of the short cantilever beam structure will increase. Although the gob roof is fully bulked, the deformation of the roadway roof may increase due to its own weight.

It can be concluded that the roof cutting height affects the bulking and filling degree of the gangues in the gob. When the roof is not adequately cut, the filling of the roadway side is incomplete, and the stability of the roadway is reduced. In contrast, an excessive roof cutting height increases the construction difficulties and the weight of the roof cutting roof structure. A reasonable roof cutting height can cause the roadway side to be filled with bulking gangues. In the field, the stability of the roadway and the workability and economy of operation should be comprehensively considered to design the roof cutting height. Considering the construction cost and amount of mining work required, the roof cutting height of 9.0 m for the test site is reasonable.

1. (2)

   Pressure of the roof strata.

In the field experiment, 173 hydraulic supports were installed along the mining wall in the S1201 working face. The KJ216 mine pressure monitoring apparatus were installed on the hydraulic supports. The loads of the hydraulic support which is at a distance of 5 m from the roof cutting line were monitored. This monitored position was within the affected area of roof cutting. Each scheme was conducted for 50 m along the mining direction.

The monitored loads of the hydraulic support under different roof cutting schemes are shown in Fig. 11. We can understand that the loads of the hydraulic support are related with the roof cutting height. As the roof cutting height increases, the loads of the hydraulic support decrease. The average load in the second scheme decreased by appropriately one third compared with that when the roof cutting height is 7.5 m. As the roof cutting height increased, more pressure was prevented from transferring to the coal rib side, but instead the pressure was released in the gangues. In addition, the roof cutting height affects the bulking degree of the gangues in the gob. A reasonable roof cutting height is favorable for the bulking of the gob roof and for decreasing the rotational deformation of the main roof. Therefore, the pressure decreased on the hydraulic support and increased on the gangues after enhancing the roof cutting effects.

Measured loads of the hydraulic support under different roof cutting height schemes

1. (3)

   Stabilizing process of the gangues.

In the process of roadway formation, the gob roof rock mass experiences two major dynamic variation processes, i.e., caving and compacting. The compacting process can reflect the stability of the caved gangues. According to the compaction characteristics of the broken gangues and the movement laws of the main roof, the caved gangues are far away from the active mining face, the compacting amount of the gangue rib increases and the real-time bulking coefficient gradually decreases. In the practical engineering of automatic roadway formation, considering that the volume of the caved roof rock in the gob is difficult to be accurately measured, it is simplified as the ratio of the height of the stacked roof rock after fragmentation and the height of the intact roof rock before caving. First, the fluorescent paint was used to mark the height H₀ in the blast holes in front of the mining face. As the panel was retreated, the roof caved and the gangues were gradually compressed under the action of the roof weight. The height of the marked roof rock after fragmenting H_n was measured every 10 m with mining at the monitoring position. The real-time coefficient can be then calculated by H_n/H₀. The compacting process of the gangues can be described by a rheological model (Tan et al. 2015; Gao et al. 2019). The basic form of the equation is expressed as

where, \(K_{{\text{A}}}\) is the bulking coefficient of the stable roof rock; \(K_{l}\) is the real-time bulking coefficient at the position where the distance between the gangue and active mining face is l; a, b and c are constants.

Under different roof cutting conditions, the real-time bulking coefficient and the required distance for the gangue to be stable are different. According to the measured data and Eq. (16), the fitted curve for the bulking coefficient under different roof cutting heights are shown in Fig. 12. From the fitted results, the stable bulking coefficients of the roof siltstone under roof cutting heights of 7.5 m, 9.0 m and 10.5 m are 1.387, 1.381 and 1.380, respectively, which are not very different from one another. The main reason is that the charging structure in the siltstone section is the same for the three roof cutting schemes. From the fitted curves for the three roof cutting heights, the a, b and c in the Eq. (16) can be obtained. For a specific roof stratum, the K_A is a constant value. When the K_l is equal to the K_A, it indicates that the gangues in the monitored position are stable, and the distance between the gangue and active mining face l in the Eq. (16) can be then calculated. In the first scheme, the required distance for the gangue to be stable is appropriately 155 m. As the roof cutting heights increased to 9.0 and 10.5 m, the required distances decreased to 118 and 96 m, respectively. Therefore, as the roof cutting height increased, the required time and distance for the gangue to stabilize was reduced.

Variation of the bulking and stabilizing processes of the gangues

4.3.2 Evaluation of roof cutting degree effects

1. (1)

   Structural characteristics and deformation of the roadway surroundings.

The charging structure of the blasting can directly influence the crack propagation and hole connection degree. In the field test, the amounts of explosives in the hole can be used to distinguish different roof cutting degrees. In the numerical simulation, the parameters of the prebuilt roof cutting interface was changed to reflect the effects of the roof cutting degree. As the interface parameters increase, the simulated roof cutting degree is weakened.

When the holes are well connected by cracks, the caved gangues at the gob are neatly stacked, as shown in Fig. 13a. In the first scheme, the contact constraint between the gob roof and caved gangues is small and the drag frictional force acting on the roof structure is small. The roof deformation is the smallest and the maximum roof sag is 214 mm. When the roof cutting degree is reduced by half, the contact friction induced by the caved gangues and applied on the short cantilever beam structure is strengthened, which causes the roof deformation to increase. The maximum roof sag is 735 mm, which is 2.4 times greater than that in the first scheme. As the roof cutting degree continues to be reduced, the restriction is strengthened again and the upper roof rock layer causes a large lateral extrusion force on the roadway roof, as shown in Fig. 13c. At this time, the movement of the roadway roof is almost synchronous with the upper main roof above the gob and the maximum deformation of the roadway roof reaches 1024 mm. If no measures are taken to strengthen the roadway support, it is likely that the roadway roof will be cut off in its entirety.

Structural characteristics and deformation of the roadway surroundings for different roof cutting degrees

From the simulated results, it can be concluded that the roof cutting degree influences the contact and restriction degree between roadway roof and gangues, which then affects the roadway stability. When the splitting holes are well connected, the disturbance of the gob roof caving on the roadway roof is small, which is favorable for the stability of the roadway. Therefore, a presplit blasting test is suggested to be performed in advance to determine reasonable blasting parameters and to reduce the influence of gob roof caving on the stability of the retained roadway.

1. (2)

   Periodic caving span.

The roof cutting degree mainly influences the restraint degree between the gob roof and roadway roof. According to the blasting damage theory, the range of detonation damage varies for different explosive charging conditions. In the field experiment, the variation of the roof cutting degree was manifested by using different quantities of explosives. For the three blasting schemes, the average charging amount of a single blast hole increases sequentially, and the restraint between the gob roof and roadway roof is then weakened. It was found that the periodic caving span was obviously affected by the roof cutting degree.

The periodic caving span is associated with the movement of the upper main roof behind the active mining face. The periodic pressure span of the upper gob roof for different roof cutting degrees is shown in Fig. 14. When the average explosive charging amount is 1200 g per blast hole, the average periodic pressure span is 9.7 m and the maximum periodic pressure span is appropriately 12.5 m. As the explosive charging amount increased to 1600 g and 2800 g, the average periodic pressure span increased to 12.7 and 16.5 m, respectively. Generally, as the roof cutting effects are enhanced, the periodic pressure span shows an increasing trend.

Influence of roof cutting degree on the periodic caving span

The influence of roof cutting degree on the periodic pressure span of the main roof can be explained using the key stratum theory (Qian et al. 1996; Xu et al. 2010). Assuming that there are n rock layers above the working face and the load on these rock layers is evenly distributed, the thickness of each layer is h_i (i = 1, 2,…,n), the volume force is \(\lambda_{i}\) (i = 1, 2,…,n) and the elasticity modulus is \(E_{i}\) (i = 1, 2,…,n). The key rock layer that controls the working face pressure is the kth layer and m layers are controlled by the kth layer. When the roadway roof is not cut, the load induced by the kth layer is derived to be:

The periodic caving span of the main roof can be expressed by considering a cantilever-type fracture of the key rock strata.

where L is the periodic caving span of the main roof, \(h_{k}\) is the thickness of the kth layer, and \(R_{\text{T}}\) is the tensile strength of the rock mass in the key layer.

When the roof cutting degree is enhanced, the caved roof rock is more bulked. Before the key stratum reaches the critical fracture criterion, the bulked gangues support the roof rock stratum. This reduces the effective stress (q) acting on the key structure that controls the roof periodic weighting, and therefore, increases the caving span (L).

4.3.3 Evaluation of roof cutting angle effects

1. (1)

   Structural characteristics and deformation of the roadway surroundings.

The roof cutting direction should be optimized to reduce the caving disturbance,. In the study, three schemes of roof cutting angle are performed, and the angles between the roof cutting direction and vertical direction are 0°, 10° and 20°, respectively. The simulation results are shown in Fig. 15.

Structural characteristics and deformation of the roadway surroundings for different roof cutting angles

When the roof cutting is perpendicular to the roadway roof, the deformation of the roadway roof is the largest and the maximum deformation reaches 745 mm. Under this condition, a significant drag force is exerted on the roof cutting short cantilever beam structure during the caving process, large deformations in the roadway roof. In addition, the stabilized gangues only support the top main roof and cannot provide obvious diagonal support for the roadway roof. When the roof cutting angle is changed to 10°, the overall deformation of the roadway roof is reduced. The bulked gangues not only support the upper main roof but also provide a diagonal support to the roof cutting structure. However, when the roof cutting angle is continuously increased to 20°, the deformation of the roof is increased to 531 mm. Therefore, a reasonable roof cutting angle for the study site is 10°.

1. (2)

   Caving speed of the gangues.

The roof cutting angle can affect the interaction force between the roadway roof and gob roof. Field tests indicated that the roof cutting angle has the greatest influence on the caving speed of the gob roof. The variation of the maximum and average areas of the unfilled room (\(\delta_{\text{M}}\) and \(\delta_{\text{A}}\)) for different roof cutting angles is shown in Fig. 16.

Influence of roof cutting angle on the caving speed of gob roof

In the scheme A1, when the roof cutting line was perpendicular to the roof, the caving speed of the gob roof was the slowest. The maximum and average areas of the unfilled room were 6.0 and 4.9 m², respectively. When the roof cutting line tilted 10° to the gob area, the gob roof caved more easily. The maximum area of the unfilled room (\(\delta_{\text{M}}\)) reduced 30% to 4.2 m² and the average area of the unfilled room (\(\delta_{\text{A}}\)) reduced 58% to 2.1 m². Furthermore, the gob roof fell closely behind the hydraulic supports in scheme A3 with an average area of the unfilled room of 1.8 m². A long hanging roof may cause a strong dynamic pressure impact once it caves. Therefore, it can be seen that a reasonable roof cutting angle is favorable for gob roof caving and for reducing the risk of dynamic pressure impact.

Field engineering application was conducted in the Ningtiaota coal mine, located in Shenfu-Dongsheng coalfield of China. The coal seam mined at the S1201 mining panel is the Jurassic Middle Yan’an Formation coal seam, with a thickness of 3.95–4.45 m and a designed mining height of 4.17 m, which belongs to the category of thick coal seam. The coal seam has a dip angle of 0°–2° and a burial depth of 80–160 m. The width of the test mining panel is 295 m. The working face adopts a fully-mechanized coal mining technology and a caving method to manage the roof. The test site is in the S1201 head entry, with a width of 6.0 m and a height of 3.75 m. The entry roof was supported by rock bolts and anchor cables, and the solid rib was supported by cuttable glass steel rock bolts. During the field test, the energy-accumulated device was used to control the blasting energy and class 2 coal mine permissible emulsion explosives were applied to conduct blasting. The size of the emulsion explosive was φ27 × 300 mm, and each explosive weighed 200 g. Five DSC tubes were installed in each hole. The final number of explosives in each hole that presenting the roof cutting degree was 15 and the sealing length was 1.5 m, as shown in Fig. 17.

Schematic diagrams of the charging structure used in the field application

The optimum designed roof cutting height was designed to e 9.0 m and roof cutting angle was 10°. As shown in Fig. 18a, after adopting the DSC technique, two directional cracks were generated. The gob roof caves along the designed position and a new rib consisting of gangues is formed (Fig. 18b and c). The final cross-section of the formed roadway was satisfactory and fully met the requirement of next mining panel, which verified that the parameters of DSC technique and roof cutting used in the field were reasonable.

Application effects of the DSC technique for roof cutting and roadway formation

The automatic roadway formation method is an emerging non-pillar mining method that can effectively save coal resources and reduce drivage ratio. In this method, the natural roof rock is fully utilized to form a roadway by roof cutting. In our study, a directionally single cracking technique was introduced to conduct roof cutting. The mechanism of the DSC technique and the influence of roof cutting were comprehensively evaluated by mechanical analysis, numerical simulation and field experiment. The main conclusions are as follows:

1. (1)

   To explore the cracking mechanism of the DSC technique, the crack propagation laws and damage process in a roof rock using the conventional blasting and DSC technique were compared. It was found that, in the conventional blasting, the randomly propagated cracks caused the integrity failure of the roadway roof. In contrast, when roof cutting was conducted using the DSC technique, the effective stress in the energy accumulated direction was more concentrated, the crack propagated in a single desired direction, and therefore, the integrity of the roadway roof was well maintained.

2. (2)

   The influence of roof cutting on the behavior of roadway surroundings was explored by discrete element method and field experiment. The roof cutting height, degree and angle were chosen as experimental variables to evaluate the roof cutting effects. It was found that the roof cutting height mainly affected the bulking and filling degree of the gangues in the gob. Increasing the roof cutting height was favorable for reducing the roof pressure on the hydraulic supports. The roof cutting degree influenced the contact and restriction degree of roadway roof and gangues. When the roof splitting holes were well connected, the caving disturbance of the gob roof on the roadway roof was reduced, and the periodic pressure span showed an increasing trend. The roof cutting angle was related with the caving timeliness. A reasonable tilting angle of the roof cutting to the gob direction was better for the roadway stability.

3. (3)

   The optimized roof cutting technique was tested in the field application. Each hole was installed five DSC tubes with explosives. The gob roof caves along the roof cutting position and the caved gangues support the overlying strata. The experimental effects in the field verified the rationality and effectiveness of the DSC technique and the evaluated roof cutting parameters.