Research on Macro-Micro Failure Mechanism of Interbedded Sandstone in the Deep of Lannigou Gold Mine

At present, more and more Chinese underground mines are entering the deep mining stage. Deep high-stress and deterioration of lithology cause a great burden on support and mining safety [1–3]. The Lannigou gold mine is China’s largest Carlin-type gold mine, despite its difficulty to mine, beneficiate, and smelt the ore. The ore bodies and surrounding rocks are mainly sandstone and claystone [4], and the quality of the ore rock is poor. As the ore bodies is controlled by the fault zone and subjected to multiple geological periods, tectonic action, fractured structures, bedding, and joint fissures are well developed. The ore bodies and surrounding rocks are relatively fragmented, particularly sandstone sections with interlayers, the self-stabilization ability of which is poor. As a result, the mine is facing problems such as roadway deformation, slab bottom heave, and roof fall [5–7]. Therefore, studying macroscopic and microscopic failure mechanisms of interbedded sandstone offers certain guiding significance for exploration engineering.

Through many physical experiments, scholars have discovered that the influence of different types of interlayers on the mechanical properties of rocks is anisotropic. (1) A weak interlayer greatly weakens the strength of the intact rock [8]. (2) Taking the anisotropic inclination of the interlayer as example, when the interlayer changes from 0° to 90°, the strength of the sample shows a U-shaped change trend that first decreases and then increases [9]. When the angle between the plane and the principal stress is either 0° or 90°, the strength of the specimen is higher and when the interlayer inclination angle ranges from 30° to 45°, the strength is lower [10, 11]. (3) The strength characteristics of rocks are closely related to their failure modes [12]. The macroscopic failure modes of the anisotropy test have been classified. When the weak face angle is 0° or 90°, the specimen undergoes tensile failure because of vertical separation of the matrix and the interlayer. When the weak face angle is 15°, 30°, or 45°, the specimen undergoes composite failure caused by a mixture of multiple failure modes. When the weak plane dip angle is 60° or 75°, the rock specimen will show slip and shear failure either along with the layer or across the layer [13–15]. However, rock is a heterogeneous material containing many micropores and microstructure surfaces [16, 17]. Rock failure leads to the anisotropy of its microscopic damage. The proportion of different mineral components in the rock, porosity, and cementation type will affect its strength and control the damage characteristics of rock under load conditions [18–21]. To further study the macroscopic and microscopic failure mechanism of rock, previous studies used scanning electron microscopy (SEM) to explore the microscopic morphology and crack distribution of the rock damage process. For example, when brittle rock is subjected to load, microcracks are generated both from the inside and boundary of the crystal. This generation of microcracks causes an increase of rock porosity, thereby reducing its strength [22, 23]. Under impact load, mineral particles undergo a single intergranular failure, and the fractured section shows clear directionality. The relatively independent primitive micropores of the acid-etched sandstone are penetrated, causing macroscopic rupture [24, 25]. Dry-wet cycles increase the stress on the microstructure, causing irreversible fatigue damage; moreover, microcracks are also caused by uneven shrinkage of hard minerals. Based on the above research, the microscopic failure mechanism of rock differs according to different conditions [26, 27]. Therefore, SEM can be used to analyze the microscopic characteristics of rock after failure under triaxial compression and to determine the macroscopic physical properties of rock samples.

Although the above studies have enriched the anisotropy theory of interbedded rocks and the understanding of microscopic failure mechanism of rocks, problems remain. Predecessors have mostly focused on the macroscopic failure characteristics of interbedded sandstone under uniaxial compression, but it is difficult to restore the original rock stress state of rock in deep mining roadways. Also, the microscopic morphology characteristics of anisotropic rocks after failure have not been explored in depth. To fill these gaps, this paper takes the interbedded sandstone in the deep part of the Lannigou gold mine as research object. Through triaxial compression tests, the macroscopic physical characteristics of sandstone with interlayer (such as peak stress, peak strain, and deformation parameters) are explored under different confining pressure conditions. SEM is employed to analyze the microscopic morphological characteristics of the fracture surface under different macroscopic failure modes. In this research, the macroscopic and microscopic connection and failure mechanism of sandstone with interlayer after destruction are explored. Thus, a theoretical basis for the stability and support of roadway surrounding rock during the deep mining of Lannigou gold mine is provided.

The Lannigou gold mine is located on the southwestern edge of the Yangtze paraplatform and the north side of the Youjiang fold belt. This gold mine is a fault-controlled gold deposit (see Figure 1(a)). The ore bodies and surrounding rocks are mainly sandstone and claystone. Strong geological tectonic movement causes the regional tectonic stress to change many times, resulting in stratum turn over and causing the principal stress and the stratum to present different directions. Moreover, the tectonic movement caused a thrust of the fault, resulting in intrusion of argillaceous rocks and formation of a rich barrier with argillaceous interlayer structure. Finally, a roadway is formed with apparent interbedded surrounding rocks and broken rock inclinations (see Figure 1(b)). This structure has soft and broken engineering geological characteristics, weathers easily, disintegrates, and becomes muddy in contact with water, identifying it as a typical poor rock mass. In addition, the mine uses ramps to open. High-stress concentration areas form easily in different fault occurrence positions and ground stress is an important factor in determining the stability of the surrounding rock [28, 29]. As the mining continued to extend into deeper layers, there were many incidents of splintering, roof falling, and bottom drum in the roadway, which induced striking safety hazards to daily mining activities [30, 31]. Predecessors have shown that from the field measurement results of crystal stress changes with buried depth (see Figure 1(c)) that the maximum principal stress of the existing roadway (600 m underground) has exceeded 20 MPa. With increasing mining depth, the stress reaches 30 MPa. The surrounding rock of the roadway is in a high-stress environment [32].

To ensure homogeneity of test rock specimens, specimens were collected from the surrounding rock of the roadway of the Lower Triassic Xuman Formation in the mines of Lannigou gold mine 600 m underground. The lithology is sandstone and argillaceous sandstone with well-developed bedding. The test preparation process is described in the following: (1) The specimen is processed into a cylindrical standard test piece with a size of $50mm×100mm$ after drilling and grinding (see Figure 2(b)). The error of the test piece meets the requirements of the “Engineering Rock Mass Test Method Standard”. (2) Specimens are manually drilled with four inclination angles of 0°, 30°, 60°, and 90° (see Figure 2(a)), and the core angle β is defined as the number of angles between the drilling direction and the interlayer (see Figure 2(b)). (3) The longitudinal wave velocity of the prepared rock specimen is tested. The test found that at $β=0°$⁠, the wave velocity is smallest (2450 m/s). At $β=90°$⁠, the longitudinal wave velocity is largest (3180 m/s). Longitudinal wave velocity gradually increases with increasing β and anisotropy is apparent. (4) According to longitudinal wave velocity results, specimens with the same interbed angle and wave velocity concentration are selected as test rock specimens.

In the College of Civil Engineering, Guizhou University, the triaxial compression experiment was carried out using the rock mechanics test system (RMT-301) and developed by the Institute of Rock and Soil Mechanics, Chinese Academy of Sciences. Based on the original rock stress environment, the design confining pressures are 0 MPa (uniaxial), 10 MPa, 20 MPa, and 30 MPa for four cases. First, stress control is used to apply axial pressure and confining pressure at the same time at a rate of 0.05 MPa/s to reach the hydrostatic pressure level, while keeping the confining pressure unchanged. Then, displacement is controlled to be axially loaded at a rate of 0.002 mm/s until the rock specimen is broken. SEM was used to observe the micromorphology of broken surfaces.

Interlayer and confining pressure jointly control the macroscopic failure mode of anisotropic rock specimens. The red dashed line in Figure 3 represents tensile failure, the yellow dotted line represents shear failure, and the white line shows the position of the interlayer. The thickness and length of the dashed line on the failure surface are used to represent the development of the above two types of cracks. When the red dashed line is dominant and the yellow dashed line is less apparent, the specimen shows tensile-shear failure. Otherwise, it shows shear-tension failure. The number and severity of shear cracks and tensile cracks are identical, which is called composite failure.

For specimens with $β=0°$⁠, tensile-shear failure occurs under the condition of a confining pressure of 0 MPa. Multiple main cracks penetrate the matrix and interlayer axially from the end of the specimen, dividing them into several small cylinders. In the continuous loading of the small cylinder, the middle part of the small cylinder undergoes shear failure, which shows the characteristics of instability of the compression rod. Moreover, when the main crack encounters the interlayer along the propagation path, it forms a secondary crack that extends laterally along the interlayer. The interlayer with $β=0°$ has the most apparent crack propagation-inducing effect. With increasing confining pressure, the rock undergoes shear-tension failure, the main crack is reduced to one, and the axial expansion through the two ends gradually changes towards the lateral direction. The fracture angle increases. When the main crack passes through the interlayer, many small cracks are formed and the cracks will cause the rock fragments to fall off. Therefore, the main crack at the interlayer is wider than that at the matrix. Tensile cracks are formed from the sliding surface and expand in axial direction.

For specimens with $β=30°$⁠, composite failure occurs under the condition of a confining pressure of 0 MPa. The failure mode is more complicated. From the end to the middle of the specimen, the formed tensile cracks are mainly axially penetrating. After encountering the middle interlayer, local slip failure occurs along the interlayer and cracks. Under restricted confining pressure, the crack lateral expansion trend continues to penetrate the matrix and interlayer axially, causing tensile failure. When the crack extends to the bottom interlayer, slip failure occurs again along the interlayer. As the confining pressure increases, the specimen undergoes shear-tensile failure, resulting in a multilayer shear slip failure, where the fracture surface is mainly concentrated at the top and the bottom of the interlayer. Additionally, when the weak surface of the end slips, there are more axial tension cracks. The tensile crack on the upper part of the specimen extends upward to the top, and cracks on the lower part of the specimen are caused by the combined effects of slip failure and tensile failure, thus dividing the sample into multiple pieces.

For specimens with $β=60°$⁠, composite failure occurs under the condition of confining pressure of 0 MPa. Tensile cracks at the top of the sample are clearly visible. When the cracks extend to the middle and lower interlayer, more shear cracks are generated along the interlayer. Relative occurs locally slip between the shear crack and the interlayer plane. The length of each shear crack is short, but their number is large. With increasing confining pressure, the specimen undergoes shear-tensile failure, undergoing single-layer to multilayer sliding shear failure along the interlayer. At the beginning of the top slip crack, many small cracks diverge to the surroundings. Small cracks appear jagged. Layered cracks tend to slip and nonlayered cracks stretch the matrix. Under the action of large confining pressure, the interior of the shear surface is mixed with many rock fragments. There are many random small cracks under the shear cracks, which penetrate both the matrix and the interlayer.

For specimens with $β=90°$⁠, tensile-shear failure occurs under the condition of confining pressure of 0 MPa. Its rupture shape shows clear characteristics of instability of the compression rod. The tensile surface penetrated the specimen along the weak plane. Part of the “rod” is sheared and broken in the middle of the body. With increasing confining pressure, the instability characteristics of the pressure bar that rupture along the interlayer gradually decrease and the specimen gradually undergoes shear tension. The shear slip surface penetrates both the matrix and the interlayer, and the shear fracture angle gradually increases, eventually forming destructive shear failure in a diagonal line.

The failure mode of anisotropic interbedded sandstone under different confining pressures is relatively complicated, as it is the result of a combination of multiple destruction methods. The damage mode can be summarized into three modes: tension-shear failure, shear-tension failure, and composite failure. Various failure modes are controlled by the relationship between interlayer angle and confining pressure. Among them, there are two failure modes (i.e., tension-shear failure and composite failure) in the case of uniaxial compression. With increasing interlayer angle, the failure mode of the specimen is dominated by tensile failure. Tensile-shear failure (0°) that penetrates the matrix and the interlayer occurs and gradually develops into composite failure (30° and 60°) controlled by both shear failure and tensile failure. Finally, it is transformed to mainly tensile failure, along the weak side of the interlayer tensile-shear failure (90°). In case of triaxial compression, the failure mode is shear-tension failure, the specimens mainly undergo shear-tension failure through the matrix and interlayer, while shear-tension failure mainly occurs because of slippage on the weak surface. As increasing confining pressure gradually limits the cracking and sliding of the weak face of the interlayer, the effect of the weak face of the interlayer on the failure mode of the anisotropic specimen gradually weakens, resulting in shear failure being the ultimate failure mode.

The prepeak deformation process is usually divided into the original crack closure stage (from o to a), elastic deformation stage (from a to b), stable crack expansion stage (from b to c), and unstable crack expansion stage (from c to d) [33–35]. According to Figure 4, under different confining pressures, the postpeak stage of stress-strain curves of interbedded brittle rocks shows different characteristics. Therefore, to facilitate analysis of the underlying mechanism, in this paper, the postpeak stress drop stage (from d to e) is added (Figure 5). Under different confining pressures (⁠$σ3$⁠), uniaxial and triaxial compressive stress-strain curves of typical specimens are shown in Figure 4.

All specimen roughly went through the above five stages. The initial loading of the o–a section compacts the original defects of the specimen. When the specimen has no confining pressure, the curve has shown a clear concave upward trend. When the confining pressure is limited, the o–a segment is consistent with the a–b segment, which shows linear elastic characteristics. After the b-segment crack is closed, the curve becomes a straight line and the specimen shows elastic deformation. As the confining pressure increases at this stage, the elastic deformation characteristics of the specimen are more apparent, and the curve change trend of the anisotropic specimen tends to be uniform. The weak part of the b–c section specimen is affected by the loading stress. Cracks generate and expand. At this stage, the slope of the curve gradually decreases with confining pressure, and the yield stage gradually becomes more apparent, the faster the specimen enters the c–d stage. The characteristics of the curve change at this stage are consistent with the b–c segment. The cracks in the c–d segment penetrate through a large area, the bearing capacity of the specimen reaches the peak value, and irreversible damage occurs. With increasing confining pressure, the brittle failure characteristics at the peak stress weaken, and the specimen gradually transforms to ductile failure. In section d–e, specimen cracks further penetrate and the bearing capacity is greatly reduced. Under the condition of zero confining pressure, the curve with obvious strain-softening characteristics quickly drops after the specimen reaches the peak stress. After confining pressure is applied, friction emerges between specimen fragments, so that there is still a certain bearing capacity after the peak stress. Furthermore, stress gradually decreases with the decreasing rate of strain, and the residual strength increases with increasing confining pressure. It is also observed that in the vicinity of the c–d and d–e sections, the curves of selected confining pressure rock specimens show sawtooth fluctuations. However, this fluctuation phenomenon gradually decreases with increasing confining pressure.

The a–b stage represents the elastic characteristics of brittle rock. The failure of brittle rock is mainly controlled by internal cracks, which are abundant in interlayers and each sample contains multiple interlayers. Therefore, at zero confining pressure, the specimen shows multiple brittle failures, resulting in multiple stress drops and the fluctuation phenomenon. However, with increasing confining pressure, the cracks in the interlayer and the rock are restricted, so the fluctuation phenomenon decreases.

The specimen is restricted by the confining pressure and can preliminarily close the original rock cracks and cavities, which makes the specimen show clear elastic deformation. When cracks occur, the specimen will enter the yield stage relatively quickly, causing the specimen to change from brittle failure to ductile failure. After destruction, it has a higher residual strength.

Figure 6 shows the change of specimen strength with $β$ under different confining pressures. With increasing $β$⁠, the compressive strength of the specimen shows a “spoon”-shaped pattern of slowly decreasing-decreasing-increasing. Among them, when $β$ is 0° or 90°, the compressive strength is relatively large. At about 60°, the compressive strength is significantly smaller. The compressive strength of rock samples with $0°<β<30°$ shows a slower decrease trend. At $30°<β<60°$⁠, the compressive strength of the rock specimen shows a rapid decrease trend. At $60°<β<90°$⁠, the compressive strength of the rock specimen shows a rapid increase trend.

Figure 7 shows $S60°σ3$ are all the largest. Average values of $Sβσ3$ decrease with increasing confining pressures (⁠$Sβ0MPa=0.21$⁠, $Sβ10MPa=0.13$⁠, $Sβ20MPa=0.10$⁠, and $Sβ30MPa=0.04$⁠). Specifically, the interlayer has the greatest weakening effect on strength when $β=60°$ and increasing confining pressure limits the sliding of interlayer.

Figure 8 shows the variation of $σf$ and $εf$ with $σ3$ for samples with different inclination angles. Fitting each value showed that the degree of fit follows a linear relationship above 0.9, and the fitting effect is good. $σf$ and $εf$ increase linearly with increasing $σ3$⁠.

Comparison of $σf$ when $σ3$ is 0 MPa and $σ3$ is 30 MPa, when $σ3=0$ MPa, shows that the values are scattered, and the difference between maximum value and minimum value is 23 MPa. However, at $σ3=30$ MPa, most specimens roughly overlap at 155 MPa, and $β=60°$is a special case. However, compared with other $σ3$⁠, $σf$ increases significantly when $σ3=30$ MPa, indicating that higher confining pressure can limit the slip shear failure of the weak plane at $β=60°$⁠. Furthermore, it can weaken the lateral deformation trend, increase the stability of the sample and the integrity of the rock sample, and enhance the resistance to axial pressure, which results in an increase of $σf$⁠. The anisotropy of $σf$ is apparent when $σ3$ is low, and the mechanical effect of the interlayer is weak when $σ3$ is high.

The increasing rate of $σf$ with increasing $σ3$ is $30°>90°>60°>0°$⁠, and the increasing rate of $εf$ with increasing $σ3$ is $30°>60°>90°>0°$⁠. This is because when $β=30°$⁠, less slip failure occurs in the uniaxial compression test, and the slip phenomenon is partially weakened after application of confining pressure. However, shear and slip failures still exist; therefore, $σf$ and $εf$ increase the most. At $β=90°$⁠, under uniaxial compression conditions, the external interlayer is detached, and cracks occur at the internal lithologic boundary, weakening sample integrity. However, confining pressure limits the separation of the lithologic boundary, which significantly improves the integrity of the rock. Therefore, the strength of the sample increases large because $β=90°$ specimens generally show compression rod phenomenon during the failure process of different confining pressures, the main failure mode is tension failure, and the strain difference is smaller than if the main failure mode is shear and slip failure. At $β=60°$⁠, the slip phenomenon is most apparent and has a limited effect on the slip failure at a confining pressure range of 20 MPa to 30 MPa. However, compared with the whole process of 0 MPa to 30 MPa, the limit on the slip failure is very restricted. Therefore, the increase rate of $σf$ is small, while that of $εf$ is still large.

The elastic modulus is taken as the slope of the approximate straight-line segment of the stress-strain curve [40].

According to the horizontal comparison shown in Figure 9, the elastic modulus at β =0° is the smallest, while that of $β=90°$ is the largest. The elastic modulus increases with increasing $β$⁠. Furthermore, linear fitting of E showed that the fitting $R2$ of elastic modulus under different confining pressures is greater than 0.95, indicating that the elastic modulus has a linear relationship with $β$ under the same confining pressures. The longitudinal comparison in Figure 9 shows that the elastic modulus of anisotropic specimens increases significantly when confining pressure is limited (by 17% on average in the process of 0 MPa to 10 MPa). However, when the confining pressure ranged from 10 MPa to 30 MPa, the elastic modulus of each sample changed a little. The elastic modulus of 0°, 30°, 60°, and 90° increased by -2%, 5%, 4%, and 7%, respectively. The confining pressure significantly influences the elastic modulus of the specimen, but with increasing confining pressure, the influence decreases. When the confining pressure ranges from 10 MPa to 30 MPa, the elastic modulus of $β=0°$ decreases because damage at the interlayer of the specimen is caused at the initial compaction stage during the application of confining pressure. Although elastic modulus decreases, the change is lowest, indicating that the increasing confining pressure has the least influence on the elastic modulus of the specimen with $β=0°$⁠. The reason for the increase of elastic modulus of other samples is that the increase of confining pressure exerts a binding effect on the transverse deformation of shear failure and peel tension failure. This has a greater impact on elastic modulus than interlayer damage.

The relationship between ultimate stress circle and strength line (see Figure 10).

By comparison, the envelope fitting effect is best for each specimen when the Mohr’s circles $σ3=0MPa$⁠, $σ3=10MPa$⁠, and $σ3=30MPa$ are selected (see Figure 11).

The strength parameters $φ$ and $c$ of the anisotropic sample can be obtained via strength curve fitting (see Figure 11). At $β=0°$⁠, they were 30.55° and 17.90 MPa; at $β=30°$⁠, they were 33.85° and 13.90 MPa; and at $β=90°$⁠, they were 32.83° and 14.74 MPa, respectively. The fitting effect was poor at $β=60°$ and the fitting limits of the two groups were 13.30 MPa and 23.26°, as well as 11.99 MPa and 28.57°, respectively. By analyzing the maximum strength parameters, the relationship between $φ$ and $c$ can be obtained as shown in Figure 12.

Figure 12 shows that the fitting curve of $φ$ and $c$ depicts an upward concave parabola with increasing $β$⁠, which is similar to that of $σf$⁠. At stratification angles of 0° and 90°, the $φ$ and $c$ are relatively large; at $β=60°$⁠, the $φ$ and $c$ are clearly small (taking the maximum value for the analysis, the actual $φ$ and $c$ of $β=60°$ are smaller than the analysis value). The minimum value and maximum value of $φ$ and $c$ of anisotropic samples are 14.91% and 34.59%, respectively. The variation of $φ$ and $c$ is similar to that of $σf$⁠, which is because they are closely related to the failure mode. Taking $β=60°$ as example, it can be calculated by Equation (3) that $βw$ is 59.29°, which is approximately equal to the interlayer angle. At $β=60°$⁠, the rock specimen mainly slips along the interlayer, and the interlayer is the main resistance to shear failure; therefore, the shear strength parameter is naturally low. For other interlayer specimens, $βw$ is also ~60°, but at a macroscopic level, the matrix and interlayer jointly resist shear failure; finally, composite failure of split tension and slip shear occurs. Therefore, the strength parameters of the specimens are relatively large.

The anisotropy of interbedded sandstones is formed under specific geological background conditions. X-ray diffraction (XRD) was used to conduct the analysis and the results showed that the main components of the matrix are quartz, and the main components of the interlayer are muscovite and clay minerals. However, the mineral contents of both components are basically identical, with little differences in the mineral content (see Figure 13).

From a macroscopic perspective (see Figure 14(a)), sandstone has distinct interlayers, changing between regular and developed interlayers and is darker than the matrix because of the inclusion of clay minerals and organic matter. To further observe the microscopic characteristics of the interlayer and the substrate, both components were assessed by SEM.

At the microscopic level (see Figure 14(b) and Figure 14(c)), the structural characteristics of the two substrates are significantly different. The matrix morphology is rough and uneven, and quartz and other mineral particles show clear contours and a high degree of crystallization. However, the particles are not distributed randomly, and there are a few rod-shaped primary fractures in the matrix. The interlayer surface shows a wavy pattern, and the direction of the wave is directional. The interlayer shows high homogeneity, a lack of large mineral particles, and a low degree of entropy. Magnified observation (see Figure 14(c), Figure 14(d), Figure 14(f), and Figure 14(g)) shows that the cementing material of the matrix is a flake or thin film, which forms porous cementation between particles. The matrix is dominated by quartz grains although there are large cracks in the matrix. In the loading process, quartz provides the main force particles. Contact between particles is close, the roundness is high, and most particles are fully spherical, with relatively high compressive strength. In the interlayer, many clastic cements are distributed on the quartz surface, and quartz particles have a lower degree of crystallization and content. Mineral particles are relatively uniform, small, and circular, and the particles are in a discrete state without contact. At the microscopic level, many circular micropores can be observed between particles, and the interlayer is bonded by clay minerals. The cementation mode is basement cementation, with weak connection ability. There is fine mica in the clay minerals, resulting in further reduction of the bearing capacity of the interlayer, leading to the failure of the interlayer in the process of specimen loading.

The main characteristics of interbedded sandstone are summarized as follows: quartz is the skeleton mineral in the matrix and has a high degree of crystallization and high strength, while the interlayer is mainly composed of mica and clay minerals, with many micropores between particles and an overall low strength. The interlayer will be destroyed first when stress is applied to the specimen.

The macroscopic properties of rock are closely related to its internal microstructures and mineral particles. Therefore, the relationship between microscopic fracture and macroscopic fracture of rock can be analyzed by observing the characteristics of its microscopic morphology. Based on the above macrofailure mode analysis, this relationship can be roughly divided into three failure modes of anisotropic sandstones containing interlayers under two stress states. Under uniaxial conditions, specimens mainly undergo tensile-shear failure and composite failure, while under confining pressure, only shear and tensile failure occur in the sample. Therefore, microscopic analysis was conducted from the three perspectives of tensile-shear failure, composite failure, and shear-tension failure (high-stress failure). The control effect of the interlayer inclination and confining pressure on the evolution of the microfracture surface is analyzed. The failure characteristics of anisotropic sandstone are discussed from the microscopic level by connecting the fracture characteristics of rock, the stress environment of rock, and the macroscopic failure mode.

Under uniaxial conditions, tensile-shear failures dominated by tensile failures occur at 0° and 90°. Figure 15(b) shows that the microfracture surface is stratified, and there are complete mineral particles and apparent primary cracks in the upper layer. The surface of mineral particles is smooth and the outline of primary cracks is clear, indicating that the rock sample fracture surface is damaged less in the process of tensile separation. Figure 15(c) and Figure 15(d) show that the damage of mineral particles can mainly be observed on the particle surface, and the inner wall of the primary fracture is smooth. Only tension-shear damage exists at the fracture mouth, which causes the mineral to be pulled out from the matrix and leave pits (see Figure 15(e) and Figure 15(g)). Mineral particles are affected by drawing and minerals are destroyed at weak edges and corners. Tensile and shear microcracks are mainly developed in cemented minerals around mineral particles, which are mainly clay and clastic. This is because the strength of mineral particles is higher than that of clastic minerals, and cemented minerals contain many micropores and microcracks (see Figure 15(f)). In addition, after mineral particles pull out the matrix, pits will remain. The microcracks expand from the pit to the surrounding area, forming a fracture group with a pit at the center and surrounding area.

Under uniaxial conditions, specimens with $β=30°$ and $β=60°$ undergo composite failure with a mixture of two failure modes. Figure 16(b) shows that the fracture line of the fracture surface of composite failure is long, indicating that the specimen produces directional failure, with a relatively single fracture failure mode. Further observation (see Figure 16(c)) shows that step patterns exist on the fracture surface, which formed with the mineral particles, indicating that this is not sufficient to cause significant damage to the mineral particles during the formation of the microfracture surface under shear damage. Rod-shaped debris exists above the step pattern, which is caused by dislocation and rolling between slip planes, and rod-shaped debris is formed. This further indicates that the specimens with composite failure also have shear and slip failure at the microscopic level. Figure 16(d) shows that there are small stepped patterns on the surface of mineral particles, while the overall mineral surface is relatively smooth. Moreover, pits formed indicating that the specimen with composite failure also shows the characteristics of tensile failure. Figure 16(e), Figure 16(f), and Figure 16(g) show that intergranular failure mainly occurred in the specimen, and microcracks formed around the crystal. In addition (see Figure 16(f) and Figure 16(g)), because of weak positions or apparent edges and corners in certain crystals, failure occurred in the stress process, but the positions where microcracks formed were mostly located outside the crystals.

Under triaxial compression, shear failure mainly occurs. As shown in Figure 17(a), compared with other failure modes, under high-stress conditions, a large number of white debris particles are attached to the macroscopic fracture surface of the specimen [41]. As shown in Figure 17(b), the shape of the original pit hole is not clearly defined, and the surrounding structure is loose. In Figure 17(c), the original hole is broken to produce a large number of debris particles in the process of filling the microscopic fracture surface. Furthermore, the macroscopic fracture surface shows self-similarity and there are more subtle clastic particles. Figure 17(d) shows that the fracture surface around the sliding groove has a clear direction, and the groove surface has many cone and strip clastic particles. The specimen is damaged in the process, as it shows apparent slip and shear breakage, and more side slips can be seen in the shape of debris particles. Under the condition of triaxial compression, the overall failure of the specimen is more severe. As shown in Figure 17(e), microcracks are more widely distributed and more disorderly (see Figure 17(f)). The crystal near the surface of the fracture is seriously damaged, mainly showing transgranular failure, while the particles far away from the fracture surface mainly show intergranular failure. After the destruction of external crystal particles, secondary destruction of particles with relatively complete shapes occurs (Figure 17(f)).

Based on the three failure modes under different confining pressures, a schematic figure of microcrack propagation is obtained (see Figure 18).

Parts of the pores of the specimen with $β=0°$ are compressed to form close pores and compact cracks. After pores are compacted, cracks are not easily continuing to expand and have high stability. However, other pores are subjected to transverse tension, and cracks form at both ends of the long axis. These cracks continue to propagate vertically and gradually connect and penetrate, and microcracks dominated by tensile failure occur. The combination of compact cracks and close cracks forms a weak surface, which has a guiding effect on the continuous propagation of tensile cracks, causing vertical cracks to change and local shear cracks to emerge, thus forming shear failure. The microcrack generation mechanism of $β=30°$ and $β=60°$ specimens is the same as that of $β=0°$ specimens. However, because of the weak faces of the interlayer, slip shear failure occurs during the cracking process, providing the macroscopic performance for composite failure. Based on the microcrack generation mechanism of the $β=0°$ specimen, the β =90° specimen also has the “pressure-bar effect” produced by vertical interlayers. Therefore, tensile failure is more apparent in the macroscopic form and finally formed tension shear failure. When triaxially compressed, the weak surface exerts little influence (the conclusion above), and slip shear does not occur easily. Therefore, the influence of slip on the microscopic damage is not considered, and interlayers with different inclination angles are ignored. All original pores are restricted by the confining pressure and become smaller. However, they cannot be completely closed, and only compact pores are formed, resulting in compact cracks. In the process of gradual compression, the compact cracks continue to expand. Therefore, the specimen shows more severe damage on the macroscopic and microscopic fracture surfaces. Transverse cracks and vertical cracks affect each other, causing the specimens to cross and penetrate. Both tensile failure and shear failure emerged. Because the specimen is subject to the end effect, and the crack extends to the surrounding surface of the specimen in a shorter path, the crack is more likely to break around the specimen. This causes more apparent shear failure, finally forming a shear-tension in the form of macroscopic appearance.

Figure 19 shows a 600 m underground horizontal roadway with a clear interlayer structure, and the structural dip $β=60°$ as shown in Figure 19(b). Observation of the roadway shows that the roof of the section is high from the left to the right, and there is rock peeling at the foot of the right side (see Figure 19(a)).

Based on the above experiments, the instability mechanism of the corresponding roadway is explored from two perspectives: First, the triaxial compression test results show that the roadway at $β=60°$ is affected by the interlayer and has a tendency to slip, which causes the roadway to be subjected to the shear force along the weak surface, causing an asymmetrical instability mode of the roadway. Under high confining pressure, the specimen compression is also affected by the weak surface, which leads to both shear failure and slip deformation. In essence, the strength of the specimen is not uniform, and the weak surface leads to the dominant occurrence of the slip phenomenon, which shows asymmetric deformation in the roadway instability. In Figure 19(b), several cracks are clearly visible in the tunneling face. These cracks are perpendicular to the weak surface of the slip and there is no shear slip trend. Therefore, it can be inferred that this crack is tensile. Based on triaxial tests, tensile cracks were found in the specimen. Because of the weak properties and more original fractures, there were more tensile cracks in the interlayer and their penetration was more apparent. This phenomenon is reflected in the roadway (Figure 19(b)) where the cracks in the interlayer are more apparent than those in the matrix. So, the roadway is mainly affected by slip and shear. The shear-tension failure of $β=60°$ is the same as the failure mode of the specimen.

Secondly, SEM test showed that the interlayer has many dense pores and poor physical properties, and thus, it can be inferred that it is fragile. Combined with the scene pictures, the rock in the interlayer zone is significantly more damaged than in the matrix zone (see Figure 19(c)). Moreover, the anchor rod located in the interlayer area is weak, and because of its crushing characteristics, the anchoring force of the anchor rod is greatly weakened, resulting in drawing failure and further making the surrounding rock more likely to peel and lose stability [42, 43]. Combined with the SEM analysis of shear-tensile failure, directional grooves can be found on the fracture surface. In the roadway, the macroscopic grooves tend to be the same as the interlayer (Figure 19(b)). Under high-stress conditions, considerable debris can be found on the fracture surface, and microcracks are short and disorderly. Macroscopic performance is generated by microscopic accumulation, especially in the tunnel interlayer. Rock particles are broken, shear cracks along the interlayer tendency are cut many times by tensile cracks, thus forming short and disorderly cracks with high strength matrix. Damage can be found inside and in the surrounding area.

In this case, studying the roadway with interlayer with $β=60°$ showed that the instability characteristics of the roadway have a certain degree of self-similarity with the laboratory test. As this paper only analyzes the case based on this inclination angle, the research on the roadway with interlayer in other dip angles still needs considerable on-site observations and statistics.

• (1)

  Different confining pressures yield different macroscopic failure modes of specimens. Without confining pressure, the specimens will undergo tensile-shear failure across the matrix and interlayer (⁠$β=0°$⁠) and along the weak surface of the interlayer (⁠$β=90°$⁠), as well as local slip along the interlayer and local composite failure across the matrix and interlayer (30° or 60°). Under high confining pressure, the specimen mainly undergoes shear failure which mainly penetrates through the matrix and the interlayer and locally slips along the interlayer

• (2)

  The peak strength of the specimen changes in a “spoon” shape with $β$⁠, and $ε$ and $φ$ are consistent with the peak strength. The specimen is restricted by confining pressure, which can initially close the cracks and cavities in the original rock, and exerts a restraining effect on the weak surface with a tendency of lateral sliding. Therefore, the peak strength, peak strain, residual strength, and elastic modulus of the specimen increase with increasing confining pressure. However, with increasing confining pressure, the increase in peak strength of the specimen decreases, and part of the interlayer will be damaged at the same time

• (3)

  The interlayer affects not only the strength of the specimen but also the macroscopic failure mode. $Sβσ3$ shows that with increasing confining pressure, the influence of the interlayer on the anisotropy of the strength of rock specimens gradually decreases. It reflects that with increasing confining pressure, the anisotropy of the interlayer to the macroscopic failure mode gradually decreases. The macroscopic failure mode of the specimen changes from tensile-shear failure and composite failure under uniaxial compression to shear-tension failure under triaxial compression

• (4)

  During the stress process of the specimen, the interlayer is mainly composed of mica and clay minerals and is destroyed first. When tensile-shear failure occurs, the crystal surface is damaged, and microcracks will expand to the surrounding area with the pit at the center. When composite failure occurs, the fracture surface is directional, the crystal will be damaged on the surface or the weak part of the crystal, and microcracks mainly form around the crystal. When shear-tension failure occurs, a considerable number of debris particles are generated on the fracture surface, and clear slip grooves appear. The distribution of numerous microcracks is disorderly, and transgranular damage and intergranular damage occur at the same time

All data are the latest data obtained through experiments and can be available from the corresponding author upon reasonable request.

The authors declare that there are no conflicts of interest.

This work was supported by the Guizhou Province Science and Technology Support Program Project (grant number QIANKEHE Support [2021] General 516), the National Natural Science Foundation of China (grant numbers 41962008, 52164006, and 51964007), the National Natural Science Foundation for Young Teachers of Guizhou University (Guizhou University Cultivation grant number [2020] 81), and Scientific and Technological Innovation Talents Team in Guizhou Province (grant number [2019] 5619).