Deformation and Failure Characteristics of Bimaterial Samples Consisting of Sandstone and Cemented Coal Gangue–Fly Ash Backfill under Uniaxial Loading

Abstract

Based on the acoustic emission (AE) system and the digital scattered-spot deformation monitoring system, uniaxial compression tests were conducted on composite samples consisting of sandstone and cemented coal gangue–fly ash backfill (CGFB) to investigate their deformation and failure characteristics. The results showed that the average uniaxial compressive strength of the composite samples was 83.09% higher than that of the pure CGFB samples and 92.28% lower than that of the pure sandstone samples. In the composite samples, damage occurred in the CGFB part, and they showed obvious plastic damage characteristics. On both sides of the intersection, the sandstone and the CGFB deformed synergistically in the absence of a macroscopic failure. After a macroscopic failure, the interface effect promoted sandstone deformation and restrained CGFB deformation, transforming the sandstone and the CGFB on both sides of the intersection into a nonsynergistically deformed state. The interface effect had the most obvious influence on the horizontal deformation of the sandstone and CGFB monitoring points near the intersection. The failure of the CGFB samples induced sandstone springback deformation with a springback capacity of 0.0089 mm in the vertical direction and 0.0055 mm in the horizontal direction, which led to the further rupture and failure of the CGFB.

1. Introduction

As the main energy source in China, coal has made important contributions to national economic construction and social development. However, the surface subsidence resulting from coal mining affects the ecological environment and surficial construction (structures) and brings many serious hidden dangers [1,2,3,4,5,6]. To protect their infrastructure, many coal mines control surface subsidence using filling methods. This technology processes solid waste, such as coal gangue and fly ash, into a paste-like slurry which is pumped to the underground workings through a pipeline to fill goaves. The proportion of fly ash in coal gangue–fly ash backfill (CGFB) is about 10%–40%, and the proportion of coal gangue is about 50%–70% [7]. Using coal gangue, fly ash, and other industrial waste as a cement filling material can also alleviate the problem of land occupation and environmental pollution caused by the solid waste from coal production [8].

To control mining costs, some coal mines in China adopt the strip-paste-filling mining technology, the basic principle of which is to use a paste cement material to construct interphase filling strips to support the overlying strata, to control surface subsidence, and to realize green mining in construction structures [9,10,11,12,13]. If the interval between CGFB bodies is greater than or equal to the roof caving interval, the roof will collapse to fill the goaf, and the goaf will form a roof–CGFB composite system that supports the overlying strata [9]. The mechanical properties of the combined system determine its ability to support the overlying strata as shown in Figure 1. Therefore, it is important to study the mechanical characteristics of the surrounding rock–CGFB composite system to effectively control surface subsidence and to ensure green and safe mining.

Many studies have been conducted on the mechanical properties of backfill. Zhao et al. [14] carried out uniaxial compression experiments on backfill samples with different ash and sand ratios and analyzed the mechanical properties and synergistic deformation characteristics of combinations of tailing-sand-cemented filling materials with different ash–sand ratios. Yin et al. [15] studied the deformation and failure patterns of mixed-aggregate backfill samples with different coarse-aggregate replacement rates, revealing the energy distribution evolution law of the deformation stage before the stress–strain curve peak of the sample. Hou et al. [16] carried out uniaxial compression tests of cemented backfill at different loading rates and explored the energy dissipation evolution process of the compression damage of cemented backfill, concluding that this process can be divided into four stages: initial damage, stable development of the damage, damage acceleration, and damage failure.

Wang et al. [17] conducted uniaxial compression and rheological tests on different types of backfills, concluding that backfill viscosity is a positive linear function of the uniaxial compressive strength and could, thus, be used to predict the uniaxial compressive strength. Behera et al. [18] analyzed the relationship between the strength development and microstructural evolution of fly-ash-filled pastes using electron microscopy–X-ray energy spectroscopy. Shao et al. [19] investigated the wave velocity, porosity, strength, and damage mode of filled samples with different laminar ratios and analyzed the effect of the laminar structure on the mechanical properties and damage mode of the filled bodies. Fang and Fall [20] investigated the effect of the maintenance temperature on the shear properties of the cemented backfill–rock interface. Zhang et al. [21] analyzed the effect of the initial air content in a new slurry on the compressive strength of backfill through the microscopic force analysis of cemented paste backfills with different initial aeration contents and established a corresponding mathematical model.

These studies are important for understanding the mechanical properties of backfill. However, there have been relatively few studies on rock–backfill composite structures. Chen et al. [13] studied the mechanical properties of the combined hard roof-CGFB system and concluded that the overall strength of the system is determined not only by its internal fill but also by the interaction between the roof and the fill. Tan et al. [22] explored the mode and mechanism of the pressure-bearing action of the composites of backfills and surrounding rocks through mechanical experiments on samples of backfills, surrounding rocks, backfill-wrapped surrounding rock composites, and surrounding-rock-wrapped backfill composites. Zhan et al. [23] conducted shear experiments to investigate the effect of dry and wet rock wall conditions on the shear properties of the backfill–surrounding rock interface.

These results provide an important reference for successfully carrying out experimental research on the deformation and failure characteristics of sandstone–CGFB composite samples under uniaxial compression. However, the results mainly focus on the mechanical properties of surrounding rock–backfill composites when the combination form and external conditions change. The deformation and failure characteristics as well as the energy evolution law of the surrounding rock–backfill body have not been studied. Therefore, in this paper, we carried out uniaxial compression–deformation failure evolution tests based on the acoustic emission (AE) system and the digital scattered-spot deformation monitoring system and studied the mechanical properties, deformation failure characteristics, and energy evolution law of sandstone–CGFB composite samples. The goal was to provide a theoretical basis for understanding and implementing strip-fill mining and for effectively controlling surface subsidence.

2. Materials and Methods

2.1. Sample Preparation

The rock samples used in this experiment were pieces of sandstone taken from the Liangjia Coal Mine in Yantai, Shandong Province, China. All samples were taken from the same block to prevent discrete rock samples from affecting the experimental results. In accordance with the mass ratio of cement, fly ash, and coal gangue of 1:4:6, the mass fraction was 78% of that of the CGFB samples. The cement used was P.042.5 silicate cement from Shandong Rizhao Third Cement Factory. The fly ash was from the Qingdao Huangdao Power Plant, Shandong Province. The coal gangue was from the Liangjia Coal Mine and was crushed and screened to less than 5 mm.

For this experiment, we prepared three pure sandstone samples, three CGFB samples, and three sandstone–CGFB composite samples.

To process and prepare sandstone samples, first, the rock block was cut into rectangular samples measuring 50 mm × 50 mm × 100 mm and square samples measuring 50 mm × 50 mm × 50 mm (length × width × height, respectively) using a cutting machine. Subsequently, a stone grinding machine was used to polish the six end surfaces of the samples to create smooth end surfaces with nonparallelism of less than 0.05 mm and an axial deviation of less than 0.25° on the corresponding end surfaces [24].

CGFB samples were prepared by mixing cement, fly ash, coal gangue, and water evenly according to the mass ratio and were loaded into a 50 mm × 100 mm mold. The samples were then placed in a curing chamber at 20 °C under a relative humidity of 95% and were then demolded and maintained for 28 d after 24 h of molding.

Sandstone–CGFB composite samples were prepared by placing 50 mm × 50 mm × 50 mm sandstone samples at the end of a 51 mm × 50 mm × 100 mm mold. Next, the filling slurry was poured directly into the mold and placed in a curing chamber at 20 °C under 95% relative humidity for 24 h and was then demolded.

Finally, the sandstone–CGFB composite samples were placed in a maintenance box for 28 d. After maintenance was completed, the samples were sandpapered to 50 mm × 50 mm × 100 mm. In the samples, the sandstone and the CGFB were bonded into a single unit by cement–fly ash in the paste slurry. The sandstone:CGFB ratio in the composite samples was 1:1.

Pure sandstone samples, CGFB samples, and sandstone–CGFB composite samples were numbered A1–A3, B1–B3, and C1–C3, respectively, as shown in Figure 2.

2.2. Experimental Program

Figure 3 shows the loading and monitoring system of this experiment, including a Shimadzu AG-X250 electronic universal testing machine produced by Shimadzu, Kyoto, Japan, a MISTRAS series PCI-2 AE system, and a digital scattering strain measurement system. The loading system, the AE system, and the digital scattering strain measurement system were synchronized to ensure that the three systems had the same time parameters [25,26,27,28].

The loading system was a Shimadzu AG-X250 electronic universal testing machine. It can achieve conventional compression, tensile, and other mechanical tests with a maximum load of 250 kN. The test was controlled with displacement loading until the coal sample was damaged. To better simulate the CGFB long-term pressure-bearing process, the loading rate was kept at 0.005 mm/s.

The deformation characteristics of the sandstone–CGFB composite samples were monitored, and three-dimensional strain field data were obtained using the digital scattering strain measurement system, an optical non-contact three-dimensional deformation measurement system for the measurement and analyzation of object surface morphology, displacement, and strain. Before the experiment, the scattered field was produced by manual spraying: first, white matte paint was evenly sprayed on the surface of each group of samples. Then, black matte paint was sprayed on the samples. The spray was allowed to fall randomly and formed a scattered field of black spots [29] as shown in Figure 4. During the experiment, the system charge-coupled device camera (5 million pixels) acquired images of the sample surfaces at a frequency of 4 frames/s. These images were transferred to a computer for processing, and the desired strain maps of the samples were finally obtained.

The MISTRAS series PCI-2 AE system was used to monitor the AE characteristics of the uniaxial compression process of the composite samples. The R3α type transducer was selected, and the resonant frequency was set to 20–100 kHz; the main amplification was set to 40 dB; the threshold value was set to 45 dB; the floating threshold was set to 6 dB; and the sampling frequency was set to 106 times/s. The AE sensor was fixed to the surface of the sample using adhesive tape. Petroleum jelly was applied between the sensor and the surface to reduce the acoustic impedance difference between the contact surfaces and to reduce the energy reflection loss of this interface to ensure that the sensor could easily receive the AE signal. Before the experiment, the AE sensor was tested with broken lead coupling to ensure that the sensor amplitude signal was above 90 dB [30].

3. Results and Discussion

3.1. Strength Characterization

Table 1. Uniaxial compression experiment results of sandstone, CGFB, and sandstone–CGFB composite samples.

Category	Number	Uniaxial Compressive Strength, MPa	Elasticity Modulus, MPa	Peak Strain	Density, g/cm3
Pure sandstone samples	A1	49.835	3060.25	0.0203	2.615
	A2	64.372	3354.67	0.0262	2.611
	A3	57.016	3133.25	0.0300	2.620
	Average value	57.074	3182.72	0.0255	2.615
Pure CGFB samples	B1	2.271	605.65	0.0041	1.896
	B2	2.310	611.02	0.0053	1.923
	B3	2.639	631.15	0.0054	1.957
	Average value	2.407	615.94	0.0049	1.925
Sandstone–CGFB composite samples	C1	4.433	845.30	0.0081	2.254
	C2	4.718	1127.04	0.0063	2.210
	C3	4.067	885.19	0.0062	2.377
	Average value	4.406	952.51	0.0069	2.280

presents the uniaxial compression experiment results and the densities of the samples. Figure 5 shows the uniaxial compression stress–strain curves, and Figure 6 shows the comparative histograms of the uniaxial compressive strength, elastic modulus, and peak strain.

Figure 5 shows that the uniaxial compressive stress–strain curves of the sandstone-CGFB composite samples were closer to those of the pure CGFB samples, both of which went through an initial compression-density stage, a linear elastic stage, a plastic-yielding stage, and a postpeak damage stage. The plastic deformation phase of the stress–strain curve of the pure sandstone samples was not obvious; the phase was brittle after the peak, and the corresponding postpeak stress–strain curve steeply decreased in a straight line. These results showed that the stress–strain characteristics of the sandstone–CGFB composite samples were mainly determined by their filling part; i.e., the filling strength determined the overall strength of the composite samples, which was verified in the subsequent deformation field evolution analysis of the composite samples. In addition, the deformation damage of the composite samples mainly occurred in the filling body with no significant damage occurring in the sandstone.

As shown in Figure 6 and Table 1, the uniaxial compressive strength, elastic modulus, and peak strain of the sandstone–CGFB composite samples were greater than those of the pure CGFB samples and less than those of the pure sandstone samples, while Poisson’s ratio of the sandstone–CGFB composite samples was greater than that of the pure sandstone samples and less than that of the pure CGFB samples. The average uniaxial compressive strength of the sandstone–CGFB composite samples was 4.406 MPa, which was 83.09% higher than that of the pure CGFB samples and 92.28% lower than that of the pure sandstone samples. The average elastic modulus of the sandstone–CGFB composite samples was 952.51 MPa, which was 54.64% higher than that of the pure CGFB samples and 70.07% lower than that of the pure sandstone samples. The peak strain of the composite samples was 0.0069, which was 40.82% higher than that of the pure CGFB samples and 72.94% lower than that of the pure sandstone samples. Compared with the mechanical properties of the pure sandstone and pure CGFB samples, the strength, elastic modulus, and peak strain of the composite samples were more influenced by the CGFB.

The mechanical properties of the sandstone–CGFB composite samples were mainly determined by their CGFB parts. The filling configuration depended on factors such as water secretion, dry shrinkage, and temperature changes, which caused nonuniform deformation between the gangue aggregate and the cement paste, which, in turn, formed internal defects, such as initial bond cracks [31,32,33,34]. The pure CGFB samples were larger than the CGFB part in the composite samples and, thus, had a higher probability of initial defects, therefore reducing the material strength. However, the strength of the sandstone–CGFB composite samples was greater than that of the pure CGFB samples because of the interaction between the sandstone and CGFB. On the one hand, axial stress acted indirectly on the CGFB body through the sandstone, and the compression deformation of the sandstone ensured that the sandstone–CGFB composite samples adapted to the increase in the axial stress. This reduced the damage from axial stress on the CGFB body and limited the development and evolution of internal cracks and other defects, thus increasing the overall strength of the composite samples. The rupture of the composite samples induced the rebound deformation of the sandstone, which aggravated the further rupture and destruction of the CGFB and reduced its strength, thus reducing the overall strength of the assemblage sample. However, this strength-weakening effect was limited, mainly because the sandstone had been compressively deformed before the rebound deformation in order to accommodate the increase in axial stress. On the other hand, the bonding force at the intersection of the sandstone–CGFB composite samples and the friction force between the sandstone and the CGFB restrained the lateral expansion of the CGFB samples. Under the joint action of the bonding and friction forces at the sandstone–CGFB interface, the lateral expansion of the CGFB samples decreased, the bearing capacity increased, and the overall compressive strength of the composite samples increased. Therefore, due to the interaction between the sandstone and the CGFB, the overall strength of the sandstone–CGFB composite samples was greater than the strength of their CGFB.

3.2. Analyses of Deformation Damage Characteristics

3.2.1. Deformation Field Evolution Analysis

Three sets of uniaxial compression repetition experiments were conducted on the sandstone–CGFB composite samples, and the experiment results were basically consistent. C1 was selected for deformation damage characterization. Figure 7 shows the selection of the characteristic points of the uniaxial compression stress–strain curve for the sandstone–CGFB composite samples. Six characteristic points (a–f), corresponding to prepeak stress (0%, 50%, and 90%), peak stress, and postpeak stress (90% and 95%), respectively, were selected in the curve to analyze the deformation and damage characteristics of the sandstone–CGFB composite samples. The digital scattering strain measurement system recorded pictures of the strain field of sample C1 during the test. Figure 8 shows the evolution characteristics of the maximum principal strain field of the sandstone–CGFB composite samples corresponding to the six characteristic points.

As can be seen in Figure 7 and Figure 8, point b was in the stress–strain curve’s linear elastic stage, the maximum strain field distribution of the sandstone–CGFB composite samples was relatively uniform, the initial damage area had a small deformation concentration phenomenon, the amount of deformation was small, and the sample as a whole did not appear to be in the deformation localization zone. Point c was in the plastic-yielding stage of the stress–strain curve and was close to the peak point when the maximum principal strain field showed obvious nonuniformity characteristics and when a crack, crack ①, was generated at the bottom of the CGFB, accompanied by deformation localization. The maximum principal strain at this point was 0.0315 and was generated by the initial damage point of the CGFB in the initial damage area. Point d was at the peak point of the stress–strain curve, and the localized zone of the sandstone–CGFB composite samples continued to develop and expand. The maximum principal strain was 0.0348, the localized zone near crack ① expanded to the intersection, and a new crack, crack ②, was generated near crack ①. Points e and f were in the plastic-softening phase of the stress–strain curve. The deformation localization zone in the e-point crack ① area extended to the intersection. A new crack, crack ③, formed at the bottom of the assemblage, and a slight localization phenomenon occurred in the crack ③ area at which the maximum principal strain was 0.0794, which was 128.17% higher than that at the peak point. The localized zone of deformation in the sample of the f-point composite further developed and expanded, and the maximum strain was 0.0817. The localized zone of deformation near crack ① in the sample expanded laterally, and the localized-zone area increased. The localized zone near crack ③ continued to develop and penetrated the localized zone near crack ②. The elongation, intersection, and connection of the deformation localization zone of the sandstone–CGFB composite samples at points e and f were accompanied by the continued destruction of the initial damage and the initiation, expansion, and penetration of the new cracks, leading to the destabilization of the composite samples as shown in the enlarged area in Figure 7.

In summary, the evolution of the deformation localization zone of the sandstone–CGFB composite samples was related to the destruction of the initial damage of the CGFB and the initiation and expansion of new cracks. The intersection and penetration of the deformation localization zone within the CGFB samples led to the destruction of the composite samples. Because the sandstone samples were much stronger than the CGFB samples, the sandstone was not significantly damaged during the experiment, and the maximum strain field was distributed more uniformly and without localization. Therefore, sandstone could be considered an elastomer during the experiment. The CGFB materials were characterized by wider strips and lamellar connections in the localized-zone area, which were analyzed to be related to the lower strength of the CGFB materials, the loose internal structure, and the exfoliative ductile damage.

3.2.2. Analysis of the Displacement Evolution of the Deformation Localization Zone

For the calculation of the strain field of the sandstone–CGFB composite samples, sample continuity was assumed. However, the insides of the samples were discontinuous after the end of localization; therefore, other indicators were needed to further analyze the intrinsic relationship between the evolution of deformation localization and the bearing capacity of the samples. Localized-zone displacement evolution analysis was used to quantitatively investigate the displacement evolution characteristics of the deformed localized zone. The localization zone of the deformation of the sandstone–CGFB composite samples before the final damage was identified [35] (Figure 9a). Displacement misalignment analysis [36], as shown in Figure 9b, was used to calculate the displacement misalignment of the deformation localization zone of the sandstone–CGFB composite samples on both sides of the deformation localization zone, where a was the distance (2 mm) on both sides of the deformation-localization-zone identification line, M1 and M2 were the center points of the selected pixel points, and u and v were the displacement components of the selected pixel points. Figure 10 shows the displacement misalignment analysis results which stipulated that the displacement misalignment along the counterclockwise direction was positive.

Figure 10 shows that the evolution of the displacement misalignment in the deformation localization zone of the sandstone–CGFB composite samples could be divided into three stages: microvariation, linear growth, and nonlinear variation. The displacement misalignment evolution of the deformation localization zone of the sandstone–CGFB composite samples was influenced by the generation, development, expansion, and penetration of the deformation localization zone.

In the microvariation stage, the overall deformation of the sandstone–CGFB composite samples was uniform, and the displacement misalignment of deformation localization zones A–D was almost zero.

In the linear growth phase, the deformation localization zone of the sandstone–CGFB composite samples expanded to the intersection development as a whole, and the displacement misalignment of deformation localization zones A–D grew linearly. Influenced by the formation time and the development expansion of the deformation localization zone, the displacement misalignment of deformation localization zones B and C first entered the linear growth stage, and the growth rate was similar and relatively large. The displacement misalignment of deformation localization zone B moved counterclockwise, while the displacement misalignment of deformation localization zone C moved clockwise.

During the nonlinear growth phase, with the increase in axial strain, the micro-rupture developed further. The misalignment curves of each region showed different degrees of “nonlinearization” due to the redistribution of the stress field inside the composite samples. In the initial damage area of the sandstone–CGFB composite samples, after the pressure-dense and linear-elastic stage, the macroscopic damage did not extend significantly, and the displacement misalignment of deformation localization zone A tended to be smooth. Due to the upward expansion and development of crack ① under axial stress and the shear sliding between microcracks causing significant increases in the displacement misalignment (Figure 8), the displacement misalignment of deformation localization zone B increased rapidly. Due to the bulging of the CGFB between localized zones B and C, the displacement misalignment of the localized zone C changed from clockwise to counterclockwise, and the growth rate of the misalignment reduced. At 304 s, due to the region of deformation localization zone C producing more serious deformation damage, the monitoring points in the region could not be identified, and the localization-zone displacement misalignment curve ended here. Because the cracks in the area of deformation localization zone D continued to develop and evolve under axial pressure, the displacement misalignment of deformation localization zone D had a nonlinear growth trend, and the misalignment was between A and B.

3.2.3. Evolution of Displacement Variation Amount of Sandstone and CGFB near the Intersection

To monitor the horizontal displacement of the sandstone and CGFB samples at the intersection, four monitoring points were set up at the upper and lower sides of the sandstone-CGFB intersection. The deformation and failure characteristics of the composite samples were analyzed according to the monitoring data. Monitoring point R1 was 3 mm from the upper side of the intersection and 23 mm from the left side of the sample centerline; F1 was 3 mm from the lower side of the intersection and 23 mm from the left side of the sample centerline. R2 and F2 were on the sample centerline and were 3 mm from the upper and lower sides of the intersection, respectively, as shown in Figure 11.

The interface effect [33] influenced the synergistic deformation of the sandstone and CGFB samples at the intersection of the composite samples. Because the modulus of elasticity of the sandstone was larger than that of the CGFB and because Poisson’s ratio of the sandstone was smaller than that of the CGFB, the deformation of the sandstone and CGFB samples at the intersection was not uniform. Therefore, to maintain the overall stability of the composite samples, the derived stresses in the sandstone region at the intersection promoted their deformation, and the derived stresses in the CGFB sample region limited their deformation.

Figure 12 shows the evolution curves of the horizontal variation amounts of monitoring points R1, R2, F1, and F2 at the intersection of the sandstone–CGFB composite samples, and the displacement to the right was specified as positive. Figure 13 shows the monitoring points near the sandstone–CGFB intersection against the macroscopic fractures.

At the beginning of loading, the displacement of each measurement point grew negatively, and, after loading for up to 40 s, the displacement started to grow positively. The displacement to the right of the measurement points was positive, and that to the left was negative. Therefore, it can be seen in Figure 12 that the displacement of each measurement point was leftward first, and the maximum displacement amounts of measurement points F1, F2, R1, and R2 to the left were 0.12 mm, 0.15 mm, 0.16 mm, and 0.17 mm, respectively. Subsequently, each measurement point gradually moved to the right and showed positive growth in the displacement as seen in Figure 12. The randomness of the sample deformation at the beginning of loading led to the gradual movement of each measurement point to the right after it had moved to the left, and this phenomenon was numerically expressed as the displacement decreased first and then increased. The measurement points on the CGFB side peak increased before the measurement points on the sandstone side. From 40 to 180 s of sample loading, the measurement points on both sides of the sample intersection grew at the same rate. The displacement of the measurement points on the CGFB side was larger than that on the sandstone side at the same time, and the displacement of F1 on the side was larger than that of F2 in the middle. During loading from 180 to 200 s, the growth rate of the measurement points on both sides of the intersection peaks tended to be the same. The displacement of R1 and R2 on the sandstone side tended to be the same, and the displacement of F1 and F2 on the CGFB side tended to be the same. The displacement of R1, R2, and F2 showed a slow rise and then a slow decline from 200 to 443 s. The displacement of F2 was significantly larger than that of R1 and R2; the displacement of F1 increased rapidly after a short fluctuation. No significant internal damage occurred in the sandstone or the CGFB in the sandstone–CGFB composite samples from 40 to 180 s. Therefore, both sides of the intersection maintained synergistic deformation. From 180 to 200 s, the displacement on both sides of the intersection grew rapidly due to the accelerated misalignment of deformation localization zone B. Subsequently, under the influence of Poisson’s effect, the displacement of F2 on the CGFB side was larger than that on the sandstone side and, finally, tended to be stable, and the displacement of F1 continuously increased until the sample was destabilized and destroyed due to the expansion of the deformation localization zone, penetration, misalignment, and macroscopic fracture ② development (Figure 13). The CGFB was destabilized due to the development of internal fractures, which induced the rebound deformation of the sandstone, and the final displacement of R1 and R2 produced a small decrease.

In summary, when the sandstone and the CGFB did not undergo macroscopic damage, the sandstone and the CGFB on both sides of the intersection deformed synergistically. In contrast, when the CGFB underwent macroscopic damage first, the sandstone and the CGFB on both sides of the intersection transformed into a nonsynergistically deformed state at which time the interface effect promoted the deformation of the sandstone and inhibited the deformation of the CGFB.

To further study the interface effect on the deformation of the sandstone–CGFB composite samples, monitoring points R3–R6 and F3–F6 were set at 1 mm from the right side of the sandstone–CGFB composite samples at a spacing of 4 mm as shown in Figure 14.

Figure 15a shows the horizontal displacement evolution curves of measurement points R3–R6. During the period of sample loading from 0 to 40 s, measurement points R3–R6 moved to the left and gradually reached peak values of −0.132, −0.172, −0.189, and −0.205 mm, respectively. The peak value of the measurement points gradually increased with increasing distance from the intersection. From 40 to 180 s of sample loading, R3–R6 moved to the right, and the displacement grew at a uniform rate. The growth rate of the measurement point displacement decreased gradually with increasing distance from the intersection. The growth rates of R3–R6 peaked from 180 to 200 s of sample loading. The growth rates of R4–R6 gradually decreased from 200 to 443 s, and the displacement reached its peak at 300 s. After that, the displacement of measurement points R4–R6 slowly decreased because the instability rupture of the assemblage induced the rebound deformation of the sandstone. The displacement of R3 increased during this period because the interface of the assemblage inhibited the horizontal rebound deformation of the sandstone. The horizontal displacement evolution trend of monitoring points R3–R6 on the sandstone side was similar throughout the loading period, and the displacement of the measurement points showed an increasing trend from top to bottom. Compared with R6, the increases in R5, R4, and R3 were 45.8%, 90.8%, and 231%, respectively. The results suggested that the interface effect most obviously promoted the deformation of the sandstone near the boundary in a horizontal direction.

Figure 15b shows the displacement evolution curves of measurement points F3–F6 in the horizontal direction. From 0 to 30 s of sample loading, measurement points F3–F6 gradually moved to the left with peaks at −0.132, −0.125, −0.116, and −0.101 mm, respectively, and they gradually decreased with increasing distance from the intersection. From 30 to 180 s, F3–F6 moved to the right, and the displacement grew at a uniform rate. The growth rate of measurement point displacement increased gradually with increasing distance from the intersection. The displacement growth rate of F3–F6 peaked from 180 to 200 s of sample loading. The displacement of measurement point F3 stabilized at 0.52 mm from 200 to 443 s. The displacement of F4–F6 increased continuously, reaching 0.663, 0.914, and 1.063 mm at the time of sample destabilization. The results revealed that the interface effect had the strongest inhibitory effect on the horizontal deformation of the filling near the boundary; therefore, F3 maintained a high stability under the interface effect.

3.2.4. Analysis of Rebound Deformation Characteristics of the Sandstone and the CGFB

In this experiment, the strength of the CGFB samples was much lower than that of the sandstone samples, and, when the sandstone–CGFB composite samples reached their strength limit, the CGFB samples ruptured. At this point, the sandstone samples were still in an elastic state, and the destruction of the CGFB samples induced the rebound deformation of the sandstone samples. Therefore, the CGFB samples were the rupture bodies, and the sandstone samples were the rebound bodies [27]. To further reveal the interaction mechanism between the sandstone and CGFB samples, monitoring points were arranged at 1 mm from the boundary of the sandstone and CGFB samples to monitor the changes in the heights (H_r and H_f) and widths (W_r and W_f) of the sandstone and CGFB samples as shown in Figure 16. Figure 17 shows the variation curves of axial stress, H_r, and H_f versus time, and Figure 18 shows the variation curves of axial stress, W_r, and W_f versus time.

As can be seen from Figure 17, during the failure process of the sandstone–CGFB composite samples, H_r showed a fluctuating decrease followed by a fluctuating increase, and H_f showed a nonlinear decreasing trend. The changes in the H_r and H_f of the sandstone–CGFB composite samples were mainly influenced by the degree of initial failure inside the sandstone and the CGFB samples as well as the expansion evolution of the new cracks.

Before the peak point of the loading curve, H_r showed a fluctuating decreasing trend following a short fluctuation, and H_f showed a nonlinear decreasing trend following a gentle fluctuation. The results showed that no macroscopic damage had occurred in the sandstone during the experiment, which could be considered an elastomer. The sink shrinkage caused by the secretion of the filling material, the chemical shrinkage in the hardening process, the physical shrinkage, and the pores formed during the casting process all caused initial damage to the internal structure of the filling body at different levels, and the uneven distribution and continued evolution of the damage inside the filling body created the nonlinear characteristics in front of the H_f peak.

The characteristics of the H_r and H_f evolution in the postpeak phase corresponded to the characteristics of the stress–time curve of the sandstone–CGFB composite samples. Before loading to 199.5 s, the stress–time curve of the sandstone–CGFB composite samples showed a slow decreasing trend, and the CGFB still had a strong load-bearing capacity providing support for the sandstone. The H_r value fluctuated around 49.6995 mm with no obvious rising or falling trend. At 199.5 s, when the axial stress was 93% of the peak stress, the stress–time curve of the sandstone–CGFB composite samples fell faster, while H_r changed from a “hold” to a fluctuating rise with a rise of 0.0089 mm. H_f was still on a downward trend. However, the H_f–time curve reached an inflection point, and the rate of the H_f decline accelerated.

As per the above results, the destruction of the CGFB samples in the composite samples induced a rebound deformation of the sandstone. As the CGFB body showed obvious plastic damage characteristics, the CGFB body still had a strong supporting effect in the postpeak stage. As the sandstone was still in the elastic stage without macroscopic damage at this time, H_r fluctuated upward from the beginning of the rebound deformation phenomenon to the end of monitoring. After the rebound deformation of the sandstone, part of the elastic energy stored in the sandstone overcame its own damping to do work and to be consumed; the other part worked on the filling body, intensifying its rupture and movement and accelerating the rate of the H_f decrease. Due to the exfoliative damage of the CGFB body, the height and energy release of the CGFB body changed during the destruction process, which further affected the sandstone height. H_r fluctuation significantly increased in the rebound deformation stage of the sandstone compared with that before the peak of the loading curve. With increased CGFB damage, crack ③ penetrated crack ① to form a spalling damage area at the bottom of the CGFB (Figure 19). The monitoring points in this area were deformed or spalled when loaded to 304.5 s, and the H_f data could not be monitored further.

W_r showed a trend of fluctuating upward and then downward, and W_f showed a nonlinear increasing trend (Figure 18). Before loading to the peak of the stress–time curve, with respect to the axial stress and Poisson’s effect, W_r fluctuated slightly upwards (0.0021 mm). W_f showed a nonlinear, accelerated rising trend, increasing by 1117 mm. Since Poisson’s ratio of the CGFB was larger than that of the sandstone, the lateral deformation of the CGFB was larger, and W_f was larger than W_r. Before loading up to 244.5 s, W_r fluctuated downwards, indicating that the CGFB damage induced the rebound deformation in the horizontal direction of the sandstone, and the rebound deformation was 0.0055 mm. When loading up to 378 s, due to the continued deformation of the initial damage area of crack ② and the continued expansion of crack ③ (Figure 19), the area where the monitoring point was located was deformed or underwent spalling damage, and the W_f data could not be monitored further.

3.3. Evolutionary Analysis of Deformation Energy Characteristics

Assuming that there is no heat exchange between the sandstone–CGFB composite samples and the outside world when the external force is working, the prepeak phase has the following relationship according to the first law of thermodynamics [37]:

$$U=U_e+U_d$$

(1)

where $U$ is the total work performed by the external force; $U_{\mathrm{d}}$ is the unit dissipation energy, which is used to inflict internal damage and which helps in the plastic deformation of the assembly; and $U_{\mathrm{e}}$ is the elastic strain energy that can be released by the unit. The formula for calculating $U_{\mathrm{e}}$ is [37]

$$U_{\mathrm{e}}=\frac{1}{2E_{\mathrm{u}}}\left[{\sigma }_1^2+{\sigma }_2^2+{\sigma }_3^2-2\mu \left({\sigma }_1{\sigma }_2+{\sigma }_2{\sigma }_3+{\sigma }_1{\sigma }_3\right)\right]$$

(2)

where $E_{\mathrm{u}}$ is the unloading elastic modulus of the sample and where $\mu$ is Poisson’s ratio of the sample. Using the initial modulus of elasticity $E_0$ instead of the unloaded modulus of elasticity, the strain energy per unit volume of each part in the uniaxial compression experiment can be expressed as [38]

$$U={\displaystyle {\int }_0^{{\epsilon }_1}{\sigma }_1}d{\epsilon }_1$$

(3)

$$U_e=\frac{1}{2E_0}{\sigma }_1^2$$

(4)

$$U_{\mathrm{d}}={\displaystyle {\int }_0^{{\epsilon }_1}{\sigma }_1}d{\epsilon }_1-\frac{1}{2E_0}{\sigma }_1^2$$

(5)

where ${\epsilon }_1$ is the axial strain of the sample and where ${\sigma }_1$ is the axial of the sample.

According to the energy calculation formulas of each part above, the change in the value of the energy density before the peak of the sandstone–CGFB composite samples can be calculated during the loading process. Figure 20 shows the characteristic curves of the energy density evolution before the peak of the sandstone–CGFB composite samples. Figure 21 shows the energy distribution before the peak of the sandstone–CGFB composite samples.

Figure 21 also reveals the correspondence between the energy evolution before the peak and the stages of the stress–strain curve.

Initial compacting stage: The input energy density, the elastic strain energy density, and the dissipated energy all grew slowly with the increase in the strain. The external input energy was mainly converted into elastic energy, with dissipated energy accounting for a relatively small amount. With the increase in the axial stress, the growth of the dissipated energy was accelerated. At this time, the dissipated energy was mainly used for compacting internal pores, compacting the interface between the sandstone and CGFB body, closing the initial small cracks, and overcoming the frictional slippage between contact surfaces or aggregates. At this time, the elastic energy was mainly stored inside the composite samples.

Linear elastic deformation stage: The external energy input was mainly converted into elastic energy. In the early stages of the linear elastic deformation, the input energy density curves of C1, C2, and C3 of the sandstone–CGFB composite samples and the elastic energy density curves were nearly coincident, while the growth of the dissipated energy was almost stagnant. With the increase in the strain, the growth of the dissipated energy density accelerated, and the growth of the elastic energy density slowed at the end of the linear elastic deformation stage. Although this stage was dominated by the elastic strain, due to the internal defects of the CGFB and sandstone, the stress continuously expanded the internal microcracks to produce new cracks, and this process dissipated a small part of the input energy. However, most of the input energy was stored in the form of elastic energy inside the composite samples.

Yield damage stage: When stress increased beyond the elastic limit of the composite samples, the composite entered the yielding stage. The input energy density, the elastic energy density, and the dissipated energy density continued to grow with increasing strain, but the growth rate of the elastic energy density gradually slowed, and the proportion kept decreasing. The growth rate of the input energy density was slightly higher than that of the elastic phase. The dissipated energy density grew rapidly, and the growth rate and the proportion of the dissipated energy increased. When the stress peaked, the dissipation energy density values of C1, C2, and C3 exceeded the elastic strain energy density. The energy dissipation in the yielding stage was mainly caused by plastic deformation, the generation of new cracks, and the expansion of old cracks. The closer the area was to the damage, the more the energy dissipated. As the internal damage increased, the limit of the energy storage lowered, and the elastic energy stored inside the sample was released, causing overall damage.

As can be seen in Figure 21, most of the input energy in the prepeak stage was converted into the elastic strain energy stored in the sandstone–CGFB composite samples, and the percentages of the prepeak elastic energy in composite samples C1, C2, and C3 were 65.5%, 62.7%, and 58.4%, respectively. The rest of the energy was converted into dissipated energy, and the percentages of the dissipated energy in front of the C1, C2, and C3 peaks were 34.5%, 37.3%, and 41.6%, respectively. This phase could be considered an accumulation of energy. In the postpeak stage, after the sample was destroyed, elastic energy was released and converted into dissipated energy together with the input energy of the testing machine to make the internal cracks penetrate and to form a macroscopic rupture surface, therefore causing further damage to the CGFB of the composite samples. The destruction of the sandstone–CGFB composite samples reduced their ability to store energy, and the elastic energy stored before the peak was not completely released but was still partly retained, stored in the residual strength stage. The residual CGFB body skeleton still played a certain support role for the external loads, so the composite samples after the peak showed more obvious plastic damage characteristics. In summary, the destruction process of the sandstone–CGFB assemblage samples could be seen as the result of the energy drive.

3.4. AE Characteristics Analysis

The AE energy is the area under the AE event signal detector envelope [39] that reflects the relative energy and intensity, and it can be used to identify the wave source-type and monitor continuous-type AE signals. The AE energy reflects the strength of the event signal activity and can characterize the scale of the rupture occurring within the samples. The R3α-type sensor was selected, and the resonant frequency of the sensor is 20–100 kHz. The AE energy during the progressive failure of the sandstone–CGFB composite samples was analyzed by comparing the stress–time curves. Figure 22 shows the uniaxial compressive stress and the AE energy plot for each sample.

Figure 22 shows that, with increasing axial stress, the AE energy signals of the sandstone–CGFB composite samples had obvious phase characteristics. According to the number and peak magnitude of the AE energy signals, the progressive failure process of the composite samples under uniaxial compression was divided into four stages: the initial active stage (stage I), calm stage (stage II), secondary active stage (stage III), and residual stage (stage IV).

The initial active stage included the stress–time curve’s compressive dense stage, the linear elastic stage, and the early part of the plastic-yielding stage. During this stage, the sandstone–CGFB samples’ contact intersection was compressed and dense, the failure inside the samples was mainly in the form of small-scale initial damage, and new fissures were formed and expanded. More elastic energy was released while the fissures evolved, which led to an active AE energy rate intensity and a higher frequency, and the peak fluctuation was large. The maximum values of the C1 and C3 AE energy rates of the sandstone–CGFB composite samples all appeared in this stage. The energy ratio share of this stage, calculated from the accumulated-energy ratio curve of each sample in Figure 22, was 20%–40%.

The calm stage was in the late plastic-yielding phase of the stress–time curve. During this stage, the AE signal was stable, and the AE energy rate intensity was low. The samples entered a period of unstable crack expansion, and microcracks formed inside the samples and expanded steadily. This stage was shorter than the initial active and secondary active stages. This stage could be used to gather the precursor information of the destabilization of the composite samples under the load. The energy ratio share of this stage was 1%–4%.

The secondary active stage was in the stress–time curve peak and the early-stage postpeak damage phase of the sandstone–CGFB composite samples. The AE signal was enhanced, and the AE energy rate fluctuated significantly and had a large peak. Microcracks propagated unstably and formed new macroscopic cracks in the composite samples. The propagation of macroscopic cracks led to the overall instability of the composite samples, though the composite samples still had some load-bearing capacity after the peak of the stress–strain curve, and a macroscopic rupture with friction, and a dislocation slip released more elastic energy. The intensity and frequency of the AE energy rate at this stage for samples C1 and C3 were lower than they were in the initial active stage, mainly due to the intersection and penetration of a large number of macroscopic cracks, and the AE signal was severely decayed. The energy ratio share of this stage was 55%–70%.

The residual stage corresponded to the late stage after the peak of the stress–time curve. The AE energy rate was lower and more stable without obvious fluctuations. This phenomenon indicated that the three-dimensional spatial and temporal evolution of the cracks in the sandstone–CGFB composite samples under uniaxial compression during the residual stage had basically ended, that the internal structure of the samples had been completely damaged, and that their mechanical properties had completely changed. The energy ratio share of this stage was 2%–6%.

4. Conclusions

This investigation aimed to study the mechanical properties, deformation failure characteristics, energy evolution law, and AE characteristics of sandstone–CGFB composite samples under uniaxial loading. To make the experimental results better for use as a practical engineering reference, we obtained our test materials from the engineering site. The sandstone was taken from the Liangjia Coal Mine in Yantai, Shandong Province, China. The fly ash was from the Qingdao Huangdao Power Plant. The coal gangue was from the Liangjia Coal Mine. The test results are useful for the analysis of the performance of surrounding roof–CGFB combinations in the case of immediate roof collapses in strip-fill mining. The main findings of this study were as follows:

(1)

The stress–strain curve of the sandstone–CGFB composite samples was closer to that of the pure CGFB samples, and the average uniaxial compressive strength of the sandstone–CGFB composite samples was 83.09% higher than that of the pure CGFB samples and 92.28% lower than that of the pure sandstone samples. The average elastic modulus of the sandstone–CGFB composite samples was 54.64% higher than that of the pure CGFB samples and 70.07% lower than that of the pure sandstone samples. In the composite samples, all the damage occurred in the CGFB part. This was mainly spalling damage, and obvious plastic damage characteristics were shown.

(2)

The failure of the sandstone–CGFB composite samples occurred mainly in the CGFB part. The deformation localization zone first appeared in the nearby area of the initial damage. The evolution of the displacement misalignment on both sides of the deformation localization zone of the sandstone–CGFB composite samples was affected by the formation, expansion, and penetration of the deformation localization zone. The amount of displacement misalignment in the deformation localization zone in the stage of nonlinear change could be used as an important indicator for the prediction of the deformation failure of the composite samples. The failure of the CGFB samples induced a sandstone springback deformation with a springback capacity of 0.0089 mm in the vertical direction and 0.0055 mm in the horizontal direction, which led to the further rupture and failure of the CGFB.

(3)

The interface effect influenced the synergistic deformation of the sandstone and CGFB samples at the intersection of the sandstone–CGFB composite samples. When the sandstone and the CGFB were not macro-damaged, the sandstone and the CGFB on both sides of the intersection were deformed synergistically. In contrast, when the CGFB was macro-damaged, the sandstone and the CGFB on both sides of the intersection were transformed into a nonsynergistically deformed state at which time the interface effect promoted the deformation of the sandstone and inhibited the deformation of the CGFB. The interface effect had the most obvious influence on the horizontal deformation of the sandstone and fill monitoring points near the intersection.

(4)

The input energy was mainly transformed into elastic strain energy in the prepeak pressure-density stage and the linear elastic stage, and a small part was transformed into dissipated energy. The ability of the input energy to transform into dissipated energy increased in the yielding stage of the plastic zone, and the trend of increasing elastic strain energy slowed. The prepeak elastic energy share of the samples was 58.4%–65.5%, and the prepeak dissipation energy share of the samples was 34.5%–41.6%. The input energy and the stored elastic strain energy were jointly transformed into dissipated energy in the postpeak damage stage. This led to the rapid penetration of internal cracks and the formation of a macroscopic rupture surface, therefore damaging the CGFB part of the composite samples and leading to the destabilization and failure of the composite samples.

(5)

The AE energy signals of the sandstone–CGFB composite samples showed obvious time-dependent characteristics. According to the number and peak size of the AE energy signals, the progressive damage process of the sandstone–CGFB composite samples under uniaxial compression was divided into the initial active, calm, second active, and residual stages. The energy share of each stage was 20%–40%, 1%–4%, 55%–70%, and 2%–6%, respectively. The change in the AE signal energy rate corresponded to the stress–time curve of the integrated body samples, and the change in the AE signal frequency and amplitude could be used as precursor information related to the destabilization of the combined body samples under a load.