1) The moisture content of the waste rock used for filling is 3.63%, and the gradation is measured by manual screening, as shown in Figure 2. The particle size of the waste rock ranges from 0.1 to 10 mm, and the range of –2 mm is relatively high, nearly 70%.

It was determined in the field that the filling slurry concentration realizes gravity transportation at about 74%, when the slurry diffusivity was more than 80 cm, with good workability and no segregation and precipitation (Xiao et al., 2019; Huang et al., 2021). Therefore, this value was taken as the standard of filling slurry concentration. According to the filling materials in the mine, experiments of fly ash on the effect of the backfill strength and the optimization of the backfill ratio were carried out. The experiment scheme of fly ash on the effect of the backfill strength is shown in Table 2, and the optimization experiments of the backfill ratio are carried out based on the results of the experiments in Table 2.

The materials of each experiment scheme placed in the mixing bucket were weighted. The mixer was used to mix the slurry evenly. Then, the slurry was loaded into the mold of φ80 × 200 mm. Because of the low initial strength of the backfill, it was easy to be damaged after demoulding. Therefore, it was demoulded after 24 h and moved to the constant temperature and humidity box for maintenance, with the temperature set at 20 ± 1°C and humidity at 95%. After curing to the predetermined age, the uniaxial compressive strength was measured by WES-100 hydraulic universal testing machine, and representative backfill blocks were selected and made into standard specimens. The microstructure characteristics were observed by scanning electron microscope after gold spraying (Hu et al., 2019; Xie, et al., 2020) to analyze the variation mechanism of backfill strength. The flow chart of experiments is shown in Figure 7.

To analyze the effect of fly ash on the strength of backfill, group A and group B experiments were conducted, respectively. The results of 28, 56, and 90 d uniaxial compressive strength of backfill are shown in Figure 8.

When the cement content was 9%, the 28 d uniaxial compressive strength of the backfill without fly ash was 0.57 MPa and the 56 d uniaxial compressive strength was 0.97 MPa. As can be seen from Figure 8, the strength of the backfill in both groups A and B is significantly improved under the condition of adding fly ash. The 28 d uniaxial compressive strength is greater than 1.65 MPa. The 56 d uniaxial compressive strength is greater than 2.24 MPa. That is, the addition of fly ash in the slurry has an overall promoting effect on the strength of the backfill.

As shown in Figure 8A, when the filling aggregate was all waste rock, the strength of the backfill gradually increased with the increase of fly ash to cement mass ratio, and both of them increased approximately linearly. The 56 d strength was 2.24 times the 28 d strength (mean value), while the strength grew less from 56 to 90 d during the maintenance period. The strength development of backfill containing fly ash was mainly concentrated in the maintenance period from 28 to 56 d, after which the growth trend slowed down. As shown in Figure 8B, when the filling aggregate was waste rock and tailings with the ratio of 3:1, after the fly ash-cement mass ratio exceeded 3 (the fly ash-tailings mass ratio was about 3:2 at this time), the strength of the backfill suddenly decreased, and fly ash had a great negative effect on the strength at this time, which is different from the study of Dong et al., 2018. Generally, the increase of fly ash content promotes the strength of backfill in the middle and late stages, and the main reason for fly ash inhibiting the strength of backfill in this experiment is as follows: as the tailing particles were fine, the continuous increase of fly ash content led to more fine particles in the backfill, which hindered the hydration of cementitious materials and affected the growth of the backfill strength. Therefore, fly ash has a two-way effect on the late strength of the backfill under the condition of tailings. It is necessary to determine the appropriate filling ratio of tailings and fly ash.

1) From the results of group A, it can be seen that under the condition that the filling aggregate is all waste rock, the strength of the backfill shows an increasing trend with the increase of fly ash content. However, when the amount of filling aggregate is too much, the bonding quality of the interface between slurry and aggregate will be reduced, weakening the integrity of the backfill. Therefore, the group C experiment was conducted to determine the appropriate ratio of filling aggregate to fly ash: the filling slurry concentration was 74%, the cement content was 6.5%, the filling aggregate was all waste rock, and the mass ratios of waste rock-fly ash were 1:1, 2:1, 3:1, and 4:1, respectively. In addition, due to the poor activity of the fly ash, 3% content CaO was added to the slurry

As shown in Figure 9, the 28 d uniaxial compressive strength of the backfill gradually decreased with the increase of waste rock content, while the 56 d uniaxial compressive strength showed a trend of increasing and then decreasing. As fly ash mainly affected the strength of the backfill in the middle and late stages, the fly ash was not heavily hydrated at 28 d. The waste rock was not completely wrapped by the gel and the integrity was poor; therefore, the strength at 28 d decreased with the increase of waste rock; while at 56 d, the fly ash was hydrated to produce a large amount of gel and wrapped waste rock, which showed that the strength of the backfill was higher than that at 28 d, and increased first with the increase of waste rock content. When the waste rock-fly ash mass ratio exceeded 2:1, too much waste rock was not wrapped. The 56 d strength of the backfill started to decrease, which was the optimal value of the waste rock to fly ash, that is, the early strength of the backfill met the demand and had no inhibitory effect on the middle and late strength.

2) Group D experiment was carried out based on the waste rock-fly ash mass ratio of 2:1, and tailings were added to the filling aggregate to determine the optimum tailings content. Group D experiment scheme: the cement content was 6.5%, the mass ratio of waste rock-fly ash was 2:1, the mass ratios of waste rock-tailings were 1:1, 2:1, 3:1, and 4:1, respectively, and CaO content was 3%

As can be seen from Figure 10, with the increase of waste rock to tailings mass ratio, i.e., with the decrease of tailings content, the 28 d uniaxial compressive strength of the backfill increased linearly. The 56 d uniaxial compressive strength showed a trend of first increase and then decrease. The minimum strength of the backfill was 1.16 MPa when the mass ratio of waste rock-tailings was 1:1, which was better than that of the backfill without fly ash; the 56 d uniaxial compressive strength reached the maximum when the waste rock-tailings mass ratio was 3:1. At this time, the number of tailings had no inhibiting effect on both 28 d and 56 d uniaxial compressive strength.

3) Based on the results of group C and group D, the optimal mass ratio of waste rock-tailings-fly ash was determined to be 6:2:3. On this basis, group E experiments were conducted to determine the optimal cement content to meet the filling demand of the mine: the mass ratio of waste rock-tailings-fly ash was 6:2:3, CaO content was 3%, and the cement content was 5, 8, 11, and 14%, respectively.

From Figure 11, it can be seen that the uniaxial compressive strength of the backfill at 28 and 56 d increases linearly with the increase of cement content, and the fitting equations are as follows: 28 d : ⋅ y = 0.373 x - 0.26 (   R 2 = 0.989 ) , (1) 56 d : ⋅ y = 0.904 x - 1.19 ( R 2 = 0.994 ) . (2)

where y is the uniaxial compressive strength of the backfill, MPa; x is cement content. The slope of the 56 d fitting equation was significantly greater than that of 28 d, i.e., its strength development was significantly higher than that of 28 d. The increase of cement content had a more obvious effect on the 56 d strength of the backfill. This was mainly since 28–56 d was the main hydration stage of fly ash, and the increase of cement content led to the increase of generated OH⁻, which promoted the hydration of a large amount of fly ash during 28–56 d.

4) When the filling slurry concentration was 74%, the mass ratio of waste rock-tailings-fly ash was 6:2:3, CaO content was 3%, and cement content was 5%; the uniaxial compressive strength of the backfill at 28 and 56 d was 1.70 and 3.11 Mpa, respectively, which met the demand of filling of the mine. Based on the market price of building materials at the mine site, the filling cost was calculated as follows: the price of 32.5 ordinary Portland cement bought from local cement plant was RMB 340 per ton; the waste rock was produced from underground mining, and the transportation and crushing cost was RMB 20 per ton; all tailings were purchased from other mines at a cost of RMB 18 per ton; fly ash was taken from nearby thermal power generation at a price of RMB 70 per ton; CaO was purchased from local enterprises at a price of RMB 240 per ton. After calculation, with the filling slurry concentration of 74%, the mass ratio of waste rock-tailings-fly ash of 6:2:3, CaO content of 3%, and cement content of 5%, the cost of filling slurry was RMB 80.32 per cubic meter. Compared with the current filling ratio under the condition of 74% concentration of filling slurry, 10% cement content, and 3:1 mass ratio of waste rock-tailings, it can save RMB 0.56 per cubic meter. In addition, the strength of the backfill can be increased by about 1 MPa at 28 d and 2 MPa at 56 d. It not only makes a large amount of use of fly ash and reduces the cost of filling but also improves the strength of the backfill significantly, which improves the safety factor and economic benefits of mining.

To analyze the reasons for the strength variation of the backfill, the backfill of the stope, group A3, and group B3 were selected, respectively. Their images were magnified to 2000 times to observe the microstructure by scanning electron microscope. The images of the backfill are shown in Figure 12.

The backfill is a porous structure composed of cementitious material hydrated colloid, Ca(OH)₂ crystals, solid particles, and pores, etc. The relationship between its strength and porosity is as follows: S = S 0 ⁡ exp ( − b p ) , (3) where S is the ideal state strength of the backfill, MPa; S 0 is the strength of the backfill when the porosity is 0, MPa; b is a constant related to the cementitious material and maintenance age; p is the internal porosity of the backfill. It can be seen that the strength decreases exponentially as the porosity of the backfill increases, which can be analyzed visually by the structural characteristics of the backfill pores (Cao and Song, 2018; Nakata et al., 2018; Wang and Qiao, 2019). The reaction process of cement and fly ash hydration in the backfill: C₃S, C₂S, C₃A, and C₄AF in cement react with water to generate Ca(OH)₂ and C-S-H, calcium aluminate, and other gel materials, in which Ca(OH)₂ is the intermediate product to promote the mutual reaction of ions in the solution and then reacts with a large amount of SiO₂ and Al₂O₃ in fly ash to promote the generation of gel materials. The gel material can bind to the aggregate and fill the pore structure after hardening and is the key component affecting the strength of the backfill (Yu et al., 2018).

From Figure 12A, the 28 d scanning electron microscope of the backfill can be seen that the cement was hydrated to generate part of the flocculent C-S-H gel, accompanied by a small amount of fibrous AFt and plate-like Ca(OH)₂ crystals. There were still some inert particles around the gel that have not been wrapped. The overall structure was relatively loose, with a more developed pore structure (Zhang et al., 2021). The strength of the backfill was low at 0.67 MPa, which did not meet the demand of the mine; Figure 12B shows the structure of the backfill at 28 d under the condition of adding fly ash and waste rock. The hydration of cement and fly ash was obvious, generating a large number of flocculent C-S-H gel, and the hydration products were staggered and well wrapped around the aggregate. Compared with Figure 12A, the structural integrity and connectivity were enhanced. The pores were filled with C-S-H gels. The strength was well developed, reaching 2.24 MPa. Figure 12C shows the structure of the 28 d backfill under the condition of adding fly ash, waste rock, and tailings. Due to the increase of fine particles in the slurry caused by the addition of fly ash and tailings, and the poor hydration condition of fly ash, the amount of C-S-H gel was reduced compared with Figure 12B. A large number of fine particles of tailings were not connected and the distribution was more dispersed, which led to the increase in the porosity of the backfill and the inability to form a dense structure (Liu et al., 2018; Fu et al., 2020). The macroscopic performance of the backfill strength is reduced to 1.90 MPa. Comparing Figure 9D and Figure 9A, it can be seen that during 28–56 d, the flocculent gel further developed and closely connects with the aggregate. Besides, the tightness of the backfill increased significantly and the porosity decreased, i.e., a large amount of hydration reaction of fly ash occurs during 28–56 d.

Based on the results of scanning electron microscope analysis and the hydration characteristics of cement and fly ash, the strength variation process of the backfill containing fly ash can be divided into three stages: cement hydration period, fly ash infiltration period, and slurry hardening period, as shown in Figure 13.

During the hydration period of cement, C₃S, C₂S, C₃A, and C₄AF in cement were hydrated to generate Ca(OH)₂, C-S-H, and other gel materials. The slurry is in a strong alkali state, and the surface of fly ash was wrapped by gel particles after hydration, which led to slow OH⁻ infiltration, thus reducing its hydration rate and less participation in the early stage of hydration. During the fly ash infiltration period, although the gels were gathered on the surface of fly ash particles, their structure was not stable and was continuously destroyed and reorganized under the osmotic pressure of the slurry solution to promote OH⁻ infiltration (Zhang et al., 2020; He et al., 2021). In addition, the pores between the cementitious particles also provided channels for OH⁻ immersion. The OH⁻ reacted with a large amount of SiO₂ and Al₂O₃ in fly ash to accelerate the hydration of fly ash in the middle and late stages, which continuously generated C-S-H and calcium aluminate gels, and it was the main stage of hydration reaction occurring in fly ash. However, when tailings were added to the slurry, the hydration conditions of fly ash deteriorated due to its finer particles and larger specific surface, which could be uniformly distributed around the fly ash and reduced the contact area between fly ash and Ca(OH)₂ (Lee and Kim, 2017; Liu et al., 2018). The fly ash itself had no strength, thus presenting the results of group B: when the fly ash-cement mass ratio exceeded 3, the late strength of the backfill decreased instead. During the hardening period of the slurry, the gel particles generated by the hydration reaction of cement and fly ash combined and continuously formed a structurally stable gel and wrapped coarse and fine aggregates to form a dense structure with reduced porosity, which macroscopically showed an increase in the strength of the filled body. Therefore, to play the role of fly ash in the backfill, firstly, it was necessary to determine the appropriate cementitious material ratio to generate the right amount of OH⁻ during the hydration period of cement to maximize the activation of fly ash and increase its hydration; secondly, the gel structure produced by hydration needed to be in excellent ratio with the aggregate in the stage of slurry hardening period, with good connection performance to reduce the porosity and ensure the quality of the backfill.