2.1 Sample preparation

The high-water material used in the present research was provided by the Yangzhou Zhongkuang Building New Material Technology Co., Ltd. This two-component material has been widely used for underground mines, one component of which is made of calcium aluminate, calcium sulfoaluminate, and the other component is mainly gypsum and lime. It should be noted that the one-component slurry will not harden within 24 h before the mixture of the two components. According to GB/T 39336-2020 and GB/T 39337-2020 (Technical requirement of roadside bag filling for gob-side entry retaining with high-water material [GB/T 39336-2020]; Technical requirement of fully mechanized bag filling mining with superhigh-water material [GB/T 39337-2020]; Feng et al., 2010; Yan, 2004; Zhang & Li, 2021), 57 standard specimens with a diameter of 50 mm and a height of 100 mm were prepared with the recommended water-to-solid ratios (i.e., 1.0, 1.5, and 2.0), the preparation procedure of which is illustrated in Figure 1.

As listed in Table 1, all these specimens were divided into two groups according to the loading modes (i.e., the uniaxial compression and tri-axial compression). It can be seen from Table 1 that there are three identical specimens prepared and tested to investigate the effects of the water-to-solid ratio (i.e., 1.0, 1.5, and 2.0) and the curing time (1, 3, 7, 14, and 28 days) as per GB/T 50081-2019 (Standard Test Method for Physical and Mechanical Properties of Concrete). For the triaxial compression tests, only one specimen was prepared to investigate the mechanical response of the high-water material with variable parameters. For ease of reference, each specimen was assigned a name consisting of three components: the Arabic numeral representing the loading mode, followed by the number indicating the water-to-solid ratio, and a final number distinguishing identical specimens from one another. For example, specimen 1-1.0-3 refers to the third specimen prepared for uniaxial compression testing at 7 days, with a water-to-solid ratio of 1.0.

2.2 Experimental equipment

As illustrated in Figure 2, the MTS-E45.605 mine rock mechanics testing system was adopted for uniaxial compression tests, for which the DH3820Net static strain-testing system was applied to monitor the deformation during the whole test.

Triaxial compression tests were performed on the GCTS RTR1000 rock mechanics integrated testing system, the test setup of which is illustrated in Figure 3.

A Sigma 300 high-resolution field-emission scanning electron microscope (Figure 4a) was applied to explore the microstructural changes of the high-water material under different loading states. Moreover, the AP-608 automatic porosity and permeability tester shown in Figure 4b was adopted to evaluate the porosity. The mineral component testing was conducted in the ARL9900 X-ray fluorescence spectrometer illustrated in Figure 4c.

2.3 Experimental plan

Based on the calculation results of Equation (1), it can be determined that the lateral confinement provided by FRP materials with an elastic modulus of 27 GPa, a layer thickness of 0.5 mm, and a tensile peak strain of 1.3% for 1, 2, 3, 4, and 5 layers is approximately 7, 14, 21, 28, and 35 MPa, respectively. Therefore, the confining pressure gradients for the experimental groups were set at 35, 28, 21, 14, 7, and 0 MPa. The control group specimens were loaded to a confining pressure of 35 MPa at a loading rate of 5 MPa/min. During the testing process, circumferential strain and bleeding amount of water were monitored.

After the completion of the triaxial tests, specimens were extracted, and thin circular slices with diameters of 5–10 mm and thicknesses of 1–5 mm, as well as standard cylindrical specimens with diameters of 25 mm and heights of 50 mm, were obtained from the middle position along the axial direction of the specimens. Additionally, powder samples with particle sizes less than 200 mesh were prepared by grinding. The thin circular slices were observed under a Sigma300 high-resolution field emission scanning electron microscope to analyze the microstructural changes. The porosity and permeability of the cylindrical specimens were measured using an AP-608 automatic porosity permeability tester. Furthermore, the mineral composition content of the powder samples was analyzed using the ARL9900 X-ray fluorescence spectrometer. More detailed information about the experimental contents and plan can be seen in Figure 5.

3.1 Uniaxial compression tests

Figure 6 depicts the typical failure modes of the unconfined high-water material under uniaxial compression. It is apparent that these noticeable cracks generally developed from both ends toward the middle of the specimen. These specimens with shorter curing times and higher water-to-solid ratios usually exhibited the ductile deformation associated with the minor failures, while these samples with longer curing times and lower water-to-solid ratios featured the brittle failure. Taking the specimen with a curing time of 1 day for comparison, the cracks were mainly concentrated at the upper and lower ends, which can be seen in Figure 6. As the curing time increased, the number of cracks consequently increased and the locations of which developed from the middle height to the ends.

To further explore the failure modes of the unconfined high-water material, the failed specimens were manually split, and the inner surface of which is illustrated in Figure 7. It is interesting that there is no more obvious damage observed from the inner surface of the specimens, rather than limited pores, non-uniformly distributed within the surface.

The representative stress–strain curves of the high-water material under uniaxial compression are shown in Figure 8, for which all specimens shared the same curing time of 1 day. When the water-to-solid ratio of the high-water material increases from 0.5 to 1.0, the uniaxial compressive strength decreases to 3.08 and 1.81 MPa, respectively. When other parameters are constant, the peak strengths of the high-water material with a water-to-solid ratio of 1.0 are 4.71 MPa at 1 day, 5.61 MPa at 3 days, 5.88 MPa at 7 days, 6.00 MPa at 14 days, and 6.29 MPa at 28 days. These observations agree well with previous research on the influence of the water-to-powder ratio and curing time.

According to GB/T 50081-2019, the error analysis was conducted in the present research to verify the consistency of test data. As listed in Table 2, the differences between the maximum value, minimum value, and the middle value of each set of parallel specimens are within the values of 15%, suggesting the consistency of the high-water material.

3.2 Triaxial compression loading

Under the lateral constraint condition, the volume of the high-water material exhibited a continuous decrease within the loading process. The compressed shapes of the specimens under different confining pressures are shown in Figure 9a, for which there is no obvious cracks are observed on the surface of the specimens. Figure 9b depicts the internal sectional view of representative specimens, while Figure 9c,d shows the lateral expansion view of the specimens. Observation reveals that there were no cracks on the surface or inside of the high-water material specimens after loading.

Figure 10 shows the relationship between the water bleeding rate and the circumferential strain of the high-water material at the given confining pressure. It is obvious that the increased confining pressure results in the enlarged circumferential strain. Despite monitoring failures, the mechanical response pattern of the circumferential strain to the confining pressure could still be observed. The volume change exhibited a distinct trend of initial slow contraction, followed by rapid compression, and finally slow and sustained densification.

The confining pressure-bleeding rate/circumferential strain relationships for high-water material specimens with different water-to-solid ratios and curing times of 3 days are shown in Figure 10a. For the specimen with a water-to-solid ratio of 1.0 (2-1.0-1), drainage began at 25.61 MPa with a flow rate of 0.024 mL/s (0.477 mL/MPa). Monitoring failed when the circumferential strain reached 4.27% at 13.11 MPa, and the flow rate decreased to 0.001 mL/s (0.013 mL/MPa) at 32.78 MPa. Slow drainage, along with particle outflow, continued until 35 MPa, resulting in 3.45 mL of water, with a water bleeding rate of 2.51%. For the specimen with a water-to-solid ratio of 1.5 (2-1.5-2), drainage started at 11.08 MPa with a flow rate of 0.051 mL/s (1.013 mL/MPa). Monitoring failed at 13.88 MPa after the circumferential strain reached 4.35%, and the flow rate decreased to 0.002 mL/s (0.042 mL/MPa) at 16.68 MPa, with 6.42 mL of water collected at 35 MPa, the water bleeding rate of which is 4.26%. In the case of the specimen with a water-to-solid ratio of 2.0 (2-2.0-1), drainage began at 7.01 MPa with a flow rate of 0.095 mL/s (1.896 mL/MPa). Monitoring failed when the circumferential strain reached 4.28% at 4.67 MPa, and the flow rate decreased to 0.003 mL/s (0.065 mL/MPa), collecting 18.60 mL of water at 35 MPa, with a water bleeding rate of 11.92%. The variation in the water-to-solid ratio affected (1) the pressure at which drainage began, (2) the flow rate, and (3) the final amount of water collected. Higher water-to-solid ratios required lower pressures to start draining, which resulted in greater final amounts of water being collected.

The confining pressure-bleeding rate/circumferential strain relationships for high-water material specimens with a water-to-solid ratio of 1.5 at different curing times are depicted in Figure 10b. For the specimen with a curing time of 1 day (2-1.5-1), drainage began at 9.01 MPa at a flow rate of 0.059 mL/s (1.193 mL/MPa). Monitoring failed at 8.70 MPa after the circumferential strain reached 4.27%, and the flow rate decreased to 0.004 mL/s (0.076 mL/MPa) at 13.29 MPa, with 6.76 mL of water collected when loaded to 35 MPa, with a water bleeding rate of 4.41%. For the specimen with a curing time of 3 days (2-1.5-2), drainage started at 11.08 MPa at a flow rate of 0.051 mL/s (1.013 mL/MPa). Monitoring failed at 13.88 MPa after the circumferential strain reached 4.35%, and the flow rate decreased to 0.002 mL/s (0.042 mL/MPa) at 16.68 MPa, with 6.42 mL of water collected when loaded to 35 MPa, with a water bleeding rate of 4.26%. For the specimen cured for 7 days (2-1.5-8), drainage began at 14.85 MPa with a flow rate of 0.041 mL/s (0.823 mL/MPa). Monitoring failed at 18.24 MPa after the circumferential strain reached 4.65%, and the flow rate decreased to 0.003 mL/s (0.057 mL/MPa) at 21.34 MPa, with 6.13 mL of water collected when loaded to 35 MPa, with a water bleeding rate of 4.17%. For the specimen cured for 14 days (2-1.5-9), drainage began at 16.78 MPa with a flow rate of 0.058 mL/s (0.7 mL/MPa). Monitoring failed at 12.78 MPa after the circumferential strain reached 4.24%, and the flow rate decreased to 0.003 mL/s (0.034 mL/MPa) at 24.77 MPa, with 5.94 mL of water collected when loaded to 35 MPa, with a water bleeding rate of 4.04%. For the specimen cured for 28 days (2-1.5-10), drainage began at 17.50 MPa with a flow rate of 0.035 mL/s (0.425 mL/MPa). Monitoring failed at 10.294 MPa after the circumferential strain reached 4.30%, and the flow rate decreased to 0.005 mL/s (0.060 mL/MPa) at 28.32 MPa, with 4.99 mL of water collected when loaded to 35 MPa, the water bleeding rate of which is 3.47%.

Increasing the curing time resulted in higher specimen drainage envelope pressures and initial flow rates. All the specimens exhibited slow drainage and particle outflow at 35 MPa.

Analysis of the loading process of the high-water material specimens revealed that various phenomena correspond to specific confining pressure conditions. As the curing time increased and the water-to-solid ratio decreased, the confining pressure required to initiate the water expulsion increased, whereas the flow rate decreased. In addition, changes in the flow rate occurred at higher confining pressure. The relationship curves of the curing time, water-to-solid ratio, and confining pressure at the critical points are shown in Figure 11.

The relationship curves among the gradient confining pressure, water output, and circumferential strain for high-water materials under lateral constraints are shown in Figure 12. For specimens 2-1.5-2, 2-1.5-3, 2-1.5-4, 2-1.5-5, 2-1.5-6, and 2-1.5-7 (with a curing time of 3 days and a water-to-solid ratio of 1.5), the water outputs under confining pressures of 35, 28, 21, 14, 7, and 0 MPa were 6.75, 6.06, 5.78, 2.24, 0, and 0 mL, respectively. The peak values of circumferential strain were 4.35%, 4.29%, 4.26%, 4.24%, 3.98%, and 0, respectively. Analysis of the experimental data revealed that the water output increased with increasing confining pressure. When the confining pressure reached a certain threshold, the samples began to bleed water, and the flow rate initially increased rapidly and then decreased, indicating that under no lateral constraint and low confining pressure conditions, the high-water material samples did not expel water. In addition, the circumferential strain decreases significantly as the confining pressure decreases.

4.1 Component testing

The components of high-water material specimens under different confining pressure conditions were analyzed using the ARL9900 X-ray fluorescence spectrometer. The obtained diffraction spectra are shown in Figure 13. By utilizing X-ray diffraction data processing and analysis software, the content of minerals such as ettringite, gypsum, and CaSO4·0.5H2O, CaO, and Al2O3 in the high-water material under various confining pressure conditions was determined. It was observed that under confining pressure loading conditions ranging from 0 to 35 MPa, the content of ettringite remained relatively constant. The results of the analysis are summarized in Table 3. The forms of these minerals in the high-water material specimens mainly exist in two forms: free water in the pore structure and structural water in the ettringite crystal structure. Based on the content analysis of ettringite, it can be concluded that the main source of bleeding in high-water materials under 0–35 MPa confining pressure conditions is the free water in the pore structure of the high-water rapid-setting material.

4.2 Microstructure testing

The microstructural characteristics of the mine-based high-water materials under lateral constraints are shown in Figure 14. A comparison of the microstructural characteristics of these materials under different test conditions revealed two main changes: the microscopic morphology and pore characteristics.

The extraction and classification of the microscopic morphology of ettringite are illustrated in Figure 15. The SEM analysis results in Figure 14 reveal that the microscopic morphology of ettringite exhibits four typical characteristics: loose fibrous (Morphology I), dense fibrous (Morphology II), rod-like (Morphology III), and dense granular (Morphology IV).

The microscopic morphological changes in high-water material specimens under 0–28 MPa confining pressure mainly involve the transformation of needle-like ettringite (Phase I) into clustered ettringite (Phase II) as the pores are compressed under the confining pressure. However, the microscopic structure remains unchanged. The morphological changes of ettringite under 21 and 28 MPa confining pressure conditions are not significant, indicating that the closure degree of the pores in the specimens has reached a relatively high level. Therefore, it is believed that during this stage, the reason for water bleeding of high-water materials under lateral constraint is that the pores of the high-water materials are compressed and tend to stabilize under the confining pressure, causing the free water in the pores to be squeezed out. This is manifested as continuous bleeding of the specimen, with a faster initial bleeding rate during loading, followed by a decrease in the bleeding rate and reaching a relatively stable state.

When the confining pressure is increased from 28 to 35 MPa, the analysis of the microscopic structure of high-water materials reveals that at this stage, the continuous clustering of needle-like ettringite (Phase I and Phase II) under the confining pressure continues. Some portions of the ettringite exhibit a rod-like structure (Phase III), and there are areas where the ettringite structures cluster into dense masses (Phase IV). At this stage, the specimen continues to bleed, but at a slower rate. The morphological changes in ettringite may lead to the conversion of some structural water into free water, which is then expelled under the influence of confining pressure. The possibility of the transformation of structural water in ettringite crystals into free water will be further explored in Section 4.2. Therefore, at this point, the reason for water bleeding of high-water materials under lateral constraint is that the free water in the pores is the main source of maintaining the bleeding phenomenon of high-water rapid-setting material specimens. There may be a conversion of some structural water in ettringite into free water, which, together with the residual free water in the pores, is expelled under the influence of the confining pressure.

4.3 Assessment of porosity and permeability

Owing to its small molecular diameter (which provides more precise measurements), helium gas was used to test the volume porosity and permeability of the high-water materials. Additionally, the results of the porosity and permeability tests are presented in Table 4. Notably, the volume porosity measurement of sample 2-1.5-4 was inaccurate owing to damage at the bottom during postprocessing.

A comparison of samples 2-1.0-1, 2-1.5-2, and 2-2.0-1 (curing time of 3 days), as shown in Figure 16a, and samples 2-1.5-1, 2-1.5-2, 2-1.5-8, 2-1.5-9, and 2-1.5-10 (water-to-solid ratio of 1.5), as shown in Figure 16b, reveals that the porosity, permeability, and water discharge of high-water materials increase with the water-to-solid ratio and decline with the curing time.

When comparing samples 2-1.5-2, 2-1.5-3, 2-1.5-4, 2-1.5-5, 2-1.5-6, and 2-1.5-7 (with a curing time of 3 days and a water-to-solid ratio of 1.5), as depicted in Figure 17, the volume porosities of the high-water material samples under confining pressures of 35, 28, 21, 14, 7, and 0 MPa were found to be 55.38%, 55.24%, 53.56%, 55.22%, 56.01%, and 58.13%, respectively. Their surface porosity and permeability values were 48.30%, 52.31%, 53.86%, 54.81%, 55.36%, and 57.63% and 7.883, 7.187, 6.724, 8.168, 8.431, and 10.854 md, respectively. The bleeding rates recorded were 4.26%, 4.02%, 3.84%, 1.49%, 0, and 0, respectively. Analysis of the porosity of the high-water material samples under various lateral constraint conditions revealed that the porosity decreases with increasing confining pressure. Notably, when the confining pressure reached 21 MPa, the rate of decrease in the volume porosity decreased. This trend correlated with the observed bleeding rate, which started rapidly and then slowed during drainage, indicating a shift in the material's behavior under stress.

5.1 Law of free water bleeding

It is believed that the free water is internally and externally stored in the specimen. With the increase of the confining pressure, free water near the surface of the specimen gradually dissipates, and the extent of dehydration expands toward the interior. When the confining pressure reaches 21 MPa, free water originally near the surface has been bled out. It should be noted that there is still a certain amount of free water within the specimen, as shown in Figure 18a,b. As the confining pressure increased from 0 to 35 MPa, drainage began at 11.08 MPa with a flow rate of 0.051 mL/s. When the pressure increased to 16.68 MPa, the flow rate decreased to 0.002 mL/s. Correspondingly, the bleeding rates of the high-water material at the given confining pressures (i.e., 0, 7, 14, 21, 28, and 35 MPa) are 0, 0, 1.49%, 3.84%, 4.02%, and 4.26%, respectively. Based on the amount and rate of water bleeding, the bleeding of free water can be divided into three stages: the air expulsion stage, the rapid bleeding stage, and the slow bleeding stage. In the present research, the threshold points for these three stages are 11.08 and 16.68 MPa, which can be seen in Figure 18c. That is, the pores within the internal structure of the high-water material are rapidly compressed at the initial stage. Once the air within the pores is expelled, the original free water begins to bleed out, with a relatively fast flow rate. As the confining pressure reaches a certain level, the porosity of the specimen stabilizes, and the proportion of original free water undergoes a significant decrease. Consequently, the flow rate slows down as the rate of pore closure decreases.

5.2 Mechanism of structural water removal

5.2.1 Classification and removal sequence of structural water in ettringite crystals

The structural water of the ettringite crystals (3CaO·Al2O3·3CaSO4·32H2O) within the high-water material can be classified into four groups: the interlayer water, the calcium polyhedron secondary coordinated water, the calcium polyhedron primary coordinated water, and the aluminum hydroxide ions (Al-OH)−, accounting for 45.9% of the ettringite crystals in mass (Jakob et al., 2019; Jiménez & Prieto, 2015). As illustrated in Figure 19, the positions of structural water within the ettringite crystal are different from each other (Hartman et al., 2006; Li & Qiao, 2023; Shimada & Young, 2001).

When the ettringite is influenced by the external factors, both the interlayer water and calcium secondary coordinated water are removed in order of occupancy. Two interlayer H2O molecules may be adsorbed on the surface of the ettringite crystal or exist within defects in the crystal, making them easier to detach first. At this stage, the ettringite structure resembles a trigonal prism with two main vertices and two additional vertices. The H2O molecules at the main vertices act as primary coordinated water, connecting to the calcium atoms in a double-bonded form with shorter Ca–O bonds. The H2O molecules at the additional vertices serve as secondary coordinated water, linking to the calcium atoms in a single-bonded form with longer Ca–O bonds. After the removal of interlayer water, the 12 secondary coordinated water molecules in ettringite start to be removed, reducing the total crystal structural water from 30 H2O molecules to 18 H2O molecules. As previously indicated, the loss of calcium secondary coordinated water does not alter the crystal structure (Skoblinskaya & Krasilnikov, 1975a, 1975b). However, the coordination number of the calcium polyhedron decreases from 8 to 6, strengthening the connections between the calcium atom and two primary coordinated water molecules and four aluminum hydroxide ions (Al-OH)−. The primary coordinated water molecules shift closer to the calcium atom, enhancing the bonding strength of the aluminum octahedron's OH– bonds, ultimately transforming the calcium polyhedron from an undecahedron to a tetragonal bipyramid. Once all the calcium secondary coordinated water is removed, the subsequent dehydration sequence of ettringite does not continue in the order of occupancy. Instead, the ettringite first sheds some of the primary coordinated water from the calcium polyhedron, followed by the removal of aluminum hydroxide ions (Al-OH)−, with both types of water being removed simultaneously. After the removal of some primary coordinated water molecules, a significant change occurs in the coordination number of the calcium atom, causing distortion in the central column and disrupting the long-range order of the crystal structure.