To investigate the impact of hydraulic conditions on the seepage characteristics of loose sandstone, this study employed optimized methods to prepare loose sandstone samples. Subsequently, seepage experiments were conducted under different injection pressures, flow rates, and flow volumes. The permeability, porosity, particle size distribution, and other parameters of the rock samples were obtained. By analyzing the response of seepage characteristics to pore and particle size characteristics, the influence of different hydraulic conditions on the seepage characteristics of loose sandstone was explored. The results indicated that improvements in the parameters of hydraulic conditions had different effects on various rock samples. For rock samples with developed seepage channels, increasing the value of each hydraulic condition parameter could expand the channels and discharge particles, and improve permeability. For rock samples with a larger number of small pores, increasing each hydraulic condition parameter caused particles to crack under pressure, drove particles to block holes, and thus reduced permeability. In this experiment, the permeability parameter had a significant positive response to the proportion of pores larger than 0.1 µm and a significant negative response to the proportion of particles smaller than 150 µm.
In the fields of oil extraction and solution mining, loose sandstone is a common resource-bearing rock mass, and the seepage characteristics of this type of rock are directly related to the process design and efficiency of resource extraction [1-5]. During the processes of extraction and injection in rock formations, problems often occur, such as increased rock permeability leading to imbalanced extraction and injection or decreased permeability leading to low mining efficiency [6-8]. In the above situations, a common countermeasure in industrial practice is to use agents or equipment to adjust the permeability of rock strata to improve production. For example, in oil exploitation, dispersed gel particles with a certain particle size are used to temporarily plug the high permeability area to better drive the production of the reservoir [9, 10], and unblocking agents are used in situ to unblock clogged channels to improve the efficiency of uranium leaching [11-13]. Implementation of the above methods provides immediate improvement in the seepage effect, and the cost can be controlled, effectively solving problems related to abnormal seepage in rock formations. However, it is important to note that while the solutions for abnormal seepage are relatively effective, they are still confined to a reactive, postevent stage. The time and costs associated with these solutions continue to impede the enhancement of production efficiency. The objective of this study is to address these issues by starting from the mechanisms of permeability changes. It delves into the influence of hydraulic conditions on the seepage characteristics of loose sandstone, sheds light on the microscopic aspects of permeability changes during the extraction process, and aims to resolve challenges such as the difficulty in monitoring ore body permeability changes and suboptimal dewatering effects.
Hydraulic parameters, such as water injection pressure, flow rate, and flow volume, are important process parameters in rock layer pumping operations [14-19]. In engineering, it is often necessary to establish an empirical relationship between the above parameters and the permeability of rock layers to grasp the changes in the seepage characteristics of rock layers. By purposefully adjusting hydraulic condition parameters, the seepage characteristics of rock layers and mining efficiency can be improved. Relevant methods have been widely used on project sites and have proven to be effective [20, 21]. However, it is not possible to apply the empirical relationship between hydraulic conditions and seepage characteristics obtained by monitoring data from a single mining area to other projects, and relationships are not consistent for different mining areas in the same project. Therefore, the priorities are to summarize the influence of hydraulic conditions on the seepage characteristics of loose sandstone and apply experimental methods to reveal the internal mechanism. So as to grasp the fundamental principles of changes in the seepage characteristics of loose sandstone from the root and then scientifically improve the efficiency of rock seepage operations.
In the study of the principles of the influence of hydraulic conditions on seepage characteristics, the commonly considered microscopic characteristics and mechanisms include pore distribution characteristics, particle distribution characteristics, and interactions between particles and pores [22-25]. The characteristic parameters of pore distributions can effectively reveal the pore development of rocks with seepage, and the change characteristics and mechanisms of seepage characteristics can be derived from them [26-29], while the analysis of particle characteristics focuses on the influence of particle size distribution on pore structure and the influence of free particles on seepage channels [29, 30]. Research has shown that pore and particle characteristics basically determine rock permeability, and the two characteristics are interrelated . Thus, the combination of research on the characteristics of pore distributions and particle size distributions of rock and seepage experiments conducted from the microscopic perspective can fully reveal the relationship between hydraulic conditions and seepage characteristics of loose sandstone.
However, in the laboratory, there is a clear lack of microscopic research on the seepage characteristics of loose sandstone due to its unique lithological characteristics. The cohesion of loose sandstone mainly comes from the capillary action between unsaturated rock particles as well as fine-grained soil [32, 33]. This type of force has poor stability and fails in a saturated state, making it difficult to cut and form loose sandstone, and rock particles easily disintegrate and diffuse in water to damage seepage pipelines. In addition, natural loose sandstone is prone to significant experimental errors due to the large degree of randomness of the particle distribution. Thus, few seepage experiments have been conducted on loose sandstone in the laboratory. Some scholars have adopted numerical simulation or the overall seepage method of large rock samples to obtain the changes in seepage characteristics [34-38]. However, this type of method can only be used to evaluate parameters such as permeability changes and cannot intuitively obtain key parameters such as porosity and particle distribution before and after the experiment, making it difficult to explain the principles for the changes in the seepage characteristics of loose sandstone. In recent years, some studies used lateral wrapping or special fixtures to protect loose sandstone samples [39-41], successfully achieving small-scale seepage experiments, and their rock sample protection methods are worthy of reference. However, according to the figures from relevant studies, the success of such experiments was due to the good integrity of the sandstone samples. However, particle spalling was obvious at the exudation end of the rock sample after these experiments, which led to significant experimental errors and made it difficult to carry out additional experiments. Therefore, due to the inherent defects of the samples, experimental research on the seepage characteristics of loose sandstone is still a major challenge, and there is an urgent need for an innovative method to protect rock samples to support the development of such experiments.
To investigate the patterns and mechanisms by which different hydraulic conditions influence the porosity and particle characteristics of loose sandstone, consequently altering its seepage characteristics, a special optimized preparation method for loose sandstone was designed and used to prepare rock samples. Seepage experiments were conducted on rock samples under different injection pressures, flow rates, and flow volumes. Key parameters, such as permeability, porosity, and particle size distribution, were obtained before and after the experiment through seepage meters, nuclear magnetic resonance spectroscopy, and particle size distribution instruments, and the responses and principles of hydraulic conditions, seepage characteristics, porosity distribution characteristics, and particle size distribution characteristics were comprehensively analyzed. Finally, the influence relationships and principles of the influence of different hydraulic conditions on the seepage characteristics of loose sandstone were summarized, providing a theoretical reference for engineering practice and assisting in improving the efficiency of operations related to the seepage of resources containing loose sandstone.
2. Materials and Methods
2.1. Experimental Materials and Instruments
In this experiment, loose sandstone samples A and B were obtained from the 400 m underground seam of uranium mine A in Xinjiang and the 500 m underground seam of uranium mine B in Inner Mongolia. The results of mineral phase microscopy identification of two types of rock samples are shown in Table 1. The dry density of sample A was 1.69 g/cm³, and sample B had a dry density of 1.62 g/cm³. The particle size distribution curves are shown in Figure 1.
Seepage experiments were conducted using an HKY-1 long-core seepage monitoring device made by Hai'an County Petroleum Scientific Research Instrument Co., Ltd, as shown in Figure 2(a). The main structure of the instrument is shown in Figure 2(b), including a control computer, monitors, constant flow pump, pressure pumps, experimental chamber X, confining chamber Y, return pressure chamber Z, and multiple pressure measurement points.
In Figure 2(b), X and Y are not connected, and the core is tightly clamped by driving the confining pressure pump to inject liquid into Y. X and Z are connected, and the piston inside the chamber is moved up to block the channel by driving the return pressure pump to inject liquid into Z. The channel opens when the water pressure between X and Z is equal. This structure allowed the instrument to set the return pressure to pressurize the entire experimental chamber to simulate the internal water injection pressure of the rock sample.
Among the five pressure measurement points, the third and fourth measurement points were the confining pressure and return pressure measurement points. The fifth measuring point was the external pressure measuring point of the rock sample. When the value of this measuring point was not 0, it indicated that the side of the sample was in a connected state, and it was necessary to increase the clamping confining pressure.
The first and second measuring points were the pressure measurement points at the injection and discharge ends, and the difference between the two was the seepage pressure difference. By recording the seepage pressure difference during the seepage process ΔP, the real-time permeability of the sample was calculated using the following formula:
where the parameters were permeability k, mD; flow rate Q, m3/s; sample length L, m; sample bottom area A, m2; and seepage pressure difference ΔP, Pa. Pure water was used as the percolation solution in the experiment, and the viscosity μ = 1.01 × 10-3 Pa·s was taken at room temperature.
In the experiment, the characteristic parameters of the sample pore distribution during the experiment were obtained through a NIUMAG MesoMR medium-size nuclear magnetic resonance imaging analyzer, as shown in Figure 3(a), and the characteristic parameters of the sample particle size distribution were obtained through a Malvern Panalytic Mastersizer 3000 laser diffraction particle size analyzer, as shown in Figure 3(b).
2.2. Sample Preparation of Loose Sandstone
Figure 4 illustrates the full disintegration process of loose sandstone samples in a saturated state, rendering it challenging to perform seepage experiments on natural samples. Nevertheless, this also signifies that these rock samples exhibit minimal cementation characteristics, making the use of the particle ratio method for artificial sample preparation a viable option.
In response to the difficulties in preparing loose sandstone samples, the tendency for loose sandstone samples to disintegrate in water, and the large experimental errors, a special method was used to optimize the preparation of loose sandstone samples in this study. The specific steps were as follows:
The natural rock sample was completely broken and granulated while maintaining the original particle size distribution.
The particle size distribution parameters of the original rock samples were obtained by using the particle size distribution meter.
According to the particle size distribution parameters, the particles were proportioned and mixed evenly one by one during sample preparation.
A standard loose sandstone sample was prepared according to the steps shown in Figure 5(a).
In step 4, the auxiliary materials for sample preparation included thermoplastic pipes, dense mesh filters, and porous fixing sheets. The initial diameter of the thermoplastic pipe was 60 mm. For clamping and shaping purposes, a diameter of 50 mm and a height of 50 mm cylindrical mold were used to preshape the thermoplastic pipe, and thus, the resulting rock sample was a uniformly sized cylinder. The pore size of the dense mesh filter at both ends of the rock sample was 150 µm. This allowed fine particles to seep out of the rock sample, while coarse particles were retained to maintain the skeletal structure of the rock sample. The aperture of the porous fixing sheet was 1 mm, which played a role in seepage and clamping shaping. The above auxiliary materials could continuously fix the shape of the rock sample and ensure that particles did not collapse during the seepage process. Figure 5(b) shows the rock sample during the saturation process, which had a stable structure and did not disintegrate. Figure 5(c) shows the comparison between artificial and natural rock samples. In the unwrapped state, both types of rock samples showed significant particle detachment.
2.3. Experimental Plan and Design
This study focused on loose sandstone obtained from on-site sampling of projects A and B. Stable and uniform rock samples were made through special sample preparation and solidification methods. The influences of different hydraulic conditions on the seepage characteristics of loose sandstone were investigated. Based on the initial dry density of the rock samples with a diameter of 50 mm and a height of 50 mm, the weights of the rock samples were 165.91 g in Group A and 159.04 g in Group B. The particle size distributions of the rock samples are shown in Figure 1. The specifications for auxiliary materials for sample preparation are shown in Table 2.
Seepage experiments were designed and conducted on loose sandstone specimens A and B for various hydraulic conditions such as injection pressure, flow rate, and flow volume as key parameters.
To investigate the influence of different injection pressures and flow rates on the seepage characteristics of rock samples under the basic conditions of a total flow rate of 2000 mL, two injection pressures and three flow rates were designed, and a total of twelve experimental groups were prepared, as shown in Table 3(a). To investigate the influence of different flow rates on the seepage characteristics of rock samples, the basic conditions were 10 MPa injection pressure and 5 mL/min flow rate, and a total of eight experimental groups were prepared, as shown in Table 3(b). The experimental contents of Groups 5 and 9 were the same, and they shared the same experimental group.
3. Seepage Experiment and Response Analysis
The temperature of the experimental chamber was maintained at 25℃, and the pressure and flow rate were recorded every 10 seconds during the experiment. Seepage experiments were conducted according to the preset groups, and the mass and volume were measured before and after the experiment. The results are shown in Table 4. In terms of dry mass, the mass loss after the experiment was significantly greater for Group A than Group B, indicating that the loss of particles below 150 µm in the seepage experiment was greater for Group A than Group B. There was a significant increase in wet mass after the experiment that was more significant for Group A than Group B. This indicated that although Group A had a greater particle loss, its pore water storage capacity was larger. The volumes of all the rock samples decreased after the experiment, and the decrease was more significant for Group A than Group B. This may have been related to greater particle loss and more pronounced pore compression in Group A.
The pressure difference between pressure measurement points 1 and 2 was used as the seepage pressure, and the pressure change curves of samples A1, A2, A3, B1, B2, and B3 (injection pressure of 1 MPa, flow rate of 1, 5, and 10 mL/min, and total flow volume of 2000 mL) were plotted as shown in Figure 6(a). The seepage pressure of the sample showed an upward trend, indicating that not only particle loss but gradual blockage of seepage channels by particles occurred, resulting in an increase in seepage pressure. At the same flow rate, the pressure value was significantly higher for the Group B sample than the Group A sample, and the difference between different flow rates within the group was greater. This indicated that the degree of pore development of the Group A sample was greater than that of Group B, and the liquid more easily passed through the sample. Both sets of pressure curves exhibited fluctuations, and the faster the flow rate was, the more intense the curve fluctuations. This indicated that there was a cycle of blocking and unblocking in the pores during the seepage process. According to the characteristics of the curves, the frequency of this cycle was higher in the earlier stages of seepage.
Figure 6(a) shows the fits to the permeability change data for the corresponding samples according to equation (1) and the specification parameters of the rock samples, as shown in Figure 6(b). The permeability changes in both groups of samples were characterized by a rapid decrease in the early stage of seepage followed by gradual slowing. This feature indicated that the blockage of the pores during the seepage process occurred extensively in the early stage of the seepage experiment and then gradually stabilized. As the flow rate increased, the permeability of Group B tended to be consistent, while the permeability of Group A differentiated into three different values and was positively correlated with the flow rate. This indicated that the permeabilities of Group A samples had significant positive responses to the flow rate, while those of Group B had smaller responses.
By comprehensively analyzing the influence of hydraulic conditions, such as injection pressure, flow rate, and flow volume on the permeability of loose sandstone, permeability tests were conducted on the samples before and after the seepage experiment. The test conditions were as follows: confining pressure of 1 MPa, flow rate of 1 mL/min, flow volume of 10 mL, and no return pressure. The permeability changes under different hydraulic conditions are shown in Figures 7(a)–7(c), and the permeability change rates before and after the experiment are shown in Figures 7(d)–7(f). The fitting functions are shown in Table 5. Due to different return pressure settings, there were differences in permeability between Figure 6 and Figure 7.
According to Figures 7(a)–7(c), the seepage experiment had opposite effects on the permeabilities of the two groups of samples. Under the same conditions, the permeabilities were larger for Group A than Group B samples, and this characteristic was more obvious after the experiment. This may have been related to the particle change characteristics of the two groups of samples. According to Table 4, the dry mass change characteristics of Group A showed that there was a greater loss of fine particles, leading to more developed permeation channels after the permeation experiment. Although Group B also had particle loss, Figure 6(a) shows that the particle blockage caused by the seepage process of the Group B sample was more prominent. According to Figures 7(d)–7(f), the permeabilities of Groups A and B had significant responses to flow rate and flow volume, and these responses were more significant under the condition of 10 MPa injection pressure. This indicated that increasing the water injection pressure, flow rate, and flow volume in the seepage experiment had a significant effect on the permeabilities of Groups A and B. However, Group A had a positive response, and Group B had a negative response. As shown in Table 4, this may have been related to changes in pores and particles after hydraulic conditions changed.
Based on the above analysis, changes in hydraulic conditions, such as water injection pressure, flow rate, and flow rate, affected the pore and particle characteristics of loose sandstone in Groups A and B and thus significantly influenced permeability. Therefore, completely understanding the variation of characteristic parameters such as porosity and particle size distribution of rock samples under different hydraulic conditions and studying the response of seepage characteristics to the above parameters can effectively explain the relationships and internal mechanisms of different hydraulic conditions affecting the seepage characteristics of loose sandstone.
4. Responses and Principles
4.1. Analysis of the Responses of Seepage Characteristics to Pore Characteristics
Pore structure is the main factor affecting the permeability of rock samples [42, 43]. To investigate the influence of hydraulic conditions such as water injection pressure, flow rate, and flow volume on the porosity of loose sandstone, a nuclear magnetic resonance imaging analyzer was used to evaluate the porosities of the rock samples before and after the seepage experiment. The fundamental principle of this porosity acquisition method involves saturating the rock sample and subsequently employing an analyzer to selectively excite the hydrogen element within the rock sample, causing it to absorb and release energy (referred to as relaxation) in order to derive the relaxation time T2. The pore diameter is then determined using equation (2), and the corresponding pore volume at this diameter is calculated based on the energy intensity generated during excitation. Ultimately, this process yields the characteristic parameters of pore distribution .
where the parameters were relaxation time T2, ms; relaxivity ρ2, μm/s; in accordance with the lithological characteristics of loose sandstone and referencing the table in the experimental instrument manual, this parameter has been set to 10 µm/s; relative surface area S/V μm−1, this parameter is inversely proportional to pore diameter.
The testing environment was 25℃, and the sampling frequency was 200 kHz. The porosity distribution and cumulative curves of Groups A and B are plotted as shown in Figures 8(a) and 8(b), where A0 and B0 are the mean values of the porosity curves of each sample before the experiment. In Figures 8(a) and 8(b), new pores were detected of A and B samples in the section of pore diameters less than 0.05 µm after the seepage experiment, which is due to the hydraulic invasion of the sample micropores under the injection pressure of the seepage experiment, resulting in the detection of smaller pores. This was also the main reason for the increase in the wet mass change rate in Table 4. In Figure 8(a), the first and second peaks of A0 were significantly higher than those of the other samples after the experiment, indicating that the pore compression of Group A is more pronounced after the seepage experiment. However, this phenomenon was not found in Figure 8(b), which was also the reason for the larger volume change rate of Group A in Table 4.
To explore the response of permeability characteristics to pore characteristics, the variations in total porosity under different hydraulic conditions were plotted, as shown in Figures 9(a)–9(c). The cumulative porosities of Group B samples were generally greater than those of Group A, but the permeability performances of the two were opposite. Parameter correlation analysis showed that the correlation between total porosity and permeability was not significant. This was because the total porosity could not directly reflect the permeability of the sample. During the seepage process, pores with diameters less than 0.1 µm require extremely high pressures under the influence of molecular gravity to achieve effective seepage; these types of pores are referred to as capillary pores . Consequently, this study centers on pores with diameters exceeding 0.1 µm, categorizing them as effective pores capable of facilitating efficient seepage, and the distribution and cumulative curves of effective porosity were plotted as shown in Figures 8(c) and 8(d). The variation curves of effective porosity under different hydraulic conditions were plotted as shown in Figures 9(d)–9(f), and their fitting functions are shown in Table 6. The image shows that the effective porosity was greater for Group A than Group B, which was one of the reasons the permeability was greater for Group A than Group B, and this also facilitates the escape of unbound particles in the Group A samples, leading to increased mass loss (refer to Table 4). The correlation coefficient of the relationship expression Ae in Table 6 was only 0.1 because the initial porosity and permeability of the A4 rock sample were the greatest in each group, as shown in Figure 7(b) and Figure 9(b), which made the permeability, porosity, effective porosity, and other values of the A4 experimental group after the experiment significantly larger than those of the other groups.
In Figures 9(d)–9(f), with the increases in injection pressure, flow rate, flow volume, and other hydraulic condition parameters, the effective porosity of Group A showed an overall upward trend, while the effective porosity of Group B did not change significantly. Its value fluctuated near the initial effective porosity before the experiment, indicating a significant positive response for the effective porosity to the hydraulic condition for Group A and no obvious response for Group B. Figure 10 shows the data for the permeability versus effective porosity and curve fits to the data. With increasing effective porosity, the permeabilities of rock samples in Groups A and B increased linearly, which indicated a positive response, and this was still true when the initial value of sample A4 was large. Therefore, Groups A and B permeabilities had significant positive responses to effective porosity. However, the influence of hydraulic conditions on Group B effective porosity was not obvious.
To reveal the differences between Groups A and B in the responses of hydraulic conditions, pore characteristics, and permeability, the distributions and cumulative curves of the porosity change rate were plotted based on the difference in porosity before and after the experiment, as shown in Figures 11(a)–11(f) and Figures 11(g)–11(l).
In Figures 11(g)–11(l), after the experiment, the porosity showed an overall increase for Group A and a decrease for Group B. The final cumulative results of the porosity change rate showed that the cumulative porosity change rates of samples A and B had a certain response to hydraulic conditions such as injection pressure, flow rate, and flow volume, with positive responses for Group A and negative responses for Group B.
In Figures 11(a)–11(f), although the volumes of small mesopores in Group A samples decreased after the experiment, the volumes of micropores and macropores significantly increased. However, there was no significant change in the pore sizes of Group B samples, and the changes were mainly concentrated in micropores and small pores with diameters less than 0.1 micrometers, which may have been related to the lower proportion of large pores in Group B. Therefore, the change in effective porosity was more significant in Group A than in Group B, which was why the response of Group B to hydraulic conditions was not significant.
According to the correlation coefficients between the effective porosity and permeability of Groups A and B samples, the correlations were weak, indicating that other factors besides the effective porosity had a strong impact on the permeability of samples, and Group B was more affected by other factors. Based on the pressure variation curves in Figure 6(a), there were pressure fluctuations in the seepage processes of samples A and B, with those for Group B being more pronounced, indicating a significant decrease in permeability due to pore blockage in Group B. Therefore, it was necessary to analyze particle distribution characteristics on the basis of sample porosity analysis to fully grasp the response law of seepage characteristics to hydraulic conditions and the underlying mechanisms.
4.2. Analysis of the Response of Seepage Characteristics to Particle Size Distributions
The above analysis indicated that the problem of pore blockage by seepage was evident in both groups of samples. The problem was more prominent for Group B, indicating that the particle distribution characteristics had a significant impact on the seepage characteristics of the rock samples. On the premise of retaining the particle size characteristics, the rock samples after the experiment were broken and granulated, and the particle size distributions of the rock samples were obtained using the particle size distribution instrument. The sampling quality was controlled in the range of 0.4–0.6 g to meet the shading requirements of wet sampling. Three weighed samples and five tests were taken for a single sample, and the average value of 15 tests was taken as the particle size distribution parameter.
The particle size distributions and cumulative distributions of Groups A and B after the experiment are shown in Figures 12(a) and 12(b). The aperture of the dense mesh filter screen was 150 µm; therefore, particles smaller than 150 µm are more prone to be lost from the samples, thereby influencing the rock sample structure. Consequently, these particles are considered as the key research target of this study. 0–150 μm was defined as the target particle size that could be lost, and the target particle size distributions and cumulative curves were drawn as shown in Figures 12(c) and 12(d).
In order to provide a visually quantitative assessment of the particle distribution characteristics in the rock sample, based on the cumulative curves in Figures 12(a) and 12(b), the uneven coefficient value (Cu) was used to evaluate the particle distributions, and the Cu variation curves under different hydraulic conditions were plotted as shown in Figure 13(a)–13(c) . The change curves of the target particle proportions are shown in Figures 13(d)–13(f), and the fitted functions are shown in Table 7. The smaller the value of Cu is, the more concentrated the particle size and possibly the greater the porosity. The calculation formula for Cu is:
where dn, μm. The particle diameter corresponds to the cumulative proportion of particles reaching n%.
Based on the equation above, the Cu values for A0 and B0 are 6.37 and 5.34, respectively. This indicates that the B group samples have a higher overall porosity, as confirmed by the data in Figure 8. Correspondingly, the B group exhibits a more concentrated particle size distribution, primarily around 400 µm. This suggests a stable rock structure with a low proportion of fine sand, leading to reduced particle loss, a conclusion supported by the data presented in Table 4.
To explore the response relationship between hydraulic conditions, pore characteristics, particle distribution characteristics, and seepage characteristics, the relationship curves in Figures 7, 9,, 13 were paired to form coordinates and fitted with straight lines to obtain their correlation coefficients. Finally, the Cu—porosity response and the target particle proportion—permeability response was obtained with larger correlation coefficients. A plot of the relationship of Cu—porosity and target particle proportion—permeability was drawn, and its trend was fitted, as shown in Figures 14(a) and 14(b). The porosities of samples in Groups A and B showed negative responses to Cu values, which was consistent with the logic that a smaller Cu value led to a greater porosity. However, the slope of the fitting functions for Groups A and B was different, indicating that it was still important to further analyze the specific characteristics of the particle distributions of the two types of loose sandstone. The permeabilities of samples A and B both showed a negative response to the target particle proportion, which indicated that further increasing the number of 0–150 μm particles under a certain particle distribution would have a negative impact on the permeability of the rock sample. The reason was that after the particle distribution was determined, the skeletal structure of the rock sample was basically determined. Increasing the proportion of fine particles would increase the blockage of the seepage channel. According to the slope value and correlation coefficient of the fitted function, the sample structure of Group B was more susceptible to the influence of the proportion of fine particles, and the response was clearer.
The particle proportion of the target particle size in Group A was significantly higher than that in Group B (see Figures 12(c) and 12(d)), which was the main reason for the significant differences in seepage and pore distribution characteristics between the two groups of rock samples. A larger proportion of fine particles resulted in a smaller proportion of coarse particles in Group A as well as a more porous rock mass skeletal structure, which led to a greater proportion of large pores in Group A (see Figures 8(c) and 8(d)), and more developed seepage channels, in turn, led to a greater tendency for loss of Group A particles, resulting in an increase in porosity and permeability after the experiment. Group B had a smaller proportion of small particles and a larger proportion of medium particles, which led to a greater total porosity of Group B but also made Group B have a larger proportion of small holes, fewer medium and large holes, and a more solid pore structure (see Figure 8(b)). Therefore, Group B had a smaller effective porosity than Group A and was less permeable, and particles were not easily lost, which was more vulnerable to the effects of plugging by particles. Based on the analysis results in section 4.1, it was concluded that the seepage characteristics of Group A samples were more affected by pore characteristics, while the seepage characteristics of Group B samples were more affected by particle distribution characteristics.
The difference in particle distribution characteristics resulted in fundamental differences in the seepage characteristics of samples A and B, and the particle change characteristics during the seepage process determine the direction of changes in the seepage characteristics of the two groups. To reveal the response mechanism of hydraulic conditions, particle distribution characteristics, and permeability characteristics of Groups A and B, the absolute mass change curve is used to analyze the particle size change of rock samples, and the particle size distribution curve is mapped according to its mass to calculate the absolute mass change of particles of different sizes before and after the experiment to explore the particle change characteristics of Groups A and B samples before and after the seepage process. The calculation formula was
where Δmd corresponds to the absolute mass change of particles with the same diameter; M0 and M correspond to the total masses before and after the experiment, respectively; and pd0 and pd correspond to the percentages of particles with the same diameter, d μm, before and after the experiment, respectively.
The absolute mass changes and cumulative curves of the rock samples at different particle sizes after the experiment are shown in Figures 15(a)–15(f) and Figures 15(g)–15(l). It should be noted that in Figures 15(a)–15(f), there were some irregular fluctuations in the change in particle mass above 150 µm. The reason was that the segment value of the instrument increased exponentially when sampling. In combination with sampling and testing errors, there may have been a large particle size difference between particles in the same segment in the later stages of testing. The fluctuation pattern gradually converged in development to the left.
The maximum pore size of the Group A sample was approximately 13 µm (see Figures 8(a) and 8(b)), but the particles in Group A also decreased in the range of 13–20 μm (see Figures 15(a)–15(c)), while the loss of particles from Group B was mainly less than 6 µm (see Figures 15(d)–15(f)). This was because Group A had a significant expansion of seepage channels under pressure during the seepage process, with a larger proportion of large pores and a greater degree of compression shrinkage. As shown in Figure 6(b), the permeabilities of Group A samples significantly increased with increasing seepage flow rate, indicating the expansion of seepage channels.
In addition, in Figure 15, samples from Groups A and B showed mass improvements within a certain particle size range, as shown in the 20–40 μm range of Group A in Figures 15(g)–15(i) and the 6–40 μm range of Group B in Figures 15(j)–15(i). These regular upward curves indicated the splitting of large particles into small particles during the seepage process. This type of particle splitting could be divided into two situations, as shown in Figure 16. In case A, the seepage experiment, the inlet side produced axial pressure on the rock sample due to liquid retention, and some particles were crushed due to weak strength or excessive pressure. In case B, loose sandstone itself contained a certain amount of hard cohesive soil particles, which had weak permeability and did not crack easily. However, in environments with high water pressure (such as a high water pressure environment inside a rock sample caused by water injection pressure), the liquid invades the particle pores, causing complete saturation, and the particles decompose due to the disappearance of capillary action . In Figures 11(a)–11(f), obvious new pores were found in the 0–0.35 μm section (high-pressure water invaded the internal pores of cohesive soil particles and rock sample micropores, allowing nuclear magnetic instruments to detect this part of the pores). Analysis of Figures 15(g)–15(l) showed that the particle cracking behavior of the Group A sample was similar under different hydraulic conditions, indicating that the Group A sample had a greater particle strength and a smaller content of hard cohesive soil, and its responses to water pressure and seepage pressure were average. For Group B samples, as shown in Figure 15(j), as the flow rate and axial pressure increased, the formation of particles in the range of 6–40 μm significantly increased. Meanwhile, compared with Figures 15(j) and 15(k), as the injection pressure increased to 10 MPa, although the loss of 0–6 μm particles from Group B was more severe, more 6–150 μm particles were generated accordingly, indicating that the degree of cracking of Group B particles had a significant positive response to axial pressure and water pressure.
An overall analysis was conducted on the characteristics of particle mass changes in Groups A and B under different hydraulic conditions. Figures 15(g)–15(l) showed that on the particle diameter line of 150 µm, the cumulative changes in particle mass in Group A were all less than 0, and there was a significant negative response to hydraulic conditions such as injection pressure, flow rate, and flow volume. The cumulative changes in particle mass of Group B were greater than 0 near the particle diameter 150 µm line. The cumulative changes in particle mass of Group B did not have a significant response to flow rate and flow volume but had a positive response to injection pressure. At an injection pressure of 10 MPa, Group B generated more particles with diameters of 80–120 μm, achieving an earlier regression of the positive value of the cumulative mass change.
In terms of the changes in different particles as shown in Figures 15(a)–15(f), after the experiment, the loss of particles with particle diameters of 0–150 μm in Group A sample was significant, while the fluctuation of particles with particle diameters greater than 150 µm was small. This was because Group A had well-developed seepage channels, and low pressure was required for seepage, resulting in a large loss of fine particles and a small amount of generation. For Group B samples, although particles with diameters 0–6 μm were lost, more particles with diameters 6–150 μm were produced. In addition, the mass fluctuations of particles with diameters greater than 150 µm were significant, indicating a more obvious particle cracking phenomenon during the seepage process. This was because the seepage channels were not developed, and the pressure required was strong, resulting in a small loss of particles and formation of a large amount of particles.
Based on the above analysis, the positive response of hydraulic condition—permeability—effective porosity was effective for Group A rock samples, while Group B rock samples had a positive response only for permeability—effective porosity. The response mechanism of permeability to effective porosity was that effective seepage could only pass through pores with diameters greater than 0.1 µm. When more such pores were present, the seepage channels were more developed, and the seepage effect was better. The response of effective porosity to hydraulic conditions of Groups A and B differed because the increase in hydraulic conditions significantly increased the proportion of effective porosity in Group A, but the proportion of macropores in Group B limited change, and the change in effective porosity was not obvious. This is a significant reason depicted in Figure 6 for the notably higher and more variable permeability in the samples of Group A compared with Group B.
Both Groups A and B had a negative response of permeability to the target particle proportion, and the response mechanism was that the increase in fine particles led to more frequent pore plugging and a decrease in permeability. Due to the presence of more developed seepage channels in Group A samples (which made particles more prone to be lost), there was a significant negative response between the hydraulic conditions and the proportion of target particles of Group A. The response of the proportion of target particles to the hydraulic conditions in Group B was not significant due to the underdeveloped seepage channels. This led to less loss of fine particles and higher pressure required for seepage. Thus, large particles were more prone to cracking, resulting in a less significant change in the proportion of target particles, and this is the reason behind the relatively low and less pronounced variability in permeability observed in the samples of Group B in Figure 6. In addition, although the Group A samples had a negative response to the proportion of effective target particles, the correlation coefficient was only 0.48 (see Figure 14(b)) since the effective porosity of Group A was greater, and fine particles could be discharged through the expanded pore channels during the seepage process. The Group B sample had a smaller effective porosity, which made it difficult to discharge particles, led to plugging problems, and the required injection pressure was high, making large particles crack into fine particles under axial pressure and further reducing permeability. Therefore, the permeability of Group B exhibited a strong negative response to the proportion of effective target particles, with a large inclination of the fitted straight line and a correlation coefficient of 0.73 (see Figure 14(b)).
Based on the above analysis, a comprehensive explanation of the underlying principles can be provided for the response of seepage characteristics to hydraulic conditions. For Group A samples, increasing the injection pressure promoted the expansion of the pore channels, and increasing the flow rate and flow volume promoted particle discharge, thereby increasing the number of seepage channels in Group A and its permeability. For Group B samples, increasing the injection pressure promoted the cracking of cohesive particles, increasing the flow rate increased the axial pressure and caused particles to break, and increasing the flow volume made the particle blockage problem more severe, thereby reducing the porosity of Group B, exacerbating the particle blockage problem, and decreasing its permeability.
In this study, an optimized method was used to prepare samples of loose sandstone and conduct seepage experiments under different injection pressures, flow rates, and flow volumes. The parameters of the permeabilities, porosities, and particle size distributions of rock samples were obtained and studied, and the influences of different hydraulic conditions on the seepage characteristics of loose sandstone were explored.
Based on the responses of seepage characteristics to pore distribution characteristics and particle distribution characteristics, the following conclusions were drawn:
Due to the differences in pore and particle characteristics, different rock samples had different responses to hydraulic condition parameters. For example, the permeability of Group A responded positively to various hydraulic condition parameters, while Group B responded negatively.
The permeability of the rock sample had a positive response to the proportion of pores with diameters greater than 0.1 µm and a negative response to the proportion of particles with diameters less than 150 µm. The principle was that pores with diameters greater than 0.1 µm could form effective seepage channels, and the increase in such pores was more conducive to seepage. Under the determined pore structure, an increase in the proportion of particles with diameters of 150 µm led to more frequent blockage of the pore channels, which was not conducive to seepage.
The main roles played by various hydraulic conditions in different rock samples were different: increasing injection pressure increased permeability by expanding pores in Group A and reduced permeability by cracking cohesive particles to block pores in Group B. The increasing flow rate increased the permeability by expelling fine particles and expanding channels in Group A and reduced the permeability by fracturing particles through axial pressure to block channels in Group B. The increasing flow volume increased permeability by expelling fine particles to expand the channels in Group A and reduced the permeability by driving particles to block the channels in Group B.
This study tested the porosity of rock samples before and after the seepage experiment, and the changes in porosity of rock samples during the seepage experiment effectively reflected the impact of seepage on the process of rock samples, which has particular reference significance for engineering sites. Monitoring particle leakage during the experimental process was also an effective means to dynamically grasp the seepage state. Improving the experimental content and equipment based on the above experimental requirements is a feasible direction for future studies.
Data availability statement
All data in this study have been included in the paper and presented in the form of graphs and tables. All retained samples, unprocessed original experimental data, and supporting photos during the experimental process can be obtained by contacting the corresponding author.
Conflicts of Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
The study is supported by the National Natural Science Foundation of China (Key Program). Subject No: 52034001 and the Independent Scientific Research Project of China Nuclear Inner Mongolia Mining Co., Ltd. Subject No: 4Y00-FW-GKJT-23-0507.