Study on the creep permeability of mining-cracked N2 laterite as the key aquifuge for preserving water resources in Northwestern China

This research focused on the impact of mining on the permeability of key aquifuge (N2 laterite) that is widespread in the arid and semi-arid areas of northwestern China and is critical for preserving water resources. The impact of mining stress recovery on the permeability of cracked N2 laterite was assessed for parts of northwestern China that included the Jingle laterite and Baode laterite. The mineral compositions and swelling properties of the laterite at both locations were examined, and analytical results showed that the laterite contained abundant clay minerals. The Baode laterite exhibited higher expansibility than Jingle laterite. The triaxial creep permeability performance of laterite specimens with a prefabricated crack width of 1.0, 1.5, and 2.5 mm were tested. The results indicated that strain of cracked laterite all exhibited transient creep following each level of loading, and then unstable creep and stable creep. With the increase of loading, the transient creep deformation corresponding to each level of loading decreased, the unstable creep deformation produced by identical loading gradually and incrementally increased. The nonlinear power function equation was selected to fit creep grading curves which have high precision. The cracks within the laterite gradually closed with the stress recovery, and permeability gradually recovered. During the stress recovery, the narrower cracks exhibited a smaller change in permeability. However, for narrow cracks in mining soil, permeability recovered after mining stress when permeability was closer to initial permeability, and the Baode laterite showed greater recovery than that of the Jingle laterite.


Introduction
Coal occupies a dominant position in China's energy structure and is an important primary energy source for China. Coal resources account for about 70% of the nation's energy structure based on China's Development and Reform Commission (2012) (http://www.nea.gov.cn/), and in long-term forecasts, a Chinese coal-based energy structure will not be supplanted (Qiao et al. 2014). Because of the depletion of coal resources in the eastern mining area, coal resource development focus in China has quickly shifted to western China where the environment is more vulnerable. In 2013, coal production from the five northwestern provinces reached 17.9 9 10 8 t, accounting for 48% of national output. Future coal production in the western area will account for more than 70% of national coal output. The main coal seams in the western Jurassic coal fields are juxtaposed with a loose aquifer in the upper layer, clay aquifuge layer, bedrock, and a variety of other geologic characteristics, which typically results in five categories of geotechnical engineering conditions in area (Li et al. 2000), of which sand was the most common type of stratum. The strata featured a modern aeolian and Salawusu group sand layers with an underlying yellow clay and Neogene Pliocene laterite (N 2 laterite), and weathered bedrock and bedrock layers. The N 2 laterite is the regionally distributed strata that overlies the Jurassic coal seams in western China (Fig. 1) and is the primary aquifuge. The Salawusu group and the bedrock contained water, and bedrock group was a coal-containing strata that included mudstone, siltstone, and sandstone (Fig. 2). Since northwestern China is an arid and semi-arid region, and water is in shortage scarce resource, the ecological geotechnical environment is vulnerable. When the regional topography and ground conditions are suitable, shallow groundwater supplies are very valuable water resources, and have great significance for maintaining the ecological environment of the overall region (Xu et al. 2004;Li et al. 2017;Zhang et al. 2017). These are represented by the upper Pleistocene Salawusu group (Q 3s ) sand aquifer which has a large distribution as the Maowusu desert beach, runoff water formed by rainfall at loess ravine region, lakes and reservoir water (Fig. 1). Coal mining in recent decades has resulted in a series of environmental issues in western China (Han et al. 1992;Fan and Jiang 2004;Yang et al. 2006;Miao 2008;Wang et al. 2008;Song 2009;Ma et al. 2010;Wang et al. 2011;Huang et al. 2012). The impact of coal mining on groundwater resources can be divided into three types (Wang et al. 2014;Qiao et al. 2017) including (1) not influenced zone: coal seams are buried deeply; water flowing fractured zone does not develop to clay aquifuge layer (N 2 laterite layer) after mining, and coal mining almost has no impact on sand aquifer at the region with a thick development of clay aquifuge layer, and basically ground surface ecological environment is not affected; (2) zone of serious water leakage: coal seams are buried shallowly; water flowing fractured zone develops into sand aquifer and above, which results in quick loss of sand aquifer, water level rapidly decreases, and ground surface ecological environment severely deteriorates in a short time; (3) zone of general water leakage: after coal mining, water flowing fractured zone develops into clay aquifuge layer, where the clay aquifuge layer could lose a certain impermeability because of generated cracks, and later recovery of stress that leads to closure of the cracks in the clay and recovery of permeability to some extent. Water levels in the sand aquifer exhibited a significant decrease at the beginning of coal mining, and the ground surface ecological environment deteriorated. However, recovery of permeability from closure of the clay cracks allowed for water levels in sand aquifer to also recover, and ground surface ecological environment to be revitalized. Figure 3 shows a cross-sectional schematic for a typical mining operation in western China. The N 2 laterite over Jurassic coal layers distributed regionally in western China is of special natures. Most of current research on N 2 laterite is about its geologic components and engineering geology properties, and the research on the creep behavior and the permeability of N 2 laterite is hardly seen. At present, investigators have conducted numerous studies on common soil rheology and creep problems including experiments, soil creep modeling, and the numerical application of soil creep strain-stress model (Singh and Mitchell 1968;Garlanger 1972;Ladd 1977;Butterfield 1979;Mesri et al. 1981;Kabbaj et al. 1988;Zhan et al. 1993;Nash and Ryde 2001;Lan 2002;Chen et al. 2005Chen et al. , 2006Yu et al. 2016). Experimental studies on mining stress recovery and creep permeability of special regional N 2 laterite are ongoing.
In this paper, stress recovery refers to the process of the cover stress re-establishment in the mined soil mass, which is gradually recovered to the self-weight stress of the overlying strata at the end of the mining process. This paper utilized double linkage triaxial testing machine for hydro-mechanical coupling in soft rock to conduct tri-axial creep permeability change testing on prefabricated N 2 laterite specimens (taken from Jingle laterite and Baode laterite from western Jingle County and Shenmu County, China, respectively). The research focused on the creep deformation and permeability change processes of N 2 laterite during mining stress recovery. The research outcomes can provide fundamental theory to predict the selfrecovery of general water leakage zone which may occur within sand aquifers. In addition, the resulting data may have significance for interpreting the process mechanisms of inflow reduction and leakage change after mining inrush occurs in western China.
2 Mineral composition and expansibility analysis of clay specimen

Clay mineral composition
There are a lot of factors influencing clay expansibility such as clay content, clay mineral type, organic matter content, and exchangeable citation group. A D/Max-3B type of X-ray diffractometer was used to quantitatively analyze the clay mineral compositions of laterite specimens taken from two locations, and related testing results were shown in Fig. 4 and Table 1.
X-ray diffraction testing results showed that Jingle and Baode laterite contained abundant montmorillonite, kaolinite, illite, and clay minerals such as illite-montmorillonite mixture and chlorite. Montmorillonite comprised 23 and 21%, respectively, of the mixtures and was most influential on the permeability of clay. The mineral crystal structure of montmorillonite determined its high expansibility after water absorption. The water expansion characteristic of montmorillonite reduced the crack of clay, and accordingly resulted in a smaller permeability. In situ permeability test showed that cracks in Jingle laterite and Baode laterite did not develop under natural conditions, and after contact with water the montmorillonite minerals expanded to fill the internal cracks of soil and resulted in an aquifuge effect. However, after large-scale coal mining, a variety of cracks in soil would develop; corrosion would readily occur because of expansion of the clay layers, followed by a corresponding increase in permeability.
Mineral identification and molecular formula in Fig. 4 and Table 1 are as follows.

Expansibility of clay
Expansibility is one of the important physical properties of clay. On the one hand, the porosity of clay deteriorates after clay absorbs water and expands making it relatively impermeable. On the other hand, the change of clay strainstress field due to coal mining results in the partial rupture of clay, and the expansibility of clay regenerates the cracks in clay that can be accessible to water. The expansibility experiment of aquifuge clay is significant for understanding clay aquifuge layers in mining operations.
The indexes to characterize the expansibility of clay mainly include limited expansion (confined expansion), free expansion, and expansion force. The magnitude of limited expansion is mainly associated with compactness, natural moisture content and structural constitution of soil, but free expansion is the expansibility of clay without structural impact and represents the expansive tendency of clay; therefore, these two indexes were selected to evaluate the expansibility of N 2 laterite.
In this study, based on the ''Standard for soil test method'' (GB/T50123-1999), the loose, dry soil samples were prepared. The ratio of the volume increment to the original volume is calculated when the soil sample is stabilized in pure water. The free expansion ratio is based upon the formula as follows: where, d ef is the free expansion ratio (%); V we is the volume of the specimen in water; V 0 is the initial sample volume. The confined expansion ratio test of N 2 laterite was based on the ''Standard for soil test method'' (GB/T50123-1999), Firstly, the prepared standard samples (diameter to height ratio of 1: 2) were dried and were placed into plastic bag. The gas in the plastic bag was discharged through the air pump so that the plastic bag was in close contact with the sample. And then, the sample volume was measured by drainage method. The specimens were wrapped with plastic film, and the amount of water was added according to the proportion. The specimens were placed in the steel mould with side limit. Finally, the axial deformation after confined expansion was tested by dial indicator. The confining expansion ratio was calculated by using Eq. (2): where, d e is the confining expansion ratio (%); z is the axial deformation after confined expansion; h 0 is the initial sample height. In this study, confined expansion ratio and free expansion ratio were tested and the results were shown in Table 2. A difference in expansibility was documented between the Jingle laterite and the Baode laterite. Because of higher compactness and more expansive minerals, the Baode laterite exhibited better expansibility under confined conditions and better expansive tendency than those of the Jingle laterite.   and 70-85 m for Baode laterite. The basic mechanical properties of specimens were provided in Table 3.
In this experiment, specimens of cracked Jingle laterite and Baode laterite were used to simulate the cracked key clay aquifuge layer generated in the process of coal mining. The undisturbed soil samples were cut into the dimensions of u38 mm 9 76 mm. The initial cracks of the specimens were artificial, and a crack was located at the middle of the specimens. A total of six groups of cracked specimens were prepared with three groups of Jingle laterite and Baode laterite, respectively, having crack widths of 1.0, 1.5, or 2.5 mm, and crack lengths of 28 mm (Fig. 5).

Experimental equipment
Experimental instruments in this research included a double linkage tri-axial testing machine for hydro-mechanical coupling in soft rock that was jointly developed by the Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, and the Changchun Chaoyang Experimental Factory. The rheometer is especially suitable to test the properties of soft rock and hard soil under various stress conditions. In addition to common functionalities of the triaxial experimental machine, such as uniaxial and tri-axial compression experiment, the machine can also perform mechanical experiments of identical axial compression, different confining pressure, identical hydraulic pressure simultaneously for two specimens, confining pressure control, porous water pressure control, measure pore water pressure, and check the saturation degree of specimens. Figure 6 shows the system theory for a dual linkage soft rock seepage-stress coupling triaxial rheometer. The instrument mainly consists of four components to achieve above functionalities including an axial loading part (primary machine), confining loading part (pressure chamber and loading unit), pore water loading part, and control part (computer and controller). A pad was used to uniformly transfer force between the upper and lower tri-axial pressure chambers.

Experimental methodology and procedure
In order to more closely reflect the actual situation during experiments, the experimental soil specimens were initially saturated. Three-isometric hydrostatic pressures (r 1 = r 2 = r 3 = 0.2 MPa) (in situ horizontal effective stress) were applied, in the process of saturation, at the two ends of specimens to provide backpressure for saturation and to avoid disturbance against specimens. At first, isotropic pressures were increased to 0.2 MPa with a loading rate of 0.139 kPa/min, then a backpressure of 0.1 MPa was applied at the top and bottom of specimens, which was maintained for 24 h. Saturation status was judged by checking the variation of pore pressure coefficient B, and when Skempton B coefficient reached 0.85, the specimens were regarded as saturated. The burial depth of laterite ranges from 0 to 225 m, so the maximum of loads is about 4.5 MPa. In this study, experimental soil specimens were loaded in various loads of 1.5 MPa, 2.5 MPa, 3.5 MPa, and 4.5 MPa as shown in Fig. 7. Each loading was maintained for about one day. After deformation of specimens stabilized, a steady state method was used to test their permeability coefficients under the condition of keeping confining pressure and axial pressure constant (the steady state method referred to application of a hydrostatic pressure of 0.2 MPa on the top of specimen. The amount of water effluent at the bottom of the specimen was measured using a high precision soap-bubble flowmeter, and the permeability coefficient was calculated after the effluent stabilized before applying the next loading. Stress and deformation were monitored during implementation of experiments. This meant that mining cracked N 2 laterite had a larger deformation at the beginning of stress recovery, i.e. closing of crack was more apparent; but the later deformation was smaller and smaller until it became stable. In addition, under the application of identical loads, a larger crack width occurred for identical soil specimens and there was greater transient creep deformation. This indicated that a larger mining crack width during soil stress recovery. Under application of identical loading and with the same crack width, the Baode laterite exhibited greater deformation than Jingle laterite. This was mainly because of the differences in mechanical properties. On the one hand, the Baode laterite contained abundant clay minerals. On the other hand, the void ratio of Baode laterite was larger than that of Jingle laterite. Therefore, the Baode laterite was more condensable than Jingle laterite under equivalent conditions. The creep testing curves under different stress levels show that the creep process of cracked N 2 laterite presents nonlinear characteristics. In this study, the nonlinear power function creep model was used to describe the creep behavior of cracked N 2 laterite in three axis creep test. The model can be expressed by the given formula equation.
where, e is creep strain; A, m and n are coefficients to be determined; r is loading confining pressure; and t is reference time.
Polynomial regression analysis method was used to fit empirical equation, which got corresponding parameters of each specimen under different stress levels (Table 4). From Table 4, it was seen that the fitting curves agreed well with experimental results when the applied loads are at high levels (2.5 MPa, 3.5 MPa, 4.5 MPa). However, the fitting curve showed large difference with experimental results. The most important reason is that the closure degree of the cracks within the laterite is bigger when the applied load is 1.5 MPa. In general, the correlation coefficients (R 2 ) of most of the fitting results were greater than 0.9, which showed that the fitting precision is very high. The change of creep deformation of cracked N 2 laterite with time can be well reflected by using this nonlinear power function  Fig. 9. The creep test results and fitting curves of model for Jingle and Baode laterite specimens were shown in Fig. 10. The relationship between the permeability and confining pressure of cracked N 2 laterites at two locations were presented in Fig. 11. Figure 11 showed that there was an exponential relationship between permeability and confining pressure of cracked N 2 laterites that can be approximately described by Eq. (4).
where, K is permeability coefficient; K 0 and k are coefficients to be determined; and r is loading confining pressure. Table 5 showed that the corresponding values of K 0 and k for different conditions. The experimental and fitting results showed that the permeability coefficients of cracked Jingle and Baode laterites gradually decreased with the increase of loading, and the decrease in magnitude of the permeability coefficient was gradual under the application of identical loading. This indicated that the mining cracked laterite had a higher variation of permeability at the beginning of stress recovery, i.e. the extent of crack closure was more evident; but the variation of permeability gradually stabilized because deformation was minimized. In addition, in similar soil specimens, larger crack widths corresponded to larger variations of permeability during stress recovery. When stress completely recovered, the soil specimens with larger crack widths exhibited larger permeability coefficients. This indicated that smaller mining soil crack widths resulted in smaller permeability coefficients after stress recovery, and larger crack widths resulted in more severe soil damage and limited permeability recovery. Under the application of identical loading for soil specimens with similar crack widths, the permeability coefficient of Baode laterite was about less than one-third of Jingle laterite. The permeability of mining cracked Baode laterite was slightly higher than that of Jingle laterite after mining recovery. The reason for this is that engineering geological properties of the laterite at both locations are different. The Baode laterite contained richer clay minerals and exhibited higher expansibility than Jingle laterite.

Conclusion
(1) Through analyzing the mineral compositions and expansibility of Jingle laterite and Baode laterite, this research shows that both Jingle laterite and Baode laterite contained abundant montmorillonite, kaolinite, and illite, as well as some clay materials such illite-montmorillonite and chlorite. Montmorillonite comprised 23 and 21%, and was the most abundant mineral. Jingle laterite and Baode laterite exhibited differences in expansibility. Baode laterite had a higher expansibility than Jingle laterite under confining conditions because it was more compact and was comprised of more expansive minerals. (2) Specimens for Jingle laterite and Baode laterite exhibited transient creep following each level of loading, and then unstable creep and stable creep. With the increase of loading, the transient creep deformation corresponding to each level of loading decreased, the unstable creep deformation produced by identical loading gradually and incrementally increased. The nonlinear power function equation was selected to fit creep grading curves which have high precision. (3) Mining cracks in Jingle laterite and Baode laterite were gradually closed and permeability recovered following the recovery of mining stress. In the process of stress recovery, smaller crack widths and lower variations in permeability were observed; however, permeability was closer to the initial permeability in specimens that had smaller soil cracks after recovery of mining stress. Baode laterite with mining cracks exhibited higher recovery ability than that of Jingle laterite.