The shale specimens were taken from an outcrop of the lower Silurian Longmaxi formation, located in the Changning region of the Sichuan Basin. The total organic carbon (TOC) content was 4.18% and the vitrinite reflectance (Ro) was 2.36%, the uniaxial tensile strength was 9.2 MPa, the elastic modulus was 16.0 GPa, and Poisson’s ratio was 0.31. The shale was mainly composed of quartz and illite, with the addition of calcite, feldspar, dolomite, chlorite, pyrite, and other minerals. The average value of the total amounts of brittle minerals was about 70.54%, which is similar to typical shale gas basins in North America.

To reduce the dispersion of the samples of the test results, they were drilled from the same piece of shale. The drilling direction was perpendicular to the bedding plane, and samples with smooth ends and no surface cracks were selected for processing into cylinder specimen (100 mm in diameter and 200 mm in length, axial and radial pressure were applied respectively), as shown in Figure 1. Before the fracturing experiment, an axial central borehole with an 8-mm diameter was drilled to a depth of 140 mm. An injection pipe was inserted into the borehole to ensure the tightness of the system during the experiment.

The deviation of height and diameter in samples was controlled at ±0.3 mm, and the deviation of flatness was controlled at ±0.02 mm in the process. The shale samples were tested by a sonic wave before the fracturing experiment, the samples with relatively closer wave velocity were selected. The basic physical parameters of experimental shale samples are shown in Table 1.

The supercritical CO₂ fracturing shale and increased permeability experimental device were independently developed by Chongqing University (Figure 2A). The experimental device was specifically designed for supercritical CO₂ fracturing and was composed of six subsystems: an air supply system, CO₂ injection system, triaxial loading and control system, temperature control system (oil bath), acoustic emission monitoring system, and a data collecting system.

Figure 2 shows the test system. This test was developed to study the fracture law of shale caused by supercritical CO₂ and explore the influence law of supercritical CO₂ on shale initial cracking pressure, furthering research on self-developing supercritical CO₂ cracking using an increasing permeability test device (Zhou et al., 2016), for shale gas reservoirs. The system is composed of six parts: Gas source supply system, CO₂ boosting control system, triaxial servo loading system, temperature control system, acoustic emission monitoring system, and data acquisition system. It can simulate an experiment of fracture propagation laws in the process of supercritical CO₂ fracturing shale under the coupling of multiple fields and collect the stress-strain relationship and acoustic emission signals in real time during the process of the shale fracture.

1) Figure 2A is a High precision plunger pump, which is mainly used to provide high-pressure CO₂ fluid. The plunger pump can provide continuous, stable and pulse free flow, and can carry out constant pressure and constant flow operation. The maximum pressure is 50 MPa and the maximum capacity is 266 ml. Supercritical CO₂ is a critical state in which the temperature and pressure of CO₂ meet the requirements of 31.04°C and 7.38 mpa respectively. The plunger pump can fully meet the pressure control range and pressure mode required by the experimental process. The plunger pump adopts a constant pressure mode to load.

To reach the supercritical state of CO₂, the experiment temperature was kept constant at 308 K. To simulate the influence of different stress conditions on failure modes and initiation pressures of crack, horizontal stress remained constant, the initial stress ratio (the ratio of horizontal stress and vertical stress) was changed by changing vertical stress during the experiment. The confining pressure (horizontal stress) was constant at 10 MPa, and the axial (vertical stress) was set at 10, 20, 30, and 40 MPa, respectively. In the process of supercritical CO₂ fracturing, acoustic emission signals were monitored simultaneously, and the experimental scheme is shown in Table 2.

Fracturing and re-fracturing experiments were carried out to simulate the fracturing and re-fracturing process. Therefore, supercritical CO₂ injection was divided into two stages, the injection rate of CO₂ fluid was constant at 20 ml. min^–−1 at each stage. First, CO₂ was injected continuously until the pressure was no longer elevated, and after the completion of fracturing, when the pressure inside the specimen decreased, the pump pressure stopped immediately. Second, when the internal pressure of the crack achieved balance, CO₂ fluid continued to be injected.

Before and after the fracturing test, the specimens were photographed using a high definition digital camera and CT scanning to obtain surface and internal crack characteristics of the samples before and after fracturing. The CT scans of each specimen, along with samples in a vertical direction (200 mm) of 0.75 mm thickness to form a layer.

The acoustic emission count-time curve and pump pressure-time curve during shale fracturing experiments are shown in Figure 3. The overall trend of the pump pressure curve and acoustic emission curve of the four specimens was similar. The process can be divided into three stages:

The initial pressure of failure of shale can be defined as peak pressure at this stage. Carbon dioxide fluid, an effective fracturing fluid, first gathered in an artificial cavity pocket with the gradual injection of fluid, the pressure of carbon dioxide in the fracturing chamber grew sharply, before the carbon dioxide fluid started to squeeze into the pore and fissure system around the wellbore. When the pressure reached a certain value, the pressure of fracturing fluid growth rate became relatively slow due to the diffusion of pressure in cracks and pores. The small amount of acoustic emission signals produced in this phase indicate that the native cracks of shale began to open gradually under the effect of carbon dioxide fluid along with the formation of micro cracks, but shale had not yet formed macroscopic cracks through this process. Carbon dioxide fluid pressure still rose and its increase rate began to increase faster after the pore and fissure space was full of carbon dioxide with a continuous injection of fluid until the shale failed, and pump pressure vs. time curve reached the first peak value. A large number of acoustic emission counts were found during the peak value of pressure, which indicates that fractures were generated and extended dramatically, and transfixion cracks were formed.

The pressure of fracturing fluid decreased rapidly after the closure of the fracturing pump, as crack coalescence formed in the shale specimens and the seepage channel had been generated, which resulted in the faster spread of the fluid inside the crack and pressure dropping quickly in the fracture until finally, the pressure remained at a certain value. The acoustic emission signal was weak and the counts were relatively small (or even none), which showed that hardly any new cracks were generated at this stage and mainly show the closure of the generated fracture.

The fracturing fluid of supercritical carbon dioxide was continually injected into the opened cracks by opening the plunger pump after the pressure of the fracturing chamber reached the balance value. The pressure of the fracture increased dramatically at the same time, the fractured cracks reopened after the fracture stop closure, and crack pressure reached the second peak value, which reached Pr. The AE counts of repeated fracturing reduced significantly compared to the initial fracturing, but it still had a strong acoustic emission signal, which indicates that the original cracks continue to propagate and extend at the second fracturing process and many secondary cracks were generated.

The results of supercritical CO₂ fluid fracturing shale are shown in Table 3. In combination with Table 1 and Table 2, it can be concluded that the initiation pressure of shale is always higher than horizontal stress; however, the stress of repeated fracturing is close to horizontal stress (10 MPa). With the decrease of λ, the initiation pressure of shale, the stress of repeated fracturing, and the amount of fracturing fluid decreased. Among them, the initiation pressure of Specimen 1 was highest (13.7 MPa), and the stress of repeated fracturing on Specimen 4 was the smallest (9.4 MPa).

According to first strength theory, during the supercritical CO₂ fluid fracturing process when the fluid pressure exceeds the stress required for the cracking of the rock at the hole wall, cracks are generated at the wall of the hole, and the occurrence of cracks is related to the type of fluid, the size of ground stress and the physical and mechanical properties of the rock. In this paper, with a constant confining pressure and the greater the vertical stress, the smaller the initial stress ratio and the lower the stress of the rock cracking at the hole wall, which results in a decrease in the initiation pressure with the decrease of the initial stress ratio (Zhao et al., 2015). The relationship between the initiation pressure and the initial stress ratio of the shale specimens is shown in Figure 4, which shows that the initiation pressure increases with the increase of the initial stress ratio and that there is a linear relationship between the initiation pressure and the initial stress ratio, which can be fitted to y = 3.7x + 10 (R ² = 0.998).

When the shale samples initial pressure relief, the samples will reach the second pressure peak, continuing to inject CO₂. The reopening pressure can be obtained. As shown in Table 3. The reopening pressure of the cracks are lower than the initial pressure of the crack, which is due to damage to the shale specimen and forms a certain crack network during the initial fracturing as the strength of the shale itself was reduced. In the second fracturing, the crack continued to propagate based on the existing crack. This can also be verified in the CT scanning image where the initial stress ratio was different, as the number of fractures formed and the crack volume space were different in terms of fracturing. The different crack volumes led to a difference in the amount of CO₂ fluid entering the crack pore space, so different initial stress ratios in addition to the impact of initiation pressure also lead to a difference in the amount of fracturing fluid.

From the obtained CT scanning images of the shale samples, each specimen (S₁–S₄) was selected from two groups of CT scanning images in the vertical and parallel direction of the axial section before and after fracturing, with a total of 16 images shown in Figure 5.

The CT scanning image of the initial stress ratio at 1 is shown in Figure 5A. Some new cracks were created at the level of the vertical wellbore axis, while transverse cracks were found at the level of the parallel wellbore axis, which suggests that cracks mainly extend from the center of the wellbore to the horizontal direction of the wellbore, at the same vertical stress and horizontal stress (Figure 5A).

In Figures 5B–D, there were no obvious cracks in the shale specimens and the shales kept integrity before fracturing. In comparison to Figure 5A, some cracks were detected in both the vertical and parallel wellbore sections after fracturing in (Figures 5B–D), and cracks and fractures were similar to the end face of the specimens, which indicates that cracks spread from top and bottom, and the cracks in the specimens were connected. Furthermore, it was noted that the cracks mainly occurred along the direction of maximum principal stress, thus, cracks in the specimens after fracturing confirmed that supercritical CO₂ fracturing can form a complex network of cracks in shale specimens.

To better understand the process of supercritical CO₂ fracturing, the mechanism of the fracturing is described. From Figures 5B–D, we can see that cracks were first produced at the wall of the hole, and tensile failure occurred along the direction of maximum principal stress and accompanied by shear failure. The mechanical failure of the fracturing may relate to the developed stratification of shale and the characteristics of supercritical CO₂ with low viscosity and high diffusivity. Compared to water, the low viscosity and high diffusion of supercritical CO₂ make it easier to enter the micropore of the shale and make the cracks extend and turn along the weak bedding surface. These research findings are consistent with those of Chen L. Q. et al. (2015), who obersved that, due to the special properties of supercritical CO₂ fracturing, it forms a complex fracturing networks more easily. In the process of crack propagation—due to the increase of free space—the phase change of CO₂ from supercritical states to gas states was likely to have occurred. In addition, the compressibility of gaseous CO₂ is much larger than supercritical CO_2, so the phase change will produce gas explosions steeply, and is favorable for the production of complex fracture networks in shale specimens.

As shown in Figures 5B–D, it is interesting to note that new cracks at the same height of the shale specimens varied from different initial stress ratios. Specimen 2 had the largest number of cracks and the most complex fracture networks, while there were fewer cracks and fractures in Specimen 4, which had monotonous fracture networks. Under different stress state conditions, the crack initiation pressure and propagation direction varies with stress states and eventually forms different crack distributions and quantities, which show that the complexity of fracturing networks is affected by the initial stress ratio. In conclusion, these results show that cracks extend along the direction of maximum principal stress, and the cracks in shale form the main crack parallel to the direction of maximum principal stress after fracturing by supercritical CO₂ fluid.