Propagation Characteristics of Supercritical Carbon Dioxide Induced Fractures under True Tri-Axial Stresses

Supercritical carbon dioxide (SC-CO2) fracturing is a non-aqueous fracturing technology, which has attracted considerable attention on exploiting shale gas. In this study, shale specimens and artificial sandstone specimens were used to conduct SC-CO2 fracturing and water fracturing experiments to investigate the characteristics of SC-CO2 induced fractures. An acoustic emission (AE) monitoring device was employed to monitor the AE energy release rate during the experiment. The experiment results indicate that the breakdown pressure of SC-CO2 fracturing is lower than that of water fracturing under the same conditions, and the AE energy release rate of SC-CO2 fracturing is 1–2 orders of magnitude higher than that of water fracturing. In artificial sandstone, which is homogeneous, the main fracture mainly propagates along the directions perpendicular to the minimum principal stress, no matter if using SC-CO2 or water as the fracturing fluid, but in shale with weak structural planes, the propagation direction of the fracture is controlled by the combined effect of a weak structural plane and in-situ stress.


Introduction
Shale gas is one of the unconventional resources which is stored in organic matter-rich shale formations. Technically recoverable shale gas reserves in China, estimated by the Energy Information Administration (EIA), are 36.1 × 10 12 m 3 at standard temperature and mainly distributed in the Sichuan Basin, Tarim Basin, and Ordos Basin [1,2]. Due to the very small porosity (2% or less) and ultra-low permeability (0.1 to 0.0001 mD or even less) of shale, it is difficult to exploit shale gas by conventional oil and gas exploitation methods. However, in the past 20 years, the breakthrough of horizontal drilling and hydraulic fracturing technology has triggered the "Shale Oil and Gas Revolution" all over the world [3]. Presently, hydraulic fracturing with multiple horizontal wells has been widely used in the exploitation of shale gas [4][5][6]. However, with the widespread use of hydraulic fracturing technology, its drawbacks also emerge. The hydraulic fracturing needs to consume a large amount of water and a typical shale gas well injects 2-4 million gallons of water into a deep shale reservoir, which is not conducive to the exploitation of shale gas reservoirs located in water shortage areas [7,8]. The flow-back water contaminated with secondary substances, which are added to the water to enhance fracture initiation and propagation, requires disposal [3,9,10]. The injection of water will alter the distribution of

Experimental Apparatuses
The fracturing experiments were conducted by a true tri-axial fracturing system which could perform the SC-CO 2 fracturing experiment and the water fracturing experiment on a cubic specimen of 300 × 300 × 300 mm 3 (see Figure 1). The system consists of SC-CO 2 generating device, a water injection device, a true tri-axial hydraulic loading device, and an AE detecting device. stored in a buffer tank. Thirdly, the liquid CO2 stored in the buffer tank is heated by a water bath heater to 35-100 °C, and the CO2 changes to a supercritical state when its temperature and pressure exceed 31.1 ℃ and 7.38 MPa, respectively. The water injection device can produce high pressure water with constant pressure ranging from 0-60 MPa or a constant injection rate ranging from 0.01-60 mL/min for water fracturing.
The true tri-axial hydraulic loading device can provide confining pressure to simulate in-situ stress for the experiment. The confining pressure can be applied along X-, Y-, and Z-directions independently by three hydraulic pumps, which are controlled by a servo control cabinet. The pressure is transmitted to the specimen by a loading board in each direction. The maximum pressure applied in each direction can reach up to 30 MPa.
The AE detecting device is composed of an AE meter, a differential preamplifier, and AE probes. The size of the AE probe is Φ 22 × 36.8 mm, and the detection frequency range is 15-70 KHz, with a resonant frequency of 40 KHz. Eight AE probes were place on four surfaces of the specimen, except the upper and lower surfaces, to monitor acoustic emission features in the experiment (see Figure 2a).

Specimen Preparation
Eight specimens 300 × 300 × 300 mm 3 in size were used to conduct the experiment. Three of these specimens were shales and the rest were artificial sandstones. The shales were obtained from outcrops of the Changning Block in Sichuan Province, China, which belonged to black shale of a The SC-CO 2 generating device can produce SC-CO 2 with temperatures from 35-100 • C and pressure from 10-80 MPa, and the forming process of SC-CO 2 is as follows: Firstly, the refrigeration converts high pressure CO 2 gas from CO 2 cylinder to liquid CO 2 stored in a storage tank. Second, the triple plunger pump pressurizes the liquid CO 2 to 10-80 MPa and the high pressure liquid CO 2 is stored in a buffer tank. Thirdly, the liquid CO 2 stored in the buffer tank is heated by a water bath heater to 35-100 • C, and the CO 2 changes to a supercritical state when its temperature and pressure exceed 31.1°C and 7.38 MPa, respectively. The water injection device can produce high pressure water with constant pressure ranging from 0-60 MPa or a constant injection rate ranging from 0.01-60 mL/min for water fracturing.
The true tri-axial hydraulic loading device can provide confining pressure to simulate in-situ stress for the experiment. The confining pressure can be applied along X-, Y-, and Z-directions independently by three hydraulic pumps, which are controlled by a servo control cabinet. The pressure is transmitted to the specimen by a loading board in each direction. The maximum pressure applied in each direction can reach up to 30 MPa.
The AE detecting device is composed of an AE meter, a differential preamplifier, and AE probes. The size of the AE probe is Φ 22 × 36.8 mm, and the detection frequency range is 15-70 KHz, with a resonant frequency of 40 KHz. Eight AE probes were place on four surfaces of the specimen, except the upper and lower surfaces, to monitor acoustic emission features in the experiment (see Figure 2a). In order to model the wellbore, two coaxial holes were drilled in the specimens (the first hole with diameter of 40 mm and depth of 150 mm; the second hole with diameter of 14 mm and depth of 15 mm), and then a stainless steel tube with a diameter of 14 mm was inserted into the center of the unset specimen and it was bond to the borehole wall by cement with a cementing length of 150 mm, and an open hole section (OHS) measuring 15 mm was preserved (see Figure 2b). The orientation of the bedding in the shale was perpendicular to the direction of vertical stress (σv) (see Figure 2c).

Experimental Procedure
The specimen was put on the true tri-axial loading frame, and the eight AE sensors were connected to the data acquisition system were fitted inside the loaded boards and placed in direct contact with the specimen surfaces. After that, the stresses were independently loaded on the specimens by the true tri-axial hydraulic loading device along the X-, Y-, and Z-directions in the Cartesian coordinate. The minimum horizontal stresses (σh), maximum horizontal stresses (σH), and vertical stress (σv) were applied in the X-axis, Y-axis, and Z-axis directions, respectively (see Figure  2a). All experiments were conducted under the same stress states of σv = 9.8 MPa, σh = 8.5 MPa, σH = 10.5 MPa. After the confining pressure was loaded, the injection valve was opened and the fracturing fluid entered into the wellbore, and the acoustic signal was monitored by the AE detecting device. During the experiment, the injection valve was closed when the specimen was broken. After the experiment, all specimens were separated along the fracture traces on the surface to investigate the morphology of fracture plane. Table 2 provides the parameters used in the experiment. As for water fracturing, red ink (Bose, type 855) was mixed into the water in order to mark the crack induced.

Specimen Preparation
Eight specimens 300 × 300 × 300 mm 3 in size were used to conduct the experiment. Three of these specimens were shales and the rest were artificial sandstones. The shales were obtained from outcrops of the Changning Block in Sichuan Province, China, which belonged to black shale of a Longmaxi Formation, and beddings and joints existed in these shales. The artificial sandstones were made from a mixture of cement (P.C 32.5R), quartz sand (40-70 mesh), and water, with a mass ratio of 1:1:0.3, and the making process is as follows: The mixture was stirred evenly, and then it was poured into the a mold which could make cubic specimens 300 × 300 × 300 mm 3 in size. After the mixture solidified, the rock specimen was taken out of the mold, and one artificial sandstone was made; then the artificial sandstones were maintained for 28 days at temperature of 25 • C. The mechanical properties of the shales and artificial sandstones were obtained by a uniaxial compressive strength (UCS) test and a splitting tensile strength (STS) test (Table 1). In order to model the wellbore, two coaxial holes were drilled in the specimens (the first hole with diameter of 40 mm and depth of 150 mm; the second hole with diameter of 14 mm and depth of 15 mm), and then a stainless steel tube with a diameter of 14 mm was inserted into the center of the unset specimen and it was bond to the borehole wall by cement with a cementing length of 150 mm, and an open hole section (OHS) measuring 15 mm was preserved (see Figure 2b). The orientation of the bedding in the shale was perpendicular to the direction of vertical stress (σ v ) (see Figure 2c).

Experimental Procedure
The specimen was put on the true tri-axial loading frame, and the eight AE sensors were connected to the data acquisition system were fitted inside the loaded boards and placed in direct contact with the specimen surfaces. After that, the stresses were independently loaded on the specimens by the true tri-axial hydraulic loading device along the X-, Y-, and Z-directions in the Cartesian coordinate. The minimum horizontal stresses (σ h ), maximum horizontal stresses (σ H ), and vertical stress (σ v ) were applied in the X-axis, Y-axis, and Z-axis directions, respectively (see Figure 2a confining pressure was loaded, the injection valve was opened and the fracturing fluid entered into the wellbore, and the acoustic signal was monitored by the AE detecting device. During the experiment, the injection valve was closed when the specimen was broken. After the experiment, all specimens were separated along the fracture traces on the surface to investigate the morphology of fracture plane. Table 2 provides the parameters used in the experiment. As for water fracturing, red ink (Bose, type 855) was mixed into the water in order to mark the crack induced.  Table 2 provides the fracturing results of the eight specimens. As several repetitive experiments conducted under the same experimental conditions showed similar trends, only the representative results of the four experiments (specimen SC-1, SC-5, W-2, and W-3) are presented in this section.

Fluid Pressure During Fracturing
The fluid pressure at the well head, monitored by a pressure transducer during the experiment, is shown in Figure 3. The fluid pressure-time curve can be divided into four stages, as follows: Fluid injection and pressure rise stage (A-B), specimen rupture stage (B-C), fracture propagation stage (C-D), and stop injection and pressure decay stage (D-E). There are some differences between the fluid pressure-time curve of SC-CO 2 fracturing and water fracturing.   Table 2 provides the fracturing results of the eight specimens. As several repetitive experiments conducted under the same experimental conditions showed similar trends, only the representative results of the four experiments (specimen SC-1, SC-5, W-2, and W-3) are presented in this section.

Fluid Pressure During Fracturing
The fluid pressure at the well head, monitored by a pressure transducer during the experiment, is shown in Figure 3. The fluid pressure-time curve can be divided into four stages, as follows: Fluid injection and pressure rise stage (A-B), specimen rupture stage (B-C), fracture propagation stage (C-D), and stop injection and pressure decay stage (D-E). There are some differences between the fluid pressure-time curve of SC-CO2 fracturing and water fracturing. As for water fracturing, when the control valve was opened, the water entered into the well quickly, with a rapid increase of fluid pressure (A-B). When the fluid pressure reached the breakdown pressure of the specimens, the fracture occurred, and more space was created for the water to store, which led to a rapid decrease of fluid pressure (B-C). The fluid pressure decreased by 73.2% in W-3, and the decrease degree is larger than that in W-2, in which the fluid pressure decreased by 53.9%. This means that more fracture volume was created in W-3 (shale) at the specimen rupture stage due to the existence of bedding planes in the shale. At the fracture propagation stage (C-D), the fluid pressure changed slowly, and the fluid flow in and out gradually reached a balanced state. As for water fracturing, when the control valve was opened, the water entered into the well quickly, with a rapid increase of fluid pressure (A-B). When the fluid pressure reached the breakdown pressure of the specimens, the fracture occurred, and more space was created for the water to store, which led to a rapid decrease of fluid pressure (B-C). The fluid pressure decreased by 73.2% in W-3, and the decrease degree is larger than that in W-2, in which the fluid pressure decreased by 53.9%. This means that more fracture volume was created in W-3 (shale) at the specimen rupture stage due to the existence of bedding planes in the shale. At the fracture propagation stage (C-D), the fluid pressure changed slowly, and the fluid flow in and out gradually reached a balanced state. When water appeared on the outside of specimen, the control valve was closed, and the fluid pressure dropped rapidly (D-E).
The fluid pressure-time curve of SC-CO 2 fracturing was different from water fracturing at the fluid injection and pressure rise stage. When the control valve was opened, the SC-CO 2 entered into the well and changed to a gas at the first time under the low pressure, and the fluid pressure rose rapidly at first, which was followed by a slow rise in fluid pressure due to the complex phase transition process of CO 2 from gaseous to supercritical. The fluid pressure increased rapidly again when the CO 2 in the well was supercritical (A-B). At the specimen rupture stage (B-C), the slope of the pressure curve and the pressure decrease degrees (the pressure decreased by 31.6% in SC-1 and the pressure decreased by 46.7% in SC-5) for SC-CO 2 fracturing were smaller than that in water fracturing under the same conditions. As a consequence of the high compressibility of SC-CO 2 , when the pressure dropped, the SC-CO 2 expanded to retard the decrease of pressure. When the fracture extended to the surface of specimen, a jet phenomenon caused by the CO 2 escaping from the rock and transitioning from supercritical to gas occurred, and the control valve was turned off.

Comparison of Breakdown Pressure
The breakdown pressure is the peak value of the fluid pressure, and the breakdown pressure data of the eight specimens are listed in Table 2. The average breakdown pressure of artificial sandstone fractured by SC-CO 2 is 13.88 MPa, which is almost 10% lower than the average breakdown pressure of artificial sandstone fractured by water. The average breakdown pressure of shale fractured by SC-CO 2 is 23.57 MPa, and the breakdown pressure of shale fractured by water under the same condition is 26.60 MPa. It is clear that whether using shale or artificial sandstone to conduct fracturing experiment, the breakdown pressure of experiment using SC-CO 2 as fracturing fluid is lower than that using water as fracturing fluid, which is consistent with the studies of Zou et al. [26], who used layered tight sandstone to conduct fracturing experiments, and Zhang et al. [22], who used shale to conduct experiments.
Since diffusivity of SC-CO 2 is greater than that of water in the same porous medium, the amount of SC-CO 2 penetrating into the area around the open hole section is larger than that of water at the pressure rise stage (A-B), which leads the pore fluid pressure around the open hole section during SC-CO 2 fracturing to be higher than the pore fluid pressure around the open hole section during water fracturing. According to the study of Zhang et al. [22], the high pore fluid pressure caused by SC-CO 2 penetration reduces the breakdown pressure.

Acoustic Emission Characteristics During Fracturing
In this section, AE energy, which is residual elastic energy released by an acoustic emission source (rupture section of the specimen) and measured from surface of the material after propagation attenuation, and AE cumulative energy, which can reflect total elastic energy released during fracturing process, are used to investigate the fracturing initiation and propagation process in SC-CO 2 fracturing and water fracturing. The AE energy is mainly affected by three factors during fracturing, as follows: Intensity of the AE source, propagation medium of the elastic wave, and fracturing fluid [13].
The AE probe converts elastic waves generated by acoustic emission sources into electrical signals, and the energy of an electrical signal is a proportional to the square of the voltage, so the AE energy generated by each event can be calculated using the following formula [44]: where the subscript i is the channel number of the recorded voltage transient V(t) and t 0 and t 1 are the start and end of the voltage transient record, respectively. The AE energy value is summed over all eight channels to get the total energy measured for each event.
The AE energy and cumulative energy varying with time during SC-CO 2 fracturing and water fracturing are shown in Figure 4. The AE energy release rates of the water fracturing had the same features. The AE energy signal was severe (energy surge), in the tens of seconds, before and after the breakdown point of the specimen. At the fracture propagation stage, the AE energy signal was weak at first and it became strong again when the fractures extended to the surface of the specimen. The AE energy signal still existed after stopping the fluid injection (shut-in) due to the fracture closure. During the process of SC-CO 2 fracturing, the AE energy release rate was intense (energy surge) at the specimen rupture stage. The AE energy signal was relatively severe during the whole stage, which was different from water fracturing in that the AE energy signal was weak at the early stage of fracture propagation. After stopping fluid injection, the AE energy signal still existed, which was similar to water fracturing.
During the process of SC-CO2 fracturing, the AE energy release rate was intense (energy surge) at the specimen rupture stage. The AE energy signal was relatively severe during the whole stage, which was different from water fracturing in that the AE energy signal was weak at the early stage of fracture propagation. After stopping fluid injection, the AE energy signal still existed, which was similar to water fracturing.
Comparing the AE energy release rate between SC-CO2 fracturing and water fracturing, the AE energy release rate of SC-CO2 fracturing was 1-2 orders of magnitude higher than that of water fracturing, signifying that more energy can be converted into kinetic energy to promote fracture growth [45]. The AE energy release rates and AE cumulative energy in shale (W-3 Figure 4b) and artificial sandstone (W-2 Figure 4a) using water as fracturing fluid were of the same magnitude, and there was only a simple fracture in W-2 ( Figure 5b) and W-3 ( Figure 6b). However, the AE energy release rate and AE cumulative energy in shale (SC-5 Figure 4d) using SC-CO2 fracturing was one order of magnitude higher than that in artificial sandstone (SC-1 Figure 4c), which meant more ruptures were created in shale. Through the observation of fracture morphology in SC-5 ( Figure 6a) and in SC-1 (Figure 5a), that more ruptures were created in shale was because SC-CO2 could wreck the weak planes in shale and the artificial sandstone was homogeneous without a weak plane. In conclusion, AE energy release rate and AE cumulative energy can reflect the complexity of the fracture created, and more AE energy releases during SC-CO2 fracturing compared to water fracturing under the same conditions.

Fracture Propagation and Morphology
The spatial morphology of fractures induced by SC-CO2 or water in artificial sandstone and shales are shown in Figures 5 and 6, respectively. As shown in Figure 5a, the artificial sandstone fractured by SC-CO2 had a main fracture with multi-fracture branches, and the main fracture initiated from the open hole section and extended to the specimen surface along approximately perpendicular directions to the minimum principal stress (σh). Through the observation of fracture morphology Comparing the AE energy release rate between SC-CO 2 fracturing and water fracturing, the AE energy release rate of SC-CO 2 fracturing was 1-2 orders of magnitude higher than that of water fracturing, signifying that more energy can be converted into kinetic energy to promote fracture growth [45]. The AE energy release rates and AE cumulative energy in shale (W-3 Figure 4b) and artificial sandstone (W-2 Figure 4a) using water as fracturing fluid were of the same magnitude, and there was only a simple fracture in W-2 ( Figure 5b) and W-3 ( Figure 6b). However, the AE energy release rate and AE cumulative energy in shale (SC-5 Figure 4d) using SC-CO 2 fracturing was one order of magnitude higher than that in artificial sandstone (SC-1 Figure 4c), which meant more ruptures were created in shale. Through the observation of fracture morphology in SC-5 ( Figure 6a) and in SC-1 (Figure 5a), that more ruptures were created in shale was because SC-CO 2 could wreck the weak planes in shale and the artificial sandstone was homogeneous without a weak plane. In conclusion, AE energy release rate and AE cumulative energy can reflect the complexity of the fracture created, and more AE energy releases during SC-CO 2 fracturing compared to water fracturing under the same conditions. directions approximately perpendicular to the minimum principal stress (σh). The morphology of the fracture surface was flat, and there was no obvious fracture bifurcation phenomenon in the specimen. In conclusion, the SC-CO2 fracturing can produce more complicated fractures than conventional water fracturing, and the main fracture mainly propagates along directions perpendicular to the minimum principal stress no matter if using SC-CO2 or water as a fracturing fluid when using artificial sandstone as a fracturing material. The fracture morphology in shale is different from that in artificial sandstone. As shown in Figure 6a, the shale fractured by SC-CO2 has complex fracture networks connected with multibedding planes and joint planes. Observing from the upper surface of the specimen, a main fracture initiated from the open hole section and extended along directions approximately perpendicular to the minimum principal stress (σh), connecting with a fracture propagated along the joint plane and a fracture branch initiated from the well connecting with the joint plane. Viewing from the side surface,

Fracture Propagation and Morphology
The spatial morphology of fractures induced by SC-CO 2 or water in artificial sandstone and shales are shown in Figures 5 and 6, respectively. As shown in Figure 5a, the artificial sandstone fractured by SC-CO 2 had a main fracture with multi-fracture branches, and the main fracture initiated from the open hole section and extended to the specimen surface along approximately perpendicular directions to the minimum principal stress (σ h ). Through the observation of fracture morphology inside the specimen, the fracture branch surfaces were found to be oblique to the main fracture surface at a large angle, and were far from the open hole section, which meant that the fracture branches were caused by main fracture bifurcation when it extended to the specimen surface. As shown in Figure 5b, the artificial sandstone fractured by water only had a simple hydraulic fracture, and the fracture also initiated from the open hole section and extended to the specimen surface along directions approximately perpendicular to the minimum principal stress (σ h ). The morphology of the fracture surface was flat, and there was no obvious fracture bifurcation phenomenon in the specimen. In conclusion, the SC-CO 2 fracturing can produce more complicated fractures than conventional water fracturing, and the main fracture mainly propagates along directions perpendicular to the minimum principal stress no matter if using SC-CO 2 or water as a fracturing fluid when using artificial sandstone as a fracturing material. simple in the two specimens, but the fracture in artificial sandstone which propagated along the direction approximately perpendicular to the minimum principal stress (σh) was vertical. The fracture in shale, which is dominated by the effect of the bedding plane, was mainly horizontal with obvious diversion phenomenon caused by in-situ stress. It is clear that a weak structural plane and in-situ stress compete for the dominance of the propagation direction of fractures [21].

Conclusions
In this study, a series of SC-CO2 fracturing and water fracturing experiments were conducted on cubic shale with bedding planes and homogeneous artificial sandstone and combined with AE monitoring to investigate the characteristics of SC-CO2 fracturing. Based on this experiment, the following conclusions can be drawn: 1. The fluid pressure-time curve of SC-CO2 fracturing is different from water fracturing. The pressure rise stage of SC-CO2 fracturing takes 7-9 min due to the complex phase transition process of CO2 in this stage, and the pressure rise stage of water fracturing only takes 1-2 min. At the specimen rupture stage and pressure decay stage, the pressure drop process of SC-CO2 is relatively flat compared with water fracturing due to the high compressibility of SC-CO2. The fracture morphology in shale is different from that in artificial sandstone. As shown in Figure 6a, the shale fractured by SC-CO 2 has complex fracture networks connected with multi-bedding planes and joint planes. Observing from the upper surface of the specimen, a main fracture initiated from the open hole section and extended along directions approximately perpendicular to the minimum principal stress (σ h ), connecting with a fracture propagated along the joint plane and a fracture branch initiated from the well connecting with the joint plane. Viewing from the side surface, two fractures propagated along the bedding plane and connected with the joint plane, and a complicated fracture network formed. Taking off the rocks on the top of bedding plane ( Figure 5 top right-upper rocks removed), the fractures were more complicated below the bedding plane. As shown in Figure 6b, the shale fractured by water only had a single horizontal fracture, and no vertical fracture occurs. Since bedding plane is well developed in shale, and in the meantime, weak cementing bedding plane occurs at the open hole section, the fracture is prone to extend along horizontal bedding plane. Fracture diversion phenomenon occurred in this sample, and the direction of fracture extension diverted to the direction perpendicular to the minimum principal stress (σ h ) due to influence of in-situ stresses. In fact, the fracture propagation and morphology are significantly influenced by a weak structural plane (bedding planes, joint planes, and natural fractures) and in-situ stresses [27,35,46,47], and the fracture propagation direction is controlled by combined effect of a weak structural plane and in-situ stresses. Comparing fracture propagation and morphology in SC-5 and W-3, the fractures created by SC-CO 2 fracturing were more complicated than fractures created by water fracturing, and the SC-CO 2 induced fractures were prone to connect with the weak structural plane (bedding plane and joint plane) to form a more complex fracture network than water induced fractures due to the low viscosity and high diffusivity of SC-CO 2 , which is consistent with the experimental study of Zhang et al. [22].
Comparing fractures in artificial sandstone ( Figure 5a) and shale (Figure 6b) induced by SC-CO 2 , the existing of the bedding plane and the joint plane increases the complexity of fractures and is conducive to the formation of complex fracture network in shale by SC-CO 2 fracturing. As for fractures created by water in artificial sandstone ( Figure 5b) and shale (Figure 6b), the fractures were simple in the two specimens, but the fracture in artificial sandstone which propagated along the direction approximately perpendicular to the minimum principal stress (σ h ) was vertical. The fracture in shale, which is dominated by the effect of the bedding plane, was mainly horizontal with obvious diversion phenomenon caused by in-situ stress. It is clear that a weak structural plane and in-situ stress compete for the dominance of the propagation direction of fractures [21].

Conclusions
In this study, a series of SC-CO 2 fracturing and water fracturing experiments were conducted on cubic shale with bedding planes and homogeneous artificial sandstone and combined with AE monitoring to investigate the characteristics of SC-CO 2 fracturing. Based on this experiment, the following conclusions can be drawn:

1.
The fluid pressure-time curve of SC-CO 2 fracturing is different from water fracturing. The pressure rise stage of SC-CO 2 fracturing takes 7-9 min due to the complex phase transition process of CO 2 in this stage, and the pressure rise stage of water fracturing only takes 1-2 min. At the specimen rupture stage and pressure decay stage, the pressure drop process of SC-CO 2 is relatively flat compared with water fracturing due to the high compressibility of SC-CO 2 .

2.
Under the same in-situ stress condition, the breakdown pressure of SC-CO 2 fracturing is about 10% lower than that of water fracturing no matter if using shale or artificial sandstone as fracturing materials, because the percolation effect of SC-CO 2 can greatly increase pore pressure, which leads to the decrease in breakdown pressure.

3.
The AE energy surge phenomena mainly occur at the specimen rupture stage due to the severe rupture of the specimens. The AE energy release rate of SC-CO 2 fracturing is 1-2 orders of magnitude higher than that of water fracturing, signifying that more energy can be converted into kinetic energy to promote fracture growth.

4.
By observing the fracture morphology in shale and artificial sandstone fracturing by SC-CO 2 and water, the main fracture mainly propagates along the directions perpendicular to the minimum principal stress in artificial sandstone, which is homogeneous no matter if using SC-CO 2 as fracturing fluid; but in shale, the weak structural plane and in-situ stresses compete for the dominance of the propagation direction of the fractures.