Experimental study on the effect of CO₂ storage on the reservoir permeability in a CO₂-based enhanced geothermal system

Abstract

A CO₂-based Enhanced Geothermal System (CO₂-EGS) has dual benefits of heat extraction and CO₂ storage. Mineralization storage of CO₂ may reduce reservoir permeability, thereby affecting heat extraction. Solutions require further research to optimize and balance these two benefits. In this study, CO₂ storage and heat extraction were simulated by alternating cyclic injection of water and supercritical CO₂ into fractured granite. By analyzing the changes of ion composition in water samples and the minerals of fracture surface, the mechanisms controlling the fracture permeability with and without proppant were obtained. The results suggest that monticellite and vaterite were formed besides montmorillonite, calcite and illite after increasing the injection cycles. This promotes mineralization storage of CO₂ but reduces reservoir permeability. Without proppant, the permeability decreased in three stages and the reduction rate exhibited a sharp-slow–fast–slow trend. While the use of proppant caused an increase of two orders of magnitude in permeability. Therefore, increasing the non-contact area of the main fracture and the CO₂ flow velocity can avoid a large decrease in permeability, which will increase the heat extraction and mineralization storage of CO₂. The findings provide solutions for the CO₂ emission reduction and the efficient exploitation of hot dry rock.

Introduction

The ever-increasing CO₂ emissions have a significant impact on climate change and lead to many extreme climate events, including floods, wildfires, and heat waves (Meinshausen et al. 2022; Guo et al. 2019). Therefore, many studies have focused on low-carbon production technology, CO₂ capture and storage technology, and how to achieve carbon emission reduction by government legislative intervention and market regulation (Subramanian et al. 2019; Frank et al. 2019; Deng et al. 2019; Ager and Lapkin 2018). Renewable energy is central to invigorating the transition to a low-carbon and sustainable energy system (IEA 2021). In the long term, renewable energy consumption per capita increases by 1%, the per capita carbon emissions will be reduced by 0.259% (Li and Haneklaus 2021). As a renewable energy, hot dry rock (HDR) geothermal energy has huge reserves. Compared to the solar energy and wind energy, the exploitation of HDR is characterized by stability and seasonal independence (Hou et al. 2018; Liao et al. 2020). To increase the heat extraction and reduce the carbon emission, a novel solution has been proposed that using CO₂-based Enhanced Geothermal Systems (CO₂-EGSs) for HDR exploitation (Pruess 2006).

At present, how to improve the heat extraction rate is one of the main difficulties in HDR exploitation. The flow rate of CO₂ is significantly affected by reservoir permeability, which is a key factor affecting the heat extraction rate. Existing studies have shown that the permeability is influenced by the fracture aperture, fracture surface roughness, pressure and temperature of the heat extraction fluid, and the reservoir temperature (Shu et al. 2020; Guo et al. 2017; Kamali-Asl et al. 2018). Accordingly, a series of solutions have been proposed to improve reservoir permeability. Such as the use of proppant and the reservoir stimulation by microwave heating. The former is a common strategy for increasing the fracture aperture, but a high seepage pressure can redistribute the proppant’s action range and reduce the propping effectiveness (Huang et al. 2021). Fracture permeability is very sensitive to the rock mass’s temperature (Song et al. 2021). When granite is heated above 600 °C, its permeability can increase by about two orders of magnitude (Liu et al. 2020). Therefore, reservoir stimulation by microwave heating may be an effective method. In addition, it is necessary to evaluate the enhancement effects of permeability under different methods. It has been found that permeability is independent of the average aperture and fracture size within a specific contact area. Moreover, the permeability can be predicted based on the change in the resistivity ratio between the saturated rock and the fluid (Sawayama et al. 2021).

The injection of CO₂ may result in CO₂-water–rock interactions, which have a significant impact on the reservoir permeability. In the EGS reservoir with granite-dominated rock, the hydrothermal alteration of granite causes the dissolution of primary minerals and the crystallization of newly formed phases on fracture surfaces, which can improve the porosity. While the crystallization of illite and calcite decreases the permeability (Ledésert et al. 2010). In the two possible types of granite alteration, vein alteration can form carbonate, quartz and some clay minerals, which significantly change the granite overall porosity. And these processes are greatly influenced by the reservoir temperature and the chemical characteristics of fluid. In addition, chemical dissolution and mechanical creep of the protrusions on the fracture surfaces will reduce the aperture and permeability (Caulk et al. 2016). However, long-term CO₂-water–rock interactions may cause silicification and embrittlement of granite, which results in the extension of shear failure and the increase of permeability (Dempsey et al. 2015). Through numerical analysis, we found that the permeability is relatively less sensitive to mechanical and hydraulic effects, and the thermal effect has short- or medium-term influences on the permeability, while the chemical effect has long-term and significant influences on the permeability (Tao et al. 2019).

Several experimental studies have been conducted to analyze CO₂-water–rock interactions by placing block or powdered samples in a reaction kettle containing CO₂ solution (Na et al. 2015; Liu et al. 2003; Ueda et al. 2005; Ré et al. 2014). In addition, by injecting a mixture of CO₂ and water, one study analyzed the effect of this method on the production sustainability of a two-phase geothermal reservoir (Kaya et al. 2018). However, for the influences of chemical effect on the permeability of CO₂-EGS reservoirs, few studies have comprehensively considered the two-phase flow of CO₂ and water coupled Thermo-Hydro-Mechanical-Chemical (THMC) effect. In this regard, based on our previous research (Wu and Li 2020; Wu et al. 2021), a novel experimental method of alternating cyclic injection of water and supercritical CO₂ (CO₂ (sc)) into fractured granite was adopted in this study. By increasing the number of injection cycles, the influences of the coupled THMC effect on the fracture surfaces’ morphology and mineral crystallization were studied, and the mechanisms controlling the fracture permeability with and without proppant were obtained. The results of this study provide technical support for increasing the heat extraction rate of the CO₂-EGS and the CO₂ emission reduction.

Materials and methods

Experimental installation

A schematic diagram of the triaxial seepage system used in this study to produce the coupled THMC effect is shown in Fig. 1. The experimental system consists of four units, including the CO₂ (sc) preparation unit (Unit I), the water injection/pressurization unit (Unit II), the high-temperature seepage unit (Unit III) and the system control and data acquisition unit (Unit IV). In the preparation of CO₂ (sc), the CO₂ cylinder is connected to the booster pump, the air compressor and the booster pump are opened to pressurize CO₂, and the pressurized CO₂ is stored in the high pressure gasholder. Then, the preheater is opened to inject the CO₂. The pressure regulating valve is used to adjust the injection pressure, and the inlet pressure sensor of Unit II is observed to reach the set value. The pressure is maintained for continuous injection for 1 h, until the temperature of CO₂ flowing through the preheater does not exceed 0.5 °C, then CO₂ reaches the supercritical state. During the experiment, pressures are exerted on the rock sample by the axial pressure pump and the confining pressure pump in unit II. Water is injected through the water injection pump, and the injection water is measured in real time. The pressure sensors and temperature sensors are used to collect the pressures and temperatures of CO₂ at injection port and flow outlet, respectively. The temperature control probe and the temperature measuring probe in the triaxial seepage chamber are used to measure the heating temperature and the rock sample temperature, respectively. Finally, all the pressures and temperatures during the experiment are collected and stored in real time through the data collection system.

Fig. 1

Schematic diagram of the triaxial seepage experimental system for producing the coupled thermo-hydro-mechanical-chemical effect

The hydro effect is realized by the Unit I, Unit II, and Unit III. Unit I can prepare CO₂ (sc) with a pressure of 7.5–20 MPa and a temperature of 40–100 °C. The thermo and chemical effects are achieved primarily using Unit III, in which the granite can be heated to a maximum temperature of approximately 300 °C using a high-temperature oil bath. Therefore, CO₂-water–rock interactions may occur on the surfaces of the fractures in the granite. The fracture water sample during each injection cycle can be collected using Unit IV to analyze its ionic composition and content. The mechanical effect is realized using Unit II and Unit III. The axial pressure pump and confining pressure pump in Unit II supply the triaxial pressure on the rock sample, and the pressure provided by both units can reach 60 MPa.

Experimental design and rock sample preparation

Two groups of experiments were designed in this study, including heat extraction experiments with and without proppant in the fractures. The granite used in this experiment was from the high heat flow granite in Zhangzhou, Fujian Province of China, and the fractured granite samples were prepared using the Brazilian splitting method (Shams et al. 2023). Since the first group of experiments was perfected based on our previous carbon storage studies, the rock samples and experimental conditions were consistent with those used in our previous study (Wu et al. 2021). Figure 2 shows the first group of experimental rock samples with a diameter and height of 50 mm, namely, fractured granite samples CA-1 and CA-2 without proppant.

Fig. 2

Fractured granite samples CA-1 and CA-2 without proppant

In the first group of experiments, to analyze the influence of the number of alternating injection cycles on the permeability, rock samples CA-1 and CA-2 were subjected to six and nine injection cycles, respectively, and each cycle lasted for 3 days. In this study, using the condition of hot dry rock at Bad Urach in Germany as the background, the rock samples were subjected to a confining pressure of 30 MPa and an axial pressure of 35 MPa. The rock samples were heated to 200 °C (simulated reservoir temperature), and the CO₂ injection pressure and temperature were 7.5 MPa and 40 °C [i.e., CO₂ (sc)], respectively. Similar to the experimental method for the rock sample CA-1, we collected fracture water samples from rock sample CA-2 during each injection cycle and analyzed the ion compositions and contents. Figure 3 shows the fracture water samples collected during each injection cycle and the pH testing process.

Fig. 3

Fracture water samples collected during each injection cycle for rock sample CA-2 and pH testing

Figure 4 shows fractured granite samples CA-3 and CA-4 with proppant. The proppant was sintered from high-quality bauxite and was 300–600 μm in size. In addition, it has the advantages of a high strength (> 52 MPa), a wear resistance at high temperatures, and is insoluble in water. The proppant was added in a different way than in a real reservoir. First, the proppant was spread evenly over the fracture surfaces in half of the rock sample. This part of the rock sample was then fitted to the other part. Finally, high temperature resistant (~ 260 °C) Teflon seal tape was wrapped tightly around the rock sample to prevent the proppant from falling off. The rock sample was then placed in the core holder of Unit III.

Fig. 4

Fractured granite samples CA-3 and CA-4 with proppant

Using the first group as the control experiments, the second group of experiments was designed to explore whether the mineral dissolution or that of the mineral precipitation affected the fracture permeability more under CO₂-water–rock interactions. Therefore, in the two groups of experiments, the same CO₂ (sc) injection conditions, pressure and temperature were applied to the rock sample. To increase the zone affected by the proppant, the heights of rock samples CA-3 and CA-4 were double the height (i.e., 100 mm) of the rock samples in the first group.

Fracture permeability calculation

A permeability test of a gas with a high compressibility in a rock sample needs to consider two aspects. One is the possible gas slip phenomenon (Klinkenberg effect) under a lower average flow pressure, and the other is the inertia effect of high-speed flow (Caulk et al. 2016). CO₂ (sc) with a high flow pressure was used as the working fluid in the experiments, so the Klinkenberg effect could be ignored. The Reynolds number (Re) of the fluid flow has been widely used to evaluate the influence of the inertia effect in flow systems, so it was necessary to first estimate the Reynolds number of the CO₂ (sc) flow in the fractured granite. In the experiments, the rock sample was a cylindrical fractured granite sample with a diameter of 50 mm and a height of 50 mm (or 100 mm). According to the definition of the fluid flow state in hydraulics, the Reynolds number of the CO₂ (sc) flow in the fracture was calculated as follows (Kolditz 2001):

$$Re = \frac{{2v_{\text{f}} w\rho_{\text{f}} }}{\mu }$$

(1)

where $v_{\text{f}}$ is the average flow velocity of the CO₂, w is the average fracture aperture, $\rho_{\text{f}}$ is the density of the CO₂, and $\mu$ is the dynamic viscosity of the CO₂.

In this study, the maximum average fracture aperture in the two groups of experimental rock samples was about 0.318 mm. Owing to the effect of the confining pressure, the fracture aperture was less than this value. Other parameters such as the average velocity $v_{\text{f}}$ and density $\rho_{\text{f}}$ of the CO₂ (sc) were taken as the maximum values, and the dynamic viscosity $\mu$ was taken as the minimum value. Using Eq. (1), the maximum Reynolds number was estimated to be 3.8.

When Re > 1, the fluid flow exhibits the characteristics of non-Darcy flow. However, when 1 < Re < 10, the inertial force producing the nonlinear effect in the fracture flow is very weak (Ranjith and Viete 2011). Therefore, the influence of the inertial force on the flow could be ignored in this study, and it is believed that the process of the CO₂ (sc) flow in the fractured granite satisfied Darcy's law. Thus, the permeability of the fracture can be obtained as follows (Caulk et al. 2016):

$$k = \frac{{b^2 \rho_{\text{f}} g}}{12}$$

(2)

$$\frac{{p_{{\text{in}}}^2 - p_{{\text{out}}}^2 }}{{2p_{{\text{out}}} L}} = \frac{{12\mu Q_{\text{f}} }}{{b^3 w\rho_{\text{f}} g}}$$

(3)

where k is the permeability, b is the equivalent hydraulic aperture, g is the acceleration due to gravity, $p_{{\text{in}}}$ and $p_{{\text{out}}}$ are the pressures of the CO₂ at the inlet and outlet, respectively, L is the length of the rock sample, and $Q_{\text{f}}$ is the volume flow rate of the CO₂.

Results and discussion

Chemical properties of water in fractures

The chemical properties of the water samples from rock sample CA-1 can refer to our previous study (Wu and Li 2020). From the initial stage to the 6th injection cycle, the pH of the water sample was reduced from 6.5 to 4.7. For other ions, the concentrations of ${\text{Cl}}^-$, ${\text{NO}}_3^-$ and ${\text{SO}}_4^{2 - }$ did not change significantly in each injection cycle. The ${\text{Na}}^{ + }$ concentration increased from 4.7 mg/L to 26.9 mg/L, and ${\text{K}}^+$ concentration increased from 5.2 mg/L to 106.1 mg/L, and then decreased to 24.9 mg/L. The ${\text{Ca}}^{2 + }$ concentration fluctuated continuously during the six injection cycles, indicating that Ca-containing minerals dissolved and precipitated. The ${\text{Mg}}^{2 + }$ concentration increased slowly from 1.4 mg/L to 4.1 mg/L. Figure 5 shows the changes in the pH of the fracture water samples from rock sample CA-2 during each injection cycle. Similar to rock sample CA-1, after the first injection cycle, the pH changed from approximately neutral to slightly acidic. Then, as the number of injection cycles increased, the pH cyclically increased and decreased slightly. This indicates that the continuous dissolution of a small amount of CO₂ caused the dissolution and precipitation of minerals on the fracture surfaces, which led to the constant generation and consumption of ${\text{H}}^+$. The specific forms of mineral dissolution and precipitation were preliminarily identified based on the changes in the ion compositions and contents of the water samples.

Fig. 5

Changes in the pH of the fracture water samples from rock sample CA-2 during each injection cycle

The changes in the ion composition and content of the water samples from rock sample CA-2 are shown in Fig. 6. It can be seen that during injection cycles 0–2, the ${\text{Mg}}^{2 + }$ content did not change significantly, while the ${\text{Na}}^{ + }$ and ${\text{Ca}}^{2 + }$ contents increased slowly, and the ${\text{K}}^+$ content decreased significantly. This indicates that the decrease in the pH caused the dissolution of feldspars such as albite and anorthite, releasing the corresponding cations into the fracture water, generating illite precipitation under the K-rich conditions, and reducing the ${\text{K}}^+$ content. The following geochemical reactions may be present in the above processes.

$$2\,{\text{potassium}}\,{\text{feldspar}}\,{ + }\,{\text{2H}}^+ \, + \,9{\text{H}}_2 {\text{O}} \to \,2{\text{K}}^+ \, + \,{\text{kaolinite}}\,{ + }\,\,{\text{4H}}_{4} {\text{SiO}}_{4}$$

(4)

$$3\,{\text{albite}}\, + \,{\text{K}}^{ + } \, + \,2{\text{H}}^{ + } \, \to \,{\text{illite}}\,+{\text{3Na}}^{ + } \, + \,6{\text{SiO}}_6$$

(5)

$$3\,{\text{anorthite}}\,{ + }\,{\text{2K}}^{ + } \, + \,4{\text{H}}^{ + } \, \to \,{\text{illite}}\,{ + }\,{\text{3Ca}}^{_{2 + } }$$

(6)

Fig. 6

Changes in the ion content of the water samples from rock sample CA-2

During injection cycles 2–5, the ${\text{Mg}}^{2 + }$ content increased slightly, and the rate of increase of the ${\text{K}}^+$ content exhibited a slow–fast–slow trend, indicating that the biotite dissolved as the ${\text{Na}}^{ + }$ and ${\text{Ca}}^{2 + }$ contents increased, and small amounts of ${\text{Mg}}^{2 + }$ and ${\text{K}}^+$ were generated. The rapid increase in the ${\text{K}}^+$ content indicates that the dissolution of potassium feldspar may also have occurred. The rate of increase of the Na⁺ content gradually increased, while that of the ${\text{Ca}}^{2 + }$ content gradually decreased. It shows that calcite, illite, and kaolinite may have precipitated as the potassium feldspar, albite, and calcium feldspar dissolved. In addition, the ${\text{Mg}}^{2 + }$ content initially increased and then decreased gradually, which indicates that dolomite precipitation (containing Ca/Mg) also occurred. Relevant geochemical reactions may include:

$${\text{biotite}}\, + \,3{\text{Al}}^{3 + } \, + \,{\text{Si}}^{4 + } \, + \,6{\text{H}}_{2} {\text{O}}\, \to \,{\text{K}}^{ + } \, + \,{\text{Al}}_{4} {\text{Si}}_{4} {\text{O}}_{{10}} {\text{(OH)}}_8 \, + \,3{\text{Fe}}^{2 + } \, + \,3{\text{Mg}}^{2 + } \, + \,6{\text{H}}^{ + } \, + \,2{\text{F}}^{ - }$$

(7)

$${\text{anorthite}}\,{ + }\,{\text{H}}^{ + } + \,{\text{HCO}}_3^- \, + \,{\text{H}}_{2} {\text{O}}\, \to \,{\text{kaolinite}}\, + \,{\text{calcite}}$$

(8)

$$2\,{\text{albite}}\,{ + }\,{\text{2H}}^{ + } \, + \,9{\text{H}}_{2} {\text{O}}\, \to \,{\text{kaolinite}}\,{ + }\,{\text{2Na}}^{ + } \, + \,4{\text{H}}_{4} {\text{SiO}}_{4}$$

(9)

$$2\,{\text{potassium}}\,{\text{feldspar}}\, + \,2{\text{H}}^{ + } \, + \,9{\text{H}}_{2} {\text{O}}\, \to \,{\text{2K}}^{ + } \, + \,{\text{kaolinite}}\,{ + }\,{\text{4H}}_{4} {\text{SiO}}_{4}$$

(10)

$$3\,{\text{anorthite}}\, + \,2{\text{K}}^{ + } \, + \,4{\text{H}}^{ + } \, \to \,{\text{illite}}\,{ + }\,{\text{3Ca}}^{2 + }$$

(11)

$$\begin{gathered} {2}\,{\text{plagioclase}}\,{ + }\,{1}{\text{.5}}\,{\text{potassium}}\,{\text{feldspar}}\,{ + }\,{0}{\text{.8}}\,{\text{Mg}}^{2 + } \, + \,0.25\,{\text{Fe}}^{2 + } \, + \,2.2{\text{CO}}_2 + \,4{\text{H}}_{2} {\text{O + }} \rightleftharpoons \,1.4{\text{Na}}^{ + } + \, \hfill \\ 0.8{\text{K}}^{ + } \, + \,{\text{HCO}}_3^- \, + \,{\text{kaolinite}}\, + \,{\text{illite}}\,{ - }\,{\text{montmorillonite}}\,{ + }\,{4}{\text{.4}}\,{\text{quartz}}\,{ + 6}\,{\text{dolmite}} \hfill \\ \end{gathered}$$

(12)

During injection cycles 5–6, the contents of the four cations increased at different rates. The rate of increase of the ${\text{Na}}^{ + }$ content was slow, while that of the other three cations were fast. This demonstrates that the precipitation of calcite, illite, and dolomite consumed the ${\text{Ca}}^{2 + }$, ${\text{K}}^+$, and ${\text{Mg}}^{2 + }$ in the fracture water, respectively, and promoted the dissolution of anorthite and potassium feldspar. Furthermore, owing to the high ${\text{Na}}^{ + }$ and ${\text{K}}^+$ contents, the dissolution of biotite was promoted. The dissolution of these minerals increased the rates of increase of the ${\text{Ca}}^{2 + }$, ${\text{K}}^+$, and ${\text{Mg}}^{2 + }$ contents, but decreased the rate of increase of the ${\text{Na}}^{ + }$ content, indicating that the precipitation of Na containing cancrinite may have occurred. The possible geochemical reactions involved include Formulas (7), (8), (10), (11) and (12).

During injection cycles 6–8, the ${\text{Mg}}^{2 + }$ and ${\text{Ca}}^{2 + }$ contents gradually stabilized, indicating that the precipitation consumption and dissolution generation of ${\text{Mg}}^{2 + }$ and ${\text{Ca}}^{2 + }$ became balanced. Nevertheless, the ${\text{K}}^+$ content decreased as the ${\text{Na}}^{ + }$ content increased, indicating that the K-rich solution promoted the dissolution of albite and generated illite. The consumption of ${\text{K}}^+$ promoted the dissolution of potassium feldspar and released ${\text{K}}^+$ into the fracture water. In addition, the precipitation of cancrinite consumed ${\text{Na}}^{ + }$ and reduced its content. The possible geochemical reactions involved include Formulas (5) and (10).

During injection cycles 8–9, the ${\text{Na}}^{ + }$ and ${\text{K}}^+$ contents continued to increase, while the ${\text{Mg}}^{2 + }$ content did not change significantly, and the ${\text{Ca}}^{2 + }$ content slightly decreased. This indicates that only calcite was generated. The dissolution of albite did not consume ${\text{K}}^+$, and the kaolinite was precipitated during this process. The possible geochemical reactions involved include Formulas (8), (9) and (10).

Analysis of mineral changes on the fracture surfaces

The analysis of X-Ray Diffraction (XRD) and the observations of Scanning Electron Microscopy (SEM) revealed the compositions of the minerals on the fracture surfaces of rock sample CA-1, which were described in our previous study (Wu et al. 2021). The mineral compositions on the fracture surfaces of rock sample CA-2 obtained by XRD are shown in Fig. 7. Since rock samples CA-1 and CA-2 were prepared from two parts of a standard rock sample, the two samples had the same primary mineral composition (Fig. 7a). Compared to injection cycle 0, new minerals were formed after injection cycle 9, including cancrinite, vaterite (a thermally unstable phase of CaCO₃), dolomite, kaolinite, illite, and monticellite (silicate minerals containing Ca and Mg) (Fig. 7b).

Fig. 7

XRD analysis of mineral compositions on fracture surfaces of rock sample CA-2. a is before the 1st injection cycle (Wu et al. 2021), b is after the 9th injection cycle

The results are generally consistent with the analysis of changes in the ion contents of the water samples presented in Sect. "Chemical properties of water in fractures". The difference is that monticellite is present in the XRD analysis results. Therefore, the rates of increase of the ${\text{Ca}}^{2 + }$ and ${\text{Mg}}^{2 + }$ contents of the water samples slowed down or even slightly decreased, which indicates the formation of monticellite. The specific minerals were further qualitatively analyzed using SEM.

After injection cycle 9, blocky samples were collected from the fracture surface in rock sample CA-2 and three areas were selected for SEM combined Energy Dispersive Spectroscopy (SEM–EDS) analysis of the surface minerals. In area 1 (Fig. 8), Mg, Al, K, and Si are the minor elements at points 1–1 and 1–2 (generally Wt % ≤ 5% can be regarded as minor elements, and Wt% ≤ 0.5% can be regarded as trace elements). Therefore, the minerals at the above two points were identified as calcite according to their compositions. The difference is that the calcite at point 1–1 is the mineral base (i.e., the primary mineral) on the fracture surface, while the calcite at point 1–2 is a small granular mineral crystallization, which is obviously a new mineral formed via chemical reactions. Therefore, the ${\text{Ca}}^{2 + }$ at and around point 1–2 may have been derived from the dissolution of primary calcite or anorthite.

Fig. 8

SEM–EDS analysis results of the mineral components in area 1

At point 1–3, first, the morphology shows that it is the mineral base of the fracture, which has many dissolution pores on the surface, so it may be a feldspar mineral. Then, according to the element composition, without considering the trace element K, the atomic number ratio of the other elements was calculated to be O:Ca:Si:Al:C:Na = 15.8:2.2:4.8:3:4:1. It can be concluded that the mineral is mainly oligoclase, and there may be few calcite crystals.

At point 1–4, the mineral occurrence is rod-like crystal clusters, indicating that it is a new mineral formed in the experiment. Without considering the trace elements and based on the contents of the different elements present, it can be identified as the thermally stable calcite phase of CaCO₃.

The SEM–EDS analysis results of the mineral components in area 2 are shown in Fig. 9. At point 2–1, the mineral base of the fracture surface is the same as that at point 1–3. The mineral has obvious dissolution pores and occurs as highly crystalline rods with granular mineral particles around them. In addition, the atomic number ratio of each element is O:Ca:Si:Al:C:Na = 17.25:3.6:2.6:2.96:4.57:1, so it is concluded to be oligoclase with a small amount of calcite crystal grains.

Fig. 9

SEM–EDS analysis results of mineral components in area 2

At point 2–2, without considering the minor and trace elements, the atomic number ratio of each element was calculated to be O:Ca:C:Si = 12:2:6.38:1. Therefore, the mineral particles at this point are mainly calcite, and there is a quartz mineral base nearby. The minerals and elements at points 2–3 and 2–2 were found to be consistent in type, but the content of Si increases and the content of Ca decreases at point 2–3. Therefore, the mineral at point 2–3 is the primary quartz on the fracture surface with a small amount of calcite crystals attached to it.

Figure 10 shows the SEM–EDS analysis results of the mineral components in area 3. It can be seen from the morphology that the mineral at point 3–1 is the mineral base of the fracture surface. The analysis of the mineral’s composition shows that the Al and Si contents are relatively high, so it was concluded to be aluminosilicate. In addition, it also contains K, Fe, and Mg, which indicates that it is biotite.

Fig. 10

SEM–EDS analysis results of mineral components in area 3

At point 3–2, the element contents reveal that the O and Ca contents are high, but there is no C present. The ratio of the atomic number of each element calculated has no specific rule, and the minerals occur as irregular granules. Therefore, it is inferred that the minerals are Ca, Al, Mg, K, and Fe oxides, with a high CaO content. The types of elements in the mineral at point 3–3 are the same as those at point 3–2, but the Ca content is slightly lower, while the Al and Si contents are higher. Moreover, the mineral is needle like, so it was concluded that the aluminosilicate at this point is scolecite.

According to the above SEM–EDS results, the influence of the new minerals on the morphology of the primary mineral base of the fracture surface after the experiment can be analyzed. Figure 11 shows the SEM images of the fracture surface of rock sample CA-2 before and after the experiment. Before the experiment (Fig. 11a–c), the morphology of the primary minerals on the surface of the main fracture is relatively flat, and that of the minerals in the microfractures and micropores is clear and smooth. After the experiment, highly crystalline granules and rod-like calcite were observed near the microfractures (Fig. 11d and e). In the microfractures and micropores (Fig. 11f), highly crystalline rod-like clusters of calcite and hairball squama of montmorillonite were formed. Furthermore, highly crystalline rod-like calcite crystal clusters (Fig. 11g and h) and blocky carbonates such as CaCO₃ and CaMg(CO₃)₂ (dolomite) were formed on the main fracture surface and between the narrow minerals (Fig. 11i).

Fig. 11

SEM images of the fracture surface in rock sample CA-2 (a–c) before and (d–i) after the experiment

It can be seen that the CO₂-water–rock interactions in the fracture resulted in obvious mineral dissolution and precipitation. At the same time, the types and morphology of mineral crystallization on the surface of main fracture, micro-fractures and micropores were different. There are two main reasons for this result. On the surface of the main fracture, the difference of CO₂ velocity caused by concave-convex mineral particles changes the CO₂-water–rock interaction time. While for the surface of unconnected micro-fractures or micropores, the difference in mineral crystallization is mainly caused by residual trapping effect (Wu and Li 2020). This mineral crystallization changes the roughness of the fracture surface, affects the flow of CO₂, and has a significant impact on the permeability (Huang et al. 2019; Wang et al. 2016; Fang et al. 2018). Therefore, we further analyzed the coupled thermo-hydro-mechanical–chemical effect on the evolution of the fracture permeability.

Permeability evolution in the absence and presence of proppant

In the absence of proppant, the permeability evolution of rock sample CA-1 with increasing number of injection cycles is shown in Fig. 12. It can be seen that the overall permeability decreased as the number of injection cycles increased, but the reduction rate of the permeability varied greatly in the different stages.

Fig. 12

Permeability evolution of rock sample CA-1 with increasing number of injection cycles

It can be seen that during injection cycles 0–1, the permeability decreased at the highest rate and within the largest range. This stage was mainly affected by pressure dissolution (Lu et al. 2018; Yasuhara et al. 2004). On the surfaces of the fracture, some convex particles had asperity contact under the action of the pressure (see Fig. 13a). The contact part of the particles was squeezed and a large number of microfractures formed, which resulted in a crushed zone (see Fig. 13b). In this zone, the mineral dissolution was enhanced. The crushed particulate solute slowly diffused into the main fracture channel and generated mineral precipitation near the particles (see Fig. 13c). As the pressure dissolution continued, the contact area between the particles gradually increased, and the effect of the pressure dissolution gradually weakened. Therefore, the average hydraulic aperture b of the fracture decreased to $b - \Delta b$, where $\Delta b$ is the average width of the pressure dissolution zone. It can be seen that the coupled thermo (T)-hydro-mechanical (σ)–chemical effect affected the hydraulic aperture during this process. The variations in the hydraulic aperture before and after the pressure dissolution are shown in Fig. 14.

Fig. 13

Schematic diagram of the pressure dissolution effect on the asperity contact particles of the fracture surface. a convex particles contact, b crushed zone is formed by squeezing, c mineral dissolution and precipitation (adopted from Yasuhara et al. 2004)

Fig. 14

Schematic diagram of the variations in the hydraulic aperture in the absence of proppant before (left) and after (right) pressure dissolution (adopted from Yasuhara et al. 2004)

During injection cycles 1–3 and 4–6, the reduction rate of the permeability tended to decrease to different degrees. During the injection cycle 6, the pressure dissolution caused weakening and disintegration of the particle contact positions, resulting in three distinct stages of permeability variation. In the first stage (i.e., injection cycles 0–1), the dissolutions of albite, anorthite and potassium feldspar mainly occurred at the particle contact positions on the fracture surface, and the potential precipitations of calcite, dolomite, kaolinite and illite were generated on the fracture’s free surface. In the two subsequent stages (i.e., injection cycles 1–3 and 4–6), the mineral dissolution and precipitation on the free surface gradually became apparent. The significant mineral precipitation reduced the fracture aperture, which led to a decrease in the permeability. Owing to the non-isothermal flow of the water and CO₂, the matrix heat conduction affected the CO₂ dissolution and the other chemical processes near the fracture’s free surface, thus affecting the permeability.

The permeability evolution of rock sample CA-2 with increasing number of injection cycles is shown in Fig. 15. Compared with rock sample CA-1, when the two rock samples had a similar initial permeability, their permeabilities after the experiment were also relatively close. Moreover, there were also three stages of permeability evolution. In the first stage (i.e., injection cycles 0–1), the pressure dissolution had an intense effect on the permeability. The difference is that in the two subsequent stages (i.e., injection cycles 1–5 and 5–9), the reduction rate of the permeability was relatively slow at the beginning of each stage, but it gradually increased as the number of injection cycles increased. In general, the reduction rate of the permeability exhibited a sharp-slow–fast–slow trend. These results indicate that the contact particles on the fracture’s surface underwent strength weakening stage (expansion of the crushed zone), a dissolution and disintegration stage, and a stabilization stage (i.e., re-contact of mineral particles) under the THMC effect.

Fig. 15

Permeability evolution of rock sample CA-2 with increasing number of injection cycles

Figure 16 is a schematic diagram of the variations in the hydraulic fracture aperture in the presence of a proppant before and after pressure dissolution. The contact form of the fracture surface changed from particle–particle contact to particle-proppant-particle, particle-proppant-free surface, and free surface-proppant-free surface contacts.

Fig. 16

Schematic diagram of the variations in the hydraulic fracture aperture in the presence of a proppant before (left) and after (right) pressure dissolution

When the pressure did not exceed the proppant’s failure strength (about 52 MPa), the first two contact forms were more likely to cause pressure dissolution at the proppant-particle contact positions, resulting in a small reduction in the fracture aperture. For the third contact form, the pressure dissolution on the free surface was not obvious because the granite’s strength was slightly higher than that of the proppant. Under certain conditions, the pressure may be higher than the failure strength of the proppant. At this time, the proppant is gradually damaged, resulting in a significant reduction of the fracture aperture and thus a decrease in the permeability.

In the presence of a proppant, the permeability evolutions of rock samples CA-3 and CA-4 with increasing number of injection cycles are shown in Fig. 17. It can be seen that the permeabilities of the two samples still decreased. However, there was no order of magnitude change in the decrease in the permeability. Therefore, the ability to maintain the high permeability of the reservoir is significantly better than that without proppant. Moreover, the permeability had a relatively small reduction rate during injection cycles 0–2 for rock sample CA-3 and during injection cycles 0–3 for rock sample CA-4. Therefore, during the initial injection cycle stage, the addition of proppant reduced the asperity contact portions on the fracture surface, which in turn significantly reduced the pressure dissolution effect. However, as the number of injection cycles increased (injection cycles 2–4 for CA-3 and injection cycles 3–5 for CA-4), the convex particles in contact with the proppant on the fracture surface disintegrated under the pressure, resulting in a small decrease in the fracture aperture and a relatively significant decrease in the permeability.

Fig. 17

Permeability evolution of rock samples CA-3 and CA-4 with increasing number of injection cycles

During the subsequent injection cycles, the reduction rate of the permeability of rock sample CA-3 gradually tended to zero, while that of rock sample CA-4 initially decreased in injection cycles 5–6, and then increased significantly in injection cycles 6–8. These results show that the particle-proppant-free surface contact form dominated the fractures in rock sample CA-3, while the particle-proppant-particle contact form was dominant in rock sample CA-4. In addition, it was also found that neither condition (with and without proppant) resulted in a significant decrease in the permeability when the pressure dissolution of the fracture surface stabilized.

Conclusions

In a CO₂-EGS reservoir, CO₂ (sc) injection results in CO₂-water–rock interactions and generates the mineralization storage of CO₂. The coupled THMC effect during this process has a great influence on the permeability. In this study, CO₂ storage and heat extraction were simulated by alternating cyclic injection of water and supercritical CO₂ into fractured granite. The influence of the coupled THMC effect on reservoir permeability was studied, and the mechanism of the above effect on the reservoir permeability with and without proppant was analyzed. The following conclusions were obtained:

• (1) As the number of injection cycles increased, the minerals dissolved on the fracture surface in the granite were still mainly feldspars including albite, anorthite, and potassium feldspar. As the ${\text{Na}}^+$ and ${\text{K}}^+$ contents increased, a small amount of biotite was also dissolved. However, in addition to cancrinite, dolomite, kaolinite, montmorillonite, calcite and illite, the newly formed minerals included monticellite and vaterite (a thermally unstable phase of CaCO₃) as well after increasing the number of injection cycles. It promotes mineralization storage of CO₂ but reduces reservoir permeability. This indicates that the duration of the CO₂-water–rock interactions had a significant influence on the form of the mineral dissolution and precipitation. Therefore, selecting a reasonable CO₂ (sc) injection pressure and injection flow rate can shorten the above process and reduce the influence on the fracture’s permeability.

• (2) After the dissolution of the primary mineral base, owing to the difference in the CO₂ flow velocity and the effect of the residual trapping in the micropores/microfractures, the type of mineral precipitation varied with the positions of the fracture surface. Highly crystalline rod-like calcite and hairball squama montmorillonite were formed in the relatively clear and smooth microfractures and micropores, while highly crystalline granular or rod-like calcite crystal clusters were formed near the microfractures. Also, on the main fracture surface and between the narrow minerals, highly crystalline rod-like calcite crystal clusters and other blocky carbonate mineral crystallization were formed. These carbonate crystallizations formed by mineralization storage of CO₂ significantly affect porosity and permeability of fractured granite.

• (3) In the absence of the proppant, the pressure dissolution had a significant impact on the fracture surfaces in the initial stage, and the permeability decreased sharply. As the number of the injection cycles increased, the asperity contact particles on the fracture surfaces were dissolved and disintegrated continuously, which caused the reduction rate of the permeability to exhibit a sharp-slow–fast–slow trend. While in the presence of the proppant, the contact form on the fracture surfaces was changed, which decreased the effect of the initial pressure dissolution. In the middle stage, the mineral particles on the fracture surfaces in contact with the proppant dissolved under the pressure, leading to a small decrease in the permeability. However, compared to the condition without proppant, the use of proppant caused an increase of approximately two orders of magnitude in permeability. Therefore, increasing the non-contact area of the main fracture can avoid a large decrease in permeability, which will increase the heat extraction and mineralization storage of CO₂.