Review Article Review of the Generation of Fractures and Change of Permeability due to Water-Shale Interaction in Shales

In shale development, water-based liquids are injected into the formations. In this process, water can interact with shales, especially with clay content. The interaction can lead to some phenomena, including clay swelling, reduction of mechanical properties of shales and fractures, generation and propagation of fractures, particle detachment, and permeability change. All the phenomena can impact productivity during the development, thereby impacting our investment and return on investment (ROI). So far, many researchers have put their time and efforts into this topic, and many articles have been published. However, some discrepancies still exist in shale reservoirs regarding the role of the interaction between water and shale, especially the impact of clay swelling. Some believe that clay swelling causes formation damage, mainly impairing shale permeability. Others state that fractures can be induced because of clay swelling, leading to the enhancement of shale permeability. So far, few articles have reviewed the various views on this interaction. Additionally, the relationship between each phenomenon is not discussed. In this paper, we try to draw a clear picture of water-shale interaction by reviewing the published studies, mainly focusing on experimental methodology and experimental results. Based on the review, we summarized the influencing factors as well as the mechanisms about the formation of fractures and change of permeability due to water-shale interaction. In water-shale interaction, the induced fractures are generated by the combined effects from clay swelling, reduction of mechanical properties of shales and fractures, and stress anisotropy. Shale permeability can be enhanced if the generated fractures can form an effective flow channel. However, if the generated fractures cannot serve as an effective flow channel, shale permeability will be impaired by clay swelling, water blocking, stress-sensitive, etc.


Introduction
The unconventional resource is becoming more and more important in the USA. The unconventional oil from tight oil reservoirs is estimated to exceed 12 million barrels per day [1]. Shale is an important resource in unconventional plays. Lots of companies have great interests and invest plenty of resources into developing shale. Shale is low permeable; hydraulic fracturing is the key to improve shale productivity [2]. Water is the major component of the fracturing fluid [3], and less than 10% of fracturing fluids are recovered [4]. Besides, aqueous solutions, working as EOR methods, are injected into shale plays to enhance oil recovery [5] Once aqueous solutions are injected into shales, water can interact with shale, which cause clay swelling, changing of mechanical properties of shales, and changing of petrophysical properties, thereby impacting oil productivity and oil recovery.
Clay swelling commonly occurs in water-shale interaction. It is widely accepted that clay swelling causes the impairment of formation permeability in conventional reservoirs [6][7][8]. However, many experimental results show that clay swelling can induce fractures in shales [9][10][11][12][13][14][15]. The generated fractures may result in permeability enhancement. No articles thoroughly explain the role of water-shale interaction in the development of shales so far. This paper was aimed at filling this research gap by summarizing the research results and analyzing the impacting factors.
Changing of mechanical properties of shales and fractures is another result of water-shale interaction. Due to water imbibition, the strength of shale can be weakened [16][17][18]; thus, fractures can be induced more easily. Besides, water-shale interaction can weaken the mechanical properties of fractures [19][20][21][22][23]. Studies show that stress intensity factor, crack extension force, and subcritical fracture growth index (SCI) can be impacted by clay swelling and watershale interaction [19,20]. The change in the mechanical properties of fractures is beneficial for fracture generation, as well. However, no articles have summarized those changes and discussed the effect of those changes in waterrock interaction.
Water-shale interaction can alter the permeability dramatically by the generation of fractures [12,24,25], particle detachment [11,12], plugging of flow channels [26], and pore spaces [6], as well as water blocking [24]. We review previous works on the effect of water-shale interaction on permeability change in this paper. The experimental methodologies that were used in the published articles are discussed. The advantages and disadvantages of those experimental methodologies are analyzed. Besides, the relationship between the generated fractures, particle detachment, and permeability change is stated. The factors that could influence particle detachment and permeability change are summarized.
The rest of this paper is extended in four parts: Part 2 briefly introduces the characteristics of shales, clay minerals, and clay swelling. Part 3 discusses the role of water-shale interaction in fracture generation and the possible mechanisms. Part 4 discusses the role of water-shale interaction in permeability change and the possible mechanisms. Part 5 concludes this paper and provides a further experimental approach.

Characteristics of Shale, Clay Minerals, and
Clay Swelling 2.1. Characteristics of Shales. Shale belongs to sedimentary rock and is most abundant on earth [27]. Shale is characterized as layered, fissile fine-grained, heterogeneous, and anisotropic. Shale's mineralogical composition controls the lithological properties of shale [28]. The typical minerals that are usually found in shales include clay minerals, quartz, feldspars, and carbonates [29,30]. Organic matter is also a common component for shales [31,32]. Clay minerals form the load-bearing framework in shales [33,34]. Clay minerals are the key factor in the water-shale interaction and will be detailed discussed in Subsection 2.2. Shale has a low permeability, ranging from nano-Darcy to micro-Darcy [35,36].
To accurately evaluate the permeability of shales is difficult. Many factors can impact permeability, like porosity, fluid viscosity, grain size and shape, and tortuosity [37]. Natural fractures and beddings largely exist in shales [38][39][40]. Therefore, laboratory-measured permeability underestimates the reservoir permeability in shales [41]. How to accurately measure the permeability of shales is an interesting topic and worthy to put efforts on. In this review paper, we will only focus on the change of permeability due to watershale interaction.

Clay Minerals in Shales and Clay
Swelling. Clay is a layered silicate mineral called phyllosilicate. Typically, clays are fine crystalline particles with two-dimensional arrays of silicon/oxygen tetrahedra or aluminum (or magnesium)/ oxyhydroxy octahedra. There are five sorts of clays: illite, montmorillonite, chlorite, kaolinite, and attapulgite [42]. Most often, based on properties, crystal structure, and capacity to accommodate water within clay structure, clay minerals are divided into three main groups: smectite, illites, and kaolinite [43]. Montmorillonite is included in the smectite family [44]. Based on the swelling potential, smectite, illite, and kaolinite are marked as high, moderate, and low swelling potential clays [43]. The swelling potential is the potential volume change for clay minerals. Clay mineralogy plays the most important role in volume change, even in small fractions [45]. Clay can expand up to 20 times its original volume [46]. The process by which clay minerals in shale absorb water is called clay hydration [47]. For most studies, clay hydration is clay swelling. Simply put, clay swelling is the result of increased space between the layered structure of clay minerals and the adsorbed cations [48,49]. Clay swelling consists of two stages: intracrystalline swelling and osmotic swelling [49]. The hydration of exchangeable cations in the interlayer space leads to internal crystallization, resulting in enhanced space between the clay layers (as shown in Figure 1). Concentration differences between ions in solution and ions in the space between clay layers cause osmotic swelling (shown in Figure 2).
Clay swelling can be affected by initial water content, adsorbed water, clay fraction, and confining pressure. Swelling occurs only when the water balance inside the shale is disturbed when it is in contact with the fluid. Studies show that dry shales are more reactive to water than water-saturated shales [50,51]. The initial water content in shales can influence swelling rate as well as swelling potential. A study by Chenevert [52] showed that the swelling rate decreased sharply with increasing water absorption time. The highest expansion rates were recorded in the first few hours. Studies conducted by Al-Mhaidib and Al-Shamrani [53] and Sabtan [54] showed that swelling volume and swelling ratio decreased with increasing initial water content. Gomez-Gutierrez et al. [55] stated that the more water adsorbed, the larger the swelling volume, which was supported by Bryson et al. [56]. For here, the samples were not confined. Sabtan [54] measured the free swelling potential for 30 clay shale samples with different clay fraction. In his study, swelling volume increased as the clay fraction increased. A similar trend was achieved by Gomez-Gutierrez et al. [55] and Bryson et al. [56]. Confining pressure could negatively influence the swelling volume of clay minerals [57]. To combine those influencing factors, Lyu et al. [58] developed an MLR model to predict clay swelling. The relationship between swelling potential (S), water content (W), clay fraction (C), and confining pressure (P) can be expressed as S = 30:0247 − 0:274 W + 0:0455C − 9:1778 log ðPÞ.

Geofluids
The temperature could influence clay swelling, as well. Wang et al. [59] and Huang et al. [60] employed a temperature range from −10°C to 23°C to investigate the swelling of shales. Their results showed that when the temperature is lower than 10°C, the expansion potential of shale decreases with the increase of temperature. The temperature could positively influence the swelling volume of shales when the temperature was higher than 10°C. Li et al. [61] studied the effect of temperature change on fracture generation in water-shale interaction. They compared the CT images for the tests under 80°C with those under 20°C. They found that fractures were more likely to develop in the tests using 80°C. The combined effects from high temperature and watershale interaction led to the propagation of fractures. Additionally, as cool water is injected into hot formations, the water-cooling effect can induce fractures by the increase in pore pressures and the shrinkage of formations [62,63]. The thermal effect is beneficial to the generation of fractures in water-shale interaction. However, so far, no experiments have been conducted to support this statement when samples are under stress anisotropy.
The types and concentrations of salts in water-based liquids also have an impact on clay swelling. The ions of Na + , K + can influence the clay swelling. KCl is a commonly used swelling inhibitor [64,65]. Wong [18] used water and brine (1% NaCl solution) to test the swelling potential of La Biche shale samples. He found that clay swelled more in the water when compared with brine.

The Influence of Water-Shale Interaction on Fracture Formation and Propagation
This section introduces the experimental methodologies and results related to fracture formation and propagation due to water-shale interaction. The interaction-induced fractures and the testing conditions are discussed. Based on those experimental results, the mechanisms about the role of water-shale interaction in fracture formation and propagation are discussed.

Induced Fractures in Experimental Studies.
To study the interaction between water and shale, spontaneous imbibition tests were carried out. Figure 3 shows the imbibition tests performed by Morsy and Sheng [66]. Other researchers conducted the similar tests [9,[66][67][68][69][70][71][72][73][74][75]. From their results, the imbibition process was controlled by capillary actions, osmotic fluid flow, and clay minerals. To represent the imbibed water, the weight difference was used in most tests. Dehghanpour et al. [9] conducted imbibition tests for samples from five different shale formations. In their tests, they compared the imbibition profiles of oil and brines. The brine intake of all samples was significantly higher than the oil intake of the same samples. Besides, they found that excess water intake was due to water adsorption by clay minerals by comparing the inhalation profiles of oil and brine with oil intake. The water-shale interaction-induced microfractures even disintegration were observed in their tests. Samples from Fort Simpson had the highest clay content; therefore, disintegration was observed; samples from Otter Park had the least clay content, and only microfractures were observed. Figure 4 shows the generated fractures in the tests which were conducted by Morsy et al. [71] and Morsy and Sheng [66]. From their results, clay swelling can induce fractures. The above results indicate the following: (1) clay content is the main reason for the water uptake in the tests; (2) clay swelling can generate fractures in the imbibition tests; and (3) whether or not disintegration can occur is related to the amount of clay content in the samples. However, those tests were conducted without any restrains, and the mechanisms behind the generation of fractures were not discussed in these studies.
To examine the role of confining pressure in fracture generation because of clay swelling, several researchers introduced confining pressure into their tests [11,25,[76][77][78][79]. In their tests, a core holder capable of holding the core sample and a pump for applying confining pressure are required. Figure 5 shows a schematic diagram of the      [76] and Wang et al. [25]. In their tests, the sample can imbibe water from the open ends of the core holder.
To observe the generated fractures, a CT scanner was employed by Bin et al. [76] and Wang et al. [25]. The general test procedure is that (1) the core samples are scanned before the tests; (2) the sample is placed in the core holder and exposed to confining pressure for imbibition tests; and (3) after imbibition tests, pressures are released, and the samples are scanned again to observe the generated fractures. For this experimental methodology, elastic strain energy is stored inside the samples as long as the confining pressure is applied. The release of confining pressure can result in a sudden release of strain energy, thereby causing the generation of fractures or even the failure of samples [80][81][82]. In their tests, it is hard to tell whether the fractures are generated by water-shale interaction or the sudden release of strain energy. It is better to monitor the process of the generation and propagation of fractures when the water imbibition tests are running under confining pressures. Two methods were employed: (1) first is monitoring the pressure change in the tests [11]. In their tests, A shale sample is exposed to a designed upstream pressure of water under confining pressure. Downstream pressure is maintained at atmospheric pressure. The upstream pressure is recorded. A pressure drop can be achieved if the fractures are generated. (2) Second is using an X-ray core holder system [77][78][79]. Figure 6 shows the schematic of the apparatus in their tests. CT images are achieved without releasing the confining pressure. In these ways, the effect of the sudden release of strain energy on fracture generation can be eliminated.
Bin et al. [76] and Wang et al.'s [25] results show that fractures were induced and developed (shown in Figure 7). In Bin et al.'s [76] results, fractures grow and connect to form complex fracture networks (shown in Figure 7(a)). In Wang et al.'s [25] tests, both isolated and connected fractures were observed on the CT images. In both studies, the reason for the generated fractures was clay swelling due to water-shale interaction. However, based on the description of their experimental procedures, both studies dried the cores and scanned the samples to get the CT images. Once the confining pressure is released, the CT scanning cannot accurately show the welling induced fractures. The effect of the sudden release of confining pressure, which caused the sudden release of strain energy inside the samples, cannot be differentiated from the effect of clay swelling. Therefore, it is difficult to identify whether the generated fractures were caused by clay swelling or by the sudden release of confining pressure. In Bin et al.'s [76] study, another set of tests was performed to characterize the distribution of pore-fracture structures in the samples during imbibition testing. The distribution of T 2 spectra was analyzed. The longer the transversal relaxation time, the larger the size of the porefracture structure should be. From Figure 8, the peak value of the transversal relaxation time increased in their tests, which means the size of the pore-fracture structure increased due to water-shale interaction. Combined with the results from CT images and those from the T 2 spectra, it is found that under the action of confining pressure, the expansion of clay can produce fractures.
Roshan et al. [11] explored the underlying mechanism of water uptake by using partially saturated samples to conduct tests. Free and confined water imbibition tests were done. For the confined test, a shale sample was placed under confining pressure of 1000 psi. The sample was exposed to distilled water by upstream pressure of 500 psi for 40 h. Figure 9 shows the induced fracture in the test. The fracture was formed by taking a longer time under confining pressure when compared with the tests without restrain. They stated that the formation of fractures under confining pressure is due to the decrease in the mechanical strength of the rock and the swelling of the clay. Besides, they indicated that this fracture was formed along with weak structures (beddings and laminations).
By comparing the results from Bin et al. [76], Wang et al. [25], and Roshan et al. [11], we can see that fractures can be generated due to the effect of water-shale reaction in shales under confining pressure. However, the formation of such fractures may take a longer time. The main factor that can contribute to the generation of such fractures is clay swelling. Other factors, like the degradation of the mechanical strength of shale, can also be beneficial for fracture generation in water-shale interaction.
Zhang and Sheng [77,78] and Zhang et al. [79] employed CT images to observe fractures. Isotropic confining pressure was applied in their tests. Samples from Mancos shale formation were used. The constant pore pressure (0.03 MPa) was used for all the tests, and confining pressure (0.1 MPa, 2.0 MPa, and 20.0 MPa) was applied for the tests. Figure 10 shows the CT images from Zhang and Sheng [78]. They observed that at lower confining pressures, the samples developed more fractures. The generated fractures under confining pressures could close and reopen, which means the generation of the fractures was largely impacted by water-shale interaction and the applied confining pressure. Besides, Zhang and Sheng [78] found that swelling strain decreased due to confining pressure. The larger the confining pressure, the more the swelling strain was reduced. Ewy and Stankovic [83] reached similar results: increasing confining pressure leads to a decrease in swelling. Sufficient confining pressure could prevent clay swelling from occurring. A threshold confining pressure exists. Above this threshold, swelling decreases with the increasing confining  4 Geofluids pressure or is prevented by confining pressure completely. The threshold for confining pressure changes from shale to shale. In conclusion, in water-shale interaction, confining pressure has a negative impact on fracture formation. In fracture formation, stress anisotropy plays a positive role. Liu and Sheng [10], Liu and Sheng [12], and Liu et al. [13] investigated the role of clay swelling in fracture formation in shales with stress anisotropy. In their tests, stress anisotropy is applied using anisotropic core support. The schematic of the test system is shown in Figure 11. The results indicate that clay swelling and stress anisotropy can facilitate the fracture initiation and development when shales meet water. More fractures can be formed with larger stress anisotropy (shown in Figure 12).
Liu and Sheng [10] also observed the dynamic changes of the generated fractures in water-shale interaction under stress anisotropy. They and Zhang and Sheng [78] reached the same conclusions: (1) water-shale interaction does have a positive effect on fracture formation for samples; (2) fractures can be induced much easier in water-shale interaction when samples are under stress anisotropy.

The Influencing Factors for the Generation of Fractures in Shales.
Wang et al. [84] found that clay mineral types and fractions are critical for water-shale interaction. Shale reacts with water stronger with more clay minerals. Clay swelling can change the stress distribution, leading to a stress concentration around fracture tips, thus causing the fractures to grow. This statement is supported by the simulation work done by Liu et al. [13]. Clay swelling can cause the redistribution of the stress in the models. Stress concentration was observed around the generated fractures to cause the generation and propagation of the fractures.
Natural fractures are everywhere in shales [39,40,85]. Natural fractures are suspected as a factor in productions in shales. Additionally, natural fractures serve as the sites for fracture creation and propagation in shales [10,39,86,87]. The locations of natural fracture also can impact the fracture formation and distribution [86]. Natural fractures can facilitate the formation and growth of new fractures in water-shale interaction (shown in Figure 13).
Bedding and lamination are ubiquitous in shales [40]. Roshan et al. [11] observed that the induced fracture was formed along with the bedding (shown in Figure 9). Moradian et al. [88] also observed that if the bedding directions are 60°and 90°, fractures with bedding directions were dominant. This result is same as the results from Liu and Sheng [10] and Makhanov et al. [68]. Water absorption parallel to the lamination is higher than perpendicular to the same lamination [89]. The enhancement of the imbibition rate was caused by the generation of fractures. This observation indicated that lamination was beneficial for the generation of fractures in water-shale interaction. By those results, bedding and lamination is another positive factor to influence the generation and propagation of fractures in water-shale interaction.

Water entrance Water entrance
Core holder Sample Pressure pump Figure 5: The schematic of the apparatus that was used by Bin et al. [76] and Wang et al. [25] to conduct imbibition tests with confining pressure.    Figure 9: Testing results from [11]. 6 Geofluids atmospheric conditions. They found that the generation of fractures mainly occurred in the areas between organic matter and inorganic minerals. The cohesion between the mineral particles could be weakened after the water-shale interaction. The nonclay mineral particles are exfoliated to form inorganic pores, which gradually develop into microfractures between the nonclay mineral particles and the clay mineral particles. Li et al. [92] observed that microfractures were generated along with the interfaces between organic and inorganic matter as well as between different components. Similar results were observed by Wang et al. [84]. Organic matter can provide space for fracture development caused by clay swelling and capillary pressure [93]. Organic matter is another positive factor for the generation and propagation of fractures in water-shale interaction. The elasticity, hardness, and strength of shales are deteriorated, which is termed as "shale softening." The properties of fractures, like fracture conductivity, stress intensity factor ðK c Þ, crack extension force ðG c Þ, and subcritical fracture growth index (SCI), are reduced.
Shale's mechanical strength can be weakened under the effect of water-shale interaction [16][17][18]. Cheng et al. [16] compared the shale strength of dry and water-saturated samples. The results indicated that the compressive strength, Young's modulus, and the anisotropy of the water-saturated samples were reduced compared to the results of the dry samples (as shown in Figure 14). The average compressive strength of water-saturated samples was 28.9% lower than that of the dry samples, with a maximum decrease of 54% for water-saturated samples. Likewise, the mean and maximum reductions in Young's modulus of the watersaturated samples were 26.1% and 62%, respectively. Two reasons why we need to compare Young's modulus are as follows: (1) the brittleness of shales can be used to determine the ability to be fractured of the shales [94], and brittleness is Axial pressure Figure 11: The schematic of the test system (Liu and Sheng [10] and Liu and Sheng [12]. 7 Geofluids a function of Young's modulus. The shale is more brittle with a higher Young's modulus. Brittle shale has a higher potential for fracture initiation and propagation and the for-mation of a complex fracture network than ductile shale. (2) Decreased Young's modulus leads to deterioration of fracture conductivity [95]. Once the compressive strength of  Figure 13: Results for EF-6. The testing condition was 1350 psi axial and 500 psi confining pressure [10]. 8 Geofluids the shales is weakened by water-shale interaction, it is easier to meet the failure criterion and fractures could develop in shales. The results from Wong [18] indicated that Young's modulus decreased with increasing swelling. Besides, the behavior of shale under larger swelling was like a ductile material. Similar results were achieved by Yang et al. [96].
Talal [17] compared the effect of Na + , K + , Ca 2+ on the compressive strength of shales. The results indicated that K + had a strengthening effect on shale strength, while Na + and Ca 2+ ions had a weakening effect on shale strength. Akrad et al. [95] measured Young's modulus of shales before and after exposure to 2% KCl slickwater and freshwater. The results showed that potassium ion ðK + Þ caused the reduction of Young's modulus of shales, no matter it was "soft" mineral with Young's modulus below 30 GPa or "hard" mineral with Young's modulus above 30 GPa. The highest reduction occurred in Eagle Ford shale (70%), which resulted in a 39% loss in fracture conductivity. 52% of the reduction was observed for Bakken shale, causing a 14% loss in fracture conductivity. For these two shales, high calcite (77%) and low clay content (8% for Eagle Ford and 4% for Bakken) were measured. For the clay-rich samples, like Lower Bakken (clay content of 47%) and Haynesville (clay content of 57%), a lesser decrease was observed in Young's modulus (22% for Lower Bakken and 6% for Haynesville), resulting in a lesser reduction in fracture conductivity (5% for Lower Bakken and 1% for Haynesville). These results indicated that clay swelling might not cause permeability impairment as bad as assumed. The reduction of fracture conductivity was also reported by Pedlow and Sharma [97] and Jansen et al. [98]. Li et al. [92] examined the effect of water-shale interaction on the tensile strength of shales. Tensile strength is critical for predicting fracture initiation and growth. The smaller the tensile strength, the easier the fractures can be generations. As shown in Figure 15, as the water content increases from 4.45% to 11.7%, the water-shale interaction reduces the tensile strength by 4.4% to 51.7%, which means that fractures are more likely to develop due to the decrease in tensile strength. In conclusion, the change of the mechanical properties of shales due to water-shale interaction can facilitate fracture generation in shales.
The properties of fractures are affected by water-shale interaction. Based on the equilibrium law, stress intensity factor (K c ) or crack extension force (G c ) has a critical value.
Once those values are reached or exceeded, the fracture can propagate [99]. As stress corrosion occurs, fracture can develop when the K or G is far below the critical values, which is known as subcritical fracture growth [99]. Stress corrosion means that the strained Si-O bonds at crack tips are more reactive to environmental agents than the unstrained bonds of crystalline silicates because of the reduced strain-induced overlap of atomic orbitals [99]. A weakened state is produced due to strained bond-environmental agent reaction, resulting in that the bonds can be broken at a lower stress than the unstrained bonds. A general expression to represent the weakening effect of silicate and quartz in water is proposed as the following [100][101][102][103][104]:

Geofluids
It is first-order chemical reaction when strong Si − O bonds hydrolyze to weaker hydrogen-bonded hydroxyl groups attached to silicon atoms. Thus, water-shale interaction can result in stress corrosion.
Chen et al. [19] investigated the change of fracture mechanical properties in clay-rich shales under water-shale interaction. Their results (shown in Figure 16) indicated that even at low smectite content, clay-rich shales had a strong water-weakening effect, resulting in a 54% reduction in fracture toughness and 77% reduction in subcritical fracture growth index for water-saturated samples. The carbonaterich shale barely had water-weakening effects. Besides, K IC for wet samples was reduced by more than 50% compared to dry samples for clay-rich shales. For samples without clay minerals [105], K IC dropped less than 20% with increasing water saturation. The reduction in subcritical fracture growth index in clay-rich samples could reach 77%, while the reduction in subcritical fracture growth index for samples without clay minerals was negligible or less than 50% [21][22][23].
Their findings suggest that water-shale interactions lead to the weakening of the subcritical fracture properties of clay-rich shale, which may lead to subcritical fracture growth and failure more prone to clay-rich shale than clay-poor shale.
Chen et al. [20] conducted comparison work for Woodford shale, Mancos shale, and Marcellus Shale with different temperatures, pH, and varying fluid salinities. The generated fractures were traced by scanning electron microscopy (SEM) (shown in Figure 17). The clay fabrics and grain boundaries are shown in Figure 18.
It can be seen from Figure 17 that intergranular fractures are generated in the clay-rich shale, forming zigzag traces at the grain scale. As can be seen in Figure 18, clay fabrics and grain boundaries and pits are observed on the fracture surfaces of the Woodford and Mancos Shales (red arrows in Figures 18(b) and 18(d)). The fracture trace observations were compared with the clay structure and grain boundaries, and the fractures were formed by clay swelling.
According to Chen et al. [19], Chen et al. [20], Nara et al. [21,22], and Waza et al. [23], two points are reached: (1) clay swelling is primarily responsible for the generation of the fractures in water-shale interaction; (2) the reduction in stress intensity factor (K c ), crack extension force (G c ), and subcritical fracture growth index (SCI) can facilitate the generation and propagation of fractures in water-shale interaction. However, those tests were conducted without restrains.
Through the comparison, the degradation of mechanical properties of shales and fractures due to water-shale interaction has a positive effect on the generation of fractures in shales.

Mechanisms about the Generation and Development of
Fractures in Water-Shale Interaction. The initiation of swelling-induced fractures is due to swelling pressure [106], which is the difference in the hydrostatic pressure of the water inside and outside of the clay membrane. Once the swelling pressure is large enough, it can break the natural cementation of shale and allow the generation of swellinginduced fractures. A similar explanation is stated by Steiger [107] which suggests that interactions between water and  The fractures induced by the water-shale interaction under stress conditions are more likely caused by the combined effects from clay swelling, reduction of mechanical properties of shales and fractures, and stress conditions. The initiation of fractures can be triggered by clay swelling through the formation of new pores and microfractures. Meanwhile, clay swelling can induce nonuniform stress, resulting in the local stress concentration around fractures. The reduction of mechanical properties of shales and fractures can facilitate the generation of fractures by lowering the fracture mechanical properties. Fractures can propagate under the combined effect of clay swelling, stress anisotropy, shale, and reduced mechanical properties of fractures. The characteristics of shale natural fractures, bedding, and organic and inorganic matter interface can be used as the place for the generation of induced fractures. In this process, clay swelling, redistribution of stress, and the reduction of mechanical properties of shales and fractures are caused by water-shale interaction. Therefore, water-shale interaction is critical in the formation of fractures under stress conditions. However, new experimental methodology or even new apparatus is needed to quantitatively reflect the contribution of water-shale interaction in the generation of fractures in shales.
Another mechanism that may induce fractures in watershale interaction is mechanical failure. The mechanical failure is caused by increased pore pressure caused by the  [110] found that although oil had no affinity for absorption in clays, some microfractures still were created with oil adsorption. This result revealed that the pore pressure due to fluid imbibition reduced the effective stress and thus created fractures. This kind of fractures is generated by mechanical failure of samples. A similar conclusion was reached by Santos et al. [51]. However, it is difficult to tell the amount and importance of the fractures caused by mechanical failure due to fluid imbibition. Besides, it is hard for us to separately study the effect of mechanical failure due to fluid imbibition in water-shale interaction. The reason is that once the fluid is imbibed into samples, the fractures caused by fluid-shale interaction and fractures caused by the mechanical failure may occur at the same time, and we cannot tell whether the fractures are caused by the interaction or mechanical failure.

The Change of Permeability due to Water-Shale Interaction
In this part, the permeability change due to water-shale interaction in the experiments is summarized. Based on those experimental results, the mechanisms about the change of permeability due to water-shale interaction are analyzed.

Experimental Methodologies That
Were Used and the Experimental Results. Most researchers believe that watershale interaction causes formation damage, mainly permeability impairment in shales; however, many studies showed that induced fractures were formed in water-shale interaction. Unlike conventional reservoirs, enhancing oil productivity in shales mainly relies on the conductivity of fractures, even though the conductivity of such induced fractures is lower than that of propped fractures in hydraulic fracturing [98,111]. The matrix permeability and conductivity of natural fractures can be damaged due to the interaction between water and shale, especially clay swelling [47,112]; however, shale permeability can be enhanced by the induced fractures [9,11,24]. What is the final effect on permeability change is an interesting question. Based on the methodology, the permeability tests can be divided into two types: gas permeability measurement during water imbibition and pressure monitor during water imbibition. The gas permeability measurement during water imbibition was popularly used by lots of researchers [6,24,25,75,77]. The general testing procedure is the following: the sample is exposed to water-based fluid initially. After a certain period, the sample is removed from the fluid and the gas permeability is measured. Then, the sample is resumed to the fluid to continue the imbibition test. For this method, this procedure may be repeated several times to achieve the gas permeability for different time spots in the fluid imbibition tests. In the studies from Aksu et al. [6], Shen et al. [24], and Zhou et al. [75], they did not mention clean the samples before measuring permeability. In Wang et al. [25], they dried the core samples before measuring permeability. In Zhang and Sheng [77], they vacuumed the sample before measuring the permeability. This methodology can quantitatively reflect the effective gas permeability change in the tests. However, gas permeability cannot truly represent absolute permeability in shales. Besides, different methodologies are employed to measure shale permeability [113]. Pressure monitor during water imbibition was not as widely used as gas permeability measurements [11,12]. For this methodology, the upstream pressure or pressure difference is monitored in the tests. The increase of pressure can reflect the enhancement of shale permeability; otherwise, the impairment of shale permeability can be shown by the decrease of pressure. This methodology can indirectly and qualitatively reflect the change of shale permeability. However, it is still unsolved to quantitatively reflect the change of shale permeability in the tests. Permeability is a function of multiple factors, like stress, pore throat, pore size, and liquid viscosity. This research gap needs to be filled.
Shen et al. [24] measured the permeability change in water imbibition tests for sandstones and shales. The pulse-decay permeability methodology was used to measure samples' permeability. Before they conducted imbibition tests, permeability for dry samples was measured. Then, the imbibition tests were conducted. After a certain period of imbibition tests, the imbibition cells were removed from the imbibition liquid, and permeability tests were conducted as soon as the cells were removed. Figure 19 shows the permeability change vs. the increased water content in their tests. In the early stage, the decrease of shale permeability (Y1, Y2, and Y3) was due to the reduction of effective flow channel caused by water blocking and stress sensitivity. It means that the applied stress conditions have an impact on the permeability change [24]. At the middle stage, samples' permeability (Y1, Y2, and Y3) increased due to the generation and propagation of new induced fractures caused by clay swelling. Shale permeability (Y1, Y2, and Y3) decreased, at the last stage, which is due to water blocking and stress sensitivity. A failure occurred in the sample Y4, resulting in a permanent increase in the sample's permeability. The failure was caused by clay swelling. They stated that the three critical factors to control shale permeability are clay swelling, stress sensitivity, and water blocking. Whether the permeability can be increased depends on the comparison of these three factors.
Shales' wettability, hydration stress, and permeability change in slickwater treatment are studied by Yuan et al. [114]. The testing methodology is similar to Shen et al.'s [24]. The permeability was damaged by the narrowed flowing channels due to clay swelling at the beginning of the test. Under the effect of swelling stress and capillary pressure, fractures were induced. However, the fractures are not interconnected to form effective flow channels. Therefore, the contribution from these fractures to permeability was negative. As water-shale interaction continues, fractures propagated and interconnected because of the increased hydration stress and wettability change. Thus, shale permeability was recovered and even higher than the original value. A pressure build-up method was used by Zhou et al. [75] to measure the samples' gas permeability. They stated that matrix permeability and fracture permeability were 12 Geofluids decreased due to clay swelling and water blocking. Besides, they observed that the samples' permeability could be increased due to the generation of microfractures. In the pictures provided by them, the generated microfractures propagated through the core samples and formed an effective flow channel, thereby increasing the sample's permeability.
All the above three studies used effective gas permeability to represent the change of shale permeability. However, it is widely accepted that gas permeability is higher than liquid permeability [115][116][117]. Moreover, no accurate formula was proposed to represent the gas and liquid permeability correlations for various shales. The measured gas permeability cannot correctly represent the permeability change in water-shale interaction. Roshan et al. [11] conducted a test under confining pressure (1000 psi). In the test, the injection pressure is 500 psi  Figure 19: (a) Change of shales' permeability in water imbibition tests [24]. (b) Change of shales' permeability in water imbibition tests [114]. 13 Geofluids and the injected fluid is DI water. The upstream pressure was recorded. No pressure change was observed until a fracture was generated (shown in Figure 9). Similarly, in their permeability test, they recorded the change of the upstream pressure in the imbibition test with the downstream pressure at atmospheric conditions. Figure 20 shows the pressure curve from their test. In the test using 10 wt.% NaCl solution, the upstream pressure was almost constant, whereas, in the test using DI water, the pressure changed from 8 psi to 7 psi. After close observation of the tested sample, they found that particle detachment by clay swelling occurred and caused this pressure drop.
To study permeability change due to water-shale interaction, one set of experiments was performed by Liu and Sheng [12]. In their tests, the upstream pressure was monitored to represent the change of permeability. Figures 21  and 22 show the experimental results.
From Figure 22, pressure data decreased once fractures were generated or propagated. Multiple fractures were generated and interconnected to form flow channels and cause the failure of the sample, resulting in the permanent recovery of shale permeability. In another test, an isolated fracture was generated and did not form an effective flow channel for DI water; thereby, the pressure increased in the test. The results indicated that if the fractures are isolated and could not form an effective flow channel, the fractures did not contribute to the recovery of shale permeability. However, once the fractures form effective flow channels, shale permeability can be recovered. It is seen from Figure 22 that particle detachment occurred on the artificial fracture surface. The detached particles were washed away by the DI water, enhancing the shale permeability.

4.2.
Mechanisms about Permeability Change due to Water-Shale Interaction. From those studies, shale permeability can be enhanced if the generated fractures can form an effective flow channel. However, if the generated fractures cannot serve as an effective flow channel, shale permeability will be impaired by clay swelling, water blocking, stress-sensitive, etc. Gas permeability cannot be used to represent the permeability change in water-shale interaction. Pressure monitor cannot quantitatively show the permeability change in water-shale interaction. It is unsolved to quantitatively reflect permeability change by water-shale interaction. More efforts are needed.
Particle detachment affects permeability changes. The permeability of the shale can be increased when the separated particles can be washed away from the shale. Grain   14 Geofluids detachment can lead to pore or pore-throat plugging problems, thereby impairing shale permeability. Mohan et al. [118] found that kaolinite and muscovite are more easily detached from the shale surface when swelling occurs. The effects of pH, ion exchange, and brine salt concentration can affect particle separation. Higher pH, lower ionic strength, and low salinity can facilitate separations [119][120][121][122][123]. Whether particle detachment is beneficial to shale permeability depends on whether the detached particles can be transported from the shale.

Concluding Remarks
Water-shale interaction is a complex phenomenon. This reaction can cause clay swelling, reduction of mechanical properties of shales, generation and propagation of fractures, Upstream pressure (psi) Constant flow rate of 2 (mL/min) Figure 22: Another testing result from Liu and Sheng [12].  particle detachment, permeability change, etc. Other conditions, like stress condition, shale characteristics, and reacting fluid, could influence the reaction. Figure 23 summarizes these phenomena and the relationship between each other.
Clay swelling is the key phenomenon occurring in watershale interaction. Clay swelling can trigger the generation of fractures, cause the detachment of particles, influence the magnitude of the reduction of mechanical properties of shales and fractures, and impact permeability change. Experiments showed that fracture can be generated in water-shale interaction. The most likely mechanism for the generation of fractures in water-shale interaction is that the induced fractures are generated by the combined effects from clay swelling, reduction of mechanical properties of shales and fractures, and stress anisotropy. However, other mechanisms can induce fractures as well, like the swelling pressure, the mechanical failure caused by the imbibition of water. Whether clay swelling can directly induce fractures under stress conditions is still questionable. To answer this question, a new experimental methodology and apparatus are needed. Ideally, the process of the generation of clay swelling inducing fractures under stress conditions should be monitored and captured in the reaction between water and shale. The effect of clay swelling and stress conditions should be separated. Measuring gas permeability cannot accurately reflect the permeability change by water-shale interaction. Using liquid to measure the liquid permeability is difficult because of the injectivity issue in low permeable shales. Pressure monitoring can qualitatively reflect permeability change in shales. However, it is better to quantitatively reflect the permeability change in watershale interaction under stress anisotropy. More efforts are needed.

Data Availability
Data are available by contacting the authors.

Additional Points
Highlight. (i) Water-shale interaction can cause clay swelling, change of mechanical properties of shales, generation of fractures, and permeability change. (ii) Fractures are generated by the combined effects from clay swelling, reduction of mechanical properties of shales and fractures, and stress anisotropy. (iii) Both the induced fractures and particle detachment can impact shale permeability. (iv) Stress condition, shale characteristics, and the reacting fluid could influence water-shale interaction.

Conflicts of Interest
The authors declare that they have no conflicts of interest.  Figure 23: Phenomena appearing in water-shale interaction, the impact between these processes, and the other influencing factors. 16 Geofluids