Triaxial Experimental Study of Natural Gas Hydrate Sediment Fracturing and Its Initiation Mechanisms: A Simulation Using Large-Scale Ice-Saturated Synthetic Cubic Models

Abstract

The efficient extraction of natural gas from marine natural gas hydrate (NGH) reservoirs is challenging, due to their low permeability, high hydrate saturation, and fine-grained sediments. Hydraulic fracturing has been proven to be a promising technique for improving the permeability of these unconventional reservoirs. This study presents a comprehensive triaxial experimental investigation of the fracturing behavior and fracture initiation mechanisms of NGH-bearing sediments, using large-scale ice-saturated synthetic cubic models. The experiments systematically explore the effects of key parameters, including the injection rate, fluid viscosity, ice saturation, perforation patterns, and in situ stress, on fracture propagation and morphology. The results demonstrate that at low fluid viscosities and saturation levels, transverse and torsional fractures dominate, while longitudinal fractures are more prominent at higher viscosities. Increased injection rates enhance fracture propagation, generating more complex fracture patterns, including transverse, torsional, and secondary fractures. A detailed analysis reveals that the perforation design significantly influences the fracture direction, with 90° helical perforations inducing vertical fractures and fixed-plane perforations resulting in transverse fractures. Additionally, a plastic fracture model more accurately predicts fracture initiation pressures compared to traditional elastic models, highlighting a shift from shear to tensile failure modes as hydrate saturation increases. This research provides new insights into the fracture mechanisms of NGH-bearing sediments and offers valuable guidance for optimizing hydraulic fracturing strategies to enhance resource extraction in hydrate reservoirs.

1. Introduction

Marine natural gas hydrate (NGH) deposits are primarily composed of clay, silt, and coarse sand, and are characterized by small media particle sizes and a high proportion of pore spaces, occupied by solid hydrates. These characteristics result in extremely low reservoir permeability, categorizing them as unconventional low-permeability gas reservoirs [1,2]. To address this production challenge, hydraulic fracturing has become a key stimulation technique and has been widely employed in enhancing the productivity of low-permeability reservoirs, such as shale gas and tight sandstone formations [3,4]. Applying hydraulic fracturing to NGH reservoirs aims to artificially create high-permeability flow channels, significantly improving the gas recovery efficiency and offering a promising direction for hydrate exploitation [5,6]. A series of meticulously designed indoor hydraulic fracturing simulation experiments have confirmed the feasibility of fracturing within hydrate deposits [7,8], yet a comprehensive understanding of the fundamental mechanisms governing fracture initiation and propagation in these complex media remains limited.

Studies reveal that hydraulic fractures in unconsolidated formations often result from a combination of fluid invasion and shear failure, leading to the development of multiple subparallel fractures [9]. Additionally, in formations with mud–sand interlayers, fracture propagation tends to follow the permeable interface, suggesting the potential for interface-controlled fracturing in hydrate-bearing heterogeneous systems [10]. For methane hydrate-bearing sediments, tensile failure has been confirmed as the primary failure mode, with fracture formation significantly enhancing permeability [11]. In silty hydrate sediments, both tensile and shear failure mechanisms coexist, producing complex fracture networks, whose propagation is not strictly governed by the in situ stress field [12,13,14].

Recent studies have focused on site-specific hydrate systems, such as the Shenhu area in the South China Sea, examining how factors like clay mineralogy, hydrate saturation, and in situ stress influence fracture behavior [15,16,17]. Complementary numerical simulations using the Discrete Element Method (DEM) and Cohesive Zone Models (CZMs) have provided insights into the roles of hydrate saturation, reservoir heterogeneity, and fracturing parameters in shaping fracture morphology and propagation [18,19].

Although significant progress has been made, conducting direct mechanical experiments on in situ hydrate samples remains highly challenging. This is primarily due to the stringent conditions required to maintain hydrate stability, including high pressure, low temperature, and high saturation conditions [20]. To address these challenges, ice has been widely adopted as a mechanical surrogate for hydrates in laboratory-scale simulations, due to its comparable crystallographic structure and mechanical behavior [21]. This surrogate provides a practical and effective means of investigating the mechanical response of hydrate-bearing sediments under well-controlled experimental conditions.

In this study, first we establish the scientific rationale for employing ice as a surrogate material for natural gas hydrates by summarizing their key similarities. Ice and hydrate share comparable crystallographic structures, mechanical strength, and deformation characteristics under similar loading conditions. Furthermore, both substances exhibit analogous morphological features in the way they occupy pore spaces within granular sediment matrices, making ice a reliable analog for simulating hydrate-bearing systems in laboratory environments. Building on this foundation, this study systematically investigates the fracturing behavior of hydrate analogs using large-scale, ice-saturated synthetic cubic samples. Through a series of controlled triaxial experiments, we simulate hydrate fracturing conditions, including variations in in situ stress, perforation geometry, and fracturing fluid properties. The experimental results are then analyzed in conjunction with theoretical models, based on elastic and plastic failure criteria, aiming to deepen our understanding of fracture initiation and propagation mechanisms in hydrate-bearing sediments. The insights gained from this study contribute to the development of more effective hydraulic fracturing strategies for the sustainable exploitation of hydrate resources.

2. Experimental Methods

2.1. Experimental Materials

Natural gas hydrates (NGHs) are crystalline compounds formed by water and hydrocarbon molecules, stabilized through van der Waals forces under low-temperature and moderate-pressure conditions. These structures are thermodynamically unstable at standard atmospheric temperature and pressure, making it difficult to synthesize and maintain NGHs within core samples under laboratory conditions [22,23]. As a result, a systematic investigation of the mechanical behavior of NGH-bearing sediments, particularly across a range of hydrate saturations, poses significant experimental challenges.

To address these limitations, ice is commonly adopted as a surrogate material in physical simulations of hydrate-bearing systems. This surrogate is grounded on two fundamental justifications: (1) the physical properties of ice are quantitatively similar to those of methane hydrate, and (2) both materials exhibit comparable morphologies in regard to their pore-scale distribution within sediment matrices. A detailed comparison of the relevant physical parameters is presented in

Table 1. Comparison of physical properties of ice and methane hydrate [25].

Item	Ice	NGH I	NGH II
Lattice diameter (0 °C)	4.52	11.97	17.14
Volume expansion coefficient (k−1)	1.5 × 10−4	1.5 × 10−4	1.7 × 10−4
Isothermal Young’s modulus at −5 °C (109 Pa)	9.5	8.4	8.2
Poisson’s ratio	0.33	0.33	0.33
Adiabatic bulk modulus at 0 °C (Pa)	12 × 10−11	14 × 10−11	14 × 10−11
Longitudinal sound velocity (km·s−1)	4	3.8	3.8
The enthalpy of formation of a crystal lattice compared to gas at 0 °C (kJ·mol−1)	−51.01	−50.2	−50.2
Lattice energy at 0 K (kJ·mol−1)	−47.3	Slightly lower than ice	Slightly lower than ice
Residual entropy at 0 K (J·kmol−1)	3.43	Slightly lower than ice	Slightly lower than ice
Density at 0 °C (g·cm−3)	0.912	0.91	0.883

. The data confirm that ice and NGH (both structure I and structure II) share similar mechanical stiffness, Poisson’s ratio, acoustic properties, and thermodynamic behaviors under cryogenic conditions. Furthermore, Figure 1 illustrates the morphological resemblance between ice-saturated samples and NGH-bearing sediments, especially in regard to how both materials fill pore spaces and bind sediment grains into a cohesive framework [24].

While these similarities provide a sound basis for using ice in mechanical simulation experiments, it is important to acknowledge the limitations of this approach. Ice does not replicate the exact phase behavior, thermal sensitivity, or gas–water interactions intrinsic to methane hydrates in natural geological settings. Additionally, differences in the dissociation kinetics and interfacial behavior may affect the dynamic response of ice-based analogs under certain conditions. Nevertheless, ice remains a practical and effective surrogate for use in mechanical studies focused on stress–strain responses, fracture behavior, and failure mechanisms. Within these constraints, the use of ice allows researchers to conduct controlled and reproducible laboratory investigations that contribute valuable insights into the geomechanical processes governing hydrate-bearing sediments.

2.2. Sample Preparation

In designing hydrate analog samples that accurately replicate the mechanical and physical properties of hydrate-bearing sediments, it is essential to consider the factors influencing rock behavior, the representative mechanical parameters, and the practical feasibility of parameter testing. Based on a detailed analysis of the mineral composition of the target hydrate reservoir, an artificial core skeleton was developed using a tailored mixture of quartz sand, illite, and calcium carbonate powder, combined with cement as the binder. Porosity and uniaxial compressive strength were selected as the key parameters for evaluating the similarity between synthetic and natural hydrate reservoir cores, with the corresponding experimental results summarized in Appendix A. Through the use of systematic testing and optimization, a core mixture with a sand-to-cement mass ratio of 84:16 and a water addition rate of 33% of the total mixture mass were identified as optimal. The granular component of the skeleton was composed of 36.7% calcium carbonate powder, 33.3% quartz sand, and 30.0% illite, closely resembling the mineralogical composition of sediments in the target area. The prepared artificial core skeletons exhibited a uniaxial compressive strength of approximately 2.5 MPa, a porosity range of 35–40%, and a permeability of 1.0–11.0 mD, effectively matching the geomechanical and petrophysical characteristics of the target hydrate reservoir.

The hydrate analog cores were fabricated by embedding ice into this optimized artificial sediment matrix. Ice saturation was controlled by calculating the required water injection volume, based on the porosity, bulk volume, and densities of water and ice. The mass of water required for injection into cores with different ice saturations is calculated using Equation (1):

$$m_w=0.92{\rho }_w\Phi V_{f}S$$

(1)

where m_w represents the mass of injected water in grams (g); Φ is the porosity of the core skeleton, which is dimensionless; V_f denotes the apparent volume of the core skeleton in cubic centimeters (cm³); ρ_w is the density of water in kilograms per cubic meter (kg·m⁻³); and S is the preset saturation, as a percentage (%).

Due to gravitational drainage during the early freezing phase, especially in medium-to-high saturation samples, the actual injected water volume was adjusted to exceed the calculated amount by 2–3% to compensate for leakage.

Ice saturation was verified using two methods. The mass method calculates the ice saturation by determining the mass difference between the core skeleton and the core after water injection and freezing, using Equation (2):

$$S_1={\displaystyle \frac{m_b-m_g}{\Phi {\rho }_b{V}_f}}$$

(2)

where S₁ represents the ice saturation of the core measured using the mass method; m_b is the mass of the ice-bearing core in grams (g); m_g is the mass of the dry core in grams (g); and ρ_b is the density of ice in grams per cubic centimeter (g·cm⁻³).

The porosity method estimates the ice saturation of the core by measuring the difference in pore volume between the core skeleton and the core after water injection and freezing, as calculated by Equation (3):

$$S_2={\displaystyle \frac{\Phi V_f-{\Phi }_b{V}_b}{\Phi V_f}}$$

(3)

where V_b is the volume of the core after saturation with ice in cubic centimeters (cm³) and Φ_b is the porosity of the core sample after saturation with ice, which is dimensionless.

Figure 2 presents a comparison of the measured saturation values, using both methods. The results demonstrate high consistency, validating the effectiveness of both approaches. Given its simplicity and repeatability, the mass method was adopted for saturation control during sample preparation.

Based on the aforementioned methods, a total of 36 synthetic samples with varying saturations and dimensions of 300 mm × 300 mm × 300 mm were fabricated. Figure 3 illustrates the schematic diagram and physical appearance of the molded samples.

2.3. Experimental Device and Procedure

The large-scale physical simulation system used for the hydrate sediment fracturing experiment is shown in Figure 4. This system consists of four primary components: a low-temperature simulation unit, a triaxial stress loading system, a fracturing fluid injection system, and a data acquisition system. The low-temperature simulation unit maintains a stable testing environment, with a minimum controllable temperature of −10 °C, enabling experiments under hydrate-relevant thermal conditions. The triaxial stress loading system, which includes a triaxial press, hydraulic station, and control cabinet, provides independent and synchronized control of the three principal stresses, with a pressure resolution of 0.1 MPa. The fracturing fluid injection system consists of a triplex plunger pump, water tank, mass flowmeter, and high-pressure fluid pipelines. The pump can reach a maximum pressure of 100 MPa at a rated displacement of 15 L·min⁻¹, while the piping is rated to withstand pressures up to 20 MPa, and the water tank stores the fracturing fluid, which is dyed with a red tracer for post-fracture visualization. The data acquisition system includes a signal acquisition box, pressure and flow sensors, a data transmission interface, and data acquisition software. It enables the real-time, multi-channel monitoring of parameters, such as maximum and minimum principal stresses, vertical stress, injection flow rate, and fluid pressure, with a sampling frequency of 1024 Hz.

The basic process in terms of the hydraulic fracturing experiment is as follows:

(i) Sample installation: Place the core sample in a triaxial press test chamber in a low-temperature environment. Connect the pre-embedded perforation tubing inside the sample to the external fracturing fluid pipeline using threaded couplings;

(ii) Stress loading: According to the experimental plan, set the loading pressure values for each principal direction. Simultaneously initiate all three pistons via the control cabinet to apply uniform loading at a rate of 0.1 MPa·s⁻¹, until the desired stress state is reached;

(iii) Fracturing: When stress stabilization is confirmed, initiate fracturing by injecting fluid, mixed with a red tracer, into the sample, using the triplex plunger pump. At the same time, activate the data acquisition system to monitor the real-time injection pressure and fracture evolution curves;

(iv) Post-fracture analysis: After fracturing, remove the sample and visually assess the dye distribution on the surface to preliminarily identify the fracture paths. The sample is then dissected to observe the internal fracture morphology, orientation, and propagation characteristics.

2.4. Experimental Design

In this study, the simulated wellbore was designed with four perforation channels to replicate the conditions of a horizontal well. The triaxial stress loading method is illustrated in Figure 5a, and the experimental setup is shown in Figure 5b. Various perforation geometries were applied, including 180° directional, 60° helical, 90° helical, and 90° fixed-plane configurations, as illustrated in Figure 5c. The open-hole configuration, which lacks a dedicated perforating gun, follows the same layout as the 180° directional perforation.

The experimental was designed to systematically explore the influence of five key parameters on the fracture initiation pressure and fracture morphology: ice saturation, fracturing fluid viscosity, injection rate, in situ stress conditions, and perforation patterns. The detailed test plan is summarized in

Table 2. Experimental plan for triaxial hydraulic fracturing of samples.

Group Number	Sample ID	${\mathit{\sigma }}_{\mathit{v}}\mathit{/}{\mathit{\sigma }}_{\mathit{H}}\mathit{/}{\mathit{\sigma }}_{\mathit{h}}$ (MPa)	Injection Rate $\mathit{Q}$ (L·min−1)	Viscosity $\mathit{\eta }$ (mPa·s)	Ice Saturation $\mathit{S}$ (%)	Perforation Patterns
A1	1#	3/1.5/1.5	14	1	0	180-degree directional hole layout
	2#				20
	3#				40
	4#				60
	5#				80
A2	6#	3/1.5/1.5	14	5	0	180-degree directional hole layout
	7#				20
	8#				40
	9#				60
	10#				80
A3	11#	3/1.5/1.5	10	30	0	180-degree directional hole layout
	12#				20
	13#				40
	14#				60
	15#				80
A4	16#	3/1.5/1.5	8	60	0	180-degree directional hole layout
	17#				20
	18#				40
	19#				60
	20#				80
A5	21#	3/1.5/1.5	6	80	0	180-degree directional hole layout
	22#				20
	23#				40
	24#				60
	25#				80
B	26#	3/1.5/1.5	5	1	40	180-degree directional hole layout
	27#		6
	28#		8
C	29#	4/2/1	14	1	40	180-degree directional hole
	30#	5/3/1
	31#	5/3/1.5
D	32#	3/1.5/1.5	14	1	40	60-degree helical hole layout
	33#					90-degree helical hole layout
	34#					180-degree directional hole layout
	35#					90-degree fixed-plane hole layout
	36#					open-hole perforation-free pipe

.

3. Experiment Results and Analysis

3.1. Effect of Ice Saturation and Fracturing Fluid Viscosity

Figure 6 presents the injection pressure curves over time for samples with varying ice saturations under a fracturing fluid viscosity of 1.0 mPa·s. The fracturing curves of the samples can be broadly categorized into three stages. The first stage involves the filling of the simulated wellbore with fracturing fluid, which is pumped into the wellbore through pipelines and comes into contact with the sample at the perforation. During this stage, the injection pressure remains relatively constant. The second stage involves the initiation of fractures. As the injection rate exceeds the seepage capacity of the sample, fluid accumulation leads to a rapid increase in pressure. The increased pressure induces tangential or vertical tensile stresses around the wellbore. Once the tensile stress exceeds the tensile strength of the sample, a primary fracture is initiated, accompanied by a sudden drop in injection pressure. The third stage involves fracture propagation. Continuous fluid injection drives the expansion and propagation of fractures. In some cases, secondary fractures are generated, causing fluctuations in the pressure curve during propagation.

Figure 7 illustrates the relationships between the fracture initiation and propagation pressures and ice saturation when the fracturing fluid viscosity is 1 mPa·s. The fracture initiation pressure exhibits a non-monotonic trend: it first decreases, then increases, and finally decreases again, as the ice saturation increases. Within the saturation range of 0–40%, the decrease in the fracture initiation pressure is primarily influenced by the inherent strength of the core, as inferred from strength tests conducted on cores with varying ice saturations. When the ice saturation ranges from 30% to 60%, the resistance of the samples to failure is primarily determined by the strength of the ice, the cementation strength between the sediment particles and ice, and the strength of the rock matrix. An increase in ice saturation enhances the cementation strength between the sediment particles and ice, while reducing fracturing fluid leak-off, leading to a notable increase in the fracture initiation pressure. Once the ice saturation exceeds 60%, the plasticity of the samples intensifies, with most pores being filled with ice, resulting in reduced porosity and permeability. As the injection pressure increases, stress concentration zones are more likely to form, thereby decreasing the fracture initiation pressure. There is no apparent pattern between the fracture propagation pressure and ice saturation, except that when the ice saturation reaches 60%, the fracture propagation pressure significantly rises due to the increased strength.

When the fracturing fluid viscosity is 1 mPa·s, most of the samples with ice saturations of 0%, 20%, 40%, and 60% exhibit the formation of transverse and torsional fractures. In contrast, samples with ice saturations of 20% and 80% develop longitudinal fractures upon fracturing, as shown in Figure 8. It is evident that there exists a correlation between the morphology of fractures and ice saturation.

Figure 9 shows the injection pressure curves for samples with varying ice saturations under a fracturing fluid viscosity of 5.0 mPa·s. And as illustrated in Figure 10, a notable increase in the fracture initiation pressure is observed when the fracturing fluid viscosity is 5 mPa·s compared to 1 mPa·s. Within the ice saturation range of 0–40%, the fracture initiation pressure decreases with increasing ice saturation, while in the 40–80% range, a positive correlation is observed. The propagation pressure follows a similar trend, although anomalies at 60% saturation may result from the heterogeneity of the ice distribution.

The fracture morphology remains largely consistent with that observed at a fracturing fluid viscosity of 1 mPa·s. The samples with low ice saturation predominantly exhibit horizontal fractures, transitioning to through-going fractures as the ice saturation reaches 80%, as shown in Figure 11. Specifically, at 20% and 60% ice saturation, single-wing fractures are observed, further underscoring the heterogeneity of the samples.

Figure 12 shows the injection pressure curves for samples with varying ice satura-tions under a fracturing fluid viscosity of 30.0 mPa·s. To quantify these observations, Figure 13 further explores the relationship between fracture pressures and ice saturation at the same fluid viscosity. It is evident that the fracture initiation pressure increases significantly with ice saturation compared to the case with 1 mPa·s viscosity. However, the fracture propagation pressure shows no clear trend, suggesting a more complex interaction between ice content and fracture development under higher viscosity conditions.

Building upon these findings, Figure 14 provides visual evidence of the resulting fracture morphology under 30 mPa·s fluid viscosity. It can be observed that most of the fractures tend to propagate vertically. This implies that not only does ice saturation influence the magnitude of fracture pressures, but together with fluid viscosity, it also plays a critical role in determining the directionality of fracture propagation.

Figure 15 shows the injection pressure curves for samples with varying ice satura-tions under a fracturing fluid viscosity of 60.0 mPa·s. The pressure profiles reflect distinct responses influenced by ice content, indicating potential changes in fracture behavior. To further analyze these trends, Figure 16 illustrates the variation of fracture initiation and propagation pressures with increasing ice saturation. Notably, both pressures exhibit a trend of initially decreasing and then increasing, suggesting that ice saturation has a nonlinear but consistent influence on the fracturing process at this viscosity level.

Building upon this pressure behavior, Figure 17 compares the fracture patterns formed at 60 mPa·s viscosity. It highlights that higher viscosity fluids promote the development of more complex and diverse fracture geometries, even under the same ice saturation. This comparison across different viscosities underscores the significant role of fluid viscosity in modifying fracture morphology alongside ice saturation effects.

Figure 18 shows the injection pressure curves for samples with varying ice satura-tions under a fracturing fluid viscosity of 60.0 mPa·s. To examine the influence of even higher viscosity, Figure 19 presents the pressure responses at 80.0 mPa·s. At this level, the fracture initiation pressure shows a non-monotonic trend, first increasing and then decreasing with rising ice saturation. Meanwhile, the propagation pressure lacks a clear pattern, indicating more complex fracture dynamics under higher viscosity conditions.

Further insights are provided in Figure 20, which shows the fracture morphologies observed at 80.0 mPa·s. Under these conditions—particularly at higher ice saturations—secondary fractures become more prevalent. Additionally, some samples exhibit asymmetric fractures on either side of the wellbore. This asymmetry is likely due to the natural heterogeneity within the ice-saturated samples, which causes fractures to propagate preferentially along structurally weaker planes. These observations highlight how increasing fluid viscosity, in combination with ice saturation, contributes not only to pressure behavior variations but also to increasingly complex and irregular fracture patterns.

3.2. Effect of Injection Rate

Figure 21 presents the variation in the injection pressure over time at different injection rates for a specimen with an ice saturation of 40% and a fracturing fluid viscosity of 1 mPa·s. As depicted in Figure 22, the initiation pressure and propagation pressure exhibit a trend of initially decreasing and then increasing with the increase in the fracturing fluid injection rate. At lower injection rates, the infiltration rate of the fracturing fluid into the rock sample slows down, resulting in a smaller seepage velocity gradient. In this scenario, a higher total pressure is required to initiate fractures within the specimen. Additionally, the fluid loss from the fracturing fluid enhances the pore pressure, while the fracture propagation rate is insufficient to release the fluid pressure within the fractures. Consequently, higher initiation and propagation pressures are observed during the fracturing tests conducted at low injection rates.

As the injection rate increases, the dynamic impact of the fluid on the wellbore wall becomes more pronounced. Once the fluid energy density exceeds a critical threshold, the elevated strain rate within the wellbore region promotes faster fracture nucleation. Therefore, a moderate increase in the injection rate enhances the fracturing efficiency. Nonetheless, the results also suggest that beyond a certain critical injection rate, further increases yield diminishing returns in terms of fracture performance.

The fracture morphology, shown in Figure 23, supports these observations. At injection rates of 5, 6, and 8 L·min⁻¹, the samples predominantly exhibit vertical fractures. However, when the injection rate reaches 14 L·min⁻¹, more complex patterns, including transverse and torsional fractures, are observed. This suggests that higher injection rates reduce fluid loss and promote the development of simpler, more through-going fractures. In contrast, lower injection rates result in more irregular fracture geometries and incomplete propagation, due to an insufficient fracture-driving force and a greater filtration loss.

3.3. Effect of Horizontal In Situ Stress Difference

Figure 24 shows the injection pressure–time curves for the samples with 40% ice saturation and a fracturing fluid viscosity of 1 mPa·s under different horizontal in situ stress differences. Compared with previous experimental groups, the fracturing curves under varying horizontal stress conditions exhibit distinct features during the fracture initiation stage. In this set of tests, the injection pressure rapidly dropped after reaching the initiation threshold, but then briefly increased, before entering the stable propagation stage. This transient rise in pressure is attributed to uneven internal stress distribution caused by the horizontal in situ stress differential, which inhibits local fracture propagation, thereby requiring additional pressure for fracture advancement. Notably, Sample #29 displays a unique pressure response. After reaching the first pressure peak, the injection pressure does not exhibit a sharp decline, but instead rises again to a second peak, before fracture propagation begins. Its initiation pressure is approximately 3 MPa higher than that of the other samples, while its propagation pressure is nearly double. This anomalous behavior is presumed to result from the simultaneous formation of two transverse fractures within the specimen, which demand greater energy for initiation and continued propagation.

Figure 25 further quantifies the relationship between the horizontal in situ stress differential coefficient and both the fracture initiation and propagation pressures. The data reveal that the horizontal stress differential has a relatively minor effect on the fracturing pressure itself. However, when the horizontal differential coefficient is fixed at 1.0, increasing the vertical stress significantly increase both the initiation and propagation pressure, indicating the critical role of vertical confinement. Despite the differences in the stress conditions, all the specimens consistently formed transverse and torsional fractures, with curvilinear propagation paths observed in most cases. Figure 26 illustrates the fracture morphology under different stress states, confirming that the in situ stress regime influences the fracture trajectory and complexity.

3.4. Effect of Perforation Patterns

Figure 27 illustrates the variation in the injection pressure over time for a sample with an ice saturation of 40% and a fracturing fluid viscosity of 1 mPa·s under different hole arrangement methods. The relationship between the fracture initiation pressure, propagation pressure, and perforation pattern is illustrated in Figure 28. Overall, the influence of the perforation pattern on the fracturing pressure is relatively minor compared to other factors, such as fluid viscosity or saturation. However, specific trends can still be observed. In particular, spiral perforation patterns tend to produce more tortuous fracture trajectories. As the spiral phase angle increases, both the initiation and propagation pressure also rise, as demonstrated by Samples #33 and #35. Compared to uniform (planar) perforation layouts with the same phase angle, spiral perforations introduce greater resistance to fluid flow along the fracture path, due to increased curvature and geometric complexity, necessitating higher fluid energy to drive fracture growth.

Figure 29 illustrates the fracture morphologies associated with various perforation patterns. Regardless of the configuration, fractures typically propagate symmetrically from opposing wings of the perforation zone. However, the perforation pattern clearly influences the orientation of the resulting fractures. For example, 90° helical perforations produce vertically intersecting fractures, while 90° fixed-plane perforations yield horizontally aligned fractures. As the helical angle increases, the fracture path becomes more irregular and multidirectional, reflecting the geometric steering effect imposed by the perforation layout.

These findings highlight the critical role of perforation design in guiding the fracture orientation, which is essential for optimizing fracture connectivity and enhancing reservoir stimulation performance.

4. Discussion

4.1. Analysis of Fracture Morphology During Hydraulic Fracturing

At lower saturation levels and fracturing fluid viscosities, transverse and torsional fractures are predominantly formed. However, as these parameters increase, longitudinal fractures become more common. At low injection rates, longitudinal fractures are formed, along with significant fluid loss, leading to incomplete fracture propagation and complex fracture morphologies. As the injection rate increases, fractures propagate more rapidly, resulting in the formation of transverse fractures, torsional fractures, and more secondary and derivative fractures, within the samples. Under varying horizontal in situ stress differences, transverse fractures are consistently observed in regard to the sample morphologies. The perforation pattern significantly affects the direction of fracture propagation. Specifically, a 90° helical perforation pattern leads to the formation of vertically intersecting fractures, whereas a 90° fixed-plane perforation pattern results in transverse fractures. As the helical perforation angle increases, the fracture morphology becomes more complex. In our statistical analysis of the fracture types observed in the experiments, it was found that longitudinal fractures and secondary fractures accounted for a significant proportion of the fractures, 35% and 38%, respectively, as illustrated in Figure 30.

4.2. Hydrate Constitutive Model and Fitting

Research indicates that hydrate sediments exhibit distinct elastic–plastic characteristics [26,27], and a piecewise function is adopted to establish a constitutive equation for accurately describing the stress–strain relationship.

4.2.1. Nonlinear Constitutive Equation for Hydrate Sediments

The nonlinear constitutive equation for hydrate sediments can be broadly categorized into three stages: the linear elastic stage, the plastic yield stage, and the softening stage. The constitutive equation for the linear elastic stage can be described using Hooke’s Law. Classical models that delineate the constitutive relationship in the plastic stage typically assume stress to be either independent of strain or related through a linear stress–strain function, which is not compatible with the plastic behavior exhibited by hydrates. Given the linear decrease in stress during the softening phase and the inherent residual strength of the material, a nonlinear function is required to describe the stress–strain relationship during the plastic yield and softening stages.

Hence, the constitutive equation for hydrate sediments is given by Equation (4):

$$\left\{\begin{array}{l}\sigma =E\epsilon ,\epsilon \le {\epsilon }_s\\ \sigma =m\ln {\displaystyle \frac{\epsilon }{{\epsilon }_s}}+{\displaystyle \frac{{\sigma }_c-{\sigma }_s-m\ln \frac{{\epsilon }_c}{{\epsilon }_s}}{{\epsilon }_c-{\epsilon }_s}}(\epsilon -{\epsilon }_s)+{\sigma }_s,{\epsilon }_s\le \epsilon \le {\epsilon }_c\\ \sigma ={\sigma }_c{\displaystyle \frac{\epsilon /{\epsilon }_c}{n{(\epsilon /{\epsilon }_c-1)}^2+\epsilon /{\epsilon }_c}},{\epsilon }_c\le \epsilon \end{array}\right.$$

(4)

In the equation, m represents the yield coefficient, which characterizes the plastic yield behavior of the material and is obtained by fitting experimental data. Moreover, n represents the softening coefficient, which characterizes the softening behavior of the material and is obtained by fitting experimental data. As can be seen from the above equation, when the strains are ε_s and ε_c, respectively, the corresponding stresses are σ_s and σ_c, satisfying the continuity condition of the piecewise function.

Based on the uniaxial compression stress–strain curves obtained under varying saturation conditions, the relevant parameter values for the constitutive equation were determined using a standard least squares fitting method to minimize the error between the experimental data and the model’s predictions, as shown in

Table 3. Values of the parameters related to the constitutive equation for hydrates.

Saturation S, %	40	60	80
$\mathrm{yield} \mathrm{stress} {\sigma }_s$, MPa	0.323	0.319	0.235
$\mathrm{yield} \mathrm{strain} {\epsilon }_s$, %	0.47	0.32	0.62
$\mathrm{peak} \mathrm{stress} {\sigma }_c$, MPa	2.3640	2.7154	2.2504
$\mathrm{peak} \mathrm{strain} {\epsilon }_c$, %	3.578	3.828	6.090
elastic modulus $E$, MPa	72.266	109.840	35.236
yield coefficient $m$	0.72	0.77	0.68
softening coefficient $n$	0.97	0.95	1.09

.

4.2.2. Total Deformation Theory

In the field of geological engineering, the process of hydraulic fracturing of formations belongs to the realm of typical low strain rate mechanical behavior, where both the inner boundary (i.e., the wellbore wall) and the outer boundary (deep within the formation, far from the wellbore) can reasonably be considered as fixed constraint boundaries. Under the condition of proportional loading of the principal stresses around the wellbore, total deformation theory [28] (also known as perfect plasticity theory) emerges as an effective method to describe the material constitutive relationship during this process.

To simplify the complexity of the stress distribution, the stress state surrounding the wellbore in the formation is often idealized as the stress distribution within an infinite, homogeneous, and isotropic plate containing a circular hole, as visually depicted in Figure 31. Here, R and R_S represent the radii of the wellbore and the plastic zone, respectively, in meters (m); p_i denotes the fluid pressure within the wellbore in MPa; σ_x and σ_y are the terrestrial stresses in the X and Y directions, respectively, also in MPa; σ_ϴ and ${\tau }_{\theta r}$ stand for the hoop stress and shear stress, respectively, in MPa. This model disregards the heterogeneity and anisotropy of actual formations, yet it provides a sufficiently accurate and manageable framework for predicting and evaluating the key mechanical parameters and effects during hydraulic fracturing in many engineering applications.

When attempting to solve the stress distribution around the wellbore [29], the plastic radius R_S is first calculated using Equation (5). Equation (5), being a transcendental equation, can be solved numerically. If the plastic radius R_S is less than or equal to the wellbore radius R, only elastic deformation occurs in the formation, indicating the presence of an elastic deformation zone solely around the wellbore. Conversely, when the plastic radius R_S is greater than the wellbore radius R, both elastic and plastic deformation occurs in the formation, signifying the coexistence of both elastic and plastic deformation zones around the wellbore.

$$\begin{array}{l}{\sigma }_{xx}+{\sigma }_{yy}-2({\sigma }_{xx}-{\sigma }_{yy})\cos 2\theta -4{\sigma }_{xy}\sin 2\theta \\ ={\displaystyle \frac{2\vartheta {\epsilon }_s}{\sqrt{3}}}(1-{\displaystyle \frac{R_s^2}{R^2}})-{\displaystyle \frac{m}{\sqrt{3}}}{\ln }^2{\displaystyle \frac{R_s^2}{R^2}}+{\displaystyle \frac{4}{\sqrt{3}}}(\vartheta {\epsilon }_s-{\sigma }_s)\ln {\displaystyle \frac{R_s}{R}}\\ +2(1-\alpha )p_i-{\displaystyle \frac{2}{\sqrt{3}}}{\sigma }_s+{\displaystyle \frac{\alpha (1-2\nu )}{(1-\nu )}}{p}_0\\ +{\displaystyle \frac{\alpha }{(1-\nu )}}\left[p_0-(p_0-p_i){\displaystyle \frac{\ln \frac{R_e}{R_s}}{\ln \frac{R_e}{R}}}\right]\end{array}$$

(5)

When $r\ge R_s$, the stress field distribution in the elastic zone is given by:

$$\left\{\begin{array}{l}{\sigma }_r={\displaystyle \frac{{\sigma }_{xx}+{\sigma }_{yy}}{2}}(1-{\displaystyle \frac{R_s^2}{r^2}})+{\displaystyle \frac{{\sigma }_{xx}-{\sigma }_{yy}}{2}}(1+{\displaystyle \frac{3R_s^4}{r^4}}-{\displaystyle \frac{4R_s^2}{r^2}})\cos 2\theta \\ +{\sigma }_{xy}(1+{\displaystyle \frac{3R_s^4}{r^4}}-{\displaystyle \frac{4R_s^2}{r^2}})\sin 2\theta +{\displaystyle \frac{R_s^2}{r^2}}{\sigma }_{rs}+{\displaystyle \frac{\alpha (1-2\nu )}{2(1-\nu )}}(1-{\displaystyle \frac{R_s^2}{r^2}})(p_{rs}-p_0)\\ {\sigma }_{\theta }={\displaystyle \frac{{\sigma }_{xx}+{\sigma }_{yy}}{2}}(1+{\displaystyle \frac{R_s^2}{r^2}})-{\displaystyle \frac{{\sigma }_{xx}-{\sigma }_{yy}}{2}}(1+{\displaystyle \frac{3R_s^4}{r^4}})\cos 2\theta \\ -{\sigma }_{xy}(1+{\displaystyle \frac{3R_s^4}{r^4}})\sin 2\theta -{\displaystyle \frac{R_s^2}{r^2}}{\sigma }_{rs}+{\displaystyle \frac{\alpha (1-2\nu )}{2(1-\nu )}}(1+{\displaystyle \frac{R_s^2}{r^2}})(p_{rs}-p_0)\\ {\sigma }_z={\sigma }_{zz}-\nu \left[2({\sigma }_{xx}-{\sigma }_{yy}){\displaystyle \frac{R_s^2}{r^2}}\cos 2\theta +4{\sigma }_{xy}{\displaystyle \frac{R_s^2}{r^2}}\sin 2\theta \right]\\ +{\displaystyle \frac{\alpha (1-2\nu )}{2(1-\nu )}}(p_{rs}-p_0)\end{array}\right.$$

(6)

When $R\le r\le R_s$, the stress field in the plastic zone is given by:

$$\left\{\begin{array}{l}{\sigma }_r={\displaystyle \frac{m}{2\sqrt{3}}}({\ln }^2{\displaystyle \frac{R_s^2}{r^2}}-{\ln }^2{\displaystyle \frac{R_s^2}{R^2}})+{\displaystyle \frac{\vartheta {\epsilon }_s}{\sqrt{3}}}({\displaystyle \frac{R_s^2}{r^2}}-{\displaystyle \frac{R_s^2}{R^2}})+{\displaystyle \frac{2}{\sqrt{3}}}(\vartheta {\epsilon }_s-{\sigma }_s)\ln {\displaystyle \frac{r}{R}}\\ +\alpha \left[p_0-(p_0-p_i){\displaystyle \frac{\ln \frac{R_e}{r}}{\ln \frac{R_e}{R}}}\right]+(1-\alpha )p_i\\ {\sigma }_{\theta }={\displaystyle \frac{m}{2\sqrt{3}}}({\ln }^2{\displaystyle \frac{R_s^2}{r^2}}-{\ln }^2{\displaystyle \frac{R_s^2}{R^2}})+{\displaystyle \frac{\vartheta {\epsilon }_s}{\sqrt{3}}}({\displaystyle \frac{R_s^2}{r^2}}-{\displaystyle \frac{R_s^2}{R^2}})+{\displaystyle \frac{2}{\sqrt{3}}}(\vartheta {\epsilon }_s-{\sigma }_s)\ln {\displaystyle \frac{r}{R}}\\ +\alpha \left[p_0-(p_0-p_i){\displaystyle \frac{\ln \frac{R_e}{r}}{\ln \frac{R_e}{R}}}\right]+(1-\alpha )p_i\\ -{\displaystyle \frac{2}{\sqrt{3}}}\left[m\ln {\displaystyle \frac{R_s^2}{r^2}}+\vartheta {\epsilon }_s({\displaystyle \frac{R_s^2}{r^2}}-1)+{\sigma }_s\right]\\ {\sigma }_z={\displaystyle \frac{m}{2\sqrt{3}}}({\ln }^2{\displaystyle \frac{R_s^2}{r^2}}-{\ln }^2{\displaystyle \frac{R_s^2}{R^2}})+{\displaystyle \frac{\vartheta {\epsilon }_s}{\sqrt{3}}}({\displaystyle \frac{R_s^2}{r^2}}-{\displaystyle \frac{R_s^2}{R^2}})+{\displaystyle \frac{2}{\sqrt{3}}}(\vartheta {\epsilon }_s-{\sigma }_s)\ln {\displaystyle \frac{r}{R}}\\ +\alpha \left[p_0-(p_0-p_i){\displaystyle \frac{\ln \frac{R_e}{r}}{\ln \frac{R_e}{R}}}\right]+(1-\alpha )p_i\\ -{\displaystyle \frac{1}{\sqrt{3}}}\left[m\ln {\displaystyle \frac{R_s^2}{r^2}}+\vartheta {\epsilon }_s({\displaystyle \frac{R_s^2}{r^2}}-1)+{\sigma }_s\right]\end{array}\right.$$

(7)

When $R\ge R_s$, the stress components of the elastic shaft lining can be obtained as follows:

$$\left\{\begin{array}{l}{\sigma }_r=p_i\\ {\sigma }_{\theta }={\sigma }_{xx}+{\sigma }_{yy}-2({\sigma }_{xx}-{\sigma }_{yy})\cos 2\theta -4{\sigma }_{xy}\sin 2\theta -p_i\\ +{\displaystyle \frac{\alpha (1-2\nu )}{2(1-\nu )}}(p_i-p_p)\\ {\sigma }_z={\sigma }_{zz}-\nu \left[2({\sigma }_{xx}-{\sigma }_{yy})\cos 2\theta +4{\sigma }_{xy}\sin 2\theta \right]\\ +{\displaystyle \frac{\alpha (1-2\nu )}{2(1-\nu )}}(p_i-p_0)\end{array}\right.$$

(8)

And when $R<R_s$, the stress components of the plastic shaft lining can be derived as:

$$\left\{\begin{array}{l}{\sigma }_r=p_i\\ {\sigma }_{\theta }=p_i-{\displaystyle \frac{2}{\sqrt{3}}}\left[m\ln {\displaystyle \frac{R_s^2}{R^2}}+\vartheta {\epsilon }_s({\displaystyle \frac{R_s^2}{R^2}}-1)+{\sigma }_s\right]\\ {\sigma }_Z=p_i-{\displaystyle \frac{1}{\sqrt{3}}}\left[m\ln {\displaystyle \frac{R_s^2}{R^2}}+\vartheta {\epsilon }_s({\displaystyle \frac{R_s^2}{R^2}}-1)+{\sigma }_s\right]\end{array}\right.$$

(9)

4.2.3. Sensitivity Analysis of Elastic–Plastic Constitutive Model Parameters

To investigate the impact of uncertainty on the model’s predictions, a sensitivity analysis of the parameters used in the elastic–plastic constitutive model was conducted. The relevant parameter values for the sensitivity analysis are shown in

Table 4. Values of model input parameters corresponding to laboratory experiments. Input parameter ranges for the elastic–plastic model used in sensitivity analysis.

Serial No.	$\mathit{m}$	${\mathit{\epsilon }}_{\mathit{s}}$	${\mathit{\sigma }}_{\mathit{s}}$	${\mathit{\epsilon }}_{\mathit{c}}$	${\mathit{\sigma }}_{\mathit{c}}$
1	[0.65, 0.80]	0.5	0.3	4.9	2.50
2	0.70	[0.3, 0.7]	0.3	4.9	2.50
3	0.70	0.5	[0.2, 0.4]	4.9	2.50
4	0.70	0.5	0.3	[3.6, 6.2]	2.50
5	0.70	0.5	0.3	4.9	[2.25, 2.75]

.

Figure 32a shows the stress distribution around the wellbore for different yield coefficients (m). As the yield coefficient increases from 0.65 to 0.8, the plastic radius decreases from 13.620 mm to 13.258 mm, indicating that the smaller the yield coefficient, the more pronounced the plastic characteristics. Both the radial and circumferential stresses in the plastic zone decrease slightly as the yield coefficient increases, while the elastic zone behaves in the opposite manner. At the wellbore, the radial stress remains unchanged, while the effective circumferential stress ranges from 0.545 MPa to 0.556 MPa. Thus, the yield coefficient has a minor effect on fracture initiation. Figure 32b presents the stress distribution around the wellbore for different yield strains. As the yield strain increases from 0.3% to 0.7%, the plastic radius decreases from 14.251 mm to 12.760 mm, indicating that a smaller yield strain results in more pronounced plastic characteristics. At the wellbore, the effective circumferential stress decreases from 0.829 MPa to 0.250 MPa as the yield strain increases. Figure 32c illustrates the stress distribution for different yield stresses. As the yield stress increases from 0.2 MPa to 0.4 MPa, the plastic radius decreases from 14.218 mm to 12.796 mm, indicating that the plastic characteristics reduce as the yield stress increases. At the wellbore, the effective circumferential stress increases from 0.427 MPa to 0.653 MPa with increasing yield stress. Figure 32d depicts the stress distribution for different peak strains. As the peak strain increases from 3.6% to 6.2%, the plastic radius increases from 12.829 mm to 13.900 mm, indicating that the plastic characteristics become more pronounced with increasing peak strain. Additionally, the peak strain mainly affects the stress near the wellbore, especially the circumferential stress in the initial segment of the elastic zone. As the peak strain increases from 3.6% to 6.2%, the effective circumferential stress at the wellbore increases from 0.280 MPa to 0.702 MPa. Therefore, a higher peak strain makes fracture initiation more difficult. Figure 32e shows the stress distribution for different peak stresses. As the peak stress increases from 2.25 MPa to 2.75 MPa, the plastic radius decreases from 13.876 mm to 13.174 mm, suggesting that the plastic characteristics decrease as the peak stress increases. At the wellbore, the effective circumferential stress decreases from 0.694 MPa to 0.423 MPa as the peak stress increases. Therefore, a higher peak stress makes the material more brittle and reduces the circumferential stress at the wellbore.

4.3. Exploration of the Mechanism of Fracture Initiation Pressure

4.3.1. Criterion for Rock Failure

During hydraulic fracturing, the main failure mechanism of the formation involves the dual effects of tensile and shear failure. In particular, in the analysis of the mechanical behavior of hydrate deposits, their shear strength characteristics follow the widely validated Mohr–Coulomb criterion. In view of this, this article adopts two classic criteria—the maximum tensile stress criterion and the Mohr–Coulomb criterion, to accurately predict the initiation pressure of hydrate deposits during the elastic–plastic stage, thereby ensuring scientific and accurate analysis [30,31,32].

• Maximum tensile stress criterion

This criterion states that failure occurs when the minimum effective principal stress, σ₃, reaches the tensile strength, S_t, of the rock, i.e.,

$${\sigma }_3=-S_t$$

(10)

2.

Mohr–Coulomb criterion

The Mohr–Coulomb strength criterion posits that rock failure occurs along a plane wherein the shear stress and normal stress attain the most unfavorable combination. Primarily utilized to describe the shear failure in rocks, its mathematical expression is given in Equation (11).

$${\sigma }_1={\displaystyle \frac{1+\sin \phi }{1-\sin \phi }}{\sigma }_3+2C{\displaystyle \frac{\cos \phi }{1-\sin \phi }}$$

(11)

4.3.2. Calculation of Fracture Initiation Pressure

• Tension-induced cracking mode

During hydraulic fracturing operations, when the effective circumferential stress on the wellbore wall of the formation decreases below its inherent tensile strength, the formation will undergo a cracking phenomenon. This mechanism, the tension-splitting model, is key to understanding the process and effects of hydraulic fracturing.

For horizontal wells, the initiation pressure p_f of the elastic wellbore $R\ge R_s$ can be obtained through a precise mechanical analysis and calculation, as shown in Equation (12). The plastic wellbore $R<R_s$ initiation pressure p_f can be calculated using Equation (13).

$$p_f={\displaystyle \frac{3{\sigma }_{yy}-{\sigma }_{xx}-\frac{\alpha (1-2\nu )}{1-\nu }{p}_0+S_t}{1+\alpha -\frac{\alpha (1-2\nu )}{1-\nu }}}$$

(12)

$$p_f={\displaystyle \frac{\frac{2}{\sqrt{3}}\left[m\ln \frac{R_S^2}{R^2}+\vartheta {\epsilon }_S(\frac{R_S^2}{R^2}-1)+{\sigma }_S\right]-S_t}{1-\alpha }}$$

(13)

2.

Shear-induced cracking mode

During the process of hydraulic fracturing, when the effective radial stress and effective circumferential stress on the wellbore wall of the formation reach the critical conditions of the Mohr–Coulomb criterion, fracture initiation occurs within the formation. The fracture initiation pressure p_f for elastic wellbore walls ($R\ge R_s$) and plastic wellbore walls ($R<R_s$) in horizontal wells can be calculated using Equations (14) and (15), respectively.

$$p_f={\displaystyle \frac{(1+\sin \phi )\left[3{\sigma }_{yy}-{\sigma }_{xx}-\frac{\alpha (1-2\nu )}{1-\nu }{p}_0\right]+2C\cos \phi }{(1-\alpha )(1-\sin \phi )+(1+\sin \phi )\left[1+\alpha -\frac{\alpha (1-2\nu )}{1-\nu }\right]}}$$

(14)

$$p_f={\displaystyle \frac{2\frac{(1+\sin \phi )}{\sqrt{3}}\left[m\ln \frac{R_S^2}{R^2}+\vartheta {\epsilon }_S(\frac{R_S^2}{R^2}-1)+{\sigma }_S\right]-2C\cos \phi }{(1-\alpha )2\sin \phi }}$$

(15)

Based on the aforementioned theories, combined with the fracturing experiments, this study explores the initiation mechanism of hydrates. The input parameters for the model are presented in

Table 5. Values of model input parameters corresponding to laboratory experiments.

Parameter	Value	Parameter	Value
${\sigma }_v,{\sigma }_H,{\sigma }_h$ (MPa)	3, 1.5, 1.5	well deviation angle $\psi$ (°)	90
azimuth angle $\Omega$ (°)	90	wellbore circumference angle $\theta$ (°)	0
$R_e$ (mm)	150	$p_0$ (MPa)	0.1
$r$ (mm)	7.5	$\nu$	0.25
cohesive strength $C$ (MPa)	$0.735\times S+0.1435$	internal friction angle $\phi$ (°)	33
$S_t$, MPa	$S_t=0.131e^{1.2122\times S}$	$m,\vartheta ,{\sigma }_s$	along with the selected constitutive model

. The determination of the tensile strength, cohesion, and internal friction angle relates to Appendix A.

Figure 33 depicts the predicted fracture initiation pressures using the elastic fracture initiation model under varying effective stress coefficients (with other parameters held to be constant). The fracture initiation pressures, calculated based on the two criteria, range from 1.24 MPa to 4.37 MPa, significantly lower than the experimentally measured values of 7.12 MPa to 17.63 MPa. Therefore, the elastic fracture initiation model is not suitable for predicting the fracture initiation pressure during hydraulic fracturing of hydrate-bearing sediments.

Figure 34 depicts the variation in the initiation pressure and plastic radius under tensile and shear fracture modes of the plastic wellbore wall with different saturations, as a function of the effective stress coefficient. As the effective stress coefficient increases, the plastic radius decreases, whereas it gradually increases with the saturation level. When the initiation pressure is relatively low, the samples are prone to shear fracture. When the hydrate saturation reaches 80%, tensile fracture occurs in the sample. As the initiation pressure rises, the fracture mode of the sample shifts from shear to tensile. Based on the calculation results, the full-quantity theoretical model effectively reveals the fracture mechanism of hydrates.

5. Conclusions

(1) Ice has demonstrated its strong potential as a mechanical analog for methane hydrates in laboratory investigations, owing to its comparable elastic and tensile properties, similar occurrence modes within sediment matrices, and analogous surface morphologies. Its availability, stability under ambient conditions, and controllable saturation make it well-suited for preparing hydrate analog cores. Nevertheless, it is important to acknowledge that while ice-based analogs provide meaningful insights into fracture behavior, they do not fully replicate the complex thermal, multiphase, and geochemical interactions present in natural gas hydrate-bearing sediments.

(2) Fracturing simulations conducted on large-scale synthetic cores exhibit pressure–time responses that align with the three typical stages observed in field operations: fluid filling, fracture initiation, and fracture propagation. The fracture initiation pressure responds nonlinearly to changes in ice saturation, decreasing initially, increasing at moderate levels, and declining again at high saturations. Viscosity exerts a notable influence on the initiation pressure when below 30 mPa·s, but this influence diminishes at higher viscosities. The initiation and propagation pressures also vary with the injection rate, exhibiting a trend of an initial decline, followed by an increase, suggesting the existence of an optimal injection range. In contrast, variations in horizontal in situ stress differentials and perforation patterns have relatively limited effects within the parameter space explored.

(3) The fracture morphology varied with the experimental conditions. Low injection rates favored longitudinal fractures due to fluid loss, while higher injection rates induced transverse and torsional fractures. The perforation geometry influenced the fracture direction; spiral patterns yielded more complex paths. Although the classification was qualitative, future work will incorporate image-based quantification and 3D reconstruction.

(4) A nonlinear elastic–plastic constitutive model was developed and applied to predict fracture initiation. The sensitivity analysis showed that parameters such as the yield strain, yield stress, peak strain, and peak stress substantially affect plastic zone development and fracture pressure. The plastic model more accurately captured the experimental behavior than the elastic assumptions.

(5) The laboratory-scale experiments provided fundamental insights into the mechanical response and fracture behavior of hydrate analog materials. However, the finite boundary conditions and simplified sample structure in the test setup inevitably differ from the complex, heterogeneous conditions encountered in natural hydrate-bearing formations. In addition, the absence of repeated trials, in situ thermal monitoring, and full field strain measurements limit the extrapolation of the results to field applications. Future research will incorporate numerical modeling (e.g., FEM, THM coupling) and improved experimental instrumentation to bridge the scale gap and enhance the field relevance of laboratory findings.