1 Introduction

In recent years, the application of hydraulic fracturing techniques has successfully unlocked unconventional tight reservoirs, such as shale, coal, tight sandstone (Bennour et al. 2015; Hou et al. 2016; Ren et al. 2018; Chen et al. 2020). Multilayer formations can be commonly seen in many tight or unconventional reservoirs. It is well-known that some areas in this specific formation contain weak geological interfaces, lamina, or layers that can act as strong hydraulic fracture barriers and are responsible for hydraulic fracture height containment. In addition, the alternation of fracture propagation path along different layers can even cause the generated hydraulic fracture that deviates from the original in-situ stress direction, also affecting the intended hydraulic fracture design purpose while increasing the risk for the proppant injection and placement (Cruz et al. 2018). To increase the fractured hydraulic volume, the penetration of hydraulic fracture across multilayers from multiple perforation clusters on the horizontal well is critical in hydraulic fracture design. Thus, understanding the fracture initiation and propagation path in a multilayered rock is essential in the oil and gas industry.

The hydraulic fracture propagation in multilayered reservoirs may be more uncertain than in heterogeneous rocks. The authors found that the layer mechanical properties, in situ stress differences, thickness, and treatment parameters, such as fracturing fluid viscosity or pump rate, can affect the final fracture geometry in multilayered formations. However, several questions remain regarding the critical factors affecting the vertical hydraulic fracture growth as they propagate through laminated formations or multiple layers. Various numerical simulations have been performed to study the vertical fracture propagation in vertical wells in multilayered media (Sun et al. 2020). Yang (2012) proposed a Pseudo-3D Model for hydraulic fracture propagation in multilayered lithologies. Moreover, she discussed the possibility of a unified fracture design method for multilayered fracturing development. Afsar (2015) investigated the vertical hydraulic fracture propagation in multilayered reservoirs with rhythmic limestone-marl alternations. They indicated that fracture propagation in layered rocks is inhibited by stress barriers such as lithological contacts, thicknesses, and heterogeneities within marl layers and stress barriers. Ouchi et al. (2017) established a fully 3-D poroelastic model to predict vertical fracture propagation. They demonstrated that several mechanical behaviors, including fracture turning, kinking, and branching, can result from the existence of bedding planes, layer interfaces, and even smaller-scale heterogeneities. Ju et al. (2019) applied a hydromechanical coupling model to demonstrate the propagation behavior of hydraulic fractures in multilayered reservoirs. They believed that interfaces disrupt stress continuity, reselect the propagation direction of hydraulic fractures in numerous perforations, induce hydraulic fracture growth, and enhance fracturing-induced damage and contact slide events. Considering the differences between the actual complex situation and numerical simulation conditions, true triaxial laboratory experiments of hydraulic fracturing are common and direct methods for investigating the mechanism of hydraulic fracture initiation and propagation in multilayered media (Garcia et al. 2013; Sesetty and Ghassemi 2015; Shi et al. 2017; Gao et al. 2018). Compared to numerical simulations, actual downhole and rock properties rather than idealistic assumptions can be considered in mineback experiments and subsurface cored rocks (Lhomme et al. 2002; Srutarshi et al. 2015; Li et al. 2018, 2022; Hou et al. 2019). Athavale and Miskimins 2008 investigated the vertical fracture behavior in a homogeneous cement block. They believed that the material property contrasts between different layers and the stress contrasts set up inside the block are regarded as the primary causes of complex fracture growth in the laminated block. Tan et al. (2017) performed triaxial fracturing experiments on laminated shale outcrops. Moreover, the fracture initiation and vertical propagation behavior were also investigated. They concluded that simple fractures, fishbone-like fractures, fishbone-like fractures with natural fracture openings, and multilateral fishbone-like fracture networks could be observed in the experimental results. Given the variations in mechanical properties, in situ hydraulic fracture path monitoring can also be used to monitor the containment of a hydraulic fracture. The solution to increasing the fracture volume in this particular formation for hydraulic fracturing on horizontal wells is still uncommon, according to previous research, extensive experimental and numerical discussion of fracture mechanisms such as hydraulic fracture turns (bends), kinks, and arrests at the layer boundary in multilayered rock for vertical wells have occurred.

Large-scale multilayered rock samples were subjected to true triaxial hydraulic fracturing studies to better understand the complicated fracture behavior of multilayered rock in a horizontal well. Additionally, the effects of the treatment parameters on the hydraulic fracture initiation and vertical propagation behavior were investigated. These parameters include the perforation characteristics, fracturing fluid viscosity, and injection rate. Moreover, a convenient Grid measurement method was introduced to describe the fracture area and analysis the hydraulic fracturing effectiveness. This study aims to predict the potential of the stimulated reservoir volume technique in multilayered reservoirs under specific fracture treatment designs and better understand the initiation and propagation of fractures in multilayered reservoirs in horizontal wells.

2 Experimental methodology

2.1 Equipment apparatus

The true triaxial apparatus is laboratory testing equipment used to investigate fracture behavior under actual geological stress regimes, see Fig. 1(1). The experimental device is comprised the confined pressure loading system, hydraulic servo loading system, pump pressure control system, oil–water separator, and data acquisition system. In the test frame, a 300 mm cubic specimen was placed. According to the experimental design, axial and horizontal strains could be evenly applied thanks to the installation of a ball seat and oil-pressure pump package in each horizontal and vertical clamp. The specimen was surrounded by four flat jacks, which loaded the confining stress. The specimen's pre-set wellbore was used to inject the fracturing fluids. The triaxial servo controller controlled the confining pressure loading system, which had a load capacity of 80 MPa and an axial stress range of 0–120 MPa. The air compressor pump was attached to one end of the pressure control system, whereas the fracturing fluid container was connected to the other. The injection mechanism can constantly pump up to 800 mL of different types of fracturing fluid.

Fig. 1
figure 1

Experimental instrument and prepared sample for true triaxial hydraulic fracturing

Full size image

2.2 Samples preparation

Because large outcrop samples, particularly multilayer rock samples, are challenging to obtain in the field, fracturing experiments primarily use artificial rocks. The artificial large-scale rock samples comprised of three layers, with the pay zone in the middle and barriers in the outside layers. Cement and quartz sand were mixed in different ratios in the pay zone and boundary layers. Table 1 lists the basic mechanical properties of the multilayer rock samples. 1. The tensile strength of the sample layer rock was 4.2 MPa, while the sample boundary rock had a tensile strength of 5.3 MPa. The Poisson's ratio was 0.21, and the elastic modulus was 37.2 GPa for the layered rock, while the Poisson's ratio was 0.21, and the elastic modulus for the boundary rock was 29.8 GPa.

Table 1 The mechanical rock properties of the pay zone layer and boundary layer
Full size table

Finally, during the hydraulic fracturing experiments, eight tight specimens with dimensions of 300 mm × 300 mm × 300 mm were used (Fig. 1(2)). The interface between the boundary and pay zone layers was bonded with a cuboid using putty powder, considering the weak layer interface effect for the hydraulic fracture path. Scale law was used to design the experimental conditions. A wellbore with an outside diameter of 160 mm was bored near the center of the sample, parallel multilayer orientation, and during the sample production procedure. A steel pipe is then installed in the drilled wellbore and the perforation tunnel is 30 mm long. The crucial framework was sealed with a high-strength adhesive agent to ensure no fracturing fluid leakage occurred during the triaxial fracturing experiment. The distribution perforation type with one group of two oriented perforation clusters was applied for most samples. For sample #4, two groups with four oriented perforation clusters were used to compare the effect of perforation number on fracture behavior. The direction of the entire perforated tunnel follows the vertical stress direction, which can guide the fracture perpendicular to the horizontal well. The cluster spacing was 5 cm for sample #4, which matched the field’s multiple perforations in the horizontal well. After 24 h, the specimens were unmolded and cured. All samples were kept for 14 days at approximately 25 °C under room humidity.

2.3 Experimental process

Four steps typically comprise the hydraulic fracturing process: (1) a multilayer rock specimen was placed into a real triaxial cell; (2) the lowest horizontal stress was first set to preset values, and then the maximum horizontal stress and vertical stress were applied; (3) fracturing fluid was injected until the rock sample was fractured; subsequently, the injection monitoring parameters were established by the experimental design; and (4) the fracturing fluid in the results from the injection pressure was then examined. The experimental plan for this study is presented in Table 2. The in-situ stress state met the strike-slip faulting requirement (σH > σv > σh). The slick water had a viscosity of 1 mPa s, gel fracturing fluid had a viscosity of 50 mPa s, and guar fluid had a viscosity of 100 mPa s. These features can be used to observe the effects of treatment parameter differences on fracture initiation and propagation paths for multilayered rocks.

Table 2 Experimental design for hydraulic fracturing on multilayered rock
Full size table

3 Experimental results

After the hydraulic fracturing treatment, the block was split off along the fracture area, and the fracture morphology and injection pump pressure curve were obtained for analysis. Moreover, the 3D rebuilt model was used to analyze the influence of the mechanical properties and the treatment parameters on fracture propagation.

3.1 Fracture morphology

Figure 2 illustrates the typical fracture morphology of the multilayer rock in a horizontal well after hydraulic fracturing. Figure 3 shows the 3D rebuilt model after the hydraulic fracturing for multilayered rocks. A transverse curved hydraulic propagation path can be observed for almost all multilayer rock samples, indicating that the in situ stress condition is critical in the fracture propagation direction. However, secondary and tertiary fractures along the interface can also be observed for all the test samples, revealing a high possibility of an inclined and non-planar fracture during hydraulic fracturing in the multilayer rock. The experimental results noted significant fracture differences for samples with slickwater as a low-viscosity fracturing fluid. When the low-viscosity fracturing fluid was applied, fishbone-shaped fractures, composed of transverse and axial fractures, were observed in samples 1, 2, 3, and 4. The intersection of the transverse and axial fractures can be found in the horizontal well, whereas the axial fracture can propagate towards the top and bottom boundaries of the sample. Because two groups of perforation clusters were set for sample #4, two parallel transverse fractures could be observed for this sample. For samples with the gel fracturing fluid (samples 5 and 6), the transverse hydraulic fracture direction can still follow the maximum principal horizontal stress direction but reorient to a certain degree with the maximum horizontal stress direction. A local axial fracture can be seen but only propagates in one direction of the sample. The fracture geometry of the samples with gua fracturing fluid is also composed of one transverse fracture and one axial fracture. For sample 7, the transverse fracture initiated along the maximum horizontal principal stress was approximately 90°. Reorientation of the transverse fracture can be observed in sample 8. In particular, the transverse and axial fractures almost merged, and the induced hydraulic fracture propagated along the layer interface with the lowest mechanical strength. However, there is a small number of reorientation hydraulic fractures for samples with a highly viscous fracturing fluid.

Fig. 2
figure 2

HF geometry of multilayer rock after hydraulic fracturing

Full size image
Fig. 3
figure 3

Hydraulic fracture geometry of multilayered rock after hydraulic fracturing

Full size image

Comparisons of samples 1 and 3 indicate the effect of the horizontal principal stress contrast on the fracture propagation path for multilayer rock in the horizontal well. When the horizontal principal stress contrast is minor, a large reoriented direction of the hydraulic fracture can be generated for sample 3. Moreover, the fracture initiation direction of the secondary and tertiary fractures differed significantly from that of the initial fracture. The generated transverse fracture is also reoriented to a certain degree with the horizontal principal stress direction. The change in the perforation numbers for the horizontal well can also affect the fracture geometry in the horizontal well. Compared with samples involving fewer perforations, the fracture geometry with more perforations tends to be more complex. This is owing to the competition of the fracturing fluid and redistribution of stress during the close fracture propagation when the same fracturing fluid type was used. Comparisons of the fracture geometry of samples 1, 2, 5, 6, 7, and 7 revealed that the pump rate change could also affect the fracture propagation path with the same viscous fracturing fluid. An increase in the pump rate generated a more complex fracture path in the horizontal well. In addition, increasing the number of perforations can increase fracture complexity near the horizontal well. For sample #4, both interfaces between the pay zone and boundary layers were activated, and more shear fractures along the interface could be generated. Moreover, the initiation positions for the secondary and tertiary fractures of sample 4 tend to occur near the horizontal well, whereas the initiation position for sample 2 tends to occur at the layer interface. Among all the samples, this fracture system was the most complex. This fracture behavior can be explained in two ways. First, increasing the number of perforations can provide two initiation sites. Additionally, even though uneven fracture propagation is visible, a sufficiently high pump rate can deliver sufficient fracturing energy for two competing fractures to propagate simultaneously, whereas the stress shadowing between adjacent perforations increases the fracture interaction. Consequently, fracturing treatment factors can affect the hydraulic fracture morphology of multilayered rocks.

3.2 Fracturing pump pressure

The fracturing pump pressures for the multilayered rock are shown in Fig. 4. The characteristics of the injection pressure curves obtained under different treatment conditions were significantly different. A comparison of the injection pressures for all samples indicates that the pressure curve can identify the three stages. For all samples, no sharp injection pressure increase or fall-off behavior occurred during multilayer rock fracturing, indicating that the fracture propagation process was not very fast. In contrast, the injection pressure fluctuated in all four stages. During the fracture extension process, the up- and down-trend fracturing pressure curves produced multiple significant reductions, which related to multiple breakdown points and microfracture propagation.

Fig. 4
figure 4

Injection pressure versus time curves for hydraulic fracturing experiments on multilayered rocks

Full size image

Compared to the injection pressure curves with a low-viscosity fracturing fluid, the injection pressure increased rapidly to reach the breakdown pressure using a highly viscous fracturing fluid. A relatively steady increase in wellbore pressure was observed for samples with a low-viscosity fracturing fluid, and the fluid pressure within the wellbore exceeded the minimum principal horizontal stress. After approximately 370 s in the test, a breakdown was finally observed, after which a decreasing trend in wellbore pressure was observed. To understand the effect of the pump rate on the fracture injection pressure of multilayered rock, the injection pressure results of the three groups of samples were obtained. It shows the increase in the pump rate can increase the breakdown pressure. The breakdown pressure for sample 2 was 68.14 MPa using low viscous fracturing fluid, respectively. The treatment pressure for sample 6 using gel fracturing fluid was 73.59 MPa, respectively. The fracturing breakdown pressure increased to 76.46 MPa for sample 8 under the same fracturing conditions, which was the highest for all the samples using a low-viscosity fracturing fluid. When a highly viscous fracturing fluid is applied, the breakdown pressure is the highest when the other conditions remain the same. The injection pressure comparison also shows that a low in-situ stress contrast can increase the breakdown pressure, approximately 60.11 MPa. The perforation increase can significantly decrease the breakdown pressure, and the decline rate was about 117.7%. In comparison, the change in pump rate can strongly affect the breakdown pressure; the breakdown pressure using the low pump rate was only 58.72 MPa, which is 86.1% of the sample using the high pump rate. A large fluctuation in the injection pressure curve can be observed, demonstrating the twist and reorientation of hydraulic fractures, and the two distinct peak pressure values for the injection pressure curve of sample 3 also suggest tense fracture competition, which matches the observed complex fracture system.

3.3 Fracture area comparison

The experimental results indicated that induced complex fractures were generated in samples after hydraulic fracturing under the layer influencing. To accurately and quantitatively estimate the reservoir stimulated volume, the measurement of the fracture surface area is necessary. The current test methods are divided into the Geomagic method (a commonly used three-dimensional graphical area measurement method) and Grid measurement method. The traditional Geomagic method needs to reconstruct the fracture with the help of the three-dimensional scanner, and the fracture morphologies and roughness can be obtained. Raw coordinate data points of the main fracture area can be obtained by scanning, but the scanning area is much smaller than the fracture area, thus resulting in measurement error. Additionally, each digital surface should be reconstructed by the Geomagic Studio software. However, this method is time-consuming and strongly dependent on the Geomagic scanning accuracy, especially for large-size samples. Thus, a new and convenient method with the square grid paper was presented. This method can unify the scale for all fracture surfaces and operate efficiently. In this work, the reconstructed surface can calculate the actual area, and the projected fracture surface is the cross-sectional area of each sample minus the hole area. The detail method can be drawn as follows: A grid paper was selected, and each grid length and width are 1cm2 or 0.5cm2, covered with paper on each wave surface, and then the cloth will be spread flat. Thereafter, the number of the covered grid can be recorded, thus obtaining the fracture area can be estimated. In general, the denser the squares more accurate. Figure 5 shows the fracture area estimation method for different samples. A higher grid number signifies that the main fracture surface is larger. Table 3 shows the comparison of the fracture area estimation using different methods. The relatively flat and curved fracture surfaces were selected and calculated. According to the comparison of the results using two different methods, a good match can be found for the estimated results regardless of the fracture surface types.

Fig. 5
figure 5

Fracture morphology estimation with the grid measurement method

Full size image
Table 3 The comparison of the fracture area estimation using different methods
Full size table

4 Discussion

The experimental findings following the true triaxial hydraulic fracturing experiments for multilayer rocks in the horizontal well are described in Table 4. Almost all samples can produce transverse hydraulic fractures, which are perpendicular to the horizontal well. The presence of secondary and tertiary curved hydraulic fractures further increased the complexity of the fracture. As can be observed, these secondary or tertiary hydraulic fracture extensions tend to be along the slip interface for most samples. Generally, the fracture geometry of a horizontal well in multilayered rock can be classified as a fishbone-type fracture. Figure 6 illustrates the fracture morphology of the multilayered rock on the horizontal well under different conditions. Three types of fishbone fracture geometries for multilayered rocks are related to the fracture behavior between the fracture and interface. If the fracturing fluid energy is sufficiently large, the fracture can penetrate or deflect across the interface, and axial fractures may occur. Occasionally, axial fractures occurred at the interface and were generated near the perforation position. A deviation in the transverse fracture can also be observed due to the interface barrier effect, indicating that more non-planar fractures can be created. Thus, the impacts of the layer mechanical strength differential on the fracture complexity in multilayered rocks are substantial. When the in situ stress and treatment parameters are identical, an increase in fracture height is observed because the interface slip may cause fracture propagation along the interface and absorbed fracturing energy.

Table 4 Experimental results after hydraulic fracturing on multilayered rock
Full size table
Fig. 6
figure 6

Fracture morphology for multilayered rock on the horizontal well under different geological and treatment conditions

Full size image

It is a known fact that bedding characteristics and interactions between induced and natural fractures can significantly influence fracture propagation. However, various treatment settings affect the interaction between the hydraulic fracture and the interface layers. The morphology of the hydraulic fractures corresponding to the highly viscous fracturing fluid revealed that the secondary and tertiary fractures tended to penetrate the multilayers. When a low-viscosity fracturing fluid is used, there is a high possibility of fracture propagation along the layer interface. Considering this condition, fishbone-shaped fractures tend to be created, while slippage likely occurs between layers' interfaces. With an increase in fracturing fluid viscosity, interfacial slippage is an important containment mechanism, and there are more chances for slippage to occur when a low fracturing fluid is used. The fracture near the wellbore was complex, with reoriented secondary fractures for some samples and rough fracture surface. Significant tortuosity and fracture width differences near the wellbore may pose a high risk for proppant injection. To reduce the possibility of layer mechanical strength dominating the fracture initiation geometry in the horizontal fracturing of multilayered rocks, the hybrid fracturing scenario can be applied as a better choice for the fracturing operation for this specific formation. To avoid a potential sand bridge near the wellbore, injecting highly viscous fracturing fluid can increase the chances of fracture penetration into the far field while decreasing the risk of near-wellbore tortuosity. Fracturing fluids with low viscosity after injecting a highly viscous fracturing fluid can enlarge the lateral fracture transformation and improve fracture complexity. The fracture toughness can be used to justify the fracture propagation path:

$$K_{I} = - \left( {2\sigma \sqrt {a/\pi } + \frac{(1 - 2v)}{{\pi (1 - v)}}P_{0} (r)\sqrt {\pi a} } \right)$$
(1)

where a represents the fracture's half-length, p0 represents the pore pressure in the region, v represents the Poisson's ratio, and σ represents the uniaxial compressive stress.

Because of the interface existence and layer mechanical properties difference, the fracture complexity is complex for almost all samples. Thus, a more significant challenge exists for the multilayered formation during the hydraulic fracturing stimulation.

The application of the fracture area was used to describe the stimulation effectiveness, which was also recorded in Table 4. Figure 7 illustrates the fracture area estimation under different conditions. The high in situ principal horizontal stress difference has strong control over the fractures, which increases the probability of hydraulic fractures penetrating the interface and connecting layers. When the in situ principal horizontal stress contrast is small, a random hydraulic fracture is easy to create and tends to form along the layer interface. The small horizontal principal stress difference benefits fracture complexity but limits the fracture length in the lateral and vertical directions. The total fracture area with a horizontal stress difference of 3 MPa was about 1053.83 cm2, which is larger than the sample with a horizontal stress difference of 7 MPa. When the in situ stress difference is greater than 5 MPa, the mechanical properties of the layer can have a profound, strong effect. Changes in the pump rate and perforation clusters can change the fracture complexity from simple to complex. The increase in perforation number can greatly increase the fracture area, and the total fracture area with 4 perforations was about 1942.43 cm2, while the fracture area with 2 perforations was about 1609.61 cm2. As a high injection rate is adopted during hydraulic fracturing in the horizontal well of multilayered rock, the fracturing fluid has a high chance of leaking into the layer interface, causing more interface slipping behaviors when the hydraulic fracture reaches the interface. The fracture area with the pump rate of 20 ml/min is much larger than the sample with the pump rate of 5 ml/min using the same fracturing fluid. A high injection rate increases the net pressure within the fracture and boosts fracture propagation. However, there was a slight difference in the fracture area using different fracturing fluid viscosity. The total fracture area was about 992.83 cm2 using slick water fracturing fluid. Comparitively, the total fracture areas were about 1002.57 and 1041.61 cm2, respectively. The main reason is that the low fracturing fluid is sometimes inadequate to create enough high fracturing energy to open or slide the layer interface, which is different from the total fracture area change rule in conventional natural fractured rock. Although some fractures are formed due to the high filtration capability of low viscous fracturing fluid, the total fracture area is not large, while the induced fracture gets more tortuous and the fracture surface gets rougher. Unlike in heterogeneous rocks, the existence of layers increases the possibilities of a complex fracture system. Moreover, multiple fractures during one stage cannot be initiated easily in such a formation environment. The uniform initiation of all clusters is difficult to achieve in multilayered rocks. To improve fracture initiation and overcome the adverse effects of fractures due to fluid competition, the spacing between fractures must be optimized to form an effective fracture network in layered formations. In addition, a temporary plugging diversion technique can be used. This increase in the friction force at the local position can be used to control the order of fracture initiation. Specifically, a high pump rate alleviates the stress shadow effect between the two induced fractures and creates a simultaneous fracture propagation path with multiple perforation clusters. Thus, uniform fracture propagation in a horizontal well is difficult to achieve in multilayered formations without suitable treatment parameters.

Fig. 7
figure 7

Fracture area estimation under different conditions

Full size image

5 Conclusions

This study focused on the sensitivity analysis of a series of true triaxial fracturing experiments on multilayered rocks. Research findings can be obtained.

  1. 1.

    The fracture geometry of multilayered rocks tends to be complex during hydraulic fracturing in a horizontal well. Fracture diversion and slip interfaces tend to occur and cause diversion and reorientation of non-planar fractures. The transverse and secondary fractures were composed of a final fishbone structure fracture system, and three types of fracture systems can be defined.

  2. 2.

    A complex fracture system can be created under a low-principle horizontal stress contrast or using a low-viscosity fracturing fluid for multilayered rocks. The increased fracturing fluid viscosity provides more changes for the layer interface damage and promotes fracture penetration directly across the interface, resulting in a high breakdown pressure. Thus, an ideal vertical planar fracture growth should be obtained with a sufficiently high viscous fracturing fluid, and a low fracturing fluid can be used to improve the fracture complexity.

  3. 3.

    The convenient grid method using special paper was presented to describe the simulated fracture effectiveness. Fracture deflection, diversion, arrest failure modes and a high breakdown pressure can be frequently observed, which is also related to the pump rate and perforation number. The large pump rate and perforation number are beneficial for larger fracture areas, but the decrease of fracturing fluid viscosity is not always beneficial for larger fracture areas. A high pump rate and fracturing fluid viscosity increase fracture energy and induce fracture extension. In contrast, multiple perforations can yield a parallel fracture system and more possible initiation positions but produce high breakdown pressure.

  4. 4.

    The penetration transverse fracture failure mode dominates all types in the near-wellbore zone with an oriented perforation type. However, tortuous fractures can still be observed near the wellbore while fracturing multilayered rocks, which could be a risk for proppant plugs. Consequently, it is necessary to adopt different proppant techniques during stimulation according to multilayered geological conditions.