In recent years, the Mahu oilfield (MHO) in the Junggar Basin has been discovered under the guidance of accumulation theory in sag areas. The Mahu oilfield is an extremely large tight conglomerate oilfield with reserves up to a billion tons^[1,2]. A series of scientific and technical difficulties exist in the development technology of the Mahu oilfield.

Typical overseas marine tight oil/shale plays are characterized by stable deposition, simple geologic structure, wide distribution, large thickness, high thermal maturity, medium depth, high pore pressure, high oil saturation, high gas-oil ratio, and good fluidity^[3,4]. However, the Mahu oilfield has its special characteristics of deposition and accumulation. First, the lithology of the Mahu oilfield is more complex. With quick changes in reservoir lithology laterally and strong heterogeneity, the superior "sweet spot" is difficult to be drilled in. Second, it is difficult for oil wells to obtain high yield due to remote source accumulation and low oil saturation. Third, the reservoir has relatively poor reservoir quality, undeveloped natural fractures, and large horizontal stress difference, which are difficult to induce complex fracture network in formation per traditionally theoretical understanding. It is necessary to demonstrate whether multi-stage fracturing in horizontal well is suitable for the Mahu oilfield under such complex geological conditions. Additionally, it is also necessary to determine whether or not volume fracturing can generate complex fracture networks in this place.

North America mainly focuses on the fast way to retrieve quick return on investment. Based on this business model, the developers usually pursue high production and fast return of investment in the early stage, and maintain stable low production for a long time in the later stage, and then replace the low-production sessions with new blocks. China has insufficient resources and strong demand for oil and gas, thus, the philosophy of development should be pressure drawdown management from beginning to extend the stable production period with acceptable rates, and improve the oil recovery as far as the requirements for return of investment^[5]. The Mahu oilfield has complex lithology, strong heterogeneity, low oil saturation, large horizontal stress difference, undeveloped natural fractures and low reserve utilization rate. In order to fully develop the Mahu oilfield, we should focus on the way to enhance the effective reserve utilization, EUR and recovery factor. To meet this purpose, we should base on the tridimensional tight-spacing staggered development (TSSD) technology. The key of this technology is the way to determine the optimum well spacing and well locations in tridimensional space. As traditional theory of fracturing mechanism and interwell interference of tight conglomerate are not appropriate and should be innovated. We should focus on the innovation of relevant technologies like spatial well pattern design, proactive interference control and utilization, customized fracturing optimization and multi-component synergy. Additionally, those innovative technologies should be continuously revised and improved during the application process in the oilfield development.

In this paper, first, we introduce the geological characteristics of the Mahu oilfield. Second, the theoretical understanding and key technologies of the TSSD combined with horizontal wells are emphatically expounded. Then, we review the test history of the Mahu oilfield and analyze the development effects under different lateral spacing. Finally, the key issues and future research directions that should be paid attention to are put forward.

Currently, the Mahu oilfield mainly developed the oil-bearing series of the Triassic Baikouquan Formation and the Permian Wuerhe Formation. These two set of layers are conglomerate deposits dominated by front sandy clastic flow in the fan delta with a small amount of distributary channel conglomerate. The conglomerate has poor particle separation and large particle size difference (2-64 mm). It presents a complex modal structure. These characteristics lead to complex pore structure. Fine-grained conglomerate and small-grained conglomerate are more conducive to forming high quality reservoirs in the Mahu oilfield. The reservoir space mainly consists of residual intergranular pore and feldspar dissolution pore, with typical characteristics of small pores and throats^[6,7]. The production wells are nearly unproductive without fracturing. The absolute content of clay minerals as binder is low (0.24%-4.57%). The reservoirs of the upper Wuerhe Formation in the southern Ma1 well area and the lower Wuerhe Formation in the northern Ma2 well area have medium-strong to strong water sensitivity.

The main development layer is the Triassic Baikouquan Formation with a depth of 2500-4000 m and sedimentary thickness of 110-140 m. It is divided into three longitudinal segments, namely, Baiyi (T₁b₁), Baier (T₁b₂) and Baisan (T₁b₃). The oil layers distribute among all the 3 segments, with a thickness of 6-25 m (average is 11.5 m), and a single oil layer thickness of 2-10 m with an extension of 1500-3000 m. The reservoir has a porosity of 7.7%-11.8% and a gas permeability of (0.3-1.0)×10^-3 μm². It belongs to low-porosity and low to ultra-low permeability reservoir.

The vertical distance between the Triassic Baikouquan Formation and the Permian Fengcheng Formation (the main source rock) exceeds 200 m. The accumulation of crude oil experienced a long-distance migration. The small pore throats due to tight lithology result in low oil saturation (41%-67%). The reservoir lithology is mostly basaltic tuff with strong plasticity, mixed with sandstone (12%-45%) and mudstone (1.8%-6.1%), rare brittle minerals and undeveloped natural fractures. There exists a large horizontal principal stresses difference (12.16-38.21 MPa). It is difficult to form a complex fracture network after fracturing^[8]. Reservoir pressure coefficient is 0.92-1.82, with an average of 1.41. Most reservoirs possess abnormal high pressure. The surface crude oil density is 0.82-0.85 g/cm³, the viscosity of crude oil at 50 °C is 2.0-17.3 mPa·s, and the gas-oil ratio is 180-400 m³/m³.

The Mahu oilfield possesses unique characteristics of deposition, reservoir and accumulation. There is a lack of development experience that can be directly used for the Mahu oilfield. The early development shows that the traditional development theory cannot guide the efficient development of tight conglomerate oilfield. There exists inherent difference between coarse-grained sedimentary conglomerate and fine- grained sedimentary shale. Thus, the development theory of North American cannot be fully applied to the Mahu oilfield.

A series of tests have been carried out to seek for efficient development technology in tight conglomerate oil field. Based upon, we innovatively put forward the theories of "steered-by- edge" fracturing mechanism (SBEFM) and proactive fracturing interference mechanism (PFIM). They have successfully guided the development of Ma131 well (referred to as the pilot area). We have concluded a series of theoretical understandings of high-efficiency development in tight conglomerate oilfield at different developing stages (Table 1). This paper mainly focuses on the two theoretical understandings of SBEFM and PFIM which are closely related to production and efficiency improvement.

Stress field and rock fabric characteristics are critical to the formation of complex fractures during fracturing. The fracture network can be further complicated through technological optimization and manual control. Taking shale as an example, the smaller the horizontal principal stress difference, the more developed the stratified fractures/natural fractures, and the higher the brittleness index, the more easily the complex fracture network will be generated. However, the horizontal principal stress difference of the reservoir in pilot area is much larger (larger than 10 MPa). The stratified fractures/ natural fractures are undeveloped with rare brittle minerals. The Mahu tight conglomerate is not conducive to the generation of complex fracture network. Based on the reservoir parameters of the Mahu oilfield, we simulated various scenarios using Unconventional Fracture Modeling model (UFM), including horizontal principal stress differences (0.5, 1.0, 3.0, 5.0, 10.0 MPa), wellbore spacing (100, 150, 200, 300 m), and cluster spacing (10, 20 m). The results indicated that only double-wing fractures were created in all the scenarios without considering the presence of natural fractures. If several "virtual natural fractures" were added into the model, the created hydraulic fractures were complicated only in the case that the horizontal principal stress difference was 0.5 MPa and cluster spacing was 10 m (Fig. 1). Based on above simulations, we concluded that the pilot area could not create complex fracture network using the current fracturing model.

Core observation and analysis showed that the conglomerate sections under different diameters are dominated by “steered-by-edge” fractures and there are only a few conglo-merate-piercing fractures (Fig. 2). Through in-depth researches and experiments, we put forward the SBEFM theory of complex network in the tight conglomerate. The theory is as follows, the mutual-supported structure of gravels in conglomerate, which is different from the stratified fractures/natural fracture of shale, is conducive to the formation of complex fracture network. The strong heterogeneity due to various lithology and gravel diameters is also a favorable factor for the formation of complex fracture networks. The conglomerate with less cementation and more intergravel pores can form effective reservoirs. Its plastic brittleness in mechanical measurement is not only related to lithology, but also to its fabric characteristics such as gravel diameter, cementation and mutual support structure. When the composition of Anshan basaltic tuff in the parent rock increases, the plasticity of conglomerate is enhanced. The brittleness of the conglomerate increases as the granite ratio in the parent rock rises. It is also possible that although the lithology of the gravel itself is brittle, the gravel is prone to dislocation or slippage along the interface between the gravels under the action of external force, which makes it exhibit partial plasticity in mechanical experiments. The contact surface between the gravels is weak. Hydraulic fractures are mainly generated and extend along the weak surface around gravel. Due to the combination of tensile fracture and shear slip, hydraulic fractures can branch and complicate in the process of extension. When more dissolution pores exist in the gravel and the cementation content are higher, fractures through the gravel can be more easily generated. For example, in the upper Wuerhe Formation of Mahu 1 well area, the proportion of dissolution pores in the gravel is 30%-53%, which may form fractures. Due to the relatively poor connectivity of dissolution pores within the gravel, the probability to form high pore pressure is low and the occurrence of tensile fractures is rare. Thus, “steered-by-edge” fracture is still the most important form in conglomerate reservoir. In the absence of fault induction or traction, the strong heterogeneous conglomerate is not easy to form simple main fractures, but more likely to form short and complex fractures with large tortuosity during hydraulic fracturing (Fig. 3).

Microseismic monitoring and production characteristics in pilot area also indicated that fracturing has created a complex fracture network. Therefore, the current models about complex fracture network fracturing are not suitable for tight conglomerate reservoirs.

Interwell interference will affect the production performance of adjacent wells (old wells), which could be positive, neutral, negative. The duration of the interference is uncertain, which could be transient, short period, long period or even permanent^[10]. Operators and researchers prefer to reduce or prevent the interwell interference. In the development and test of the Mahu oilfield, interwell interference occurred to different degrees in different well spacing tests. The maximum observed interference distance could even be 2 km. There may exist three types of mechanisms of interwell interference resulting from volumetric fracturing in tight conglomerate.

(1) Type I: fracturing fluid front does not arrive but compressive stress front has reached (Fig. 4a, 4b). During fracturing, the strain or movement generated in formation squeezes the casing of the adjacent well. Then, the casing produces local elasticity or even plastic deformation, which eventually increases the liquid pressure in the closed wellbore. If the adjacent well has been fractured, the fluid in the hydraulic fracture will be compressed when the compressive stress zone has reached. It will generate pressure surge and increase the fluid pressure in the wellbore. The interference can be observed on the pressure gauge. Type I interference is usually transient or short-time and does not receive sufficient attention. We find that Type I interference mechanism has important theoretical value for drilling optimization and the wellbore integrity through the coupled study on stratum, fracture, cement sheath, casing and wellbore.

(2) Type Ⅱ: fracturing fluid has arrived and fractures have connected to each other (Fig. 4c). When the fracturing fluid reaches a certain location but the proppant does not, it will generate pressure interference. The pore pressure decreases gradually when the well is backflow-produced. The hydraulic connectivity between unpropped fractures at the far end weakens gradually with the increase of effective stress. Therefore, the effect of Type Ⅱ is usually short-term.

(3) Type Ⅲ: fracturing fluid and proppant both have arrived and fractures have connected (Fig. 4d). In this case, the effect of interwell interference is long-term unless the interwell fracture connection is lost due to the failure of the propped fracture network. Evident pressure responses can be observed when the production system changes or when interwell interference is tested. When the wellbore is situated near the fault/fracture belt, fracturing can activate or open fault/fracture belt. Type I or Ⅱ also can occur, which will cause the interwell interference over a long distance.

According to the theory of artificial reservoir, some situations between Type Ⅱ and Type Ⅲ should be the optimal goal. In these situations, stress and pressure interference can produce a variety of positive effects that complicate hydraulic fractures. It includes: "Stress shadow" effect occurs among fractures between wells. This effect can make multiple fractures interact with each other and increase the complexity of fracture network; Mass and pressure transfer of fracturing fluid can reduce the horizontal stress difference and improve the complexity of fractures between wells. Based on the formation pressure monitoring and dynamic stress field simulation during the fracturing in pilot area, it was found that the formation pressure could be increases by more than 6.0 MPa, and the horizontal stress difference at the fracture tip could be decreased by more than 10 MPa; The coherent effect similar to wave field occurs when stress zones overlap with each other. It reduces or even eliminates the dynamic stress barrier formed by the torsion of the adjacent well compressed stress zone in principal stress direction. It inhibits the bursting of tensile stress zone along the principal stress direction to a certain extent. This is conducive to the further complication of fractures. All of the above in this section are the theoretical basis of deploying small spacing staggered well pattern and actively using the spatial stress field interference to generate the whole complex fracture network.

The development of tight conglomerate oilfield is a complex systematic project. Many key technologies were formed during the construction of pilot area (Table 2). This paper focuses on four key innovative technologies that mostly influence oil recovery.

Based on theoretical understanding and analysis of North American tight-spacing infill well pattern, the development effects of various well layout modes in the Mahu oilfield were analyzed considering different economic parameters and full life cycle economic model. The optimal deployment technology of TSSD well pattern was formed, including the following technical points.

(1) One-time well deployment. The distribution of stress field is complicated due to the uneven pressure drop during production. A "stress vortex" may be formed and drives the hydraulic fracture advance along the edge of the vortex. It would result in inadequate reconstruction and serious interwell interference. North American usually form small well spacing pattern through gradually increasing the density of the wells from different stages. The distribution of stress field becomes more complex and difficult to control. As the well pattern density and the production time increase, the stress field is more difficult to restore or reconstruct. The normalized economic indexes of the infill wells with various spacing are generally lower than those of the old ones. However, the small spacing wells with one-time deployment pattern not only fully utilize the reserves, but also provide favorable conditions for the original stress field to fully utilize the positive effect of stress interference. Its overall development effect is better than that of infill well pattern.

(2) Coordinated design of well spacing. Based on the current single well prediction model and full life cycle economic model, the range of geological reserves to be controlled by a single well is estimated. The initial well spacing is designed according to the horizontal segment length. According to the fracturing simulation and reservoir engineering evaluation, the average length of the hydraulic fractures (i.e., the length of the fracturing fluid reached in the fracture) and mean effective propped fracture length (i.e. the length of the proppant reached in the fracture) are determined. Considering fracture arrangement pattern, the well spacing range will be designed based on Type Ⅱ well interference (Fig. 4c). The well spacing is adjusted according to reservoir conditions. If the reservoir is thin, the well spacing can be increased appropriately. If the reservoir is thick, the well spacing can be reduced appropriately.

(3) Subdivided development layer system. Flow units are divided based on geomechanics. According to the average vertical height of the hydraulic fracture network, the maximum flow range and effective flow range are determined. For single-layer reservoir, according to its thickness and geome-chanical flow unit, several horizontal well patterns are divided vertically, and the landing position of each pattern is optimized. In the case of vertical multiple thin layers overlapping each other, the combination of multiple thin layers and the optimal landing layer in each combination are optimized. The division principle is: based on the current technical conditions, the vertical effective utilization should be achieved with the least layers.

(4) Stereo staggered well pattern. The W-type stereo staggered mode can reduce the negative vertical interference and strengthen the positive vertical interference between wells (Fig. 5). By optimizing the fracturing process of TSSD platform, the overall fracturing can be achieved and the complexity of the fracture network can also be enhanced.

The spatial stress field affected by interwell interference can be controlled and actively utilized. It plays an active role in the formation of complex fracture networks, multiple collaborative optimization and full life cycle management of tight conglomerate rocks. The proactive utilization technology of spatial stress field in stereo well pattern is formed in combination with the TSSD. It includes the following technical points.

(1) Inhibit adverse interference near well zone. In the process of multi-fracture cutting, the inner side fracture is prone to distortion, tortuosity and flexural friction increase. It will result in increased difficulty in sand adding and operation risk. The complexity of near-wellbore fracture network can be reduced by using current limiting perforation to improve discharge rate per hole, high viscosity liquid to restrain filtration loss, and quickly increasing discharge rate.

(2) Enhance the favorable interference on the plane. In the same layer of well pattern, staggered fracture arrangement should be adopted. We designed the fracture half-length as 2/3-3/4 of the well spacing (i.e., making the hydraulic fracture cross), and conducted zipper fracturing. In this way, the stress vortex in the same fracturing segment between wells is misaligned. It can avoid the approaching or connecting of "inrush zone" of two stress vortices between wells, which will result in adverse interference of simple overlong fracture. In addition, the dynamic stress barrier zone parallel to the direction of the wellbore should be avoided to cause insufficient interwell reconstruction. The compressive stress zone and tensile stress zone in the fracturing zone would be locally superposed by zipper operation. It can produce coherent effect which is similar to wave fields. This is helpful to eliminate the dynamic stress barrier zone formed by the maximum horizontal principal stress steering and the inrush zone of two adjacent stress vortex edges. During fracturing, the dynamic horizontal stress difference decreases, especially near the fracture tip. By designing fracture viscosity and displacement, the minimum horizontal principal stress at the fracture tip can increase by more than 10 MPa, which is conducive to complicating the fracture network far from the wellbore. The designed fracture half-length (2/3-3/4 of well spacing) is favorable for the formation of stress interference in the middle zone between two wells. It can also reduce the horizontal stress difference and is conducive to improving the complexity of hydraulic fracture network. Fractures far from the wellbore should be as complex as possible, which can suppress the length of hydraulic fractures and reduce the adverse interference between wells.

(3) Enhance vertical favorable interference. By adopting the staggered well arrangement, the effect of the staggered fractures in the vertical direction can be similar with that on the plane. According to the stress profile, the vertical operation procedure is determined. If the fracture is easy to extend downward, the fracturing procedure should be from upper well pattern to lower well pattern. If the fracture is likely to extend upward, fracturing sequence from lower well pattern to upper well pattern is recommended. Based upon, the positive effect of vertical stress interference can be fully utilized. In addition to vertical stress interference, the vertical channeling and interference of fracturing fluid also can lead to further increase of pore pressure, which is also beneficial for the formation of complex fracture network.

(4) Overall utilization of spatial stress interference. In the large-well cluster stereo platform operation, the operation mode of "from outside to inside, from bottom to top (or from top to bottom, depending on the stress profile), space zipper” is adopted. It is conducive to the formation of the positive interference of the overall spatial stress field.

In the pilot area, the consumption of fracturing fluid and sand in single well of T₁b₂ segment are 40% higher than that of T₁b₃ segment on average. But there is no significant difference in their production. Due to the relatively thin T₁b₂ segment, a significant portion of fracturing fluid and proppant entered the non-reservoir when conducting the large liquid and sand volume operation. In addition, according to the statistical analysis of production wells in the Mahu oilfield, there was no obvious positive correlation between fluid volume and production in some wells. In other words, the high volume of fracturing fluid did not lead to high production. The amount of fracturing fluid and proppant, fracture height and length should be optimized according to well spacing and reservoir conditions. Meanwhile, the amount of fracturing fluid and sand should also be reduced as much as possible based on the guarantee of maximum producing reserves. Therefore, we proposed the “optimum-size and distribution” fracturing optimization technology in the tight conglomerate formation (Fig. 6). It includes the following technical points.

(1) The design of fracturing fluid viscosity, displacement, the amount of fracturing fluid and sand and pumping program should optimized according to the reservoir thickness. Fracture height and shape should be controlled to make fracturing fluid and proppant mainly distributed within the reservoir, which is the optimum distribution vertically.

(2) According to the well spacing, fracture length and fracture layout design, the fluid volume and sand volume should be customized section by section. Optimize fluid properties and ratio of fluid to sand to improve the matching degree between average effective propped fracture length and average hydraulic fracture length.

(3) The uniformity of reconstruction along the wellbore and the uniformity of fracture patterns distribution between wells can be strengthened by some fracture control techniques, such as limiting the current, dynamic temporary plugging, variable load or pulse load. It not only can inhibit the adverse interference between wells caused by simple long fractures, but also avoid the blank area with less propped fractures between wells.

(4) According to the full life circle economic model of oil well, the optimal balance between production and benefit should be sought to determine the optimum fracturing size.

Geological engineering integration is a complex system engineering which aims at achieving global optimum through multiple collaborative optimization. It consists of the combination of multidisciplinary theories and multi-domain technologies, as well as the optimization of technical solutions, management processes, and the full life cycle of key nodes.

TSSD well pattern is favorable for multiple collaborative optimization. The mode of "small well spacing and large well cluster" can effectively improve the operating and management efficiency of drilling, completion and industrialized fracturing. Further cost reductions can be achieved through process optimization. Since the design stage, multiple collaborative optimization mechanism should be fully considered, including well spacing and hydraulic fracture length, well layout and interwell fracture layout, multi-well operation sequence in the same layer, synchronous pressure control, backflow and production after fracturing, etc. We should aim at maximizing the utilization of interwell resources, complicating fracture network, reducing the adverse impact of pressure interference and adverse effect of stress sensitivity to well productivity.

In pilot area, the stereo development and multiple collaborative optimization techniques were adopted as the main operation measure. It includes large well cluster factory, stereo staggered well layout, inter-well staggered fracture layout, platform integrated fracturing, zipper staggered construction, overall balance of shut in well, well group synchronous backflow, pressure control and coordinated production. The results showed that the operation efficiency and production effect were greatly improved. All the production indicators reflected that the key factors and links had achieved better collaborative optimization.

In order to explore the efficient development technology in tight conglomerate oilfield, we carried out a series of tests in the Mahu oilfield, such as well type and well spacing tests. The Mahu oilfield has gone through five stages, including advance water injection in vertical wells, depletion development and water injection in vertical wells, large well spacing and small well spacing development in horizontal wells, and small well spacing stereo development.

4.1.1. Advance water injection in vertical wells

The first experiment of advance water injection in vertical wells was applied in the Ma2 well area. In 2011, vertical well pattern (250 m×433 m) with five-point injection-production was constructed. Under the influencing factors such as high sensitivity of reservoir and fast lateral change of oil reservoir, the oil well showed the characteristics of low oil production, serious water channeling and ineffective water injection during production. Their outputs were only half that of wells adopting depletion development. Therefore, advance water injection in vertical wells did not achieve the desired effect.

4.1.2. Depletion development and water injection in vertical wells

In 2015, two fault blocks in the Ma18 well area were tested with depletion development mode in diamond-shaped vertical well pattern. The short wheelbase between the two wells is 200 and 300 m. The long wheelbase between the two wells (fracture extension direction) is 346 and 520 m. During fracturing, there was obvious interwell interference in long axis direction. During production, the output decreased greatly and the production effect was not good. After that, two well groups of water flooding development tests were carried out. The central well was injected to form a diamond-shaped reverse nine-point injection-production pattern. The test indicated that the two injection wells showed the characteristics of water absorption by fractures. Rapid water channeling occurred in the artificial fracture direction, but no reaction of water injection was observed in other directions. It indicated that the reservoir was not suitable for water flooding.

4.1.3. Horizontal wells development with large well spacing

Since 2013, the Mahu oilfield has entered the stage of horizontal well development test. Horizontal well spacing, segment length, distance between fracture intervals, fracturing scale and other parameters were tested in several blocks. The Ma131 well area experienced most tests. In 2015, 6 horizontal wells test with the well spacing of 400 m were carried out in the fault block T₁b₂ in Ma133 well area. There were 2 wells in each of the 1200, 1600 and 2000 m horizontal sections. There was interference during fracturing, manifested as a sharp rise in adjacent well pressure. In 2016, the 500 m well spacing test was carried out again in the fault block T₁b₂. 400 m and 500 m well spacing tests were carried out in fault block T₁b₃ in the Ma131 well area. The results showed that the fracturing interference still happened in some sections of the two fault block test wells, and the influence range was up to 2 km. When the well was disturbed, water content and pressure rose rapidly. After drainage and pressure relief for 9-50 d, the output could be restored to the level without interference. It has no obvious influence on the later production effect.

4.1.4. Horizontal wells development with small well spacing

Based on the above test results, in 2018, horizontal well patterns with well spacing of 300 m and 260 m were deployed. The well spacing in Ma131 and Ma133 well areas, whose reservoirs showed good physical properties, both were 300 m. The well spacing in Xia72 wells with poor physical properties was 260 m. Meanwhile, 200 m well spacing tests were carried out respectively in the Ma133 area and Xia72 area.

Well spacing tests were also carried out in other blocks such as Ma18 and Fengnan4.The spacing has been reduced from 500, 550 m to 200, 300 m. The results showed that there was interference during fracturing at different well spacing, but it was not obvious after production. Single well production and cumulative production were not strongly correlated with well spacing when it was 300, 400 and 500 m. The production effect did not deteriorate due to shorter well spacing.

4.1.5. Small well spacing stereo development

Further improvements are highly desirable, such as single well production, reserve utilization rate, recovery rate, drilling and completion efficiency, and effective reductions in investment. In June 2018, based on the theory of "SBEFM" and "PFIM" and numerical simulation, a pilot area was designed in Ma131. It adopted a series of core technology, including big well clusters, multiple pay zones, small well spacing, long well section, staggered pattern, fine perforation clustering, zipper fracturing and factory operation. A total of 12 horizontal wells were deployed on a large platform of T₁b₂ (5 wells) and T₁b₃ (7 wells). The well spacing is 100, 150 m and is in W-shaped staggered arrangement. The horizontal segment is 1800 m long and the spacing between clusters is 10 and 20 m. We used factory platform operation, zipper-type large-scale fracturing and synchronous production.

The technology of horizontal well combined with multi- stage fracturing has been tested for 6 years. With the improvements of engineering technology and construction organization, the horizontal lateral length has increased from 800 m in 2013 to 1428 m in 2019. The horizontal segment length in pilot area has reached 1800 m. The average horizontal segment length increased by 1000 m and the longest one reached 3000 m. The maximum depth of the well is 6130 m. So far, the development test of the Mahu oilfield has been basically completed. The optimal well type, well spacing and fracturing technology suitable for the Mahu oilfield are obtained.

4.2.1. Development and design of TSSD pattern

In pilot area, the optimization technology of TSSD well pattern was adopted. Two sets of oil reservoirs (T₁b₃ and T₁b₂¹) with a span of 35-40 m in the longitudinal direction were developed (Fig. 5). A total of 12 horizontal wells were deployed, including 7 wells in T₁b₃ layer with well spacing of 100 m, with reservoir thickness of 12.5 m, porosity of 10.5%, and permeability of 1.1×10^-3 μm². There are 5 wells in T₁b₂¹ with a well spacing of 150 m, reservoir thickness of 8.0 m, porosity of 10.2%, and permeability of 1.5×10^-3 μm². The two sets of reservoirs are developed with staggered well pattern. The horizontal section length is 1800 m, the fracture half-length is 80-90 m, and the main cluster spacing is 20 m. The staggered fracture spacing in the longitudinal direction is 10 m (Table 3).

The 7 wells in T₁b₃ have a produced oil-bearing area of 1.5 km², proven geological reserves of 70.0×10⁴ t, and average single-well draw up on reserves of 10.0×10⁴ t. The five wells in T₁b₂¹ have a draw up on oil-bearing area of 1.4 km², proven geological reserves of 33.0×10⁴ t, and average single-well draw up on reserves of 6.6×10⁴ t. The draw up on superimposed oil-bearing area is 1.5 km², the proved geological reserves is 103.0×10⁴ t, and the average proved geological reserves per well is 8.6×10⁴ t.

Twelve wells were designed with factory type platform. The well heads of the platform are arranged in a row, and each four wells make up a group. The group spacing is 60 m, and the well spacing within the group is 10 m. Three rigs operated at the same time. We drilled the first spud and second spud, then concentrated on completing the horizontal segment in the third spud.

4.2.2. Implementation of small well spacing stereo development platform

We adopted multiple collaborative optimization technology in the Mahu oilfield. Drilling in the pilot area began in July 2018 and was completed in June 2019. The average completion depth is 4932 m. The average horizontal segment length is 1720 m. The average reservoir penetration rate is 96.8%. The average drilling cycle is 62 d. Compared with the average level in this block, the average well depth increased by 50 m. The horizontal segment length increased by 160 m. Penetration rate increased by 9.6%. Drilling cycle is reduced by 32.6 d. Drilling efficiency has been greatly improved. Multiple collaborative optimization techniques have achieved remarkable results.

“Optimum-size and distribution” hydraulic fracturing optimization technology was also used. On June 11, 2019, factory fracturing was carried out. Zipper fracturing was performed at the same level on the same set of platforms. The fracturing stages range from 24 to 33, with an average of 28 stages. The number of clusters is 51-168, with an average of 94. Cluster spacing is 11-30 m, with an average of 21 m. The amount of added sand is 1670-3330 m³, with an average of 2350 m³. The amount of fracturing fluid is 23 986-57 661 m³, with an average of 35 944 m³ (Table 4). The fracturing was completed on August 28, then it entered the balance period of shut in well with an average of 28 d. After the wellhead pressure was balanced, we opened the well and started production.

4.2.3. Effects of small well spacing stereo fracturingThe proactive utilization technology of spatial stress field of small spacing stereo well pattern was adopted. The fracturing operation in pilot area was conducted according to the sequence of "fracturing the sides first and then the middle between platforms, and cross-fracturing between wells within the platform". It was found that the increase of the minimum horizontal principal stress was caused by the massive filling of the fracturing fluid and proppant along the direction of the minimum horizontal principal stress. The maximum horizontal principal stress was less affected and the horizontal stress difference decreased. The positive intervention of stress field is realized by the "pressure holding" effect formed by large amount of liquid injection. The shutdown pressure showed that the fracturing fluid increased the formation pore pressure. So the shutdown pressure also increased significantly (Fig. 7). The sand was added along the direction parallel to the direction of the minimum horizontal principal stress, which increased the minimum principal stress. The horizontal stress difference in the central well was reduced from 15-20 MPa to 8-15 MPa. Pressure monitoring showed that the formation pressure increased by 6 MPa during fracturing, which further increased the probability of forming a complex fracture network. Smaller well spacing and stereo well pattern resulted in more fracturing fluid per pore volume. This is the key to the significant rise in pore pressure.

The horizontal well M1244 and the vertical well Ma15 in the middle position are selected as the joint monitoring wells for downhole microseismic monitoring. The geophone of well M1244 went down into the horizontal segment and was duly dragged in accordance with the principle of "keeping a proper distance from the fracture stage", so that two monitoring wells were within the effective monitoring range (with the radius of 600 m). The results showed that the microseismic response in the initial stage was more concentrated. The created fracture was relatively simple in the early stage under the large horizontal stress difference. In the later stage and central zone, when pore pressure of fracturing zone increased, the rock strength limit forming the fracture network was reduced. According to the initial evaluation of shutdown pressure, the minimum horizontal principal stress increased by 5-10 MPa, while the horizontal stress difference decreased. Both of these factors are beneficial to the formation of branch fractures and complex fracture network. The monitoring interpretation results indicated that the stimulated reservoir volume of the post-fractured well or segment was larger (Fig. 8) and the fracture aspect ratio was smaller (Fig. 9). The fractures in the same layer interlaced with each other and covered the whole control area in the pilot area. The network complexity increased (Fig. 10). The proactive utilization of spatial stress field was effective.

Similarly, compared with the complex fracture network generated in the small well spacing pilot area, the hydraulic fractures formed under well spacing of 400 m are relatively simple (Fig. 11). There are large areas between wells that have not been covered by hydraulic fractures. These resources may not be effectively utilized. This demonstrates the importance of optimum-size and distribution fracturing techniques for tight conglomerate and the need to reduce well spacing.

4.3.1. Comparative analysis of operation parametersSix typical wells with different well spacing were selected for comparison (Table 5). It is found that over time, the drilling and completion cycle gets shorter and well spacing is getting smaller.

4.3.2. Comparative analysis of typical well production

The production of the pilot area is shown in Table 6. For the sake of comparison, production system and horizontal segment length are normalized (Fig. 12). It shows that the average single well production of the 7 wells in T₁b₃ layer with a spacing of 100 m is higher than that of the 5 wells in T₁b₂¹ layer with a spacing of 150 m.

The average output of the pilot area is 34.5% higher than that of the whole region. It is nearly the same as the top 10% wells in the region, the overall performance was excellent (Fig. 13). Besides, we also compared these wells with typical horizontal wells with distances of 200, 300, 400, 500 m respectively. It is found that the well with small spacing has a shorter oil breakthrough time with 1-3 d after the well is opened. By comparing the output in the first 180 d, it is found that, on the whole, the cumulative production of single well with small well spacing is higher than that with large well spacing (Fig. 14a). After normalization, the daily oil production per 1000 m horizontal segment with small well spacing is generally higher than that with large well spacing (Fig. 14b).

However, due to the reduction of well spacing, the geological reserves controlled by the single well decrease. Thus, the production indicators become better, which is obvious in the pilot area.

4.3.3. Comparative analysis of pressure changes

Wellhead pressures at different well distances in the same layer of same fault block are compared (Fig. 15). Although the initial pressures are different, the pressure drop trends under different well spacing are basically the same. The absolute value of oil pressure with small well spacing is higher. It shows that a large amount of liquid is injected in a small range under small well spacing, which enhances production. The pressure drop rate does not increase as the well spacing decreases, reflecting the strong ability of stable production.

4.3.4. Comparative analysis of reserve occupancy

The average geological reserves occupied by each well under different well spacing are compared. The larger the well spacing, the larger the reserves occupied by per 1000 m horizontal segment (Fig. 16). The geological reserves occupied by single well decreases with the decrease of well spacing. It also reflects that the smaller well spacing can make full use of geological reserves.

4.3.5. Production trend analysis

Production statistics were carried out on 36 horizontal wells whose operating time was more than 1 year in Ma131 well area. The initial monthly decline rate was 2.0%-2.9%, and the reduced annual decline rate was 21.5%-29.8%. In the first year, this typical decline rate of North American tight oil is about 40%-50%. The lower decline rate in the Mahu oilfield is due to the adoption of controlled pressure production. However, it effectively extended the production time of higher yield and significantly increased the final recoverable reserves of a single well.

Research and application in North America have shown that the cumulative production of the first 180 d of a tight oil well was a good indicator of the long-term performance^[11]. We analyzed the production characteristics of horizontal wells of the first 180 d in pilot area. It was found that production peaked in the fifth month and then entered a slow decline stage. Currently, the production level and decline trend are closer to the top 10% wells with the best production in this region (Fig. 17). The cumulative production per well in pilot area is (2.02-3.82)×10⁴ t. Average cumulative production per well is 3.07×10⁴ t.

4.3.6. Economic benefit analysis

Relationship models between well spacing and recovery and rate of return on investment are established (Fig. 18). There is a negative correlation between recovery and well spacing. The best solution depends on economic conditions. Smaller well spacing at low oil price is not economical because the predicted final recoverable reserves from single well are less. However, the higher the oil price, the better performance with the small well spacing.

The prediction results adopting decline method show that the final recovery rate of the 100, 150 m well spacing is more than 25% (Fig. 19a), which is better than that of the large well spacing. It is calculated according to the current return rate on investment calculation method of PetroChina Co., LTD. The analysis is based on the actual cost of settlement in the current year for large well spacing wells. The price is determined by the service company in 2020 for small spacing wells.

Small spacing well pattern in North American tight oil is formed through gradually multiple refinements. With the decrease of well spacing and the normalization of single well (unit length), the final recoverable reserves gradually decrease on the whole^[12]. However, the average cumulative production of 180 d per well in pilot area does not decrease significantly due to the reduction in well spacing. It reflects the advantages of the overall use of primary well pattern.

(1) The decline law of single well and platform in pilot area should be continuously observed. The production in pilot area has been with good performance for more than 8 months. It is necessary to continuously track and evaluate its future decline law, and to predict the final recoverable reserves and ultimate recovery in a single well.

(2) The quantitative description and evaluation of the artificial fracture network in pilot area should be carried out as soon as possible. Due to its complex and special sedimentary characteristics, tight conglomerate makes the formation mechanism and distribution of artificial fractures extremely complex, which is different from fine-grained sandstone and shale. The coring of highly deviated wells, the data acquisition and analysis of the logging system should be carried out in the future. It is also necessary to analyze and quantitatively evaluate the details of artificial fracture networks in tight conglomerate.

(3) The selection of optimum-size and distribution hydraulic fracturing needs further improvement. There is no obvious advantage in the production effect of the three wells with dense cutting, large fluid volume and high intensity fracturing. The cluster number and fluid amount should be designed properly to improve single well production and decrease the cost. This is the main direction of future technological exploration.

(4) The theory and key technology of efficient development of tight conglomerate oilfield need to be popularized. In 2020, we selected five well areas as role models, including Ma18, Fengnan4, Madong 2, Ma 2 and Aihu 2. Their advanced theory and key technology will be applied to layer T₁b₁, T₁b₂, T₁b₃, P₃w₁, P₃w₂ and so on. The main factors that affect productivity should be evaluated in depth. In the future, we will evaluate the effect of reservoir thickness, horizontal stress difference and pore pressure coefficient on production in tight conglomerate reservoirs with small well spacing.

(5) The mechanical mechanism of tight conglomerate and the optimal fracture network theory should be continuously explored. At present, the borehole trajectory in pilot area is designed parallel to the horizontal maximum principal stress direction. However, just several wells whose horizontal direction intersects with the horizontal maximum principal stress direction at a small angle achieve higher yield. Will changing the direction of well layout lead to higher production? What is the appropriate angle between horizontal well direction and in-situ stress? What is the proper well spacing? Which is more economical, to fully utilize reserves of the primary dense well pattern to pursue the high primary recovery or to use the gas injection method to improve the recovery in the later stage of the primary large spacing well pattern? All these problems require in-depth theoretical research and practical exploration.

In this paper, we put forward the “steered-by-edge” fracturing mechanism in tight conglomerate. It is the main pattern for forming complex hydraulic fracture network in the tight conglomerate with rich intergravel pores, large horizontal stress difference and undeveloped natural fractures. The traditional complex fracture network models for shales are not suitable for this special reservoir.

The proactive interwell interference theory in tight conglomerate oilfield is suggested for this region. During factory fracturing under tight well spacing, interwell interference is an effective mean to increase the complexity of fracture network. It is also the key to enhancing single well production and more importantly to largely increasing the recovery factor. It may be difficult to utilize the proactive interwell interference technology into fields with multi-generations of infilled-drilling wells in order to form complex fracture network, which have complex induced stress field due to pressure depletion. The one-time tight spacing well pattern is considered better than the progressive infilled drilling approach.

The “optimum size and distribution” concept of hydraulic fracturing design and optimization is put forward. According to the principle of "staggered fracture arrangement, zipper fracturing, right-size and overall utilization", the stereo staggered fracturing sequencing is implemented. The results show that the stress field can be proactively utilized to enhance the complexity of fracture network, maximize the effective utilization of reserves, and consequently achieve maximum recovery factor and EUR from a single well to whole platform

Through the pilot project, it is determined that horizontal well with multistage fracturing is the suitable main development technology in the Mahu oilfield. The further field tests indicate that the tridimensional tight-spacing staggered development (TSSD) technology can be a game changer for efficient development of the tight conglomerate oilfield in Mahu.

We are grateful for comments from Prof. Liu He, Prof. Zou Caineng, Prof. Wu Qi, Prof. Kuang Lichun, Prof. Wang Yuanji, Prof. Huo Jin, Prof. Zhi Dongming and other experts.