The Tarim Basin is a composite and superimposed sedimentary basin that formed on a cratonic basement (Jia, 1997; He et al., 2005). Multiple findings, such as differences in the composition of the basement (Yang et al., 2014), the presence of a W-E-oriented high magnetic anomaly zone (Yuan, 1996), and primary paleomagnetic results, suggest that the Tarim Craton was finally assembled in the Early Neoproterozoic (Zhao et al., 2021). The Neoproterozoic evolution of the basin can be divided into three stages: 1) the rifting stage (760–700 Ma); 2) the rifting to depression transitional stage (660–600 Ma); and 3) the postrift depression evolution stage (580–540 Ma) (Shi et al., 2018).

From the Cambrian to Middle Ordovician, the Tarim Craton was subjected to an extensional tectonic setting, corresponding to the opening of the Proto-Tethys Ocean (He et al., 2005). Since the Late Ordovician, the southern margin of the Tarim Craton has transformed into a compressional setting, corresponding to the closure of the Proto-Tethys Ocean (Matte et al., 1996; Jia, 1997; Zhang et al., 1999; Xu et al., 2006). This compressional setting caused the formation of the intracratonic strike-slip fault zone (in the Ordovician carbonate sequence) (Deng et al., 2018; Huang, 2019; Sun et al., 2021; Chen et al., 2022). In situ laser ablation inductively coupled plasma–mass spectrometry (LA–ICP–MS) U–Pb dating of the calcite cements in these fault zones indicates that the strike-slip fault (or possibly a certain strike-slip fault) initiated at ∼460 Ma (Yang et al., 2021). There are three stages of key tectonic events for strike-slip faults during the Paleozoic: 1) end of the Middle Ordovician; 2) end of the Ordovician; and 3) end of the Middle Devonian (Huang, 2019). During the Late Devonian–Permian, the northern margin of the Tarim Craton experienced scissor-like, east-to-west closure of the South Tianshan Ocean and subsequent collision (Jia, 1997; Chen et al., 1999; Gao et al., 2009). The intracratonic strike-slip fault zone reactivated correspondingly in this stage (Huang, 2019). At the southern margin of the Tarim Craton, a passive margin was formed due to the opening of the Paleo-Tethys Ocean (Xu et al., 2006). From the Jurassic to the Eocene, a multiphase extensional setting occurred due to the evolution of the Neo-Tethys Ocean (He et al., 2005). During the Cenozoic, the far-field effect of the India–Eurasia collision influenced the whole basin (Sobel and Dumitru, 1997), while the F_Ⅰ17 fault zone was deeply buried.

The stratigraphic column in the Fuman oilfield contains sedimentary layers from the Nahua system (Neoproterozoic) to the Quaternary, with a total thickness of ∼11 km. The uppermost ∼2,400 m of Cenozoic strata contains Quaternary, Neogene, and Paleogene strata, which are in unconformably contact with each other (Figure 2). These sequences are composed of conglomerates, sandstones, and mudstones. They exhibit multiple rhythms and generally coarsen upward. Mesozoic strata (approximately 1,400 m thick) are composed of Lower Cretaceous, Lower Jurassic, and Triassic strata. The Lower Cretaceous strata are characterized by fine-grained sandstones interbedded with mudstone. In particular, the lower part of this sequence has a high feldspar content while the upper part has a high lithic content. Unconformably overlain by the Lower Cretaceous strata, the Lower Jurassic strata only have a thickness of <150 m, and the lithology consists of sandstones interbedded with mudstones. The Triassic strata with a thickness of ∼770 m consist of mudstones, silty mudstones, and lithic sandstones. At the bottom of this sequence, gravels and fine-grained lithic sandstones unconformably overlie the Paleozoic sequence.

The upper Paleozoic sequence consists of Permian igneous rocks and clastic rocks, Carboniferous marine clastic rocks interbedded with limestone, and Upper Devonian sandstone and siltstone (Figure 2). The thickness of this sequence is in the range of 1,300–1,400 m. The lower Paleozoic sequence is mainly characterized by lower Cambrian and upper Cambrian to Middle Ordovician carbonates, middle Cambrian gypsiferous rocks, Upper Ordovician mudstone, and Silurian clastic rocks. The Sinian (Ediacaran) strata consist of clastic rocks and dolomite in sequence. At the bottom of the sedimentary sequence, Nahua coarse clastic rocks are distributed sporadically in Neoproterozoic rifts. In general, the stratigraphic composition can be subdivided into 11 tectonostratigraphic units based on the regionally distributed unconformities.

In recent well production, approximately thirty wells confirm that the reservoirs are widely distributed along the F_Ⅰ17 fault zone. Fault-enlarged caves, and fracture-enlarged pores in the fault zone are the main types of reservoir spaces. Similar to the features of strike-slip faults, the reservoir distribution also exhibits rapid changes in structural style along strike and with depth (e.g., Deng et al., 2022). As these wells have been successively putt into production, a mass of dynamic data provides a comprehensive understanding of the interwell connectivity. On the other hand, the seismic inversion results based on a high-resolution seismic reflection survey also help us determine the reservoir connectivity. Combined with pressure changes, interwell interference analysis, changes in oil pressures and production, and seismic inversion of reservoirs, the Tarim Oilfield Company (CNPC) has confirmed four connected well groups in the southwest part of the 3D seismic survey area (Figure 1). These four connected well groups contain (Ⅰ) ManS4, ManS4-H3, ManS4-H2, and ManS4-H1; (Ⅱ) ManS2, ManS1, and ManS1-H2; (Ⅲ) ManS1-H1, ManS3-H8, ManS3-H7, and ManS3-H6; and (Ⅳ) ManS3 and ManS3-H2 (Figure 3). Subsequently, the reservoirs penetrated by these wells are supposedly connected and are thus classified into a unified reservoir unit.

In map view, the negative structural relief penetrated by wells ManS401H and ManS4 shows a typical extensional structure (Figures 7A, 8A). In sectional view, the rocks between two boundary faults (two independent principal displacement zones; PDZ1 and PDZ2) subside, forming a rhombic graben (Figure 6A). The graben has a length of ∼4.0 km and a width of ∼1.0 km (Figure 7). The maximum offset of the graben is up to 200 m (Figure 8B). Within the rhombic graben, a series of nearly North–South-striking normal faults are linked to the two northeast–southwest boundary faults, forming a classic negative flower structure. Left-stepping strike-slip faults with extensional settings indicate that the kinematic evolution of this fault segment underwent left-lateral slip, forming a typical extensional strike-slip duplex (Woodcock and Fischer, 1986). Otherwise, some of the north–south striking faults are not hard-linked, forming a series of relay ramps (Figure 7A).

The total length of this segment is approximately 18 km. Five major positive relief parts are present along the principal displacement zone (Figures 7A, 8A). The seismic cross-sections show that the rocks are uplifted between the two boundary faults (Figure 6B), indicating a local compressional setting. In strike-slip deformation, local compression zones are caused by either the left-lateral slipping of right-stepping shear faults or the right-lateral slipping of left-stepping shear faults (e.g., Sylvester, 1988). As mentioned previously, the extensional strike-slip duplex indicates that the F_Ⅰ17 fault zone mainly experienced left-lateral strike-slip deformation. Therefore, only the right-stepping shear faults [synthetic (R) shears] can form a compressional region in the overlapping position. This series of shear faults mainly strikes NE47°, and their lengths are 4.5 km (R1), 5.0 km (R2), 4.0 km (R3), 3.7 km (R4), and 2.8 km (R5). Otherwise, there are another series of shear faults linking the neighboring R shears. They strike NE65°, and their lengths are 1.3 km (P1), 1.5 km (P2), 1.2 km (P3), 0.9 km (P4), 0.7 km (P5), and 0.8 km (P6) from south to north (Figure 8). In summary, this fault segment is characterized a series of elongated blocks bounded by various arrangements of R and P shears, which is a usual pattern in the Riedel simple shear model.

In this segment, the strikes of faults change from NE45° to NE35° with a length of ∼30.2 km (Figure 7A). In sectional view, a single east-dipping transtensional fault is developed in the Cambrian–Ordovician sequence (Figure 6C). Five of these kinds of faults constitute a right-stepping fault zone. These faults do not overlap with each other. In the relief–distance plots, the relief values of this segment are distinctly smaller than those of the extensional strike-slip duplex and Riedel left-lateral shear segments (<80 m), indicating a smaller deformation zone.

On the structural map of the base of the Silurian, the distribution of en echelon normal faults shows a horsetail splay pattern (Figure 7B). Correspondingly, the initial position of horsetail splays exactly overlies the eastern end of PDZ2. Therefore, this segment is a typical horsetail splay structure in which the strain distribution diverges toward the northeast.

At the northeast margin of the F_Ⅰ17 fault zone, there is almost no relief between the two sides of the faults. The length of this segment is approximately 14.8 km and is mainly characterized by ‘string of beads’ seismic responces. The seismic cross-section shows that two independent single faults developed separately in the upper Cambrian–Middle Ordovician sequence (Figure 6D). These two faults are soft-linked at this position but hard-linked in the adjacent area, showing the lateral propagation growth and linkage process.

In fault zone width–distance plots, the width is closely related to fault segmentation (Figure 8; Supplementary Table S1). Focusing on the base of the Tumuxiuke Formation (DO₃ t), the fault zone width exhibits rapid change along strike, ranging from 362 to 2,518 m (Figure 8C). Each peak value corresponds to the overlapping or stepping position of two neighboring faults. For instance, the width of the extentional strike-slip duplex between PDZ1 and PDZ2 is up to 1,684 m (box 1 in Figure 8). The width of the pop-up structure caused by two overlapping synthetic shears (R) is up to 2,033 m (box 2 in Figure 8), while the width of the relay ramp caused by two right stepping transtentional faults is up to 1,592 m (box 3 in Figure 8). These two plots (relief–distance and fault zone width–distance) show high positive correlations, namely, the regions with greater relief (both positive and negative) also have wider fault zones.

After validation of borehole data, we reveal that the reservoir widths in the extensional strike-slip duplex, Riedel shear zone, and right-stepping horsetail splay range from 505 m to 1,594 m (Figure 8). In the reservoir width–fault zone width plot, several types of trend lines can generally fit the relationship between them, such as linear, logarithmic, polynomial, and power law (Figure 9; Supplementary Table S2). The determination coefficient (R ²) is an indicator of the fitting degree between the estimated value of the trend line and the actual data. The value of R ² is in the range of zero to one. The higher the value of R ² is, the better the fitting degree. Therefore, a logarithmic function relation [y = 740.53In(x)–3,906.6] is the best fitting relation, which has the highest value R ² value (0.6015). In other words, this plot reflects that the reservoir distribution is closely related to the fault zone width and that the reservoir width is positively related to the fault zone width.

In terms of reservoir connectivity, faults are the main channels of fluid migration. The seismic cross-sections parallel to the strike of the fault show that the fault surfaces are characterized by a chaotic seismic reflection configuration, in contrast to the continuous reflection of the adjacent unbroken rocks (Figure 10). The established structural framework of the F_Ⅰ17 fault zone indicates that the reservoirs penetrated by wells ManS4-H3, ManS4-H2, and ManS4-H1 are connected by a synthetic shear fault (R1) (Figures 10A, 11). Although the reservoir penetrated by well ManS4 is not located at R1, it is hard-linked with this synthetic shear fault. The reservoirs penetrated by wells ManS2 and ManS1 are connected by a synthetic shear fault (R2) (Figures 10B, 11). Although the reservoir penetrated by well ManS1-H2 is not linked by R2, it is located at the same pop-up structure as the wells that ManS2 and ManS1 penetrated, and this structure is characterized by a highly permeable fracture system. The reservoirs penetrated by wells ManS1-H1, ManS3-H8, ManS3-H7, and ManS3-H6 are linked by R4 (Figures 10C, 11). The reservoirs penetrated by wells ManS3 and ManS3-H2 are also linked by the same fault surface (Figures 10D, 11). This shear fault is nearly parallel to the principal displacement zone; thus, we interpreted it as a low-angle shear (Y).

Field work (e.g., Christie-Blick and Biddle, 1985) and physical and numerical analog modeling (Dooley and Schreurs, 2012; Asaoka et al., 2016) of intraplate strike-slip tectonics suggest that a long-distance strike-slip fault is composed of secondary associated structural elements (Sylvester, 1988). Furthermore, the occurrence of these elements has a certain sequence, such as in a Riedel shear model (Tchalenko, 1970; Dooley and Schreurs, 2012). For instance, synthetic shear (R), secondary synthetic shear (P), and low-angle shear (Y) appear successively, and antithetic (R′) shear is usually rare in nature (Keller et al., 1982; Sylvester, 1988; Venâncio and da Silva, 2023). A typical pattern of a principal displacement zone is composed of R, P, and R′ shears (e.g., Figure 11D). According to our interpretation, the Riedel left-lateral shear segment is mainly composed of R and P shears. Therefore, we suggest that the formation of the Riedel shear segment has three kinematic evolutionary stages: 1) the R shears initiated and laterally grew (Figure 11A); 2) the P shear initiated and linked to the neighboring R shears (Figure 11B); and 3) with the progressive growth of the R and P shear system, the present-day PDZ formed (Figure 11C). In structures associated with strike-slip faults, both releasing bends and extensional strike-slip duplexes may form rhombic grabens. Genetically, a releasing bend was formed due to progressive movement of a bent strike-slip fault (Christie-Blick and Biddle, 1985; Song et al., 2013). An extensional strike-slip duplex was formed due to the shearing of two stepping boundary faults. For the graben in the southwestern part of the F_Ⅰ17 fault zone, there are two independent principal displacement zones rather than a bend fault (Figures 11A, B). Each fault within the graben has a relatively small displacement (secondary linked fault). Therefore, we suggest that the extensional strike-slip duplex is a more appropriate definition. For horsetail splays, normal faults usually initiate after the formation of the PDZ. Therefore, the kinematic evolution of the horsetail splay is slightly later than the evolution of the R and P shear systems (Figure 11C).

Since the reservoir in the Fuman oilfield has been defined as a fault-controlled fracture-cave carbonate reservoir (e.g., Huang et al., 2022), the geometry of the fault zone directly influences the reservoir distribution. The reservoir spaces are mainly located in the fault core and damage zone within the strike-slip deformation zone (Bruhn et al., 1994; Caine et al., 1996; Berg and Skar, 2005). In the F_I17 fault zone, the large-scale reservoirs (indicated by a wider reservoir distribution) are mainly in the overlapping or stepping zone (Figure 8). In fault zone width–reservoir width plots, a positive correlation is presented, indicating that the reservoir distribution is controlled by the lateral segmentation of the strike-slip fault (Figure 9).

In a fault zone, fractures, connected pores, and induced fractures are potential channels for fluid migration, which has been exemplified in several cases (e.g., Zeng et al., 2022; Fu et al., 2023; Gong et al., 2023). When there are two or more faults (relay or overlap, soft- or hard-linked), the increased structural complexity increases the numbers of faults and fractures and gives a wider range of orientations than for single, isolated faults (e.g., Sibson, 1996; Peacock and Parfitt, 2002; Kim et al., 2004; Fossen et al., 2005). Thus, the overlapping zones provide a wider range of fluid flow patterns (e.g., Fossen and Rotevatn, 2016), promoting hydrocarbon migration.

During the intermittent stage of faulting, most fractures and induced fractures are sealed, causing most damage zones to have no capacity for migrating fluids. In the fault core and damage zone–fault core transition areas (Billi et al., 2003), several formation microimagers (FMIs) indicate brecciated limestones with high porosity and permeability (e.g., Deng et al., 2022). Therefore, the neighboring pores within the same fault cores are more likely connected with each other. This interpretation is consistent with the previous model, suggesting that the fluid usually migrated through the connected pores (Sun et al., 2012). According to the established evolution model of the F_Ⅰ17 fault zone, the reservoirs that connect with each other, such as the reservoir connected groups I–IV, are mainly located at the same synthetic shear fault surface (Figure 12). The connected pores are mainly located at the fault cores and damage zone–fault core transition areas, namely, around the fault slip surfaces. Therefore, different shear faults along the strike provide independent connected reservoir units. We suggest that the unique fault segmentation (lateral growth and linkage of segmented shear faults) controls the reservoir distribution and connectivity.