Due to frequent tectonic movements, rift basins in China show characteristics of multi-stage superimposition, resulting in multi-directional and multi-stage superimposed deformation of faults^[1] to form complex fault zones. Oil and gas exploration practices have shown that the fault reservoir with a high proportion of reserves and production is the main reservoir type in rift basins^[2,3,4]. Currently, the remaining resources controlled by fault zones still play an important role in the reserve growth and resource strategy of mature exploration areas, so fault reservoirs are still important targets for further exploration in the future^[4,5]. As faults are the key link throughout the whole process of hydrocarbon generation, migration, accumulation and preservation in petroliferous basins, they are always attractive for geologists and have been increasingly and extensively studied. With the progress of oil and gas exploration, certain achievements have been made in researches on fault system and hydrocarbon accumulation^[6,7], fault sealing and fluid migration/accumulation^{[8,9,10,11,12,13,14,15]}, and fault activity and hydrocarbon preservation^[16,17,18]. However, the oil and gas exploration is risky in complex fault zones, where the phenomenon of hydrocarbon differential accumulation is obvious^[19]. Moreover, failures in actual oil and gas exploration of fault traps are mainly attributed to the faulting history and lateral/vertical sealing of faults^[20]. Generally, there are two main mechanisms for formation of fault traps. First, cross fault traps are formed by superimposed deformation under multi-stage stress fields in multi-directions, especially the superimposed deformation due to multi-stage extension in different directions^[21]. Specifically, superimposed deformation caused by revived old faults and new faults, or interaction of multi-directional faults in the same stage gives rise to the cross fault trap. Second, fault traps are formed under the control of segment growth of single fault. According to the relationship between faults and formation occurrence, the fault in the same direction as formation dip is defined as synthetic fault, and the fault in the direction opposite to formation dip is defined as antithetic fault. For antithetic faults, the footwall of an isolated fault has moved up relatively under the action of fault tilting, and the fault trap is formed at the position with a large throw. With the continuous activity of faults, isolated faults begin to grow in segments, and then multiple isolated fault traps gradually connect to form a composite fault trap. For synthetic faults, the hanging wall and footwall of a synthetic fault cannot form fault trap in the isolated growth stage due to the differential activities of the fault. Along with the segment growth and connection of isolated faults, transverse anticlines are generally developed at the segment growth points of the hanging wall to form fault traps. The segment growth of fault has experienced three stages: isolated nucleation, soft connection, and hard connection^{[22,23,24,25,26,27]}. Differential fault activities lead to different formation mechanisms of various fault traps, which result in different hydrocarbon accumulation patterns. In this paper, a system of comprehensive evaluation on hydrocarbon-bearing effectiveness of fault traps in rift basins which considers the reliability of fault interpretation, the effectiveness of fault trap formation time, and the lateral and vertical sealing of faults has been worked out, based on the results of researches on rift basins and the laws of fault growth and evolution.

The contour map of throw of an isolated fault plane is ellipse on the whole, with the throw largest at the center and gradually decreasing towards the periphery, and becoming 0 at the tips^[28]. The throw-distance curve is semi-ellipse. Two isolated faults overlap to form a relay ramp. Due to the energy consumption on the relay ramp, the growth of the fault throw is relatively slow, the total throw within the range of the relay ramp is relatively small, and the throw-distance curve shows the shape characterized by "two highs and one low". The throw-distance curve is an effective means to determine the fault type. The low value area is the linkage point of fault segments, while high value area indicates isolated faults^[26,27]. At the linkage position of the hanging wall with small fault throw, anticline structure is formed, which is called transverse anticline^[29]. This feature is outstanding on the seismic profile parallel to the fault strike. Therefore, the throw-distance curve and the seismic profile parallel to the fault strike can be used to characterize the fault segmentation effectively.

According to the throw-distance curves and seismic profile along the fault strike for Nandagang synthetic fault and Koucun antithetic fault in southern Qikou sag, the position of the segment growth of the synthetic fault indicates the development position of trap (Fig. 1a, 1b), and there are four transverse anticlines at the segment growth points on the seismic profile of hanging wall along the fault strike (Fig. 2a). In contrast, the antithetic fault traps are mostly formed between the segment growth points on the footwall (Fig. 1c, 1d), and there are two anticline zones between the segment growth points on the seismic profile of footwall along the fault strike, indicating the positions of the fault traps (Fig. 2b).

Fault growth is a dynamic process, and it is induced by progressive deformation of fractures^[30]. Fault evolution is the accumulation of multiple sliding events^[31]. It is necessary to restore the distribution of faults in different geological periods. There are mainly two methods to restore the history of fault formation and evolution: vertical throw subtraction^[32,33] and maximum throw subtraction^[34]. The statistical data of faults in China and abroad show that there is a power exponential relationship between the maximum displacement and the extension length in the process of fault segment growth^[35,36]. In the process of fault segment growth, the maximum displacement increases linearly with the extension length, that is, the fault extension length increases correspondingly in the process of fault displacement accumulation. Therefore, the maximum throw subtraction method can reflect the history of fault growth and evolution more veritably^{[26-27, 32]}.

The reliability (spatial validity) of fault interpretation directly affects trap identification, mainly in two aspects. First, if several segmented faults are interpreted as a large fault, the trap area will be overestimated, leading to drilling failure at the position without trap^[25]. Second, faults which are difficult to identify due to the lower seismic resolution are called sub-seismic faults (including seismic fault tips and low-order faults). Low-order faults are often combined with large-scale faults to form fault traps. The accurate interpretation of seismic fault tips directly affects the judgment of whether the cross fault is connected, and then affects the judgment of effective fault trap.

2.1.1. Application of displacement/separation transform method to enhance the interpretation reliability of fault combination

Referring to the data available around the world^[37,38] and 3D seismic data in Bohai Bay Basin and Songliao Basin, the displacement/separation transform method is proposed to enhance the interpretation reliability of fault combination. Fault displacement (D) refers to the sum of the displacements of two faults in the center of the overlapping fault segment, and separation (S) refers to the vertical distance between two faults in the center of the overlapping fault segment. When the ratio of D to S is smaller than 0.27, the fault segment is in the stage of lateral overlap, that is, the stage of "soft linkage". When the ratio of D to S is 0.27-1.00, the fault segment is in the stage of initial rupture, that is, the stage of "transitional linkage". When the ratio of D to S is larger than 1.00, the fault segment is in the stage of complete rupture, that is, the stage of "hard linkage".

The f1 segment of Nandagang fault in Qikou sag has four segment growth points (Figs. 1a and 3), which control the formation of four fault traps (Fig. 1a). Also, the f2 segment links to the f1 segment to form the cross fault trap in Guantao Formation (Fig. 3). Statistics of the ratio of D to S of the fault growth points and f1 and f2 overlapping segment show (Table 1): First, the ratio of D to S of all segment growth points in f1 segment are more than 1, which may indicate the stage of hard linkage and high certainty of the fault combination. Second, the ratio of D to S of the f1 and f2 overlapping segment is 0.4, indicating the stage of transitional linkage. Thus, it is uncertain whether f2 is connected to f1.

2.1.2. Prediction of the extension length of fault tips using the displacement gradient method

Seismic fault tip refers to the part of fault between the fault tip which can be identified seismically and the actual fault tip. Its extension length can be predicted by the ratio of seismic resolution to displacement gradient at fault tip (R/G), namely, the displacement gradient method^{[25, 39]}. The throw-distance curve shows that f2 tends to propagate towards f1 (Fig. 4). As it is uncertain if the f2 and f1 overlap, it is necessary to use R/G to predict the extension length of the tip of f2 and judge whether f2 intersects with f1. Spectrum analysis shows that the Guantao Formation has a dominant frequency of 30 Hz, wave velocity of 2400 m/s and vertical resolution of about 20 m. The displacement gradient at the fault tip calculated from the throw-distance curve was 0.089, and the extension length at the tip of f2 predicted by the R/G method was about 223 m (Fig. 3). Therefore, f2 doesn’t intersect with f1 to generate a cross fault trap (Fig. 5), and consequently, they fail to create effective closed boundary. The seismic profile through the overlapping segment also verifies that f2 of Nandagang fault fully penetrates Sha-1 Member but fails to cut through the bottom of Guangtao Formation (Fig. 6). Thus, there is no cross fault trap in Guantao Formation.

When an antithetic fault begins to form, nose-like structure (fault trap) would be formed in its footwall. The active period of antithetic fault is the formation period of the trap, and the formation finalizes in the period when the fault activity is ended. For synthetic fault, under the effect of segment growth and linkage, fault trap can be formed in the hanging wall only when segment growth faults enter the stage of hard linkage, so the initiation of hard linkage represents the beginning of trap formation. Therefore, it is necessary to evaluate the temporal validity of fault trap formation. The maximum throw subtraction method can effectively restore the history of fault formation and evolution, so as to determine the validity of the fault trap.

Hydrocarbons in Qikou sag have mainly accumulated in two stages: the end of Dongying Formation deposition, and the end of Guantao Formation deposition to the present. The second stage represents the main reservoir-forming period in this area^[39]. The throw-distance curve and distribution of ancient faults in the hydrocarbon accumulation period were restored by inversion with the maximum throw subtraction method. It can be seen (Fig. 1e, 1f) that, traps ①, ②, ③ and ④ along the Nandagang syntropic fault are effective, whereas the trap ⑤ is ineffective. Actual production test showed that traps ①, ② and ③ produced oil and gas, but the trap ⑤ failed to produce oil and gas, indirectly proving the reliability of the back-stripping result and also verifying that the formation period of synthetic fault trap is controlled by the hard linkage period of fault segment growth. The Koucun fault is a typical antithetic fault. The inversion by the maximum throw subtraction method shows that the Koucun fault was a large connected fault at the vital moment of reservoir-forming. Hence, traps ① and ② are effective (Fig. 1g, 1h). Large oil and gas reserves have been found in the trap ②, further confirming that the formation time of fault trap is effective.

The lateral sealing of faults affects the hydrocarbon-bearing properties of fault traps and decides the oil-water distribution pattern of fault traps. The evaluation of fault lateral sealing is one of the important parts in pre-drilling risk assessment of fault traps. Essentially, fault sealing refers to the differential permeability between the fault zone and surrounding rocks^[40]. According to the factors causing differential permeability, fault sealing can be classified into 3 types: (1) lithology juxtaposition sealing^{[41,42,43,44]}; (2) fault rock sealing, including mudstone smearing, layered silicate-frame fault rock, and cataclastic rock sealing; and (3) cement sealing^[45,46].

Knipe^[42] proposed that the reasonable prediction of the lithologic juxtaposition between two sides of the fault and the shale content of the filling material in the fault zone was the core of the fault sealing study, and shale thickness and fault throw jointly controlled the shale gouge ratio (SGR). SGR is a means to calculate the shale content with the minimum error^[47]. The quantitative relationship between the maximum height of hydrocarbon column and the SGR of fault zone can be used to quantitatively evaluate the lateral sealing capacity of fault^[47,48].

Take the southern Qikou sag as an example, the quantitative evaluation of fault rock sealing can be done in 4 steps: (1) Refine the interlayer distribution and oil-water units of known reservoirs and determine the height of hydrocarbon column, oil-water contact (OWC), trap amplitude and other data (Fig. 7); (2) Determine the across fault pressure difference (AFPD) of oil-water units by the depth-pressure profile of two sides of fault (Fig. 8); (3) Determine the SGR value at each point on the fault plane by using the TrapTester software (Fig. 9); (4) Build the SGR-AFPD quantitative relationship chart by the statistical method with consideration to the actual height of hydrocarbon column, AFPD value and SGR value (Fig. 10).

The evaluation results show that the minimum SGR value of sealing fault in southern Qikou sag is 20%, and the faults with SGR lower than this value may have high risk of leakage.

Under the same stratigraphic sequence conditions, with the increase of fault throw, the maximum SGR of the fault plane decreases, and the minimum SGR increases; finally, they balance out and reach a stable level (Fig. 11a). However, the SGR values are quite different under different sand-formation ratios: the lower the sand-formation ratio, the higher the SGR value is. In order to facilitate the evaluation, based on statistics of SGR-throw under different sand-formation ratios (Fig. 11a) and the sealing capacity quantitative evaluation model (Fig. 10), the criterion for assessing the sealing capacity of faults in Qikou sag has been set up to predict the height of hydrocarbon column sealed by the fault based on sand-formation ratio and fault throw data (Fig. 11b).

In the brittle caprock interval, faults mainly reflect the process of rupture, and a series of faults and fractures can often be seen both macroscopically and microscopically, which destroy the integrity of caprocks. With the increase of fault throw, fractures increase in density gradually, till interconnecting to form faults. Oil and gas migrate vertically through faults^{[16, 49]}. The vertical connectivity of fractures is controlled by strain (a function of caprock thickness and fault throw). Therefore, the concept of caprock juxtaposition thickness (CJT) is proposed to quantitatively characterize the vertical connectivity of fractures^[13]. The larger the CJT value, the poorer the vertical conductivity is. In the brittle-ductile caprock interval, faults are dominated by typical mudstone smear structure, but have a large number of sub-seismic faults and fractures seen microscopically. With the increase of fault throw, faults and fractures may connect vertically^[24]. Whether faults and fractures connect also depends on fault throw and caprock thickness. Thus, the shale smear factor (SSF) is used to predict the degree of smear^[50,51]. The larger the SSF, the more likely vertical adjustment or damage of oil and gas reservoir will be^[26]. In the ductile caprock interval, faults are dominantly subject to plastic deformation mechanism, but the evaluation method of shale being ductile has not been reported. Based on the research results of gypsum salt rock, the ductile caprock has good integrity, and faults cannot break through the regional caprock of this kind and are generally sealed. The caprock brittle-ductile change is fundamental for the evaluation of vertical sealing of faults.

Fine reservoir dissection is an important basis for the establishment of criteria for evaluating vertical fault sealing, and its key is to determine the migration and accumulation history of the reservoir in concern. Quantitative grain fluorescence (QGF) is an effective technique to identify ancient and present oil-water contacts and restore the evolution history of reservoirs. Generally, ancient oil layers have QGF index over 4, and ancient water layers have QGF index below 4. Extract samples of present oil layers have the quantitative grain fluorescence (QGF-E) intensity of over 40 pc, while extract samples of water layers have QGF-E intensity below 20 pc generally^[52]. According to the distribution pattern of oil and gas and the analysis of geochemical tracing evidences, it is considered that fault control on the vertical "petroleum system" appears in three forms (Fig. 12): (1) Oil and gas accumulate under the caprock only, and there is no oil/gas show in shallow layers. For example, the fault segment controlling trap ① of the Nandagang fault is sealed vertically (Fig. 12a). (2) Oil and gas accumulate above the regional caprock, deep formations have oil/gas shows in logging, but no hydrocarbon accumulation to form pool. For example, Well Zhuangqian-12 encountered a fault trap with QGF index larger than 5, confirming the existence of ancient reservoir in the deep formation. This indicates that the fault controlling the fault trap is not sealed vertically (Fig. 12b). (3) Oil and gas accumulate both above and below the caprock. For example, the trap ③ of the Nandagang fault has QGF index generally greater than 5, and oil/gas shows in logging, proving the existence of ancient oil reservoir. But affected by late fault reactivation, some oil and gas have adjusted vertically, resulting in the hydrocarbon accumulation both above and below the caprock (Fig. 12c).

According to the statistics of 31 fault reservoirs in southern Qikou sag, the SSF method is suitable to evaluate the vertical sealing of faults in the caprock of the middle part of the first member of Shahejie Formation (middle Sha-1 sub-member) of Paleogene, and the critical value of SSF is 3.5 (Fig. 13a), while the CJT method is suitable to evaluate the vertical sealing of faults in the caprock of the second member of Paleogene Dongying Formation (Dong 2 Member), and the critical value of CJT is 70-80 m (Fig. 13b). Although rock mechanic parameters were not used to discriminate the brittleness-ductility of caprocks, the vertical sealing evaluation based on reservoir dissection indirectly reflects the differences of mechanical properties of caprocks at different depths, so different methods are suitable for evaluating the vertical sealing of faults in the caprocks.

Traps bounded by the synthetic fault and the antithetic fault were quantitatively evaluated using the critical SSF for vertical sealing of middle Sha-1 Member. For the Nandagang synthetic fault, both the trap ① and trap ② have SSF below the critical value 3.5, indicating the fault is vertically sealed (Fig. 14a, 14b). The trap ③ has SSF greater than 3.5, so a part of hydrocarbons has migrated to shallower layers. Since the SSF at structural high of the trap is less than 3.5, hydrocarbons accumulate beneath the caprock, which resulted in failure of Well Qinan-3 (Fig. 14a, 14b). The trap ④ has SSF larger than 3.5, so hydrocarbons have migrated to shallow layers, and hydrocarbon pool cannot be formed beneath the caprock. For the Koucun antithetic fault, the trap ① has fault throw greater than that of trap ② (Fig. 1c), and the SSF greater than 3.5, so vertical leakage of the fault has led to the migration of hydrocarbons to shallow formations. The trap ② has SSF smaller than 3.5, in other words, the trap is vertically sealed, so oil and gas accumulate beneath the caprock (Fig. 14c, 14d).

The validity evaluation of fault traps in rift basins involves three aspects: (1) Temporal-spatial validity of fault traps. As far as the process of fault growth and linkage is concerned, under the same stratigraphic sequence conditions, the stronger the fault activity (the larger the fault throw), the more likely it is to form effective fault traps. It is necessary to determine the reliability of fault trap interpretation and the validity of formation time. (2) Lateral sealing validity of fault. Under the same stratigraphic sequence conditions, the larger the fault throw is, the stronger the lateral sealing capacity of the fault is. Reservoir dissection has confirmed that fault reservoirs have a risk of lateral leakage. The faults in Qikou sag have lateral sealing SGR of 20% which is lower than the critical value, so they have high risk of lateral leakage. (3) The possible damage of later fault reactivation to hydrocarbon accumulation, that is, the vertical sealing validity of fault. The larger the fault throw is, the easier the caprocks will be destroyed. If the caprock is destructed, oil and gas would adjust vertically or dissipate. Considering all of the above factors, it is necessary to establish a comprehensive evaluation system for hydrocarbon-bearing validity of fault traps in rift basins to guide the prediction of evaluation of interpretation reliability and formation time effectiveness of fault trap, evaluation of lateral sealing effectiveness of fault, and evaluation of vertical sealing effectiveness of fault (Fig. 15).

On the basis of 3D seismic data, the displacement/separation transform method can be used to effectively judge the linkage stage of fault plane, and the displacement gradient method can be used to determine the extension length of fault tip and identify the growth and linkage state of fault plane, so as to clarify the validity of present fault trap interpretation.

According to the mechanism of fault differential activity, it is proposed that the antithetic fault trap is formed and developed in the whole faulting period, while the synthetic fault trap begins to form in the stage of hard linkage. The maximum throw subtraction method can effectively restore the process of fault formation and evolution, and in turn the validity of fault trap formation time can be sorted out.

The sandstone-mudstone interbedded formations of rift basins are mainly sealed by fault rock, and the height of hydrocarbon column sealed by fault rock depends on the minimum SGR distribution location and sealing capacity. Based on the SGR-AFPD quantitative relationship of known reservoirs in southern Qikou sag, the evaluation method of fault rock sealing has been established, and the critical value of SGR for fault lateral sealing is 20%. Moreover, the quantitative evaluation chart based on the relationship between fault throw-sand-formation ratio and hydrocarbon column height has been constructed, which can quickly evaluate the validity and capacity of fault lateral sealing.

In the brittle caprock interval, rupture deformation is dominant, and fracture connectivity determines the extent of leakage, so the caprock juxtaposition thickness (CJT) can be used to quantitatively evaluate the vertical sealing. In the brittle-ductile caprock interval, smearing deformation is dominant, and the continuity of smearing determines the extent of leakage, so the shale smear factor (SSF) is used to quantitatively evaluate the vertical sealing. Based on the distribution pattern of oil and water and the results of QGF tracing tests, the criterion for evaluation of fault vertical sealing has been established. The SSF method is applicable to evaluate the vertical sealing of faults in the middle Sha-1 sub-member in southern Qikou sag. Faults with a SSF value of less than 3.5 are sealed vertically, which is conducive to hydrocarbon accumulation in deep layers. The CJT method is applicable to evaluate the vertical sealing of faults in Dong-2 Member in southern Qikou sag. The critical CJT is 70-80 m. Faults with CJT greater than that range are vertically sealed, which is conducive to the preservation of oil and gas.