Identification of fractures in tight-oil reservoirs: a case study of the Da'anzhai member in the central Sichuan Basin, SW China

The Da'anzhai Member of the Jurassic Ziliujing formation in central Sichuan is a typical tight-oil reservoir with porosity and permeability less than 2% and 0.1 × 10–3 μm2, respectively. Fractures in this formation are well developed in micro- and nano-scale. However, the factors that control the fracture distribution are unclear. Additionally, the uncomprehensive and ineffective identification and evaluation of fractures in the early stage of tight-oil development makes it difficult to meet the requirements of tight-oil development. In our work, we used cores, thin sections, and a scanning electron microscope (SEM) to study the influence of the microscopic rock composition, including the shelly grains, calcite grains, and clastic grains, on the fracture development. We found that the microscopic composition of shelly grains and calcite grains separately control the development of inter-shelly fractures and shelly fractures, and intergranular fractures, and tectonic fractures. Except for a small number of dissolution fractures found in mudstone, the fractures are not well developed in the formations with clastic grains. According to the characteristics of the development degree of fracture and the resolution of the well-logs, the fractures are divided into large scale, small scale, and micro-scale. By a newly established level-by-level constraints method, we systematically identified the scale, occurrence, filling characteristics, and development degree of fractures in the Da'anzhai member by well-logs. Moreover, a quantitative model is also proposed for identifying the angles and development degree of fractures. The results show that the scale of fractures can be effectively identified by the shapes and values of resistivity logs; the occurrence, development, and filling characteristics of fractures can be semi-quantitatively evaluated by the relative amplitude difference between the matrix resistivity (Rb) and formation resistivity (RT). The results are consistent with the interpretation results by formation micro-resistivity imaging (FMI) log, which further demonstrates that the level-by-level constraint method by conventional well-logs can be used to systematically and effectively predict the fracture characteristics in tight-oil reservoirs.


Geological background
The tectonic basin of Sichuan is generally diamond-shaped. It can be divided into three tectonic regions bounded by the anticlines of the Huaying Mountain and the Longquan Mountain: (1) the high-steep slope tectonic zone of the southeastern Sichuan, (2) the low-steep depression tectonic zone of the western Sichuan, and (3) the lowlying uplift tectonic zone of the central Sichuan 35 . The oil and gas region is located in the low-lying uplift tectonic zone of central Sichuan, which is between Longquan Mountain and Huaying Mountain (Fig. 1). Low-amplitude folds dominate the structure of the oil and gas region in various directions and by a few faults 5 . The Da'anzhai member, the main oil-bearing area of the tight-oil reservoirs in central Sichuan, has sedimentary facies typical of continental inland freshwater lacustrine facies deposits. The primary sedimentary environment is a beach facie with shelly grains, and it has a complete longitudinally regression-transgression sedimentary cycle 6 . The lithofacies is mainly interbedded of lacustrine mudstone and shelly limestone with different thicknesses, which have undergone multiple diageneses such as sediment compaction, dissolution cementation, and recrystallization 6 . The average helium porosity and alcohol porosities are 1.77% and 1.04% respectively. The porosity range is 0.26% to 4.31%. The geometric mean of permeability is 0.02 × 10 -3 μm 2 with a range of 0.002 μm 2 to 12.62 × 10 -3 μm 2 . The characteristics of ultra-low porosity and low permeability demonstrate that the Da'anzhai member is a typical tight-oil reservoir. www.nature.com/scientificreports/ In the early stage of the oil exploration and development, the Da'anzhai member was initially considered to be a fracture-type reservoir. Fractures are the reservoir spaces and percolation channels. Thus, the key point for oil exploration was to find fractures with large-scale 7 . With the promotion of oil exploration, both the pores and fractures were found to be oil-bearing, and the Da'anzhai member should not be simply regarded as the fracturedtype reservoir. The pores with various types contribute to long-term stable production. Based on the coexistence of the microfractures and matrix pores, the Da'anzhai reservoir was initially defined as a matrix fracture-pore reservoir. In the matrix state, micropores and microfractures are reservoir spaces, and the connections among microfractures provide the channels for the oil migration 36 .

Basic characteristics of the fractures in the Da'anzhai member
Characteristics of the fractures. The 178 cores at the production site indicate that the fractures in the Da'anzhai member are mainly structural. The structural fractures in the Da1 sub-member account for 57% of the total fractures, followed by compression-dissolution and interlayer fractures. The Da13 and Da3 sub-members have more than 60% structural fractures, 25% compression-dissolution fractures, and 10% interlayer fractures.
About 80 percents of the fractures are horizontal and low-angle. The Da1, Da13, and Da3 sub-members account for about 55% of the horizontal fractures. The Da1 and Da3 sub-members have about 25% of the lowangle fractures. With 36% high-angle fractures, the high-angle fractures are more common in the Da13 sub-member. The statistics of the 52 cores show that the filling situation of the fractures are 55% closed, 23% semi-filled, and 21% open. The closed fractures and the semi-filled fractures are mainly filled with mud, followed by calcite.
The development degree of fractures with large-scale are analyzed by the thin sections and cores. As shown in Fig. 2, the linear density mainly ranges from 5 to 10 fractures per meter. The linear density has a wide range distribution with the maximum linear density approaching 40 fractures per meter. The surface rate of fracture observed from thin sections is mainly in ranges of 0.2-0.6%.
The characteristics of fracture development in the Da'anzhai member are controlled by the composition and structure of the rocks 7 . The primary organisms contributing to the formation consist of bivalves, gastropods, ostracods, and other shelly microorganisms. A small number of internal clastic particles are included, such as sand and rock debris. The main rock types are shelly limestone, clastic limestone, grain limestone, mudstone, and sandstone. The images of the thin sections and scanning electron microscopy show that ( Fig. 3): (1) For the shelly limestone with shell contents over 50%, the inter-shell fractures are often developed along the edge of the shells. The shell size and numbers greatly affect the numbers and scales of fractures. The shelly limestone is tight with mainly inter-shell fractures and shell fractures developed. The development and orientation of the inter-shell fractures affect the seepage capacity of the oil; (2) For the shelly-clastic limestone, the split of the shell inevitably results in the connection of the inter-shell fractures. The split degree of the shell affects the development of the micro-nano scale fractures; (3) For the grain limestone with high calcite content, the types of the minor mesh-intergranular fractures primary develop. The width of these kinds of fractures ranges from 300 nm to 1000 nm. The grain limestone has a strong rock brittleness. It is easy to form structural fractures as to the intergranular fractures, along the weak surfaces under the crustal stress. The size of the crystal grains will affect the scale of the intergranular fractures.
(4) For the mudstone and sandstone, the primary and secondary intergranular pores develop, and few fractures are observed. The dolomite content in the Da'anzhai member is generally low. The dolomitization process only enlarge the intergranular pore boundary and enhance the pore connectivity, whereas it does not have an obvious influence on the fractures.
Classification of the fractures. The Da'anzhai member is in the middle-deep diagenetic stage with a medium-deep burial environment. In the early diagenetic stage, strong compaction makes the loss of the primary porosity. There are clay mineral metasomatic particles, calcite, and a small amount of siliceous cementation. During the middle-deep burial diagenesis stage, the dissolution and compressional dissolution is intense. The edges of large structural fractures are dissolved, and the compression-dissolving fractures form in the limestone or at the boundary of limestone and mud 6 . Considering the impact of different scale fractures on hydrocarbon migration and conventional logging resolution, the Da'anzhai member's fractures are classified as largescale, small-scale, and micro-scale fractures. www.nature.com/scientificreports/ As shown in Table 1, the large-scale fractures are visible in the cores, and they are primarily non-fabric controlled fractures. The number of large-scale fractures is small. The small-scale fractures are visible in the thin sections. This kind of fracture includes the structural and dissolution fractures that run through the thin sections and inter-shell fractures with large openings. These fractures are more plentiful than large-scale fractures and are essential for connecting large-scale and micro-scale fractures. The micro-scale fractures have poor connectivity. Scanning electron microscopy shows that irregular and discontinuous fractures appear on the edges of the calcite crystals and shells. The fractures are mostly shell-controlled and matrix-controlled. When the micro-scale fractures are well developed, they interlace and play an essential role in matrix permeability.
Overall, the micro-scale and small-scale fractures in the Da'anzhai member are highly developed. The oil from the source rocks, such as mud shale, enters the small-scale fractures from the large-scale fractures and then enters the micro-scale fractures. The fractures at different scales are effective channels for oil percolation.
Level-by-level constraints in fracture identification. For the Da'anzhai member, the conventional well-logs mainly contain AC and dual laterologs, and the CNL and DEN logs are always unavailable. It is challenging to identify fractures systematically by using conventional well-logs. Thus, the level-by-level constraint method is proposed to identify the fracture scale, occurrence, filling characteristics, and development degree.   www.nature.com/scientificreports/ (1) Exclusion of thin layers: Different logging instruments have different resolutions for measuring the thickness of a rock layer. The well-log only reflects the real formation information when the layer thickness is greater than the instrument's resolution. Generally speaking, the resolution of the gamma-ray log(GR) is about 30 cm and the resolutions of AC are 60 cm for limestone and 100 cm for mudstone. The resolutions of CNL and dual laterologs are approximately 40 cm and 80 cm, respectively. Considering the well-logs' resolutions, a rock layer with a thickness greater than 1 m was selected for fracture identification. The logging values of the thin interbedded layers of the Da13 sub-member are affected by the surrounding rock, so they cannot represent the actual response characteristics of the rock layer.
(2) Exclusion of mud impact: For mud bands, its response is similar to large-scale fractures with high values of AC and CNL and low values of formation resistivity(R T ) and flushed zone resistivity(R XO ). These mud bands can be excluded using a combination of GR and AC. For the mud-filled, the resolution of GR is commonly much larger than the fracture width. Thus, GR has no obvious response to the mud-filled fractures, that is the mud-filled fractures have a low value of GR. The well-logs are characterized by a smooth box or microdentation box type shape. However, the values of GR for the mud bands increase significantly and AC has a dentation-finger shape. The above characteristic as to the mud bands and mud-filled fractures are shown in well HC125-17-H ( Fig. 4). At the depth of the 1481 m to 1483 m, the fractures are well-developed. These fractures are mud-filled as displayed in the core. The lithology is fine-grained limestone. The well-log shows a low value of GR with approximately 23 API. Correspondingly, the values of AC and CNL increase significantly, while the R T values decrease significantly. A dissolution fracture at the depth of 1486 m shows a similar logging response to the mud-filled fracture. The logging response of the dissolution fracture has the characteristics of the low GR and RT, and high AC, while no obvious DEN changes. At the depth of 1488-1490 m, a typical mud band develops. Compare to the low GR values of the mud-filled fracture and dissolution fracture, the GR value for the mud band is high. Except for the different GR characteristics, the mud band has similar characteristics of AC, CNL, and R T to mud-filled fracture and dissolution fracture.
(3) Influence of wellbore stability: During the drilling process, wellbore instability may occur due to factors such as geological structure, stratum in-situ stress, and weak structural planes. For the intervals with borehole expansion, AC, DEN, and CNL values will be affected. The values of the CNL and AC will increase, while the DEN values will decrease. When the borehole expands dramatically, the values of R T and R XO will also decrease, indicating a similar well-log response to fractures. The influence caused by the borehole expansion can be excluded by using the caliper log (CAL).

Secondary constraints (lithologic constraints).
(1) Lithology constraints on fractures After considering the first-level constraints, lithological constraints are imposed on the analysis of fracture development. First, it is necessary to identify the lithologic characteristics of the intervals where fractures are found. The lithology is the background of the log response and the most fundamental factor for controlling the development of fractures. The main lithology of the Da'anzhai member is limestone, mudstone, sandstone, and a small amount of dolomite. www.nature.com/scientificreports/ For the sandstone, calcareous sandstone, and the sandy limestone filled with quartz in the Da'anzhai member, intergranular pores are mainly found among quartz grains, clay minerals, and calcite. The foamy intragranular pores are found within quartz grains. The fractures are not developed and are rarely found in the sandstone. For the mudstone in the Da'anzhai member, the reservoir space has intragranular and intergranular pores in clay minerals. The pore size is in the range of 0.1 μm to 1 μm. The pore morphology is irregular, which is related to the clay minerals and hard particles. The development degree of fracture is low, which is mainly found in the Da13 sub-member. The primary type of fractures in the mudstone are pressure-solution fractures, formed by differential compaction and chemical solution pressure among components. Because of the distortion of the well-log and the low width of the pressure-solution fractures, the pressure-solution fractures in the mudstone cannot be identified.
According to the analysis above, we mainly identified fractures in limestone and dolomite in the Da'anzhai member. Since the dolomite only occupies a small amount of the Da'anzhai members, we focus on the fracture identification in limestone.
(2) Characteristics of primary lithology response in fracture zones From the lithology analysis of the Da'anzhai member, it can be seen that the limestone in the Da'anzhai member is widely distributed, and its structure is mainly composed of calcite crystals and shells, including shells from bivalves and gastropods. The original material is carbonate minerals, including aragonite and calcite. In the process of burial diagenesis, the original aragonite and high-magnesium calcite are converted into low-magnesium calcite. The lithology becomes more homogeneous, and the shale content decreases, which in general, leads to high values of the RT. The well-development of micro-nano-scale reservoir spaces leads to porosity and permeability less than 2% and 0.1 × 10 -3 μm 2 , respectively. The ultra-low porosity and low permeability lead to the high RT value and low AC value. In general, due to the high calcite, low shale contents, and ultra-low porosity and permeability, the RT value of limestone is generally greater than 5000 Ω•m, the AC values are approaching 48 μs/ft, the GR values are low. For the thick limestone layer, the conventional well-logs have a box-shaped form with smooth or microdentation change. Figure 5 shows that the scale and origin of fractures can be judged after applying the lithology constraints.
The types of fractures that can be identified need to be determined, and they need to be classified as large, small, or micro scale. After this, within the constraints of large-scale fractures, the fracture filling and fracture occurrence are identified. Finally, the development degree of the micro-scale and large-scale fractures is determined. The purpose of level-by-level constraints is to exclude the influence of the non-fracture features on the logging curve and to locate the fracture characteristics to be investigated at a comparable level.

Identification of fracture characteristics. Scale identification of fractures.
(1) Logging response characteristics The interlaminar fractures, which cannot be effectively identified due to the limited resolution of conventional well-logs, are formed in the thin lime-mud interbeds of the Da13 sub-member. For large-scale structural fractures and dissolution fractures, the mud invasion into the opened fractures will result in a relative decrease of resistivity. The decrease in resistivity is related to fracture width, filling, and radial extension. The R T values for the large-scale fractures decrease dramatically, having a dentate shape or a finger shape at the background of high resistivity values. The reduced resistivity is lower than 3,000 Ω▪m. www.nature.com/scientificreports/ exhibiting a dentate drop compared to the matrix rock background, which often drops to a gap shape, and the R T curve becomes a "platform notch type", as shown in Fig. 6.
In limestone, the main types of micro-scale fractures are intergranular fractures and interlaminar fractures. Similar to pores, micro-fractures are spaces in reservoirs with poor connectivity that are mainly found in pure grain limestone and shell limestone with little shale content. Through calibration of a large number of core and thin sections to well-logs, the micro-scale fractures are characterized by low values of GR and AC. The R T values are equal to or close to the resistivity of matrix rock. The R T log shows "finger-type" and "platform double-track type" shapes. Finger type: In thick limestone, the R T curves for micro-scale fractures show a finger peak, and the R T values are equal to or close to the R T values of the matrix rock. Due to the difference in lithology and thickness, the R T values often have high, medium, and low amplitudes. The shape of the GR curve is smooth box-type with GR value increasing slightly. The AC curve is smooth and box-shaped. Compared with adjacent rock strata with good connectivity or high mud content, the R T curve is "finger-type" because of the poor connectivity of the micro-fractures, as shown in Fig. 7a,b. Platform double-track type: Double rails appear in the lateral resistivity, with a smooth box-shaped curve and a high resistivity value of 6,000 Ω▪m. Micro-fractures are formed by stress release in dense intervals. The general shape of the GR and AC curves is box-shaped. The curves have micro-dentation with values increasing slightly, as shown in Fig. 7c,d. The double-rail phenomenon of micro-scale fractures is caused by the deformed high-angle fractures along weak structural planes. At the same time, due to the widespread oil-bearing in the Da'anzhai member, the mud invasion in a limestone formation has a low resistivity, forming a "platform double rails" phenomenon for the micro-scale fractures. Identification of fracture occurrence.
(1) Logging response characteristics Within the constraints on large-scale fractures, an analysis of logging response characteristics shows that the response for high-angle and oblique fractures is unclear. In contrast, the response for low-angle fractures and horizontal fractures is obvious. AC and R T values decreased, and their curves are finger-shaped. Figure 9 shows that the GR log curve is box-shaped and smooth for low angle and horizontal fractures. The R T curve is often reduced to a spike-like or dentate shape. There is either no amplitude difference or a slight negative amplitude difference. The AC values tend to increase, yielding an AC curve with a spike-like or dentate shape. According to their logging curves, the RT values are lower for the oblique fractures, and the AC values are slightly higher. For high-angle fractures in the Da'anzhai member, due to their low development degree, intense filling, and small opening, the well-logs show matrix rock characteristics.
(2) Cross-plot identification   www.nature.com/scientificreports/ As the horizontal fractures have a finger-like decline on the R T log, high-angle fractures have no obvious response. The fracture occurrence can be distinguished by using the relative difference between logarithmic matrix rock resistivity (logR b ) and logarithmic formation resistivity (log R T ). On the cross plot of (logR b -logR T )/logR b versus (logR b -logR T ), the high-angle fractures, and the horizontal open fractures form a distinct partition. The relative difference of the matrix rock resistivity for the horizontal fractures is larger, (logR b -logR T )/logR b > 0.15. The relative difference between logR T and logR b for horizontal closed fractures is close to 0. It should be pointed out that (logR b -logR T )/logR b increases when the horizontal fracture opens is filled with mud or when the corresponding cores develop dissolution cavities, and the data points will be distributed in the horizontal fracture range. The (logR b -logR T )/logR b versus R T cross plot shows the differences among horizontal fractures, high-angle fractures, and closed fractures, as shown in Fig. 10.

Identification of fracture filling conditions.
(1) Degree of fracture opening We found that the R T values of closed fractures are greater than 800 Ω·m, and (logR b -logR T )/logR b is less than 0.1. The opening degree of fractures is classified as a high-degree opening, low-degree opening, half-open, and closed. As shown in Fig. 11 www.nature.com/scientificreports/ mud-filled fractures are higher than those of the non-filled open fractures. The GR values of the non-filled fractures are slightly higher than those of the filled fractures, and their R T values are low. The specific classification criteria are shown in Fig. 12 and Table 2.

Identification of the development degree of fractures.
(1) logging response characteristics The development degree of fracture, especially the open fractures will affect the logging response reservoir effectiveness. Therefore, in the process of quantitative identification of the development degree of fracture,   it is a large-scale fracture, a small-scale, or a micro-scale fracture, the higher the development degree of a fracture is, the more distinct the decrease of the resistivity is. The AC values will increase with an increasing development degree of fractures. Figure 13 shows that for large-scale fractures, increasing linear density leads to increasing AC values and decreasing R T values, which can be reduced to several hundred Ω•m. Surface fracture rates and well-logs can be used to quantitatively analyze the development degree of micro-scale and small-scale fractures.    www.nature.com/scientificreports/ In general, the development degree of micro-and small-scale fractures has no obvious effect on GR and AC curves, while an obvious effect on the R T . Increasing surface fracture rates lead to decreasing absolute values of R T . The R T values of micro-and small-scale fractures are lower than that of matrix rock, and their R T log has a dentate shape.
(2) Quantitative identification of the developmental degree Based on the different characteristics of resistivity of large-scale and small-scale fractures with different development degrees, we selected the difference between the R b and the R T values to analyze the linear density of large-scale fractures. Moreover, we selected the difference of amplitude between the R T and R XO values to identify the development degree of small and micro-scale fractures. Due to the low resolution of the well-logs, it was difficult to identify the surface fracture rate of micro-scale and small-scale fractures quantitatively by conventional logging curves. The higher the difference between the R T and R XO values, the higher the development degrees of the microand small-scale fractures. The boundary value between the high and the low development of fractures for logR T -logR XO is 0.1. For the development of large-scale fractures, we found that the linear density has a good positive correlation with (logR b -logR T )/logR b . The higher the linear density, the greater the reduction of the R T values is and approach to the matrix resistivity. For the same AC, the higher the linear density of fractures is, the lower the R T value is. The specific identification criteria are shown in Fig. 14 and Table 3.

Discussion
Based on the identification criteria for fracture scales and fillings and the models of fracture angles, development degree, openness, porosity, and permeability (Table 4), the fracture parameters can be calculated. Taking the well LQ2 as an example, the interpretation correctness and the applicability of the level-by-level constraint method is discussed as shown in Fig. 15. At the depth of 2101 m to 2102 m, the resistivity is extremely high, the AC values and the CNL values are low, and the DEN values are high, showing obvious compact lithologic response characteristics for conventional welllogs. By the level-by-level constraint method, we identified micro-scale fractures as the predominant fractures in this layer. The static images of the FMI show high-brightness and compact intervals, while the dynamic images   Table 4. The models of fracture interpretation. For the value of the fracture cementation index m f , the m f of the large-scale fractures is 1.04, and the m f of the structural fractures is 0.99 (derived from the rock data).

Model type Formula Parameter description
Porosity At 2102 m, the R T curve shows a marked decline with a dentate shape, the CNL values show a significant increase, and the AC values also increase. By using the level-by-level constraint method, the fractures are identified as large-scale open fractures with low angles and no filling. The development degree of fracture can be characterized as five fractures per meter, average fracture porosity of 0.002%, and fracture width of 10 μm. Both the static and dynamic FMI images show six horizontal open fractures distributed in black bands. The logging interpretation by conventional well-logs is consistent with the interpretation by FMI. Additionally, the comparison shows that the level-by-level constraint method based on conventional well-logs is suitable for interpreting the fracture intervals.
At the depth of 2102 m to 2108 m, the CNL values increased continuously, and the R T values decreased significantly compared to the values from 2101 to 2102 m, but there are similar fracture response characteristics. The increased range of the CNL values and the decreased range of the R T values are lower than those of fractures at 2102 m. The average value of GR is higher than that from 2101 to 2102 m. The conventional well-logs reflect an increase in mud contents. During the identification process by the level-by-level constraint method, the fractures can not be identified due to mud contents and lithology constraints. The static FMI shows a high resistivity and dense lithology of fractures from 2012 to 2018 m, and low resistivity bands can be seen under high resistivity. On the dynamic FMI, dark and black bands are parallel, which is obvious muddy bands. The core shows that the lithology is mesoclastic limestone, and an argillaceous strip can be observed.
Overall, the fracture identification by conventional well-logs at different depths in well LQ2 is consistent with FMI. It shows that the level-by-level constraint method can systematically and effectively interpret the fractures of tight-oil reservoirs in the Da'anzhai member of central Sichuan based on conventional well-logs.

Conclusion
This paper proposes an effective method to evaluate the scale, occurrence, and development degree of the fractures using conventional well-logs. The main conclusions are as follows: (1) The composition and structure of the rocks in the Da'anzhai member will affect the development of the fractures. The composition of the shells controls the development of the inter-shell fractures and the shell fractures. The calcite grain composition controls intergranular fractures and the development of tectonic www.nature.com/scientificreports/ fractures. The fractures in sandstone and mudstone are undeveloped, and only a small number of dissolution fractures are found in mudstone.
(2) Large-scale fractures of the Da'anzhai member can be identified from both AC and RT logs. The R T values of large-scale fractures are lower than 3,000 Ω·m, and the R T curves often have dentate or finger-like shapes. The resistivity curves of small-scale fractures have a platform-notch shape with values of about 6,000 Ω·m. The micro-scale fractures have high resistivity values in finger shape. The smaller the scale is, the closer the resistivity is to the resistivity of the matrix value.
(3) High-angle fractures of the Da'anzhai member are poorly developed and often filled. The high-angle fractures have an unobvious response in the well-logs. The resistivity log for horizontal fractures and low-angle fractures show finger-like decreases, and the AC values increase.
(4) Increasing the linear density of large-scale fractures leads to a decrease in the dentate character. The RT value decreases with the increasing surface fracture rate for micro-scale fractures.
(5) The level-by-level constraint method can be used to identify the fractures systematically and effectively in the tight-oil reservoirs of the Da'anzhai member, which is of great significance to the development of tight-oil in old oilfields.