TOC prediction and grading evaluation based on variable coefficient △logR method and its application for unconventional exploration targets in Songliao Basin

The prediction of total organic carbon (TOC) content and grading evaluation of shale formation are very much significant and essential for reservoir description of rolling exploration and development in the new shale exploration area (Shuangcheng) in Songliao basin, China. In order to improve exploration efficiency and obtain continuous TOC content curve of wells, the variable coefficient △logR technique was developed for TOC estimating which is based on the two of acoustic time difference and deep lateral resistivity logging curve and the variable scale coefficient (K) between them as well as another scale coefficient (A) between TOC and △logR. A prediction model of TOC was established for the well which TOC is measured by evaluation of side wall cores, then apply it to other wells to verify the reliability of the model. The application result of eleven exploration Wells in Shuangcheng area show that the TOC of shale is linearly correlated with △logR, and the maximum prediction accuracy k value varies with wells, so it is necessary to determine the undetermined coefficient k according to a single well, but the A value having no big change from one well to another in similar sedimentary facies and thermal evolution degree of shale. The average relative error of TOC between prediction model and core measurement is 10.6% which verifies the accuracy of this method. On this basis of TOC prediction, we establish shale grading evaluation criteria for the study area. In the establishment process, not only the relationship between TOC and S1, but also vitrinite reflectance (Ro) are considered. The shale in Shuangcheng area can be divided into three types (Class I: TOC > 3.5% and Ro > 0.9%; Class II: TOC 2%–3.5% and Ro > 0.9; Class III: TOC < 2% or Ro < 0.9%), and achieved shale classification on the well profile with TOC and Ro which are easy to predict and reliable. According to the relationship between the thickness of shale of disparate classes and the total thickness of shale in different zones, the thickness of shale of disparate classes in each well is predicted.


Introduction
Organic matter in shale is not only the parent material for directly generating oil and gas resources, but also an indicator of the quality of shale reservoirs and the design of pressure protocols (Passey et al., 2010;Sondergeld et al., 2010;Wang et al., 2015). So, the study of geochemical characteristics of organic matter present within the rock is required to delineate the potential unconventional targets within shales. In the current years, source rock and geochemical data are mainly exclusive datasets produced by internal and/or external laboratories (Aziz et al., 2020). Such data are quite expensive and time taking that are difficult to manage and investigate geographically or statistically (Evenick, 2020). Therefore, by using logging data in addition to identify and evaluate source rocks by experiment testing methods (such as the TOC analysis, rock pyrolysis, elemental analysis, chromatography-mass spectrometry analysis, etc.) to identify the distribution of hydrocarbon source rock and accurately predicate interval shale geochemistry index is necessary (Banerjee et al., 2000;Ghassal et al., 2018;Wang et al., 2018;Cai et al., 2019).
In recent years, some scholars have attempted to evaluate the oilbearing property (chloroform bitumen A) and maturity of shale utilizing logging data (Liu et al., 2014;Zhao et al., 2018), but total organic carbon (TOC) is still the recognized geochemical index that can be accurately evaluated by logging data. TOC content is generally measured by traditional methods such as the evaluation of side wall cores and formation cuttings in a geochemical lab. Data obtained from these experimental techniques are not continuous and also a time-consuming process (Aziz et al., 2020). Therefore, many scholars tried to use different logging-based methods to predict continuous TOC content and improve exploration efficiency. The logging-based methods can be generally divided into three categories. The first category involves establishing a regression analysis model directly with logging curves and geochemical indices which may include one or several radioactive logs (GR,U), porosity logs (Δt, DEN, and CNL), and lateral resistivity logs (Schmoker, 1979;Fertl and Rieke, 1980;Schmoker, 1981;Schmoker and Hester, 1983;Mendelzon and Toksoz, 1985;Fertl and Chilingar, 1988;Zhu et al., 2003;Herron et al., 2011;Hu et al., 2011); The second category involves constructing new parameters that respond to the TOC content, such as the carborlog method (Carpentier et al., 1991), △logR technique (Passey et al., 1990) and several improved △logR models (Liu et al., 2011a;Liu et al., 2015;Hu et al., 2015;He et al., 2016;Zhao et al., 2017;Liu et al., 2019;Zhu et al., 2019); The third category involves predicting the TOC by artificial intelligence, such as neural network, support vector machine, Gaussian process regression and other non-linear methods (Kamali and Mirshady, 2004;Kadkhodaie et al., 2009;Khoshnoodkia et al., 2011;Tan et al., 2015;Shi et al., 2016;Kadkhodaie et al., 2017;Yu et al., 2017;Shalaby et al., 2019).
Field applications reveal that the multiple regression and machine learning methods can provide a high accuracy for training samples Khoshnoodkia et al., 2011;Mahmoud et al., 2017;Yan et al., 2017;Shalaby et al., 2019). However, when the trained models were applied to new wells or new areas, their accuracy decreased because these methods have no theoretical basis and suffer from overftting (Passeyet al., 1990). On the contrary, the ΔlogR method is based on the petrophysical model derivation and includes fewer well logs (redundant information), and its TOC quantification performance is relatively good. Many scholars have focused on the improvement of the ΔlogR model to enhance its TOC prediction accuracy. In addition to replacing the level of organic metamorphism (LOM) using an easily accessible parameter (Tmax or Ro) , almost all of the modifcations have focused on how to overlay the sonic transit time and the resistivity logs. Liu et al. (2015), Liu et al. (2014) noted that using a fixed overlay-coefficient for all of the wells is not always appropriate, because the k value that minimizes the error of the TOC prediction may be different for different wells in different Basins. Hu et al. (2011) also noted that using a fixed overlay-coefficient may lead to large errors in shale formations with large lithology changes. Wang et al. (2015) calculated the overlay-coefficient from theoretical models, in which several key parameters must be determined from the rock minerals and a complex petrophysical model. Although such modifcations have different theoretical bases, their essence is to determine a proper method of overlaying the logging curves and producing a higher correlation between the'ΔlogR'and the core TOC. However, the current methods of determining the overlaycoefficient based on the rock minerals, empirical parameters (sonic transit time of the rock minerals), and complex petrophysical calculations, which are even less accessible than the measured TOC, weaken the applicability or the accuracy of the ΔlogR model (including the modified ΔlogR models).
In this study, we improve the application and accuracy of △logR model related with conventional geochemical data to carry out the prediction of TOC and classification evaluation of shale. A simple and feasible method, the V-ΔlogR model is proposed based on three revisions to the ΔlogR model. i.e., replacing the fixed overlay-coefficient with a variable one, removing the maturity, and replacing the baselines with a single one. Among these, a critical improvement is treating the overlay-coefficient (k) as a variable and determining it from the measured TOC values or from well logs. The application of our improved model in the Songliao basins indicates that the V-ΔlogR model can accurately predict the TOC of shale with large lithological and mature variations. Because it uses actual geological data to determine the overlay-coefficient in the model, the V-ΔlogR model offers more reliable results for test wells than other commonly used models (e.g., the modified Schomoker, multiple regression analysis, modified ΔlogR). The V-ΔlogR model eliminates the need for mineral and maturity data constraints and the overlay-coefficient can be determined using only the well logs, which significantly expands the applicability of the ΔlogR method.
2 Methodology 2.1 Background Passey et al. (1990) proposed a method for calculating the TOC by overlaying a properly scaled porosity curve (generally the sonic transit time curve) over a resistivity curve and baselined them in clay-rich, organics-poor, and fine-grained rock intervals (Liu et al., 2020). The mathematical expressions that were used by Passey for the calculation of ΔlogR summarized in Eqs 1, 2 and widely used for evaluating the unconventional resources (Mahmoud et al., 2017).
Where ΔlogR is the curve separation measured in logarithmic resistivity cycles; R is the resistivity measured in Ω·m using the logging tool; Δt is the measured sonic transit time in μs/ft; and Rb is the resistivity corresponding to the Δtb value when the curves are baselined in nonsource,clay-rich rocks. The LOM can be obtained from core sample analysis, such as the vitrinite refiectance or Tmax (Passeyet al., 1990).
The ΔlogR separation in the source rock and non-source rock intervals can be distinguished using the natural gamma-ray curve (Passeyet al., 1990). For clay-rich rocks, they proposed an empiricalequation for calculating the TOC from ΔlogR: where TOC is the total organic carbon content measured in wt%; and LOM is the maturity.

V-△logR model
The V-△logR technique was proposed and improved by Liu et al. (2011b); Liu et al. (2014); Liu et al. (2019). This model includes three revisions to the ΔlogR model. 1) Replace the fixed overlaycoeffcient (0.02) with a variable one, and determine it according to the actual geological data (TOC and well logs). 2) Remove the maturity (LOM) from the model. 3) Determine the baseline from the ΔlogRV curve, instead of determining it from the sonic and resistivity curves.
Here, ΔlogRv is the curve separation measured in logarithmic resistivity cycles; R is the deep lateral resistivity (Ω·m); p is the sonic transit time, bulk density, or neutron log. k is a variable that ranges from 0 to 1; baseline is the baseline value of the ΔlogRv curve; b is the background value of the TOC content; and α is a coeffcient calculated using a mathematical tool.

A variable overlay-coeffcient
In the V-ΔlogR model, the overlay-coeffcient (k in Eq. 3) is a variable, which was determined according to the actual geological data. In this method, k is the logging response of the non-organic elements, which should be eliminated from the ΔlogR separation as the porosity and resistivity logs also respond to other non-organic elements, such as rock minerals, formation water salinity, and formation pressure Zhu et al., 2019). Under geological conditions, non-organic

FIGURE 1
Flowchart of calculating TOC content of shale by the V-△logR model.

Frontiers in Earth Science
frontiersin.org elements can be very complex and may vary significantly from place to place. Thus, the value of k should also change with the geological conditions. Based on the above analysis, it can be concluded that a proper k value can remove the non-organic response in the ΔlogR separation. Due to the strong heterogeneity of hydrocarbon generation and expulsion process of source rocks, and the factors that interfere with TOC logging response under geological conditions are complex (including lithology, physical properties, fluid salinity, etc.), the K value that leads to the best TOC prediction effect often changes with different wells Chen et al., 2017;Liu et al., 2019). For this reason, it is often necessary to determine k value based on a single well, rather than using a fixed k value uniformly for entire area of interest. In cases where TOC for core test is available, the coefficient k can be determined by an exhaustive method (based on core and logging) (Liu, 2011a;Liu et al., 2015). If measured TOC data cannot be obtained, the variable coefficient △LogR model provides a method to calculate TOC independently of measured TOC data. i.e., logR-AC (using only logging curves) intersection method Liu et al., 2019). In this study, we first set the average value of acoustic logging curve and logarithmic resistivity curve as 2, then we determine k value with the exhaustive method or the "logR-Δt" crossplot method (see the example analysis below). Once the k value is determined, combining with the gamma-ray curve to distinguish the △logR of the shale and the non-shale section, we calculate △logRv according to Formula (3).

2.2.2
Removing maturity from the model Passey et al. (1990) noted that ΔlogR is linearly related to the TOC content and is a function of maturity. However, our analysis indicates that maturity can be removed from the model. First, the increase in the  Generalized stratigraphic column for deep formation in the study area.

Frontiers in Earth Science
frontiersin.org resistivity response in mature and clay-rich continental shales is not signifcant (probably related to high clay contents, high bound water contents, and overpressure) . Second, it is cumbersome to determine the LOM of each depth point, and may be inaccurate in cases where test data is unavailable. Third, the 10 (2.297-0.1688 × LOM) term in Eq. 2 is approximately a fixed value if the LOM changes little throughout a well profile, and it can be replaced by a fixed value calibrated from the TOC measured from the core samples. In a case where the maturity flfluctuate significantly over a long profile, the profile can be divided into several subsections that each contain small maturity changes.

A single baseline
In the V-ΔLogR model, the baseline appears around the valley of the ΔLogRv curve and corresponds to the organic-lean and clay-rich intervals. This method is essentially consistent with Passey's approach (Passeyet al., 1990), except that it avoids determining the baselines twice (i.e., for the porosity log and resistivity curve). Therefore, the baseline values are determined which are generally occur in fine-grained non-oil-producing zones according to the shape of △logR curve.

Coefficient α and b
The coefficients α and b are undetermined coefficients. Chen et al. (2017) pointed out that the coefficient α is related to the depositional environment of shale. When the maturity is basically the same, the α value of lacustrine shale is larger than that of delta front shale. Liu et al. (2019) proposed a prediction method for α in formation with strong sediment phase change, which can achieve the prediction of α value in the absence of experimental test TOC b corresponds to the background value of organic carbon content (△TOC) in the △log technique proposed by Passey. In this study, the α and b coefficients in Eq. 2 were determined by Linear regression analysis. In order to avoid the introduction of large systematic errors when applied to other wells, it is recommended that the value of b should not exceed the lower limit of the TOC of the source rock.

Geologic setting and lithostratigraphy
The Songliao basin is located in the northeastern part of China ( Figure 1A), and the Shuangcheng area is located in the eastern fault depression zone in the northern part of the Songliao Basin ( Figure 1B). The study area described in terms of structural elements in the Shuangcheng sag which is sub-element within the Shuangcheng area. The strata with clastic rocks in the Shuangcheng sag include the Yingcheng Formation and the Denglouku Formation, The fourth member of the Yingcheng Formation develops black shale considered to be rich potential source rocks for hydrocarbons ( Figure 2). Since 2017, a suite of organic-rich shale has been discovered in the southern sag of the Shuangcheng region (Huang et al., 2015;Yin et al., 2018). This set of shale has strong hydrocarbon-generating capacity, because it was discovered in the sandstone of the Denglouku Formation, and the shale is oil, and the natural productivity of a single well can reach up to 91 t/d, The shale rich in organic matter is developed in the fourth member of the Yingcheng Formation. The shale is deposited in the freshwater environment of continental semi-deep lake-deep lake, with a thickness of 20 m-120 m. Previously, this set of shale was mainly studied as a source rock, and it was not used as a shale oil exploration object until 2019.

Geochemical and well logging responses of shale
The total organic carbon, Rock eval pyrolysis and vitrinite reflectance (Ro) data involved in this study were collected from Key Research Laboratory on Tight Sandstone and Shale Oil Reservoir Forming. The data were obtained from core samples of 11 Wells in the Shuangcheng area which are used to determine the origin, content, and maturity level of the organic matter present in a shale. The depth of the data ranges from 1,241.00 m to 1796.35 m, and the well locations are shown in Figure 3C.
In order to avoid the volatilization of light hydrocarbons in shale, Rock eval pyrolysis and TOC tests of core samples were carried out soon after drilling and coring. Analysis data from two typical exploration wells were shown on Supplementary Table S1. These data revealed that the TOC of shale in Shuangcheng area ranged from 0.78% to 6.38%, with an average of 2.86% ( Figure 4A). S1 ranged from 0.05 mg/g-1.65 mg/g (average 0.58 mg/g), and S2 ranged from 0.36mg/g-16.36 mg/g (average 6.35 mg/ g). The measured Ro is between 0.82% and 1.31%, indicating that shale has entered a mature evolution stage, and Ro increases regularly with the increase of depth ( Figure 4C). Figure 5 shows geochemical analysis data and conventional logging curves of typical exploration wells in Shuangcheng area. Shale segments correspond to higher gamma values (typically greater than 140API), which can be used to distinguish shale from non-shale.
The acoustic time of shale is relatively high, mainly between 70 μs/ ft-100 μs/ft. The density of shale is mainly between 2.46 g/cm 3 and 2.52 g/cm 3 , which is slightly less than that of sandstone. However, the density curve was greatly affected by hole collapse (e.g., 1,366 m-1378 m), so it was not used to predict the total organic carbon. The deep lateral resistivity of shale is lower than that of sandstone, but the resistivity of shale does not show the obvious positive anomaly as Marine shale, and even shows a slight negative anomaly, even when the shale is mature and contains liquid hydrocarbons (e.g., 1,460 m-1480 m chloroform bitumen A 0.049%-0.361%).

FIGURE 5
Organic heterogeneity and logging response of shales in the Yingcheng Formation of the study area.
Frontiers in Earth Science frontiersin.org

Establishment of the model
Two typical wells (well S59 and well S70) in the targeted area with core TOC measurements were randomly selected to establish and vertify the prediction model of TOC, and the remaining 9 wells were used for model validation. The key point to establish the model is to obtain the parameters of the V-ΔlogR methods described in process 3.3. Therefore, we determine the parameters in Eq. 3 of the V-ΔlogR model separately. Combined with the TOC evaluation standard of Lacustrine shale (SY/T 5735-1995), the value b should not exceed the lower limit of organic matter abundance of source rock (< 0.4%). As the value b has little influence on the prediction accuracy of TOC of shale. We mainly discuss the coefficients k, baseline and α in the V-ΔlogR model. Frontiers in Earth Science frontiersin.org

Determination of the optimal k value
Coefficient k was determined by exhaustive method (based on the core sample and logging data) and "logR-AC" intersection (using logging data only). Figures 6A, B are the crossplots of the coefficient R2 and k between the predicted TOC and core TOC determined by the exhaustive method proposed by Liu et al. (2015).
The figure shows that as the k value increases, the determination coefficient R2 increases first and then decrease, so, this shows that there is an optimal K value which makes the predicted TOC agree with the core TOC best an optimal k value that gives the highest coincidence between predicted TOC and core TOC. The k value corresponding to the peak values of the determined coefficients R2 in Figures 6A, B are 0.65 and 0.12, respectively. The obvious difference between them indicates that it is necessary to determine the k value according to every single well in Shuangcheng area for accurate prediction of shale TOC; Figures 6C, D show the k predicted by "LogR-AC" intersection for S59 and S70 Wells. The data points in the figure are the acoustic time difference and log resistivity corresponding to the points near the baseline (baseline ± 20%; if there is a significant deviation from the trend, the point should be excluded). The lithology is no organic matter or low organic matter shale or silted shale. The slope of the fitting formula in Figures 6C, D (absolute value, negative sign indicates that a curve decreases as the other curve increases) is basically consistent with the k value corresponding to the peak value in Figures 6A, B, indicating that the k value predicted by "LogR-AC" intersection is reliable. Therefore, we applied this approach to all Wells in the targeted area and ultimately determined the optimal k value for each well, and the coefficient k predicted by LogR-AC intersection method for other wells is shown in Supplementary Table S2.

FIGURE 7
The relationship between core TOC and TOC predicted by V-△LogR model.

FIGURE 8
Grading evaluation of shale in the study area: (A) Crossplot of S1, TOC, Ro of samples of shale from the study area, indicting that S1 increases with the increase of TOC; (B) Crossplot of S1/TOC,TOC, Ro, S1/TOC decreases rapidly when TOC exceeds 3.5%, indicating that hydrocarbon expulsion has occurred; (C) The crossplot of S1 and Ro shows that the inflection point of S1 from increasing to decreasing indicates that the hydrocarbon expulsion threshold is Ro = 0.9%; (D) The crossplot of S1/TOC and Ro shows that the inflection point of S1/TOC from increasing to decreasing indicates that the hydrocarbon expulsion threshold is Ro = 0.9%.

Frontiers in Earth Science
frontiersin.org

Determination of the baseline
Substitute the calculated k value into Eq. 3 to calculate the △logR value of each well, and then uses the morphology of △logRv curve to determine the baseline of the V-△logR. The thickness of the target layer is relatively thin (generally less than 200 m), lead to the fluctuation of the acoustic time difference and resistivity curve is not obvious, so the baseline value is relatively stable. Taking well S59 as an example, the baseline position corresponds to a small △logRv value, and the lithology is mainly low organic matter shale. The baseline value for well S59 was 1.63 (the red line 5 in Figure 5). Baseline values for other Wells are shown in Supplementary  Table S2.

Determination of the coefficient α
The coefficient α is determined according to the least square method. Figures 6E, F show that the coefficients α obtained by Comparison of TOC predicted by logging on well profile and TOC of core and the identification of different grades of shale.

Frontiers in Earth Science
frontiersin.org 09 least squares fitting method for S59 (K = 0.65) and S70 (K = 0.12) are 8.58 and 8.94, respectively. The shale of the 4th member of Yingcheng Formation was deposited in a relatively stable lacustrine environment, which may be the reason for the relatively close α value. Figures 5G, H show that the TOC calculated by the model is in good agreement with the TOC of the core.

Model verification
In the model validation process, coefficient A was calculated using the value of the adjacent well, while K was calculated using the logging curve of the well (Supplementary Table S2). The prediction model was applied to nine retention wells. The model of well S59 was applied to well Sx67, S71, and Sx74 in the same eastern slope zone. The prediction model established in S70 well in the western slope zone was applied to S66 well, S661 well, S68 well, S72 well, S76 well and S83 well, which also belong to the western slope zone. The cross plot of TOC calculated by V-△logR model and TOC predicted by core for all Wells in the study area shows that TOC calculated by model and TOC predicted by core are distributed around Y=X, with a determination coefficient of 0.83 and an average relative error of 11.06% (Figure 7).
In Supplementary Table S2, the determination coefficients of the predicted TOC and core TOC of the single well model ranged from 0.57 to 0.93, and the relative average errors ranged from 7.13% to 25.31%. In addition to the S74 well prediction error is relatively large (R2 = 0.57 and MRE = 25.31%), the overall anastomosis effect is also good. In general, the established prediction model can accurately predict the TOC of surrounding wells, and the TOC of prediction model is consistent with the TOC of core in well profile.

Grading evaluation of shale
Shale grading evaluation usually considers reservoir oil-bearing property evaluation and recoverable factors (Lu et al., 2012). Since there is little difference in the content of mineral components in shales in the study area, this shale grading evaluation is mainly carried out based on the difference in oil content of shales. Rock pyrolysis index S1 represents the amount of hydrocarbons that have been generated and remain in the shale. Although S1 lacks heavy components and some light components are lost during sample collection, it is still one of the more intuitive and reliable indicators to characterize shale oil content (Lu et al., 2012;Li et al., 2015). Given the difficulty of accurately predicting S1 from well logging data, a feasible approach is to establish relationships with more readily available indicators (including TOC, organic matter type and Ro) that are closely related to S1, and then use these easily available indicators to characterize shale oil content and to carry out shale grading evaluation. For example, Lu et al. (2012) carried out shale grading evaluation through the relationship between TOC and S1.
The shale maturity parameter (Ro) was also used in grading evaluation in addition to the relationship between S1 and TOC. By discussing the relationship between S1 and TOC and Ro, shale in Yingcheng Formation can be divided into three categories. Type I (TOC > 3.5%, Ro > 0.9%) shale is characterized by a slight increase in S1 or a steady increase in TOC ( Figure 8A), and most samples have high S1 values (Y is generally greater than 0.7 mg/g). Samples in this interval have high organic matter abundance and hydrocarbon generation ability. When TOC is greater than 3.5%, S1/TOC shows a downward trend ( Figure 7B), indicating that the hydrocarbon generation content in these shale has exceeded the adsorption capacity, and the excess hydrocarbon is expelled , which also indicates that the oil has relatively good mobility. In Figure 8A, some samples with TOC greater than 3.5% and still with low S1 values were expelled from type I shale. Because of their The S1 of type II shale (TOC between 2% and 3.5%) increases rapidly with the increase of TOC, indicating that the oil content (hydrocarbon generation capacity) of shale increases with the increase of TOC. Class II shale also include shale with TOC > 3.5 and Ro less than 0.9. The saturation degree of adsorbed hydrocarbon of type II shale should be lower than that of type I shale and the mobility should be lower than that of type I shale. The S1 of Class III (TOC < 2.0%) shale increases slowly with the increase of TOC, and shalewith different evolution degrees have lower S1 values, mainly because they have lower organic matter abundance and hydrocarbon generation capacity.

Thickness distribution of shale
Based on the calculation of TOC of 11 wells in the study area, combined with the shale grading limits in Figure 7A, the distribution of different grades of shale is identified on the well profile, and the thickness is calculated. As shown in the typical interwell profile in Figure 9, the development frequency of type I shale is relatively low, and the top, middle and bottom of the fourth member of the Yingcheng Formation are all developed. The continuous thickness of single layer is generally no more than 5 m, and the thickness ratio revealed by drilling (cumulative thickness of type I shale/total thickness of shale) is generally 10%-20%. The thickness of class II shale monolayer is generally 10 m-20 m, and the cumulative thickness of a single well is 40%-60%. The monolayer thickness of shale of Class III is generally less than 5 m, and the cumulative thickness of a single well is generally 20%-30%.
By statistical analysis of the relationship between the thickness of different grades of shale and the total thickness of shale, we established a method conducive to predict the thickness of different grades of shale. As shown in Figure 10, the determination coefficient between the thickness of different grades of shale and the total shale thickness exceeds 0.91, indicating that the thickness of different grades of shale can be predicted by the total shale thickness. The quantitative relationship between the two was established based on the zonal zone. For areas lacking drilling, the thickness of shale in each grade was predicted based on the total shale thickness predicted by the combination of well vibration   (Figure 11). The results show that the thickness of class I shale is generally 5 m-15 m, class II shale is generally 20 m-80 m, and class III shale is generally 10 m-40 m.

Conclusion
In this study, V-△logR method was applied to TOC prediction and grading evaluation of lacustrine shale in Shuangcheng area, Songliao Basin. The prediction model of total organic carbon (TOC) is established by using the acoustic and the deep lateral resistivity curve and the variable scaling coefficient (k) between them and another scaling coefficient (A) between TOC and △logR. Although the shale in Shuangcheng area is rich in organic matter and in the oil generation window exhibits high natural gamma and high acoustic time difference, however, it shows a unique logging response of medium to low resistivity. The application results show that it is necessary to determine the undetermined coefficient K for individual wells, even in the formation with stable sedimentary facies and low maturity fluctuations, because the value of K that achieves the highest TOC prediction accuracy varies significantly but the undetermined coefficient A varies little from well to well. According to the very close relationship between oil content (S1) and TOC and considering the degree of thermal evolution (Ro), shales in Shuangcheng area are Frontiers in Earth Science frontiersin.org divided into three categories. The TOC of well logging prediction provides a reliable basis for the evaluation of shale grading on well profile and the prediction of shale thickness distribution at each grade on the plane. The application results of 11 exploration wells in Shuangcheng area show that the variable coefficient △logR method can accurately predict TOC of lacustrine shale and guide shale grading evaluation.

Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.