Predicting Source Rock Distribution of the Ninth Member of the Upper Triassic Yanchang Formation Based on Well-Logging Parameters and TOC Values in the Longdong Area, Ordos Basin, China Authors: Xiao Hui, Xuan Ke, Yalin Qi, Shuyong Shi, Jing Zhu, Yunpeng Wang Published: 2024 ABSTRACT In recent years, new oil reservoirs have been discovered and exploited in the ninth member (Chang 9 Member, T3y9) of the Upper Triassic Yanchang Formation (T3y) in the Longdong area, Ordos Basin. Some studies have shown that the crude oils of the Chang 9 Member may originate from the Chang 9 source rock in some areas, which may be related to the distribution of the source rock. However, the distribution of the Chang 9 source rock in the Longdong area is still unclear, which hinders further exploration and development of petroleum. In this study, we established a multiple linear regression model for predicting total organic carbon (TOC) based on the relationship between well-logging parameters and measured TOC values of shale core samples from 30 wells in the study area. The results show that the Chang 9 shale is mainly composed of gray and dark mudstones, which mainly belong to the interdistributary bay and front delta depositional subfacies. The TOC values of the shale core samples from this member vary in a range of 0.11–4.8%, with an average value of 0.96%. Compared with traditional and improved Δlog R models, our model shows a higher accuracy of TOC prediction with R2 = 0.9181, which meets the requirements for predicting the distribution of the Chang 9 source rock. In the map of the Chang 9 source rock predicted by our model, the thickness of the source rock (TOC ≥ 1.0%) varies in the range of 1–12 m, showing a decreasing trend from northeast to southwest in the Longdong area. The crude oil in the northeastern areas enjoys a high ratio of 17α(H)-C30 rearranged hopane and C30 hopane (C30*/C30), and the thickness of the Chang 9 source rock is also greater than in other areas. It is speculated that the Chang 9 Member tight oil in the northeast area is mainly from the Chang 9 source rock, while the oil in other areas is from the Chang 7 source rock. In our study, we presented a method for predicting the source rock distribution, which can be widely used for exploring the tight oil of the Chang 9 Member in the study area. Introduction The Ordos Basin is an important oil-bearing basin in China, in which the Upper Triassic Yanchang Formation (T3y) is the most important oil-bearing layer.1,2 Oil exploration and development to date have mainly focused on the seventh member (Chang 7 Member, T3y7) and the upper members (Chang 1–6 Members, T3y1–T3y6).1,3,4 After several decades of continuous breakthrough and development of tight oil, a number of oil reservoirs were also discovered in the ninth member (Chang 9 Member, T3y9) of the Yanchang Formation in the Zhidan, Jiyuan, and Longdong areas of the basin.3,5,6 However, the origin of the crude oil from the Chang 9 Member remains controversial.7−11 Some scholars suggested that the crude oil mainly originated from the Chang 7 source rock due to its high total organic carbon (TOC), large thickness, and wide distribution in the basin,3,5,8,10 while other scientists believed that the crude oil in some areas might originate from the Chang 9 source rock.1,9,12 In comparison with the Chang 7 shale, part of the Chang 9 shale enjoys high abundance of 17α(H)-C30 rearranged hopane (C30*), with the ratio of C30* and C30 hopane (C30) higher than 1.0.7,8,11 Therefore, it is considered that the crude oil with low C30*/C30 values is mainly originated from the Chang 7 shale, while the oil with high C30*/C30 values is derived from the Chang 9 shale.8−10,13,14 The Longdong area is located in the southwestern part of the Ordos Basin (Figure 1a), which is considered as the next favorable area for tight oil exploration of the Chang 9 Member.3 Numerous research works have focused on the sedimentary facies, petroleum reservoir evaluation, and petroleum accumulation conditions.1,3,15,16 The crude oil recovered from the Chang 9 Member could be related to the source rock distribution of this member. Compared with the Chang 7 source rock, the distribution of the Chang 9 source rock in the Longdong area remains unclear. (a) Structural units of the Ordos Basin; (b) stratigraphic column of the Upper Triassic Yanchang Formation. Parts (a) and (b) were reproduced from ref (23). Copyright 2016 Elsevier. Total organic carbon is an important component of mudstone, which is also crucial for the evaluation of the source rock. Based on the well-logging parameters and the corresponding TOC values, various TOC prediction models have been established, with the traditional (or improved) Δlog R,17 multiple linear regression,18−20 and neural network models21,22 being widely used for the TOC prediction. Although there are so many models for the TOC prediction, the choice of model depends on the accuracy of the model and the actual geological properties of the source rock. In this study, we first observed drill cores of the Yanchang Formation from 30 wells in the Longdong area, Ordos Basin, with respect to their lithology, sedimentary structure, and shale development. The TOC values of the shale core samples were also measured. Subsequently, the well-logging parameters, including the acoustic time difference (AC), resistivity (RT), natural γ response (GR), and density (DEN), corresponding to the TOC value of each sample were collected. Finally, we constructed traditional and improved Δlog R, and multiple linear regression models and compared the results of the three models for the TOC prediction. The aim of this study was to select an appropriate model for predicting the distribution of the Chang 9 source rock in the Longdong area in the Ordos Basin. At the same time, accurate prediction of TOC is also important for further oil exploration of the Chang 9 Member in the Ordos Basin. Geological Settings and Stratigraphy Geological Settings The Ordos Basin, covering an area of 3.7 × 105 km2, is tectonically located on the western margin of the North China Craton (Figure 1a).24 The basin can be divided into six structural units: the Weibei Uplift in the south, the Yimeng Uplift in the north, the Jinxi Fold Belt in the east, the Western Edge Thrust Belt and the Tianhuan Depression in the west, and the Yishan Slope in the central basin.23 The study area (Longdong area), which crosses the Tianhuan Depression and the Yishan Slope, is located in the southwestern part of the basin (Figure 1a). The Upper Triassic Yanchang Formation is an oil-producing strata that can be divided into ten members (Chang 1–10 Members, T3y1–T3y10) from top to bottom (Figure 1b). The thickness of this formation varies between 700 and 1200 m in the basin. All members are mainly composed of sandstone, siltstone, and mudstone, in which the Chang 7 Member is mainly composed of mudstone with high TOC.25−27 Previous studies have shown that the Chang 7 Member is the dominant source rock in the Mesozoic and formed during the transgression of the lakes.25,27,28 It is widely distributed in the Ordos Basin, with thicknesses varying between 20 and 60 m.29,30 In addition, the shale from the Chang 9 Member is considered as the secondary source rock, mainly composed of deltaic plain shale and deltaic frontal shale, with an average TOC of less than 2.0%.31 Previous studies showed that the Chang 9 source rock is only distributed in the Zhidan and Ganquan areas, and the thickness is generally less than 25 m,30,32,33 but the distribution in the Longdong area is not clear. Observation of the Drill Cores and the Sedimentary Facies The lithologies of the Chang 9 Member are mainly composed of sandstone, fine sandstone, siltstone, and mudstone (Figure 2a–c) in the Longdong area, Ordos Basin. The Chang 9 shale is composed of gray and dark mudstone, indicating a low content of organic matter (Figure 2d–h). It is also worth noting that the shale in the form of mudstone interbeds in the Longdong area is thin and not well-developed. The interdistributary bay and delta font are the main types of mudstones in most areas (Figure 2), and the mudstone formed in the semideep lacustrine is only common in the northeastern part of the study area (Figure 3). Plant fossils were also found in some core samples (Figure 2d,e), indicating a shallow deltaic sedimentary environment. Pictures of core samples of the Chang 9 Member in the Longdong Area, Ordos Basin. (a) Yuan 427, 2432.5 m, mud graveled sandstone, delta font; (b) Yuan 457, 2327–2331 m, sandstone interbedded with 1 m dark mudstone, delta font; (c) Le 46, 1413.5–1415.5 m, fine sandstone interbedded with dark mudstone, delta font; (d) Mu 23, 2516 m, gray mudstone, stem fossil, delta font; (e) Bai 442, 2338.4 m, gray mudstone, interdistributary bay; (f) Yuan 541, 2294.5 m, plant fossil, delta font; (g) Chang 23, 1726.6 m, gray mudstone, delta font; and (h) Bai 445, 2350 m, mud graveled sandstone and dark mudstone, interdistributary bay. Sedimentary facies of the Chang 9 Member in the Longdong area, Ordos Basin. This figure was reproduced from ref (3). Copyright 2018 Elsevier. The distribution of the Chang 9 source rock is mainly controlled by the sedimentary facies. During the depositional period of this member, three major sedimentary systems were developed in the northwest, northeast, and southwest of the basin (Figure 3). Previous studies have shown that the Chang 9 Member was mainly formed in shallow delta fronts and plain sedimentary environments.16,32,34 However, the Chang 9 Member in the northeastern areas was formed in a semideep lacustrine environment, resulting in the source rock being thicker than in other areas in the Longdong area, Ordos Basin.2,30 Experiments and Methods Total Organic Carbon Analysis All shale core samples, including the Chang 6, Chang 7, Chang 8, and Chang 9 Members, were collected from 30 wells in the Longdong area. The TOC value was measured by the LECO CS-800 analyzer, and the carbonates were removed with hydrochloric acid (HCl) before measurement. Well-Logging Surveys The well-logging surveys were conducted by the Research Institute of Petroleum Exploration and Development, PetroChina Changqing Oilfield Company. Some common well-logging parameters such as GR, RT, DEN, and AC, which are sensitive to the TOC value, were recorded and processed with the corresponding measured TOC values. Traditional Δlog R Model The Δlog R model, the most common model for the TOC prediction, was proposed by Passy based on the well-logging data of RT and AC.17 The Δlog R can be expressed by the following equation:1 The predicted TOC can be expressed as2 After the combination of eqs 1 and 2, the predicted TOC can also be expressed as3where a, b, c, d, e, and f are the fitting coefficients. RT and AC are the resistivity (Ω·m) and acoustic time difference (μm/s), respectively. RTbase and ACbase are the baseline values of resistivity and acoustic time difference. Then, the data of RT and AC corresponding to the measured TOC values were brought into eq 3 using Origin software. The TOC prediction model is obtained.4 The predicted TOC calculated by eq 4 shows a positive correlation with measured TOC with a R2 value of 0.5802 (Figure 4a). Relationship between measured TOC and predicted TOC by (a) traditional and (b) improved Δlog R models. Improved Δlog R Model The improved Δlog R model also considered the DEN and GR response to the TOC value based on a traditional model.19,21 The improved Δlog R model can be expressed as5 After the combination of eqs 1 and 5, the predicted TOC can be also expressed as6where a, b, c, d, e, f, g, and h are the fitting coefficients. GR and DEN are the natural γ response (API) and density (g/cm3), respectively. Then, the data of GR, DEN, and Δlog R corresponding to the measured TOC were brought into eq 6 by using Origin software. The TOC prediction model is obtained.7 As shown in Figure 4b, the predicted TOC calculated by eq 7 shows a positive correlation with the measured TOC with a R2 value of 0.7711. Multiple Linear Regression Model Some common well-logging parameters, such as GR, RT, DEN, and AC, are sensitive to the TOC value:21,22,35 (1) The organic-rich source rock shows a positive correlation with GR (Figure 5a) due to its strong adsorption for radioactive elements such as uranium (U) and thorium (Th). The higher the radioactivity of the source rock, the more sensitive the GR response. (2) The organic-rich source rocks belong to the noneasily conductive materials, which lead to the RT showing an increasing trend with the TOC (Figure 5b). (3) The density of solid organic matter is about 1 g/cm3, which is close to that of water. The organic-rich source rocks generally have a low density (Figure 5c); hence, the DEN curves can be used for the TOC prediction. (4) Organic-rich source rock has a low sonic velocity. The AC usually shows a positive correlation with the TOC (Figure 5d). Linear relationships between measured TOC and well-logging parameters of the Chang 6–9 shales in the Longdong area, Ordos Basin. (a) Natural γ response (GR) versus measured TOC; (b) resistivity (RT) versus measured TOC; (c) density (DEN) versus measured TOC; and (d) acoustic time difference (AC) vs measured TOC. The TOC prediction based on a single well-logging parameter is not accurate enough. To further improve the accuracy of the TOC prediction, a multiple linear regression model for the TOC prediction is constructed. Due to the weak correlation between measured TOC and AC, the model was created considering RT, GR, and DEN. The predicted TOC value can be calculated by the following equation:8 In eq 8, TOC* is the predicted TOC, GR is the natural γ response (API), RT is the resistivity (Ω·m), and DEN is the density (g/cm3) of the shale. As shown in Figure 6, the multiple linear regression model improves the accuracy of the TOC prediction with an R2 value of 0.9181. Relationship between the measured TOC and the predicted TOC by the multiple linear regression model. Results and Discussion Total Organic Carbon of the Chang 9 Shale Total organic carbon is an important indicator for evaluating the hydrocarbon generation capacity of shale. The TOC of the Chang 9 shale varies in the range of 0.11–4.8%, with an average value of 0.96% (Figure 7). This is because the Chang 9 shale was mainly formed in a deltaic sedimentary environment (Figure 3). According to China’s industrial criterion for continental source rock,36 the shale is considered as the source rock when the TOC value is higher than 0.4% and good to very good source rock with the TOC value in the range of 0.6–1.0%. The Chang 9 shale is mainly composed of good to very good source rock with 66% (N = 21); however, the proportion of shale with TOC higher than 2.0% is 3.1% (N = 1, Figure 7). Histogram of the frequency distribution of the TOC of the Chang 9 source rocks in the Longdong area, Ordos Basin. Comparisons of Traditional, Improved Δlog R, and Multiple Linear Regression Models Based on the results of the traditional and improved Δlog R and multiple linear regression models for the TOC prediction, the R2 values of the three models are 0.5802, 0.7711, and 0.9181, respectively (Figures 4 and 6). The traditional Δlog R model considers the influence of TOC on AC and RT, while the improved Δlog R model also considers the influence of TOC on DEN and GR. The improved Δlog R model improves the accuracy of TOC prediction compared to the traditional model; however, the R2 value (R2 = 0.7711) of the model (Figure 4b) is lower than that of the TOC prediction model based on GR values (R2 = 0.7781, Figure 5a). The determination of the baseline values of RTbase and ACbase is subject to uncertainties and randomness, which affects the accuracy of the TOC prediction. The core observation results also show that the Chang 9 Member is characterized by thin and multilayer mudstone interbeds in the Longdong area (Figure 2). Due to the weak correlation between TOC and AC (Figure 5b) and the uncertainty of baseline selection, the traditional and improved Δlog R models may not be suitable for the TOC prediction of this member. In comparison, the multiple linear regression method can take the response of multiple well-logging parameters to TOC into account and improve the accuracy of the TOC prediction. In recent years, this method has also widely been used for the TOC prediction due to its higher accuracy than traditional and improved Δlog R models.18−20 Besides, the measured TOC values show a good correlation with RT, GR, and DEN, with R2 values of 0.7781, 0.6431, and 0.4841, respectively, but a poor correlation with AC with R2 values of 0.1926 (Figure 5). The multiple linear regression method (R2 = 0.9181, Figure 6) based on RT, GR, and DEN parameters (Figure 5a–c) is more suitable for the TOC prediction of the Chang 9 Member in the Longdong area, Ordos Basin. Application for the Distribution of the Chang 9 Source Rock The multiple linear regression model has a higher accuracy with an R2 value of 0.9181 (Figure 6), which is more suitable for the TOC prediction of the Chang 9 source rock in the Longdong area of the Ordos Basin. Based on the above considerations, we chose the multiple linear regression model for the TOC prediction. According to the industrial criterion for continental source rock in China, mudstone with TOC ≥ 0.4% is considered as the source rock. Generally, the excess hydrocarbons are expelled when the generated hydrocarbons meet their own adsorbed and free capacity.37 Regarding the Yanchang source rock, Yao et al.2 found that the generated hydrocarbons of the Yanchang Formation are expelled when the TOC value exceeds 1.0%. More importantly, the measured TOC values have a poor correlation with the well-logging parameters when the TOC value is below 1.0% (Figures 4–6). From the above considerations, the Yanchang source rock with a TOC above 1.0% is the effective source rock as the hydrocarbons can expel and form conventional oil reservoirs. Then, we input the machine language “0.049 × GR + 0.046 × RT – 4.433 × DEN + 5.939 ≥ 1.0%” into the Gxplorer software. The software can automatically calculate the thickness of the Chang 9 source rock for each well. For example, we use the multiple linear regression model to predict the TOC and thickness of the Chang 9 source rock from well Bai 445 in the Baibao area. The results show that the effective source rock (TOC ≥ 1.0%) is multiple mudstone interbeds with a thickness of tens of centimeters, and the total thickness in the Chang 9 Member of well Bai 445 is 11.75 m (Figure 8). Prediction results of the thickness of the Chang 9 source rock from well Bai 445 in the Baibao area. The thickness of the source rock was calculated by a multiple linear regression model. Finally, we use the kriging method to construct the distribution map of the Chang 9 source rock in the Longdong area. In this way, we obtain the final distribution of the Chang 9 source rock in the Longdong area of the Ordos Basin by combining sedimentary facies analysis and the prediction of the thickness of the Chang 9 source rock (Figure 9). Distribution of the Chang 9 source rock (TOC ≥ 1.0%) in the Longdong area, Ordos Basin. Distribution of the Chang 9 Source Rock and Potential Triggering Factors Our core observation results show that the Chang 9 source rock in the Longdong area is not well-developed. The main lithology of the Chang 9 Member consists of sandstone or siltstone interbedded with thinly bedded mudstone (Figure 2). In general, the TOC and thickness of the Chang 9 source rock are significantly lower than those of the Chang 7 source rock (Figure 7), mainly due to the differences in sedimentary environments in the Longdong area, Ordos Basin (Figure 3). As shown in Figure 9, the distribution characteristics of the Chang 9 source rock are mainly controlled by the sedimentary environment,16 and the thickness shows a decreasing trend from northeast to southwest. The thickness of the Chang 9 source rock in the northeastern areas (e.g., Huachi and Baibao) is higher than in other areas and varies between 8 and 12 m. In the Qingcheng and Qingyang areas, the thickness of the Chang 9 source rock is between 3 and 7 m, compared to 1 to 3 m in other areas. It is worth noting that the crude oils from the Chang 9 Member in the northeastern parts of the Longdong area have a high C30*/C30 ratio, while the C30*/C30 ratio of crude oils in other parts is lower.8,10 The C30*/C30 ratio of the source rocks of Chang 7 and 9 Members shows significant differences, which may be related to the sedimentary environments.7,11,38 Some studies have shown that the Chang 9 source rocks of, which formed in semideep lacustrines, have a high abundance of C30* in the Huachi and Baibao areas. In comparison, the Chang 7 source rock with a low abundance of C30* content was mainly formed in deep lacustrines.9,10 In addition, our results showed that the Chang 9 source rock with a large thickness was mainly distributed in the Huachi and Baibao areas (Figure 9). From the perspective of C30*/C30 and the distribution of the Chang 9 source rock, it can be suggested that the tight oil of the Chang 9 Member in the northeastern areas is derived from the Chang 9 source rock, while the oil in other areas is mainly derived from the Chang 7 source rock. Conclusions (1)The shale of the Chang 9 Member is mainly composed of gray and dark mudstones, which are generally formed in a deltaic sedimentary environment. Total organic carbon (TOC) varies in the range of 0.11–4.8%, with an average value of 0.96%.(2)The determination of baseline values of resistivity (RT) and acoustic time difference (AC) is subject to a certain uncertainty that affects the accuracy of the TOC prediction. The TOC prediction value based on the traditional and improved Δlog R models differs greatly from the measured TOC values, with R2 values of 0.5802 and 0.7711, respectively.(3)The TOC value shows good correlations with well-logging parameters such as natural γ (GR), RT, and density (DEN), leading to constructing a multiple linear regression model for the TOC prediction. Compared with traditional and improved Δlog R models, the R2 of our model can reach 0.9181, which is more suitable for the TOC prediction of the Chang 9 source rock in the Longdong area, Ordos Basin.(4)The map of predicted source rock TOC shows that the distribution of Chang 9 source rock (TOC ≥ 1.0%) is mainly controlled by the sedimentary environment. The thickness varies in the range of 1–12 m and shows a decreasing trend from northeast to southwest in the Londong area, Ordos Basin.(5)From the perspective of the relative abundance of C30 rearranged hopane and the distribution of the Chang 9 source rock, it is speculated that the tight oil of the Chang 9 Member in the northeastern areas derives from the Chang 9 source rock, while the oil in other areas mainly from the Chang 7 source rock. (1)The shale of the Chang 9 Member is mainly composed of gray and dark mudstones, which are generally formed in a deltaic sedimentary environment. Total organic carbon (TOC) varies in the range of 0.11–4.8%, with an average value of 0.96%. The shale of the Chang 9 Member is mainly composed of gray and dark mudstones, which are generally formed in a deltaic sedimentary environment. Total organic carbon (TOC) varies in the range of 0.11–4.8%, with an average value of 0.96%. (2)The determination of baseline values of resistivity (RT) and acoustic time difference (AC) is subject to a certain uncertainty that affects the accuracy of the TOC prediction. The TOC prediction value based on the traditional and improved Δlog R models differs greatly from the measured TOC values, with R2 values of 0.5802 and 0.7711, respectively. The determination of baseline values of resistivity (RT) and acoustic time difference (AC) is subject to a certain uncertainty that affects the accuracy of the TOC prediction. The TOC prediction value based on the traditional and improved Δlog R models differs greatly from the measured TOC values, with R2 values of 0.5802 and 0.7711, respectively. (3)The TOC value shows good correlations with well-logging parameters such as natural γ (GR), RT, and density (DEN), leading to constructing a multiple linear regression model for the TOC prediction. Compared with traditional and improved Δlog R models, the R2 of our model can reach 0.9181, which is more suitable for the TOC prediction of the Chang 9 source rock in the Longdong area, Ordos Basin. The TOC value shows good correlations with well-logging parameters such as natural γ (GR), RT, and density (DEN), leading to constructing a multiple linear regression model for the TOC prediction. Compared with traditional and improved Δlog R models, the R2 of our model can reach 0.9181, which is more suitable for the TOC prediction of the Chang 9 source rock in the Longdong area, Ordos Basin. (4)The map of predicted source rock TOC shows that the distribution of Chang 9 source rock (TOC ≥ 1.0%) is mainly controlled by the sedimentary environment. The thickness varies in the range of 1–12 m and shows a decreasing trend from northeast to southwest in the Londong area, Ordos Basin. The map of predicted source rock TOC shows that the distribution of Chang 9 source rock (TOC ≥ 1.0%) is mainly controlled by the sedimentary environment. The thickness varies in the range of 1–12 m and shows a decreasing trend from northeast to southwest in the Londong area, Ordos Basin. (5)From the perspective of the relative abundance of C30 rearranged hopane and the distribution of the Chang 9 source rock, it is speculated that the tight oil of the Chang 9 Member in the northeastern areas derives from the Chang 9 source rock, while the oil in other areas mainly from the Chang 7 source rock. From the perspective of the relative abundance of C30 rearranged hopane and the distribution of the Chang 9 source rock, it is speculated that the tight oil of the Chang 9 Member in the northeastern areas derives from the Chang 9 source rock, while the oil in other areas mainly from the Chang 7 source rock.