The pore structures of shale in the Shahejie formation of Dongying depression were quantitatively characterized by FIB-SEM three-dimensional imaging technology. FIB-SEM was used to obtain high-resolution SEM images and through image registration, geometric correction and image segmentation, the organic pores and inorganic pores in the shale were distinguished. The pore space was converted into a structured pore network model via maximal ball method. Then, the pore size distributions and volume contributions of shale organic pores and inorganic pores, and the coordination number of the total pore network model were statistically analyzed via the pore network model. The results showed that due to the low thermal maturity of organic matter in the Shahejie formation, fewer organic pores were observed in the shale samples, and the number of pores was dominated by inorganic pores. Statistical analysis of the pore network model indicated that the studied shale samples considerably included mesopore (size of 2–50 nm) which accounted for 61.1%, but their contribution to the total pore volume was small (4.2%). Macropores (>50 nm) accounted for a relatively small proportion (38.9%) in number, but they provided the main storage space (95.8%) for the shale oil and gas. The volume contribution of pores to the total pore space at a certain bin size is controlled by the combination of pore number and the pore size. The coordination numbers of the two samples were mainly 0 and 1, indicating that the pore connectivity was poor and isolated pores accounted for most part. These results are critical for further reliable petrophysical simulations based on shale digital rocks as well as for the accurate understanding of their petrophysical properties.

Investigating the pore structure of shale rocks plays an important role in understanding the mechanism of shale oil accumulation and migration; thus, it has a notable influence on the exploration and development of shale oil resources. In the present study, the microstructure of oil shale samples extracted from the Shahejie Formation in the Dongying Depression was investigated through the joint usage of focused ion beam (FIB) and scanning electron microscopy (SEM). Herein, we selected four representative samples. The SEM images of the sample surfaces milled by FIB indicated that pores were found in both the organic and inorganic matter. Organic pores were less developed in the samples because of the low thermal maturity of the organic matter. Continuous 3D digital rock images of shale samples were generated by FIB grinding and SEM in situ imaging. Based on multithreshold segmentation, the shale rocks were separated into different parts including kerogen, inorganic matter, and pores. The pore connectivity and pore size distributions were analyzed via Avizo software and in‐house developed MATLAB scripts. The pore size and corresponding volume distributions suggested that mesoscale pores (2‐50 nm) were abundant in the samples but they contributed limitedly to the total pore volume. Macropores (>50 nm) accounted for a small proportion but they had a greater contribution to the overall pore space volume. The coordination numbers of the subject four samples were mainly 0 and 1, which means that isolated pores accounted for the vast majority of the pores. These observations are critical to conduct reliable petrophysical simulations in the oil shales and understand their intrinsic properties accurately.

Reservoir permeability is an important parameter for reservoir characterization and the estimation of current and future production from hydrocarbon reservoirs. Logging data is an important means of evaluating the continuous permeability curve of the whole well section. Nuclear magnetic resonance logging measurement results are less affected by lithology and have obvious advantages in interpreting permeability. The Coates model, SDR model, and other complex mathematical equations used in NMR logging may achieve a precise approximation of the permeability values. However, the empirical parameters in those models often need to be determined according to the nuclear magnetic resonance experiment, which is time-consuming and expensive. Machine learning, as an efficient data mining method, has been increasingly applied to logging interpretation. XGBoost algorithm is applied to the permeability interpretation of carbonate reservoirs. Based on the actual logging interpretation data, with the proportion of different pore components and the logarithmic mean value of T2 in the NMR logging interpretation results as the input variables, a regression prediction model is established through XGBoost algorithm to predict the permeability curve, and the optimization of various parameters in XGBoost algorithm is discussed. The determination coefficient is utilized to check the overall fitting between measured permeability versus predicted ones. It is found that XGBoost algorithm achieved overall better performance than the traditional models.

Complete and thorough core cleaning is a critical prerequisite for the precise measurements of most rock's petrophysical parameters. In shale, the oil cleaning process, aimed to remove the volatile hydrocarbons, is often complicated by the requirement for intact solid organic. Evaluation of shale's cleaning methods needs to take structural integrity of organic matrix into account but neglected in the existing researches. Here, we develop a novel evaluation method using a modified ESH (extended slow heating) pyrolysis cycle, which starts at a lower initial temperature of 150°C for 10 minutes and then slowly increases to 650°C by 10°C/min. Hydrocarbons on the ESH pyrogram were divided into light free hydrocarbon (S_A), FHR (fluid‐like hydrocarbon, S_B), and solid organic matter (S_C). We propose a set of quantitative evaluation criterions comparing the results of pyrograms, for different types of the hydrocarbons, at different cleaning conditions. We showed that a modified pyrogram achieves complete cleaning with S_A and S_B removed while S_C remains almost intact. The modified pyrogram achieves complete removal of FHR in the second stage of pyrogram, while earlier researches often report residual FHR. The introduced method improves the accuracy in the identification of production potential in kerogen‐rich shale reservoirs up to about 3% of the total pore volume. Further, the new approach allows a quantitative assessment for the cleaning quality without altering the sample's organic matrix. Future studies on the petrophysical properties of the hydrocarbon‐bearing reservoir rocks may benefit from the thorough hydrocarbon removal achieved through the modified pyrogram methods proposed in this study.

The interlayer is a key geological factor that regulates reservoir heterogeneity and remaining oil distribution, and its accurate identification directly affects the reservoir development effect. To address the strong subjectivity of traditional identification methods and the insufficient recognition accuracy of single machine learning models under imbalanced sample distributions, this study focuses on three types of interlayers (argillaceous, calcareous, and petrophysical interlayers) in the W Oilfield, and proposes an accurate identification method integrating the Synthetic Minority Over-Sampling Technique (SMOTE) and heterogeneous ensemble learning. Firstly, the corresponding data set of interlayer type and logging response is established. After eliminating the influence of dimension using normalization, the sensitive logging curves are optimized using the crossplot method, mutual information, and effect analysis. SMOTE technology is used to balance the sample distribution and solve the problem of the identification deviation of minority interlayers. Then, a heterogeneous ensemble model composed of the k-nearest neighbor algorithm (KNN), decision tree (DT), and support vector machine (SVM) is constructed, and the final recognition result is output using a voting strategy. The experiments show that SMOTE technology improves the average accuracy of a single model by 3.9% and effectively improves the model bias caused by sample imbalance. The heterogeneous integration model improves the overall recognition accuracy to 92.6%, significantly enhances the ability to distinguish argillaceous and petrophysical interlayers, and optimizes the F1-Score simultaneously. This method features a high accuracy and reliable performance, providing robust support for interlayer identification in reservoir geological modeling and remaining oil potential tapping, and demonstrating prominent practical application value.