Density logging is a key technique for calculating reservoir physical parameters, identifying lithology, and evaluating oil and gas reserves. Due to factors such as borehole conditions and poor tool contact, density curves often suffer from local data gaps, distortion, or noise interference. To address these issues, this paper proposes a density logging curve reconstruction method that integrates the K-nearest neighbors algorithm and the Transformer algorithm (KNN-Transformer). The method first employs KNN to retrieve samples with temporal sedimentary characteristics similar to the target segment within a multi-dimensional logging feature space. By calculating the Euclidean distance between the target segment and historical samples across multi-dimensional features such as acoustic travel time, natural gamma ray, and resistivity, the K most similar neighboring samples are selected to construct an enhanced geological prior input set, thereby improving the geological representativeness of the input data. Subsequently, the multi-head self-attention mechanism of the Transformer algorithm is utilized to establish long-range dependencies between arbitrary positions in the depth sequence, effectively integrating local similarity constraints with global sequential patterns. This achieves a synergistic representation of local features and global structures. Experimental results show that the KNN-Transformer algorithm achieves a mean absolute error (MAE) of 0.017 0 and a coefficient of determination (R2) of 0.953 3 for density curve reconstruction. Compared to typical algorithms such as support vector regression (SVR), linear regression, and long short-term memory (LSTM), the value of MAE is reduced by 30% to 60%. The method demonstrates higher reconstruction accuracy for both the overall trend and local details of the density logging curve, along with better stability and correctness at lithological interfaces and in complex intervals. This approach effectively recovers missing sections, corrects distortions, and suppresses noise, significantly improving both the numerical accuracy and geological plausibility of the reconstructed curves. It provides a reliable technical pathway for high-quality logging data reconstruction under complex reservoir conditions.

Lithofacies identification constitutes a critical step in the fine characterization of hydrocarbon reservoirs, and its accuracy directly affects the reliability of reservoir evaluation results. Existing identification methods are deficient in suppressing high-frequency noise in logging data and fail to accurately capture the longitudinal long-range dependencies of strata. Therefore, this paper proposes a hybrid deep learning model integrating Random Forest (RF) and Transformer (RF-Transformer for short), aiming to improve the accuracy and efficiency of shale lithofacies identification in heterogeneous reservoirs and provide technical support for fine reservoir characterization. The model first uses the RF module to evaluate the feature weights of logging curves (such as natural gamma ray, acoustic travel time, and resistivity) to screen key parameters, suppress high-frequency noise, and construct high-quality feature input vectors. Then, it employs the Transformer module to calculate the correlation weights of logging curves in parallel by virtue of the global context-aware capability of its self-attention mechanism, thus deeply mining and reconstructing the longitudinal long-range dependencies of strata. A dataset consisting of 3 800 field-measured samples from the southern Sichuan shale gas field, including 6 typical lithofacies types and 8 conventional logging curves, serves as the basis for model performance comparison and case application analysis. The results show that: ① The RF-Transformer model achieves an identification accuracy of 91.51%, which is 12.90%, 23.60%, and 47.54% higher than those of the Transformer, Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN) models respectively, and outperforms traditional machine learning models such as K-Nearest Neighbor (81.09%) and Decision Tree (77.28%). ② The model only requires about 25 iterations to reach a convergent state, with its convergence speed 8~10 times faster than that of existing models. ③ It successfully screens 6 key logging curves including natural gamma ray, acoustic travel time, and shallow lateral resistivity, and effectively eliminates redundant features such as deep lateral resistivity as well as non-geological noise. ④ In case applications, the predicted shale lithofacies profiles exhibit high longitudinal continuity and smoothness, which are highly consistent with actual geological stratification characteristics, enabling precise delineation of the boundaries of shale lithofacies transition zones. It is concluded that the RF-Transformer model realizes efficient and accurate shale lithofacies identification while ensuring both strong noise resistance and robust temporal dependency capture capability, thereby providing reliable technical support for the fine characterization of heterogeneous reservoirs. Future research needs to focus on optimizing the model for compatibility with logging interpretation software.

Under the backdrop of digital subsurface and intelligent field development, together with sustainable development planning, reliable and continuous well-log measurements are increasingly essential for reservoir evaluation and geological interpretation. Density (DEN) logging is critical for reservoir evaluation and geological interpretation, providing fundamental constraints for lithology/porosity-related assessment and integrated subsurface characterization. However, the DEN curve often contains missing intervals or distortions caused by borehole conditions and tool/environmental interference. This study proposes an RF-Transformer framework for DEN reconstruction that couples (i) Random-Forest-based feature screening to suppress redundant or low-contribution channels and (ii) a Transformer encoder with mask-aware self-attention to capture both local fluctuations and long-range depth dependencies. Experiments were conducted on logging data from nine vertical wells in the Lianggaoshan Formation (Sichuan Basin, China) with a unified sampling step of 0.125 m. Under a well-wise split protocol, RF-Transformer achieved RMSE = 0.0126 g/cm3, MAE = 0.0079 g/cm3, R2 = 0.9863, and r = 0.9932, outperforming Random Forest, Decision Tree, KNN, LightGBM, LightGBM-NN, and a base Transformer. The pass rate reached 92.86% under an error tolerance of +/- 0.02 g/cm3, demonstrating robust reconstruction in long missing sections and lithological transition zones. The proposed workflow provides an effective route for repairing density logs in complex reservoirs and for improving the continuity of multi-log interpretation.

CO₂ cooling damage significantly impairs reservoir recovery near the wellbore in a spatially non-uniformity way. Nuclear Magnetic Resonance (NMR) T2 tomography, an advanced technique for studying non- uniform fluid mobilization, was employed in this study. We first optimize the T2 tomography sequence to address challenges of low SNR and loss of microporous signals through short echo time (TE) scanning, achieving superior microporous imaging compared to conventional methods. Subsequent cyclic CO₂ huff-n-puff experiments on shale plugs, with controlled puff-rate adjustments, were conducted to investigate how CO₂ cooling influences oil mobilization in pores. Results demonstrated that: (1) The non-uniformity of cooling manifests as stronger temperature reduction closer to the bottom-end, leading to an overall 36% production decline during the first HnP cycle. (2) Faster puff-rates intensify cooling in plug's bottom-end, elevating residual oil saturation. (3) While rapid gas release enhances initial marginal recoveries, it later triggers severe pore blockage as cooled oil droplets from smaller pores migrate and accumulate in larger pores (T2 >1 ms), severely impairing long-term recovery. (4) Slower puff-rates enhance total recovery, as mild cooling exerts a weaker inhibitory effect on late-stage oil mobilization. This research provides critical insights for mitigating CO₂ cooling damage and optimizing gas injection strategies in shale reservoir development. Keywords: shale, CO₂ cooling effect, NMR, tomography

The occurrence and mobility of shale oil are critical issues in exploration and development. Shale reservoirs exhibit a complex fluid state, with oil and water present in various forms. The presence of organic matter and clay minerals within the reservoir framework further complicates the fluid’s occurrence and mobility. Utilizing two-dimensional nuclear magnetic resonance (NMR) experiments, in this study, core samples from the Shengli Oilfield’s shale oil reservoirs were analyzed. We conducted pyrolysis-NMR and CO2 huff-n-puff-NMR joint measurement experiments to assess the shale oil mobility. The results indicated that CO2 huff-n-puff was the most effective in the initial cycle, with diminishing returns in subsequent cycles, and NMR signal changes were predominantly observed in the movable oil fraction. The selected samples showed an average recovery rate of 26.9%, suggesting good mobility of shale oil in the study area. Based on the experimental results, a fluid component identification template for the study region was established, which mainly consists of the following five parts: movable oil, adsorbed oil, asphaltene, clay-bound water, structural water, and kerogen. This research provides valuable insights for the efficient development of shale oil reservoirs.

In the field of log interpretation of tight sandstone and shale reservoirs, it is difficult to quantitatively evaluate the fracture distribution parameters (fracture azimuth angle and fracture dip angle) near the borehole. Multi-component induction logging can measure the apparent resistivities of the reservoir in different directions by an orthogonal coil system, which lays a foundation for the quantitative evaluation of the fracture distribution parameters near the borehole. First, we derive the finite element equations for numerical calculation from the Maxwell equations for the time-harmonic field and construct a three-dimensional (3D) vector finite element numerical algorithm for the fractured formation. Second, we quantitatively analyze the multi-component induction logging detection characteristics of the fracture azimuth and dip angles. Third, we establish a fast extraction method of the fracture azimuth angle based on coordinate system rotation and an analytical calculation method of the fracture dip angle using the combination of cross components and coplanar components of multi-component induction logging. Finally, we use field data to test the fracture distribution parameter extraction method, and the fracture dip and azimuth angles calculated by multi-component induction logging are consistent with those calculated by borehole imaging logging, which realizes the fast extraction of the distribution parameters of fractures in tight sandstone and shale reservoirs.

Two-dimensional (2D) nuclear magnetic resonance (NMR) inversion operates with massive echo train data and is an ill-posed problem. It is very important to select a suitable inversion method for the 2D NMR data processing. In this study, we propose a fast, robust, and effective method for 2D NMR inversion that improves the computational efficiency of the inversion process by avoiding estimation of some unneeded regularization parameters. Firstly, a method that combines window averaging (WA) and singular value decomposition (SVD) is used to compress the echo train data and obtain the singular values of the kernel matrix. Subsequently, an optimum regularization parameter in a fast manner using the signal-to-noise ratio (SNR) of the echo train data and the maximum singular value of the kernel matrix are determined. Finally, we use the Butler-Reeds-Dawson (BRD) method and the selected optimum regularization parameter to invert the compressed data to achieve a fast 2D NMR inversion. The numerical simulation results indicate that the proposed method not only achieves satisfactory 2D NMR spectra rapidly from the echo train data of different SNRs but also is insensitive to the number of the final compressed data points.

裂缝是致密砂岩储层高产、稳产的一个关键因素,由于受常规测井分辨率低、致密砂岩岩性和物性变化快以及裂缝发育规模小且产状多变等因素影响,常规测井裂缝识别符合率偏低.以川西坳陷新场气田须家河组二段致密砂岩为例,在总结区域裂缝类型和裂缝测井响应特征的基础上,创新性地提取电阻率降低因子和时差增大因子等多种裂缝敏感因子,以及电阻率维度与声波维度约束指标,在放大裂缝信号的同时压制干扰影响,并首次采用灰色关联确定敏感因子权重系数,加权计算裂缝指示曲线,利用阈值约束实现裂缝识别,进一步通过与裂缝响应模式库对比,聚类识别裂缝类型.与X5和X101等5口井电成像对比,裂缝识别符合率达到90％以上,为研究致密砂岩裂缝展布规律、储层类型评价以及产能预测等提供了可靠依据.

位于巴西深J油田的盐下碳酸盐岩岩性及孔隙结构的识别对于区域储层的分布规律十分重要,但由于海外资料的获取量有限,如何最大限度地利用已有资料进行精准有效的岩性识别成为难题.本文在已有的各项地质信息资料的基础上,先利用成像测井资料提取出储层内主要岩性典型的结构标准图版,然后探索各类岩性在常规测井曲线上的响应规律,并提取出识别岩性的敏感参数,建立起常规测井参数的交会图法识别规律.确认其应用在全井段识别中取得效果后,进一步利用数据挖掘软件中的决策树技术,组合成交会图-决策树模型法.将此模型应用于实际井资料的处理,利用薄片资料和成像图像进行验证,得到了更高的岩性识别符合率,证实了模型在仅有常规测井资料中的适用性.

In this paper, we propose a method for obtaining two-dimensional T1–T2 spectrum from simultaneous inversion of the MRIL-Prime tool, dual-TW logging data, in order to improve the accuracy in identifying gas-bearing reservoirs. This paper was accomplished by analyzing the theoretical feasibility of the method, verifying its effectiveness by numerical simulation, and then applying the method to actual logging interpretations to identify gas-bearing reservoirs. The practical application results show that this method can circumvent misidentification of reservoirs due to the presence of large pores—a known issue with using a one-dimensional differential spectrum—and effectively identify gas-bearing reservoirs with low resistivity.

花古101区块储层致密、裂缝发育、非均质性强,导致产能预测存在较大误差.从平面径向流模型出发利用摩尔斯分类筛选法对产能敏感因素进行灵敏度分析,再结合各相关参数的变化范围得到其对产能的影响程度排序:渗透率>厚度≥压差>表皮系数>原油黏度>体积系数>供液半径>井眼半径,渗透率影响程度最大,裂缝是控制产能大小的重要因素.直接利用电成像测井资料与间接利用非均质参数实现对裂缝发育的综合定量评价.建立了声波时差和自然伽马分形维数差值与电成像孔隙度谱二维劳伦兹系数的相关性,结果表明分形维数差值越大,二维劳伦兹系数越小,地层非均质程度越强,裂缝越发育.考虑裂缝计算地层总渗透率对产能进行预测,计算结果与试油产能对应较好,实现了对花古储层产能疑难诊断分析.

Norm smoothing is commonly used in nuclear magnetic resonance (NMR) T2 inversion and the choice of a suitable regularization parameter is a key step for obtaining a satisfactory inversion result, which is usually achieved by repeating T2 inversion multiple times. However, a greater number of inversions result in a slower speed for the inversion process. In this paper, we propose a rapid norm smoothing T2 inversion method achieved using a new selection method for the regularization parameter. First, the singular value decomposition (SVD) method is used to calculate singular values of the kernel matrix to compress the echo train data. Subsequently, a suitable regularization parameter is calculated based on the signal-to-noise ratio (SNR) of the echo train and the maximum singular value of the kernel matrix, which avoids the repetitions of the T2 inversion. Finally, a rapid T2 inversion is obtained using the Butler–Reeds–Dawson (BRD) method. Numerical simulation and logging data inversion results show that the new method can rapidly provide reasonable T2 spectra for data with different SNRs and is insensitive to the amount of the compressed data.

岩石岩电实验表明电阻率增大系数与含水饱和度关系呈现非阿尔奇现象.基于岩石导电与渗流过程的相似性与差异性,分别建立纯岩石渗流与导电模型,尝试结合渗流特征探索岩石电阻率增大系数与含水饱和度的关系.建立岩石渗流模型,通过模型分析、公式推导和渗流实验数据,得到不同含水饱和度条件下水相渗流路径迂曲度比值表达式;建立纯岩石等效导电模型:将岩石孔隙空间划分为参与渗流的孔隙空间与不参与渗流的孔隙空间,2部分并联导电,结合渗流模型结论推导电阻率增大系数与含水饱和度的关系式,提出了差异渗流导电公式.该公式在一定条件下可退化为经典阿尔奇公式,其呈现的电阻率增大系数与饱和度关系趋势解释了岩石中存在的非阿尔奇现象.

塔河油田奥陶系洞穴型储层是该区最重要的油气生产层段,对于洞穴型充填物性质和充填程度的判断是评价该类储层是否具有储集空间和有效性的关键.由于充填特征的复杂性,使得利用常规测井资料评价洞穴型储层充填特征仍然面临着巨大的困难.本文通过分析塔河油田洞穴型储层不同充填物、不同充填程度的洞穴在常规测井、成像测井的测井响应特征,结合钻井、录井等方面的响应,利用泥质含量(Vsh)-浅侧向电阻率(RS)交会图识别洞穴充填物性质,包括:未充填,机械沉积、重力崩塌堆积和化学沉积物.由于研究区以机械沉积充填为主,利用无铀伽马(Kth)对洞穴型储层充填程度进行定量计算,提出了适合研究区储层洞穴充填程度计算的相关公式,提高了洞穴型储层充填程度的测井评价精度.对塔河奥陶系洞穴型储层93层进行解释,其中洞穴充填物性质识别准确率为86％,洞穴充填程度表征的准确率为92％.

The sedimentary facies, diagenetic facies and pore structures of the Upper Triassic Yanchang Formation Member 8 (Chang 8) reservoir in the Zhenjing region are studied by using core observation, thin section analysis, well logging and drilling data. The petrophysical faices of Chang 8 oil layers are analyzed to evaluate and classify the relative permeability. The petrophysical facies are classified into four types according to sedimentary facies, diagenetic facies and pore structure. The prison model of water relative permeability and universal formula of oil relative permeability are chosen to calculate water relative permeability and oil relative permeability, respectively. The results show that the calculated oil and water relative permeability based on the classification of petrophysical facies can be used to fully match with the production data, which verifies the accuracy of the proposed method to be used to evaluate the oil and water layers of Chang 8 reservoir.

The invention relates to a porosity analysis method. According to the method, abnormal points are quickly recognized and removed during establishment of the correspondence between logging parameters and core analysis porosity of a known reservoir, feature data which can represent the overall trend of the logging parameters and the core analysis porosity are selected, the more accurate correspondence between the logging parameters and the core analysis porosity of the known reservoir is established quickly, and the porosity of a to-be-analyzed reservoir is analyzed accurately according to logging parameters of the to-be-analyzed reservoir by using the correspondence. According to the method, only one abnormal datum is deleted each time, the overall correspondence trend of the logging parameters and the core analysis porosity is not affected, the data are prevented from being deleted by mistake, and the working efficiency is improved.

Based on the analysis of capillary pressure curves of Paleogene in A Sag, integrated with NMR, testing oil and fracturing data, effective porosity, absolute permeability, displacement pressure, pore throat mean, sorting coefficient and the maximum saturation of mercury are selected as the key parameters for the classification of reservoir. Based on 6 key parameters, a comprehensive reservoir classification index curve is constructed to classify Paleogene reservoirs of A Sag. The accuracy of evaluating reservoir productivity is 88.42% according to 95 oil test results. This reservoir evaluation method overcomes the shortcomings of the reservoir classification method based on conventional logging and capillary pressure curve of cores and improves the accuracy of reservoir classification.

On the basis of core data acquisition from Chenggu×well in the Archean buried hill reservoir,the porosities,permeabilities,densities,resistivities and wave velocities of granite and lamprophyre rocks in this reservoir are measured through crossplots,and furthers,from which analysed and obtained are the physical parameters of the reservoir rocks.The experiment matched equations are helpful in fractured reservoir identification and prediction.It is found from the experiments that lamprophyre is a better reservoir because its porosity is bigger and sonic wave velocity is slower;The elastic wave of the granite is faster and its compacted core porosity is lower.After being saturated by fluids,the lamprophyre is more sensitive than the granite.The wave velocity of the former changes with the increase of resistivity and stress,but the Poisson's ratio of the dried cores increases as the stress increases.The saturated water's Poisson ratio decreases as the stress increases.The wave velocity and Poisson ratio are nearly normal values as the pressure goes up.The experimental results provide a reference for the exploration and development of Archean buried hill reservoir in Dongying depression.

The NMR numerical simulation signals were used to compare and analyze NMR relaxation spectral inversion results by singular value decomposition(SVD)and iterative Tikhonov regularization method under the conditions of different numbers of pre-assigned relaxation bins,different modes of pre-assigned relaxation bins and different SNR.The results of the study indicate that SVD algorithm is suitable for the high signal-noise ratio(SNR 50);iterative Tikhonov regularization method is suitable for the lower signal-noise ratio of data inversion.Nuclear magnetic resonance logging data are inversed by the two algorithms with good application effect.