Well-log Data Research Articles

Machine Learning Based Reservoir Characterization and Numerical Modeling from Integrated Well Log and Core Data

In recent years, artificial intelligence applications have proved useful for analyzing petrophysical data and characterizing subsurface formations. Conventional reservoir characterization employs core data measurements and local correlations between porosity and permeability as input data for reservoir property modeling. The vertically and laterally sparse nature of core and well-log data respectively present challenges in developing three-dimensional models of subsurface properties. In addition, a strong correlation between porosity and permeability as well as reliable core measurements are not always available. The complex interrelationship within the dataset is further evidenced through the application of parametric and nonparametric approaches, which are commonly utilized in conventional regression methods. The parametric method yielded an RMSE value of 0.335 and the nonparametric, 0.418, necessitating the implementation of an algorithm capable of modeling the complex relationship within the dataset. The proposed approach uses an improved machine learning (ML) workflow to generate and evaluate the performance of different porosity and permeability models from integrated well-log and core data, comparing the performance to traditional methods. The workflow combines a suite of unsupervised and supervised ML methods in establishing the relationship between porosity and permeability for different zones. Unsupervised ML algorithms such as K-means, K-median, and hierarchical clustering are applied to the data to generate groups with similar trends. Supervised ML regression methods such as Gaussian process regression (GPR), neural network regression, support vector machine (SVM) regression, and ensemble regression are implemented to predict permeability based on gamma ray, bulk density, and deep resistivity logs. This method demonstrated enhanced dataset modeling, as indicated by the decrease in RMSE values when using the GPR following the hierarchical clustering technique. This recorded RMSE values as low as 0.125, which is roughly 34% lower than those observed with traditional methods. Two of the predicted models were assessed using a history-matching process, which showed an almost perfect fit for nearly all the wells, thereby confirming the accuracy of the regression.

Intelligent Prestack Multi-trace Seismic Inversion Constrained by Probabilistic Geological Information

Seismic inversion uses observed data including well-logging and seismic data to infer rock parameters. However, it still suffers from non-unique solutions. The non-uniqueness increase when extrapolating away from well location. To mitigate this effect, we incorporate geological information into a deep learning (DL) framework in the form of generated probabilistic labels. We transform geological information into reservoir heterogeneity (RH) weights and geological pattern constraint (GPC) weights to generate probabilistic labels. The RH weights are used to capture the variability in accuracy during the spatial extrapolation of well data. The RH weights are used to balance the constraint weights between well and seismic data in the DL model. The GPC weights are used to characterize the changes of geological patterns from well locations to non-well locations. Based on variations in local prestack seismic waveforms, the DL model learns extrapolation patterns of well data with similar geological patterns. By combining the rock parameters derived from well logs with these two types of weights, we can generate a series of probability labels. Two types of geological information and well-logging data are deeply integrated into a supervised 2D-ResUNet framework. In this way, we develop a novel DL-based prestack multi-trace seismic inversion method. We validate our method on a 3D synthetic model and a 3D field dataset. The results show that our method exhibits great potential for characterizing the spatial variation of subsurface rocks and outperforms traditional deep learning methods.

Delineation of the reservoir petrophysical parameters from well logs validated by the core samples case study Sitra field, Western Desert, Egypt.

In the northern section of the Western Desert, there are many extremely profitable petroleum and natural gas deposits in the Abu EL-Gharadig Basin. This study aims to highlight the hydrocarbon potential of Abu Roash F Formation, which stands for high organic content unconventional tight reservoirs, and Abu Roash G Formation which stands for conventional sand reservoirs, in Sitra field located in the central-western part of the Abu EL-Gharadig Basin. The research employed well-log data from four wells to ascertain petrophysical properties combined with core samples of two wells for a comprehensive examination and description of lithology. Initially, we commenced the execution of petrophysical analysis, encompassing log quality control procedures. Subsequently, we identified and revealed zones of interest and hydrocarbon indicators in both formations. Additionally, we ascertained the three most influential parameters, shale Volume, effective Porosity, and water saturation, which serve as defining factors for reservoir quality. Subsequently, an examination of the core samples, which encompassed lithologic description, lithofacies analysis, paleoenvironmental interpretation, petrographic analysis, and porosity assessment is conducted. For the sake of a more accurate interpretation, we conclude our research with cartographic maps created to evaluate the geographical distribution of hydrocarbon potential based on petrophysical characteristics, Distribution of the net-to-gross ratio among wells by correlating the litho-saturation models (rock models) for the four wells. The foregoing results declare that The Abu Roash F carbonate-rich rocks are a contender for unconventional tight oil reservoir potential with thin secondary porosity and high organic content, which normally requires a kind of hydraulic fracturing for prospective oil extraction, Furthermore, the upper section of Abu Roash G formation, particularly in well sitra8-03, has highly favorable conventional reservoir characteristics.

Comparison of indentation test results and well-logging data for estimation of uniaxial compressive strength and Young’s modulus

Indentation test results and well-logging data are indirect methods of estimating Young’s modulus (E) and uniaxial compressive strength (UCS). The current study compared the results of indirect methods for a well in an oil field in southern Iran. The indentation tests were conducted by an indenter with a diameter of 260 µn on over 500 drill cuttings. The axial load versus displacement curve was used to determine the indentation modulus (IM) and critical transfer force (CTF) for each drill cutting. Indentation tests were performed perpendicularly and parallel to the weakness surface and were determined CTFV, IMV, CTFP, and IMV. The shear and compressive wave velocity (VS and VP ) for the well-logging data were determined and the UCS and static Young’s modulus (Esta.) were determined. Empirical equations for UCS-CTFV, UCS-CTFP, UCS-IMV, UCS-IMP, Esta-CTFV, Esta-CTFP, Esta-IMV, and Esta–IMP were proposed. It was determined that the relationships between UCS-CTFV and Esta-CTFV were more valid and meaningful than the other relationships. Therefore, it is suggested that the indentation test be performed perpendicular to drill cuttings with a weakness surface. The results showed that the values obtained for CTFV and IMV were 20% higher than for CTFP and IMP.

High-resolution acoustic-impedance inversion based on a deep-learning-aided representation model of nonstationary seismic data

Building a high-resolution acoustic-impedance (AI) model based on nonstationary seismic data plays a key role in reservoir predictions. However, the common AI inversion methods are faced with two main shortcomings. First, because of the nonstationary feature of seismic data, a multiparameter inverse problem should be considered to not only estimate AI but also to estimate a time-varying wavelet or a quality factor ( Q) model, leading the problem to be more ill posed. Second, the resolution of the estimated AI is limited due to the band-limited nature of nonstationary seismic data. To address these issues, we develop a new nonstationary seismic AI inversion method. Notably, we develop a deep-learning-aided representation model to replace the nonstationary convolution model to solve the forward problem in inversion. Benefiting from multisource information (seismic data and well-log data) and the powerful nonlinear function fitting ability of deep learning, this model can map high-resolution AI to band-limited nonstationary seismic data without requiring a time-varying wavelet or a Q model. A new deep-learning architecture is developed for processing the spatio-temporal seismic data with better accuracy. In addition, total-variation regularization is adopted to enforce a physically reasonable AI model. The results of our 3D synthetic and field data experiments clearly demonstrate that our method has significant advantages over other common methods in building a high-resolution AI model.

Sequential binary classification of lithofacies from well-log data and their uncertainty quantification

Understanding and identifying the composition of various lithofacies in the subsurface is essential for successful reservoir characterization in hydrocarbon exploration. However, conventional methods such as core sampling and manual well-log interpretation are labor-intensive. As a result, many scientists are conducting research to use machine learning to study lithofacies more effectively and efficiently. However, as researchers are becoming more dependent on machine learning, an uncertainty analysis of machine-learning models is crucial to determine the reliability of the prediction results. Machine-learning algorithms that use ensemble methods provide an easy method for an uncertainty analysis but algorithms that do not use ensemble methods have difficulty in quantifying the level of uncertainty. In our research, we introduce a method known as sequential binary classification (SBC), which helps not only classify lithofacies but also to quantify and visualize the regions of uncertainty of the machine-learning models. SBC provides a method for using any classification algorithm of the user’s choice to construct an ensemble, which allows the user to quantify uncertainty readily. Our research uses the SBC algorithm to classify and quantify the uncertainty from the well-log data obtained from the North Sea near Norway. The results show that most of the lithofacies that exist in the region of interest share similar characteristics, which results in high uncertainty among the various lithofacies, and SBC helps to visualize these high uncertainties. We additionally demonstrate how SBC alleviates the class imbalance issue among the various lithofacies in the area, which is a very common problem in well-log data analytics.

Characterization of Natural Gas Hydrate Constrained by Well and Seismic Data in Qiongdongnan Basin

This study investigates the natural gas hydrates within the Qiongdongnan Basin by integrating well-log and seismic data. Through pre-stack inversion and rock physics analysis, key parameters such as P-wave and S-wave impedances were utilized to distinguish hydrate-bearing formations from other geological bodies. A low-frequency model was constructed using the Inverse Distance Weighting (IDW) algorithm to improve the precision of parameter inversion. This study employs a multi-constraint inversion strategy, incorporating hard constraints from multiple wells and soft constraints from geological frameworks, ensuring reliable inversion results. Findings indicate that hydrate reservoirs are characterized by increased wave velocity and density due to hydrate accumulation, providing insights into the spatial distribution and characteristics of hydrates. This research enhances the understanding of hydrate reservoirs and offers valuable data for exploration in the Qiongdongnan Basin.

Calculation and application of elastic modulus of mineral components in tight sandstone based on adaptive method

Abstract For the petrophysics modelling of tight sandstones, the elastic modulus of their sandstone and mudstone components are often substituted with those of quartz and clay, which affects model accuracy. To solve this problem, we innovate an adaptive approach for modelling the rock physics characteristics of tight sandstone. First, based on the relationship between P- and S-wave velocities from well logs and the elastic modulus of the rocks, the equivalent elastic modulus of tight sandstone under saturated conditions is calculated. Next, The Lee model and the Gassmann equation were jointly used to determine the equivalent elasticity modulus of tight sandstone matrix. The upper and lower limits of the equivalent elasticity modulus for mudstone and sandstone are established using the mudstone content curve, the least squares method and the Voigt–Reuss–Hill (V-R-H) model. We used the random-walk algorithm to accurately calculate the equivalent elastic modulus of the sandstone and mudstone components. Finally, using the accurately obtained elastic modulus of sandstone and mudstone, the equivalent elastic modulus of the matrix is calculated. The Kuster–Toksöz (K-T) model is subsequently applied to compute the dry frame bulk modulus and shear modulus of the rock. Following this, the Brie model is used to calculate the elastic modulus of the mixed fluid, thereby completing the construction of the rock physics model for the study area. The results demonstrate that our improved petrophysics model can predict the S-wave velocity curve with ≤ 15% errors relative to the true curve. When sweet spot prediction was performed using reservoir-sensitive parameter (Vp/Vs) derived from the petrophysics template, the agreement between the predictions and well-log data was 80%. Thus, our petrophysics modelling method can be used to predict tight gas reservoirs effectively and will aid efforts to improve petroleum exploration works in the study area.

The influence of major faults and fractures on the development of non-matrix porosity system in a pre-salt carbonate reservoir, Santos Basin – Brazil

Faults and fractures are central for characterizing the permeability distribution in carbonate reservoirs since they act as pathways for diagenetic fluids that often favor intense rock dissolution and permeability. Usually, high permeability zones and fractures are not easily recognized in seismic data due to limited resolution and they are often associated with higher concentrations of hydrocarbons or even significant fluid losses during drilling, thus creating a challenge for hydrocarbon exploration. In the Santos Basin, southeast Brazil, the pre-salt carbonate reservoirs from the Barra Velha Formation (BVE) are the main hydrocarbon producers in Brazilian Atlantic margin and well-known for being extremely heterogeneous, exhibiting complex dual-porosity systems. In this study, we built a conceptual model of these fracture zones and non-matrix porosity formation that helped narrowing the understanding of these complex systems. The characterization of faults and fractures was carried out using seismic, well-logs, and borehole image data to understand the influence of these structures in the porosity formation along the Barra Velha Formation. In the study area, three fault sets were defined (F1, F2, and F3) from seismic data. F1 represents to the larger faults, while the F3 fault set represents the smaller faults related to the reactivation of F1; both sets being oriented NE-SW. The F2 fault set is associated with the rift formation and is oriented to NNE-SSW. These three fault sets compartmentalized the studied area into different domains, each exhibiting distinct fracture sets. The natural open fractures were formed during the reactivation of rift faults and are oriented mainly NW, NNE-NNW, NE, and ENE and were identified across the entire study area, but with different intensity values. The fracture intensity closely relates to the distance from major faults where the wells with the highest fracture intensity occurs located 150–590 m from the larger F1 fault set. Scan-lines were conducted throughout the area to determine the fault width, and an average value of 1.2 km was established. Seven non-matrix porosity classes were characterized and classified into stratigraphically concordant and discordant non-matrix pore types at well scale through borehole image interpretation. The Barra Velha Formation exhibit higher occurrence of stratigraphically discordant non-matrix porosity related to fractured zones while stratigraphically concordant non-matrix porosity is mainly controlled by the paleotopography of the study area. Overall, non-matrix porosity formation tends to follow an orientation that suggests a preferential dissolution flow towards NE and ENE directions. Intervals with higher silica content shows a positive correlation with both fracture intensity and stratigraphically discordant non-matrix porosities. This work provides a conceptual model about the fractures and non-matrix porosity distribution in pre-salt carbonate rocks that can help address some of associated structural and stratigraphic uncertainties during field appraisal and development.

Exploring comparative heterogeneity management for precise water saturation assessment in carbonate formations: Dean-Stark measurements and beyond

Accurate determination of water saturation is a critical aspect of reservoir characterization and development potential of a hydrocarbon field. The inherent heterogeneity in carbonate reservoirs significantly influences the Archie equation coefficients and, consequently, water saturation calculations. In this study, 160 direct core sample water saturation measurements using the Dean-Stark method have been used. The primary goal was to identify the most suitable reservoir heterogeneity management approach for water saturation calculation. The research focuses on the Dalan and Kangan formations (equivalent to Khuff Formation), known as some of the most hydrocarbon-rich (gas) reservoirs globally, located in the western part of the Persian Gulf. To manage reservoir heterogeneity, we employed various techniques, including Winland R35, flow zone indicator, Lucia, pore types, and electrofacies, utilizing porosity and permeability core data, thin section studies, and well-logging data to classify the studied intervals into more homogeneous categories. Subsequently, we measured Archie parameters and, for the first time, water saturations derived from Dean-Stark measurements were compared with Archie saturations from various heterogeneity management methods. Additionally, the influence of geological and petrophysical parameters on the accuracy of water saturation calculations in different rock types is discussed. Our results highlighted the significance of geometrical attributes of pore types, pore throat radius, and permeability as key factors affecting the precision of water saturation calculations. Moreover, our investigation revealed that Lucia's rock typing methodology exerted a substantial influence on the control of permeability and the distribution of water saturation within the reservoir. Our analysis revealed that the electrofacies method provided the most precise estimates for water saturation, with a mean difference of 0.07 (v/v) units compared to the Dean-Stark method. Subsequently, the accuracy of predicted water saturation in rock types determined by the Lucia method, with a mean error of 0.09 (v/v), ranked second among the rock typing methods. Winland R35 and flow zone indicator methods exhibited the highest uncertainty in water saturation estimation, with mean differences of 0.23 and 0.21 (v/v), respectively, compared to the Dean-Stark method.

Efficient self-attention based joint optimization for lithology and petrophysical parameter estimation in the Athabasca Oil Sands

Accurately identifying lithology and petrophysical parameters, such as porosity and water saturation, are essential in reservoir characterization. Manual interpretation of well-log data, the conventional approach, is not only labor-intensive but also susceptible to human errors. To address these challenges of lithology identification and petrophysical parameter estimation in the Athabasca Oil Sands area, this study introduces an AutoRegressive Vision Transformer (ARViT) model for lithology and petrophysical parameter prediction. The effectiveness of ARViT lies in its self-attention mechanism and its ability to handle data sequentially, allowing the model to capture important spatial dependencies within the well-log data. This mechanism enables the model to identify subtle spatial and temporal relationships among various geophysical measurements. The model is also interpretable and can serve as an assistive tool for geoscientists, enabling faster interpretation while reducing human bias. The interpretable nature of the model should assist geoscientists in conducting faster quality checks of the predictions, ensuring that errors are not propagated to subsequent stages. This study adopts a multitask learning approach, jointly optimizing the model's performance across multiple tasks simultaneously. To evaluate the effectiveness of the ARViT model, we conducted series of experiments and comparisions, testing it against traditional artificial neural networks (ANN), Long Short-Term Memory (LSTM), and Vision Transformer (ViT) models. To showcase the versatility of ARViT, we apply Low-Rank Adaptation (LoRA) to a different smaller dataset, showing its potential to adapt to different geological contexts. LoRA not only helps in model adaptability but also helps to reduce the number of trainable parameters. Our findings demonstrate that ARViT outperforms ANN, LSTM, and ViT in estimating lithological and petrophysical parameters. While lithology prediction has been a well-explored field, ARViT's unique blend of features, including its self-attention mechanism, autoregression, and multitask approach along with efficient fine tuning using LoRA, sets it apart as a valuable tool for the complex task of lithology prediction and petrophysical parameter estimation.

Seismic reflection data interpretation and petrophysical evaluation of the Meyal area, the Potwar Basin, Pakistan

Despite extensive exploration efforts in the Potwar Basin, a thorough understanding of the structural framework and reservoir characteristics within the Meyal Field remains critical for pinpointing potential hydrocarbon zones. This study aims to integrate geology and geophysics to emphasize the structural and stratigraphic interpretation of the Meyal Field, including reservoir characterization using seismic and well-log data. The seismic interpretation of six seismic lines focuses on four key reflectors: the Kamlial, Chorgali, Sakesar, and Salt Range Formations. The Eocene Chorgali and Sakesar Formations are of primary importance for exploration and production. Structural analysis revealed that the Meyal anticline is a plunging structure bounded by a back thrust to the north and a fore thrust to the south, suggesting a favorable location for hydrocarbon accumulation. Time-depth contour maps derived from seismic data further delineate potential sites for future investigations. The well correlation indicated an uplift toward the Meyal-2 compared with Meyal-5, signifying a shallowing trend attributed to thrust tectonics. This tectonic regime has rendered the area highly prospective for hydrocarbon exploration, with thrusting and oblique-slip faulting enhancing the reservoir qualities of Eocene Formations. The petrophysical evaluation revealed favorable reservoir characteristics, including an average porosity ranging from 0% to 12% with an effective porosity of approximately 7.5%, water saturation up to 42%, and hydrocarbon saturation reaching 58% within the pay zone of the reservoir. These findings suggest that the Sakesar and Chorgali Formations within the Meyal oil field hold promise as productive hydrocarbon reservoirs. The integrated study provides valuable insights into the structural framework and reservoir characterization, highlighting the potential of the Eocene Formations as productive hydrocarbon reservoirs supported by favorable structural configurations and petrophysical properties.

Seismic scattering inversion for multiple parameters of overburden-stressed isotropic media

Subsurface reservoirs are compressed by overburden stress resulting from the gravity of overlying masses, and the resulting stress changes significantly affect the seismic reflection responses generated at the reservoir interfaces. Several exact reflection coefficient equations have been well established to delineate the role that in situ stress plays in altering the energy transition and amplitude of seismic reflection responses. These exact equations, however, cannot be effectively used in practice due to their intricate formulations and the difficult geophysical estimation for the third-order elastic constants (3oECs) embedded in reflection coefficients. Based on the theories of nonlinear elasticity and elastic wave inverse scattering, we derive an approximate seismic reflection coefficient equation for overburden-stressed isotropic media in terms of the P-wave modulus, shear modulus, density, and a defined stress-related parameter (SRP). The SRP is a combined quantity of elastic moduli, 3oECs, and overburden stress, which can be naturally treated as a dimensionless stress-induced anisotropy parameter. Its inclusion effectively eliminates the need for 3oECs information when using our equation to estimate the desired reservoir properties from seismic observations. By comparing our equation to the exact one, we confirm its validity within the moderate-stress range. Then, we introduce a Bayesian inversion approach incorporating the new reflection coefficient equation to estimate four model parameters. In our approach, the Cauchy and Gaussian distribution functions are used for the a priori probability and the likelihood distributions, respectively. The synthetic tests from two well-log data sets and a field example demonstrate that four parameters can be reasonably inverted using our approach with rather smooth initial models, which illustrates the feasibility of our inversion approach.

Research on porosity prediction method for petroleum reservoirs based on improved gene expression algorithm

Abstract Porosity, an essential parameter of reservoir properties, plays a crucial role in reservoir evaluation. Indirect prediction of reservoir porosity using well-logging data is a fundamental task in assessing hydrocarbon reserves. Multivariate stepwise regression and artificial neural network methods are currently typical prediction approaches. However, these methods suffer from drawbacks such as high workload, extensive training sample requirements, and high computational costs. Genetic Expression Programming (GEP), a novel genetic algorithm, exhibits remarkable function discovery capabilities, high efficiency, and the ability to function without requiring large training samples. This paper proposes a method that utilizes GEP to mine a porosity calculation model from well-logging sample data and subsequently compute the porosity. The effectiveness and efficiency of this method are validated through actual well-logging data.

Carbonate rock physics model using different approaches to estimate rock frame stiffness

Deepwater carbonate reservoirs in the Brazilian pre-salt have emerged as significant hydrocarbon plays in the region and globally. Seismic monitoring of these reservoirs is crucial to provide information for well placement assessments, to reduce uncertainty in reservoir models, and to enhance model-based decision making. To integrate seismic data quantitatively, a petroelastic model (PEM) is required to estimate elastic properties. The modeling of the rock frame stiffness is an essential part in the study of PEM. However, this becomes more complex for carbonate rocks given their diverse pore-types. One way for estimating stiffness is to model each distinct pore-type and include them into the overall rock frame (different inclusion models). In this research, an alternative approach is explored to estimate the rock frame stiffness. A proxy model is employed, and its coefficients are subsequently calibrated with well-log information. Two PEMs were developed using different approaches for rock frame stiffness modeling. The first was inspired by inclusion models and incorporated two pore types (compliant and stiff pores) for the modeling process. The second considered a mathematical regression model as a proxy to relate reservoir porosity to the rock frame stiffness. These two PEMs were tested on the Brazilian pre-salt carbonate rocks of the Barra Velha (BVE) formation using well-log data from three wells and core sample measurements for one well. The accuracy of the PEMs’ performances was assessed by measures (such as, mean absolute error and root mean square error) and visual comparisons of their predictions. The results showed that both PEMs predicted velocities (compressional and shear) within an acceptable match with the actual well-log data. Similarly, when tested on the core samples, the PEMs produced comparable results, which indicates the validity of the proxy, not only at the well-log scale but at the core scale. The visual comparison and the accuracy analyses of the results for all three wells confirmed that the two PEMs had comparable predictions. This is significant given that the proxy simplifies the modeling process and facilitates the representation of various pore-types in the prediction of elastic properties in carbonate rocks. Overall, the proxy for the rock frame stiffness modeling is a computationally less expensive model compared to the inclusion models which involve modeling of various pore-types. It also offers a straightforward model which enables the quantitative integration of seismic data in a multidisciplinary team of geosciences and petroleum engineers (for instance, early engagement of petroleum engineers in 3D/4D seismic history matching).

Effects of overconsolidation on seismic rock physics — A comparative study between unconsolidated sands and weakly cemented sandstone

In seismic rock physics, models typically assume rocks are normally consolidated, often neglecting the effects of overconsolidation. Recent studies reveal that overconsolidated sands, resulting from stress release, show different physical property variations than normally consolidated sands. However, similar research is lacking for weakly cemented sandstones, a common reservoir analog. This study aims to systematically compare the effects of overconsolidation and stress direction on unconsolidated sandstones and weakly cemented sandstones. It also assesses the impact of overconsolidation on seismic interpretation, monitoring, and rock physics modeling applied to the two types of rocks. Experimental measurements of porosity, P-wave, and S-wave velocities were collected under different stress paths between loading and unloading. For weakly cemented sandstone, well-log data from the Norwegian Sea and Barents Sea were compared with experimental data. Analyses focused on the velocity-porosity relationship, VP/VS ratio vs. AI, and VP/VS ratio vs. effective stress.

Seismic Facies-Controlled Porosity Prediction in a Tight Sandstone Reservoir Based on the XGBoost Algorithm

Porosity is one of the most important properties for evaluating hydrocarbon reservoirs. There is a strong nonlinearity between seismic data and rock physical properties. Machine learning methods possess the powerful ability to capture complex nonlinear relationships between these two parameters, holding significant potential for porosity prediction in tight sandstone reservoirs. In this study, we propose a seismic facies-controlled porosity prediction method based on the extreme gradient boosting algorithm. Through seismic meme inversion method,we obtain high-resolution P-impedance data. In the inversion process, lateral variations of seismic waveforms were utilized instead of the variogram function. We create labels for model training using porosity curves and borehole side P-impedance data, applying them to the extreme gradient boosting regressor. To demonstrate its feasibility, we apply the model to a real 3D case from an oil field in the Songliao Basin. Our application results demonstrate that this method can provide porosity predictions for tight sandstone reservoirs, maintaining consistency with seismic waveforms and adhering to seismic facies control patterns. The method uses only acoustic travel time, density, gamma rays, and spontaneous potential well-log data and post-stack 3D seismic data as inputs, allowing for quick porosity predictions in the field. Compared to the seismic facies-controlled inversion combined with support vector machine and multi-Layer perceptron methods, the method proposed in this study provides more reliable porosity predictions, offering insights and references for the exploration and development of tight sandstone reservoirs.

Petrophysical parameter estimation using a rock-physics model and collaborative sparse representation

The estimation of petrophysical parameters is key in the identification of underground reservoirs. Current petrophysical parameter estimation methods are typically constrained by the choice of particular rock-physics models, necessitating the use of distinct models for various reservoirs. Furthermore, the inherent pronounced nonlinearity of these models presents significant challenges to the solution process in reservoir parameter inversion. Although linearized petrophysical inversion methods can simplify the solving process, they can introduce errors due to linearization. To address these limitations, we develop a petrophysical parameter estimation method driven by a rock-physics model and collaborative sparse representation (CSR). In this approach, the rock-physics model governs the fundamental pattern of the petrophysical parameters, with the CSR introducing perturbations to minimize model errors. Our method maximizes the use of well-log data and the rock-physics model, thereby reducing errors associated with linearized rock-physics (LRP) models and enhancing the method’s adaptability. We use the joint dictionary method to learn features and correlations among multiple parameters from the existing well-log data. This learned dictionary is then used as a CSR regularization constraint and applied to refine the LRP inversion objective function. Finally, the objective function is minimized using the block coordinate descent method to predict the petrophysical parameters. The effectiveness of this method in enhancing the adaptability and accuracy of petrophysical parameter estimation is confirmed through synthetic and field data tests.

Control and prediction of bedding-parallel fractures in fine-grained sedimentary rocks: A case from the Permian Lucaogou Formation in Jimusar Sag, Junggar Basin, Western China

The fine-grained sedimentary rocks have numerous bedding-parallel fractures that are essential for the migration, enrichment, and efficient development of oil and gas. However, because of their variety and the complexity of the factors that affect them, their spatial prediction by the industrial community becomes challenging. Based on sample cores, thin sections, and well-logging and seismic data, this study employed a multi-scale data matching approach to quantitatively predict the development of bedding-parallel fractures and investigate their spatial distribution. Bedding-parallel fractures in the Lucaogou Formation in Jimusar Sag frequently occur along preexisting bedding planes and lithological interfaces. Unfilled bedding-parallel fractures inside or near source-rocks exhibit enhanced oil-bearing capacity. They were identified on micro-resistivity scanning images by the presence of regularly continuous black or nearly black sinusoidal curves. Overall, the developmental degree of bedding-parallel fractures was positively related to the brittle mineral and total organic carbon contents and negatively related to single reservoir interval thickness. The maintained porosity of the reservoir matrix contributed to a thorough response to factors affecting the development of bedding-parallel fractures. Here, an effective and objective method was proposed for predicting the development and distribution of bedding-parallel fractures in the fine-grained sedimentary rocks. The method was based on the matched reservoir interval density, reservoir interval density, and matched sweet spot density of bedding-parallel fractures. The prediction method integrated the significant advantages of high vertical resolution from logging curves and strong lateral continuity from seismic data. The average relative prediction error was 8% in the upper sweet spot in the Lucaogou Formation, indicating that the evaluation parameters for bedding-parallel fractures in fine-grained sedimentary rocks were reasonable and reliable and that the proposed prediction method has a stronger adaptability than the previously reported methods. The workflow based on multi-scale matching and stepwise progression can be applied in similar fine-grained sedimentary rocks, providing reliable technological support for the exploration and development of hydrocarbons.

Intelligent seismic AVO inversion method for brittleness index of shale oil reservoirs

The brittleness index (BI) is crucial for predicting engineering sweet spots and designing fracturing operations in shale oil reservoir exploration and development. Seismic amplitude variation with offset (AVO) inversion is commonly used to obtain the BI. Traditionally, velocity, density, and other parameters are firstly inverted, and the BI is then calculated, which often leads to accumulated errors. Moreover, due to the limited of well-log data in field work areas, AVO inversion typically faces the challenge of limited information, resulting in not high accuracy of BI derived by existing AVO inversion methods. To address these issues, we first derive an AVO forward approximation equation that directly characterizes the BI in P-wave reflection coefficients. Based on this, an intelligent AVO inversion method, which combines the advantages of traditional and intelligent approaches, for directly obtaining the BI is proposed. A TransU-net model is constructed to establish the strong nonlinear mapping relationship between seismic data and the BI. By incorporating a combined objective function that is constrained by both low-frequency parameters and training samples, the challenge of limited samples is effectively addressed, and the direct inversion of the BI is stably achieved. Tests on model data and applications on field data demonstrate the feasibility, advancement, and practicality of the proposed method.