Reservoir Permeability Prediction Research Articles

Thermomechanical coupling seepage in fractured shale under stimulation of supercritical carbon dioxide

As a waterless fracturing fluids for gas shale stimulation with low viscosity and strong diffusibility, supercritical CO2 is promising than the water by avoiding the clay hydration expansion and reducing reservoir damage. The permeability evolution influenced by the changes of the temperature and stress is the key to gas extraction in deep buried shale reservoirs. Thus, the study focuses on the coupling influence of effective stress, temperature, and CO2 adsorption expansion effects on the seepage characteristics of Silurian Longmaxi shale fractured by supercritical CO2. The results show that when the gas pressure is 1–3 MPa, the permeability decreases significantly with the increase in gas pressure, and the Klinkenberg effects plays a predominant role at this stage. When the gas pressure is 3–5 MPa, the permeability increases with the increase in gas pressure, and the influence of effective stress on permeability is dominant. The permeability decreases exponentially with the increase in effective stress. The permeability of shale after the adsorption of CO2 gas is significantly lower than that of before adsorption; the permeability decreases with the increase in temperature at 305.15 K–321.15 K, and with the increase in temperature, the permeability sensitivity to the temperature decreases. The permeability is closely related to supercritical CO2 injection pressure and volume stress; when the injection pressure of supercritical CO2 is constant, the permeability decreases with the increase in volume stress. The results can be used for the dynamic prediction of reservoir permeability and gas extraction in CO2-enhanced shale gas development.

A novel deep learning method based on 2-D CNNs and GRUs for permeability prediction of tight sandstone

The accurate calculation of tight sandstone reservoir permeability is crucial for optimizing production and maximizing natural gas recovery. Traditional physical model-based permeability calculation methods heavily rely on core experimental parameters. However, petrophysical experiments are costly, prompting the development of fast and reliable reservoir permeability prediction methods. In this paper, a novel deep learning model that combines two-dimensional convolutional neural network (2-D CNN) model and gated recurrent neural network is proposed to predict tight sandstone reservoir permeability. The 1-D conventional well logs data is converted into the 2-D geological feature map to fit the model's input dimension. The TimeDistributed layer wrapping technology is used to couple the 2-D CNN and GRU to extract comprehensive features of geological feature maps. By training the model, a depth nonlinear mapping between the spatial and temporal features in the well logs data and permeability is established. The model is divided into two subnetworks, 2-D CNNs and GRUs, which are used to compare and analyze the functions of different modules in the proposed model. Additionally, two traditional machine learning algorithms, SVM and XGBoost, are also introduced as comparison models. Four regression evaluation metrics are used to quantify the predictive performance of different models. The test results in a blind well demonstrate that the proposed method outperforms the comparison models in predicting permeability, with a coefficient of determination (R2) of 0.9305, a root mean square error (RMSE) of 0.0895, a mean squared error (MSE) of 0.0080, and a mean absolute error (MAE) of 0.0725. This model can provide insight into the petrophysical characteristics of reaction reservoirs based on conventional well logs data and can accurately estimate reservoir permeability under limited core data, which has great application potential in improving the exploration accuracy of unconventional oil and gas resources.

A Prediction Method of Compacted Rock Hydraulic Permeability Based on the MGEMTIP Model

The permeability of the fluid-bearing rock is an important parameter for reservoir prediction. The Kozeny-Carman (K-C) formulation based on electrical measurements effectively characterizes the permeability-resistivity relationship of rocks with a single mineral composition or high porosity. The complex pore structure and mineral composition of compacted reservoirs affect induced polarization (IP) characteristics, indirectly limiting the applicability of conventional electrical K-C models. The permeability of fluid-bearing rocks is an important parameter for reservoir prediction. The theoretical chargeability of the modified generalized effective medium theory of induced polarization (MGEMTIP) model includes the effects of various conductive minerals. Due to the disconnection assumption of the disturbed medium in the MGEMTIP, there is a significant difference between the theoretical chargeability and the measured chargeability, and the difference is a sensitive parameter of rock permeability. A semi-empirical reservoir permeability prediction model is proposed based on the MGEMTIP. Theoretically and experimentally, the prediction model based on MGEMTIP is compared with the two electrical K-C models. Under the condition that the rock does not contain low-resistivity minerals, the prediction model based on MGEMTIP is theoretically equivalent to the K-C model. The experimental results show that this prediction model is more suitable for low-porosity and low-permeability rocks containing low-resistivity minerals, and the prediction results can be effectively restricted to the same order of magnitude. From the perspective of differences between model assumptions and natural rocks, the prediction model provides a semi-empirical relationship between complex mineral IP characteristics and permeability. Combined with the geological information of the survey area, the permeability prediction model can provide a theoretical basis for reservoir permeability prediction based on electromagnetic exploration.

A multiple-input deep residual convolutional neural network for reservoir permeability prediction

Permeability plays an essential role in reservoir-related studies, including fluid flow characterization, reservoir modeling/simulation, and management. However, operational constraints and high costs limit the wide access to the direct measurements of reservoir permeability. Over the years, various machine learning (ML) techniques have been utilized to predict reservoir permeability from widely-available reservoir-related datasets (such as well logs). Nevertheless, experience in real case studies reveals that there is still room to improve the process of permeability prediction in both well and field scales. In this study, an innovative Deep Learning (DL)-based approach is developed for fast and accurate reservoir permeability prediction from conventional well logs. By integrating a multiple-input module, convolutional network, deeper bottleneck, and residual structures, we developed a novel Multiple-Input deep Residual Convolutional Neural Network (named MIRes CNN) for this purpose. It is simultaneously fed by two different types of datasets, Numerical Well Logs (NWLs) and Graphical Feature Images (GFIs). To construct this MIRes CNN model, we integrated a Single-Input deep Residual one Dimensional CNN (SIRes 1D-CNN) to handle the NWL dataset and a Single-Input deep Residual two Dimensional CNN (SIRes 2D-CNN) to treat with GFIs. The significant advantages of the proposed architecture are its capability to efficiently exploit mixed data, reduce both overfitting problem and computational cost, and add more flexibility to the process compared to the classical single-input deep neural networks. The other novelty of this work lies in its generation of an image from geophysical well logs, which offers some physical insight. We compared the proposed MIRes CNN model with two Single-Input deep Residual CNN methods (i.e., SIRes 1D-CNN and SIRes 2D-CNN) and two baseline methods (i.e., Random Forest (RF) and Group Method of Data Handling (GMDH)). The statistical results concluded that the proposed multiple-input method outperforms single-modality-based techniques and the baseline methods for all employed performance indicators. The study also demonstrates the importance of using physically meaningful log-derived images for CNN model training and the crucial role played by the modified architectures (e.g., residual and deeper bottleneck architectures) in enhancing the accuracy and performance of the model.

Reservoir Permeability Prediction Based on Analogy and Machine Learning Methods: Field Cases in DLG Block of Jing’an Oilfield, China

Abstract Reservoir permeability, generally determined by experimental or well testing methods, is an essential parameter in the oil and gas field development. In this paper, we present a novel analogy and machine learning method to predict reservoir permeability. Firstly, the core test and production data of other 24 blocks (analog blocks) are counted according to the DLG block (target block) of Jing’an Oilfield, and the permeability analogy parameters including porosity, shale content, reservoir thickness, oil saturation, liquid production, and production pressure difference are optimized by Pearson and principal component analysis. Then, the fuzzy matter element method is used to calculate the similarity between the target block and analog blocks. According to the similarity calculation results, reservoir permeability of DLG block is predicted by reservoir engineering method (the relationship between core permeability and porosity of QK-D7 in similar blocks) and machine learning method (random forest, gradient boosting decision tree, light gradient boosting machine, and categorical boosting). By comparing the prediction accuracy of the two methods through the evaluation index determination coefficient (R2) and root mean square error (RMSE), the CatBoost model has higher accuracy in predicting reservoir permeability, with R2 of 0.951 and RMSE of 0.139. Finally, the CatBoost model is selected to predict reservoir permeability of 121 oil wells in the DLG block. This work uses simple logging and production data to quickly and accurately predict reservoir permeability without coring and testing. At the same time, the prediction results are well applied to the formulation of DLG block development technology strategy, which provides a new idea for the application of machine learning to predict oilfield parameters.

Coupled Seepage Mechanics Model of Coal Containing Methane Based on Pore Structure Fractal Features

The paper applies fractal theory to the structure of fractal coal pores and calculates the fractal dimension and integrated fractal dimension for each pore section >100 nm, 100 nm > d > 5.25 nm, and <2 nm. In the experiment, we performed the full stress–strain-seepage experiment of methane-bearing coal, revealed the deformation–seepage characteristics of methane-bearing coal under load, and deduced the dynamic prediction mechanical model of methane-bearing coal permeability based on pore heterogeneity, followed by practical verification. The results show that the permeability change in methane-bearing coal is an external manifestation of coal pore deformation, and the two are closely related and affected by changes in the effective stress coefficient. The derived fractal-deformation-coupled methane permeability mechanics model based on coal pore heterogeneity has high accuracy, a general expression for the stress–strain-permeability model based on coal heterogeneity is given, and the fractal Langmuir model is verified to be highly accurate (>0.9) and can be used for coal reservoir permeability prediction.

Application of Machine Learning for Shale Reservoir Permeability Prediction

Due to ultra-low permeability, the characterization of shale reservoir is always being a challenge. The traditional models are insufficient to estimate the ultra-low permeability of shale reservoirs. Based on Machine Learning, we proposed a simple mathematical approach to predict the permeability of shale reservoirs. Machine-learning techniques are good options for generating a rapid, robust, and cost-effective permeability prediction because of their strengths to deliver the variables. Additionally, used the Kozeny’s equation with power mean approach to constraint the estimated permeability for more reliable. To do this, we used a pure shale well-log downloaded from open source. The results show that the predicted permeability is well correlated with the neutron log and significantly match with the other well-logs.

A new predrilling reservoir permeability prediction model and its application

The accurate prediction of sandstone permeability from variables such as porosity, composition, and texture is one of the major problems in petroleum geology. However, the application of geological variables related to diagenesis in permeability prediction models is relatively rare, especially the anomalous variations of clay content and type. This study investigates the differential control of sedimentation and diagenesis on permeability using a comprehensive analysis of cores, thin sections, routine core analysis, X-ray diffraction, and scanning electron microscopy from sandstone samples in the third member of the Weizhou Formation (Ew3), Weizhou 12-X oilfield, Weixinan Sag, Beibu Gulf Basin, China. An innovative sandstone reservoir permeability prediction model is proposed by introducing the clay factor β confining the effect of clay minerals on permeability, predicting the spatial distribution of diagenetic index (ID) using the Diagenetic Modeling System (DMS), and improving the Walderhaug permeability prediction model. The permeability of the well evaluated in the study was predicted with the relative and absolute errors changing from 0.01 mD to 192.65 mD and from 0.00 to 1.00, with an average values of 38.6 mD and 0.33, respectively, based on the sandstone reservoir permeability prediction model in Ew3. The permeability predicted by the permeability prediction model matches well with the measured permeability in Ew3. The results indicate that permeability is successfully predicted by the modified Walderhaug permeability prediction model for most sandstone reservoirs with strong diagenesis (such as anomalous changes in clay mineral types and content) under the background of a single distributary channel sedimentary microfacies. In general, anomalously high permeability zones are preferentially distributed in dominant sedimentary-diagenetic facies.

Permeability Predictions for Tight Sandstone Reservoir Using Explainable Machine Learning and Particle Swarm Optimization

High-precision permeability prediction is of great significance to tight sandstone reservoirs. However, while considerable progress has recently been made in the machine learning based prediction of reservoir permeability, the generalization of this approach is limited by weak interpretability. Hence, an interpretable XGBoost model is proposed herein based on particle swarm optimization to predict the permeability of tight sandstone reservoirs with higher accuracy and robust interpretability. The porosity and permeability of 202 core plugs and 6 logging curves (namely, the gamma-ray (GR) curve, the acoustic curve (AC), the spontaneous potential (SP) curve, the caliper (CAL) curve, the deep lateral resistivity (RILD) curve, and eight lateral resistivity (RFOC) curve) are extracted along with three derived variables (i.e., the shale content, the AC slope, and the GR slope) as data sets. Based on the data preprocessing, global and local interpretations are performed according to the Shapley additive explanations (SHAP) analysis, and the redundant features in the data set are screened to identify the porosity, AC, CAL, and GR slope as the four most important features. The particle swarm optimization algorithm is then used to optimize the hyperparameters of the XGBoost model. The prediction results of the PSO-XGBoost model indicate a superior performance compared with that of the benchmark XGBoost model. In addition, the reliable application of the interpretable PSO-XGBoost model in the prediction of tight sandstone reservoir permeability is examined by comparing the results with those of two traditional mathematical regression models, five machine learning models, and three deep learning models. Thus, the interpretable PSO-XGBoost model is shown to have more advantages in permeability prediction along with the lowest root mean square error, thereby confirming the effectiveness and practicability of this method.

A method to predict heterogeneous glutenite reservoir permeability from nuclear magnetic resonance (NMR) logging

Permeability is an important input parameter in evaluating formation validity and predicting deliverability; its value is always estimated from porosity or other formation parameters based on the statistical methods. However, glutenite reservoir permeability prediction faces a great challenge due to the strong heterogeneity. In this study, to establish a reasonable model to predict heterogeneous glutenite reservoir permeability, 483 core samples were recovered from the Permian Jiamuhe Formation, Shawan Depression, northwestern Junggar Basin, China for routine physical property experiments; 21 of them were chosen to apply for laboratory nuclear magnetic resonance (NMR) measurements. After introducing the hydraulic flow unit (HFU) theory, these 483 core samples are classified into eight categories based on the flow zone indicator (FZI) difference, and applicative relationships of core-derived porosity and permeability are respectively established for every type of core sample. Meanwhile, based on the deeply analyzing of the expression of FZI and the classic Timur-Coates permeability-based model, a theoretical model of calculating FZI from NMR data is proposed. The involved input parameters are calibrated by using the experimental data of 21 core samples. After the established models are extended into field applications, the consecutive FZI curve is estimated from NMR logging to classify formations, and the corresponding relationships are used to predict permeability. Comparisons of the predicted FZI and permeability with core-derived results verify the reliability of the proposed models. They can be well used to precisely predict permeability in our target heterogeneous glutenite reservoirs.

Fast prediction of reservoir permeability based on embedded feature selection and LightGBM using direct logging data

Permeability estimation plays an important role in reservoir evaluation and hydrocarbon development, etc. Traditional physical model-based methods have problems with being time consuming and high cost. The applications of machine learning are currently becoming more and more extensive, however, there are still several limitations to previous machine learning-based permeability estimation methods, such as a limited number of samples, a requirement of prior knowledge, and some parameters needing to be calculated indirectly. In this paper, a hybrid reservoir permeability prediction approach, which is based on a certain scale of permeability dataset, embedded feature selection (EFS) and a light gradient boosting machine (LightGBM), is proposed. First, EFS is used to select features from the raw dataset. Then a LightGBM is adopted to predict the permeability. The influence of feature selection threshold, the base learners’ number and dataset size on prediction results is also investigated. In addition, different feature selections and prediction models are compared. The proposed hybrid approach is also verified on other datasets. The experimental results show that the proposed approach can effectively predict the reservoir permeability based on limited direct logging data.

Machine learning in reservoir permeability prediction and modelling of fluid flow in porous media

Reliable data on the properties of the porous medium are necessary for the correct description of the process of displacing hydrocarbons from the reservoirs and forecasting reservoir performance. The true permeability of the reservoir is one of the most important parameters which determination is time-consuming, costly and require skilled labor. The paper describes the methodology for determining the permeability of a porous medium, based on machine learning. The results of laboratory experiments, available in the database (terrigenous reservoirs with permeability in the range from 12 to 1132 md), are used to train the neural network, and then to predict the reservoir permeability. Comparison of the predicted and calculated permeability values showed a fairly good match between them with the determination coefficient of 0.92. The last task considered in this paper is to obtain an analytical expression describing a fluid flow in a porous medium using machine learning. This procedure enabled to obtain a resultant equation of fluid flow in a wide range of reservoir parameters and pressure gradients, which can be used in reservoir simulators.

A parametric study of machine learning techniques in petroleum reservoir permeability prediction by integrating seismic attributes and wireline data

One of the recipes for the big data and artificial intelligence paradigms is multi-dimensional data integration for improved decision making in petroleum reservoir characterization. Various machine learning (ML) techniques have been applied. However, there is still ample room for improvement. This paper carries out a rigorous parametric study to investigate the comparative performance of common and sophisticated ML techniques in the estimation of the permeability of a carbonate reservoir in the Middle East. The study integrates seismic attributes and wireline data for improved permeability prediction. The effects of tuning hyperparameters on the performance of the techniques are also studied. The techniques are evaluated on two versions of the seismic-log integrated data: globally-averaged and depth-matched. The results show that using the depth-matched dataset gives marginal improvement on the permeability prediction as reflected in the higher correlation coefficient and lower errors than the globally-averaged version. The outcome of this study will assist users of ML techniques to make informed choices on the appropriate ML techniques to use in petroleum reservoir characterization for more accurate predictions and better decision making especially when faced with limited and sparse data.

Mechanism of Permeability Evolution for Reservoir Sandstone with Different Physical Properties

Permeability of sandstones with different properties taken from Chongqing reservoirs has been measured and deeply discussed under increasing deviatoric stress. Corresponding to the distinct features in the stress-strain behaviors, the permeability of sandstones is found to evolve with a clear permeability decrease in the initial closure region, a constant permeability value in the elastic region, a permeability increase in the crack initiation and propagation region, a sharp permeability increase in the crack growth region, and a decrease permeability in the residual stage. The results also show that the variation patterns of permeability are similar for two reservoir sandstones under combination of confining pressure and water pressure; however, the strength and permeability are smaller for the sandstone with mud than that without mud, deeply indicating that mud-like materials have a relatively great impact on the mechanical properties and permeability, so mud components cannot be ignored for prediction of reservoir permeability. Furthermore, a statistical damage constitutive model considering hydraulic-mechanical coupling process is presented to calculate the damage variable D, illustrating that larger water pressure will result in a relatively smaller damage variables D and corresponding maximum, which explains that the permeability increases more rapidly and is larger for the sandstone without mud than that with mud, and sandstone damage related to corresponding circumferential crack strain and permeability has been investigated, also implying the evolution mechanism of permeability for two sandstones with different physical properties. Therefore, it is worth pointing out that rock physical properties have a great influence on the reservoir permeability under complex extraction conditions and cannot be ignored, which is necessary to improve the recovery ratio and productivity.

An ensemble of multi-model regression framework based on fuzzy clustering using inference system architecture for reservoir permeability prediction

An ensemble of multi-model regression framework based on fuzzy clustering using inference system architecture for reservoir permeability prediction

An ensemble of multi-model regression framework based on Fuzzy clustering using Inference System architecture for Reservoir Permeability prediction

One of the critical engineering problems in optimisation reservoir development is petroleum reservoir description and characterisation. Also, the successful applications of fuzzy inference system (FIS) and ensemble learning method in reservoir characterisation have been reported. In this study, we proposed an ensemble of multi-model regression framework based on FIS architecture to tackle the challenge of permeability prediction using logs data properties. The study demonstrates the capability of the ensemble model when tested in well log properties which is practical data of Oligocene geological types from Cuu Long basin. Empirical results indicate that our proposed algorithm framework is efficient and has the significant improvement compare to each existing standard single model.

Inversion of the permeability of a tight gas reservoir with the combination of a deep Boltzmann kernel extreme learning machine and nuclear magnetic resonance logging transverse relaxation time spectrum data

In view of the low accuracy of the existing nuclear magnetic-resonance (NMR) logging permeability model in tight sandstone reservoirs, we derive a relationship between the NMR [Formula: see text] spectrum and permeability based on the transverse relaxation theory of NMR and the Kozeny-Carman equation. We determined the reasons for the low accuracy of the model through the theoretical analysis. We have developed the deep Boltzmann kernel extreme learning machine (DBKELM) to improve the deep-learning algorithm and to predict the reservoir permeability based on NMR logging with a deep Boltzmann machine (DBM). We use the permeability data of 200 rock specimens in a tight gas reservoir in a certain area and the corresponding [Formula: see text] spectra from NMR logging for modeling. We apply the model to the evaluation of permeability in this area. The results show that the accuracy of the deep-learning algorithm is higher than that of the existing NMR logging permeability model and the shallow layer machine-learning model. Furthermore, the accuracy of the DBKELM that we have developed is higher than that of the DBM, which indicates that DBKELM is more suitable for the prediction of reservoir permeability. Therefore, deep-learning theory can be effectively used in oil exploration and development, and it can improve the interpretation accuracy of reservoir parameters. These findings contribute to the interpretation of reservoir parameters.

Prediction of reservoir permeability from porosity measurements for the upper sandstone member of Zubair Formation in Super-Giant South Rumila oil field, southern Iraq, using M5P decision tress and adaptive neuro-fuzzy inference system (ANFIS): a comparative study

This study focus on the determining the reservoir permeability from measured porosity using different techniques namely linear regression, adaptive neuro-fuzzy inference system, and M5 decision trees. The upper sandstone member of Zubair Formation in Rumila oil field in south of Iraq was taken as an example to reveal the efficacy of these techniques for simulating porosity and permeability relationship for this heterogeneous reservoir. The models performances were evaluated using root mean squared error (RMSE) and coefficient of determination (R2) error statistics. Results demonstrated that ANFIS model with 0.380 (RMSE) and 0.844 (R2) for the testing phase perform the best among the other models. The results of M5 decision trees model in testing phase were very similar to that of ANFIS with 0.378 (RMSE) and 0.842 (R2). The complex structure of ANFIS if it compared with a very simple structure and very easy to interpret M5 model make this technique is a best choice for predicting reservoir permeability form core porosity in the study area. The study implies that both M5 and ANFIS were promising tool for identifying porosity–permeability relationship in the study area.

A Novel Homogenous Hybridization Scheme for Performance Improvement of Support Vector Machines Regression in Reservoir Characterization

Hybrid computational intelligence is defined as a combination of multiple intelligent algorithms such that the resulting model has superior performance to the individual algorithms. Therefore, the importance of fusing two or more intelligent algorithms to achieve better performance cannot be overemphasized. In this work, a novel homogenous hybridization scheme is proposed for the improvement of the generalization and predictive ability of support vector machines regression (SVR). The proposed and developed hybrid SVR (HSVR) works by considering the initial SVR prediction as a feature extraction process and then employs the SVR output, which is the extracted feature, as its sole descriptor. The developed hybrid model is applied to the prediction of reservoir permeability and the predicted permeability is compared to core permeability which is regarded as standard in petroleum industry. The results show that the proposed hybrid scheme (HSVR) performed better than the existing SVR in both generalization and prediction ability. The outcome of this research will assist petroleum engineers to effectively predict permeability of carbonate reservoirs with higher degree of accuracy and will invariably lead to better reservoir. Furthermore, the encouraging performance of this hybrid will serve as impetus for further exploring homogenous hybrid system.

Reservoir permeability prediction by neural networks combined with hybrid genetic algorithm and particle swarm optimization

ABSTRACTReservoir characterization involves describing different reservoir properties quantitatively using various techniques in spatial variability. Nevertheless, the entire reservoir cannot be examined directly and there still exist uncertainties associated with the nature of geological data. Such uncertainties can lead to errors in the estimation of the ultimate recoverable oil. To cope with uncertainties, intelligent mathematical techniques to predict the spatial distribution of reservoir properties appear as strong tools. The goal here is to construct a reservoir model with lower uncertainties and realistic assumptions. Permeability is a petrophysical property that relates the amount of fluids in place and their potential for displacement. This fundamental property is a key factor in selecting proper enhanced oil recovery schemes and reservoir management. In this paper, a soft sensor on the basis of a feed‐forward artificial neural network was implemented to forecast permeability of a reservoir. Then, optimization of the neural network‐based soft sensor was performed using a hybrid genetic algorithm and particle swarm optimization method. The proposed genetic method was used for initial weighting of the parameters in the neural network. The developed methodology was examined using real field data. Results from the hybrid method‐based soft sensor were compared with the results obtained from the conventional artificial neural network. A good agreement between the results was observed, which demonstrates the usefulness of the developed hybrid genetic algorithm and particle swarm optimization in prediction of reservoir permeability.