Improved Reservoir Characterization Research Articles

A novel approach to classify lithology of reservoir formations using GrowNet and Deep‐Insight with physic‐based feature augmentation

AbstractManual interpretation of geophysical logging data can be a tedious and time‐consuming task in the case of the nonlinear behavior of well‐logging signals. This study aims to enhance lithology classification of reservoir formations through advanced machine learning (ML) and deep learning (DL) techniques, introducing and comparing three novel algorithms, GrowNet, Deep‐Insight, and blender, against traditional models like random forest (RF) and support vector machine (SVM). Data from the South and North Viking Graben regions, encompassing 12 lithological facies, was preprocessed through cleaning, normalization, transformation, and imputation of missing values using regression models. The data set was enhanced with physic‐based features and balanced using SMOTE and NearMiss algorithms. Deep‐Insight converted tabular data into images for a convolutional neural network (CNN), significantly improving classification accuracy compared to conventional models such as decision trees (DTs). GrowNet and blender models leveraged hybrid approaches for enhanced performance. These hybrid approaches successfully addressed data imbalance and enhanced model learning, outperforming classic methods. The GrowNet and blender models for lithology classification successfully increased the penalty score and accuracy compared to the FORCE 2020 competition. Additionally, introducing the class prediction error plot visualizes multiclass classification performance more effectively than using a confusion matrix. These novel models in multiclass classification contribute to the petroleum industry by providing more accurate and reliable lithology classification, thereby improving reservoir characterization and exploration efficiency.

Identifying payable cluster distributions for improved reservoir characterization: a robust unsupervised ML strategy for rock typing of depositional facies in heterogeneous rocks

The oil and gas industry relies on accurately predicting profitable clusters in subsurface formations for geophysical reservoir analysis. It is challenging to predict payable clusters in complicated geological settings like the Lower Indus Basin, Pakistan. In complex, high-dimensional heterogeneous geological settings, traditional statistical methods seldom provide correct results. Therefore, this paper introduces a robust unsupervised AI strategy designed to identify and classify profitable zones using self-organizing maps (SOM) and K-means clustering techniques. Results of SOM and K-means clustering provided the reservoir potentials of six depositional facies types (MBSD, DCSD, MBSMD, SSiCL, SMDFM, MBSh) based on cluster distributions. The depositional facies MBSD and DCSD exhibited high similarity and achieved a maximum effective porosity (PHIE) value of ≥ 15%, indicating good reservoir rock typing (RRT) features. The density-based spatial clustering of applications with noise (DBSCAN) showed minimum outliers through meta cluster attributes and confirmed the reliability of the generated cluster results. Shapley Additive Explanations (SHAP) model identified PHIE as the most significant parameter and was beneficial in identifying payable and non-payable clustering zones. Additionally, this strategy highlights the importance of unsupervised AI in managing profitable cluster distribution across various geological formations, going beyond simple reservoir characterization.

Reservoir Rock Discrimination Based on Integrated Image Logs and Petrographic Analysis: A Case Study from the Early Miocene Nukhul Carbonate, Southern Gulf of Suez, Egypt

AbstractThe discrimination of rock types within the limestones and dolostones of the Nukhul Formation in the West Younis Field (Gulf of Suez Basin, Egypt) presents significant challenges due to their multi-scale compositional and diagenetic heterogeneity, diverse pore types, complex microstructures, and limited core data. This study aims to characterize the carbonate reservoir of the Early Miocene sediments and establish distinct reservoir rock types by employing textural analysis, geological interpretations (i.e., structural interpretation, fracture analysis, reservoir characteristics) using advanced imaging tools, and petrophysical measurements to model porosity/permeability profiles across the reservoir. A new dataset was obtained from the latest exploratory well in the West Younis Field, incorporating microresistivity and acoustic image logs, well logs, nuclear magnetic resonance (NMR) tools, and drill cutting petrographic analysis. The integration of these datasets provided a comprehensive understanding of the properties of the Early Miocene carbonate reservoir. Based on image logs, the carbonate facies were divided into four reservoir units. Petrographic evaluation further classified two facies (A and B) based on diagenetic factors controlling reservoir quality. The results revealed the occurrence of multiple phases of dolomitization, which influenced the reservoir quality. Early-stage dolomitization enhanced reservoir quality, while late-stage idiotopic dolomite crystal growth diminished it. The study also provided comprehensive information on the original rock fabric/texture, diagenetic processes, porosity types and origins, as well as the spatial distribution of pores (permeability index) within this complex carbonate reservoir. By employing an integrated technique, this study successfully differentiated the carbonate reservoir into distinct rock types, leading to improved reservoir characterization and field development. Additionally, the findings contribute valuable insights for the development and exploration of the Early Miocene carbonate section in the southern Gulf of Suez.

New insights into the Kingia Sandstone facies distribution, and implications for hydrocarbon prospectivity, north Perth Basin, Western Australia

The Kingia Sandstone has proven to be a prolific play fairway within the north Perth Basin, with primary porosity preservation observed down to at least 4600 mSS. However, the distribution of the Early Permian depositional and diagenetic facies has remained enigmatic. This is especially true given recent exploration results. Recent wells targeting the Kingia Sandstone have highlighted facies variability. As such, an opportunity exists to leverage legacy well data to investigate the interplay between sediment supply and accommodation space, resulting in improved reservoir characterisation. Through the integration of well logs, core and petrography studies, new insights into the spatial and temporal trends governing the presence and effectiveness of the Kingia are proposed. Implications of this work include Kingia Sandstone facies models that may be extended beyond known well control to help reduce subsurface uncertainty.

Improving Reservoir Characterization and Prediction Via Machine Learning-Driven Integration of Subseismic Geologic Concepts, Geophysical Attributes, and Wells

Summary Reservoir dynamic performance is affected by subseismic (below the resolution of seismic imaging) geologic features such as submeter thin barriers and baffles, or high-permeability streaks. Here, we present an application of machine learning that has the potential to improve and significantly accelerate the workflow for generating reservoir realizations and scenarios with relevant subseismic features consistent with both well-log and 3D/4D seismic data and/or attributes. We leverage unsupervised machine learning methods, in particular, a modification of diffusion probabilistic modeling (DPM).

Reservoir Geomechanics: A Data-driven Approach

Reservoir geomechanics is a crucial aspect of optimising and developing oil and gas activities, especially in maximising production. Recent technological advancements have revolutionised reservoir geomechanics studies, including integrating data-driven approaches. This review examines and integrates machine learning, data science, and data twin in reservoir studies. The primary aim is to identify the benefits, limitations, significant advancements, potential challenges, opportunities, and research gaps of data-driven approaches to reservoir geomechanics. Additionally, this study aims to create opportunities for further research to address these challenges. The review identifies cost-effectiveness, improved reservoir characterisation, and reduced operational risks as the benefits of integrating data-driven approaches in reservoir geomechanics. However, the review also highlights the significant challenges of data-driven approaches, such as insufficient datasets, limited interpretability, and limited transferability of models. By shedding light on these issues, this review provides a foundation for future research toward finding solutions to these challenges.

Rock Typing Approaches for Effective Complex Carbonate Reservoir Characterization

For highly heterogeneous complex carbonate reef reservoirs, rock typing with respect to depositional conditions, secondary processes, and permeability and porosity relationships is a useful tool to improve reservoir characterization, modeling, prediction of reservoir volume properties, and estimation of reserves. A review of various rock typing methods has been carried out. The basic methods of rock typing were applied to a carbonate reservoir as an example. The advantages and disadvantages of the presented methods are described. A rock typing method based on a combination of hydraulic flow units and the R35 method is proposed. Clustering methods for rock typing are used. The optimum clustering method is identified, and for each rock type, the permeability–porosity relationships are built and proposed for use in the geomodelling stage.

Artificial Intelligence Unleashes Potential of Relative Permeability

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 207855, “Unleashing the Potential of Relative Permeability Using Artificial Intelligence,” by Abdur Rahman Shah, SPE, Schlumberger; Kassem Ghorayeb, SPE, American University of Beirut; and Hussein Mustapha, Schlumberger, et al. The paper has not been peer reviewed. _ To unlock the potential of large relative permeability (Kr) databases, the work flow proposed in the complete paper integrates data analysis, machine learning (ML), and artificial intelligence (AI). The work flow allows for the automated generation of a clean database and a digital twin of Kr data, using AI to identify analog data from nearby fields by extending the rock-typing scheme across multiple fields for the same formation. Introduction Accurate Kr curves are critical because they can improve reservoir characterization, reduce uncertainty in history matching and production forecasting, and provide robust and reliable field development plans. However, preparing special core analysis (SCAL) data, particularly Kr curves, as an input for reservoir simulation traditionally has been a highly manual, time-consuming, labor-intensive process. Furthermore, no automated tools are available currently that perform a full quality review of Kr data using analytical or numerical methods. As a result, the quality and representativeness of Kr data used in reservoir simulation models is frequently inadequately checked. Furthermore, no tools exist to generate Kr curves based on analog field data in fields where Kr data is either missing or scarce. Integrated Work Flow The solution discussed in this paper aims to address challenges and automate processes using a combination of conventional algorithms and ML. It begins with a thorough quality check and cleanup of Kr laboratory data before using analog data to build Kr curves for a specific field or reservoir. The latter is especially important considering the challenges that arise when attempting to build on analog field data in the presence of large, corporate, regional, or country-scale databases. The approach classifies rocks using ML-proxy rock typing, which is subsequently used to find Kr equivalents. After that, these analogs can be used to fill up any data gaps for reservoir simulation in any discipline.

Time‐lapse PS‐wave time‐shifts and quantitative pressure‐saturation discrimination applications

AbstractWe show that time‐lapse (4D) time‐shifts from PS‐wave data are valuable as they are very sensitive to pressure changes, whereas 4D time‐shifts from PP‐wave data are sensitive to both pressure and saturation changes. This gives another opportunity from using 4D amplitudes alone to understand and discriminate areas of pressure changes from areas of saturation changes in field data. 4D PS time‐shifts sense transmission related changes taking into consideration propagation through the altered medium, whereas amplitude changes are reflection or backscattered signals from the boundary. Therefore, these two attributes will be complimentary in 4D interpretation. Rock‐physics‐based time‐shift cross‐plots are generated to analyse how pressure and saturation changes in a reservoir impact PP‐ and PS‐wave time‐shifts. The developments of 4D time‐shifts are shown fixing the reservoir properties and varying reservoir layer thickness. Synthetic seismic data from a wedge is modelled to understand differences between 4D amplitude and time‐shift changes and the value of PS time‐shifts. Cases with pressure changes only and saturation changes only are also analysed with the wedge model. The method is next applied to two 4D field data sets. One is in the North Sea and the other offshore Brazil. In these field data sets, with different levels of noise in the 4D data, we see interpretable results from PS‐wave time‐shifts that can help in 4D signal understanding and improve reservoir characterization and monitoring practices. We believe that with modern acquisition and processing, this technology is realizable and provides quantitative and accurate 4D interpretations. It is shown that the use of PS‐wave data can lead to a larger chance of success in placing development wells.

Improved reservoir characterization by means of supervised machine learning and model-based seismic impedance inversion in the Penobscot field, Scotian Basin

The present research work attempted to delineate and characterize the reservoir facies from the Dawson Canyon Formation in the Penobscot field, Scotian Basin. An integrated study of instantaneous frequency, P-impedance, volume of clay and neutron-porosity attributes, and structural framework was done to unravel the Late Cretaceous depositional system and reservoir facies distribution patterns within the study area. Fault strikes were found in the EW and NEE-SWW directions indicating the dominant course of tectonic activities during the Late Cretaceous period in the region. P-impedance was estimated using model-based seismic inversion. Petrophysical properties such as the neutron porosity (NPHI) and volume of clay (VCL) were estimated using the multilayer perceptron neural network with high accuracy. Comparatively, a combination of low instantaneous frequency (15–30 Hz), moderate to high impedance (7000–9500 gm/cc∗m/s), low neutron porosity (27%–40%) and low volume of clay (40%–60%), suggests fair-to-good sandstone development in the Dawson Canyon Formation. After calibration with the well-log data, it is found that further lowering in these attribute responses signifies the clean sandstone facies possibly containing hydrocarbons. The present study suggests that the shale lithofacies dominates the Late Cretaceous deposition (Dawson Canyon Formation) in the Penobscot field, Scotian Basin. Major faults and overlying shale facies provide structural and stratigraphic seals and act as a suitable hydrocarbon entrapment mechanism in the Dawson Canyon Formation's reservoirs. The present research advocates the integrated analysis of multi-attributes estimated using different methods to minimize the risk involved in hydrocarbon exploration.

Surveillance Optimizes Development Plan for a Deepwater Reservoir

_ This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 210274, “Effective Use of Surveillance To Optimize the Development Plan for a Deepwater Reservoir,” by Cengiz Satik, SPE, Matthew Burgess, SPE, and Matthew Carter, Chevron, et al. The paper has not been peer reviewed. _ A three-phase strategy was adopted for a deepwater reservoir development as a result of large uncertainties associated with fault compartmentalization and aquifer support. The first phase began with primary production from several wells, while water injection was implemented a few years later as a second phase. The third and final phase involved infills. Data collected during drilling were used to improve reservoir characterization. The complete paper describes how a surveillance, analysis, and optimization plan resolved key subsurface uncertainties and optimized the development plan and presents lessons learned and best practices. Introduction The field is in the central Gulf of Mexico Green Canyon. It produces from a series of stacked subsalt reservoirs consisting mainly of lower-slope to basin-floor turbidite sands. Uncertainty in flow behavior exists in prospective areas near low-confidence faults, affecting well performance and ultimate recovery predictions. To improve reservoir characterization and enhance field-performance predictions, a full-factorial 3D static Earth model was built by integrating well, surveillance, and seismic data. Dynamic simulation was used to help calibrate the well and field historical performance, forecast future recovery, and highlight and test potential opportunities. Despite achieving acceptable history-matched simulation models, specific development scenarios presented large recovery uncertainties and varying economic outcomes. In this work, results from reservoir surveillance modeling and execution were used to narrow the range of recovery uncertainty and improve field-development decisions. Dynamic reservoir-surveillance techniques, such as interference and pulse testing, used in this study provided key evidence that reduced fault compartmentalization uncertainty significantly and eliminated the need for a costly infill well to recover remaining resources. Furthermore, increased confidence in recent seismic-imaging enhancements has since corroborated these results. Opportunity The reservoir is the deepest producing reservoir in the field. It consists of two distinct and vertically separated zones. Southwest of the historical production from Reservoir 1, existing fault interpretation from seismic imaging placed three faults (Faults 1, 2, and 3), creating several compartments within the area (Fig. 1). An infill drilling program was progressed to drain the resources from these compartments. Well 1 was completed only within Zone 1 (upper zone). As shown in Fig. 1, Well 2 was a new drill between the interpreted Faults 2 and 3. It was completed as an intelligent well perforated along both zones of the reservoir. The length and transmissibility of all three faults, as well as the reservoir heterogeneity within the area, were highly uncertain. Consequently, previous development plans placed a potential infill well between Fault Blocks 1 and 2 to drain the volume within the compartment bounded by Faults 1 and 2. However, reservoir simulation studies indicated that recoverable resources from this future infill well were highly dependent on the sealing nature of Faults 1 and 2.

A review of the applications of magnetic susceptibility measurements for improved reservoir characterization

In the past, the application of magnetic susceptibility of diamagnetic and paramagnetic minerals to the study of rock magnetism was believed to be limited, due to both of their susceptibility signals being small compared to those of minerals in other magnetic. However, recent works have proved the usefulness of diamagnetic and paramagnetic susceptibility for reservoir studies. This paper summarizes results from previous studies showing the application of magnetic susceptibility measurements for characterizing different reservoir properties. Studies on the magnetic susceptibility of crude oils from different oil field regions around the world have shown good correlations of the susceptibility values with crude oil densities as well as other physical and chemical properties. In other studies, the measurements of magnetic susceptibility at the low and high fields have also been applied on rock core samples collected from different types of oil and gas reservoirs such as clastic shoreface, carbonate, shale and oil sand reservoirs. The magnetic susceptibility measurements for core samples without damaging the core by using the probe magnetic technique is probably used as a rapid, reasonable screening method for the initial estimation of core samples. The susceptibility of rock samples does not only show good correspondence with other reservoir characterizing methods, such as downhole gamma ray, spontaneous potential logs and core permeability, but it also shows some advantages over the traditional ones. The results suggested that the measurements of magnetic susceptibility could be used as an independent and improved method for distinguishing crude oil from different types of reservoirs, identifying main lithologies and predicting the permeable zone of a reservoir as well as estimating clay mineral contents.

Reconstruction of the Diagenetic Environments of Tight Sandstone Reservoirs: A Case Study from the Tengger Formation in the Baiyinchagan Sag, Erlian Basin, Northern China

Abstract The Lower Cretaceous Tengger Formation located in the Baiyinchagan Sag of the Erlian Basin comprises mainly deeply buried tight sandstone. The identification of high-quality reservoirs in these thickly stacked and heterogeneous units requires a comprehensive understanding of the diagenetic environmental history of the rocks. This paper reports an integrated study involving thin-section petrography, scanning electron microscopy, X-ray diffraction, fluid-inclusion analysis, and vitrinite reflectance analysis of Tengger Formation sandstones with the aim of characterizing the diagenetic conditions of the reservoir rocks and providing guidance for future petroleum exploration. Observed mineral assemblages, the distribution of authigenic minerals, and the distribution and nature of pores suggest the presence of two types of diagenetic environment, acidic and alkaline, which have varied over time and vertically through the rock column. Acidic conditions are indicated by quartz overgrowths and dissolution of both feldspar and carbonate cement. In contrast, alkaline conditions are indicated by the precipitation of carbonate cement, feldspar overgrowths, quartz dissolution, and occurrences of authigenic illite and chlorite. Changes in pore fluid chemistry controlled the evolution of the diagenetic environment. The early diagenetic environment from 110 Ma to 107 Ma was syndepositional and thus controlled by the chemistry of water in depositional centers, which is interpreted to have been weakly alkaline. Significant burial that occurred at 107 Ma induced pulses of hydrothermal fluids and petroleum into the reservoir rocks, which caused a shift to an acidic diagenetic environment. From 103 Ma to 70 Ma, subsequent episodes of uplift and burial caused periodic alternation between acidic and alkaline diagenetic environments. Three distinct episodes of oil and gas charging interpreted from petrography and the homogenization temperatures of fluid inclusions played a critical role in the enhancement of porosity through time. From 70 Ma to the present, acidic diagenesis gradually weakened because of the consumption of organic acids during the process of interaction between rocks and fluids. This study demonstrates the importance of understanding the diagenetic history of reservoir rocks and provides the basis for improved reservoir characterization and optimized hydrocarbon exploration of the Tengger Formation.

Geoscience Poster G9: Optimising value in the mature Turrum Field: integrating modern seismic, high-resolution sequence stratigraphy and production data in a 3D geological model

Poster G9 The Turrum Field in the offshore Gippsland Basin has been a significant producing field since first development in 2004 and remains a keystone gas resource for the future as many of the Gippsland assets continue to deplete. The Turrum Field is a faulted anticlinal structure with multiple, stacked, Paleocene-aged fluvial to lagoonal reservoirs that stratigraphically sit below the overlying Marlin Field gas cap. This field was developed in two phases with drilling campaigns in 2004 and 2015 and includes a dedicated platform (Marlin B) for the higher pressure, higher CO2 gas. Leveraging the 2018 reprocessed Northern Margins 3D seismic survey with a modern broadband processing workflow and wider bandwidth and higher signal-to-noise ratio has led to a seismic image with greater resolution and allowed improved reservoir characterisation. The uplift in seismic data together with the colour blending capabilities in GeoTeric®, has led to the seismic identification of large-scale fluvial reservoir systems. These products have been integrated with a revised field-wide sequence stratigraphic framework, petrophysical analysis and incorporated into a full-field 3D geological model of the Turrum Field. The resultant geological model has improved our understanding of reservoir connectivity and will enable rapid integration of production and surveillance data to optimise gas deliverability. This will also provide increased confidence for future field depletion planning including the identification of potential bypassed gas opportunities prior to end of field life. To access the poster click the link on the right. To read the full paper click here

Synthetic well log modeling with light gradient boosting machine for Assam-Arakan Basin, India

Synthetic logs can help in the identification of petrophysical and geological properties of reservoirs, where the actual well logs are incomplete or absent. This study presents new and novel machine learning (ML) algorithms to generate pseudo well logs using the readily available conventional wireline logs. The machine learning algorithm works well with conventional wireline logs because they can deduce implicit relationships and efficiently detect trends in the logs. Moreover, identifying ups and downs (or peaks and valleys) of the subsurface characteristics is essential for any log. In this work, we have elucidated the application of Light Gradient Boosting Machine (LightGBM) and compared its performance with Category Boosting (CatBoost) and Random Forest (RForest). Grid search cross-validation (GridSearchCV) was done to optimize the hyperparameters of the algorithms towards a better balancing between accuracy and speed. To check the robustness of the algorithms, a number of input and output logs, considering variation in trends of logs with depth, were chosen to train the network. The well logs from Assam-Arakan Basin (India) include neutron porosity (NPHI) log, photoelectric (PE) log, caliper (CALI) log, density correction (DRHO) log, gamma-ray (GR) log, and bulk density (RHOB) log along with depth. We have evaluated the well log models and machine learning algorithms using determination coefficient (R2), mean square error (MSE), and root mean square error (RMSE) for training, testing, and validation. It has been shown that the use of machine-learning algorithms to generate synthetic logs improves reservoir characterization since they are computationally efficient, accurate, and cost-effective. The results showed that the LightGBM outperformed all the other algorithms in terms of speed and accuracy, in addition to its simplified hyperparameter tuning.

Optimising value in the mature Turrum Field: integrating modern seismic, high-resolution sequence stratigraphy and production data in a 3D geological model

The Turrum Field in the offshore Gippsland Basin has been a significant producing field since first development in 2004 and remains a keystone gas resource for the future as many of the Gippsland assets continue to deplete. The Turrum Field is a faulted anticlinal structure with multiple, stacked, Paleocene-aged fluvial to lagoonal reservoirs that stratigraphically sit below the overlying Marlin Field gas cap. This field was developed in two phases with drilling campaigns in 2004 and 2015 and includes a dedicated platform (Marlin B) for the higher pressure, higher CO2 gas. Leveraging the 2018 reprocessed Northern Margins 3D seismic survey with a modern broadband processing workflow and wider bandwidth and higher signal-to-noise ratio has led to a seismic image with greater resolution and allowed improved reservoir characterisation. The uplift in seismic data together with the colour blending capabilities in GeoTeric®, has led to the seismic identification of large-scale fluvial reservoir systems. These products have been integrated with a revised field-wide sequence stratigraphic framework, petrophysical analysis and incorporated into a full-field 3D geological model of the Turrum Field. The resultant geological model has improved our understanding of reservoir connectivity and will enable rapid integration of production and surveillance data to optimise gas deliverability. This will also provide increased confidence for future field depletion planning including the identification of potential bypassed gas opportunities prior to end of field life.

Improving reservoir characterization and time-lapse seismic through joint inversion of PP- and PS-wave seismic data

Using synthetic and field data examples, we find that joint prestack amplitude variation with angle (AVA) inversion of PP- and PS-wave data can significantly improve estimation of P-impedance, S-impedance, and density. For reservoir characterization, improvements in these parameters can better identify reservoir rock and fluid properties. For reservoir monitoring, time-lapse (4D) changes in P-impedance, S-impedance, and density can lead to inversion of saturation and pressure changes. We see that in the joint inversion, 4D S-impedance is better estimated and not coupled to 4D P-impedance. These claims are first demonstrated on synthetic data, and then shown on an onshore unconventional play from Colorado and offshore 4-D-4C data set from the North Sea. Joint inversion of PP- and PS-wave data requires a higher level of care compared with PP-waves because the two wave modes need to be acquired, processed, and merged properly. This has diminished the use of converted waves in the past. However, modern acquisition and processing on land and offshore data make this technology quantitatively more accurate and realizable. As such, we also provide best practices for a successful project. We indicate that joint inversion can lead to a larger chance of success in placing exploration and development wells.

Enhancing Seismic Data Interpretation through Unsupervised Machine Learning and Data Analytics for Improved Reservoir Characterization

Enhancing Seismic Data Interpretation through Unsupervised Machine Learning and Data Analytics for Improved Reservoir Characterization

Analytical Work Flows Enable Continuous Waterflooding Optimization for a Mature Field

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper SPE 198586, “A New Continuous Waterflood Operations Optimization for a Mature Oil Field by Use of Analytical Work Flows That Improve Reservoir Characterization,” by Atul Yadav and Anton Malkov, SPE, Wintershall, and Essam Omara, Suez Oil Company, et al., prepared for the 2019 SPE Gas and Oil Technology Showcase and Conference, Dubai, 21–23 October. The paper has not been peer reviewed. In the complete paper, the authors present a novel approach that uses data-mining techniques on operations data of a complex mature oil field in the Gulf of Suez that is currently being waterflooded. Evidence is presented about how salinity data can be used to further justify the linkages between different wells obtained from cross-correlation analysis. The results presented in this research can be adapted to any waterflooded field to optimize recovery at frequent intervals where injection and production data are available continuously. Introduction Mature oil fields typically present challenges of increased water production and water handling. Considering the geological complexity and associated field-performance behavior, reservoir characterization to optimize water flooding is a major challenge. An integrated reservoir study was con ducted to minimize reservoir uncertainties and increase understanding of the field’s performance behavior. The acceptable history-matched model was used to estimate remaining oil potential, maintain and increase current production levels, and optimize the water-injection rate. Generally, history-matched models need to be updated throughout the life of producing fields as new subsurface data are acquired. Such integrated reservoir modeling studies, however, can be time-consuming and do not necessarily enable quicker decision-making around operational activities. The continuous recording of production and injection data presents new opportunities to apply novel analytical techniques to understand interwell connectivity in the reservoir. The current ability to store and analyze data, coupled with advances in the ability to interpret big data sets, has helped create an independent toolkit that provides analysis without the geological model. In addition, geological information such as pre-existing faults and the commingled or disconnected nature of production between different layers can be integrated to obtain and improve analyses from the analytical models. The authors analyze the results using Pearson’s cross-correlation analysis measure to obtain a qualitative analysis of the field. They also apply Spearman’s rank correlation analysis for the discussed field (henceforth named GOS for purposes of this paper) that helps compare injection and production data. The objective is to present a comparison between the analytical and the stream-lined approach to show consistency in reservoir characterization. The effective injector/producer pairs identified form an important component of the field development.

Theory-guided data science-based reservoir prediction of a North Sea oil field

Data science-based methods, such as supervised neural networks, provide powerful techniques to predict reservoir properties from seismic and well data without the aid of a theoretical model. In these supervised learning approaches, the seismic-to-rock property relationship is learned from the data. One of the major factors limiting the success of these methods is whether there exists enough labeled data, sampled over the expected geology, to train the neural network adequately. To overcome these issues, this paper explores hybrid theory-guided data science (TGDS)-based methods. In particular, we build a two-component model in which the outputs of the theory-based component are the inputs in the data science component. First, we simulate many pseudowells based on the well statistics in the project area. The reservoir properties, such as porosity, saturation, mineralogy, and thickness, are all varied to create a well-sampled data set. Elastic and synthetic seismic data are then generated using rock physics and seismic theory. The resulting collection of pseudowell logs and synthetic seismic data, called the synthetic catalog, is used to train the neural network. The derived operator is then applied to the real seismic data to predict reservoir properties throughout the seismic volume. This TGDS method is applied to a North Sea data set to characterize a Paleocene oil sand reservoir. The TGDS results better characterize the geology and well control, including a blind well, compared to a solely theory-based approach (deterministic inversion) and a data science-based approach (neural network using only the original wells). These results suggest that theory and data science can complement each other to improve reservoir characterization predictions.