Ensemble Learning for Predicting TOC from Well-Logs of the Unconventional Goldwyer Shale

Precise estimation of total organic carbon (TOC) is extremely important for the successful characterization of an unconventional shale reservoir. Indirect traditional continuous TOC prediction methods from well-logs fail to provide accurate TOC in complex and heterogeneous shale reservoirs. A workflow is proposed to predict a continuous TOC profile from well-logs through various ensemble learning regression models in the Goldwyer shale formation of the Canning Basin, WA. A total of 283 TOC data points from ten wells is available from the Rock-Eval analysis of the core specimen where each sample point contains three to five petrophysical logs. The core TOC varies largely, ranging from 0.16 wt % to 4.47 wt % with an average of 1.20 wt %. In addition to the conventional MLR method, four supervised machine learning methods, i.e., ANN, RF, SVM, and GB are trained, validated, and tested for continuous TOC prediction using the ensemble learning approach. To ensure robust TOC prediction, an aggregated model predictor is designed by combining the four ensemble-based models. The model achieved estimation accuracy with R2 value of 87%. Careful data preparation and feature selection, reconstruction of corrupted or missing logs, and the ensemble learning implementation and optimization have improved TOC prediction accuracy significantly compared to a single model approach.

Petrophysical Modelling in Absence of Core Data in a Small, Marginal Field in Mumbai Offshore Basin, India

The best petrophysical models are based on direct measurements from the core. Unfortunately, core is not available in many cases, either for economic, logistical, or historical reasons. In this study, we needed to construct a detailed Field Development Plan (FDP) for the small, marginal B-9 field in the western offshore basin, India that did not merit core acquisition. The objective is to propose a workflow for building a petrophysical model with limited datasets instead of a typical FDP workflow. After analysing the assumptions, limitations, and uncertainties involved in the petrophysical model, we used advanced petrophysical logs to reduce uncertainty and create a robust petrophysical model. We carried out a log-based petrophysical study to determine the volume of shale, porosity, saturation, and permeability. The advanced petrophysical logs like spectroscopy, nuclear magnetic resonance (NMR), formation pressures, and well testing data are utilised to calibrate the petrophysical model. Spectroscopy data is used to calibrate mineralogical volumes and grain density, while porosity is calibrated from NMR data. We calibrated log-derived permeability results with NMR permeability and mobility from well test data. We used heterogeneous rock analysis (HRA) on petrophysical outputs to carry out Petrophysical Rock Typing (PRT). This has helped in establishing the porosity-permeability relationship and saturation-height model for each PRT. In absence of irreducible water saturation ( S wirr) information from the core, NMR-derived S wirr is calculated and then utlised to calibrate the saturation model. Log-derived permeability and saturation are estimated which agrees well with the available testing data. This provided a robust petrophysical model that served as a basis for geological static and reservoir dynamic models. The gas-down-to and water-up-to methods are employed to establish the contacts. The resulting saturation height model agreed well with the saturations derived from the log, which gave us confidence in our dynamic model.

Data mining based investigation of the impact of imbalanced dataset over fractured zone detection

Several studies have been conducted in recent years to discriminate between fractured (FZs) and non-fractured zones (NFZs) in oil wells. These studies have applied data mining techniques to petrophysical logs (PLs) with generally valuable results; however, identifying fractured and non-fractured zones is difficult because imbalanced data is not treated as balanced data during analysis. We studied the importance of using balanced data to detect fractured zones using PLs. We used Random-Forest and Support Vector Machine classifiers on eight oil wells drilled into a fractured carbonite reservoir to study PLs with imbalanced and balanced datasets, then validated our results with image logs. A significant difference between accuracy and precision indicates imbalanced data with fractured zones categorized as the minor class. The results indicated that the accuracy of imbalanced and balanced datasets is similar, but precision is significantly improved by balancing, regardless of how low or high the calculated indices might be.

AUTO-NAVIGATION ON PRESSURE AND SAMPLING LOCATION IN WIRELINE AND LWD: BIG DATA CHALLENGES AND SOLUTIONS

Measuring formation pressure and collecting representative samples are the essential tasks of formation testing operations. Where, when and how to measure pressure or collect samples are critical questions which must be addressed in order to complete any job successfully. Formation testing data has a crucial role in reserve estimation especially at the stage of field exploration and appraisal, but can be time consuming and expensive. Optimum location has a major impact on both the time spent performing and the success of pressure testing and sampling. Success and optimization of rig-time paradoxically requires careful and extensive but also quick pre-job planning. The current practice of finding optimum locations for testing heavily rely on expert knowledge. With nearly complete digitization of data collection, the oil industry is now dealing with massive data flow giving rise to the question of its application and the necessity to collect. Some data may be so called “dark data” of which a very tiny portion is used for decision making. For instance, a variety of petrophysical logs may be collected in a single well to provide measures of formation properties. The logs may include conventional gamma ray, neutron, density, caliper, resistivity or more advanced tools such as high-resolution image logs, acoustic, or NMR. These data can be integrated to help decide where to pressure test and sample, however, this effort is nearly exclusively driven by experts and is manpower intensive. In this paper we present a workflow to gather, process and analyze conventional log data in order to optimize formation testing operations. The data is from an enormous geographic distribution of wells. Tremendous effort has been performed to extract, transform and load (ETL) the data into a usable format. Stored files contains multi-million to multi-billions rows of data thereby creating technology challenges in terms of reading, processing and analyzing in a timely manner for pre-job planning. We address the technological challenges by deploying cutting-edge data technology to solve this problem. Upon completion of the workflow we have been able to build a scalable petrophysical interpretation log platform which can be easily utilized for machine learning and application deployment. This type of data base is invaluable asset especially in places where there is a need for knowledge of analogous wells. Exploratory data analysis on worldwide data on mobility and some key influencing features on pressure test and sampling quality, is performed and presented. We further show how this data is integrated and analyzed in order to automate selection of locations for which to formation test.

RESERVOIR SCALE CHEMOSTRATIGRAPHY AND FACIES MODELING USING HIGH SAMPLE RATE GEOPHYSICAL SCANS OF WHOLE CORE

Many unconventional reservoirs exhibit a high level of vertical heterogeneity in terms of petrophysical and geo-mechanical properties. These properties often change on the scale of centimeters across rock types or bedding, and thus cannot be accurately measured by low-resolution petrophysical logs. Nonetheless, the distribution of these properties within a flow unit can significantly impact targeting, stimulation and production. In unconventional resource plays such as the Austin Chalk and Eagle Ford shale in south Texas, ash layers are the primary source of vertical heterogeneity throughout the reservoir. The ash layers tend to vary considerably in distribution, thickness and composition, but generally have the potential to significantly impact the economic recovery of hydrocarbons by closure of hydraulic fracture conduits via viscous creep and pinch-off. The identification and characterization of ash layers can be a time-consuming process that leads to wide variations in the interpretations that are made with regard to their presence and potential impact. We seek to use machine learning (ML) techniques to facilitate rapid and more consistent identification of ash layers and other pertinent geologic lithofacies. This paper involves high-resolution laboratory measurements of geophysical properties over whole core and analysis of such data using machine-learning techniques to build novel high-resolution facies models that can be used to make statistically meaningful predictions of facies characteristics in proximally remote wells where core or other physical is not available. Multiple core wells in the Austin Chalk/Eagle Ford shale play in Dimmitt County, Texas, USA were evaluated. Drill core was scanned at high sample rates (1 mm to 1 inch) using specialized equipment to acquire continuous high resolution petrophysical logs and the general modeling workflow involved pre-processing of high frequency sample rate data and classification training using feature selection and hyperparameter estimation. Evaluation of the resulting training classifiers using Receiver Operating Characteristics (ROC) determined that the blind test ROC result for ash layers was lower than those of the better constrained carbonate and high organic mudstone/wackestone data sets. From this it can be concluded that additional consideration must be given to the set of variables that govern the petrophysical and mechanical properties of ash layers prior to developing it as a classifier. Variability among ash layers is controlled by geologic factors that essentially change their compositional makeup, and consequently, their fundamental rock properties. As such, some proportion of them are likely to be misidentified as high clay mudstone/wackestone classifiers. Further refinement of such ash layer compositional variables is expected to improve ROC results for ash layers significantly.

ADDRESSING RESERVOIR HETEROGENEITY BY INTEGRATION OF GEOCHEMISTRY AND PETROPHYSICAL LOGS IN CARBONATE PROSPECTS

Formation evaluation challenges in highly fractured, stacked reservoirs with multiple source rocks and structural complexities that have complicated charging histories are common in the Middle East. Finding additional pay zones, understanding the contribution of individual oils to the overall production, or evaluating the compartmentalization within the reservoir by resolving the heterogeneity of the reservoir rocks are to name but a few. This work tries to understand the challenges posed by the subsurface complexities and attempts to find answers through physical evidence, using both onsite data acquired during drilling and data gathered through organic and inorganic laboratory measurements. Formation evaluation challenges are mostly attributed to formation heterogeneity, which we have aimed to address through the integration of petrophysical and geochemical data within this work. This project encompasses the integration of petrophysical and geochemical analyses of the reservoir rocks. Geochemical data have provided the ability to make maturity, richness, and other character interpretations and will be combined with important petrophysical properties of the carbonate intervals to predict reservoir heterogeneities. These interpretations could support perforation interval selection on subsequent wells in the field through the understanding of the mobility of the oils and, ultimately, production allocation. Best practices for thermally extracting hydrocarbons from drill cuttings, quality-controlling advanced mud gas data, and interpretive processes together with the entire workflow followed will also be elaborated. The analysis has the objectives of establishing results to support completion decisions through understanding reservoir quality, reservoir fluid communication, and compartmentalization specific to the basin studied. The petrophysical reservoir properties such as hydrocarbons in place, mobility of the oils, porosity, permeability, fracture intensity, geomechanical properties (brittle vs. ductile), and production allocation will be tied in to geochemical analyses to this extent. The focal point of the work is ascertaining and characterizing both the reservoir properties using a number of integrated analytical techniques on DST oil samples of 12 offset wells and rock cuttings, as well as petrophysical logs and advanced mud gas data. The concepts, tools, and methods that have been demonstrated for evaluating crude oils, natural gases, and petrophysical characteristics of the rocks are applicable to many problems in petroleum production and field development as well as exploration efforts.

Analysis of permeability based on petrophysical logs: comparison between heuristic numerical and analytical methods

A series of novel heuristic numerical tools were adopted to tackle the setback of permeability estimation in carbonate reservoirs compared to the classical methods. To that end, a comprehensive data set of petrophysical data including core and log in two wells was situated in Marun Oil Field. Both wells, Well#1 and Well#2, were completed in the Bangestan reservoir, having a broad diversity of carbonate facies. In the light of high Lorenz coefficients, 0.762 and 0.75 in Well#1 and Well#2, respectively, an extensive heterogeneity has been expected in reservoir properties, namely permeability. Despite Well#1, Well#2 was used as a blinded well, which had no influence on model learning and just contributed to assess the validation of the proposed model. An HFU model with the aim of discerning the sophistication of permeability and net porosity interrelation has been developed in the framework of Amaefule’s technique which has been modified by newly introduced classification and clustering conceptions. Eventually, seven distinct pore geometrical units have been distinguished through implementing the hybridized genetic algorithm and k-means algorithm. Furthermore, a K-nearest neighbors (KNN) algorithm has been carried out to divide log data into the flow units and assigns them to the pre-identified FZI values. Besides, a cross between the ε-SVR model, a supervised learning machine, and the Harmony Search algorithm has been used to estimate directly permeability. To select the optimum combination of the involved logging parameters in the ε-SVR model and reduce the dimensionality problem, a principle component analysis (PCA) has been implemented on Well#1 data set. The result of PCA illustrates parameters, such as permeability, the transit time of sonic wave, resistivity of the unflashed zone, neutron porosity, photoelectric index, spectral gamma-ray, and bulk density, which possess the highest correlation coefficient with first derived PC. In line with previous studies, the findings will be compared with empirical methods, Coates–Dumanior, and Timur methods, which both have been launched into these wells. Overall, it is obvious to conclude that the ε -SVR model is undeniably the superior method with the lowest mean square error, nearly 4.91, and the highest R-squared of approximately 0.721. On the contrary, the transform relationship of porosity and permeability has remarkably the worst results in comparison with other models in error (MSE) and accuracy (R2) of 128.73 and 0.116, respectively.