Integration of NMR and Conventional Logs for Vuggy Facies Classification in the Arbuckle Formation: A Machine Learning Approach

Abstract

Summary Diagenetic effects in carbonate rocks can enhance or occlude depositional pore space. Reliable identification of porosity-enhancing diagenetic features (e.g., vugs and fractures) is essential for petrophysical characterization of reservoir properties (e.g., porosity and permeability), construction of geological and reservoir models, reserve estimation, and production forecasting. Challenges remain in characterizing these diagenetic features from well logs as they are often mixed with variations in mineral and fluid concentrations. Herein, we explore a data-driven approach that is based on a comprehensive well log data set from the Arbuckle Formation in Kansas to classify vuggy facies in carbonate rocks. The available well log data include conventional logs (gamma ray (GRTC), resistivity (RT), neutron/density porosity (NPHI/RHOB), photoelectric factor (PE), and acoustic slowness) and nuclear magnetic resonance (NMR) transverse relaxation time (T2) logs. We parameterized the measured T2 distribution using a multimodal lognormal Gaussian density function and combined the resulting Gaussian parameters with conventional logs as inputs into three supervised machine learning (ML) algorithms; namely, support vector machine (SVM), random forest (RF), and artificial neural network (ANN). The facies labeling data used in this study were based on visual examination of vug sizes from core samples, which include five classes; namely, nonvuggy, pinpoint-size, centimeter-size, fist-size, and super-vuggy. In total, 80% of the data set was used as the training set, and a fivefold cross validation was used for hyperparameter tuning. We conducted a detailed comparison of the above three ML algorithms on the basis of different combinations of features. The highest classification accuracy achieved on the holdout testing set is 84% using SVM on a combination of conventional logs and selected NMR Gaussian parameters as inputs. In general, inclusion of conventional log data improves the prediction accuracy compared with using NMR data alone. Feature selection improves the performance for SVM and ANN but is not recommended for RF.

Carbonate reservoir characterization with integrated rock, special well logs, and machine learning methods

“…Despite some success, these models may not perform well in complex porous systems. Integrating special logs and conventional logs in machine learning models have been tried in other geological settings (Hamada & Elshafei, 2009;Zhu et al, 2016;Xu et al 2020). Therefore, a literature gap is identified, where machine learning models using inputs from both conventional and special logs are tested to improve porosity estimation in the Brazilian Pre-Salt carbonates.…”

“…The research on artificial intelligence and machine learning grew steadily in the early 2000s, with the popularization of broadband internet and increasingly cheaper computational cost. A major crossover point in the petroleum industry is pinpointed around the year 2011 (Xu et al 2020), from which the growth of publications on machine learning and artificial intelligence has become close to exponential. Several important factors contributed to this phenomenon, including the development of specialized open-source libraries (e. g. scikit learn, pytorch, keras, tensorflow), availability of hardware to run experiments on parallel computing, advancement of deep-learning methods, publication of complete datasets by major oil companies and governments, among others.…”

“…An innate difficulty when integrating special and conventional logs is usually related to their data structure. Converting the T2 transverse relaxation time waveform into discrete values allows for different statistical approaches (Hamada & Elshafei, 2009;Xu et al, 2020). The purpose of this study is to reduce porosity estimation uncertainty, especially related to secondary porosity, without attempting to generate a pore-size distribution using the T2 waveform.…”

“…Moreover, recent advances in data-science and machine learning, which includes the development of sophisticated algorithms, faster computer processing, the popularizing of open-source software, and specialized libraries, led to its growing adoption within the geosciences and petroleum engineering communities (Xu et al, 2020, Dramsch, 2020.…”