--> Shale Lithofacies Classification and Modeling: Case Studies From the Bakken and Marcellus Formations, North America

AAPG Annual Convention and Exhibition

Datapages, Inc.Print this page

Shale Lithofacies Classification and Modeling: Case Studies From the Bakken and Marcellus Formations, North America

Abstract

Lithofacies classification, assigning a rock type to specific rock samples on the basis of petrography or measured physical properties, is fundamental to subsurface investigations. Clastic and carbonate lithofacies have been studied extensively for depositional and diagenetic environment studies. However, research in black shale lithofacies is relatively rare, most being based on either single well study or descriptive analysis. To have broad applications, shale lithofacies should be meaningful, mappable and predictable at core, well and regional scales, in terms of maturity, mineral composition and Total Organic Carbon (TOC). The utility of different petrophysical approaches to shale lithofacies classification and prediction are demonstrated with examples from the prolific Bakken and Marcellus shale oil and/or gas resources. Core data (XRD, TOC), basic and advanced logs (such as Pulsed Neutron Spectroscopy, Dipole Sonic and Spectral Gamma) are used to investigate the petrophysical and geomechanical characteristics of shale. Core parameters calibrated with advanced logs as well as a series of multi-mineral and crossplot solutions were used to define six different shale lithofacies units. Facies pattern recognition and rock typing from basic logs (such as gamma, resistivity, porosity and photo-electric) used techniques such as Artificial Neural Network, Support Vector Machine and Self-Organizing Map trained on a foundation of core data and advanced logs. After classification and prediction of shale lithofacies in all wells, including uncored wells and wells without advanced logs geostatistical approaches such as Sequential Indicator Simulation were applied to generate 3D static geocellular models for each play. The stochastic facies models were used for detailed geological interpretation of each shale lithofacies and compared with production data for integrated reservoir characterization. The study shows that mineralogy (especially, presence of biogenic silica), kerogen type and thickness of different shale units contribute to hydrocarbon production for both plays.