Date of Graduation
2012
Document Type
Thesis
Abstract
The Marcellus Shale, marine organic-rich mudrock deposited during Middle Devonian in the Appalachian basin, is considered the largest unconventional shale-gas resource in United State. Although homogeneous in the appearance, the mudstone shows heterogeneity in mineral composition, organic matter richness, gas content, and fracture density. Two critical factors for unconventional mudstone reservoirs are units amenable to hydraulic fracture stimulation and rich of organic matter. The effectiveness of hydraulic fracture stimulation is influenced by rock geomechanical properties, which are related to rock mineralogy. The natural gas content in mudrock reservoirs has a strong relationship with organic matter, which is measured by total organic carbon (TOC). In place of using petrographic information and sedimentary structures, Marcellus Shale lithofacies were based on mineral composition and organic matter richness and were predicted by conventional logs to make the lithofacies 'meaningful’, ‘predictable’ and ‘mappable’ at multiple scales from the well bore to basin. Core X-ray diffraction (XRD) and TOC data was used to classify Marcellus Shale into seven lithofacies according to three criteria: clay volume, the ratio of quartz to carbonate, and TOC. Pulsed neutron spectroscopy (PNS) logs provide similar mineral concentration and TOC content, and were used to classify shale lithofacies by the same three criteria. Artificial neural network (ANN) with improvements (i.e., learning algorithms, performance function and topology design) was utilized to predict Marcellus Shale lithofacies in 707 wells with conventional logs. To improve the effectiveness of wireline logs to predict lithofacies, the effects of barite and pyrite were partly removed and eight petrophysical parameters commonly used for a conventional reservoir analysis were derived from conventional logs by petrophysical analysis. These parameters were used as input to the ANN analysis. Geostatistical analysis was used to develop the experimental variogram models and vertical proportion of each lithofacies. Indictor kriging, truncated Gaussian simulation (TGS), and sequential indicator simulation (SIS) were compared, and SIS algorithm performed well for modeling Marcellus Shale lithofacies in three-dimensions. Controlled primarily by sediment dilution, organic matter productivity, and organic matter preservation/decomposition, Marcellus Shale lithofacies distribution was dominantly affected by the water depth and the distance to shoreline. The Marcellus Shale lithofacies with the greatest organic content and highest measure of brittleness is concentrated along a crescent shape region paralleling the inferred shelf and shoreline, showing shape of crescent paralleling with shoreline. The normalized average gas production rate from horizontal wells supported the proposed approach to modeling Marcellus Shale lithofacies. The proposed 3-D modeling approach may be helpful for (1) investigating the distribution of each lithofacies at a basin-scale; (2) developing a better understanding of the factors controlling the deposition and preservation of organic matter and the depositional model of marine organic-rich mudrock; (3) identifying organic-rich units and areas and brittle units and areas in shale-gas reservoirs; (4) assisting in the design of horizontal drilling trajectories and location of stimulation activity; and (5) providing input parameters for the simulation of gas flow and production in mudrock (e.g., porosity, permeability and fractures).
Recommended Citation
Wang, Guochang, "Black shale lithofacies prediction and distribution Pattern analysis of middle Devonian Marcellus Shale in the Appalachian Basin, northeastern U.S.A." (2012). Graduate Theses, Dissertations, and Problem Reports. 9975.
https://researchrepository.wvu.edu/etd/9975