Predicting gas content in high-maturity marine shales using artificial intelligence based seismic multiple-attributes analysis: A case study from the lower Silurian Longmaxi Formation, Sichuan Basin, China

Abstract

Abstract Gas content is the most important index used to evaluate shale gas reservoirs. Compared to the North American marine shales, the marine shales of China are older, characterized by high thermal maturity, and have experienced multiple stages of tectonism. There is no clear positive correlation between the gas content and TOC content for marine shales in China; therefore, it is not feasible to indirectly predict the gas content by predicting TOC content, as has been extensively done for the North American shales. In this study, the geophysical responses of reservoirs with high gas contents, the factors that influence the gas content have been determined based on well log interpretation and seismic rock physics analysis. The seismic attributes and elastic parameters that are sensitive to the gas content have also been investigated. Additionally, a pre-stack seismic inversion has been carried out through a combination of well log and 3D seismic data to obtain the data volumes of the seismic elastic parameters, which are highly sensitive to variations in gas content. A seismic multiple-attribute analysis based on neural networks of AI (artificial intelligence) technology has been applied to determine the type and the combination of elastic parameters for computing the gas content data volume. This kind of analysis has been carried out based on pre-stack inversion of seismic data. The results suggest that the lower Silurian Longmaxi Formation, a high-quality gas-rich reservoir, is characterized by its low density, low Poisson's ratio and low velocity. A combination of six parameters or attributes optimally predicts gas content. The studied block, W202, generally has a high gas content. Two classes of sweet spots have been predicted in the study area based on these results. The application of neural networks has provided a precise prediction for the spatial distribution of gas saturation in the shale reservoir, serving as a basis for locating and designing horizontal wells for shale gas development.