Estimation of Adsorption Capacity of CO2, CH4, and their Binary Mixtures in Quidam Shale using LSSVM: Application in CO2 Enhanced Shale Gas Recovery and CO2 Storage

Abstract

Carbon dioxide enhanced shale gas recovery depends strongly on adsorption properties of carbon dioxide and methane. In this work, Least Squares Support Vector Machine (LSSVM) optimized by Particle Swarm Optimization, has been proposed to learn and then predict adsorption capacity of methane and carbon dioxide from pure and binary gas mixtures in Jurassic shale samples from the Qaidam Basin in China based on input parameters pressure, temperature, gas composition and TOC. A literature dataset of 348 points was applied to train and validate the model. The predicted values were compared with the experimental data by statistical and graphical approaches. The coefficients of determination of carbon dioxide adsorption were calculated to 0.9990 and 0.9982 for training and validation datasets, respectively. For CH4 the numbers are 0.9980 and 0.9966. The model was extrapolating reasonable trends beyond measurement ranges. More extensive datasets are needed to properly parameterize the role of shale properties.