Assessment of CO2 Storage as Hydrates in Saline Aquifers Using Machine Learning Algorithms

Abstract

Global warming is one of the most serious issues the world is currently facing. The major reason is attributed to emission of greenhouse gases and in particular carbon dioxide, CO₂. The most promising methods that could allow significant reduction in CO₂ emissions are capture and geological storage of CO₂. One major concern against storage of CO₂ is the possibility of its leakage. One process that could lead to more reliable trapping of CO₂ is hydrate formation – that leads to trapping of CO₂ in the solid form. In this study, Machine Learning algorithms and reservoir simulation software were used to conduct sensitivity studies on some of the main reservoir parameters, to understand which characteristics have most impact on stability of CO₂ storage in the form of hydrates. The hydrate formation curve calculated by HydraFlash software was considered to be a benchmark for experiments conducted in this study.