[H-SC06] 地球温暖化防⽌と地学（CO2地中貯留・有効利⽤、地球⼯学）

[HSC06-11] Development of a permeability reduction model using deep learning for CO₂ hydrate storage

*Alan Junji Yamaguchi¹、Toru Sato¹、Takaomi Tobase²、Xinran Wei³、Lin Huang³、Jia Zhang³、Jiang Bian³、Tie-Yan Liu³ (1.The University of Tokyo Graduate School of Frontier Sciences Department of Ocean Technology, Policy, and Environment、2.Chigasaki Research Institute, Electric Power Development Co., Ltd.、3.Microsoft Research Asia)

キーワード：Carbon capture and storage, CO2 hydrate, effective permeability coefficient, machine learning

Global warming is an important environmental concern and carbon capture and storage (CCS) emerges as a very promising technology. Captured carbon dioxide (CO₂) can be stored in aquifers onshore or offshore seabed regions. There is, however, a small risk that stored CO₂ will leak due to natural disasters. One possible solution to this is the natural formation of CO₂ hydrates under high pressure and low temperature conditions. Its stability under these conditions acts as a cap rock to prevent CO₂ leaks. It is of utmost necessity to understand and evaluate the permeability change due to the hydrate formation. Numerical simulation on different spatial scales has been very important for this purpose. The main objective of this study is to create a novel framework for permeability reduction due to CO₂ hydrate formation. A multiscale approach is considered to link a large reservoir scale hydrate formation model with a microscale model by using machine learning. Detailed information from the hydrate shape can be obtained from the microscopic range, and be used to predict the new permeability reduction coefficient. Initial results have shown that this approach can be considered to obtain the permeability change due to CO₂ hydrate formation with good accuracy.