[2C09-14] AI

3:00 PM - 3:15 PM JST(6:00 AM - 6:15 AM UTC)

[2C10] Machine Learning-Based Prediction of Reaction-Induced Permeability Changes in Fracture Networks

*Ippei Chikura¹, Shunichi Okamoto^2,1, Takumi Saito¹ (1. University of Tokyo Graduate School of Engineering, 2. Central Research Institute of Electric Power Industry)

Keywords:

Fracture network,Machin learning,Permeability changes

Safety assessment of geological disposal of high-level radioactive wastes requires groundwater flow and radionuclide transport analyses under numerous scenarios in fractured rock systems. Permeability evolution caused by geochemical reactions strongly influences groundwater flow and radionuclide migration. However, reactive transport modeling (RTM) is computationally expensive because it simultaneously solves flow, solute transport, and chemical reactions. In addition, the controlling factors of permeability evolution in fractured media remain poorly understood due to the complex interplay among flow, mass transfer, and geochemical reactions. This study develops machine-learning-based surrogate models for permeability prediction using RTM results obtained from discrete fracture network (DFN) simulations. The proposed approach aims to accelerate reactive transport simulation and identify key factors governing permeability evolution in fractured rock systems.