C01-05

10:30 AM - 10:45 AM JST(1:30 AM - 1:45 AM UTC)English

[C02] Multiphysics Deep Learning for Spatiotemporal Carbon Storage Prediction Under Geological Uncertainty

○Vo Thanh Hung¹ (1. Waseda University)

Keywords:

CCS,Deep learning,Geological uncertainty

This work presents a deep learning–accelerated framework for emulating strongly coupled multiphysics processes in subsurface storage systems, enabling rapid field-scale screening and digital-twin workflows. High-fidelity numerical models that resolve multiphase flow, geomechanical deformation, and reactive transport remain the benchmark for assessing storage performance and integrity, yet their computational cost constrains uncertainty quantification, real-time optimization, and continuous monitoring. This study addresses this limitation by developing a multi-output surrogate that approximates the response of a fully coupled multiphysics simulator for fluid injection in heterogeneous, layered geological formations, delivering fast predictions of key state variables relevant to plume evolution, pressure–stress response, and deformation risk.