Physics-Based Digital Twin & AI

Research focus

High-fidelity multiphysics simulation is valuable for understanding complex reactor behavior in detail, but its computational cost limits direct use in real-time or repeated prediction. Continuous prediction of nuclear-system states for operation and safety decisions requires a computational framework that can use physics-based analysis results much more rapidly.

MONAMI Lab builds reduced-order, surrogate, and AI-based predictive models on multiphysics and high-fidelity simulation and connects them with experimental and operational data to predict reactor states and behavior efficiently. High-fidelity results can be used to calibrate and train lower-dimensional models, while models of different fidelity are linked to balance computational accuracy and speed.

We aim to extend conventional offline analysis into a physics-based nuclear digital twin capable of state estimation, anomaly prediction, and near-real-time simulation, ultimately establishing an intelligent analysis environment for reactor design, operation, and safety assessment.