REACTWIN: A Digital Twin Framework for Fault Diagnosis and Resilience Monitoring in Nuclear Reactor Systems

Abstract

Nuclear power plants supply approximately 20% of U.S. electricity and nearly 50% of domestic carbon-free generation, making reactor component health a matter of national energy security. Existing condition monitoring methods rely on scheduled offline inspections and SCADA-based threshold alarms that are temporally sparse, contextually blind, and fundamentally reactive. This paper proposes REACTWIN, a conceptual digital twin framework for real-time fault diagnosis and resilience monitoring of nuclear reactor component systems. The framework integrates physics-based multi-physics reactor models, IoT-enabled sensor data pipelines, and ensemble machine learning classifiers within a continuously synchronized virtual-physical environment. Five critical fault modes are formally characterized and mapped to corresponding sensor modalities and physics simulation modules. The proposed architecture is evaluated against existing approaches through a structured design principle analysis grounded in published nuclear engineering standards and prior digital twin literature. REACTWIN addresses a clear methodological gap in the reactor monitoring literature and establishes a deployable pathway aligned with U.S. Department of Energy Light Water Reactor Sustainability program objectives and NRC regulatory requirements.