Digital Twins of Nuclear Power Plants

A consortium of the University of Michigan, Idaho National Laboratory, Argonne National Laboratory, Kairos Power and Curtiss-Wright is developing digital twins for nuclear power plants with a $5.2 million grant from the Department of Energy’s ARPA-E. The first digital twins are applied to Kairos Power’s fluoride-salt-cooled high-temperature reactor (FHR).

The approach combines machine-learning algorithms with high-fidelity physics models, avoiding the “black box” risk and keeping predictive robustness for the preventive maintenance of critical components. The system is designed to support both operating plants and advanced reactors still in design, cutting time and cost through upfront simulation that steers targeted experimental testing. The project also includes a modular software package for predictive maintenance, operational control and supervision, meant to be adaptable by other nuclear vendors.

For the industrial client this technology is a potential tool for optimising maintenance intervals and supporting load-follow operation, increasingly required where intermittent renewables have high penetration. The ASME page carries no direct reference to standards or certification paths, but the subject stays relevant for clients following how the rules on AI and digital twins in nuclear evolve.