AI and digital engineering could reduce deployment schedules by up to 21% (likely lowering the cost of advanced nuclear by ~20%), accelerate the commercial viability of hydrogen power at scale and significantly reduce the risks of deploying clean power generation. INL will use DeepLynx for the digital thread, Multiphysics Object-Oriented Simulation Environment (MOOSE) for fast response surrogates, the Framework for Optimization of Resources and Economics (FORCE) for economics, automatic building of computer-aided design (CAD) meshing tools for finite element analysis integration, ALCHEMY for autonomous CAD, and RAVEN for statistical analysis and model optimization. The semiautonomous design will use new multimodal AI technologies fused with generative, surrogate, nonlinear optimization and other models to achieve higher performance. This grand challenge will collaborate closely with large fission and fusion power programs to verify this technology on real-world projects.
Near-autonomous and remote operation of advanced microreactors demonstrations are in the works at INL, representing an essential economic milestone for the nuclear industry. These operating methods are positioned to demonstrate the world’s first autonomous microreactor by building on past successes (i.e., the first nuclear reactor digital twin [AGN-201], the Microreactor Agile Non-nuclear Experimental Testbed [MAGNET], and the Nearly Autonomous Management and Control System framework for advanced reactors) and by exploring both physics-informed and data-driven AI approaches. We will address this challenge using INL tools such as Optimization for Real-time Capacity Allocation (ORCA) and DeepLynx. INL will also collaborate with space and defense fission programs to realize a demonstration of this technology.
INL is investing in material and fuel irradiation capabilities at its experiment testing facilities to enable breakthroughs in the development and qualifications of advanced fuels and materials for nuclear systems. The lab will design an autonomous lab platform to accelerate irradiation testing, reduce the barrier to entry and increase throughput at this critical juncture in next-generation nuclear reactor deployment. This challenge necessitates a holistic end-to-end integration of nuclear facilities, including a test reactor, high-performance computing, digital engineering and twinning frameworks, predictive massively parallel high-fidelity modeling and simulation, and big data collection throughout the experiment process.
Applied to cyber and disaster resilience, AI will inform real-time responses to cyber and physical threats, influence resilient critical infrastructure designs, and mitigate long-term risks to cyber and physical assets. As industry moves to a more integrated environment with millions of embedded control devices, the use of AI/Ml to distinguish physical and cyber threats from operational anomalies will be critical for response, engagement and building trust in this new environment. INL is developing AI/ML capabilities to extract threat intelligence from real-time cyber monitoring solutions by optimizing integrated resource planning to ensure a more resilient future energy grid and authoring multiple datasets in critical infrastructure protection that will enable enduring AI-assisted analyses. Key to these analyses are appropriate knowledge exchange and risk analysis along with imagery and multimodal, multivariate dataset cohesion. Future efforts will center on maturing knowledge models and developing high-fidelity, AI-driven modeling and simulation capabilities that enhance real-time threat detection and identification (including from the AI itself), disaster mitigation and recovery, cyber and disaster resilience, and omnidirectional risk assessments and impact analyses.