Overview

• In Progress in Nuclear Energy, the Michigan-led team reports a model predictive control approach that autonomously adjusts nuclear microreactor output.
• They modeled a High-Temperature Gas-Cooled microreactor based on a simplified Holos-Quad (Gen 2+) design and manipulated control drums to vary power.
• Integrated with the PROTEUS high-fidelity reactor physics tool, the controller tracked 20%‑per‑minute ramps within 0.234% of target in simulation.
• Extensive sensitivity testing showed robust performance across varied inputs, and the method intentionally avoids AI to remain fully explainable for regulators.
• The work, funded by the U.S. DOE Office of Nuclear Energy, points to applications in remote, military, maritime, and grid settings, with experimental demonstrations and licensing still ahead.