ANS is committed to advancing, fostering, and promoting the development and application of nuclear sciences and technologies to benefit society.
Explore the many uses for nuclear science and its impact on energy, the environment, healthcare, food, and more.
Explore membership for yourself or for your organization.
Conference Spotlight
2026 ANS Annual Conference
May 31–June 3, 2026
Denver, CO|Sheraton Denver
Latest Magazine Issues
Mar 2026
Jan 2026
Latest Journal Issues
Nuclear Science and Engineering
April 2026
Nuclear Technology
February 2026
Fusion Science and Technology
Latest News
Going Nuclear: Notes from the officially unofficial book tour
I work in the analytical labs at one of Europe’s oldest and largest nuclear sites: Sellafield, in northwestern England. I spend my days at the fume hood front, pipette in one hand and radiation probe in the other (and dosimeter pinned to my chest, of course). Outside the lab, I have a second job: I moonlight as a writer and public speaker. My new popular science book—Going Nuclear: How the Atom Will Save the World—came out last summer, and it feels like my life has been running at full power ever since.
Quinton J. Williams, Ryan H. Stewart, Todd S. Palmer, Camille J. Palmer, Chad Pope, Ashley Shields, Christopher Ritter
Nuclear Science and Engineering | Volume 200 | Number 1 | March 2026 | Pages S391-S405
Research Article | doi.org/10.1080/00295639.2025.2455908
Articles are hosted by Taylor and Francis Online.
Nuclear reactor digital twins (DTs) have been proposed for use as a safeguards technology to efficiently monitor new and novel reactors as they come online. A safeguards DT needs to be capable of detecting misuse and diversion as they occur, requiring physics models to be accurate and efficient. Mathematical surrogate models are capable of achieving the necessary efficiency and can largely maintain the accuracy of higher-order models given a quality training sample. The Multiphysics Object-Oriented Simulation Environment (MOOSE) code framework is specifically equipped to generate training samples and create surrogate models using full-order reactor physics models. Utilizing an operational AGN-201M reactor’s specifications, two surrogate types were trained on samples of variable size, and using Cartesian products, Latin hypercube sampling, and quadrature sampling, each was compared and evaluated on accuracy when compared to a full-order Monte Carlo model. Both surrogate types were able to capture reactivity changes within 0.05 $ of the Monte Carlo model while reducing the computation costs by eight orders of magnitude.