Artificial intelligence

How Artificial Intelligence Can Improve Advanced Nuclear Reactors

Researchers are looking for the ideal characteristics of molten salt, which can serve as both coolant and fuel in advanced nuclear reactors. Credit: Argonne National Laboratory

Technology developed at Argonne may help narrow the field of candidates for molten salts, according to a new study.

Scientists are looking for new materials to advance the next generation of nuclear power plants. In a recent study, researchers from the US Department of Energy’s (DOE) Argonne National Laboratory showed how artificial intelligence can help identify the right types of a key component for advanced nuclear reactors.

The ability to absorb and store heat makes important for and national climate goals. Molten salts can be used as both coolant and fuel in nuclear reactors that generate electricity without emitting greenhouse gases. They can also store large amounts of energy, which are increasingly needed on a power grid with fluctuating sources such as wind and solar power.

If you heated the on your kitchen table at 801 C (1,474 F), it would melt and you would have molten salt. However, to manufacture and store energy, not just any salt. Scientists are exploring different combinations of salts to achieve the exact properties needed to effectively cool and power a nuclear reactor for decades. These properties include lower melting temperatures, good consistency and the ability to absorb large amounts of heat, among others.

Which molten salt planes will provide the desired characteristics for a nuclear reactor? The potential variations are nearly endless. The study aimed to determine whether driven by machine learning could guide and refine real-world experiments at the Advanced Photon Source (APS), a DOE Office of Science user facility in Argonne. The results were recently published in the journal Physical examination B.

“We used the experimental results from APS to validate our simulation. At the same time, the simulation results provided us with more details of the salts to study further. They work with each other,” said Jicheng Guo, a chemical engineer at Argonne and the journal’s lead author. “It allows us to study several compositions at the same time.”

Researchers are using APS’s powerful X-rays to better understand specific salt mixtures by looking closely at their structures. But the time and cost associated with real-world experiences make it desirable to narrow the field of candidates who undergo inspection.

“The possible compositional space for molten salts is huge,” said Nathan Hoyt, Argonne researcher and co-author of the paper. “So it would be impossible to try to take experimental data for every possible composition.”

At the facility’s 6-ID-D beamline, a technique called high-energy X-ray diffraction captures the patterns generated when X-ray beams scatter across a sample of molten salt.

“APS is unique for these types of measurements,” said Chris Benmore, senior physicist at APS and co-author of the paper. “The high-energy X-rays it generates are very good for observing the structure of molten liquids, glasses, and amorphous materials in general.”

Machine learning involves training a computer to analyze a situation based on existing data. But in this case, the researchers didn’t have an abundance of validated examples to show the pattern. Building on previous modeling that explored heat-resistant materials, the researchers used what’s called active learning to create a transferable model for analyzing molten salts.

Rather than being adapted to one or two specific compositions of molten salt mixtures, the transferable model can be applied to mixtures in composition space. The model makes principle-based predictions; in other words, rather than a set of predefined answers. the the simulations were run using high-performance computing resources at the Argonne Leadership Computing Facility (ALCF), a user facility of the DOE Office of Science, and using the Bebop cluster at the Laboratory Computing Resource Center of Argonne.

“We didn’t train the model with examples of this sweet spot composition, where you get the right melting point,” said Ganesh Sivaraman, Argonne computational scientist and corresponding author of the paper. “Our model successfully predicted this sweet spot, even without the corresponding training input.”

Now that the researchers have shown that this approach can work, the next step is to work with even more complex data.

“A molten salt reactor is quite a dynamic environment. Conditions change over time, and sometimes impurities can get into the salt,” Guo said. “We want to introduce a tiny amount of these impurities to see if the model can predict how this affects the overall structure of molten salts and their properties.”

Co-authors with Guo, Hoyt, Sivaraman and Benmore are Logan Ward, Yadu Babuji, Mark Williamson and Ian Foster from Argonne and Nicholas Jackson from the University of Illinois Urbana-Champaign.

More information:
Jicheng Guo et al, Compositionally transferable machine learning potential for molten salts of LiCl-KCl validated by high-energy X-ray diffraction, Physical examination B (2022). DOI: 10.1103/PhysRevB.106.014209

Quote: Hot Salt, Clean Energy: How Artificial Intelligence Can Improve Advanced Nuclear Reactors (2022, December 15) Retrieved December 18, 2022 from artificial intelligence. html

This document is subject to copyright. Except for fair use for purposes of private study or research, no part may be reproduced without written permission. The content is provided for information only.

Leave a Reply