A research team at Oak Ridge National Laboratory (ORNL) has developed a new machine learning method to predict the properties of molten salts, materials widely used in the nuclear energy field. The study, published in Chemical Science, used ORNL's Summit supercomputer to reach a level of accuracy comparable to quantum chemistry methods.
According to Oak Ridge National Laboratory website, the research enables precise modeling of both the solid and liquid states of molten salts, focusing on their thermodynamic properties, which are critical for high-temperature nuclear applications. These include nuclear fuel processing and supporting long-term reactor performance.
“The exciting part is the simplicity of the approach,” said Luke Gibson of ORNL in the article. “In fewer steps than existing approaches, machine learning gets us to higher accuracy at a faster rate.”
Modeling molten salt behavior has traditionally required significant time and resources. The new method allows researchers to obtain accurate predictions more efficiently. This may assist in advancing nuclear reactor design, safety, and waste management.
The study highlights how machine learning and increased computational capabilities are being applied to materials science. This work represents a step forward in applying artificial intelligence to improve understanding and use of complex materials in the energy sector.
