[2C01-07] Data Science, Fuel Research with Machine Learning

[2C05] New Developments in Nuclear Fuel Research through Integration with Data Science

(5)Fabrication and Characterization of Metallic Uranium Compounds (UFe₃B₂, USiNi) Prepared via Spark Plasma Sintering

*Yifan Sun¹, Yuji Miyawaki², Yuji Ohishi², Hiroaki Muta², Shun Fujieda², Ken Kurosaki^1,3 (1. Kyoto Univ., 2. Osaka Univ., 3. Univ. of Fukui)

Keywords:Advanced nuclear fuels, Machine learning, Thermal conductivity

To accelerate the discovery of advanced nuclear fuels exhibiting superior thermal conductivity, we constructed a machine learning (ML) model to predict the thermal conductivity of uranium-based compounds. To evaluate the reliability of our ML model, we need to compare the predicted thermal conductivities with experimental data. In this study, dense UFe₃B₂and USiNi samples were fabricated through spark plasma sintering and their thermal conductivities were assessed using laser flash analysis. A comparison of the measured and predicted thermal conductivities of UFe₃B₂ revealed that the model accurately predicted both the thermal conductivity and its temperature dependence. Conversely, for USiNi, the model only accurately predicted the temperature dependence but overestimated the thermal conductivity. Overall, our preliminary findings highlight the potential of ML as a transformative tool in the discovery of novel nuclear fuels.