A dataset for thermal stratification phenomenon in lead-cooled fast reactors

BackgroundLead-cooled Fast Reactors (LFR), as promising Generation-IV nuclear energy systems, feature excellent passive safety and economic efficiency. However, thermal stratification easily occurs in the upper plenum after emergency shutdown, seriously endangering reactor operation. Currently, relevant data are scattered in various literatures with inconsistent formats, limiting the development of numerical simulation and data-driven methods.PurposeThis study aims to integrate and standardize existing data to establish a unified, reusable benchmark dataset, supporting turbulence model calibration, system code optimization, and data-driven thermal stratification prediction.MethodsFirstly, publicly published relevant literatures available at home and abroad from 2004 to 2026 were systematically retrieved and after rigorous screening, experimental and numerical simulation data from 32 published literatures (covering typical reactors and experimental facilities developed in different countries and regions including China, Russia, South Korea and Europe, as well as various working conditions such as steady-state operation and typical accident transients) were collected. Then, curves in literatures were digitized via Origin software (with triple independent extractions for error reduction). Finally, a structured metadata framework was established, and data were standardized through unit unification and CSV format conversion.ResultsTriple independent extractions are verified the reliability and repeatability of the method. The established dataset integrates multi-source data including vertical temperature distribution, mass flow rate, and interface position, achieving high data consistency and reliability.ConclusionsThis standardized dataset proposed in this study resolves data fragmentation and heterogeneity, providing benchmark support for turbulence model evaluation and system code optimization. It also lays a foundation for machine learning-based rapid prediction of thermal stratification, facilitating LFR thermal-hydraulic safety analysis.