Severe accidents (SA), although highly improbable, are the major contributors to the societal risk associated with nuclear power. These events involve core meltdowns, potentially leading to releases of radioactive materials to the environment. To mitigate these risks, advanced simulation tools, such as MELCOR, are employed to analyze accident scenarios. However, due to the complexity of physical phenomena under accident conditions, both input parameters and model formulations are subject to large uncertainties, which must be rigorously quantified to ensure reliable predictions. In this direction, this paper presents the first application of inverse uncertainty quantification (IUQ) to SA analyses. Specifically, the CIRCE IUQ methodology is adopted, and fission product release (FPR) is the targeted phenomenon. Following the SAPIUM guidelines, a comprehensive FPR database is setup and qualified through an adequacy analysis. The Revised CORSOR-Booth in MELCOR is used to simulate these tests, and results are benchmarked against experimental data. The cesium release fraction is selected as quantity of interest, and the CIRCE algorithm is applied to determine the cesium diffusion coefficient uncertainty band. Finally, a statistical validation process is conducted by propagating the derived uncertainties through the PHEBUS FPT-1 test. Results confirm that CIRCE, along with SAPIUM guidelines, provides a consistent and realistic uncertainty characterization, demonstrating its applicability to SA simulations. Despite the challenge of requiring a large and representative experimental database, often lacking for many SA phenomena, this study highlights the significant added value of IUQ in enhancing the predictive capabilities of SA codes. It supports the integration of IUQ into Best Estimate Plus Uncertainty methodologies for SA analyses, especially for phenomena featured by sufficient experiments. Furthermore, IUQ could be a valuable tool for code developers, helping identifying key uncertain parameters and guiding research efforts toward areas with the greatest impact on model fidelity.
Inverse Uncertainty Quantification for Severe Accident Analyses
G. Tinfena;M. Angelucci;S. Paci;
2025-01-01
Abstract
Severe accidents (SA), although highly improbable, are the major contributors to the societal risk associated with nuclear power. These events involve core meltdowns, potentially leading to releases of radioactive materials to the environment. To mitigate these risks, advanced simulation tools, such as MELCOR, are employed to analyze accident scenarios. However, due to the complexity of physical phenomena under accident conditions, both input parameters and model formulations are subject to large uncertainties, which must be rigorously quantified to ensure reliable predictions. In this direction, this paper presents the first application of inverse uncertainty quantification (IUQ) to SA analyses. Specifically, the CIRCE IUQ methodology is adopted, and fission product release (FPR) is the targeted phenomenon. Following the SAPIUM guidelines, a comprehensive FPR database is setup and qualified through an adequacy analysis. The Revised CORSOR-Booth in MELCOR is used to simulate these tests, and results are benchmarked against experimental data. The cesium release fraction is selected as quantity of interest, and the CIRCE algorithm is applied to determine the cesium diffusion coefficient uncertainty band. Finally, a statistical validation process is conducted by propagating the derived uncertainties through the PHEBUS FPT-1 test. Results confirm that CIRCE, along with SAPIUM guidelines, provides a consistent and realistic uncertainty characterization, demonstrating its applicability to SA simulations. Despite the challenge of requiring a large and representative experimental database, often lacking for many SA phenomena, this study highlights the significant added value of IUQ in enhancing the predictive capabilities of SA codes. It supports the integration of IUQ into Best Estimate Plus Uncertainty methodologies for SA analyses, especially for phenomena featured by sufficient experiments. Furthermore, IUQ could be a valuable tool for code developers, helping identifying key uncertain parameters and guiding research efforts toward areas with the greatest impact on model fidelity.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
