Assessment of DRACCAR predictive capabilities during bundle reflood under low-power conditions: A systematic uncertainty and sensitivity analysis
Abstract
Accurately predicting the thermal evolution of fuel rods during a Loss-of-Coolant Accident (LOCA) is essential for nuclear safety assessments, yet nonlinear coupling among reflood phenomena remains a major source of predictive uncertainty. This work performs a systematic uncertainty propagation and sensitivity analysis using DRACCAR, applied to two low-power, low-injection-velocity tests (9178, 9271) from the OECD/NEA RBHT Phase II benchmark. By propagating uncertainties in thirteen thermal-hydraulic closure laws and five experimental boundary conditions, the methodology quantifies the resulting uncertainty bands on the local maximum cladding temperature and quench time evaluated at each instrumented thermocouple location, and identifies, via Partial Rank Correlation Coefficients, the closures most responsible for experimental discrepancies. These demanding conditions are exploited as a diagnostic framework to identify which closure laws control the predictive response of the code. Results show that this discrepancy is primarily driven by three closures: post-CHF interfacial friction, steam-droplet interfacial heat transfer, and steam-to-wall convection. Increasing interfacial coupling systematically reduces the error, whereas intensifying steam-to-wall convection amplifies it, suggesting that enhanced steam-to-wall heat transfer disproportionately heats the steam core without cooling the cladding at the limiting axial locations. Rod power emerges as the single most influential parameter, with a PRCC reaching 0.90 against the cladding temperature error metric despite its narrow experimental uncertainty range. These results indicate that, beyond refining the dominant closure laws identified here, the extreme sensitivity to rod power reflects an intrinsic difficulty of resolving the multiphase flow finely under very-low-power conditions.
Related
- When does boundary-condition uncertainty outweigh thermal and vegetation effects in street canyon flow? A global sensitivity analysis of input uncertainty
- Predictive Uncertainty for Neural CAE Surrogates
- Hurricane Laura-induced loads on wind turbines -- Part I: High-fidelity modeling and sensitivity to turbulent inflow conditions
- Assessment of Super- and Hyper-Kamiokande Sensitivity to the DSNB
- Gaussian process regression with physics-guided features for WLAM bead geometry: Repeated validation, algorithmic baselines, and predictive-interval assessment
BibTeX
@article{valencia2026assessment,
title = {{Assessment of DRACCAR predictive capabilities during bundle reflood under low-power conditions: A systematic uncertainty and sensitivity analysis}},
author = {J.E. Luna Valencia and Tony Glantz},
journal = {Nuclear Engineering and Design},
year = {2026},
month = sep,
doi = {10.1016/j.nucengdes.2026.115203},
url = {https://asnr.hal.science/hal-05748599v1},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Assessment of DRACCAR predictive capabilities during bundle reflood under low-power conditions: A systematic uncertainty and sensitivity analysis
AU - J.E. Luna Valencia
AU - Tony Glantz
PY - 2026
DA - 2026/09/11
JO - Nuclear Engineering and Design
DO - 10.1016/j.nucengdes.2026.115203
UR - https://asnr.hal.science/hal-05748599v1
AB - Accurately predicting the thermal evolution of fuel rods during a Loss-of-Coolant Accident (LOCA) is essential for nuclear safety assessments, yet nonlinear coupling among reflood phenomena remains a major source of predictive uncertainty. This work performs a systematic uncertainty propagation and sensitivity analysis using DRACCAR, applied to two low-power, low-injection-velocity tests (9178, 9271) from the OECD/NEA RBHT Phase II benchmark. By propagating uncertainties in thirteen thermal-hydraulic closure laws and five experimental boundary conditions, the methodology quantifies the resulting uncertainty bands on the local maximum cladding temperature and quench time evaluated at each instrumented thermocouple location, and identifies, via Partial Rank Correlation Coefficients, the closures most responsible for experimental discrepancies. These demanding conditions are exploited as a diagnostic framework to identify which closure laws control the predictive response of the code. Results show that this discrepancy is primarily driven by three closures: post-CHF interfacial friction, steam-droplet interfacial heat transfer, and steam-to-wall convection. Increasing interfacial coupling systematically reduces the error, whereas intensifying steam-to-wall convection amplifies it, suggesting that enhanced steam-to-wall heat transfer disproportionately heats the steam core without cooling the cladding at the limiting axial locations. Rod power emerges as the single most influential parameter, with a PRCC reaching 0.90 against the cladding temperature error metric despite its narrow experimental uncertainty range. These results indicate that, beyond refining the dominant closure laws identified here, the extreme sensitivity to rod power reflects an intrinsic difficulty of resolving the multiphase flow finely under very-low-power conditions.
SN - 0029-5493
SN - 1872-759X
ER -