U&S analyses in a BWR NPP LB-Loca transient with ATF

• A LB-LOCA scenario in a BWR/4 reactor is simulated using Zr and FeCrAl-C35M as fuel rod cladding materials. • An uncertainty and sensitivity analysis has been performed on the maximum peak cladding temperature reached. • Two uncertainty quantification methods are employed using the DAKOTA v6.19 tool: the Polynomial Chaos Expansion (PCE) approach and Wilks’ method. • During LB-LOCA, FeCrAl-C35M cladding shows lower PCT rise, slower cooling, and delayed quench temperature compared to Zr. • A second-order quadrature PCE was applied, requiring 64 TRACE simulations, which is comparable in computational cost to Wilks’ method based on 59 runs. Regarding the Advanced Technology and Accident Tolerant Fuels (ATF) framework, there are ongoing international projects focused on developing methodologies for Loss-of-Coolant Accident (LOCA) evaluation in nuclear power plants, including the implementation of Best-Estimate Plus Uncertainty (BEPU) approaches. This paper presents the Uncertainty and Sensitivity (U&S) analysis of the maximum Peak Cladding Temperature (PCT) reached during a Large-Break LOCA (LB-LOCA) in a Boiling Water Reactor (BWR) using the TRACE v5p7 code and the DAKOTA v6.19 tool. The FeCrAl-C35M alloy, considered an ATF cladding concept, has been modeled in TRACE as a user-defined material. Under LB-LOCA conditions, FeCrAl cladding shows a lower increase in PCT, a slower cooling rate, and a delayed quench temperature compared to conventional Zircaloy cladding. Two uncertainty quantification methods are applied: the Polynomial Chaos Expansion (PCE) approach and Wilks’ method. PCE is used as a propagation technique that represents the model response as a linear combination of orthogonal polynomials. The polynomial coefficients are determined from a set of TRACE simulations using two non-intrusive spectral projection schemes: the Tensor Product Rule Quadrature and Smolyak’s Sparse Grid Integration. The number of required TRACE simulations depends on the number of uncertain input variables and on the parameters defining the PCE integration scheme, such as the polynomial order and the quadrature order or sparse grid level. Once the model has been constructed, large-sample response statistics can be obtained without additional TRACE simulations. This study assesses the implementation of the PCE approach within the DAKOTA tool to apply BEPU methodologies for the evaluation of LOCA accidents.