Search

The topic of this paper is the development of sensitivity and uncertainty analysis capability to the CASMO-4/CASMO-4E – SIMULATE-3 code sequence in the context of the OECD/NEA benchmark ‘Uncertainty Analysis in Best-Estimate Modelling for Design, Operation and Safety Analysis of LWRs’ (UAM). The developed capability uses a two-step approach. In the first step, Uncertainties in nuclear data are propagated to two-group cross sections, diffusion coefficients, and assembly discontinuity factors. This is carried out using deterministic, perturbation-theory-based uncertainty analysis methodology. In the second step, a global covariance matrix, characterizing the uncertainties of the group constants, is formed, and the uncertainties are propagated through a full core SIMULATE calculation using a stochastic approach. This system enables the analysis of nuclear data related uncertainties in assembly-homogenized group constants, assembly discontinuity factors, and pin powers, as well as full core results such as multiplication factor and power distribution. The mathematical background of the deterministic uncertainty analysis methodology is reviewed and the main conclusions related to the implementation are summarized. Numerical results are presented for the full core Three Mile Island model in exercise I-3 of the UAM benchmark at hot zero power with all rods out and inserted. The computational efficiency of the calculations is discussed.

This paper describes the methods used in the Serpent 2 Monte Carlo code for producing homogenized group constants for nodal diffusion and other deterministic reactor simulator calculations. The methodology covers few-group reaction cross sections, scattering matrices, diffusion coefficients and poison cross sections condensed in infinite and B 1 leakage-corrected critical spectra, as well as the calculation of discontinuity factors, pin-power form factors, delayed neutron parameters and total and partial albedos. Also included is a description of an automated burnup sequence, which was recently implemented for the handling of restart calculations with branch variations. This capability enables covering the full range of local operating conditions required for the parameterization of group constants within a single run. The purpose of this paper is to bring the methodological description provided in earlier publications up to date, and provide insight into the developed methods and capabilities, including their limitations and known flaws.

We survey prediction capabilities of ENDF/B-VII.1 , JEFF-3.2 and JENDL-4.0u nuclear data libraries (NDLs) for the application of generating two-group homogenized assembly constants for a steady state diffusion model in the context of UAM-LWR (Uncertainty Analysis in Best-Estimate Modeling for Design, Operation and Safety Analysis of LWRs) Benchmark. We consider two different fuel assembly test cases representing a PWR. State of uncertainty quantification in each NDLs is presented for the application. We expect small differences between the NDLs due to the use of expert judgment in the evaluation processes, and identify several order-of-magnitude differences between the NDLs for significant contributors to uncertainty. We also quantify the contribution from cross-material correlations to the uncertainties.

This paper describes the implementation of classical perturbation theory based sensitivity and uncertainty analysis to the reactor physics code CASMO-4 in the context of the UAM (Uncertainty Analysis in Best-Estimate Modelling for Design, Operation and Safety Analysis of LWRS) benchmark. The theoretical background as well as practical guidelines for similar work are reviewed and the developed methodology is described in detail. A technique is proposed for handling a discrepancy between the code and covariance libraries regarding the amount of reactions present in the cross-section model. Numerical results for PWR and GEN-III MOX pin-cell test problems are presented and compared to TSUNAMI-1D. The results are in accordance with theoretical considerations and reflect the characteristics of the developed methodology.