Nuclear data uncertainty propagation to the main physical parameters of NUR research reactor

Highlights

• S/U analysis was applied on NUR nuclear research reactor

• The sensitivity coefficients were calculated and confirmed by the direct perturbation method.

• The total uncertainty in $k_{eff}$ due to cross-section covariance data was obtained.

• The main nuclide-reaction pairs contributing to the k_eff uncertainty were identified.

Abstract

An accurate nuclear reactor model with less statistical uncertainty is not enough to have an accurate assessment of the calculated physical parameter value. It is worthwhile to note that another source of uncertainties originating from nuclear data uncertainties must be quantified. To realize such an assessment, sensitivity and uncertainty analysis based on first-order perturbation theory implemented in the SCALE code system was applied to the IV.N (an optimized one in series of configurations) core configuration of the Algerian light water Nuclear Uranium Reactor (NUR). The impact of nuclear data uncertainties on the effective multiplication factor was analyzed. Criticality calculation was performed by using the KENO V.a Monte Carlo code of the SCALE code system and the ENDF/B libraries with 238 and 252 energy group structures. The results were compared to the available measurement and validated. The accuracy of this criticality calculation was judged sufficient to do a sensitivity and uncertainty calculation for the same model.

First, sensitivity and uncertainty calculations were performed using the TSUNAMI3D module of the SCALE code system and ENDF/B-VII.0 library with the covariance data in 44 and 56 energy group structures. Next, the model was updated to the latest library used in SCALE code system version 6.2.3; ENDF/B-VII.1 with 252 energy group and the corresponding 56 group covariance data.

The uncertainty results of NUR effective multiplication factor due to uncertainties in nuclear data were determined. The top contributor is $\bar{\nu }$ (the average number of neutrons per fission) of 235U for all cases. The total uncertainty in $k_{eff}$ due to uncertainties in the cross-section data was determined to be around 0.84% using 56 group covariance data. This uncertainty was higher than the uncertainty when using 44 group covariance data.

Introduction

An accurate nuclear reactor model depends on the identification and minimization of the whole errors committed in the modeling approximation, the numerical solution and the input parameters entered (geometries, materials …). Errors in the input parameters, mainly the nuclear data (microscopic cross sections, fission spectra, neutron yield $\bar{\nu }$ , and scattering distributions …) are considered as the most significant source of uncertainties in the reactor physics calculations.

Sensitivity and Uncertainty (S/U) analysis methods were used to quantify the impact of nuclear data uncertainties on effective multiplication factor and reactivity responses. Those methods were first developed for application to fast reactor studies in the 1970s (Weisbin et al., 1976). Several works have revitalized and updated the available S/U computational capabilities for computing sensitivity coefficients such as the adjoint-weighted perturbation capability in MCNP6 (Kiedrowski and Favorite, 2010), a discrete ordinates sensitivity formulation of the first-order perturbation theory in the SUSD3D code (I. Kodeli, 2001) and a fine group calculation from TSUNAMI-3D (Rearden, 2009).

In this work, sensitivity and uncertainty analysis (Broadhead et al., 2004) based on first-order perturbation theory implemented in TSUNAMI-3D of SCALE code system (Oak Ridge National Laboratories, 2018) was applied on IV.N core configuration of NUR research reactor in order to know the nuclear data uncertainty propagation on effective multiplication factor.

First, for the optimized reactor core configuration “IV.N″, the criticality calculation was performed using the KENO V.a Monte Carlo code of the SCALE code system. The main calculated physical parameters out coming from the model were compared to experimental data (Meftah et al., 2006; Madariaga et al., 1989) and show a good agreement (Sellaoui et al., 2014).

Then, sensitivity and uncertainty calculations were performed by TSUNAMI3D using ENDF/B-VII.0 library with the covariance data in 44 and 56 energy groups. Next, the model was updated to the earliest library; ENDF/B-VII.1 with 252 energy group and the corresponding 56 group covariance data (Oak Ridge National Laboratories, 2018).Sensitivity coefficients obtained by sensitivity and uncertainty analysis were confirmed by the direct perturbation sensitivity calculation. Those sensitivities of a computed $k_{eff}$ value to cross-section data were coupled with cross-section-covariance data to produce uncertainty in $k_{eff}$.