Impact of cross-section generation procedures on the simulation of the VVER-1000 pump startup experiment in the OECD/DOE/CEA V1000CT benchmark by coupled 3D thermal-hydraulics/neutron kinetics models

In the framework of joint effort between the Nuclear Energy Agency (NEA) of OECD, the United States Department of Energy (US DOE), and the Commissariat a l'Enerige Atomique (CEA), France a coupled three-dimensional (3D) thermal-hydraulics/neutron kinetics benchmark was defined. The overall objective of OECD/NEA V1000CT benchmark is to assess computer codes used in analysis of VVER-1000 reactivity transients where mixing phenomena (mass flow and temperature) in the reactor pressure vessel are complex. Original data from the Kozloduy-6 Nuclear Power Plant are available for the validation of computer codes: one experiment of pump start-up (V1000CT-1) and one experiment of steam generator isolation (V1000CT-2). Additional scenarios are defined for code-to-code comparison. As a 3D core model is necessary for a best-estimate computation of all the scenarios of the V1000CT benchmark, all participants were asked to develop their own core coupled 3D thermal-hydraulics/neutron kinetics models using the data available in the benchmark specifications and a common cross-section library. The first code-to-code comparisons based on the V1000CT-1 Exercise 2 specifications exhibited unacceptable discrepancies between two sets of results. The present paper focuses on the analysis of the observed discrepancies. The VVER-1000 3D neutron kinetics models are based on cross-section data homogenized on the assembly level. The cross-section library, provided as part of the benchmark specifications, thus consists in a set of parameterized two group cross sections representing the different assemblies and the reflectors. The origin of the observed large discrepancies was found mainly to lie in the methods used to solve the diffusion equation. The VVER reflector properties were also found to enhance discrepancies by increasing flux gradients at the core/reflector interface thus highlighting more the difficulties in some codes to handle high exponential flux gradients. This paper summarizes the different steps applied to analyze the neutronic codes and their predictions as well as the impact of cross-section generation procedures.