Statistical methods used for code-to-code comparisons in the OECD/NRC PWR MSLB benchmark

Abstract

The ongoing pressurized water reactor (PWR) main steam line break (MSLB) benchmark problem, sponsored by the Office for Economic Cooperation and Development (OECD), the United States Nuclear Regulatory Commission (US NRC), and the Pennsylvania State University (PSU) consists of three exercises, whose combined purpose is to verify the capability of system codes to analyze complex transients with coupled core/plant interactions; to test fully the 3D neutronics/thermal-hydraulic coupling; and to evaluate discrepancies between the predictions of coupled codes in best-estimate transient simulations. Exercise two is intended to test core response to imposed system thermal-hydraulic conditions. For this exercise, the participants are provided with transient boundary conditions and two cross-section libraries. Results are submitted for six steady-state cases and two transient scenarios. The boundary conditions, the details for each case, and the output requested are described in the final specifications for the benchmark problem. To fully analyze the data for comparison in the final report, a suite of statistical methods has been developed, to serve as a reference in the absence of experimental data. A corrected arithmetical mean and standard deviation are calculated for all data types: single-value parameters, 1D axial distributions, 2D radial distributions, and time histories. Each participant's deviation from the mean and a corresponding figure-of-merit are reported for the purposes of comparison and discussion. Selected mean values and standard deviations are presented in this paper for several parameters at specific points of interest: for the initial steady-state 2, at hot-full power, radial and axial power distributions are presented, along with effective multiplication factor, power peaking factors, and axial offset. For the snapshot taken at the time of highest return-to-power in transient Scenario 2, parameters presented include axial and radial power distributions, total and fission power levels, power peaking factors, and axial offset. Additionally, some features of the time histories for total power and fission power in Scenario 2 are presented and discussed.