连续能量核数据敏感性分析及其在AP1000中的应用

AP1000 is the typical third-generation nuclear power technology. The sensitivity analysis of continuous-energy nuclear data for AP1000 is the basis of uncertainty quantification which plays an important role in subsequent safety analysis for AP1000. Multi-group approximations exist in sensitivity analysis based on multi-group energy nuclear data. Therefore, it's necessary to calculate sensitivity coefficients based on continuous-energy nuclear data. In this paper, adjoint flux was calculated based on iterated fission probability (IFP) in Monte Carlo forward calculation and then sensitivity coefficients of k_eff to nuclear data were obtained through first-order perturbation theory. The sparse-matrix method was used to reduce the large memory usage in IFP, while the overlapping block method was used to improve counting efficiency and large statistics fluctuation. The function module of sensitivity analysis was developed in Monte Carlo code NECP-MCX. The module developed above was verified based on the VERA 2B problem. Both energy-dependent and energy-independent sensitivity coefficients calculated by NECP-MCX are all within 3σ of those obtained by MCNP, which indicates the accuracy of the NECP-MCX. In addition, the sensitivity coefficients related to the scattering cross section fluctuate greatly, which is because when the magnitudes of the scattering term and the total term in the sensitivity coefficient formula are the same, the subtraction of these two terms will result in a value with a larger statistical fluctuation. Furthermore, the sensitivity analysis of continuous-energy nuclear data was performed with enhanced NECP-MCX for AP1000 and the nuclear data that are most sensitive to k_eff were analyzed, which are ²³⁵U-ν, ²³⁵U-σ_f, ²³⁵U-σ_t, ²³⁸U-σ_γ, ¹H-σ_elas and ¹⁰B-σ_t, where ²³⁵U-ν, ²³⁵U-σ_f and ²³⁵U-σ_t are mainly related to the production of neutron, ²³⁸U-σ_γ and ¹⁰B-σ_t are dominant in the disappearance of neutron and ¹H-σ_elas is dominant in the slowing down of neutron. In this paper, the sensitivity coefficient of k_eff to nuclear data was derived based on first-order perturbation theory. The module of continuous-energy nuclear data sensitivity analysis was implemented in Monte Carlo code NECP-MCX. The verification of the module was conducted based on VERA 2B problem. Finally, the sensitivity coefficients of k_eff to continuous-energy nuclear data were calculated for AP1000. The most sensitive nuclear data to k_eff were obtained from the results. In the future, based on the obtained sensitivity coefficients and the covariance information in the nuclear data, the uncertainty of the AP1000 core k_eff calculation caused by nuclear data can be obtained through sandwich rule for subsequent safety analysis.