Cross-section sensitivity analysis and uncertainty quantification of the ASPIS Iron 88 shielding benchmark experiment

Abstract: Nuclear data is the major source of error in shielding calculations, which leads to inaccurate neutron fluence and energy spectrum. We establish a correlation model between cross sections and integral responses based on perturbation theory, and improve the efficiency of the sensitivity solution by adjoint flux weighting. Error propagation is obtained by estimating statistical moment of Taylor stochastic expansion for the integral response. Numerical validation using the ASPIS Iron 88 shielding benchmark, which demonstrates comparable accuracy between self-developed code and the SUSD3D program. The method effectively identifies the differential impacts of various cross-section data, and is not strictly dependent on transport calculation accuracy. Perturbation theory and the sandwich rule can efficiently quantify the uncertainty introduced by nuclear data.