Improvement of the CLUTCH method for sensitivity analysis of k-eigenvalue to continuous-energy nuclear data in NECP-MCX

Abstract

The Iterated Fission Probability (IFP) method and Contributon-Linked eigenvalue sensitivity/Uncertainty estimation via Track-length importance Characterization (CLUTCH) method are commonly implemented in Monte-Carlo codes to calculate the sensitivity coefficients of keff to continuous-energy nuclear data. However, these methods are plagued by the problem of large variance of sensitivity coefficients of the scattering reactions with relatively small cross sections. This problem means that more particles are needed to accurately calculate these sensitivity coefficients. To address this problem, a novel technique called CLUTCH coupled with forced collisions (CLUTCH-FC) method is proposed to reduce the variance of sensitivity coefficients of those reactions. The forced collision method is performed with specified reactions rather than specified cells in tradition to increase the sampling of specified reactions and a new rule for the setting of cutoff weight of neutrons is also proposed to make the best use of forced collisions. Three methods, viz. IFP, CLUTCH and CLUTCH-FC methods, are implemented in the Monte-Carlo code NECP-MCX. The verification is conducted in Godiva, Jezebel and TMI-1 problems by comparing the sensitivity coefficients calculated by IFP and CLUTCH methods with those calculated by the direct numerical perturbation (DNP) method. The numerical results indicate that 1) the sensitivity coefficients calculated by the IFP and CLUTCH methods corroborate well with those calculated by the DNP method, and 2) the CLUTCH-FC method is efficient in improving the tally efficiency of sensitivity coefficients of scattering reactions with relatively small cross sections by comparing results of CLUTCH-FC and CLUTCH methods.