COMPUTING ADJOINT-WEIGHTED PARAMETERS WITH REACTOR MONTE CARLO CODE RMC

摘要

In Monte Carlo simulations, many quantities are adjoint-weighted parameters including kinetic parameters and sensitivity coefficients. In this work, three algorithms based on the iterated fission probability (IFP) method, namely, the non-overlapping blocks algorithm, multiple overlapping blocks algorithm, superhistory algorithm, which are implemented in the Reactor Monte Carlo code RMC to compute adjoint-weighted kinetic parameters and eigenvalue sensitivity coefficients with regard to nuclear data, are investigated in terms of accuracy, efficiency, memory consumption, estimation of variance, etc. All the results computed by the three algorithms of RMC agree well with the MCNP6 results. The multiple overlapping blocks algorithm performs the most efficient calculations, however, it consumes the most memory and its computational time increases the most. The superhistory algorithm is as efficient as the non-overlapping blocks algorithm, however, both algorithms are less efficient than the multiple overlapping blocks algorithm by a factor of a block size. On the other hand, the non-overlapping blocks algorithm consumes less memory than the multiple overlapping blocks algorithm by a factor of a block size while the superhistory algorithm consumes the least memory. Additionally, all the three algorithms almost produce the same estimation of variance at the condition of same effective particle histories.