Efficiency Improvement of Local Power Estimation in the General Purpose Monte Carlo Code MCNP
Progress in Nuclear Science and Technology
The MCNP general purpose Monte Carlo code was modified to improve the efficiency of detailed power distribution estimation in a reactor core for a large number of tally volumes. The standard MCNP F7 tally for fission energy deposition in a subvolume turns out to take a relatively long CPU time and becomes prohibitively slow when it is applied to all of the large number of fuel pins in a full size reactor, not to mention if it is applied to a number of subvolumes of each fuel pin. The recently
... pin. The recently introduced FMESH tally uses a Cartesian mesh independent of the material geometry and is therefore much faster, but still generates a lot of overhead. In the proposed modifications the addressing of a non-zero contribution to the deposited energy in a certain subvolume of a fuel pin with regard to the array where all tally contributions are stored is optimized. This turns out to reduce the Monte Carlo simulation time dramatically compared to the F7 tally and still appreciably (about 50 %) compared to the FMESH tally. Another efficiency improvement is the calculation of the potential contribution to the deposited fission energy in a fissile medium as a sum over contributions from all fissile nuclides in the medium in advance of the Monte Carlo simulation and to store this quantity as a function of energy in memory like is done in MCNP for cross section. However, this turns out to result in only a minor improvement in computation time.