Improper source convergence benchmarks performed thanks to 5 different sampling methods implemented in Moret 4
J. Miss, O. Jacquet, ANS 2002 Winter meeting, 17-19 november 2002, Washington (USA).
The OECD/NEA working group on Sources Convergence for criticality safety is an international committee composed of scientists who deal with criticality safety studies, and researchers who develop Monte Carlo code systems. Such a group composition is of particular interest, because it makes it possible, thanks to benchmarks, to test and compare the performances of computer codes, and also to make recommendations for users. Since a system, described in a criticality study, is composed of units which are very loosely coupled, the true sources distribution can be very difficult to achieve. In such a case, the conventional Monte Carlo method can lead to an erroneous estimation underprediction) of k-effective. The expert group has selected four benchmarks having
loosely coupled system and close to nuclear safety assessment configurations: (i) a checkerboard storage, (ii) a burnup credit array of pincells, (iii) a slab problem, and (iv) a lattice of 5x5x1 metal spheres. The group for comparison has chosen simulation parameters and output format. The group is expected to propose appropriated
methods to deal with this issue. This paper is aimed to present and discuss the results achieved using the MORET4 code, in which one can choose among five different sampling methods: conventional, stratified, superhistory, matrix and importance strategies.