Toward a More Realistic Analysis of Neutron Clustering

The results of neutron Monte Carlo (MC) transport calculations are subject to random fluctuations about their expected values. The term “neutron clustering” refers to situations in which these fluctuations exhibit particularly strong spatial correlations in iterated-fission-source calculations. Various idealized models of the MC process have been developed to study this phenomenon. Over time, these models have evolved to more realistically reflect the algorithms used in MC codes. This paper continues along this path by including the possibility that some neutrons will not terminate in an event that can potentially produce new neutrons and by considering an algorithm without replacement (WOR) for selecting the neutron source sites. It is shown that sampling source sites WOR versus with replacement can greatly reduce the degree of clustering.