An Orthonormally weighted standardized time series for the error estimation of local tallies in Monte Carlo criticality calculation

Ueki, Taro  

An orthonormally weighted standardized time series (OWSTS) was investigated for the statistical error estimation in the Monte Carlo calculation of local tallies. Unlike the original implementation of a standardized time series, the computation of standard deviation via OWSTS can be made free of the grouping of iteration cycles into batches. Its characteristic aspect is the application of an arbitrary number of weighting functions to a standardized series of tallies such that asymptotically independent and unbiased estimates are produced based on the statistics of Brownian bridge. In the present work, a trigonometric set of weighting functions is extended and applied to local power tallies in a pressurized water reactor. Numerical results are demonstrated that the OWSTS error estimation is unbiased for a sufficiently large number of iteration cycles.