Extending Embedded Monte Carlo as a novel method for nuclear data uncertainty quantification

Abstract

The purpose of this paper is to introduce a new approach to compute nuclear data uncertainties called Embedded Monte Carlo (EMC) and compare it to the well-established Total Monte Carlo (TMC) method. While the TMC methodology involves generating numerous random nuclear data library samples and conducting separate Monte Carlo simulations for each, this approach calculates nuclear data uncertainties by subtracting statistical uncertainties from the total uncertainties of each simulation. The EMC method addresses the challenge of statistical uncertainty where each batch represents a new random sample, thereby embedding the propagation of uncertainties within a single calculation and reducing computational costs. This technique also enables the calculation of nuclear data uncertainties by leveraging a combination of history and batch statistics in eigenvalue calculations. This paper demonstrates the potential of the EMC method using OpenMC with an analysis performed on two different benchmarks by propagating the uncertainty on three input parameters: the average neutron multiplicity ⊽, the prompt neutron fission spectrum (PFNS) χ and the 239Pu density.