A new approach for uncertainty quantification in predictor-corrector quasi-static Monte Carlo transient simulation

Abstract

Two different approaches are widely accepted for transient Monte Carlo (MC) simulation namely the Dynamic Monte Carlo (DMC) and the Predictor-Corrector Quasi-Static Monte Carlo (PCQS-MC). The MC transport code iMC developed at the Korea Advanced Institute of Science and Technology incorporates both approaches. In this paper, an original method for properly assessing the uncertainty of PCQS-MC is proposed and demonstrated using the iMC code. A detailed description of applying the PCQS method for transient MC calculation is presented with an emphasis on the origins of PCQS-MC uncertainties. The implementation of a quasi-static method, i.e., calculation of point-kinetics (PK) parameters, incurs an additional uncertainty for PCQS-MC calculation alongside the conventional stochastic MC uncertainty. Such quasi-static treatment-driven stochasticity cannot be recognized through the conventional MC sampling scheme, insinuating a significant underestimation of uncertainty without proper measures. To verify the findings, null-transient simulations have been performed for GODIVA and C5G7 benchmarks, where the former and latter problems require continuous and multi-group energy treatment respectively. For an improved uncertainty evaluation, a new sampling scheme referred to as the PK-sampling scheme is proposed where cycle-wise PK correction is made. Proper estimation of uncertainty can be made from the heap of cycle-wise corrected power through a unique screening process based on the null hypothesis testing. The proposed PK sampling scheme is tested for C5G7-TD transient benchmarks using iMC where the result plainly attests to the effectiveness of the new sampling approach.