LAUSR

Abstract In this study we use independent replica calculations in order to assess statistical error propagation in whole-core actinide-burner High Temperature Reactor model. We use the MCB5 code for modeling a single irradiation cycle with reactivity control. We made 100 simulations of representative neutron precision, each having different initial random number generator seed. We analyze the real-to-apparent tally variation for neutron multiplication factor and nuclear reaction rates. While k eff uncertainty is well-predicted, local reaction uncertainties have exhibited under-prediction of several times since the very beginning of irradiation. Additionally, we correlate the increasing dominance ratio with the emergence of numerical instability and source convergence problems at a high burnup. Our conclusions indicate that the neutron cycle-to-cycle correlation must be considered by code users dealing with computationally expensive HTR models.