Stochastic mixing of bound thermal scattering data in MONK

Abstract

The Monte Carlo neutronics code MONK® uses run-time Doppler broadening to achieve highly accurate temperature interpolation. However the tabulated S(α,β) data which are used to compute the secondary energies and angles in bound thermal scattering interactions do not lend themselves to interpolation, and it is therefore recommended to use the closest tabulated temperatures. In some cases this can lead to significant step changes in reactivity at the mid-point between two tabulated temperatures. The implementation and testing of a stochastic mixing approach to avoid this approximation are described. Results from two test cases are presented: the first decouples the effect of moderator density and demonstrates that the stochastic mixing removes the step changes in reactivity seen when using the nearest temperature approach; and the second is a more physically-realistic model at reactor operating conditions, which demonstrates that the nearest temperature approach does not introduce significant errors in thermal LEU systems if the moderator density is varied with temperature in a physically-realistic way.