A Monte Carlo based nodal diffusion model for criticality analysis, and, Application of high-order cross section homogenization method to two-group nodal diffusion

In the first part, an accurate and fast computational method is presented as an alternative to the Monte Carlo or deterministic transport theory codes currently used to determine the subcriticality of spent fuel storage lattices. The method is capable of analyzing storage configurations with simple or complex lattice cell geometry. It is developed based on two-group nodal diffusion theory, with the nodal cross sections and discontinuity factors determined from continuous-energy Monte Carlo simulations of each unique node (spent fuel assembly type). Three different approaches are developed to estimate the node-averaged diffusion coefficient. The applicability and the accuracy of the nodal method are assessed in two-dimensional geometry through several benchmark configurations typical at Savannah River Site. It is shown that the multiplication constant of the analyzed configurations is within 1% of the MCNP results.; In the second part, the high-order cross section homogenization method, recently developed by McKinley and Rahnema, is implemented in the context of two-group nodal diffusion theory. The method corrects the generalized equivalence theory homogenization parameters for the effect of the core environment. The reconstructed fine-mesh (fuel pin) flux and power distributions are a natural byproduct of this method. The method was not tested for multigroup problems, where it was assumed that the multigroup flux expansion in terms of the perturbation parameter is a convergent series. Here the applicability of the method to two-group problems is studied, and it is shown that the perturbation expansion series converges for the multigroup case. A two-group nodal diffusion code with a bilinear intra-nodal flux shape is developed for the implementation of the high-order homogenization method in the context of the generalized equivalence theory. The method is tested by using as a benchmark a core configuration typical of a BWR in slab geometry, which has large variations in the flux distribution across the core. There is a very good agreement between the nodal calculation and the fine-mesh reference calculation: the node-integrated group flux is within 0.5% of the reference solution in all nodes. The reconstructed fine-mesh flux (or equivalently the power distribution) in the core approximates the reference value very well.