Advancements to generalized equivalence theory for preserving cross-sections of whole core simulations

Abstract

The conventional two-step approach, without the use of an equivalence method, can introduce significant error for the simulation of non-LWRs (Light Water Reactors), especially fast reactors, due to the large leakage and high anisotropic neutron distribution. To improve upon the accuracy of the two-step approach, a 3D whole core model is simulated with a transport method to generate region-wise homogenized cross-sections (XS). These XS can then be used in a diffusion whole core solver during the second step to extend the application to cycle and transient analysis. Discontinuity factors (DFs) are then introduced to improve the accuracy during the simulation of the second step. With properly generated DFs from Generalized Equivalence Theory (GET), the region-averaged solutions from the first transport step can then be reproduced by the second step diffusion solver. The Monte Carlo method was selected to perform the whole core transport simulation to generate region-wise XS. However, simulating whole core problems with Monte Carlo may result in poor statistics near the peripheral region especially for partial current tallies. This paper introduces an advancement to GET to reproduce region-wise solutions for select regions when the reaction rates and surface currents have good statistics in these regions but poor statistics in other regions. A reference high-fidelity model was constructed using the Serpent 2 Monte Carlo code based on the EBR-II benchmark evaluation and verification was carried out using the TriPEN-4 method in PARCS. The results show that it is possible to reproduce the exact eigenvalue and power distributions of whole core problems in a feasible manner.