Abstract
HTR-10 is a 10 MWt prototype pebble-bed reactor (PBR) that presents a doubly heterogeneous geometry for neutronics calculations. An appropriate unit-cell treatment for the associated fuel elements is vital for creating problem-dependent multigroup cross sections. Considering four unit-cell options for resonance self-shielding correction in SCALE6, a series of HTR-10 core models were established using the CSAS6 sequence to systematically investigate how they affected the computational accuracy and efficiency of PBR criticality calculations. Three core configurations, which ranged from simplified infinite lattices to a detailed geometry, were examined. Based on the same ENDF/B-VII.0 cross-section library, multigroup results were evaluated by comparing with continuous-energy SCALE6/CSAS6 and MCNP5 calculations. The comparison indicated that the INFHOMMEDIUM results overestimated the effective multiplication factor (k eff ) by about 2800 pcm, whereas the LATTICECELL and MULTIREGION treatments overestimated k eff values with similar biases at approximately 470-680 pcm. The DOUBLEHET results attained further improvement, reducing the k eff overestimation to approximately 280 pcm. The comparison yielded two unexpected problems from using SCALE6/CSAS6 in HTR-10 criticality calculations. In particular, the continuous-energy CSAS6 calculations in this study present a non-negligible discrepancy with MCNP5, potentially causing a k eff value overestimate of approximately 680 pcm. Notably, using a cell-weighted mixture instead of an explicit model of individual TRISO particles in the pebble fuel zone does not shorten the CSAS6 computation time in the high-fidelity HTR-10 model because of the presence of graphite balls and the complex modeling of peripheral reflectors. © 2013 Elsevier Ltd. All rights reserved.