A recently developed crystallite-scale regeneration model [D. Bhatia, M.P. Harold and V. Balakotaiah, Catalysis Today 151 (2010) 314] is extended and new data are reported that provide insight about cyclic NOx storage and reduction (NSR). The model is based on the concept of NOx spillover from Pt to BaO and diffusion in the barium phase during storage and the reverse process during regeneration. The model is shown to predict the main features of NOx storage, such as the increase in NOx breakthrough time with increased Pt dispersion for fixed Pt loading. The increase in NOx storage with Pt dispersion is a result of (i) an increase in exposed Pt area which leads to a higher intrinsic NO oxidation activity, and (ii) an increase in the interfacial perimeter between Pt and BaO which promotes the rate of NOx spillover. These effects outweigh the known increase in activity with crystallite size. The model is used to simulate the complete lean-rich cycles in order to elucidate the effects of Pt dispersion on various cycle-averaged variables such as NOx/H-2 conversion and N-2/NH3 selectivity. The simulations show that a higher stored NOx diffusivity is needed to satisfactorily predict experimental conversion and selectivity trends. This finding suggests the possible involvement of enhanced diffusion, likely of the reductant, during the regeneration. The model is used to study various storage and regeneration timing protocols, such as the use of shorter cycle times to achieve a high cycle-averaged NOx conversion and NH3 selectivity for low dispersion catalysts. The model also predicts incomplete storage phase utilization both at the crystallite and reactor scales. For example, a reactor containing high Pt dispersion catalyst tends to utilize the storage phase effectively at the crystallite scale but can have significant axial storage non-uniformities, whereas a reactor containing low dispersion Pt catalyst tends to have a more axially uniform storage but poorer local utilization. (C) 2012 Elsevier B.V. All rights reserved.