SERPENT2 and CMS5 depletion modelling of the PRATIC SMR startup core benchmark

Soluble-boron-free SMR cores rely strongly on burnable absorbers and control rods for reactivity management, making depletion predictions particularly sensitive to neutronic modelling approximations. This work compares the two-step CMS5 deterministic code system with a continuous-energy SERPENT2 Monte Carlo model of the PRATIC startup cycle, from assembly to three-dimensional full-core depletion. Both All Rods Out and pseudo-critical configurations are considered; in the latter, SIMULATE5 critical rod insertions are imposed on the SERPENT2 model. The deterministic energy condensation and scattering anisotropy order are investigated. At the infinite-medium assembly level, P 3 scattering is sufficient and CASMO5 is converged in energy at 586/95 groups, the gain over 586/19 groups being mainly confined to the gadolinium depletion window. At the core scale, refining the condensation beyond 586/19 groups brings no benefit for the CASMO5 MxN calculation or for SIMULATE5 cross-section generation. The SIMULATE5 solution is driven by the number of energy groups at the core scale: four- and eight-group results agree closely and are closer to SERPENT2 than two-group results at beginning and end of cycle. The largest reactivity discrepancies with SERPENT2 occur during gadolinium depletion, reaching about 385 pcm in 2D CASMO5 MxN and 600 pcm in 3D SIMULATE5, the latter reflecting the additional nodal diffusion approximation. They are reduced neither by finer condensation nor by more core groups. The black reflector boundary condition lies systematically closer to SERPENT2 than the infinite reflector. The SERPENT2 model provides a comparison basis for the PRATIC benchmark and illustrates the computational constraints of full-core Monte Carlo depletion.