For the past twelve years, CEA has been developing a deterministic multi-purpose neutron transport code, APOLLO3, which is starting to be used for reactor studies. A classical two-step APOLLO3 calculation scheme is based on a first stage of two-dimensional infinite lattice calculations in fine transport, generating multi-parameter cross-section libraries used in the second stage of 3D core calculations. In the case of a large power reactor, the core calculation requires approximations that can differ in accuracy, depending on the type of application.

The reference calculation schemes of the SHEM-MOC type and the industrial schemes of the REL2005 type, still in use at the lattice stage by CEA and its industrial partners, EDF and Framatome, were developed in the mid-2000s, based on the methods available in the APOLLO2.8 code. Since then, new methods have been implemented in the APOLLO3 code, which have been individually verified and validated, demonstrating their ability to improve the quality of results at the lattice stage. These include new self-shielding methods, subgroups and Tone, the use of surface line sources in flux calculations using the method of characteristics, flux reconstruction for burnup calculations and a new 383-group fine energy mesh.

The aim of this thesis is to define and validate two new lattice calculation schemes for LWR applications to be used in future calculation tools at CEA and its partners. The goal is to integrate all or part of the new calculation methods, while aiming for reasonable calculation times for the reference scheme, and compatible with fast-running routine usage for the industrial scheme. The calculation schemes implemented will be validated in 2D on geometries taken from the VERA benchmark. Validation will be carried out using an innovative approach involving continuous-energy or multi-group Monte Carlo calculations and a perturbation analysis.