Part I of this study introduced a novel methodology for the optimization of energy group structures to be used in diffusion calculations of Sodium cooled Fast Reactors. The goal of the optimization is to speed up calculations, particularly in transient analyses, while maintaining an acceptable accuracy of the results. The proposed methodology is based by either direct or simulated annealing search techniques. The capabilities of the method were preliminarily demonstrated on a single core state of the Superphenix reactor by using the DYN3D nodal diffusion code. The scope of Part II is the further demonstration of the methodology efficiency through its application to more challenging "real-life" cases. In this respect, a static Superphenix neutronic benchmark comprising 13 different core states and a transient test initiated by an increase of the core inlet temperature at the Phenix reactor are considered. For both Superphenix and Phenix reactor cores, 2-to 12-group optimal condensed energy group structures are identified using the 24-group structure as a starting point. The obtained optimal structures are thus employed in DYN3D and compared to the reference 24-group DYN3D solutions. The outcomes show that also for a broader range of core configurations and under transient conditions the optimal energy group structures allow for significant acceleration of the DYN3D performance with practically negligible degradation of the solution accuracy. (c) 2021 Elsevier Ltd. All rights reserved.