Building on the strong belief that the advancement and consistent adoption of cutting-edge simulation tools is critical to the future of nuclear power, three-dimensional thermal-hydraulic methods in the form of computational fluid dynamics (CFD) have made enormous advancement and promise to transform the way we approach the design of more efficient and reliable systems. The success of these methods hinges on the accuracy and predictive ability of the underlying models, which must, at the same time, limit the computational cost and allow optimal scalability. A large effort at the Massachusetts Institute of Technology has been devoted to the development of a second-generation of multiphase-CFD (M-CFD) closures and to leveraging the continuous progression in the experimental techniques. Among the numerous objectives, the central challenge that has driven the overall approach is the prediction of departure from nucleate boiling. This work focuses on deriving the fundamental meso-scale mechanisms from the CFD-grade experiments and incorporates them in the M-CFD framework as subgrid-scale models. A more complete representation of lateral lift force and near-wall effects are proposed, in combination with direct numerical simulation–driven understanding of bubble-induced turbulence effects. The improved description of the multiphase flow distribution is coupled to a novel representation of boiling heat transfer, which aims at introducing all the physical mechanisms that are encountered at the boiling surface. Starting from the improved representation at the wall, this work concentrates on the micro-hydrodynamics of the thin liquid film on the heated surface, which governs the critical heat flux limit.