Computational Fluid Dynamics (CFD) offers the opportunity to investigate physically and geometrically complex systems with high fidelity. Its applicability to multiphase flow, and particularly boiling heat transfer, is currently limited by the lack of appropriate closure models to describe all relevant phenomena. In this paper, we present an original subcooled flow boiling modeling framework for CFD, which aims at consistently and accurately characterizing the key physics that affect heat transfer at the boiling surface. The new framework introduces a fully mechanistic representation of heterogeneous boiling that improves numerical robustness and reduces sensitivity to closure coefficients. The proposed formulation is inspired by new experimental insight, and significantly extends the existing boiling models by capturing the effects of (i) the microlayer on surface evaporation, (H) the boiling surface, and (Hi) bubbles sliding along the boiling surface. A new statistical treatment of the location and mutual interactions of bubbles on the surface allows for mechanistic prediction of the dry surface area, an important quantity that affects the boiling heat transfer coefficient. This approach lends itself naturally to extension to very high heat fluxes, potentially up to the critical heat flux. An assessment and sensitivity study of the model is presented for a range of mass fluxes (500-1250 kg/m(2)/s), heat fluxes (100-1600 kW/m(2)), inlet subcoolings (5, 10, 15 K), and pressures (1, 1.5, 2 bars), demonstrating improved robustness and predictive accuracy at all tested conditions in comparison to traditional heat partitioning approaches, including high heat fluxes, where classic models often fail to converge. Lastly, the framework proposed here should not be viewed as another heat partitioning model, but rather as a general platform that allows incorporation of advanced models for each physical phenomenon considered, leveraging the growing insight generated by modern experimental diagnostics for boiling heat transfer. (C) 2017 Elsevier Ltd. All rights reserved.