We develop a two-fluid Euler-Euler CFD framework based on the PolyMAC numerical scheme (Gerschenfeld and Gorsse, 2022) in CEA’s open-source TrioCFD code (Angeli et al., 2015). Interfacial momentum closure terms are selected and validated using bubbly adiabatic experiments on vertical flows (Colin et al., 2012; Hibiki et al., 2001). The local experimental bubble diameter is enforced to avoid the use of an interfacial area transport equation, as in Sugrue et al. (2017). Independently, it is shown that in a high-pressure developed boiling pipe flow, changing the entrance temperature while measuring flow characteristics at the outlet is equivalent to changing the distance from the inlet where the flow characteristics are measured. This enables us to simulate the DEBORA experiment (Garnier et al., 2001), an ascending boiling R12-freon flow in a tube, using a 3D map of the experimental diameter, avoiding distortions due to interfacial area modeling. We demonstrate that atmospheric-pressure closure terms are not able to reproduce measured void fraction profiles. We then consider that bubbles are deformed, i.e. non-spherical, in nuclear reactor conditions, characterized by high pressures, void fractions and flow velocities. Taking this deformation into account, we propose a new set of momentum and energy closures that is found to be independent of the bubble diameter. This enables us to run simulations without prior knowledge of the bubble diameter nor the need for an interfacial area transport equation or population balance model, as in system-scale codes extensively used in the nuclear industry (NRC, 2010; Berry et al., 2018). Void fraction predictions are more precise than with the baseline set of closures.