Two-phase turbulence is studied using DNS of upward turbulent bubbly flows in a plane channel. Fully deformable monodispersed bubbles are tracked by the Front-Tracking algorithm implemented in TrioCFD code on the TRUST platform. Two sets of fluid properties are used. Firstly, two simulations are performed with virtual fluids at a low void fraction of 3% and for a Reynolds friction number of 127 to benchmark our code against . Good agreements are obtained for both deformable and spherical cases. A third simulation closer to pressurized water reactor (PWR) conditions was performed at higher void fraction and Reynolds number to push the limits of DNS capabilities.
DNS results are averaged (i) to provide reference profiles for an up-scaling methodology towards RANS two-fluid models and (ii) to analyze the equilibrium between buoyancy, surface tension, viscous and turbulent shear at statistically steady-state. Surface tension forces and turbulence are essentials to capture the equilibrium. Their accurate modeling is the key to velocities and void fraction predictions in averaged codes. Our analysis reveals the important role of surface tension, not only in the determination of the bubble shapes, but also as a source of local imbalance of the momentum transfer between phases. More advanced models considering interfacial energy are necessary to predict these flows.