Empirical Correlations and CFD Simulations of Vertical Two-Phase Gas-Liquid (Newtonian and Non-Newtonian) Flow Compared Against Experimental Data of Void Fraction and Pressure Drop

Gas-Newtonian liquid two-phase flows (TPFs) are presented in several industrial processes (i.e. oil-gas industry). In spite of the common occurrence of these TPFs, their understanding is limited compared to single-phase flows. Different studies on TPF have focus on developing empirical correlations based in large sets of experiment data for void fraction and pressure drop which have proven to be accurate for specific condition that their where developed for, which limit their applicability. On the other hand, scarce studies focus on gas-non-Newtonian liquids TPFs, which are very common in chemical processes. The main reason for it is due to the characterization of the viscosity, which determines the hydraulic regime and flow behaviours on the system. The focus of this study is the analysis of the TPF for Newtonian and non-Newtonian liquids in a vertical pipe in terms of void fraction and total pressure drop using computational fluid dynamics (CFD) and compared directly with experimental measurements and empirical relationships found in literature. A vertical tube of 3.4 m with an internal diameter of 0.1905 m was used. Superficial liquid and gas velocities ranged from 0.32 to 2.34 and from 0.08 to 1.59 m×s-1, respectively. The mixture Reynolds number and void fraction ranged from 2000 to 69000 and from 0.11 to 0.52, respectively. The two-phase CFD model was implemented in Star CCM+ using the volume of fluid (VOF) model. It was found a relatively good agreement between the experimental measurements, the CFD results and the empirical relationships. In terms of void fraction, for Newtonian and non-Newtonian liquids, the empirical correlations perform much worse than the CFD simulations, error of 48 and 25 %, respectively, against the experimental data. In terms of pressure drop, for Newtonian and non-Newtonian liquids, the empirical correlations perform much worse than the CFD simulations, error of 29 and 19 %, respectively, against the experiment data. This shows that CFD can be used to predict relatively well void fraction and pressure drop compared against empirical correlations and they can be used for design and scale-up processes.