A Comparison of Volume of Fluid and Euler–Euler Approaches in Computational Fluid Dynamics Modeling of Two-Phase Flows With a Sharp Interface

Abstract

Abstract In aero-engines, it is important to predict the behavior of shear flows in the different parts such as bearing chambers or gearboxes. In bearing chambers, the thickness distribution of wavy films is well studied as two-phase flows are still very hard to predict depending on the case. Experimental studies remain very expensive to carry out and computational fluid dynamics (CFD) still struggles with two-phase flow prediction especially when a sharp interface between the two phases must be modeled. CFD is used to predict the oil film thickness distribution and interface velocity at different engine operating conditions. Currently, Reynolds-averaged Navier–Stokes (RANS) CFD uses a semi-empirical method of turbulence damping, which is inaccurate for wavy films and so impacts the modeling of bearing chambers and gearboxes. With the objective of improving RANS models from large eddy simulation (LES) methods, the volume of fluid (VOF) and Euler–Euler methods for two-phase flow modeling are investigated in this study. The VOF approach assumes a single set of momentum equations for the two phases and volume fractions are 1 or 0 everywhere except in the interface region. An alternative to VOF is the Euler–Euler method with interface sharpening for shear flows. This approach assumes one set of momentum equations per phase but a shared field of pressure. The VOF and Euler–Euler approaches are compared in this study using LES with the CFD code OpenFOAM v6. The case study is based on experimental work investigating stratified flow in a horizontal channel that will be further detailed in this paper. In this study, a simplified 3D periodic channel filled with two distinct phases—air and water—is used. A flow regime is studied in which flows are fully developed and the water phase has a much smaller velocity than the air phase in order to obtain a shear flow. Numerical results are compared with experimental measurements from the literature. With OpenFOAM, the VOF solver used for the study is interFoam and the Euler–Euler solver used is reactingMultiphaseEulerFoam. Velocity profiles, shear–stress profiles, and kinetic energy profiles are compared with experimental measurements for the assessment of the two flow solvers. Maps of vorticity magnitude are also provided to support the comparisons between the Euler–Euler and the VOF approaches as well as an appropriate vortex identification method.