In aero engines, the oil and air interaction within bearing chambers creates a complex two-phase flow. Since most aero engines use a close-loop oil system and releasing oil out is not acceptable, oil-air separation is essential. The oil originates from the engine transmission, the majority of which is scavenged out from the oil pump. The remainder exits via the air vents, where it goes to an air oil separator called a breather. In metal-foam-style breathers separation occurs by two physical processes. Firstly the largest droplets are centrifuged against the separator walls. Secondly, smaller droplets, which tend to follow the main air path, pass through the metal foam where they ideally should impact and coalesce on the material filaments and drift radially outwards, by the action of centrifugal forces. Although these devices have high separation efficiency, it is important to understand how these systems work to continue to improve separation and droplet capture.
One approach to evaluate separation effectiveness is by means of Computational Fluid Dynamics. Numerical studies on breathers are quite scarce and have always employed simplified porous media approaches where a momentum sink is added into the momentum equations in order to account for the viscous and/or inertial losses due to the porous zone [1]. Furthermore, there have been no attempts that the authors know of to model the oil flow inside the porous medium of such devices.
Normally, breathers employ a high porosity open-cell metal foam as the porous medium. The aim of this study is to perform a pore-level numerical simulation on a representative elementary volume (REV) of the metal foam with the purpose of determining its transport properties. The pore scale topology is represented firstly by an idealized geometry, namely the Weaire-Phelan cell [2]. The pressure drop and permeability are determined by the solution of the Navier-Stokes equations. Additionally, structural properties such as porosity, specific surface area and pore diameter are calculated. The same procedure is then applied to a 3D digital representation of a metallic foam sample generated by X-ray tomography scans [3]. Both geometries are compared against each other and experimental data for validation. Preliminary simulations with the X-ray scanned model have tended to under predict the pressure drop when compared to in-house experimental data. Additionally, the few existing studies on flow in metal foams have tended to consider laminar flow; this is not the case here and this also raises the question that Reynolds-averaged turbulence models might not be well suited to flows at such small scales, which this paper considers.
