CFD Methods for Shear Driven Liquid Wall Films

Shear driven liquid wall films or physically similar two phase flow phenomena can be found in a number of different industrial and engineering applications. Gear boxes, bearing chambers or combustors in aero engines, heat exchanger ducts, oil and gas production and transport in petrochemical industry are just a few examples where this phenomenon is present and has been studied for decades. The most common approach of modeling shear driven film flows consist of empirical correlations derived from simple experiments. This approach is reasonable but highly case dependent. The problem lies in the difficulty of achieving experimental data for cases of practical importance. For a more global approach in this respect, CFD can be a useful tool. Therefore the study presented in this paper is dedicated to explore the potential of modern CFD methods. All available multiphase flow models are analyzed for their applicability for subcritical shear driven wall films (no mass transfer, no droplet shedding/deposition from/to film). VOF is suggested to be the only available multiphase flow model applicable to shear driven flows. However, further investigations have revealed that VOF method in its original form is not suitable for the flow conditions leading to high interaction between the phases i.e. where the motion of slow moving heavier phase is dictated by the fast moving lighter phase. This shortcoming in the VOF method is explained by means of a false momentum transfer between the phases. The focus then turns to find the improvement possibilities in the VOF method. Two approaches can be found in literature addressing the improvement possibilities in VOF method. The approach of physically justified modification of the turbulence quantities at the gas-liquid interface is adopted in this paper and is referred to as interface treatment. The approach is applied to a simple test case where the liquid phase acts as a wall. The results achieved for this test case are compared to the validation data where remarkable improvements are observed when compared to the VOF method without interface treatment. The interface treatment is then applied to a case of more practical importance where improvements are clearly evident again. Due to the lack of quantitative information on the interfacial waves, outlet boundary conditions cannot be well defined at this point. Therefore the later case is only seen as a motivation for further investigation of this approach.