Two-phase flows are regularly involved in the heat and mass transfer in industrial processes. To ensure the safety and efficiency of such processes, an accurate prediction of the flow field and phase distribution by means of Computational Fluid Dynamics (CFD) is required. Nowadays, Direct Numerical Simulations (DNS) of large-scale two-phase flow problems are not feasible due to the computational costs involved. Therefore, the Euler-Euler framework is often employed for large-scale simulations, which involves macro-scale modelling of turbulence, mass and momentum transfer. The research activities at Helmholtz-Zentrum Dresden-Rossendorf (HZDR) focus on general closure models for multiphase flows that are closer to physics and include less empiricism. As part of this effort, an Algebraic Interfacial Area Density model (AIAD) is developed for the morphology detection in the two-fluid Euler-Euler approach. Drag models for free surface flows are often based on experimental correlations, their applicability thus being limited to certain flow regimes. In this paper, a modified free-surface drag model based on local shear stress is investigated that avoids this limitation. For this purpose, the algebraic morphology detection mechanism of the AIAD model is revised. In DNS of free surface flow a dampening of the gas side turbulent fluctuations in the near surface region was found by previous investigators. This effect has also been accounted for in Euler-Euler simulations by means of dampening functions. In this work, the significance of turbulence dampening, in case of free surface flows, is examined quantitatively for the k-omega turbulence model. Model validation is performed with the commercial CFD code ANSYS CFX by means of experimental data of countercurrent, supercritical stratified air-water flow. The revised morphology detection mechanism is seen as an improvement with respect to the detection of sharp interfaces. Satisfactory quantitative agreement is achieved for the modified free surface drag model based on experimental pressure difference, liquid levels and interfacial shear stress. Furthermore, it is demonstrated that turbulence dampening has to be accounted for in the k-omega model to qualitatively reproduce the mean flow and turbulence quantities from the experiment. More CFD grade experimental data is required for further model validation. (C) 2015 Elsevier Ltd. All rights reserved.