RANS simulations are known to suffer from serious deficiencies in the prediction of jet in a crossflow (JCF) because of the high complexity of this kind of flow. Particularly, the coherent structures resulting from the interaction of the two flow streams are characterized by a highly unsteady and anisotropic behavior which hardly stresses the hypotheses underling common eddy viscosity models (EVMs). Direct numerical simulation (DNS) and large eddy simulation (LES) methodologies are still excessively computationally intensive to be used as ordinary design tools. Therefore, the development of reliable RANS turbulence models for film cooling flows deserved a great deal of attention from the gas turbine community. Computations presented in this work were carried out using a modified k-ε turbulence model specifically designed for film cooling flows. The model, due to Lakehal et al., is based on the usage of an anisotropic eddy viscosity. The model has been implemented in the framework of a CFD commercial package through the user subroutine features. Computational model is developed following the suggestions of Walters and Leylek concerning the correct representation of the problem geometry and the location of the boundary conditions. The predictive capabilities of the model concerning the ability to capture the main flow structures as well as heat transfer features are investigated. Comparison of computed adiabatic effectiveness profiles with experimental measurements is provided in order to quantitatively validate the model. Results obtained with standard EVMs, particularly a two layer standard k-ε model, are also shown in order to reveal the improvements in the predictive capabilities resulting from the modified models.
