Numerical study on the sensitivity of film cooling cfd results to experimental and numerical uncertainties

Film cooling is used in a wide range of industrial and engineering applications; one of the most important is in gas turbine cooling. The intent of film cooling is to provide a layer of cool film between the surface and the hot gas. Predicting film-cooling characteristics, particularly at high blowing ratios where the film is likely to be detached from the surface, is a challenge due to the complex three-dimensional and possibly anisotropic nature of the flow. Despite the growth of more sophisticated techniques for modeling turbulence, such as large eddy simulation (LES), the most commonly used methods in design are Reynolds-Averaged Navier Stokes (RANS) methods that employ a two-equation turbulence model for specifying the eddy viscosity. Although these models have deficiencies, they continue to be used throughout industry because they offer reasonable turnaround time as compared to LES or other methods. This paper studies in detail two cases, one of high blowing ratio (off-design condition) of 2.0 and low blowing ratio of 0.5, and compares RANS-based computational fluid dynamics (CFD) results with experimental data for flow field temperatures and centerline, lateral, and span-averaged film effectiveness for a 35-degree circular jet. The effects of mainstream turbulence conditions, boundary layer thickness, and numerical dissipation are evaluated and found to have minimal impact in the wake region of separated films (i.e., they cannot account for the discrepancy between measured and predicted CFD results in the wake region). Analyses of low blowing ratio cases are in good agreement with data; however, there are some smaller discrepancies, particularly in lateral spreading of the jet.