Coupling of Mainstream Velocity Fluctuations With Plenum Fed Film Cooling Jets

Modern gas turbine engines require film cooling to meet efficiency requirements. An integral part of the design process is the numerical simulation of the heat transfer to film cooled components and the resulting metal temperature. Industry design simulations are frequently performed using steady Reynolds averaged Navier-Stokes (RANS) simulations. However, much research has shown limitations in the use of steady RANS to predict film cooling performance. Prediction errors are typically attributed to poor modelling of turbulent mixing. Recent experiments measuring time-accurate film cooling jet behavior have indicated unsteady jet motions in sweeping and separation-reattachment modes contribute to the dispersion of the cooling jet along the cooled surface and the resulting time-averaged distribution. This study identifies the physical phenomena acting on film cooling jets issuing from fan-shaped film cooling holes, including acoustic resonance, which drive the unsteady behavior. Turbulent velocity fluctuations in the stream-wise direction cause corresponding fluctuations in the film cooling jet blowing ratio, which in turn reduces the time-averaged film cooling performance compared to the steady behavior that would be predicted with time-averaged blowing ratio. The plenum film cooling supply geometry acts as a Helmholtz resonator. An unsteady RANS (URANS) simulation including unsteady forcing is compared to experimental data. Helmholtz frequency excitation causes film cooling jet motions that qualitatively match the experiment. Resonant behavior causes the periods of lower blowing ratio to contribute to coolant dissipation rather than increased surface coverage. Results from URANS simulations demonstrate that replicating the unsteady jet motion is an important step in film cooling predictions. Starting with a steady baseline prediction, the URANS model used in this study is observed to reduce the overprediction of lateral average effectiveness by more than 50%, underlining the advantages of modeling the unsteady components of the Navier-Stokes equations.