Hybrid RANS/LES Turbulence Modeling

Research Area: Acoustics

Two predominant statistical approaches to modeling turbulent flows are Reynolds-Averaged Navier-Stokes (RANS) and Large-Eddy Simulation (LES) models. RANS, while computationally inexpensive, models all scales of turbulence. Modeling of all scales of turbulence can result in erroneous simulations for complex, engineering relevant flows. LES, resolves all but the smallest scales of turbulence, leading to simulations that are high fidelity, but computationally too expensive within the engineering design phase. Hybrid turbulence models provide a practical solution by blending RANS and LES characteristics. This results in solutions that are higher fidelity than traditional RANS calculations at a fraction of the cost of LES calculations.

A novel hybrid RANS/LES turbulence model based on partial-averaging closure has been proposed for the simulation of unsteady turbulent flows. Wavenumber Adaptive Simulation blends RANS and LES by scaling turbulent viscosity locally and directly based on the ratio of unresolved-to-total turbulent kinetic energy. Unsteady RANS is postulated to resolve a fraction of the turbulent kinetic energy spectrum, while a properly resolved LES simulation will capture more than 80%. By introducing a characteristic length scale based on the local cell size, partial integration of a Von-Karman turbulent kinetic energy spectrum is performed for the estimation of unresolved-to-total turbulent kinetic energy. This ratio is used to compute a convex combination of RANS and LES turbulent viscosity for the closure of the Wavenumber Adaptive Simulation model. By scaling turbulent viscosity, additional scales of turbulence are resolved based on what the mesh can locally support.