Abstract:
Current work investigates the sensitivity of DES type hybrid methods toward turbulence length scales and anisotropy of grid. Different length scales available from literature are implemented into a compressible, finite volume GPU accelerated Navier-Stokes solver. A novel length scale is proposed based on the properties of available length scales and the grid requirements in mildly separated flows. Predictive capabilities of RANS and DDES method with the conventional and proposed length scales is assessed by conducting static and dynamic stall simulations on SC1095 and modified VR12 airfoils. Two different grids are used to examine the boundary layer resolving capabilities of the turbulence length scales. RANS based Spalart-Allmaras model and DDES method with conventional length scale predicted excessive eddy viscosity in the boundary layer of flows operating in near-stall regime. Proposed length scale demonstrated good predictive capabilities in mildly separated flows by reducing eddy viscosity levels at the outer region boundary layer. Three dimensional dynamic stall simulations are also conducted on flows over modified VR12 airfoil at moderate reduced frequency. When using proposed length scale, DDES method captured multiple lift peaks before stall and enabled flow to enter deep stall. The predictions agreed well with experimental data and captured cycle-to-cycle variation of integrated aerodynamic quantities. This work further employs specialized treatment to model laminar-turbulent transition and very low Mach number flows.