Sensitivity of Rotor Aeroelastic Predictions with Two-Equation Turbulence Models

Abstract

The capabilities of modeling complex rotor aeromechanics, principally via Computational Fluid Dynamics - Computational Structural Dynamics (CFD-CSD) coupling, have significantly improved with advancements in both computational hardware and software. Of particular interest is the accurate turbulence modeling for rotating blades in a variety of conditions, which remains the primary uncertainty in the prediction of rotorcraft aeromechanics. Recent collaborative investigations where different solvers and turbulence models are applied have yielded for some flight conditions dissimilar results while in other flight conditions, they produce comparable results. An investigation to further elucidate the root causes of this behavior has been undertaken with the series of two-equation models based on the Kok k-? and Menter k-? SST equations. The role that Delayed Detached Eddy Simulation (DDES) plays when added to these turbulence models is also of interest. These studies have been performed for both rotating and nonrotating configurations, applying a common solver, mesh, and time step. Both prescribed motion and aeroelastic applications have been examined for rotor flight conditions where the dichotomy between the two models was previously observed. Results indicate that the differences in the models occur when separated flows are present. Purely aerodynamic differences observed with rotor prescribed motion are compensated with structural dynamics when fully trimmed aeroelastic predictions are computed. The Kok k-? turbulent eddy viscosity exhibited numerical growth that resulted in solver instability when large separation was encountered. The addition of the SST correction mitigated this growth, but DDES did not. Thus for these two turbulence models, if the same options (SST and/or DDES) are applied, differences can be attributed to the solvers, solver options, and/or meshes used in the simulation.