Use of detached eddy simulation for aerodynamics and aeroacoustics of blade sections

Cengiz, Kenan
Investigation of noise generation mechanisms due to turbulence necessitates resolution of eddies in space and time. Among the broad-band noise simulation tools, direct numerical simulation (DNS) is the most comprehensive one. However, it is prohibitively expensive. At the other extreme, unsteady Reynolds-averaged Navier-Stokes (URANS) based solvers, which are widely used in industry, can merely be reliable for attached flows. Besides, the inherent time-averaging procedure destroys the unsteadiness of eddies in most of the scales. Moreover, large-eddy simulation (LES) is still expensive in Reynolds numbers of industrial interest. At that point, use of hybrid RANS/LES approaches come to aid for capturing broad range of spectrum at acceptable costs. As a non-zonal variant, detached-eddy simulation (DES) has increasingly become useful in determination of noise generation mechanisms. There are commercial codes with DES modules. However, because of highly dissipative and dispersive low-order schemes of such codes, direct simulation of noise requires extremely fine meshes. Therefore, the aim of this thesis is to develop a low-dissipative low-dispersive high-order finite volume code to solve the compressible Navier-Stokes equations with DES capability, which will enable resolving eddies that are responsible for aeroacoustic noise generation around bodies. Thanks to an enhancement over DES, the model becomes more viable in attached flow problems, with a swifter switch to LES mode towards the outer boundary layer. Several validation studies reveal the solvers low-dissipation qualities. Finally, noise from a wing section is investigated. This is an important step towards design of quieter wind turbine blades. Ffowcs Williams and Hawkings acoustic analogy is utilized for prediction of the noise at far locations. Both the aerodynamic and aeroacoustic results show good agreements with the benchmark data, for markedly less computational cost than an LES study.