Turbulence-resolving simulations for engineering applications

Doktorsavhandling, 2015

Most fluid flows of industrial interest are turbulent and their accurate representation may be of vital importance for the design process of new products. To date, steady RANS methods are usually employed for the simulation of turbulent flows of everyday engineering problems. These methods base the description of turbulence characteristics on mean-flow gradients and only provide a solution for the mean flow. However, there are applications that require instantaneous flow information, for which the use of unsteady, turbulence-resolving simulation techniques is indispensable. In this thesis, the latter have been applied to predict two flow problems of industrial importance. Additionally to providing the flow solution, the simulation method's capability of producing input data for subsequent multi-disciplinary analysis was evaluated. In the first case, hybrid RANS/LES methods were used for simulating the complex flow around a three-element airfoil with deployed high-lift devices. Instantaneous flow properties were extracted from the simulation via a sampling surface and served as input for a subsequent aeroacoustic analysis of the airfoil using acoustic analogies. It was found that the chosen hybrid RANS/LES simulation technique was well-suited for computing the flow. Moreover, decoupling the flow simulation and the noise propagation enables aeroacoustic analysis and farfield-noise prediction for complex geometries at relevant Reynolds numbers. The slat was confirmed to be a major contributor to high-lift noise. Careful placement of the sampling surface, so as to enclose all turbulent noise sources, seems to be of paramount importance, in particular for using the Kirchhoff analogy. The second case dealt with LES simulations of the atmospheric boundary layer above and inside forest regions. Also from these simulations, instantaneous turbulence data were extracted and used in subsequent fatigue-load calculations for a wind turbine. It was expected that the presence of a forest leads to stronger atmospheric turbulence and increased wind shear, compared to flow over low-roughness flat terrain. By simulating the atmospheric boundary layer with and without a forest, this expectation could be verified and it was possible to quantify the effect of the forest on the wind-turbine fatigue loads. It could be shown that typical loads are increased by a factor of almost three in terms of root-mean-square values and equivalent fatigue loads.