Computational Fluid Dynamics and Computational Aeroacoustics for Turbomachinery Applications with emphasis on High Speed Propellers and Vertical Axis Wind Turbines

Abstract

Turbomachinery aerodynamic and aeroacoustic modelling is recognized to be one of the main areas of interest in the frame of industrial engineering, due to the large number of applications concerning the rotating fluid machines. Numerical modelling is today widely integrated in the design and development process of industrial components, as well as Computational Fluid Dynamics (CFD) is a tool well integrated into the industrial development and production life-cycles. This has been made because of two main factors: the increase in the performance of relatively cheap computing systems and the progress made by development of integrated CAE virtual environment even more oriented to design and optimization. Furthermore well established and validated CFD methodologies able to deal with the high complexity of geometries and multiphysics approaches are available and practicable within the industrial environment. Nevertheless since for many of applications such as aerodynamics, heat exchange, etc., CFD is largely used, there are still some areas where it represents a challenge, in particular for its industrial applicability. One of these areas is the Computational AeroAcoustics (CAA), whose interest is rapidly growing due to the increasing demand of quick and efficient numerical tools for the design-to-noise approach to airframe components. The major difficulty of CAA is that noise is a secondary effect of other physical phenomena, and noise prediction has to deal with highly complex topics such as turbulence modelling and transition to turbulence prediction. Moreover the sound pressure compared with the hydrodynamic pressure is several orders of magnitude smaller making its numerical prediction very difficult. The goal of the present work is to perform investigations on two rotating fluid machines: the vertical axis wind turbines (VAWT) and the high speed transonic cruise propellers. The investigations performed were focused on the aerodynamic and aeroacoustic theoretical and numerical modelling in order to achieve the numerical results that actually miss in the literature improving the know-how for two classes of problems which play a central role in the modern industry: the renewable energies and the advanced transportation systems. VAWT modelling was approached from a theoretical point of view, by the implementation of a modern unsteady aerodynamic model for non-stationary airfoil aerodynamic coefficients prediction (the Beddoes Leishman model) into a classical theory, the Double Streamtube Paraschivoiu model. The acoustic modelling of WT was approached on a general basis with the modelling of the airfoil self-generated problem. It was chosen as it is one of the most prominent noise sources for wind turbines and it represents a big challenge for CFD. The preliminary literature review pointed out how there are different airfoil self-generated noise typologies, according to the laminar/turbulent characteristics of the flow field, airfoil aerodynamic parameters and far-field flow conditions. Different CFD numerical approaches were used for the different noise generation mechanisms, including transient RANS/LES simulations with a particular focus for novel turbulence modelling methodologies, involving transitional models and Embedded LES. The acoustic pressure flow computations were performed with the Ffowcs Williams Hawkings (FW-H) acoustic analogy and pressure time signals post-processed with an in-house built acoustic tool. Deep explorations of acoustic signal noise processing, Fast Fourier Transformation techniques and noise signal corrections have been necessary in order to achieve good agreement between experimental and computational data. The reference test-case was the NACA 0012 airfoil, chosen for the wide amount of experimental data available in literature in a significant range of velocities and angles of attack. The problem of propeller aerodynamic and aeroacoustic modelling was approached with the CFD and CAA simulation of a subsonic propeller, the NACA 4-(3)(08)-03, chosen for the experimental results available in literature for the 2-bladed and 7-bladed configurations. This test-case allowed us to set up properly the numerical simulation, the CAD model and the computational grid requirements for such kind of geometries. In a second stage the attention was focused on the NASA SR2 and SR3 geometries, both in the 8-bladed configuration. Experimental data in transonic conditions for wind tunnel and flight test were available for these two geometries, and numerical simulations were performed for 5 different data sets, for asymptotic Mach numbers between 0.6 and 0.9 and rotational speeds between 6000 and 9000 rpm. Results of the simulations are compared with experiments, showing the ability of this approach to predict noise with a discrepancy within a few dB for the different simulated conditions and microphones locations. Particular attention is given to the set of corrections to be applied to acoustic experimental data in order to be consistently compared with free field CFD results. The CFD simulation strategy has been refined to perform complete aerodynamic and aeroacoustic calculations with highly competitive computational cost.