Aerodynamic noise

Noise prediction methods are necessary in aspects of aerodynamic and hydrodynamic engineering. Predictive models of noise from rotating machinery ingesting turbulence is of much interest and relatively recently studied. This thesis presents a numerical method processed in a series of three codes that was written and edited to receive input for geometrical features of rotating machinery, as well as, adjustments to turbulent operating conditions. One objective of this thesis was to create a platform of analysis for any rotor design to obtain five parameters necessary for noise prediction; 1) the hydrodynamic inflow angle to each blade section, 2) chord length as a function of radius, 3) the cylindrical radius of each blade section, 4) & 5) the leading edge as a function of span in both the rotor-plane and as a function of axial distance downstream. Another objective of this thesis was to use computational fluid dynamics (CFD), specifically by using a Reynold’s-Averaged Navier-Stokes (RANS) Shear Stress Transport (SST) 𝑘 − 𝜔 model simulation in ANSYS Fluent, to obtain the turbulent kinetic energy distribution, also necessary in the noise prediction method presented. The purpose of collecting the rotor geometry data and turbulent kinetic energy data was to input the values into the first of the series of codes and run the calculation so that the output spectra could be compared to experimental noise measurements conducted at the Stability Wind Tunnel at Virginia Tech. The comparison shows that the prediction method results in data that can be reliable if careful attention is payed to the input parameters and the length scale used for analysis. The significance of this research is the noise prediction method presented and used simplifies the model of turbulence by using a correlation function that can be determined by a one-dimensional function while also simplifying the iterations completed on rotor blade to calculate the unsteady forces.

This project is intended to demonstrate the current state of knowledge in the prediction of the tonal and broadband noise radiation from a Sevik rotor. The rotor measurements were made at the Virginia Tech Stability Wind Tunnel. Details of the rotor noise and flow measurements were presented by Wisda et al(2014) and Murray et al(2015) respectively. This study presents predictions based on an approach detailed by Glegg et al(2015) for the broadband noise generated by a rotor in an inhomogeneous flow, and compares them to measured noise radiated from the rotor at prescribed observer locations. Discrepancies between the measurements and predictions led to comprehensive study of the flow in the wind tunnel and the discovery of a vortex upstream of the rotor at low advance ratios. The study presents results of RANS simulations. The static pressure and velocity profile in the domain near the rotor's tip gap region were compared to measurements obtained from a pressure port array and a PIV visualization of the rotor in the wind tunnel.

This dissertation will consider the sound radiation from forward-facing steps and a three dimensional cylindrical embossment of very low aspect ratio mounted on a plate. Glegg et al (2014) outlined a theory for predicting the sound radiation from separated flows and applied the method to predicting the sound from forward-facing steps. In order to validate this theory it has been applied to the results of Catlett et al (2014) and Ji and Wang (2010). This validation study revealed that the original theory could be adjusted to include a mixed scaling which gives a better prediction. RANS simulations have been performed and used to support the similarities between the forward-facing step and the cylindrical embossment. The simulations revealed that the cylindrical embossment exhibits a separation zone similar to that of the forward-facing step. This separation zone has been shown to be the dominant source of noise on the forward-facing step in previous works and therefore was expected to be the major source of sound from the cylindrical embossment. The sensitivity of this separation zone to the different parameters of the flow has been investigated by performing several simulations with different conditions and geometries. The separation zone was seen to be independent of Reynolds number based on boundary layer thickness but was directly dependent on the height of the cylinder. The theory outlined in Glegg et al (2014) was then reformulated for use with a cylindrical embossment and the predictions have been compared with wind tunnel measurements. The final predictions show good agreement with the wind tunnel measurements and the far-field sound shows a clearly defined directionality that is similar to an axial dipole at low frequencies.

The aim of this thesis is to develop a theory for non stationary propulsor flow noise. The model which is proposed is based on Amiet's paper "Acoustic Radiation from an Airfoil in a Turbulent Stream" [1], which describes broad band noise when a simple model of airfoil interacts with a turbulent flow, under the assumption of stationarity. The Karhunen-Loeve method provides a set of modes which describe the turbulent flow without the assumption of stationarity. A method is described to obtain broad band noise calculations when the mean turbulent flow varies with time and produces non stationary turbulence. A comparison of the numerical results obtained with the results from the paper of reference [1] shows the characteristics of time varying sound radiation. The various mathematical formulae will give a starting point to the analysis of real time varying flows, which are not considered in this thesis.

Ji and Wang (2010) propose that the dominant source of sound from a forward facing step is the stream wise dipole on the face of the step and that sources acting normal to the flow are negligible. Sound radiation normal to flow of forward facing steps has been measured in wind tunnel experiments previously by Farabee and Casarella (1986, 1991) and Catlett (2010). A method for evaluating sound radiation from surface roughness proposed in Glegg and Devenport (2009) has been adapted and applied to flow over a forward facing step which addresses the sound normal to the flow that was previously unaccounted for. Far-field radiation predictions based on this method have been compared with wind tunnel measurements and show good agreement. A second method which evaluates the forcing from a vortex convected past surface roughness using RANS calculations and potential flow information is also evaluated.