Buzz-saw noise prediction for axisymmetric and drooped turbofan intakes

Abstract

As the aerospace industry grows and the number of aircraft movements increases, the noise impact of air travel is under scrutiny in the media and subject to increasingly stringent regulatory requirements. High-bypass ratio turbofan engines used on modern commercial airliners feature a large diameter fan to produce a significant proportion of the overall thrust. During the departure phase, the fan is a major contributor to overall aircraft noise. At high engine speeds used during take-off and climb, the velocity of the fan blade tip relative to the incoming flow can reach supersonic speeds, generating a system of shock waves that propagate through the intake and radiate to the forward arc as noise. The propagation of the pressure field generated by a transonic fan is inherently complex. The shocks produced at the leading edge of the fan blade tips differ from blade to blade due to variation in the manufactured blade geometry and the ‘untwist’ of each blade under load. The shock pattern then becomes increasingly irregular as it travels through the intake through a highly nonlinear propagation process. In addition, the pressure field is attenuated by an acoustic liner fitted to the intake wall and modified through interaction with the distorted intake flow. At the end of the intake, the irregular shock pattern is comprised of tones at harmonics of the blade passing frequency and the shaft rotation frequency. This noise type is known colloquially as ‘buzz-saw noise’ due to it’s similarity to the noise produced by a circular saw. Models which accurately predict the noise produced by an engine are essential for the development of future engine technologies. Existing numerical and analytic models can individually capture a subset of the physical effects associated with buzz-saw noise. These methods include the classical analytic models of linear duct acoustics, semi-analytic models for nonlinear propagation in simplified ducts, linear Computational AeroAcoustics (CAA) and Computational Fluid Dynamics (CFD). While each of these methods has their merits, no single model can capture the nonlinear propagation, liner attenuation, interaction with a distorted flow field and radiation to the far-field. A final factor with considerable impact on a model’s prediction performance is the way a boundary layer is represented when modelling liner absorption. In many analytic and computational models, the boundary layer is modelled by an infinitely thin vortex sheet. This can result in an over-prediction of the liner attenuation for upstream noise propagation. In reality, a finite thickness boundary layer refracts the upstream propagating wave away from the liner, reducing the level of attenuation. In this work, a method for improving attenuation predictions is developed, by defining an ‘effective impedance’ which is informed by eigenmode solutions of the Pridmore-Brown equation in a duct with a finite-thickness boundary layer. The key aim of this PhD thesis is to develop fast and accurate buzz-saw noise prediction models which include all of the aforementioned physical features. The methods are ultimately intended to be applicable to three-dimensional intake geometries with distorted mean flow fields. To achieve the key aim, two coupled prediction models are developed and validated, initially against test data from an axisymmetric static rig intake. These methods incorporate a source prediction from RANS CFD (HYDRA) model with a linear CAA prediction (ActranTM) for the propagation and radiation. The difference in the two methods is the method by which nonlinear effects are captured; one method employs a one-dimensional nonlinear model (FDNS) to apply a nonlinear correction to the CAA predicted levels, while the other captures the nonlinear effect from the CFD within the source description provided to the CAA model. Both methods provide very realistic results in comparison to the rig intake data. Near the end of this thesis, the most appropriate of the two coupling methods is selected and extended to produce a prediction of buzz-saw noise in a realistic ‘drooped’ rig intake. The model couples a prediction from a large-scale unsteady RANS prediction to a three-dimensional CAA model of the intake, to include all effects of nonlinear propagation, liner attenuation with consideration of a finite thickness boundary layer, flow distortion and radiation to the far-field.

More information

Identifiers

ORCID for Rie Sugimoto: orcid.org/0000-0003-2426-2382

Catalogue record

Export record