A three-dimensional cylindrical model for non-linear propagation prediction in lined intake ducts with uniform flow

Abstract

Aircraft community, cabin and cockpit noise are important design requirements for aircraft manufacturers. One of the main aircraft noise sources is the engine fan noise. It can be divided into two main components, Forward and Rearward fan noise. One of the tonal components of the forward fan noise is called the Buzz Saw Noise (BSN). It is mainly generated at take o conditions when the relative Mach number at fan blade tips become supersonic giving rise to shock waves that propagate upstream in the air intake duct. Given the high amplitude of such shock waves the propagation is highly non-linear and thus non-linear propagation theory have to be implemented for the BSN prediction. The state of the art in-duct propagation codes used in the aircraft industry, which are mainly used for acoustic liner optimization, do not take into account non-linear e ects. It is thus important to develop nonlinear codes in order to improve acoustic liners performance prediction at supersonic relative Mach number conditions. In a recent publication [1] a validation of a one dimensional analytical code for the non-linear propagation in lined intake ducts with ow against a scaled test rig measurements was presented. The comparison between the predictions and the measurements showed satisfactory results in both the hard-wall and the lined con gurations. However, it has been noticed that the one dimensional code underestimates the energy transfer from the Blade Passing Frequency (BPF) and it's harmonics to the other Engine Orders (EO) which is one of the main phenomena associated with non-linear in-duct propagation. In order to improve the predictions, a three-dimensional cylindrical model has been developed. This new model is similar to a model presented by Fernando and al. [2] [3] but takes into account uniform ow conditions. A linear code has also been developed in order to take into account liner attenuation. This linear code was coupled to the non-linear model using a split-step approach [4] similar to that used for the one-dimensional model proposed in [5] and validated in [1].