The prediction of jet noise from CFD data

This paper describes a methodology for the prediction of jet noise based on data from a two-equation turbulence model. The noise model is an acoustic analogy based on the linearized Euler equations. Equivalent sources are included in both the continuity and momentum equations. Mean flow acoustic interaction effects are based on high and low frequency asymptotic solutions to both Lilley's equation as well as the linearized Euler equations. The range of validity of these solutions is discussed. Comparisons are made between predictions and measurements for a high subsonic unheated jet. It is shown that reasonable predictions can be made at all observer angles without recourse to a second source mechanism. All the observed effects on the noise spectrum can be explained by mean flow acoustic interaction effects. It is argued that traditional convective amplification effects are relatively weak. Their existence depends on the particular choice of model for the two-point statistics of the noise sources. However, mean flow acoustic interaction effects result in a predicted behavior equivalent to convective amplification. It is shown that this amplification is not due to the source motion relative to the observer: but, it is due to the sound radiating into a mean flow that is in motion relative to the observer. In addition, it is argued that "self" and "shear" noise are not separate noise mechanisms: they are mathematical constructs associated with the reduction of the linearized Euler equations to a single wave equation.