Review of Computational Aeroacoustics in Propulsion Systems

Abstract

Recenteffortsfocusingoncomputationalaeroacousticsinpropulsionsystemsarereviewed.Dife cultiesassociated with a e nite difference solution of the time-dependent governing equations and boundary treatments are briee y discussed. Success and limitations of the large-eddy-simulation (LES) approach in which the sound source and the radiation e eld are simultaneously obtained are presented. It is suggested that LES be limited to the near e eld, and that other techniques be used to extend the near e eld to the far e eld. Several extension techniques are given. Approximate techniques for fast prediction of the source regime are reviewed. This is followed by discussing the coupling between the engine internal e ow and the jet plume noise. Apart from empirically relating noise to mean-e ow parameters, 5 earlyattemptsforthe predictionofjetnoisewerebased onmodeling the time-dependentsound sourcein theneare eldbysemi-analytical solutions. Acoustic analogy or asymptotic methods are then used to calculate the noise e eld associated with this source solution. 6i12 Recently, the direction for jet-noise prediction has shifted toward the application of computational aeroacoustics (CAA) to calculate the jet' s unsteady e ow and its radiated sound. We reviewherein e rst-principles approaches forthe prediction of noise in propulsion systems, with focus on jet noise and internal en- gine noise. In Sec. II, the numerical issues associated with CAA are discussed. In Sec. III, the large-eddy-simulations (LES) approach will be presented. For the three-dimensional case, current computer capabilities make direct LES of near and far e eld prohibitive. It is more practical for the three-dimensional case to restrict LES to the near e eld, and to use extension techniques to obtain the far-e eld sound. Such extension techniques will be discussed in Sec. IV. Ap- proximate techniques to obtain fast prediction of the source region arediscussedinSec.V.InSec.VI,wediscussengines' internale ow, which precedes the jet plume exit and plays a key role in controlling far-e eld noise. streamwise development of the jet can be split into three regimes. In the potential core regime, the shear layer, formed at the nozzle lip, spreads and reaches the centerline of the jet, marking the end of the potential core. The mean-e ow centerline velocity is constant within thiscore.This is followed by the transitional regime until the fully developed regime is reached. A computation domain needs to extend60D downstream before the centerline velocity has consid- erably decayed. Here, D is the nozzle diameter. In measuring the acoustic e eld, the microphone is usually placed at a circle centered at the jet exit of at least a 40 D radius. Thus, the computational do- main must extend radially to about 40 D. Though the mean e ow of the jet may be axisymmetric, its unsteady structure is three dimen- sional, and, in supersonic jets, the e rst helical mode could dominate over the axisymmetric mode. Threedistinct scales can be identie ed. In the acoustic e eld, the disturbance scales with the acoustic wave- length. In the jet e ow, the structure can be classie ed into two. One scales with the nozzle diameter and is sometimes called the jet col- umn mode. 14 The other, sometimes called the shear-layer inability mode,scales with the initialmomentumthicknessofthe shear layer,