Aeroacoustics

Computational aeroacoustics (CAA) is a branch of multiphysics modeling and simulation that involves identifying noise sources that are induced by fluid flow and propagation of the sound waves then generated.

Noise sources originate from various types of flow, such as:

Turbulent flow over solid bodies (bluff body flows)
Turbulent boundary layer flows (for example, automobile, aircraft components)
High-speed turbulent shear flows (for example, free jet flow)
High-speed impinging flows (for example, jet impingement, rocket exhaust noise)
Structural vibration that is induced by fluid flow (fluid-structure interactions)
High-speed rotating flows (for example, rotorcrafts or turbomachinery)
Turbulent combustion (reacting flows)
Blast waves (explosions)

A typical CAA simulation requires the following components:

Navier-Stokes equations for fluid flow
High-resolution turbulence models
Analytical or computational acoustic wave propagation models

The noise signatures at the locations of interest exhibit corresponding noise spectra—that is, the intensities of sound pressure level over a range of frequencies. The noise characteristic can be tonal noise with a distinct peak at a frequency (such as engine noise; jet impingement noise, or Noise, Vibration, and Harshness (NVH)) or broadband noise spread over a frequency range (typical of turbulence-induced noise).

The sound pressure level (SPL measured in decibel, dB) is:

Figure 1. EQUATION_DISPLAY

S P L = 20 {log}_{10} \frac{p_{r m s}}{p_{r e f}} d B

(4658)

where:

$p_{r m s} =$ root mean square pressure
$p_{r e f} =$ reference pressure (usually 20 µPa)

The mechanism of noise generation differs according to the underlying physics of the flow. When modeling noise generation using computational methods, it is essential to capture the noise sources and frequencies that are relevant to the acoustic analysis. Resolution of the noise sources relies on the fidelity of the turbulence modeling.

The computational aeroacoustics methods in Simcenter STAR-CCM+ target the following applications:

Prediction of near-field noise by Large-Eddy Simulations (LES) or Detached-Eddy Simulations (DES): Although LES and DES simulations can predict noise generation in the near-field, they are not generally suitable for predicting noise propagation to far-field locations. The flow fluctuations that generate noise are orders of magnitude lower than the hydrodynamic flow properties. Without adequate mesh resolution, sound waves quickly dissipate away from the source region. The mesh density that is required to preserve this acoustic signature at far distances restricts the feasibility of pure CFD for real world applications.
It is more practical to consider noise generation and propagation phenomena separately, adopting an appropriate hybrid CFD/CAA method. In this method, LES or DES is coupled (one-way) with a noise propagation aeroacoustic model.
Prediction of mid-field noise by acoustic wave equations, Linearized Euler Equations (LEE), or Acoustic Perturbation Equations (APE) (see Acoustic Wave Model): High fidelity LES/DES can be combined with other accurate mesh-based CAA methods (typically requiring high-order space and time discretization) such as Linearized Euler Equations (LEE) or Acoustic Perturbation Equations (APE), but not without extreme computational burden.
Prediction of far-field noise by the Ffowcs Williams-Hawkings equation (FWH) (see Ffowcs Williams-Hawkings Model): The Ffowcs Williams-Hawkings equation considers all fundamental components of noise sources (monopole, dipole, and quadrupole). The surface integral methods such as those of the FWH method are a practical approach for noise propagation in the far-field, being able to superimpose all sources of noise over an enclosed surface and provide acoustic signatures by accurate meshless analytical solutions of a surface integral equation.

The noise propagation models in Simcenter STAR-CCM+ are based on Lighthill's equation Eqn. (4705) [901], Curle's equation, and the Ffowcs Williams-Hawkings equation (FWH). Lighthill's equation derives the acoustic field from the Navier-Stokes equations and some basic assumptions, containing the definition of quadrupole noise sources from turbulent flows. Simcenter STAR-CCM+ uses LEE and Curle for noise source modeling only, not for acoustic wave modeling (LEE), or for propagation (Curle).

The FWH method is based on the assumption of sound propagation in free space, so it cannot consider any embedded structure in the propagation region. On the other hand, a Boundary Element Method (BEM) for the solution of the Helmholtz equation (a wave equation in the frequency domain) by a boundary integral equation can provide solutions of acoustic field with embedded structures. See [882], [886], [913].