Part 2: Simulating Aeroacoustics: From CFD Sources to Acoustic Propagation

In Part 1, we looked at what aeroacoustics means and why it is such a good example of multiphysics. The basic idea is simple: moving air creates pressure fluctuations, and some of those fluctuations become sound. The engineering challenge begins when we try to predict that sound accurately enough to guide design decisions.

That is where aeroacoustic simulation becomes important. Engineers do not always need the most expensive simulation possible. They need the method that best answers these questions: Are we screening multiple designs? Investigating a specific noise source? Studying how sound propagates through a duct? Are we trying to understand what a listener will hear at a specific location? Do we need to guarantee that our design is compliant with regulations? Do we want our product to have a specific sound associated with it?

The First Question: What Are You Trying to Predict?

Aeroacoustic workflows usually start with the same practical question: where does the noise come from, and where does it go? The source is often tied to turbulence, rotating blades, separated flow, jets, or pressure fluctuations near surfaces. The propagation path may include free-field radiation, ducts, cavities, mean flow, temperature gradients, absorbing materials, or nearby structures.

Once the source and propagation path are clear, the simulation strategy becomes easier to choose. In practice, most aeroacoustic workflows fall into four broad families.

1. Direct Methods: Let CFD Resolve the Sound

Direct methods use the computational fluid dynamics (CFD) solver to compute the acoustic solution directly. In other words, CFD does it all: the same simulation model accounts for both the turbulent flow field and the acoustic contributions. That makes the approach conceptually attractive because there is no separate acoustic model to manage.

The trade-off is computational cost. Accurately resolving acoustic waves requires fine CFD meshes and transient simulations. With a standard Navier–Stokes-based CFD code, that often means running long LES or WMLES simulations in the noise-critical regions. Lattice-Boltzmann-based methods can often handle this class of aeroacoustic problem more efficiently, but direct simulation is still usually reserved for cases where high fidelity is essential, and the compute budget allows it.

2. Integral Methods: Use CFD Near the Source, Then Predict Far-Field Noise

Integral methods, such as the Ffowcs Williams–Hawkings approach, use the CFD solver to compute the transient flow behavior and then apply an acoustic analogy to calculate acoustic pressure at observation points. The method generally separates the source region from the propagation path and the receiver, which makes it useful when the main interest is far-field noise.

The major advantage is efficiency: velocity or pressure fluctuations computed by CFD can be propagated to the far field without refining the CFD mesh throughout the entire domain. The drawback is setup sensitivity. Engineers must know how to define the measurement or integration surfaces because separating acoustic content from hydrodynamic fluctuations is not always straightforward.

3. Hybrid Methods: CFD for the Flow, Acoustics for the Sound

Hybrid methods are where many day-to-day aeroacoustics problems live. They rely on a practical assumption: the flow generates sound, but the sound does not significantly change the flow. When the one-way coupling assumption holds, the fluid and acoustic computations can be decoupled without significant loss of accuracy.

The workflow is then split into clear stages: CFD computes the flow field, aeroacoustic sources are extracted from the CFD results using an analogy such as Lighthill or Möhring, and an acoustic solver propagates the sound by solving the wave-based equations using methods such as finite elements or the Discontinuous Galerkin Method.

When the assumption holds, hybrid methods can deliver accuracy close to direct methods while giving each solver a focused role. CFD resolves the dynamic fluctuations in the noise-critical flow region, often without requiring a fully compressible formulation. The acoustic solver then handles propagation effects such as resonators, absorbing materials, structural vibrations, and complex receiver environments. The trade-offs are that it is a two-model workflow and it still requires an unsteady CFD solution, such as LES, in the noise-critical region. Overall, hybrid methods perform best when engineers want to understand how geometry and materials affect noise.

4. SNGR-Type Methods: Faster Screening with Synthetic Turbulence

Sometimes the engineering question is not “What is the exact sound field?” but “Which design is likely to be quieter?” That is where semi-empirical or SNGR-type methods become useful. They are designed for situations where engineers need to explore multiple design variants but do not have time to run full unsteady CFD simulations for each one.

Conceptually, SNGR says: use a steady simulation to learn the statistics of the turbulence, then construct synthetic turbulence that matches those statistics. In practice, engineers run a RANS simulation, analyze the mean flow and turbulence statistics, assume or approximate a turbulence spectrum, and generate synthetic time-dependent velocity fluctuations that are fed into the acoustic solver.

The benefit is speed. Because the CFD requirement is a lighter RANS simulation, SNGR-type methods are useful for quickly comparing alternatives, such as whether design A is likely to be quieter than design B. The price is lower accuracy at low frequencies and, like hybrid methods, a two-model workflow.

In some cases, the noise source can also be described analytically when the source follows a repeatable pattern. Fan and rotor noise are good examples: peaks occur at blade-pass frequencies, while the amplitudes vary with geometry and operating conditions. Simple CFD calculations can help define those amplitudes and make the acoustic prediction more practical.

Choosing the Right Method

So, how do specific tools turn all this theory into something usable? If you are new to aeroacoustics, start with Part 1 to learn the basic concepts of airflow, pressure fluctuations, and sound.

If you are ready to see these methods become a practical simulation workflow, continue to Part 3, where we look at Cradle CFD and Actran in action.

Written by Vicky Tsianika, Product Management Director, and co-authored by Nicolas Driot, Senior Principal Product Manager.