Duct noise and pressure loss are two faces of the same thing

In ventilation systems, noise complaints and high pressure loss usually originate in the same places: sharp turns, abrupt contractions, part-closed dampers, badly positioned fans.

The cause is shared. When the flow separates and generates fluctuation, mechanical energy is converted to heat (pressure loss) and pressure fluctuations radiate (noise).

Two source types

Fan noise. A tonal component at the blade passing frequency and a broadband component. Set by blade count and speed; the clearance between the volute cutwater and the blade directly affects the tonal component. See the rotating machinery note.

Flow-generated noise. From turbulence at dampers, grilles, bends and contractions. It depends very strongly on velocity — enlarging the duct section lowers the velocity and reduces the noise markedly.

Silencers are not free

An acoustic silencer works by narrowing the section and adding surface, so it adds pressure loss. If that added loss forces the fan to run faster, fan noise rises and part of the gain is given back. Silencer design is therefore as much a flow calculation as an acoustic one.

The modelling approach

Solving the noise directly — computing the pressure waves the flow generates — is expensive and requires a scale-resolving approach.

What usually suffices in industrial practice is finding the sources: regions of high turbulent kinetic energy and velocity fluctuation are the noise sources. That can be read from a cheap RANS solution and is enough to guide geometric improvement.

If an absolute sound level is required, measurement or a detailed acoustic analysis is needed; CFD's comparative strength on this subject exceeds its absolute strength.

References

1. ANSI/AMCA Standard 210 / ANSI/ASHRAE Standard 51 — Laboratory Methods of Testing Fans for Certified Aerodynamic Performance Rating. AMCA International / ASHRAE.
2. Spalart, P. R., Jou, W.-H., Strelets, M. & Allmaras, S. R. (1997). Comments on the feasibility of LES for wings, and on a hybrid RANS/LES approach. In Advances in DNS/LES, Greyden Press, Columbus, OH.