How to Control HVAC Noise in Auditoriums

By Sarah Okonkwo · May 1, 2026

1) Project overview: what, where, who, and why

In late 2024, the City of Redbrook commissioned an acoustic and AV refresh for the Redbrook Civic Auditorium, a 620-seat, fan-shaped venue used for lectures, chamber music, and town meetings. The room was built in 1998, with a balcony, a shallow stage house, and a constant-volume HVAC system that had never been revisited beyond routine mechanical maintenance.

SonusGearFlow was brought in as the audio project documentarian and technical liaison between the mechanical contractor (Horizon Mechanical), the acoustical consultant (Riverton Acoustics), and the AV integrator (ClearSignal Systems). The “why” was simple: the auditorium had become known for a persistent low-frequency rumble and midband hiss that forced operators to run speech reinforcement hotter than desired. The city wanted improved intelligibility for council meetings and quieter ambient conditions for unamplified performances—without replacing the entire air-handling plant.

The scope was deliberately focused: diagnose noise sources, implement practical HVAC noise-control measures, and verify results with repeatable measurements. A secondary goal was to ensure any changes didn’t break thermal comfort or violate ventilation requirements.

2) Challenges and requirements at the outset

The noise complaints were consistent, but the causes were not. During initial walk-throughs, three distinct issues presented themselves:

Broadband “air rush” noise from ceiling supply diffusers at higher flow rates (most noticeable during cooling calls).
Low-frequency rumble that varied with fan speed and sometimes intensified when the auditorium doors were closed.
Structure-borne vibration audible as a faint buzz near the stage-left wall where a duct riser passed close to the proscenium.

At the outset, the city set target requirements aligned with typical auditorium expectations, without chasing studio-grade quiet:

Background noise target: NC-25 to NC-30, measured in the seating area with HVAC operating in “event mode.”
Speech intelligibility: maintain or improve STI, with the operational aim of reducing average A-weighted SPL needed for meetings by 3–6 dB.
Schedule: 10-week window from approval to commissioning, with a hard stop before a spring concert series.
Constraints: keep the existing rooftop air-handling unit (AHU) and duct mains; modifications had to fit above existing ceilings with limited plenum access in the balcony.

Baseline measurements were sobering. With the room empty, HVAC in normal daytime mode, and stage lighting off, we measured NC-38 to NC-42 in mid-house seating. The spectrum showed elevated energy around 125–500 Hz (air turbulence and duct-borne noise), plus a distinct hump around 63 Hz that correlated with fan operation. For reference, “quiet enough for speech” is often achievable around NC-30; NC-40 makes soft talkers and Q&A sessions noticeably harder.

3) Approach and methodology chosen

We approached the problem the same way we’d approach a gain-structure issue: measure first, isolate variables, implement changes that address root causes, then verify.

The methodology combined three tracks:

Acoustic measurement: 1/3-octave SPL and NC curves at multiple seats (front, mid, balcony) using a Class 1 measurement mic and logging SPL meter. We took readings with HVAC in multiple states: fans low/high, cooling/heating, and with the economizer open/closed.
Mechanical diagnostics: duct static pressure measurements, diffuser face velocity checks, and inspection of duct lining, turning vanes, and flex connections. We used a handheld anemometer at diffusers and a manometer at test ports.
Operational review: check building automation system (BAS) sequences. The existing “event mode” simply locked the space at 72°F but did not reduce airflow; it sometimes increased fan speed to recover temperature quickly after door openings.

From these, we built a short list of likely contributors: excessive air velocity at diffusers, inadequate attenuation between AHU and auditorium, and a mechanical vibration path through rigid duct connections near the stage.

4) Step-by-step execution narrative

Week 1–2: Baseline measurements and source localization

We mapped noise across 12 listening positions. Mid-house seats averaged 41 dBA with HVAC active; balcony seats were slightly worse due to closer proximity to a supply branch. When the AHU fan was temporarily forced to 60% (maintenance override), NC dropped from roughly 40 to 33, strongly implicating airflow and fan-generated duct-borne noise.

We also used a simple but effective technique: a mechanic’s stethoscope and handheld vibration meter on duct hangers and nearby framing. A clear spike appeared on the stage-left riser when the fan ramped above 75%. That pointed toward structure-borne transmission.

Week 3: Quick fixes and confirmatory tests

Before specifying hardware, we tested operational changes:

Reduced maximum fan speed in event mode from 100% to 75% (temporarily) and observed temperature recovery impact.
Adjusted discharge static pressure setpoint down by 0.3 in. w.g. to reduce turbulence at diffusers.

The room got quieter quickly—down to around NC-33 mid-house—but temperature drifted during a full-load rehearsal simulation (doors opening, 300 people equivalent heat load modeled by BAS settings). That validated the problem but confirmed we needed physical noise control, not just throttling airflow.

Week 4–6: Implement attenuation and airflow improvements

With the team aligned, the mechanical contractor executed three primary modifications:

Add duct silencers on the main supply and return runs feeding the auditorium.
Replace problematic diffusers and re-balance to reduce face velocity and regenerated noise.
Introduce vibration isolation where ductwork coupled into the building structure near the stage.

The silencers were selected as rectangular, passive, packless media silencers rated for low pressure drop. The supply silencer was sized at 48” x 24” x 60” equivalent with a published insertion loss favoring 125–1000 Hz, where our spectrum was most elevated. A slightly smaller unit was used on return to reduce fan noise recirculation. Both were installed above the lobby ceiling where access allowed and where the duct mains were straight enough to avoid compounded turbulence.

For diffusers, we replaced eight high-throw, 2’x2’ lay-in diffusers that had been selected for air distribution but were noisy at the required CFM. They were swapped for low-noise, perforated-face diffusers with better acoustical performance at similar flow. During re-balance, we targeted diffuser face velocities under 500 fpm in occupied zones, down from measured peaks closer to 750–900 fpm during cooling calls.

At the stage-left riser, we found a rigid duct section hard-coupled through a wall penetration with minimal clearance. The fix was twofold: add a short canvas/fabric flex connector at the connection point, and re-hang the riser with spring isolators rather than rigid trapeze hangers in that segment. We also ensured the penetration was properly sleeved and sealed with non-hardening acoustical sealant so the duct wouldn’t “short” vibration into the structure.

Week 7–8: Seal leaks, line critical sections, refine BAS event mode

After the major installs, we did a smoke pencil inspection and found two significant duct leakage points near access panels that created high-velocity hiss. Sealing those with mastic and improved gasketing reduced localized noise at the balcony considerably.

We also added internal duct liner to a short, unlined branch that turned sharply above the rear seating. Rather than lining the entire system, we targeted only the sections where turbulence and reflection were worst—particularly after elbows without turning vanes. In one elbow, we installed turning vanes to reduce flow separation, which can generate midband noise.

Finally, we refined BAS sequences. Event mode was changed to:

Pre-cool or pre-heat the room for 45 minutes at higher airflow before doors open.
During the event, cap fan speed at 80% and widen the temperature deadband slightly (±1.5°F) to avoid aggressive fan ramps during quiet program material.
After the event, return to normal ventilation and recovery.

Week 9–10: Verification measurements and operator handoff

We repeated the same measurement grid, same HVAC states, and compared 1/3-octave spectra before and after. We also measured during a live council meeting to verify real operational conditions, including occupancy and door activity.

5) Technical decisions and trade-offs made

Noise control in auditoriums is rarely about a single “silver bullet.” The trade-offs on this project were practical and explicit:

Silencers vs. pressure drop: Adding attenuation can increase static pressure and fan energy. We selected silencers with modest pressure loss (target under 0.15 in. w.g. each at design flow) and confirmed the AHU had headroom. This kept the system from compensating by spinning the fan faster—negating gains.
Diffuser replacement vs. architectural constraints: The ceiling grid and lighting layout limited diffuser size changes. We prioritized diffuser types with published NC performance data at the required CFM and kept the same 2’x2’ footprint to avoid ceiling rework.
Targeted lining vs. full duct treatment: Full lining can help but raises concerns about maintenance, fiber shedding, and cost. We used liner only where measurements and geometry suggested meaningful benefit: after high-velocity transitions and at specific elbows.
BAS changes vs. comfort: Operators wanted “set it and forget it” comfort control. We used pre-conditioning to reduce the need for aggressive airflow during quiet periods, accepting a slightly wider deadband during events to stabilize acoustics.
Isolation vs. access: Spring isolators and flex connectors improve vibration control but can complicate service access. We documented locations and ensured maintenance clearances were preserved.

6) Results and outcomes with specific details

The final results were measurable and audible:

Background noise: Mid-house seating improved from NC-41 (baseline) to NC-28 to NC-30 in event mode with steady-state cooling. Balcony seats improved from NC-42 to NC-30 to NC-32. Front seating landed around NC-27.
Spectral improvements: The 125–500 Hz bands dropped by 8–12 dB depending on location. The 63 Hz hump reduced by 4–6 dB after isolation work and BAS fan-speed smoothing, though it did not disappear entirely (fan and building structural modes can be stubborn at very low frequencies).
Operational gain: For council meetings using the existing reinforcement system (a DSP with automatic mixing and a pair of flown loudspeakers), operators reported running the system 4–5 dB lower on master output to achieve the same intelligibility in the back rows. The reduced HVAC noise floor also improved the perceived quality of recordings made from the board feed plus room mics.
Comfort and ventilation: CO₂ levels remained within acceptable thresholds during a 2-hour event with near-capacity attendance, and the room maintained temperature within the new deadband. Pre-conditioning proved critical; without it, the system would demand high airflow during the first 20 minutes of occupancy.

Timeline-wise, the work stayed inside the 10-week window. On-site mechanical modifications took 9 working days spread across two weeks, mostly during off-hours. Total direct construction cost (mechanical only) came in around $68,000, with roughly half attributable to silencers and associated duct modifications.

7) Lessons learned and what could be done differently

Three lessons stood out:

Don’t skip operational testing early. The temporary fan cap quickly proved the noise was airflow-related and helped justify silencers and diffuser changes. It also revealed the comfort implications, which prevented an overly simplistic “just slow the fan” plan.
Balance is a noise-control tool. The diffuser replacements mattered, but the re-balance was equally important. In auditoriums, a few overfed diffusers can dominate perceived noise. Establishing velocity targets and enforcing them during TAB (testing, adjusting, balancing) made the improvements consistent.
Low-frequency issues need structural thinking. The rumble wasn’t solved by attenuation alone. The isolation changes and smoothing of fan ramps reduced it. If the budget allowed, a deeper dive into fan vibration isolation at the AHU curb and potential duct breakout transmission would likely have delivered another couple dB at 63 Hz.

If we could redo one part, we would insist on earlier access to the BAS programming and trend data. Having a week of trended fan speeds, static pressure, and damper positions correlated to noise logs would have shortened diagnosis and reduced the number of iterative site visits.

8) Takeaways applicable to other projects

This project reinforced a repeatable playbook for controlling HVAC noise in auditoriums:

Measure in 1/3-octave and map the room. A single SPL reading is not enough. NC curves and spatial variation tell you whether the issue is diffuser-generated, duct-borne, or structure-borne.
Separate duct-borne from regenerated noise. Silencers help when noise is coming down the duct; diffusers, elbows, and transitions matter when the noise is being created locally by turbulence and high velocity.
Use velocity targets and verify them. “Quiet diffuser” claims only hold at specified CFM and pressure conditions. Confirm face velocities and static pressure at the branches feeding occupied zones.
Plan for pressure drop and control behavior. If you add attenuation but the fan ramps up to maintain static pressure, you can erase your gains. Coordinate silencer selection with fan capability and BAS strategy.
Address vibration paths explicitly. Flex connectors, isolation hangers, and properly sleeved penetrations are often inexpensive compared to chasing low-frequency rumble with acoustic band-aids.
Commission like an audio system. Establish a baseline, change one variable at a time when possible, and verify with repeatable measurements under real operating conditions, including occupancy and event mode settings.

Auditoriums don’t need to be anechoic to sound professional, but they do need a controlled noise floor. When HVAC noise is treated as a system-level engineering problem—airflow, mechanics, controls, and acoustics working together—the improvement is not subtle. In Redbrook’s case, the room went from “always fighting the building” to a venue where the PA supports the program instead of competing with it.