An attenuator selection sheet usually contains two numbers that people look at and one that decides whether the selection works. The two are insertion loss and pressure loss. The one is airway velocity, and it drives both of the others plus a third quantity, flow generated noise, that does not appear on many selection sheets at all.

Sizing an attenuator is not an acoustic exercise with a mechanical footnote. It is a simultaneous acoustic and mechanical problem in which the same geometry sets every outcome, and the two disciplines that own the constraints frequently do not review the selection together.

What the geometry is doing 

A rectangular splitter attenuator is a casing containing absorptive splitters separated by airways. Sound entering an airway is attenuated by absorption at the lined faces as it travels the length of the unit.

Three geometric parameters set the acoustic result.

Length determines how much interaction occurs. For a given cross section, doubling the length roughly doubles the attenuation in the frequency range where the unit is working well, though the relationship flattens as attenuation becomes large.

Airway width determines how often sound meets a lined surface. Narrower airways attenuate more, particularly at mid and high frequencies, because a wave travelling down a narrow lined passage interacts with the absorptive faces far more often than one travelling down a wide passage.

Splitter thickness determines the low frequency limit. A thin absorptive layer cannot absorb a wavelength much longer than itself. Attenuation at 63 Hz and 125 Hz is bought with splitter thickness and length, which is why low frequency requirements make attenuators large rather than making them slightly longer.

The result is the characteristic shape of an attenuator insertion loss curve. Performance is modest at the bottom of the range, peaks somewhere in the mid frequencies, and falls away again at the top, where sound can propagate down the centre of a wide airway as a beam without meeting the lining.

Our published indicative data for the Sonic acoustic attenuator range shows that shape clearly, with insertion loss at 63 Hz running from around 3dB on a short, wide-airway unit to around 19 dB on a long thick-splitter unit, which is the spread that makes airway width and length the real variables. Those are published ranges across a product family rather than a selection for any particular duty, and a project selection needs configuration specific data at the design flow.

Velocity is the parameter that connects everything 

Face velocity is airflow divided by the gross cross sectional area of the casing. Airway velocity, sometimes called core velocity, is airflow divided by the open area between the splitters. Because splitters occupy a substantial share of the cross section, airway velocity is always the higher figure, and it is the one that governs the mechanical behaviour.

Pressure loss across the unit is dominated by the dynamic pressure at the airway, which for air at about 1.2 kg/m3 is half the density times the square of the velocity. At an airway velocity of 10 m/s that dynamic pressure is about 60 Pa. At 15 m/s it is about 135 Pa. A fifty per cent increase in velocity produces something in the order of a 125 per cent increase in loss.

The total loss also includes entry contraction, friction along the splitters and exit expansion, each with its own coefficient, so the working figure has to come from the manufacturer’s published resistance data rather than from the dynamic pressure alone. The point of the arithmetic is the sensitivity, not the absolute value.

Flow generated noise is more sensitive still. Flow noise from obstructions in an airstream is commonly modelled as a dipole type source whose sound power rises with roughly the sixth power of velocity, which corresponds to about 18dB for a doubling of velocity. ISO 14163 defines this regenerated sound as flow noise caused by the flow conditions in the silencer, and notes that performance departs from the laboratory case when the inlet flow distribution is not uniform. This is why an attenuator can be selected with adequate insertion loss and still fail to deliver the required level downstream. The unit is attenuating the fan noise correctly and adding its own.

One widely used manufacturer guideline holds that regenerated noise is generally not a governing concern while the pressure loss across the unit stays at or below about 87 Pa, which is 0.35 in WG in the United States source the guideline comes from. That is a rule of thumb rather than a standard, and it is a useful early screen rather than a substitute for the manufacturer’s self noise data.

The two failure modes 

The first failure mode is selecting from the insertion loss column.

An engineer needs 25dB at 250 Hz, opens the selection table, finds the shortest unit that delivers it, and specifies that unit. The unit is short because its airways are narrow, and its airways are narrow because that is how short attenuators reach mid frequency performance. In the duct it produces a pressure loss two or three times the allowance the mechanical designer carried, and the fan cannot make the duty.

What happens next is usually one of three things, and all three are expensive. The fan is upsized, which increases the source sound power and often the low frequency content, moving the acoustic problem rather than removing it. The duct is enlarged to reduce velocity, which requires plantroom space that no longer exists. Or the attenuator is swapped for a lower resistance unit late in the programme, and the acoustic performance is quietly reduced to whatever the pressure budget allows.

The second failure mode is the mirror image, selecting for minimum pressure loss.

A mechanical designer, working to a tight fan energy target, selects the lowest resistance unit that carries an acceptable A-weighted insertion loss figure. Wide airways give low resistance. Wide airways also give poor mid and high frequency attenuation for a given length, and if the requirement was concentrated in one octave band, the broadband figure conceals the shortfall.

Both failures come from the same cause. One number was optimised and the other was accepted as whatever it turned out to be.

The consequences that land outside the acoustic scope 

Pressure loss is not only a fan selection issue. It is an ongoing energy and capacity issue.

Fan shaft power is the product of volumetric flow and total pressure rise divided by total efficiency. For a plant handling 10 m3/s, an additional 200 Pa of resistance at a total efficiency of 0.65 requires roughly 3.1 kW of additional shaft power. On plant that runs continuously, that is in the order of 27 MWh per year for a single duct.

Where an attenuator is installed on a data centre cooling plant, a water treatment blower or an industrial extraction duct that runs at high utilisation, that figure is worth putting in front of whoever signs off the operating cost model, because it is the part of the acoustic decision that keeps being paid after handover.

Resistance also interacts with the plant as a whole. A duct network behaves approximately as a square law resistance, so adding an attenuator moves the operating point along the fan curve. If the fan is not reselected, the delivered flow falls. On a generator or process cooling duty, a reduction in delivered airflow is a reliability question, not an acoustic one.

The conditions the laboratory number assumes 

An insertion loss figure is produced under a defined test arrangement. ISO 7235 sets out the laboratory measurement procedure for ducted silencers and air terminal units, producing insertion loss, flow noise and total pressure loss together. That grouping is deliberate. The three quantities are not independent, and quoting one of them without the flow condition it applies to is quoting an incomplete measurement.

Several installation conditions cause the installed result to fall short of the laboratory figure, and they are worth checking during design review rather than during commissioning.

Non uniform approach flow, caused by a bend, a damper or a fan discharge immediately upstream, raises both pressure loss and self noise, and can reduce insertion loss.

Breakout through the casing lets sound leave the attenuator through its own walls into the plantroom, bypassing the airways entirely, which matters when the attenuator is upstream of a duct that then passes through an occupied space.

Flanking around the unit, through the surrounding builder’s work opening, through unsealed penetrations in a plantroom wall, or along the duct wall itself, sets a ceiling on the achievable result in exactly the way an opening does in an enclosure.

Direction of flow relative to the direction of sound propagation changes the result, which is why forward flow and reverse flow data are reported separately.

What a defensible selection looks like 

An attenuator is the point where the acoustic model and the mechanical model have to agree, and the sizing decision is where any disagreement between them is settled.

The productive sequence is to establish the required attenuation in octave bands first, then set the velocity from the pressure and self noise limits rather than from the duct size that happens to exist, then find the length and airway configuration that meets the acoustic requirement at that velocity, and only then check whether it fits. Where it does not fit, the honest options are more space, a different treatment location, or treatment at the source, and all three are cheaper to discuss at design stage than after the plant is installed.

The number that should never be the starting point is the one at the top of the insertion loss column.