Silencer / Exhaust Blower Sizing for PSA Regeneration Step: Noise Limit 85 dBA at Property Boundary

1. Why the "85 dB(A) at Property Boundary" Specification Is Misleading for PSA Exhaust

A PSA regeneration train has at least four acoustic sources, and a procurement specification that simply writes "noise limit 85 dB(A) at property boundary" treats them as one. In practice they have different dominant frequency bands, different directivity patterns, different time profiles within the regeneration cycle, and therefore require entirely different mitigation. Spending a fixed lump-sum silencer budget on the wrong source is the single most common reason a PSA unit still fails its first regulatory noise survey.

The four sources, ranked by typical A-weighted sound power level at 1 m under nominal PSA conditions:

The depressurisation valve is almost always the loudest source, but procurement specifications typically silence the blower because the blower is a rotating-equipment line item with a documented noise curve in the vendor data sheet, while the depressurisation valve is a small instrument item that never gets its own acoustic line item. The procurement specification ends up asking for a silencer on the blower exhaust and a "schedule 40 vent pipe" on the depressurisation valve, and the property-boundary noise survey fails on the depressurisation pulse. We have seen this exact sequence on three of the last seven 1000 Nm³/h hydrogen PSA retrofits our procurement team reviewed.

2. Acoustic Source Characterisation: Choked-Flow Valve Exhaust vs Blower Broadband

2.1 Depressurisation Valve: Choked-Flow Acoustic Power from First Principles

When the depressurisation valve opens on a 7 barg to atmospheric blowdown, the gas accelerates through the valve trim to Mach 1 at the vena contracta. The acoustic power radiated from a choked jet scales as the eighth power of the jet exit velocity, which is fixed at sonic velocity for the local stagnation temperature (approximately 320 m/s for nitrogen at 40 °C). Therefore the acoustic power of a choked-flow depressurisation valve is set almost entirely by the mass flow rate (which is set by the upstream pressure and the valve Cv) and is largely independent of any downstream silencer geometry until the gas is decelerated below sonic. The correct silencer for a choked depressurisation valve is a venturi diffuser or multi-stage diffuser that recovers pressure and reduces exit velocity below 0.3 Mach, NOT an absorptive silencer, which the choked jet blows straight through.

Approximate acoustic power LwA (dB re 1 pW) for a single choked-flow depressurisation valve at 7 barg to atmospheric on nitrogen, estimated from LwA = 10·log10(η·ṁ·v²) with η ≈ 1×10⁻⁴:

2.2 Regeneration Blower: Tonal Plus Broadband with Characteristic Spectrum

A Roots blower (the most common PSA regeneration blower) has a fundamental noise source at the blade-pass frequency (BPF) equal to N_blades × N_shaft_rpm / 60. For a 3-lobe Roots blower at 3000 rpm, BPF = 150 Hz. The second harmonic at 300 Hz and the broadband noise above 1000 Hz from gear-mesh and bearing sources add another 6-10 dB to the overall level. A screw blower has a higher BPF (typically 800-1500 Hz) but lower tonal amplitude because the rotor profile is helical rather than impinging lobes. A centrifugal blower has the lowest broadband level but adds a high-frequency whistle from the impeller-tip Mach number.

The procurement pitfall: vendor noise curves are quoted as "free-field sound pressure level at 1 m from the blower housing" assuming the inlet and outlet piping are acoustically treated. In a PSA the outlet piping is typically 4-8 m of 4" pipe routing back to the adsorber, and every elbow and tee is an acoustic re-radiator. The actual sound power radiated from the un-treated outlet pipe can exceed the housing-radiated level by 4-6 dB. Specify the noise limit at the property boundary, not at the blower housing. Our internal procurement checklist (see Blog #156 activated-alumina acid-spill recovery for the related procurement discipline pattern) flags this distinction as a mandatory line in any blower RFQ.

3. The Three Silencer Topologies and When Each One Applies

3.1 Reactive Silencer (Choked-Flow Diffuser)

A reactive silencer is a multi-chamber diffuser with no absorptive material. It reduces acoustic power by expanding the gas through successive chambers, decelerating the jet below sonic, and dumping the residual acoustic energy into turbulence that decays into heat. For choked-flow depressurisation valves, a 3-chamber reactive diffuser achieves 18-25 dB insertion loss at the BPF-equivalent broadband of 500-2000 Hz. Cost is low (USD 1500-3000 for a 2" valve), pressure drop is moderate (0.2-0.5 barg), and maintenance is trivial — empty the chambers every 2 years to remove adsorbed water. The reactive silencer is the correct first choice for any depressurisation valve operating at inlet pressure greater than 1.5×atmospheric, which is essentially every PSA depressurisation valve in industrial service.

3.2 Absorptive Silencer (Lined Duct)

An absorptive silencer uses a layer of acoustic absorption material (typically mineral wool or fibreglass, density 32-48 kg/m³) lining a straight duct section of 3-5 diameters length. The acoustic power is converted to heat by viscous shear in the porous material at the boundary layer. Insertion loss is 15-30 dB at high frequency (above 1000 Hz) but falls to 5-8 dB below 250 Hz. Absorptive silencers are the correct choice for subsonic continuous-flow noise sources such as blower outlet piping and purge vent stacks. They are the WRONG choice for choked depressurisation valves because the choked jet erodes the lining within weeks and turns the silencer into a maintenance liability. The two most common procurement errors we audit on buyer-submitted RFQs are (a) specifying an absorptive silencer on a choked-flow valve, and (b) specifying a reactive silencer on a subsonic blower outlet.

3.3 Hybrid Silencer (Reactive First Stage Plus Absorptive Second Stage)

A hybrid silencer combines a 1-2 chamber reactive diffuser (to handle the choked-flow component and decelerate the jet) with a downstream absorptive section (to mop up the broadband residual and the low-frequency tonal from the blower BPF). Hybrid silencers are the correct choice when the same exhaust pipe carries both a depressurisation pulse (1-5 s, broadband, 500-4000 Hz) AND a continuous low-flow regeneration exhaust (subsonic, 30-120 s, tonal at blower BPF). Most medium-to-large PSA installations use hybrid silencers on the regeneration-blower outlet piping and reactive-only silencers on the depressurisation valves. Procurement cost: USD 4500-9000 for a 4" hybrid silencer rated for 0.5 barg pressure drop and 25 dB overall insertion loss.

4. Octave-Band Insertion-Loss Methodology

Specifying a silencer by its broadband A-weighted insertion loss is fundamentally wrong, because a silencer that achieves 30 dB at 1000 Hz may only achieve 8 dB at 125 Hz. Yet the property-boundary noise contribution from a blower BPF at 150 Hz can dominate the broadband A-weighted total once distance attenuation has reduced the higher-frequency components. The correct procurement metric is octave-band insertion loss at the centre frequencies 63 / 125 / 250 / 500 / 1000 / 2000 / 4000 / 8000 Hz, and the silencer datasheet must specify the insertion loss in each of those 8 bands independently.

Octave-band insertion loss requirements for a typical 500 Nm³/h hydrogen PSA unit with 60 m propagation distance to the nearest property boundary (using ISO 9613-2 outdoor propagation with -2.5 dB per distance doubling and soft-ground attenuation):

Note the negative values in the "required IL" column above 500 Hz: this means distance attenuation alone brings the source below the 35 dB target, so NO silencer is needed at those bands. The silencer procurement specification should focus on 63-500 Hz where the acoustic source is competitive with distance attenuation. A silencer datasheet that only specifies broadband A-weighted insertion loss is hiding the low-frequency weakness.

5. ANSI S1.13 Measurement Protocol and the L_eq vs L_max Question

The single most common compliance failure on a PSA noise survey is the measurement methodology, not the actual silencer selection. ANSI S1.13-2005 (R2010) "Measurement of Sound Pressure Levels in Air" specifies a continuous-equivalent A-weighted level L_Aeq,T measured over the full PSA regeneration cycle, with the microphone positioned 1.2 m above ground at the property boundary and at least 3.5 m from any reflective surface. The standard is unambiguous on these four points, but procurement specifications frequently call for L_Amax (instantaneous maximum) instead of L_Aeq,T, which produces two failure modes:

1. L_Amax over-states the depressurisation pulse by 4-7 dB relative to L_Aeq,T, because the pulse is 1-5 s long in a 240 s cycle. A 105 dB(A) pulse becomes a 75 dB(A) L_Aeq,T contribution. If the regulatory limit is set in L_Amax the operator must silence the pulse to meet the limit, which requires a 30-40 dB reactive silencer on every depressurisation valve and significant capital cost. If the regulatory limit is set in L_Aeq,T (the ANSI S1.13 default) the pulse can usually be left un-silenced because the time-averaged contribution is small.
2. L_Amax under-states the continuous blower tonal by 1-3 dB relative to L_Aeq,T, because a 90 dB(A) continuous tone adds another 6 dB to L_Aeq,T over 60 s of continuous operation, but shows up as only 90 dB(A) L_Amax. Setting the limit in L_Amax leaves the continuous noise un-controlled and creates a night-time noise complaint problem even when the regulatory survey passes.

The OSHA 29 CFR 1910.95 occupational limit uses an 8-hour L_Aeq (90 dB(A) PEL, 85 dB(A) AL). The EU 2003/10/EC occupational directive uses the same metric with 87/80 dB(A) thresholds plus a 3 dB exchange rate (more conservative than OSHA's 5 dB). The GBZ 2.2-2007 Chinese occupational health standard uses 85/80 dB(A) with a 3 dB exchange rate. All three occupational frameworks apply to the PSA building interior (operator stations, control rooms), NOT to the property boundary. For the property boundary the governing standard is typically the local general-purpose noise regulation, which varies widely by jurisdiction — for instance the Chinese GB 22337-2008 for industrial enterprises at boundary is 65 dB(A) daytime and 55 dB(A) nighttime (Class 1 zone, sensitive area).

6. Procurement Specification: 10 Mandatory Line Items

A PSA silencer procurement specification must include the following 10 line items. Our internal procurement team rejects any RFQ that does not address all 10.

1. Source-by-source list — depressurisation valves, blower outlet, vent stack, compressor piping — with each source's acoustic power LwA in octave bands 63-8000 Hz.
2. Property-boundary noise limit — daytime and nighttime, in dB(A) L_Aeq,1h, with reference to the local regulation (GB 22337-2008, EU 2002/49/EC, OSHA 29 CFR 1910.95, etc.).
3. Distance to nearest property boundary in 4 cardinal directions, with terrain and ground-type description (hard / soft ground).
4. Propagation model — ISO 9613-2 with ground factor G=0.99 (hard) or G=0.68 (mixed), 10 °C / 70 % RH standard atmosphere.
5. Octave-band insertion loss requirement per source, in all 8 standard octave bands.
6. Silencer topology — reactive / absorptive / hybrid, with justification per source.
7. Pressure drop budget in barg at the design mass flow, with the constraint that the depressurisation cycle time does not increase by more than 5 %.
8. Material specification — 304L / 316L stainless, with chloride-service restriction for sour-gas service (H₂S-bearing feed).
9. Maintenance interval — inspection access, drain ports, expected lining life in absorptive silencers (typically 5-7 years).
10. Vendor acoustic test certificate — ISO 7235 laboratory test per silencer, with octave-band insertion loss traceable to NIST or PTB calibration.

For a related procurement discipline pattern — RFQ line items that have to be airtight before the LOI goes out — see Blog #159 on API 521 depressurisation vent sizing, which covers the analogous 10-line spec for emergency depressurisation vents.

7. Case Study: 2000 Nm³/h Hydrogen PSA, 85 dB(A) at 50 m Boundary

A 2000 Nm³/h hydrogen PSA with 4 beds, 7 barg adsorption / 0.05 barg desorption, regenerating on a 240 s cycle with 4 depressurisation pulses per cycle and 60 s continuous blower exhaust per bed, was specified with "85 dB(A) at property boundary, 50 m distance, daytime operation." Initial silencer specification (absorptive on the blower only, per the buyer's instrument-engineer's default) was estimated to deliver 95 dB(A) at the boundary, 10 dB above the limit, because the absorptive silencer does nothing for the depressurisation pulse. Re-specification with hybrid silencer on the blower and reactive diffuser on each depressurisation valve delivered an operating result of 60 dB(A) L_Aeq,1h at the boundary, with 25 dB margin against the 85 dB(A) limit. The hybrid silencer cost USD 6800 (vs USD 2400 for the absorptive-only original spec), a USD 4400 incremental investment that averted a regulatory rework.

The same case study demonstrated a related issue that has come up on three PSA retrofits in 2026: molecular sieve attrition dust loading the blower. The hybrid silencer included a 50-mesh wire-mesh pre-filter on the blower inlet to catch dust before it deposits on the blower rotor lobes. Without this filter, dust loading on a 3-lobe Roots blower over 6 months reduces the rotor-to-housing gap, increases BPF amplitude, and degrades the silencer insertion loss by 4-6 dB. The fix is to specify low-attrition molecular sieve (4A / 5A / 13X with COA d50 ≥ 2 mm and dust index ≤ 0.02 %w) and pair it with the blower pre-filter. For a related discussion of dust-induced blower degradation and the wire-mesh selection logic, see Blog #153 on spent molecular sieve bed sectioning.

8. Vacuum-Pump Operating Envelope: When the "Blower" Is Actually a Liquid-Ring or Dry Pump

Vacuum-swing adsorption (VSA) units and sub-1 barg PSA units use a liquid-ring vacuum pump or dry claw pump instead of a positive-displacement blower. The acoustic profile is different. A liquid-ring pump has a dominant tonal at the impeller-vane pass (typically 800-1500 Hz) plus a broadband water-splash component at 2000-5000 Hz, with overall L_wA typically 92-98 dB(A) at 1 m. A dry claw pump has a higher tonal (1200-2000 Hz) but lower broadband, with overall L_wA 88-94 dB(A). The silencer procurement logic is the same as for a blower (absorptive or hybrid, octave-band spec), but the pressure drop budget is tighter — liquid-ring pump suction pressure is typically 0.3-0.5 bara, and silencer pressure drop above 30 mbar (0.03 barg) directly degrades the achievable vacuum and the regeneration cycle efficiency.

Two procurement pitfalls specific to vacuum-pump service:

1. Moisture carryover from the liquid-ring seal water — silencers must be self-draining with a bottom drain port to prevent condensate accumulation, which would otherwise produce a low-frequency gurgling noise that the silencer does not address. Condensate accumulation also corrodes 304L stainless over 12-18 months. Specify 316L for liquid-ring-pump service with chloride-content seal water above 50 ppm.
2. Dry claw pump rotor cooling airflow — the cooling fan on a dry claw pump is on the same shaft as the rotor and contributes an additional 78-85 dB(A) broadband component at 1 m, typically above the silencer frequency band. Specify acoustic enclosure on the cooling fan inlet rather than trying to silencer it downstream.

9. Pressure Drop, Back-Pressure, and the Cycle-Time Penalty

Every silencer has a pressure drop, and on the depressurisation side of a PSA, every additional millibar of back-pressure lengthens the time required to depressurise the bed to the regeneration pressure. For a 7 barg to 0.05 barg blowdown the natural depressurisation time constant (with no silencer restriction) is approximately 8 s. A 0.3 barg silencer pressure drop extends this to approximately 11 s, a 38 % cycle-time penalty that translates directly into a 5-7 % reduction in product flow. The silencer procurement specification must quantify this penalty and bound it at less than 5 %.

On the regeneration-blower outlet side the pressure drop is recovered as compression work, so the penalty shows up as blower-motor kW rather than battery load. A 0.5 barg silencer pressure drop on a 60 m³/h regeneration flow costs approximately 1.0 kW of blower power, equivalent to USD 800/year in electricity at USD 0.10/kWh and 8000 operating hours/year. Over a 10-year PSA life that is USD 8000 in operating cost, which is comparable to the silencer capital cost. The total-cost-of-ownership comparison should include this operating penalty.

10. Maintenance, Inspection, and 5-Year Lining Life

Absorptive silencer linings degrade over time from moisture, dust, and chemical attack. Typical lining life in clean dry compressed-air service is 7-10 years; in PSA regeneration exhaust with molecular sieve dust loading it is 3-5 years; in sour-gas PSA service with H₂S it is 5-7 years if the lining is 316L stainless-wrapped; in chloride-bearing feed-gas service the lining must be Hastelloy or equivalent to avoid chloride-stress-corrosion-cracking of the stainless wool wrap. Inspection interval should be annual visual inspection from the inlet and outlet, plus a 3-year internal inspection via access door to verify lining integrity.

Reactive silencers require almost no maintenance — the chambers are empty, the only failure mode is external corrosion, and the inspection interval can be the same as the vessel external-inspection interval (typically 5-10 years per ASME Section VIII). Condensate drain ports must be cleared annually.

11. Frequently Asked Questions

12. Next Steps

For PSA silencer and exhaust blower sizing support, contact our engineering team with your regeneration cycle parameters (cycle time, depressurisation set-point, regeneration pressure, mass flow rate, and nearest-property-boundary distance). We will run an octave-band acoustic source characterisation, develop a silencer topology recommendation, and provide a 10-line-item procurement specification aligned with the local regulatory framework (OSHA, EU 2003/10/EC, GBZ 2.2-2007, GB 22337-2008, or other).

For complementary coverage on related PSA topics, see our Blog #162 on emergency adsorbent discharge system design, Blog #163 on adsorber vessel internal corrosion, and the full product details on low-attrition molecular sieve, activated alumina, and alumina powder.

WhatsApp: +86 133 2522 2240  ·  Email: barry@aluminaworld.com

Standards referenced: ANSI S1.13-2005 (R2010), ISO 1996-1:2003, ISO 1996-2:2007, ISO 7235:2003, ISO 9613-2:1996, OSHA 29 CFR 1910.95, EU Directive 2003/10/EC, GBZ 2.2-2007, GB 22337-2008, API 521-2020, ASME Section VIII.