Air source heat pumps measured in UK homes ran 4 dB louder on average than their datasheets, and up to 8 dB louder, across ten week-long case studies, according to the UK government’s review of heat pump noise. The sound that reaches a neighbour’s bedroom is mostly a set of discrete tones between 50 Hz and 500 Hz, and a 2025 field study recorded a 110 Hz compressor tone that passed through a 15 mm wooden fence with minimal attenuation. Meanwhile, 2.9 million heat pumps were sold across 21 European countries in 2025, each one a new fixed sound source outside a home. This roundup looks at the outdoor condenser as a neighbourhood noise source, which is a different angle from our breakdown of low-frequency HVAC and subwoofer noise transmission inside buildings. We aggregated data from the UK Department for Energy Security and Net Zero (DESNZ), the Microgeneration Certification Scheme (MCS), the University of Salford, the European Heat Pump Association (EHPA), AHRI, Ofgem, the Bavarian Environment Agency and three peer-reviewed studies published in 2026.
TL;DR
- In-situ heat pump levels averaged 4 dB above datasheet values, with a maximum gap of 8 dB (DESNZ Research Paper 2023/046).
- A unit with 55 dB(A) sound power needs 22.2 m to stay 6 dB(A) under the night limit of a purely residential zone (Bavarian Environment Agency 2011).
- Adding a low-frequency tone to broadband noise raised annoyance by the equivalent of 0 to 7 dB (Yonemura et al., IJERPH 2021).
- A 110 Hz compressor tone was more audible at the neighbouring property, and a 15 mm wooden fence barely reduced it (University of Salford, Field Assessment of ASHP Noise 2025).
- England raised the permitted outdoor unit size from 0.6 m3 to 1.5 m3 and allowed two units on detached houses from 29 May 2025 (MCS 2025).
- Heat pump noise-associated arousals rose from 5.50 to 7.60 per night with a tilted window instead of a closed one, P < 0.050 (Benz et al., Noise and Health 2026).
- Raising heat pump noise from 36.5 to 46 dB(A) strongly increased annoyance in a 50-person listening test, β = 2.312, p < 0.001 (Acun et al., Acta Acustica 2026).
- 75% to 85% of UK heat pump owners are satisfied with noise, but 15% to 45% report intrusive or annoying noise (Kawai et al., Noise and Health 2026).
- European heat pump sales rose 13% to 2.9 million units in 2025, with stock reaching 29.3 million (EHPA Market Data 2025).
- US air-source heat pump shipments reached 2,908,644 units from January to August 2026, up 5.7% (AHRI August 2026 Statistical Release).
1. How Many Outdoor Condensers Are Being Added
Every heat pump sold adds a compressor and a fan that will run through winter nights, often a few metres from someone else’s window. The numbers below are the scale behind the noise question: 2.9 million heat pumps were sold in 21 European countries in 2025, a 13% rise, after a sharp fall in 2024. Two of the fastest-growing markets, Germany and the UK, have many terraced and semi-detached homes, where an outdoor unit is rarely far from a neighbour.
The United States shipped a similar number of air-source heat pumps in just the first eight months of 2026. Air-source heat pumps still trail central air conditioners in shipments, but both put an outdoor condenser in a back garden or side passage, so the siting question is the same on both sides of the Atlantic.
| Metric | Value | Source |
|---|---|---|
| Heat pumps sold in 21 European countries in 2025 | 2.9 million (up 13% on 2024) | EHPA Market Data 2025 |
| Total installed heat pump stock in those 21 countries | 29.3 million | EHPA Market Data 2025 |
| Largest national market by units and fastest-growing market in 2025 | France 528,000 units sold; Germany sales up 50% | EHPA Market Data 2025 |
| Residential heat pump sales in 16 EHPA-tracked countries, 2025 vs 2024 | 2.62 million vs 2.38 million (average growth 10.3%) | EHPA preliminary data, reported by pv magazine 2026 |
| UK heat pump sales growth in 2025 | 27% (roughly 125,000 units) | EHPA preliminary data, reported by pv magazine 2026 |
| US air-source heat pump shipments, January to August 2026 vs same period 2025 | 2,908,644 vs 2,751,248 (up 5.7%); central air conditioners 3,481,613 | AHRI August 2026 Statistical Release |
| Low-carbon heating systems funded by the UK Boiler Upgrade Scheme since launch, and share of grants for air source heat pumps | 80,098 systems; 97.6% of grants for air source heat pumps | Ofgem Boiler Upgrade Scheme Annual Report 2025-26 |
| Systems installed with Boiler Upgrade Scheme grants in Scheme Year 4 (April 2025 to March 2026) | 31,104 systems (£233.1 million in grants) | Ofgem Boiler Upgrade Scheme Annual Report 2025-26 |
Context note: EHPA publishes two different country sets. The full-year EHPA market data covers 21 countries, while the earlier preliminary release reported by pv magazine covers 16 residential markets, so the totals and growth rates should not be mixed. US figures come from the AHRI monthly shipment release, and UK grant figures from the Ofgem Boiler Upgrade Scheme annual report.
2. Where the Hum Sits in the Spectrum
The broadband whoosh of the fan and the tonal hum of the compressor behave like two different machines. Air source heat pump noise is dominated by discrete tones between 50 Hz and 500 Hz, which is exactly the range that light walls, fences and open windows handle worst. A tone is a single, steady pitch that stands out from the background sound, and people notice it at lower levels than a smooth broadband noise.
Fan speed swings the A-weighted output far more than compressor speed does, so slowing the fan cuts the broadband part of the sound sharply while the compressor tone changes much less. That is one reason a unit in quiet night mode can still be described as a hum rather than as silence.
| Metric | Value | Source |
|---|---|---|
| Frequency range where air source heat pump noise is dominated by discrete tones | 50 Hz to 500 Hz | DESNZ Research Paper 2023/046 |
| Main emission bands of the two sound sources inside the outdoor unit | Compressor: mainly around 200 Hz, tonal; fan: mainly 250 Hz to 2 kHz, broadband | Kawai et al., Noise and Health 2026 |
| Change in A-weighted sound power from fan speed (75 to 500 rpm) vs from compressor frequency (45, 59 and 80 Hz) in a measured 5.8 kW unit | Fan: up to 26 dBA; compressor: up to about 4 dBA | Kawai et al., Noise and Health 2026 |
| Low-frequency tone identified at a new-build estate in Nottinghamshire | 110 Hz (plus a high-frequency tone near 14 kHz) | University of Salford, Field Assessment of ASHP Noise 2025 |
| Where heat pump tones occur, and the mains hum frequency of electric motors | Integer multiples of compressor speed and of fan blade frequency; motor hum at 50 Hz | Bavarian Environment Agency (LfU), Low-Frequency Noise from Air Heat Pumps 2011 |
Context note: the fan-versus-compressor comparison comes from a single 5.8 kW unit measured with 13 microphones on a 2 m hemisphere, so it shows the mechanism rather than a market average. The tone mechanism is explained in the Bavarian Environment Agency guidance on low-frequency noise from air heat pumps. Most recent available data for that document: Bavarian Environment Agency, 2011.
3. Datasheet vs Real-World Sound Power
Every planning calculation for a heat pump starts from one number: the sound power level on the product datasheet. Sound power is the total acoustic energy a machine radiates; sound pressure is what a microphone, or an ear, measures at a given spot. Across ten week-long UK case studies, measured levels averaged 4 dB above datasheet values and reached 8 dB above them, which is enough to turn a borderline compliant installation into a clearly audible one.
The spread between products is even wider than the gap between datasheet and reality. Two units with the same heating output can differ by up to 25 dBA, so the model chosen can matter more than many siting measures. Yet one field case shows that the official calculation can also work well when the inputs are right.
| Metric | Value | Source |
|---|---|---|
| In-situ heat pump levels compared with datasheet values, ten week-long UK case studies | 4 dB higher on average; 8 dB higher at most | DESNZ Research Paper 2023/046 |
| Scatter of A-weighted sound power between units of the same heating capacity (database of 38 units) | Up to 25 dBA; heating capacity plus efficiency explain only about 40% of the variation (R² about 0.4) | Kawai et al., Noise and Health 2026 |
| Typical outdoor unit sound power in a 2011 German survey, and the state of the art at that time | Mostly 62 to 67 dB(A), scatter about ±15 dB(A); state of the art about 50 dB(A) | Bavarian Environment Agency (LfU) 2011 |
| MCS calculated levels compared with levels measured at a Nottinghamshire field site | Good agreement; compliance status unchanged between the 2019 and 2025 calculation methods | University of Salford, Field Assessment of ASHP Noise 2025 |
Outlier note: the 4 dB average gap comes from a small set of case studies gathered in a literature review, while the Salford and Heat Pump Association field result covers one estate. Both can be true: the calculation method is sound, but the sound power value fed into it is often optimistic.
4. The Rulebook: Limits, Background Assumptions and Tonal Penalties
Permitted development means an installation that needs no planning application, provided it meets fixed conditions. In England the noise condition is checked with the MCS calculation, and in 2025 the rules changed shape: the outdoor unit size cap rose from 0.6 m3 to 1.5 m3 and the 1 metre boundary rule was removed, so noise is now the main test left. Larger units can mean slower, quieter fans, but closer siting puts the low-frequency part of the sound nearer to neighbours.
The UK limit also looks generous next to much of Europe. A tonal penalty is an extra number of decibels added to the measured level when an audible tone is present, to reflect the extra annoyance; the MCS method has none.
| Metric | Value | Source |
|---|---|---|
| England permitted development noise limit at the neighbour’s assessment position, and the night-time background that local planning authorities actually estimate | 42 dB(A), which includes an assumed 40 dB(A) background and leaves about 37 dB(A) for the heat pump; real night background about 30 dB(A) in suburbs and 18 to 20 dB(A) in very rural areas | DESNZ Research Paper 2023/046 |
| Strictest night-time limits in other European countries, in dB(A) | Spain 25; Norway 28; France, Poland, Sweden and Finland 30; Austria, Germany and Denmark 35; Italy 40 | DESNZ Research Paper 2023/046 |
| Penalty or limit reduction applied when an audible tone is present | BS 4142: 2, 4 or 6 dB; Netherlands: +5 dB; Sweden and Finland: limit lowered by 5 dB; Germany: 0, 3 or 6 dB; MCS 020: none | DESNZ Research Paper 2023/046 |
| Maximum outdoor unit volume for a house in England, before and after 29 May 2025 | 0.6 m3 before; 1.5 m3 after | MCS, New Permitted Development Rules in England 2025 |
| Heat pump units allowed on a detached house under permitted development, and units covered by the new MCS 020 a) spreadsheet tool for full planning | 2 units (previously 1); tool covers up to 4 units | MCS, New Permitted Development Rules in England 2025 |
| Sound power value the MCS calculation must use | The highest specified level; low noise mode values may not be used | MCS 020 Planning Standards, Issue 1.3 (2019) |
Context note: the MCS announcement states that MCS 020 a) applies in England from 29 May 2025, while Scotland, Wales and Northern Ireland keep MCS 020 until further notice. The sound power rule is quoted from the MCS 020 Issue 1.3 text. Most recent available data for that clause: MCS, 2019.
5. Distance, Fences and Walls: What Actually Reduces the Hum
Low-frequency sound has long wavelengths, so it bends around thin barriers and passes through light walls that would stop a hiss. A University of Salford field study found that a wooden fence with boards about 15 mm thick gave minimal attenuation of the low-frequency tones, and those tones were more prominent at the neighbouring property than at the unit itself.
Distance and placement still work, but the margins are large. A German guidance table shows that a unit with 55 dB(A) sound power needs more than 20 metres to stay 6 dB(A) below the night limit in a purely residential zone, while a corner placement adds about 3 dB(A) before anything else is done.
| Metric | Value | Source |
|---|---|---|
| Attenuation of low-frequency tones by a wooden fence with boards about 15 mm thick | Minimal | University of Salford, Field Assessment of ASHP Noise 2025 |
| Two closely spaced units running together, measured vs predicted by simple logarithmic decibel addition | Adequate prediction at most microphones; some positions measured above the prediction; no significant low-frequency interaction confirmed | University of Salford, Field Assessment of ASHP Noise 2025 |
| MCS reduction terms for a unit 4 m away on the ground against one wall, behind a solid brick wall | Distance and directivity: -17 dB; brick wall barrier: -5 dB (22 dB in total: 17 + 5) | MCS 020 Planning Standards, Issue 1.3 (2019) |
| Distance needed for a unit with 55 dB(A) sound power to stay 6 dB(A) below the night limit | 22.2 m in a purely residential zone; 12.4 m in a general residential zone | Bavarian Environment Agency (LfU) 2011 |
| Reduction from an acoustic screen vs a partial enclosure | Screen: about 10 to 15 dB(A); partial enclosure: about 5 to 10 dB(A) | Bavarian Environment Agency (LfU) 2011 |
| Corner placement compared with placement against a single wall | About 3 dB(A) louder | Bavarian Environment Agency (LfU) 2011 |
Context note: the Bavarian figures in the LfU low-frequency noise guidance are overall A-weighted reductions, and the same document warns that required distances can double where reflections are unfavourable. A-weighted numbers can hide how little a light screen does below 200 Hz, which is the point of the Salford field assessment. Most recent available data for the distance table: Bavarian Environment Agency, 2011.
6. Annoyance and Sleep in the Laboratory
Controlled studies now separate how loud a heat pump is from how it sounds. In a 40-person sleep laboratory study, noise-associated arousals rose from 5.50 to 7.60 per night when the bedroom window was tilted open instead of closed (P < 0.050), while total sleep time did not change significantly. An arousal is a brief shift towards lighter sleep visible on brain-wave recordings, not necessarily a full awakening. In other words, the hum fragments sleep before it shortens it, a pattern also seen in hospital ICU noise and sleep arousal data.
Listening tests point the same way. In a 50-person study, annoyance tracked loudness, roughness and tonality, and changes in operating mode, such as a ramp from minimum to maximum capacity after defrost, were more annoying than steady running. Character matters as much as level, which mirrors findings on multirotor drone psychoacoustic annoyance.
| Metric | Value | Source |
|---|---|---|
| Indoor heat pump level with the window tilted vs closed, for 45 dB(A) played outside at 0.5 m | LAeq 30 dB(A) tilted vs 21 dB(A) closed; peaks up to 41 dB(A) tilted | Benz et al., Noise and Health 2026 |
| Noise-associated arousals per night, window tilted vs closed (40 adults) | 7.60 ± 5.90 vs 5.50 ± 3.60 (P < 0.050) | Benz et al., Noise and Health 2026 |
| Total sleep time: quiet baseline night vs heat pump night with tilted window vs with closed window | 418.20 vs 408.20 vs 405.70 minutes (P = 0.417, not significant) | Benz et al., Noise and Health 2026 |
| Participants who noticed the heat pump sound during the daytime session | 33 of 40 (11 described it as air conditioning or ventilation) | Benz et al., Noise and Health 2026 |
| Effect on annoyance of raising heat pump noise from 36.5 to 46 dB(A) at the listener | β = 2.312, p < 0.001 | Acun et al., Acta Acustica 2026 |
| Correlation between tonality and annoyance (3,518 ratings) | r = 0.305, p < 0.001 | Acun et al., Acta Acustica 2026 |
| Effect on annoyance of a ramp from minimum to maximum capacity, compared with steady running | β = 0.959, p < 0.001 | Acun et al., Acta Acustica 2026 |
| Extra annoyance, expressed in equivalent decibels, from adding a low-frequency tone (40 to 400 Hz) to broadband noise | 0 to 7 dB | Yonemura et al., IJERPH 2021 |
Context note: the window data reflect a sound reduction index of about 32 dB for the closed window against 11 dB for the tilted one with an 80 mm gap (Benz et al. 2026), and our acoustic glazing statistics cover how glazing and openings change sound insulation more broadly. The Acun et al. study used recordings of a single prototype unit. The Yonemura et al. tonal study found that the tonal adjustment exceeded the ISO 1996-2 value by 2 to 4 dB for tones above 100 Hz. Most recent available data for tonal adjustment: Yonemura et al., 2021.
7. Complaints, Satisfaction and the Annoyed Minority
Field evidence on neighbour impact is thin, and the numbers that exist are reassuring at the population level. Only 5 of 130 respondents living within 50 m of an air source heat pump could hear a neighbour’s unit indoors during the day, and the same share at night. But satisfaction surveys of owners keep finding a stubborn minority who call the noise intrusive, especially in social housing.
The older Swedish data explains why A-weighted limits may miss the problem. There, 14.7% of residents were annoyed at indoor levels of only 27 to 33 dBA, while the C-weighted readings, which keep more of the low-frequency energy, were 49 to 50 dBC. A 2020 German forecast cited in the 2026 systematic review by Kawai et al. expects the potential for conflict over low-frequency noise in residential areas to increase strongly as heat pumps spread.
| Metric | Value | Source |
|---|---|---|
| Neighbours within 50 m of a heat pump who could hear a neighbour’s unit indoors | 4% by day (5 of 130); 4% at night (5 of 129) | DESNZ Research Paper 2023/046 |
| UK heat pump owners satisfied with noise levels, across several surveys | Typically 75% to 85% | Kawai et al., Noise and Health 2026 |
| UK owners reporting intrusive or annoying noise | 15% to 45%, depending on study and context | Kawai et al., Noise and Health 2026 |
| Residents annoyed by heat pump or ventilation noise indoors (Swedish survey, 279 people) | 14.7% at 27 to 33 dBA (49 to 50 dBC); 20% at 33 to 36 dBA (52 to 60 dBC) | Persson Waye and Rylander 2001, reported in Kawai et al., Noise and Health 2026 |
Context note: the household figures come from 139 responses to 3,050 letters, so they describe a small, self-selected sample. The Swedish survey is more than 20 years old. Most recent available data for long-term annoyance by indoor level: Persson Waye and Rylander, 2001.
Summary: Residential Heat Pump Condenser Low-Frequency Noise by the Numbers
| Metric | Value | Source |
|---|---|---|
| Heat pumps sold in 21 European countries in 2025 | 2.9 million (up 13%) | EHPA Market Data 2025 |
| US air-source heat pump shipments, January to August 2026 | 2,908,644 (up 5.7%) | AHRI August 2026 Statistical Release |
| Share of UK Boiler Upgrade Scheme grants for air source heat pumps | 97.6% of 80,098 systems | Ofgem Boiler Upgrade Scheme Annual Report 2025-26 |
| Systems installed with Boiler Upgrade Scheme grants, April 2025 to March 2026 | 31,104 | Ofgem Boiler Upgrade Scheme Annual Report 2025-26 |
| In-situ levels vs datasheet values | 4 dB higher on average; up to 8 dB | DESNZ Research Paper 2023/046 |
| Outdoor unit volume cap for houses in England from 29 May 2025 | 1.5 m3 (was 0.6 m3) | MCS, New Permitted Development Rules in England 2025 |
| Low-frequency attenuation of a 15 mm wooden fence | Minimal | University of Salford, Field Assessment of ASHP Noise 2025 |
| Distance for a 55 dB(A) unit to stay 6 dB(A) under the pure residential night limit | 22.2 m | Bavarian Environment Agency (LfU) 2011 |
| Indoor level, window tilted vs closed | LAeq 30 vs 21 dB(A) | Benz et al., Noise and Health 2026 |
| Noise-associated arousals per night, tilted vs closed window | 7.60 vs 5.50 (P < 0.050) | Benz et al., Noise and Health 2026 |
| Total sleep time, quiet night vs heat pump night with tilted window | 418.20 vs 408.20 minutes (P = 0.417, not significant) | Benz et al., Noise and Health 2026 |
| Effect on annoyance of raising heat pump noise from 36.5 to 46 dB(A) | β = 2.312, p < 0.001 | Acun et al., Acta Acustica 2026 |
| Effect on annoyance of a ramp from minimum to maximum capacity vs steady running | β = 0.959, p < 0.001 | Acun et al., Acta Acustica 2026 |
| Correlation between tonality and annoyance | r = 0.305 | Acun et al., Acta Acustica 2026 |
| Extra annoyance from a low-frequency tone, in equivalent dB | 0 to 7 dB | Yonemura et al., IJERPH 2021 |
| UK owners reporting intrusive or annoying noise | 15% to 45% | Kawai et al., Noise and Health 2026 |
Methodology and Sources
- DESNZ, Review of Air Source Heat Pump Noise Emissions, Permitted Development Guidance and Regulations, Research Paper 2023/046 (main report republished January 2024): gov.uk publication page
- MCS, New permitted development rules in England: air source heat pump installations must now comply with MCS 020 a) (2025): mcscertified.com
- MCS 020 Planning Standards, Issue 1.3 (2019), copy hosted by Mid Sussex District Council: MCS 020 PDF
- University of Salford Acoustics Innovation Institute with the Heat Pump Association, Field Assessment of Air Source Heat Pump Noise (September 2025): project page
- Kawai C, Pachale U, Pieren R, Schäffer B. Heat Pump Noise: Scoping Overview of Emission Characteristics and Systematic Review of Human Effects. Noise and Health 2026: doi.org/10.4103/nah.nah_106_26
- Benz SL et al. Impact of Noise from Heat Pumps on Sleep, Noise Annoyance, and Concentration in Healthy Adults in a Laboratory Setting. Noise and Health 2026: doi.org/10.4103/nah.nah_147_24
- Acun V, Graetzer S, Radivan M, Torija AJ. Human response to air source heat pump noise: influence of background noise, operating conditions and acoustic characteristics. Acta Acustica 2026: article PDF
- Yonemura M, Lee H, Sakamoto S. Subjective Evaluation on the Annoyance of Environmental Noise Containing Low-Frequency Tonal Components. IJERPH 2021: doi.org/10.3390/ijerph18137127
- Bavarian Environment Agency (LfU), Low-frequency noise from biogas plants and air heat pumps, Part III (February 2011): LfU PDF
- European Heat Pump Association (EHPA), Market Data 2025: ehpa.org/market-data
- EHPA preliminary 2025 residential sales data, reported by pv magazine (March 2026): pv-magazine.com
- AHRI, August 2026 US Heating and Cooling Equipment Shipment Statistical Release: AHRI PDF
- Ofgem, Boiler Upgrade Scheme Annual Report April 2025 to March 2026 (31 July 2026): ofgem.gov.uk
- University of Salford Future Homes Acoustics team, Overcoming heat pump noise barriers (ACR Journal, March 2026), used for context on ongoing research: acrjournal.uk
- Data watch: the 42 dB(A) limit, the 40 dB(A) background assumption and the in-situ gap come from the 2023 DESNZ review, written before the May 2025 rule change; the MCS 020 a) text itself was not reviewed here, so the 2019 Issue 1.3 wording is quoted for the sound power rule. The DESNZ household survey (139 responses) and the Salford field study (one estate) are small samples. EHPA’s 21-country (2.9 million) and 16-country (2.62 million) figures cover different markets and are not comparable; pv magazine’s list of the 16 countries names Switzerland twice, so the exact country set should be checked against EHPA. The Swedish annoyance data (2001) and the Bavarian guidance (2011) are the most recent available sources for their specific metrics. Kawai et al. rate the certainty of health evidence as very low, and the Acun et al. listening test used one prototype unit. The University of Salford is also coordinating a seven-country neighbourhood modelling study of heat pump clusters under IEA HPT Annex 63, and a follow-up field study on multiple units in a small residential area; neither has published results yet.
Last updated: October 24, 2026. We update this roundup quarterly, and the next refresh is expected when AHRI publishes its full-year 2026 shipment release and the University of Salford publishes its multi-unit neighbourhood field study.