Home Energy Storage Noise and Placement: A Siting Guide
After eleven years of commissioning residential lithium systems, I can tell you that the callbacks that cost the most money are rarely about capacity. They are about noise. A 10 kWh home energy storage unit that performs perfectly on paper becomes a problem the first night a family hears it cycling at 2 a.m., and by then the wall is finished and the conduit is set. Noise and siting are not cosmetic details — they are engineering decisions that determine whether a system survives its first summer and its first complaint.
This guide is written from the field. It covers where residential batteries make noise, how that noise travels through a building, what numbers actually matter on a spec sheet, and how I choose a mounting location that keeps both the acoustics and the thermal envelope inside their limits. Whether you are a homeowner planning a garage installation or an integrator standardising a fleet of units, the physics is the same: sound is energy, structure carries sound further than air, and heat is the reason a badly sited pack dies early.
Where the Noise Actually Comes From
A residential battery is often described as “solid state” in the colloquial sense — no moving parts, therefore silent. That is only half true. A modern lithium pack contains at least three audible sources, and the loudest one is usually not the cells.
Transformer and inductor magnetostriction
Any inverter or bidirectional converter contains magnetic components. When a transformer core magnetises and demagnetises at the switching frequency, the steel laminations physically change dimension by a few parts per million. That is magnetostriction, and it produces the classic 100 Hz or 120 Hz hum — twice the line frequency, plus harmonics at 200/240 Hz and above. Measured at one metre, a decent hybrid inverter sits around 25–30 dB(A) in standby and 35–45 dB(A) under full transfer. The cheap units reach 50 dB(A), and 50 dB(A) through a shared wall at night is a complaint waiting to happen.
The annoying part is that magnetostrictive hum is tonal. Broadband noise at 40 dB(A) is easy to ignore; a pure 100 Hz tone at 32 dB(A) is not, because the ear and the brain lock onto a steady pitch. A-weighted measurements under-report this annoyance by roughly 5 dB, which is why I always ask vendors for both the dB(A) figure and an unweighted one-third-octave spectrum.
Cooling fans
Fans are the second source and the one most likely to wake people. A 92 mm axial fan at 2,000 rpm generates 30–38 dB(A) at one metre, and the acoustic signature is broadband with blade-pass tones. Most residential packs only run fans during high-rate charge or discharge, so the pattern matters: a unit that is silent for six hours and then ramps to 38 dB(A) for forty minutes is more disturbing than one that holds a constant 30 dB(A).
In my own commissioning I log a 24-hour fan duty cycle before handover. If the fans run more than about 15% of the time in mild weather, something is wrong — usually an oversized charge rate, a hot room, or poorly applied thermal interface material between the cells and the cold plate.
Contactors and relays
The third source is rarely in the datasheet: the DC contactor. Pre-charge and disconnect events produce a sharp 60–75 dB(A) transient lasting 20–60 milliseconds. It is brief, but a mechanical “clunk” at 3 a.m. travels through structure extremely well. Some integrators switch to solid-state DC contactors specifically for bedroom-adjacent installs; they cost more, generate heat, but are acoustically invisible.
How Sound Travels: Airborne vs Structure-Borne
Understanding the path matters more than the source level. Airborne sound from a cabinet falls off by about 6 dB for every doubling of distance in free field, and roughly 3–5 dB in a reflective room. Moving a unit from 1 m to 4 m from a listener therefore buys you only around 8–12 dB. That is useful but rarely sufficient on its own.
Structure-borne sound is the real enemy. When a cabinet is bolted directly into studs, the cabinet panel vibrates, the stud vibrates, and the drywall on the other side of the wall becomes a loudspeaker diaphragm. A stud wall with 12.5 mm plasterboard has a mass-spring resonance typically between 40 and 80 Hz — squarely in the range of transformer hum. At resonance the wall amplifies rather than attenuates, and you can easily measure 5–8 dB more on the bedroom side than the source side would predict.
This is why decoupling beats distance. A resilient mounting approach using a rubber-in-shear isolator with a static deflection of 3–6 mm gives an isolation efficiency above 90% for frequencies above roughly 25–30 Hz. Practically, I specify EPDM or neoprene grommets with a Shore A hardness of 45–60, sized so each isolator carries 25–40% of its rated load. Four isolators on a 110 kg cabinet means each one sees roughly 27 kg static plus dynamic load, and the assembly should still allow 2–3 mm of free movement.
Practical attenuation numbers
- Relocate from 1 m to 2 m in a reflective garage: 3–5 dB(A) reduction.
- Add 15 mm acoustic plasterboard plus a resilient channel on the shared wall: 8–12 dB improvement at 100 Hz.
- Mount on rubber isolators instead of rigid brackets: 6–10 dB reduction in structure-borne component.
- Add a sealed, gasketed enclosure with 50 mm mineral wool lining: 10–15 dB airborne reduction, but only if you leave ventilation area open or you will cook the pack.
Combine the last three and a 45 dB(A) unit becomes effectively inaudible in the adjacent room, typically landing below 25 dB(A) — under the WHO night-time guideline of 40 dB(A) and close to the typical rural bedroom floor of 20–25 dB(A).
Reading the Spec Sheet Honestly
Vendor acoustic data is where I see the most optimism. Here is how I normalise it before I believe it.
First, check the measurement distance and the standard. A figure quoted to IEC 60704 or ISO 3744 at 1 m in a hemi-anechoic space is not comparable to a number measured in a tiled garage. If a datasheet just says “<30 dB” with no distance, treat it as marketing. Second, check the operating point. Many units are quoted in standby or at 0.2C; a 1C discharge can be 8–12 dB louder. Third, check whether the figure is sound power (Lw, in dB) or sound pressure (Lp, in dB(A) at a distance). Sound power is the honest physical quantity, but it is bigger numerically and looks bad, so vendors quote pressure instead.
As a rule of thumb for a 10 kWh residential unit: Lp of 25–30 dB(A) at 1 m is excellent, 30–40 dB(A) is normal and fine in a garage or utility room, 40–50 dB(A) requires siting care, and anything above 50 dB(A) should not go on a wall shared with a bedroom.
Siting: The Decision Tree I Use
A good home energy storage placement satisfies five constraints simultaneously. Most failed installs I have inspected broke exactly one of them, and it was almost always thermal.
1. Thermal envelope first
Lithium iron phosphate cells used in stationary storage are specified for a charge window of roughly 0 °C to 45 °C and discharge from -10 °C to 55 °C, but those are cell-surface limits, not ambient guarantees. Charging below 0 °C causes lithium plating, which is irreversible capacity loss and a safety risk; a good BMS blocks charge below about 2–5 °C and engages a heater. That heater draws 60–200 W, and if the room is at -5 °C you may spend 5–8% of stored energy just staying warm enough to accept a charge.
Cycle life, meanwhile, is dominated by temperature. At 25 °C a quality LFP pack delivers 6,000 cycles to 80% capacity; at 35 °C the same chemistry typically lands near 3,500–4,000; at 45 °C it can fall below 2,000. Arrhenius behaviour means every 10 °C above roughly 25 °C roughly halves calendar and cycle life. An attic that reaches 55 °C on a summer afternoon will destroy a pack in three to five years — which is exactly why I reject attic installs unless the space is conditioned and the warranty explicitly allows it.
For garages, I measure worst-case ambient for a full week before committing. In a detached, uninsulated garage in a temperate climate, summer peaks of 35–40 °C are common; in a hot climate, 45–50 °C is realistic. If peak ambient exceeds 35 °C, either add ventilation, add a small ducted cool path, or move the unit indoors.
2. Structure and mounting
Wall-mounted units weigh 60–130 kg; floor-standing 10–20 kWh cabinets run 120–250 kg. That is not a drywall job. I require the load path to reach structural framing or solid masonry: minimum 16 mm plywood backboard screwed into at least two studs at 400 mm centres, or M10/M12 sleeve anchors into concrete with 45–60 N·m torque. Seismic zones additionally require restraint against lateral movement — check local code, but a 1.5–2.0 g horizontal design load is a reasonable starting assumption for a rack-mounted enclosure.
Stud walls transmit sound; masonry walls transmit less but reflect more. A solid concrete or block wall is my first choice for acoustics, followed by a stud wall with resilient channel, followed by a plain stud wall — which I will only use if the other side is a garage, hallway, or non-sleeping space.
3. Fire and electrical code clearances
Installations in most jurisdictions reference UL 9540 and UL 9540A for thermal runaway propagation, and NFPA 855 for installation. Internationally, IEC 62619 governs industrial and stationary cell safety, and UN 38.3 still applies to transport. Typical residential requirements: 900 mm (3 ft) clearance in front of the unit for service, no installation in sleeping rooms or egress paths, and for garages a protection barrier or a minimum 450 mm elevation above the floor to avoid impact damage from vehicles and pooling liquids.
Do not treat these as bureaucratic. The 900 mm service clearance is what lets a technician actually disconnect a fused DC string, and the 450 mm elevation is what keeps a pack out of a 40 mm flood.
4. Moisture and ingress
Outdoor or semi-outdoor installs need an appropriate ingress rating and, more importantly, a sensible microclimate. IP65 on the enclosure means nothing if the cable glands are at the bottom facing prevailing rain and the conduit is not sealed. I specify a drip loop, glands on the underside, a breather vent with a hydrophobic membrane to equalise pressure without ingesting moisture, and a 3–5° forward slope so water never sits on the lid gasket. Inside, a 10–20 W anti-condensation heater controlled by a humidity switch set around 60% RH prevents the corrosion that kills BMS boards long before the cells age out.
5. Acoustic adjacency
Finally, map the room on the other side of the wall. My ranking, best to worst: detached garage, utility/laundry room, plant room, hallway, garage under a bedroom, living room, and last, any wall shared with a bedroom. If the only viable location shares a wall with a bedroom, spend the money on decoupling and mass rather than on a quieter inverter — the isolation is cheaper and more reliable than the premium hardware.
Installation Details That Change the Result
Small choices during installation produce most of the difference between a quiet system and a noisy one.
- Isolate mechanically, never rigidly. Rubber grommets or spring hangers at every mounting point. One rigid bolt short-circuits the whole isolation scheme — it is a single-point acoustic bridge.
- Break the conduit path. Rigid metal conduit tied to both the cabinet and the building structure carries vibration. Use a flexible conduit section or a rubber-lined clamp within the first 300 mm.
- Seal the perimeter. A 2 mm gap under an enclosure door leaks more sound than the whole panel. Compression gaskets with 20–30% deflection are standard.
- Do not block ventilation. Every 10 dB of acoustic absorption you gain by stuffing mineral wool around a unit costs you roughly 3–6 °C of internal temperature rise. Leave the specified free area and route the air path through a lined baffle box instead.
- Torque to spec. Loose busbar joints heat, and hot joints cause the BMS to derate, which lengthens high-current operation, which runs the fans longer. A 6 N·m versus 4 N·m difference on a module terminal is the difference between a 4 K rise and a 25 K rise at 100 A.
- Commission with a sound level meter. I take readings at 1 m and in the adjacent room at 22:00 with the house quiet. Target is below 30 dB(A) in the adjacent room; if we are at 35 dB(A) or above, we fix it before handover, not after.
Commissioning and Handover Checklist
Before I sign off a residential installation, the following are verified and recorded. This is the same list I use whether the pack is a standard product or a custom battery solution built for an unusual site.
- Insulation resistance: 500 V megger, phase-to-earth and string-to-earth above 100 MΩ. Anything under 10 MΩ is a stop-work.
- Cell voltage spread at full charge: below 30 mV after a 2-hour rest. Above 50 mV indicates a balancing or connection problem.
- Capacity verification at 0.2C: should return at least 95% of nameplate on a fresh pack.
- Fan and contactor operation: confirm the acoustic profile matches the datasheet and that no tone is dominant.
- Thermal imaging under 0.5C: no terminal or busbar more than 15 K above the coolest point in the string.
- Ambient logging: place a data logger in the room for seven days and confirm peak ambient stays inside the warranty envelope.
- Torque marks on every power connection, documented in the handover pack with the measured values.
That last item sounds trivial. In practice, a documented torque record and a seven-day ambient log resolve almost every warranty dispute I have been pulled into, because they distinguish a product defect from a siting problem. And a siting problem is always cheaper to fix at design stage than after the drywall is closed.
Frequently Asked Questions
How loud is a typical home energy storage system?
A well-designed 10 kWh residential unit measures 25–30 dB(A) at one metre in standby and 35–45 dB(A) during high-rate charge or discharge. For reference, a refrigerator is about 40 dB(A) and a quiet library is 30 dB(A). Tonal hum near 100 Hz or 120 Hz is more annoying than the number suggests, so ask for an unweighted spectrum as well as the A-weighted figure.
Can I install a home battery in the garage?
Yes, in most jurisdictions, provided the unit carries the relevant listing, is mounted on a structural surface, is elevated roughly 450 mm above the floor or protected by a barrier, and keeps about 900 mm of service clearance. Check that summer garage temperature stays below the manufacturer’s limit — 35 °C ambient is a practical ceiling, and above that you will lose significant cycle life or trigger thermal derating.
Why is my battery humming at 100 Hz?
A steady hum at 100 Hz (or 120 Hz in 60 Hz regions) is magnetostriction in the inverter transformer or output inductor. It is normal at low levels. If it is loud enough to hear in an adjacent room, the cause is usually structure-borne transmission: a rigid mount coupling cabinet vibration into the wall framing. Rubber isolators and a flexible conduit section typically cut it by 6–10 dB.
Should I put the battery in the attic to save space?
Almost never. Unconditioned attics commonly reach 50–60 °C in summer. Every 10 °C above about 25 °C roughly halves lithium cycle life, so an attic installation can reduce a 6,000-cycle pack to fewer than 2,000 cycles, and most warranties exclude it. If the attic is the only option, it must be conditioned, ventilated, and explicitly permitted in writing by the manufacturer.
Does a noisy installation mean the battery is faulty?
Not necessarily. Fan noise during heavy charge or discharge is expected. What is not expected is a new or worsening tonal hum, a grinding fan bearing, or contactor chatter. A change in the acoustic signature over time is a genuine diagnostic signal: it usually means a loosening connection, a failing fan bearing, or a BMS that is spending longer at high current because a cell string has drifted out of balance.
How much clearance does a residential battery need?
Typically 900 mm (3 ft) in front for service access, plus whatever the manufacturer specifies for side and top ventilation — commonly 150–300 mm. Never install in a sleeping room or block an egress route. Outdoor units additionally need a drip loop, sealed glands, and clearance from direct sun where possible, because solar gain alone can add 10–15 °C to an enclosure.
Can I reduce noise with an acoustic enclosure?
Yes, but only if you preserve the airflow. A lined enclosure with 50 mm mineral wool gives 10–15 dB of airborne reduction, but blocking ventilation can raise internal temperature by 3–6 °C or more. The correct approach is a baffled enclosure: sound-absorbing lining on the inside surfaces with labyrinth air paths at inlet and outlet, sized so the pressure drop stays within the fan curve.
What is the best wall type for a quiet installation?
Solid masonry or concrete is the best acoustic substrate because its high mass does not resonate in the 40–80 Hz band where transformer hum lives. Second best is a stud wall built with resilient channel and acoustic plasterboard, which delivers 8–12 dB improvement at 100 Hz. A plain stud wall directly attached to a bedroom is the worst option and should be avoided unless you invest in mechanical decoupling.
Further Reading
References
