Lithium-ion battery fires are no longer just a domestic problem and conventional smoke detectors often miss the early warning signs. We're a family-run fire and security firm based in Southend-on-Sea, Essex and we've designed and maintained commercial Fire Alarm Systems since 1979, holding NSI Gold Approved status since 2005. So we've written this as a working brief for anyone responsible for a self-storage site, a logistics depot or a waste and recycling facility.
It's gone 11pm at a self-storage facility, motion-sensor lighting clicking off unit by unit as the last visitor leaves. Somewhere inside one of those units, a customer's forgotten e-bike battery is quietly overheating. Nobody's there to smell it. Nobody's there to see the vapour building in the dark. By the time a corridor sensor picks up smoke, the fire is already spreading cell to cell. That's faster than a response built around smoke alone can catch up with.
Lithium-ion battery fires used to read as a household problem, something that happened in somebody else's house. Domestic e-bikes charging overnight in hallways. E-scooters left by the front door. That assumption isn't wrong, but it's increasingly incomplete and out of date.
UK fire brigades attended lithium-ion battery fires at a rate of 4.8 a day in 2025. That's nearly 1,760 incidents, up 147% since 2022 (QBE European Operations, FOI research, 11 May 2026). QBE's own analysis splits the incidents by location: 46% in homes, 31% outdoors and 23% in commercial premises. That means more than half of these fires now happen somewhere other than a living room. And getting on for a quarter of them happen on premises where a business - not a homeowner - carries the legal responsibility. The commercial figure comes from a single insurer's FOI analysis, not official Home Office statistics. Treat it as the strongest available indicator of how much of this risk now sits on commercial premises, rather than a definitive market-wide number. Even read conservatively, a 147% rise in attended fires over three years puts commercial premises squarely inside a fast-growing risk category - not a detail your Fire Risk Assessment can afford to overlook.
In London alone, the Fire Brigade attended 521 lithium-ion battery fire incidents in 2025, up 28% year-on-year. That resulted in 109 injuries and three deaths (London Assembly Research Unit, 6 March 2026). This is no longer a niche domestic hazard. It's a workplace problem and one that needs the same forward planning as your Fire Alarm servicing contract already gets.
In practical terms, a lithium-ion battery fire begins when a cell overheats, is damaged or short-circuits until it can no longer shed heat fast enough. Past that point, it tips into thermal runaway - a self-sustaining chain reaction that behaves differently from a conventional fire and is correspondingly harder to detect, contain and extinguish.
A lithium-ion cell packs a lot of energy into a small space, sitting right next to a flammable liquid electrolyte. Damage, overcharging, overheating or a manufacturing defect can push a cell into generating more heat than it can shed. Past a certain point, the reaction becomes self-sustaining. That's thermal runaway. Each failing cell heats its neighbours in turn and the fire cascades cell to cell through the pack.
Before flames appear, a failing cell usually off-gasses. It vents a cloud of flammable, toxic vapour - electrolyte solvents mixed with hydrogen. This is an early warning sign of a battery in trouble, often minutes before visible fire. It's also the signal that conventional point smoke detection was never designed to catch in time.
The fire that follows is chemically driven. It can generate its own oxygen supply, which is why smothering agents and conventional firefighting methods often struggle against it. It can burn hotter and spread faster than an ordinary fire. And it can look extinguished, then reignite hours or days later, because heat is still propagating somewhere inside the pack.
"Thermal runaway caused by these types of batteries burns differently, takes much longer to tackle and can require up to 10 times more water to contain."
Adrian Simmonds, Risk Manager, QBE Insurance (QBE press release, 11 May 2026)
That difference - in behaviour, not just in scale - is why early detection matters more for lithium-ion risk than for almost any other fire type you have to plan around as a Responsible Person.
Logistics and last-mile depots run high-density overnight charging for e-cargo bike and van fleets, often unattended for hours at a stretch. Dozens of packs charging simultaneously, out of sight, is precisely the scenario thermal runaway thrives in.
Waste and recycling sites face a different version of the same problem: batteries hidden inside discarded electricals, crushed or damaged during collection and processing. Those damaged cells go on to ignite bulk combustible waste around them. Battery-related fires in refuse vehicles and waste facilities exceeded 1,200 in 2023/24, a 71% year-on-year rise. That's up from around 700 the year before (Environmental Services Association and Material Focus). Those incident counts are underpinned by reported data.
Self-storage carries a harder problem still: unknown contents. If you operate a self-storage site, you often have no reliable way of knowing what's in a unit. It could be an e-bike, a stack of power banks or nothing of concern at all - until something goes wrong.
Aviva recorded a 7% year-on-year rise in customer claims for lithium-ion battery fires. A separate Aviva business survey found that only 15% of UK organisations had carried out a workplace Fire Risk Assessment specifically covering lithium-ion batteries. Both figures come from Aviva's own claims book and survey panel rather than a national census. But they line up with the wider claims trend reported across the insurance sector. They point to a stark compliance gap: most UK businesses haven't yet assessed a risk their own insurers are actively pricing for. Every renewal that passes without one risks a premium priced on the insurer's assumptions about that risk, not the site's actual position.
Allianz UK's claims data adds further scale, drawn from two separate releases. Its residential claims data puts the average lithium battery fire claim at £50,000. A separate motor-trade release puts two commercial losses at £5 million and £1.5 million, both attributed to stored EV batteries. Numbers on that scale are precisely what insurers factor into a commercial premium, assessed risk or not.
Insurers are several steps ahead of most Responsible Persons on this. The Regulatory Reform (Fire Safety) Order 2005 (FSO 2005) is the law behind that term. It defines a Responsible Person as whoever has control of a premises in connection with a trade, business or other undertaking. That's typically you, as employer or building owner. That gap between what's assumed and what's actually in place doesn't close on its own. And it's you who carries the risk in the meantime, not the insurer. That's where a BS 5839-1:2025-compliant detection strategy comes in. It's a practical way to start closing that gap between what insurers assume and what's actually in place.
BS 5839-1:2025 (the British Standard governing Fire Detection and alarm system design in non-domestic premises) is a code of practice, not law. It's voluntary - the legal duty sits with FSO 2005. But Approved Document B references BS 5839-1:2025 directly — one reason it functions in practice as the recognised benchmark of good practice for Fire Detection design. The 2025 edition, published 30 April 2025, replaced BS 5839-1:2017+A2:2024. It sets out eight system categories - M, L1 to L5, P1 to P2. The appropriate category for a given space is determined by Fire Risk Assessment, not guesswork.
For battery risk specifically, the detector-selection depth matters. A single point smoke detector, positioned for a conventional fire, is a poor match for this threat. It announces itself first as off-gassed vapour, rather than smoke. No single detector solves this on its own. We design a properly justified BS 5839-1:2025 strategy for a charging area or battery storage room around a combination of technologies. Multi-sensor detectors combine smoke, heat and CO (carbon monoxide) sensing, detecting faster and with fewer false alarms. Aspirating smoke detection (ASD) samples air continuously through pipework and can pick up very early smoke in high-value stores or warehousing. ASD systems are tested to BS EN 54-20. That standard grades sensitivity into three classes, Class A for very early warning through to Class C for general Fire Protection. We match the class specified to how early a warning the space actually needs.
Off-gas and VOC (volatile organic compound) detection deserves a specific caveat here. It's a genuinely strong technology for early warning. It detects electrolyte vapour and hydrogen at the point a cell first vents, potentially minutes ahead of any smoke. But it is not a BS 5839-1:2025 primary detector. No dedicated UK detection standard for it has been published yet. We specify it honestly: as a supplement to a compliant detection design, integrated alongside conventional and multi-sensor detection, not as a replacement for it.
Design, installation, commissioning and maintenance to BS 5839-1:2025 sits within the scope of BAFE SP203-1. That's the scheme covering Fire Detection and alarm work, with certification bodies including NSI, SSAIB, BSI and NICEIC. We're BAFE SP203-1 Registered, delivered through our NSI Gold Approved status - NSI is a UKAS-accredited certification body licensed by BAFE to conduct that audit. Two of our fire system designers work under that NSI Gold approval. Like BS 5839-1:2025 itself, SP203-1 registration is voluntary, not a legal requirement. The legal duty sits with FSO 2005 instead and it comes down to three things:
FSO 2005 is outcome-based. It doesn't name BS 5839-1:2025 or specify testing frequencies. But non-compliance that places people at risk of death or serious injury carries an unlimited fine and up to two years' imprisonment. Your local Fire and Rescue Authority enforces this. That's why we build every detection strategy around BS 5839-1:2025, installed and maintained by a BAFE SP203-1 registered company. It's a practical way to show those duties have been met, not just a box-ticking exercise.
"I would just like to thank you and all the team...for an absolutely fantastic service and a great job well done."
Kirsty Powell, Director, Chef's Choice Ltd
That's the NSI Gold-approved standard Chef's Choice got from a straightforward office Fire Alarm installation - the same standard behind a detection strategy built around battery risk.
It's our policy not to offer a service we're not qualified to provide. So when we recommend a detection strategy for your site, it's built around what BAFE SP203-1 and BS 5839-1:2025 require, not what's easiest to sell.
FSO 2005 requires "appropriate fire-fighting equipment" on premises, but it never mandates portable extinguishers by name. That distinction matters: some extinguishers on the market are marketed as suited to lithium-ion fires, but no recognised UK test currently backs that claim up.
Start with a basic point of accuracy: lithium-ion battery fires are not Class D. Class D covers combustible metals - magnesium, sodium, titanium - not lithium cells and treating them as equivalent is a genuine misclassification, not a technicality. Lithium-ion battery fires don't fit neatly into the traditional A-F fire classes and industry sources report that BS ISO 3941:2026 introduces a new "Class L" category now being recognised across the industry for them. But that's a classification, not a performance rating. No BS EN 3 performance rating for extinguishers against lithium-ion fires has been published to date. So any claim that a product is "Class L rated" for extinguishing performance isn't backed by a recognised UK test. The rationale for a distinct class mirrors the mechanism explained earlier. Lithium-ion fires exhibit the faster heat release and cell-to-cell fire growth of thermal runaway. That's behaviour the existing extinguisher classes (A, B, C, D, F) were never designed to describe. It's precisely why a new class exists, even without a matching performance rating yet.
The trade body guidance is blunt about the limits. The FIA's "Guidance on Li-Ion Battery Fires" (December 2020) states plainly that "complete extinguishment may not be possible" with portable extinguishers. Manufacturer guidance goes further: portable extinguishers won't put out a fire in a lithium-ion battery pack any larger than an e-scooter, and won't extinguish an Electric Vehicle (EV) fire at all.
Sprinklers, where fitted to the commercial anchor standard BS EN 12845:2015+A2:2026, have a genuine role - they cool and control fire spread, buying critical time. But they don't "extinguish" a battery pack in thermal runaway either. A burning cell keeps generating its own heat and oxygen internally, regardless of what's applied from outside. We won't tell you otherwise: fire-fighting equipment and suppression systems reduce risk and slow spread. They don't guarantee an outcome.
None of this is a one-off fix. A detection strategy that's compliant on day one drifts out of alignment as charging habits change, device numbers grow and sensors age. That's why the maintenance relationship matters as much as the initial install. A detector justified for last year's charging pattern won't necessarily catch next year's failure.
BS 5839-1:2025 calls for twice-yearly inspection and testing of Fire Detection and alarm systems, on top of weekly user checks. Off-gas and VOC sensors carry their own servicing profile. Some modern units run largely calibration-free with long service lives. Older electrochemical sensors typically need calibration and replacement on a two-to-five-year cycle. Catalytic or infrared sensors sit somewhere between the two. None of these figures are fixed - always check the specific manufacturer's servicing schedule rather than assuming a single interval applies across every sensor type.
Here's what we'd suggest as a next step, in order:
Start that review with our free business Fire and Security survey or call our team directly on 01702 447800. Get it right and the next failing cell is far more likely to announce itself as off-gas at 11pm - not as smoke in a corridor after the fact.
No. BS 5839-1:2025 is the British Standard covering Fire Detection and alarm system design in non-domestic premises. It's a voluntary code of practice, not a legal requirement in itself. Your legal duty sits with the Regulatory Reform (Fire Safety) Order 2005. It requires a suitable and sufficient Fire Risk Assessment (Article 9), general fire precautions (Article 8) and maintained Fire Safety equipment (Article 17). Non-compliance that puts people at risk of death or serious injury carries an unlimited fine and up to two years' imprisonment. In practice, though, BS 5839-1:2025 functions as the recognised benchmark because Approved Document B references it directly. It's the standard your insurer is likely to expect you've already considered.
Because it's designed to catch smoke and a failing cell gives off a different warning sign first. Before flames appear, a cell in trouble typically off-gasses. It vents a cloud of flammable, toxic vapour made up of electrolyte solvents and hydrogen, often minutes before any visible fire. Conventional point smoke detectors weren't designed to catch that early enough. A properly justified BS 5839-1:2025 strategy for a charging or storage area typically combines multi-sensor detectors (smoke, heat and carbon monoxide) with aspirating smoke detection. That gives continuous early sampling. Off-gas and VOC detection is a genuinely strong supplementary technology. But it isn't yet a BS 5839-1:2025 primary detector - there's no dedicated UK detection standard governing it. It should sit alongside compliant detection, not replace it.
Not on its own and not reliably. Lithium-ion battery fires aren't Class D - that classification covers combustible metals such as magnesium and titanium, not lithium cells. Industry sources report that BS ISO 3941:2026 introduces a new Class L classification for lithium-ion battery fires. But that's a classification, not a performance rating - there's currently no BS EN 3 performance rating for extinguishers against lithium-ion fires. The FIA's own December 2020 guidance states plainly that "complete extinguishment may not be possible" with portable extinguishers. Manufacturer guidance goes further, warning that portables won't extinguish anything larger than an e-scooter. Sprinklers, where fitted to BS EN 12845:2015+A2:2026, cool and control fire spread rather than extinguish a pack in thermal runaway. Plan around early detection and containment, not guaranteed extinguishment.
Three sectors stand out in the evidence. Logistics and last-mile depots run high-density overnight charging for e-cargo bike and van fleets, often unattended for hours at a stretch. Waste and recycling sites face batteries hidden inside discarded electricals that get crushed or damaged during collection and processing. Battery-related fires in refuse vehicles and waste facilities exceeded 1,200 in 2023/24, a 71% year-on-year rise (Environmental Services Association and Material Focus). Self-storage carries a harder problem again: operators often have no reliable way of knowing what's in a unit. It could be an e-bike, a stack of power banks or nothing of concern - until something goes wrong.
BS 5839-1:2025 calls for twice-yearly inspection and testing of Fire Detection and alarm systems, on top of weekly user checks. Off-gas and VOC sensors, where fitted, carry their own servicing profile. Some modern units run largely calibration-free with long service lives. Older electrochemical sensors typically need calibration and replacement on a two-to-five-year cycle. Catalytic or infrared sensors sit somewhere between the two. Always check the specific manufacturer's schedule rather than assuming one interval fits every sensor type. None of this is a one-off fix. A strategy that's compliant on day one drifts as charging habits change and device numbers grow. That's why a recurring servicing and calibration arrangement matters as much as the initial installation.
