Battery stock may look like identical boxes, yet be in different risk states
Warehouse safety begins not with what to do after a fire is visible, but with how quickly normal stock can be distinguished from abnormal stock. Lithium-ion batteries in particular may have a history of impact, crushing, water immersion, overcharge, manufacturing defects, or internal short circuits even when they look intact. Conversely, the mere presence of batteries does not make every warehouse the same high-risk facility. The actual risk changes with the cell chemistry, format, packaging, individual capacity, total stored quantity, state of charge, rack arrangement, sprinkler conditions, and whether charging takes place.
First, distinguish a logistics warehouse that stores packaged new products from an energy storage system (ESS), in which batteries are electrically connected and charged or discharged. NFPA 855 and UL 9540A primarily address the installation of stationary ESS and thermal-runaway propagation testing. The gas release, reignition, propagation, and detection principles identified in those materials are useful for warehouse risk assessment, but they do not mean that ESS numerical requirements can be applied unchanged to general inventory warehouses. Korean facilities must also check fire-safety laws, building and hazardous-material conditions, insurance criteria, manufacturer instructions, and consultation with the fire authority having jurisdiction.
The first step is to record the condition of inventory
Early detection begins before sensors are installed. Registering only the product name on receipt makes it difficult to judge what is dangerous and to what extent when an abnormality occurs. At a minimum, it is advisable to connect the inventory system to chemistry and cell format, packaging unit, rated energy, quantity, manufacturing lot, state-of-charge range, receipt date, whether impact or water immersion occurred in transit, and whether the item is returned, used, or intended for disposal. Safety information and manufacturer emergency-response materials must also be immediately available to site personnel and firefighters.
It cannot be simplified into a claim that risk always rises at the same rate as state of charge rises, but stored energy may affect heat release and propagation in an incident and should be considered. NFPA research materials describe how high state of charge and high-temperature storage affect degradation, and cite industry practices using roughly half charge for long-term storage performance. A uniform charge rate must not, however, be specified for every product. Leaving cells fully discharged for a long time can also cause cell degradation and problems during recharging. Warehouse policy should apply the manufacturer's specified storage charge level, temperature, humidity, recharge interval, and transport requirements by product group.
It is practical to divide inventory at least into normal new products, uninspected returns, externally damaged items, items suspected of water immersion or impact, recalled items, and items awaiting disposal. Mixing different states in the same rack and workflow can return an abnormal item to normal outbound stock or lead a worker to attempt charging without knowing the risk. Barcode or WMS status values must match the physical marking and isolation location.
Separate damaged items from the normal logistics flow immediately upon discovery
Receiving and returns areas are where risks first become apparent. Dents, cracks, swelling, melting marks, corrosion, damaged wires, and leakage, as well as unusual heat, electrolyte odor, hissing or crackling sounds, are abnormal signals. UL recommends inspections to identify such external and internal damage indicators early. FDNY likewise advises stopping use and charging if deformation, overheating, leakage, unusual odors, or sounds are present.
Inspectors must not automatically move a damaged item by hand or open its packaging merely because it has been found. If heating or gas release is already under way, impact during movement and worker exposure can increase the risk. Site procedures must include the person responsible for deciding whether movement is possible, the handling equipment to use, the control radius, protective equipment, isolation container, and notification criteria. Only items assessed as safe to move should be sent to a designated isolation area away from normal stock, combustibles, ignition sources, and the building's main traffic routes.
UK Environment Agency guidance for waste-battery facilities calls for damaged batteries to be managed in suitable fire-resistant containers with inert filler, with temperature changes checked regularly in a cool, dry, ventilated isolation location. This, too, is not a universal prescription for every new-product warehouse. An expert must confirm that the container is suitable for the battery's size and condition, potential gas release, and transport and disposal rules. The key is not to leave damaged items temporarily in any vacant space, but to manage them through an exception flow with an owner and a removal deadline.
Temperature, gas, and smoke observe different moments
It is dangerous to assume that one sensor can find every thermal runaway first. A battery abnormality can progress through internal heating, cell venting and gas release, fumes or smoke, and flame, but the order and rate vary by product and failure mode. Early warehouse detection should therefore layer multiple measurements of different physical quantities and ensure that alarms lead to real action.
1. Operating data and work records
For charging batteries or connected equipment, BMS cell voltage, current, temperature, insulation faults, and charger errors are the closest status information. Yet packaged individual inventory is often not connected from its BMS to a central system. Relying on the BMS alone then creates a monitoring gap. That is why work records such as forklift impacts, drops, water immersion, and reasons for return must be viewed with alarm data.
2. Temperature detection
Fixed thermal-imaging cameras, infrared sensors, and rack or zone temperature sensors can help find localized overheating or a temperature rise relative to surroundings before visible smoke. But boxes, pallets, and rack structures can shield cell heat, delaying observation of internal heating. Sunlight, heaters, and forklift exhaust heat can also create false alarms. Rather than using one absolute temperature value, combine the zone baseline temperature, rate of rise, and deviation from adjacent points, and test fields of view and blind spots in the actual layout.
3. Vent-gas detection
DOE's energy-storage safety strategy explains that cell venting can occur before full thermal runaway and may be missed by smoke or heat detectors, so gas sensors measuring one or more components such as volatile organic compounds, hydrogen, carbon monoxide, and carbon dioxide can provide early warning. A single specified gas is not, however, a universal indicator across all battery chemistries and failure stages. Interfering sources such as cleaning solvents, vehicle exhaust, packaging materials, and hydrogen generated during charging also exist. The detection target and alarm value must be set using test data for stored products, ventilation rate, sensor cross-sensitivity, and mounting height.
4. Smoke and fire detection
Aspirating or spot smoke detectors alarm when fumes and smoke reach the detection location. Heat detectors and flame or radiant-heat detectors respond to different fire stages and environments. High ceilings and airflow within racks can dilute or move smoke, so ceiling-only detection can be delayed. NFPA storage-facility research mentions smoke, gas, and leak detection together with automatic sprinklers, compartmentation, ventilation, and continuous notification because the combination of protection systems matters more than a single device. A fire-safety designer and the authority having jurisdiction must decide detector type and placement using the actual ceiling height, HVAC, racks, and storage height.
For alarms, correlation and delivery paths matter more than sensor count
Evacuating the entire facility immediately because a single temperature sensor briefly spikes can reduce response capability through frequent false alarms. Conversely, waiting until two types of alarms both sound can miss a fast incident. Good logic distinguishes the meaning of a single alarm, its duration, its rising trend, and other signals in the same zone. For example, a persistent temperature deviation or rising gas concentration can start a site-verification stage; concurrent gas and temperature rise, or confirmed smoke, can automatically shift to a higher stage. Actual thresholds are not universal numbers; they must be verified using product tests, sensor performance, space volume, and ventilation conditions.
An alarm must not remain only on the fire-control-room display. It should show the zone address, rack number, inventory type and quantity, damage history, adjacent combustibles, and whether it is an isolation area. It must reach the responsible person and fire-alarm receiving equipment outside working hours, with backup paths for communication failure and power loss. Whether monthly functional checks alone are sufficient should also be reviewed. Regularly check sensor contamination, blocked sampling piping, expired calibration, and blind spots caused by rack changes.
Design ventilation separately for normal and emergency operation
A cool, dry, ventilated environment is a general storage principle, but normal HVAC does not guarantee safe handling of thermal-runaway gases. If flammable gas released by a battery accumulates in an enclosed space and then ignites, a deflagration hazard can result. UL 9540A installation-level testing evaluates heat and gas release, deflagration and reignition hazards, and the effectiveness of protection systems. DOE materials likewise link gas detection with fire alarms and treat explosion protection as a separate design issue.
Emergency ventilation is not the simple function of always running fans at maximum. It must not send gas toward workers, egress routes, adjacent rooms, or spaces containing ignition sources. Analyze together the ignition potential of fans, dampers, and electrical equipment; discharge location; make-up air; operation during power loss; and the relationship to fire-service tactics. Ventilation may lower concentrations in some incidents but can affect oxygen supply or contaminant spread under other conditions. Whether to start automatically and the stop conditions must therefore be determined through fire and explosion risk analysis and consultation with the fire authority having jurisdiction.
Separation and compartmentation create time after detection
If there is only early warning and rapid propagation to adjacent pallets, people have no time to decide. Separate normal inventory from returned and damaged items, and keep batteries away from ordinary combustibles, ignition sources, charging areas, and egress paths. Distinguish different chemistries and packaging formats so they can be identified. Do not store batteries at entrances or access routes to firefighting equipment.
Rack spacing, storage height, pallet-level separation, fire-resistant compartments, and sprinkler layout cannot be set by one arbitrary number. Just as NFPA 855 treats battery storage in a separate chapter and NFPA research points to code gaps and fire-service access issues in large warehouses, stored quantity and packaging, ceiling height, structure, and test results change the design. When changing rack layouts or increasing battery inventory, reassess whether the existing sprinkler design and detection coverage remain valid.
Charging adds failure potential and heat-generation factors beyond storage. If charging is necessary, do not perform it temporarily at normal storage racks; separate it into a dedicated area with compatible chargers, electrical protection, ventilation, detection, separation from combustibles, and work supervision. Do not charge batteries whose damaged or returned condition has not been confirmed.
Incident-escalation stages must make everyone take the same actions
Attention stage
When a slight deviation from one sensor, packaging damage, or an impact history is found, mark the location and hold shipment and charging. A trained person checks remote data and surrounding conditions. Do not arbitrarily open or move an item whose safety has not been confirmed. Record the cause and action result, then inspect the same lot and adjacent inventory.
Alert stage
If multiple signs appear, such as persistent temperature rise, vent gas, odor, abnormal sound, or fumes, stop work in the affected zone and move people outside the control line. Cut power and charging only where designed procedures exist, and apply notification criteria for the fire-protection manager and fire authority. Avoid having people hurriedly carry a battery in a hazardous state. Track changes with cameras and remote sensors, but do not send people back in to confirm.
Emergency stage
When smoke, flame, a rapid rise in temperature or gas, a popping sound, or cell ejection is confirmed, activate the building alarm and evacuation procedures and report to 119. Employees' role is not to attempt initial suppression, but to prioritize life safety, access control, and accurate information transfer. Provide firefighters with the zone plan, battery type and quantity, state-of-charge information, safety information, power-isolation locations, detection trends, and adjacent hazards. Manual operation of automatic sprinklers and smoke-control or ventilation equipment follows the pre-plan and incident command.
FDNY materials warn that batteries can continue generating heat and reignite hours or days after visible fire is extinguished. Do not therefore immediately organize inventory or resume operations after extinguishment. Follow the observation time, thermal-imaging and gas monitoring, debris isolation, contaminated-water and waste handling, and re-entry approval procedures determined by fire authorities and experts. Reassess inventory with the same transport or impact history as the incident lot.
Ten checks when applying this on site
Are battery type, capacity, quantity, state of charge, and location connected to real-time inventory information?
Are returned items and items suspected of damage, water immersion, or impact immediately locked out of the normal outbound flow?
Is the isolation area separated from combustibles, ignition sources, and egress routes, and monitored regularly?
Have temperature-detection blind spots been tested with heat generation inside packaging and racks considered?
Does the selected gas sensor suit the release components of that chemistry and on-site interfering sources?
Do smoke, heat, and gas alarms transmit the zone address together with inventory information?
Does ventilation operation avoid spreading gas to other work areas or egress routes?
After storage changes, are sprinkler, compartmentation, and fire-service access conditions still maintained?
Are there alternative procedures for off-hours notification, power loss, communication failure, and sensor failure?
Have evacuation, reporting to 119, on-site information provision, and reignition monitoring been trained using real scenarios?
The goal of early detection is earlier decision-making
An early-detection system for a battery warehouse is not a purchase list for one thermal-imaging camera or gas sensor. It is an operating system that knows inventory condition, physically separates damaged items, makes temperature, gas, and smoke signals complement one another, and connects ventilation, compartmentation, sprinklers, and alarm delivery. The first task is not to put the same equipment in every facility, but to investigate stored products and workflow and decide who stops what and how far people evacuate for each abnormal signal. When those decision criteria are verified against actual test data and jurisdictional requirements, sensors finally create time that can be used before an incident escalates.