Technology
Thermal Runaway Detection
Thermal runaway is a self-sustaining exothermic cascade inside a lithium-ion cell. Once initiated, the reaction generates heat faster than the cell can dissipate it, and the process becomes irreversible. Detection that arrives after initiation can only support containment.
The engineering objective is therefore to detect the precursor chemistry, not the runaway itself. This page sets out the initiation sequence, the signals available at each step, and where interruption is still possible.
How thermal runaway initiates
The cascade begins with SEI decomposition at moderately elevated temperature, which exposes fresh anode surface to electrolyte and drives further exothermic reaction and gas generation. As temperature climbs, the separator softens and then melts, allowing direct contact between electrodes and a large internal short. Cathode decomposition releases oxygen, which sustains combustion inside a sealed cell, and the vent opens under pressure.
Gas generation therefore precedes and accompanies every stage. The first vent — the release of hydrogen and electrolyte vapour through the cell's pressure relief — occurs before the energetic phase and is the last clean opportunity to intervene.
Propagation to adjacent cells is driven by conducted and radiated heat plus ejected hot material. Interrupting propagation after the first cell has gone into runaway is a containment problem; preventing the first cell from getting there is a detection problem.
What each detection method can actually see
Ranking detection methods by the stage at which they respond gives a clear picture of the intervention window each one provides.
| Stage | Physical change | Detected by | Intervention still possible? |
|---|---|---|---|
| Degradation | SEI growth, lithium plating, trace H₂ | Molecular gas sensing | Yes — wide window |
| First vent | H₂ and electrolyte VOC release, pressure relief opens | Molecular gas sensing, pressure | Yes — isolate, ventilate, pre-arm suppression |
| Exotherm | Cell and casing temperature rise | Thermistors, thermal imaging | Marginal |
| Smoke | Aerosolised particulate | Smoke and aspirating detection | Containment only |
| Runaway and propagation | Fire, jet flame, cell-to-cell spread | Fire alarm, flame detection | Containment only |
How large is the intervention window?
The available window depends on chemistry, state of charge, cell format and how the fault developed. Slow-developing faults in LFP installations can off-gas for hours before any thermal signature; energetic internal shorts in high-energy NMC cells compress the sequence into minutes.
In both cases the ordering is preserved: gas first, heat second, smoke third. Detecting at 10 ppm hydrogen at the module rather than at percent-LEL in the room is what converts a window measured in seconds into one measured in minutes or hours.
Designing the response, not just the alarm
An early alert should drive a graded response defined in advance. Typical staging: at first detection above baseline, log and notify, reduce charge rate on the affected string. At sustained rise, isolate the string and increase ventilation. At confirmed vent signature (H₂ plus electrolyte VOC), pre-arm suppression on the identified zone and dispatch with a specific rack and module reference.
Battery Sense supports this with dry potential-free contacts for direct interlocks alongside Modbus, CAN and 4–20 mA telemetry, so the response can be implemented in existing control logic rather than requiring a new supervisory system.
- Stage 1: baseline departure — log, notify, derate charging
- Stage 2: sustained rise — isolate string, increase ventilation
- Stage 3: vent signature confirmed — pre-arm suppression on the localised zone, dispatch
- Stage 4: thermal confirmation — full shutdown and suppression release
Frequently asked questions
Can thermal runaway be stopped once it starts?
Once a cell has entered self-sustaining runaway, the practical objective is preventing propagation to neighbouring cells and limiting damage. Stopping the event requires intervening earlier, during off-gassing and at first vent.
How long before thermal runaway does a battery start off-gassing?
It varies from minutes to hours depending on chemistry, state of charge and failure mode. Slow-developing LFP faults sit at the long end; energetic internal shorts in high-energy cells at the short end. Detection sensitivity and sensor proximity determine how much of that window you actually capture.
Is temperature monitoring useless for thermal runaway detection?
No — it is a necessary confirmation layer and essential for thermal management. It is simply not an early-warning layer, because the measurable temperature rise at accessible mounting points lags the chemistry.
What does UL 9540A tell us about detection?
UL 9540A characterises fire propagation behaviour of energy storage systems under test, including vent gas composition and quantity. It informs installation design and suppression, and its vent-gas data supports the case for chemical detection as the earliest available signal.
References and further reading
Put molecular detection on your battery assets
Send us your chemistry, enclosure and comms constraints. Our engineers will map BD-100 placement at cell, module or rack level and model the detection window you gain.
Related reading
Battery Off-Gassing Guide
The precursor chemistry in full detail.
Battery Fire Prevention
Turning early detection into prevented incidents.
Hydrogen Sensor for Batteries
The instrument that sees the precursor.
Battery State of Safety
Quantifying margin to runaway as an operational index.
BESS Monitoring
Propagation control in containerised storage.
Off-Gassing vs Thermal Imaging
Which sees the event first, and by how much.
Battery safety intelligence, monthly
Incident analysis, off-gassing research, standards updates (NFPA 855, UL 9540A, IEC 62485-2) and field data from live BESS, UPS and residential deployments. No marketing filler.