How Long Is the Off-Gas Window, Really?

The vendor figure is a range, not a number. Li-ion Tamer's own literature gives between 2 and 30 minutes from cell venting to thermal runaway, and says the severity of the abuse sets where in that range an event lands. Peer-reviewed timings cluster well below the ceiling. And the failure modes a vehicle deck actually sees are the ones that sit at the short end — including one with no window at all.

What the vendors actually claim

Less than the headline, and more carefully than the headline. The off-gas category — Li-ion Tamer from Xtralis and Honeywell being the reference product — detects the volatile organic compounds released as electrolyte solvent vents from a failing cell, before reaction gases and before any thermal signature. Xtralis's own material puts the warning at between 2 and 30 minutes and states directly that the severity of the abuse factor greatly affects the time between venting and runaway.

The validation behind it is real but specific. The system delivers consistent early warning across cell manufacturers, chemistries and form factors when tested using the UL 9540A recommended failure method — which is overheating. A DNV GL joint-industry programme found the system able to prevent thermal runaway, without a published rate. In one reported test series the earliest detection of imminent failure came 12 minutes before the event. None of that is a 30-minute guarantee; it is a 30-minute best case under the slowest abuse mode.

What the measurements say

Under slow heating the window is real and it is minutes, not tens of minutes. Golubkov and colleagues, working on large automotive cells, recorded a full sequence under overheating: case swelling at 120°C, cell failure at 176°C, the burst disc opening at 203°C for a first minor venting, then exothermic reactions evolving into runaway at a critical temperature of 248°C with a second major venting. Two venting events, and the first is the one a solvent sensor is built to catch.

The same paper records the other case. In a separate experiment an internal short circuit at an average cell temperature of 170°C triggered a sudden heat release that caused venting and thermal runaway immediately — the two were not separated by any usable interval. Comparative abuse studies say the same thing in aggregate: under overheating a delay window of roughly 350 seconds precedes safety-valve activation, while under nail penetration runaway follows the internal short almost at once, with temperatures exceeding 350°C and violent gas release.

Sensor-side, the published lead times sit in the same band. Electrolyte solvent vapour dominates the release before runaway and reaction gases dominate after it, which is why solvent detection leads: one reported series had DMC detectable at about 390 seconds against 1,312 seconds for hydrogen, and a purpose-built DMC sensor has been reported alarming about 15 minutes before onset. A hydrogen detector at the top of an energy-storage cabin gave about 145 seconds. All of these are overheating-mode figures. All of them are under the ceiling.

The abuse mode sets the window — and a car deck skews short

This is the correction to the version of this post that stood here before, which took the 30-minute ceiling as the working number. The ceiling belongs to slow thermal abuse — the overcharge, the failed cooling loop, the UL 9540A heater. That is the failure population of a stationary battery room, and it is the population the off-gas category was validated against.

A vehicle deck adds a second population. Cargo shifts in heavy weather. A vehicle takes collision damage before it is loaded. A pack is struck by a lashing point or by the vehicle above it on a liftable deck. Those are mechanical abuse modes, and mechanical abuse is the nail-penetration end of the range — seconds between the internal short and runaway, or in the Golubkov case, nothing. Importing a BESS lead time onto a car deck assumes the deck fails the way a battery room fails, and it does not.

The marine caveats, with sources this time

Ventilation does not just dilute the signal; it moves it. EMSA's FIRESAFE II work identified airflow on open ro-ro and weather decks as capable of both delaying detection and delocalising it — the alarm arrives late and points somewhere other than the source. A deck-level solvent sensor is reading a plume that has been carried, not a plume that has spread.

The source is sealed until it is not. A traction pack is validated gas-tight and releases through a calibrated burst path, so the clock a deck sensor can see starts at the aperture rather than at the cell. Whatever fraction of the venting-to-runaway interval elapses inside the enclosure is invisible from outside it.

Background is a narrower problem than the old version of this post suggested. Carbonate solvent vapour — DMC, EMC, EC — is not what an internal-combustion vehicle emits; petrol and diesel hydrocarbons are different species. A solvent-specific sensor therefore discriminates better against a mixed deck than a broad VOC sensor does. What it still has to survive is dilution to ppm level in a compartment of thousands of cubic metres — and the industry's move to a 50% state-of-charge ceiling reduces vent gas volume, which shrinks the concentration available to detect.

What to specify

• Lead time by abuse mode, not a single number. Ask for the overheating figure and the mechanical-abuse figure separately; a supplier who offers one number is offering the ceiling.
• Which failure method the validation used. UL 9540A's overheating method is the slow end by construction — it says nothing about a struck pack.
• What the scheme does in the zero-window case. If there is no solvent lead time, the first evidence is heat and gas at once, and localisation has to come from a different modality.
• Whether the 12-minute or 15-minute figures quoted are best cases from a single series — they usually are — and what the median across the test set was.
• The concentration the sensor needs at the deck, against the vent volume of a pack held at the state of charge the carrier actually accepts.

Sources

• 1. Xtralis / Honeywell — Li-ion Tamer product literature and FAQ: warning of between 2 and 30 minutes of a thermal runaway event; the severity of the abuse factor greatly affects the time between cell venting and thermal runaway; consistent early warning across cell manufacturers, chemistries and form factors when tested with the UL 9540A recommended failure method (overheating); tested in a DNV GL joint-industry programme in which the system was able to prevent thermal runaway. Trade reporting of the system's testing gives an earliest detection of imminent failure 12 minutes before the event.
• 2. Golubkov, A. W. et al. — "Thermal runaway of large automotive Li-ion batteries", RSC Advances (2018) 8(70):40172–40186: under overheating, case swelling at 120°C, cell failure at 176°C, burst disc opening at 203°C (first, minor venting), critical temperature 248°C with second, major venting at 247°C; in a separate experiment an internal short circuit at an average cell temperature of 170°C triggered a sudden heat release that immediately caused venting and thermal runaway; cell temperatures up to 1000°C; gas release up to 25 mol/kWh (600 L/kWh); main venting gases CO₂, CO and H₂.
• 3. Comparative abuse-mode studies — overheating versus nail penetration: under overheating a critical delay window of approximately 350 s precedes safety-valve activation; under nail penetration thermal runaway is triggered almost immediately after internal short-circuit formation, with temperature escalation exceeding 350°C and violent gas release. Drawn from Energy (ScienceDirect S0360544223014214) and related abuse-comparison work.
• 4. Sensor-side lead times and mechanism: electrolyte vapours dominate gas release before thermal runaway with reaction gases dominating after (Cell Reports Physical Science / ScienceDirect S2666386423005362, comparative venting-gas study across trigger methods); DMC detectable at ~390 s against ~1,312 s for H₂ in one reported series; a mixed-phase bismuth oxide DMC sensor reported alarming approximately 15 minutes before runaway onset (Sensors and Actuators B, ScienceDirect S0925400526003291); an H₂ detector at the top of an energy-storage cabin reported warning about 145 s before runaway (Journal of Energy Storage, ScienceDirect S2352152X23009076).
• 5. Marine constraints: EMSA FIRESAFE II identified ventilation in open ro-ro spaces and weather decks as able to both delay and delocalise detection — see FIRESAFE II: Open-Deck Detection Lessons; the sealed-pack egress constraint is set out at Does the Off-Gas Ever Leave the Battery Pack?; the state-of-charge effect on vent volume is at The Argument Over State-of-Charge at Loading.