Industrial fire protection has, for most of its history, begun at the moment a flame appears. A detector sees the fire, a panel raises the alarm, and a suppression system discharges. The whole chain starts after combustion, by which point the question is no longer whether a fire will happen but how much it will cost to control. Here, we address radiant-energy fire detection, the family that senses the heat and light of a developing fire, rather than gas or smoke detection, which work on different principles. This approach worked when hazards were predictable and suppression was fixed: A sprinkler over a known risk, a deluge system over a fuel tank. The weakness shows when the hazard does not stay where it is supposed to, or when the most dangerous phase happens before any flame appears. Two sectors now expose that weakness daily, and they do it in opposite ways: Battery energy storage and waste.
Heat before flame: The battery problem
A lithium-ion cell entering thermal runaway doesn’t start with a flame. It heats up and the cell undergoes a self-sustaining reaction, venting gas and rising in temperature. If that heat is not removed, it propagates to the next cell and the next battery and/or container. By the time there is open fire, propagation is under way and suppression applied then does little to stop it. The decisive act is earlier, during the temperature rise, before ignition, and is the phase a flame detector cannot see because there is no flame yet to detect. The standards have started to recognise this. NFPA 855, the standard for stationary energy storage, named thermal imaging explicitly in its 2026 edition, where the previous edition was far narrower, and UL 9540A underpins it as the test method for thermal runaway propagation.
The unpredictable hazard: Waste
Waste facilities present the opposite problem. Where a battery fire is hidden in its early phase, a waste fire is hidden in its location. A tipping hall or refuse-derived fuel store is a large open volume where the next ignition could occur anywhere: Spontaneous combustion in compacted material, a discarded lithium battery in the domestic stream, or a leakage of reactive mixed chemicals. Material recovery facilities are among the most frequently fire-affected industrial sites anywhere, and the incidence is rising as more batteries enter the waste stream. Fixed suppression cannot cover this efficiently, because you cannot put a sprinkler over every square metre of a tipping floor and expect it to find a smouldering pocket two bays away. The hazard moves and the protection has to be able to move with it.
Closing the loop: From detection to interception
The technology that answers both problems works on a simple principle. A thermal imaging IR system scans the protected area continuously, identifying a heat anomaly regardless of its cause, before flame. Instead of sending that information only to an alarm panel, the system passes the coordinates of the hotspot to a controller, which drives a remote-controlled extinguishing monitor to that exact position and delivers a targeted water jet to cool the source. The monitor aims at the threat rather than to a predefined zone, which means earlier intervention, far less water and far less collateral damage. In a battery installation it means cooling a container in early runaway before propagation, and in a waste hall it means reaching a smouldering pocket before it becomes a fire that closes the facility. The principle is the same across industry sectors: Detect the heat, locate it and cool it, before there is a fire to fight.
The false-alarm problem, and the honest position on ‘intelligence’
Any detection system that cries wolf will eventually be switched off. This is the single most common reason early-warning systems fail in practice: The operators lose trust in the alarms, mute them and the protection becomes decorative. A thermal system on a busy site will see sunlight on metal, the hot exhaust of a passing loader, the reflection off glass. A system that cannot tell these apart from a developing fire is worse than useless, because the operator learns to ignore it. The answer is signature handling: The system recognises the heat patterns that are not fires, applies confirmation logic before it acts, and allows defined areas to be excluded or tested independently. A word of caution on language, because the industry has earned some scepticism. Many systems are now marketed as AI or ‘self-learning’, and in most cases what sits behind the label is a large library of pre-loaded signatures and threshold logic. That can be very good, but it is not learning in any meaningful sense, and the honest measure of one of these systems is not the marketing term attached to it, but whether it reliably tells a fire from a high-temperature vehicle exhaust and goes on doing so as site conditions change.
Where the standards sit, and where the technology has gone
Flame detector (Radiant Energy sensing) approval was built around one question: Is there a flame, and has the panel been told. FM Approvals evaluates radiant-energy detectors under Approval Standard 3260, dividing the field into flame detectors and spark or ember detectors, with video-based units under a parallel standard, 3232. Both describe a device that sees a fire already in progress, then signals a panel. The UAE Fire and Life Safety Code of Practice inherits the same logic, defaulting detection to NFPA 72. There is no category for a system that detects a rising heat signature before flame, and none for a detector whose output is anything other than an alarm, so a camera that locates a sub-ignition hotspot and aims a monitor at it has no standard to refer it to. It is approved as a flame or video-image detector, a classification that describes the sensor it is built from rather than the job it performs. NFPA 855 naming thermal imaging is the first real crack in that wall, and the wider framework has not yet followed.
The Middle East requirement, and where this is heading
For industrial and oil and gas facilities across the region, approval is not optional and the conditions are particularly hazardous. A system must satisfy the Authority Having Jurisdiction, which in the UAE means alignment with the Fire and Life Safety Code of Practice and Civil Defence acceptance, and increasingly it has to function in or around potentially explosive atmospheres, which brings hazardous-area certification for the electrical equipment on top of the fire-performance approval. This raises the bar usefully, filtering out systems that perform under carefully controlled test conditions but cannot be certified for the environment where the client needs them. A coupled thermal-imaging and suppression system can hold full FM and Civil Defence approval today, certified under the flame-detector standards that exist, while delivering well beyond the minimum those standards describe. The hazards have changed, the technology has been solved by moving the point of intervention earlier, and the standards are following but slowly, so a buyer who waits for the rulebook to catch up will be protected only to yesterday’s requirement.
References:
NFPA855: Standard for the installation of Stationary Energy Storage Systems
UL9540A: Standard for Test Method for Evaluating Thermal Runaway Fire Propagation in Battery Energy Storage Systems
FM3260: Radiant Energy-Sensing Fire Detectors for Automatic Fire Alarm Signaling
FM3232: Video Image Fire Detectors for Automatic Fire Alarm Signaling
NFPA72: National Fire Alarm and Signaling Code
This feature appeared in issue 70 of Fire Middle East magazine.
