No matter what we throw at fire detection, from drones to advanced prediction models and watch towers, predicting when and where blazes will start remains challenging. And not all fires are created equal.
Scientists from South Korea's Hanyang University have taken a new approach, creating an electronic sticker that attaches to a variety of surfaces including tree bark to transmit continuous data via Bluetooth. It doesn't just detect the point of ignition but can identify the different kinds of fires, helping first responders tackle fires faster and smarter.
This flexible sensor targets the deep-ultraviolet (DUV) light emitted by fire, and it can be safely attached to tree trunks and leaves – as well as power cables or industrial equipment, making it useful beyond wildfire detection.
Its power is in the sensor that focuses on the spectrum known as "solar-blind" DUV. While sunlight obviously contains UV radiation, wavelengths between 200 and 280 nanometers are almost entirely absorbed by Earth's atmosphere before they reach the ground. Essentially, this means that if DUV radiation is detected at ground level, it's likely the result of combustion.
Basically, the sensor ignores sunlight and most background light "noise" but get triggered almost instantly when it detects DUV, or the hallmarks of fire.
The sticker contains a zinc tin oxide nanocomposite photodetector, which creates and transmits an electrical-signal alarm when DUV light is present. It's also self-powering and durable, able to bend and flex on leaves or stay on tree trunks for months. In testing, it was still 96.7% effective after 180 days.
However, the researchers wanted it to do more than just sound an alarm but detect different kinds of fires – critical for how we respond to blazes. The team tested the device using a blowtorch, solid-fuel flame and gas stove, discovering that each type produced its own DUV light "footprint". Using machine learning, the sensor could then identify not just the type of flame it's detected but determine how far away the fire was.
Because wildfires, once alight, provide challenging variables – including wind and fuel source (the flammability of trees varies greatly) – this distance detection could better inform those charged with putting the blaze out as to the direction of the burning in real-time.
The unpredictability of wildfires once they've started traditionally puts first responders in a defensive position, making these destructive events hard to contain.
It's still early days, but as the prevalence of wildfires continues to grow, a forest full of plant-safe sensors has the potential to detect trouble before smoke, cameras or other early alarm systems identify emerging blazes.
"This integrated approach provides a practical route to reliable fire monitoring, relevant to early-stage fire monitoring concepts for wildfire and industrial safety applications," the team, led by Ho-Chun Yu, a professor in the Department of Electronic Engineering at Hanyang University, notes.
The research was published in Science Advances.
Source: Hanyang University via Scimex
