Researchers develop early warning system to detect fires in lithium-ion batteries

Researchers at the US National Institute of Standards and Technology (NIST) have innovatively developed an early warning system aimed at detecting potential fires in lithium-ion batteries. The nature of these types of fires is particularly hazardous, with the batteries capable of emitting jets of flame reaching temperatures as high as 1,100°C, akin to a blowtorch, within merely a second. This poses significant risks, especially as by the time traditional smoke alarms are triggered, the fires could already be in advanced stages and far hotter than typical flames.

Lithium-ion batteries, commonplace in everything from personal gadgets to large transport vehicles, have a history of unexpectedly combusting. The incidents can occur in varied locations such as the holds of ships or the passenger compartments of aircraft, where the risks are considerably more dangerous without prompt intervention. Therefore, a system that can provide preemptive alerts, even by a few minutes, could vastly mitigate potential disasters.

The foundation of this early warning system lies in recognising a specific sound emitted during a battery’s failure. This sound, described as a “click-hiss” akin to opening a soda bottle, occurs when a safety valve in the battery’s casing breaks to alleviate pressure caused by internal chemical reactions. While this auditory signal is a clear indicator of impending failure, the challenge has been distinguishing it from similar everyday noises, like a dropped paper clip.

To address this, the researchers employed machine learning technology. They collected audio footage from 38 battery explosions, modifying the pitch and speed to generate over 1,000 auditory samples. This set of data was then used to train a software algorithm to accurately identify the unique sound of a safety valve breaking. The resulting system proved effective, detecting overheating batteries with a success rate of 94 percent.

Wai Cheong “Andy” Tam, a member of the NIST research team, put the algorithm through rigorous testing by subjecting it to a variety of noises such as footsteps, door closings, and the sound of opening cans. The algorithm remained largely unfazed, indicating a high level of accuracy and reliability.

With lithium-ion batteries becoming ubiquitous in more environments, this early warning system holds significant promise. Once the technology is fully developed, it could potentially be integrated into homes, office spaces, and electric vehicle parking garages, providing valuable minutes of warning before a battery failure leads to a fire. In tests, the sound of the safety valve offered approximately two minutes of advance notice, although further research will continue with a broader range of battery types to refine the system’s capabilities.

NIST articulates the system’s significance with a light-hearted note, suggesting that a “healthy relationship” with lithium-ion batteries might be achieved through careful auditory monitoring. The ongoing enhancements in this system reflect a proactive approach to improving safety standards associated with these vital yet volatile power sources.

Source: Noah Wire Services