Sensors to detect smouldering cables before they catch alight
They say that where there's smoke there's fire, but when it comes to electrical systems, by the time the smoke is detected, it's often too late. To raise the alarm early, a team of researchers from the Karlsruhe Institute of Technology and Karlsruhe University of Applied Sciences have developed hybrid sensors that detect gases given off by overheated plastic cables before too much damage can occur.
Electrical fires are usually detected by someone smelling burning or noticing that the insulation is off color. However, when a situation reaches that stage, valuable components may already have suffered damage or a dangerous fire may already be on the verge of breaking out. What the Karlsruhe team was looking for was more of an early warning system that catches the overheating as early as possible.
What they came up with a new type of sensor that detects the gases given off by overheating cables, then analyzes them to determine if they indicate a possible fire. The sensors are made up of four areas constructed of different metal oxides that alter their resistance at different temperatures when in contact with gases. By cyclically heating and cooling each area, comparison of simultaneous measurements of the resistance of each one produces a profile that can identify a particular gas.
In addition to the sensors themselves, each unit includes as its basic design microelectronics and associated algorithms. These not only detect the gases, but measure them and carry out data analysis to eliminate false positives, such as from carbon dioxide.
The team says that once the technology is mature, it can not only be used for fire detection, but also for seeking out toxic molds, explosive gases in fertilizer silos, and natural gas leaks.
"Hybrid sensors can be used anywhere as separate systems or in a network. They may also be combined with classical safety technology, such as infrared cameras," says Dr. Hubert Keller, Simulation and Measurement Project Head of KIT's Institute for Applied Computer Science.
The team's research was published in Sensors & Transducers Journal.