Smoke alarms have helped halve the number of deaths from home fires in the last 40 years. But, according to Bruce Warmack, a researcher at the Oak Ridge National Lab, the standard detector could be a lot better. And anyone who’s ever struggled to rip the batteries out of their device while cooking a non-fire-starting meal can definitely agree.
About 2,500 people a year still die from residential fires. In many cases the problem was not the lack of an alarm, but that it wasn’t working properly. Either it had run out of power, or the residents had deliberately disabled it (PDF). People often complain about false alarms--most frequently because of something burning on the stove, or moisture coming from the bathroom. One study in North Dakota found that almost half the installed alarms in one community did not work, and that 86% had been disabled purposely because of "nuisance" events.
Warmack points to other problems, too. One: Modern homes, and furnishings made of petrochemical materials, burn more quickly than years ago. Residents now have an estimated three minutes to exit safely, down from 17 minutes in 1975.
And two: Elderly, hard-of-hearing, and young people often fail to respond to standard high-pitched alarms. One study in Australia found that an alarm sounding for 30 seconds did not wake 78% of children aged 5 to 15.
"The technology has been only incrementally advanced in the last 40 years. They have worked pretty well, but there are some shortfalls," Warmack says. A better alarm would tell the difference between nuisances and real fires, alert inhabitants to danger more quickly, and wake up a larger section of the household.
Warmack’s new design includes a microcontroller that’s programmed based on years of previous event data. Sensors feed live information to the controller, allowing it to tell where it is on a "discriminant space" map. If it’s dangerous, and the data shows similarities with previous dangerous events, the alarm will sound.
Warmack says trials have shown that the detector can sense smoldering fires--which can take more than an hour to become flames--30 minutes earlier than standard alarms. It also almost completely eliminates unnecessary nuisance alarms.
The new design includes both standard aerosol sensors, plus a carbon monoxide sensor and a temperature gauge--all of which feed into the discriminant matrix. It also has an amplified low-frequency alarm that research has shown is better at waking up the elderly and hard-of-hearing.
Warmack has taken his prototype to several manufacturers, including Kidde and Bosch, who he hopes will commercialize it. Because microcontrollers cost as little as $1 to manufacture, he doesn’t think their inclusion will add much to the end-price. We’ll have to wait and see.