30 July 2004

Where there's smoke, there's (not always) fire

An inside look at Smoke Detectors

Part 4. Intelligent alarm systems and false alarms

Part 1 of this series covered flame and heat detectors (The Citizen Scientist, 9 July 2004).

Part 2 covered photoelectric smoke detectors (16 July 2004).

And Part 3 (23 July 2004) discussed ionization smoke detectors.

This final installment covers intelligent smoke and fire detection systems. Reducing false alarms is also covered.

Intelligent smoke alarm systems

Intelligent fire detection systems have various advantages over discrete, stand-alone residential detectors, because they incorporate complex algorithms that can correct calibration drift, poll the system, and so forth. These capabilities include:

• Drift Compensation. This is the ability to compensate for individual detector chamber value fluctuations over time due to increased contaminants inside the detector chamber. As the background chamber value increases due to dust and other contaminants, the difference between the background level and the alarm threshold decreases. If a slow, gradual increase is noted for a specific detector, a new baseline for "normal" can be established. This will ensure that the detector responds properly, while limiting false alarms.

•  Detector Verification. This is the ability to rescan an individual detector that has signaled a momentary alarm indication. This ability helps to cut false alarms due to single transient events. One such event occurred at a large commercial fire alarm manufacturer for which I worked. The human resources department had a photoelectric smoke detector just above the photo-ID camera. When a new photo-ID badge was being made, the flash of the camera coincided precisely with the interrogation of the detector. The resultant rapid increase in incident light on the internal photo-sensor gave a fictitious alarm indication. The result was that the entire building was immediately evacuated. If detector verification were used, then the specific detector giving an alarm condition would have been resampled (polled) several more time in a few seconds. The fire panel would have determined that the transient alarm indication was just a glitch, and the signal would be ignored.

•  Detector Sensitivity Adjustment. An intelligent fire panel can make a sensitivity adjustment based upon time of day or day of the week. For example, the alarm threshold can be adjusted for a lower sensitivity during normal working hours, when many people are present, and adjusted to a much more sensitive threshold during nighttime and weekends, when few people are present. This will reduce the false alarm rate while still providing adequate protection.

•  Automated Detector Test. An intelligent fire panel can automatically test addressable smoke detectors by placing the detectors in a simulated alarm condition and looking at the returned value from each chamber. This testing goes on in the background throughout the day or night to assure functionality of the entire system. If any problems are noted during the automated detector test, a warning condition will be indicated for the particular detector.

For more information on commercial smoke and fire detection systems, here are some of the major manufacturers:

Edwards Systems Technology

Simplex Grinnell

Hochiki

System Sensor

Honeywell Fire Group

Other manufacturers can be found by searching on key words.

False alarms and how to reduce them

Although most smoke detectors work very well in many circumstances, they are prone to false alarms and nuisance alarms. To be semantically correct, here is the System Sensor definition of both terms:

False Alarm. "A false alarm is when a detector goes off due to dust, steam, fibers, or other non-smoke phenomena."

Nuisance Alarm. "A nuisance alarm, on the other hand, is when a detector mistakenly goes off due to smoke such as from welding or cigarettes."

Some well known nuisance alarms are caused by cigarette smoke, smoke from bread toasting, dust, high humidity/condensing water vapor, pollen, baby powder, and smoke from heated cooking oil. Anyone who has ever fried chicken on the stove probably knows what I am talking about.

If you are doing some home repairs, such as installing drywall, and you need to sand the joint compound, or vacuum up what was already sanded, keep in mind that the resultant dust may trigger a nearby smoke detector. Smoke detectors are also made with a fine mesh screen to keep out tiny insects which can cause "bugs" in the system (literally!). Tiny insects have set off many a smoke detector.

NIST has an entire section of technical papers devoted to studying and preventing false alarms and nuisance alarms .

As mentioned before, Ionization detectors become more sensitive when they are dirty. It is advisable to clean your smoke detectors at least once a month. A dirty detector has less headroom between the normal chamber background level, and the alarm threshold. If you have a smoke detector that frequently activates its alarm with no apparent reason, try cleaning it with either a vacuum cleaner, or compressed air.

Testing smoke alarms

You should test your smoke detectors at least once a month, preferably once a week. In commercial fire alarm systems, NFPA 72 dictates that the system should be tested twice a year. There are several methods for doing this. Some detectors include a reed switch that can be activated (closed) by a small magnet. The testing process consists of placing the fire panel in a special "test" mode and using a telescoping rod with a magnet on the end to activate the detector.

Some people prefer to use " Canned smoke " to test the system. The canned smoke contains a hydrocarbon spray that simulates smoke particles. It is a bit ironic that some cans of canned smoke contain flammable material that could actually start a fire. Other people use a burning incense stick, or a magnet to test the smoke detectors.

Most intelligent fire alarm systems have an automated detector test function that periodically places a simulated alarm condition on the detector. The chamber value returned is compared to its expected alarm value. This automatic testing helps to weed out any defective smoke detectors before there is a problem. Even with automated testing, NFPA 72 dictates that the fire system must be tested twice a year.

In summary, I hope you understand that there is a lot more to fire detection systems than meets the eye. Although smoke detectors can detect smoke, they also react to a wide variety of other similar material. There is no true "Smoke detector" that will only detect smoke that I know of, but if used in accordance with NFPA guidelines, they can be more effective and discerning of smoke.

For more information

Additional information about smoke is available at libraries and on the web, including the following hyperlinks:

Particle Size Chart

History of NBS/NIST Research on Fire Detectors

Radiation Measurements

The Properties and Climate of Atmospheric Haze

Comparison of aerosol size distributions, radiative properties, and optical depths determined by aircraft observations and Sun photometers during SAFARI 2000