This project will develop the technology to significantly reduce residential fire deaths through demonstration of economical advanced detection systems and the proposal of enabling codes and standards provisions for residential applications. A systems approach with research in sensors and sensor modeling, data fusion, human interaction, and implementation strategies will enable early warning fire detection to significantly reduce egress time while also reducing nuisance (false positive) alarms. The ability to discriminate between the early stages of an unwanted fire and a range of nuisance sources that present themselves at much lower signal levels than current alarm conditions will require some combination of sensitive chemical, thermal, and particulate sensing techniques. NIST will measure particle concentrations, chemical species, and temperatures, and assess flame detection and other discriminating phenomena in the very early stages of fires including residential electrical fires and nuisance source events. Data gathered will be used to develop discriminating detection algorithms.
Objective: By 20151, to develop the measurement science that enables (1) commercialization of economical detection systems with more rapid response time and lower nuisance activations, and (2) development of draft codes and standards provisions for residential applications.
What is the new technical idea? The combination of early warning fire detection and nuisance alarm resistance in residential smoke alarms could cut the fire death rate in half. Technology innovations needed for early warning fire detection to significantly reduce egress time will require a systems approach with research in sensors and sensor modeling, data fusion, and implementation strategies. A significant increase in fire sensitivity exacerbates the nuisance alarm problem. Thererfore, it is critical that any research to increase the sensitive of smoke alarms be conducted concurrently with reducing nuisance activiations. The ability to discriminate between the incipient stage of a fire and a range of nuisance sources that present themselves at much lower signal levels than current alarm conditions may require a combination of sensitive chemical, thermal, and particulate sensing techniques and discriminating alogarithms. The range of identified fire sensing techniques is broad (e.g., thermal, gas species, particulate) with particulate detection likely the best for early fire detection. Since light scattering measurements are easily 5 orders of magnitude more sensitive than typical sensitivities of current commercial alarms, we have decided to focus on evaluating light scattering as the detection source in the next generation smoke alarms.
Current residential smoke detection research focueses on upholstered furniture/mattress fires. The fire losses from residential furniture fires may decrease due to the development of new regulations2; therefore it is imperative to evaluate the new detection approachs with the next most significant fire losses in residential fires: electrical failure or malfunction. Electrical failure or malfunction was indicated as contributing to ignition in 44,800 residential structural fires and 472 fire deaths in 2009 (Hall, Home Electric Fires, 2012), whereas cooking is the leading cause of home fires and injuries and responsible for 300 home fire deaths in 2009. Early detection of electrical malfunction or failure via sensing of early degradation products could significantly impact the fire losses and deaths attributed to these ignition sources. Flame detection may provide a “fool proof” way to discriminate cooking activities from real fire hazards at the earliest possible moment for unattended cooking. Beyond advances in sensing technologies, a key relationship that needs to be understood is the interaction of humans to information presented by residential fire detection systems (e.g. obvious nuisance alarm – ignore or disable, failure to awaken a sleeping individual, tolerance of a “learning” phase of an advanced alarm, etc.) For instance, an intelligent fire alarm system that knows when cooking appliances are turned on, what time it is, and where occupants are located and were located in the immediate past may be used to inform occupants much differently than a fire alarm system that does not have this information. This project is aligned with NIST’s Strategic Roadmap on the Reduced Risk of Fire in Building and Communities (NIST SP 1130, 2012).
What is the research plan? This project is separated into two research tasks whose outputs will be used to demonstrate the capabilities of discriminating algorithms to provide early detection and discrimination of nuisance sources.
Task 1 will continue to develop the measurement science to discriminate smoke and nuisance aerosols from light scattering measurments, gas species, and thermal signatures. In Year 1 (FY12), we designed a multi-angle, multi-wavelength scattering apparatus, conducted exploratory tests with selected components, and procured the equipment to build it. In Year 2 (FY13), we built the smoke nephelometer, aerosol polarimeter (SNAP). This purpose of this research tool is to perform multi-angle, mult-wavelength light scattering measurements of fire smokes and other aerosols for smoke detection research, and other fire smoke measurements3. In Year 3 (FY14), we will continue to make measurements with SNAP of very early fire environments and select nuisance sources in addition to gas measurements (carbon monoxide and carbon dioxide), thermal measurments, humidity and particle size distributions. To determine if the onset of a fire can be accurately detected, an effort will be made to determine if it is possible to accurately discriminate between background particles (e.g., nuisance aerosols or smolder smokes) and a mix of background particles and soot4. The ability to sense early indications of soot particles would provide critical information for very early warning fire detection. The response of a commercial two-wavelength photoelectric detector, as compared to SNAP, will be measured for a range of aerosol size distributions5.
Task 2 will develop new fire and nuisance scenarios to challenge advanced detection systems, and continue development of nuisance source test standards. In Year 1 (FY12), we completed aerosol exposure experiments of 8 residential smoke alarms to provide input into the development of standardized nuisance tests. We assisted the Oak Ridge National Laboratory by evaluationg their prototype advanced detector that was funded by the US Fire Adminstration and the US CPSC. Tests were conducted in theNIST fire emulator / detector evaluator (FE/DE) and the small smoke box. Oak Ridge used the results to tune their detection algorithm for the prototype. In Year 2 (FY13), we analyzed the kitchen nuisance alarm and fire tests conducted and published6. We concluded a test series focused on evaluating the performance of the Southwest Sciences SBIR flame detector and a commercial detector head with integrated flame detection to small kitchen range top fires. The focus was on evaluating flame detection to discriminate cooking activities from real fire hazards at the earliest possible moment after transition to flaming. In Year 3 (FY14), we will focus on the precursor signatures from cooking fires and the ability discriminate normal cooking activities from conditions just prior to ignition. We will also continue to conduct to characterize the degradation products of electrical wiring and components to enable the development of advanced detection schemes that are able to sense electrical fire precursors before initiation of flaming.
As a conclusion to this project, in FY14, we will also conduct a series of room-scale fire detection experiments covering a range of fire sources to assess the capability of discriminating alogithms to provide early detection and discrimination of nuisance sources.
 The project is currently in Phase 1. The long term vision is to enable development of commercially available, economical advanced detection systems that provide the earliest practical, highly reliable alarm to fire or impending fire events.
 CPSC 16 CFR 1634 proposed residential furniture regulation, CBHFTI TB117 residential furniture smoldering ignition regulation, deployment of fire safe cigarettes.
 SNAP has the capability to measure light scattering at multiple fixed angles from up to four different wavelengths of laser light (400nm, 635 nm, 785 nm, and 980 nm) approximately simultaneously in the upper section of the device. In the lower section of the device is designed to measure the light scattering of polarized light at angles from ~5o to 150o at a fixed laser wavelength of either 405 nm, 635nm, or 850 nm.
 Soot is formed as a by-product of combustion. The more accurately the system can detect the onset of soot formation, the more accurately the system may be at detecting the fire at the very early stage of fire development.
 Previously gathered data from full-scale fire detection tests were analyzed to assess to what extent earlier fire detection could potentially impact residential fire safety. That analysis will be used to focus on detection strategies that will improve detector performance, and will provide a scientifically sound basis for new smoke detector fire test standards.
 Cleary T, et al. NIST TN 1784 “Smoke Alarm Performance in Kitchen Fires and Nuisance Alarm Scenarios”, 2013.
Impact of Standards and Tools:
 (1) “Effective January 1, 2016, smoke alarms and smoke detectors used in household fire alarm systems installed between 6 ft (1.8 m) and 20 ft (6.1 m) along a horizontal flow path from a stationary or fixed cooking appliance shall be listed for resistance to common nuisance sources from cooking.” (2) “Effective January 1, 2019, smoke alarms and smoke detectors used in household fire alarm systems shall be listed for resistance to common nuisance sources.”
Sample sensor test board used during tests of residential smoke alarms. Photo credit: NIST
Start Date:October 1, 2011
Lead Organizational Unit:el
Related Programs and Projects:
100 Bureau Drive, M/S 8664