Analysis of the Response of Smoke Detectors to Smoldering Fires and Nuisance Sources

Transcription

1 Analysis of the Response of Smoke Detectors to Smoldering Fires and Nuisance Sources Prepared for: Maryland Industrial Partnerships and USI Electric Prepared by: Jewell T. Feng and James A. Milke Department of Fire Protection Engineering University of Maryland, College Park January 2012

2 Table of Contents List of Figures... 2 List of Tables... 2 Executive Summary... 3 Literature Survey... 3 Fire Hazard and Nuisance Alarms... 3 Photoelectric Sensors... 4 Ionization Sensors... 4 Ionization versus Photoelectric Sensors and Nuisance Alarms... 5 New Types of Smoke Detection Technologies... 5 Current Regulations Regarding Photoelectric and Ionization Smoke Alarms... 6 Experimental methodology... 7 Orientation Sensitivity Testing... 7 Experimental Setup... 7 Procedure... 7 Orientation Sensitivity Results... 8 Nuisance Source Testing... 8 Sources... 8 Aggressive Nuisance Scenarios... 8 Experimental Program for Nuisance Sources... 9 Cooking Related Tests... 9 Bathroom Related Tests... 9 Instrumentation Procedure Toast Steam Onion Hamburger Vegetable Oil Cement Dust Results Individual Source Results Toast Steam Onion Hamburger Oil Dust Conclusion

3 Acknowledgements Works Cited Appendices Appendix A: Test Data Toast Tests Onion Tests Hamburger Tests Oil Tests Dust Tests Steam Tests Appendix B: Obscuration graphs Toast Tests Onion Tests Hamburger Tests Oil Tests Dust Tests Steam Tests Appendix C Manufacturer s Promotional Materials for IoPhic- based alarms List of Figures Figure 1 - Room layout... 9 Figure 2 - Steam test configuration Figure 3 - Edible toast Figure 4 - Inedible toast Figure 5 - Onions sliced Figure 6 - Onions quartered Figure 7 - Onions at 0, 5, 10, 15, 20, 25, and 30 minutes Figure 8 - Hamburgers at 0, 5, 10, 15, and 20 minutes Figure 9 - Portland cement dust spray Figure 10 - Comparison between UL smoldering test profile and nuisance sources List of Tables Table 1 - Total probability of activation Table 2 - Transition times to 'aggressive nuisance' phase Table 3 - Probability of activation during 'pure nuisance' phase Table 4 - Probability of activation during aggressive nuisance' phase

4 Executive Summary This report describes the test methods, results, and analyses used for this project. The intent of this project is to investigate the nuisance alarm immunity of a new smoke alarm technology (IoPhic). It was completed by the Department of Fire Protection Engineering at the University of Maryland (UM) with the sponsorship of the Maryland Industrial Partnerships (MIPS) and USI Electric. The National Fire Protection Association (NFPA) estimates that in 2010 there were 482,000 structure fires reported to public fire departments in the United States. Those fires resulted in 2,755 civilian deaths, 15, 420 injuries, and $9.7 billion in property damages. Residential structural fires comprise a significant portion of the number of casualties experienced in the U. S. In 2010, residential fires were responsible for 85% (2,640) of those civilian deaths (1). One of the solutions promoted over the latter part of the 20 th century to the substantial life loss in residential structure fires has been the installation of home smoke alarms. In the fire incidents that accounted for 50% of the home fire deaths in the U.S., the smoke alarms did not function due to a missing or disconnected battery (2). Occupants often disconnect their smoke alarms when the smoke alarm reacts too frequently to nuisance sources. In the last two decades, studies have continuously analyzed the impact of nuisance alarms on the two major detection technologies (photoelectric and ionization). The common theme of this research is that ionization- based smoke alarms are more likely to react to nuisance sources than photoelectric smoke alarms (3). Thus, reducing the number of nuisance alarms in ionization- based alarms is projected to keep more smoke alarms in service which should lead to a further reduction in the number of fatalities in residential homes. This project tested 10 smoke alarms representing all the technologies available to the American consumer. Residential bathrooms and kitchens are the most common locations for nuisance alarms (3). Therefore, the nuisance sources used in this project were chosen for their likelihood to occur at these locations. Based on the results of this test series, IoPhic- based smoke alarms are observed to have nominally equivalent nuisance alarm immunity to photoelectric smoke alarms in kitchen scenarios, and are more resistant to nuisance sources near bathrooms than photoelectric smoke alarms. They are more resistant to nuisance alarms than all other smoke alarms utilizing an ionization sensor. Literature Survey Fire Hazard and Nuisance Alarms NFPA estimates that in 2010 there were 482,000 structure fires reported to public fire departments in the United States. Those fires resulted in 2,755 civilian deaths, 15,420 injuries, and $9.7 billion in property damages. Residential structural fires comprise a significant portion of the number of casualties experienced in the U.S. In 2010, residential fires were responsible for 85% (2,640) of those civilian deaths (1). One of the solutions promoted over the latter part of the 20 th century to the substantial life loss in residential structure fires has been the installation of home smoke alarms. 3

5 From records show that the probability of dying in a residential fire is roughly doubled when a home has smoke alarms that do not function (2). Further, the likelihood of casualties (fatal and non- fatal) is appreciably reduced in homes with working smoke alarms than those without (4). 24% of the home fire deaths during this period fall into the fatal and non- fatal category, and during 50% of those incidents, the smoke alarms did not function due to a missing or disconnected battery (2). An estimated 95% of all alarms reported by smoke alarms during the 1980 s in the U.S. were unnecessary (5). Reducing nuisance alarms will in turn reduce the number of households with nonfunctioning smoke alarms, and ultimately reduce the number of fatalities in residential homes. CPSC s National Smoke Detector Project reported 32% of removed batteries were due to nuisance alarms and 19% of all homes have nonfunctioning smoke alarms (6). Nonfunctioning smoke alarms are caused more often by the occupant disconnecting the alarm than a product defect or dead (or disconnected) battery. Photoelectric Sensors Photoelectric sensors use a light source and a photocell to detect airborne particles. When a particle enters the sensor chamber, it obstructs the light source and scatters light onto the photocell. The incident light on the photocell causes it to generate a current. The amount of scattered light and hence the magnitude of the current generated by the photocell is linearly proportion to the number of smoke particles and the square of the diameter of the particles as depicted in Equation 1 (7). LS n i d i 2 Equation 1 As such, large diameter soot particles are more influential in scattering light than the small soot particles generated during flaming fires. With the prevalence of large particles in smoldering fires, the photoelectric sensors are inherently sensitive to smoldering fires (7). This is a disadvantage for nuisance sources that are inclined to scatter light due to either particle size or optical properties (such as those found in water vapor). Ionization Sensors Ionization sensors use an ionization chamber and a source of ionizing radiation to detect airborne particles. The chamber has two oppositely charged walls, and the source of radiation in contemporary smoke alarms is Americium Under clear conditions, the charged radioactive particles are attracted to the charged walls, and generate a current. When airborne particles enter the sensor, they affect the movement of the ions to reduce the current generated. This drop in current is measured and used to gauge the presence of particles are in the chamber. The ionization potential is proportional to the number of smoke particles and their diameter as indicated in Equation 2 (7). ΔMIC α Σd i n i Equation 2 4

6 Small particles are produced in substantial quantities in flaming fires, thereby causing the increased sensitivity of ionization alarms to flaming fires. Ionization versus Photoelectric Sensors and Nuisance Alarms Reducing nuisance alarms requires an analysis of how the two different smoke alarm sensors respond to nuisance sources. Due to the fundamental differences in how photoelectric and ionization sensors operate, they have different responses to nuisance sources. During the last two decades, studies have continuously analyzed the impact of nuisance alarms on the two major detection technologies (photoelectric and ionization). Three of those are mentioned here. A U.S. Consumer Product Safety Commission (CPSC) report from 1994 indicated that while 87% of smoke alarms in U.S. homes have ionization sensors, 97% of nuisance alarms are reported by those units (6). A second study in 2000 installed smoke alarms in Alaskan Eskimo homes between 10 and 15 ft from the cooking appliances (8). This study found that 92% of homes with ionization alarms recorded nuisance alarms while only 11% with photoelectric alarms recorded nuisance alarms. Furthermore after 6 months, 19% of the ionization alarms were disconnected as opposed to only 4% of the photoelectric alarms. Finally, a 2007 study in Washington State reported that 9 months after installation, ionization alarms had sounded in 78% of the installed homes while only 39% of the photoelectric alarms had sounded 9. While having fires in 78% of the homes in a community in a 9- month period would be surprisingly, leading to the expectation that most of these alarms were due to nuisance sources, the proportion of nuisance alarms in these incidents is unknown. Also 9 to 15 months after initial installation of new smoke alarms, 20% of ionization alarms were disconnected and only 5% of photoelectric alarms. The most common reason for these disconnections was reported to be nuisance alarms (9). The problem has remained consistent over this time period. Ionization sensors have been more prone to nuisance alarms and nuisance alarms are a significant contributor to the problem of nonfunctioning alarms in residential homes which results in loss of life and property damage every year. New Types of Smoke Detection Technologies The effort to reduce nuisance alarms and simultaneously improve detection of smoldering fires has driven the development of new detection technologies such as the combination photoelectric/ionization alarm, the carbon monoxide- ionization alarm, and the algorithm- assisted ionization alarm (IoPhic). The primary advantage of combination photoelectric/ionization alarms is their ability to respond quickly to both flaming fires (with the ionization sensor) and smoldering fires (with the photoelectric sensor). Unfortunately, this also makes the alarm vulnerable to nuisance alarms from both sensors if the smoke alarm uses an or type of logic in order to provide an alarm. Carbon monoxide- ionization alarms can use the input from the carbon monoxide sensor to provide a second data point in determining whether or not a fire event is occurring to assist the ionization sensor with detecting smoldering fires. Finally, IoPhic alarms use algorithms to better identify smoldering fires, and nuisance sources based on their respective signatures. 5

7 Current Regulations Regarding Photoelectric and Ionization Smoke Alarms The national code for the installation of residential smoke alarms near nuisance source areas (e.g. kitchens and bathrooms) is NFPA 72 National Fire Alarm and Signaling Code (3). According to section : (4)*Smoke alarms and smoke detectors shall not be installed within an area of exclusion determined by a 10 ft (3.0 m) radial distance along a horizontal flow path from a stationary or fixed cooking appliance, unless listed for installation in close proximity to cooking appliances. Smoke alarms and smoke detectors installed between 10 ft (3.0 m) and 20 ft (6.1 m) along a horizontal flow path from a stationary or fixed cooking appliance shall be equipped with an alarm- silencing means or use photoelectric detection. (5)*Smoke alarms and smoke detectors shall not be installed within a 36 in. (910 mm) horizontal path from a door to a bathroom containing a shower or tub.(3) Based on the historical performance of home smoke alarms and the previously mentioned studies, a preference for photoelectric alarms near kitchen nuisance sources is reflected in NFPA 72. It acknowledges that bathrooms are a source of nuisance alarms; however it does not recommend that a specific type of alarm be used in this location. The appendix for section (5) mentions that it may be possible that photoelectric alarms are more susceptible to nuisance alarms from steam sources, but does not make any recommendations due to a lack of conclusive research on the subject at the time of publication. Some regional fire codes are more restrictive than NFPA 72 on both of these points. For example, under Massachusetts General Law chapter 148 section 26F, one- or two- family residences subject to the law may install only photoelectric alarms within 20 ft of kitchens and bathrooms containing a bathtub of shower. In addition to this requirement, all alarms in these residences must be either dual alarms using both a photoelectric sensor and an ionization sensor, or two alarms must be installed (one photoelectric and one ionization) (10). Based on the installation instructions included in ten different smoke alarm units, smoke alarm manufacturers have inconsistent recommendations for home smoke alarm installation. Recommendations range from 5 ft separation from cooking appliances to 20 ft separation from cooking appliances. At least one manufacturer states only that alarms should not be installed near cooking areas. Some manufacturers recommend that the alarms be a minimum of 3 ft from a doorway to a kitchen, or bathroom. Others require a 10 ft separation from bathroom doorways. One strategy to reduce nuisance alarms is to use photoelectric alarms near areas where nuisance sources are commonly found. This limits the flexibility of ionization- only alarm use in homes. Ionization alarms are generally less expensive than photoelectric alarms making them attractive to homeowners. As noted previously, ionization smoke alarms are also more effective than photoelectric smoke alarms at identifying flaming fires, providing additional escape time for occupants involved with this type of fire which has faster hazard development than smoldering fires. 6

8 Experimental methodology This experimental program was designed to compare the sensitivity of different smoke detection technologies on the market to household nuisance sources. The technologies used in this series were: Combination photoelectric/ionization sensors Carbon monoxide- ionization sensors Photoelectric sensors Ionization sensors IoPhic sensors A total of ten different smoke alarm models were used to cover the range of alarm technologies, and in order to avoid individual unit defects, alarm units were randomly selected from lots of 24 units from several major smoke alarm manufacturers. Two models were used for each technology with the exception of the carbon monoxide- ionization sensors (one was used) and the photoelectric sensors (three were used). The nuisance sources were chosen and designed to be aggressive in order to provide a challenging test of nuisance resistance. In addition to the selection of aggressive sources, the alarms were placed in close proximity to the source, being closer than the 10 ft separation prescribed by NFPA 72 (for kitchens). Close proximity of the nuisance sources was chosen in order to test this recommendation and investigate the possible advantages of the new technology in an aggressive nuisance setting. It was also necessary to address the orientation of the units during testing. Due to the asymmetrical physical characteristics of the unit housings, flow into the unit is not uniform from all directions for all alarm models. Before full scale nuisance testing could begin, determining the least- favorable orientation was necessary in order to identify the physical limitations of different models and account for them. Orientation Sensitivity Testing Experimental Setup The orientation sensitivity testing was conducted using a recirculating smoke generation box modeled after the smoke density sensitivity chamber (smoke box) used in UL268 (11). Smoke was generated using a 6 Volt power supply and 3 ft of PVC insulated 22 AWG wire similar to the Hot Wire Test from NFPA 76 Annex B.2 (12). Electric fans were used to induce a recirculating flow of smoke past the smoke alarms at approximately 98.4 ft/min (0.5 m/s). A red helium- neon laser was used to measure smoke obscuration and ensure that smoke generation was rapid enough to be recognized immediately as a trouble alarm. Procedure Each of the ten different smoke alarm units was tested four times with 90- degree rotations between trials. If one orientation was less responsive than the others by more than 1 7

9 sec., then it was retested two more times to establish its average time delay compared to the other three orientations. Orientation Sensitivity Results Two of the smoke alarm models demonstrated significant, i.e. 4 sec. or greater increases in response times for a given orientation, and an additional two models demonstrated minor, i.e. 1 to - 4 sec., increases in response times for a given orientation. Physical inspection of the unit housings in conjunction with the test results indicated that if a model had an unfavorable orientation, it also had a battery compartment that blocked airflow into the unit from that orientation. Therefore the side of the alarm with a battery compartment was deemed to be the least favorable side for full scale testing. During nuisance testing this side was directed away from the source in order to position all alarms in their most sensitive orientation. Nuisance Source Testing Sources Six nuisance sources were used to assess the sensitivity of different alarm technologies to a range of common problem scenarios. Because a prevalence of nuisance sources are present in kitchens and bathrooms and fire codes have requirements for alarm placement near kitchens and bathrooms, the majority of the nuisance sources were selected with kitchen and bathroom locations in mind. Making toast and cooking onions, hamburgers, and vegetable oil all produce aerosols. These sources are common in a kitchen environment where the majority of nuisance alarms occur. It is important to note that the measured obscuration at the time of smoke alarm activation is dependent on the fuel source given the nature and quantity of the particles produced from these sources. Therefore an alarm may activate at different obscuration thresholds with different sources of airborne particles (13). Steam is a common bathroom nuisance. Showers generate large quantities of hot water vapor which can rise to the ceiling and condense on smoke alarm housings and inside smoke alarm chambers causing nuisance alarms. Portland cement dust was selected as an aggressive potential household nuisance during intermittent incidents such as home remodeling, construction, or cleaning activities. Aggressive Nuisance Scenarios Due to the nature of nuisance sources (particularly in the kitchen) it is important to look at the potential for progression from a nuisance alarm to a serious trouble alarm. Once food has received enough heat input it can become a fuel source and an ignition point for a serious fire. The toast, onion, hamburger, and vegetable oil tests included periods of heating that go beyond a typical nuisance scenario and begin to enter an aggressive nuisance scenario. This is emphasized by the close proximity of the sources to the smoke alarms. The source is not yet an incipient smoldering fire, but the food is no longer edible and the individual who was cooking it has likely had a lapse in attention to their duties. Therefore alarm activations to this type of 8

10 source would not be entirely unwarranted. This should be kept in mind when interpreting results from this period of each test. Experimental Program for Nuisance Sources Cooking Related Tests All cooking- related tests were performed in a room with nominal dimensions of 18 ft x 24 ft with a ceiling height of 6 ft formed by acoustic drop ceiling. Five smoke alarm units were located at a radial distance 7 ft from the corner of the room, and 6 ft from the nuisance source origin (hotplate or toaster), as depicted in Figure 1. Units were spaced 1 ft apart center- to- center. Figure 1 - Room layout Bathroom Related Tests All bathroom- related tests were performed at a 6 ft ceiling height above a hotplate in the same room as the other nuisance source tests. Units were arranged in a 3 ft diameter circle with the origin centered over the hotplate and steam source. They were spaced nominally ft apart center- to- center as seen in Figure 2. Alarms were placed in their most sensitive orientation as determined during Orientation Sensitivity Testing. 9

11 Figure 2 - Steam test configuration Instrumentation The instrumentation for all tests consisted of a National Instruments Compact DAQ (cdaq- 9174) connected to an optical density meter (ODM), and a video camera. The ODM was designed via recommendations from UL 268 (11) using a low voltage automotive spot- light lamp rated at 6 volts DC and a Weston Photronic photocell. The centerline of the ODM was located ft below the ceiling. The path of the ODM s light beam was nominally directly underneath the arc of smoke alarms. The video camera was placed on the floor below the alarms to record alarm activations. The optical density measurement was recorded by the data acquisition system at 1 Hz. Procedure Toast Two slices of white sandwich bread were inserted into a Hamilton Beach two slice vertical toaster set to the darkest setting. The toaster was turned on, and allowed to run its normal toasting cycle, which lasted nearly 4 minutes. The first cycle resulted in dark, but edible toast as seen in Figure 3. After this first cycle, the toaster was immediately turned back on to run a second cycle. The second cycle resulted in a substantial amount of smoke issuing from the toaster and inedible charred bread as seen in Figure 4. 10

12 Figure 3 - Edible toast Figure 4 - Inedible toast Steam 1.0 L of water was brought to a vigorous boil before the test starts. At the start of the test (time = 0 min) the lid was removed for 0.33 min then replaced. This was repeated after 2 minutes of total test time, and again at 4, 6, and 8 minutes of total test time. Onion A nominally 1 pound onion was sliced into quarters as seen in Figure 5, and then the slices were again quartered as seen in Figure 6 to produce pieces no larger than 1 x 2.5. A light coating of approximately L of vegetable oil was used to sauté the onions over high heat in a 1 ft diameter frying pan. The onions were gently flipped at 5 minute intervals for 30 minutes to simulate a realistic cooking scenario. The progression of the onion source is illustrated in Figure 7. Figure 5 - Onions sliced Figure 6 - Onions quartered 11

13 Figure 7 - Onions at 0, 5, 10, 15, 20, 25, and 30 minutes Hamburger Three lb. (nominal) frozen hamburger patties were thawed and cooked in a 1 ft diameter frying pan over high heat. They were gently flipped every 5 minutes or the duration of the 20 minute test. The hamburger progression is depicted in Figure 8. Vegetable Oil 0.3 L of vegetable oil was placed in a 1 ft diameter frying pan over high heat and allowed to heat and vaporize for 30 minutes. Cement Dust A powder coating gun set to psi provided a fine spray of Portland cement dust as illustrated in Figure 9. Setting the hot plate to high heat and introducing the spray to the hot gas current from the hot plate created a visually uniform front of dust particle flow to the ceiling and the alarm locations. 12

14 Figure 8 - Hamburgers at 0, 5, 10, 15, and 20 minutes Figure 9 - Portland cement dust spray 13

15 Results Data for each test can be found in Appendix A: Test Data and Appendix B: Obscuration graphs. The number of alarm activations for each technology was normalized by the number of models using each technology and by the number of tests. This yielded an average probability of activation for a single alarm during a single test for each technology. Using the results from the onion test with the photoelectric smoke alarms as an example: Number of activations: 1 Number of tests performed: 15 Probability of activation: 100 * 1/15 = 7% Table 1 lists the probability of activation for each technology to a given nuisance source and to the total test series. Table 1 - Total probability of activation Technology Toast Steam Onion Hamburger Oil Dust Total Ion + Photo Ion + CO Ion Photo IoPhic On inspection of the % obscuration per foot per time graphs it became apparent that some of the nuisance sources provided an appreciable amount of smoke. In some cases, the rate of increase in obscuration was similar to that in the UL217 smoldering test profile as seen in Figure 10 (during the toast test, the obscuration actually exceeded the test profile determined in UL 217) (14). 1 The majority of these activations occurred during the aggressive nuisance phase of the test (after min). 14

16 Figure 10 - Comparison between UL smoldering test profile and nuisance sources This observation suggested that the tests should be split into a pure nuisance phase, and an aggressive nuisance phase. The threshold obscuration value chosen for separating the two phases of each test was based subjectively on the average obscuration exceeding 0.15% obscuration per foot, in part accounting for the systematic errors in the ODM. Over a period of 30 minutes the ODM would record a systematic increase from the baseline (pre- test) measurement due to the heating cycle of the lamp and light sensor. The mean value for that increase with a 95% confidence interval is 0.05%±0.05% obscuration per foot for a 30 minute measurement. In addition, in those tests where this threshold point was achieved, this was also supported through the observation that the food source was deemed no longer edible afterward. The transition time for each source using the 0.15% obscuration per foot criterion is presented in Table 2. Table 2 - Transition times to 'aggressive nuisance' phase 0.15 %/ft - Time (min) Onion Oil Hamburger Toast It is important to note that the times noted in Table 2 are appreciably less than 30 minutes. Therefore the systematic error of the ODM is less than 0.05% obscuration per foot at these times. The increase in obscuration is therefore due to a visible increase in aerosol particles in vicinity of the smoke alarms. Using this threshold value of 0.15% obscuration per foot, the results for pure nuisance alarm activations are as follows in Table

17 Table 3 - Probability of activation during 'pure nuisance' phase Technology Toast Steam Onion Hamburger Oil Dust Total Ion + Photo Ion + CO Ion Photo IoPhic The results for activations in the aggressive nuisance phase are indicated in Table 4 (note that the steam and dust tests do not fit the criteria for this category). Table 4 - Probability of activation during aggressive nuisance' phase Technology Toast Steam Onion Hamburger Oil Dust Total Ion + Photo 0 n/a n/a 20 Ion + CO 0 n/a n/a 35 Ion 20 n/a n/a 35 Photo 100 n/a n/a 27 IoPhic 20 n/a n/a 23 Individual Source Results Toast During the toast tests, all of the technologies saw at least one alarm. However, a distinction is evident between the units with units with ionization sensors, and the IoPhic and photoelectric units. One IoPhic unit did not alarm during the toast test at all, and the photoelectric units only alarmed during the second aggressive nuisance phase of the testing. Steam During the steam test the ionization units performed much better than the photoelectric units. Nearly half of the photoelectric units alarmed to steam exposure. This has implications for the recommendation of using photoelectric alarms to avoid nuisance alarms in bathroom locations. Onion During the onion test, half of the IoPhic units and one photoelectric unit alarmed, predominantly during the aggressive nuisance phase of the tests (after minutes). This may be due to the close similarity between the UL 217 smoldering test profile and the measured obscuration profile of the onion nuisance test. Hamburger During the hamburger test, the IoPhic and photoelectric units were the most resistant to nuisance alarms, while the ionization + CO unit alarmed during every test. The typical ionization 16

18 and the combination ionization/photoelectric units also alarmed during the majority of testing. The majority of these activations occurred during the aggressive nuisance phase. Oil During the oil test, the IoPhic and photoelectric units were the most resistant to nuisance alarms, while the ionization + CO unit alarmed to during every test. The typical ionization and the combination ionization/photoelectric units also alarmed during the majority of testing. These results appear similar to the results from the hamburger test, however during the oil test the activations were more evenly split between the pure nuisance and aggressive nuisance phases. This is likely due to the lower rate of increase in obscuration compared to the hamburger test. Dust During the Portland cement dust tests, the combination ionization/photoelectric and normal photoelectric units alarmed to the majority of tests. With the exception of one activation by a typical ionization unit, the ionization- based units were not sensitive to the airborne dust particles. Conclusion The IoPhic based units were significantly less likely to respond to nuisances than combination ionization/photoelectric, ionization + CO, and typical ionization units respectively. The IoPhic units also responded to only 85% of the sources that caused alarms with typical photoelectric units. Of particular interest in this comparison is the advantage IoPhic units demonstrated over typical photoelectric units with regard to steam sources. Based on the results of this test series, IoPhic based smoke alarms have a nominally equivalent performance with respect to nuisance alarms as typical photoelectric smoke alarms in the kitchen, and are more resistant to nuisance sources near bathrooms than photoelectric smoke alarms. They are also more resistant to kitchen pure nuisance alarms than all other units utilizing ionization sensors. Acknowledgements Appreciation is extended to the University of Maryland (UM), Maryland Industrial Partnerships (MIPS), USI Electric, and the Fire Testing and Evaluation Center (FireTec) for their financial and technical support of this project. Special thanks go to Mr. Eric Gonzales of USI Electric for their oversight and guidance. Special thanks also goes to Dr. André Marshall, Olga Zeller, and the FireTec team from the Department of Fire Protection Engineering for their technical expertise and assistance during project execution. 17

19 Works Cited 1. Karter, M J. Fire Loss in the United States During s.l. : NFPA Fire Analysis and Research Division, Ahrens, M. Smoke Alarms in U.S. Home Fires. s.l. : NFPA Fire Analysis and Research Division, NFPA 72 National Fire Alarm and Signaling Code. s.l. : National Fire Protection Association, Milke, J A, et al. Performance of Smoke Detectors and Sprinklers in Residential and Health- Care Occupancies. s.l. : Department of Fire Protection Engineering University of Maryland, May The Latest Statistics on U.S. Home Smoke Detectors. Hall, J R. s.l. : Fire Journal, 1989, Vols. 83:1, CPSC. Smoke Alarms Pilot Study of Nuisance Alarms Associated with Cooking. s.l. : U.S. Consumer Product Safety Commission, Fabian, T Z, et al. Smoke Characterization Project: Final Report. Northbrook, IL : Underwriters Laboratories Inc, April 24, Ionization and Photoelectric Smoke Alarms in Rural Alaskan Homes. Fazzini, Thomas M, Perkins, Ron and Grossman, David. 173, 2000, West J. Med, pp Randomized Controlled Trial of Ionization and Photoelectric Smoke Alarm Functionality. Mueller, B A, et al. 14, 2008, Injury Prevention, pp Services, Department of Fire. A Guide to the Massachusetts Smoke & Carbon Monoxide Requirements when Selling a One- Or Two- Family Residence UL268. Smoke Detectors for Fire Alarm Systems. s.l. : Underwriters Laboratories Inc. 12. NFPA 76 Standard for the Fire Protection of Telecommunications Facilities. s.l. : National Fire Protection Association, Milke, J A, Mowrer, F W and Gandhi, P. Validation of a Smoke Detection Performance Prediction Methodology, Volume 3. Evaluation of Smoke Detector Performance. Quincy, MA : Fire Protection Research Foundation, UL217. Single and Multiple Station Smoke Alarms. s.l. : Underwriters Laboratories Inc., Cleary, T. Residential Nuisance Source Characteristics for Smoke Alarm Testing. Gaithersburg, MD : National Institute of Standards and Technology, Building and Fire Research Laboratory. 16. Bukowski, R and al, et. Performance of Home Smoke Alarms Analysis of the Response of Several Available Technologies in Residential Fire Settings. s.l. : National Institute of Standards and Technology, Fire Research Division, NIST Technical Note

20 Appendices Appendix A: Test Data Toast Tests Test Unit :25 3:28 3:55 3:55 7: :17 2:50 3:18 3:40 6: :52 3:13 2:47 4:45 6: :40 2:52 3:50 4:40 7: :48 3:03 3:25 5:22 7: :10 4:54 6:03 6: :05 4:47 6:05 6:05 5: :47 5:50 6:38 5:50 5: :45 5:22 6:42 6:42 5: :05 5:18 6:35 6:31 5:30 Onion Tests Test Unit : : : : : : : :28-19

21 Hamburger Tests Test Unit :19 18:46 19: :39 18:49 20: :35 18: :22 18:14 18:25 18: :23 19:22 20: : : :57 9: : Oil Tests Test Unit :28 29:57 28: :55 17:00 25: :29 13:00 19: :41 9:00 15:41 20: :19 10:12 15:48 17: : : : :

22 Dust Tests Test Unit :40 0: : :36 0: : :44 0: : :48 : : : : :42-09:35 4: : : : :30 - Steam Tests Test: s1 unit activation - x x x - - Test: s2 unit activation x x x Test: s3 unit activation Appendix B: Obscuration graphs In the following graphs, the vertical line indicates the time of transition to the aggressive nuisance phase of the test. 21

23 Toast Tests test1 Obscura^on (%/b) test7 test13 test19 test25 test31 test43 test55 avg Time (min) transiton Onion Tests Obscura^on (%/b) Time (min) test3 test9 test15 test21 test27 test33 test45 test51 avg transiton 22

24 Hamburger Tests 0.6 Obscura^on (%/b) test4 test10 test16 test34 test46 test52 test58 avg transiton Time (min) Oil Tests 0.6 Obscura^on (%/b) test5 test11 test17 test35 test41 test47 test59 avg transiton Time (min) 23

25 Dust Tests Obscura^on (%/b) test6 test12 test36 test42 test48 test54 test Time (min) Steam Tests Obscura^on (%/b) test_s1 test_s2 test_s Time (min) 24

26 Appendix C Manufacturer s Promotional Materials for IoPhic- based alarms Begins on the next page. 25

27 26

28 27

29 28

30 29