FOREIGN RRL

1
Smoke Alarm Research
The Fire Research Division is conducting research to provide the technical basis
for modifications and/or additions to standard test methods used to certify
residential smoke alarms by developing performance metrics related to the
hazards posed by fires, and to the susceptibility to nuisance alarms.

Smoke alarms are now installed in 96% of all U.S. homes. While 4% of homes do
not have smoke alarms, approximately 20% of homes with smoke alarms have
non-operational smoke alarms. It is estimated that if every home had working
smoke alarms, U.S. residential fire deaths could drop by 36% (~1100 lives saved
per year.) U.S. fire statistics for home structure fires from Years 2000-2004 reveal
that 34% of civilian deaths occurred in homes with operating smoke alarms, 22%
of civilian deaths occurred in homes with smoke alarms present, but that failed to
operate, and 43% of civilian deaths occurred in homes with no smoke
alarms. The statistics point to three major problems or obstacles to further
reducing residential fire deaths through smoke alarm usage, leading to the
following questions: First, how can 100 % of homes have smoke alarms?
Second, how can the percentage of working smoke alarms be significantly
increased? Third, how can smoke alarms, when they are functioning, be more
effective in warning occupants? Research points to two main reasons for non-
working smoke alarms: missing or dead batteries and intentional power source
interruption. A reduction in nuisance alarms would tend to reduce intentional
disabling of alarms. There is a need for research to improve the performance
measurement of smoke alarms including performance for a range of fire
scenarios, and the susceptibility to nuisance alarms.

This page provides a summary of recent NIST research related to smoke alarms.

Cleary T.G., and Chernovsky, A. "Smoke Alarm Performance in Kitchen

Fires and Nuisance Alarm Scenarios," NIST Technical Note 1784 (January,
2013)
Experiments were conducted to assess the performance of various residential
smoke alarms to kitchen fires and nuisance alarm cooking scenarios. A structure
representing a kitchen, living room and hallway was constructed to conduct the
experiments. Eight different residential smoke alarms types, two photoelectric
models (P1 and P2), two ionization models (I1 and I2), two dual sensor
photoelectric/ionization models(D1 and D2), and two multi-sensor, intelligent
models (M1 and M2) were used in this study. The data gathered provided insight
into the susceptibility of alarm activation from exposures to typical cooking events
and alarm times for actual kitchen fires. The effects of alarm technology and
installation location on the propensity of an alarm to activate were examined. In
the kitchen fire experiments, all smoke alarms responded before hazardous
conditions developed. An ionization alarm (I1) tended to respond first compared
to other co-located alarms. Results show smoke alarms placed greater than 6 m
from the kitchen range may provide less than 120 s of available safe egress time,
which suggests the importance of a more central alarm location closer to the
kitchen for this configuration. Experiments were conducted to determine an
alarm's propensity to activate when exposed to particulates generated from eight
FOREIGN RRL

typical cooking activities including toasting, frying, baking and broiling. In most
cases, the propensity to nuisance alarm decreased as the distance from the
cooking source increased. Two alarms, I1 and D2, experienced more nuisance
alarm activations across the eight cooking activities than the other alarms. The
remaining alarms experienced about the same combined nuisance alarm
frequency by averaging all cooking events for installation locations outside the
kitchen. Experiments showed combustible materials typically found on a counter
top can spread flames to overhead cabinets, and a single empty 0.6 m wide 1.0
m tall wood-framed, pressboard cabinet can produce a peak heat release rate
nearly sufficient to flashover a small room. Alternatively, protective metal barrier
on the bottom and side facing the range tended to limit the spread of flames to
the cabinet and reduce the heat release rate.

Cleary, T.G., "An Analysis of the Performance of Smoke Alarms," presented

at 10th International Symposium on Fire Safety Science, University of
Maryland, USA, June 19-24 (2011).
Test results from the NIST 2008 Smoke Alarm Sensitivity Study were used in a
smoke alarm performance analysis to examine the effects of pre-movement time,
reduced travel speeds through smoke, and smoke optical density limit on
occupant survivability given different smoke alarm installations. Smoke alarm
installations that meet the requirements in the current National Fire Alarm Code
NFPA 72 were considered. Alarm times from commercially-available
photoelectric, ionization, and dual photoelectric/ionization alarms were used in
the analysis to examine the effects of smoke alarm type on the predicted
survivability for a range of fire and egress scenarios. Fire scenarios included both
flaming and initially smoldering upholstered chair mock-ups. Egress scenarios
considered occupants located in, or remote from the room of fire origin. Reduced
travel speed through smoke was included in the analysis. Prior to occupant
movement and as an occupant travels to the exit, the fractional effective dose
from toxic gas and heat exposure were computed to determine survivability. The
concept of relative effectiveness as performance metric for smoke alarms is
introduced. The relative effectiveness is the fraction of occupants that
successfully escape a given fire and egress scenario. It is computed by
considering a frequency distribution for the pre-movement time and determining
the cumulative fraction of occupants that successfully escape. Thus, the relative
effectiveness of a smoke alarm type or installation requirement can be averaged
over a large number of fire and egress scenarios. The pre-movement frequency
distribution was modeled as a log-normal function. Experimental studies suggest
that the median value of the distribution relates to characteristics of the
population and a geometric standard deviation of 1.6 characterizes the width the
distributions. The distribution median was varied to examine relative
effectiveness skewed to more vulnerable populations (those slower to react).
Travel speed was modeled as a function of smoke optical density which predicts
reduced travel speed as thicker smoke is encountered. Model results showed
photoelectric alarms had the lowest relative effectiveness values for flaming fires,
while ionization alarms had the lowest relative effectiveness values for
smoldering fires. These trends were expected based the results of previous
studies. It was observed that there can be a steep increase in relative
FOREIGN RRL

effectiveness, depending on the smoke alarm type and fire scenario, as the
smoke optical density limit was increased from 0.25 m-1 to 0.50 m-1. However,
the ranking of smoke alarms tend to remain the same. Given the magnitude of
statistically significant mean values of relative effectiveness for all flaming and
smoldering fires considered, the model results suggest that there is a benefit from
a combination of alarm technologies, and that vulnerable populations who may
require significantly more time to escape, regardless of the fire scenario, would
benefit the most from dual alarms or side-by-side photoelectric and ionization
alarms.

Cleary, T.G., "Full-scale Residential Smoke Alarm Performance," 14th

International Conference on Automatic Fire Detection, University of
Duisburg-Essen, Duisburg, Germany (2009).
A series of 24 full-scale fire experiments was conducted in a multi-room structure
to examine the effects of alarm type (photoelectric, ionization, and dual sensor),
alarm location, fabric type (cotton and polyester), polyurethane foam density,
ignition scenario (smoldering or flaming), and room configuration on smoke alarm
performance. The fire source was a chair mock-up consisting of a seat and back
cushion of a specific fabric and foam density, resting on a metal frame. Each fire
progressed for a time sufficient to produce multiple hazards (smoke, heat, toxic
gases) throughout the compartment. Photoelectric, ionization, and dual
photoelectric/ionization alarms were co-located at multiple locations to facilitate
comparisons of each type of alarm. In the room of fire origin, a smoke optical
density of 0.25 m-1 was reached before a fractional effective dose of 0.3 for
either toxic gases or heat exposure. The available safe egress time (ASET) for
both flaming and smoldering fires was sensitive to the imposed optical density
limit. Further study is needed to deduce the impact of visibility-limiting smoke
levels on the time needed to egress residential fires to justify any particular
optical density limit value.

Cleary, T.G., "Results from a Full-Scale Smoke Alarm Sensitivity

Study," Presented at the Fire Protection Research Foundation's 13th annual
Suppression and Detection Research & Applications Symposium (SUPDET
2009), February 24-27, 2009, Orlando, FL, and published in Fire Technology
(2010)
A series of 24 full-scale experiments was conducted during the summer of 2008
to examine the effects of alarm type (photoelectric, ionization, and dual sensor),
alarm location, fabric type (100 % cotton and 100 % polyester), polyurethane
foam density, ignition scenario, and room configuration, on smoke alarm
performance. A two-level, fractional factorial design of eight experimental
configurations was developed around the five factors: fabric type, foam density,
fire location, ventilation, and ignition scenario. A structure, designed to represent
a single-story home or apartment, was constructed inside the Large Fire
Laboratory at the National Institute for Standards and Technology for the
experiments. The fire source was a chair mockup consisting of a seat and back
cushion of a specific cover fabric and foam density, weighing between 5.5 kg and
8.3 kg. It rested on a metal frame and was subjected to a small propane gas
FOREIGN RRL

flame, or an electric cartridge heater to initiate smoldering. Each experimental

configuration was conducted three times. Smoldering fires were allowed to
progress until they naturally transitioned to flaming fires except for one test that
was terminated early due to time constraints. The smoldering to flaming transition
times ranged from (81 to 182) min. Each fire progressed for a time sufficient to
produce multiple hazards (smoke, heat, and toxic gases). All alarms tested were
purchased from retail outlets and activated at their preset levels. Photoelectric,
ionization, and dual photoelectric/ionization alarms were co-located at multiple
locations to facilitate comparisons of each alarm type, and different designs of the
same type of alarm. For smoke alarms in the room of fire origin, it was observed
that each of the five factors had an effect on the measured alarm times that was
primarily a result of fire growth rate (fabric type, foam density, and ignition
scenario), or smoke dilution (fire location and ventilation). The photoelectric alarm
responded quicker on average than ionization alarm in two of four smoldering fire
configurations, responding before the ionization alarm in all 6 trials, while the
ionization alarm responded before the photoelectric alarm in two of three trials for
the other two configurations. The ionization alarm responded quicker on average
than photoelectric alarm in all four flaming fire configurations, and responded
before the photoelectric alarm in all 12 flaming fire trials. One dual alarm had the
fastest average alarm time for all four smoldering fire configurations, and
responded first in 11 of the 12 trials. It also yielded faster average alarm times
than the other dual alarm in seven of eight configurations, and was the first dual
alarm to respond in 22 out of 23 trials where dual alarms were present.

Cleary, T.G., "Performance of Dual Photoelectric/Ionization Smoke Alarms

in Full-Scale Fire Tests," Presented at the Fire Protection Research
Foundation's 13th annual Suppression and Detection Research &
Applications Symposium (SUPDET 2009), February 24-27, 2009, Orlando,
FL, and published in Fire Technology (2010)
Data from two full-scale residential smoke alarm fire test series were analyzed to
estimate the performance of dual sensor photoelectric/ionization alarms as
compared to co-located individual photoelectric and ionization alarms. Dual
alarms and aggregated photoelectric and ionization alarm responses were used
to estimate dual alarm performance. It was observed that dual alarms with
equivalent or higher sensitivity settings performed better than individual
photoelectric or ionization alarms over a range of flaming and smoldering fire
scenarios. In one test series, dual alarms activated 539 s faster than ionization
alarms and 79 s faster than photoelectric alarms on average. In another test
series, individual alarm sensor outputs were calibrated against a reference
smoke source in terms of light obscuration over a path length (percent smoke
obscuration per unit length) so that alarm thresholds could be defined by the
sensor outputs. In that test series, dual alarms, with individual sensor sensitivities
equal to their counterpart alarm sensitivities, activated 261 s faster on average
than ionization alarms (with sensitivity settings of 4.3%/m smoke obscuration for
the ionization sensors) and 35 s faster on average than the photoelectric alarms
(with sensitivity settings of 6.6%/m, for the photoelectric sensors.) In cases where
an ionization sensor was the first to reach the alarm threshold, the dual alarm
activated 67 s faster on average than the photoelectric alarm. While in cases
FOREIGN RRL

were a photoelectric sensor was the first to reach the alarm threshold, the dual
alarm activated 523 s faster on average than the ionization alarm. Over a range
of ionization sensor settings examined, dual alarm response was insensitive to
the ionization sensor setting for initially smoldering fires and fires with the
bedroom door closed, while dual alarm response to the kitchen fires was very
sensitive to the ionization sensor setting. Tests conducted in the National Institute
of Standards and Technology (NIST) fire emulator/detector evaluator showed that
the ionization sensors in off-the-shelf ionization alarms and dual alarms span a
range of sensitivity settings. While there appears to be no consensus on
sensitivity setting for ionization sensors, it may be desirable to tailor sensor
sensitivities in dual alarms for specific applications, such as near kitchens where
reducing nuisance alarms may be a goal, or in bedrooms where higher smoke
sensitivity may be a goal.

Bukowski, R.W., et. al., "Performance of Home Smoke Alarms, Analysis of

the Response of Several Available Technologies in Residential Fire
Settings," NIST Technical Note 1455-1 (February 2008 Revision)
This report presents the results of the project and provides details of the
response of a range of residential smoke alarm technologies in a controlled
laboratory test and in a series of real-scale tests conducted in two different
residential structures. The data developed in this study include measurement of
temperature and smoke obscuration in addition to gas concentrations for a range
of fire scenarios and residences. The results are intended to provide both insight
into siting and response characteristics of residential smoke alarms and a set of
reference data for future enhancements to alarm technology based on fires from
current materials and constructions. Smoke alarms of either the ionization type or
the photoelectric type consistently provide time for occupants to escape from
most residential fires, although in some cases the escape time provided can be
short. Consistent with prior findings, ionization type alarms provide somewhat
better response to flaming fires than photoelectric alarms, and photoelectric
alarms provide (often) considerably faster response to smoldering fires than
ionization type alarms. Escape times in this study were systematically shorter
than those found in a similar study conducted in the 1970's. This is related to
some combination of faster fire development times for today's products that
provide the main fuel sources for fires, such as upholstered furniture and
mattresses, different criteria for time to untenable conditions, and improved
understanding of the speed and range of threats to tenability.

2.
FOREIGN RRL

2/25/08 http://smokealarm.nist.gov/ 1 Questions and Answers Clarifying Findings of NIST Home

Smoke Alarm Study1,2 1. In how many scenarios contained in the July 2004 NIST Technical Note
1455 on "Performance of Home Smoke Alarms: Analysis of the Response of Several Available
Technologies in Residential Fire Settings" does the available time for egress exceed the time required
for egress? The difference between the available safe egress time (ASET) and the required safe
egress time (RSET) is dependent on three important factors: (1) the time to activate the alarm, (2)
the time to untenable (unsafe) conditions, and (3) the estimated egress time. The time to activate
the alarm is a function of detector sensitivity, detector placement, and characteristics of the fire. The
time to untenable conditions varies with the characteristics of the fire, the location within the
structure and the selection of tenability criteria (NIST defined "untenable" as the temperature, heat
flux, CO, or smoke conditions3 at the 5 ft level outside of the room of origin that would affect
particularly sensitive populations: the young, elderly, asthmatics, and sufferers of other lung
conditions). The required egress times (RSET) were estimated to range between 10 s and 140 s. This
large variability4 (See Table 26 of the report2 ) was meant to account for differences in the time of
day, the capability of the occupants, the design of the residence, the character and location of the
fire, and the location of the occupant relative to the location of the fire and smoke alarms. For any
given test, the times to activate the alarms and to reach one of the tenability criteria (most often,
smoke obscuration) were measured, from which the value of ASET was determined. Table 1 shows
the number of tests (out of a total of 32 tests) when the available time was less than the time
required for safe egress. For the 12 smoldering tests, this number increased from one instance to
five instances for the photoelectric alarm, and from two instances to nine instances for the
ionization alarm, over the range of values assumed for RSET. For the 16 flaming chair and mattress
fires, this number increased from one instance to eleven instances for the photoelectric alarm, and
from zero to five instances for the ionization alarm as the time needed for safe egress was increased.
Both the ionization and photoelectric detectors provided more than enough time for safe egress
from the four scenarios involving kitchen fires for all assumed values of RSET. 1 Reported in February
26, 2004 Tech Beat article "Current Smoke Alarms Pass Life-Saving Tests." 2 Bukowski, R. W.,
Peacock, R. D., Averill, J. D., Cleary, T. G., Bryner, N. P., Walton, W. D., Reneke, P. A., and Kuligowski,
E. D. Performance of Home Smoke Alarms, Analysis of the Response of Several Available
Technologies in Residential Fire Settings, Natl. Inst. Stand. Technol., Tech. Note 1455 (2004) 3 Life-
threatening components of fire -- Guidelines for the estimation of time available for escape using fire
data, ISO/TS 13571:2002." International Organization for Standardization, 2002. 4 The estimated
evacuation times listed in the NIST report for nighttime fires (90 s to 140 s) are more conservative
than the times found in actual studies of nighttime residential evacuation when occupants were
sleeping, which reported times ranging from 36 s to 119 s, including waking time, time to call the fire
department and for the family to escape outside the house. See Nober, E. H.; Peirce, H.; Well, A. D.;
Johnson, C. C.; Clifton, C., “Waking Effectiveness of Household Smoke and Fire Detection Devices,”
Fire Journal, Vol. 75, No. 4, 86-91,130, (1981). 2/25/08 http://smokealarm.nist.gov/ 2 Table 1.
Number of tests (out of 32) with ASET < RSET RSET = 10 s RSET = 65 s RSET = 140 s Fire Type Number
of Tests Photoelectric Alarms Ionization Alarms Photoelectric Alarms Ionization Alarms Photoelectric
Alarms Ionization Alarms Smoldering 12 1 2 4 6 5 9 Flaming 16 1 0 4 1 11 5 Cooking 4 0 0 0 0 0 0 With
alarms placed on every level plus in bedrooms or in every room, the number of tests where ASET
was less than RSET is smaller. For example, with alarms placed in every room, ASET was less than
RSET in two tests for photoelectric alarms and three tests for ionization alarms with an RSET value of
65 s. Most current building codes require smoke alarms at every level in existing homes and at every
FOREIGN RRL

level plus bedrooms in newly constructed homes. No codes require smoke alarms in every room but
this is included as (theoretically) the best performance achievable. 2. Do ionization detectors provide
enough time to save lives? How did NIST come up with the "three minute warning" criteria that was
cited in a February 26, 2004 Tech Beat article "Current Smoke Alarms Pass Life-Saving Tests"? There
are numerous combinations of fire scenarios and smoke alarm placement locations. In the NIST
experiments, the following variables were included: smoke alarm location; room of fire origin,
smoldering, flaming, or cooking fire; and bedroom doors open or closed. Table 2 (taken from
Appendix A of NIST TN 1455) lists the averages and standard deviations for the measured times to
first alarm and to untenable conditions in these tests. In all cases, the average ASET was positive,
although the average ASET for the photoelectric alarm in the case of the flaming fires was less than
the maximum assumed for the RSET. The average estimated time available for safe egress in the
smoldering scenario was greater than the RSET by much more than one standard deviation in the
case of the photoelectric alarms. For the ionization alarms in the smoldering situation, the average
ASET was about one minute greater than the longest assumed RSET, but in ten of the twelve tests
was less than the longest RSET assumed (See last column of Table 1). The general trends from the
NIST experiments are consistent with previous studies showing that: (1) both ionization and
photoelectric alarms provide enough time to save lives for most of the population under many fire
scenarios, (2) ionization alarms may not always provide enough time for the most sensitive
populations with mobility limitations to escape a smoldering fire, and (3) photoelectric alarms may
not always provide enough time to escape a flaming furniture fire for this same population. Table 2.
Average time to first alarm and time to untenable conditions, with standard deviations. Time to First
Alarm (s) Time to Untenable Conditions (s) Available Safe Egress Time (s) Photoelectric Alarms
Ionization Alarms Photoelectric Alarms Ionization Alarms Smoldering Fires 2219 ± 1061 4010 ± 1120
4316 ± 1256 2136 ± 1011 276 ± 331 Flaming Fires 94 ± 33 47 ± 36 217 ± 67 129 ± 74 177 ± 69
Cooking Fires 738 ± 103 681 ± 475 1477 ± 249 739 ± 148 796 ± 241 2/25/08
http://smokealarm.nist.gov/ 3 For the NIST test series, the average ASET (assuming placement of
detectors in all rooms) for ionization alarms ranged between 177 s for flaming fires and 276 s for
smoldering fires; for photoelectric alarms the ASET averaged between 129 s and 2136 s. For the
purpose of the Tech Beat article, 180 s (or 3 minutes) was selected as representative of the amount
of time available, and to convey to the reader the importance of leaving their residence quickly after
first hearing a smoke alarm of any design. 3. Are ion and photo detectors qualitatively similar since
the February 26, 2005 Tech Beat article states that "Ionization smoke alarms respond faster to
flaming fires, while photoelectric smoke alarms respond quicker to smoldering fires"? Why does
NIST describe the 30-45 minute of photos in smoldering fires as "quicker" and in the same sentence
describe the 20-30 seconds benefit of ion detectors as "faster"? No, the ionization alarms and photo
alarms are qualitatively different because they respond to different aspects of a fire. The Tech Beat
article was correct in stating that "ionization alarms respond faster to flaming fires, while
photoelectric smoke alarms respond quicker to smoldering fires." While the photo detectors sensed
the smoldering fires on average 30 minutes earlier than the ionization detectors, the threat from a
smoldering fire grows much more slowly than from a flaming fire. That is, the margin of safety
associated with a 30 minute earlier warning in a slow growing smoldering fire may not be necessarily
any more significant than a 30 s earlier warning for a fast growing flaming fire. 4. Should "necessary
escape time" be replaced with "available escape time" in the February 26, 2004 NIST news release
where it states that "The researchers determined the necessary escape times by considering the
time that the alarms sounded in various locations and the development of untenable (unsurvivable)
FOREIGN RRL

conditions"? Yes. Necessary escape time refers to the time required for occupants to safely evacuate
from a building. This is usually termed the required safe egress time (RSET). Available safe egress
time (ASET) is the elapsed time between when occupants are notified of a fire and when conditions
along the path of their egress become sufficiently untenable that escape may no longer be possible.
The correct term in the press release should have been the available safe egress time rather than
necessary escape time. The press release has been corrected effective March 8, 2007. 5. Why does
NIST appear to install "un-modified” detectors in areas which prevent comparison between the
unmodified and modified detectors? Where in the report does NIST present this comparison and
associated data in a manner that justifies its analysis of "modified" detector response? Is the manner
in which NIST models the response of "modified" detectors valid? The unmodified smoke alarms
used in the study consisted of a photoelectric model and two ionization types, all purchased from
retail establishments by NIST for use in the test series. Some of the identical models of smoke alarms
were modified by their manufacturers so that the 2/25/08 http://smokealarm.nist.gov/ 4 voltage
produced by the sensor could be monitored to determine the response of the sensor to the changing
environment in a continuous manner, rather than registering a single alarm point. This analog signal
provided a means to monitor the environment during the build up period prior to the detector
alarming, and had no effect on the sensitivity of the smoke alarms or the reported detector response
times. Thus, it was appropriate to compare detector response times at different locations for both
un-modified and modified detectors. A more appropriate term for "modified" would have been
"continuously monitored" detectors. It was not necessary to monitor the output voltage of all of the
smoke alarms since the monitored alarms were located judiciously. Chapter 2 of the July 2004 NIST
report discussed the alarm calibrations in detail. Section 2.8 discusses the response of the
"modified" alarms. All of the alarms were calibrated multiple times throughout the test series. Any
changes in the response of the alarms were incorporated into the analysis. NIST spot-tested
"unmodified" alarms in the fire-emulator/detector-evaluator (FE/DE)5 to compare alarm point
smoke levels to the equivalent alarm level for the "modified" alarms. The "unmodified" (off-the-
shelf) ionization alarms had sensitivities near the low level sensitivity setting for the "modified"
ionization alarms used in the analysis; otherwise, the alarm levels were comparable. Beyond simple
alarm point calibration, the minimum sensitivity of smoke alarms is established in US and
International standards by performance in fire tests. Specifically, smoldering smoke tests utilize
cotton wicks or wood pieces on a hotplate (UL217/2686 and EN54/ISO TS7240-9 7 ) as sources, and
these were used by NIST as detailed in the report. The Underwriters Laboratories (UL) smoldering
test with wood on a hotplate was developed by UL in the late 1970‟s to mimic the smoldering
mattresses and furniture (both the smoke characteristics and temporal increase) in the original
Indiana Dunes Tests.8 Since UL uses this smoke to quantify smoke alarm points, NIST chose the
nominal sensitivities stamped on the back of typical alarms to be the alarm points for this study.
While NIST used the unique FE/DE apparatus as described in the preceding paragraph to calibrate
the test alarms, the output has been correlated with the UL apparatus. Both US and International
standards also include flaming sources in their fire tests. These include cellulosic (wood and paper –
flaming wood is used by both UL and EN/ISO, paper only by UL), liquid hydrocarbon (heptane), and
plastic (polyurethane foam – EN/ISO only). UL stopped conducting its flaming plastic (polystyrene
foam) test in 1999. 5 Cleary, T. G., “Fire Emulator/Detector Evaluator: Design, Operation, and
Performance.” Proceeding of the International Conference on Automatic Fire Detection "AUBE '01",
March 25-28, 2001, Gaithersburg, MD, Beall, K.; Grosshandler, W. L.; Luck, H., Editors, Natl. Inst.
Stand. Technol., NIST SP 965; February 2001. 312-323 pp, (2001) 6 UL 217: Standard for Safety Single
FOREIGN RRL

and Multiple Station Smoke Alarms, and UL 268: Standard for Smoke Detectors for Fire Protective
Signaling Systems, 4th ed., Underwriters Laboratories Inc., Northbrook, IL., 1996 7 EN 54:
Components of Automatic Fire Detection Systems, Part 9, Fire Sensitivity Test, European Committee
for Standardization, Brussel, 1982. 8 Bukowski, R.W., Waterman, T.E., and Christian, W.J., “Detector
Sensitivity and Siting Requirements for Dwellings: A Report of the NBS „Indiana Dunes Tests‟” NFPA
No. SPP-43 Nat. Fire Prot. Assn., Quincy, MA, 1975. 2/25/08 http://smokealarm.nist.gov/ 5 6. How
relevant is the calibration of detectors with smoke from flaming hydrocarbon and smoldering cotton
for smoke from smoldering plastic? How likely is it that the ionization detector’s response is
overestimated and the photoelectric detector’s response is underestimated if the detectors are not
calibrated with smoke caused by smoldering plastic? Differences in characteristics of the smoke from
a smoldering plastic (which is not used for calibration) and from smoldering cotton and for wood on
a hot plate (which are used for calibration) is an active area of research and does have implications
for the design and standard test methods for smoke alarms. Given smoke from, for example,
smoldering upholstered furniture, an ionization alarm of a given design will respond based primarily
on the smoke concentration and size distribution. A photoelectric alarm‟s response will also depend
on its design, the smoke concentration, size distribution and optical properties. It is possible, and
likely, that a photoelectric alarm would respond before an ionization alarm (exactly what was
observed multiple times in the NIST Home Smoke Alarm study) to a rising concentration of such a
smoke. This does not imply any underestimate or overestimate of the various alarm type
calibrations. The flaming hydrocarbon smoke calibrations were performed to allow further study of
these fire tests, specifically modeling of the flaming fire tests, and the response of the smoke alarms.
These calibrations were never used in any of the analysis in NIST TN 1455. 7. How was a detection
time of 1830 seconds determined for both types of detectors (ionization and photoelectric) in the
smoldering bedroom fire involving the 2-story house, especially since many of the detectors did not
respond? Does this raise questions about the adequacy of the methodology used to determine
detection times? The method NIST used to determine the detection time in the smoldering fire
involving the 2- story house was in error. The results of the smoldering mattress test (test SDC 21)
should not have been included in the analysis since the fire development did not allow enough
alarms to respond by the time the test concluded. In addition, test SDC 13 was inadvertently
excluded from the original analysis. The report will be revised to reflect these changes, although this
is unlikely to change any key conclusions or recommendations. 8. Was the “end of test” assumed as
the time to untenability, since the smoking bedroom test did not reach untenable conditions? For
most tests, one or more criteria for untenable conditions were reached prior to the end of test. For
six of the tests, untenable conditions were not reached outside the room of fire origin. For these
tests, the time to untenable conditions was assumed to be as long as the time to the end of test. This
provides a conservative estimate of the time to untenable conditions. 2/25/08
http://smokealarm.nist.gov/ 6 9. Has NIST concluded that adding smoke alarms to the bedrooms
provides no benefit in smoldering mattress fires with detectors placed in the room of fire origin,
since the test results show that for an installation with detectors in every room and in every
bedroom, the ASET is the same as when they are only on every level? It appears the same conclusion
can be drawn about the benefit of placing detectors in the room of fire origin from the flaming living
room fire test. No. The results from the single-story manufactured home show the value of adding
smoke alarms in the bedroom. NIST did not place any detectors in the room of origin (bedroom or
living room) for the 2-story tests. The two-story home was an open floor plan. Alarms were placed in
the foyer and den, with no doors blocking the smoke path. In the analysis, this was considered the
FOREIGN RRL

same space (room) as the living room. There were also smoke alarms in the entranceway,
comparable to the den mid-level detector locations. Scheduling constraints and the number of
available unmodified alarms precluded inclusion of alarms in all locations for some of the field tests.
10. Are both types of detectors (i.e., ionization and photoelectric) equally susceptible to nuisance
alarms? In the smoke alarm research, and in applications in the field, it is documented that most
common ionization detectors have a propensity to produce nuisance alarms during cooking
activities. NIST examined a broad range of activities (including cooking) that yield nuisance alarms.
The published field observations guided the nuisance alarm scenarios studied. Specifically, the
sensitivity to alarm threshold, distance from the source, background air flows, and alarm sensor
(photoelectric or ionization) were examined. Additional measurements were made with aerosol
instrumentation to provide a more fundamental understanding of nuisance alarm sources than has
been previously published. Given the scenarios examined, both photoelectric and ionization alarms
produced nuisance alarms, but NIST does not mean to imply that they are equally susceptible to
such nuisance alarms. Most field data suggest that ionization alarms have a greater propensity to
nuisance alarm than photoelectric alarms, possibly indicating that certain activities such as cooking
dominate reported nuisance alarms in the field. 11. Why did NIST run half the nuisance tests with a
large fan only on one side of the kitchen to blow smoke away from the detectors? Would this not
favor the ionization detectors in the tests? The fan was added to provide an additional variable to
the data set. About half of the manufactured home tests were conducted with all exterior doors and
windows closed and no ventilation of any kind, which is unrepresentative of situations including
open windows, HVAC flows, cooking range fans, movement of occupants, etc. Flow velocities were
monitored just below the ceiling at three locations. Air speeds at the ceiling were typically below 0.1
m/s without the fan and up to 0.5 m/s with the fan. The fan had a tendency to break up the plumes
generated from the cooking activities, and to dilute the aerosols. To the extent that the fan 2/25/08
http://smokealarm.nist.gov/ 7 diluted the nuisance aerosols and distributed them throughout the
connected rooms and subsequently impacted the nuisance alarms produced, it was precisely the
result intended. 12. Have ionization detectors been de-sensitized over time (since the late 1980s)
and are they relatively poor at detecting the kind of smoke given off by today’s synthetic
furnishings? NIST is not aware of any definitive data on the actual sensitivities of the detectors over
the past 20 or 30 years. All of the smoke alarms used in the current study met the sensitivity
requirements specified in the 2002 version of UL 217, and those in the 1975 study met the
requirements of the applicable UL standard at that time.9 The average ionization alarm sensitivity
based on smoke obscuration was found to be 4.2 %/m in the current study and 6.8 %/m in the 1975
study. In other words, the ionization detectors were, on average, slightly more sensitive in the
current study. Photoelectric alarms were, on average, slightly less sensitive in the current study (6.8
%/m versus 4.8 %/m in 1975). 13. Given the clear evidence that ionization alarms should not be used
near kitchens since they often are intentionally disabled to avoid nuisance alarms, why does HUD
allow them to be installed in manufactured homes? NIST is a non-regulatory agency and does not set
the rules for smoke alarm placement. HUD released the final rule "Manufactured Construction and
Safety Standards: Smoke Alarms Rules and Regulations." 10 It was based on the requirements of
NFPA 501. Concerning kitchens the final rule requires: “At least one smoke alarm must be installed in
each of the following locations: (i) To protect both the living area and kitchen space. Manufacturers
are encouraged to locate the alarm in the living area remote from the kitchen and cooking
appliances. A smoke alarm located within 20 feet horizontally of a cooking appliance must
incorporate a temporary silencing feature or be of a photoelectric type.” 11 This requirement should
FOREIGN RRL

reduce nuisance alarms, which should reduce the negative effects (disabling of alarm) of repeated
nuisance alarms. This requirement is consistent with the requirements of the National Fire Alarm
Code, NFPA 72.12 9 The UL standard requires that alarm point based upon smoke obscuration be
within the range of 1.6 %/m to 12.5 %/m. This has not changed. In the 1980s, manufacturers were
allowed to shift production windows (based on a sample of 24 alarms originally submitted by a
manufacturer for testing) by 1.6 %/m to reduce susceptibility to nuisance alarms. The production
window is the sensitivity range the manufacturer must meet during production after Listing with UL.
A shift in the production window to a less-sensitive range is allowed to ensure that alarms produced
represent the least sensitive range. 10 67 FR 12811 (March 19, 2002) 11 Manufactured Home
Construction and Safety Standards, 24 CFR 3280.208(b)(1)(i) (2002). 12 “NFPA 72, National Fire
Alarm Code, 1999 Edition,” Volume 5 of the National Fire Codes, Natl. Fire Protection Association,
Quincy MA 2002. 2/25/08 http://smokealarm.nist.gov/ 8 14. Did NIST consider (1) the different
ignition methodologies and (2) unlikelihood of change in materials as being factors that
changed/decreased the time to untenable conditions in the recent NIST tests compared with the
Indiana Dunes8 tests of the mid1970s? NIST considered a number of factors for the increased fire
growth rates, including ignition methodology and materials of construction. Although the ignition
methodologies were not identical between the NIST tests2 and the original Indiana Dunes tests,8
NIST does not feel they were sufficiently different to fully account for the difference in time to
untenable conditions. The smoldering ignition technique for both test series utilized a heated wire
loop in contact with the item ignited for a similar time period (typically 2 minutes). For flaming
ignition, a larger wastebasket ignition source was used in the original Indiana Dunes tests compared
to the single matchbook ignition source used in the current NIST study. By design, the amount of
paper ignition source in the wastebasket in the original Indiana Dunes tests was controlled so that
the ignition source was fully consumed within 2 min of ignition. For particularly difficult to ignite
items of furniture, additional newspaper was added until ignition was achieved. NIST did not attempt
to adjust the ignition times reported in the original Indiana Dunes tests to account for these difficult
ignitions. While the 1975 study also used actual upholstered furniture and mattresses, these were
purchased from a (charity) resale shop from donated articles. Such items purchased in 1975 would
have been new in the early 1960s or even late 1950s and represented materials and constructions of
that period with typically natural materials. The chairs used in the present study were purchased
from a furniture rental store and, while used, were only a year or two old and of synthetic materials.
The mattresses were purchased new. Thus, the materials were certainly significantly different. While
the internationally accepted ISO tenability criteria used for the NIST study2 were not identical to
those used in the 1975 study,8 the criteria were equivalent or slightly higher in the current study,
which, if anything, would have a tendency to increase the time to untenable conditions for the
current study. Table 3 shows a comparison of tenability criteria used in the two studies. For the
temperature and smoke obscuration criteria, values used in the current study were similar to or
slightly higher than those used in the 1975 study. For CO concentration, the range for the FED-based
model used in the current study includes the value used in the 1975 study, but is also quite a wide
range, depending Table 3. Comparison of Tenability Criteria Used in the 1975 and Current Studies
1975 Study8 Current Study2 Temperature T ≥ 66 °C T ≥ 88 °Ca Gas Concentration CO ≥ 0.04 %
volume fraction CO ≥ 0.02 % – 0.3 % volume fractionb Smoke Obscuration O.D. ≥ 0.23 m-1 O.D. ≥
0.25 m-1 a – value for flaming fires calculated from ISO TS 13571 equation for convected heat b –
range of average values calculated from ISO TS 13571 equation for asphyxiant gases with tenability
times for flaming fires and smoldering fires 2/25/08 http://smokealarm.nist.gov/ 9 on duration of
FOREIGN RRL

the fire. For nearly all tests, the smoke or temperature criterion was met prior to the CO criterion.
The conclusion of faster fire growth rates for flaming fires in the current test series was based on a
comparison of gas temperature histories near ceiling level in the room of fire origin. With the
exception of one room in each of the two test homes in the original Indiana Dunes tests, ceiling
heights were similar in both tests, so comparisons of this gas temperature provides a relative
indication of fire growth rate. From Table 32 of TN 1455,2 flaming fires reached a near-ceiling gas
temperature of 65 °C more than 7 times faster in many of the current tests than in the original
Indiana Dunes tests (130 s versus 970 s). Including an adjustment for delayed ignition for some of
the flaming fire in the original Indiana Dunes tests brings this ratio down to a factor of four. Fire
development in the current tests was generally similar to the growth of heat release rate
determined in other recent studies for upholstered furniture13 and mattresses14. The same trend
was not evident for smoldering fires. 15. Why does NIST in its February 26, 2004, Tech Beat article
highlight the increase in growth rate of fires and fail to point out that this finding applies only to
flaming fires? Why does NIST fail to mention that smoldering fires are not growing faster? The Tech
Beat article was designed to highlight trends that were different between the two studies, especially
one which was so dramatic and critical to life safety. The executive summary of the NIST report
provides a more detailed description of the findings, including the statement: “...the smoldering fire
scenarios are very difficult to reproduce experimentally and tenability times in the present study
have an uncertainty (based upon one standard deviation) which overlaps the uncertainty from the
1975 study. Therefore, caution should be exhibited in drawing conclusions based upon comparisons
of smoldering tenability times between the two studies.” 16. Do the data indicate that the increase
in fire growth rate, which may be far less than NIST estimates due to the different ignition
methodologies for the flaming fire tests, cannot be the reason for the increase in fatalities when the
detector operates? NIST did not study fatalities with working smoke alarms in the study. Average
times to untenable conditions in the current NIST study for flaming and smoldering furniture fires
were found to be 17 % and 49 %, respectively, of those found in the 1975 Indiana Dunes tests, 4 as
can be seen in Table 4 based on a revised analysis of the NIST data (as noted in Question 7, test SDC
13 was added and test SDC 21 was removed from the analysis). On the other hand, the average time
for the cooking fires to reach untenable conditions was 23 % longer in the current study. Since the
cooking materials were similar in the two studies, the shorter time to untenable 13 Cleary, T. G.,
Ohlemiller, T. J, and Villa, K. M., “The Influence of Ignition Source on the Flaming Fire Hazard of
Upholstered Furniture.” Natl. Inst. Stand. Technol., NISTIR 4847, 1992. 14 Ohlemiller, T. J, and Gann,
R. G.., “Estimating Reduced Fire Risk Resulting From an Improved Mattress Flammability Standard.”
Natl. Inst. Stand. Technol., Tech. Note 1446, 2002. 2/25/08 http://smokealarm.nist.gov/ 10
conditions in the furniture fires supports the NIST statement that a major factor in the increase in
fire growth rate is due to differences in modern furniture materials and construction compared to
furniture manufactured four decades ago. 17. Why is NIST highlighting the difference in tenability
criteria, when the obscuration criterion that matters for the smoldering case is essentially the same
as in the Indian Dunes study? Tenability criteria are important since it is the difference between the
alarm time and the time to untenable conditions that determines the available safe egress time. As
noted in the response to Question #15, above, these times were quite different for the current study
and the original Indiana Dunes tests, even with similar tenability criteria in the two studies (see
Question #13). 18. NIST states in its February 26, 2004 Tech Beat article that "The tests also showed
how closed bedroom doors and proper placement of smoke alarms affect one's chance of survival. In
both cases, the time to escape untenable conditions was increased, providing that the individual was
FOREIGN RRL

not in the room where the fire originated." This finding would help a very small group of
homeowners who (1) can be rescued from the bedroom window and (2) have interconnected
detectors including the bedroom. What is the basis for this finding since the approach is harmful to
the larger group of homeowners that has battery-powered smoke alarms and needs to hear the
detector to exit the home safely? In the July 2004 Technical Note 1455, NIST concluded that the
available safe egress time was often quite short. In many cases, available escape time would be
sufficient only if households follow the advice of fire safety educators, including sleeping with doors
closed while using interconnected smoke alarms to provide audible alarm in each bedroom, and pre-
planning and practicing escape so as to reduce pre-movement and movement times. It is this
interconnection that insures all alarms respond to a fire event. Further, building codes require two
ways out of a sleeping room, one of which is generally a window. With the bedroom door closed
there is more time in which to use the window exit should the primary exit be blocked. The latest
version of NFPA 72 requires the installation of fire alarms at more locations in order to improve
audibility in bedrooms where occupants sleep with the door closed, and to provide warning to the
occupants of bedrooms with closed doors when the fire starts in that bedroom. However, audibility
of smoke alarms remains an issue, particularly for sleeping children and Table 4. Comparison of
alarm times and times to untenable conditions for 1975 and current studies 1975 Tests (s) Current
Tests (s) Alarm Times Flaming 146 ± 93 47 ± 35 Smoldering 1931 ± 1103 2042 ± 876 Tenability Times
Flaming 1036 ± 374 177 ± 69 Smoldering 4419 ± 1790 2148 ± 1023 2/25/08
http://smokealarm.nist.gov/ 11 adults impaired with alcohol or other drugs.15 For existing
residences that do not fall under the "new construction" requirements of NFPA 72, the following
approaches are suggested to reduce problems associated with inaudibility: placing smoke alarms in
bedrooms, interconnecting smoke alarms, changing alarm tones, and providing better home fire
escape planning. 19. Why does NIST conclude that increased usage of smoke alarms between 1975
and 2000 can be credited with decrease in home fire deaths by nearly a half when data indicate that
(1) the rate of reduction of fire deaths was relatively independent of the rate of increase in smoke
alarm usage, and (2) the rate of death and injury per 100 cigarette fires increased as smoke alarm
usage tripled? This is not a NIST finding. NIST is citing data based on a 2004 study by the National
Fire Protection Association16. A newer white paper from a broad coalition of public and private
organizations, including NFPA, also supports this conclusion17 . 20. What fraction of fatal residential
fires is initially smoldering in nature and occurs when people are asleep? While nationally-collected
data on fire incidents do not specifically classify fire sources as smoldering and/or flaming, one can
get an indication from the reported cause of death (reported for example as smoke inhalation,
burns, or a combination), the extent of damage, and victim location. An analysis of U.S. data from
1986 to 1990 shows that two-thirds to three-quarters of fire deaths in the United States were due to
smoke inhalation18. It is estimated that more than one-fourth of home fire deaths involve an
extended period of initial smoldering19. According to an analysis of home structure fires from 1999
to 2002, 29 % of civilian fire deaths are due to fires from smoking materials20. Intentionally set fires
and fires originating from heating equipment were responsible for 19 % and 11 % of civilian fire
deaths, respectively. Although kitchens were seen as the leading area of origin for home structure
fires (34 %) and for civilian home fire injuries, 21 % of reported home fires and 51 % of home fire
deaths occur between 11 pm and 7 am when people are unlikely to be cooking in the kitchen and
are likely to asleep in their bedrooms. 15 A Review of the Sound Effectiveness of Residential Smoke
Alarms, CPSC-ES-0502, U.S. Consumer Product Safety Commission, Washington, DC (2004). 16
Ahrens, M., “U.S. Experience with Smoke Alarm and Other Fire Detection/Alarm Equipment.”
FOREIGN RRL

National Fire Protection Association, Quincy, MA (2004). 17 Home Smoke Alarms and other Fire
Detection and Alarm Equipment, Public/Private Fire Safety Council, available from National Fire
Protection Association, Quincy, MA (2006) 18 Gann, R. G., Babrauskas, V., Peacock, R. D., and Hall, J.
R., “Fire Condition for Smoke Toxicity Measurement,” Fire and Materials, Vol. 18, 193-199 (1994). 19
Home Smoke Alarms and other Fire Detection and Alarm Equipment, Public/Private Fire Safety
Council, available from National Fire Protection Association, Quincy, MA (2006) 20 Ahrens, M. “U.S.
Fires in Selected Occupancies, National Fire Protection Association, Quincy, MA (2006) 2/25/08
http://smokealarm.nist.gov/ 12 21. A true understanding of the historical tests shows the ionization
alarm to be deficient whenever synthetic material is smoldered for 30 minutes or more. The recent
NIST study is consistent with this finding. Why has NIST ignored the results of research conducted in
Norway (1991), Australia (1986) and England (1978) which concluded that ionization detectors
provide inadequate warning and escape times for smoldering fires while photoelectric detectors
provide a more effective alternative for such fires? NIST had no intention of overlooking research
conducted elsewhere. The shortening in time to untenable conditions from a fire (either flaming or
smoldering) in modern, synthetic materials indicates the need to determine the ability of the
standard test methods to ensure safe performance of modern (and legacy) residential fire alarms.
Both NIST and UL are involved in research that will assess whether or not changes are required in
the standard to accommodate the changing threat. 22. Why do dual ion/photo alarms appear to
perform worse than individual photo or ion alarms in tables 23, 24, 27 and 28 of the report? The
dual ion/photo alarms appear to perform worse than individual photo or ion alarms in the tables
because the dual-alarms were often located further from the fire source than the individual photo or
ion alarms listed in tables 23, 24, 27 and 28. The affected tables and figures 206-208 have now been
revised by removing the instances where a particular alarm type was not colocated. It is stressed
that the individual alarm times reported in the appendix of the report have not been changed and
remain available for direct comparison of the individual alarms and the dual ion/photo alarm in
every case where these alarms were co-located. Table 5 shows examples of the comparison of alarm
time for the two ionization alarms, the single photoelectric alarm and the dual ion/photo alarm for
when all the alarms were co-located and functioning. The specific test number is indicated in
parentheses. The alarm times come directly from Appendix A of NIST TN 1455-1 (2008). Examples
from the manufactured home (single-story) and the two-story home are presented. The shaded
block indicates the alarm (either one of the ionization alarms, the photoelectric alarm or the dual
photo/ion alarm) that reached the alarm threshold first at the particular location of the alarms. Thus,
the values presented in the dual ion/photo column refer to either the photo or ionization sensor,
whichever one sounded first in the dual sensor device. NIST TN 1455-1 was not meant to provide a
detailed comparison of individual alarms to dual technology alarms. Rather, it was intended to
provide an overall assessment of the performance of smoke alarms placed according to the broad
placement categories specified in building and fire codes. 2/25/08 http://smokealarm.nist.gov/ 13
Table 5. Sample comparison of individual ionization and photoelectric alarm times with collocated
dual photo/ion alarms Fire Scenario Location of Alarms Alarm Time (seconds) Ion #1 Ion #3 Photo #1
Dual Photo/Ion Flaming Chair in Living room (SDC02) Util. Hall 70 78 122 66 Master Bedroom 106
124 154 102 Smoldering Chair in Living Room (SDC01) Util. Hall 6087 6112 5262 5282 Master
Bedroom 6262 6287 5447 5512 Smoldering Mattress in bedroom (SDC04) Util. Hall 3393 3413 3418
3398 Master Bedroom 3523 3518 3523 3503 Flaming Mattress in bedroom (SDC05) Util. Hall 117 117
127 107 Master Bedroom 147 147 167 147 Kitchen Fire (SDC12) 2nd bedroom 977 797 839 797
Bedroom hall 737 795 847 657 Flaming Chair in Living room (SDC25)* Den 122 124 152 118
FOREIGN RRL

Smoldering Chair in Living Room (SDC23)* Upstairs Hallway 4824 4886 1542 1508 Kitchen Fire
(SDC24)* Den 1290 1500 1194 876 Upstairs Hallway 1554 1554 880 898 *Two-story home test 23.
Did the modifications to the dual-sensor smoke alarms that were used in the test affect their ability
to react to smoke conditions, and did the modified dual-sensor smoke alarms meet the UL listing
criteria? The photoelectric and ionization sensing chambers in the dual ion/photo alarms were
calibrated in the same manner as the individual ionization and photoelectric alarms (see answer to
question 5 of this FAQ). Additionally, in the computation of alarm times the same alarm levels were
specified for either the individual photoelectric alarm and dual alarm photoelectric sensor, or the
individual ionization alarms and the dual alarm ionization sensor as specified in table 5 of the report.
Given that the alarm levels specified in table 5 fall within the range allowed by UL 217, the modified
dual ion/photo alarms would meet the UL requirements just like the individual photoelectric and
ionization alarms. In dual ion/photo alarms that meet UL 217 requirements, manufacturers are
allowed to set the sensitivity of the photoelectric and ionization sensing chambers to any allowable
values so long as the dual alarm passes all applicable sensitivity tests. Thus manufacturers are free to
set alarm sensitivities in dual photoelectric/ionization alarms less sensitive than in individual sensor
alarms in order to reduce nuisance alarms.

Supplementary Questions and Answers Clarifying “Detector Sensitivity and Siting Requirements for
Dwellings,” Phase I (NBS GCR 75-51) and Phase II (NBS GCR 77-82) 1. Does the National Institute of
FOREIGN RRL

Standards and Technology (NIST) have regulatory powers and does it certify smoke detectors? No.
NIST is a non-regulatory agency of the U.S. Department of Commerce. NIST does not “certify” smoke
detectors. 2. Could working ionization detectors fail to detect the type of smoke most commonly
produced by the early fire? If so, state the conditions under which this may occur. Both ionization
and photoelectric sensors detect particles. Photoelectric technology is sensitive to the mass density
of particles of a diameter of the same order as the wavelength of the light source in its sensor.
Ionization technology is sensitive to the sum of all particle diameters and so it can respond to smaller
particle diameters present in greater numbers. Because smoke from a smoldering fire produces
relatively fewer particles and the particles are relatively larger, it is possible for an ionization-type
detector to alarm at a time well after smoke from a smoldering fire is sensed by a photoelectric
sensor. 3. Did NIST ever compare the performance of a single ionization detector with a
“throughout-thehome” fire detection system with a detector (heat, ionization, or photoelectric) in
every room? If so, what did NIST find from its studies? No. A pioneering Canadian study published in
1962 (McGuire and Ruscoe) postulated the performance of a single smoke alarm outside the
sleeping room(s) based on the judgment of experts reviewing fire reports. In 1974 this study was
cited as the basis for a minimum requirement for a single smoke alarm outside the bedrooms in
homes. In 1975, NIST (then NBS) funded tests to examine if one smoke alarm was sufficient. The
tests indicated that it was not sufficient for multilevel homes, leading to the “every level”
requirement common to most modern regulations. The 1975 tests examined the performance of a
heat detector in the room of fire origin, and in the second year (1976) looked at heat and smoke
detectors in the room of fire origin, both equivalent to an “every room” alarm placement. These
studies concluded that, while detectors in every room provided some increase in escape time, the
increase was incremental and not justified by the cost as a minimum code requirement. The 2004
NIST tests also examined several installation strategies, including “every level”, “every level plus
bedrooms”, and “every room.” Due mostly to the systematically lower escape times observed in the
more recent tests, it appears that there is justification to apply the "every level plus bedroom"
strategy (that is currently the minimum requirement for new homes) to existing homes. Both the
Phase I and Phase II tests utilized heat, ionization, and photoelectric technologies and concluded
that particulate sensors (either ionization or photoelectric) were adequate when placed
appropriately and that heat sensors did not provide any additional protection. 1/12 4. Did NIST ever
conclude from any of its studies that the warning provided by a single ionization detector “would be
so rapid that an occupant would be able to correct the pre-fire condition before any fire actually
developed”? No. Over the years numerous claims were made by marketers of ionization and
photoelectric technologies to demonstrate the superiority of their products. Some ionization
manufacturers claimed that their detectors could sense “invisible particles of combustion that are
present before any visible smoke or flames.” While the latter statement may be true in some flaming
fire situations (such as burning alcohol fuels), NIST (NBS) never attempted to reproduce nor
commented on these claims. 5. Did the operating times when the fires were smoldering in Phase I of
the Indiana Dunes Tests1,2 average more than one hour? If so, state the observed conditions when
the fires were smoldering. The average test duration for the smoldering fires in the Phase I study
was 5266 s ± 1907 s. In all of these tests, the test item smoldered for an extended period of time,
averaging 4298 s ± 1929 s before transitioning to flaming. For the Phase I smoldering tests, the
average time to first alarm was 2489 s ± 1324 s for ionization alarms and 1927 s ± 1065 s for
photoelectric alarms. 6. Did the ionization detectors in the Phase I smoldering tests remain silent for
over an hour? If so, why did they remain silent? In the Phase I smoldering tests, the average time to
FOREIGN RRL

first alarm was 2489 s ± 1324 s for ionization alarms and 1927 s ± 1065 s for photoelectric alarms.
The time to first alarm was greater than 3600 s in 4 of the 23 tests for the ionization alarms and in 2
of the 23 tests for photoelectric alarms. The reasons for the delay with both alarm technologies are
associated with the slow rate at which smoldering fires produce smoke and the much longer
transport time to the ceiling due to the low buoyancy of a smoldering plume. In addition, because
the number of particles that eventually reach either detector is relatively smaller from a smoldering
source, the technology that is based upon number density (i.e., ionization detectors) is slower to
respond. 7. Were most Phase I smoldering tests terminated shortly after one hour? If so, why were
they terminated? The average test duration for the smoldering fires in the Phase I study was 5266 s
± 1907 s with tests as long as 8520 s. Tests were terminated after one or more pre-established
tenability limits were exceeded, since any devices responding after that time would not affect the
escape time provided. In Phase I, the first 36 and the last of the 40 tests exceeded a tenability limit
prior to termination. Tests 37 (overloaded electrical motor), 38 and 39 (overloaded extension cord)
did not exceed a tenability limit and these tests were terminated after the test items burned out on
their own. 1 Bukowski, R.W., Waterman, T.E., and Christian, W.J., “Detector Siting and Sensitivity
Requirements for Dwellings.” IIT Research Institute Report prepared for Natl. Bur. Stand. (U.S.), NBS
GCR 75-51 (1975) 2 Harpe, S.W., T.E., and Christian, W.J., “Detector Siting and Sensitivity
Requirements for Dwellings – Phase 2.” IIT Research Institute Report prepared for Natl. Bur. Stand.
(U.S.), NBS GCR 77-82 (1977) 2 8. Was there a one-hour “average” time to warn of a smoldering fire
when using ionization detectors in the Phase I tests? Was this based on “prematurely terminated
tests”? If not, state why that is not the case? As stated above, the average time to first alarm was
2489 s ± 1324 s for ionization alarms. No tests were prematurely terminated since in every case
(except the three electrical fires mentioned above) test termination did not occur until tenability had
been exceeded. For the smoldering tests, the average time from first alarm to the onset of
untenable conditions was 1926 s ± 1454 s for the ionization alarms. Note that many of the smoke
detectors (ionization and photoelectric) were located well away from the origin of the fire and not in
the path of egress. For those tests where not all these alarms activated prior to test termination (8 of
23 smoldering tests for the ionization detectors and 2 of 23 smoldering tests for the photoelectric
detectors), it is possible that the time to alarm would have exceeded one hour, which could increase
the average time to warn of a smoldering fire. 9. Did the Phase I report find that “photoelectric type
detectors seem to respond better to the smoldering type fires and the ionization detectors seem to
respond slightly better to the flaming fires”? What was the technical basis for this finding? Yes, at
any given detector location, most photoelectric detectors responded sooner than most ionization
detectors to smoldering fires and most ionization detectors responded sooner than most
photoelectric detectors for flaming fires. On average, the first ionization alarm responded 147 s ± 92
s before the first photoelectric alarm in the flaming fires (in all cases, an ionization alarm responded
before a photoelectric alarm). On average the first photoelectric alarm responded 562 s ± 800 s
before the first ionization alarm in the smoldering fires (in 20 of the 23 smoldering tests,
photoelectric alarms responded before an ionization alarm). The technical basis for this result lies
with the differences in the character of smoke from a smoldering versus a flaming fire, and from the
different operating principles of the detectors. Both ionization and photoelectric sensors detect
particles. Photoelectric technology is sensitive to the mass density of particles of a diameter of the
same order as the wavelength of the light source in its sensor. Ionization technology is sensitive to
the number density of particles and so responds to smaller particle diameters present in greater
numbers. Smoke from a smoldering fire produces relatively fewer particles and the particles are
FOREIGN RRL

relatively larger, as compared to particles from a flaming fire which are much more numerous
though generally smaller. 10. Did the Phase I report conclude that “In general, all detectors
responded well to fires” and “A residential smoke detector of either the ionization or photoelectric
types with small lag time would provide more than adequate life saving potential under most real
residential fire conditions when properly installed”? What was the technical basis for this
conclusion? Yes. For the “every level” and “every room” criteria, escape times provided by both
smoke sensing technologies (ionization and photoelectric) were positive for nearly all tests. In the
end, NIST (NBS) pointed out that the adequacy of the time provided would depend on the time
needed by a specific group of people in a specific household. The report presents escape times in a
probability plot (percent of the tests in which a specific escape time was provided for a specific
installation location, sensor type, or sensitivity). An independent analysis of the Phase I data by a
State Fire Protection 3 Board considering detector legislation selected a 3 min (minimum) escape
time criterion3 . On this basis they concluded that every level smoke detectors (either ionization or
photoelectric) provided 3 min of escape time in 89 % of the tests (average escape time +18.6 min), a
single smoke detector outside the bedroom provided 3 min of escape time in 35 % of the tests
(average escape time +4.5 min), a rate-of-rise heat detector in every room provided 3 min of escape
time in 19 % of the tests (average escape time -0.6 min) and a fixed temperature heat detector in
every room provided 3 min of escape time in 11 % of the tests (average escape time -2.2 min). 11.
What criteria did the Phase I study use to determine “life saving potential”? How were the criteria
judged to be “more than adequate”? “Life saving potential” was judged by the time for escape
provided in advance of any untenable condition appearing anywhere along the “primary escape
path,” which was defined as the path from any room to one of the doors to the exterior. While
building regulations require a second means of egress from any residential room (normally a
window) the researchers did not judge escape out a window as a success. The tenability criteria
selected were the occurrence of a smoke optical density of 0.07 per foot, a temperature of 150 °F, or
a CO concentration of 0.04 % averaged over one hour. The substantiation for these limits including
literature references is provided in Appendix D of the reports. For CO, the report observes that the
available data is all for exposure to a constant CO level for a given time and the observed CO levels in
the tests rose almost linearly with time. This is the basis for using a (one hour) time averaged
concentration. Smoke was always the first tenability limit exceeded (in two tests the CO limit was
approached but not exceeded before the smoke limit was reached.) All three criteria were
considered conservative, based on the literature cited in the report. This is especially true of smoke
since it only relates to loss of visibility and because escape paths in homes are generally short and
are generally familiar to the occupants. 12. What criteria did the Phase I study use to determine
“most real residential fire conditions”? The tests involved real detectors available for purchase
installed in the locations cited in the national standard in real homes in which real contents were
ignited in ways that are observed in data collected in real fires. On this basis the researchers felt that
the tests represented real fire scenarios to the maximum extent possible. 13. Did the report(s) from
the Indiana Dunes Tests find or conclude that the smoke detector (ionization or photoelectric) is
“the ONLY type of detector” capable of providing adequate warning of a flaming fire? If so, state the
technical basis for this finding or conclusion. The report did not state that “the ONLY type of
detector” capable of providing adequate warning of a flaming fire is the smoke detector (ionization
or photoelectric). It presents escape times in a probability plot (percent of the tests in which a
specific escape time was provided for a specific installation location, sensor type, or sensitivity). An
independent analysis of the Phase I data by a State Fire Protection Board considering detector
FOREIGN RRL

legislation selected a 3 min (minimum) escape time criterion. On this basis they concluded that every
level smoke detectors (either ionization or photoelectric) provided 3 min of escape time in 89 % of
the tests (average escape time +18.6 min), a 3 Wilson, R., Computer Analysis of Data on Fire
Detectors Available for Purchase in Massachusetts, Data from the National Bureau of Standards
Indiana Dunes Tests Record, Contract No. 4-36092, Massachusetts Fire Prevention Fire Protection
Board, January 1976. 4 single smoke detector outside the bedroom provided 3 min of escape time in
35 % of the tests (average escape time +4.5 min), a rate-of-rise heat detector in every room provided
3 min of escape time in 19 % of the tests (average escape time -0.6 min) and a fixed temperature
heat detector in every room provided 3 min of escape time in 11 % of the tests (average escape time
-2.2 min). 14. Is the heat detector (mechanical or electrical) used as the trigger for fire sprinkler
systems? Above what average temperature (°F) do the ideal and typical detectors trigger? Heat
detectors are not typically used to trigger fire sprinkler systems. While there have been attempts at
commercialization of such systems, they have not proved viable. Rather, they are most typically
separate systems. Heat detectors are most often rated to operate at fixed temperatures from 135 °F
to 175 °F. Some fire sprinklers utilize heat sensors of a type similar to those used in the mechanical
heat detectors used in Phase II (eutectic solder). The most common fire sprinklers have a rated
activation temperature of 165 °F with some residential use sprinklers rated as low as 135 °F. In Phase
I the response of fixed temperature heat detectors was estimated from thermocouples at
temperatures of 135 °F and 150 °F. Actual rate-of-rise heat detectors were present beginning with
test 14. In Phase II actual (fixed temperature) mechanical heat detectors with activation
temperatures of 135 °F were installed in the fire room through test 63 at which time all devices
supplied by the manufacturers were expended. 15. How does the heat detector (either type) in a fire
sprinkler system perform in the two major types of fires (smoldering fires and flaming fires)? Is the
heat detector reliable for this application? Heat detectors are not typically used to trigger fire
sprinkler systems. Rather, they are most typically separate systems. Heat detectors are most often
rated to operate at fixed temperatures from 135 °F to 175 °F. Some fire sprinklers utilize heat
sensors of a type similar to those used in the mechanical heat detectors used in Phase II (eutectic
solder). A heat detector or a fire sprinkler cannot detect a smoldering fire while it is smoldering since
it does not raise the room temperature appreciably. The purpose of residential fire sprinklers is
distinctly different from residential smoke alarms however. From the National Fire Protection
Association standard, NFPA 13D4 , these sprinklers are “expected to prevent flashover (total
involvement) in the room of fire origin,” thus suppressing the fire and limiting the spread of fire and
toxic conditions in rooms remote from the fire. Activation of heat detectors or fire sprinklers is often
significantly later in the fire than smoke alarms. In the Phase I tests, heat detector activation was
well after the time to untenable conditions in all fires and after the recorded transition to flaming in
all but one of the smoldering tests. Neither the Phase I nor Phase II tests included residential fire
sprinklers since these were not developed until several years later. There were rare residential
sprinkler systems that utilized commercial sprinkler heads, but these did not represent a significant
force in the market. In the NIST 2004 study5 so-called “tell tale” sprinklers (commercially available
fire sprinklers typical of those installed in residential systems but not connected to a water supply)
with a temperature rating of 155 °F were installed in the fire room 4 “NFPA 13D, Installation of
Sprinkler Systems in One- and Two-Family Dwellings and Manufactured Homes, 2007 Edition,”
Volume 2 of the National Fire Codes, Natl. Fire Protection Association, Quincy MA 2007. 5 Bukowski,
R. W., Peacock, R. D., Averill, J. D., Cleary, T. G., Bryner, N. P., Walton, W. D., Reneke, P. A., and
Kuligowski, E. D. Performance of Home Smoke Alarms, Analysis of the Response of Several Available
FOREIGN RRL

Technologies in Residential Fire Settings, Natl. Inst. Stand. Technol., Tech. Note 1455 (2004) 5 for all
tests. Activation times recorded for flaming fires ranged between 126 s and 247 s, and for
smoldering fires they did not activate until after transition to flaming. Specific times are documented
in the report. 16. Did the Dunes study find that the heat detector (mechanical or electrical) is not a
reliable alarm device to warn of a flaming fire? What was the technical basis for this finding? The
tests showed that heat detectors in every room were far less effective than smoke alarms on every
level in the majority of tests (see answer to question 10). 17. Did the Phase I study find that the
smoldering fires involving heat detectors did not produce “any significant heat”? If so, why was that
the case? Is this representative of typical fires in a residential dwelling to evaluate detector
performance? Smoldering fires by their nature do not produce any significant heat as smoldering is a
very slow combustion process. At the time, NFPA data showed smoldering fires might make up 2/3
of all (fatal) residential fire starts as estimated from fire reports where the ignition source was
smoking materials and the first item ignited was upholstered furniture or mattresses. For this reason
the test scenarios in the 1975/76 tests were approximately 2/3 smoldering ignitions and 1/3 flaming
ignitions. 18. Did the Phase I study find that the flaming fires involving heat detectors did not “raise
the ceiling temperatures in the fire rooms to the modest 135 °F which was the set temperature[s]
for the heat detectors”? If so, why was that the case? Is this representative of typical fires in a
residential dwelling to evaluate detector performance? An independent analysis (Patton 1992) listed
27 tests (12 in Phase I and 15 in Phase II) that reached or exceeded 200 °F. Some of these began as
flaming and others were smoldering and transitioned to flaming. Several conditions must occur to
raise the temperature of the air below the ceiling. First there must be sufficient heat release from
the fuel, second the plume entrains cooler ambient air that reduces the temperature by mixing, and
third is heat lost to the ceiling by convection (initially as much as 90 % of the energy in the plume is
lost to the ceiling). These processes are typical of fires in compartments. 19. Did the Phase II study
find that in the six flaming fires involving heat detectors “the ceiling temperatures went above 200 °F
but never above 500 °F”? If so, why was that the case? Is this representative of typical fires in a
residential dwelling to evaluate detector performance? The analysis cited above (Patton 1992)
indicated that all 27 tests that reached or exceeded 200 °F did not exceed 500 °F. This should be
expected since the tenability criterion for temperature was 150 °F (at the 5 foot level) that was
never exceeded prior to test termination based on exceeding the smoke criterion. Since the
activation temperature of all of the heat detectors was 135 °F (150 °F for the thermocouple reading
to simulate thermal inertia) this would have provided an opportunity for the heat detectors to
respond had the temperature exceeded their activation temperature. 6 20. Were any of the heat
detectors (electrical or mechanical) removed from the fire rooms before lighting the fires in the six
Phase II flaming fire tests? If so, why were they removed? None of the heat detectors were
removed. In the Phase II tests actual mechanical heat detectors were used. Since these are not sold
in retail stores they were obtained from the manufacturers. Each of two manufacturers (that utilized
a horn or bell as the sounding device) provided a case of 24 units. One of each was included in each
Phase II test starting with test 41 and ending with test 63 at which time all 24 samples had been
used. Both devices used eutectic solder sensors which have a characteristic that the activation
temperature can decrease if they are exposed to an elevated temperature without activating. Thus
each unit could be used in only one test even if it did not activate. 21. In how many of the six Phase II
flaming fire tests were there no electrical heat detectors? Mechanical detectors? If so, what was the
technical basis for this condition(s)? Officially there were 15 flaming tests in Phase II (44, 45, 49, 50,
58, 59, 63, 64, 66, 68, 69, 70, 71, 74, and 76). The mechanical heat detectors were present in 7 (44,
FOREIGN RRL

45, 49, 50, 58, 59, and 63). Electrical heat detectors were not used in Phase II since the mechanical
devices were present. Estimated response of heat detectors from thermocouple readings was not
made although a thermocouple was present in the room of fire origin and the data were published
in the report. 22. How many smoldering fire tests were conducted in the Phase I test series? 23 of 37
tests (not counting the three electrical fires) in Phase I were smoldering. 23. Were nine ionization
detectors installed in each of the Phase I smoldering fire tests? If not, state how many were installed
in each of the tests. The smoke alarms used in the Phase I tests were arranged on one large board
and two “half” boards. The large board had 4 ionization, 4 photoelectric, and a dual gate (an older
technology that used an early ionization detector paired with an electrical resistance circuit)
detector. One half board had 3 ionization and 2 photoelectric detectors, and the other half board
had 2 ionization, 2 photoelectric, and a dual gate detector. The alarm boards were located in the first
and second floor hallways (outside sleeping rooms) and one of the half boards was moved to the top
of the basement stairs for fires in the basement. The half boards were constructed to assess any
differences between wall mounting and ceiling mounting (one half board on the wall and one on the
ceiling at the same location). For some first floor fires the half boards were in the first floor hallway
and for some on the second floor hallway to examine wall vs. ceiling at that location. The large board
was used at the other hall location. In some tests the half boards were switched (wall for ceiling) to
eliminate systematic errors. These detectors; 9 ionization, 8 photoelectric, and two dual gate (plus
heat detectors in the fire room), were present in all tests. Not all detectors were operational in all
tests. For tests 28 through 40, one of the detector locations (clock No. 25, a photoelectric detector)
was taken out of service due to physical breakage of the detector. 7 24. Did the Phase I study find
that: (a) not a single ionization detector sounded within five minutes of the initiation of the Phase I
smoldering fire tests? (b) Only one time did an ionization detector sound within 10 minutes of the
initiation of the Phase I smoldering fire tests? (c) Only 26 times (or about 16 percent of possible
times) did the ionization detector sound within the first 30 minutes of the initiation of the Phase I
smoldering fire tests? (d) The ionization detector never sounded 44 times at the time of the
termination of the Phase I smoldering fire tests? State the technical basis for each of the above with
the responses. (a) Neither ionization detectors nor photoelectric detectors alarmed within the first
five minutes in any test. The shortest alarm times in the smoldering tests were 473 s in test 6 for an
ionization and 468 s in test 6 for a photoelectric alarm (b) Only tests 5 and 6 had alarm times of less
than 10 min for both ionization detectors and photoelectric detectors. (c) A total of 21 ionization
detectors of a possible 207 alarmed within the first 30 min of test initiation. A total of 30
photoelectric detectors of a possible 166 alarmed within the first 30 minutes of test initiation. (d) A
total of 21 ionization detectors out of a possible 207 did not alarm prior to test termination. A total
of 5 photoelectric alarms out of a possible 166 did not alarm prior to test termination in all the tests.
It is important to note that the number of detectors that alarm within a particular time may not be
that valuable. More important is a comparison of alarm time to the time to untenable conditions. In
all tests, multiple detectors of both types provide alarms well before untenable conditions. For the
smoldering tests, the average time from first alarm to the onset of untenable conditions was 1926 s
± 1454 s for the ionization alarms and 2511 s ± 1667 s for the photoelectric alarms. 25. Did the Phase
I study find that the ionization detector did not respond to the smoldering fires in the Phase I tests
so long as they remained smoldering fires? If so, why was that the case? No, the Phase I study did
not find this. In all tests, at least one ionization alarm responded well before the transition to
flaming. 26. Did the Phase I study find that “long before an hour has passed (average time to
operate) potentially deadly levels of carbon monoxide and other toxic gases were created” in the
FOREIGN RRL

smoldering fire tests? Is this statement a reasonable representation of knowledge among experts at
the time the tests were conducted? Today? The tenability limit used for the Phase I study for CO was
a time averaged (one hour) concentration of 400 ppm (based on a review of literature available at
the time). This limit is roughly consistent with limits in today’s literature6 . In only 2 of the 40
experiments carbon monoxide concentrations 6 Most typically, tenability is judged using the ISO
Technical Specification 13571. For CO, this is a fractional effective dose calculation, FED, of FED =
([CO] 35000 Δt t1 t 2 ∑ ) where the CO concentration, [CO] is in ppm and the time, t, is in min.
Incapacitation is considered to have occurred at FED values between 0.3 and 0.5; lethality at a value
of 1. The 400 ppm one hour average limit is equivalent to an FED of 0.68. 8 approached but did not
exceed the time averaged (one hour) concentration of 400 ppm before the optical density reached
0.07 per ft. Nevertheless, the occurrence of the critical optical density preceded the occurrence of
critical carbon monoxide levels in all of the experiments. Today, the guidance regarding CO toxicity is
more conservative, with an accepted tenability limit of 175 ppm for one hour for the most sensitive
population. We are reexamining the data from Phase I and the more recent NIST home smoke alarm
study5 to determine how often this limit may have been exceeded inside the room of origin and
outside the room of origin along the egress path prior to alarm. 27. Did the Phase I study find that
the ionization detector sounded only after the smoldering fire was heating up and converting, or had
already converted, to a flaming (hot) fire? If not, why was this not the case? No, in all tests, at least
one alarm of each type, ionization and photoelectric, responded well before the transition to
flaming. On average, the time from first alarm to the transition to flaming was 1962 s ± 1678 s for
ionization alarms and 2445 s ± 1967 s for photoelectric alarms. In all cases, the time to first alarm
preceded the transition to flaming in the smoldering tests. The minimum time from first alarm to the
transition to flaming was 40 s for the ionization alarms and 41 s for the photoelectric alarms (both in
test JR-22). 28. Is the type of smoke created in the standard smoldering fire test developed by
Underwriters Laboratories (i.e., where Ponderosa Pine sticks on a hot plate are raised closed to their
autoignition temperature of about 700 °F) similar to the type of smoke created by a burning
cigarette on upholstered furniture or bedding? If not, state the major differences. At the time of
publication of the study (1975/76) there was no smoldering test in the UL standard (UL217). The
Dunes tests clearly showed the importance of response to such fires. UL conducted a study to
develop a suitable test. They smoldered cotton-stuffed mattresses similar to those used in the Dunes
tests in their fire test room to develop smoke vs. time profiles. They then experimented with
different techniques and materials to generate the observed conditions in a reproducible manner.
Heating ponderosa pine wood strips on a temperature-programmed hotplate obtained the desired
result7 . Recently, UL has conducted tests following the NIST Home Smoke Alarm Project and
concluded that smoldering and flaming polyurethane foam smoke has characteristics different from
the smoke in their standard test fires8 . UL is now investigating the possibility of adding additional
standard tests using polyurethane foam. 7 Harpe, S.W., and Christian, W.J., “Development of a
Smoldering-Fire Test for Household Smoke Detectors,” Fire Journal, Vol. 73, No. 3, 1979 8 Smoke
Characterization Project Final Report, Underwriters Laboratories Inc. 333 Pfingsten Road,
Northbrook, IL 60062, April 24, 2007) 9 29. Does the industry standard used to certify fire alarms in
flaming fires allow the smoke obscuration to go as high as 37 percent? If so, why is that the case
when the International Association of Fire Chiefs set 4 percent smoke obscuration as the maximum
for travel along an exit path on the basis of the live fire tests conducted by the Los Angeles Fire
Department in the aftermath of the December 1958 school fire? The UL fire tests are not tests of
sensitivity but rather reproduce a range of conditions observed in real fires. The paper fire produces
FOREIGN RRL

a high smoke density (about 30 % per foot at the ceiling and 37 % at the wall mounting location) for
a short duration (about 10 s). Detecting this fire assures that the device will not ignore a high initial
pulse of smoke due to internal time delays. The basic sensitivity (alarm threshold) is measured in the
sensitivity test in a test compartment where the alarm threshold must be within the range of
sensitivities that passed the fire tests and that which passed the false alarm tests, but not outside
the range on 0.5 to 4.0 percent per foot of grey smoke (black smoke sensitivity can go to 10% but
the mass density ratio of black to grey smoke is about 5:1, meaning that the mass density of 10 %
black smoke is equal to that of 2 % grey smoke). 30. Why were detectors with a sensitivity of 1 %/ft
used in Phase I and Phase II testing when typical sensitivities of smoke alarms used by consumers are
2 %/ft? The studies used detectors that covered a range of sensitivities, nominally 1 %/ft or 2 %/ft in
order to evaluate the life safety impact of detector sensitivity. Actual sensitivities were measured
and reported in the appendices of each report. 31. Has any study conducted by NIST (or the
predecessor National Bureau of Standards) concluded that a combination of both the ionization and
the photoelectric smoke detector provides more than adequate life saving potential under most real
residential fire conditions when properly installed? If so, what is the technical basis for this
conclusion? The subject of combination (ion and photo) detectors was never raised in the 1975/76
test reports. However in the 1979 report (NBSIR 79-1915) on the results of similar tests conducted in
mobile (manufactured) homes, conclusion 6 stated, “Based on the series of fire tests conducted in
this program and the evaluation criteria discussed in section 2.3, a properly functioning smoke
detector of either the ionization or photoelectric type should provide an alarm in sufficient time to
permit an alert and mobile occupant to escape from the mobile home. While either type detector
provided enough time for escape, the use of a detector which combines both ionization and
photoelectric sensors in the same unit (or one of each) could provide significant improvement in
alerting the occupants of a mobile home to either a flaming or smoldering fire.” 32. Has any
publication or presentation by a NIST researcher suggested that a “combination of both the
ionization device and the photoelectric smoke detector” may provide more than adequate life saving
potential under most real residential fire conditions when properly installed? If so, provide complete
citations for all publications that make this suggestion. Yes. See: Bukowski, R.W., Investigation of the
Effects of Heating and Air Conditioning on the Performance of Smoke Detectors in Mobile Homes,
NBSIR 79-1915, Nat Bur Stand, Gaithersburg, MD, 1979. 10 Also, there is a fact sheet jointly
produced by the USFA, HUD, and NIST which states: “There are two kinds of smoke alarms --
ionization and photoelectric. The ionization activate quicker to fast, flaming fires and the
photoelectric are quicker for slow, smoldering fires. Either one will provide enough time to get out,
but having a mix of the two types is a good idea. Models with both sensors are better than single
sensor units, but of course they cost more.” This fact sheet can be found at:
http://www.fire.nist.gov/factsheets/Smoke&CO.pdf and
http://www.usfa.dhs.gov/downloads/pdf/smokeco-mh.pdf 33. Does the ionization detector provide
more than adequate life saving potential under most real residential fire conditions when properly
installed? If not, cite specific limitations of the ionization detector for flaming fires and for
smoldering fires. In the 1975/76 tests the answer is clearly yes. In the 2000 tests the performance of
both types is often close to the margin, possibly due to the much more rapid development of fires in
modern furniture as opposed to the detectors. If the furniture produces conditions that exceed
tenability limits in a much shorter time, then detectors that meet the current UL test standards may
be inadequate to provide sufficient warning to occupants intimate with the fire or who need 3
minutes or more to escape. For example, victims located in the room of fire origin are considered
FOREIGN RRL

intimate with the fire and may not have adequate warning based on the established tenability
criteria for smoldering or flaming fires, especially if the room of fire origin does not contain a smoke
alarm. Impaired occupants, small children, and mobility-limited occupants incapable of self-rescue
may be unable to evacuate safely. In summary, the NIST studies conclude that both ionization and
photoelectric alarms provide enough time to save lives for most of the population under many fire
scenarios; however, ionization alarms may not always alarm even when a room is filled with smoke
from a smoldering fire, exposing the most sensitive populations with mobility limitations to an
undetermined risk. Photoelectric detectors can provide a lot more warning time than ionization
detectors in a smoldering fire; at the same time a smoldering fire can take a longer period to become
dangerous. Ionization detectors can provide a little more time than photoelectric detectors in a
flaming fire; in this case there can be little time to spare. Changes in furnishing materials and
construction over the past decades have reduced the time available for safe egress in any fire. NIST
is currently conducting research to assess whether or not modifications may be needed in the
standard test method for certifying residential smoke alarms to accommodate the changing threat.
34. Does the photoelectric detector provide more than adequate life saving potential under most
real residential fire conditions when properly installed? If not, cite specific limitations of the
photoelectric detector for flaming fires and for smoldering fires. If the room of fire origin is to be
explicitly addressed in the adequacy of life saving potential then, photoelectric detectors would
most likely be deemed adequate for smoldering fires. For flaming fires, the photoelectric detector
would tend to alarm after an ionization alarm, so the rate of fire growth would determine the
outcome. 11 In summary, the NIST studies conclude that both ionization and photoelectric alarms
provide enough time to save lives for most of the population under many fire scenarios; however,
ionization alarms may not always alarm even when a room is filled with smoke from a smoldering
fire, exposing the most sensitive populations with mobility limitations to an undetermined risk.
Photoelectric detectors can provide a lot more warning time than ionization detectors in a
smoldering fire; at the same time a smoldering fire can take a longer period to become dangerous.
Ionization detectors can provide a little more time than photoelectric detectors in a flaming fire; in
this case there can be little time to spare. Changes in furnishing materials and construction over the
past decades have reduced the time available for safe egress in any fire. NIST is currently conducting
research to assess whether or not modifications may be needed in the standard test method for
certifying residential smoke alarms to accommodate the changing threat. 35. Why is the ionization
detector, which senses the products of combustion and not the large smoke particles caused by a
fire, called a “smoke detector”? If the ionization detector does not sense smoke, is it still technically
appropriate to call it a smoke detector? Why? All US Codes and Standards consider both ionization
and photoelectric detectors as smoke sensing fire detectors responding to particles produced by
combustion. The official NFPA definition of smoke (found in NFPA 90A) is: Smoke. The airborne solid
and liquid particles and gases evolved when a material undergoes pyrolysis or combustion, together
with the quantity of air that is entrained or otherwise mixed into the mass. In particular, the
ionization alarm senses smoke particles of all sizes that enter its sensing chamber. Its sensitivity is
linear with the particle diameter. The photoelectric, light-scattering, detector’s sensitivity is
proportional to the smoke particle diameter raised an exponent that decreases from a power of 6 to
a power of two as the diameter increases from smallest to largest sizes9 . 36. Did the results from
one or more studies conducted by NIST indicate that “more than 50 percent of the time” a new
ionization detector with a fresh battery will fail to warn in a timely manner? No. 9 Mulholland, G. W.,
FOREIGN RRL

and Liu, B. Y. H. “Response of Smoke Detctors to Monodisperse Aerosols,” Journal of Research of the
National Bureau of Standards, Vol. 85, No. 3, 223-238, May/June 1980.

4
FOREIGN RRL

Statement for the Record National Institute of Standards and Technology to the Boston City Council
Committee on Public Safety August 6, 2007 This statement provides technical information on smoke
detector technology based on research conducted by the U.S. Commerce Department’s National
Institute of Standards and Technology (NIST), especially on the differences between ionization and
photoelectric technology. NIST is a non-regulatory federal research agency that specializes in
measurement and basic standards, and has been engaged in fire research for more than a century.
Our mission is to promote U.S. innovation and industrial competitiveness by advancing
measurement science, standards, and technology in ways that enhance economic security and
improve our quality of life. Prior to 1988 NIST was known as the National Bureau of Standards (NBS).
NIST staff has been involved in research and standards development related to smoke detectors and
fire alarms for approximately four decades. We interact with the National Fire Protection Association
(NFPA), Underwriters Laboratories (UL), the U.S. Fire Administration (USFA), the Consumer Product
Safety Commission (CPSC), and other federal agencies on technical matters concerning fire sensing
and fire alarm technologies. The output of NIST research on smoke alarms is freely available for
downloading from the NIST web site (http://smokealarm.nist.gov/). Smoldering fires are inherently
different from flaming fires, and the operating principle for a photoelectric detector is distinct from
the operating principle of an ionization detector . Smoldering can occur only in a porous solid; for
example, polyurethane foam, shredded paper, or cotton. The rate of smoldering is limited by the
ability of air to penetrate the porous fuel, and hence the rate at which heat is released from a
smoldering fire is low, as is the rate of carbon monoxide (CO) production. The smoke layer from a
smoldering fire grows slowly and smoke can accumulate well below the ceiling, especially in rooms
other than the room of fire origin. 1/5 Another important aspect of a smoldering fire is that it
produces smoke particles that are relatively large (greater than one micrometer). Flaming fires can
occur in all types of fuels, including porous and non-porous solids, liquids, and gases. Air has easy
access to a flaming fire, which means that the rate at which heat is released is high. Smoke and fire
gases, including CO, can be produced at a higher rate in a flaming fire than in a smoldering fire. The
smoke layer is hotter and can build up near the ceiling of the room of fire origin as well as elsewhere
in a building. Flaming fires produce a very large number of smoke particles that can be relatively
small (less than one micrometer). The above discussion distinguishes between smoldering and
flaming fires; the following describes the different characteristics of photoelectric and ionization
detectors, and why they react the way they do in smoldering and flaming fires. The bottom line is
that each type of detector has its advantages and disadvantages. An ionization detector responds to
the movement through an electric field of ions produced by a small radioactive source in the smoke
alarm chamber. When smoke particles move into the chamber they inhibit the motion of the ions,
altering the electrical current. Fires that produce a large number of particles (i.e., flaming fires) are
sensed more easily with an ion detector. A photoelectric detector operates on the principle of light
scattered from the surface of particles. Because large particles have much more surface area than
small particles, a photoelectric detector is more sensitive to the large particles produced in a
smoldering fire. The sensitivity requirements of the UL standard are identical for both types of
detectors, and neither detector responds to carbon monoxide or heat. The general trends from a
2004 NIST experimental study into the behavior of smoke alarms are consistent with several
previous scientific studies showing that properly installed and maintained ionization and
photoelectric alarms provide enough time to save lives for most of the population under many fire
scenarios. However, ionization detectors have been shown to sometimes fail to alarm in a
smoldering fire even when visibility in the room is significantly degraded by smoke. Most
FOREIGN RRL

photoelectric detectors alarm substantially sooner in these situations. In the NIST experiments the
photoelectric detectors sensed smoldering fires on average 30 2/5 minutes earlier than the
ionization detectors. The same study demonstrated that ionization detectors responded, on average,
50 seconds earlier than photoelectric detectors during flaming fire experiments. The relative margins
of safety associated with a 30 minute earlier warning in a slow growing smoldering fire compared to
a 50 second earlier warning for a fast growing flaming fire is difficult to determine. In the 2004 NIST
study, average times to untenable conditions for flaming and smoldering furniture fires were found
to be shorter by 17% and 47%, respectively, than those found in the 1975 NBS study (also known as
the Indiana Dunes study). On the other hand, the average time for the cooking fires to reach
untenable conditions was 120% longer in the current study. Since the cooking materials were similar
in the two studies, NIST concluded—on the basis of the shorter time to untenable conditions in the
furniture fires—that a major factor in the increase in fire growth rate is due to differences in modern
furniture materials and construction compared to furniture manufactured four decades ago. The
reduction in time to untenable conditions from a fire (either flaming or smoldering) in modern,
synthetic materials indicates the need to determine the ability of standard test methods to ensure
safe performance of modern (and legacy) residential fire alarms. Both NIST and UL are involved in
research that will assess whether or not modifications are required in the standard to accommodate
the changing threat. An important conclusion from the 2004 NIST study was that the available safe
egress time provided by a smoke alarm would be sufficient, in many cases, only if households follow
the requirements in NFPA’s National Fire Alarm Code (NFPA 72) for new construction, which requires
the installation of fire alarms at more locations in order to improve audibility in bedrooms where
occupants sleep with the door closed, and to provide warning to the occupants of bedrooms with
closed doors when the fire starts in that bedroom. NFPA 72 also requires two ways out of a sleeping
room, one of which is generally a window. With the bedroom door closed there is more time in
which to use the window exit should the primary exit be blocked. Audibility of smoke alarms remains
an issue, particularly for sleeping children and adults impaired with alcohol or other drugs. For
existing residences that do not fall under the "new construction" requirements of NFPA 72, or for
new residences where the state or local building regulatory agency has not adopted the
requirements, the following approaches are suggested to 3/5 reduce problems associated with
inaudibility: placing smoke alarms in bedrooms, interconnecting smoke alarms, changing alarm
tones, and providing better home fire escape planning. Interconnection of the smoke alarms ensures
all alarms respond to a fire event. Nationally-collected data on fire incidents do not specifically
classify fire sources as smoldering and/or flaming; however, NFPA estimates that more than one-
fourth of home fire deaths involve fires with an extended period of initial smoldering. In NIST's
smoke alarm research, and in applications in the field, it is documented that most common
ionization detectors have a propensity to produce nuisance alarms during cooking activities. NIST
examined a broad range of activities (including cooking) that yield nuisance alarms. The published
field observations guided the nuisance alarm scenarios studied. Specifically, the sensitivity to alarm
threshold, distance from the source, background air flows, and alarm sensor (photoelectric or
ionization) were examined. Additional measurements were made with aerosol instrumentation to
provide a more fundamental understanding of nuisance alarm sources than has been previously
published. Given the scenarios examined, both photoelectric and ionization alarms produced
nuisance alarms, but NIST does not mean to imply that they are equally susceptible to such nuisance
alarms. Most field data suggest that ionization alarms have a greater propensity to nuisance alarm
than photoelectric alarms, possibly indicating that certain activities such as cooking dominate
FOREIGN RRL

reported nuisance alarms in the field. To reduce the impact of nuisance alarms, NFPA 72 requires
that smoke alarms not be located directly in cooking areas, and that any alarm located within 20 feet
of a cooking appliance be photoelectric. Studies have shown that this should be reasonably effective
except where some cooking techniques (blackening, deep frying, and broiling especially in a broiler
that is not clean) are employed. However, photoelectric detectors may also alarm to these more
egregious cooking styles. The second biggest nuisance alarm culprit is steam from showers where
both detector types are equally susceptible. In summary, the research conducted by NIST staff leads
to the conclusion that both ionization and photoelectric alarms provide enough time to save lives for
most of the population under many fire scenarios; however, ionization alarms may not always alarm
4/5 even when a room is filled with smoke from a smoldering fire, exposing the most sensitive
populations with mobility limitations to an undetermined risk. Photoelectric detectors can provide a
lot more warning time than ionization detectors in a smoldering fire; at the same time a smoldering
fires can take a longer period to become dangerous. Ionization detectors can provide a little more
time than photoelectric detectors in a flaming fire; in this case there can be little time to spare.
Changes in furnishing materials and construction over the past decades have reduced the time
available for safe egress in any fire. NIST is currently conducting research to assess whether or not
modifications may be needed in the standard test method for certifying residential smoke alarms to
accommodate the changing threat.