GROSS

IT AL ON METHOD

FOR MEASURING

SMOKE

167

D. Gross/

J. J. Loftus,2 and A. F. Robertson3

Method for Measuring Smoke from Burning Materials*

S.O.I.
to

T
V
CT

f-I.

Average smoke accumulation rate (see Appendix 2) Maximum rate of increase in specific optical density, per minute, measured over a 2-min period Smoke obscuration hazard index (see Appendix 2) . Time to reach a "critical" specific optical density (see Appendix 2) Light transmittance, per cent Volume of smoke chamber Attenuation coefficient in Bouguer's law Micron

REFERENCE: D. Gross, J. J. Loftus, and A. F. Robertson, "Method for l\lleasuring Smoke from Burning IVlaterials," Symposium on Fire Test j\.1ethods-Restraint .. & Smoke 1966, ASTM STP 422, Am. Soc. Testing Mats., 1967, p. 166 ABSTRACT: Based on a study of possible smoke-measuring methods. a laboratory test has been developed for the photometric measurement of smoke from burning materials. The method assumed the applicability of Bauguer's law to the attenuation of light by smoke, and smoke quantity is therefore reported in terms of optical density rather than light absorptance or transmittance. Optical density is the single measurement most characteristic of a "quantity of smoke" with regard to visual obscuration. Experiments have been performed on a variety of building finish materials under both flaming and nonflaming (smoldering) conditions, and the results are reported in terms of (a) maximum smoke accumulation, (b) maximum rate of smoke accumulation, and (c) the time period to reach an arbitrary "critical" smoke level. .: KEY 'VORDS: density smoke, combustion, pyrolysis, aerosols, fire tests, optical

Nomenclature
A
D Dm Ds

F
Fo

L
1/

,.

Surface area of specimen Optical density == 10glo(Fo/ F) Maximum value of specific optical density Specific optical density = D(V)/(AL) Transmitted light flux Incident light flux Proportionality constant Length of light path Particle number concentration Radius of particle

When accidental fires occur, the smoke generated is often considered to represent the major life danger to the building occupants. Although large concentrations of carbon monoxide and perhaps other toxic products may be present, the obscuration of vision by dense smoke often prevents the direct and logical escape of occupants, or rescue by firemen, during the few minutes available prior to spread of the fire, and the onset of lethal conditions. The smoke-limiting requirements in current building codes have been established in an attempt to regulate and to reduce the potential lightobscuration hazard from smoke generated by the interior building finish materials applied to walls, ceilings, and floors. These requirements are commonly based on the results from tests devised principally for measuring comparative surface flammability. Unfortunately the relationship between the results of such tests and the visual-obscuring qualities of smoke are not well established. In a recent study, an attempt was made to relate the "tunnel" smoke density ratings for a variety of building materials with visual and photoelectric observations in a room in which the smoke from the tunnel was collected [1].4 The immediate objectives of the present study are to investigate the suitability of a laboratory method to measure smoke quantity under prescribed and standardized exposure conditions and to evaluate the appropriate optical properties of smokes which obstruct human vision in building fires, without regard to the chemical nature of the smoke or the fundamental processes of its generation. Measurement of Smoke Aerosols are collections of small particles suspended in a gaseous medium, and include smoke, dust, fog, and haze. Smoke is generally considered to be the gaseous products of burning organic materials in which small solid and liquid particles are also dispersed. Another definition limits smoke to solid particles, such as carbon and ash, suspended in air. Under
j;cC

'" The work reported in this paper was supported by the Federal Housing Administration as part of a technical study of test methods and performance criteria for wall systems. I-a Physicist, chemist, and chief, respectively, Fire Research Section, Building Research Division, Institute for Applied Technology, National Bureau of Standards, \Vashington, D. C. Nfr. Gross is a personal member of ASTM. 166

'The paper.

italic numbers

in brackets

refer to the list of references

appended

to this

1_'
1

,ic

168

FIRE TEST METHODS

GROSS ET AL ON METHOD FOR MEASURING SMOKE

169

typical conditions of nonflaming thermal decomposition, however, it has been fairly well established that the nongaseous portion of wood smoke consists, not of carbon particles, but primarily of homogeneous spherical tarry droplets [2J, and of liquid droplets of organic substances of fairly high boiling points, above 100 C [3]. On the other hand, some materials produce smoke containing a large fraction of carbon particles. Test methods for measuring smoke are generally of two types: (I) those in which light transmittance measurements are made on the smoke aerosol directly, and (2) those in which the smoke particles are collected on a suitable filter paper which can be either weighed or measured for light transmittance. Of the first group, those employing unaided visual methods, such as the use of the Ringelmann chart, are generally not considered adequate for laboratory measurements. Instrumental visual methods include the smokescope, umbrascope, and others, in which visual comparisons of the unknown smoke are made relative to reference shades or gray glasses. Light scattering is a useful technique for detecting the presence of small quantities of smoke, particularly for particles which are not readily accessible or where particle size determination is desired [4J. Scattering measurements are generally limited to particles whose size is of the order of the wavelength of visible light. However, where the object is the measurement of smoke as it relates to visibility, the light attenuation method appears to be the most direct and practical approach. The most common method for smoke suspension measurements today employs a light source and a photoelectric cell, arranged so that the electrical output of the cell may be used as a measure of the attenuation of light by smoke. The readings are commonly expressed as per cent absorptance, or in terms of the Ringelmann scale (0 to 5). Attempts have been made to relate the per cent light transmittance to an optical density scale and to a quantitative smoke concentration, that is, mass per unit volume [2,5,6]. Such methods are considered generally valid and will be discussed in more detail later. There does not seem to be technical justification for integrating the area under a per cent light absorptance versus time curve[1J, [ASTM Test for Surface Burning Characteristics of Building Material (E 84 - 67)J or for making allowance for residue deposits on lenses at the end of a test by simple subtraction of light absorptance readmgs. In the second type of smoke measurement, a known volume of gas is filtered through a known area of filter paper, and the resultant spot is classified according to its degree of blackness, and commonly referred to as "smoke shade." It has been assumed by some that Bouguer's law could be applied to smoke shade spots, and that optical density, 10glO(l00IT), based on per cent light transmittance is an appropriate measure [7]. Others have shown a preference for light reflectance measurements of

smoke shade (ASTM Test for Smoke Density in the Flue Gases from Distillate Fuels (D 2156 - 65)). However, considering all types and colors of smoke, a direct relationship between such measurements and light attenuation measurements through disperse smoke aerosols is not obvious. Physical Optics of Light Transmission In formulating the requirements for a suitable laboratory smoke meter, primary consideration should be given to the optical properties of the smoke, but it is necessary also to understand the nature of the visible target, the amount and distribution of light, and certain psychophysical properties related to human vision (for example, contrast level, response time, and adaptation to levels of illumination). The problem is a very complex one, involving elements of both physics and psychology. Considerable aids to understanding various aspects of the problem are the very complete study of particulate clouds in a book by Green and Lane [8J, and the comprehensive survey of available knowledge on vision through the atmosphere given in a book by Middleton [9]. To reduce the problem to a level of possible solution, certain simplifications and assumptions must be made, since it is not possible at present to evaluate quantitatively complications due to eye irritations or respiratory effects, hysteria, and associated physiological and psychological factors. Considering only the limited problem of visibility through smoke, a criterion sometimes used is based on the assumption that when the "visibility" (preferably "visual range") of a hand-lamp illuminated sign drops to 4 ft, a room is smoke-logged to a degree that would seriously impede the escape of occupants [5]. It was inferred that this limit was reached when the light transmitted over a 4-ft. path was reduced to 0.25 per cent of the value in the absence of smoke. This density of smoke was later found to correspond to the perception of a lO-w lamp at a distance of 11 ft, under idealized conditions involving dark-adapted observers stationed outside a smoke-filled room.5 On the other hand, it has been reported that observers within a smoke-filled room became "apprehensive about personal safety" when smoke concentration reached levels corresponding to approximately 40 per cent and 80 per cent light transmittance at a distance of 10 ft, depending on whether or not self-contained breathing apparatus was used.6 These two different findings represent a very wide range in smoke concentration. To relate the visibility of an object to the obscuration caused by the presence of a "quantity of smoke" (or other particulate cloud), it is necessary to apply the appropriate laws of extinction and of reduction
5
G

G. Williams-Leir, See page 93.

private communication.

GROSS ET Al 170 FIRE TEST METHODS

ON METHOD FOR MEASURING SMOKE

171

in contrast. A "quantity of smoke" can be completely defined only when the total weight of material in suspension, its physical properties, and the state of its dispersion are completely known. Since it is usually impractical to determine all this information, even in the laboratory, it becomes necessary to substitute for it a single measurement which is as characteristic of the smoke as possible. The property most frequently selected for this purpose is its optical density; this defines the attenuation of a beam of light passed through the smoke [10]. If we consider a volume of smoke through which a parallel beam of light i; passed, then the law of extinction, properly known as Bouguer's law (sometimes also as Lambert's law or Beer's law) is given by:
F

Optical density is also proportional to the optical path length L, so that it is appropriate to write

AL

D=~y

D. may be termed the "specific optical density," and represents the optical density measured oVer unit path length within a chamber of unit volume produced from a specimen of unit surface area. Thus

D. = D L, = :L loglo ()

(4)

Foe-uL

........

(I)

where:
F"

= = rJ = L =
F

transmitted flux, incident flux, attenuation coefficient, and path length.

For a monodisperse aerosol, the attenuation coefficient is usually expressed as rJ = Kn'r217, where K is a proportionality constant, r is the radius of each particle and 17 is a particle number concentration, that is, the number of particles per unit volume. Optical density is defined as: D
==

F IoglO(Fa)

= 2.303 rJL

Klrr2

nL 2.303 .,.

(2)

This relation and Bouguer's law are strictly applicable only for monochromatic light for which the optical density is independent of the light intensity and the receiver sensitivity. However, it has been shown that the relation can be applied to light from a tungsten lamp passing through wood smoke by the use of an effective attenuation coefficient [2]. When two smoke clouds are compared by optical density measurements (reduced to a common path length), a comparison is thus being made of the attenuation coefficient, rJ, and is the very property required to determine visibility through smoke. Where visibility is of prime concern, it may be unnecessary to relate particle size distribution, refractive index, coagulation, or other particle properties to particle number concentration, even though they have a large effect on optical density. Throughout this paper, the terms "smoke accumulation," "smoke quantity," and "smoke density" refer to optical density levels, and not to mass or number concentration. If the smoke is produced from a surface of area A and collected in a closed chamber of volume V, the optical density of the generated smoke should be directly proportional to A and inversely proportional to V.

Smoke measurements expressed in terms of D. , which is dimensionless, have the advantage of representing a smoke density independent of box volume, specimen size or photometer path length, provided a can" sistent dimensional system is used. During the smoke accumulation process, therefore, the change of D. with time will ideally depend only upon the thickness of the specimen, its chemical and physical properties, and the exposure conditions. Since the ratio VIAL may be a fairly large number, the magnitudes of D. can be much greater than the optical density (D) values. In a closed chamber under prescribed exposure conditions, Dm is used to designate the maximum value attained by D . However, there are definite limitations to the uSe of specific optical density for extrapolation and comparison with other box volumes, specimen areas and photometric systems, and for extension to human visibility. The degree to which such extet1sions are valid depend upOn a number of major assumptions: I. The distribution of st110keis unifOrm. 2. The smoke generated is independent of the amount of (excess) air available, and of any "edge effects" due to thermal conduction, convection currents, or radiation. 3. For any given smoke, the optical density is a linear function of concentration. 4. Coagulation (agglomeration), deposition, and other changes in the smoke after its generation are similar regardless of the specimen size, or the size and shape of the chamber. S. Human and photometric vision through light-scattering smoke aerosols, expressed in terms of optical density, are similar. Experimental Methods Rohm and Haas Test Chamber The test apparatus used in the initial phase was developed by the Physics Laboratory of the Rohm and Haas Co. [11], and was made available by them for our use. The 12 by 12 by 3D-in. XP2 test chamber

I
172 FIRE TEST METHODS

j,
j

GROSS ET At ON METHOD FOR MEASURING SMOKE

173

\\'as instrumented with a 3-w light source, a barrier-layer photoelectric cell with visual eurreetiun lilter, and a meter to measure light transmittance horizontally across the 12-in, light beam path. In the prescribed test. a propane burner operating at a pressure of 40 psi and using air supplied frum outside the chamber, provided a flaming expusure to a I by I by II[ -in. specimcn placed on a supporting metal screcn. Thc chambcr was closed and unventilated during the 4-min test period exccpt for I-in. high ventilatiun openings around the bottom. A principal feature uf this test methud is that procedural changes and the exploration of variables can be made quickly amI ceonomically.
1.0.
I

j !
j

I ,

I
I ,
1
;,

t
J

!
1
J

NO IGNITION

--1

RAPID

IGNITION

I ,
J j

I-

00.8

o
"--

0:::

I ,

-0.6 >I(f)

W 0..

i i
j

~ 04
..J <J:

I- 0.2 0..

I
J

200

400
FURNACE
of

600
/Ilm/board

800
wil/z/nll

1000
/zclllcd

1200
fll/'l/II(,(,.

' jj
-j

TEMPERATURE,oC

FIG. I--.\II/<)/(,

f'}'(!dIlClioli

i
;
FIG. 2-Smoke lest c/zamber.

It became apparent, after considerable experimentation, that several changes in the equipment were desirable in order to permit (I) measurement of the maximum quantity of smoke accumulated, (2) surface exposure of the specimen rather than complete immersion, (3) pyrolytic as well as flaming exposure, and (4) a higher degree of reproducibility fur all types of materials. The critical importance of the type of exposure used was demonstrated by empluying an electrically heated furnace in place of the propane burner and perform ing a series of tests on unfinished hardboard specimens O\'er a wide range of temperatures. It is evident from Fig. I that smoke accumulation from pyrolytic decomposition was low at temperatures below 300 C but that peak smoking occurred when the temperature was approximately 380 C. At higher temperatures, self-ignition with t1aming combustion occurred, reducing the smoke accumulation. At very

high temperatures, the smoke accumulation again increased. Although not explored, a relation of similar form would be expected as a function of irradiance ]eve I on a single exposed surface of ce11ulosie materials. Thcse considerations led to the construction of a new chamber which was completely closed to prevent smoke loss, was of greater capacity to permit burning of larger size specimens, used a vertical rather than horizontal photometer beam to minimize measurement differences due to smoke stratification, and employed an electrically heated furnace and a sman auxiliary pilot flame for close control of irradianee on the exposed specimen surface.

174

FIRE TEST METHODS

GROSS ET AL ON METHOD FOR MEASURING SMOKE

175

NBS Smoke

Test Chamber

A completely closed cabinet, comprising a box 3 ft wide, 3 ft high, by 2 ft in depth was developed (see Fig. 2). The 3 by 3-in. specimen was arranged for support in a frame limiting thermal exposure to the surface of 2}i 6 by 2!X 6 in. size. Provision was made for two types of thermal exposure. One involved simple irradiance of the specimen with a flux of 2.5 w/cm2 yielding nonflaming pyrolysis (smoldering) conditions. The second involved a similar exposure but with the addition of a small pilot flame impinging on the lower portion of the specimen to initiate flaming combustion. A vertical photometer path over the full 3-ft box height was used. The photometer system incorporated a collimating lens at the source and a small aperture diaphragm at the focal plane of the receiving lens. This effectively limited the field of view of the phototube to the photometer beam itself and reduced interference from outside illumination. A 1P39 vacuum phototube, with an S-4spectral response, was connected in series with a 67.5-v battery and a suitable resistor including a shunt for the potentiometer input to provide a signal of at least 1 v with full illumination. A potentiometric type recorder with four decades of sensitivity increase was used for continuous measurement of light transmission, ranging to below 0.0 I per cent transmission, corresponding to an optical density of about 4. In the most sensitive range it was necessary to make corrections for the dark current of the phototube. The 30-w light source used was operated from a constant voltage transformer at a color temperature of about 2400 K. Further details on the construction of the equipment and test procedure used are included in the Appendix. In some of the work reported, specimens were burned in a box of approximately 250 ft3 volume. This box was used primarily for the purpose of confirming the validity of optical density as a quantitative means for measuring smoke. In this case, the photometer made use of a 5-ft path length.

terior finishes within buildings, it was considered desirable to reduce, so far as possible, the release of smoke from the rear of the specimen. This was accomplished by the use of an aluminum foil wrapper and an asbestos board backing as described in the Appendix. Maintaining a constant specimen irradiance level, the rate of smoke production was found to depend upon the spacing between the front face of the furnace and the specjmen. The reason for this has not been clearly identified, but appears to be associated with adequate space for free convection of air past the specimen during pyrolysis. The spacing selected for standardizing the test procedure was 1 Y2 in. The volume of the smoke cabinet was chosen to provide adequate air for complete combustion of the specimen. The volume selected provides
TABLE I-Results
Test

oj replicate tests on hardboard, nonjlaming exposure.

Time to = 16, min

1 2.

4. 5 .. 6. 7. 8 .. 9. 10. Mean. Standard deviation. Coefficient of variation,

556 618 678 645 595 532 506 579 618 507 583 58

68 52 50 46 49 42 46 54 65 51 52 8.3

5.10 5.50 5.29 5.34 5.54 6.65 5.94 6.14 5.70 6.51 5.77

10%
optical density increase

16%
per min, based

0.53 9.2%
on a 2-min time

R", is the maximum

period.

Experimental Results
Prior to performance of extensive tests with the smoke chamber, it was necessary to explore the effects of many procedural variables on the results obtained. These experiments will not be discussed in detail, but those considered most important will be mentioned briefly. Standardization of Specimen Exposure

Significant variations in experimental results were found to occur depending on the way in which the specimen was backed. This was caused by the flow of smoke through the specimen and its escape from the rear surface. Since the method was expected to be used principally for in-

an air-fuel ratio, on a weight basis, of over 20: 1 based on the combustion of a yt-in. thick hardboard specimen of unity specific gravity. This is about four times the air requirement for complete oxidation of the fuel. Measurements were made of CO, CO2, and O2 following both flaming and smoldering combustion of specimens in the cabinet. Typical results for smoldering and open-flaming combustion, respectively, of hardboard showed: O2 content 19 and 17 per cent, CO content 0.4 and 0.5 per cent, and CO2 content 0.45 and 2.5 per cent. Although in no case was the specimen completely consumed, it appeared that extinguishment of open flaming during such tests resulted primarily from fuel depletion at the burning layer and thermal shielding by char formation rather than lack of available oxygen. Further studies to definitely establish this point appear desirable before the results are applied to situations involving considerably higher ratios of volume to exposed surface area.

l
_~iiiiiIIii-"' _

JJ,j

ii'

176

FIRE TEST METHODS

GROSS ET AL ON METHOD FOR MEASURING SMOKE

177

A few measurements confirmed the fact that significant variations in smoke production were caused by changes in moisture content of the specimen. For the sake of standardizing experimental conditions, it was decided to condition dried specimens to equilibrium with air at 73 F and 50 per cent relative humidity. Initial experiments with ignited specimens were performed with short duration exposure (Yi to 2 min) to the pilot flame. Considerable variability in the time of cessation of flaming of the specimens was noted, resulting in large differences in smoke production. It was decided, therefore, that the pilot flame should be applied continuously during the exposure period for specimens tested in the flaming combustion condition.
40
(/) W :J:

variation, only very slight changes in smoke measurements were observed. However, for purposes of standardization, the IP39 vacuum phototube having an S-4 surface and low-loss base was used. Care was taken to select a tube with a very low dark current. Smoke Stratification A few measurements were made to evaluate the magnitude of smoke stratification. This was accomplished by arranging a horizontal smoke photometer assembly so that it could be slowly raised or lowered inside

'0 c..-HARDBOARD POLYSTYRENE NON-FLAMING FLAMING I

"
>-

600
AREA

100 IN2 V0 6.6CHAMBER 250 6.6 250 SYMBOL 250 18 IB 6.6 18 VOLUME FT3

SPECIMEN

U
w

Z 30

I
I

500
2 MIN.

~IMIN.
I

,....-I~MIN.
I

l(/)

u z o ..J u
Iex: I I

I I I I

~ 20
(/)

/ ,
I I

a ..J u i= 300 a.. o

~400

PLYWOOD NON-FLAMING

>
1.0

w
a.. (/)

U 200

FIG. 3-Verlical
fl{lI11ing.

distribution of smoke, fiberboard, nonfiaming-polystyrene,

Test reproducibility was investigated by performing replicate nonflaming tests on a selected hardboard material over a period of four months. The test results are summarized in Table 1 and indicate that even very high optical density values can be reproduced with reasonable precIsion. Spectral Response of Photometer

10
TIME, MINUTES

20

30

FIG. 4-EfJect

of specimen area and chamber volume on smoke buildup.

Experiments were performed to determine the degree to which smoke measurements would be changed by variations in the spectral range of the photometer. These were performed using the same type of incandescent tungsten light source and with phototubes of three spectral sensitivity ranges: vacuum phototubes with S-1 and S-4 spectral response, and a barrier-layer photocell corrected to approximate the average human eye. For an incandescent lamp operating at a color temperature of 2870 K, the wavelengths corresponding to the peak responses were at approximately 8300, 5100, and 5800 A, respectively. Within this range of

the smoke chamber and light transmittance readings were taken during two tests. As shown in Fig. 3, stratification was quite significant during the first few minutes of test. Although not shown, greater mixing and a close approach to uniform smoke distribution was noted as the smoke quantity increased. Scattering Effects Although the photometer optics used were selected to provide freedom from off-axis illumination, it was considered possible that errors in smoke measurement might still be present due to scattering from smoke surrounding the direct beam path. This was examined in two ways: (a) by employing a beam-limiting aperture which permitted only the direct

1
178 FIRE TEST METHODS GROSS ET AL ON METHOD FOR MEASURING SMOKE 179

photometer beam to pass to the detection unit, and (b) by employing a shutter of such size as to interrupt only the direct photometer beam. These tests were performed with the box filled with smoke produced from smoldering cellulosic material. In both cases, the decrease in optical density, attributable to scattered light, was less than 1 per cent.
Size and Volt/me Effects

1
!

Early work had verified the general validity of reporting smoke accumulation in terms of optical density. It now appeared likely that the specific optical density of smoke produced from specimens of different sizes, and in boxes of differing volumes, should show good agreement as a function of time until smoke accumulation stops or photometer dark current limits measurement. As shown in Fig. 4, there is fair agreement in the specific optical density curves even for a volume ratio of almost 14 and, in one case, a surface area ratio of 15. These results, although limited in number, tend to confirm the validity of the measurement
TABLE 2-Smoke measurements on clear spruce of various thickness.
min-t Exposure Nonllaming Exposure

maximum smoke produced increased with thickness, but not in proportion to thickness above YI in. In some tests, the maximum quantity of smoke may yield optical densities within the chamber which may exceed the measurement capability of the photometric system. The existence of such limitation may be verified by noting the extent to which an approach has been made to the phototube "dark current," which is the reading obtained when the light source is totally shielded or turned off. By mixing the smoke into an ad700

600

o r-500 I(/) Vi

Z
<X

310 Maximum Rate, Rm ~laximum Rate, Maximum 45 21 147 4.8 16, min 4.2 16,to I/)1/1 4.3 39 36 19 18 6275 378 .....5.1Smoke, liS Flaming 5.5 421 to 145 4.6 14 Time Time Ds49 Maximum =

w
...J

0400

Ig, 300

u
4Q. (/)

FIBERBOARD

~ 200

method used and the usefulness of the concept of specific optical den. sity as a means for quantitative measurement of smoke under specific burning conditions. When specimens of YI-in.-thick hardboard of various widths (YI, Y2, 1, and 20/16 in.) were tested under nonflaming exposure, a fairly good agreement was obtained in the specific optical density curves for the 1 in. and full-size specimens. However, the D. curves for specimens of smaller width, YI in. and Y2 in., were noticeably different, due to thermal edge effects. Smoke is produced by thermal degradation or burning and is generally considered to result from reactions at the specimen surface. As heat penetrates into the specimen, products of combustion, including condensible vapors, also originate from various depths, the quantity depending mainly upon diffusion, ablation, and char formation effects. To explore this effect, measurements were made of the smoke produced by various thicknesses of matched clear spruce specimens, and the results are shown in Table 2. For this range of thickness, smoke accumulated at a fairly constant rate (slightly faster for the thinner sections). The

30
FIG. 5-Buildup

40

50

60

70

80

TIME. MINUTES

and disappearance of smoke, nonflaming exposure.

joining chamber of sufficiently large volume, the change in optical density resulting from the dilution (which is inversely proportional to the volume increase) will be measurable by the photometer in the smoke chamber. The procedure for accomplishing this measurement is described in the Appendix, and permits determination of the peak specific optical density (maximum smoke accumulated) on a large majority of heavy smoke-producing materials. A smoke aerosol within a confined space gradua!!y disappears due to coagulation, to settling, and to condensation on the chamber wa!!s. The rate of disappearance depends upon the relative motion of the partic!es

180

FIRE TEST METHODS

GROSS ET AL ON METHOD FOR MEASURING SMOKE

181

600

as well as upon concentration and particle size. Typical curves for smoke buildup and disappearance are given in Fig. 5, from which it may be noted that the rate of disappearance is greater for higher concentrations. Influence of Ambient Atmosphere

500
>-

t(/) Z400
w

a
--I

~
~300 tel.

It appeared likely that changes in the oxygen concentration of the ambient atmosphere of the specimen would influence smoke production significantly. Accordingly, tests were performed with fiberboard specimens pyrolyzed in atmospheres diluted with nitrogen and the results are
300

o u ~200 u w
el.

(/)

>-

!=
(/)

I2 Z w w

~ 200 4 FIG. 6--Smoke

u a::
a.

8
oxygen

12
atmosphere,

16

20
fiberboard, nonflaming

TIME,MINUTES

o.

U t-

--I

5 w.

u Z
~
:::E

10 lt(/)

~ 100 300

U w a.
(/)

20 Z t50

a::

t(/) Z

>-

~UNFINISHED NON-FLAMING

t2 ~ w 5

w 200

o
--I

u a::
a.
exposure.

01 o
FIG. 8-Effect

/-r4
of surface

'ALKYD

PAINT

8
coatings

12
TIME,MINUTES on smoke

16
buildup

20
for plywood,

1100

~ U
to.

o u
~ 100

w U Z ~
:::E

nonflaming

10 ~ (/)

w
(/)

a.

20~

a::

t50

shown in Fig. 6. It is evident that when smoke is produced from fiberboard by nonflaming pyrolysis, the quantity generated as measured by photometric means is an increasing function of the oxygen concentration of the ambient atmosphere. Similar studies for flaming combustion were not made. Smoke Production from Typical Materials

o o

."

4
TIME, for a typical

6
MINUTES cellulose

8
fiberboard.

10

[JIOO

12
of (a) expo-

FIG. 7-Smoke buildup sure and (b) paint coating.

Effects

In Fig. 7 are shown smoke production curves for the finished (front) and unfinished (rear) surfaces of a typical cellulose fiberboard. These tests were performed under both flaming and nonflaming exposure conditions, and illustrate the wide range of smoke levels possible from a

min-I

Thickness,

--

E5.0 -'" 310 ~"" 371 ~E 146 ~~ 112 231 0.7 3.7 Rill. 56 28.. ~QE 26: 25: 28' 59. 18, 7' 3' 2 I(j( 71 .... .... Nominal 4.7.3 3920 5.6 7.0 9.4 35: 12. 15 3.8 89 6 NR'25 .- c: 0.250 '.0.070 :- " '.-, Flaming ,orExposure 0.75 9.314 4673I 34Max. ... ..... ,,' ...... '.' .. 0.0030.25 4.310 4079 4.21.4 3327 4.8.0 3489 9.7 -""oJ 5113 10.3 6.5.7 21 1.7 . ..4.5 1656 '"0.004 '.'-,. ........ 0.0040.18 4.4 4.2 4.7 4.8 4.3 3.2 3.9 0.8 0.. 16 11 4 4.6 335 4.0 125 .. a;U 2 ... 20: 15 9 3 62 18 4 63 80 34 33 38 ............... ...... 0.035 .50 ......... 9.6 .4 2346: 4.9 4.3 141 ...27 3355 .~'~.g35( .3 .. 331 ............. 0.055 .25 0.011 .75 Rate ~~ 132 Actual, in. Asbestos-Cement board .........

42 67 . 0.25 29: 34: 48( 41: 2H84 25 32: 108 81 36 56 209 155 138 64 75

- --

0.25 .50 0.25 0.25 .375 60~ + +

TABLE 3-SlImll1ary
.75
" " " "

of test results."

23 92 15 24 36 40 17 44 54 Max. ::-lonflaming min-t "" ~Rm. 2 Exposure 22 50Rate, 60 21 26 61 34 9 4950 63 0

--

24 15 37

o""-~.'.E45 to .. 0.4 0.0 0 . ASTM Smoke .;:PM 0,>5.0.80 =:~ 4.6 4.8 2.6 "" 3.7 ..Density 3.1 7.7 5.1 3.7 3.4 NR'.."~E152- Smoke MethodE 162 5.4 2195o ... 0.7 3.8 1.8 5.5 1.7 3.3 2.7 4.1 ... '" 1.2 ... 5.1 ..184 E 8110 8.7 8772c:: to 3.3 11II >5OD223 4.1 160 4.8 0.6 2.7 4.6 3.3 4.5 0.8 2.6 2.4 5.0 3.9 6.7 1.4 4.3 5.0 3.2 3.0 5.0 0.79 0.33 0.45 107 0.2 0.1 0.0 1.9 1.3 0.3 6.0 ~E ,~'3 ASTM Method Smoke Chambc Rating t Dcposi

"

--.. .. ---~"" .... 60~

Ib/fl' jlb/ft'

-n

;;;
m -<
C/>

m
-<

:;:
m
-<

:r
'" C/>

7.8269 ..71 2.1 170 300 4.6 0.8 1.2 0.1 4 0.6 11.49 93 24 1.3 43 17 NR'1400 '"... 7.3'".9 174 ...434 16 117 0.35 0.512 0.83.. 0.8159 80 ..6082 ... 5.0110 41 0.6198 75 2.6 18.8 2.0 2.4 21134 65 0.2 456 0.3 1.4 261 2.8 320 156 12 .. 4.0 .5.8 4.8 2.5 4.5 4.3 .45 16 49 ... 29 84 1.8 76 60 50 29 47 .. 22 8 90 88 0.9178 87 3.9 7.3 2.7 361 215 1.1 4.7 4.2 19 12 2 6 4.8... .79 618 0.42 5.3243 ... ...... 372 0.3 0.5 20.. 1.7 7.1 0.5 0.9 4.9 1.2 9.2 0.1 2.3 319 105 1.5 3.8 ... 4.4 4.9 3.1 3.7 3.6 4.1 4.0 42 2 3 1.6 .23 35 17 33 .54 66 "... . ' ....... ........ .. "" .. '. '., Floor Coverillgs: Red ................ tardan 1. ........................ Acrylic ................................ " ...... " . Wool. oak................
"

...32 0.064 0.076 0.25 . . 0.50 . . 0.16 236 350 304 660 20 ... ..... 49 218 107 100 .. >660 395

.... ... ... ... ...

0.25 0.78

.. 0.125

0.156 0.75 0.5 0.25 0.18 0.3 0.046 0.156 0.219 1.0 0.094 0.026 0.024

C1

'"

o
C/> C/>

o z
:;: m
-<

o
'"

:r
-n

o '" ~
C/>

:;: C

'"

Z
C1

" Results
b

apply

to listed thickness. before and after dilution (volume ratio

C/>

Ratio of optical densities drift, where appropriate. , Not reached.

1:4).

Values

corrected

for smoke

deposit

on window

and zero

:;: o '"
m

co

'"

184

FIRE TEST METHODS

GROSS ET At ON METHOD FOR MEASURING SMOKE

185

single interior wallboard, the front surface of which is prefinished with a factory-applied paint coat. Figure 8 further illustrates the effect of surface coatings in modifying the smoke production of an exterior grade plywood base material, y,j-in. thick. It is evident that thin surface coatings (in the thickness range 0.003 to 0.035 in.) can have a very large effect on the maximum smoke generated and on its rate of release. Table 3 summarizes flaming and smoldering test results for a wide variety of typical interior finish materials in terms of the maximum smoke accumulation, D1II , the maximum rate of smoke accumulation (averaged

during dilution was partially accounted for as suggested in the Appendix. It might be noted that deposition, condensation, or other loss of suspended smoke particles upon circulation into the large cooler chamber, would result in high values for the optical density ratio. Values of the optical density ratio considerably beJow 4 were noted for smokes of very low optical density, 0.15 (per foot) and less. Thus, the general validity of the logarithmic extinction relation, Eq 2, appears reasonably well established for a wide range of smoke types and levels.
100 om
lLJ '"

o
~ ~"O 00 0 .0 0
00 o
'"

50

::!: (/)

u. x

o
x
>(/) V

ox ~o0 00

toleQ
""

u ,..

in C 3
"" ..J

t-

0 Il. a: W Z t-

0o<f' 0x 2 000 0 oJ' o "::o

0 8 0 ''''' o 0'(0 )(. 00

00

"Ox

0
00

x
o

!:: 20

00
o

~
dO

CHAMBER VOLUME RATIO V. x

x.

o
lLJ

x 0 x

4 U
i=
g,

-I 10

U
j;:

U
lLJ

10

20

50

100

200

500

1000

GEOMETRICAL
00 OPTICAL

FACTOR AVL

0.5
DENSITY,

I
DO' PRIOR TO DILUTION,( PER FOOT)

I.~

FIG. 10-Specific optical density versus geometrical factor for five selected light transmittance values.

x Cellulosic materials noncellulosic materials FIG. 9-0ptical density ratio in smoke dilution tests.

Discussion For many of the materials tested, a very uniform rate of smoke buildup was obtained. This is evidenced by the almost linear slopes of specific optical density versus time curves as seen in Figs. 4 to 8. The smoke-generating process often terminated fairly abruptly, with a rapid approach to the final reading, representing maximum accumulated smoke. During and after a typical test, smoke disappeared (due to settling and coagulation), the rate being greater at higher concentrations, see Fig. 5. No adjustments were made to the measurements for these losses, since it may be presumed that a similar, though generally not identical, process of smoke buildup (generation less disappearance) will occur in a room or chamber of other size. Where visibility through smoke is of interest, it is not necessary to convert optical measurements into concentration units, either by weight

over a 2-min time period), R"" and the time period to reach a specific optical density of 16. For tests in which the maximum smoke accumulation in the test chamber approached the "dark current" limitation of the phototube, the smoke was diluted by mixing into an adjoining chamber of 54-ft;; volume. Care was taken to ensure that the peak smoke level had been achieved prior to making a reading. In other tests, the dilution process was also accomplished, and this permitted application of the inverse relation between concentration and volume, and, in fact, provided a generaJ verification of it as shown in Fig. 9. Here, the ratio of the optical density for the initial condition (test chamber volume only = 18 ft3) to the optical density for the final condition (total volume of both chambers = 72 ft3) is plotted a~ a function of the opticaJ density of the smoke prior to dilution. The condensation and seWing of smoke

jIi

186

FIRE TEST METHODS

GROSS ET AL ON METHOD FOR MEASURING SMOKE

187

ar number af particles. On the cantrary, the decrease in light intensity in

terms af aptical density is the precise type of attenuation coefficient desired. As previously mentioned, the terms maximum smake quantity, and smake density shauld be understaad to. refer to. optical density levels, and not to. mass ar number concentratian. With respect to defining the hazard when a persan is trapped in a roam where smoke is being generated or accumulated, three aspects were considered important: 1. Maximum Smoke Accumulation represents the maximum accumulated quantity from a given area af material under the prescribed expasure >-w W z 0 :>:: 10.0. z <t .... (maximum specific aptical density). 2. Maximum Smoke Accumulation Rate represents haw quickly the smoke level increased as averaged aver a 2-min periad. 3. Time to Reach a Prescribed Smoke Density represents the time periad prior to. attaining a critical smoke level. The critical optical density thraugh which an exit sign may still be visible can vary aver a wide range depending an characteristics of the light saurce, the general illuminati an level, and the degree af irritatian to., and dark adaptatian af the abserver's eyes. Assuming a uniformly distributed smake, the relatian between Ds and the geametrical factor V / AL is shawn in Fig. 10 far five values af light transmittance ranging from 80 to. 2.5 per cent, carrespanding to. aptical densities of 0.1 and 1.6, respectively. Far a selected critical transmittance (or optical density), the limiting value af specific aptical density Ds can be determined far the viewing distance L, the exposed surface area af the specimen A and the valume af the chamber V. Far camparative purpases, it was assumed that a light transmittance af 16 per cent over the viewing distance wauld be critical. The limiting value af Ds for a 12.5 by 20 by 8-ft roam, a viewing distance af 10ft, and a surface area af lOW wauld be 16. The times to. reach this arbitrary level of smake far the materials tested have been included in Table 3 tagether with maximum smake accumulatian and maximum smake accumulatian rate. It may be nated that the abserved times to. reach this level ranged fram as little as 0.1 min to. aver 10 min. For a given expasure an a building finish material, measurement of the maximum smake accumulation, the maximum smoke accumulation rate, and the time to. reach a prescribed smake density, may be abtained fram a single test, and it is tempting to. devise a farmula far cambining these properties into. a single classificatian index. However, further study seems desirable in arder to. estimate the relative weighting af these factors in terms af the smake accumulatians and mavements resulting from natural or forced ventilation, and typical apenings in roams and buildings. For selected cellulasic, plastic and campasite materials, the tatal
::::> 1 .....1

smoke praduced under bath flaming and nanflaming exposure was campared with the gravimetric smake depasit measurements from a standardized flame-spread test (ASTM Methad E 162 - 67) and with the smake density rating from the standard "tunnel" test (ASTM Meth<?d E 84 67). The data far matched specimens are listed in Table 3. Far a particular smake, it is reasonable to. expect a carrelatian between the optical density of the smoke aerasal and gravimetric smake cancentratian (mass per unit valume); such a relatian has been demanstrated, for example, for smake generated by burning aakwaad sawdust at ele-

c.:I

0 08 0 30.0..0 I W <I Z 20.0. 00 en ::!: .... a: V1 Z w .... Vi

0 00 L.

0 0 0

o.~C>
0.

20.0. 30.0. 40.0. 50.0. 60.0. 70.0.

MAXIMUM SMOKE ACCUMULATED, Dm Flaming exposure 0 NonfJaming exposure FIG. II-Comparison of smoke data by two test methods

vated temperatures [2] and far smoke generated in the burning of coal [6]. However, using materials of widely varying chemical compositian and physical structure, and under both flaming and pyrolytic exposure, the type of smake generated wauld be expected to. range from organic fL) liquid droplets af small size (,.....,0.2 to 'solid carbon particles of large size (> 1 fL) and irregular shape. For this range af smake types and densities, it is nat reasonable to expect any simple or general carrelatian between optical density af the smake aeroso.l and weight af the deposited smoke layer. Where light transmissian measurements are made directly on the smake aerosal, as in Method E 84, a better correlation cauld be expected. Such a comparison is shown in Fig. 11, using bath flaming and

I',
i'
~'

188

FIRE TEST METHODS

GROSS ET Al ON METHOD FOR MEASURING SMOKE

189

non flaming smoke chamber test results. While a fair correlation was obtained for the smoldering tests, no satisfactory correlation was noted for the flaming tests. For these materials, the maximum smoke accumulation values for the flaming tests were always less than those for the smoldering tests, although the differences were considerably less for the noncellulosic materials. However, since the "smoke density rating" from the "tunnel" test is based on the integrated light absorptance versus time relation (rather than optical density versus time), the results, especially at high smoke levels, are not strictly comparable. In any event, lack of close correlation between the smoke chamber test and either of the preceding flame spread tests, results primarily from the fact that the smoke produced during a flame spread test is invariably generated under a complex combination of strong and weak flaming, and high and low temperature pyrolysis. The controlled smoke chamber test is designed to permit the separation of smoke generated by flaming versus nonflaming exposure, and to evaluate more readily the effects of a wide variety of test conditions for both the specimen (for example, surface coatings, treatments, moisture content) and its surroundings (for example, ventilation, oxygen concentration). That such evaluation is necessary is apparent from the widely different results obtained under flaming versus smoldering exposure of cellulosic and plastic materials. Ideally, for a selected exposure in the test chamber, a single test permits extrapolation to surface areas and to chamber volumes of other size. However, since the rate of smoke disappearance due to settling and coagulation depends upon mass concentration, the results will not always be identical in chambers of different size (different values of V / A), and this is a limiting factor particularly in using the measured value of D", as the true total smoke with confidence. Thermal edge effects must also be considered in any extension to other scale sizes. In addition, further studies may be necessary to establish the extent to which flaming combustion in the smoke chamber can be taken to represent burning under excess air conditions typical of very large volumes. ~ Finally, it should be clearly noted that conditions in the smoke test chamber are obviously not identical to the situation involving the human eye, seeking a light source in a smoke-filled room. In an actual situation, the light source is not (normally) concentrated in a parallel vertical beam, and the human eye has a much wider viewing angle and a slightly different spectral response from the lens/aperture/phototube combination used. In addition, certain important properties related to human visioncontrast level, response time, and adaptation to levels of iIluminationhave not been considered in this investigation. Additional studies relating to human vision through smokes are obviously desirable.

Summary

Based on a study of possible smoke-measuring methods, a laboratory test has been developed for the photometric measurement of smoke from burning materials. The test utilizes a closed chamber of 18 fta volume containing an electrically heated furnace which provides an irradiance of 2.5 w/cm2 (2.2 Btu/sec ft2) on the surface of a nominal 3-in. square specimen. Features of the smoke chamber include the following: (a) each specimen is exposed to the same controlled radiant exposure, and generates smoke primarily from the exposed surface, (b) the results apply to either flaming or smoldering conditions, (c) the photometer light path is vertical to reduce measurement errors resulting from smoke stratification, and (el) the chamber is sufficiently large to reduce effects resulting from premature consumption of air. Details of the test equipment are included to permit duplication by others. Results are presented which illustrate the wide variation in smoke production to be expected (a) between materials, and (b) for the same material under different thermal exposure and reduced oxygen concentration. The effects of stratification, scattering, and photometer spectral response on the measurement were briefly studied, and typical results were obtained to illustrate the relationship between chamber volume, optical path length, and exposed surface area of specimen. The method assumes the applicability of Bouguer's law to the attenuation of light by smoke, and smoke quantity is therefore reported in terms of optical density rather than light absorptance. Optical density is the single measurement most characteristic of a "quantity of smoke" with regard to visual obscuration. To take into account the optical path length L, the volume of the chamber V, and the surface area of material producing smoke A, a specific optical density is defined as Ds = V /LA [log1o(100/T)], where T is the per cent light transmittance. Thus, for a selected exposure in the test chamber, and within the limitations discussed, a single test permits rough extrapolation to surface areas and to chamber volumes of other size. Experiments have been performed on a variety of building finish materials under both flaming and nonflaming (smoldering) conditions, and the results are reported in terms of: (a) total maximum smoke accumulation, (b) maximum rate of smoke accumulation over a 2-min. period, and (c) the time period to reach a "critical" specific optical density of 16 under the test condition. This study was concerned with the limited problem of measuring the optical density of smoke as it relates to the obscuration of human vision in building fires. No detailed attempt was made to measure and analyze the fire-generated toxic combustion products which are likely to be the major and direct danger to life.

i
~

190

FIRE TEST METHODS

GROSS ET At ON METHODS FOR MEASURING SMOKE

191

Acknowledgments
The authors wish to thank A.

J.

Bartosic,

and the Rohm and Haas Co. the

for the loan of their smoke chamber, and to gratefully acknowledge financial support provided by the Federal Housing Administration.

APPENDIX 1
Smoke Test Equipment C flamber The smoke chamber consisted of a 16 gage sheet metal box, 3 ft by 2 ft by 3 ft high. As shown in Fig. 12, openings were provided to accommodate a photometer (C and l) with a 3-ft vertical light path, power and signal lead wires, air and gas supply tubes, an exhaust blower (B) and damper, an aluminum foil safety blowout panel (D), and a hinged door with a window (E). The chamber was tightly closed and usually not ventilated during test. It was supported on an angle-iron frame (L), on which were mounted the electric (N, 0, and P), and gas and air controls (S and Q). A multi range meter and timer, or alternately, a recorder, were used for taking data. The interior of the chamber and all parts used therein were either anodized black or painted with a flat black paint resistant to corrosive decomposition products. Furnace and Control System

To provide uniform irradiance on the surface of the nominal 3 in. by 3-in.-square specimen, an electrically powered furnace with a 3-in.-diameter opening was used. As shown in Fig. 13, a 525-w heating element' (D) was mounted within a 3-in. inside diameter by 3%-in. outside diameter by 1VB-in. long ceramic tube (C), bored out to 3Jh2-in. inside diameter, VB-in. deep to accommodate the heating element. Behind the heating element were mounted a JIiG-in.-thick asbestos paper gasket (F), three YrG-in. stainless steel spacing washers (G), and two .!2-in. stainless steel reflectors (H) and (/). The heating element assembly was centered with respect to the front %-in. asbestos board (B) and the %-in. asbestos board centering disk (1) by means of a 6-32 stainless steel screw (E), and the adjustment of the nuts on the end of this screw provided the proper spacing of the furnace components. Pyrex glass wool (W) is used to fill in the spaces in the heating element assembly. Two spacing rings (K) of %-in.-thick asbestos board, a rear cover (L) of %-in.-thick asbestos board, and a 4-in. outside diameter by O.OS3-in. wall by 4Ys-in. long stainless
7 A Silex Percolator Element (for example, Silex-Bloomfield Type EK6, or Eagle Electric, Cat. No. 385) has been found convenient for this purpose, if the ceramic projection shoulder and the steel connector prongs, which extend out beyond the base diameter, are removed. Certain commercial equipment and materials are identified in this paper in order to adequately specify the experimental procedure. In no case does such identification imply recommendation or endorsement by the National Bureau of Standards, nor does it imply that the materials or equipment identified is necessarily the best available for the purpose.

A-Chamber B-':'Exhaust blower C-Photometer light source D-Blowout panel E-Hinged door with window F-Air pressure gage G-Gas flowmeter H-Blower and damper lever I-Photometer J-Pilot burner lever FIG. 12-Smoke

K-Service openings L-Support frame M-Temperature controIJer N-Main power switch O-Internallight switch P-Autotransformers Q-Gas, air shut-off valves R-Electric ignitor switch S-Gas, air control valves

chamber assembly.

steel welded sanitary tube (A), ISO-grit polished inside and outside, completed the furnace assembly. Three sheet metal screws (M), No. 6 by 112in., were used around the periphery at each end. Appropriate holes were provided in the centering disk for asbestos-covered copper lead wires to the heating element and in the rear cover for a motor base plug and six Y2-in. ventilation

192

FIRE TEST METHODS GROSS ET Al ON METHOD FOR MEASURING SMOKE

193

holes on a 2Y2-in.-diameter circle. The furnace was supported on a stand shown in Fig. 14. Although the materials and dimensions used are given in detail, they are not considered critical, provided the construction can withstand continuous operation and provided the geometry of the furnace opening is not materially altered. As shown in Fig. 12, the control system consisted of a temperature controller (M), two autotransformers (P), and a sensing thermocouple placed

BLC film-viewer lamp powered by a voltage regulating transformer. Two adjustable resistors in series provide coarse (100 ohms, 25 w) and fine (10 ohms, 25 w) adjustment for zeroing, and reduced the frequency of lamp replacement. The light source was mounted in a box (C, Fig. 12) extending above the top of the smoke chamber, which contained the source and a sevendiopter collimating lens (see Fig. 15). A glass window, gasketed for smoke tightness, was mounted permanently in the ceiling of the chamber. Another box (I, Fig. 12) containing the photometer was located directly below the source and attached to the bottom of the smoke chamber. Below a similarly mounted glass window in the chamber floor were, in order, a sevendiopter lens forming an image of the source, a iJ.6-in. iameter circular stop d

t'
5" 0.0.
x

20 TAPPED HOLES SPACED

3 EQUALLY WALL STEEL TUBE 2" LONG

If' 0.0. SPACING STOPS ~"THICK WITH lac 32 SET SCREWS

["x ["x {

STEEL ANGLE FRAME

7
A-Stainless B-Asbestos C-Ceramic D-Heating E-Stainless F-Asbestos G-Stainless steel tube H, I-Stainless steel reflectors board J-Asbestos board tube K-Asbestos board rings L-Asbestos board cover element, 525 w steel screw M-Sheet metal screws paper gasket W-pyrex glass wool steel spacing washers (3) FIG. 13-Furnace section.
FIG. 14-Furnace

18'~/8'
support.

within and close to the surface of the ceramic core of the furnace opening. The temperature set point of the controller was arranged so that a radiometer, placed at the same location as the specimen, measured the prescribed irradiance level. The two autotransformers provided high and low voltage levels (rather than on and off), and were adjusted to minimize power fluctuations to the heating element.
Photometric System

in the focal plane of the lens and a IP39 single-stage vacuum phototube having an 8-4 spectral sensitivity response. The photo tube was specially selected for its very low dark current, less than 0.001 jLa.This lens and stop combination did not permit the receiver to register rays departing from parallel by more than a few degrees, and reduced to a negligible amount the effect of "beam-broadening" caused by smoke particles scattering light from the original beam. The photo tube circuit load comprised the input resistance of the recorder, and an adjustable load resistor arranged to provide a convenient signal level. When the full-scale output for the clear, smoke-free condition was adjusted to 1 v or more using the resistor adjustments, tenfold reductions in light transmittance could be accommodated without appreciable loss in accuracy, by decade range changes of the recorder down to O.I-mv full scale. This permitted the recording of reliable optical densities of about 4, corresponding

The light path was arranged vertically to reduce errors in measurement due to smoke stratification effects. The light source was a 30-w, 120-v, 8-11, Type

30 /,COARSE ZERO ADJUST

WATT

FILM LAMP"

CONSTANT VOLTAGE TRANSFORMER REAR I I \ \ OF CHAMBER 110 V

VIEWER

100D
----.. FINE

IOn
I

2"DIA.
7 DIOPTER LENS

~~ I

----

811--~----

L1 1- I
I
I
I I

3" DIA.GLASS WINDOW _TOP OF

5u_~_
CHAMBER

5"

5"

/"t'

DIA. STEEL

RODS

I
~BOTTOM OF CHAMBER

5"
I SIGNAL OUTPUT '/, I
I

ALL

SURFACES PAINTED
./

I"
"2

DIA. STEEL

RODS

\ \\
\ \\ I

FLAT

BLACK

~"DIA. HOLE ---.::.\ 8

PHOTOMETER
PLAN

LOCATION
VIEW

lI"x

II"X~rALUMINUM TOP AND

BASE BOTTOM

PLATE

(2)

FIG. I5-Photometer

details.

(;)

o '"
'"
m

'"

-i

~ o z ~ m
SPRING BENT FROM

.032" x 3"x 3"

o "
."

-i :r

o
I"
"80.0.
STAINLESS

o '" ~ m
> '"

I"
32

~w I Ii);
11

c '"
(;)

WALSJr I"
TUBE --~

I"
8" DIA. STEEL /' RETAINING ROD SPECIMEN RETAINER

STEEL

41 ~
of specimen holder and pilot burner. FIG. I6-Details

~ o A
m

'"

PI LOT

BURNER

ARRANGEMENT

GROSS ET Al ON METHOD FOR MEASURING SMOKE

197

198

FIRE TEST METHODS

,
j
rate of temperature and absorptivity. Test Procedure

GROSS ET At ON METHOD FOR MEASURING SMOKE

199

to transmittance values of 0.01 per cent of the incident light. At the lowest levels of light transmittance a correction becomes necessary for the dark current of the vacuum phototube, since this represents zero light transmittance. The photometer system was checked occasionally and any accumulated smoke and dust deposits on the lenses, mirror, and phototube removed. The photo tube should be inspected, and replaced if necessary, when there is an indication of a shift in dark current between tests or of excessive zero shift during test. A check of photometer linearity was readily made using either wire screens of known open area, or calibrated neutral density gelatin filters. 'Specimen Holder The 3 by 3-in. specimen was placed in a holder (see Fig. 16), designed for rapid positioning and for maintaining, by means of the furnace support, the specimen surface 1 Yz in. in front of and parallel to the furnace opening. The furnace support also served as a positioning mount for a radiometer, Fig. 17, which established the prescribed irradiance level at the specimen surface just prior to its exposure. The stainless steel specimen holder was fabricated by bending and brazing (or spot welding) to give a 2!J1u by 2!J1uin. exposed area. The back, edges, and front non exposed surfaces of the specimen were covered with a single sheet of aluminum foil (thickness 0.001 in. or greater) to prevent smoke passage at any but the exposed specimen surface. Behind the specimen was placed a 3 by 3 by Yz-in.-thick sheet of asbestos millboard (conforming to Federal Specification HH-M-35I). A phosphor bronze spring and a steel pin were used to maintain a snug assembly. When the proper spacing (9%-in.) was maintained between the spacing stops of the furnace mount (Fig. 14), the loaded specimen could be quickly and accurately positioned by placing it on the support bars and sliding the radiometer or another holder to the limit of its travel. Radiometer The desired irradiance level (2.5 w / em") at the specimen surface was measured by means of a circular foil radiometer of the type described by Gardon (12]. It is of simple construction, has a sufficiently rapid time-constant, and produces a millivolt output which is nearly proportional to the irradiance level. The use of a reflective heat shield, with aperture, on the front of the radiometer and a finned convector supplied with compressed air on the rear, help to maintain the radiometer body at a more constant temperature and to minimize effects due to variable convective and radiative losses. The body temperature was monitored by means of a thermometer placed in a well. (After brazing the copper wire to the center of the receiving plate, it is necessary to remove all flux.) The receiving surface was spray-coated with an infrared-absorbing black paint containing a silicone vehicle.' The absorptivity of the paint to infrared energy is approximately 0.93 for a film thickness of I mil. Details of the radiometer construction are shown in Fig. 17. The air-cooled radiometer was calibrated by placing it at suitable distances from a radiant energy source and measuring its electrical output as a function of the irradiance level. The latter was determined calorimetrically by measuring the 'Type
gan,

rise of a copper disk of known weight, area, specific heat,

Ill.

8X906 Flat Black Paint, Midland Industrial Finishes Co., Inc., Wauke-

All specimens, prepared in the 3 by 3-in. size, were predried for 24 hr' at 140 F and then conditioned to equilibrium with an ambient of 73 5 F and 50 5 per cent relative humidity. The specimen was representative of the material or composite as intended for use and was prepared by the intended application procedures. Where the intended application of a finish material was not specified, or could be any of several, the following procedures were established for preparing specimens for test: (a) Surface finish materials, in either liquid or sheet form, including those intended to control and reduce the smoke produced by supporting base materials, are tested in the assembly or assemblies proposed for use. In the absence of specific information, the finish material is applied to the smooth surface of y.:!-in.-thick tempered hardboard sheet using recommended (or practical) application techniques and spreading rates. (b) Liquid films, such as sealers and adhesives, and other materials intended for application to noncombustible base materials intended for application to noncombustible base materials or being tested for their inherent smoke contribution are applied to the smooth surface of ~-in. thick asbestos cement board of 120 lb/ft" density, using recommended spreading rates. (c) Materials intended for air-backed applications, such as suspended ceilings or hanging drapes, are mounted in a specimen holder providing a Vain. air space behind the specimen. This is accomplished by use of an asbestos board back fitted with Ys-in.-thick, y.:!-in.-wide asbestos strips at the borders. To perform a test, the electrically powered furnace and associated controls were turned on. The radiometer was placed in a specimen holder and positioned in front of the furnace. The compressed air supply to the radiometer convector was turned on and the flow rate (or pressure) was adjusted to correspond to the value used for calibration. The controller temperature setting was adjusted to produce a millivolt output of the radiometer corresponding to an irradiance of 2.5 w / em". The two autotransformers were adjusted to suitable voltage levels (generally between 80 and 100 v) to minimize cyclic variations in irradiance. The photometer light source and the recording meter were turned on. Using the lamp and load resistor adjustments, the output reading was set to full scale on a convenient range (1 v or higher). The zero reading on the most sensitive range was verified by shorting the meter input. A preconditioned specimen in a cool specimen holder was mounted, using a single sheet of aluminum foil (0.001 in. or thicker) along the back edges and unexposed periphery of the front surface of the specimen, care being taken not to puncture the foil. The specimen was backed with a sheet of Y2-in. asbestos millboard and assembled into the holder snugly, using the spring and pin. For nonflaming (smoldering) tests, the air supply to the radiometer was turned off and the loaded specimen holder was placed on the bar supports and moved into position at time zero by displacing the radiometer. The door was closed. Values of light transmittance versus time were recorded, making full-scale range changes (in decade steps for maximum convenience) as appropriate. Observations were made of characteristic smoking or burning patterns, the color and nature of the smoke, and so forth. At very low light

1 200 FIRE TEST METHODS

levels, the window was covered to avoid stray light effects. The test proceeded until a minimum light transmittance value was reached. The "dark current" light transmittance was recorded by switching off power to the photometer light source and setting the recorder at a suitable high sensitivity. The zero reading was verified again by shorting the meter input. The door was opened a small amount and the exhaust fan turned on to clear the chamber of smoke. The specimen was discarded and the chamber completely cleared of smoke. With the photometer light on, the final light transmittance value under clear air conditions in the chamber was recorded, making the appropriate meter range changes. The glass window was cleaned (ethyl alcohol generally satisfactory), and the meter reading was again adjusted to 100 per cent transmittance in preparation for the succeeding test. For tests employing pilot ignition, gas was supplied to the pilot burner (see Fig. 16) at a rate of 350 Btu/hr (0.35 SCFH of 1000 Btu/ft" natural gas, or equivalent). The horizontally oriented gas jet was ignited using an electrically powered platinum "hot wire." The lighted pilot burner was positioned to impinge on the specimen surface when the loaded specimen holder displaced the radiometer in front of the furnace. If the pilot burner was blown out during a test, it was reignited using the platinum hot wire. The test proceeded as before. For tests in which the photo tube dark current was reached, the smoke was mixed after 20 min into an adjoining large chamber (54 fta) by means of a blower and short connecting ducts, and the transmittance of the diluted mixture was measured. When necessary, corrections were applied for smoke deposition and settling, by extrapolating the transmittance increase back to the time at which the dilution was initiated. Test Results The result of a smoke measurement test of a material is a curve of specific optical density versus time. For each test, the following information was noted: (a) Identification of the material, including data such as density, thickness, and type of base material (if used). (b) Test conditions, including irradiance level (2.5 w/cm'), flaming (pilot) or non flaming (no pilot) exposure, and so forth. (c) Important visual observations of specimen, color, and nature of smoke, and test chamber conditions both during and after test. The reduction in the light transmittance caused by smoke wascollverted to specific optical density using the relation:

GROSS ET AL ON METHOD FOR MEASURING SMOKE

201

windows was made except in instances where the smoke dilution measurements were made. The important test results summarized were: (1) the peak specific optical density, (2) the maximum 2-min rate of optical density, and (3) the tim'e to reach a specific optical density of 16.

APPENDIX 2
In preparing this paper it has seemed premature to include suggestions on the most useful way in which an overall smoke obscuration hazard index

DmO.9Dm-

0"'1 0.7Dm
W Y:
~

---.

o
IO.5Dm --

t50

t70

t90

D.

= LA [ IOglO v

COO) ]

132

IOglO

COO)

TIME

where T is the per cent transmittance and L = 3 ft, A = 0.0456 ft', and V = I S ft" for the standard chamber. For each test, a record was retained of: (a) the phototube dark current (with photometer light source off and recorder set at a suitably high sensitivity), and (b) the final transmittance value (after removing specimen and clearing chamber of smoke). To correct for light transmittance readings approaching the dark current value, the ratio between them was used as the corrected light transmittance, and the optical density then computed. No correction for smoke deposits on

FIG. IS-Points on typical smoke accumulation curve used in developing the Smoke Obscuration Index (S.O.I.). might be developed on the basis of the measurements reported. There are, obviously, numerous ways in which this could be done, but probably no way is uniquely appropriate for all fire situations. It might be useful, nevertheless, to indicate one such possibility. It will be assumed that the hazard to visibility is directly related to the product of the maximum observed smoke accumulation and the observed average rate of accumulation, and, inversely, to the time to reach a critical

.~

I
1
202 FIRE TEST METHODS

~
.1 700

GROSS ET AL ON METHOD FOR MEASURING SMOKE

203

smoke accumulation level, corresponding to D. of 16. These are identified by the symbols D . , R, and I, , respectively. If the average accumulation rate is defined as the average of the linear rates for each of the four 20 per cent smoke intervals between the ]0 and 90 per cent levels, Fig. ] 8, then the average accumulation rate, R, may be written as:

R=~

20 D(

lao -

110

+ I +--+--I) I
100 -

600

lao

170 -

150

190 -

170

where 110 , I"" , etc., indicate the time in minutes at which the smoke accumulation reaches ]0, 30, etc., per cent of the maximum indicated accumulation

D .
TABLE 4-Computed " Lb/ft'.in. . ................. tyrate. board) insulating grade) ished smoke obscuration index for various specimens.
88 3366 0.42" Smoke 0.25

500

.0.50 Ib/ft' S.D.!0.6 372Thickness,.0.25 S.D.!. 434 ...... 56.0Density, Smoldering.0.78 72.041 67 1.6 112 156 11.0 16 27.075 80 Flaming21.02.1 56 Specimen .0.22 64.02.4 13215Dm 11.0 178 9.5 107 287 117 660 49 .0 208 300 595 >970.0 33.0 1.6 >660 1.3 ..... 0.25 >1270.0 D

o
w ~ ~

If) 400

o
(f) 300

200

!S
<c.

\'(t:.O

OI>.l<.

The smoke obscuration DmR

hazard index is then defined as:

100

S.O.l.

100lc

-- --- +--+ I I
2000lc Dm2
(

lao -

110

100 -

lao

170 -

/50

+ I)
/90 -

170

where the factor of ]1 ]00 is used to yield index numbers usually greater than one. This index assumes the dimensions of reciprocal minutes squared. Table 4 shows the result of applying this method of deriving an overall index to selected materials included in Table 3. A comparison of the 5.0./. values with the smoke accumulation curves for these specimens in Figs. ]9 and 20 suggests that the ranking achieved is intuitively correct although it is evident that significant emphasis has been placed on the time at which the smoke has been released and for some materials early rapid smoke accumulation can overshadow the total quantity and rate of accumulation. It must again be emphasized that these indexes are specimen rather than material properties, since specimen thickness and surface finish may be important uncontrolled variables.

600

1200

TIME, SECONDS

FIG. 19-5moke accumulalion curves for aclive flaming conditions for specimens of various materials.

.. ~, <j

j 204 FIRE TEST METHODS

I
J I
(

'00

I I

600

!
'00

!
0'
w
>: o :; tf) 300 '00
J,

600

"'"

IZOO

"00

1800

ZIOO

"00

TIME. SECONDS

FIG. 20--Smoke various materials. References

accumulation

curves for smoldering exposure of specimens of

[1] Bono, 1. A. and Breed, B. K., "Study of Smoke Ratings Developed in Standard Fire Tests in Relation to Visual Observations," Bulletin of Research No. 56, April, 1965, Underwriters' Laboratories, Inc., Northbrook, Ill. [2] Foster, W. W., "Attenuation of Light by Wood Smoke," British Journal of Applied Physics, Vol. 10, 1959, pp. 416-420. [3] Akita, K., "Studies on the Mechanism of Ignition of Wood," Report of the Fire Research Institute of Japan, Vol. 9, March, 1959, p. 43. [4] Van de Hulst, H. c., "Light Scattering by Small Particles," Wiley, New York, 1957. [5] Kingman, F. E. T., Coleman, E. H., and Rasbash, D. J., "The Products of Combustion in Burning Buildings," British Journal of Applied Chemistry, Vol. 3, 1953, pp. 463-468. [6] Hurley, T. F., and Bailey, D. L. R., "The Correlation of Optical Density with the Concentration and Composition of the Smoke Emitted from a Lancashire Boiler," Journal of the Institute of Fuel, Vol. 31, 1958, pp. 534-540. [7] Hemeon, W. C. L., et aI, Air Repair, Aug., 1953, pp. 22-28. [8] Green, H. L., and Lane, W. R., "Particulate Clouds: Dusts, Smokes and Mists," Van Nostrand, New York, 1957. [9] Middleton, W. E. K., "Vision Through the Atmosphere," University of Toronto Press, Toronto, 1952. [10] Axford, D. W. E., Sawyer, K. F., and Sugden, T. M., 'The Physical Investigation of Certain Hygroscopic Aerosols," Proceedings Royal Society, Vol. A195, 1948, pp. 13-33. [11] Anon, "A Method of Measuring Smoke Density," NFPA Quarterly, Vol. 57, 1964, pp. 276-287. [12] Gardon, R., "An Instrument for the Direct Measurement of Intense Thermal Radiation," Review of Scientific Instruments. Vol. 24, 1953, pp. 366-370.

"J

;j

;it

.. ~I
t
f'

.',

~ ~!