STP44709S.27319

A. F. Robertson1 and Daniel Gross1

Fire Load, Fire Severity, and Fire Endurance

REFERENCE: Robertson, A. F. and Gross, Daniel, "Fire Load, Fire Severity, and Fire Endurance," Fire Test Performance, ASTM STP 464, American Society for Testing and Materials, 1970, pp. 3-29.

ABSTRACT: A review is presented of fire studies beginning with the work of Ingberg at the National Bureau of Standards, who attempted to relate the severity of a fire endurance test in the laboratory to the conditions existing during actual building fires. He showed the importance of weight of combustibles per unit floor area as a major factor. He recognized the importance of ventilation in controlling fire behavior but did not specify it as a separate variable. Fujita in Japan is credited with emphasizing the importance of ventilation. His work has been followed and enlarged by others around the world. Ventilation parameters, compartment geometry, and fuel arrangement have been shown to exert a powerful influence. The radiance from a burning building is dependent to a large extent on the nature of the ventilating openings. Fire severity is not well defined, since it depends on the interaction of the temperature-time curve developed during a fire and the thermophysical properties of the materials exposed. There is a great need for further research on the influence of fuel arrangement, building geometry, and ventilation on fires in buildings.

KEY WORDS: fires in buildings, burn-out, fire severity, fire endurance, fire ventilation, experimental fires, evaluation, tests

The structural failure of buildings during accidental fires was so frequent and disastrous to adjoining property owners at the turn of the century that both the insurance industry, through the Underwriters' and the Factory Mutual Laboratories, and the public, through the National Bureau of Standards (NBS), built new laboratories and initiated studies on the fire performance of building construction elements. The first of these investigations involved cooperative fire tests of building columns [I].2 While the fire test procedure used at that time was not based on a national standard, it was very similar to methods used today, ASTM Methods of Fire Tests of Building Construction and Materials (ASTM

1 Physicist, Office of Fire Research and Safety, and physicist, Fire Research Section, respectively, National Bureau of Standards, Washington, D.C. 20234. Mr. Gross is a personal member ASTM.

2 The italic numbers in brackets refer to the list of references appended to this paper.

Copyright̂ 1970 by ASTM International www.astm.org

Copyright by ASTM Int'l (all rights reserved); Mon Nov 23 23:44:50 EST 2015Downloaded/printed byUniversidad Catolica De La Santisima Concepcion - UCSC (Universidad Catolica De La Santisima Concepcion - UCSC) pursuant to License Agreement. No further reproductions authorized.

4 FIRE TEST PERFORMANCE

designation: E 119-67), etc. The present fire endurance test involves erection of a portion of a building either within a furnace, as in the case of columns, or as part of the enclosing walls or top of such a furnace. The structure is loaded, if appropriate, and oil or gas fires are initiated within the furnace and controlled to follow a standard temperature-time curve.

The length of time during which the specimen remains structurally stable without the development of through openings or of excessive temperature rise on the unexposed surface is defined as the fire endurance of the construction. The adoption of a single temperature-time curve for these tests was based on the recognized need for a performance evaluation and the obvious need for economy in testing. Nevertheless, it was recognized early that such a procedure did not closely simulate the many types of thermal exposure likely to be encountered in a building fire.

This paper reviews the published evidence of the manner in which these test methods were justified. Early work related to such justification was confined primarily to the study of temperatures developed within the building or furnace. Recent developments have shown the need to recognize that one of the important effects is radiant emission from the fire. In this paper radiation effects are treated as one aspect or measure of fire severity.

Early NBS Work

In 1922, the National Bureau of Standards reported the first of a series of burn-out tests in specially constructed buildings [2]. It is clear from this reference that the objective of the study was to relate fire test exposure conditions to those existing during fires in occupied buildings. These first studies were conducted in a brick building 16 by 30 by 9 ft high. Further studies [3, 4] in a brick building 30 by 60 by 9 ft high were described briefly in 1926. The latter of these, describing burn-out of office occupancies involving both wood and steel furniture and filing cabinets, reported that the decrease after peak temperatures were developed was much more gradual in an actual building fire than that following a typical furnace test. It was further stated that the way in which the two exposure conditions could be related must wait for further study. Another report [5] defines the objectives of the studies by saying:

The tests to obtain information on the intensity and duration of fires in buildings are conducted to determine the temperatures to which it is proper to subject building materials and constructions in fire test furnaces and to serve as a guide in applying the results of such furnace tests to building design.

The first report of such correlation appears to have been published as part of the Report of the Committee on Protection of Records in the


ROBERTSON AND GROSS ON LOAD, SEVERITY, AND ENDURANCE

Proceedings of the National Fire Protection Association [6] in May 1927. Here results were tabulated for equivalent fire duration of office and file room occupancies using wood furniture or shelving and fire loads (weight of combustible per unit floor area) varying from 10 to 60 psf. This reference contains the following statement:

The durations given in the table apply to air temperature and flame effects on record containers and thin partitions, the average room temperatures developed in the tests being considered as affecting records or thin partitions down to 300°F. (or 150°C.). As concerns effects on heavier partitions and walls, as well as on interior incombustible structural members whose strength is not appreciably reduced by temperatures throughout the section of 300°C. (572°F.), the periods can be reduced by 10 to 15 minutes for durations of 2 hours or less and by IM hoiu-s for the 7 and 8 hour periods.

Photographs of the occupied buildings and descriptions of the tests were published in the Quarterly of the National Fire Protection Association [7] in January 1927. This same journal a year later [8] carried the most complete published report on these tests and again included a table correlating fuel load with fire endurance during a standard test.

The two buildings mentioned above were constructed specially for these studies and were furnished to simulate the type of occupancy desired. One such arrangement is shown in Figs. 1 and 2. The first

-<:

FIG. 1—The interior of small test building representing office occupancy with metal furniture on concrete floor. The exposed papers were intended to represent the probable maximum of exposed material in an occupancy of this type.


FIRE TEST PERFORMANCE

FIG. 2—The interior of the small test building after a fire test with office occupancy, metal furniture on concrete floor. This was an exposure fire test, the rack in the right background of the picture being filled with kindling wood surrounded with a shield which was drawn back after it was burning briskly, thus simulating the condition of an exposure fire.

shows conditions prior to starting an office occupancy fire; the latter shows conditions after the fire. In this test the fire start had been made more severe by burning a kindhng fire in a shielded grate or tall metal basket and then, when the fire was well under way, removing the metal shield from around the kindling. This was described as an exposure fire starting method. It is clear from the description of these experiments in the reference that the type and distribution of combustible furniture and contents as well as the ventilation had an important influence in modifying fire behavior. For example, in the reports, statements such as the follovidng were common:

During this test the window shutters were opened by an amount deemed sufBcient to give the maximum fire intensity when the fire was at its height and left at such openings while the interior cooled down.

It now seems surprising that the importance of ventilation control did not suggest the necessity of including in the reports some quantitative indication of its magnitude.

In selection of a fire endurance period corresponding to a given building bum-out experience [6, 8, 9] it was assumed that the furnace


ROBERTSON AND GROSS ON LOAD, SEVERITY, AND ENDURANCE 7

cooling process after termination of the fire test comprised a part of the thermal exposure. While recognizing the technical questions posed [8] it was assumed that by matching areas below the average burn-out temperature curve with that below a combined heating and cooling curve in a furnace, the severity of furnace specimen exposure would correspond to fire.^ Figure 3 presents the derived relationship between fire endurance and fire load together with the results of burn-out tests on which it was based.

7

6

5

A

3

2

1

FIRE LOAD, kg/m2 50 100 150 200

1 1 1 1

-

-

/ "

X / *

f \ \ \ \

250 1

• •

« /

• X

X

1 1

-

-

-

-

-

-

-

30 40 50

FIRE LOAD, Ib / f t ^

FIG. 3—Laboratory fire endurance test period corresponding to experimental temperature time results from 15 burn-out experiments performed by Ingberg. The points marked x and • correspond to a match of areas above base temperatures of 300 and 150 C, respectively. The solid line passes through points recommended by Ingberg in Refs 8 and 9.

During the period from 1928 to 1940, surveys were made of the fire load (that is, combustible contents) of residences, offices, schools, medical buildings, and a few mercantile buildings [9]. In 1947, an enlarged survey was made of the combustible contents of mercantile and manu-

3 The areas measured in this matching process were either above 150 C, representing temperatures at which records or thin partitions would be damaged, or above 300 C, corresponding to the lowest temperature at which thicker partitions or walls may be assumed to be influenced by the thermal exposure conditions.



O) • *

•̂ en :§ w 1 o

<N

c«

s m .s 13

o a. ca K

^ o

t j

w

c o O

ID

3 w 3

S

6

c o

1

S

3

•a p.

1! £

Hi UH

« a a? § E S > <

.« S j -

jSS

^<

^ ,TJ U V

II I I

>, s 2 & g u o

' — •

135 • *

i-H

O

oo

03 00

ira CD f—(

CO

CO t-H

(D O

3 *crt

C} 60

« 2 « 5

W

a>

C/3 n

• — • — '

05 CO — K M

o

CO

00 N

o 00 t -CO • * I—1

I—(

'B, o

's 5 o a o

o

IP

—' 00

CO

£ t^

t -i n i - H

i n oo CO cq t~

CO

c u

c/:

"c 0

ID

4-*

ft

1:~

»*•

CO o P H

M CO o i n o p-H

I—I

«

'QT ( H

£ « C3 0)

§ t - (

rt ft <3J

s ^ •p

IS

11 bo ca U i

o • CA

l - l

a ID

> i >

ft O

^ w 4-»

' 3

,—̂ V to a (-. o

iH

ft tS

3 . Pi • ^ —

• * t -T-IOCD

3

CO

0 5

g o c'l"

in m

bOiJ

c c

bo C •c S

u

13

CO in

0 - * in-*' CMt-

coin 05in

ooin l-HCO ini—I

OP4

3 O

I

60

•=3 S

CD g O J 5

• ^ 111



facturing buildings [10]. These surveys involved the weighing of all the movable combustible contents of the buildings studied. In cases where furnishings were fixed in place estimates of the weight were made on the basis of dimensions. Additionally, weights were estimated for combustible flooring and exposed woodwork other than flooring. These items were reported separately as well as combined with movable fire load to show the total fire load present. The results of these surveys are summarized in Table 1 and have been used [9] as a basis of defining the fire endurance performance requirements for fire resistive buildings in building codes.

Ingberg had shown through his burn-out studies [7, 8, 9] that when combustibles were stored in metal furniture and filing cabinets, only a fraction of the material so stored would become involved in fires that might occur. Apparently at that time the use of wooden furniture and cabinets was so common that when Ref 9 was prepared no consideration was taken of the lower fire loads and the lower fire performance requirement for the structure that would have been possible by the use of metal furniture.

This work in measuring combustible loads in buildings served to close the gap between fire endurance tests in the laboratory and the experimentally measured fire severity as a function of fuel load during actual building fires. The surveys were performed from 20 to 40 years ago, and there is today some question as to the relevance of the findings at that time when compared with present trends in occupancy of buildings. As a result there are plans in several countries to update these studies. The meager evidence available to date does not suggest a need to revise the original findings.

Postwar Japanese Studies The work of Ingberg apparently represented the most progressive

thinking on this subject until sometime after World War II. At this time, the Japanese, because of their serious losses to fire during the war and the obvious need for better understanding of the problem, assumed leadership in the theoretical and experimental study of fire problems.

Fujita* started work on spread of fire between buildings by radiation and convection in 1940 and by 1948 had published ten papers on this subject. During this period his attention was directed to fire development within enclosures. His first study on this subject involved the analysis of the thermophysical problem of gas flow through the windows of a compartment involved by fire. He then proceeded to study the heat balance between a fire, the enclosing walls, and the surroundings. In

* Fujita, K., Building Research Station, Japan, private communication.



1948, he encouraged Kawagoe to continue this work with both model and full-scale fire experiments. Kawagoe [11] published an English version of his work on enclosure fires in 1958. In this paper he described both extensive fire model experiments as well as 14 large scale building fire studies. Moreover, he suggested a rational analysis of the building fire problem. He showed that the fire activity or rate of burning was a direct function of the air supply and, in many cases, could be represented by the product of window area and the square root of the window or ventilation opening height. On analysis of the combustion of wood, assumptions of the excess air flow, and the completeness of combustion, he was able to predict the fuel combustion rate and, from this, fire duration.

He summarized his work on studies of fires in actual buildings by suggesting that:

1. With small window area to fuel ratio, combustion may be prolonged and temperatures relatively moderate.

2. With large window area to fuel ratio, the period of uniform combustion can be greatly reduced but often temperatures develop which are much above those of the standard curve.

3. That while the relationship between fuel load and fire duration accepted in United States and United Kingdom may be appropriate for buildings with small windows, it was not applicable to many of the modern buildings with very large windows. In this latter case "flashy" fires were found to produce compartment temperatures significantly higher than those expected on the basis of the standard temperature-time curve.

Kawagoe and Sekine have continued these theoretical studies [12, 13, 14] and combined them with the pioneer work of Fujita by attempting to set up a balance between heat generation and loss from a compartment during a fire. In doing this they have recognized heat loss to enclosure walls, as well as to the surroundings through windows, by both radiation and convection and have assumed that the fuel burned at a fixed rate proportional to the product of window area and square root of window height. With this type of analysis the following enclosure properties assume significance:

1. Internal surface area. 2. Window area and height. 3. Wall thickness. 4. Thermal properties of wall materials.

Fuel, flame, and ventilation properties must be recognized also, and the authors proposed the following:

1. Flame emissivity. 2. Excess air fraction. 3. Rate of fuel consumption.



The last of these has been assumed controlled entirely by window ventilation and directly proportional to

A^/^^ where:

A = window area, and H = height of the window.

It was assumed that all the properties involved were independent of temperature and time. With these simplifications he was able to compute the temperature-time curve within an enclosure. This was done for various wall conductivities, as well as for various values of "opening" or "fire temperature" factors, defined as A-S/W/AT where AT was the total interior enclosure surface including window openings. It was shown that fire temperatures increased with opening factor but were related inversely to the thermal conductivity of the wall material.

Kawagoe and Sekine concluded, from studies of burn-out tests, that after peak temperature has occurred the temperature within enclosures seems to decay at a rate of about 10 or 7 C/h depending on whether peak temperatures were reached at times greater or less than one hour, respectively. They have used these cooling rates associated with the computed temperature rise curve to complete the predicted curve. The time at which the cooling process started was derived from the ratio of total fuel load to burning rate. The burning rate, of course, was assumed dependent only on the ventilation parameter. In their papers many of these curves are presented for different types of typical buildings, and it was shown that the peak temperatures developed were usually higher than those called for by the standard curve used in Japan. This curve is, for the first 4 h, very similar to the curve used in this country.

They proposed, therefore, the selection of a standard fire test exposure time such that it yielded the same area above a base temperature of 300 C as achieved by the combined heating and cooling portion of the computed curve. This treatment is quite similar to that used earlier by Ingberg in comparing furnace exposures with experimentally derived temperature time data. However the Japanese procedure neglects the specimen thermal exposure during cooling after completion of a fire test.

The techniques used in these studies [12, 13, 14] are certainly interesting and of value. However, several observations seem appropriate. The first is concerned with the assumption of uniform burning rate completely controlled by window ventilation. This is based on the early recognition in Japan of the importance of ventilation openings in controlling fire activity together with correlations developed by Thomas following study of a variety of enclosure fire experiments [15]. More recent work in the United States [16] and in Britain [17] has shown that



there is a definite influence of fuel arrangement in setting an upper limit to the combustion rate. Kawagoe by his references to these later studies recognizes but does not allow for the problems introduced by neglecting this rate-limiting factor. This may result in the prediction of temperatures significantly on the high side in many cases.

Another apparent shortcoming of the analysis in his first two papers is that in deriving the predicted temperature-time curve for a building fire he assumed that the interval from start of the fire to maximum temperature corresponds to consumption of all the fuel. At the latter time a gradual cooling process was initiated to simulate the temperature decay process during a real fire experiment.. All the evidence available suggests that the gradual cooling process during a real fire burn-out experiment is itself a portion of the fuel combustion process. Thus the predicted gradual cooling process overestimates the heat release capability of the fuel in the building and again results in the suggestion of need for more severe fire exposures than might otherwise be necessary.

In his most recent paper on this same subject [14] Kawagoe recognizes this limitation and has developed a computer solution of the heat balance equation for conditions other than constant heat release rate. He also provides for computer solution of the temperature decay curve. The need now is to develop a better understanding of the factors of importance in controlling fuel consumption rate. It is evident, however, that the work in Japan has gone a long way towards rational prediction of fire behavior in building compartments.

Recent Swedish Studies In 1963, Odeen published a thesis [18] in which he set up a heat

balance equation in a manner very similar to that of Kawagoe and Sekine. However, heat release rate was not related to window opening but arbitrarily estabhshed. He showed that compartment temperatures were a direct function of heat release rate as well as the insulating properties of enclosure walls. Such temperatures were related inversely to both the heat capacity of the inert items present in the room and the emissivity of the enclosure walls. Odeen also made estimates on the importance of fuel element geometry on the rate of heat release, on the assumption that the reaction was surface-area controlled. He concluded that for a given average rate of heat release and minimum cross-sectional dimension of fuel element the maximum fire temperatures would be little affected by shape of fuel element, but the temperature time history would experience drastic changes.

In a later paper [19] Odeen has reported experiments on fires in a small hut. Fire severity, as indicated by areas below the temperature-time curves, were measured and compared with changes in quantity of



fuel available for combustion as well as surface-to-volume ratio of the fuel. The results indicate that both temperature and fire severity were related directly to the quantity of fuel consumed. On the other hand, fire severity was related inversely to air supply rate while the maximum temperature increased with air supply in the low air feed rate range but was affected very little by air supply for higher rates. Changes in fuel element size, over the range studied, had surprisingly little effect on both fire severity and maximum temperatures observed.

Recent British Studies Work recently conducted in Britain has added much to knowledge of

fires in compartments. Planned experiments were performed in compartments of roughly 3.7 by 7.7 m (12 by 25 ft) dimensions and 3 m (10.5 ft) high. Two 1.8 by 3 m (6 by 10 ft) windows were located in one of the larger compartment walls. The controlled variables studied included:

(a) Fire load. (b) Fire load display, that is, fuel stick spacing and arrangement.

1200,

1600 u.

30 40 TIME — min

FIG. 4—Average compartment temperatures as a function of time, for compartment fires involving varying fire loads and window ventilation conditions (from Ref 20).


14 FI RE TEST PERFORMANCE

1200

."lOOO

13 800 < a: 111 O L

UJ 6 0 0

FIRE LOAD PER UNIT FLOOR AREA —lb / f t * 1-5 3 6 12

400

2

- 200 <

O 7.5 15 30 45 FIRE LOAD PER UNIT FLOOR A R E A — k g / m '

Q 4̂ ventilation O ^i ventilation

FIG. 5—Maximum average temperatures inside fire compartment as a function of fire load per unit floor area and two window ventilation conditions (from Ref 20).

(c) Window opening area (height was constant). (d) Fuel character; solid or liquid.

Dependent variables observed as a function of time included:

(a) Gas temperatures within the compartment. (b) Radiation levels both within the compartment and through win

dows. (c) Fuel consumption rate. (d) Thermal gradients in walls and ceilings.

Twenty-four full-scale fire experiments comprised a study designed for analysis by statistical methods to yield the most comprehensive findings possible. The results have been analyzed in a number of papers [20, 21, 22, 23], In the first of these it was shown that over the range of variables studied, compartment temperatures were related directly to fire load per unit floor area but inversely to area of the ventilation opening. Figures 4 and 5 are reproduced from this reference [20], Later work including windows only Vs open showed lower peak average


ROBERTSON AND GROSS ON LOAD, SEVERITY, AND ENDURANCE 1 5

RATE OF BURNING/VENTILATION AREA —lb m i n ' ' f t ' ' 1 2 3

Kerosine

O Fibre insulating board

5 1° , 5 RATE OF BURNING/VENTILATION AREA —kg nnin"'m"'

15

+ O

A

'' j-ventilation

V4-v(2ntiiaiior(

''a-vantilation

FIG. 6—Maximum average gas temperature rise inside fire compartment as a function of average burning rate per unit ventilation area, data points for other than wood-crib fires are marked (from. Ref 22).

temperatures than for V4 window openings, at least for the 60 kg/m^ (12.4 psf) fire load. Thus there are fuel and compartment variables which in particular combinations produce maximum compartment temperatures; above and below these values, compartment temperatures are



lower. This explains how interpretation of data over limited ranges in variables could yield seemingly different conclusions. It was evident that, as found by others, the area below die temperature-time curve for a given fire load generally increased with decreasing ventilation opening.

Heselden" [22] published an analysis of the heat balance during nine of these fires in which the rather shallow compartment had noncom-bustible finish. He reported that, for the experiments performed, changes in ventilation area had little effect until fuel loads of about 146 kg/m^ (30 psf), based on area of window opening were reached. If recognition is taken of the way in which window height, as well as area, serves to control ventilation, this ratio, total fuel load, L, divided by A\/H, becomes 112 kg/m^/^ (12.2 Vo/h^'^). The author commented that average gas temperature rise in the compartment can.be correlated directly with fire load and inversely with window area but found a better correlation of such temperatures directly with burning rate and inversely with window ventilation area. His plot is shown as Fig. 6. Since burning rate is a dependent variable, it seems preferable to plot temperature as a function of fire load divided by As/W iox the window. This plot is shown as Fig. 7. It is evident that with the exception of the fires involving liquid fuels or fiberboard wall linings the maximum tem-

2400

FIRE LOAD/VENTILATION ( L / A / h i j . k g / m ^ ^ ^

100 200 300 400 500

10 20 30 40 50

FIRE LOAD/VENTILATION(L/A\/H),lb/ft^'^

60

FIG. 1—Maximum average gas temperature rise for compartment fires as a function of fuel load per unit ventilation parameter Ay/iT; points marked F, G, and K refer to fiberboard, gasoline, and kerosine fuels, respectively. Symbols used correspond to different floor fuel loadings *-7.5 kg/m^ (1.55 psf), X-13 kg/m^ (3 psf), 0-30 /tg/m« (6 psf), A-60 kg/m^ (12 psf). (Experimental data from Refs 20 and 21).



peratures are related directly to L /AViT up to values of about 80 kg/m^''^ (9.1 lb/ft^/2). Further increase of this parameter had little influence on peak temperatures, apparently because the fire was ventilation limited. Figure 8 presents these same temperature data plotted with respect to fire load per unit floor area. The deterioration of correlation is immediately apparent.

10 FIRE LOAD/FLOOR AREA, kg /m^ 20 30 40 50 60

2000

3 1600 < UJ Q.

Lu 1200 -

UJ (S < a:

LlJ

X

<

800 -

400 -

1

• G

• t "K

•

• • f

_ •

1

1

X

X

X

1

1 1

<&

8 o o

1 1

1 ^

A

1 1

-

- 1000 o o uJ a: 3

800 S a. LLI 1 -

600 UJ o <t a: Ld

> - 400

X 200 <

8 10 12 FIRE LOAD/FLOOR AREA.ib/ft"^

FIG. 8—Maximum average gas temperature rise in compartment as a function of fire load per unit floor area, see caption of Fig. 7 for symbol legend. (Experimental data from Ref 22j.

Margaret Law has analyzed the data on radiation from the windows during these 24 experimental fires [23], She has shown good correlation between thermal radiation theory and the data for window irradiance and compartment temperatures. The line shown in Fig. 9, reproduced from her paper, is the theoretical relationship for window irradiance on the assumption that the effective emissivity of the cavity is unity. The correlation is surprisingly good. In plotting other total irradiance data, she also used fuel burning rate per unit window area as an independent variable. However, it seems preferable from a practical point of view to plot the data as shown in Fig. 10. It is evident that total irradiance



6-0

4-0

D Slope a 4

/c

D /

o/

o /

7o / 1 1

o/

80 o P

n/

1

E

S 2-0

Q.

O f= 1-0 <

5 0-8

^ 0 - 6

z 0-4 UJ

X 0-2 <

0-1 4 0 0 600 8 0 0 1000 2 0 0 0

MAXIMUM TEMPERATURE (Tp) — dcgK FIG. 9—Maximum intensity of window radiation as a function of maximum com

partment gas temperature (from Ref 23J.

of these fires with wood-crib fuel arrays correlates in nearly linear fashion with the fire load ventilation parameter up to the point that the fire becomes ventilation limited, after which radiation levels start to decrease. The data for fiberboard emphasize the importance of the way in which


ROBERTSON AND GROSS ON LOAD, SEVERITY, AND ENDURANCE

FIRE L0A0/VENTILATI0N(L/AyH),kg/m^^2

19

200 400 600 800 1000 1200

20 40 60 80 100 120

FIRE LOAD/VENTILATION ( L / A / H ) , I b / f t ^ ' Z

140

FIG. IQ—Maximum total intensity of radiation as a function of fuel load per unit ventilation parameter AVET, see caption of Fig. 7 for symbol legend. (Experimental data from Ref 23).

the fuel is arranged in modifying burning behavior and thus irradiance. To date there have not been sufBcient studies to suggest effective and useful ways of predicting the burning behavior of fuels in various exposure geometries.

One step towards meeting this need was reported by Gross [24] in which the free unconfined burning behavior of cubical crib-type fuel arrays was studied as a function of stick-crib dimensions and crib porosity. Of course, the burning behavior of the sticks was influenced greatly by porosity of the crib. Nevertheless, Seigel [25] has used the data as the basis of a plot relating maximum burning rate to stick size. Use of his chart suggests a burning rate of about 3 percent of the total fuel per minute for the 4.5 cm (1% in.) square sticks used for the British experiments. Figure 11 presents a plot of fuel consumption rate as a function L/A\/H from Ref 23 based on tests under ventilation-limited conditions. Since the data show mostly fuel combustion rates greater than 3 percent per minute, it is evident that the form of crib used in the



FIRE LOAD/VENTILATION ( L / A / H ) , kg /mS /z

100 200 300 400 500

FIRE LOAD/VENTILATION ( L / A N / H ) , Ib / f t^^^

FIG. 11—Average consumption rate of wooden cribs during compartment fires as a function of fire load per unit ventilation parameter L/AVn; see caption of Fig. 7 for symbol legend. (Experimental data from Ref 23j.

British tests resulted in fuel consumption rates significantly higher than would be expected on the basis of Gross's data. This again suggests that more work is needed on the influence of fuel display properties on fire behavior.

The work of Butcher and Margaret Law [26] in relating compartment fire effects with those developed during a standard fire endurance test is of considerable interest. The steel columns protected with low density mineral wool insulation were exposed to both the conditions of the compartment fire and a standard fire endurance test. By determining the maximum average temperature rise of the steel shaft during the compartment fires and selecting exposure durations which resulted in development of equal steel temperatures during a fire endurance test they were able to plot furnace fire endurance time as a function of compartment fire load per unit floor area.

This plot failed to result in any satisfactory correlation of the data but did emphasize the importance of ventilation as a complicating variable. They then proceeded to analyze the experimental data through analogy with an electrical resistance-capacitance network. Electrical resistance



was considered analogous to the thermal resistance of the insulation, and electrical capacitance was considered analogous to the thermal capacity of the steel. On this basis they were able to show rather close correlation of furnace exposure time with development of selected steel temperatures and the time constant of the insulation-steel assembly. The empirical formula they derived was the following:

t = 0.140 ^yRC where:

t = time (in seconds) in the furnace test, 6 = temperature (in deg C) reached by the steel at time *, R r= insulation thickness divided by the product of conductivity and

fire exposed perimeter in units of deg C cm s/cal, and C = product of specific heat and steel weight per unit length in units

of cal/cm deg C.

This formula should be of significant value to those wishing an estimate of fire endurance of columns protected with light weight insulation material. The authors suggest that it is probably valid for periods up to about iy2 h.

Margaret Law [27] also has shown a correlation of compartment fire test data by plotting the ratio of the maximum value of the average steel temperature rise to effective fire temperature rise against the ratio of effective fire duration to the time constant, RC, of the column assembly. It would have been helpful if data on furnace tests had been plotted in a similar manner. Although the author suggests that the shape of the fire temperature-time curve was not critical, at least for structures with appreciable thermal impedance, this claim should be explored further before acceptance.

St. Lawrence Bums

The planned fire experiments at Aultsville [28], a town that was cleared for the St. Lawrence Seaway development, were designed specifically to explore life hazards associated with building fires as well as spread of fire through radiant exposure of adjacent structures. Six homes, a school, and a community hall were involved. Three of the homes were finished in plaster walls and ceilings, while the other three were lined with fiberboard. The homes were not furnished during the experiments so the only fire or occupancy load involved was that of the building itself together with the two fuel cribs used in starting the fires. These cribs were designed to burn well and involved sticks varying in size from 1.25 cm (Vz in.) square to 4 by 9 cm (2 by 4 in.) lumber. Each crib weighed about 155 kg (340 lb) representing a fire load for



the single room in which they were placed of about 14.6 kg/m^ (3 psf) of floor area.

A large variety of measurements were made during these tests. The only ones relevant to the subject of this paper appear to be those for exterior radiation and equivalent black-body temperatures within the building. It was proposed that measured irradiance levels could be best expressed as a hypothetical irradiance. This property was defined by the ratio of measured irradiance to the geometric configuration factor appropriate in describing the sum of the solid angles subtended by the ventilation openings viewed from the radiometer. Configuration factor is defined as the ratio of the solid angle subtended by an object, as viewed from a receiver, to that occupied by a hemisphere.

The main findings from this work may be summarized as follows: 1. When fiberboard linings were involved the radiation levels from

the building were about twice those measured during fires in plaster finished buildings. The hypothetical window irradiance values observed were as high as 40 cal/cm^ s in the case of fires involving fiberboard lined buildings. This is a level about eleven times the irradiance that would be expected, on the basis of the radiation pyrometer temperature measurements, from an area confined to the windows alone.

2. Maximum radiation measurements were influenced significantly by winds, but not by the type of exterior wall cladding used.

3. The radiation pyrometer measurements made during these experiments showed equivalent black-body temperatures within the burning buildings to be significantly higher than would be expected on the basis of the standard temperature-time curve used in laboratory tests.

The immediate conclusion drawn from the experiments was that the fiberboard lined homes presented fire severities of a much more serious character than developed in the plaster finished homes. Certainly this was the case for the fire experiments conducted, but the lack of normal combustible occupancy in the homes may have tended to exaggerate the difference. Thus while in both cases the crib presented a fire load of about 15 kg/m^ (3 psf) of floor area for the single room in which it was placed, the other rooms had no occupancy fire loads. For one of the buildings studied, the weight of fiberboard used as interior finish was estimated at somewhat over a ton. If this were to be considered as an occupancy load, it would amount to a fire load of about 13.6 kg/m^ (2.8 psf) for the building as a whole. Neglecting floor and exposed trim, this would have the effect of nearly quadrupUng the readily exposed fire load. It is not surprising that this significant difference in readily available fuel should impose quite different fire severity conditions on



the buildings. From both the academic and practical point of view, it would be desirable to perform further experiments in which a normal occupancy load was used to clarify the situation.

These experiments remain today one of the few reliable published records of radiation measurements on fires involving complete buildings. The higher hypothetical irradiance values reported of up to 40 cal/cm^ s (167.3 W/cm^) were undoubtedly influenced by winds, through ventilation, and probably also by flames projecting from windows not directly visible from the radiometers. These levels are about eight times the maximum values reported by the British [23] and similar values measured by NBS [29]. The fact that the fires involved more than a single floor of the building surely contributed to the size of, and thus radiation from, the flames.

Discussion

This review of the literature relating to fire loads and fire severity has presented much of the published information available on this subject. The excellent works of the IIT Research Institute and others have not been included primarily because of lack of open-literature publications on the findings. There has been little discussion of the basis on which severity may be measured or assessed. It seems important to consider this subject here.

Severity of a building fire may be defined as a measure of its potential for damage to contents or structure. However, assessment of severity in any given situation must be of a subjective nature until more adequate definition is provided of the way in which the fire damage is caused. Thus fires may result in damage through any or a combination of the following mechanisms:

1. Growth and spread of fire to involve combustibles other than those originally ignited in the compartment of origin.

2. Overheating of fire-protective covering of structural elements with resulting mechanical failure of the protection and exposure of the structural element to direct heating by the fire.

3. Melting or otherwise destroying the fire-exposed surface of a diaphragm type fire barrier in such a way that fire ventilation is modified, fire activity'is increased, and flames and hot gases project from the opening.

4. Overheating of a structural element resulting in loss of strength and mechanical failure.

5. Excessive heat transfer through a fire barrier causing ignition of material and thus spread of fire on the unexposed side.



6. Mechanical deterioration of the structure resulting from thermally induced spalling or other similar effects.

Ways in which fire severity could be measured in these cases might include:

1. Measurement of char formation, weight loss or other degradation characteristics of selected indicator materials displayed in the fire compartment.

2. Determination of maximum compartment temperature rise or the temperature time relationship developed during the fire.

3. Measurement of integrated heat flux absorbed by a structure or a calorimeter.

4. Development of data to permit complete heat balance for the compartment in question.

All of these techniques have been used at various times for defining fire severity. However, preference has been usually given to measurement of the area under the fire temperature-time curve as was suggested by Ingberg [8]. As noted by him, the temperature within a structure exposed to a thermal transient is a direct function of the exposing temperature rise, but time enters the heat transfer equation as an exponent. Ingberg notes that in spite of this implied technical limitation, the matching of areas provided the best way of selecting a standard furnace test time to simulate a building fire exposure. The base above which temperatures were measured was either 150 C (302 F) or 300 C (572 F ) . The former was considered of importance as the lowest level at which light combustible materials would be aflFected, while the latter figure was considered the lowest level at which protected combustible constructions might be affected. Actually the accuracies of the experimental data of Fig. 3 do not seem to justify distinguishing between these two levels, and all the published comparisons between fire load and corresponding fire endurance based on his work involve an average line fitting the combined data.

The assumption that two temperature time curves enclosing equal areas result in the same fire exposure severity is of course not exact [25], It fails to be a usefiil criterion where the differences in maximum temperature bracket a temperature which is critical with respect to the physical properties of the fire protection covering. This is especially true when the thermal time constant of the member being tested is short compared to the time at which maximum fire temperatures are observed.

Seigel [25] has pointed up the great variation in temperature-time curves which may develop in an enclosure. Fig. 4. He suggests that under such a situation it is unreasonable to use a fixed standard temperature-time curve for fire test evaluation. He proposes that one solu-



tion might involve the use of a series of different temperature-time exposures representative of different occupancy fire loads. Another method which apparently would be even more relevant would involve the use of programmed heat release rates. This latter procedure would seem to more correctly credit constructions which exhibit high heat absorption properties. The fire severity would be more responsive to factors influencing it during an actual building fire. However, both of these suggestions would complicate greatly the problem of evaluating the fire resistance of construction. As a result, these suggestions seem unlikely to be well received by testing laboratories prior to valid detailed proof of their merit.

Numerous workers have attempted to measure rate of heat release during building fires. The Japanese [11] have supported the fuel and floor of rooms on load measuring equipment and thus been able to measure fuel consumption directly. The British in their recent full-scale burn-out tests have weighed individual fuel cribs and also in this way the rate of fuel consumption. Fujita,* Kawagoe [14], Odeen [18], and Heselden [22] have all attempted heat balance calculations on compartment fires. However, such calculations and measurements at best are most difficult and tedious.

The concept of an instrument which could be used during experimental fires to measure fire severity seems a most useful one. It could be probably achieved through use of a steel shaft protected with light weight insulation as suggested by Margaret Law and Butcher [26]. Such columns involving standardized low-density but relatively inert insulation could be placed within a compartment during a burn-out test. Much smaller units, possibly of spherical shape, of course, could be used for local measurements when an average over the compartment height was not desired. The resulting temperature changes of the steel, together with fire exposure temperatures, could aid in defining fire severity.

It also seems likely that more adequate understanding of the nature and character of the hazards involved with the use of fire protective materials would result from measurement of physical properties as a function of temperature. Thus, compressive tensile and flexural strength, coefficient of expansion, and dilatometric transients (all as a function of temperature) would provide useful guides of the likely problems when such materials are to be used for fire protective coverings. This has been recognized for some time by Harmathy [30] and others in Canada as well as by investigators at the Portland Cement Association [31].

Improved tests of this type should be routine for subsidiary studies in connection with standardized fire endurance tests. Thus if a wall-



board reinforcing material is found to lose strength at a temperature just slightly in excess of that to which it is exposed during a furnace test of several hours, then this reinforcement may become the critical link in determining the performance of the structure. Such a structure, when exposed to an actual building fire which may develop temperatures significantly above the furnace exposure, could be expected to show failure after rather brief fire exposure.

Before concluding the discussion of temperature history as an indication of fire severity it seems important to mention two problems related to experimental techniques:

First, the question of the influence of flame emissivity and size in heat transfer to the specimen during furnace tests has not been studied well enough. There is at present a rather general agreement by technical people in Europe that the large luminous flames in use in some of the laboratories result in much more severe thermal exposure of specimens than that provided by premixed gas flames used at other laboratories. It seems likely that this is true during the early portions of a laboratory test. However, apart from the British reports [26, 27] in which comparisons were made between furnace and compartment fire tests, there appears to be little published information relative to this problem.

Second, the way in which temperature measurements are made during fire experiments may alter significantly their usefulness in making comparisons with other data. The fire test method, ASTM Methods E 119-67, makes use of furnace thermocouples which have a time constant of about 6 min at 800 C, as determined in one of the small furnaces at the NBS and under prevailing convection conditions. Thermocouples of this type always were used by Ingberg at NBS during his building fire studies. As a result, his measurement techniques were consistent and comparisons justified. However, almost all other known building fire temperature measurements have been made with either bare thermocouples or short-time-constant shielded thermocouples which much more closely followed actual fire conditions. The lag in temperature indication under the transient heating conditions in the first 10 to 15 min of a fire may be very significant. When the ASTM Methods E 119 thermocouples are used a difference of 100 C (180 F) or more should be expected. Future experimental building fire tests should make use of sufficient ASTM Methods E 119 thermocouples to ensure that some confidence can be achieved in comparing results with furnace test conditions.

The irradiance measured at a distance from openings in the shell of a building during a fire is of great importance in defining severity with respect to ignition and spread of fire to neighboring structures. A good



start has been made in understanding factors influencing such irradiance behavior. However most of the work has been done under conditions involving only a single floor or building compartment. The effects of winds and vertical ventilation of the building are suggested by the work in Canada [28]. More work in this area is necessary, together with investigation of the influence of compartment properties and fuel display variables on the radiant severity of the fire. In making such studies research workers must remember that the radiometers used in many cases will not be looking at a plane object. The configuration factors involved with the three dimensional flames may differ greatly depending on the viewing direction.

Conclusions

This review of the developing understanding of fires in compartments or buildings has indicated some of the complexity of the problem. It suggests the following conclusions:

1. The occupancy fire load itself, though important, does not always exert controlling influence on the character of the fire that may develop.

2. The geometry and arrangement of fuel display may exert a strong influence on the character of fire behavior.

3. The availability of an oxidizer near the fuel emphasizes the major role played by ventilation in influencing fire severity.

4. It follows that fuel, fuel display variables, and character of ventilation together exert a controlling influence on fire behavior.

5. There is increased awareness of the importance of relating fuel load to a ventilation parameter when estimating the character of fire which could develop in a particular building.

6. Fire severity is not a well-defined term since the damage done by a fire is dependent on the thermophysical properties of the exposed materials or constructions. Thus changing trends in building construction involving modification in materials and architectural design features such as window sizes may change significantly the nature of fire damage.

7. Research to date has only started to clarify the importance of fuel display variables, thermal and geometrical properties of the enclosure, and ventilation parameters in afteeting fire behavior. Much more work is needed on these aspects of the problem.

Acknowledgment

This being primarily a review paper, the authors are indebted to the many workers referenced for the data on which the paper is based. Figures 4, 5, 6, and 9 are Crown Copyright and reproduced, by permission of the controller. Her Britannic Majesty's Stationery OflSce.



References [i] Ingberg, S. H. et al, "Fire Tests of Building Columns," Technologic Papers,

National Bureau of Standards, No. 184, April 1921. [2] "Intensity and Duration of Fires," Technical News Bulletin, National Bureau

of Standards, No. 68, Dee. 1922. [3] "Intensity and Duration of Fires in Buildings," Technical News Bulletin, Na

tional Bureau of Standards, No. 114, Oct. 1926, pp. 10-11. [4] "Intensity and Duration of Building Fires," Technical News Bulletin, National

Bureau of Standards, No. 116, Dec. 1926, pp. 5-8. [5] "Bureau of Standards Fire Test Activities," Quarterly, National Fire Protection

Association, Vol. 19, No. 3, Jan. 1926, pp. 234-237. [6] "Report of Committee on Protection of Records," Proceedings, National Fire

Protection Association, May 1927, pp. 295-297. [7] "Fire Tests of OfHce Occupancies at U.S. Bureau of Standards," Quarterly,

National Fire Protection Association, Vol. 20, No. 3, Jan. 1927, pp. 243-252. 18] Ingberg, S. H., "Tests of the Severity of Building Fires," Quarterly, National

Fire Protection Association, Vol. 22, No. 1, July 1928, pp. 43-61. [9] "Fire Resistance Classifications of Building Constructions," Building Materials

and Structures Report 92, Superintendent of Documents, Washington, D.C., Oct. 1942.

[10] Ingberg, S. H., Dunham, J. W., and Thompson, J. P., "Combustible Contents in Buildings," Building Materials and Structures Report 149, National Bureau of Standards, Superintendent of Documents, U.S. Government Printing Office, 1957.

[11] Kawagoe, K., "Fire Behavior in Rooms," Report No. 27, Building Research Institute, Japan, Sept. 1958.

[12] Kawagoe, K. and Sekine T., "Estimation of Fire Temperature-Time Curve in Rooms," Occasional Report No. 11, Building Research Institute, Japan, June 1963.

[13] Kawagoe, K. and Sekine, T., "Estimation of Fire Temperature-Time Curve in Rooms, Second Report," Occasional Report No. 17, Building Research Institute, Japan, March 1964.

[14] Kawagoe, K., "Estimation of Fire Temperature-Time Curve in Rooms, third report," Research Paper No. 29, Building Research Institute, Japan, 29 Oct. 1967.

[15] Thomas, P. H., "Research on Fire Using Models," Institute of Fire Engineers Quarterly, 1961.

[16] Gross, D. and Robertson, A. F., "Experimental Fires in Enclosures," Tenth Symposium (International) on Combustion, Combustion Institute, Pittsburgh, 1965, pp. 931-42.

[17] Thomas, P. H. and Heselden, A. J. M., "Fully-Developed Compartment Fires —Two Kinds of Behavior," Fire Research Technical Paper No. 18, Her Majesty's Stationery Office, 1967.

[18] Odeen, K., "Theoretical Study of Fire Characteristics in Enclosed Spaces," Bulletin No. 10, Division of Building Construction, Royal Institute of Technology, Stockholm, 1963.

[19] Odeen, K., "Experimentellt och Theoretiskt Studium an Brandforlapp i Bygg-nader" (Experimental and Theoretical study of Fire Characteristics in Enclosed Spaces), Report 23/68, National Swedish Institute for Building Research, 1968.

[20] Butcher, E. G., Chitty, T. B., and Ashton, L. A., "The Temperature Attained by Steel in Building Fires," Fire Research Technical Paper No. 15, Her Majesty's Stationery Office, 1966.

[21] Butcher, E. G., Bedford, G. K., and Fardell, P. J., "Further Experiments and Temperatures Reached by Steel in Building Fires," Behavior of Structural Steel in Fire Symposium No. 2, Proceedings of a symposium held at the Fire Research Station, Boieham Wood, Herts, on 24 Jan. 1967, Her Majesty's Stationery Office, 1968, pp. 2-17.



[22] Heselden, A. J. M., "Parameters Determining the Severity of Fire," Behavior of Structural Steel in Fire Symposium No. 2, Proceedings of a symposium held at the Fire Research Station, Boreham Woods, Herts, on 24 Jan. 1967, Her Majesty's Stationery Office, 1968, pp. 20-27.

[23] Law, Margaret, "Radiation from Fires in a Compartment," Fire Research Technical Paper No. 20, Her Majesty's Stationery Office, 1968.

[24] Gross, D., "Experiments on the Burning of Cross Piles of Wood," Journal of Research, National Bureau of Standards, Vol. 66C, No. 2, April-June 1962, pp. 99-105.

[25] Seigel, L. G., "The Severity of Fires in Steel Framed Buildings," Behavior of Structural Steel in Fire Symposium No. 2, Proceedings of a Symposium held at the Fire Research Station, Boreham Wood, Herts, on 24 Jan. 1967, Her Majesty's Stationery Office, 1968, pp. 59-63.

[26] Butcher, E. G. and Law, Margaret, "Comparison Between Furnace Tests and Experimental Fires," Behavior of Structural Steel in Fire Symposium No. 2, Proceedings of a Symposium held at the Fire Research Station, Boreham Wood, Herts, on 24 Jan. 1967, Her Majesty's Stationery Office, 1968, pp. 46-55.

[27] Law, Margaret, "Analysis of Some Results of Experimental Fires," Behavior of Structural Steel in Fire Symposium No. 2, Proceedings of a symposium held at the Fire Research Station, Boreham Wood, Herts, on 24 Jan. 1967, Her Majesty's Stationery Office, 1968, pp. 31-43.

[28] Shorter, G. W. et al, "The St. Lawrence Burns," Quarterly, National Fire Protection Association, Vol. 53, No. 4, April 1958, pp. 300-316.

[29] Gross, D., "Field Burn-Out Tests of Apartment Dwelling Units," Building Science Series, No. 10, National Bureau of Standards, Superintendent of Documents, Washington, D.C., Sept. 1967.

[30] Harmathy, Tibor, "Determining the Temperature History of Concrete Constructions following Fire Exposure," Journal of the American Concrete Institute, Vol. 65, No. 11, Nov. 1968, pp. 959-964.

[31] Cruz, C. R., "Elastic Properties of Concrete at High Temperatures," Journal of the Portland Cement Association, Vol. 8, No. 1, Jan. 1966, pp. 37-45.


STP44709S.27319

Documents

endurance test

actual building fires

materials astm

astm methods

burning building

tests of building construction

astm intl

astm stp