Publisher’s version / Version de l'éditeur:
Vous avez des questions? Nous pouvons vous aider. Pour communiquer directement avec un auteur, consultez la première page de la revue dans laquelle son article a été publié afin de trouver ses coordonnées. Si vous n’arrivez pas à les repérer, communiquez avec nous à PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca.
Questions? Contact the NRC Publications Archive team at
PublicationsArchive-ArchivesPublications@nrc-cnrc.gc.ca. If you wish to email the authors directly, please see the first page of the publication for their contact information.
https://publications-cnrc.canada.ca/fra/droits
L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.
Paper (National Research Council of Canada. Institute for Research in
Construction), 1988
READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE. https://nrc-publications.canada.ca/eng/copyright
NRC Publications Archive Record / Notice des Archives des publications du CNRC : https://nrc-publications.canada.ca/eng/view/object/?id=d5bfbc50-17ab-4691-9e96-358aed14f1f2 https://publications-cnrc.canada.ca/fra/voir/objet/?id=d5bfbc50-17ab-4691-9e96-358aed14f1f2
NRC Publications Archive
Archives des publications du CNRC
This publication could be one of several versions: author’s original, accepted manuscript or the publisher’s version. / La version de cette publication peut être l’une des suivantes : la version prépublication de l’auteur, la version acceptée du manuscrit ou la version de l’éditeur.
For the publisher’s version, please access the DOI link below./ Pour consulter la version de l’éditeur, utilisez le lien DOI ci-dessous.
https://doi.org/10.4224/40001422
Access and use of this website and the material on it are subject to the Terms and Conditions set forth at
How much fire resistance is really needed?
Ser
T H I
National Research
Conseil national
N21a
J(*r(
Council Canada
de recherches Canada
0 .
1588
C , 2
Institute for
lnstitut de
BLDG
'
Research in
recherche en
-- -
Construction
construction
-
How Much Fire Resistance is Really
Needed?
by T.Z. Harmathy
Reprinted from.
Concrete International: Design and Construction
Vol. 10, No. 12, December 1988
p.
40-44
(IRC Paper No. 1588)
NRCC 30238
I,
NRC-
FIST1I R C
L I B R A R Y
(
JUN
19
!?EO
B I B L I O T H ~ Q U E
II R C
CN%!C-
IClST .Ce document prdsente
une
mdthode pennettant de relier le comportement au feu &el des
616ments de construction h celui des essais au feu au moyen du concept de charge
d'incendie normalis&,
grace
auquel le moment
oh
une structure fait ddfaut dans les essais
normalis6s peut
&re relie awc
six
variables d'enee qui d6tenninent le potentiel destructeur
des vrais incendies.
La
technique fournit des informations
sur
la r6sistance au feu
How
Much
Fire
Resistance
9
is Really
Needed.
Presents a method for correlating the performance of building elements in real-world fires and test fires using the concept of normalized heat load, whereby the time at which a structure falls in standard tests can be related to the six input variables that determine the destructive potential of real-world fires. The technique provides information on the fire resistance required for achieving any prescribed level of failure probability.
ow much fire resistance is really needed? An important design problem, and yet this ques- tion is hardly ever asked. The building industry i s highly regulated, and the fire-resistance re- quirements specified in building codes are based mainly on tradition or on rough estimates of the expected fire severities and assumed t o be proportional t o the amount of combustible materials present in the build- ing.
/
Standard fire tests
Whether or not a wall, floor, beam, column, or other building element possesses the required fire resistance is evaluated from standard fire tests, ASTM E 119 in North America. In this test, a representative specimen of the building element is exposed, usually on one side only, to the hot gases of a test furnace. The furnace temperature is controlled to follow a prescribed curve that rises very steeply at the beginning. The fire resis- tance of the element is determined from the time at which the specimen, through structural weakness or ex- cessive heat transmission, ceases to be an effective bar- rier against the spread of fire.
Selecting building elements that fulfill the fire resis-
I tance requirements specified in the building codes is, I
however, no assurance that these building elements will satisfactorily function in real-world building fires. It has been known since the mid-1950s that the amount of combustible materials in a building is only one of sev- eral factors on which the severity (destructive potential) of fires depends. Clearly, specifying fire endurance re- quirements in terms of performance in test fires is jus- tified only if (1) the severity of real-world fires can be described as a function of all significant parameters, and (2) the relationship between the performance of
~ r e s e n t e d z t h e ~ l Symposium on Fire Resistance of Concrete structures, sponsored by Committee 216 Orlando, March 1988.
building elements in real-world fires and standard test fires is understood.
A method for correlating the performance of build- ing elements in real-world fires and test fires is pre- sented here, focusing on design formulas without de- tailing the underlying theories. Once this relationship is known, specifying fire resistance requirements in terms of performance in test fires becomes a straightforward design procedure.
Normalized heat load concept
The normalized heat load concept is a convenient tool for relating the severity of real-world fires to the results of test fires. The essentials of the concept are explained and summarized in Reference 1.
The normalized heat load H (h"R; s"K) is a quanti- fier of the destructive potential of a real-world or test fire. It is defined as
where E (Btu ft-2; J m-2) is the total heat absorbed per unit area by a building element during a fire exposure ) the thermal and &% (Btu ft-2h-"R-'; J m-2s-"K-' is
absorptivity of the surfacing material for the building element (where k is thermal conductivity, p is density, and c is specific heat) on the side exposed to fire.
Design formulas
It has been s h ~ w n ~ , ~ that the normalized heat load for a compartment involved in fire depends primarily on six variables
A, average floor area of the building compart- ments,* ft2 (m2)
A, average surface area of the compartments, ft2 (m2)
h, height of the compartments, ft (m)
thermal absorptivity of the compartment bound- aries, Btu ft-2h-"R-' (J m-2s-"K-') (where k is thermal conductivity, p is density, c is specific heat)
ventilation factor, characteristic of the rate of in- flow of air into the compartment during fire, lb h-' (kg s-I)
Keywords: failure; fire resistance; fires; fire tests; probability; reliability; *A compartment is defined as a building space that is separated most of the ventilation. time from other building spaces by closed door(s).
Table
1
-
Thermal absorptivity of a number
of common building materials
Table
2
-
Information on specific-fire
load,
estimated from Swedish data"
Material Marble Normal-weight concrete Brick Lightweight concrete density: 90.5 lb ft-3 (1450 kg rn-') Plasterboard Vermiculate plaster Wood Mineral wool
L specific fire load (mass of combustibles* per unit floor area), lb f r 2 (kg m-2)
For buildings with compartments of various sizes, the average surface area of the compartments may be esti- mated as
4
&
where A, now means total floor area divided by the number of compartments.
If the boundaries of the compartments are surfaced with different building materials, the thermal absorp- tivity should be looked upon as a surface-averaged value, to be calculated as Btu R-%-HR-l 6.67 6.44 4.46 2.73 2.18 1 .% 1.28 0.25
where the A's with numerical subscripts denote bound- ary surface areas, and the subscripts refer to the var- ious materials forming these surfaces.
The thermal absorptivities of a number of common building materials are listed in Table 1
.'
The ventilation factor and the specific fire load are random variables. In the calculation scheme to be in- troduced, the ventilation factor will be taken into ac- count with its most adverse value, which arises when the ventilation of the burning compartment is mini- mum, i.e., when the inflow of air is not augmented by drafts. This minimum value is
J rn-'s-"K-' 2,270 2,190 1.520 930 740 670 440 90 where
pa density of atmospheric air, = 0.0755 lb ft-3 at 68 F (= 1.21 kg m-3 at 20 C)
A v area of the opening through which the burning compartment is ventilated [usually broken win- dow(~), less frequently open door(s)], ft2 (m2) h v height of the ventilation opening, ft (m)
g gravitational constant = 4.17
x
10s ft h-2 ( ~ 9 . 8 m s-2)Information on the specific fire load, developed from some Swedish data4 is presented in Table 2 for a few
Ib 1
-
O.! 1.' 1.4 Oectlpmcy Dnelling Office Schml Hospid Hoteloccupancies. The arithmetic means L, and the stan- dard deviations a, are listed.
For the mean value of fire load and the most adverse (minimum) value of the ventilation factor, the normal- ized heat load on the compartment boundaries (and on the compartment as a whole) can be expressed by the following semi-empirical equation:+
where - '-erilosd Standard
-
deviation, q 456 Btu lb-' 1.06x
106 J kg-' Ib R-' 6.17 5.08 3.59 5-13 2.99 c 2 =[
0.223 Btu lb-'R-' 935 J kg-'K-' and I kgm-a-
4 - n-3-
30.1 PO 4 24.8 76 6 17.5 35 1H' normalized heat load for a compartment fire (for L ,
= L, and
4
= &,), hsR (s"K)6 semi-empirical factor, dimensionless, as defined by ,
Eq. (6) 25.1 14.6 whichever is less (6) 1.60 7.8 0.86 4.2 - where
The destructive potential of a standard test fire is
1
also quantifiable by the normalized heat load. Since in,
a test fire the furnace temperature follows a prescribed course, the normalized heat load imposed on a building'With non-cellulosic combustibles, multiply the mass of the material by the heat of combustion of the material, and divide it by the heat of combustion of wood.
'The validity of this equation has been verified by a series of compartment burnout experiments consisting of about 25 tests.' C, includes an empirical correction to account for the fact that the experimentally determined normalized heat load values were found to be, on average, 6 percent higher than the calculated values. The coefficient of variation for the error associated with the use of Eq. (5) is 0.101, i.e.. 10.l'per~ent.'~~
element in a test fire is a unique function of the dura- tion of the test, which, if the test is carried on up to the point of failure, is equal to the fire resistance of the building element. The relationship between the fire re- sistance and normalized heat load is
where
and
T fire resistance (or duration of test fire), h
H " normalized heat load on the construction in stan- dard test fire, h%R (sXK)
In Eq. (7), the fire resistance time is expressed in terms of the normalized heat load (rather than the nor- malized heat load in terms of the fire resistance time) for convenience in determining the fire resistance re- quirements. The fire resistance requirement is the value of T for which H"
2
H'. In other words. the value ofFig. 1-Correlation between the factor
P
and the probability of failure in fire P,.cedure~,'.~ the following formulas have been derived9 for the calculation of the fire resistance requirements
where*
H: = H'exp
(PJQ:
+
Q:
+
0.1012) (9) and in Eq. (9)and
In these equations, the d subscripts indicate design val- ues and
0
is a factor (dimensionless), function of the allowed failure probability,P,
(dimensionless), as shown in Fig. 1. The failure probability is, naturally, conditional on flashover.+After deciding on the permissible failure probability
P,,
the fire resistance requirement is calculated from Eq. ( 9 , (6), and (8) through (1 1).r for which the capacity of the building element for
~~~~~l~ of design and discussion
normalized heat load is equal to or larger than the im-posed normalized heat load. Given: A 10-story office building with a floor area per The fire resistance (i.e., the result yielded by a stan- Story of 4,900 ft2 (454 m2). There are 16 compartments dard fire test) is also a random quantity. The ~ o ~ f f i - 0" each story; they are 8.2 ft (2.5 m) tall and have an
cient of variation for T, VT, is somewhere between 0.01 average floor area of 280 ft2 (26 m2). The cornpart-
and 0.15 (iae., 1 and 15 percent),6 dependent mainly on ments will be lined with ~ 0 n ~ e n t i 0 n a l materials: walls
the type of construction.
VT
= 0.1 may be regarded as an all-purpose value.Taking into consideration the random nature of the 'A recent survey of the statistics of the National Fire Incident Reporting System has indicated that, given ignition, the probability that flashover will specific fire load L, and the result of fire test r, as well follow is 21 percent.
as the uncertainty associated with the use of Eq. ( 9 , 'Under the square root sign, 0.101 has been introduced to take into account and employing well-known reliability-based design pro- the error associated with the'use of Eq. (5). Ibid.
-
' _
Table 3
-
Examples of the effect of fire load (as reflected by occu ncy), ventilation (as reflected by window hei htrand permissible failure probability onfin
resistance requirements*Fires are likely to be fuel-surface controlled.
and ceiling with plasterboard and floor with wood par- The
0
factor (Fig. 1): quet. The compartments have two windows, 5 ft (1.52m) tall and 3.33 ft (1.02 m) wide.
8
= 1.64Calculate the fire resistance requirement in such a
way as to have the failure probability less than 0.05 (5 Normalized heat load in a real-world fire [Eq. (5)]: percent), given flashover (i.e., less than 1.05 percent,
given ignition*). (11.0x0.82+ 1.6)x280x5.08
H' = 456
1 , 1 0 9 ~ 1.95+0.223 m 5 x 1 0 4 x 2 8 0 ~ 5.08
Average floor area: = 1,370 hMR (45,680 s"K)
A, = 280 ft2 (26 m2)
61, [Eq. (1011:
Average surface area of the compartments [Eq. (2)]: 1.76
61, = -
A, = 2
x
280+
4x
8 . 2 ~ m = 560+
549 5.08= 1,109 ft2 (103 m2) X 1,109
x
1.95+
0.112J11.5x
104x
280 x 5.081,109
x
1.95+
0.223411.5 x 104x
280x
5.08Average thermal absorptivity of the compartment = 0.248
boundaries [Eq. (3) and Table 11:
61, [Eq. (1 1) with V, = 0.11:
XC
= - [(280+
549)x
2.18+
280x
1.2811,109 61, = 0.9
x
0.1 = 0.09= 1.95 Btu ft-'h-"R-' (665 J m-2s-"K-1)
Design value of normalized heat load [Eq. (9)]: Minimum value of the ventilation factor [Eq. (4) with
Hi = 1,370
x
exp[1.64 J0.24g2+
0.09+
0.1012]p, = 0.0755 lb ft-' (1.21 kg m-3) and g = 4.17
x
lo8 fth-2 (9.8 m ss2)]: = 2,180 h"R (72,600 s"K)
4,,,
= 0.755x
2x
5x
3.33 J4.17x
lo8x
5 Fire resistance requirement [Eq. (8)]: 1= 11.5
x
104 lb h-I (14.5 kg s-I)rd = 0.11
+
5.33x
x
2,180Specific fire load for office buildings (Table 2):
+
14.44x
x
2,1802 = 1.96h >L, = 5.08 lb ft-2 (24.8 kg m-2) Conclusion: Using building elements of 2 h fire resis-
a, = 1.76 lb ft-2 (8.6 kg m-2) tance will insure that the failure probability in fully de- veloped fires will not be more than 5 percent.
The 6 factor [Eq. (6)]: Some further calculations have been performed for
the same building to illustrate the relative importance of 6 = 11.8 J8.23/11.5
x
104 = 0.82 three design parameters: the amount of combustible materials (as reflected by the occupancy), ventilation (as reflected by the window height), and permissible fail-indicate that the amount of combustible materials, once assumed to be of overriding importance in determining the fire resistance requirements, is no more important than the other two parameters.
Poorly ventilated (ventilation controlled) fires are known to be more destructive than well ventilated (fuel- surface controlled) fires. It has been established [using
Eq. (25) in Reference 21 that among the cases analyzed
in Table 3 only two (those marked with asterisks) are likely t o lead t o fuel-surface controlled fires. Since
4. Pettersson, 0.; Magnusson, S. E.; and Thor, J., "Fire Engi-
neering Design of Steel Structures," Bulletin No. 50, Swedish Insti-
tute of Steel Construction, Stockholm, 1976.
5. Mehaffey, J. R., and Harmathy, T. Z., "Thermal Response of
Compartment Boundaries to Fire," Proceedings, 1st International
Symposium on Fire Science, Gaithersburg, 1985, pp. 111-1 18.
6."Repeatability and Reproducibility of Results of ASTM E 119
Fire Tests," Research Report No. RR: E5-1003, ASTM, Philadel-
phia, 1982.
7. Cornell, C. Allin, "A Probability-Based Structural Code," ACI
JOURNAL, Proceedings V. 66, No. 12, Dec. 1969, pp. 974-985.
8. Zahn, J. J., "Reliability-Based Design Procedures for Wood
Structures," Forest Products Journal, V . 27, No. 3, Mar. 1977, pp.
21-28.
9. Harmathy, T. Z., and Mehaffey, J. R., "Design of Buildings for
Prescribed Levels of Structural Fire Safety," Fire Safety: Science and
Engineering, STP-882, ASTM, Philadelphia, 1985, pp. 160-175.
Received and reviewed under Institute publication policies.
1
conditions in modern buildings are usually such as toresult in fuel-surface controlled fires, the fire-resistance member 'Tibor Harmathy, Of Ot-
tawa, Ontario, Canada graduated in me-
requirements shown in the table are, in general, higher
chanical engineering from the Budapest
than one would expect for buildings built during the University of Technology, and received
past few decades. a Doctor of Engineering degree from the
Vienna University of Technology. After
References several years with industrial organiza-
1. Harmathy, T. Z., and Mehaffey, J. R., "The Normalized Heat tions, he joined the National Research
Load Concept and its Use," Fire Safety Journal. V . 12, No. 1 , 1987, council to conduct research on mate-
pp. 221-237.
2. Harmathy, T. Z., "Fire Severity: Basis of Fire Safety Design," rials science and the destructive potential of compartment
Fire Safety of Concrete Structures, SP-80, American Concrete Insti- fires. He was head of the Fire Research Section, Institute
tute, Detroit, 1983, pp. 115-149. for Research in Construction, from 1980 to 1988. He is
3. Mehaffey, J . R., and Harmathy, T. z., "Assessment of Fire chairman of ACI Committee 216, Fire Resistance and Fire
Resistance Requirements," Fire Technology, V. 17, NO. 4, NOV. Protection of Structures. He has published widely and holds
1981, pp. 221-237. five patents.
Authorized Reprint Fran Decenher 1988, issue o f
Concrete International : Design and Construct ion Published by Amrican Concrete Institute
This paper is being distributed in reprint form by the Institute for Research in Construction. A list of building practice and research
publications available from the Institute may be obtained by writing to
the Publications Section, Institute for Research in Construction,
National Reseatch Council of Canada, Ottawa, Ontario, KIA
OR6.
Ce document est distribud sous forme dc tirC-&-part par 1'Institut de recherche en construction. On peut obtenir unc liste des publications de 1'Institut portant sur les techniques ou les recherches en matibre de
bitiment cn 6crivant
B
la Section des publications, Institut de rechercheen construction, Conseil national de recherches du Canada, Ottawa