UDC 699.

81

ORDINARY MEETING
A paper to be presented and discussed at a meeting of The Institution of Structural Engineers, I 1 Upper Belgrave Street, London SWlX 8BY, on
Thursday 24 February 1983, at 6.00p.m.

1
A basis for the design of fire
0

Pr otection of building b

structures
Margaret Law, BSc, FIFireE detection and alarm systems, which are designed to operate only when a
Ove Arup & P a r t n e r s fire starts. The importance of this distinction is also discussed later.

Quantification
Margaret Law is a Technical Director of Ove Arup
Partnership, providing specialist advice on all A fire engineering design system needs to quantify the fire exposure and
aspects offire safety in buildings. She worked at the theeffectsofthatexposureonstructuralbehaviour.Certaincritical
Fire ResearchStation, Borehamwood, for a number
ofyears, carrying out research into many aspects of conditionsforloss of loadbearingcapacity,excessivedeflection, or
fire behaviour and its effects on building materials I excessive heat conduction, can be defined in terms of the attainment of a
and structures and also became involved inthe critical temperature or a given amount of degradation.
application of research results to Building
Regulations, Codes of Practice, and design guides. The quantificationof fire exposure for design purposesis based on the
After a period inthe Directorate of Research
Requirements of the DOE, assessing

the Partnership in 1974.

Synopsis
research

- -&:A
priorities in the field of building and construction,
shejoined the Research and Develooment Grouo of
study of so-called compartment fires. These can be considered to be more
representative of the potential exposure in a real fire than the standard
fire resistance test because they take account more directly
compartment size, and ventilation.

Fire exposure
of fire load,

The present system for grading the fire resistance of structural elements A building fire that is serious enough to need the attention of the fire
gives little scope for engineering design, andthecurrent regulatory brigade can be said to go though three main phases-ignition and growth,
requirements give only a broad indication of the fire safety objectives. full development, and decay. Detectors, alarms, and automatic extinction
Methods of quantifying fire exposure andstructural behaviour in systems, are designed to operate during the growth stage, while the fire is
building fires for design purposes are reviewed, and it is suggested they still small, before the structure is at risk. Large flames, high rates of
could form part of a new approach to designing structural safety. Such an heating, and most structural damage, occur during the fully developed
approach wouldbe more selective andwould take into accountthe period. The decay period begins when the fire is brought under controlor
beneficial effects of fire safety measures which are increasingly required when the fuel (fire load) is exhausted.
for life safety. It is necessaryfor public authorities to identify clearly the In this paper the term ‘real fire’ denotes a fire that occurs in real life in
fire safety objectives of the regulations, while struc!ural engineersshould a building, i.e. not an experimental fire. Laboratory experiments with
propose a formulation of performance requirements which would give representative buildings or structures are constrainedin one or more ways
design methods compatible with structural Codes andwouldhelp to even when they contain realistic fire loads, and they simulate only certain
facilitate cost-effective design. aspects of a real fire. These fires can therefore be called simulated and
form one type of experimental fire. (The term ‘natural fire’is often used
Introduction for what is really a simulated fire.) In general, an experimental fire may
Fires affect the structural performance of buildings because they change be more or less realistic, in terms of scale, fire load, enclosure, mode of
the physical and mechanical properties of the materials of construction. ignition, and means of extinction (if any). For example, the experimental
By defining the importance of these changes, which can be related to the fires in enclosures are usually designed to develop very quickly and the
estimatedfireexposure,astructuralCodecouldprovidemethods of windows are left unglazed. In a real fire the windows would break when
designing a structure to give the prescribed performance in a building fire. and if the fire becomes large, and the growth period could be very long.
At
present,
however,
performance is defined by the authorities The term ‘compartment fire’ is commonly used for the design concept
(regulatory and insurance) as the minimum time for which each element of a definedenclosurecontainingafullydevelopedfireextending
would survive if it were subjected to a standard test, and this test provides throughout the defined space (which may aberoom, a compartment,or a
anarbitraryfireexposure,anarbitraryload,andarbitraryrestraint. whole building).
Themethod of
calculating
the
behaviour of a
Therefore, at the moment, the Codes contain little scope for design and compartment fire is based on measurements in experimental fires using
are constrained to describe ways of achieving the required results in the mainly wood cribs as the fuel.
standard test. It is the purpose of this paper to review the information The standard fireis the one used in standard fire resistance tests andis
available that could form the basisof an alternative method of design for provided by afurnaceusuallygas-fired or oil-fired,controlled in a
the fire protection of building structures and to discuss ways in which a standard way.
more selective approach could be adopted to meet the requirements of
both the public and the private sectors for the protection of life and Compartment fires
property. For structural engineers the fully developed period of a fire is of major
Thisreview is concernedprimarilywiththesafetyofthebuilding importance. It has been studied by a number of workers anda summary is
structure. In order to provide adequate protection for the occupants, for givenbyThomas’.Ananalysisoffully-developedfirebehaviour is
themovablecontents,andfornearbybuildings,othermeasuresare complex, and the models that have been developed are based on a number
needed in addition-notably, control of smoke movement in a building of simplifying assumptions, notably that the temperature distribution is
and access for fire fighting. uniform throughout a fire compartment and that the fuel burns in a
The relevance of structural fire protectionin the overall fire safety plan uniform way. Nevertheless, important parameters have been defined and
is discussed later. Structural fire protectionis a passive measure, in that it a substantial data bank exists from which isitpossible to show how these
is deemed to be ‘always there’, unlike active measures, such as automatic parameters interact.

The Structural EngineerNolume 61AINo. 1lJanuary 1983 25

 Paper: Law

Fire temperature
When the combustible materials are heated and decompose in a fire, the
heatreleasedbytheburningvolatiles is absorbedbytheenclosing
surfaces of the compartment and any other structural surfaces, by the
0
1200 c
surface of the unburnt fuel and by the incoming air. Heat is lost to the
exterior in the flames and hot gases that emerge from the windows and by I C'
radiation through the windows. The rate of heat generation and the way it
is distributed has been measured experimentally. An example of a heat
balance measured in a small compartment is given in Table 12. For this
compartment, unglazed windows provided ventilation from the start of E
the Sire. c

0 1
0 60
TABLE l-Heat balance measured in experimental fires in U
compartment of 29m2 floor area with a fire load of wood cribs

Fig I . Average temperature during fully developed period measured in

Window
Fire HeatHeat loss fromgases (Yo)
e.xperimental fires in compartments' (A, is theareaofwalls,ceiling,
area load release*EffluentStructuralFeedbackWindow
floor; A is the window area; h is the window height)
(m2) (kg) (kcal/s)
surfaces
radiation
fuel
gasto

877 11.2 1900 65 15 l1 9

5.6 1900 52 26 11 11
1744 11.2 3200 61 15 11 13
5.6 2300 53 26 12 9
2.6 1600 47 30 16 7

*From rate of burning (rate of weight loss of fuel)

The table illustrates the significant amount of heat loss in the effluent
gases and shows that, with decreasing window area, a larger proportion O
oftheheatreleased will entertheenclosingstructure.Thetotalheat
released, assuming a complete burnout,is directly proportional to thesize
of the fire load but the rate of heat release may be controlled by the
ventilation. In this example, with the lower fire load, both window areas
givesufficientventilationforthefueltoburnatitsmaximum(free-
burning)ratebut,withthedoubledfireload,theburningrate is not
doubled because the window area restricts the ventilation needed. This
ventilation effect has been described as follows3:
Fig 2. Variation of rate of burning during fully developed period
R = 0 * 1A ( h . . . . (1)
measured in experimental fires in compartments' (R is therateof
burning; A , is the area of walls, ceiling floor; A is the window area)
where
R is the rate of burning (kg/s)
A is the window area (m2)
h is the window height (m) Tf= max. Tf ( l - e - 0-051p) "C . . . .(3)
where y = L, (kg/m2)
The outflow of hot gases is also proportioned to A f h . The area of oq=-a-
structural surfaces to which heatis lost is expressed by ( A , - A ) where A, and L is the
fire
load
(wood)
(kg)
is the area of the planes enclosing the compartment (walls, ceiling, floor).
For a given fire load, compartments with different values of A , , A2nd
, h, The effect of the fire on the structure depends not only on the value of
will haveadifferentheatbalance,andthusthetemperature in the Tfbut also on the duration of heating. The effective fire duration,
T, is
compartment will vary.This is illustratedinFig 1 whichshowshow given by.
temperature varies with q = (At- A)/A\/h. For low values of q (i.e. high
ventilation areas), the rate of heat release is at a maximum but the heat T = L . . . . (4)
- (S)
loss from the window is also large and the resultant temperature is low. R
For high values of q (i.e. low ventilation areas), there is little heat loss to
the outside but the rate of heat release is also small and the resultant Equation (1) suggests that the smaller the value A(h
of the lower the rate
temperature is again low. of burning and the longer the duration. Assuming a complete burnout,
The curve in Fig 1 has been derived from many experimental fires4. For therefore, the effect on the structure tends to be more severe for large
design purposes, it has been defined as follow^^^^: values of q (small A f h )

max. Tf = 6000 (l-e- 0. l?) ("C) . . . .(2) Rate of burning

fi Because R definesnotonlytherateofheatreleasebutalsothe Sire
duration, it has been the subject of much study.
This represents an upper limit of fire temperature for a given q. However, Later work4 based on many experiments showed equation ( 1 ) to be only
if the fire load is low, this value may not be attained. The importance of approximate and demonstrated that R was not simply proportional to
the fire load effect depends on A and A , also, and it can be allowed for as Afh but also depended onA, and the ratio of depth D to width Wof the
follows~: compartment, as illustrated in Fig 2. For design purposes the following

26 The Structural EngineerlVolume 61AINo. 1lJanuary1983

 Paper: Law

equation has been derived5:

R = 0.18 A f h d m (l-e - 0.036~) . .. .(5)

Thisequation is forventilation-controlledfires.Whenthere is ample

ventilation, so that the fuel is free burning, the value of R depends on L
and the type of fuel. For example, domestic furniture has a free-burning
fire duration' of about 20 min, giving T= 1200s and R = L/1200

Temperature-time curves 3 4
The temperatures discussed above are the average measured during the Time - hr
fully developed period of the fire. It is possible to calculate a complete Fig 3. Calculated temperature-time curves f o r compartment fires with
temperature-timecurveusingtheheatbalancemodel,andthemost different fireloads (from reference 8) ( q = 18.8 L/AI; L is the fire load;
comprehensive work has been done in Sweden*. This work assumes that A r is the area of walls, ceiling floor;A is the window area;h is the window
all fires are ventilation controlled, with the simple relationship for rate of height)
burning given by equation (l), and it is assumed that combustion of 1 kg
ofwoodreleases18.8 MJ. Fig 3 showssometypicalcurves.(Inthis
methodthefireload is expressedinrelation to A , as q = 18.8L/A,
MJ/m2.) The curves are calculated for walls, floor and ceiling materials
with 'normal' thermal properties, but it is possible to adapt the methodto
enclosures with thermally different properties.

Fire loads
The amounts of combustible material in buildings have been measured in
a number of survey^^.^.'^. The traditional way of describing fire loadis as
weight or heat units per unit floor area (although,in Sweden, it is related I 1 1 1 I
to A , ) . Usually these materials burn in a similar way to wood, so that, 3 4
when weight is given, it is the amount of wood that is deemed to give an Time - hr
equivalent amount of heat.
Fig 4. Standard temperature-time curve specified f o r the fire resistance
test"

Standard fire TABLE 2-Assumed equivalent severities of building fires (from Postwar
The standard fire exposureis defined as'' building studies no. 20)

T - To = 345 log1,(8t + 1)
Equivalent severity
of fire
in standard test
where (min)
f is the time (min)
Tis the furnace temperature at time I
To is the initial furnace temperature Less than60, i.e. low
fire
load
60
60-120
moderate
i.e. load
fire 120
Thefurnace is controlled so thatthetemperatureofthermocouples load
fire
120-240
high
i.e. 240
adjacent to the exposed surface of the element of construction undergoing
a fire resistance test follows the standard curve shown in Fig 4. The curve *Assuming approx. 8OOO Btu/lb for wood
is virtuallythesamethroughouttheworld,butheattransfertothe
elementcanvaryaccording to thefuelandthefurnacedesign.The
significance of these effects is discussed later.
The British report on fire grading of buildings(1946)12 gives the broad
relationship between standard fire exposure and fire load shown in Table Themeasurementsavailablefromrealbuildingfiresaremainly
2.IttakesintoaccountAmericanmeasurementsofsimulatedfiresin statistical, based on fire brigades' reports. The information includes an
offices
comparison
and
a of the
area
under
the
measured estimateofthefiredurationandtheareadamaged by fire.Inlarge
temperature- time curve with the area under the standard curve. buildings the fires are progressive and tend to grow exponentially:

A , = A. exp (at)
Real fires where
There are many uncertainties in real fires which are not A , is the floor area damaged in time t
taken into account in the compartment fire model: A . is the floor area originally ignited
-the time taken to reach full development may be much longer; a is the fire growth parameter
-the temperature distribution is not likely to be uniform, particularly
inlargespaces, so thattheremaybelocalratesofheating A general value of 4 min ( a = 0-0029 S l ) for doubling time has been
significantly different from the average; suggested13, while a special study of fires in the textile industryI4 gives a
-the firemaybeprogressive, so thatonlypartofthestructure is value of 11 minfor A , to doubleinsize. (a = 0.00105 S - ' . ) This
exposed to the fully developed fire at any one time; exponential type of fire growth means that, if the fire is not tackled early
-the fire brigade tackles the fire before all the fuel is consumed; enough, it may be too big to be controlled and extinguished by the fire
-the fuel itself may not behave like wood. brigade andit will continue to burn until virtually all the fuel
is exhausted.
Such uncertainties do not rule out the use of a design method, and they Such fires may last considerably longer than calculated for compartment
exist with the present grading system, but they are matters that should not fires, and the discrepancy can be accounted for partly by the progressive
be forgotten by the engineer. nature of large fires and partly by the cooling effect of fire hoses which

The Structural EngineeriVolume 61AiNo. 1iJanuary 1983 27

Paper: Law

prolongstheburningtime.Sincethiscoolingeffect is likely tobe

beneficial for the structure,a design based on the compartment fire model
would tend to be conservative.
Specialstudiesofindustrialfireshaveprovidedusefulestimatesof
burningratesandheatoutputperunitfireareaforavarietyof
contents15. These confirm that the wood cribs used in experimental fires
givefiresthatarerealisticandconstituteafuelofreasonablysevere
hazard. Most of the fires studied were well ventilated and free burning,
but one building had a veryhighvalue for v , of about 230m - which
meant that the airflow was insufficient to maintain flaming combustion
throughout the building. However, there was sufficient heat generated
during more than 1 h to heat up the entire building and its contents, so
thatwhen,eventually,airwasadmitted,fireflashedoverthewhole
interior and produceda minor explosion which coincided with the arrival
of the fire brigade.
The fire duration T~ for these industrial fires correlated with fire load
per unit floor area as follows. l I I I
400 800
T~ = 590 ( L ) O +30s (correlation coefficient = 0.62) Temperature - OC
(A,)
where A , is the floor area (m2) Fig 5. Design curves f o r variation of concrete strength with ternperaturgO
(a) Dense concrete (b) Lightweight aggregate concrete
A better correlation was with fire load per unit fire area

T~ = 126 ( L ) 0 * 4 8 s (correlation coefficient = 0-70) Material properties

(-4,) The effects of temperature on the thermal and mechanical properties of
where A,. is the fire area (m2) structural materials have been published in a range of papers, and wide-
rangingsummariesareavailable18,".Fordesignpurposes,simplified
Rate of burning per unit fire area correlated with bulk density of fuelrelationships
as have been devised.
follows:
Concrete
R =
__ 0 -l l e 0'34 (kg/m2.s) For concrete, the compressive strength varies not only with temperature
A, but also witha number of other factors including the rate and duration of
where e is the bulk density = L / ( A , x fuel height) (kg/m3) heating, whether the specimen was loaded or not, the type and size of
aggregate, percentage of cement paste, and the water/cement ratio. In
Studies such as these indicate that it is possible to postulate design fire general, it maybesaidthatconcreteheatedabove300°Closessome
exposures which would correlate reasonably well with experience of real compressivestrength.Fordesignpurposes,compressivestrengthasa
fires. proportionofstrength at normaltemperaturehasbeencodifiedas
illustrated in Fig 5.
The elastic modulus also decreases with temperature; for design, i t has
been suggested that the reduced value at 700°C can be taken as 40% of
External fire exposure the normal temperature value for lightweight concrete and 10% for other
It has been known for a long time that structural elements placed outside concretes. Creep becomes significant at quite low temperatures, but, for
the fapde arelikely t o receive lower fire exposure than internal elements theshort-durationtransientheatingnormallyexperiencedinbuilding
because, although they can be heated by radiation and convection from fires, is nottakenintoaccount in design.Thecoefficient of thermal
the emerging flames and by radiation from the windows, they are free to expansion is of order 10 ?C andvarieswithaggregate.Designvalues
cool to the ambient air. No satisfactory relationship with the standard fire and methods of estimating thermal deformation are given in references
exposure can be established and therefore this is an aspect of structural 20. 21, and 22.
fire safety that has been an obvious candidate for a design approach. For transientheating,animportantthermalproperty is the thermal
The discussion earlier of heat balance showed the importance of heat diffusivity k where
loss in the emerging flames and hot gases and the window radiation. Since
it is theinternalfirethatcontrolstheexternalfireexposure,theheat k = - A (m2/s)
balance approach is equally relevant. It has been shown5 that the heightz
ec
of the flame tip of the emerging flame above the top of the window is A is the thermal conductivity (W/m"C)
given by e is the density (kg/m3)
C is the specific heat (J/kg"C)
z + h = 12.8 ( E ) 2 / 3
( W ) The value ofk depends on the aggregate, porosity, and moisture content.
where W is the total width of windows For gravel aggregate concrete, it varies from about 10- h m2/s at normal
Thetemperature at theflametip is about540°C.Thetemperature temperatures to about4 x lo--' m'/s at 700°C. For lightweight concrete,
distribution along the flame axis is linear and given by: it varies fromabout 7 x 1 0 - t o 3 x 10overrhesametemperature
range2". These low values of diffusivity mean that, in building fires, the
T, - T, = 1 - 0 . 0 2 7 1 ~ 300°Ccontour,beyondwhichtheconcreteretainsvirtuallyallits
Ti0 - T, R strength, may be at only a small depth below the heated face.

where Steel
I is the distance along flame axis from window plane (m) The effect of temperature on yield strength varies with the type of steel.
T i s the temperature at distance I ("c) For mild steel, it is reduced to half at about 500°C. A design curve for
Ti, is the temperature rise at I = 0 ("C) effective yield stress of mild steel is shown in Fig 6". Similar curves are
availableforvariousreinforcingandprestressing steels"). The stress-
Since Ti - T, = 520°C at theflametip, Ti(, canbederived.Design induced strains of steel start to be affected by creep at temperatures over
manuals have been published16. l 7 which show how to estimate the rates about 450°C. However, for the heating rates normally encountered in
of heating at different positions within and adjacent to theemerging building fires, these design curves can be assumed to include the effect of
flames. creep. Thus it is not normally necessary to take into account the stress and

28 The Structural EngineerlVolume 61A/No. 1IJanuary 1983.

 Paper: Law

temperature history of the structure. The elastic modulus decreases with

temperature;
because at elevated
temperaturesthe
stress-strain
relationship follows a curve, the recommended values for design apply to
the tangent modulus at infinitely small stress23. The coefficient of thermal
expansionvarieswithtemperature,butanapproximatevaluecanbe
taken as 1 a 4 x 10 -s//"C, which is similar to the value for concrete.
The thermal diffusivity of steel is significantly greater than that for
concrete and, except for very massive sections, unprotected steel rapidly
heats up in a fire. The specific heat is a function of temperature but an
approximate value can be taken as 520 J/kg"C, which is about half the
value for concrete.

Masonry
The thermal properties of concrete bricks and blocks are similar to those
of concrete having similar constituents and mix designIY. The properties
of clay bricks are not available in such detail as for concrete. A rise in
temperaturegives a decreasein
compressive
strengthand
causes
expansion. Values of thermal conductivity depend on density and increase
somewhatwithtemperature.Forexample,forbrickofdensity2100
\ \

\
kg/m3,the
conductivity
increases
from 0.93 W/m"C at
normal \
temperatureto 1e07 W/m"C at 700°C'8.Specificheatvarieswith '\
temperature but an approximate value is 1000 J/Kg "C, similar to the 'l
,
value for concrete. I I I
400 800
Wood Temperature - 'C
The tensile andcompressivestrengthsofwoodalsodependonits
temperature. However, because of the insulation provided by the charred Design curve f o r variation of effective steel strength with
surface layers, the uncharred portion of wood in a building fire may be
assumedtoretainvirtuallyallitsnormaltemperaturestrength.For
design,anominalreductionof10%inthestrengthoftheuncharred
section has been suggested19. The rate of charring, which is the major
feature of interest, depends to some extent on species but mainly on the
rate of heating24. For rates of heating, I, between 20-3300 kW/m2, the
charring rate25 has been found to be3 e 6 7 x 10 - 4Imm/s. Measurements
of char depths in timber beams subjected to the standard fire resistance
test showed a char depth of 18mm after 30 min exposure which suggests
there was a heating rate of about 30 kW/m2 in these standard furnace
tests26. Notional charring rates can be used to design for standard fire
resistance ratings2'-

Heat transmission
Theheattransferfromthefiretothestructurecanbecalculatedas
follows:
I = (ac + or) (TI - T 3 ) (W/m2)

where
T, is the ambient gas temperature at time t ("C)
T 5 is the surface temperature at time t ("C)

ac is the heat transfer coefficient for convection (W/m2"C) t I l 1

ar id the heat transfer coefficient for radiation (W/m 2oC) 1000 2000
1

and f r is the resultant emissivity of flames, combustion gases Fig 7. Time f o r average temperature of protected steel to reach 550°C in
the standard fire resistance test (derived fromreference 23) (A is the
where e r = 1 thermalconductivityofinsulatingmaterial; d is thethicknessof
l/El + l / e s -1 insulating material; P, is the perimeter of steel; A s is the cross-sectional
area of steel)
where E, is theemissivityoftheflameand is theemissivityofthe
surface
theconcretecanbecalculatedor,forheatinginfireresistancetests,
Tt can be taken from the temperature-time curve. Values of25 W/m2"C obtained from charts20. With steel elements, it is assumed that the steel
for ac and 0.85 for have been suggested for compartment fires, while ac temperature is uniform across the section,so that the steel is acting purely
= 25 W/m2"C and e t = 0 . 5 7 are suggested for standard fire resistance as a thermal capacity, while the insulating material is acting as the thermal
tests8. As a conservativeapproximation, it canbeassumedthatthe resistance. Equations and charts have been devised for the standard fire
surface temperature of the exposed structure follows the temperature- resistance test and for compartment fires8> 28 which take into account the
time curve for the fire. mass and exposed surface area of the steel sections as well as the fire
Heat transfer within the structure can be calculated using standard heat exposure. Fig 7 illustrates this approach.
conductionequations
and
certain
simplifying assumptions.
With A relationship between the fire exposure in the standard fire resistance
concrete,forexample,wherethecriticalcondition is usuallythe test and in compartment fires has been developed in terms of attainment
temperature of the reinforcement, it can be assumed that theissteel of the of a certaincriticaltemperaturewithinanelement of structure.For
same temperature as the adjacent concrete. Temperature gradients within protected steel elements and reinforced concrete elements, where failure is

The Structural EngineerlVolume 61A/No. IlJanuary 1983 29

Paper: Law

due t o the steel reaching a critical temperature, the relationship is given and use the results of fire resistance tests, such as they are, plus whatever
by'Y: research data are available, to check his assumptions.
Design guidance is mainly confined t o ways of passing fire resistance
te = K L, (min) . . . .(6) tests; since these are carried out on single elements, there is much less
d A ( A t- A ) guidanceforassessingthebehaviourofframeworksandmultistorey
structures.Guidance for
frameworks has been basedmainly
on
where te is the equivalent fire exposure in the standard test and K depends calculations. However, there is a particular difficulty in calculating fire
on the furnace design and is of order unity. resistance in this country because the conditions for the reloadtest are n o t
Thus the temperature attained at a critical part of an element exposed defined: there is not a prescribed cooling curve (timber elements are hosed
to a compartment fire can be assessed where measurements are available down to prevent further charring but the other elements are not). For
from a standard test of duration te on the same element. practical reasons, the load is maintained on some elements through the
A design method for the estimation of heat transfer to external fire 24-h cooling period, while it is removed from other elements at the end of
exposed elements is a l s o a ~ a i l a b l e ' ~ . ' ~ . the test and reapplied 24 h later.I t has not been found possiblet o measure
the usefulness of this test, which is probably why i t is not used in other
Mechanical behaviour countries.
Mostoftheinformationthathas been
available to measurethe
mechanical behaviour of structures exposed to fire is provided by the Design guidance
results of standard fire resistance tests of elements of construction. I t is Interim guidance'() for concrete construction stresses the importance of
important, therefore, to understand how thetest is carried out and what is good detailing of elements which, together with suitable proportioning of
measured. the members, will meet the objectives of satisfactory performance for u p
to 2-h fire resistance without being unduly sensitive t o the actual rates of
Fire resistance tests heating that might be encountered in a real fire.
A loadbearing element of construction can fail in a standard test under Forlargerperiods o f fireresistance, i t canbeworthwhileusinga
any one of three criteria method of analysis based onlimit state principles, taking into account the
-stability effects of temperaturerise on the material properties. LMethods of analysis
-integrity have been developed for flexural members (beams and slabs) where the
-insulation mode of failure may be in flexure, shear, or combined flexure and shear.
Stability failure is failure to support the test load or, f o r floors, flat roofs, Analysis is carried out after the temperature of the tensile steel and the
and beams, a deflection exceeding span/30. Integrity failure for floors average temperature of the concrete
has been estimated and
the
and walls is the formation of significant cracks or other openings through corresponding reductions in strength have been determined. Met hods of
which flame or hot gases pass. Insulation failure for floors and walls, is analysis are available for statically determinate members (rarein practice
an average temperature rise of 140°C on the unexposed face or a local but forming a basis for analytical treatment), for members monolithically
temperature rise of 180°C. The fire resistance is the duration of the test built into supporting structures but not fully continuous (the most likely
period or the time t o failure under one of these criteria, whichever is condition for members designed as simply supported), and f o r continuous
shorter, except that, for stability, the fire resistance is reduced by 20% if structures. There are certain circumstances when the flcxural action of a
the element fails t o support the load 24 h later. (This 'reload' test is not beamorslabcanbereplaced by membraneaction,and it has been
applied in other countries.) It is interesting to note that a wall or floor suggested that a deflection of span/20 may be acceptable30. u p to now,
could fail on insulation or integrity while still supporting the load. there has been less success in devising analytical methods for the design of
Theseparticularcriteriaforfailurehavebeen in existenceforsome compression members-walls and columns-and reliance must be based
time. The possibility of changes to the deflection criterion is frequently on experience from fire resistance tests and good detailing. The majority
discussed, some support being voiced for a limiting rate of defle~ti~jn of methods developed so far are directed towards achieving certain levels
rather than a limiting deflection, while it has beensuggestedthat,for of fire resistance but, in principle, the same approach could be adopted
someconstructions,adeflectionofasmuchasspan/20wouldbe forcompartment fire
exposure
and extended t o frameworks and
acceptable.Theinsulationcriterion is rare!y discussed,althoughits multistorey struc:ures'".
origins are obscure; it may have been intended to prevent fire spreading No discussionofconcretebehaviour in fire is completewithouta
by conduction to papers. The author has not been able to find an example mention of spalling. The usual approach is to detail the element to avoid
of a real fire where there was fire spread by conduction through a fire- spalling but there is is no doubt that, if design rules could be established
resistant wall. that showed that the element would perform adequately even if spalling
The load appliedis required to produce 'stresses of the same nature and occurred, this would be a great improvement.
ofthesameorderofmagnitudeaswouldbeproduced at normal Design rules for steel elementshave been devised for estimating the
temperatures in the full-size element by the maximum permissible loads critical steel temperature for columns and beams'3, and the heat transfer
which the element is capable of carrying when designed in accordance tothe steel forboththestandardfireandcompartmentfirecanbe
with the requirements of the appropriate British Standards or Codes of determined8. 23. Calculating the
fire
resistance
generally gives
Practice;loadsotherthanthesemaybeappliedonlybyagreement conservative results: one reason is that characteristic values of initial steel
between the sponsor of the test and the testing laboratory. The load shall strength are used in calculations, whereas the test element is a random
be maintained constant throughout the test period'. Thus it is the reserves sample.However, correction
factorshave been devised3'. Some
of strength designed to withstand extreme loads at normal temperature information is availableforframeworksandaspecialstudy o f fire-
that are exploited for fire safety. The vertical edges of loadbearing walls exposed portal frames has been published3'.
are tested free from restraint. Floors, roofs, and beams, are supported Guidance for timber is mainly related to good detailing and nominal
and restrained in ways that 'as far as possible are similar to those which rates of charring27, while for masonry the information is mainly in the
would be applied in service'; when these cannot be defined, the elements formoftablesofconstructionsdeemed-to-satisfythevariousfire
are tested simply supported. resistance requirements.
There is some variation in results between test houses, thought to be
due to variations in modes o f heating, loading, and support conditions. Behaviour in real fires
Great efforts are being madeto achieve more uniform results, although i t The most familiar structural failures are those experiencedin single-storey
mightseem more useful t o attempt to definewhat is beingmeasured. unprotectedsteel-framedbuildingssubjectedtosevereormoderately
However, the purpose of the test is to form the basis of the grading system severe fires-a not unexpected result, since unprotectedsteel generally has
used by regulators; it is not intended t o be used by engineers for design a low fire-resistance. However, once designed to provide a given period of
purposes. Thus, the initial properties of the structural materials are not fireresistance, all structuralmaterialsappear t o performequally well
measured; the changes that take place when the temperatures within the during real fires. I t is rare that a properly detailed building structure,built
element increase, are not monitored in a way to give design information; t o comply with the regulations, suffers a major collapse during real a fire,
the desire to satisfy the reload test means that the test may be terminated although of course i t may need extensive repair o r even replacement if the
beforethemodeoffailurebecomesclear.Therefore,atpresent,the firehasbeen very severe. A concretestructuremaysufferextensive
design engineer needs to postulate a model for structural behaviourin fire spalling and some deflection but still serves its purpose during the fire.

30 'The Structural EngineerlVolume GlA/No. 1IJanuary 1983

 Paper: Law

The effects of expansion may cause serious damage (e.g. in some long- TABLE 3-Some comparative risks of fatality
lasting fires where concrete slabs have been heated to a significant depth
and caused damage remote from the area where the fire occurred33) but
Activity
Fatal
accidentrate*
person
108h
exposure
major failure due to expansion of a fire-protected structure is rare while
the fire is in progress. In general, the performance of building structures
as loadbearing elements and barriers to fire spread is good. (Where fire Staying at home-average
breaches a wall or floor designed as a fire barrier, it is sickness) usually through (excluding 3
gaps such as open doors or unprotected ducts.) This would suggest person
able-bodied
that for Total l
the current fire grading regulations may be about (UK) car probably by Travelling
right but are 57
conservative. (USA) car by Travelling 95
660 (UK) Motor cycling
Fire safety measures (UK) hotels in Fires 1
Structural fire protection is only0.1 (UK)
one part of the dwellings
package in
of fire safety Fires
measures used in a building. There are two broad groups of measures: (USA) aviation
Scheduled
flights 240
-fire prevention, designed to reduce the chance of a(USA) disease for Average
fire occurring; 110
0.01 designed to mitigate the effects
-fire protection, disaster
of a fire should it Natural
nevertheless occur.
Fireprotectionmeasuresmaybepassive or active.Passivemeasures *In these units a mean lifetime of 70 years corresponds to 160.
include structural fire protection, layout of escape routes, fire brigade 70 years = 6. l x 105h.108/6.1 x lo5 = 160
accessroutes,andcontrolofcombustiblematerialsofconstruction.
Active measures include detection and alarm, fire extinction, and smoke
control, all of which may be operated manually or automatically. When
deciding what structural performanceis needed, it would seem reasonable building the potential for large life-lossis greater. Therefore, to maintain
to take into account special measures designedto reduce the chance ofa public confidence the national authorities have two main objectives:
fire occurring, or measures designed to reduce the fire severity. -maintenance of a low average risk per person, which means paying
Earlydetectionleadingtoearlyfire-fightingdecreasestheriskthat attention to dwelling fires;
there will be a large fire. Automatic detection and extinction measures -reduction of the risk of fire disasters that will cause large life-loss or
decrease the risk that there will be a large fire. Although such automatic economic loss, or both42.
devicesweredevelopedoriginally toprotectbuildingcontents,they Recentstudiesindicatethat,because so higha proportionoffire
indirectly protect the structure as well. In recent years, the combination of casualtiesindwellingsresultsfrompersonalincapacityorbehavioural
automatic sprinklers and a designed smoke-control system has been used factors, there is little scope for addition to the Building Regulations that
to protect people escaping from fire in large buildings”. Since there are couldfurtherinfluencelifesafetyindwellings.Inthecaseofother
methods and data available for assessing the effectiveness of automatic occupied buildings, the number of casualties by type of building is too
fire
detectorsand sprinkler^'^. j6, there is scope for
explicitly few to enable firm conclusions to be drawn38.
recognising their beneficial effects in the determination of the level of If we examine the origins of regulations for structural fire protection
structural fire protection needed. and control of building materials,we see that the regulatory requirements
Atpresent in regulationsthelimitof7000m3compartment size for have arisen from the need to allay public concern, and this is done by
shops is doubled when there is an automatic sprinkler installation. Thisis reducing the risk of major fire spread in cities and by protecting a person
a recognition that sprinklers provide fire containment. In London, all and his propertyfromthefoolishactsofhisneighbours.Thusthe
largebuildingsotherthanflatsandmaisonettesarerequiredtohave requirementsareessentiallydesignedtoprovidefirecontainment-by
automatic sprinkler installations, except for very small compartments, compartmentationandwidestreets-andalsotoprovidestabilityof
and in return some relaxation of smoke-control measures and of water majorstructuralcomponents.Theinsurancecompanies,concerned to
pressure in rising mains may be accepted3’. There is, as yet, no explicit reduce the maximum likely loss of a building structure or a building’s
recognitionthatextrastructuralsafetycouldbeprovidedbyactive contents, or both, also require containment and structural stability. In
measures as a substitute for an increase in structural fire protection. It is addition, they have placed major emphasis on the use of fire extinction
interesting to note that, from a national point of view, a Home Office measures: many of the public fire brigades were originally established by
studyshowsthatsome90%ofindustrial floor spacecouldbecost- theinsurancecompanies,whoalsopioneeredtheuseofautomatic
effectivelycoveredbysprinklers and25%byautomaticdetectors. detectors and sprinkler systems for property protection.
Sprinklersshouldbecosteffectiveinthelargershops,butnotin It is difficult to justify many of theexisting regulations on the grounds
hospitals, offices, schools, public houses, and restaurants, or in all but that they save life but they d o have the effect of reducing property loss.
the largest storage buildings (because the probability of fire occurring is
For the individual owner, of course, there may be benefits but, Levels of grading
from a regulatory point of view, where there is no benefit in relation to The level of fire protection required depends ona number of factors that
firelosses,itwouldseem thatrequirementsforsprinklersshouldbe are not all stated explicitly. The 1946 Report onfire grading of buildingd2
justified either because they are needed for life safety or because they formed the starting point for current regulations. It considered fire load
result in savings elsewhere. to be an important factor, as illustrated by Table 2. This relationship, in
terms of fire exposure, can be expressed as:
Fire grading
The present regulatory system is based on grading of elements in the te = -L (min)
standardfire-resistancetest.Thistestdoesnotrepresentthefire A,
exposure, the loading (floor loads, structure, wind, etc.) or the restraints
experienced in real fires, but thisis not necessarily a drawback if it can be
whereas, from the studies of compartment fires, equation (6) suggests:
assumed that the test ranks performance of the elements in the same order
and degree as their performance ina real fire when they form part of the
t = L (min)
building structure. A regulatory system can then select the grading needed
accordingtothecircumstancesandtheobjectives.It is thelackof i A ( A , - A )
definition of the circumstances and the objectives that provides much of
the dissatisfaction with the present system. The level of grading chosenis not necessarily the equivalent fire exposure,
however, because it also includes allowances for uncertainties and the
Objectives of the regulations consequences of failure. Another matter of concern in the report was the
The level of fire risk to anindividual person is fortunately low compared need to reduce property loss, and this resulted in recommended limits to
with other risks, as illustrated in Table 339, 40, 41. compartment size.
Most of the life loss caused by fire involves few people in any one The selection of three levels of fire load, as in Table 2, can be discerned
incident, and these losses occur mainly in dwellings. However, in a large in the current regulations where residential, office and assembly fall in the

The Structural EngineedVolume 61A/No. 1IJanuary 1983 31

Paper: Law

low group, shop and factory in the moderate group, storage and general performance required. While it is the responsibility of the authorities to
inthehigh.Inaddition,thedetailedgradingandcompartmentation set the levels required, these levels should be expressed in a way that is
requirementsvaryaccordingtotheheight,areaandvolumeofthe suited to a design approach andgives design methods that are compatible
building and vary again according to the building's use. It is not easy to with the structural Codes. Thus the structural engineer needs to take the
explain the reasons behind all of these differences, although some guesses initiative if the changes are to be to his benefit.
can be made. Whether the performanceis defined ona probabilistic basis, by various
It would seem reasonable for grading to increase with height in order to limit states, or by a modification of the present grading system, there are
have increased structural safety for tall buildings, when the consequences certainaspectstobeincludedwhichwouldinvolveamoreselective
of failure could be serious. There are also requirements that are related to approach and, formulated properly, would give a system more responsive
the height of a fire brigade ladder, since, apparently, it was assumed that to changes in fire behaviour and building uses in the future.
abovethisheight a personcouldnotberescuedandneededextra A definition of objectives could include one or morc of the following:
protection. However, since escape routes must now be designed so that
people can escape unaided, some reappraisal may be needed. -protection of people while they escape
The increase of fire grading with volume is presumably intended to take -protection of people who cannot escape readily
account of fire severity increasing with compartment size, but it can give -protection of property within buildings
anomalies because, for a given floor area, the designer can be penalised -protection of access for fire-fighters
for providinga high ceiling. This enclosure of extra space does not add to -protection of adjacent buildings from fire spread
the fire load and, indeed, from heat balance considerations, an increase in -protection of adjacent buildings from structural collapse
wall surface area could cool the fire. The floor area would be a better
guide to thesize of the fire load and the number of people who could beIt inwould then be possible to define which elements need to have structural
the building. fireprotectiontomeettheseobjectivesandtobeselectiveaboutthe
The need to protect fire-fighters may be taken into account, but is not
it performance required. It might be that only the escape route structure
clear how this is done in the fire gradings required. needs protection. For a tall building, once people had escaped, the sole
The regulations make a distinction between dwellings or institutions, objectivemightbe to preventcollapseofthebuildingon to its
where there.is a so-called sleeping risk or there are disabled people, and neighbours; thus the loadbearing capacity of the walls and floors could be
buildings where people work or play and can be expected to be able of more importance than their compartment properties and a reduced
bodied.Thus,extraprotectionisgiventopeoplewhodonotescape standard of insulation and integrity would be acceptable.
because they are asleep or disabled and may have to stay in the building On the basis of the work described, a burnout fire could be defined,
until the fire is exhausted. using agreed values of fire load per unit floor area and the compartment
The categorisation of buildings according to fire load is a reasonable characteristics.Smallbuildingsandsingle-storeybuildingswouldnot
approach, based on the total burnout principle; in practice, the treatment necessarily need to withstand a burnout, but for many larger buildings it
is broad brush, grouping together, for example, carparks with a low fire might be thought necessary to have a notional value of fire resistance to
loadandwarehousesthatmayhave a veryhighfireloadunderthe give Some protection, even if all other safety measures fail. For these
heading of storage. buildingsitwouldprobablybeconsideredthatthemainelementsof
At the moment, there are three sets of regulations in Great Britain with construction should be able to at least withstand a notional burnout and
the same broad objectives but with the requirements differing in detail. that, in tall or important buildings, they should surviveby a comfortable
Thus, it ought,to be possible to calibrate one set of regulations against margin. Once the building had been designed to survive a burnout, the
another and assess the importance of the various rules. However, there is extra margin of safety would not necessarily be provided by increasing the
n o evidence available that buildings in England and Wales, or London or structural fire protection; other protection measures, such as automatic
Scotland,performsignificantlydifferentlyinrealfires.It is likely, sprinklers or automatic detectors, could be used. The benefits of such
therefore, that all the regulations are on the conservative side. This is active measures would therefore need to be quantified.
encouraging, since it indicates that there is scope for change and that a Thepresentgradesoffireresistancerequiredbytheregulatory
more analytical approach would yield some benefit without prejudice to authorities
must
include,
though
not
explicitly,
allowances for
meeting the fire safety objectives, In any new approach, however, the fire uncertainties in loading, material properties, fire behaviour, fire-fighting
safety objectives need to be stated explicitly. andfireresistancetesting.Suchuncertaintiesremainedunquestioned
until attempts were made to introduce calculation methods. Defining all
New approach to fire safety theuncertaintiesmaybedifficult,butthere is scopeforusingthe
During recent decades, it has become clear that the tradional approach to statistical data available and, as a first step, statistical models could be
fire safetyis not enough on its own. With the higher standards living
of in calibrated
againstthe
existing
gradingsystem.
The standard
fire
thedevelopedcountriescomedemands,bytheconsumer,forhigher resistance test itself, if it is to be extended to be a design tool, should be
standards of personal safety. The changes in furnishings, the increased studied by engineers and modified as necessary.
numbersofhigh-risebuildingsandlargeassemblybuildings,have Thereareadvantagesinamoreselectiveapproachnotonlytothe
directed the attention of the public authorities to life safety provisions, indesigner and client but also to the authorities. By having defined fire
particular to automatic detection and extinction systems, smoke control loads for different uses, for example, the building design can respond
systems, control of furnishings, and efforts to improve fire prevention. more quickly than at present to significant changes in amount and type of
These changes in approach to the design of fire safety have at least fire load and the introduction of new hazards (or the withdrawal of old
three major implications for building structures: ones). The definition of objectives and the performance needed would
encourage the production of buildings that are more fit for the purpose of
--If life safety is assured by other measures, in what circumstances fire safety. The protection needed for the fire-fighters would be identified
should structural fire protection also be provided? more clearly and their major role as protectors of property would be
--If measuresareinstalledthatreducethechanceof a largefire recogni~ed'~.
occurring,canthepresent levels ofstructuralfireprotectionbe It is oftensuggestedthat a structuralfireprotectionCodeshould
reduced or eliminated? include design for repairability. This would be, in principle, a departure
-What is beingspentonfireprotection?Whichcombinationof from the usual approach, which is to design for an ultimate limit state,
measures will give the optimum return on investment in terms of life and it is difficult to see howit would be done in practice. A separate Code
and property saved? for the assessment and repair of fire-damaged buildings would, however,
be very useful.
If questions such as these are to be answered fully (whether by engineers
or politicians), the risks, the benefits, the potential fire exposure, and the Conclusions
effectiveness of
the
safety
measures, must be quantified.
Much There is sufficient information available to produce design models for
information is available, but even if these factors cannot all be quantified structuralbehaviourinbuildingfires.Thedefinitionofperformance
as yet, it is still important that the questions be asked. requirement depends on a clear definition of the fire safety objectives and
On the assumption that a physical model for assessing fire exposure quantification of the safety levels required. Although this identification
and structural behaviour exists, the designer then needs a definition of the andquantification is ultimately the
responsibility
of
the
public

32 The Structural EngineerlVolume 61A/No. l/January 1983

 Paper: LawLibrary additions

authorities, the engineer should take the initiative so that any changes in Structural Engineers and the Concrete Society, London, Institution
the present system facilitate methods of design that are compatible with of Structural Engineers, 1978
structural Codes and give cost-effective solutions. If, as seems likely, the 21. ‘FIP/CEB report on methods of assessment for the fire resistance of
standard fire resistancetest is to be retained, it should be modified giveto concretestructural
members’,Wexham Springs,
Cementand
more design information to the engineer. Concrete Association, 1978
22. ‘Design of concrete structure for fire resistance. Preliminary draft of
an appendix to the CEB-FIP model Code; Bulletin d’1nformation
no. 145, Paris, Comite Euro-International du Beton, 1982
References of safety fire the 23.recommendations
‘European
for steel structures
1 . Thomas, P. H.: ‘Fires in model rooms: CIB research programmes’, Level 1 . Calculation of the fire resistance of load bearing elements
BRE Current Paper CP 32/74, Borehamwood, 1974 and structural assemblies exposed to the standard fire’, European
2. Heselden, A. J. M.: ‘Parameters determining the severity of fire’, Convention
Constructional
for Steelwork (in press)
the
Symposium no. 2 Behaviour of structural steel in fire, London, 24. Thomas, P. H., Simms, D. L., and Law, Margaret: ‘The rate of
HMSO, 1968 burning of wood’, Joint Fire Research Organisation Fire Research
3. Fujita, K.: ‘Characteristics of fire inside a non-combustible room Note 657, Borehamwood, 1967
and prevention of fire damage’, Japanese Ministry of Construction 25. Butler, C. P.: ‘Notes on charring rates in wood’, Joint Fire Research
Building Research Institute, Report 2(2), Tokyo Organisation Fire Research Note 896, Borehamwood, 1971
4. Heselden, A. J. M., and Thomas, P. H.: ‘Fully developed fires in 26. Lawson, D. I . , Webster, C. T., and Ashton, L. A.: Fire endurance
single compartments’(CIBReportno. 20), Joint Fire Research of timber beams and floors, National Building StudiesBulletin No.
Organisation Fire Research Note 923, Borehamwood, 1972 13, London, HMSO, 1951
5 . Ove Arup & Partners: ‘Design guide for fire safetyof bare exterior 27. BS5268: Code of practice f o r the structural use of timber: Part 4:
structuralsteel.TechnicalReports’,AmericanIronand SteeI Fire resistance of timber structures, London,BritishStandards
Institute, Constrado, London, 1977 Institution, 1978
6. Law,Margaret:‘Firesafety of externalbuildingelements-the 28. Law, Margaret: ‘Nomograms for the fire protection of structural
design approach’, Engineering Journal, American Institute of Steel steelwork’, Fire Prevention Science and Technology, No. 3,
Construction, Second Quarter, 1978, pp 59-74 pp 19-27, 1972
7. Theobald, C. R., and Heselden, A. J. M.: ‘Fully developed fires 29. Law,Margaret:‘Prediction of fireresistance’,paperno. 2 of
with furniture in a compartment’, Joint Fire Research Organisation Symposium no. 5: Fireresistance requirements for buildings-a new
Fire Research Note 718, Borehamwood 1968 approach, London, HMSO, 1973
8. Pettersson, O., Magnusson,S-E,andThor,J.: Fire engineering 30. Fire resistance of concrete structures, report of a joint committee of
design of steel structures, Publication 50, Swedish Institute of Steel the Institution of Structural Engineers and the Concrete Society,
Construction, Stockholm, 1976 London, the Institution of Structural Engineers, 1975
9. Law,Margaret,andArnault,P.:‘Fireloads,naturalfiresand 31. Petterson, O., and Witteveen,J.:
‘Onthe
fire
resistance
of
standard fires’,International Conference on Planning and Design of structural steel elements derived from fire tests or by calculation’,
Tall Buildings Preprints: Reports Vol. Ib-81972, ASCE- l ABSE Fire Safety Journal, 2 (1979/80), 73-87
10. Fire safety in constructional steelwork, European Convention for 32. The behaviour of steel portal frames in boundary conditions, Fire
Constructional Steelwork, CECM-III-74-2E, 1974 and steel construction, Croydon, Constrado, 1980
1 1 . BS476: Fire tests on building materials and structures: Part 8: Test 33. Kordina, K., Krampf, L., and Seiler, H. F.: ‘An examination of the
methods and criteria f o r the fire resistance of elements of building effects of a big fire in some concrete buildings’, Fire Prevention
construction, London, British Standards Institutions, 1972 Science and Technology, No. 14, pp4-17
12. ‘Fire grading of buildings. Part 1. General principles and structural 34. Morgan, H. P.: ‘Smokecontrolmethods in enclosedshopping
precautions, Postwar building studies no. 20, London, HMSO, 1946 complexes of one or more storeys: a design summary’, BRE report,
13. Baldwin, R., and North, M. A.: ‘The number of sprinkler heads London, HMSO, 1979
opening in fires’, Joint Fire Research Organisation Fire Research 35. Rogers, F. E.: ‘Fire losses and the effect of sprinkler protection of
Note 886, Borehamwood, 1971. buildings in a variety of industries and trades’, BRE Current Paper
14. Ramachandran, G.: ‘Economic value of automatic fire detectors’, CP 9/77, Borehamwood, 1977
BRE Information Paper IP27/80, Garston, 1980 36. Morgan,H. P . , andChandler, S. E.:‘Fire sizes andsprinkler
15. Theobald, C. R.,and Thomas, P. H.: ‘Studies of fires in industrial effectivenessinshoppingcomplexesandretailpremises’, Fire
buildings. Part l : The growth and development of fire. Part2: The Surveyor, 10(5), 1981, pp23-28
burning rates and durations of fires’, Fire Prevention Science and 37. ‘Sprinklers in high-risebuildings-whythe GLC introduced new
Technology, No. 17, September 1977, pp4-16 requirements’, Fire Prevention, No. 120, pp17-18
16. Fire-safe structural steel. A design guide, Washington DC, 38. ‘Future fire policy. A consultative document’, HomeOffice,
American Iron and Steel Institute, 1979 Scottish Home and Health Department, London, HMSO, 1980
17. Law, Margaret, and O’Brien, Turlogh:Fire safety of bare external 39. Kletz,T.A.: Symposium on loss prevention in the chemical
structural steel, Croydon, Constrado, 1981 industry, Inst. Chem. Eng, Newcastle-upon-Tyne, 1971
18. Lie, T.T.: Fire and buildings, London, Applied Science Publishers 40. Sowby, F. D.: Symposium on transporting radioactive materials,
Ltd., 1972 April I964
19. Malthotra,H.L.:‘Reportonthework of technicalcommittee 41. Fry, J. F.: Inst. Fire Engrs Q,30, 1977
44-PHT “Properties of materials at high temperatures”, Materiaux 42. Thomas, P. H.: Proceedings of the seventh triennial CIB congress,
et Constructions, 15, No.86, pp 161-170 Construction Research International, September 1977
20. Design and detailing of concrete structures for fire resistance, 43 Ward,W.M.:‘Canmore
9 lives andpropertybesaved?’, Fire
interimguidancebyajointcommitteeoftheInstitution of Engineers Journal, December, 1978

Library additions Mr R. F. Stevens. HULL, D.

An
introductioncomposite
to
Continued from Page 24 HARPER, D. R. Building. The process and the materials. (Cambridge: University Press, 1981 .)
product. (Lancaster: Construction Press, 1978.) PresentedbyMessrs J. M. Howard and A. G.
GORMAN, D. J. Free vibration analysis of rec- Presented by Professor E. Happold. Samani.
tangular plates. (NewYork:Elsevier, 1982.) HOMEOFFICE. Domestic nuclear shelters: INSTITUTION CIVIL
OF ENGINEERS.
Presented by Messrs J . D. Drake, V. J. Coghlin, technical guidance. A Home Office Guide. (Lon- Bridge aerodynamics. Proceedingsofacon-
B. Taylor and F. E. Weare. don: HMSO, 1982.) Presented by Mr R. Mar- ference held at the Institution ofCivil Engineers,
HARDOY, J. E. and SATTERTHWAITE, D. shall. London, 25-26 March 1981. (London: Thomas
Shelter: need and response. Housing, land and HORNER, P. C. Earthworks. ICE Works Con- Telford, 1981.) Twocopies.Onepresented by
settlement policies in seventeen third world na- struction
Guide.
(London:ThomasTelford, thepublishersandtheotherby Mr A.W.
tions.(Chichester:Wiley, 1981.) Presentedby 1981 .) Presented by Mr Y. K. Cheng. Carpenter.

EngineerlVolume
The Structural 61A/No. IlJanuary1983 33