Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Steel Structures 7 (2007) 227-237 www.kssc.or.kr
Behavior of Steel-framed Buildings in a Fire
Khalifa S. Al-Jabri*
Associate Professor, Department of Civil and Architectural Engineering, College of Engineering,
P.O. Box 33, Sultan Qaboos University, Al-Khod, PC 123, Oman
Abstract
The rules presented in the current codes of practice for the design of steel-framed buildings in fire conditions are based onresults of isolated member tests carried out in a laboratory that followed a prescribed standard fire curve. The laboratory testingconditions were obviously different from the real situation wherein the structure is subjected to natural fires. It has been knownfor many years, from observations of accidental fires, that structural members behave better in fire when they constitute partof a structural arrangement than when they are tested in isolation. These observations have been confirmed by results fromexperimental fire tests conducted on full-scale multi-story steel-framed buildings. It has been demonstrated that members thatform part of the structure can withstand much higher temperatures than those tested singly. This has raised doubts concerningthe conservative design approaches provided by current fire engineering design codes. The attack on the twin towers of theWorld Trade Centre in New York on 11 September 2001 has prompted close examination of the way in which buildings canfail in fires and has brought into the public eye the hazards that fires can pose to major building structures. This paper examinesthe effects of fires on multi-story steel-framed buildings that provide vital knowledge of the behavior of real buildings. Thismay allow for the construction of safer buildings in the future.
Keywords: Steel-framed building, composite floor, fire, Cardington frame, tensile membrane action
1. Introduction
All common building materials lose strength when
heated to high temperatures. Figs. 1 and 2 show the
deterioration of steel and concrete mechanical properties
with increasing temperatures. Although steel does not
melt below 1,500oC, at a temperature of around 600oC, its
yield strength declines to about one-third of its yield
strength at ambient temperature. At 800oC, its yield
strength is reduced to 11%, and at 900oC, to 6%. The
elastic modulus of steel is similarly reduced with increasing
temperatures, but at a higher rate. Due to internal
cracking and chemical changes, concrete also loses
strength and stiffness as temperature increases, as seen in
Fig. 2. Since concrete has much lower thermal conductivity
than steel, a concrete encasement is often used as a fire
protection for steel. The degradation of structural materials’
stiffness and strength at high temperatures may, in some
incidents, cause the structure to collapse under severe fire
conditions. Materials such as steel obviously need to be
designed to withstand the effects of a fire in order to
ensure the safety of people and property. The design must
ensure no additional threat, to either escaping occupants
or fire fighters, of the possible collapse of the steel
structure.
Using applied fire protection remains the most common
way of satisfying structural fire resistance requirements.
Typical types of protection materials consist of boards,
sprays, and intumescent paints, with the choice of material
depending on cost, appearance, durability, and the required
architectural features. The required thickness is based on
the principle of ensuring that the steel remains below
550oC for the specified fire resistance period without the
*Corresponding authorTel: +968-2414-1333; Fax: +968-2441-3416E-mail: [email protected]
Figure 1. Degradation of the mechanical properties ofsteel at elevated temperatures.
228 Khalifa S. Al-Jabri
necessary consideration of the type of structural steel
element. This assumes that fully stressed steel members
will fail when they reach a temperature of approximately
550oC. This temperature is known as the “limiting
temperature,” according to the British design code
(BS5950: Part 8, 1990). Although the philosophy behind
specifying the protection thickness is adequate, it is not
necessarily the only solution, and it can be extremely
conservative in misestimating the structural capability,
since it ignores the inherent fire resistance of the overall
stiffness of the structure.
From an engineering point of view, however, it is
probably more economical to design a structure so that it
will withstand fire without protection rather than to
design it for normal conditions and then apply protection.
To address this issue, there has been much interest in
understanding the behavior of different structural members
in fire, either in isolation or as part of a more complete
structure. In the beginning, research was conducted on
the actual behavior of isolated bare-steel beams and
columns, which resulted in the first ever fire design code,
BS 5950: Part 8 (1990). The fire occurrence is treated as
an accidental limit state, with its own associated loads
and materials. Despite current design codes providing a
more scientific basis for the provision of fire resistance to
steel-framed structures, they are based on isolated
member tests that did not take into consideration the
interaction between members (BS5950: Part 8, 1990;
EC3: Part 1.2, 1993; and EC4: Part 4, 1994).
Observations from real fires (Newman, 1991; Burgess,
2002; and Burgess, 2005) and recent fire tests on the
Cardington full-scale test frame (Moore and Lennon,
1997) have demonstrated the great importance of structural
continuity and interaction between structural members in
enhancing the performance of steel-framed buildings in
fires. Structural continuity may not have a significant
effect on the behavior of structural members at ambient
conditions, but in a fire, it can play an important role in
enhancing the survival time of the structure. Many
aspects of structural behavior occur due to the interaction
between members, and cannot be predicted or observed
from isolated tests. Some of these features include local
buckling in the vicinity of the joints, deformations due to
adjacent structures’ restraint of thermal expansion, the
redistribution of internal forces, the behavior of the
structures during the cooling phase, and the tensile
membrane action of the composite floor at large
displacements. Furthermore, unlike the standard fire
curve wherein the fire regime follows a pre-defined
temperature-time relationship, a natural fire is characterized
by three phases: a growth phase, a fully developed phase,
and a decay phase (Fig. 3). It is necessary to evaluate not
only the effect on the structural resistance during the
heating phase, but also the high cooling strains in the
beams and adjacent joints induced by the distortional
deformation of the heated elements during the decay
phase (Wald et al., 2004).
In view of the above, it is apparent that current fire
engineering design codes (BS5950: Part 8, 1990; EC3:
Part 1.2, 1993; and EC4: Part 4, 1994) do not address the
actual behavior of buildings in fires, since buildings do
not act as a series of individual members. In reality,
structures have significant reserve strength to withstand
temperatures much higher than those described as
limiting temperatures in the design codes. As a result,
concepts have started to emerge that can lead to the
development of more rational and economical design
methods for steel-framed structures in fires. This may
eventually reduce or even eliminate reliance on
traditional fire protection methods, and may allow
construction of safer, more efficient, and more effective
fire-resistant buildings in the future.
Figure 2. Degradation of the mechanical properties ofconcrete at elevated temperatures.
Figure 3. Natural fire and ISO834 standard fire.
Behavior of Steel-framed Buildings in a Fire 229
2. Review of Current Fire Engineering Design Methods for Steel-framed Structures
Existing codes (BS5950: Part 8, 1990; EC3: Part 1.2,
1993; and EC4: Part 4, 1994) for the design of steel-
framed structures in fires have introduced design methods
that treat fire as one of the basic limit states. These design
codes were developed, however, mainly from standard
fire tests on isolated columns and beams. In these tests,
the columns were 3.0 m high and the beams were 4.5 m
long, and the failure criteria were governed largely by the
size of the furnace. A beam element was assumed to have
failed when the deflection exceeded span/30, whereas
column failure was deemed to have occurred when the
column could no longer carry the applied load. Based on
the results of such tests, design methods such as the
limiting temperature method and the moment resistance
method were proposed.
2.1. The limiting temperature method
BS 5950: Part 8 (1990) uses the concept of the load
ratio as a measure of the applied load that a member can
resist at the time of a fire. Because the limiting temperature
is virtually independent of the heating time, it depends on
the load that the member carries. The load ratio is defined
as the ratio between the applied load or moment at time
of a fire and the load or moment of resistance at 20ºC.
The load ratio is a useful concept because it allows
different-sized elements to be considered in the same
way. For instance, a 200 mm-deep beam will fail at
approximately the same temperature as a 400 mm-deep
beam if they are both working at the same load ratio. In
practical designs, the load ratio will vary from 0.45 to
0.55. Load ratios much higher than 0.6 are very rare,
although the maximum value can be as high as 0.7 for an
element that carries purely the dead weight of the
structure. Using the concept of the load ratio, it is
possible to design some members that can withstand a
fire without fire protection, and to use a lesser degree of
fire protection for other members. A member that fails at
550oC will require more protection than if the same
member fails at 700oC.
After the load ratio is calculated, the limiting temperatures
(or the maximum allowable temperatures) can be
obtained from Table 5 in BS5950: Part 8. This Table is
reproduced in Table 1. To allow for the use of unprotected
steel, the limiting temperatures are compared with the
design temperatures that are tabulated in BS5950: Part 8
and that correspond to the expected maximum temperature
of the section for a given period of fire resistance (i.e., 30
and 60 minutes). If the limiting temperature is less than
the design temperature, the beam does not require fire
protection. Despite the simplicity of the load ratio
method, it is very conservative, because it does not take
advantage of the full capacity of the member.
2.2. The moment resistance method
The moment resistance method is based on the plastic
design of a structural element, as shown in Fig. 4. If the
temperature distribution across a beam is known, the
moment resistance can be calculated. This is a powerful
but simple tool that until recently was used only by
researchers. The cross-section is divided into groups of
elements of approximately equal temperatures. The
reduced strength of various elements of the cross-section
can be calculated at elevated temperatures using the
strength reduction factors given in Table 1 in BS5950:
Part 8. The plastic neutral axis of the cross-section is
determined and the moment of resistance (or the moment
capacity) of the section is determined by multiplying the
reduced strength of each element by the distance from the
neutral axis and summing up all the elements in the
section. The applied moment is calculated using the fire
limit state load factor. If the moment of resistance
(capacity) is greater than the applied moment, the section
is adequate and protection is not required.
Table 1. Limiting temperatures for the design of members (From BS5950: Part 8)
Description of member Limiting temperature (oC) at a load ratio of:
0.7 0.6 0.5 0.4 0.3 0.2
Members in compression
λ 70 510 540 580 615 655 710
λ > 70 but λ ≤ 180 460 510 545 590 635 635
Members in bending supporting a concrete or composite slab
a) Unprotected beams or beams protected with ductile protection 590 620 650 680 725 780
b) Other protected beams 540 585 625 655 700 745
Members in bending not supporting concrete
a) Unprotected beams or beams protected with ductile protection 520 555 585 620 660 715
b) Other protected beams 460 510 545 590 635 690
Members in tension 460 510 545 590 635 690
λ is the slenderness, i.e., the effective length divided by the radius of gyration.
230 Khalifa S. Al-Jabri
3. Full-scale Experimental Tests on Steel-framed Structures in a Fire
On the 23rd of June 1990, a fire broke out in the
partially completed 14-story building in the Broadgate
development (Anon, 1991). The fire began at a large
contractors’ hut on the first floor, and smoke spread
undetected throughout the building. The fire detection
and sprinkler system was not yet operational after
working hours. The fire lasted for 4.5 hours, including 2
hours when the fire exceeded 1,000oC. The direct losses
from the fire exceeded £25 million, but only a fraction of
such loss (£2 million) represented the structural frame
and floor damage. The major damage was to the building
fabric as a result of the smoke. Moreover, the structural
repairs after the fire took only 30 days. The structure of
the building was a steel frame with composite steel deck
concrete floors, and it was only partially protected at this
stage of the building’s construction. During and after the
fire, despite significant deflections in the elements exposed
to the fire, the structure behaved well, and none of the
columns, beams, and floors collapsed.
The Broadgate phase 8 fire provided the first opportunity
to examine the influence of a fire on the structural
behavior of a modern fast-track steel-framed building
with composite construction. Prompted by the evidence
from Broadgate, the Building Research Establishment
(BRE) built an 8-story composite steel and lightweight
concrete frame at their large-scale test facility at
Cardington. The frame was subjected to six full-scale fire
tests [2 by BRE and 4 by British Steel (now CORUS)] in
1995 and 1996, which allowed observation and recording
of the behavior of the structure during a fire.
These tests were commonly known as Cardington frame
fire tests. The structure was designed as a composite
building typical of contemporary medium-rise commercial
buildings in the United Kingdom, with roof-mounted
services. The plan dimensions were 45 × 21 m, which
provided a footprint area of 945 m2. The general frame
arrangement is shown in Fig. 5. These tests included: (1)
a restrained beam test, (2) a plane frame test, (3) a first
corner test, (4) a second corner test, (5) a large compartment
test, and (6) a demonstration test. A brief summary of
these tests is given below, whereas and a detailed
description is presented elsewhere (Armer and Moore,
1994; Lennon, 1997; O’Connor and Martin, 1998; and
Bailey et al., 1999).
The restrained beam test comprised a heated 8 m × 3 m
area on the seventh floor. A gas furnace was used to heat
the 9 m 305 × 165UB40 secondary test over the central 8
m of its length.
The plane frame test was conducted using a gas-fired
furnace to heat up a 21 m × 2.5 m area on the fourth floor,
and included three connected spans of primary beams.
The first corner test was carried out in a 10 m × 7 m
area on the second floor. The fire was generated by firing
wooden cribs with densities of 45 kg/m2. The internal
beams were left unprotected whereas the columns and
edge beams were protected. Lightweight concrete block
walls were used in the construction of a compartment.
The second corner test was conducted in a compartment
built from fire-resistant board. In this test, a 9 m × 6 m
area on the third floor was heated by burning wooden
cribs with densities of 45 kg/m2. The columns were fully
protected whereas the internal and edge beams were all
left unprotected.
The large compartment test was carried out in a
Figure 4. Stress distribution analysis at the fire limit state (BS 5950: Part 8, 1990).
Figure 5. General beam framing arrangement of theCardington test frame.
Behavior of Steel-framed Buildings in a Fire 231
compartment built with a fire-resistant wall across the
21 m width of the building, with fire-resistant board used
to block off the entrance to the passenger hoist. As in the
two previous tests, the fire in the compartment was
generated by burning wooden cribs with densities of 45
kg/m2 in a 21 m × 18 m area on the third floor.
The demonstration office fire test was carried out on a
compartment constructed using concrete blockwork. The
fire load consisted of actual office furniture and documents
that weighed as much as 45 kg/m2 of wood. The test area
was 18 m wide by up to 10 m deep. The columns and
beam-to-column connections were protected whereas the
primary and secondary beams were left unprotected.
Figure 6 shows the floor plan of the Cardington frame
that marks the locations of the fire tests.
4. Behavior of Structural Members in a Fire
4.1. Single beam behavior
Figure 7 compares the beam behavior of the Cardington
frame (Robinson, 1998) in the standard fire test and the
restrained beam test. The beam tested was 356 × 165 UB,
with load ratios of 0.37 and 0.4 in the standard test and
the Cardington test, respectively. In the standard test, the
beam tested was 4.5 m long, with simply supported end-
conditions without axial restraint, whereas in the Cardington
frame test, the unprotected 9 m beam was connected to
the other members in the frame.
The beam behavior in the isolated member test indicated
that the deflection began slowly as the temperature rose
and progressively accelerated up to the termination point
of span/30 at a temperature of approximately 705oC. In
the Cardington frame test, however, the beam behaved
differently, with an almost constant deflection rate
throughout, as shown in Fig. 7. The beam showed no sign
of ‘runaway’ even when the beam temperature was
875oC, at which point the test was terminated because of
a breakdown in the deflection measuring instrumentation.
Had the deflection continued at the same rate, a
temperature of over 1,000oC would have been needed to
achieve the standard test criterion of span/30 deflection
without protection. This comparison raises an obvious
concern on whether or not standard fire tests reflect the
real performance of structural members in a fire.
4.2. Behavior of joints
Two types of joints were used in the Cardington frame:
flexible (partial-depth) end-plates and fin-plates. Flexible
end-plates were used for beam-to-column joints and fin-
plates for beam-to-beam joints. These joints are usually
considered pinned joints, which are assumed to transfer
only shear forces and to have sufficient flexibility to
allow rotation. Observations from Cardington fire tests
(Al-Jabri et al., 1999; Al-Jabri and Hago, 2003; and Wald
et al., 2004) indicate that the joints showed signs of being
subjected to high tensile forces (Fig. 8). In the flexible
end-plate joints, the plates were fractured on one side
whereas the other side remained intact, as shown in Fig.
9(a); and in the fin-plates (the beam-to-beam joints), the
bolts were sheared [Fig. 9(b)]. The fracture resulted from
the high tensile forces that developed during the cooling
of the connected beam under the large rotations
associated with the pinned joints. During the cooling, the
steel contracted, which caused a reduction in the vertical
deflection of the beams, whereas a significant tensile
force was applied to the weld that connected the partial-
depth end-plate to the beam flange (Fig. 8). This tensile
strain on cooling was relieved by the plate fracture or the
bolt shear of the joint (Fig. 9). The shear fracture of the
bolts in the fin-plate joints suggests that flexible end-plate
joints act more reliably in a fire than fin-plate joints. The
behavior of joints during structural cooling has to be
further investigated. The suggested initial solution to this
situation is the design of the joint to have significant
ductile behavior so as to ensure that the shear capacity is
maintained when the what? is subjected to high tensile
forces in a fire (Sarraj et al., 2006).
The temperature of the beam bottom flange was
Figure 6. Floor plan of the Cardington test frame showingthe locations of the fire tests.
Figure 7. Comparison of the deflection behavior of abeam in the standard test and in the Cardington frametest.
232 Khalifa S. Al-Jabri
considerably higher than the temperature of the joint. At
the maximum temperature, the joint temperature was
almost 200oC lower than the limiting temperature of the
beam. The temperature of the bottom bolts was higher
than the temperature of the top bolt, due to the shielding
by the composite slab, which acted as a heat sink to the
top part of the joint. The end-plate was hotter than the
bolts at the same level because of the ratio of the bolt
diameter to the end-plate thickness.
Also, observations of the joint behavior at Cardington
showed that the bolts and welds did not suffer from
premature failure during the heating phase of the fire.
During the cooling phase, however, a number of bolts
that acted in a single shear in the fin-plate joints suffered
from shear failure due to the tensile forces generated by
thermal contraction, as shown in Fig. 9(b).
4.3. Local buckling of beams
Most of the internal beams showed signs of local
buckling during the heating phase in the lower flange and
in a part of the web in the vicinity of the joints, as shown
in Fig. 10. This behavior was caused by the restraint to
thermal expansion and the negative moment caused by
the rotational restraint from the joint. The restraint to
thermal expansion was provided by the surrounding
cooler structure and the structural continuity of the test
frame. This took place in the beam due to the inability of
the lower flange of the beam to transfer the high axial
forces induced in the beam to the adjacent beam-columns
after the closure of the gap in the lower part of the joints
(Wald et al., 2004). Conservatively, therefore, the joints
should be assumed as ‘pinned’ and the connected beams
as simply supported, allowing larger mid-span deflections
to develop than when the beams are semi-rigidly
connected. Local buckling was found as but a minor
concern in isolated member fire tests (Al-Jabri et al.,
1999).
4.4. Behavior of columns
Observations from the Cardington fire tests show that
the internal and external columns were subjected to high
moments, which caused local squashing of the columns,
as shown in Fig. 11, although no collapse occurred because
the structure had the ability to carry the column load
using an alternative load path (i.e., the tensile membrane
action in the composite floor). These moments were
caused by the expansion of the connecting beams, the
expansion of the heated floor relative to the other floors,
and the induced P-δ effects during a fire. Bailey (2000)
pointed out that if these moments were simply included
within the present member design procedure outlined in
BS5950: Part 8 or EC3: Part 1.2, the calculations would
show that the columns would fail during the fire due to
local plasticity. In steel-framed braced structures, however,
bracing causes the existence of a significant degree of
redundancy. Large localized stresses are caused by
induced column moments that result in the formation of
plastic hinges at the floor level without causing overall
collapse. It is suggested that due to the good inherent fire
resistance of the composite slab, it provides horizontal
restraint to the columns at the floor level during a fire.
Therefore, the influence of the expanding beams on the
overall stability of the column must be checked during a
fire, although it is assumed that the localized plasticity
can be accommodated.
To investigate the stability of columns during a fire in
view of the column behavior observed from the Cardington
Figure 8. Tensile forces induced on the joints during thecooling phase of the structure.
Figure 9. Typical failure modes of joints in the Cardington frame tests.
Behavior of Steel-framed Buildings in a Fire 233
tests, a simple analytical model was developed by Bailey
(2000). In the model, columns were subjected to both
axial loads and moments caused by the expansion of the
connected beams. Although the effect of floor slabs was
ignored in the analysis, the heated beams were restrained
axially to provide lateral restraint to the columns. The
effects of many parameters that influence the column
behavior, such as the beam-to-column heating rate, the
beam and column sizes, the beam-to-column connection
rigidity, the axial restraint at one end of the heated beams,
the span of the beams, the base rigidity of the heated
column, and the column axial load were studied. The
results of the analysis indicated that instability occurred
in the column, even though it was forced into a double
curvature and restrained at the floor level. This instability
was caused by the P-δ effect in the column, which was
enhanced due to the enforced deflected shape of the
column from the expansion of the connecting beams. The
analysis also showed that parameters such as the beam-
to-column heating rates, the beam cross-section size, the
span of the beams, the end-rigidity of the heated column,
and the column axial load have significant influence on
column instability. The parameters that had a nominal
effect on the behavior of the column, however, included
the column cross-section size, the beam-to-column
connection rigidity, and the horizontal restraint to the
heated beams.
4.5. Behavior of the composite floor
The composite flooring system used in the Cardington
full-scale frame comprised steel downstand beams that
acted compositely with a floor slab and that were constructed
using a trapezoidal steel deck, lightweight concrete, and
an anticrack A142 steel mesh. The overall depth of the
slab was 130 mm, and the mesh was situated 15 mm
above the steel deck. The results of the isolated fire tests
conducted on the beam-to-column composite joints (Al-
Jabri et al., 1998) showed that the composite slab above
the joints caused a 20-30% reduction in the beam’s top
flange temperatures in comparison with the beam’s
bottom flange temperature. This suggested that the concrete
slab acted as an insulation and a heat sink to the top of the
beam, which enhanced the joint performance at an
elevated temperature.
Observations from the Cardington fire tests and other
large building fires have shown that the behavior of the
composite floor slab plays a crucial role in providing
enhanced fire resistance to the structure. Also, it has been
confirmed that the performance of a steel frame with a
composite flooring system is significantly better than that
suggested by current fire design methods (Martin and
Moore, 1997 and Bailey et al., 1999) due to the presence
of tensile membrane action in the composite slab during
the fire at large displacements. At the fire limit state, large
displacements of the structure are acceptable provided
that the fire is contained within the compartment of origin
so that the risk of the fire spreading throughout the
building would be low. During a fire, when significant
numbers of unprotected secondary steel beams are
damaged, the lightly reinforced composite slab acts as a
membrane supported by cold perimeter beams and
protected columns. Due to the failure of unprotected steel
beams to carry any load, the composite slab utilizes its
full bending capacity to span the adjacent cooler members.
With increasing displacement, the composite slab acts as
a tensile membrane that carries the loads in the
reinforcement. In the case of simply supported edges, the
supports will not anchor these tensile forces and a
compressive ring will form around the edge of the slab.
Failure will only occur at large displacements with the
fracture of the reinforcement. Fig. 12 shows the tensile
membrane action of a composite slab with no horizontal
restraint around its perimeter.
Current fire engineering design methods completely
ignore the beneficial effect of the tensile membrane
action at large displacements that occurs in the composite
slab during a fire. To take advantage of this behavior
when designing structures in a fire, a new design method
has been developed (Bailey et al., 2000; Bailey and
Moore, 2000a; Bailey and Moore, 2000b; Bailey, 2001;
and Bailey, 2003) to calculate the performance of steel-
Figure 10. Local buckling of primary and secondarybeams in the vicinity of the joints.
Figure 11. Squashing of the columns due to large moments.
234 Khalifa S. Al-Jabri
framed buildings with composite flooring systems that
are subject to fire. The method is valid for both square
and rectangular slabs and conforms to the mode of
behavior observed in the Cardington full-scale fire tests.
This method uses a simple energy approach to calculate
the load-carrying capacity of a composite flooring
system. The energy of the lightly reinforced composite
slab is based on the yield-line approach that has been
modified to account for the enhancement caused by in-
plane forces. The developed method has been extended
further to incorporate the membrane action of the slab
and the beam systems that act compositely (Bailey, 2004).
The basic principle of the design method (Bailey and
Moore, 2000a; Bailey and Moore, 2000b; and Burgess,
2005) is summarized below.
(a) The entire floor slab is subdivided into a series of
square or rectangular slab panels (Fig. 13) that incorporate
a number of unprotected composite beams. The size of
the slab panels may be governed by the fire compartment
size.
(b) During a fire, each slab panel area supports the
applied load via the membrane action of the composite
slab and the flexural strength of the grillage of the
composite beam within the panel area. Protected beams
or beams designed to have sufficient resistance to support
an applied load for the duration of a fire are assumed to
support the perimeter of the slab panel. The columns are
either protected or designed to withstand the applied load
for the fire resistance period.
(c) The fire resistance period is defined in BS5950: Part
8 and EC4: Part 1.2, and the total load on the structure at
the fire limit state is calculated.
(d) The maximum allowable vertical displacement for
each slab panel is determined, the beam with the highest
load ratio is identified, and the load carried by the beam
for the specified fire resistance period is calculated.
(e) The flexural strength at the fire limit state of the
composite slab is calculated, considering only the mesh
reinforcement and the concrete components using the
yield-line theory. The slab strength can be enhanced due
to the membrane action based on the maximum allowable
vertical displacement.
(f) The load-carrying capacity of the slab panel is
calculated by summing up the capacity of the composite
slab and the capacity of the grillage of the unprotected
composite beams.
(g) If the load-carrying capacity of the slab panel and
the grillage of the unprotected composite beams are
greater than the applied load, the beams can be left
unprotected.
This method is commonly known as the BRE design
method, which has been developed further (Newman et
al., 2000) into a series of design tables that allow the
designer to leave large numbers of secondary beams
unprotected in buildings requiring 30- and 60-minute fire
resistance, although some compensation features, such as
increased mesh size and density, may be required.
The developed method was validated against the results
of the Cardington fire tests and of the 15 small-scale tests
conducted on horizontally unrestrained slabs that were
subjected to large vertical displacements (Foster, 2004).
A comparison of the results of the two tests shows that
the design method compares well with experimental
results and is generally conservative. It has been
recommended that the method be further developed to
account for the bond characteristics of reinforcement and
the change in the failure mode for some orthotropically
reinforced slabs. A number of numerical studies and
computer simulations have also been performed (Huang
et al., 2003a; Huang et al., 2003b; Izzuddin and Elghazouli,
2004a; Izzuddin and Elghazouli, 2004b; Burgess, 2005;
and Foster et al., 2006) to investigate the influence of the
Figure 12. Tensile membrane action of a floor slab withno horizontal restraint around its perimeter.
Figure 13. Division of the floor plate into slab panels.The floor plate is divided into (a) square panels and (b)rectangular panels.
Behavior of Steel-framed Buildings in a Fire 235
tensile membrane action in the composite slab on the
behavior of composite buildings in a fire based on the
results of the Cardington fire tests. These studies confirm
that while exposed steel temperatures in composite
buildings remain below 400oC, the much cooler concrete
slab plays only a trivial role in the load-carrying
mechanism, apart from generating the thermal curvature
of the composite beams. For steel temperatures higher
than 500oC, the significance of the slab progressively
increases. At very high temperatures, the floor slab
becomes the main load-bearing element and the floor
loads above the fire compartment are carried largely by
tensile membrane forces developed mainly in the steel
anticracking mesh or reinforcing bars.
5. Overall Frame Behavior during a Fire
The Cardington frame tests showed that the composite
frame suffered from considerable deformation without
any form of collapse (Fig. 14), even though the unprotected
beams reached temperatures of up to 1,100oC. At this
temperature level, BS5950: Part 8 indicates that only 3%
of the member’s design strength remains. The results of
the experiment on the Cardington frame demonstrated
with no doubt the major contribution of the composite
floor to the survival of the frame in a fire. The floor
performed very well in all the tests, which supports the
results of previous small-scale tests (Al-Jabri et al., 1999)
that showed that this type of floor system has good
inherent fire resistance. Results of analytical studies
conducted on the Cardington frame tests (Gillie et al.,
2001; Gillie et al., 2002; Huang et al., 2002; and O’Connor
et al., 2003) confirmed that the effects of thermal
expansion dominated the response of the structure and
that the material degradation and the gravity loading were
of secondary importance. It was also found that at extreme
temperatures, the significant load-carrying mechanism was
the tensile membrane action in the reinforcement mesh
and that gravity loading can influence the magnitude of
the tensile forces produced. The results also suggested
that one method of helping maintain structural integrity in
composite structures during extreme fires is to ensure that
a sufficient degree of ductile reinforcement is present in
the concrete floor slabs. Columns were found to be more
critical than beams and will need protection in multi-story
buildings. The behavior of joints during cooling due to
restraint to thermal expansion has to be further investigated.
The survival of steel-framed structures from the severe
fire conditions experienced in the Cardington project or in
buildings subjected to real fire accidents raises a number
of fundamental issues as to whether or not current design
methods that were based on isolated members tests reflect
the true behavior of structures in a fire. The results of
these tests confirmed that current fire engineering design
methods are too conservative. A new design method has
been developed (Newman et al., 2000) that takes into
account the inherent fire resistance of steel-framed
members in fire conditions. This method incorporates the
beneficial effects of a composite slab and beam systems
on the survival time of steel-framed structures in a fire.
6. Conclusions
This paper examined the effects of fires on the behavior
of multi-story steel-framed buildings. It can be concluded
that, even in the context of the structural fire engineering
approach of modern design codes, predictions of behavior
based on furnace tests or numerical modeling of isolated
members are unreliable. The behavior of the members
within a continuous, compartmented structure is very
different from the behavior of isolated members. Structural
continuity, restraint to thermal expansion provided by the
adjacent members, the beam-to-column joints, and the
tensile membrane action of the composite slab have
demonstrated a significant positive influence on the
bevahior of the entire structure in the event of a fire. Data
from the Cardington fire tests and the subsequent
experimental and analytical studies provide fundamental
information that is very important for researchers studying
the performance of steel-framed buildings in a fire for
many years to come so as to develop new design approaches
that take into account the interaction between structural
members in a fire. The developments that have already
taken place in the past few years have been very
significant in understanding the reality of structural
behavior in a fire, which will undoubtedly lead to the
emergence of new rational fire engineering design
methods that may allow us to construct safer, more
effective, and more efficient fire-resistant buildings in the
future.
References
Al-Jabri, K.S. and Hago, A.W. (2003). Towards a Rational
Approach to the Design of Steel-framed Buildings in a
Fire. Proceedings of the 9th Arab Structural Engineering
Conference, United Arab Emirates, pp. 971-978.
Al-Jabri, K.S.; Burgess, I.W.; Lennon, T.; and Plank, R.J.
Figure 14. Overall frame deformation after a fire in theCardington 8-story building.
236 Khalifa S. Al-Jabri
(1999). The Performance of Frame Connections in a Fire.
ACTA POLYTECHNICA-EUROSTEEL ‘99, 39(5), pp.
65-75.
Al-Jabri, K.S.; Lennon, T.; Burgess, I.W.; and Plank, R.J.
(1998). Behavior of Steel and Composite Beam-Column
Connections in a Fire. Journal of Constructional Steel
Research, 46(1-3), Paper No. 180.
Anon (1991). Investigation of the Broadgate Phase 8 Fire.
Structural Fire Engineering. Steel Construction Institute,
United Kingdom.
Armer, G.S.T. and Moore, D.B. (1994). Full-scale Testing on
Complete Multi-story Structures. The Structural Engineer,
72(2), pp. 30-31.
Bailey, C.G. (2000). The Influence of the Thermal Expansion
of Beams on the Structural Behavior of Columns in Steel-
framed Structures during a Fire. Engineering Structures,
22, pp. 755-768.
Bailey, C.G. (2001). Membrane Action of Unrestrained
Lightly Reinforced Concrete Slabs at Large
Displacements. Engineering Structures, 23, pp. 470-483.
Bailey, C.G. (2003). Efficient Arrangement for Membrane
Behavior of Composite Floor Slabs in Fire Conditions.
Journal of Constructional Steel Research, 59, pp. 931-
949.
Bailey, C.G. (2004). Membrane Action of Slab/Beam
Composite Floor Systems in a Fire. Engineering
Structures, 26, pp. 1691-1703.
Bailey, C.G. and Moore, D.B. (2000a). The Structural
Behavior of Steel Frames with Composite Floor Slabs
Subject to Fire (Part 1: Theory). The Structural Engineer,
78(11), pp. 19-27.
Bailey, C.G. and Moore, D.B. (2000b). The Structural
Behavior of Steel Frames with Composite Floor Slabs
Subject to Fire (Part 2: Design). The Structural Engineer,
78(11), pp. 28-33.
Bailey, C.G.; Lennon, T.; and Moore, D.B. (1999). The
Behavior of Full-scale Steel-framed Buildings Subjected
to Compartment Fires. The Structural Engineer, 77(8),
pp. 15-21.
Bailey, C.G.; White, D.S.; and Moore, D.B. (2000). The
Tensile Membrane Action of Unrestrained Composite
Slabs Simulated under Fire Conditions. Engineering
Structures, 22, pp. 1583-1595.
BS 5950 (1990). Structural Use of Steelwork in Buildings
(Part 8: Code of Practice for Fire-resistant Design).
British Standard Institution, London, UK.
Burgess, I.W. (2002). Fire Resistance of Framed Buildings.
Physics Education, 37(5), pp. 390-399.
Burgess, I. (2005). Performance and Design of Multi-story
Composite Buildings in a Fire. Proceedings of the KICT
2005 Annual Conference / CUFER Annual Technical
Seminar, Seoul, Korea, pp. 89-109.
EC 3 (1993). Design of Steel Structures (Part 1.2: General
Rules, Structural Fire Design). ENV1993-1-2, European
Committee for Standardization, Brussels.
EC 4 (1994). Design of Composite Steel and Concrete
Structures (Part 1.2: General Rules, Structural Fire
Design). ENV1994-1-2, European Committee for
Standardization, Brussels.
Foster, S.J.; Bailey, C.G.; Burgess, I.W.; and Plank, R.J.
(2004). Experimental Behavior of Concrete Floor Slabs
at Large Displacements. Engineering Structures, 26, pp.
1231-1247.
Foster, S.J.; Burgess, I.W.; and Plank, R.J. (2006). Modeling
the Membrane Action of Model-scale Slabs at Ambient
and Elevated Temperatures. Structures in Fire Workshop,
Aveiro, Portugal, pp. 635-646.
Gillie, M.; Usmani, A.S.; and Rotter, J.M. (2001). A
Structural Analysis of the First Cardington Test. Journal
of Constructional Steel Research, 57, pp. 581-601.
Gillie, M.; Usmani, A.S.; and Rotter, J.M. (2002). A
Structural Analysis of the Cardington British Steel Corner
Test. Journal of Constructional Steel Research, 58, pp.
427-442.
Huang, Z.; Burgess, I.W.; and Plank, R.J. (2002). Modeling
of Six Full-scale Fire Tests on a Composite Building. The
Structural Engineer, 80(19), pp. 30-37.
Huang, Z.; Burgess, I.W.; and Plank, R.J. (2003a). Modeling
the Membrane Action of Concrete Slabs in Composite
Buildings in a Fire (Part I: Theoretical Derivation).
Journal of Structural Engineering, 129(8), pp. 1093-1102.
Huang, Z.; Burgess, I.W.; and Plank, R.J. (2003b). Modeling
the Membrane Action of Concrete Slabs in Composite
Buildings in a Fire (Part II: Validations). Journal of
Structural Engineering, 129(8), pp. 1103-1112.
Izzuddin, B.A. and Elghazouli, M. (2004a). Failure of
Lightly Reinforced Concrete Members under Fire (Part I:
Analytical Modeling). Journal of Structural Engineering,
130, pp. 3-18.
Izzuddin, B.A. and Elghazouli, M. (2004b). Failure of
Lightly Reinforced Concrete Members under a Fire (Part
II: Parametric Studies and Design Considerations).
Journal of Structural Engineering, 130, pp. 19-32.
Lennon, T. (1997). Cardington Fire Tests: Survey of Damage
to an Eight-story Building. BRE Internal Report GD1286/
86, Building Research Establishment, UK.
Martin, D.M. and Moore, D.B. (1997). Introduction and
Background to the Research Program and Major Fire
Tests at BRE Cardington. National Steel Construction
Conference, London, UK, pp. 37-84.
Moore, D.B. and Lennon, T. (1997). Fire Engineering
Design of Steel Structures. Progress in Structural
Engineering and Materials, 1(1), pp. 4-9.
Newman, G.M. (1991). Structural Fire Engineering:
Investigation of the Broadgate Phase 8 Fire. Steel
Construction Institute, UK.
Newman, G.M.; Robinson, J.T.; and Bailey, C.G. (2000).
Fire Safety Design: A New Approach to Multi-story Steel-
framed Buildings. SCI-P288, Steel Construction Institute,
UK.
O’Connor, M.A. and Martin, M.D. (1998). Behavior of a
Multi-story Steel-framed Building Subjected to a Fire
Attack. Journal of Constructional Steel Research, 46(1-3),
Paper No. 169.
O’Connor, M.A.; Kirby, B.R.; and Martin, D.M. (2003).
Behavior of a Multi-story Composite Steel-framed
Building in a Fire. The Structural Engineer, 81(2), pp. 27-
36.
Robinson, J. (1998). Fire--A Technical Challenge and a
Market Opportunity. Journal of Constructional Steel
Research, 46(1-3), Paper No. 179.
Sarraj, M.; Burgess, I.W.; Davison, J.B.; and Plank, R.J.
Behavior of Steel-framed Buildings in a Fire 237
(2006). Finite Element Modeling of Fin-plate Steel
Connections in a Fire. Structures in Fire Workshop,
Aveiro, Portugal, pp 315-326.
Wald, F.; Simõnes da Silva, L.; Moore, D.; and Santiago, A.
(2004). Experimental Behavior of Steel Joints in a
Natural Fire. ECCS-AISC Workshop, Amsterdam, The
Netherlands.
Related Documents
