Académique Documents
Professionnel Documents
Culture Documents
ABSTRACT: When exposed to fire, steel loses stiffness and strength; to limit this loss of strength and stiffness,
external fire protection is provided to the steel structural members to satisfy required fire resistance ratings.
However, the current practice of providing fire protection is based on the behavior of single elements under
idealistic situations; as such, it is conservative and does not represent the realistic fire behavior of real structures.
Recently, there has been a spurt of worldwide research activities aimed at minimizing such fire protection costs.
These studies have mainly focused on understanding the realistic fire behavior of complete structures and on
developing innovative systems with inherent fire resistance. Feasible solutions are now emerging, in which the
external fire protection may be completely removed, in certain situations, without compromising the fire safety
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 06/03/15. Copyright ASCE. For personal use only; all rights reserved.
of steel structures. This paper reviews some of the recent developments in this field and describes one particular
system in some detail. This system uses the conventional composite slab/steel beam flooring system in con-
junction with concrete-filled steel tubular columns. Research studies at the United Kingdoms Building Research
Establishment and the National Research Council of Canada suggest that it might be possible, through proper
design, to eliminate fire protection for steel in this system.
Steel sections.
The corner fire test was carried out in one corner of the test
building, enclosing an area of 6 m by 9 m in plan, as indicated
in Fig. 2. In this test, the fire representing a realistic, but worst
case, fire scenario was provided with 40 kg/m2 of wooden
cribs of the floor area. The presence of sprinklers was not
considered in any of the tests, to reflect the worst-case sce-
nario. A double glazing window, 2.5 m high along the 9 m
wall, was initially sealed before the fire started. The static
loading included a self weight of about 2.4 kN/m2, and an
imposed load of about 2.5 kN/m2, simulated using 12 sand-
bags, each weighing 11 kN.
All steel beams were unprotected and all columns were pro-
tected. The column fire protection was considered necessary
after the observation of a previous fire test, carried out in a
narrow enclosure of 2 m wide but covering the whole depth
of 21 m of the one floor. In that test, the columns enclosed in
the fire testing area were only protected to the assumed false
ceiling height, leaving a short length of about 400 mm unpro-
tected. After the fire test, it was observed that this short length
of unprotected steel was completely squashed. Although the
steel frame was still stable, the fire damage due to the squashed
column length extended to all of the floors above the test en-
closure and to the two adjacent bays of 18 m. To prevent
excessive fire damage, a decision was made to protect all of
the columns to the full height in subsequent fire tests.
Preliminary Findings
Results from the large-scale fire test program revealed a
number of phenomena that would not be observed from fire
FIG. 1. Steel-Framed Building at Cardington tests on individual elements under the standard fire exposure.
Fire Behavior All of the beams in this fire test were unprotected. Numer-
ical simulations based on the most favorable boundary con-
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 06/03/15. Copyright ASCE. For personal use only; all rights reserved.
In the corner fire test, the double glazing window was ini- dition (assuming full continuity between beam and column
tially sealed when the fire was ignited. After some growth, as connections) indicated that the secondary beam would have
indicated by the fire temperature rise in Fig. 3, the fire started experienced failure at a temperature of about 750C. In the
to die down due to the lack of oxygen. It was only after the test, the secondary beam was completely engulfed in the fire
standby fire brigade was asked to break the window twice that and reached a temperature in excess of 900C. However, not
the fire had sufficient oxygen to grow, leading to the expected only did the beam not fail, as can be seen in Fig. 4, which
to support the composite slabs under the fire condition. The fire test. The tensile membrane action mechanism was found
observed behavior is possible only when the slab was holding to offer a credible explanation (Wang 1996, 1997) for the phe-
the weakened steel beam and directly transferring the applied nomenon observed during this fire test. However, for this ef-
load to the columns. However, calculations using the upper- fect to be reliably predicted, further tests on simply supported
bound yield-line analysis suggested that the bending capacity reinforced concrete slabs are being carried out to confirm the
of the slab was much lower than the applied load. Thus, the formation of this compressive ring beam.
observed beam/slab behavior could not be explained by con-
ventional bending theory (related to small deflections). Contribution from Nonstructural Members
Tensile Membrane Action in Composite Slab The 9 m edge beams on the adjacent floors were linked
together by two small posts at the two 1/3 positions. The func-
To explain the above behavior, an alternative load carrying tion of these wind posts was to secure the non-load-bearing
mechanism was thought to be active under the fire conditions walls and windows under the lateral wind load. They were not
and it was identified as the tensile membrane action in the designed to contribute to the load carrying capacity of the
composite slab. This mechanism occurs under large slab de- beam, and the beam was designed as simply supported with a
flections, in which the tensile reinforcement acts as a net to 9 m span. However, observed behavior suggested that these
resist the applied load. While it is easy to envisage this load nonstructural members made a significant contribution to the
carrying mechanism when the slab is securely connected to survival of the edge beam during the fire attack. Fig. 6 shows
the surrounding structure, it becomes somewhat more difficult the recorded maximum deflectiontemperature relationship for
when the composite slab becomes simply supported, as would the edge beam. Very little vertical deflection was observed,
and this was attributed to the restraint offered by these wind
posts. In fact, the edge beam may be regarded as a three-span
continuous beam.
ing a short length of about 400 mm unprotected. When the forced-composite slabs, and have shown quite promising re-
test was completed, it was observed that the short length of sults. In particular, theoretical results for the Cardington
the unprotected columns was completed squashed. This short- building indicate that even without the steel beams and the
ening of the columns extended the fire damage much beyond steel decking, the collapse loads of the reinforced-concrete
the narrow test enclosure to two complete bays. It also brought floor slabs were much higher than the applied load and also
down all of the floors above the test enclosure by almost the confirmed the recorded stable structural behavior. Experi-
same amount of deflection at the column grid line. Although mental studies are now being carried out to gain better under-
the steel structure was still stable after this fire test, the fire standing of tensile membrane action in reinforced-concrete
damage was disproportionally excessive to the cause. In real- slabs under increasingly applied loads. These experimental
ity, the entire two bays of the building would have to be re- studies will provide a basis to modify the theoretical models
placed. The damage from this testing exceeded the expecta- to allow the tensile membrane action to be more reliably pre-
tions; however, this test provided an important finding dicted.
concerning the behavior of the whole structure under fire con-
ditions. The conclusion from this test is that in order to limit Design Implications
fire damage to the areas immediate to the fire, columns should
remain stable. For the conventional H-profiled steel columns, Although extensive theoretical studies and more targeted
external fire protection should be applied. If fire protection is small-scale demonstration tests are necessary before the results
to be completely eliminated from the steel structure, an alter- of the Cardington large-scale fire tests can be translated into
native column system with high inherent fire resistance should reliable design guidance, two clear messages emerge from this
be used. large-scale structural fire testing program. They are as follows:
Building Integrity
1. Even though the supporting steel beams may lose most
In fire resistant design, not only should a structure remain of their strength, the conventional composite floor slab
stable under the fire condition, but it should also prevent fire system can still have sufficient load carrying capacity to
spread by other means. One condition is that the structure sustain and to directly distribute the applied load to the
should satisfy the integrity condition so that fire does not columns. As the role of steel beams in the load bearing
spread by opening through any structural member. Due to the function is minimal, at elevated temperatures it may not
large scale of these fire tests, large deflections and rotations be necessary to fire protect them.
were observed. Although the steel structure remained stable 2. The stability of the columns is vital. They are not only
after each fire test, integrity failure was observed in one test necessary to ensure structural stability, but also to restrict
(Fire 1996). In this test, the slab corner rotation was so high the structural damage to the area local to the fire affected
that it detached from the column, creating an opening to allow enclosure. Results from the Cardington tests suggest that
smoke to spread to the floor above. However, this integrity columns made from the conventional H-profiled steel
failure could have been avoided if the anticrack mesh had been sections would need to be properly protected to remain
passed around the column to ensure better anchorage of the stable. Thus, to have a completely unprotected steel
slab to the column. structure, it is necessary to replace the H-profiled steel
columns with other forms. One option is to use concrete-
Numerical Studies filled steel tubular columns.
In the United Kingdom and Europe, various advanced mod- STUDIES ON CONCRETE-FILLED COLUMNS
els (Wang and Moore 1995; Bailey et al. 1996) of structural
Recent research at the NRCC on concrete-filled steel col-
analysis have been developed to study the behavior of steel-
umns complements, to some extent, European studies in fa-
framed buildings under fire conditions. These models have
cilitating the fully unprotected steel structures. The studies at
been used to conduct various theoretical studies to contribute
NRCC were mainly focused on steel HSS, and the main aim
to the development of a rational fire safety design approach
was to develop unprotected steel columns to obtain fire resis-
for steel-framed buildings. However, these models have only
tance as long as 3 h. Both experimental and theoretical studies,
been validated against test results on isolated structural mem-
using computer models, were carried out to investigate the
bers. To build confidence in their ability to predict the perfor-
influence of concrete filling on the fire resistance and load
mance of complete structures under fire conditions, the pre-
capacity of HSS columns.
dictions from these models are now being compared against
the results from the large-scale fire tests at Cardington. NRCC Studies
These theoretical studies have identified a number of fea-
tures that should be included in numerical modeling so that Steel HSS columns are very efficient structurally in resisting
the whole building behavior may be predicted accurately. They compression loads, and are widely used in the construction of
JOURNAL OF STRUCTURAL ENGINEERING / DECEMBER 2000 / 1447
plete building is significantly higher than that of individual Kemp, K. O. (1967). Yield of a square reinforced concrete slab on
steel members. Test data also indicate that in certain situations, simple supports, allowing for membrane forces. The Struct. Engr.,
it is possible to use unprotected steel beams in steel-framed London, 45(7), 235240.
buildings. However, the stability of columns is critical to en- Kodur, V. K. R. (1999). Performance-based fire resistance design of
sure the buildings structural safety and to restrict the extent concrete-filled hollow steel columns. J. Constructional Steel Res., in
of fire damage. press.
Kodur, V. K. R., and Lie, T. T. (1995). Fire performance of concrete-
The use of concrete-filled steel columns offers a practical
filled hollow steel columns. J. Fire. Protection Engrg., 7(3), 8998.
solution for facilitating completely unprotected steel-framed Kodur, V. K. R., and Lie, T. T. (1996). Evaluation of the fire resistance
building. Using this method, fire resistance in the practical of rectangular steel columns filled with fibre-reinforced concrete. Can.
range of up to 3 h can be obtained through any one of the J. Civ. Engrg., Ottawa, Canada, 24(3), 339349.
three types of concrete filling. Kodur, V. K. R., and MacKinnon, D. H. (1998). Simplified design of
Thus, through the use of conventional composite floor slabs concrete-filled hollow structural steel columns for fire endurance. 2nd
and concrete-filled steel tubular columns, fire protection to World Conf. on Steel Constr., 115.
Lie, T. T., and Kodur, V. K. R. (1996). Fire performance of concrete-
steel might be eliminated in certain situations. However, to
filled tubular columns. Int. Conf. on Tubular Struct., American Weld-
make this system acceptable in practice, further research is ing Society, Miami, 8696.
required to resolve a number of technical issues. Rubert, A., and Schaumann, P. (1986). Structural steel and plane frame
assemblies under fire action. Fire Safety J., 10, 173184.
ACKNOWLEDGMENTS Schleich, J. B. (1996). A natural fire safety concept for buildings1.
Proc., 2nd Cardington Conf., Building Research Establishment, U.K.
This paper is written as a result of the collaboration between the two Societe des Ingenieurs et Architectes (SIA). (1984). Fire risk evaluation
writers, supported by a North Atlantic Treaty Organization collaborative and calculation procedures. SIA 81, Zurich.
research grant (CRG 961179). The first writer of this paper thanks his
Steel Construction Industry Forum (SCIF). (1991). Investigation of
former colleagues at the United Kingdoms Building Research Establish-
Broadgate phase 8 fire. The Steel Construction Institute, Ascot, U.K.
ment for providing an opportunity to participate in the innovative Car-
Thomas, I. R., et al. (1992). The effect of fire in the building at 140
dington research project. The opinions expressed in this paper are the
William Street. BHP Res.Melbourne Lab. Rep., BHP, Melbourne.
writers own and do not necessarily represent those of their past or current
Underwriters Laboratories of Canada (ULC). (1989). Standard methods
employers.
of fire endurance tests of building construction and materials. CAN4-
S101-M89, Scarborough, Ont., Canada.
APPENDIX. REFERENCES Wang, Y. C. (1996). Tensile membrane action in slabs and its application
Armer, A. S. T., and Moore, D. B. (1994). Full-scale testing on complete to the Cardington laboratory fire tests. Proc., 2nd Cardington Conf.,
multi-storey structures. The Struct. Engr., London, 72(2), 3031. Building Research Establishment, U.K.
ASTM. (1990). Standard methods of fire tests of building construction Wang, Y. C. (1997). Tensile membrane action and fire resistance of steel
and materials. E 119-88, West Conshohocken, Pa. framed buildings. Proc., 5th Int. Symp. on Fire Safety Sci.
Bailey, C. G., Burgess, I. W., and Plank, R. J. (1996). Computer sim- Wang, Y. C., and Moore, D. B. (1995). Steel frames in fire: Analysis.
ulation of a full-scale structural fire test. The Struct. Engr., London, Engrg. Struct., 17(6), 462472.
74(6), 93100. Wood, R. H. (1961). Plastic and elastic design of slabs and plates,
Bailey, C. G., and Newman, G. M. (1998). The design of steel framed Thames and Hudson, London.
buildings without applied fire protection. The Struct. Engr., London, Yung, D., and Benichou, N. (1999). NFPRF 4th Annu. Fire Risk & Ha-
76(5), 781. zard Assessment Res. Application Symp.