Journal of Constructional Steel Research 65 (2009) 17941802

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

journal homepage: www.elsevier.com/locate/jcsr

Fire behaviour of hollow structural section steel columns filled with high
strength concrete
Peter Schaumann a , Venkatesh Kodur b , Oliver Bahr a,
a
Leibniz Universitaet Hannover, Institute for Steel Construction, Appelstr. 9a, 30167 Hannover, Germany
b
Michigan State University, Department of Civil and Environmental Engineering, East Lansing, MI 48824, USA

article info abstract

Article history: The use of hollow structural section (HSS) steel columns filled with high strength concrete (HSC) is
Received 21 July 2008 becoming popular due to many advantages they offer. However, whereas the design rules for HSS columns
Accepted 12 April 2009 filled with normal strength concrete are well established, there are many uncertainties for HSS columns
filled with HSC. Results from numerical studies on the behaviour of HSS columns filled with HSC are
Keywords: presented. The studies were carried out based on both North American and European material properties
High strength concrete
for HSC and steel. Results show that required fire resistance in HSS columns can be obtained through the
Hollow structural section steel columns
Fire resistance
use of bar- or steel fibre-reinforcement in HSC.
Tubular steel columns 2009 Elsevier Ltd. All rights reserved.

1. Introduction The selected material models were implemented in the finite

Concrete-filling in hollow structural section (HSS) steel columns
has many beneficial effects and is becoming more popular in
high-rise and industrial buildings. At ambient temperature, the
load-bearing capacity of these composite columns is significantly
increased. Thus, it is possible to reduce cross-sectional dimensions,
which results in reduced construction costs and enhanced usable
space in buildings. In addition, the advantageous thermal proper-
ties of concrete lead to increased fire resistance of these composite
columns. Therefore additional external fire protection for the steel
might be superfluous.
Though the fire performance of HSS columns filled with normal
strength concrete (NSC) is well established [1], the fire design
rules cannot be extended to high strength concrete (HSC) filled
steel columns. This is mainly attributed to faster deterioration
of material properties of HSC as compared to that of NSC. The
high temperature properties of HSC are not well established and
there are variations in the constitutive models from different
sources. To investigate the effect of these variations two different
approaches dealing with HSC were considered for the numerical
analysis of HSS columns. The first one was the North American
code provisions [2,3] and the work of the researchers Kodur and
Sultan [4] as well as Cheng, Kodur and Wang [5]. The second was
the European material properties according to Eurocode 2, part 12
(EN 1992-1-2) [6] for concrete structures and Eurocode 4, part 12
(EN 1994-1-2) [7] for composite structures.
Corresponding author. Tel.: +49 511 762 2512; fax: +49 511 762 2991.
E-mail addresses: bahr@stahl.uni-hannover.de, olli.bahr@web.de (O. Bahr).
0143-974X/$ see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jcsr.2009.04.013
element based computer program BoFIRE. A series of parametric
studies was carried out on typical HSC-filled tubular steel
columns to investigate column behaviour under fire exposure.
The parameters considered in numerical studies included type
of concrete-filling, moisture content, section size, cross-sectional
shape, column slenderness, load eccentricities and concrete
compressive strength.
2. Computer program BoFIRE
All parametric studies were carried out using the transient,
nonlinear, incremental computer code BoFIRE. This program,
written by Schaumann [8] and further developed by Upmeyer [9],
is capable of predicting thermal and structural behaviour of both
steel and composite structures exposed to fire. It is based on the
limit state design principle:
R(t ) S (t ) (1)
where R(t ) = resistance at time t; S (t ) = effect of mechanical
action at time of fire exposure t.
The load-bearing capacity of structures R(t ), which are charged
by a mechanical load S (t ) while exposed to fire, depends on the
modification of the material properties, such as decreasing of
strength and elastic modulus affected by heat. Thus, the procedure
for determining the remaining load-bearing capacity of structures
is based on a numerical calculation model coupling thermal and
mechanical response at various time steps. At first, thermal re-
sponse is evaluated through heat transfer analysis. At this stage,
the fire temperature and temperature distribution of the cross-
section are computed. According to the temperature distribution,
 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802
Notations
EU European material properties
NA North American material properties
NSC Normal strength concrete
HSC High strength concrete
HSS Hollow structural section
the modification of material properties caused by temperature is
determined. Subsequently, the mechanical response is evaluated
where deformation and remaining strength of the members are de-
termined. At each time step the computed strength is compared to
the applied load on the column to check whether the structure still
has sufficient load-bearing capacity. This procedure is repeated for
various time steps until the resistance of the member is less than
the applied load, which represents failure of the column. The dura-
tion to failure is taken as the fire resistance period of the column.
2.1. Thermal response
In BoFIRE, the temperature field is calculated using the Fourier
differential equation for heat conduction:
div ( grad ) + c f = 0
where = thermal conductivity; = temperature; c = heat
capacity; = density; = derivation of temperature with respect
to time; f = heat source.
Caused by the modifications of material properties due to heat
exposure, the differential equation becomes transient in order of
the inhomogeneous temperature field. Thus, this equation has to
be solved numerically. In what follows, the basis of that method
will be described. A mathematical transformation of Eq. (2) results
in the weak formulation of the differential equation:
Z Z
grad : grad dA + q ndS
Z relationship at elevated temperatures (mechanical property) as
+ c dA = 0
where = area; = boundary of considered area; q = heat flux;
n = normal vector on the boundary.
For the solution of the weak form, bilinear shape functions on a
four node isoparametric element according to Eq. (4) are used.
1
Ni = (1 ) (1 ) .
4 The stressstrain relationship at elevated temperatures has
The approach is presented in Fig. 1(a). An example for mesh gen-
eration with BoFIRE is shown in Fig. 1(b). The computer program
BoFIRE is capable of accounting for different material properties as
function of temperature including that of fire protection materials.
2.2. Mechanical response
BoFIRE is capable of analyzing mechanical response of different
cross-sectional types and linear structures such as beams, columns
or plane frames taking second order theory into account. The
calculations are based on the Bernoulli hypothesis for plain state
of strains. Shear deformations are not considered. Due to the
nonlinear material properties, cross-sectional values and internal
forces depend on the temperature field and cross-sectional strains.
The strength is evaluated by setting up equilibrium equations.
The solution of the incremental system equation is given by
Schaumann [8] as:
1S L 1S th = K It K It0 v t0 + 1K II v t0


+ K It + K IIt 1v


1795
a b
Fig. 1. Four node isoparametric element (a) and mesh for a concrete-filled HSS
column with embedded X-shape profile (b).
where 1S L = difference between external forces per time
increment; 1S th = difference between thermal strains per time
increment; (K It K It0 ) v t0 = difference of system matrix stiffness
(elastic portion); 1K II v t0 = difference of system matrix stiffness
(geometric portion according to second order theory); (K It + K IIt )
1v = difference of deformations per time increment.
At first, the internal force variables and deformations, caused by
the external forces 1S L , are computed in one or more increments.
In a parallel calculation the temperature field is established as
(2) previously described. Because of the incremental procedure it is
possible to linearize the influence of nonlinear material behaviour
and temperature distribution.
3. HSC material properties
The numerical studies were carried out by considering both
North American and European HSC material properties. There
are some major differences in high temperature mechanical
as well as thermal properties of HSC in North American and
European recommendations. The following comparison includes
the material properties of HSC which are relevant for structural
analysis of fire-exposed members. These are the stressstrain
(3) well as the specific heat capacity, conductivity, expansion and mass
loss (thermal properties).
In contrast to HSC, the material properties of steel are well-
established and hence not presented here. For details concerning
the material properties formulations of steel see Refs. [3,7].
3.1. Stressstrain relationship at elevated temperatures
(4)
crucial influence on the fire response of structural systems.
The following comparison between the North American (NA)
and European (EU) provisions illustrates the high temperature
stressstrain behaviour of HSC for siliceous aggregate concrete
with 60 MPa strength. The ascending branch of the stressstrain
relationships is presented for 100 C and 600 C in Fig. 2 according
to the NA and EU material properties.
It is obvious that the North American formulas are more
conservative since the peak stress is significantly reduced for
temperature of 100 C. In contrast to this, the peak stress does not
diminish in Eurocode 2 for the same temperature. This is also true
of other temperatures of 200 and 300 C, which is not presented in
Fig. 2. For higher temperature of 600 C, the peak stress according
to both provisions is almost equal. However, the North American
provisions assume far more brittle behaviour since the strain at
peak stress is less than the corresponding European value. The
differences for other temperatures are less pronounced and thus
not presented. Mathematical formulation of the North American
material properties are presented in detail by Kodur et al. [5] as
well as by Schaumann, Bahr and Kodur [10,11].
(5)
1796 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802
Fig. 2. Comparison between EU and NA stressstrain relationship for siliceous HSC
with compressive strength of 60 MPa at elevated temperatures.
Fig. 3. Comparison between NA and EU material properties for specific heat
capacity.
3.2. Specific heat capacity
Fig. 3 shows the variation of specific heat capacity as a
function of temperature, as obtained from North American
recommendation and European code provisions. According to
Eurocode 2 thermal properties of normal weight concrete may also
be applied for HSC. Contrary to Eurocode 2, the North American
recommendations differ between concrete with calcareous and
siliceous aggregates. According to the latter code, the initially
assumed specific heat capacity for siliceous aggregates is below
the corresponding European values. However, the North American
formula results in a peak value at 500 C for concrete with siliceous
aggregates. This is due to the presence of quartz that transforms
at these temperatures [4]. For concrete with calcareous aggregates
two peaks occur. The small peak near 400 C is caused by the
removal of crystal water from the cement paste. The higher peak
at temperatures of about 700 C results from the dissociation
of dolomite. At temperatures exceeding 800 C, the different
formulas converge to similar values.
In general, the differences can be explained by the fact that
the North American provisions take physio-chemical processes
into account. In contrast to this, the European approaches neglect
such phenomena. However, Eurocode 2 allows considering the
beneficial effect of increased moisture content on the specific heat
capacity in the temperature range from 115 C to 200 C, which is
not presented here. According to Eurocode 4 a moisture content of
10% may occur for HSS filled with concrete.
3.3. Thermal conductivity
The different approaches for thermal conductivity can be seen
in Fig. 4. Eurocode 2 defines thermal conductivity by lower and
upper limit. Highest values of the thermal conductivity result from
the North American provisions, whereas the minimum is given by
the lower limit of Eurocode 2.
Fig. 4. Comparison between NA and EU material properties for thermal
conductivity.
Fig. 5. Comparison between NA and EU material properties for the variation of the
thermal elongation.
3.4. Variation of thermal elongation
The computed values for the variation of the thermal elongation
= 1l/l are summarized in Fig. 5, where the influence of the
type of aggregate becomes clear. Because of the transformation of
quartz in the siliceous aggregates, thermal elongation increases for
temperatures about 450 C. For temperatures exceeding 650 C
thermal expansion remains constant. Thermal elongation for
concrete with calcareous aggregates is considerably lower for
temperatures less than 800 C.
3.5. Mass loss
The results for the mass loss are presented in Fig. 6. According
to the North American material properties, the overall mass loss
for concrete with siliceous aggregates only amounts to about
5%. Nevertheless, for temperatures exceeding 700 C the mass of
concrete with siliceous aggregates is sharply reduced to 70% of its
initial mass. The loss is caused by the dissociation of dolomite in
concrete. In contrast to this, Eurocode 2 does not differ between
concrete with siliceous and calcareous aggregates with respect to
mass loss.
4. Verification of HSC material properties
A review of the literature indicates that there is very limited
fire test data on HSC-filled HSS columns exposed to fire [12,13].
The main objective of the reported tests was to determine the
fire resistance of HSS steel columns filled with different types
of HSC. The tests, which have been reported on HSS columns,
include plain HSC-filling (columns C-46 and C-47), HSC with steel
fibres (columns C-36 and SQ-11) and HSC with 4 15 mm
reinforcement bars (column SQ-14). Investigated cross-section
types are presented in Fig. 7. Full details of the fire tests are
documented in [12,13].
 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802
Fig. 6. NA and EU material properties for mass loss.
Fig. 7. Cross-section and static system of circular HSC-filled HSS columns.
The tested HSS columns were 3.81 m long with fixed end
conditions and filled with calcareous aggregate (limestone) HSC.
After the load was applied, the columns were exposed to
heating according to standard fire ULC-S101-M04 [14] until failure
occurred. This fire exposure is equivalent to ASTM E-119 or ISO-
834 standard fire. The fire resistance period of some of the tested
columns (selected from literature) is calculated with the computer
program BoFIRE. This allows verification of material properties
given by the North American and European codes. Moreover, it
facilitates to study the influence of different types of concrete
filling on fire resistance. The material properties of steel fibre
reinforced HSC-filling are taken according to the published data by
Kodur and Sultan [4], which is reproduced in the Appendix. The
test data of the different HSS columns, taken from references [12,
13,15,18], is summarized in Table 1, where C in the column
designation denotes circular cross-section and SQ denotes square
cross-section.
The test and calculated fire resistance period of the different
HSS columns is summarized in Fig. 8. A moisture content of 10%
was considered in the computer program BoFIRE, which may occur
for HSS filled with concrete according to Eurocode 4, part 12.
Regarding the numerical results for HSS columns filled with plain
HSC it is obvious that the fire resistance is overestimated to a great
extent. This is true for both the North American and European HSC
material properties. With respect to the circular HSS columns C-
46 and C-47 the gap between tested and calculated fire resistance
increases for higher concrete compressive strength. However,
the results of both numerical models are conservative for HSC-
filling with additional steel fibres (columns C-36 and SQ-11) and
reinforcing bars (column SQ-14).
The divergence between computed and test fire resistance
period for plain HSC-filling can be explained from the test
experience gained in [16]. The steel HSS column expands faster
than the concrete core under fire load, initiating cracks in the
concrete. This effect can be considerable due to lack of any
reinforcement.
In the worst case scenario, the cracks magnify and might
separate the concrete cross-section into two halves as shown in the
elevation in Fig. 9. In the initial stage of fire exposure, the load is
mainly carried by the steel section. However, after approximately
1797
Fig. 8. Comparison between tested and calculated fire resistance.
Fig. 9. Elevation of the separating concrete crack of an HSS column filled with plain
concrete [16].
15 to 20 min the load has to be transferred from the steel section
to the concrete core. This is because the steel section yields due
to attaining critical temperature, which is linked with significant
reduction of its strength.
In case of load eccentricities, which can hardly be avoided,
the gap between the concrete cross-sections cannot be closed;
this leads to local failure of the column [16]. An example of local
buckling of a concrete-filled HSS column is shown in Fig. 10. The
concrete prevents the steel tube from inward buckling. However,
after exposure to fire, outer buckling could be noticed which means
that the tube and concrete were locally separated. As the program
BoFIRE does not take such local effects into account, it might
be worthwhile to establish a three-dimensional model for this
problem. This is outside the scope of this paper. Nevertheless,
by comparison to test data it is examined if the distribution of
temperatures derived from this computer program is adequate.
For this purpose, the points at which temperatures are
evaluated in BoFIRE (MP1 and MP2) are arranged according to
the location of thermocouples in test columns. As an example, in
column C-46 with plain HSC-filling, the measured and calculated
temperatures according to the HSC material properties defined
in Eurocode 2, part 12 [6] are presented in Fig. 11. A moisture
1798 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802

Table 1
Summary of fire resistance test data on HSC-filled HSS columns [12,13,15].
Tube filling Column Cross-section Test load (kN) Concrete strength (MPa) Fire resistance (min)

Plain HSC C-46 273.1 6.35 1050 82.2 48

C-47 273.1 6.35 1050 107.0 51
Steel fibres C-36 219.1 4.78 600 98.1 174
SQ-11 203.2 6.35 900 99.5 128
Rebars SQ-14 203.2 6.35 1150 81.7 89

Fig. 12. Recorded and calculated temperatures for column C-46 with varying
moisture content (u) (calculated with EU material properties).

Fig. 10. Local buckling of concrete-filled HSS column [16].

Fig. 13. Recorded and calculated temperatures for column SQ-14 according to EU
material properties.

Fig. 11. Recorded and calculated temperatures for column C-46 according to EU
material properties.

content of 10% of the concrete weight is assumed in the analysis

since the moisture can hardly escape due to the surrounding steel
section. The comparison is carried out for 48 min, at which point
the column failed in the test. It can be seen from Fig. 11 that
the predicted BoFIRE temperature at location MP 2 is in good
agreement with the recorded temperatures in fire tests. As regards
to MP 1 located in the center of the column, a maximum difference
between computed and measured temperatures of approximately
53 C has to be noticed. Overall there is a fair agreement between
predicted and measured temperatures.
The analysis was extended to study the influence of moisture
content (u) on the temperature field and the fire resistance period.
It is evident from Fig. 12 that the computed temperatures for
MP1 decrease with increasing moisture content u. The difference
between ratios of u = 0% and u = 10% already amounts to about
120 C after 48 min and hence influences the fire resistance period.
For example, the fire resistance periods for u = 0% and u = 10%
are 89 min and 110 min, respectively. The predicted temperatures
for moisture content of u = 10% stand in best accordance with
measured temperatures.
For comparison between the European and North American
material properties, the computed temperatures for column
Fig. 14. Recorded and calculated temperatures for column SQ-14 according to NA
material properties [2,3].
SQ-14 (filled with bar reinforced concrete) are presented in
Figs. 13 and 14.
Results in Figs. 13 and 14 clearly infer that the use of
the European material properties leads to more realistic cross-
sectional temperatures for this case. In addition, prediction of
fire resistance period stands in good agreement with the real fire
endurance. On the whole, it can be concluded that the calculation
of the temperature field with the program BoFIRE is appropriate.
Therefore the great difference between tested and computed
fire resistance period of plain HSC-filling must be due to micro
cracking, which leads to faster loss of strength with temperature.
 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802
Fig. 15. Axial deformations recorded in test for specimens C-46, SQ-11 and SQ-14.
To further illustrate this, a comparison of the tested and measured
axial deformations for column with plain HSC-filling (C-46), the
bar reinforced concrete-filled HSS steel column (SQ-14) and the
fibre reinforced concrete-filled column (SQ-11) is shown in Fig. 15.
This is only a qualitative comparison as load ratios vary slightly.
Columns C-46 and SQ-11 have load ratios of 17% compared to
24% for column SQ-14, both related to load capacity at room
temperature.
The general load-bearing behaviour is the same for all three
columns: In the early stages of fire exposure, the load share of
the steel section increases significantly since it expands faster than
the concrete core. After approximately 20 min, the load has to be
carried by the concrete core due to the decreased yield strength
of the steel section leading to contraction of the whole column
and finally to failure [12]. As it can be seen in Fig. 15, the type
of concrete-filling influences the contraction phase. The decline is
more gradual for both reinforced columns, namely SQ-11 and SQ-
14, whereas the failure occurs suddenly for column C-46 with plain
HSC-filling. The latter is due to the brittle nature of HSC resulting in
cracks in the concrete core and hence local failure in the steel tube,
as explained in Section 4. Because of the induced load eccentricities
initiated by cracks, the column cannot sustain the load [16].
The following parametric studies address the effect of different
parameters on the fire behaviour of HSS columns filled with plain
HSC-filling. Referring to Fig. 8, it is clear that fire resistance of
these columns is overestimated. This can be primarily attributed
to the fact that the computer program BoFIRE is based on
Bernoullis assumption of cross-sections remaining plane. Thus, it
is not possible to capture local effects such as debonding of the
steel from the concrete core which results in premature failure
of the unreinforced columns. Therefore, subsequent studies are
qualitative. It should be noted that the aim of this study is not to
determine the exact fire resistance time.
5. Effect of section size
For steel columns, it is well known that section size and thus the
cross-sectional parameter A/V , where A is the area of the fire-
exposed surface and V is the volume of a member both per unit
length, influences fire resistance. However, the effect of A/V -ratio
has fully been studied in literature for concrete-filled HSS columns.
To illustrate the effect of section size on fire resistance of
concrete-filled HSS columns, a parametric study was carried out on
ten different circular HSS columns filled with HSC. For excluding
effects of slenderness, the selected columns had a length of only
0.50 m (short columns) with fixed end conditions. The yield
strength of the steel tube was assumed to be 235 MPa. As filling,
concrete with compressive strength of 40, 60 and 90 MPa and with
calcareous aggregates were chosen. These strengths are equivalent
to 50, 75 and 105 MPa based on cube strength of concrete as
measured in Europe.
1799
Fig. 16. Effect of varying A/V -ratio on fire resistance period of circular HSS columns
with HSC-filling.
Fig. 17. Comparison between fire resistance periods computed using EU and NA
material properties for varying A/V -ratios and concrete strength classes.
The steel sections were selected such that they contribute about
one third to the total load-bearing capacity at room temperature.
Calculation was carried out with material properties according to
Eurocodes and specified load level of 50% at room temperature. The
interrelationship between the fire resistance period of a column
and its cross-sectional parameter A/V is presented in Fig. 16.
It is obvious that a column with low A/V ratio exposes a
relatively small surface to the fire, delaying its heating. Thus, the
use of HSC as HSS-filling is in particular interesting for massive
columns since its thermal properties do not play crucial role as in
the case of leaner columns. This results in remarkably enhanced
fire resistance. Yet this advantage diminishes with increasing A/V -
ratio. For the given load level, the application of HSC might not be
reasonable for ratios exceeding a value of 15.
The parametric study is once more repeated using the North
American material properties. For direct comparison between both
approaches, each fire resistance period computed using the North
American approach is divided by the analogous value calculated
according to the European standard. The ratio for different
concrete classes is presented for varying cross-sectional parameter
A/V in Fig. 17. Two main conclusions can be drawn. First,
North American results become more conservative for increasing
concrete strengths. Second, this trend increases for decreasing
cross-sectional parameter. Therefore the greatest difference is
obtained for massive columns consisting of HSC with compressive
strength of 90 MPa. The corresponding ratio is 0.54 meaning that
the fire resistance computed using North American codes is only
about half of the resistance derived using European codes.
6. Effect of cross-section shape
The fire resistance of a column is influenced by the shape
of its cross-section. To illustrate this effect, a numerical study
was carried out. Consequently, three of the previous investigated
circular cross-sections (denoted as sets CS1 to CS3 in Table 2)
1800 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802
Table 2
Results from parametric studies on the effect of cross-sectional shape.
Cross-section A/V -ratio (1/m) Load for C90/105 (kN)
Circular Square Circular Square
CS1 C610 16 SQ610 12.5 7 15,316 20,628
CS2 C273 8 SQ273 6.2 15 3,117 4,214
CS3 C139.7 4 SQ139.7 3.1 29 813 1,103
Fig. 18. Comparison between circular and square cross-sections with same A/V -
ratio.
were taken as the basis of comparison. Since it was shown that
the A/V -ratio has crucial influence on the fire endurance, the
square HSS columns selected for the comparison had the same
A/V -ratio as the circular HSS columns. In addition, the wall
thickness of the square HSS columns was calculated such that
the resulting steel area for the square section was the same as
for circular cross-section. This was necessary to account for the
decreasing material properties of steel at elevated temperatures.
The examined columns can be found in Table 2. The HSS columns,
which had a load level of 50% at room temperature, were filled with
calcareous aggregate C90/105 concrete. The first value denotes the
compressive strength of the concrete in MPa based on cylinder
strength, whereas the second value is based on cube strength.
Second-order effects were excluded with a chosen column length
of only 0.50 m (short columns) with fixed end conditions.
The fire endurance for the different steel HSS columns with
HSC-filling is presented in Fig. 18. It is apparent that the circular
cross-sections showed slightly better fire resistance period than
the square cross-sections with same A/V -ratio. Whereas the differ-
ence in fire resistance between the massive circular C610 16 mm
and square SQ610 12.5 mm cross-sections is only 6% (CS1), it
amounts to about 11% for two other examples (CS2 and CS3).
The lower fire resistance in square columns can be explained
by the fact that square cross-section corners were heated from
two sides, which led to an increased supply of heat, higher loss
of strength and thus earlier failure. This is illustrated in Fig. 19. In
the case of square columns, the temperature at a given distance
from the edge was higher at the corner than that along the
sides caused by the heat transfer from two sides. In contrast to
this, the circular cross-section showed a uniform temperature
distribution along the radial lines. However, the investigation
illustrates that the difference between the two cross-sections
concerning fire resistance period is moderate. Therefore, mostly
circular cross-sections are considered in the later on carried out
studies. Considering that square cross-sections show lower fire
endurance about 10%, the numerical results for the circular HSS
columns in general allow fire resistance assessment of square
cross-sections with comparable A/V -ratio.
7. Effect of slenderness
Columns are mainly compression members and hence suscep-
tible to bending with respect to column slenderness and load
eccentricity. Thus, a numerical study was carried out to set a rea-
sonable limitation on eccentricity for the use of HSC as HSS-filling.
The parametric study was performed with a circular cross-section
C219.1 6 mm and hinged end conditions. A HSS column was filled
with calcareous aggregate concrete with compressive strength of
40, 60, 80 or 100 MPa based on cylinder strength. This corresponds
to compressive strength of 50, 75, 95 and 115 MPa based on cube
strength.
With regard to the tube, conventional yield strength of 235 MPa
and improved yield strength of 355 MPa were chosen for the first
two cross-sections. The third cross-section consisted of a tube with
yield strength of 235 MPa and 4 25 mm reinforcement bars.
The influence of slenderness was examined by varying the column
length from 0.1 to 6 m with an assumed imperfection of L/1000.
Both the North American and European material properties for
HSC were used. The cross-sections were heated 30 min according
to the ISO time-temperature curve. This standard curve according
to the European code Eurocode 1, part 12 [17] is very similar
to the North American standard time-temperature curve, which
is defined by the code ULC-S101 [14]. The short heating period
of 30 min already showed the unfavorable fire endurance of
slender HSC-filled HSS columns. For longer fire exposure times,
the reduction in ultimate load would be even more rapid. After
the heating process, the load was gradually increased until failure
occured.
7.1. Calculation with tube yield strength of 235 MPa
Results from the analysis are plotted in Fig. 20, which presents
the ultimate load-bearing capacity (at 30 min) as function of
column length. The HSC-filled HSS column with tube yield strength
of 235 MPa was concentrically loaded. Using the NA and EU
material properties, the ultimate load after fire exposure of 30 min
to ISO standard fire was determined.
As expected, load carrying capacity decreased with increased
column length. The North American code provisions were more
conservative for both NSC and HSC under concentric load, which
is also illustrated in Fig. 20. This is in particular true for the HSC
class C100/115 as the European results for non-slender columns
exceeded the North American predictions about 25%.
For column lengths exceeding 2 m similar ultimate loads were
obtained using both approaches. This can be attributed to the
fact that these columns fail by global buckling, which is largely
influenced by the elastic modulus. The latter is very similar for both
approaches. Thus, no great differences with respect to buckling
could be expected.
The results for moderate eccentric loads with e/h = 0.25
are given in Fig. 21. The tube yield strength was 235 MPa.
The North American code provisions still calculated the more
conservative results. Nevertheless, the difference between the
approaches was insignificant for column lengths exceeding 2 m.
Due to its low tensile strength, the beneficial effect of concrete
with high compressive strength was sharply reduced for slender
columns and load eccentricities causing bending moments. Thus,
the use of HSC as filling for HSS columns should be limited to non-
slender columns with only moderate load eccentricity.
7.2. Calculation with tube yield strength of 355 MPa
The average steel temperature of the tube was 860 C for a
fire exposure of 30 min. Due to its reduced strength and elastic
modulus, the load share of the steel tube was negligible. Thus, the
HSS column with the improved tube yield strength of 355 MPa
carried only slightly higher loads than the cross-section with a steel
tube having yield strength of only 235 MPa. Therefore the results
are not presented in detail.
 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802
5 0 5
5 5
0 0
5 5
5 0 5 7
Fig. 19. Temperature field for square SQ139.7 3.1 mm and circular C139.7 4 mm cross-section (CS3) after fire exposure of 32 min.
Fig. 20. Ultimate load for concentrically loaded HSC-filled HSS column using NA
and EU HSC material properties at 30 min fire exposure.
Fig. 21. Ultimate load for eccentrically loaded HSC-filled HSS column using NA and
EU HSC material properties at 30 min fire exposure.
7.3. Calculation with tube yield strength of 235 MPa and reinforce-
ment bars
The addition of reinforcement bars in concrete filling is helpful
in improving fire resistance. This can be seen in Fig. 22, where the
HSS column with the tube of improved yield strength of 355 MPa is
compared to the conventional tube with yield strength of 235 MPa
and 4 25 mm reinforcement bars.
With a concrete cover of 50 mm, the reinforcement bars were
to some degree protected against heating. The outcome was an
average temperature of 500 C in reinforcement bars remarkably
reducing the elastic modulus to about 60% of its initial value
according to Eurocode 4. Nevertheless, in comparison to the
directly fire-exposed steel tube with an average temperature of
about 860 C and corresponding reduction to only 8% of the initial
1801
7
2 3
Fig. 22. Comparison of ultimate loads between HSS column with tube yield
strength of 235 MPa and additional reinforcement bars and tube of improved yield
strength of 355 MPa both under eccentric load of e/h = 0.25 using European
material properties at 30 min fire exposure.
elastic modulus, the beneficial effect of the column concrete cover
is evident. Thus, the cross-section with additional bars showed
superior behaviour for slender columns since the elastic modulus
is crucial for the buckling failure.
Reasonable use of cross-sections with the improved tube is
restricted to about 3 m for the examined column. In contrast to
this, cross-sections with reinforcement bars are able to bear a
comparable amount of load up to a column length of approximately
6 m. In addition, the reinforcement bars allow the HSC to develop
its beneficial effect on the load-bearing behaviour as the difference
between the concrete strength classes are more pronounced than
for the cross-section with yield strength of 355 MPa.
8. Conclusions
Based on the information presented, the following conclusions
concerning the fire resistance of high strength concrete-filled HSS
steel columns can be drawn.
- The computer code BoFIRE is capable of modelling the fire
response of HSC-filled HSS columns with different types of
reinforcement (steel fibres and bar reinforcement).
- Local effects like gaping cracks and local plastic buckling, which
might occur in concrete-filled HSS columns without reinforce-
ment, are not covered by most conventional simulation tools.
This might lead to exaggerated results.
- High strength concrete-filling is particularly beneficial for
massive columns with low A/V -ratio as the material properties
of high strength concrete at elevated temperatures do not play
crucial role as in the case of leaner columns.
1802 P. Schaumann et al. / Journal of Constructional Steel Research 65 (2009) 17941802

- Circular concrete-filled HSS steel columns provide higher fire c cp ( ) = 3.81 106 for 0 C 400 C
resistance than square concrete-filled HSS columns of similar c cp ( ) = (0.0165 + 10.41) 106 for 400 C < 475 C
cross-section area. c cp ( ) = (0.0079 1.182) 106 for 475 C < 625 C
- The use of HSC as filling in HSS columns is beneficial from c cp ( ) = (0.2333 142.06) 106 for 625 C < 700 C
fire resistance point of view for non-slender columns with c cp ( ) = (0.1800 + 147.25) 106 for 700 C < 800 C
only moderate load eccentricity. This recommendation is the c cp ( ) = 3.25 106 for 800 C < 1000 C .
result of an investigation on fire exposure of up to 30 min.
Since the disadvantageous load-bearing behaviour of HSC- A.3. Thermal conductivity
filling increases with rising temperatures, this trend is also valid
for higher fire resistance classes. c = 1.80 0.0016 c for 0 C 500 C
- The fire resistance of HSC-filled HSS columns can be signifi- c = 1.20 0.0004 c for 500 C < 1000 C .
cantly enhanced by using reinforcement in the concrete core.
This is true for both steel fibre and conventional bar reinforce- A.4. Thermal elongation
ment.
- There are variations between the considered European and
c ( ) = 2.00 104 + 9 106 for 0 C 700 C
North American material properties resulting in different
c ( ) = 3.45 102 + 58 106 for 700 C < 870 C
fire resistance times. Numerical studies showed that North
c ( ) = 1.60 102 for 870 C < 1000 C .
American results are more conservative for increasing concrete
strengths and decreasing cross-sectional parameter.
- In general, the results for HSS columns filled with normal A.5. Variation of density
strength concrete are in fair agreement using the European
and North American material properties. In contrast to this, ( )/(20 C ) = 1.003 6 105 for 0 C 700 C
significant differences could be noticed concerning the results ( )/(20 C ) = 2.214 1.79 103 for 700 C < 785 C
for HSC-filling. One of the primary factors that produce these ( )/(20 C ) = 0.817 1.00 105 for 785 C < 1000 C .
differences is the variation in high temperature material
properties of HSC. Hence the authors recommend further
References
research on the high temperature material properties of HSC.
[1] Kodur VKR, Lie TT. Fire performance of concrete-filled hollow steel columns.
Journal of Fire Protection Engineering 1995;7(3):8998.
Appendix [2] Canadian Standards Association. CSA A23.3-M04. Code for the design of
concrete structures for buildings. Rexdale, Ontario, Canada; 1994.
[3] Canadian Standards Association. CSA S16.1. Limit states design of steel
The following formulae describe the material behaviour of HSC structures. Rexdale (Ontario, Canada); 2001.
with steel fibres at elevated temperatures according to Kodur and [4] Kodur VKR, Sultan MA. Effect of temperature on thermal properties of high-
Sultan [4] and are referred to as North American material models, strength concrete. Journal of Materials in Civil Engineering 2003;15(2):1017.
[5] Cheng FP, Kodur VKR, Wang TC. Stressstrain curves for high strength concrete
for HSC, in the main text. These material properties are defined for
at elevated temperatures. Journal of Materials in Civil Engineering 2004;16(1):
concrete strengths between 55 and 110 MPa. 8494.
[6] European Committee for Standardization. EN 1992-1-2 (Eurocode 2). Design
of concrete structures, Part 1-2: General rulesStructural fire design. Brussels;
A.1. Stressstrain relationship 2004.
" [7] European Committee for Standardization. EN 1994-1-2 (Eurocode 4). Design
H # of composite steel and concrete structures, Part 1-2: General rulesStructural
max c

fc = fc0 1 for: c c ,max fire design. Brussels; 2005.
max [8] Schaumann P. Computation of steel members and frames exposed to fire (in
German: Zur Berechnung staehlerner Bauteile und Rahmentragwerke unter
!2 Brandbeanspruchung). Bochum: Technisch-wissenschaftliche Mitteilungen
30 (max c ) Nr. 84-4. Institut fr konstruktiven Ingenieurbau, Ruhr-Universitt; 1984.
fc = fc0 1 for: c > c ,max [9] Upmeyer J. Fire design of partially encased composite columns by ultimate
max
0


130 fc0 fire loads (in German: Nachweis der Brandsicherheit von kammerbetonierten
Verbundbauteilen ueber Grenzbrandlasten). Hannover: Institute for Steel
where: Construction, Leibniz Universitaet Hannover, Shaker Verlag; 2001.
[10] Schaumann P, Bahr O, Kodur VKR. Numerical studies on HSC-filled steel
H = 2.28 0.012 fc0
0 columns exposed to fire. In: Tubular structures XI. London: Taylor and Francis
Group; 2006.
max = 0.0018 + 6.7 fc00 + 6 + 0.03 2 106


[11] Schaumann P, Bahr O, Kodur VKR. Fire resistance of high strength concrete-
filled steel columns. SIF06 - 4th International Workshop Structures in Fire;
and: 2006.
[12] Kodur VKR. Performance of high strength concrete-filled steel columns
fc0 = fc0
0
(1.0625 0.003125 ( 20)) for: < 100 C exposed to fire. Canadian Journal of Civil Engineering 1998;25:97581.
[13] Kodur VKR. Solutions for enhancing the fire endurance of steel HSS columns
fc0 = 0.75 fc0
0
for: 100 < 400 C filled with high strength concrete. AISC Steel Construction Journal 2006;43(1):
fc0 = fc0
0
(1.33 0.00145 ) for: 400 C 17.
[14] Canadian Standards Association. CAN/ULC S101-4. Standard methods of fire
such that: endurance tests of building construction and materials. Rexdale (Ontario,
Canada); 2004.
0 fc0 fc0
0
[15] Lie TT, Kodur VKR. Fire resistance of steel columns filled with bar-reinforced
concrete. Journal of Structural Engineering 1996;122(1):306.
Unit: MPa. [16] Kordina K, Klingsch W. Fire resistance of composite columns and of solid
steel columns - part I (in German: Brandverhalten von Stahlstuetzen im
Verbund mit Beton und von massiven Stahlstuetzen ohne Beton). Duesseldorf:
A.2. Specific heat capacity Studiengesellschaft fr Anwendungstechnik von Eisen und Stahl e.V.; 1983.
[17] European Committee for Standardization. EN 1991-1-2 (Eurocode 1). Actions
on structures, Part 1-2: General actionsActions on structures exposed to fire.
The specific heat is the product of specific heat capacity and
Brussels; 2002.
density. A value of 2535 kg/m3 is assumed for the density of HSC [18] Kodur VKR, Wang TC, Cheng FP. Predicting the fire resistance behavior of high
with calcareous aggregates. strength concrete columns. Cement and Concrete Composites Journal 2004;
26(2):14155.