Mechanical Bahviour of ultra-high strength concrete at elevated temperatures and fire resistance of ultra-high strength concrete filled steel tubes

ÔØ Å ÒÙ× Ö ÔØ
Mechanical behaviour of ultra-high strength concrete at elevated temperatures
and fire resistance of ultra-high strength concrete filled steel tubes
Ming-Xiang Xiong, J.Y. Richard Liew
PII:
DOI:
Reference:

S0264-1275(16)30655-4
doi: 10.1016/j.matdes.2016.05.050
JMADE 1798

To appear in:
Received date:
Revised date:
Accepted date:

2 February 2016
2 May 2016
13 May 2016

Please cite this article as: Ming-Xiang Xiong, J.Y. Richard Liew, Mechanical behaviour
of ultra-high strength concrete at elevated temperatures and fire resistance of ultra-high
strength concrete filled steel tubes, (2016), doi: 10.1016/j.matdes.2016.05.050

This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

ACCEPTED MANUSCRIPT
Mechanical Bahviour of Ultra-High Strength Concrete at Elevated Temperatures and
Fire Resistance of Ultra-High Strength Concrete Filled Steel Tubes

School of Civil and Transportation Engineering, Guangdong University of Technology, 100
Waihuan Xi Road, GuangZhou, China 510006

b

Department of Civil Engineering, National University of Singapore, 10 Kent Ridge Crescent,

NU

Singapore 117576

School of Civil Engineering, Nanjing Tech University, 30 Puzhu Road(S), Nanjing, China

MA

211800

CE
P

TE

D

* Corresponding author, Email address:

AC

c

SC
R

a

IP

T

Ming-Xiang Xiong a, b, *, J.Y. Richard Liew b, c

1


ACCEPTED MANUSCRIPT
Abstract
This paper introduces experimental study on mechanical behaviour of an ultra-high strength

T

concrete (UHSC) at elevated temperatures and then a simple calculation method to predict

IP

the fire resistance of tubular column infilled with the UHSC. The cylinder compressive

SC
R

strength of the UHSC was 166 N/mm2 at room temperature. The compressive strength and
modulus of elasticity of the UHSC were measured up to 800°C. Then the temperature-

NU

dependent mechanical properties were compared with those of normal/high strength
concretes provided in Eurocode 2 and ANSI/AISC 360-10, and with those of concretes in

MA

literature. The comparisons showed that the compressive strength and elastic modulus of the
UHSC were generally reduced less than those of normal/high strength concretes at the

D

elevated temperatures. The temperature-dependent mechanical properties were proposed for

TE

evaluating fire resistance of steel tubular columns infilled with the UHSC. The UHSC

CE
P

investigated in this project was shown to markedly improve the fire resistance in a number of

cases well documented in the literature concerning tubular columns filled with the normal-

AC

and high-strength concretes.
Keywords:

Ultra-High Strength Concrete, Elevated Temperatures, Mechanical Properties, Concrete
Filled Steel Tubular Column, Simple Calculation Method, Fire Resistance.

2


ACCEPTED MANUSCRIPT
1

Introduction

High strength concrete (HSC) has been used in high-rise buildings and the other structures

T

because of its technical, architectural, and economical advantages over normal strength

IP

concrete (NSC). However, the need for sustainable constructions around the world, which

SC
R

aims to further reduce the consumption of construction materials, requires higher-strength
concretes to be introduced. Nowadays, ultra-high strength concrete (UHSC) with

NU

compressive strength higher than 120MPa has been available with the development of
concrete technology and the availability of variety of materials such as silica fume and high-

MA

range water-reducing admixtures. However, the UHSC is mainly used in offshore and marine
structures and for industrial floors, pavements and security barriers. It has not been used in

D

building structures especially high-rise buildings. This may be due to the fact that there are

TE

design concerns on its brittleness and fire resistance. These concerns lead to the situations

CE
P

that the current standards allow the use of concrete only up to Class C90/105 for concrete
structures and Class C50/C60 for steel-concrete composite structures [1-4].

AC

To evaluate the fire resistance of structural members with the UHSC, the knowledge of the
temperature-dependent mechanical properties, such as compressive strength and modulus of
elasticity, is required. In literature, the said properties of the NSC and HSC have been
extensively studied where the compressive strength was found to be affected by the type of
aggregate [5-7]. Siliceous-aggregate concrete brought in greater strength losses than concrete
with carbonate aggregate, whereas firebrick aggregate exhibited superior performance. The
strength was also affected by heating rate [8]. Higher heating rate generally yielded lower
strength and was more likely to induce spalling. Furthermore, the loss of strength of HSC was
larger than that of NSC [2; 9-13]. The modulus of elasticity was generally governed by the
type of aggregate and the water/cement ratio [14-16]. The loss of modulus increased as the

3


ACCEPTED MANUSCRIPT
water/cement ratio increased. According to the literature [7; 8; 17], the elastic modulus is less
affected by the temperature in HSC compared with NSC. The addition of fibers is deemed to

T

affect the mechanical properties of concrete. Steel fibers generally increase both of the

IP

compressive strength and elastic modulus [18]; whereas polypropylene fibers decrease the

SC
R

compressive strength but increase the elastic modulus [19]. Overall, there is still little

information in the available literature concerning the mechanical properties of UHSC at high

NU

temperatures. Research efforts in this domain are, therefore, badly needed indeed.
Due to the brittleness, HSC is generally used in hollow steel tubes to form composite

MA

columns. Concrete filled steel tubular (CFST) column integrates the respective advantages of
steel and concrete materials thus exhibits many advantages over conventional steel or

D

reinforced concrete columns, such as high load bearing capacity, good ductility due to

TE

confinement effect, and convenience for fabrication and construction due to permanent

CE
P

formwork from steel tubes [20]. The CFST columns also have good fire resistance due to heat
sink effect of the infilled concrete and prevention of spalling of the infilled concrete by the
steel tube. Researches on the fire resistance of CFST columns started from 1970s. National

AC

Research Council of Canada (NRCC) is the pioneer in this area [21-24]. Until now, the

researches on the CFST columns with HSC have been carried on by Kodur [25], Lu et al.
[26-28] and Romero et al.[29]. However, little information is found for studies on CFST
columns with the UHSC of compressive strength higher than 120MPa.
A concept of CFST column with the UHSC was proposed for load-bearing system of the
high-rise building constructions [30; 31]. The compressive cylinder strength of the UHSC
exceeded 160MPa. This paper presents a study on the mechanical properties, such as the
compressive strength and modulus of elasticity, of the UHSC under elevated temperatures.
The temperature dependent properties were obtained through standard compression tests.

4


ACCEPTED MANUSCRIPT
With the tested mechanical properties, the fire resistance of CFST columns with the UHSC
was evaluated when they were subject to standard ISO-834 fire, and compared with that of

Basic Materials

SC
R

2

IP

T

CFST columns with the NSC and HSC.

The basic materials to produce the UHSC were Ducorit® D4 and water. Ducorit® D4 is one

of the commercial Ducorit® products. It is made from cementitious mineral powder,

NU

superplasticizer and fine bauxite aggregates with maximum sizes less than 4.75mm and 49%

MA

less than 0.6mm. The mixing proportions for the UHSC are shown in Table 1. Workability of
the fresh UHSC was tested using the slump flow test in accordance with ASTM

D

C1611/C1611M-09b. The slump flow spread was 735mm and the density was 2700 kg/m3

Standard Compression Tests at Elevated Temperatures

CE
P

3

TE

[32].

3.1 Test Specimens

AC

Spalling has been found for the HSC subject to high temperatures [33]. The spalling is
basically caused by thermal stresses due to a temperature gradient in concrete during heating,
and by splitting force due to the release of vapor above 100oC. It is believed that the present
UHSC is more likely to spall under high temperatures. With regard to this point, a series of
trial tests have been done to investigate the spalling behavior of the UHSC [34]. It was found
that the plain UHSC specimens and the UHSC specimens with steel fibers (dosage up to 1.0%
in volume) spalled around 490oC as shown in Figure 1 and Figure 2, respectively. The
spalling was so severe that the cover plate of the casing was bent and the ceiling of the
furnace was damaged. However, the UHSC specimens with 0.1% polypropylene fibers did
not spall at elevated temperature up to 800oC as shown in Figure 3. The properties of steel
5


ACCEPTED MANUSCRIPT
and polypropylene fibers are shown in Table 2. It is worth noting that the workability and
flowability of the UHSC were not affected by the addition of polypropylene fibers as the

T

UHSC is most likely pumped into hollow tubes for CFST columns. The dosage of

IP

polypropylene fibers was lower than that recommended by Eurocode 2 where more than

SC
R

2kg/m3 (0.25% in terms of volume) of monofilament propylene fiber should be included in
the HSC mixtures to prevent spalling [2].

NU

For the standard compression tests, cylinder specimens with a nominal diameter of 100mm
and a height of 200mm were prepared. The actual diameters and heights were measured

MA

before the test started. The specimens were cured in lab air where the relative humidity was
approximately 85% and the room temperature was around 30oC at daytime and 25oC at night.

D

Owing to the fact that the moisture content in the UHSC is low, the effect of moisture on the

TE

mechanical properties is deemed to be insignificant [34]. On the other hand, the moisture is

CE
P

evaporated around 100oC, it may only have minor influence at 100oC but insignificant
influences at higher temperatures. Considering these, the unsealed specimens were used.

AC

3.2 Test Setup

The compression tests were conducted by means of a servo-hydraulic testing machine with a

maximum 300mm stroke displacement and capacity of 10000 kN. The heat system was a
split-tube furnace with a two-zone configuration and an optional side entry extensometer port.
The furnace is constructed with S304 stainless steel shell and alumina insulation material.
Heating elements are coils of Fe-Cr-Al alloy 0Cr27a17mo2. A type K thermocouple is
mounted in the center of each heating zone. The external dimensions (diameter x height) are
700 x 600mm and internal heating dimensions (diameter x height) are 350 x 400mm. The
furnace can heat up to a maximum temperature of 900oC. Model 3548HI high temperature
furnace extensometer was used to measure the relative deformation in gauge length of the
6


ACCEPTED MANUSCRIPT
specimen. It is a strain gauged sensor and specified for a gauge length of 50mm and the
maximum measurable strain is thus 20%. The arms of the extensometer are alumina rods and

T

the rods were attached to the middle 1/4 height of the specimen which is the gauge length.

IP

The test setup is shown in Figure 4. Top and bottom cooling blocks were used to load the

SC
R

specimen inside the furnace. The cooling blocks were made from carbon steel which is not
resistant to high temperature. To bring down its temperature, channels were drilled inside the

NU

cooling blocks to allow for water circulating for the purpose of cooling. The concrete
specimen was protected by a steel casing in case where the crushing debris at failure would

MA

damage the furnace. Diameter of 10mm holes were drilled on surface for heat propagation;
and opening was cut at side for the pass of rods of the extensometer. The compression force

TE

D

was applied from the bottom of the loading frame by a hydraulic cylinder.

CE
P

3.3 Test Method and Procedure

In practice, different temperature–stress paths may appear in concrete and it is difficult to test
for all of them. Typically, two temperature-stress paths, unstressed and stressed, are

AC

considered to form the upper and lower bounds of the mechanical properties of concrete at
elevated temperatures. For the unstressed method, the specimen is loaded to fail with a
constant temperature; whereas the specimen is heated to fail under a constant load level for
the stressed method. The unstressed method is mostly used due to its convenience to obtain
stress-strain curves directly. However, it is difficult to obtain the stress-strain curves in the

stressed tests as the measured strain includes thermal strain and short-term creep strain [35].
Supplementary tests are usually required to measure them independently. The difference
between the unstressed and stressed test methods is mainly that the stressed test could capture
transient thermal strain. For a CFST column subjected to a fire, ignoring the transient thermal
strain could overestimate the buckling resistance of the CFST column, however the
7


ACCEPTED MANUSCRIPT
overestimation may not be much severe due to the existence of non-uniform temperature
distribution through its cross-section [36]. Nevertheless, the influence of transient thermal

T

strain should be considered for the fire resistance design of CFST columns. In present study,

IP

the unstressed test method was adopted and the effect of transient thermal strain was

SC
R

implicitly considered by a stiffness reduction factor given in Section 5.1 for the CFST
columns containing the UHSC. The validity of the said reduction factor has been established

NU

by test results.

For the unstressed tests conducted, a small compressive stress of approximately 0.05MPa was

MA

applied prior to testing in the direction of the specimen’s central axis in order to maintain the
specimen at the center of loading machine. Then the specimen was heated up to target

D

temperatures with a heating rate of 5oC/min. In fact, the heating rate varies when a structural

TE

member is subjected to a realistic fire. However, it would be rather difficult to conduct tests

CE
P

for various heating rates. With regard to this point, the heating rate herein is determined
based on that of standard ISO-834 fire against which the structural members are generally
designed. The heating rate of the ISO-834 fire is shown in Figure 5. At early 5 minutes, the

AC

heating rate drops to 25oC/min, after 25 minutes, the heating rate is approximately 5oC/min.
Hence the heating rate of 5oC/min would be representative for most fire scenarios. Especially
when the UHSC is infilled in steel tubes, the heating rate would be further lower due to the
heat sink effects of the steel tubes and the fire protection (if any). Thus if the heating rate of
5oC/min is used, the measured mechanical properties would be lower than they are in reality,
which will turn out a more conservative but safer design.

In addition to ambient temperature which was approximately 30oC, the target temperature
ranged from 100oC to 800oC at an increment of 100oC. As the UHSC is denser and more
impermeable than the NSC, a trial test was conducted to investigate the holding time of target

8


ACCEPTED MANUSCRIPT
temperature during which uniform temperature distributions can be achieved inside both the
furnace and the UHSC specimens. Figure 6 shows the recorded temperatures for a

T

100x200mm cylinder specimen heated up to 800oC in an electrical oven with a heating rate of

IP

5oC /min. It can be seen that the uniform temperature distribution can be achieved in 4 hours.

SC
R

Hence, the holding time at target temperatures were taken as 4 hours for all specimens.
After holding, the specimen was subjected to three load cycles between 0.05MPa and 15% or

NU

between 5% and 15% of the reference strength as shown in Figure 7 [37]. The holding time at
5% and 15% load levels was 60s. Then the specimen was loaded to fail. Displacement control

MA

was adopted during loading where the displacement rate was 0.4mm/min. It should be
mentioned that the full stress-strain curves were not recorded by the extensometer since the

D

sudden crush of the UHSC specimen would damage the extensometer. The extensometer was

TE

removed when at least 40% of the compressive strength at target temperature was reached.

CE
P

The 40% compressive strength was measured to calculate the modulus of elasticity. The
compression continued after the extensometer was removed until the specimen was crushed.
The peak compression force was recorded by the loading machine. In general, the peak

AC

compressive strength and the modulus of elasticity of the UHSC were obtained from the tests.
They are sufficient for the fire resistance design of CFST columns with the UHSC according
to EN 1994-1-2 [4].

4

Test Results

4.1 Compressive Strength
Spalling was not observed during heating of all the UHSC specimens owing to the addition of
0.1% polypropylene fibers. The compressive strength of UHSC at room temperature was
166MPa which was averaged from 6 specimens. 3 specimens were used for the other target

9


ACCEPTED MANUSCRIPT
temperatures. The compressive strength was taken as the peak stress on the curve of loading
head movement versus compressive stress as shown in Figure 8(a). The test values are shown

T

in Table 3. The reduction factor in this paper is defined as the ratio of strength or elastic

IP

modulus at target temperature divided by their counterpart at room temperature. The

SC
R

reduction factors of strength are shown in Figure 9. It can be seen that the strength was
sharply reduced at 100oC, and then it was partly recovered up to 300oC. It is believed that the
chemical composition of the cement paste were not noticeably changed around 100oC. Hence,

NU

the sharp reduction of strength at 100oC could be either due to the built-up internal pressure

MA

by the evaporation of free water, or the expansion of water between the C-S-H layers causing
a decrease in the surface forces. For the recovery of strength up to 300°C, it might be

D

attributed to the general stiffening of the cement gel by shrinkage, in other words, the

TE

increase of surface forces (Van der Walls forces) between the gel particles due to the removal
of water [38;39]. The temperature at which water is removed and the strength begins to

CE
P

recover depends on the porosity of the concrete [40]. Beyond 300oC, the strength decreased
as the temperature increased. The decrease of strength was attributed to the decomposition of

AC

hydration products such as C-S-H and Ca(OH)2, the deterioration of aggregates, and the
cracks due to thermal incompatibility between the aggregate and the cement paste which led
to stress concentration. At 800oC, the strength was about 30% of that at room temperature.
The comparisons between the strength reduction factors of UHSC and NSC given in EN
1992-1-2 [2] and AISC 360-10 [41] are also shown in Figure 9. The NSC is implicitly
defined as compressive cylinder strength less than 55MPa in EN 1992-1-2, and not greater
than 55MPa in AISC 360-10. The reduction factors are applicable to both NSCs with

siliceous aggregates and calcareous aggregates in AISC 360-10. It was supposed that the
strength of UHSC would reduce greater than that of NSC. However, it can be seen that,
beyond 300°C, the reduction factors of UHSC were similar with those of NSC with
10


ACCEPTED MANUSCRIPT
calcareous aggregates, but higher than those of NSC with siliceous aggregates. The reason
was due to the effect of aggregate type as discussed in Section 1. Generally, the aggregates

T

occupy 65% to 75% of the concrete volume. The effect of aggregate mainly depends on the

IP

thermal stability or integrity of aggregate at high temperatures [14]. Conventional calcareous

SC
R

or siliceous aggregates are thermally stable up to 300°C~350°C. Bauxite aggregates in UHSC
are more stable due to high melting point, and thus produced significant improvements in
heat resistance of the UHSC. The bauxite aggregates have been used for refractory concretes

NU

to achieve super fire performance [42; 43]. The comparison between the strength reduction

MA

factors of UHSC and HSC in EN 1992-1-2 are shown in Figure 10. The reduction factors are
not provided for HSC in AISC 360-10. It is clear that the strength of UHSC was reduced less

D

than that of HSC due to the effect of aggregate type. The comparisons indicated that, for stub

TE

CFST columns with the UHSC governed by compressive resistance, they would withstand

CE
P

longer time when exposed to fire than the CFST columns with the NSC and HSC.
The strength reduction factors of UHSC are compared with those of HSC in the literature as
shown in Figure 11 [7; 8; 19; 40; 44]. It can be seen that the sharp deterioration at 100°C and

AC

the recovery of strength between 100°C~300°C were also captured in previous researches. In
general, the strength of UHSC at elevated temperatures were reduced less than those of HSCs
in the literature.

The mechanical properties (compressive strength and modulus of elasticity) at elevated
temperatures and their counterparts measured after heating and cooling are also compared in
this paper. In tests for residual properties, the heating rate was 5 °C/min, the dosage of
polypropylene fibers was 0.1% by volume, and the specimens were naturally cooled down in
lab air. The reduction factors of strength and residual strength are shown in Figure 12 [34]. It

can be seen that the residual strength was reduced more than the strength when the

11


ACCEPTED MANUSCRIPT
temperature was higher than 300oC. This might be attributed to two facts. One fact is that the
residual strength was affected by further internal micro-cracking induced by differential

T

thermal strain during cooling down. Another fact is that the calcium oxide (CaO) from the

IP

decomposition of hydration products Ca(OH)2 absorbed water after cooling down from high

SC
R

temperatures. Then, it expanded and induced more cracks inside the concrete. As a result, the
residual strength was lower [15].

NU

4.2 Modulus of Elasticity

MA

The elastic modulus was generally defined as the secant modulus between the stress equal to

40% of peak stress and the stress corresponding to strain of 5x10-5 in accordance with ASTM

D

C469-02 [45]. For some stress-strain curves with the existence of turbulence, the slope of the

TE

regressed linear equation for a straight portion was taken as the modulus of elasticity as
shown in Figure 8(b). The elastic modulus of UHSC at room temperature was 61GPa as

CE
P

shown in Table 3. The reduction factors of elastic modulus at elevated temperatures are
shown in Figure 13. Similar to the compressive strength, the sharp reduction and the recovery

AC

were also observed for the elastic modulus due to the built-up internal pressure by the
evaporation of free water. Figure 13 also gives the comparisons between the modulus
reduction factors of UHSC and NSC as given in EN 1992-1-2 and AISC 360-10. The
modulus of elasticity of UHSC was less affected than that of NSC.
The reduced elastic modulus of UHSC are also compared with those of HSC in the literature
as shown in Figure 14 [7; 8; 19; 40; 44]. The sharp reduction and recovery were also
observed in some researches. Overall, the elastic modulus of UHSC was reduced less than
most of the HSCs in the literature due to the effect of aggregate. The practical implication is
that, for slender CFST columns with the UHSC governed by buckling resistance, they would
withstand longer time when exposed to fire than the CFST columns with the NSC and HSC.
12

ACCEPTED MANUSCRIPT
A comparison of the reduction factors for the elastic modulus and the residual elastic
modulus are shown in Figure 15 [34]. For the residual elastic modulus, there was no unusual

T

deterioration and recovery at the temperature range of 100°C~200°C. The elastic modulus

IP

was reduced slightly less than the residual elastic modulus at temperatures between

at and after elevated temperatures are comparable.

Simple Calculation Method for Predicting Fire Resistance of Ultra-High Strength

NU

5

SC
R

200°C~500°C, but more beyond 500°C. It could be concluded that the modulus of elasticity

5.1 Fire Resistance Calculation

MA

Concrete Filled Steel Tubes

D

Similar to the calculation of buckling resistance of a CFST column under axial load at room

TE

temperature [3], the buckling resistance of CFST columns with the UHSC exposed to fire can

CE
P

be determined in accordance with the following simple calculation method (SCM):

(1)

AC

 A f
A f 
N fi,Rd   fi   a,j a,θ   c,k c,θ 
 j 
 M,fi,c 
k
M,fi,a


The fire resistance time is determined when the buckling resistance is reduced to be equal to

the external fire load as the fire exposure time goes. To determine the buckling resistance,
cross section of the CFST column should be discretized into elements where Aa,j and Ac,k are
the areas of steel and concrete elements, and fa,θ and fc,θ are the yield strength of steel element
and the compressive strength of concrete element at temperature θ. The

is the buckling

reduction coefficient which depends on the buckling curve “c” according to EN 1994-1-2 [4].
γM,fi,a and γM,fi,c are partial safety factors for steel and concrete which are taken as 1.0 at fire
situation [4]. The determination of

involves the buckling length of the column in fire lθ,

the effective flexural stiffness (EI)fi,eff, the Euler buckling resistance Nfi,cr under fire, and the
13


ACCEPTED MANUSCRIPT
dimensionless slenderness ratio

as given in Eq.(2) ~ Eq.(6). They are temperature

 EI fi,eff   a,θ Ea,θ I a,j    c,θ Ec,θ I c,k 
k

A f

A f 

(3)

(4)

(5)

NU

    a,j a,θ   c,k c,θ  N fi ,cr
 M,fi,c 
k
 j  M,fi,a

MA



l

SC
R

N fi ,cr 

 2  EI fi,eff

(2)

IP

j

T

dependent parameters which should be determined for each time step.

D

  0.5 1  0.49    0.2   2 
1
   2  2

(6)

CE
P

TE

 fi 

As mentioned in Section 3.3, the transient thermal strains of the UHSC were not measured in
the unstressed tests, which would overestimate the flexural stiffness of column. The

AC

overestimation should be less for CFST columns with the UHSC as the transient thermal
strain of the UHSC is generally smaller than that of NSC [46]. To determine the effective
flexural stiffness properly, reduction factors φa,θ and φc,θ are taken into account respectively
for the steel and concrete as shown in Eq.(2). φc,θ is recommended as 0.8 for concrete [4], and
φa,θ can be taken as 1.0 for steel [47]. Ia,j and Ic,k are th e second moment of areas of the steel

and concrete elements about neutral axis; whereas the Ea,θ and Ec,θ are the elastic modulus of
the steel and concrete elements at temperature θ. Basically for NSC/HSC and carbon steel,
the temperature dependent mechanical properties can be referred to EN 1992-1-2 [2] and EN
1993-1-2 [48], respectively. For high tensile strength steel and the present UHSC, they can be
taken respectively from Ref. [49] and Section 4 of this paper.
14


ACCEPTED MANUSCRIPT
It is worth noting that the increase in strength of concrete caused by confinement is not taken
into account in Eq.(1) for the present fire resistance design. This has been found in researches

T

by Ibanez et al. [50], and Wang and Young [51]. The ignorance of confinement is mainly due

IP

to three facts. Firstly, the CFST column with the UHSC fails at small deformation and has not

SC
R

developed significant confinement [30]. Secondly for temperature lower than 250oC, the
steel’s thermal expansion is larger than that of concrete, there is actually no contact between
the steel tube and the concrete, thus no confinement occurs. At higher temperatures, there

NU

may be contact but the steel has been softened. Thirdly, the magnitude of confinement is

MA

related to the ratio fy/fck according to Eurocode 4 [3]. For the CFST column directly exposed
to fire, the steel temperature arises faster than that of concrete, as a result, the steel loses its

D

yield strength fy much faster than the concrete losing its compressive strength fck. The ratio

TE

fy/fck decreases rapidly, and the steel is not capable to provide sufficient confinement to the

CE
P

concrete.

5.2 Heat Transfer Analysis

AC

Temperature distribution on the cross section of the CFST column should be determined for
each time step for the determination of fire resistance. To the authors’ best knowledge, there
is no method available for calculating temperatures of a CFST column in worldwide design
codes. Herein the modified finite difference method (FDM) is adopted. The modifications are
based on the FDM used by Lie et al. [52] and Kodur et al. [53] to determine temperature
profiles of the circular and square CFST columns under fire. The main modifications are
inclusion of heat convection, development of thermal resistance between steel and concrete

interface, and introduction of square mesh network for square columns. To determine the
temperature distribution, the cross sections should be discretized as shown in Figure 16. Then
the elemental temperature, represented by the nodal temperature at center of an element, can

15


ACCEPTED MANUSCRIPT
be solved on the basis of energy conservation where the heat flowing into one element from
adjacent elements should be equal to the energy consumed by the temperature increase of the

T

element considered. For instance at node (1,1) in Figure 16(b), the temperature is calculated

i 1
i
T2,1i  T1,1i x
T i  T i y
x  y
x y T1,1  T1,1
  1,2 1,1
 h T fi  T1,1i 
 c
y
2
x
2
2
2 2

t

SC
R



IP

by solving Eq.(7).

(7)

NU

The left-hand size of Eq.(7) represents the heat flowing into node (1,1) at ith time step, the
right-hand side of Eq.(7) stands for the energy consumed by the increase of nodal

MA

temperature. The Tf is the fire temperature and t is the fire exposure time. The Δx, Δy are
element sizes. Δt is the time step. The λ, ρ, c are the temperature-dependent thermal

D

conductivity, density and specific heat, respectively. The thermal properties of NSC and HSC

TE

at elevated temperatures can be referred to EN 1992-1-2 [2], whereas their counterparts in EN

CE
P

1993-1-2 [48] can be used for steel. For the UHSC, little information on its thermal properties
is available in the literature. As the UHSC is less porous than the NSC and HSC, it should
have higher thermal conductivity and less moisture content. Considering this, the thermal

AC

properties of HSC can be used for the UHSC, except that the upper limit of the thermal
conductivity and the specific heat with a moisture content of 0% in EN 1992-1-2 may be used.
For NSC exposed to fire, the time-temperature curve usually shows a plateau at 100oC due to
the evaporation of water. However for the UHSC shown in Figure 6, there is no such plateau.
With regard to this, its moisture content is assumed to be 0%, which is validated by test
results in Figure 19.
In Eq.(7), the h is the sum of coefficients of the heat convection he and the thermal radiation
hr. The convection coefficient he can be taken as 25 W/mK for exposure to standard fire of
ISO-834 [54]. The thermal radiation coefficient hr should be calculated as follow:

16


ACCEPTED MANUSCRIPT
2
2
hr     m   f   T f  273  T  273  T f  T  546 



(8)

Φ is the configuration factor which can be taken as 1.0 by ignoring the position and shadow

IP

T

effects [55]. εm is the steel surface emissivity where it may be taken as 0.7 for unprotected
columns [48]. εf is the emissivity of the fire and taken as 1.0 [55]. σ is the Stephan Boltzmann

SC
R

constant equal to 5.67x10-8 W/m2K4. The finite difference equations for nodal temperatures at
the steel-concrete interface can be derived similar to Eq.(7), except that the thermal contact

NU

resistance at the interface should be used to replace the coefficient h. The thermal contact
resistance is considered due to the air gap existing at the steel-concrete interface. Basically, it

MA

can be taken as 100 W/mK according to Ref. [56].

D

5.3 Buckling Length of CFST Column in Standard Fire Test

TE

As mentioned in Section 5.1, the column buckling length is needed to determine the buckling

CE
P

reduction coefficient χfi. For the CFST columns in frames subject to a realistic fire, it can be
easily determined according to EN 1994-1-2 [4]. However it is difficult for the CFST
columns in standard fire tests. This is because only the mid-height of the column is exposed

AC

to fire and the other parts are unexposed. A typical setup of the standard fire test on a CFST
column with a pinned-pinned boundary condition and subject to axial compression is shown
in Figure 17. The column is partly exposed to fire with two ends outside the furnace. The
non-uniform temperature distribution yields differences in flexural stiffness along the column
length, thus the buckling length would be different from that of column exposed to a uniform
temperature. In general, the buckling length of a CFST column in standard fire test can be
calculated by solving a fourth-order differential equation of its lateral displacement. For
instance for a pinned-fixed column shown in Figure 18, it is divided into three segments
according to the exposed/unexposed parts, the fourth-order differential equation of lateral
displacement for each segment is then given as:
17


ACCEPTED MANUSCRIPT

x   L1 , L1  L2 

(10)

k32  P  EI 3

T

k22  P  EI 2

x   L1  L2 , L1  L2  L3 

(11)

SC
R

w3''  k32 w3'  0

(9)

IP

w2''  k22 w2'  0

x   0, L1 

k12  P  EI 1

w1''  k12 w1'  0

Substituting the boundary and compatibility conditions shown in Figure 18 into the general

,

,

, and their derivatives will yield the solution of buckling length.

NU

solutions of

MA

5.4 Validations for Simple Calculation Method

Standard fire test data on CFST columns subject to axial compression from Lie and Chabot

D

[23] Romero et al. [29], and the authors of this paper [57] were used to establish validity of

TE

the proposed simple calculation method. There were totally 29 tests carried out by Lie and
Chabot [23]. All columns were fixed at both ends. The concrete strength varied from

CE
P

23.8MPa to 58.3MPa. All column length was 3810mm with the length exposed to fire being
3200mm. For the tests by Romero et al.[29], there were 5 columns used for the validations.

AC

The concrete strength varied from 28.55MPa to 71.14MPa. The column length was 3180mm
and the exposed length to fire was 3000mm. All columns were fixed at one end and pinned at
the other end except for one column with both ends pinned. 4 unprotected CFST columns
infilled with the UHSC by Xiong et al. [57] were used for the validation. The column length
was 3810mm with 3000mm exposed to fire. For all the CFST columns used for validations
hereinafter, the fire resistance time was predicted based on the actual furnace temperatures
which were designed to follow the time-temperature curve of standard ISO-834 fire. Details
of the CFST columns are summarized in Table 4.
The heat transfer analysis in Section 5.2, the determination of buckling length in Section 5.3,
and the calculation of buckling resistance under fire in Section 5.1 were conducted by using
18


ACCEPTED MANUSCRIPT
MATLAB. The modified FDM is deemed to be qualified for the heat transfer analysis on
CFST columns under fire as the predicted temperatures for the CFST columns with the

T

UHSC, given in Figure 19, show reasonable agreements with the test temperatures. The

IP

moisture content could be taken as 0% as there was no clear plateau on the time-temperature

SC
R

curves of the UHSC infilled in steel tubes.

Table 4 gives the tested and predicted fire resistance time. Two cases were considered where

NU

the NSC/HSC in Refs.[23; 29] was replaced by the UHSC having fck=166MPa in Case 1; and
the UHSC in Ref.[57] was replaced by a NSC with fck=40MPa in Case 2. The mean

MA

prediction/test ratio is 1.017 and 0.889, respectively for Case 1 and Case 2. The mean value
for CFSTs with NSC/HSC is much close to unity, whereas the mean value for CFSTs with

D

UHSC shows conservative predictions. The conservativeness is approximately 11%, mainly

TE

due to the over-conservativeness from column LSH-2-1. Overall, reasonable predictions by

5.5 Discussions

CE
P

the proposed simple calculation method are generally observed.

coefficient

AC

The proposed simple calculation method is powerful as it can provide the buckling reduction
and the bucking resistance Nfi,Rd, changing with the fire exposure time. This is

difficult for conventional finite element software, such as ABAQUS, ANSYS, etc., as a full
package of heat transfer analysis and coupled thermal-mechanical analysis is needed for each
time step. This is tedious for many time steps involved for a rather long fire exposure time.
The buckling reduction coefficient and the buckling resistance for the CFST columns with
the UHSC are shown in Figure 20. It shows that the buckling resistance is rapidly reduced at
early stage of fire exposure, then decreases smoothly at later stage. The fire resistance time is
determined when the buckling resistance approaches the applied test load. Regarding the

19


ACCEPTED MANUSCRIPT
buckling reduction coefficient, it decreases as the fire exposure time continues. The reduction
factor is less than 1.0 at the column failure, indicating a global buckling failure.

T

The predictions for the CFST columns with the replaced concrete are shown in Table 4. The

IP

replacement was based on the same load level which is defined as the ratio between the

SC
R

applied axial load over the buckling resistance at room temperature calculated according to
EN 1994-1-1 [3]. For Case 1 with the NSC/HSC replaced by the UHSC, the ratio between the

NU

two predictions tu/tp varies from 1.083 to 2.674 with a mean value of 1.523. For Case 2 with
the UHSC replaced by the NSC, the said ratio is in the range of 0.682 ~ 0.849 with an

MA

average of 0.767. The comparisons show significant improvements on fire resistance when
the NSC/HSC with conventional siliceous or calcareous aggregates are replaced by the novel

TE

D

UHSC with the bauxite aggregates.

It is well known that the fire resistance time of a CFST column increases with the decrease of

CE
P

section factor and load level, and the increase of concrete contribution ratio. It is also worthy
to know if the improvement, represented by the ratio tu/tp, follows the same trend. The section

AC

factor is defined as the ratio between the exposed area and the volume of the CFST column. It
is generally used to measure the rate of temperature increase in a column. The higher the
section factor, the faster the section heats up. For a CFST column with a uniform crosssectional profile within its length, the section factor can be calculated as the ratio between the
perimeter and the area of the cross-section. The CFST columns for sensitivity study on the
section factor is shown in Table 5. British hot finished steel tubes are used. Figure 21 shows
the relationship between the improvement and the section factor. It can be seen that the
improvement is slightly more at early increase of the section factor, but sharply less for the
further increase. Nevertheless, there is an improvement (i.e. tu/tp > 1.0) as long as the NSC is
replaced by the UHSC.

20


ACCEPTED MANUSCRIPT
The effect of load level is shown in Table 6 and Figure 22. High load levels are not used as
the maximum value of the load level is 0.74 according to EN 1994-1-2 [4]. The improvement

T

is generally getting more with the increase of load level, this is contrary to the trend of fire

IP

resistance time. After its maximum is achieved, the improvement is getting less with the

SC
R

increase of load level.

Table 7 and Figure 23 shows the effect of concrete contribution ratio which is defined as:

NU

Ac f ck
N pl , Rd

(19)

MA

c 

where Ac is the cross-sectional area of concrete, Npl,Rd is the plastic resistance to compression

D

according to EN 1994-1-1 [3]. The concrete contribution ratio stands for the contribution of

TE

concrete to the resistance of section. The higher the ratio is, the larger the influence should be
due to the replacement of concrete. Herein the variation of concrete contribution ratio is made

CE
P

by the change of steel tube thickness. Figure 23 shows that the improvement is more with the

increase of concrete contribution ratio, but generally less for the further increase.

AC

Overall, the improvement is not monotonically changed with the increase of section factor,
load level, and the concrete contribution ratio. This reflects a counterbalance between the
benefit from the use of UHSC and the said parameters to affect the fire resistance time.

6

Conclusions

An experimental investigation on the mechanical properties of UHSC at elevated
temperatures is presented in this article. The mechanical properties included cylinder
compressive strength and modulus of elasticity. The experimental results were compared with
those of concretes given in design codes and in the literature. The fire resistance of CFST

21


ACCEPTED MANUSCRIPT
composite columns with the UHSC was evaluated based on a proposed simple calculation
method. The following conclusions can be drawn.

T

(1) Spalling of the UHSC was prevented during heating to 800 oC due to the addition of 0.1%

SC
R

IP

polypropylene fibers.

(2) Sharp deterioration of strength of the UHSC was observed at 100oC and then it was
partly recovered up to 300oC. The deterioration and recovery of strength were induced by

NU

the evaporation of free water and the resulted shrinkage. The deterioration around 100°C
and recovery of strength up to 300°C were also observed for HSC from previous

MA

researches.

D

(3) Strengths of the UHSC at elevated temperatures were reduced less than those of NSC and

TE

HSC as introduced in Eurocode 2 and AISC 360, and the HSC reported in the literature.

CE
P

This can be explained through the types of aggregates.
(4) Deterioration and recovery of the elastic modulus of the UHSC were observed at the

temperature range of 100oC~200oC, similar to HSC from previous researches. The elastic

AC

modulus of the UHSC were reduced less than that of NSC in Eurocode 2 and AISC 360.
(5) Compressive strength at elevated temperature was generally greater than the residual
strength at the same target temperature; whereas the elastic modulus was comparable
with the residual elastic modulus.
(6) Fire resistance of CFST columns with the UHSC were improved compared with the
NSC/HSC infilled in hollow steel tubes. By studying on 38 CFST columns from previous
researches subject to standard fire tests, the fire resistance time was averagely prolonged
by 30% ~ 50% when the NSC/HSC was replaced by the present UHSC.

22


ACCEPTED MANUSCRIPT
Acknowledgement
The authors would like to acknowledge the funding support by Singapore A*STAR for

T

research project "Steel-concrete composite systems employing ultra-high strength steel and

SC
R

IP

concrete for sustainable high-rise construction" under SERC Grant No: 092 142 0045.

Reference

NU

[1] European Committee for Standardization (CEN). EN 1992-1-1 Eurocode 2: Design of
concrete structures - Part 1-1: General rules and rules for buildings. Brussels; 2004.

MA

[2] European Committee for Standardization (CEN). EN 1992-1-2 Eurocode 2: Design of

D

concrete structures - Part 1-2: General rules-structural fire design. Brussels; 2004.

TE

[3] European Committee for Standardization (CEN). EN 1994-1-1 Eurocode 4: Design of

Brussels; 2004.

CE
P

composite steel and concrete structures - Part 1-1: General rules and rules for buildings.

[4] European Committee for Standardization (CEN). EN 1994-1-2 Eurocode 4: Design of

AC

composite steel and concrete structures - Part 1-2: General rules - Structural fire design.
Brussels; 2005.

[5] Abrams MS. Compressive strength of concrete at temperatures to 1600F. American
Concrete Institute SP25, Temperature and Concrete, Detroit, Mechigan; 1971.
[6] Sullivan PJE, Sharshar R. Performance of concrete at elevated temperatures (as
measured by the reduction in compressive strength). Fire Technol 1992; 28(3): 240-250.

23


ACCEPTED MANUSCRIPT
[7] Hammer TA. High-strength concrete phase 3, compressive strength and elastic modulus
at elevated temperatures. SP6 Fire resistance, Report 6.1, SINTEF Structure and

T

concrete, STF70 A95023, February; 1995.

IP

[8] Diederichs U, Jumppanen UM, Penttala V. Material properties of highs strength concrete

SC
R

at elevated temperatures. IABSE 13th Congress, Helsinki, June; 1988.
[9] Lie TT. Structural fire protection: Manual of Practice, ASCE Manual and Reports on

NU

Engineering Practice No.78, American Society of Civil Engineers, New York; 1992.

MA

[10] Chan YN, Peng GF, Chan KW. Comparison between high strength concrete and normal
strength concrete subjected to high temperature. Mater Struct 1996; 29: 616-619.

D

[11] Felicetti R, Gambarova PG. Effects of high temperature on the residual compressive

TE

strength of high-strength siliceous concretes. ACI Mater J 1998; 95(4): 395-406.

CE
P

[12] Phan LT, Carino NJ. Code provisions for high strength concrete strength-temperature
relationship at elevated temperatures. Mater Struct 2003; 36: 91-98.

AC

[13] Aslani F., Bastami M. Constitutive relationships for normal- and high-strength concrete
at elevated temperatures. ACI Mater J 2011; 108(4):355-364.
[14] Naus DJ. The effect of elevated temperature on concrete material and structures-a
literature review. Oak Ridge National Laboratory, U.S. Nuclear Regulatory Commission,
Office of Nuclear Regulatory Research, Washington, DC 20555-0001; 2006.
[15] Li GQ, Han LH, Lou GB. Fire resistance design of steel structures & steel-concrete

composite structures. China Architecture Building Press (in Chinese); 2006.

24