Indian Journal of Engineering & Materials Sciences

Vol. 14, August 2007, pp. 295-302

Evaluation of creep coefficient on concrete-filled steel tubular columns

Mustafa Hilmi Acar
Department of Civil Engineering, Akdeniz University, 07058 Kampus-Antalya,Turkey
Received 1 September 2006; accepted 22 March 2007
This paper presents the evaluation of creep coefficient on concrete-filled steel tubular (CFST) sustained loaded
columns. A test of the long-term behaviour of CFST columns has been carried out. Shrinkage and creep of CFST columns
are measured. A comparison of results calculated using three prediction models proposed by CEB-FIP, BAZANT-
PANULA, and ACI have shown a good agreement with test results. The study also showed that the shrinkage effect of the
concrete core in CFST columns is negligible. The creep coefficient measured and calculated by those recommanded CEB-
FIP, BAZANT-PANULA, and ACI during 200 days were in agreement for CFST columns.The creep coefficient ultimate
calculated for CFST columns has been found 2. At the beginning of test, the percentage of load taken by steel section was
38% and at the end of test, it was 56%. This is a stress transfer from the concrete section to the steel section due to the creep.
This creep coefficient ultimate evaluated is proposed for long-term behaviour calculation of CFST columns. In order to take
the effect of creep into account, applied sustained load must be multiplied by the value of (1+
ultimated
)
IPC Code: E04C3/30
Concrete-filled steel tubular columns at service load
will suffer, in the long-term, the effects of creep and
shrinkage of the concrete. Short-term static and
dynamic behaviour of CFST columns have been
studied by a number of researchers
1-6
. Long-term
behaviour of CFST columns has received little
attention.The basic issues that need to be addressed
for the long-term behaviour of CFST columns are the
effect of external tube on shrinkage of the concrete
core in CFST columns, how creep and shrinkage lead
to increased deformations in concrete, the creep and
shrinkage cause stress redistribution between the
compression region in concrete and the steel,
moreover, to observe leading to an increase in the
steel stress and a decrease in the concrete stress with
time.
Han et al.
7
have explained in their studies that the
experimental observation of creep in CFST columns
were reported elsewhere
8-12
. Furlong
8
carried out tests
on 22 circular and square columns.These columns
were subjected to various levels of constant axial
force as moments were increased. In addition eight
circular and five square columns were loaded only
axially. It was observed that after the steel tube began
to yield, the creep of concrete caused a load reduction
as high as 15%. Ichinose et al.
9
carried out long-time
measurements on axially CFST columns with circular
sections. The tests include creep tests and drying
shrinkage tests. It was found drying shrinkage strains
in the concrete-filled columns were very small
compared to those of the strains in the normal
reinforced concrete columns, whose surface is free to
interact with the environment. The creep of concrete-
filled steel SHS members was investigated by Morino
et al.
11
under long-term sustained load with three
series of test specimens: six centrally loaded
compression member specimens, a beam specimen
and two eccentrically loaded column specimens.
Creep and shrinkage measurements were carried out
on circular concrete-filled steel tube, plain cocrete
cylinders and concrete cylinders sealed with a
bituminous paint, by Terrey et al
10
. It was found that
when the concrete was placed in steel tube, the
moisture egress from the concrete may be very small
or eliminated totally, with a consequent reduction in
the magnitude and effects of creep and shrinkage. The
study had shown that creep coefficients of the order of
50-60% of those for normal exposed concrete are
applicable for concrete-filled steel tubes. Uy
performed tests on concrete-filled steel SHS columns
subjected to long-term sustained loads. The level of
stress applied to the composite columns was
approximately 0.25 of the assumed concrete strength.
A prediction model based on ACI 209R-92(ACI
Committee 1992) was shown to agree well with the
test results.
Lack of any long-term test data for CFST columns
underlines the necessity of a thorough experimental
______________
E-mail: [email protected], [email protected]
INDIAN J. ENG. MATER. SCI., AUGUST 2007


296
investigation.On the other hand, long duration-tests
are impractical. So, the creep phenomenon could be
studied, at least, in single test program.This paper
reports on an experimental study of the time-
dependent behaviour of CFST columns.
This study is an attempt to predict the time-
dependent behaviour of CFST columns by means of
three models which are proposed by CEB-FIP
13
,
Bazant-Panula
14
and ACI
15
. A single test on the
behaviour of CFST columns under long-term
sustained loading and a theoretical model of analysis
to account for shrinkage and creep effects on CFST
columns under sustained loading are presented in this
paper. A comparison of results calculated using these
prediction models show almost good agreement with
test results.

Experimental Procedure

Material properties
The length, diameter and wall thickness of the
column used in this test were 450 cm, 32.5 cm and
0.75 cm, respectively Fig. 1. The mix proportions were
as follows: cement: 400 kg/m
3
; water: 185 kg/m
3
;
sand: 786 kg/m
3
; and coarse aggregate: 989 kg/m
3
.
Mechanical properties of the concrete and the steel
are given below:

c
(18) = 32 MPa compressive strength of the
concrete age of t
o
,

c
(28) = 38 MPa compressive strength of the
concrete age of 28 days,
E
c
(18) = 3710
3
MPa elasticity modulus at the
age of t
o
,
E
c
(28) = 37.610
3
MPa elasticity modulus at the
age of 28 days,

s
= 386 MPa yield stress of the steel,
E
s
= 20610
3
MPa elasticity modulus of the steel.

Specimens
Six concrete prisms, which have a size of 7 7
28 cm, have been prepared and are protected by a
resin cover to represent the concrete inside columns.
Three concrete prisms unloaded of 7 7 28 cm
have been used to measure the shrinkage
deformations. Three other concrete prisms of 7 7
28 cm were loaded axially to measure the stress
variations in the concrete with a the stress
measurement apparatus because it was not possible to
measure the stress variations in steel and concrete of
CFST column.
Instrumentation
The purpose of this test was to determine the value
of creep coefficient to be used in the design of CFST
columns. Long-term test has resulted in cell as seen in
Fig 2. Test CFST column is articulated from both
ends. These articulations have been made from
special steel. Bottom and top support have been
joined at the bottom and top of same column by steel
plates. The CFST column was loaded 18 days after
the casting of the concrete (t
o
= 18 days). In order to
measure the deformations in the concrete and steel,
five strain gauges were installed into the section of


Fig. 1 The CFST column and installation of the strain gauges
over steel tube



Fig. 2 Concrete-filled steel tubular column under eccentric
loading
ACAR: EVALUATION OF CREEP COEFFICIENT ON CONCRETE-FILLED STEEL TUBULAR COLUMNS


297
o
concrete core at the half of the pile length, and four
strain gauges were installed on the steel column
Fig. 3. The tests were carried out in an environment in
which the conditions reflected atmospheric conditions
(temperature: 21C, humidity: 55%). When the CFST
column was installed into the test chamber, a
sustained load of 500 KN with a 7.5 cm eccentricity
was applied on it. Measurements were started just
after the application of the load. The elastic
deformations were measured by the strain gauges over
6 h. Then, time-depending deformations were
measured in the concrete and steel over seven months.
The measurements were not done separately. Only the
total deformations (creep + shrinkage) were
measured. In order to determine the creep and
shrinkage deformation one by one, the shrinkage
deformations were measured with an extensometer of
deformation measurement at the same time from a
7 7 28 cm unloaded concrete prisms Fig. 4. Stress
variations in the steel and the concrete of CFST
column were not possible to measure. Only stress
variations in the concrete have been measured on a
7 7 28 cm sealed concrete prism loaded axially in
stress measurement apparatus Fig. 5. A 7 7 28 cm
concrete prism was axially loaded at 7.5 MPa to
observe the stress variation.

Models of creep and shrinkage for concrete
These models are given as follows:

CEB-FIP (1990)
Total deformation = Shrinkage deformation+Strees/
E
co
+ Strees/E
c28
*Creep coefficient
E
co
=Modulus of elasticity at loading
E
c28
=Modulus of elasticity at 28 days

Shrinkage deformation:
s s
( , )
s
t t

=
s(tts)
(1)

where
s
is the notational shrinkage coefficient, and

s(tts)
is the coefficient to describe the development of
shrinkage with time

Creep coefficient :
o
( , ) t t =
c(tto)
(2)


Fig. 4 Extensometer of shrinkage deformations measurement



Fig. 5 Stress measurement apparatus


Fig. 3 Installation of strain gauges in the concrete core and
over steel tube
INDIAN J. ENG. MATER. SCI., AUGUST 2007


298
where
0
is the notational creep coefficient, and
c(tto)
is the coefficient which describes the development of
creep with time

American Concrete Institute ACI (1992)
Total deformation = Shrinkage deformation+strees/
E
co
* (1+ Creep coefficient)
E
co
=Modulus of elasticity at loading
Shrinkage deformation:

sh,o
sh sh,o sh
sh,o
( )
( , )
35 ( )
t t
t t
t t

-
=
+ -
(3)

where (tt
sh,o
) is time since application of load, and

sh
is ultimate shrinkage, Creep coefficient:

0.6
o
o 0.6
o
( )
( , )
10 ( )
t t
t t
t t

-
=
+ -
(4)

(tt
o
) is time since application of load,

is ultimate
creep coefficient

Bazant and Panulas model II(1978)
Total deformation = Shrinkage deformation
+ strees/E
/
* (1+ Basic creep coefficient)
E = Fictitious modulus depending on 28 days
cylinder strength, Shrinkage deformation:

1/ 2
sh,o
sh sh,o sh 1
(1/2)sh sh,o
( )
( , )=
( )
t t
t t k
t t t

+ -

(5)

where
1
k

is a coefficient which depends on ambient

relative humudity,
sh
is ultimate shrinkage, (tt
sh,o
)
is time since application of load, and t
(1/2)sh
is related
to the mix parameters Basic creep coefficient:

m
b o o o
( , ) ( ) 0.05 ( ) t t B t t t
-

= + -


n
the deflected shape to be part of a cosine wave, this
made: (i) The
trically-Loaded Pile
uation governing the bent
sed that the peak of this curve gives
(6)

where (t
o
) is time of application load, (tt
o
) is time
since application of load
The parameters B, m and n are all functions of the 28
day strength.

Numercal Analysis
The elasto-plastic behaviour of a pin-ended,
concrete-filled steel tubular column, loaded
eccentrically about one axis, is studied numerically. It
is assumed that complete interaction takes place
between the steel and the concrete and each material
is subjected to a uniaxial state of stress. The
eccentrically-loaded column is analysed by assuming
assumption greatly simplifies the analysis and gives
only slightly lower maximum loads.
The following assumptions are
uniaxial stress-strain curves are applicable for steel
and concrete, the concrete having no tensile strength,
(ii) Both stress-strain curves are reversible, (iii)
Failure due to local buckling or due to shear does not
occur and (iv) Complete interaction takes place
between the steel tube and the concrete core,and plan
sections remain plane after bending

Eccen
The differential eq
equilibrium configuration of an eccentrically-loaded
pile is derived by equating internal and external forces
and moments at a displaced section (Fig. 6). The
calculation can be greatly simplified by assuming the
deflected shape to be part of a cosine wave, in which
case equilibrium is satisfied only at the mid-length.
To determine the complete load-deflection curve of
the column, lateral deflection and axial load values
are calculated for a series of equilibrium shapes
defined by increments of curvature at the central
cross-section.
Acar
16
propo
the maximum load. Equating the external and internal
moments gives:



Fig. 6 Eccentrically-loaded column
ACAR: EVALUATION OF CREEP COEFFICIENT ON CONCRETE-FILLED STEEL TUBULAR COLUMNS


299
2
. .
d y
P y + =
2
0
dz
(7)

Since the flexural stiffness EI is not constant but is
1
y (8)
EI

a complicated function of P, y and z, analytical
integration of Eq. (7) is not possible and numerical
integration is required.
The linear strain distribution at any section can be
o c mpletely specified by the curvature and the
distance y of the neutral axis from the central axis
Fig. 7. The internal axial force and moment can be
calculated from the strain distribution by the
following equations:

j k
i si si ci ci
A
i=1 i=1
= = (, ) P dA A A F = +

j k
i i si i si ci i ci 2
A
i=1 i=1
= (, ) M GdA G A G A F y = + =

P
i
is internal axial forces, M
i
is internal
ape
rvature at any point (z,y)
where
moments, A
si,
A
ci
are areas of the steel and concrete in
a strip,
si;

ci
are stress at the centre of any steel and
concrete strip, G
i
is distance of central axis from the
centre of strip, where j and k are numbers of strips
(Fig. 7) in steel and concrete. Value of and y are
initially guessed, and then successively improved by
an iterative procedure.
Part-cosine wave-deflected sh
The total deflection and cu
(Fig.6.) given by:

o

.cos
z
y y
L
= (9)

2 2
o 2 2

cos
d y z
y
dz L L
= = - (10)

, the half cosine wave length, can be calculated from L
the condition y = e at z = L/2 for z = 0 Eq. (10) then
become:

[ ]
2
1
o o 2
o
4
= cos .

e
e
L e
-


- +

+

(11)

In this case the central deflection is related directly
)

to the central curvature. The procedure used for
calculating the load-deflection curve is given below:
(i) Choose a value of the central deflection
o
,, (ii
From Eq. (11) calculate
o
, (iii) Select a trial value of
Y
o
, (iv) Calculate P and M
io
corresponding to
o
and
Y
o
, (v) If the condition M
io
= M
eo
= P.y
o
is satisfied,
proceed to step (vi), otherwise improve the value of Y
o

by using Newton-Rhapson Method in successive
iterations and repeat step (iv) until the condition is


Fig. 7 Strain distributions
INDIAN J. ENG. MATER. SCI., AUGUST 2007


300
at equilibrium is satisfied only
ime-dependent behavior of the concrete sustained load
ted by
= [1+(t,t
o
)]
to
esults and Discussion
lastic deformations
deformations were in agreement
ime-dependent deformations
rmations were measured at the
predicted deformation values of
nts obtained at the end of the 70 day
tress transfers
tion at the concrete and steel sections
satisfied to a pre-assigned tolerance, (vi) By
successively incrementing the value of
o
and
repeating steps (ii) to (v) for each value
o
, trace the
load-deflection curve.
It should be noted th
at the mid-height of pile.

T
The behaviour under sustained load is predic
modifying the model under a short-term loading
condition. The modification consists of expressing the
strain corresponding to the compressive stress for
concrete, in terms of a function of time (t). Fig. 8
shows the modified stress-strain diagram at any time
(tt
o
) under un sustained stress (t
0
), applied at time
t
0
. It is obtained by multiplying the strains in the
reference diagram by {1+(t,t
o
)}, creep non
linearities are included in the creep coefficient
[(t,t
o
)] (Chu et al.
17
).

t

R

E
Measured elastic
with the calculated elastic deformations by numerical
analysis at the concrete and steel section compression
region. The elastic deformations indicated almost a
13% difference at the tension region. Measured elastic
deformations values in tension region are almost 13%
less than the calculated deformations values exclued
ACI values. This might have resulted from the
installation position of the strain gauges in the
concrete and steel section

T

Shrinkage deformations
The shrinkage defo
same time on 7 7 28 cm unloaded concrete prisms,
protected with a resin cover to represent concrete
inside the CFST column. Measured and calculated
shrinkage deformation levels were very low (2010
6
)
and in agreement. As a result, the shrinkage effect of
the concrete core in CFST column is negligible

eep deformations Cr
The sum of
shrinkage and creep are compared with the results of
measured deformations of experiment. The prediction
deformation values were determined from the creep
coefficient and shrinkage models proposed by the
C.E.B.-F.I.P., ACI-92 and Bazant-Panula. Time-
dependent deformations were calculated by modifying
stress-strain diagram of concrete in short-term loading
conditions in the numerical analysis. Modified stress-
strain diagram was obtained by multiplying the strains
in the reference diagram 1+(t,t
o
). In each model
code, creep coefficient values are predicted for each
interval of time, (tt
o
) = 20,40,60,.......,200 days.
Time-dependent deformations values are calculated
by dividing concrete elasticity modulus E
b
to 1+(t,t
o
)
in numerical analysis using these predicted creep
coefficients.
Measureme
th
indicated that experimentally gathered time-
dependent deformations at the compression region of
the concrete were 7-9 % larger than the deformations
calculated by the proposed prediction models (strain
gauge n 16225) Fig. 9. At a later stage of the
experiment, the test results were 7% larger than the
calculation results. . Here, the hardening effect of the
concrete was perceived in the first three months.
Afterwards, the deformations became stabilised. This
indicates a decrease in the creep deformation rate. At
the steel tube compression region, test results were
8% larger than calculation results (strain gauge n
5556) (Fig.10) and at the tension region of the
concrete and steel, calculation results were 35% larger
than the test results (strain gauge n
o
16221-5550)
(Figs 11 and 12). It was due to installation position of
the strain gauge. These tension regions are not
effective to evaluate creep coefficient



Fig. 8 Modified stress versus strain diagram of concrete
S
Stress varia
were calculated by numerical analysis. The direction
of stress transfers in CFST columns is from the
ACAR: EVALUATION OF CREEP COEFFICIENT ON CONCRETE-FILLED STEEL TUBULAR COLUMNS


301
rsal section of CFST column
ning
the total load for the steel.
concrete section to the steel section leading to an
increase in the steel stress and a decrease in the
concrete stress over time. It was not possible to
measure the stress variations in steel and concrete on
CFST column. A 7 7 28 cm concrete prism was
axially loaded at 7.5 MPa to observe the stress
variation. Decrease in the concrete stress at the time
on the stress measurement apparatus has been
observed. A comparison of observed stress value was
in agreement with calculated stress value
corresponding n16223 numbered strain gauge (center
of section). It was not possible to compare the stress
variations of the steel.

n on transve Load distributio
The load taken by each section at the begin
(t
o
=18 days) was 190 kN for steel section and 310 kN
for the concrete section. This corresponds to 38% of

Fig. 9 Variation of deformations in concrete (Compression)



Fig. 10 Variation of deformations in steel tube (Compression)

Fig. 11 Variation of deformations in concrete (Tension)



Fig. 12 Variation of deformations in steel tube (Tension)

at
e load taken by the steel section increased because
The test results provided the possibility of
t value at the infinite time
-

At the end of test (tt
o
= 200 days), it was seen th
th
of the decrease in the elasticity modulus of concrete.
This is the creep effect. Thus, the load taken by the
steel section and the concrete section was 280 kN and
220 kN respectively. The load taken by the steel
section formed 56% of the total load.

Evaluation of creep coefficient

obtaining a creep coefficien
interval. Firstly, creep coefficient has been evaluated
for (tt
o
) = 200 days from measured elastic, creep and
shrinkage deformations corresponding n16223
numbered strain gauge (center of section). A
comparison of measured results were in agreement
with calculated results from code models for (tt
o
) =
200 days.

measured calculated

total elastic shrinkage
elastic

- -


(12)
INDIAN J. ENG. MATER. SCI., AUGUST 2007


302
measured calculated

-

6 6
6
6
10 117 10 20
117 10
- - -
-
- -

0.23 for (tt

o
) = 200days
Predicted deformations calculated by CEB-FIP and
Bazant-Panula measured
se of this work was to evaluated the
ent value used in the design of concrete-
164 10

code models were near to
deformations. Choosing 20000 days for infinite
time. It was calculated creep coefficient ultimate

ultimated
2.

Conclusions
The purpo
creep coeffici
filled steel tubular columns. In order to determine the
creep coefficient value used in the design of CFST
columns, a long-term test was carried out on a real
size CFST column. Long-term test continued during a
seven months period. Elastic and time-dependent
deformations have been measured by strain gauges for
the steel and concrete. Shrinkage deformation and
stress distribution have been measured on the 7 7
28 cm sealed concrete prisms. A comparison of test
results were almost in agreement with calculated
results. The study resulted in a number of finding as
follows: (i) The shrinkage effect of the concrete core
in CFST column is negligible. (ii) The load taken by
each section at the beginning of test was 190 kN for
steel section and 310 kN for the concrete section. It is
corresponding to 38% of total load for the steel
section. At the end of test, load taken was 280 kN for
the steel section and 220 kN for the concrete section.
This is 56% of total load for the steel section. (iii)
Creep coefficient ultimate has been evaluated for the
CFST columns
ultimated
= 2. In the design of CFST
columns, in order to take the effect of creep into
account, applied sutained load must be multiplied by
the value of (1+
ultimated
)



Acknowledgements
This study has been supported by the Research
Foundations of the Akdeniz University.

References
1 Neogi P K, Sen H K & Chapman J C, J Struct Eng, 47
(1969) 187-195.
2 Knowles R B & Park R, ASCE J Struct Div, 96 (1970)
2125-2153.
3 Schneider S P, J Struct Eng, 124 (1998) 1125-1138.
4 Wang Y C, J Const Steel Res, 49 (1999) 25-41.
5 Han L H & Yang Y F, Adv Struct Eng, 2 (2001) 93-100.
6 Han L H, ZhaoL & Tao Z, Steel Compos Struct, 1 (2001)
51-74.
7 Han L.H.,Tao Z & Liu W, ASCE J Struct Eng, 130 (2004)
1392-1404.
8 Furlong R W, ASCE J Struct Eng, 93 (1967) 113-125.
9 Ichinose L H, Watanabe E & Nakai H, J Constr Steel Res, 57
(2001) 453-466.
10 Terrey P J, Bradford M A & Gilbert R I, Creep and shrikage
of concrete in concrete-filled circular steel tubes, Proc, 6
th

Int Symp Tubular Structures, Melbourne,Australia, (1994),
293-298.
11 Morino S, Kswanguchi J & Cao Z S, Proc Eng Found Conf
Steel-Concrete Composite Structure, ASCE, Irsee, (1996),
514-525.
12 Uy B, ACI Struct J, 98 (2001) 96-104.
13 CEB-FIP Model Code For Concrete Structures, Evaluation
of the time dependent behaviour of concrete, Bull Inform
No. 199, Comit Europeen du Beton/Federation
Internationale de la Precontrainte, (1990), pp. 201.
14 Bazant Z P & Panula L, J Mater Struct Res Test, 11 (1978)
307-329.
15 American Concrete Institute (ACI) Committee, Prediction of
creep, shrinkage and temperature effects in concrete
structures, ACI 209R-92 (1992), Farmington Hills, Mich.
16 Acar M H, Analyse thorique et exprimentale du
comportement des pieux .Thse de Docteur-Ingnieur,
Universit Pierre et Marie Curie,Paris,France (1983),
pp.184.
17 Chu K H, Domingo J & Carreira D J, J Struct Eng, 112
(1986) 943-959.