Thin-Walled Structures 97 (2015) 11–21

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Axial strengthening of thin-walled concrete-filled-steel-tube columns

by circular steel jackets
M.H. Lai a, J.C.M. Ho b,n
a
Department of Civil Engineering, The University of Hong Kong, Hong Kong
b
School of Civil Engineering, The University of Queensland, QLD 4072, Australia

art ic l e i nf o a b s t r a c t

Article history: External confinement in the form of steel rings, tie bars, spirals and FRP wraps has been widely adopted
Received 16 February 2015 for strengthening concrete-filled-steel-tube (CFST) columns. Previous experimental and theoretical
Received in revised form studies have proved that it can improve the strength, elastic stiffness, ductility and interface bonding of
3 July 2015
CFST columns. However, in real engineering practise, CFST columns need to be strengthened are usually
Accepted 1 September 2015
under pre-compressed axial load. The stress-lagging effect between the CFST columns and external
confinement due to pre-loading has not yet been justified. In this paper, external confinement in the
Keywords: form of circular steel jacket is proposed to improve the uni-axial behaviour of CFST columns with and
Concrete-filled-steel-tube without pre-compressed axial load. An experimental study, consisting of 5 hollow-steel-tubes and 10
Modelling
thin-walled CFST columns was conducted to examine the effectiveness of the proposed strengthening
Steel jackets
scheme. The main parameters were the concrete cylinder strength, jacket spacing and pre-compressed
Strengthening
Thin-walled structure axial load level. Test results revealed that the steel jacket could improve the uni-axial behaviour of CFST
columns and the stress-lagging could degrade this beneficial effect. In addition, a theoretical model
developed by the authors previously was adopted to predict the uni-axial behaviour of the strengthening
columns. Very good agreement has been obtained between the theoretical and experimental results.
& 2015 Elsevier Ltd. All rights reserved.

1. Introduction
Concrete-filled-steel-tube (CFST) column, which consists of a
hollow-steel-tube (HST) column filled with concrete, is widely
adopted in many structures nowadays attributed to the superior
behaviour by the composite action [1–4]. However, during the
initial elastic stage under compression, due to the different dila-
tion properties between steel tube and concrete [5,6], the corre-
sponding confining stress may become negative (i.e. hoop com-
pressive stress). This will reduce the strength, elastic stiffness and
ductility of CFST columns [7,8]. On the other hand, degradation of
confining stress, strength and ductility would occur in the post-
elastic stage owing to the inelastic outward buckling of steel tube.
To overcome the deficiencies and fully utilise the composite action
of CFST columns, various approaches were proposed, which in-
cluded internal stiffeners [9,10], tie bars [7], spirals [11], rings [5,8]
and FRP wraps [12,13]. A brief review of these approaches has
been conducted by the authors [11] and it has been concluded that
among these approaches, external confinement in the form of
n
Corresponding author.
E-mail addresses: lmh58743@hku.hk (M.H. Lai),
johnny.ho@uq.edu.au (J.C.M. Ho).
steel rings proposed by the authors [5,8,14] is one of the best
methods to improve the uni-axial behaviour of CFST columns.
However, the installation of external rings requires welding
onto steel tube, which increases the surface imperfections, making
the steel tube more sensitive to local buckling and the welding of
rings is difficult if the tube wall is thin (diameter-to-thickness ratio
over 100). Besides, none of the previous research studies have
investigated the effects of pre-existing loads on stress-lagging ef-
fect between the original CFST column and the external confine-
ment, although it was reported by Su and Wang [15] that pre-
loading would degrade the effectiveness of strengthening scheme
on the reinforced concrete (RC) columns seriously. (The stress-
lagging effect in CFST columns is referring to the slower develop-
ment of confining stress with pre-load condition.) Thus, to address
the above shortcomings, a simple and novel approach, using cir-
cular steel jacket is proposed in this paper. In this approach, cir-
cular steel jacket in the form of screw clamp, which consists of a
stainless steel band and a pressed screw thread pattern, is in-
stalled at different spacing against the steel tube. Fig. 1 shows that
the steel band contains a captive screw at one end. When the
screw is turned clockwise, the band will tighten against the ex-
ternal surface of the steel tube and vice versa. Thus, a perfect
contact between the steel jackets and the steel tube can be
produced.

http://dx.doi.org/10.1016/j.tws.2015.09.002
0263-8231/& 2015 Elsevier Ltd. All rights reserved.
12 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21

List of notations fcu Unconfined concrete cube strength

fr Total confining stress
εcc Strain corresponding to confined peak concrete frE Confining stress of external confinement (steel jacket)
strength frS Confining stress of steel tubes
εco Strain corresponding to unconfined peak concrete Fc Axial load of confined concrete
strength Fs Axial load of steel tube
εsθ Hoop strain of the steel tube Ft Axial load of CFST column
εsr Radial strain of the steel tube G Shear modulus of steel tube
εssE Hoop strain of external confinement H Height of the steel tube
εsz Axial strain of the steel tube HSC High-strength concrete
sE Stress provided by external confinement (steel jacket) HSCFST High-strength concrete-filled-steel-tube
ssθ Hoop stress provided by the steel tube HST Hollow steel tube
ssr Radial stress provided by the steel tube I Increment number
ssut Ultimate tensile stress of the steel tube K Bulk modulus of steel tube
sssE Yield stress of external confinement (steel jacket) LS Parameter reflecting the effect of external
ssy Uni-axial yield stress of the steel tube confinement
ssy,b Elastic buckling stress of steel tube LVDT Linear variable differential transducer
ssyc Compressive yield stress of the steel tube m Parameter considering the effect of concrete grade
ssyt Tensile yield stress of the steel tube n Number of steel jackets
ssz Axial stress of the steel tube Ncal Maximum calculated strength
νs Poisson’s ratio of steel tube Nexp Maximum experimental strength
ω Hardening parameter Nexp-c Maximum experimental strength for confined
Ac Contact concrete area specimens
As Contact steel area Nexp-u Maximum experimental strength for unconfined
CFST Concrete-filled-steel tube specimens
d Nominal width of the steel jacket NSC Normal-strength concrete
Do Outer diameter of the steel tubes NSCFST Normal-strength concrete-filled-steel-tube
Ec Elastic modulus of concrete S Centre-to-centre pacing of the steel jacket
Es Elastic modulus of steel tubes Sθ Deviatoric stress in hoop direction
EssE Elastic modulus of external confinement (steel jacket) Sr Deviatoric stress in radial direction
fc′ Unconfined concrete cylinder strength Sz Deviatoric stress in axial direction
fcc Confined concrete stress t Thickness of the steel tube
fccp Confined peak concrete stress tsj Nominal thickness of the steel jacket

hoop strain Eq.; (2) an actively confined concrete model by Attard

and Setunge [17]; (3) a comprehensive steel model by Prandtl–
Reuss theory; (4) Interaction of core concrete, steel tube and ex-
ternal confinement has been adopted to predict the uni-axial be-
haviour of the tested specimens. For the unconfined and confined
CFST columns without any pre-compressed loads, this model can
be used directly; Otherwise, this model needs minor modification
(i.e. the steel jackets would be effective only after the pre-com-
pressed axial load). The validity of the proposed model is verified
by comparing with the test results in this paper.

2. Experimental Program

2.1. Specimens

A total of 5 HST and 10 thin-walled CFST columns were fabri-

Fig. 1. Details of steel jackets.
In this paper, a total of 5 HST and 10 thin-walled CFST columns
were fabricated and tested under uni-axial compression (All with
diameter-to-thickness ratio over 100). The main parameters were
the concrete cylinder strength, jacket spacing and pre-compressed
axial load level. From the experiment, it can be concluded that the
proposed circular steel jacket is effective in improving the strength
and ductility of the HST and CFST columns. The stress-lagging ef-
fect between the original CFST columns and the new jackets de-
grades this improvement slightly. Finally, a theoretical model
previously developed by the authors [16] based on: (1) an accurate
cated and tested under uni-axial compressive load. Material
properties of the specimens are tabulated in Tables 1 and 2. For the
CFST columns, specimens were divided into two groups depending
on the concrete cylinder strength, fc′: (1) 5 CFST columns with fc′ of
30 MPa; (2) 5 CFST columns with fc′ of 80 MPa. Each group con-
sisted of one unconfined CFST columns and four confined CFST
columns. The nominal outer diameter and thickness of all the
specimens were 114.3 mm and 1 mm, resulting in diameter-to-
thickness ratio over 100. The measured outer diameter (Do) and
thickness (t) of steel tube were summarised in Table 1. To reduce
the end effects and minimise the slenderness ratio [11,18,19], the
specimens were fabricated to be exactly 350 mm in height (H),
 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21 13

Table 1
Lists of specimens.

Group no. Specimens Do (mm) t (mm) ssy (MPa) ssyt (MPa) ssut (MPa) fc′ (MPa) Ec (GPa) εco S (mm) Number of steel jackets n

1 CSJ60-1-114-30 111.5 0.96 316.3 370.0 443.8 31.4 19.7 0.0027 60 6

CSJ120-1-114-30 111.7 0.96 314.4 120 3
CSJ60-1-114-30_R 111.8 0.95 311.6 60 6
CSJ120-1-114-30_R 111.5 0.95 312.8 120 3
CN0-1-114-30 111.6 0.95 313.1 Unconfined
2 CSJ60-1-114-80 111.6 0.96 316.5 79.9 30.7 0.0036 60 6
CSJ120-1-114-80 111.6 0.96 315.2 120 3
CSJ60-1-114-80_R 111.6 0.95 312.5 60 6
CSJ120-1-114-80_R 111.7 0.95 312.5 120 3
CN0-1-114-80 111.6 0.96 316.5 Unconfined

Table 2 strengthening was applied to two of the columns at 50% of the

Details of HST columns. load-carrying capacity taken as that of the respective unconfined
column. The rest of them were strengthened before compression.
Specimens Es (GPa) ssyc (MPa)
The nominal cross-section area (Asj) of steel jackets was 12 mm2,
HSTSJ-1-114 219.1 342.0 with thickness (tsj) 1 mm and width (d) 12 mm. Besides, the actual
HSTSJ60-1-114 213.2 331.2 yield strength (sssE) and elastic modulus (EssE) of steel jacket were
HSTSJ120-1-114 212.8 303.8
400 MPa and 171.8 GPa, respectively. The jackets were screwed
HSTN0-1-114_1 211.2 278.0
HSTN0-1-114_2 200.5 301.6 just tight such that no initial pre-stressing was applied to the steel
Average for unconfined specimens 205.9 289.8 tube. Details of the specimens are shown in Fig. 2. Each specimen
was given a unique code. For example, CSJ60-1-114-30 represents
a CFST column (first alphabet “C”) with steel jackets (second and
third alphabets “SJ”). The spacing of the steel jackets is 60 times
the thickness of steel tube (indicated by the number after the al-
phabets). The last three numbers indicate the nominal thickness of
steel tube (1), nominal outer diameter of steel tube (114) and
concrete cylinder strength (30), respectively. CSJ60-1-114-30_R
stands for confined specimens with pre-compression. For un-
confined specimens, it starts with the letter C, followed by “N0”,
e.g. CN0-1-114-30 is the unconfined counterpart of CSJ30-1-114-
30. For HST column, it starts with “HST” and others remain the
same with CFST column.

2.2. Instrumentations and testing procedure

Fig. 2. Typical specimens.
which was about 3 times the outer diameter. In each group, cir-
cular steel jackets were installed at different centre-to-centre
spacing, S (60 and 120 mm), which was approximately 60 t and
120 t (t¼1 mm) for the four confined CFST columns. In order to
reflect the loading conditions in engineering practise, steel jacket
The instrumentations and testing procedure are the same with
the authors’ previous research [5,8]. To summarise briefly, the
SATEC Series RD Model with maximum load of 5000 kN and dis-
placement of 100 mm was adopted. Details of the test set-up and
instrumentations are shown in Fig. 3. Three linear variable dif-
ferential transducers (LVDTs) with 100 mm measuring displace-
ment were installed between the top and bottom loading platens
to record the full length axial displacement of the CFST columns.

Fig. 3. Test set-up.

14 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21
Three numbers of bi-directional strain gauges (Tokyo Sokki Ken-
kyujo Co., Ltd, type: FCA-5-11-3L), separated 120° each, were in-
stalled at around the mid-height of the external surface of the
specimens to measure the axial and hoop strains at the mid-height
of steel tube. Besides, several uni-directional strain gauges (Tokyo
Sokki Kenkyujo Co., Ltd, type: FLA-5-11-3L) were installed on the
surface of steel jackets.
All specimens were tested under displacement control. For
CFST columns, the top surface was coated with a layer of rapid
hardening gypsum before preloading was applied. After about
15 min of preloading, specimens were unloaded and the test
would begin. The loading rate was kept as 0.3 mm/min for the
whole loading process. Tests would stop when the axial dis-
placement was larger than 17.5 mm, which was about 0.05 axial
strain or when the applied load dropped to less than 50% of the
measured maximum load, whichever was earlier. For the speci-
mens subjected to pre-compression, the steel jackets would be
added after 50% of the maximum load-carrying capacity of the
unconfined counterpart by holding the cross head of the machine
(the speed of the test was set to 0 mm/min). In order to provide
uniform confining stress at the installation stage, the two middle
steel jackets were tightened simultaneously for CSJ60 specimens;
whilst the middle steel jacket was tightened first for CSJ120 spe-
cimens. After that, the other jackets were installed from the mid-
height to the ends. After the installation, the loading rate was set
back to 0.3 mm/min.
In order to obtain the material properties of concrete and steel
tube, six plain concrete cylinders (150  300 mm) and 4 steel
tensile coupons were fabricated and tested according to the British
standards [20–23]. The tensile coupons were taken at 90° from the
welding part in the longitudinal direction of the untested steel
tubes, which gave the most conservative results [24]. The average
concrete elastic modulus (Ec), cylinder strength (fc′), ultimate
strain (εco), steel elastic modulus (Es), steel tube compressive yield
stress (ssyc), steel tensile yield stress (ssyt), steel tensile ultimate
stress (ssut) are given in Tables 1 and 2. The loading rate of con-
crete cylinder tests adopted in this paper was 0.3 MPa/s [21] (be-
tween 0.2–0.4 MPa/s as recommended in the standard), which can
be converted to about 0.27 mm/min (i.e. strain rate at 0.0015%/s)
and it is closed to the loading rate adopted in CFST columns in this
paper (0.3 mm/min). For the compression test of HST columns, the
instrumentations and loading procedure are exactly the same as
those of CFST columns. However, for tensile coupon tests, the
loading rate recommended by the standard [23] is too high. To be
consistent, the authors adopted 0.0015%/s as the strain rate for
Fig. 4. Axial stress-strain curves for HST columns.
specimens within the elastic range and yield plateau. After that,
higher loading rate would be applied, i.e. 0.04%/s until fracture of
specimens, as suggested by Law and Gardner [25].
3. Experimental results and discussions
Tested results of 5 HST and 10 thin-walled CFST columns are
summarised and discussed below.
3.1. Axial load-strain curves of HST columns
The axial stress is plotted against axial strain for the HST col-
umns in Fig. 4. It should be noted that the dimension of HSTN0-1-
114 did not fulfil the requirement of slenderness limits by EC3
[26]: Do/tr21150/ssy. Thus, HSTN0-1-114 would fail by stability
(local buckling) first and the maximum stress attained was smaller
than the yield stress as reported in Tables 1 and 2. It can be seen
from Fig. 4 and Table 2 that the values of Es, ssyc and ductility of
CFST columns with steel jackets are larger than the unconfined
counterparts. More importantly, it is observed that HSTSJ-1-114
has the largest value of ssyc, which is confined by closely spaced
steel jackets (steel jackets cannot be installed in the mid-height
because of the bi-directional strain gauges) around the steel tube
and the outward buckling can be delayed in this case. However,
the value of ssyc is smaller than ssyt because the steel jackets
cannot offer resistance to column inward buckling. Since the local
buckling occurred before yielding in this study, the maximum
stress of steel tube should be calculated. For CFST columns, due to
the supporting effect by the in-filled concrete, the local buckling
resistance of steel tube is improved. Using the equations derived
by Bradford et al. [27], the theoretical elastic buckling stress for the
steel tube of CFST columns is √3 (or 1.73) times that of the re-
spective HST columns. Thus, the modified ratio in EC3 could be
adopted in this paper:
Do ⎛ 235 ⎞ 36632.9
≤ 90 3 ⎜⎜ ⎟⎟ ≈
t ⎝ σsy ⎠ σsy (1)
σsyc ≤ σsy ≤ σsyt (2)
where ssyc in Eq. (2) is based on the unconfined HST columns. The
calculated ssy is tabulated in Table 1 for all the specimens.
 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21 15

a b

c d

Fig. 5. Axial load-strain curves for CFST columns.

3.2. Axial load-strain curves of CFST columns columns at 0.015 strain. For high-strength CFST (HSCFST) columns,
it can be observed from Fig. 5c–d that the axial load drops more
The axial load is plotted against axial strain for the CFST col- rapidly than NSCFST columns in the post-elastic stage. This implies
umns in Fig. 5 and the maximum loads are recorded in Table 3. The that the confining effect provided by steel tube and jackets is not
axial load is directly obtained from the compression machine, and sufficient for high-strength concrete (HSC). Evidently from Fig. 5
the axial strain is obtained by the modified LVDTs reading, which and Table 3, the strength of CFST columns with steel jackets pro-
is adjusted for the axial deformation of the capping material [8]. It vided improved considerably. The maximum improvement in ul-
can be seen in Fig. 5a–b that for normal-strength CFST (NSCFST) timate load obtained by adding steel jackets is 12% for CSJ60-1-
columns, the axial load increases as the axial strain increases 114-30. Besides, the smaller the jacket spacing, the larger en-
during the elastic stage. For the post-elastic branch, the axial load hancement would be the load-carrying capacity and ductility. This
increases much more slowly until a peak value. Then, the strength is because with closer spacing of the jackets, the confining stress is
drops very gradually reaching about 97% of the maximum loading higher and more uniform thus enhancing the strength and de-
formability of CFST columns. All in all, steel jacket is highly effec-
for the confined specimens and 95% for the unconfined CFST
tive in improving the uni-axial behaviour of CFST columns.
Table 3
Fig. 5 and Table 3 also illustrate the effects of stress-lagging on
Comparisons between theoretical and experimental results. the uni-axial behaviour of strengthened CFST columns. CSJ60-1-
114-30 was confined by circular steel jackets before compression
Group no Specimens Nexp (kN) Nexp-c / Nexp-u Ncal (kN) Nexp / Ncal and it was the strongest with 12% improvement in axial strength.
1 CSJ60-1-114-30 510 1.12 513 0.99
CSJ60-1-114-30_R was strengthened by jackets after pre-com-
CSJ120-1-114-30 495 1.08 491 1.01 pression (i.e. 233 kN, about 50% of the maximum load of the un-
CSJ60-1-114-30_R 492 1.08 508 0.97 confined counterpart) and achieved only 8% improvement in
CSJ120-1-114- 470 1.03 486 0.97
strength. This difference was caused by the larger dilation in steel
30_R
CN0-1-114-30 456 1.00 466 0.98 than in concrete such that the confining stress became negative
2 CSJ60-1-114-80 999 1.05 995 1.00 and reduced the strength of CFST columns. However, by adding
CSJ120-1-114-80 966 1.01 967 1.00 steel jackets before compression, the extra confining stress pro-
CSJ60-1-114-80_R 980 1.03 990 0.99
vided by the steel jackets would restrain the dilation of steel and
CSJ120-1-114- 962 1.01 964 1.00
80_R offer larger confining stress. Thus, for specimens without pre-
CN0-1-114-80 955 1.00 944 1.01 compression, the maximum strength is higher. This can also be
Average value 0.99 proven numerically by adopting the confining stress evaluation
Standard deviation 0.0157
method proposed previously by the authors [28]. By that method,
Nexp-c and Nexp-u represent the ultimate load for confined and unconfined speci- the confining stress provided by steel tube and steel jackets at
mens, respectively. maximum load in CSJ60-1-114-30 were 2.39 and 1.50 MPa,
16 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21

was adopted to predict the uni-axial behaviour of the specimens in

this paper. This model consists of four components: (1) A new
hoop strain Eq.; (2) Constitutive model of confined concrete
modified from Attard and Setunge’s [17] actively confined concrete
model; (3) Full range constitutive model of steel tube under
complicated stress-state using generalised Hooke’s Law and
Prandtl–Reuss theory; (4) Model of interaction among core con-
crete, steel tube and steel jackets. Since the proposed model have
been presented earlier [16], only key equations and steps would
present herein.

4.1. Sign convention and perfect bond assumption

In this paper, compressive stress and strain are taken as posi-

tive, and vice versa. Bonding between the concrete and steel tube
is assumed to be intact and thus the following equations are given:
Fig. 6. Failure modes for test specimens.
εcz = εsz = εz (3)

respectively; whilst the stress decreased to 2.28 and 1.49 MPa

respectively in CSJ60-1-113-30_R. The total confining stress was εcθ = εsθ = εθ (4)
larger in CSJ60-1-114-30 and so the maximum strength. Therefore,
it can be concluded that the stress-lagging effect reduced the where εcz and εsz are the axial strains of concrete and steel tube,
maximum confining stress that can be provided by the jackets. It respectively; εcθ and εsθ are the hoop strains of concrete and steel
should be noted that the stress-lagging pattern of CFST column tube, respectively; εz and εθ are the axial and hoop strains of the
was quite different from the RC column. In RC column, the lagging CFST column, respectively.
effect would cause the locked-in stress problems [15] in concrete
and normally the concrete crushed before the yielding of 4.2. Three-dimensional stress–strain relationships of steel tube
strengthening materials. In CFST column, because of the stronger
confinement provided by the steel tube, yielding of strengthening The three-dimensional stress–strain relationship of steel tube
materials can always occur before failure of concrete. can be evaluated by the hoop-axial strain relationships with the
aid of Prandtl–Reuss theory. In this paper, since the maximum
3.3. Failure modes
axial strain considered is 1.5%, the uni-axial stress–strain curve of
steel-tube in compression can be assumed as linearly-elastic-per-
The failure modes of the HST and CFST columns are shown in
fectly-plastic. In the initial elastic stage, according to generalised
Figs. 6 and 7. Generally speaking, all specimens failed by local
Hooke’s Law (in incremental form):
buckling. It can be observed from Fig. 6 that, the failure of un-
confined HST column was due to outward buckling near the top ⎡ 4 2 2 ⎤
⎧ dσ i ⎫ ⎢
K + G K − G K − G⎥
surface (end effect). For the three confined HST columns, the
⎢ 3 3 3 ⎥⎧ dεszi ⎫
⎪
⎪
sz ⎪
⎪ ⎢ ⎪ ⎪
failure mode was alternatively inward and outward buckling. From 2 4 2 ⎥⎪ ⎪
Fig. 7a–b, it shows that the unconfined specimens buckled locally ⎨ dσsiθ ⎬ = ⎢ K − G K + G K − G ⎥⎨ dεsiθ ⎬
⎪ ⎪ ⎢ 3 3 3 ⎥⎪ ⎪
near the top surface of columns. For confined specimens, due to ⎪
⎩ dσsr ⎪
i
⎭ ⎢ ⎪ i⎪
2 2 4 ⎥⎩ dεsr ⎭
the effective confinement provided by steel jackets, the “end ef- ⎢⎣ K − 3 G K − 3 G K + 3 G ⎥⎦ (5)
fect” buckling was not seen. Instead, bulges were found between
two adjacent rows of jackets. Moreover, as HSC is more brittle than
normal-strength concrete (NSC), the failure mode of HSCFST col- Es
K=
umns was due to irregular local buckling, indicating brittle shear 3(1 − νs ) (6)
failure of the concrete.
To have a better visualisation of the concrete core failure mode, Es
G=
the steel tube of some failed specimens were cut and removed. It 2(1 + νs ) (7)
can be seen from Fig. 7c–f that the in-filled concrete crushed at the
location of outward folding of steel tube. Fig. 7c indicates that no During the plastic stage, the famous Prandtl–Reuss equations
shear failure plane was observed and the core concrete crushed are adopted:
locally near the top surface, indicating end effect for unconfined ⎡ 4 2 2 ⎤
K + G−ωSz2 K − G − ωSzSθ K − G − ωSzSr ⎥
NSCFST column. While for confined NSCFST column, the end effect ⎧ dσ i ⎫ ⎢ 3 3 3
was restrained by the steel jackets, thus the core concrete crushed ⎪ sz ⎪ ⎢ ⎥
⎪ ⎪ ⎢ 2 4 2 ⎥
at the mid-height as in Fig. 7d. More interestingly, it can be ob- ⎨ dσsiθ ⎬ = ⎢ K − G − ωSzSθ K + G − ωSθ2 K − G − ωSθSr ⎥
⎪ ⎪ ⎢ 3 3 3 ⎥
served from Fig. 7e that for CN0-1-114-80, an obvious shear failure ⎪
⎩ dσsr ⎪
i
⎭ ⎢
plane was formed and the measured angle of rupture was 73°. On
2 2 4 2 ⎥
⎢⎣ K − 3 G − ωSzSr K − 3 G − ωSθSr K + 3 G − ωSr ⎥⎦
the other hand, with the provision of steel jackets, the angle of
rupture decreased to 47° in Fig. 7f, indicating the effectiveness of ⎧ dε i ⎫
⎪
⎪
sz ⎪
⎪
the steel jackets. ⎨ dεsiθ ⎬
⎪ ⎪
4. Modelling of CFST columns and verifications ⎪
⎩ dεsr ⎪
i
⎭ (8)

A theoretical model developed by the authors previously [16] For perfectly-plastic material:
 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21 17

Fig. 7. Axial strains by modified LVDTs against axial strains by strain gauges curves.

3G 1
ω= 2
σsy (9)
Sz =
3
( )
2σsri− 1 − σszi− 1 − σsiθ− 1
(12)

where ssz and ssr refer to the axial and radial stresses of steel tube;
1 K and G are bulk and shear moduli of steel tube; νs is the Poisson’s
Sz =
3
(
2σszi− 1 − σsiθ− 1−σsri− 1 ) (10) ratio of steel tube taken as 0.3 in this paper; εsr refers to the radial
strain of steel tube; ω is the hardening parameter; Sz, Sθ and Sr
refer to the deviatoric stresses in axial, hoop and radial direction.
1
Sθ =
3
(
2σsiθ− 1 − σszi− 1 − σsri− 1 ) (11)
Lastly, i is the present stress or strain increment number. The yield
surface of steel tube is determined by Von Mises yield criterion:
18 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21

⎛ f ‵ ⎞m⎧
⎪ ⎡ −ε ⎞⎤
0.7
⎡ ⎛ ε ⎞⎤
εz = LS⎜ c ⎟ ⎨ εco⎢ 1 + 0.75( θ ⎟⎥ − εcoexp⎢ 7⎜ θ ⎟⎥
⎝ 30 ⎠ ⎪⎩ ⎣ εco ⎠⎦ ⎣ ⎝ εco ⎠⎦

0.7
⎡ ⎛ f ⎞⎤⎫ ⎪
+ 0.07( − εθ ) ⎢ 1 + 26.8⎜⎜ r ⎟⎟⎥⎬
⎢⎣ ⎝ f c‵ ⎠⎥⎦⎪
⎭ (15)

LS2 − LS1
LS = (S − d) + LS1
H−d (16)

LS1 = 0. 6650 (17)

LS2 = 0.6466 (18)

Fig. 8. Free body diagram for confining stress.

2 ⎧
σsy = (σsz − σsθ )2 + (σsz − σsr )2 + (σsθ − σsr )2 ⎪ 0 f c‵ ≤ 30
2 (13) m=⎨
⎪
⎩ − 0.05 f c‵ > 30 (19)
The radial stress ssr is equal to fr [16]:
σsr = fr (14) where LS is the parameter reflecting the effect of external con-
finement, from LS2 ¼0.6466 for unconfined CFST columns to
Thus, the three-dimensional stress–strain history of steel tube LS1 ¼0.6650 for confined CFST columns with S ¼d; H is the total
can be evaluated using Eqs. (5)–(14). height of the specimen; m is the parameter considering the effect
of concrete strength.
4.3. Constitutive model of confined concrete The axial stress–strain relationship of confined concrete is gi-
ven by Attard and Setunge [17], which has been proven to be
The following equation [16] is proposed to describe the re- applicable to a broad range of concrete strength from 20 to
lationship among εθ, εz, fr and fc′: 130 MPa :

a b

c d

Fig. 9. Hoop-axial strain curves for CFST columns.

 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21 19

Fig. 10. Experimental against predicted loads at interval of 0.001 axial strain.

Fig. 11. Measured against predicted hoop strains at interval of 0.001 axial strain (In absolute value).

fcc A(εz /εcc ) + B(εz /εcc )2 steel tube and also external confinement (steel jacket). Thus, fr is
= 2 equal to the sum of the confining stress from the steel tube (frS)
fccp 1 + (A − 2)(εz /εcc ) + (B + 1)(εz /εcc ) (20)
and external confinement (frE):

fr = frS + frE (23)

⎡ f ⎤
εcc = εco⎢ 1 + (17.0 − 0.06f c‵) r ⎥
⎢⎣ f c‵ ⎥⎦ (21) 2t
frS = − σsθ
where fcc is the confined concrete stress; fccp and εcc are the con- Do − 2t (24)
fined peak concrete stress and the corresponding axial strain of
For CFST columns with steel jackets, the following formula for
concrete under a constant fr; A and B are parameters that govern
the shape of the stress–strain curve. frE can be obtained from the free-body diagram as shown in Fig. 8:
In order to maintain the consistency to authors’ previous re-
2nAst
search [5,7,8,11,14], fccp is defined as: frE = − σE
H (Do − 2t ) (25)
fccp ⎛ f ⎞
= 1 + 4.1⎜⎜ r ⎟⎟ With:
f c‵ ⎝ f c‵ ⎠ (22)
⎧ εssEEssE εssEEssE ≤ σssE
σE = ⎨
4.4. Interaction modelling among concrete, steel tube and steel ⎩ σssE εssEEssE > σssE (26)
jackets
The average hoop strains of steel jackets were determined by
In a confined CFST column, the core concrete is confined by the following equation [16]:
20 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21
⎡ (H − S )2 ⎤ and predicted hoop strains for CSJ120-1-114-80_R becomes larger
εssE = εsθ ⎢ 0.9 2
+ 0.1⎥
⎣ (H − d) ⎦ (27)
where n is the number of steel jackets; sE refers to the stress
provided by the external confinement (steel jackets); εssE is the
average hoop strain of external confinement.
For the confined CFST columns with pre-compression equal to
50% of the maximum load, Eq. (27) is changed to:
⎧ 0 Ft < Npl
⎪ tube buckling would occur only at larger axial strain level. This is
⎪
εssE =⎨ ⎡ (H − S )2 ⎤ because: (1) HSC is more brittle than NSC, which expands much
⎪ (εsθ − εsθ, pl )⎢ 0.9 + 0.1⎥ Ft ≥ Npl
⎪
⎩ ⎣ (H − d)2 ⎦ (28)
where Ft and Npl are the total axial load of CFST column and the
pre-compressed axial load before adding steel jackets (equal to
233 and 472 kN for NSCFST and HSCFST columns), respectively. εsθ,
pl is the corresponding hoop strain at Npl.
4.5. Axial load against axial strain curves of CFST columns
Multiplying the axial stress of the steel tube ssz and core con-
crete fcc by the respective contact area, As and Ac, the axial loads
carried by the steel tube (Fs) and confined concrete (Fc) can be
obtained. Since the axial strain is less than 1.5% in this study, As
and Ac can be assumed to remain unchanged. Then the total axial
load of the CFST column, Ft can be calculated by using:
Fs = σszAs
Fc = fcc Ac
Ft = Fc + Fs
The generation of the axial load-strain curves of CFST columns
requires an iterative process. (1) A small increment of hoop strain
dεsθ (  0.5 με in elastic stage and  5 με beyond elastic stage) is
applied to the current hoop strain εθi , which starts from zero;
(2) The total axial strain of the next step εzi1+ 1 is assumed and then
ssz, ssθ, ssr and fr can be determined by Eqs. (3–14) and (23–28);
(3) With fr and εθi + 1, another total axial strain εzi2+ 1 can be calculated
by Eqs. (15–19). If εzi1+ 1 and εzi2+ 1 are close enough (Error o0.1%), εθi + 1
and εzi1+ 1 are taken as the true strains. If not, an iterative process
based on secant method is performed starting from step (2);
(4) Once the converged values of εθi + 1 and εzi1+ 1are determined, fcc,
Fc, Fs and Ft can be evaluated and a point on the axial load-strain
curve is obtained; (5) Repeat Steps (1) to (4) until the total axial
strain is larger than 1.5% to obtain the complete stress–strain
curve.
4.6. Comparisons between theoretical and experimental results
The predicted axial load-strain curves and the hoop-axial strain
curves were compared with the experimental curves in Figs. 5 and
9, whilst the predicted maximum loadings were compared with
the experimental maximum strengths in Table 3. Very good
agreements have been obtained with very small standard variation
(0.0157). To further assess the accuracy of the proposed model, the
experimental loads and hoop strains at interval of 0.001 axial
strain are plotted against the predicted loads and hoop strains at
the same axial strain in Figs. 10 and 11, respectively. Figs. 10 and 11
reveal that the experimental and predicted loads and hoop strains
agree very well with each other, as indicated by the closeness of
the data points to the line of equality and very high R2 of 0.990 and
0.962 respectively except for CSJ120-1-114-80_R in Fig. 11. It can
be seen from Figs. 9d and 11 that the deviation between measured
as axial strain increases. This can be attributed to the fact that the
steel tube started to buckle at small axial strain ( 0.004) for this
specimen, which was not considered in the proposed model. Ac-
tually, the same phenomenon occurred for CN0-1-114-80, but it
failed at small axial strain (  0.008). On the other hand, for spe-
cimens (1) CSJ120-1-114-30_R (With fc′¼30 MPa); (2) CSJ120-1-
114-80 (Steel jackets install before applying compression);
(3) CSJ60-1-114-80_R (Closer spacing of steel jackets), such steel
more rapidly when crack initiates; (2) By installing steel jackets
before applying any loadings, there is no stress-lagging effect and
the steel jackets can be fully utilised; (3) By providing closer
spacing of steel jackets, larger and more uniform confining stress
can be generated and thus delaying the tube buckling.
5. Conclusions
In this paper, a total of 5 HST and 10 thin-walled CFST columns
were fabricated and tested under uni-axial compression (with
diameter-to-thickness ratio over 100). The main parameters were
the concrete cylinder strength, jacket spacing and pre-compressed
axial load level. In addition, a theoretical model previously de-
veloped by the authors based on (1) an accurate hoop strain Eq.;
(2) an actively confined concrete model; (3) a three-dimensional
(29) steel model; (4) Interaction of core concrete, steel tube and steel
jackets by free-body diagram was adopted to predict the uni-axial
(30) behaviour of the tested specimens. From the experimental and
theoretical studies, the following conclusions can be drawn:
(31)
(1) Steel jackets are highly effective in improving the uni-axial
behaviour of HST and CFST columns.
(2) The enhancement for the CFST columns subjected to pre-com-
pressed axial load is smaller than specimens without pre-
compression, which may be due to the initial confining stress
provided by the steel jackets (specimens without pre-com-
pression) and the stress-lagging effect between the original
CFST columns and new jackets.
(3) The stress-lagging can degrade the beneficial effect by the steel
jackets slightly.
(4) For unconfined CFST columns, the failure mode is “end effect”
type buckling; Due to the confining effect of steel jackets, the
failure mode of confined CFST columns is outward folding be-
tween two adjacent rows of steel jackets.
(5) For HSCFST columns, due to the insufficient confining stress
provided by the steel tube and external confinement, the failure
mode is more irregular compared with NSCFST columns.
(6) The proposed theoretical model precisely predicts the axial
stress-strain and hoop-axial strain relationship of the
specimens.
Acknowledgements
The work described in this paper has been substantially sup-
ported by a grant from the Research Grants Council of the Hong
Kong Special Administrative Region, China (Project no. HKU
712310E). Technical supports for the experimental tests provided
by the laboratory staff of the Department of Civil Engineering, The
University of Hong Kong, are gratefully acknowledged.
References
[1] B. Uy, Strength of short concrete filled high strength steel box columns, J.
 M.H. Lai, J.C.M. Ho / Thin-Walled Structures 97 (2015) 11–21
Constr. Steel Res. 57 (2) (2001) 113–134.
[2] L.H. Guo, S.M. Zhang, W.J. Kim, G. Ranzi, Behavior of square hollow steel tubes
and steel tubes filled with concrete, Thin-Walled Struct. 45 (12) (2007)
961–973.
[3] L.H. Han, G.H. Yao, Experimental behaviour of thin-walled hollow structural
steel (HSS) columns filled with self-consolidating concrete (SCC), Thin-Walled
Struct. 42 (9) (2004) 1357–1377.
[4] Z. Tao, L.H. Han, D.Y. Wang, Strength and ductility of stiffened thin-walled
hollow steel structural stub columns filled with concrete, Thin-Walled Struct.
46 (10) (2008) 1113–1128.
[5] M.H. Lai, J.C.M. Ho, Confinement effect of ring-confined concrete-filled-steel-
tube columns under uni-axial load, Eng. Struct. 67 (2014) 123–141.
[6] X. Lu, C.T.T. Hsu, Tangent Poisson’s ratio of high-strength concrete in triaxial
compression, Mag. Concr. Res. 59 (1) (2007) 69–77.
[7] J.C.M. Ho, M.H. Lai, Behaviour of uni-axially loaded CFST columns confined by
tie bars, J. Construct. Steel Res. 83 (2013) 37–50.
[8] M.H. Lai, J.C.M. Ho, Behaviour of uni-axially loaded concrete-filled-steel-tube
columns confined by external rings, Struct. Des. Tall Spec. Build. 23 (6) (2014)
403–426.
[9] C.S. Huang, Y.K. Yeh, G.Y. Liu, H.T. Hu, K.C. Tsai, Y.T. Weng, S.H. Wang, M.H. Wu,
Axial load behavior of stiffened concrete-filled steel columns, J. Struct. Eng.
128 (9) (2002) 1222–1230.
[10] T. Kitada, Ultimate strength and ductility of state-of-the-art concrete-filled
steel bridge piers in Japan, Eng. Struct. 20 (4–6) (1998) 347–354.
[11] M.H. Lai, J.C.M. Ho, Effect of continuous spirals on uni-axial strength and
ductility of CFST columns, J. Constr. Steel Res. 104 (2015) 235–249.
[12] Y.M. Hu, T. Yu, J.G. Teng, FRP-confined circular concrete-filled thin steel tubes
under axial compression, J. Compos. Constr. 15 (5) (2011) 850–860 , ASCE.
[13] Y. Xiao, Applications of FRP composites in concrete columns, Adv. Struct. Eng.
7 (4) (2004) 335–343.
[14] J.C.M. Ho, M.H. Lai, L. Luo, Uni-axial behaviour of confined high-strength CFST
21
columns, Proc. Inst. Civ. Eng. Struct. Build. 167 (SB9) (2014) 520–533.
[15] R.K.L. Su, L. Wang, Axial strengthening of preloaded rectangular concrete
columns by precambered steel plates, Eng. Struct. 38 (2012) 42–52.
[16] M.H. Lai, Behaviour of CFST columns with external confinement under uni-
axial compression, PhD Thesis, The University of Hong Kong, Hong Kong, 2015.
[17] M.M. Attard, S. Setunge, Stress-strain relationship of confined and unconfined
concrete, ACI Mater. J. 93 (5) (1996) 432–442.
[18] L.H. Han, G.H. Yao, X.L. Zhao, Tests and calculations for hollow structural steel
(HSS) stub columns filled with self-consolidating concrete (SCC), J. Constr.
Steel Res. 61 (9) (2005) 1241–1269.
[19] Z.W. Yu, F.X. Ding, C.S. Cai, Experimental behavior of circular concrete-filled
steel tube stub columns, J. Constr. Steel Res. 63 (2) (2007) 165–174.
[20] BSI. BS, Testing Concrete. Method for Determination of Static Modulus of
Elasticity in Compression, BSI, London, UK, 1983 1881-121: 1983.
[21] BSI. BS, Testing concrete. Method for Determination of the Compressive
Strength of Concrete Cores, BSI, London, UK, 1983 1881-120:1983.
[22] BSI. BS, Testing concrete. Method for making test cylinders from fresh con-
crete, BSI, London, UK, 1983 1881-110: 1983.
[23] BSI. BS EN, Metallic materials. Tensile testing. Method of Test at Ambient
Temperature, BSI, London, UK, 2009, ISO 6892-1: 2009.
[24] Hancock G.J., Design of Cold-formed Steel Structures: To Australian/New
Zealand Standard AS/NZS 4600: 1996. 1998: Australian Institute of Steel
Construction.
[25] K. Law, L. Gardner, Lateral instability of elliptical hollow section beams, Eng.
Struct. 37 (2012) 152–166.
[26] BSI, BS EN, Eurocode 3: Design of Steel Structures, Part 1-1: General Rules and
Rules for Buildings, BSI, London, UK, 1993 1993-1-1.
[27] M.A. Bradford, H.Y. Loh, B. Uy, Slenderness limits for filled circular steel tubes,
J. Constr. Steel Res. 58 (2) (2002) 243–252.
[28] M.H. Lai, J.C.M. Ho, Confining and hoop stresses in ring-confined thin-walled
concrete-filled-steel-tube columns, Mag. Concr. Res. (2015) (in preparation).