Professional Documents
Culture Documents
Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws
art ic l e i nf o a b s t r a c t
Article history: To improve the behaviour of conventional T-shaped concrete-filled steel tubular (CFST) columns, multi-
Received 21 April 2014 cell composite T-shaped concrete-filled steel tubular (MT-CFST) columns are proposed in this paper.
Received in revised form Experimental study of 25MT-CFST columns, including 13 short specimens with various cross sections
5 August 2014
and material properties and 12 slender specimens with different slenderness ratios, subjected to axial
Accepted 5 August 2014
loads was conducted. The failure modes, axial load–strain curves for short specimens and axial load–
lateral deflection curves for slender specimens were investigated. The test results were compared with
Keywords: design approaches for conventional CFST columns presented in Eurocode 4, AISC specification, Australian
Concrete-filled steel tube standard AS51006, Chinese code CECS159, and Hong Kong steel code, and it was found that all the design
T-shaped cross section
codes underestimate the bearing capacity of both short and slender MT-CFST columns to some extent.
Multi-cell
& 2014 Elsevier Ltd. All rights reserved.
Composite columns
Mechanical behaviour
http://dx.doi.org/10.1016/j.tws.2014.08.003
0263-8231/& 2014 Elsevier Ltd. All rights reserved.
58 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
composite column. Yang et al. [14] conducted an experimental columns that were subjected to axial loads so as to investigate
study of welding battlement-shaped bar stiffeners and tensile the effect of cross-sectional geometries (wall thickness, c/a and b/a
bar stiffeners on tube surfaces to improve the axial compressive ratio) and material properties (concrete strength). Group II con-
behaviour of T-shaped CFST columns. But it was expensive to weld tained 15 columns that were subjected to axial compression so as
bar stiffeners on the steel plates and fold the steel plates into to study the influence of the slenderness ratio. Details of the
a T-shaped cross-sectional hollow tube. Instead of two rectangular specimens are listed in Tables 1 and 2, and the parameters (a, b, c
steel tubes or stiffeners, this paper proposes the multi-cell and t) and axis (x–x and y–y) of the composite T-section are
composite T-shaped concrete-filled steel tubular (MT-CFST) defined in Fig. 2. To avoid the effects of overall buckling and end
column for targeting these two problems. conditions, the physical length (L, not including the cover-plates)
The main objective of the present paper is to understand the of the short specimen was chosen to be 3 times the width of the
behaviour of MT-CFST column subjected to axial load. The experi- longest side of the T-section [15].
mental campaign is focused on two aspects: (1) short columns
with various cross-sectional geometries and material properties; 2.2. Material properties
(2) slender columns considering different slenderness ratios. In
addition, comparisons between experimental results and available The material properties of steel and concrete were obtained
design codes are discussed in this paper to validate the feasibility according to the requirements defined in China National Codes
of the design codes. (GB/T 50081-2002 and GB/T 228-2002). Test coupons were fabri-
cated with the steel material used for constructing the steel tubes
and were tested in tension. The test results suggest that these steel
2. Experimental investigation coupons don't have obvious yielding platform. Fig. 3 shows the
Table 2
Test data of specimens in Group II.
Table 3
Steel coupon tests results.
9 3
450
7 5
400
350
300 6
Stress (MPa)
typical stress–strain curves selected from the tension tests. The The experiments were conducted at Institute of Building
average 0.2% proof stress (σ 0:2 ) and average elastic modulus (Es) for Structures, China Academy of Building Research (CABR). The
steel coupons with the thickness of 2.4 mm are 326 MPa and specimens were tested on a 500 t capacity testing machine
208 GPa, respectively. More details about the material properties 30 days after the casting. For the sake of convenience, the faces
of steel tubes obtained from steel coupon tensile tests are of the specimens are numbered from 1 to 9, as depicted in Fig. 4.
summarized in Table 3. Both vertical and transverse strain gauges were arranged on all the
Concrete prisms (100 100 300 mm3 ) made at the time of face centres of the specimen to record the axial and transverse
the concrete casting were tested to determine the concrete strains, respectively.
60 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
In Group I, as illustrated in Fig. 6(b), short specimens were variable displacement transducers (LVDTs) were employed to
placed on the bottom rigid padding plate and were loaded directly record the vertical axial displacement, in which 4 LVDTs were
by the raising of the bottom rigid plate of the machine. 6 linear arranged symmetrically between the two rigid plates of the
machine and the other 2 were placed symmetrically between
the two cover-plates of the specimens. In order to explain the
20mm installation of LVDTs explicitly, only 3 LVDTs are shown in Fig. 6(a).
For specimens in Group II, two hinges (in Figs. 5 and 6(c)) that had
sharp knives installed at the middle were used at both ends of
40mm each specimen to ensure that the specimen bent about the
unsymmetrical x–x axis or about the symmetric y–y axis. Thus,
the effective length (Le ) of a specimen was calculated from its
physical length plus 100 mm. The vertical axial displacement and
200mm
the lateral deflection of the column were recorded via 4 LVDTs
Fig. 5. Size of the hinge. arranged at the middle of the specimen and 2 LVDTs arranged
Fig. 6. Test setup, (a) Test setup for short specimen (b) Illustrative drawing of loading for short specimen (c) Test setup for slender specimen.
Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 61
between the two rigid plates of the machine, respectively. The test failure was observed by the buckling of the tubes between one end
setup for slender specimens under axial compression is shown in and the mid-height of the specimen, followed by concrete crush-
Fig. 6(c). A computerized data acquisition system was used for ing and the outer steel tube separating from the in-filled concrete.
collecting the load and deformation data. Some specimens locally buckled outward near the bottom or near
A pre-load level of 20 kN was applied to make sure that the the top, and displayed multi-wave bulking mode, as shown in
plates of the machine and the specimens were attached firmly to Fig. 7(b) and (c). Compared with other short MT-CFST columns,
each other. When the axial load reached 80% of the estimated specimen 8 (in Fig. 7(d)) did not exhibit obvious local deformation
ultimate load, the load increment changed to half. Each increment due to the small width-to-thickness ratio and the low strain
was maintained for 2 min, and all specimens were compressed to development after the failure load. For all the tested specimens
failure. in this paper, there was no weld failure occurrence throughout the
loading process.
3. Experimental results and discussions 3.1.2. Ultimate load and ductility of short columns
To evaluate the positive interaction between concrete and steel
3.1. Behaviour of short columns under axial compression tube, the bearing capacity index (SI) is defined as
SI ¼ N us =Nu ð1Þ
3.1.1. Failure phenomenon
In general, there was similarity in the failure procedures of where Nus is the short columns' experimental bearing capacity,
specimens. No obvious deformation occurred at surfaces of the and Nu is the nominal bearing capacity which is determined in
specimens in the initial loading process. At higher levels of Eq.2.
loading, sounds were heard clearly due to the cracking of the
Nu ¼ f y As þAc ð2Þ
concrete. Once local buckling started at one surface of the speci-
men, consequential local buckling appeared at neighbouring where As and Ac are the area of steel and concrete in the composite
0
surfaces of the same height or the specimen displayed staggered cross section, respectively; fy is the yield strength of steel and f c is
buckling pattern. the cylinder compressive strength of concrete.
A general view of short MT-CTST columns after testing is shown The value of SI is listed in Table 1. It can be found that, in
in Fig. 7. It is obvious that the position of local buckling is different general, SI decreases with the increase in c/a or b/a ratio, ranging
across specimens. The final amount of local buckling in these from 1.028 to 1.073. With a SI of 1.150, the bearing capacity is
specimens is different as a consequence of differences in the greatly enhanced for specimen 8 compared with other MT-CFST
strains after the failure load. As illustrated in Fig. 7(a), typical columns in Group I. It is also clear that increasing the compressive
Fig. 7. A general view of short specimens after testing (a) Typical short specimen (b) Short specimen buckled near the bottom (c) Short specimen buckled near the top
(d) Specimen 8.
62 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
strength of the in-fill concrete has no obvious effect on the Group I are higher than 2.0, which shows good ductility perfor-
variance of SI. As shown in Table 1, SI varies slightly from 1.036 mance of the MT-CFST columns.
to 1.051.
One of the most significant characteristics of CFST columns is
3.1.3. Effect of c/a or b/a ratio of the composite section
their high ductility which is beneficial to the seismic resistance
The axial strain (ε) is defined as the ratio of axial shortening ðΔLÞ
capability of a structural system. Nevertheless, no unified formula
obtained from the LVDTS to the physical length (L) of the specimen.
is available to determine the ductility of varied cross-sectional
Fig. 8(a) and (b) shows the relationships between axial load (N) and
CFST columns. Based on the definition of the ductility performance
axial strain (ε) of specimens (1–6) in Group I. It can be seen that the
index (DI) for traditional CFST columns suggested in [3], the DI of
ultimate load and axial stiffness of MT-CFST columns increase with an
MT-CFST columns can be defined as
increase in c/a or b/a ratio of the cross section, whereas the ductility of
MT-CFST columns decreases. On one hand, the area of composite
DI ¼ ε0:90 =εy ð3Þ
cross section increases with c/a or b/a ratio, which means an increase
in the plastic resistance as presented in Eq. 2. On the other hand, with
where ε0:90 is the axial strain corresponding to 90% of the ultimate
the extending of web or flange, the steel tubes have little effect on the
load and εy is the axial strain when the composite section is at
confinement to the encased concrete, which results in a reduction in
yield. The yield strain εy is defined as the strain of the intersection
the ductility performance of the columns.
point of two lines: the initial elastic tangent of the axial load–
strain curve and the flat tangent of the ultimate point on the axial
load–strain curve. 3.1.4. Effect of the thickness of the steel tube
The value of DI is presented in Table 1, where symbol ‘-’ Fig. 8(c) depicts the effect of the thickness of the steel tube on
indicates that very large axial strain has occurred before the load the axial load (N) and axial strain (ε) curves of specimens 4, 7 and
falls to 90% of the ultimate load and represents a perfect ductility. 8. With the increase of steel tube thickness, the width-to-thickness
It can be seen that, except the DI of specimen 8 that didn't attain ratio of the steel plate for the same cross section decreases. At the
the demanding strain, the DIs of all other short specimens in same time, the yield strength of the steel increases, as shown in
2000 2500
2000
1500
Axial load (KN)
1500
1000
1000
2000 2000
1500 1500
Axial load (KN)
1000 1000
3.2.1. Test observations unsymmetrical axis x, while flexural buckling and twist about
For a centrally compressed slender MT-CFST column in Group longitudinal axis that passed through the shear centre of the cross
II, flexural buckling occurred when the column bent about the section simultaneously occurred when the column bent about the
64 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
symmetric axis y. Fig. 11(a) shows a view of the 1.8 m long speci- bending deformation, which can be seen from Fig. 11(c), exhibited
mens with three section sizes (60 100 80 mm3, 60 120 at the end of the testing. Moreover, there was only a small diffe-
60 mm3 and 60 120 120 mm3) after testing. It was found that rence in the bearing capacity between stub specimens in Group II
the failing of the 1.8 m long MT-CFST columns resulted from and corresponding ones in Group I, as shown in Tables 1 and 2.
overall buckling with obvious later deflections at the mid-height. Compared with the ultimate load capacity of short MT-CFST
For the 1.3 m long MT-CFST columns shown in Fig. 11(b), their columns subjected to axial loads, a significant reduction in the
failure was characterised mainly by global buckling, coupled with ultimate load capacity was observed for all slender specimens.
obvious local buckling on the compressive region of the column
near the top or bottom at the end of the testing. The reason might
be load eccentricity or the imperfect quality of in-filled concrete. 3.2.2. Strain analysis
Moreover, it has been proved by Yu et al. [17] that local buckling Two typical axial load versus axial and lateral strain curves in
near the end of CFST columns has little effect on the overall Group II are shown in Fig. 12, where the strains are named in the
axial load bearing capacity. In reality, as the bearing capacity of a same way as the specimens in Group I. Increasing divergence
slender specimen under axial compression was affected by both occurs after the elastic region for specimen 1-x-450, mainly due
residual stress and geometric imperfections (initial crookedness as to the influence of slight bending of the column and different
well as initial eccentricity), its behaviour was similar to that of a confinement of the concrete offered by different steel edges of the
beam-column and it belonged to the extreme point instability. composite cross section as well. It also can be found that the whole
As far as the three stub specimens (450 mm, 500 mm and section was under compression during the loading process. Owing
700 mm) in Group II were concerned, their failure mode was to the influence of initial deflection, the divergence among the
similar to the specimens in Group I, which was dominated by the measured strains for specimen 2-y-1800 takes place earlier than
section capacity under axial compression. However, an obvious that for specimen 1-x-450. Furthermore, obvious divergence was
Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 65
Fig. 12. Axial load versus axial and lateral strain curves.
observed in the inelastic region due to the bending of the slender torsional buckling of MT-CFST columns.
composite column after elastic region. It should be noticed that sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
some strains developed towards the opposite direction after the 1 2
λyw ¼ λy þ λ2w þ ðλ2y þ λ2w Þ2 4λ2y λ2w ð1 y20 =i20 Þ ð6Þ
failure load which indicated that part of the cross section reversed 2
from compression to tension followed by global buckling.
where y0 is the distance from the centroid to the shear centre of
the cross section, and the parameters i0 andλw are defined in
3.2.3. Effect of slenderness ratio Eqs. (7) and (8) respectively.
To investigate the effect of the slenderness ratio on the flexural
buckling of MT-CFST columns, the nominal flexural slenderness i20 ¼ y20 þ i2x þ i2y ð7Þ
ratio about the unsymmetrical x–x axis ( λx ) and about the sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
symmetric y–y axis ( λy ) are defined as follows: ffi
It
λw ¼ Asc i0 =
2
ð8Þ
λx ¼ Le =ix ð4Þ 2π 2 ð1 þ μsc Þ
1500 2-Y-1800
0
0 10 20 30 40
1000 Mid height deflection (mm)
6-X-1300
2000
500 6-Y-1300
6-X-1800
Axial load (KN)
6-Y-1800
1500 0
0 5 10 15 20 25
Mid height deflection (mm)
1000
Fig. 14. Axial load versus mid-height lateral deflection curves (a) Cross section
1 — 60 100 80 mm3 (b) Cross section 2 — 60 120 60 mm3 (c) Cross section
cross section: 60×100×80 6–60 120 120 mm3.
500
cross section: 60×120×60
slenderness ratios. In general, the ultimate axial load decreases as
cross section: 60×120×120
the nominal slenderness ratio increases. It should be noted that
0
1000 1200 1400 1600 1800 2000 the N–Um curves of some 1.3 m long columns tend to drop more
quickly than those of 1.8 m long columns with the same cross
Physical length of the column (mm)
section after the ultimate load. This is because the failing of the
Fig. 13. Physical length versus axial load (a) About x–x axis (b) About y–y axis. 1.3 m long columns mainly resulted from local buckling whereas
Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 67
overall buckling resulted in the final failing of 1.8 m long columns. where Es is the elastic modulus of steel; Ecm is the secant modulus
Besides, the flexural buckling loads about the unsymmetrical x–x of elasticity for concrete given in Eurocode 2 [25]; I s and I c are the
axis are larger than the flexural-torsional buckling loads about the second moments of area of steel tube section and concrete section,
symmetric y–y axis for the cross section 60 100 80 mm3 and respectively.
60 120 60 mm3, due to a fact that the nominal slenderness
ratios of the former are less than those of the latter. 4.1.2. AISC specification
The proposed composite section in this paper belongs to the
compact section in AISC specification. The design compressive
4. Comparison between tests and design codes strength of doubly symmetric axially loaded composite members
shall be determined for the limit state of flexural buckling based
Although several current design codes such as Eurocode 4 on member slenderness as follows:
(EC4) [19], AISC specification [20], Australian standard AS5100.6
P u ¼ ϕc P n ; ϕc ¼ 0:75 ð15Þ
[21], Chinese code CECS159 [22], and Hong Kong steel code (HKS)
[23] are available for the design of conventional CFST columns,
none of them covers MT-CFST columns. Based on the design rules
provided in these codes for rectangular CFST columns, compar- (1) when P n0 =P e r 2:25
isons will be made between the experimental results presented in h P n0 i
the previous section and the design codes shown in this section. In P n ¼ P n0 0:658 Pe ð16Þ
order to check the feasibility of these design codes, all limitations
are ignored. For an accurate comparison between the test results
(2) when P n0 =P e 4 2:25
and design codes, material factors and partial safety factors are
applied in the calculation of design values. P n ¼ 0:877P e ð17Þ
where
4.1. Brief review of current design codes P n0 ¼ As f y þ 0:85Ac f c' ; P e ¼ π 2 ðEIÞef f =L2e ;-
ðEIÞef f ¼ Es I s þ C 3 Ec I c ;C 3 ¼ 0:6 þ 2 As Aþs Ac r0:9is the coeffi-
The section capacity of a CFST stub column is generally
determined by summing up the strengths of the steel tube and cient for calculation of effective rigidity of filled
the in-filled concrete although it is defined slightly differently in composite compression member.
the current design codes due to the non-unified material factors
and partial safety factors. It should be noted that none of the 4.1.3. Australian standard AS5100.6
design codes takes into account the enhancement of the compres- The predicted ultimate section capacity is
sive strength of the in-filled concrete. In general, all codes predict Nps ¼ ϕs As f y þ ϕc Ac f c
0
ð18Þ
the member capacities of slender CFST columns in a similar
approach to steel columns. Since the scope of these codes is where ϕs ¼ 0:9 is the capacity reduction factor for structural steel
limited to doubly symmetrical composite cross section along the in compression and ϕc ¼ 0:6 the capacity reduction factor for
member length, only flexural buckling is considered in this section concrete in compression, respectively.
and the effect of torsional buckling is ignored. In addition, EC4 and The predicted ultimate member capacity is
HKS adopt a similar method for predicting the ultimate member N pl ¼ χ N ps r N ps
capacity of a CFST column. qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
χ ¼ ξ ½1 1 ð90=ξλÞ2 ð19Þ
4.1.1. Eurocode 4
ðλ=90Þ
2
þ1þη
The predicted ultimate section capacity of a MT-CFST stub where ξ ¼ ðλ=90Þ2
; λ ¼ λη þ αa αb ; λη ¼ 90λr is the modified
column is member slenderness; αa and αb are the coefficients defined in
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0
N ps ¼ As f y =γ s þ Ac f c =γ c ð10Þ AS5100.6; η ¼ 0:00326ðλ 13:5Þ Z 0; λr ¼ N u =Ncr is the relative
slenderness; Ncr ¼ π 2 ðEIÞef f =L2e ; ðEIÞef f ¼ ϕs Es I s þ ϕc Ec I c ; ϕs ¼ 0:9;
where γ s ¼1.1 is the partial safety factor for structural steel and
ϕc ¼ 0:6.
γ c ¼ 1:5 the partial safety factor for concrete, respectively.
The predicted ultimate member capacity of a slender MT-CFST
column is 4.1.4. Chinese code CECS159
CECS159 uses an equation similar to Eq. (10) to predict the
N pl ¼ χ N ps r N ps ð11Þ ultimate section capacity of a CFST stub column, where the partial
safety factor for concrete γ c is 1.4 and the cylinder compressive
1 0
χ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð12Þ strength f c is replaced by prism compressive strength fck.
2
ϕ þ ϕ2 λ The predicted ultimate member capacity is
N pl ¼ χ N ps r N ps
where χ is the reduction factor; λ is the non-dimensional slender-
2
ness; ϕ ¼ 0:5½1 þ αðλ 0:2Þ þ λ ; α is the imperfection factor for
2
χ ¼ 1 0:65 λ for λ r 0:215 ð20Þ
buckling curve given in Eurocode 3 [24]. The non-dimensional
slenderness λ for the bending plane is expressed as: 1 h 2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2 2
i
χ¼ 0:965 þ0:3 λ þ λ ð0:965 þ 0:3 λ þ λ Þ2 4λ for λ 4 0:215
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2λ
2
λ ¼ Nu =N cr ð13Þ
ð21Þ
where N u is the nominal section capacity given by Eq. (2); where the non-dimensional slenderness λ is defined as:
N cr ¼ π 2 ðEIÞef f =L2e is the elastic critical normal force. The effective sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
flexural stiffness (EI)eff is determined in Eq. (14). le f y As þf ck Ac
λ¼ ð22Þ
ðEIÞef f ¼ Es I s þ 0:6Ecm I c ð14Þ π Es I s þ Ec I c
68 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
4.1.5. Hong Kong code are from HKS (μ ¼0.795, σ ¼ 0.017). The predicted ultimate section
The predicted ultimate section capacity is capacities have slight differences between EC4 and HKS. Because
concrete cube compressive strength is used in HKS, it predicts
N ps ¼ As f y þ 0:53Ac f cu ð23Þ
higher strength than EC4 and gives better predictions. It seems
where f cu is the cube compressive strength of concrete. that a higher concrete compressive strength than the cylinder
The HKS approach is quite similar to the EC4 approach in compressive strength in the design formula is closer to the actual
predicting the member capacity of a CFST column, except that it condition. Therefore, further research should be done in the future
0
replaces 1.0 by 0.8 and f c by f cu in calculating the nominal section to develop an accurate design approach for MT-CFST stub columns.
capacity defined by Eq. (2).
EC4 2000
2000
EC4
AISC 1800
Predicted ultimate loads (kN)
AISC
Predicted ultimate loads (kN)
1800 AS5100.6
1600 AS5100.6
CECS159
1600 CECS159
HKS 1400 HKS
1400
1200
1200
1000
1000 800
800 600
1000 1500 2000 1000 1500 2000
Measured ultimate loads Nus (kN) Measured ultimate loads N ul (kN)
Fig. 15. Predicted ultimate section capacities versus test results. Fig. 16. Predicted ultimate member capacities versus test results.
Table 4
Comparison between predicted ultimate section capacities and experimental results.
Specimen N us (kN) N EC4 (kN) N EC4 /N us N AISC (kN) N AISC /N us N AS (kN) N AS /N us N CECS (kN) N CECS /N us N HKS (kN) N HKS /N us
1 1550.5 1206.0 0.778 1067.7 0.689 1142.1 0.737 1162.0 0.749 1242.7 0.801
2 1494.3 1129.4 0.756 994.0 0.665 1070.4 0.716 1088.9 0.729 1165.2 0.780
3 1961.0 1512.7 0.771 1345.1 0.686 1428.8 0.729 1454.4 0.742 1552.6 0.792
4 1466.7 1129.4 0.770 998.6 0.681 1070.4 0.730 1088.9 0.742 1165.2 0.794
5 1661.3 1282.7 0.772 1135.9 0.684 1213.8 0.731 1235.1 0.743 1320.2 0.795
6 2033.0 1589.4 0.782 1414.1 0.696 1500.5 0.738 1527.5 0.751 1630.1 0.802
7 1281.0 1013.4 0.791 919.2 0.718 954.1 0.745 971.7 0.759 1035.1 0.808
8 1849.3 1334.3 0.722 1153.7 0.624 1276.3 0.690 1296.4 0.701 1395.4 0.755
9 1314.5 1033.1 0.786 902.9 0.687 986.4 0.750 1002.3 0.762 1038.3 0.790
10 1628.6 1265.4 0.777 1124.1 0.690 1195.6 0.734 1216.9 0.747 1285.2 0.789
1-x-450 1564.8 1206.0 0.771 1067.7 0.682 1142.1 0.730 1162.0 0.743 1242.7 0.794
2-x-500 1418.6 1129.4 0.796 994.0 0.701 1070.4 0.755 1088.9 0.768 1165.2 0.821
6-x-700 1994.7 1589.4 0.797 1414.1 0.709 1500.5 0.752 1527.5 0.766 1630.1 0.817
Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70 69
Table 5
Comparison between predicted ultimate member capacities and experimental results.
Specimen N ul (kN) N EC4 (kN) N EC4 /N ul N AISC (kN) N AISC /N ul N AS (kN) N AS /N ul N CECS (kN) N CECS /N ul N HKS (kN) N HKS /N ul
1-2-x 1349.3 1077.6 0.799 1013.9 0.751 1008.1 0.747 1083.9 0.803 1113.4 0.825
1-3-y 1276.7 1035.9 0.811 994.5 0.779 972.5 0.762 1062.2 0.832 1071.1 0.839
1-4-x 1207.5 972.8 0.806 964.1 0.798 917.9 0.760 1031.2 0.854 1007.1 0.834
1-5-y 1198.3 909.2 0.759 929.1 0.775 861.7 0.719 995.8 0.831 942.5 0.787
2-2-x 1261.1 1043.5 0.827 1000.2 0.793 974.0 0.772 1030.0 0.817 1079.0 0.856
2-3-y 1209.2 873.2 0.722 911.3 0.754 827.1 0.684 936.5 0.774 905.8 0.749
2-4-x 1146.5 962.6 0.840 963.1 0.840 905.1 0.789 988.8 0.862 996.7 0.869
2-5-y 963.8 706.0 0.733 805.6 0.836 675.2 0.701 829.8 0.861 735.5 0.763
6-2-x 1733.6 1446.0 0.834 1353.5 0.781 1346.3 0.777 1439.4 0.830 1487.0 0.858
6-3-y 1818.8 1490.6 0.820 1370.6 0.754 1383.8 0.761 1460.7 0.803 1532.0 0.842
6-4-x 1487.8 1321.1 0.888 1296.5 0.871 1239.7 0.833 1378.5 0.927 1360.7 0.915
6-5-y 1576.1 1387.0 0.880 1328.1 0.843 1296.2 0.822 1411.1 0.895 1427.4 0.906
1 5. Conclusions
0.6 results. From the results of this study, the following conclusions
can be drawn within the scope of this study
0.4 (1) Compared with the plastic resistance under axial compression,
the bearing capacity of short MT-CFST columns were
enhanced, and all specimens exhibited good ductility.
(2) Increasing c/a or b/a ratio increases the resistance of the cross
0.2
Reduction factor about the x-x axis section under axial compression but reduces the axial ductility
of MT-CFST columns. The ultimate load of MT-CFST columns
Reduction factor about the y-y axis
increases along with a rise in the thickness of the steel tube.
0
0 0.2 0.4 0.6 0.8 1 Also, a MT-CFST column consisting of thick steel tubes is hard
Non-dimensional slenderness to buckle under axial compression.
(3) Increasing the concrete compressive strength increases the fail-
Fig. 17. Comparison of measured reduction factors about the x–x and y–y axes. ure load of MT-CFST columns but generally decreases their axial
ductility. In addition, an increase in the concrete compressive
capacity. Since none of the codes gives accurate prediction on the strength has a slight effect on the variation of axial ductility since
member capacities of MT-CFST slender columns proposed in this the concrete compressive strength is relatively high.
paper, further research should be done in the future to present a (4) In general, for slender MT-CFST columns, the failure load decreases
suitable design method for slender MT-CFST columns. along with an increase of slenderness ratio. The reduction factors
For the purpose of practical design, the measured reduction about the x–x and y–y axes for MT-CFST columns can be calculated
factor χm of a slender MT-CFST column is defined as using the same buckling curve in practical designs, and the effect
of torsional buckling can be ignored.
χm ¼ Nul =Nus ð24Þ
(5) All the design codes used in this paper are conservative in
where Nul and Nus are the measured member capacity of a slender predicting the bearing capacity of both short and slender MT-
column and section capacity of the corresponding short column, CFST columns, where HKS gives the closest predictions for
respectively. The value of χm is listed in Table 2. short MT-CFST columns and CECS159 gives the best predic-
Fig. 17 compares measured reduction factors about the unsym- tions for slender MT-CFST columns. Compared with AISC and
metrical x–x axis (χ mx ) and about the symmetric y–y axis (χ my ). AS5100.6, EC4 and HKS give less conservative predictions for
The non-dimensional slenderness λ in the figure is computed by both short and slender MT-CFST columns.
Eq. (13). It should be pointed out that different ways of calculating
the non-dimensional slenderness λ in the design codes have no
effect on the comparison of measured reduction factors χ mx and
χ my . Although torsional buckling is not considered in calculating Acknowledgements
the non-dimensional slenderness λ for the bending plane, some
measured reduction factors about the y–y axis are located rela- This programme was sponsored by the National Natural
tively higher than the reduction factors about the x–x axis. This Science Foundation of China under Grant no. 51278019.
indicates that the effect of torsional buckling is very slight and can
be ignored for MT-CFST columns. Furthermore, this is consistent References
with the fact that the values of nominal flexural slenderness λy are
very close to those of nominal flexural-torsional slenderness λyw [1] Kloppel VK, Goder W. An investigation of the load carrying capacity of
listed in Table 2. Therefore, it is appropriate to calculate the concrete-filled steel tubes and development of design formula. Der Stahlbau
1957;26(2):44–50.
reduction factors about the x–x and y–y axes using the same [2] Schneider SP. Axially loaded concrete-filled steel tubes. J Struct Eng (ASCE)
buckling curve in practical designs. 1998;124(10):1125–38.
70 Y.Q. Tu et al. / Thin-Walled Structures 85 (2014) 57–70
[3] Han LH. Tests on stub columns of concrete-filled RHS sections. J Constr Steel [14] Yang YL, Yang H, Zhang SM. Compressive behaviour of T-shaped concrete filled
Res 2002;58(3):353–72. steel tubular columns. Int J Steel Struct 2010;10(4):419–30.
[4] Ellobody E, Young B, Lam D. Behaviour of normal and high strength concrete- [15] Tao Z, Han LH, Wang ZB. Experimental behaviour of stiffened concrete-filled
filled compact steel tube circular stub columns. J Constr Steel Res 2006;62 thin-walled hollow steel structural (HSS) stub columns. J Constr Steel Res
(7):706–15. 2005;61(7):962–83.
[5] Liu DL, Gho WM. Axial load behaviour of high-strength rectangular concrete- [16] Teng ZM, Zhu JS. Reinforced concrete and masonry structure. Beijing: China
filled steel tubular stub columns. Thin-Walled Struct 2005;43(8):1131–42. Architecture and Building Press; 1992 (in Chinese).
[6] Uy B, Tao Z, Han LH. Behaviour of short and slender concrete-filled stainless [17] Yu M, Zha XX, Ye JQ, Li YT. A unified formulation for circle and polygon
steel tubular columns. J Constr Steel Res 2011;67(3):360–78. concrete-filled steel tube columns under axial compression. Eng Struct
[7] Mursi M, Uy B. Strength of slender concrete filled high strength steel box 2013;49:1–10.
columns. J Constr Steel Res 2004;60(12):1825–48. [18] GB50017. Code for design of steel structures. (in Chinese). Beijing: China
[8] An YF, Han LH, Zhao XL. Behaviour and design calculations on very slender
Planning Press; 2003.
thin-walled CFST columns. Thin-Walled Struct 2012;53:161–75.
[19] Eurocode 4. Design of composite steel and concrete structures, part 1.1:
[9] Hernandez-Figueirido D, Romero ML, Bonet JL, Montalva JM. Influence of
general rules and rules for buildings. London: British Standards Institution;
slenderness on high strength rectangular concrete-filled tubular columns with
2004.
axial load and non-constant bending moment. J Struct Eng ASCE 2012;138
[20] AISC 360–10. Specification for structural steel buildings. Chicago, Illinois:
(12):1436–45.
[10] Han LH, Zheng LQ, He SH, Tao Z. Tests on curved concrete filled steel tubular American Institute of Steel Construction; 2010.
members subjected to axial compression. J Constr Steel Res 2011;67(6): [21] Standards Australia. Bridge design, part 6: steel and composite construction.
965–76. AS 5100.6-2004. Sydney; 2004.
[11] Dai XH, Lam D, Jamaluddin N, Ye J. Numerical analysis of slender elliptical [22] CECS 159. Technical specification for structures with concrete-filled rectan-
concrete filled columns under axial compression. Thin-Walled Struct 2014;77: gular steel tube members. Beijing: China Planning Press; 2004 (in Chinese).
26–35. [23] Code of Practice for the Structural Use of Steel 2011. Hong Kong: Buildings
[12] Lu XL, Li XP, Wang D. Modelling and experimental verification on concrete- Department; 2011.
filled steel tubular columns with L or T section. Front Struct Civ Eng 2007;1 [24] Eurocode 3. Design of steel structures, Part1-1: General rules and rules for
(2):163–9. buildings. London: British Standards Institution; 2003.
[13] Du GF, Xu LH, Wen F. Experimental study on normal section compression [25] Eurocode 2. Design of concrete structures, Part 1: General rules and rules for
bearing capacity of composite T-shaped concrete-filled steel tubular columns. buildings. London: British Standards Institution; 2001.
China Civ Eng J 2009;42(6):14–20 (in Chinese).