Académique Documents
Professionnel Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: The compressive and flexural behavior of the hexagonal concrete-filled steel tubular (CFST) members is investi-
Received 3 August 2015 gated in this study. Tests are carried out on the specimens under axial compression and bending. The parameter
Received in revised form 21 March 2016 of the experimental work is the steel ratio of the cross section, and the hollow tubular members are designed for
Accepted 23 April 2016
comparison. The failure modes, load versus deformation relations and strain developments are studied. A finite
Available online xxxx
element analysis (FEA) model is established and verified with the test results. The FEA model is then used to con-
Keywords:
duct the full range analysis on the load versus deformation relation, internal force distribution and stress devel-
Hexagonal concrete-filled steel tube (CFST) opment. The parametric study on the compressive and flexural behavior is conducted to investigate the effects of
Compressive behavior various parameters, e.g. the steel ratio, the steel yield strength and the concrete strength. Simplified models are
Flexural behavior proposed to calculate the ultimate compressive and flexural strength. It is found that the ultimate compressive
Finite element analysis (FEA) strength of hexagonal CFST stub columns could be calculated accurately referring to the equations of rectangular
Ultimate strength CFST columns in DBJ/T13-51-2010 and EC4, and the plastic stress distribution method and the fiber model meth-
od achieve good predictions of the ultimate flexural strength of the hexagonal CFST beams.
© 2016 Elsevier Ltd. All rights reserved.
1. Introduction shaped CFST columns. Tu et al. [5] studied the hysteretic behavior
of T-shaped CFST columns, where the section was divided into
Concrete-filled steel tubular (CFST) member has been widely multi-cells by stiffeners. Similar research was also conducted on L-
used in buildings, bridges and other structures owing to its high shaped CFSTs by Zuo et al. [6] and Zhou et al. [7]. A series of experi-
strength, good ductility and excellent workability. The member is mental and numerical work were conducted on elliptical CFSTs,
consisted of outer steel tube and core concrete, and the cross- such as Yang et al. [8], Zhao and Packer [9], Jamaluddin et al. [10],
sectional shape of the outer steel tube could be circular, rectangular Dai et al. [11]. Experiments were also conducted by Evirgen et al.
(square), elliptical or polygonal. Recently, CFST members with hex- [12] on regular hexagonal CFST, and Ren et al. [13] on triangular, D-
agonal cross section are used in some high-rise buildings for their shaped and other specified cross sections.
aesthetic performance, as shown in Fig. 1(a) and (b), where the Up to now, research on the hexagonal CFST is rather limited. The
members act as mega columns in the mega frame-core tube struc- specifications such as EC4 [14], ACI318-14 [15], ANSI/AISC360-10 [16],
tures. Moreover, the hexagonal shape makes the column easier to DBJ/T13-51-2010 [17], AIJ-2008 [18] provide design methods of the
be connected with beams and the core tube. CFST member, but the feasibility of these methods for designing the
Previously, CFST members have attracted a great deal of investiga- hexagonal CFST is unknown. Therefore, the objectives of the paper are
tions [1–3]. Most studies were conducted on CFST members with circu- summarized as follows:
lar and rectangular (square) cross sections. The confinement effect
between the tube and concrete of the circular CFST is the strongest, (1) A series of compression and bending tests are conducted to pro-
while that of rectangular (square) section is much weaker. For the vide new experimental data on hexagonal CFST stub columns and
hexagonal CFST, the confinement effect is assumed to be between beams. The failure mode, load versus deformation relation and strain
those of circular and rectangular CFSTs. development are discussed.
Studies on CFST members with special cross sections were also (2) A finite element analysis (FEA) model is proposed to predict the
performed. Yang et al. [4] conducted compression test on 9 T- compressive behavior and the flexural behavior. The load versus defor-
mation relation, internal force distribution and stress development are
further analyzed using the verified FEA model.
(3) Parametric studies are conducted to investigate the effect of
⁎ Corresponding author. parameters, and simplified models are proposed to predict the load
E-mail addresses: lhhan@tsinghua.edu.cn, lhhanqw@gmail.com (L.-H. Han). carrying capacity of hexagonal CFST stub columns and beams.
http://dx.doi.org/10.1016/j.jcsr.2016.04.026
0143-974X/© 2016 Elsevier Ltd. All rights reserved.
W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175 163
Nomenclature
Table 1
Summary of stub column specimens.
1 C4-1 3.95 0.109 0.717 1768 1865 1851 1827 1634 1806
2 C4-2 3.95 0.109 0.717 1500 1845
3 C2.5-1 2.46 0.065 0.482 2366 1598 1663 1623 1476 1654
4 C2.5-2 2.46 0.065 0.482 1834 1492
5 C6-1 5.93 0.173 1.232 2544 2062 2221 2274 1993 2155
6 C6-2 5.93 0.173 1.232 1503 2195
7 CH4-1 3.95 - - 1587 680 637 - - -
8 CH4-2 3.95 - - 1480 690
Table 2
Summary of beam specimens.
1 B4-1 3.95 0.109 0.547 −4381 19.5 0.062 60.1 71.8 56.2 54.9 53.6 53.2
2 B4-2 3.95 0.109 0.547 −4245 19.6 0.061 56.8 68.5
3 B2.5-1 2.46 0.065 0.368 −2261 9.3 0.052 36.2 45.0 40.9 40.8 39.9 39.4
4 B2.5-2 2.46 0.065 0.368 −2991 13.9 0.059 38.8 46.1
5 B6-1 5.93 0.173 0.939 −5462 20.3 0.066 72.0 83.1 84.1 81.7 79.3 80.0
6 B6-2 5.93 0.173 0.939 −4211 18.7 0.061 71.4 83.4
used, namely, 2.46 mm, 3.95 mm and 5.93 mm, corresponding to α of the plates had a hole in the middle, and the concrete was poured
0.065, 0.109 and 0.173, respectively. through this hole.
Table 3
Material properties of steel tubes.
buckling of steel tube occurred near the middle section for all the hexag-
onal CFST columns. The phenomenon was more obvious when the steel
Label t (mm) fy (N/mm2) fu (N/mm2) Es (N/mm2) λ δ ratio (α) was smaller. For specimen C2.5-1 (α = 0.065), the outward
P-4 3.95 279 427 2.04 × 105 0.262 0.33 buckling was serious, resulting in a weld crack at the buckled zone. Fail-
P-2.5 2.46 313 438 2.04 × 105 0.287 0.31 ure modes of concrete are also shown in the same figure, and the con-
P-6 5.93 302 429 1.98 × 105 0.282 0.36
crete was crushed where the tube buckled. The severe concrete crush
was observed for specimen C2.5-1 with small α (= 0.065), as shown
in Fig. 5(b), while the minor concrete crush was found for specimen
2.2. Test setup and measurements
C6-1 with large α (=0.173), as shown in Fig. 5(c). The failure modes
of hexagonal CFST columns are similar to those of rectangular CFST col-
2.2.1. Compression test
umns. The failure mode of hollow steel tubular specimen (CH4-1) is
The compression test was conducted on a testing machine with a
shown in Fig. 5(d), where both outward and inward buckling were
loading capacity of 10,000 kN. The schematic view of the test setup
observed.
and instrumentation is depicted in Fig. 4(a). Four displacement trans-
ducers were placed on the top and bottom loading plates to measure
2.3.1.2. Load versus deformation performance. The load (N) versus axial
the longitudinal shortening. Four strain gauges were placed longitudi-
shortening (Δ/H) relations of all stub columns are depicted and com-
nally to measure the longitudinal compressive strain in the middle
pared in Fig. 6. The maximum load is defined as the ultimate compres-
cross section. Another four strain gauges were placed transversely to
sive strength (Nue). Fig. 6(a) shows the effect of steel ratio (α). Both
measure the transverse strain.
the stiffness and strength increased with the increase of α. After the ul-
timate compressive strength was reached, the descending slop was
2.2.2. Bending test steeper when α decreased. Fig. 6(b) shows the effect of core concrete,
The bending test was a two points loading one, where the middle where the stiffness and strength of the hexagonal CFST columns were
segment was in pure bending and two side segments were subjected much higher than those of the hollow steel tube counterparts. After
to combined bending and shear. The schematic view of the test setup the ultimate compressive strength, the load decrease was smooth for
and instrumentation is shown in Fig. 4(b). Two displacement transduc- the hollow steel tubes, for there was no constraint for the deformation
ers were placed at two ends to measure the supports' offsets. Three of the steel tube.
displacement transducers were placed at two loading points and the
mid-span to measure the deflection. Seven strain gauges were placed 2.3.1.3. Strain analysis. The load (N) versus longitudinal steel strain (εl)
longitudinally in the mid-span. Another three strain gauges were placed relations are shown and compared in Fig. 7, where εl is the averaged
transversely. value of longitudinal steel strains ε1l, ε2l, ε3l and ε4l in Fig. 4(a) and the
compressive strain is positive. The steel strain results reflect the local-
2.3. Test results ized strain distribution. The initial stiffness of the N-εl relation was
close to that of N-Δ/H curve, for the strain distribution was uniform
2.3.1. Compression test along the height at the beginning. Table 1 shows the ultimate strains
(εue) corresponding to the ultimate compressive strengths, which indi-
2.3.1.1. Failure modes. All the stub columns experienced a smooth load- cated the yielding of steel when the specimen reached the ultimate
ing process. Typical failure modes are shown in Fig. 5. The outward strength. The points when local buckling occurred are marked in the
Fig. 4. Test setup and instrumentation: compression test and bending test (Unit: mm).
Fig. 6. Load (N) versus axial shortening (Δ/H) relations of column specimens.
Fig. 7. Load (N) versus longitudinal steel strain (εl) relations of column specimens.
W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175 167
maximum flexural strength (Mmax) are shown in Table 2, where Mmax is of concrete, and it then moved upward slowly. The plastic strain devel-
the maximum moment recorded during the test. The results show that oped quickly when the moment increased from 0.9Mue to Mue, and the
Mue were smaller than Mmax with a mean Mue/Mmax ratio of 0.839. plane cross-section assumption was still available at this moment.
2.3.2.3. Strain analysis. Fig. 10(a) shows the moment (M) versus longitu- 3. Analytical behavior of the hexagonal CFST members
dinal steel strain (εl) relations of specimen B4-1. Due to the presence of
concrete, the steel strain in the tensile side developed faster than that in 3.1. Finite element analysis (FEA) model
the compressive side. At the ultimate point (M = Mue), majority of the
steel tube had yielded. When the load reached Mmax, the magnitude of 3.1.1. General description of the FEA model
ε1l, ε5l and ε7l all exceeded 20,000 με and ε3l was − 7722 με, which The schematic view of FEA model established in ABAQUS is shown in
showed excellent deformation capacity. Fig. 11. Solid elements (C3D8R) are applied for the core concrete, rigid
The longitudinal steel strain (εl) distributions along height at differ- plate and loading block, while shell elements (S4R) are applied for the
ent load levels are shown in Fig. 10(b), where the strain distribution steel tube. Details of the interaction and the material models can be
obeyed the plane cross-section assumption before 0.9Mue. The neutral found in Han et al. [21]. In order to achieve a better convergence, the dis-
axis was above the mid-height of the cross section due to the existence placement controlled load strategy is used in the model.
168 W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175
Fig. 10. Longitudinal steel strain development and distribution of typical beam specimen (B4–1).
Fig. 11. Schematic view of the finite element analysis (FEA) model.
W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175 169
0.496, 0.467 and 0.482, respectively, which indicates similar section the increase ratios for the ultimate compressive and flexural strength
combination of steel tube and core concrete. The load (N) is normalized are only 2.6% and 2.9%, respectively. In addition, the cold forming pro-
by the ultimate compressive strength (Nue) to exclude the influence of cess also weakens the fractural ductility of steel, which may accelerate
section size and material strength. As can be seen, the descending part the fracture of steel tube in beam specimens. As observed in the test,
of the curve of hexagonal specimen (C2.5-1) is closer to that of square specimens B2.5 and B4 with thinner tube thickness experienced frac-
specimen (sssc-1), which is consistent with the numerical results. ture of the steel tube. However, the fracture occurred when the mid-
Fig. 13 shows the longitudinal stress (σl, on the left of each figure) span displacement was about L/15, which was much higher than the
and contact stress (p, on the right of each figure) of the core concrete mid-span displacement at the ultimate point (L/193-L/92). The effect
for different cross sections when the ultimate compressive strength is of fracture on the flexural performance of beam specimens is minor.
reached. For the hexagonal and rectangular sections, σl ranges from Therefore, it is reasonable to neglect the cold forming effect in the FEA
0.95 fc′ to 1.05 fc′ as shown in Fig. 13(a) and (b), respectively. The stress model.
concentration is observed in corners of the section and the area is limit-
ed. For the circular section, σl ranges from 1.03 fc′ to 1.12 fc′ on the cross 3.1.4. Verification of the FEA model
section, as seen in Fig. 13(c). In terms of the contact stress (p), it appears Fig. 5(a) and (d) compares the observed and predicted failure modes
only in the corners of the hexagonal and rectangular sections, while it of column specimens. The FEA model gives similar results of the local
distributes uniformly along the circular cross section. buckling of steel tube in both hexagonal CFST and hollow tubular spec-
The load versus deformation relation, longitudinal stress and contact imens. The predicted N-Δ/H relations of column specimens are com-
stress distribution of the hexagonal CFST is similar to those of the rect- pared with the measured ones in Fig. 6, where the FEA model achieves
angular one, therefore the σ-ε relation of confined concrete in rectangu- reasonable estimation for both hexagonal CFST and hollow tubular
lar CFST suggested by Han et al. [21] is tentatively selected for the core specimens. The numerical ultimate compressive strengths (NuFEA) of
concrete in the hexagonal CFST in the study. all column specimens are summarized and compared with the tested
ones (Nue) in Table 1. The mean value and COV of NuFEA/Nue are 1.012
3.1.3. Effect of cold forming and 0.064, respectively.
As explained in Section 2.1.3, the hexagonal steel tubes are fabricat- The predicted failure modes of the steel tube and core concrete in
ed by cold forming process. The measured inside bending radiuses for beam specimen B4–1 are shown in Fig. 8(e) and (f), respectively. The
the corners of 90° and 135° are 4 mm and 10 mm, respectively. The outward buckling of steel tube occurs and significant concrete compres-
effect of cold forming should be discussed. The column and beam sive and tensile plastic strains are observed in the FEA model, which is
models with and without the effect of cold forming are established as coincident with the test results. In the FEA model, the mid-span curva-
shown in Fig. 14(a). The inside bend radiuses are the same to the mea- ture (ϕm) is derived by (εt–εc)/(2.414B), where εt and εc are the longitu-
sured results. The yield strength increase of the corner steel for the cold dinal strain of the bottom and top fiber of the hexagonal tube,
formed tube is considered according to AISI-S100-2007 [26]. It can be respectively; 2.414B is the gross height of the hexagonal section. The
seen from Fig. 14(b) and (c) that both compressive and flexural predicted M-ϕm relations are also compared with the tested ones in
strengths increase when considering the cold forming effect. However, Fig. 9. The numerical ultimate flexural strengths (MuFEA) and the tested
Fig. 13. Longitudinal stress and contact stress distribution of core concrete of different sections (Unit: N/mm2).
170 W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175
Fig. 14. Effect of cold forming on compressive and flexural behavior of hexagonal CFST.
ones (Mue) are listed in Table 2, and the mean value and COV of MuFEA/ the residual strength (Nr). For this particular sample, Nr equals to
Mue are 1.076 and 0.093, respectively. 0.637NuFEA.
The comparisons show that the σ-ε curve for confined concrete in The stress (σ) versus axial shortening (Δ/H) relations of steel tube
rectangular CFST members can be used for the modeling of core con- are shown in Fig. 16, where the longitudinal stress (σl), transverse stress
crete in hexagonal CFST members. The proposed FEA model provides (σt) and the Mises stress (σMises) are all depicted. In addition, stresses at
reasonable results on both the compressive and flexural behavior of two locations (1 and 2) are shown in straight and dashed lines, respec-
hexagonal CFST members. The mechanical analysis is then conducted tively. At the location 1, the steel yields at point A, with σl and σt of
using the proposed FEA model. 333.8 N/mm2 and −5.7 N/mm2, respectively. After that, σt increases be-
tween Point A and Point C, and σl decreases due to the Von-Mises yield
3.2. Analytical behavior criterion. After point C, σt tends to be steady, while σl increases slightly
due to the hardening of steel. The development of Ns in Fig. 15 can be ex-
3.2.1. Compressive behavior of hexagonal CFST plained by this stress development. It can also be seen that, the stress
A numerical sample is established in this part to study the compres- development at locations 1 and 2 are almost the same, while σt at loca-
sive behavior of hexagonal CFST stub column. The properties of the tion 1 is slightly larger than that at location 2 after point C due to the
column are as follows: H = 600 mm, B = 100 mm, t = 4 mm, fy = hexagonal shape.
345 N/mm2, fck = 41 N/mm2, α = 0.110, ξ = 0.929. The predicted The longitudinal stress (σl) and the contact stress (p) distribution of
load (N) versus axial shortening (Δ/H) relation of this sample is the core concrete at typical points are further compared in Fig. 17. At
shown in Fig.15, where the load resisted by the steel tube (Ns) and the points A and B, the central part has a uniform σl distribution with mod-
core concrete (Nc) are also depicted. Several feature points are marked erate stress concentration in the corners. At point C, the region between
and the curve is divided into four stages, i.e. OA, AB, BC and CD. the corners and central part experiences a decrease of stress due to the
The axial shortening reaches the yield strain of steel tube at point A, concrete crush. However, the stress in the corners still increases with a
while Ns and Nc increase linearly during this stage. At point B, Nc reaches maximum σl of 1.98 fc′ while that in the central part increases to about
its peak load and Ns remains nearly unchanged due to yielding, thus the 1.20 fc′. The confinement in the corners and the central part are
load (N) reaches its ultimate compressive strength (NuFEA). After point stronger.
B, the curve enters the descending stage when Nc declines due to the The contact stress (p) exists only near the corners during the whole
crush of core concrete and Ns starts to decrease. The steel tube is process, as shown in Fig. 17. The maximum contact stress (pmax)
subjected to transverse tension and longitudinal compression at this corresponding to points A, B, C and D are 2.6 N/mm2, 17.3 N/mm2,
stage. Based on the Von-Mises yield criterion, the longitudinal compres- 48.1 N/mm2 and 35.5 N/mm2, respectively. The confinement is weak
sive stress decreases for the transverse tension, leading to the decrease before the steel tube yields at point A, when the concrete is also in the
of Ns. At point C, the load decreases to 0.85NuFEA. After that, Nc continues elastic state. After that, p starts to increase due to the transverse expan-
to decrease while Ns remains steady. The point corresponding to an sion of the core concrete between point A and point C. For instance, pmax
axial shortening of 0.01 is defined as point D and the load is defined as at point C is 178% larger than that at point B. The concrete totally fails
after point C, causing the decrease of contact stress.
Fig. 16. Stress (σ) versus axial shortening (Δ/H) relations of steel tube in hexagonal CFST column.
Fig. 17. Longitudinal stress and contact stress distribution of core concrete at typical points (Unit: N/mm2).
curve is almost linear before the bottom steel fiber yields at point B. C) due to the existence of core concrete. At points D and E, the deforma-
After that, an elasto-plastic stage is shown with a decreasing slope. tion continues to develop while the strength development is not signif-
The bottom steel fiber yields (point B) before the top steel fiber (point icant. The moment at point E (63.8 kN m) is only 5.4% larger than that at
point D (60.5 kN m). At point F, the deformation develops quickly with
the load nearly unchanged, indicating a ductile flexural behavior of the
beam.
The longitudinal stress (σl) distribution of the steel tube and core
concrete along height at typical points are depicted in Fig. 19(a) and
(b), respectively. Fig. 19(a) shows that the distribution of σl is linear
before the bottom steel fiber yields at point B. At the ultimate point E,
most of the tube has yielded.
Fig. 19(b) shows the longitudinal stress (σl) distribution of concrete,
where a parabolic distribution is seen at the beginning of loading. When
the maximum compressive strain reaches − 0.003 at point D, the
maximum σl reaches −1.06 fc′. σl increases with a maximum stress of
− 1.25 fc′ at point E. After that, at point F, σl near the edge of the
compressive zone decreases due to the crush of concrete, which is
coincident with the experimental observation.
The moment (M) versus mid-span curvature (ϕm) curves of the hex-
agonal CFST beam, the corresponding hollow steel tube beam and con-
Fig. 18. Typical moment (M) versus mid-span curvature (ϕm) relation of hexagonal CFST crete beam are compared in Fig. 20. In addition, the contributions of the
beam. concrete and the steel tube components are extracted from the
172 W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175
Fig. 19. Longitudinal stress distribution in hexagonal CFST beam at typical points.
hexagonal CFST beam and shown in the same figure using dashed lines. 4. Parametric study and simplified model
The internal moment distribution shows that the steel tube provides
most of the bending resistance with excellent ductility. The hollow 4.1. Compressive behavior of hexagonal CFST
steel tube beam and the steel tube component in the hexagonal CFST
behave almost the same when ϕm is smaller than 0.2 m−1. However, 4.1.1. Parametric study
when ϕm is larger than 0.2 m−1, the strength of the hollow steel tube The parameters affecting the compressive behavior of the hexagonal
beam decreases due to local buckling, while that of the steel tube com- CFST column includes the steel ratio (α), steel yield strength (fy) and
ponent remains stable for the support of core concrete. The sectional de- concrete strength (fck). The effects of parameters on the load (N)
formation of the hexagonal CFST beam and corresponding hollow steel versus axial shortening (Δ/H) relations are shown in Fig. 21, where
tube beam at a ϕm of 0.5 m−1 are shown in the figure. For the existence the detailed parameters are shown for clarity and the effect of con-
f
of core concrete, the lateral expansion of the hexagonal CFST beam is finement factor (ξ ¼ α f y ) is also discussed. The height (H) and sec-
ck
0.96 mm, which is much smaller than that of the hollow steel tube tion size (B) of all the specimens are 600 mm and 100 mm,
beam (14.20 mm). For the concrete beam, the maximum lateral expan- respectively. The parameter ranges are: α = 0.055 – 0.218, f y =
sion recorded is only 0.09 mm. The core concrete component inside the 185 – 735 N/mm2, fck = 20 – 88 N/mm2, ξ = 0.5 - 2.0.
hexagonal CFST beam carries much more moment than the correspond- Fig. 21(a) shows the effect of steel ratio (α), where α is changed by
ing concrete beam, for the concrete beam may easily be fractured. varying the tube thickness (t). It can be seen that the stiffness and
Therefore, the flexural strength of hexagonal CFST is higher than the strength increases with the increase of α. The ductility of the curve is
sum of those of the hollow steel tube beam and concrete beam. also improved with a larger α. The effect of steel yield strength (fy) is
It should be noted that, when ϕm exceeds 0.1 m−1, the flexural shown in Fig. 21(b), where the stiffness keeps the same while the
strength of the hexagonal CFST beam decreases slightly due to the strength increases significantly. Fig. 21(c) shows the stiffness and
crush of concrete, as indicated in the stress distribution at point F in strength increases as fck increases. However, the ductility decreases
Fig. 19(b). The phenomenon is slightly different from that of the circular with a larger fck. Further analysis shows that the confinement factor ξ
CFST beam, where the flexural strength keeps increasing slightly. As actually influences the ductility development. Better ductility for the
mentioned in Section 3.1.2, when compared with the hexagonal CFST stub column specimen is achieved when ξ is larger.
column, the circular CFST column provides stronger confinement to
the core concrete. The confinement from the steel tube to the core con- 4.1.2. Discussion on the ultimate compressive strength
crete in the compressive part of the circular CFST beam is also stronger As denoted previously, the confinement effect and the stress
than that from the hexagonal one. Therefore the ductility of the circular distribution of the hexagonal CFST column is similar to those of the
CFST beam is better. rectangular one. Therefore the equations for calculating the ultimate
Fig. 20. Comparison of M-ϕm relations between hexagonal CFST beam, hollow steel tube beam and concrete beam.
W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175 173
Fig. 21. Effect of parameters on the N-Δ/H relation of hexagonal CFST column.
Fig. 22. Comparison between the calculated and numerical ultimate compressive strength.
compressive strength (Nuc) of rectangular CFST columns are utilized 4.2. Flexural behavior of hexagonal CFST
tentatively for that of the hexagonal CFST columns. Methods from
three design specifications including the Chinese code DBJ/T13-51- 4.2.1. Parametric study
2010 [17], the Japanese code AIJ-2008 [18] and the European code EC4 The parametric study is conducted on the flexural behavior of hexag-
[14] are adopted in the paper, where the confinement effect is consid- onal CFST beams based on the standard beam specimen in Section 3.2.2.
ered in Chinese and European codes. The effective length (L) and the side length (B) of all specimens are
Table 1 shows the measured (Nue) and code-calculated strength 1800 mm and 100 mm, respectively. The parameter ranges are: α =
for test specimens, where Nuc1, Nuc2 and Nuc3 are predicted using DBJ/ 0.043 – 0.331, fy = 235 – 420 N/mm2, fck = 20 – 72 N/mm2. The effects
T13-51-2010 [17], AIJ-2008 [18] and EC4 [14], respectively. The mean of the steel ratio (α), steel yield strength (fy) and the concrete strength
value of Nuc1/Nue, Nuc2/Nue and Nuc3/Nue are 1.036, 0.925 and 1.020, re- (fck) on the moment (M) versus mid-span curvature (ϕm) curves are
spectively. In addition, the numerical results (NuFEA) of samples in the shown in Fig. 23(a), (b) and (c), respectively.
parametric study are also compared with the code-calculated results Fig. 23(a) reflects the effect of steel ratio (α). As α increases, the stiff-
(Nuc) in Fig. 22, where the mean value and COV of Nuc/NuFEA are also ness and strength of the curve increases significantly. The effect of fy is
shown. All the three methods agree well with the numerical results shown in Fig 23(b). As fy increases, the strength increases while the stiff-
with a mean Nuc/NuFEA of 1.007, 0.918 and 0.996, respectively. The com- ness keeps the same. The effect of fck is shown in Fig. 23(c). The stiffness
parisons show that the methods in DBJ/T13-51-2010 [17] and EC4 [14] and strength increases moderately when fck increases. In general, the ef-
which consider the confinement effect give better predictions, while the fect of α and fy of steel tube on the flexural behavior of hexagonal CFST
predicted results by AIJ-2008 [18] are conservative. beam is significant while the effect of fck of core concrete is moderate.
Fig. 23. Effect of parameters on the M-ϕm relation of hexagonal CFST beam.
174 W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175
Fig. 25. Comparison between calculated and numerical ultimate flexural strength.
5. Conclusions
Fig. 24. Methods for predicting ultimate flexural strength of hexagonal CFST beam.
The following conclusions can be drawn based on the limited study
reported in this paper:
4.2.2. Discussion on the ultimate flexural strength
Three methods in Fig. 24, i.e. the plastic stress distribution method (1) For the hexagonal CFST stub columns, outward local buckling of
(PSDM), strain compatibility method (SCM) and the fiber model meth- steel tube was occurred when columns failed owing to the sup-
od (FMM) are used to evaluate the ultimate flexural strength of hexag- port of concrete, while inward and outward local buckling was
onal CFST beam. The PSDM is adopted by EC4 [14] for the ultimate observed for the hollow tubular ones. The compressive strength
flexural strength of composite members. The SCM is suggested by and ductility was improved for the concrete-filled specimens.
ACI318-14 [15] for reinforced concrete members, where the extreme The local buckling of steel tube was also improved owing to the
compressive strain of concrete is assumed to be −0.003. The measured increase of steel ratio.
maximum compressive strain of steel (εcue) at the ultimate point is (2) For the hexagonal CFST beams, the outward local buckling of
summarized in Table 2, where the strains were around − 0.003. The steel tube and the crushing of concrete in compressive zone
FMM is introduced by researchers such as Han [20] to predict the was observed when specimens failed. The tube fracture occurred
flexural behavior of CFSTs, where the extreme tensile strain of steel more easily for the beams with smaller steel ratio. The crushing
tube is assumed to be 0.01. Generally, due to the assumption that all of concrete and the buckling of tube was improved when the
sections enter the plastic state, the predicted results of PSDM are the steel ratio increased.
highest. The SCM and FMM both depend on the plane cross-section (3) The FEA model was established and verified for both column and
assumption, which is verified by the strain results in the paper. Howev- beam specimens. The constitutive model of confined concrete in
er, the definitions of the ultimate state in these two methods are rectangular CFST could be used for the core concrete in hexago-
different. nal CFST. The proposed model provided reasonable predictions
The tested ultimate flexural strength (Mue) of tested specimens and of the failure modes and the load versus deformation curves of
the calculated results (Muc) are shown in Table 2. The mean values of both stub column and beam specimens.
Muc/Mue for PSDM, SCM and FMM are 1.056, 1.030 and 1.026, respec- (4) For stub column specimens, the increase of steel ratio and steel
tively. Further comparison is conducted between the numerical results yield strength improved both the compressive strength and
(MuFEA) by FEA model and the ones by simplified methods (Muc) in ductility, while the increase of concrete strength improved the
Fig. 25. The results are sorted by the confinement ratio (ξ). For hexago- compressive strength but weakened the ductility. For beam
nal CFST beams with small confinement ratio (ξ b 0.5), these three specimens, the effects of steel ratio and steel yield strength
methods tend to provide conservative results. Due to the small confine- were significant on the flexural strength while that of concrete
ment ratio, the stress of core concrete will not reach fc′ at the ultimate strength was moderate.
W. Xu et al. / Journal of Constructional Steel Research 123 (2016) 162–175 175
(5) The feasibility of simplified models in current codes on the [9] X.L. Zhao, J.A. Packer, Tests and design of concrete-filled elliptical hollow section
stub columns, Thin-Walled Struct. 47 (6–7) (2009) 617–628.
predictions of ultimate compressive and flexural strengths for [10] N. Jamaluddin, D. Lam, X.H. Dai, J. Ye, An experimental study on elliptical concrete
hexagonal CFST members was discussed. It was found that the filled columns under axial compression, J. Constr. Steel Res. 87 (2013) 6–16.
ultimate compressive strength of hexagonal CFST column could [11] H.X. Dai, D. Lam, N. Jamaluddin, J. Ye, Numerical analysis of slender elliptical con-
crete filled columns under axial compression, Thin-Walled Struct. 77 (2014) 26–35.
be estimated accurately referring to the equations for the rectan- [12] B. Evirgen, A. Tuncan, K. Taskin, Structural behavior of concrete filled steel tubular
gular CFST members in DBJ/T13-51-2010 and EC4. The plastic sections (CFT/CFSt) under axial compression, Thin-Walled Struct. 80 (2014) 46–56.
stress distribution method and the fiber model method could [13] Q.X. Ren, L.H. Han, D. Lam, C. Hou, Experiments on special-shaped CFST stub col-
umns under axial compression, J. Constr. Steel Res. 98 (2014) 123–133.
provide reasonable predictions for the ultimate flexural strength
[14] Eurocode 4, Design of composite steel and concrete structures-Part 1-1: General
of hexagonal CFST beam. rules and rules for buildings, Brussels: European Committee for Standardization,
2004 [EN 1994-1-1:2004].
[15] ACI318-14, Building code requirements for structural concrete and commentary,
Acknowledgements
Detroit, USA: American Concrete Institute, 2014.
[16] ANSI/AISC360-10, Specification for structural steel buildings, Chicago, USA:
The research reported in the paper is part of the Tsinghua University American Institute of Steel Construction, 2010.
Initiative Scientific Research Program (No. 2013Z02). The financial sup- [17] DBJ/T13-51-2010, Technical specification for concrete-filled steel tubular structures,
Fuzhou, China: The Construction Department of Fujian Province, 2010 [in Chinese].
port is highly appreciated. [18] AIJ-2008, Recommendations for design and construction of concrete filled steel tu-
bular structures, In: Tokyo, Japan: Architectural Institute of Japan, 2008.
References [19] M. Elchalakani, X.L. Zhao, R.H. Grzebieta, Concrete-filled circular steel tubes subject-
ed to pure bending, J. Constr. Steel Res. 57 (11) (2001) 1141–1168.
[1] M. Shams, M.A. Saadeghvaziri, State of the art of concrete-filled steel tubular col- [20] L.H. Han, Flexural behaviour of concrete-filled steel tubes, J. Constr. Steel Res. 60 (2)
umns, ACI Struct. J. 94 (5) (1997) 558–571. (2004) 313–337.
[2] N.E. Shanmugam, B. Lakshmi, State of the art report on steel-concrete composite [21] L.H. Han, G.H. Yao, Z. Tao, Performance of concrete-filled thin-walled steel tubes
columns, J. Constr. Steel Res. 57 (10) (2001) 1041–1080. under pure torsion, Thin-Walled Struct. 45 (1) (2007) 24–36.
[3] L.H. Han, W. Li, R. Bjorhovde, Developments and advanced applications of concrete- [22] J.B. Mander, M.J.N. Priestley, R. Park, Theoretical stress-strain model for confined
filled steel tubular (CFST) structures: members, J. Constr. Steel Res. 100 (2014) concrete, ASCE J. Struct. Eng. 114 (8) (1988) 1804–1826.
211–228. [23] M.M. Attard, S. Setunge, Stress-strain relationship of confined and unconfined con-
[4] Y.L. Yang, H. Yang, S.M. Zhang, Compressive behavior of T-shaped concrete filled crete, ACI Mater. J. 93 (5) (1996) 432–442.
steel tubular columns, Int. J. Steel Struct. 10 (4) (2010) 419–430. [24] E. Ellobody, B. Young, Numerical simulation of concrete encased steel composite col-
[5] Y.Q. Tu, Y.F. Shen, Y.G. Zeng, L.Y. Ma, Hysteretic behavior of multi-cell T-shaped umns, J. Constr. Steel Res. 67 (2) (2011) 211–222.
concrete-filled steel tubular columns, Thin-Walled Struct. 85 (2014) 106–116. [25] L.H. Han, G.H. Yao, Experimental behaviour of thin-walled hollow structural steel
[6] Z.L. Zuo, J. Cai, C. Yang, Q.J. Chen, G. Sun, Axial load behavior of L-shaped CFT stub (HSS) columns filled with self-consolidating concrete (SCC), Thin-Walled Struct.
columns with binding bars, Eng. Struct. 37 (2012) 88–98. 42 (9) (2004) 1357–1377.
[7] T. Zhou, Z.H. Chen, H.B. Liu, Seismic behavior of special shaped column composed of [26] AISI-S100-2007, North American specification for the design of cold-formed
concrete filled steel tubes, J. Constr. Steel Res. 75 (2012) 131–141. steel structural members, Washington, D.C., USA: American Iron and Steel Insti-
[8] H. Yang, D. Lam, L. Gardner, Testing and analysis of concrete-filled elliptical hollow tute, 2007.
sections, Eng. Struct. 30 (12) (2008) 3771–3781.