Вы находитесь на странице: 1из 11

Journal of Constructional Steel Research 145 (2018) 341–351

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Global buckling of laterally-unrestrained Q460GJ beams with singly


symmetric I-sections
Shao-Bo Kang a,b, Bo Yang a,b, Yue Zhang a,b, Mohamed Elchalakani c, Gang Xiong a,b,⁎
a
Key Laboratory of New Technology for Construction of Cities in Mountain Area, Chongqing University, Ministry of Education, Chongqing 400045, China
b
School of Civil Engineering, Chongqing University, Chongqing 400045, China
c
School of Civil, Environmental & Mining Engineering, University of Western Australia, Australia

a r t i c l e i n f o a b s t r a c t

Article history: This paper investigates the global stability of welded singly symmetric I-shaped beams fabricated from Q460GJ
Received 31 July 2017 steel, a typical high-performance structural steel. Experimental tests and numerical simulations were carried
Received in revised form 27 December 2017 out on a total of eight laterally unrestrained steel beams under a concentrated point load. A special loading sys-
Accepted 4 March 2018
tem was designed through which the beam could develop lateral deflections at the mid-span. Load-displacement
Available online xxxx
curves and strains of beams were measured during testing. In the finite element model, initial geometric imper-
Keywords:
fections and residual stresses were considered. The model was verified through comparisons with experimental
Q460GJ structural steel results, and parametric studies were conducted to investigate the effects of non-dimensional slenderness and
Singly symmetric I-section height-to-width ratio. Meanwhile, design curves for lateral-torsional buckling of welded singly symmetric
Lateral-torsional buckling beams were calculated from GB50017-2003, GB50017-201X, Eurocode 3 and ANSI/AISC360-10. Comparisons
Experimental test between numerical results and design curves suggested that the design methods in GB50017-2003 and ANSI/
Numerical simulation AISC360-10 were unsafe for Q460GJ steel beams while GB50017-201X and Eurocode 3 could provide reasonably
conservative results for global buckling design of singly symmetric Q460GJ steel beams.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction proposed based on the test data of conventional mild steel and revisions
are needed when used for high-performance steel members.
GJ steel, a typical high-performance steel, has been widely used in In practice, singly-symmetric I-sections with thicker compressive
many construction projects in China as a result of its higher strength, flange are often used to avoid local buckling. In the past few years, a
better weldability and ductility as compared to conventional structural large number of theoretical investigations were conducted on lateral-
steel. GJ steel has lower contents of sulphur, phosphorus and carbon. torsional buckling of single symmetric I-beams. El-Mahdy and El-
For instance, the sulphur content must be less than 0.015%, and the Saadawy [3] numerically studied the effect of flange ratio on lateral-
phosphorus content less than 0.020–0.025%. Moreover, the yield-to- torsional buckling resistances and recommended an approximate
ultimate strength ratio of Q460GJ steel is not more than 0.85. In terms formula for the ultimate load of beams. Surla et al. [4] conducted numer-
of mechanical properties, the effect of plate thickness on the yield ical simulations to investigate the load-carrying capacity of singly
strength of GJ steel is not as significant as conventional steel. The lamel- symmetric stepped I-beams under uniform bending. They suggested a
lar tearing resistance of GJ steel heavy plates is also superior. To assess correction factor for stepped beams to account for the increase in
its influence on global buckling behaviour of columns, experimental the load capacity of beams with steps at both ends. Andrade et al. [5]
and numerical investigations have been carried out on Q460GJ columns investigated the elastic lateral-torsional buckling behaviour of singly
subject to compression [1]. As for steel beams, doubly symmetric symmetric tapered thin-walled open beams. Camotim et al. [6]
Q345GJ I-sections with lateral restraints have been studied under a concluded that a uniform bending moment might not be the most
concentrated point load [2]. Comparisons were made between design critical loading condition for I-section beams with unequal flanges
curves and test results to evaluate the accuracy of different national (mono-symmetric sections). Singly symmetric Q460GJ I-beams with a
codes in predicting the global buckling resistance of GJ steel columns lateral restraint at the mid-span were tested by Yang et al. [7]. It was
and beams. It was concluded that design methods are primarily reported that due to the presence of lateral restraint, initial geometric
imperfections did not significantly affect the global buckling resistance
of the beams. Nonetheless, few experimental results are available
⁎ Corresponding author at: School of Civil Engineering, Chongqing University,
when the lateral restraint is absent. Current design codes only adopt a
Chongqing 400045, China. reduction factor for the design of singly symmetric I-beams, as the
E-mail address: 195704148@qq.com (G. Xiong). non-symmetric theory has not been fully resolved [8]. Therefore, it is

https://doi.org/10.1016/j.jcsr.2018.03.005
0143-974X/© 2017 Elsevier Ltd. All rights reserved.
342 S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351

Table 1
Nominal dimensions of steel beams.

Steel beam Height of section Thickness of web Width of top flange Thickness of top flange Width of bottom flange Thickness of bottom flange Beam span

H(mm) tw(mm) b1(mm) t1(mm) b2(mm) t2(mm) L(mm)

B-270-3.0 270 8 180 16 180 8 3000


B-270-3.5 270 8 180 16 180 8 3500
B-270-4.0 270 8 180 16 180 8 4000
B-270-4.5 270 8 180 16 180 8 4500
B-360-3.0 360 8 180 16 180 8 3000
B-360-3.5 360 8 180 16 180 8 3500
B-360-4.0 360 8 180 16 180 8 4000
B-360-4.5 360 8 180 16 180 8 4500

necessary to verify the design methods for singly symmetric I-beams were selected for steel beams. As for specimens B-270-3.0 to B-270-4.5,
through experimental tests, in particular for high-performance steel the height and width of steel beams were 270 and 180 mm, respectively,
beams. and the corresponding height-to-width ratio was 1.5. In specimens B-
In this paper, a series of experimental tests was conducted to inves- 360-3.0 to B-360-4.5, the beam height was increased to 360 mm,
tigate the global buckling behaviour of singly symmetric I-shaped and associated height-to-width ratio became 2.0. The nominal thickness
beams fabricated from Q460GJ steel. Eight specimens with two types of beam top flanges was kept at 16 mm and the web thickness was
of cross section and varied spans were tested in total. The beams were 8 mm so that the width-to-thickness ratio of beam flanges and the
tested under a concentrated point load at the mid-span and a unique height-to-thickness ratio of beam webs were smaller than the values
test rig was designed to minimise the friction in the loading device. incorporated in GB50017-2003 [9] and local buckling could be prevented.
Finite element models were developed for steel beams, in which initial The nominal thickness of tension flanges was taken as 8 mm. Meanwhile,
geometric imperfections and residual stresses were incorporated. all beam sections were classified as those with compact flanges and webs
The model was validated against test results and further used for according to ANSI/AISC360-10 [12] and class 1 per Eurocode 3 [11], indi-
parametric studies on a wide range of height-to-width ratio and non- cating that local buckling would not occur before the specimens attained
dimensional slenderness. Finally, numerical results are compared with the plastic moment capacities. Transverse stiffeners were designed at the
GB50017-2003 [9], GB50017-201X [10], Eurocode 3 [11] and ANSI/ loading point and the supports to prevent local yielding of webs subject
AISC360-10 [12], and recommendations were provided for global buck- to concentrated point forces.
ling design of singly symmetric Q460GJ steel beams without lateral
restraints. 2.2. Test setup

2. Experimental Programme on global buckling of I-beams Fig. 1(a) shows the test setup for steel beams. At the mid-span, a
vertical load was applied through a special loading system comprised
2.1. Specimen design of a rigid rectangular loading frame, a hydraulic jack, a steel triangle
connected to two steel brackets (see Fig. 1(b)). During testing, the
In the experimental programme, a total of eight beams with singly applied load was measured by a load cell. Note that a special hemispher-
symmetric I-shaped cross-sections were designed and fabricated from ical hinge with a series of roller bearings was placed between the beam
Q460GJ structural steel with a nominal yield stress of 460 MPa. and the load cell so that the load could remain vertical after lateral-
Tables 1 and 2 summarise the nominal and the measured dimensions torsional buckling of the beam occurred and friction could be
of the specimens, respectively. The initial geometric imperfections minimised. Simple supports were designed at both ends of the steel
were measured prior to testing, as listed in Table 2. It is noteworthy beam, as shown in Fig. 1(c). Cylindrical rollers were placed beneath
that specimen B-360-4.0 was only used to validate the efficiency of the steel beam to ensure in-plane rotation at the end supports, whereas
the test setup, of which the initial deformations were not measured. longitudinal displacements of the beam were released. Lateral restraints
In accordance with Chinese design codes for steel structures [9,10] with steel rollers were provided at the top and bottom flanges to pre-
and Eurocode 3 [11], different buckling curves are defined based on vent out-of-plane rotation of the beam at the end support. The loading
the height-to-width ratio of cross sections. To investigate the effects of system was adopted for all steel beams except B-360-4.0. As for B-
height-to-width ratio on global stability of steel beams, two sets of 360-4.0, a conventional hemispherical hinge without roller bearings
height-to-width ratio were adopted in the experimental programme was used between the steel beam and the rectangular loading frame.
and cross sections H270 × 180 × 8 × 16(8) and H360 × 180 × 8 × 16(8) Due to the presence of significant friction in the hinge, the steel triangle

Table 2
Measured dimensions of steel beams.

Steel beam Height of Thickness of Width of Thickness of Width of Thickness of Initial geometric
section web top flange top flange bottom flange bottom flange imperfection

H(mm) tw(mm) b1(mm) t1(mm) b2(mm) t2(mm) D (mm)

B-270-3.0 270.98 8.322 180.19 16.52 180.24 8.3 1.51


B-270-3.5 270.8 8.3 179.36 16.64 178.16 8.34 0.84
B-270-4.0 271.94 8.24 180.31 16.73 177.7 8.09 1.76
B-270-4.5 272.14 8.22 178.93 16.44 180.23 7.99 0.41
B-360-3.0 362.74 8.68 180.28 16.65 178.67 8.67 2.16
B-360-3.5 362.68 8.69 180.98 16.71 179.04 8.66 1.57
B-360-4.0 361.23 8.88 179.81 16.7 180.57 8.71 –
B-360-4.5 360.97 8.76 180.7 16.62 180.42 8.68 1.27
S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351 343

Loading system

End support
Steel beam

(a) Test setup for steel beams

Rectangular
steel frame

Steel roller

Hydraulic jack

Cylindrical roller
Steel bracket

(c) Support system at each end

Steel triangle

(b) Loading system at the mid-span

Fig. 1. Test setup for steel beams.

could not move downwards simultaneously with the beam, which was the top and bottom flanges, respectively. Five strain gauges were used
likely to enhance the load capacity of the beam. for the web of steel beams B-360-3.0 to B-360-4.5.

2.3. Instrumentation 2.4. Material properties

During testing, the out-of-plane rotation of beams at the mid-span Steel coupons with thicknesses of 8 and 16 mm were extracted from
was measured by using an inclinometer, attached on the web. The ver- steel plates and tension tests were conducted according to Chinese stan-
tical and horizontal displacements were captured through a series of dard GB/T228.1-2010 [13] prior to beam tests. For each steel plate, three
linear variable displacement transducers (LVDTs), as shown in Fig. 2. coupons were tested in tension, and the gauge length of steel coupons
In order to obtain the strain distribution, strain gauges were mounted was 150 mm. Fig. 4 shows the stress-strain curve of steel plates.
at the mid-span of steel beams. Fig. 3 shows the layout of strain gauges Table 3 summarises the average tensile properties of steel coupons.
on the web and flanges of the beams. For steel beams B-270-3.0 to B- The yield strengths of 8 and 16 mm thick steel coupons were 481 and
270-4.5 with a height of 270 mm, three strain gauges were attached to 541 MPa, respectively, satisfying the requirements on yield strength
one side of the beam web, and two and three strain gauges were on for this type of high-performance steel.
344 S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351

LT-1 LT-2 750


16 mm
8 mm
600

Stress (MPa)
450

LT-3 LT-4
300

150

LT-5 LT-6 0
0 5 10 15 20 25
Strain (%)
Fig. 2. Layout of linear variable displacement transducers.
Fig. 4. Stress-strain curves of steel.

3. Experimental results and discussions


Table 3
3.1. Behaviour of I-beams Mechanical properties of steel plates.

Thickness of Yield Ultimate Ultimate Tangent


When subject to three-point bending, all steel beams failed by steel plate strength stress strain modulus
lateral-torsional buckling. Fig. 5 shows the applied load-vertical dis- t (mm) fy (MPa) fu (MPa) εu (%) E (GPa)

placement curve of the beams. At the initial stage, the applied load of 16 481 627 11.1 203.9
B-270-3.0 increased almost linearly with mid-span displacement, as 8 541 669 9.7 204.7
shown in Fig. 5(a). After attaining the load capacity of 382.4 kN, the
beam developed lateral-torsional buckling, leading to a gradual reduc-
tion in the vertical load with increasing displacement. By increasing Fig. 7 shows the load-rotation curve of steel beams. The out-of-plane
the beam span from 3.0 m to 4.5 m, both the initial stiffness and the rotation was measured by an inclinometer placed on the top flange. It is
load capacity of steel beams of 270 mm in depth decreased substantially obvious that the rotation was rather limited at the initial stage, as shown
(see Fig. 5(a)). Compared to B-270-3.0, the load capacity reduced to in Fig. 7(a). Nonetheless, the rotation started increasing rapidly once the
208.5 kN by around 45.5%. Similar behaviour was observed when the vertical load exceeded half of the load capacity. At the descending stage
beam height was increased to 360 mm, as shown in Fig. 5(b). However, of vertical load, the out-of-plane rotation developed rapidly when the
it is worth noting that B-360-4.0 could sustain a greater load capacity beam span was relatively long. It should be pointed out that an almost
than B-360-3.5, even though the clear span of the former was signifi- linear load-rotation relationship was obtained for B-360-4.0 as a result
cantly longer than the latter. It resulted from the application of conven- of the use of conventional hemispherical hinge to apply vertical load,
tional hemispherical hinge to impose vertical load. The presence of and yet its load capacity was greater than that of B-360-3.5 due to the
friction between the two parts of the hinge led to the enhanced load presence of friction in the hinge.
capacity. Fig. 6 shows the failure mode of steel beams subject to a con- Fig. 8 shows the strain profile of B-360-4.5 at various stages. Note
centrated point load at the mid-span. Apparent lateral deflections that the maximum compressive strain represents the average value
were observed after lateral-torsional buckling of steel beams, in partic- measured by SG-1 and SG-2, and the peak tensile strain denotes the
ular in B-360-3.0. average value of SG-8 and SG-10 (see Fig. 3). When a vertical load of

SG-1 SG-2
SG-3
SG-1 SG-2
SG-3 SG-4

Neutral axis SG-5 Neutral axis SG-5

SG-6
SG-7
SG-10 SG-8
SG-7
SG-9 SG-10 SG-8
SG-9
(a) Beams B-270-3.0 to B-270-4.5
(b) Beams B-360-3.0 to B-360-4.5

Fig. 3. Arrangement of strain gauges on beam sections.


S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351 345

500 750
B-270-3.0 B-360-3.0
B-270-3.5 B-360-3.5
400 600
B-270-4.0 B-360-4.0

Vertical load (kN)

Vertical load (kN)


B-270-4.5 B-360-4.5
300 450

200 300

100 150

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Vertical displacement (mm) Vertical displacement (mm)

(a) Steel beams of 270 mm depth (b) Steel beams of 360 mm depth

Fig. 5. Vertical load-displacement curves of steel beams.

B-270-3.0 to B-360-4.5 (from left to right)

Fig. 6. Failure modes of steel beams.

70.9 kN was applied to the beam, corresponding strain profile remained compression flange. At the load capacity, the maximum compressive
linear, with zero at the neutral axis and maximum values at the extreme strain was around 2800 με, 1.56 times as much as the peak tensile strain
fibres of the top and bottom flanges. Before its load capacity of 262.4 kN at the bottom flange. Fig. 9 shows the relationship of vertical load and
was reached, nonlinear strains were measured at the compression measured strain. When the load was less than 30% of the ultimate
flange, whereas the strain was fairly linear elastic away from the load, a linear relationship was observed between the applied load and

500 750
B-270-3.0 B-360-3.0
B-270-3.5 B-360-3.5
400 600
B-270-4.0 B-360-4.0
Vertical load (kN)

Vertical load (kN)

B-270-4.5 B-360-4.5
300 450

200 300

100 150

0 0
0 3 6 9 12 15 0 2 4 6 8 10
Out-of-plane rotation (degree) Out-of-plane rotation (degree)

(a) Steel beams of 270 mm in depth (b) Steel beams of 360 mm in depth

Fig. 7. Vertical load-rotation curves of steel beams.


346 S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351

200 the load capacity of steel beam B-360-4.0 was significantly affected by
70.9 kN the conventional hemispherical hinge used to apply vertical load, it is
Distance from neutral axis (mm)

141.6 kN excluded from the comparison. It can be observed that GB50017-2003


100
212.7 kN [9] overestimates the load capacity of Q460GJ steel beams and the
262.4 kN average ratio of experimental to calculated results is 0.947. ANSI/
0 AISC360-10 [12] provides a reasonably good estimation of the load
capacity, with an average load ratio of 0.992. Nonetheless, GB50017-
-100 201X [10] and Eurocode 3 [11] yield rather conservative estimations
of the load capacity, with average load ratios of 1.160 and 1.214,
respectively. Comparisons were also made between experimental
-200 results and buckling curves in design codes, as shown in Fig. 10. In the
figures, non-dimensional slenderness λ and buckling factor φ are
-300 defined in Eqs. (1) to (4), respectively. Note that reduction factor χ in
-3000 -2000 -1000 0 1000 2000 Eurocode 3 [11] is equivalent to buckling factor φ. Similar results to
Strain of B-360-4.5 (με) those in Table 4 were obtained. It is noteworthy that according to
Eurocode 3 [11], all beam sections could be categorised into class 1,
Fig. 8. Strain profiles of B-360-4.5 at different stages. and thus Mp was used to calculate the reduction factor and non-
dimensional slenderness. For the singly-symmetric sections in this
study, the plastic moment capacity Mp was smaller than the yield
300 moment My. As a result, the reduction factors determined from
Eurocode 3 are significantly greater than those from GB50017-2003
250 [9]. Moreover, the design curve in Eurocode 3 [11] is well below that
in GB50017-2003 [9]. Therefore, the test data matched the design
curve in Eurocode 3 [11] more closely than that in GB50017-2003 [9].
Vertical load (kN)

200
However, the non-dimensional slenderness of steel beams in the exper-
imental programme fell in the range of 0.9 and 1.4. More data beyond
150
this range are necessary to comprehensively evaluate the performance
of unrestrained Q460GJ steel beams.
100 SG-1
SG-2 sffiffiffiffiffiffiffiffi
50 SG-8 My MEXP
SG-10 λ¼ ;φ ¼ ðGB50017‐2003Þ ð1Þ
Mcr My
0
-15000 -10000 -5000 0 5000
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Strain of B-360-4.5 (με) 1:1M y M EXP
λ¼ ;φ ¼ ðGB50017‐201XÞ ð2Þ
Mcr 1:1My
Fig. 9. Variation of strains of steel beam B-360-4.5.

sffiffiffiffiffiffiffiffi
the tensile stresses measured by SG-8 and SG-10 on the bottom flange. Mp M EXP
λ¼ ;χ ¼ ðEurocode 3Þ ð3Þ
Nonetheless, the strains of SG-1 and SG-2 deviated from each other, as Mcr Mp
the load approached its capacity. After global buckling occurred,
the strain of SG-2 kept increasing, whereas that of SG-1 was shifted to sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tensile, due to lateral bending and torsional deformations. MnðL¼0Þ M EXP
λ¼ ;φ ¼ ðANSI=AISC360‐10Þ ð4Þ
Mcr M nðL¼0Þ

3.2. Discussions
where My is the yield moment of beam sections, Mcr is the theoretical
In order to evaluate the lateral-torsional buckling resistance of elastic lateral-torsional buckling moment, Mn(L=0) is the cross-section
Q460GJ steel beams, design values were calculated in accordance with moment resistance corresponding to a theoretical “zero-length”
different design codes for steel structures and comparisons were made member, MEXP is ultimate moment capacity obtained in the test and
between design and calculated results, as summarised in Table 4. As Mp is the plastic moment capacity of beams.

Table 4
Comparisons of test results with design values from different design codes.

Steel beam Test results GB50017-2003 [9] GB50017-201X [10] Eurocode 3 [11] ANSI/AISC360-10 [12]

Pt P03 Pt/P03 P1X Pt/P1X PEC3 Pt/PEC3 PAISC Pt/PAISC

B-270-3.0 382.4 417.9 0.915 361.5 1.058 341.5 1.120 365.5 1.046
B-270-3.5 302.6 330.8 0.915 275.6 1.098 264.4 1.144 299.9 1.009
B-270-4.0 259.0 269.4 0.961 219.5 1.180 211.4 1.225 250.3 1.035
B-270-4.5 208.5 211.7 0.985 170.7 1.221 167.3 1.246 211.5 0.986
B-360-3.0 571.8 575.4 0.994 465.8 1.228 421.1 1.358 539.8 1.059
B-360-3.5 415.0 447.1 0.928 350.6 1.184 316.5 1.311 442.9 0.937
B-360-4.5 262.4 260.6 1.007 210.3 1.248 218.5 1.201 275.1 0.954
Mean value 0.958 1.174 1.229 1.004
Coefficient of variation 4.0% 6.0% 6.9% 4.7%
S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351 347

1.2 1.2
GB50017-2003 h/b=1.43
Euler curve h/b=1.93
1.0 1.0
H270×180×8×16(8) Euler curve
H360×180×8×16(8) H270×180×8×16(8)
0.8 0.8 H360×180×8×16(8)
Buckling factor

Buckling factor
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.50 0.75 1.00 1.25 1.50 1.75 2.00 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Non-dimensional slenderness Non-dimensional slenderness

(a) GB50017-2003 [9] (b) GB50017-201X [10]

1.2 1.2
Welded or rolled section c ANSI/AISC360-10
Euler curve Euler curve
1.0 1.0
H270×180×8×16(8) H270×180×8×16(8)
H360×180×8×16(8) H360×180×8×16(8)
0.8 0.8
Reduction factor

Buckling factor
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.50 0.75 1.00 1.25 1.50 1.75 2.00 0.50 0.75 1.00 1.25 1.50 1.75 2.00
Non-dimensional slenderness Non-dimensional slenderness

(c) Eurocode 3 [11] (d) ANSI/AISC 360-10 [12]

Fig. 10. Comparison between test results and design values.

4. Numerical models for steel beams eigenvalue analysis. Residual stresses were applied to the beams
through the *INITIAL CONDITION command in ABAQUS. Figs. 12(a and
4.1. Development of numerical models b) show the distribution of longitudinal residual stress, measured on
steel beams of the same cross sections by using sectioning method.
To extend the range of non-dimensional slenderness in the study, Besides, one more beam section H450450 × 180 × 8 × 16(8) was sec-
finite element models were developed by using ABAQUS [14] for tioned to obtain the residual stress distribution, as shown in Fig. 12(c).
Q460GJ steel beams under three-point bending. Fig. 11 shows the
model with steel mesh in ABAQUS. The measured dimensions were
utilised in the model and steel stiffeners were also taken into consider-
ation at the loading point and end supports. Four-node shell elements
with reduced integration S4R were used for the beam. For each cross
section, the sizes of steel mesh in the flange and web were 8 mm
and 10 mm, respectively. The mesh size along the beam length was
25 mm. A piecewise linear stress-strain relationship was defined for
each steel, including the elastic stage, yield plateau and hardening
stage. The stresses and corresponding strains listed in Table 3 were
incorporated into the stress-strain curve. Note that the onset of the
hardening stage was determined from Fig. 4. The stress-strain curve
was converted to the true stress-strain curve and used in numerical
simulations.
Initial imperfections play a critical role in the lateral-torsional
buckling behaviour of steel beams subject to a concentrated point load
at the mid-span. In the model, initial geometric imperfections were
considered by using the *IMPERFECTION command. The measured
initial geometric imperfection was introduced to the model based on Fig. 11. Finite element model for steel beam B-360-4.5.
348 S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351

(MPa)
-100
-200
400
200
300
100
0
(MPa)
-100
-200
400
200
300
100
200

0
100

(MPa)
300

-100
-200
0

400
200
300
100
200
-100

0
100
300 0 -200
200 -100 (MPa)
100 -200
0 (MPa)
-100
-200
(MPa)

200
100
0
-100 200
-200 100
-300 0
(MPa) -100 200
-200 100
-300 0
(a) H270×180×8×16(8) (MPa) -100
-200
(b) H360×180×8×16(8) (MPa)

(c) H450×180×8×16(8)

Fig. 12. Residual stress in beam sections [15].

Details of the residual stress measurement can be found in a companion of steel beams. Fig. 13 shows the failure mode of beam B-360-4.5. Signif-
paper [15]. icant lateral deflections could be observed near the mid-span, indicating
All beams were simulated under simply-supported boundary condi- lateral-torsional buckling of the beam. Fig. 14 shows the comparison
tions. Both ends of the beams were restrained against vertical displace- between experimental and numerical load-displacement curves. The
ments, out-of-plane displacements and twists. A roller support was numerical results are in good agreement with the test results. However,
designed for one end of the beams so that longitudinal displacements there are some significant discrepancies between experimental and
could be released. A steel plate was modelled on the beam top flange numerical curves in Figs. 14(b and e), possibly induced by the different
to take account of the effect of loading plates on preventing local buckling initial geometric imperfections in the tests and simulations. Table 5
of the flange. A concentrated point load was applied to the steel plate. summarises the comparison between load capacities obtained from
experimental tests and numerical simulations. When the measured ini-
4.2. Validation of numerical models tial geometric imperfection was adopted, the average ratio of numerical
to test results is 1.039, with a coefficient of variation of 9.0%. It indicates
The modified Riks (arc-length) method was employed in the finite that the numerical model predicts the load capacity with reasonably
element analysis to simulate the lateral-torsional buckling behaviour good accuracy. It is noteworthy that as for B-270-3.5, the deviation

Fig. 13. Failure mode of steel beam B-360-4.5.


S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351 349

500 500
Experimental result Experimental result
Numerical result Numerical result
400 400

Vertical load (kN)

Vertical load (kN)


300 300

200 200

100 100

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Vertical displacement (mm) Vertical displacement (mm)

(a) B-270-3.0 (b) B-270-3.5

350 300
Experimental result Experimental result
Numerical result Numerical result
280 240
Vertical load (kN)

Vertical load (kN)


210 180

140 120

70 60

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Vertical displacement (mm) Vertical displacement (mm)
(c) B-270-4.0 (d) B-270-4.5

750 500
Experimental result Experimental result
Numerical result Numerical result
600 400
Vertical load (kN)

Vertical load (kN)

450 300

300 200

150 100

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Vertical displacement (mm) Vertical displacement (mm)
(e) B-360-3.0 (f) B-360-3.5

350
Experimental result
Numerical result
280
Vertical load (kN)

210

140

70

0
0 10 20 30 40 50 60
Vertical displacement (mm)

(g) B-360-4.5
Fig. 14. Vertical load-displacement curves of steel beams.
350 S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351

Table 5 between experimental and numerical values is up to 15.9%. Numerical


Comparisons of test with numerical results. simulations were also carried out by using an initial geometric imper-
Steel beam Pt (kN) PN1 (kN) PN1/Pt PN2 (kN) PN2/Pt fection of 1/1000 of the beam span. Correspondingly, the average ratio
B-270-3.0 382.4 411.1 1.075 388.0 1.015
becomes 0.935, with a coefficient of variation of 7.0%. Thus, it is
B-270-3.5 302.6 350.6 1.159 307.8 1.017 relatively conservative to adopt a geometric imperfection of 1/1000 of
B-270-4.0 259.0 260.8 1.007 236.7 0.924 the beam span, as used in most design codes. This value was also used
B-270-4.5 208.5 233.0 1.118 194.1 0.931 in the parametric study.
B-360-3.0 571.8 498.4 0.872 479.4 0.838
B-360-3.5 415.0 429.6 1.035 389.1 0.938
B-360-4.5 262.4 263.8 1.005 231.6 0.883
Mean value 1.039 0.935 5. Parametric study
Coefficient of variation 9.0% 7.0%

Note: PN1 is the numerical result by using the measured geometric imperfection, and PN2 is A series of parametric studies was conducted based on the verified
the result by using an initial geometric imperfection equal to 1/1000 of the beam span, as numerical model, in which the effects of non-dimensional slenderness
specified by GB50075 [16]. and height-to-width ratio were investigated. In the study, the non-

Table 6
Dimensions of steel beams in parametric study.

Cross section Height-to- Height of Thickness Width of Thickness of Width of Thickness of Beam
width cross section of web top flange top flange bottom flange bottom flange span
ratio
H (mm) tw (mm) b1 (mm) t1 (mm) b2 (mm) t2 (mm) L (m)

H270 × 180 × 8 × 16(8) 1.5 270 8 180 16 180 8 1.7–10.5


H360 × 180 × 8 × 16(8) 2.0 360 8 180 16 180 8 1.6–7.9
H450 × 180 × 8 × 16(8) 2.5 450 8 180 16 180 8 1.6–6.8

1.2 1.2
GB50017-2003 h/b=1.43
Euler curve h/b=1.93
1.0 1.0
H270×180×8×16(8) h/b=2.43
H360×180×8×16(8) Euler curve
0.8 H450×180×8×16(8) 0.8 H270×180×8×16(8)
Buckling factor

Buckling factor

H360×180×8×16(8)
0.6 0.6 H450×180×8×16(8)

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Non-dimensional slenderness Non-dimensional slenderness

(a) GB50017-2003 [9] (b) GB50017-201X [10]

1.2 1.2
Rolled or welded sections c ANSI/AISC360-10
Rolled or welded sections d Euler curve
1.0 Euler curve 1.0
H270×180×8×16(8)
H270×180×8×16(8)
H360×180×8×16(8)
H360×180×8×16(8)
0.8 0.8 H450×180×8×16(8)
Reduction factor

H450×180×8×16(8)
Buckling factor

0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 0.5 1.0 1.5 2.0 2.5
Non-dimensional slenderness Non-dimensional slenderness

(c) Eurocode 3 [11] (d) ANSI/AISC 360-10 [12]

Fig. 15. Comparison between numerical results and design values.


S.-B. Kang et al. / Journal of Constructional Steel Research 145 (2018) 341–351 351

dimensional slenderness ratio ranged from 0.6 to 2.0, and the height-to- (2) By increasing the beam span, corresponding ultimate load sub-
width ratio was 1.5, 2.0 and 2.5 for cross sections H270 × 180 × 8 × 16(8), stantially decreased. The height-to-width ratio of steel beams
H360 × 180 × 8 × 16(8) and H450 × 180 × 8 × 16(8) corresponding to had limited effect on the global buckling resistance.
curves c and d in Eurocode 3 [11]. A total of 45 beams were simulated, (3) The design curves in GB50017-2003 [9] and ANSI/AISC360-10
as included in Table 6. [12] overestimate the load capacity of steel beams when the
Fig. 15(a) shows the comparison between numerical results and the non-dimensional slenderness is relatively small. The curves in
design curve in GB50017-2003 [9]. It can be observed that when the GB50017-201X [10] and curve d for rolled or welded sections
non-dimensional slenderness is relatively small (less than 1.50), the in Eurocode 3 [11] are conservative for Q460GJ steel beams
numerical results fall below the design curve, indicating that the design with singly symmetric I-sections. Therefore, they can be used
code overestimates the lateral-torsional buckling resistance of Q460GJ for global buckling design of Q460 beams with reasonable
steel beams with singly-symmetric I-shaped cross sections. Once the conservatism.
non-dimensional slenderness exceeds 1.50, the design curve slightly
underestimates the load resistance. It is noteworthy that the influence Acknowledgments
of the height-to-width ratio of cross sections is limited. In GB50017-
201X [10], the effects of the height-to-width ratio are considered and The authors gratefully acknowledge the financial support provided
different curves are defined, as shown in Fig. 15(b). Comparisons by the Fundamental Research Funds for the Central Universities (No.
between numerical results and design curves suggest that the curves 106112017CDJQJ208849 and No. 106112017CDJXY200005) and the
underestimate the load resistance of steel beams. In Eurocode 3 [11], National Natural Science Foundation of China (No. 51578089).
buckling curve c should be used if the height-to-width ratio of welded
I-sections is less than or equal to 2.0; otherwise, curve d should be References
adopted. Cross sections H270 × 180 × 8 × 16(8) and H360 × 180 × 8 ×
[1] S.-D. Nie, S.-B. Kang, L. Shen, B. Yang, Experimental and numerical study on global
16(8) correspond to curve c, and H450 × 180 × 8 × 16(8) is for curve
buckling of Q460GJ steel box columns under eccentric compression, Eng. Struct.
d. Numerical results of all the three sections are well above curves c 142 (2017) 211–222.
and d, as shown in Fig. 15(c). Thus, the buckling curves specified [2] B. Yang, G. Xiong, K. Ding, S.D. Nie, W. Zhang, Y. Hu, G.X. Dai, Experimental and nu-
by Eurocode 3 [11] yield fairly conservative estimations of the global merical studies on lateral-torsional buckling of GJ structural steel beams under a
concentrated loading condition, Int. J. Struct. Stab. Dyn. 16 (1) (2016) 1640004.
buckling resistance of Q460GJ steel beams. Nevertheless, the design [3] G.M. El-Mahdy, M.M. El-Saadawy, Ultimate strength of singly symmetric I-section
curve in ANSI/AISC360-10 [12] considerably overestimates the lateral- steel beams with variable flange ratio, Thin-Walled Struct. 87 (2015) 149–157.
torsional buckling resistance of the beams (see Fig. 15(d)) when the [4] A.S. Surla, S.Y. Kang, J.S. Park, Inelastic buckling assessment of monosymmetric I-
beams having stepped and non-compact flange sections, J. Constr. Steel Res. 114
non-dimensional slenderness is between 1.0 and 1.5, and thus it is not (2015) 325–337.
conservative to use this curve for the global buckling design of Q460GJ [5] A. Andrade, D. Camotim, P.B. Dinis, Lateral-torsional buckling of singly symmetric
steel beams. web-tapered thin-walled I-beams: 1D model vs. shell FEA, Comput. Struct. 85
(17–18) (2007) 1343–1359.
[6] D. Camotim, A. Andrade, C. Basaglia, Some thoughts on surprising result concerning
6. Conclusions the lateral-torsional buckling of monosymmetric I-section beams, Thin-Walled
Struct. 60 (2012) 216–221.
[7] B. Yang, S.B. Kang, G. Xiong, S. Nie, Y. Hu, S. Wang, J. Bai, G. Dai, Experimental and
This paper presents the experimental and numerical results of singly
numerical study on lateral-torsional buckling of singly symmetric Q460GJ steel I-
symmetric I-beams fabricated from high-performance Q460GJ struc- shaped beams, Thin-Walled Struct. 113 (2017) 205–216.
tural steel. The beams were tested under a concentrated point load at [8] R.D. Ziemian, Guide to Stability Criteria for Metal Structures, sixth ed. John Wiley &
Sons Inc., Hoboken, New York, 2010.
the mid-span and expected to fail in a range between elastic and inelas-
[9] Ministry of Construction, Code for Design of Steel Structures, GB50017-2003, China
tic lateral-torsional buckling. Finite element models were developed to Planning Press, Beijing, 2003.
simulate the global buckling behaviour of the beams. The model was [10] Management Committee for National Standards, Code for Design of Steel Structures
validated against test results and used for parametric studies on the (Draft Version), GB50017-201X, Beijing, 2012.
[11] BSI, Eurocode 3: Design of Steel Structures—Part 1-1: General Rules and Rules for
effects of non-dimensional slenderness and height-to-width ratio. Com- Buildings, BS EN 1993-1-1:2005, British Standards Institution, London, 2005.
parisons were made with design curves in GB50017-2003 [9], GB50017- [12] American Institute of Steel Construction, Specification for Structural Steel Buildings,
201X [10], Eurocode 3 [11] and ANSI/AISC360-10 [12]. The following ANSI/AISC360-10, Chicago, Illinois, 2010.
[13] Standardization Administration of China, Tensile Testing—Part 1: Method of Test at
conclusions could be drawn from the study. Room Temperature, GB/T228.1-2010, Standards Press of China, Beijing, 2010.
[14] ABAQUS, Analysis user's Manual Version 6.13, ABAQUS Inc, 2013.
(1) All steel beams failed in lateral-torsional buckling except B-360- [15] B. Yang, S.D. Nie, G. Xiong, Y. Hu, J. Bai, W. Zhang, G.X. Dai, Residual stresses in
4.0 which developed a rather high load resistance due to the welded I-shaped sections fabricated from Q460GJ structural steel plates, J. Constr.
presence of friction in the loading device. Yielding of the tension Steel Res. 122 (2016) 261–273.
[16] Ministry of Housing and Urban-Rural Development, Code for Construction of Steel
flange was not observed prior to failure, and local buckling did Structures, GB 50755-2012, China Building Industry Press, Beijing, 2012.
not occur in the compression flange.