Journal of Constructional Steel Research 147 (2018) 408–421

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Ultimate capacity of I-sections under combined loading – Part 1:

Experiments and FE model validation
X. Yun a, L. Gardner a,⁎, N. Boissonnade b
a
Department of Civil and Environmental Engineering, Imperial College London, London, UK
b
Department of Civil and Water Engineering, Laval University, Quebec, Canada

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

Article history: An experimental and numerical study of hot-rolled steel I-sections under combined compression and bending
Received 24 November 2017 moment is presented herein. A total of two stub column tests and 12 mono-axial or bi-axial eccentric compres-
Received in revised form 16 April 2018 sion tests on HEB 160 cross-sections with two different material grades (S235 and S355) were carried out. The
Accepted 18 April 2018
tested cross-sections were of stocky proportions to enable the influence of material strain hardening on the
Available online 15 May 2018
strength and behaviour of hot-rolled steel I-sections to be investigated. The loading eccentricities for the eccentric
Keywords:
compression tests were varied in order to achieve different axial compression-to-bending moment ratios.
Combined loading Measured geometric and material properties, together with the full load-deformation histories from the test
Cross-sectional behaviour specimens, were reported. Finite element (FE) models were developed and validated against the experimentally
Experimental investigation obtained load-deformation curves, as well as the failure modes. The FE results successfully captured the
Hot-rolled steel I-sections experimental structural performance of hot-rolled steel I-sections and the validated FE models were then used
Numerical models for parametric studies in the companion paper to generate additional numerical results, considering different
Strain hardening cross-section slendernesses, material grades and combinations of loading. The experimental and numerical
results are employed in the companion paper for the assessment of the design rules given in EN 1993-1-1
(2005) and AISC-360-16 (2016) and for the extension of the deformation-based continuous strength method
to the case of hot-rolled steel I-sections under combined loading.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction a Class 3 cross-section is therefore limited to the elastic moment Mel, or

a linearly interpolated moment between Mel and Mpl [3,4]. For a Class 4
Hot-rolled steel I-sections are generally the product of choice in cross-section, local buckling occurs in the elastic range prior to reaching
structural frameworks, and have been extensively studied for many either the yield load in compression or elastic moment in bending, and
years. In terms of determining their structural design resistances, the the effective width method is used to consider the effects of local buck-
concept of cross-section classification is used in current design codes, ling explicitly.
e.g. EN 1993-1-1 [1] and AISC-360-16 [2], where four classes of cross- Steel design specifications [1,2] typically treat the plate elements of
section – Class 1 (plastic), Class 2 (compact), Class 3 (semi-compact) the cross-section individually, thus neglecting the interaction effects be-
and Class 4 (slender), are typically defined according to the susceptibil- tween adjacent elements, i.e. the ability of the less slender elements to
ity of the most slender element of the cross-section to local buckling, as provide some assistance in resisting local buckling to the more slender
illustrated in Fig. 1. A Class 1 cross-section is fully effective under pure elements, and the classification of the most slender element defines
compression and can develop its plastic moment capacity Mpl in bend- that of the overall cross-section. Previous stub column tests [5,6],
ing with sufficient rotation capacity to allow redistribution of moments beam tests [7] and beam-column tests [8,9] on I-sections have shown
in indeterminate steel frames. A Class 2 cross-section is also fully effec- that web-flange interaction effects can have a substantial influence on
tive in pure compression and can reach its plastic moment capacity Mpl ultimate cross-section behaviour and hence the determination of its ul-
but with lower (assumed to be zero in design) rotation capacity in timate resistance. Kato [10,11] and Beg and Hladnik [12] proposed slen-
bending. A Class 3 cross-section remains fully effective in pure compres- derness limits for Class 3 cross-sections in a form considering these
sion, but in bending, local buckling now occurs after the elastic moment web-flange interaction effects. Seif and Schafer [13] proposed analytical
Mel but before the plastic moment Mpl is attained; bending resistance of expressions for determining the elastic buckling stresses of hot-rolled
steel I-sections considering element interaction.
⁎ Corresponding author. Within the current cross-section classification framework, an idealised
E-mail address: [email protected] (L. Gardner). elastic, perfectly plastic material model is used, whereby the maximum

https://doi.org/10.1016/j.jcsr.2018.04.016
0143-974X/© 2018 Elsevier Ltd. All rights reserved.
 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 409

Mpl

Class 2 Class 1
Mel

Class 3

Class 4

Fig. 3. Typical measured stress-strain curves from S235 and S355 specimens.
Rotation

Fig. 1. Typical response and cross-section classification of steel sections in bending.

with the aim of investigating the effects of the material strain hardening
on the behaviour and resistance of stocky hot-rolled steel I-sections. A
total of two stub column tests and 12 mono-axial or bi-axial eccentric
stress in the cross-section is limited to the material yield stress fy, compression tests were carried out. The experimental results were sup-
neglecting the beneficial effects of strain hardening. In addition to plemented with additional data collected from the literature [3] and
highlighting the importance of element interaction effects, experimental generated by means of validated finite element (FE) models considering
results have also shown that this limitation often leads to conservative different cross-section slenderness, material grades and combinations
predictions of the resistance of stocky hot-rolled steel I-sections in both of loading. These data are then used to evaluate the current codes of
compression [14] and bending [15], due primarily to the omission of practice, including EN 1993-1-1 [1] and AISC-360-16 [2], and to extend
strain hardening. Furthermore, results have shown that linearly interpo- the CSM to the design of hot-rolled steel I-sections under combined
lating between Mel and Mpl does not offer an accurate representation of loading.
the spread of plasticity in Class 3 cross-sections [3,4].
To address these shortcomings in current codified methods, a defor-
mation based design approach called the continuous strength method 2. Experimental investigation
(CSM), which provides an alternative treatment to cross-section classi-
fication and enables the effective utilisation of strain hardening, has 2.1. Introduction
been proposed. The method was originally developed for stainless
steel structural elements [16–18], which exhibit a high degree of strain An experimental study comprising material testing, geometric im-
hardening, and the same concept has since been applied to structural perfection measurements, two stub column tests, four uniaxial bending
carbon steel, with a modified material model [19–21]. However, previ- plus compression tests and 8 biaxial bending plus compression tests on
ous developments have focused on compression and bending resis- hot-rolled steel I-sections was conducted at the University of Applied
tances, while investigations into the cross-sectional behaviour of hot- Sciences of Western Switzerland in order to investigate their cross-
rolled steel I-sections under combined loading have been vary limited. sectional behaviour and resistance under combined loading. The test
Hence, a comprehensive experimental and numerical simulation pro- specimens were categorized into two series according to their steel
gramme on hot-rolled steel I-sections under combined compression grade: Series ‘A’ (S235) and Series ‘B’ (S355). All the test specimens
and bending moment has been carried out and is presented herein, were hot-rolled HEB 160 cross-sections, which are Class 1 according
to the slenderness limits of EN 1993-1-1 [1]. It was chosen to test
B cross-sections of compact proportions in order that material strain
hardening would be experienced and hence that the influence of strain
hardening on their behaviour and strength could be studied.
tf
TF1 TF2
ri
Table 1
Measured material properties from tensile coupon tests.
TW1
Label E fy fu εsh εu fu/fy
tw (N/mm2) (N/mm2) (N/mm2) (%) (%)
H
S235 TF1 – 322 464 1.62 17.99 1.44
S235 TF2 193,500 303 455 1.27 18.83 1.50
S235 TF3 198,800 300 453 1.44 19.64 1.51
TW2 S235 TF4 195,800 316 457 1.33 17.79 1.45
S235 TW1 – 365 494 1.75 14.66 1.35
S235 TW1 192,800 386 488 1.92 15.01 1.26
S355 TF1 210,200 393 518 2.03 13.23 1.32
TF3 TF4 S355 TF2 205,100 392 494 2.55 13.56 1.26
S355 TF3 208,600 386 507 2.14 12.94 1.31
S355 TF4 213,000 388 504 2.17 13.56 1.30
S355 TW1 202,400 405 520 2.00 11.80 1.29
S355 TW1 199,500 405 519 1.79 12.74 1.28
Fig. 2. Definition of symbols and location of tensile coupons in tested cross-sections.
410 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421

coupons to measure the longitudinal strain and elongation. A constant

rate of strain (0.045%/s) was applied until fracture.
Typical measured stress-strain curves from the flange and the web
tensile coupons from the two cross-sections are shown in Fig. 3. All the
tensile coupons exhibited the anticipated sharply defined yield point,
followed by a yield plateau before the initiation of strain hardening. The
material properties obtained from all tensile coupon tests are summa-
rized in Table 1, including the modulus of elasticity E, the yield stress fy,
the ultimate tensile stress fu, the yield-to-ultimate tensile stress ratio fy/
fu, the strain hardening strain εsh (where the yielding plateau ends and
subsequently the strain hardening initiates) and the strain at the ultimate
tensile stress εu. The material properties of hot-rolled steel cross-sections
are generally fairly homogeneous, although higher yield strengths are
Fig. 4. Test setup for geometric imperfection measurements. often observed in thinner material due to the more rapid cooling. This
trend is seen in the present study, where the yield strength of the thinner
web is higher than that of the flanges (see Table 1).
Initial local geometric imperfections were measured for all the spec-
imens prior to testing. Since all the specimens presented in this paper
are short columns that are not susceptible to global buckling, only
local imperfections were considered. The measurements were con-
ducted by means of an aluminium perforated bar holding 7 equally-
spaced displacement transducers (LVDTs) which was moved laterally
across the flat faces of each specimen (see Fig. 4) in order to develop
3D geometrical representation of the sections, a typical example of
which is shown in Fig. 5. Note that the measurements of the 7 LVDTs
were started and finished 50 mm away from the ends of the specimens
in order to eliminate the possible influence of welding on the local im-
perfection measurements. The measured maximum local geometric im-
perfection amplitudes from the flange faces (ωf) and the web face (ωw)
of the specimens are reported in Tables 2 and 3.

2.3. Stub column tests

For each of the two test series (grades S235 and S355 steel), one
concentrically-loaded stub column test was performed in order to de-
termine the cross-sectional load carrying capacity under pure compres-
Fig. 5. Typical measured initial local geometric imperfection profile (×40). sion as well as to assess the influence of material strain hardening on the
cross-sectional resistance. The nominal length for each specimen was
chosen to be approximately three times the height of the cross-
section, to be short enough to avoid global flexural buckling, but suffi-
2.2. Material testing and initial geometric imperfection measurements ciently long to contain representative local geometric imperfection
and residual stress patterns [23]. The ends of the columns were milled
The basic stress-strain properties of the investigated hot-rolled steel flat and perpendicular to the longitudinal axis of the column, and
I-sections were determined through 12 tensile coupon tests, with six endplates with a thickness of 30 mm were welded to the ends of the
coupons tested for each of the two steel sections (one grade S235 and specimens to ensure a uniform distribution of load during testing and
the other S355). Tensile coupons were extracted from the web and to avoid premature end failures. The testing machine was a 3000 kN ca-
flanges of the two specimens in the longitudinal (rolling) direction. pacity hydraulic rig and the stub columns were loaded under displace-
Fig. 2 shows the locations from which the coupons were cut from the ment control up to and beyond their peak loads, with the rate of
test specimens, together with the definition of the adopted notation, displacement kept constant at 0.025 mm/s until the peak load, and the
where TF and TW indicate tensile coupons cut from the flange and the rate increased soon after. The testing setup for the stub columns is
web, respectively. The tensile coupons were 20 mm in width with shown in Fig. 6. Four LVDTs were employed at each end to determine
gauge lengths of 70 mm and 90 mm for the web coupons and the flange the average end shortening of the specimens and three strain gauges,
coupons, respectively. All the tensile coupon tests were conducted in mounted to the column mid-faces at mid-height, were used to measure
compliance with the procedures set out in EN ISO 6892-1 [22] using a the longitudinal strains. An additional LVDT was used at mid-height be-
100 kN hydraulic testing machine. Once a tensile coupon had been tween the flanges to detect the initiation of local buckling, if any, as
gripped, a 20 mm clip gauge was affixed to the mid-height of the shown in Fig. 6.

Table 2
Measured geometrical parameters and key test results of stub column specimens.

Specimen ID L B H tf tw ri A ωf ωw Nu δu Nu/Afy,wa

(mm) (mm) (mm) (mm) (mm) (mm) (mm2) (mm) (mm) (kN) (mm)

A7 500 161.44 160.80 13.00 8.61 15.00 5552 0.47 0.34 2316.3 12.27 1.25
B7 500 160.80 161.15 12.20 8.20 15.00 5238 2.65 0.89 2420.0 15.31 1.16
 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 411

Table 3
Measured geometrical parameters of specimens under combined loading.

Specimen ID Axis of bending L B H tf tw ri ωf ωw

(mm) (mm) (mm) (mm) (mm) (mm) (mm) (mm)

A2 Major axis 500 161.16 161.06 13.06 8.58 15.00 0.20 0.20
B2 500 160.65 161.30 12.00 8.00 15.00 0.68 0.34
A5 Minor axis 500 161.40 161.02 12.99 8.57 15.00 0.36 0.21
B5 500 160.80 161.15 12.20 8.30 15.00 2.51 0.75
A1 Bi-axial 500 160.96 161.19 13.07 8.60 15.00 0.26 0.26
A3 500 161.41 161.05 12.99 8.66 15.00 0.35 0.68
A4 500 161.08 161.23 12.99 8.48 15.00 1.46 0.44
A6 500 161.68 161.01 12.94 8.50 15.00 0.19 0.14
B1 500 160.80 161.20 12.15 8.20 15.00 1.12 1.04
B3 500 160.80 161.15 12.20 8.20 15.00 1.80 0.29
B4 500 160.95 160.90 12.25 8.30 15.00 0.66 0.76
B6 500 160.80 161.15 12.15 8.20 15.00 4.15 0.19

Table 2 presents the measured cross-section dimensions for the were carried out to investigate the cross-sectional behaviour of hot-
specimens tested under pure compression using the nomenclature rolled steel I-sections under combined loading. The measured cross-
from Fig. 2, where L is the length of the specimen, B and H are the section dimensions and maximum local imperfection amplitudes of
outer width and depth of the cross-section, respectively, tf is the flange the specimens are presented in Table 3, following the nomenclature
thickness, tw is the web thickness, ri is the internal corner radius and A is from Fig. 1.
the cross-sectional area. As shown in Fig. 10, the loading rig consisted of a hydraulic actuator
The strain gauge readings provided a more accurate means of deter- at the lower end of the specimen and a fixed top platten. Two hemi-
mining the initial axial stiffness of the stub columns than the LVDTs, spherical bearings were specially designed to provide pinned-pinned
since the LVDT measurements include not only the end shortening of end restraints for the test specimens. Endplates were welded to the
the specimen but also the elastic deformation of the endplates. Thus, a specimens at different eccentricities to generate a range of bending mo-
correction that combines both sets of measurements was required to ment to axial loading ratios, and then bolted to the hemispherical bear-
get the true end shortening values, following the method recommended ings of the 3000 kN test machine. Each of the two bearings contained
in [24]. Table 2 also summarizes the corrected test results of the stub two T-shaped grooves, which enabled adjustment of the specimens
columns, including the ultimate axial load Nu, the corresponding true when bolted to the endplates to achieve loading at the specified eccen-
end shortening at the ultimate load δu and ultimate load normalized tricities. The bolts were pre-tensioned in order to prevent uplift or de-
by the squash load Nu/Afy,wa, where fy,wa is the weighted average mate- tachment of the specimen endplates from the hemispherical bearings.
rial yield stress based on the tensile coupon test results given in Table 1 Tests were carried out under displacement control in order to capture
for the flanges and web and their respective areas. The stub column the post-ultimate behaviour of the specimens, at a rate of 0.025 mm/s.
specimens generally failed by inelastic local buckling at mid-height, as During testing, eight LVDTs placed at both ends of the test specimens
shown for specimen B7 in Fig. 7. The full load-end shortening curves were used to measure axial shortening and end rotation. In addition,
for the stub columns are presented in Fig. 8, while the normalized two further LVDTs were used at the mid-height of the specimens in
load-end shortening curves are depicted in Fig. 9, where Ny is the order to measure the lateral deflections in both principal directions,
yield load Ny = Afy,wa, highlighting the more prominent role of strain allowing the generated second order bending moments to be calculated.
hardening for the S235 specimen (Specimen A7) than the S355 speci-
men (Specimen B7).

2.4. Combined loading tests

In the combined loading tests, uniaxial and biaxial bending plus

compression was achieved through the application of eccentric com-
pression. For each of the two series, two uniaxial bending plus compres-
sion tests - one bending about the major axis and the other bending
about the minor axis, and four biaxial bending plus compression tests

LVDTs

Specimen

Endplate

Fig. 6. Testing setup for stub columns. Fig. 7. Typical failure mode of stub columns (B7).
412 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421

3000
The applied load and readings from the LVDTs, inclinometers and
strain gauges were recorded using a data logger at half-second intervals
2500 (2 Hz) during the tests.
A7
B7 The measurements from the strain gauges can be used to determine
2000 the calculated loading eccentricities e0, to enable an accurate evaluation
of the tests and FE results, as well as a fair assessment of the design
Load (kN)

1500
methods. For each axis of bending, the flexural strain εm due to the
bending moment can be derived through Eq. (1):

1000 εmax −εmin

εm ¼ ð1Þ
2
500
where εmax is the measured strain at the maximum compressive fibre
and εmin is the measured maximum tensile or minimum compressive
0
0 4 8 12 16 20 24
strain at the other extreme fibre. The total bending moment MT, consid-
End shortening (mm) ering both the first order bending moment M1 due to the eccentricity of
the applied force N and the second order bending moment M2 due to the
Fig. 8. Load-end shortening curves for stub column specimens. lateral deflection at mid-height Δ, can be calculated as:

1.4
MT ¼ M 1 þ M 2 ¼ Ne0 þ NΔ ð2Þ

1.2 A7 The total bending moment can also be expressed, in the elastic
B7
range, as MT = EIκ, where E is the Young's modulus, I is the second mo-
1.0 ment of area about the axis of bending and κ is the curvature which can
be determined, assuming plane sections remain plane and normal to the
0.8 neutral axis during bending, from the readings of the strain gauges. The
curvature is given by κ = ɛm / 0.5d, where d is the distance between the
N/Ny

0.6 extreme fibres in the corresponding bending axis. Therefore, the calcu-
lated loading eccentricity e0 can be determined from Eq. (3). Note that
0.4 since Eq. (3) is only applicable in the elastic range, the average value
of the eccentricities, obtained during the early stage of loading, was
0.2
taken as the calculated loading eccentricity e0 for each test. Overall,
the calculated loading eccentricities e0 were found to be in reasonably
0.0
0.00 0.01 0.02 0.03 0.04 0.05
good agreement with the measured loading eccentricities em, as
shown in Table 4, but e0 is considered to be more accurate than em
and is therefore used in the following analyses.
Fig. 9. Normalized load-end shortening response for stub column specimens.
EIðεmax −εmin Þ
e0 ¼ −Δ ð3Þ
dN
Two inclinometers, placed at each end of the test specimens, were used
for measuring end rotations. Four strain gauges, two attached to the Table 4 summarizes the key results from the specimens tested under
mid-face of each flange and the other two attached to the extreme fibres combined loading, including the ultimate load Nu, the end rotation at
of one flange at mid-height of the specimen, were used to measure the the ultimate load θu, the measured loading eccentricity em, the calcu-
longitudinal strains and to determine the calculated loading eccentricities. lated loading eccentricity e0, the mid-height lateral deflection at

Top plateau
Hinge

LVDTs
Hemispherical Le
End plates
bearings
Specimen

Hydraulic actuator

(a) Schematic diagram of test setup (b) Experimental setup

Fig. 10. Testing configuration for specimens under combined loading.

 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 413

Table 4
Summary of key test results of specimens under combined loading.

Specimen ID Axis of bending Nu em,y e0,y em,z e0,z Δu,y Δu,z θu,y θu,z MT,u,y MT,u,z

(kN) (mm) (mm) (mm) (mm) (mm) (mm) (deg) (deg) (kNm) (kNm)

A2 Major axis 1157.7 75.0 70.2 – – 15.70 – 4.90 – 99.5 –

B2 1183.3 75.0 75.1 – – 21.00 – 3.72 – 113.7 –
A5 Minor axis 943.6 – – 45.0 44.8 – 6.90 – 1.53 – 48.8
B5 1138.8 – – 45.0 44.6 – 6.40 – 1.40 – 64.9
A1 Bi-axial 1302.8 30.0 28.8 20.0 21.8 1.59 2.90 0.34 0.77 39.5 32.2
A3 1191.6 15.0 15.0 30.0 27.8 0.99 6.49 0.16 1.33 19.1 40.9
A4 879.2 75.0 75.1 30.0 29.8 2.97 3.66 0.60 0.78 68.6 29.4
A6 827.2 45.0 46.0 50.0 50.6 1.74 9.10 0.38 1.36 39.5 49.4
B1 1514.5 30.0 30.1 20.0 19.7 3.14 5.40 0.42 0.73 50.3 38.0
B3 1359.6 15.0 17.7 30.0 27.7 1.26 7.69 0.18 1.10 25.8 48.2
B4 993.5 75.0 75.5 30.0 31.2 3.19 5.48 0.67 0.88 78.2 36.5
B6 952.6 45.0 45.4 50.0 50.6 2.08 7.07 0.40 1.13 45.3 54.9

(a) Uniaxial bending plus compression (B5)

(b) Biaxial bending plus compression (B3)

Fig. 11. Typical experimental and numerical failure modes from tests under combined loading.
414 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421

1400 1200

1200 B2 e0y=75.1 mm
1000
B5 e0z=46.6 mm
A2 e0y=70.2 mm
1000
800 A5 e0z=44.8 mm
Load (kN)

800

Load (kN)
600

600
400
400
200
200

0
0 0 1 2 3 4 5 6 7
0 1 2 3 4 5 6 7 End rotation θ (deg)
End rotation θ (deg)

(a) Load-end rotation curves for specimens A2 and B2 (b) Load-end rotation curves for specimens A5 and B5

1800 1600
B3 e0y=17.7 mm
1400
1500
B1 e0y=30.1 mm B1 e0z=19.7 mm 1200
B3 e0z=27.7 mm
1200 A3 e0z=27.8 mm
A1 e0z=21.8 mm 1000
A1 e0y=28.8 mm
Load (kN)
Load (kN)

A3 e0y=15.0 mm
900 800

600 600

400
300
200

0
0
0 1 2 3 4 5
0 1 2 3 4 5
End rotation θ (deg)
End rotation θ (deg)

(c) Load-end rotation curves for specimens A1 and B1 (d) Load-end rotation curves for specimens A3 and B3

1200 1200
B4 e0y=75.5 mm
B4 e0z=31.2 mm
1000 1000
B6 e0z=50.6 mm
B6 e0y=45.4 mm
800 800
A4 e0z=29.8 mm A6 e0z=50.6 mm
A4 e0y=75.1 mm
Load (kN)

A6 e0y=46.0 mm
Load (kN)

600 600

400
400

200
200

0
0 1 2 3 4 5 0
End rotation θ (deg) 0 1 2 3 4
End rotation θ (deg)

(e) Load-end rotation curves for specimens A4 and B4

(f) Load-end rotation curves for specimens A6 and B6

Fig. 12. Measured load-end rotation curves from combined loading tests.

ultimate load Δu and the total bending moment at ultimate load MT,u, hardening on the cross-sectional resistance is carefully explored in the
where the suffixes ʻyʼ and ʻzʼ denote bending about the major axis and companion paper [25].
the minor axis, respectively.
The specimens generally failed by material yielding, though inelastic 3. Finite element modelling
local buckling was also observed for some test specimens at large defor-
mations. Representative failure modes for the uniaxial and biaxial bend- Following the experimental investigation, a numerical study of hot-
ing plus compression tests are illustrated in Fig. 11(a) and (b), rolled steel I-sections under combined loading, using the nonlinear FE
respectively. The full experimental load-end rotation curves for each analysis programme ABAQUS [26], was carried out, and is described in
specimen subjected to uniaxial or biaxial bending plus compression this section. The aims of the numerical study were to capture the phys-
loading are presented in Fig. 12(a)-(f). The influence of material strain ical behaviour observed in the experiments and to investigate the cross-
 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 415

Nodes of flange Nodes of flange

Fillet zone Increased web thickness

Nodes of web Nodes of web

Fig. 13. Representation of the fillet zones of I-sections in FE models.

sectional resistance of hot-rolled steel I-sections under combined load- flange thickness to avoid overlapping of the elements at the web-to-
ing. The FE models were firstly validated against the test results and flange junction, and these nodes were tied to their corresponding
subsequently used to perform parametric studies, described in the com- nodes at the mid-thickness of the flanges using “General multi-point
panion paper [25], to generate additional data over a wide range of constraints (*MPC)”, which ensured that both the translational and ro-
cross-section slenderness and loading combinations. tational degrees of freedom were equal for this pair of nodes. The
additional area and moments of inertia due to the fillet zones were
allowed for by increasing the thickness of the adjacent web elements
3.1. Finite element model (see Fig. 13).
The average material stress-strain responses from the flanges and
In order to represent accurately the behaviour of the test specimens, web, represented by the bilinear plus non-linear hardening model pro-
suitable FE models need to be developed. The four-noded doubly curved posed by Yun and Gardner [19], were employed in the FE models for the
shell element S4R with reduced integration was selected for the model- validation study. The engineering stress-strain curve was converted into
ling of the test specimens in this investigation, which has been shown in the format of true stress and logarithmic plastic strain, as required in
the literature to be suitable for similar problems [3,27,28]. The endplates ABAQUS [26] for the element type adopted, according to Eqs. (4) and
were modelled using the same shell elements S4R with an equivalent (5), in which σnom is the engineering stress, εnom is the engineering
thickness of 30 mm and modelled with elastic material behaviour strain and σtrue and εpl
ln are the true stress and logarithmic plastic strain,
since the endplates remained elastic during loading. The finite element respectively.
model was meshed using 20 elements along the flange width, 20 ele-
ments along the web height, with a finer mesh of 2 elements in each fil- σ true ¼ σ nom ð1 þ εnom Þ ð4Þ
let zone of the cross-section, and 100 elements along the length of the
specimen. The selected mesh size gave a sufficient degree of accuracy σ true
εpl
ln
¼ ln ð1 þ εnom Þ− ð5Þ
with acceptable computational cost. E
Particular attention was given to ensure that the properties of the fil-
let zones, as shown in Fig. 13, could be accurately represented. The The interfaces between the endplates and the specimen were
nodes at each end of the web were shifted by a distance of half the modelled using a constraint pair, where the endplates were the master

0.5fy* 0.3fy*

0.5fy* 0.5fy* 0.3fy* 0.3fy*

0.5fy* 0.3fy*
H
H

0.5fy* 0.3fy*

B 0.5fy* B 0.3fy*

(a) H/B 1.2 (b) H/B > 1.2

Fig. 14. Residual stress patterns applied to FE models.
416 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421

Table 5
Comparison of the experimental and numerical results for hot-rolled I-sections under different loading conditions.

Specimen ID Loading condition Nu,FE/Nu,test θu,y,FE/θu,y,test θu,z,FE/θu,z,test Δu,y,FE/Δu,y,test Δu,z,FE/Δu,z,test

A7 Pure compression 0.90 – – – –

B7 0.92 – – – –
A2 Major aixs bending plus compression 0.93 0.72 – 0.53 –
B2 0.95 0.78 – 1.06 –
A5 Minor aixs bending plus compression 1.00 – 0.89 – 0.84
B5 0.96 – 0.82 – 0.92
A1 Biaxial bending plus compression 0.93 0.65 1.06 1.06 1.60
A3 0.98 1.28 0.79 0.90 0.77
A4 0.96 0.77 0.45 1.01 1.25
A6 0.96 0.82 0.79 1.02 0.71
B1 0.95 0.96 1.57 0.54 0.94
B3 1.01 1.03 1.02 0.94 0.63
B4 0.98 0.78 0.85 1.42 1.27
B6 0.97 0.85 1.01 0.87 0.91
Mean 0.96 0.86 0.92 0.93 0.98
COV 0.03 0.21 0.30 0.28 0.31

surfaces and the end cross-sections of the specimen were the slave sur- web imperfection amplitudes measured in the test specimens and is
faces. This constraint provides a simple means of bonding surfaces to- also in accordance with the recommendations in [29,30], for all the FE
gether in terms of all the translational and rotational degrees of simulations in this study.
freedom. Each endplate was then coupled to a reference point located The residual stress patterns recommended by the European Con-
at an eccentricity corresponding to the calculated loading eccentricity vention for Constructional Steelwork (ECCS) [31], as shown in
e0. Note that the reference point was also offset longitudinally from Fig. 14, where compressive residual stresses are designated as posi-
the end cross-section of the specimens by 172 mm (equal to the dis- tive and tensile residual stresses as negative, were applied to the FE
tance from the end cross-section to the centroid of the hinge) for the models. The magnitude of the residual stress depends on whether
combined loading tests, in order to accurately model the effective length the height to width ratio of the cross-section is less than or equal
of the specimens Le, as shown in Fig. 10. The axial compressive force was to 1.2 or greater than 1.2 and is independent of the yield stress,
applied at the bottom reference point following the test procedure. For with the nominal stress fy⁎ = 235 MPa taken as the reference value.
the stub column FE models, which were under pure axial compression, The modified Riks method was used as the solution scheme for the
the concentric reference points were restrained against all degrees of FE simulations to allow the post-ultimate path of the modelled spec-
freedom, only allowing longitudinal translation at the bottom reference imens to be captured.
point. For the combined loading FE models, all degrees of freedom were
restrained at the reference points except for longitudinal translation at 3.2. Validation of the FE models
the bottom reference point and rotation about the axis of bending for
both reference points, to simulate pin-ended boundary conditions. Validation of the FE models was based on the comparison of the nu-
Initial local geometric imperfections were incorporated into the FE merically obtained results with the relevant data from the conducted
models in the form of the lowest elastic buckling mode shape arising tests. The ultimate load Nu,test, the end rotations at ultimate load θu,test,
under axial compression; this results in a perturbed mesh throughout and the mid-height lateral deflection at ultimate load Δu,test obtained
the cross-section and provides comparable local imperfections for all from the tests were compared with the corresponding numerical values
specimens. It has been found that the sensitivity of the numerical results Nu,FE, θu,FE and Δu,FE predicted by the FE models, as reported in Table 5.
to the initial geometric imperfections, both the shapes and amplitudes, Overall, good agreement between experimental and numerical results
is relatively low for non-slender hot-rolled steel I-sections [29]. The im- can be seen, particularly in terms of the ultimate load Nu, with the
perfection amplitude was taken as a/200 (a being the flat width of the mean value of the ratio of Nu,FE/Nu,test being 0.96 and the COV being
most slender constituent plate element in the cross-section under com- 0.03. The end rotations θu and mid-height lateral deflection at ultimate
pression), which approximately represented the average maximum load Δu less accurately captured, but acceptably predicted, and although

2500 3000

2000 2500

2000
1500
Load (kN)
Load (kN)

1500
1000
1000

500 Test Test

FE 500
FE
FE without strain hardening FE without strain hardening
0 0
0 5 10 15 20 25 0 5 10 15 20 25
End shortening δ (mm) End shortening δ (mm)

(a) A7 (b) B7
Fig. 15. Experimental and numerical load-end shortening curves for stub column specimens.
 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 417

in some cases the numerical values deviated from the corresponding behaviour of all specimens and on the ultimate capacity of some. In gen-
test values by more than 30%, the absolute differences were relatively eral, the curves obtained from the FE models with strain hardening tend
small. Good agreement between the failure modes of the tested and to show a more gradual loss of strength in the post-ultimate region,
simulated specimens was also obtained, typical examples of which are while the curves without considering the effect of strain hardening
shown in Fig. 11 (Specimens B3 and B5). tend to have a more accentuated post-peak drop in strength. There is
Figs. 15 and 16 show comparisons between the numerical load- also consistently a more significant influence of strain hardening (on
deformation curves and the corresponding experimental curves from both ultimate and post-ultimate behaviour) for the hot-rolled I-
the stub column tests and bending plus compression tests, respectively. sections under compression and major axis bending plus compression
The load-deformation curves obtained from the FE models but without than for the other loading scenarios; this is because, for a given level
considering the effect of material strain hardening (i.e. using an elastic- of outer-fibre strain, a greater proportion of the cross-sectional area en-
perfectly plastic material model) have also been plotted in Figs. 15 and ters the strain hardening regime in the former cases. Overall, it can be
16 for comparison. It can be seen from Figs. 15 and 16 that the material seen that accurate replication of the initial stiffness, ultimate load and
strain hardening has a significant influence on the post-ultimate general form of the load-deformation histories was achieved by the

1400 1400

1200 1200

1000 1000
Load (kN)

Load (kN)
800 800

600 600

400 400
Test Test
200 FE 200 FE
FE without strain hardening FE without strain hardening
0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 1.0 2.0 3.0 4.0
End rotation θ (deg) End rotation θ (deg)

(a) A1 Major axis (e0y = 28.8 mm) (b) A1 Minor axis (e0z = 21.8 mm)

1600 1600
1400 1400
1200 1200
1000 1000
Load (kN)
Load (kN)

800 800

600 600

400 400 Test

Test
200 FE 200 FE
FE without strain hardening FE without strain hardening
0 0
0.0 0.5 1.0 1.5 2.0 0.0 1.0 2.0 3.0 4.0
End rotation θ (deg) End rotation θ (deg)

(c) B1 Major axis (e0y = 30.1 mm) (d) B1 Minor axis (e0z = 19.7 mm)

1400 1400

1200 1200

1000 1000
Load (kN)

800
Load (kN)

800

600 600

400 400
Test Test
200 FE 200 FE
FE without strain hardening FE without strain hardening
0 0
0.0 2.0 4.0 6.0 8.0 10.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0
End rotation θ (deg) End rotation θ (deg)

(e) A2 Major axis (e0y = 70.2 mm) (f) B2 Major axis (e0y = 75.1 mm)

Fig. 16. Experimental and numerical load-end rotation curves for biaxial bending plus compression tests.
418 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421

1400 1400

1200 1200

1000 1000
Load (kN)

Load (kN)
800 800

600 600

400 400
Test Test
200 FE 200 FE
FE without strain hardening FE without strain hardening
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 1.0 2.0 3.0 4.0 5.0
End rotation θ (deg) End rotation θ (deg)

(g) A3 Major axis (e0y = 15.0 mm) (h) A3 Minor axis (e0z = 27.8 mm)

1600 1600

1400 1400

1200 1200

1000 1000

Load (kN)
Load (kN)

800 800

600 600

400 400
Test Test
200 FE 200 FE
FE without strain hardening FE without strain hardening
0 0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 1.0 2.0 3.0 4.0 5.0
End rotation θ (deg) End rotation θ (deg)

(i) B3 Major axis (e0y = 17.7 mm) (j) B3 Minor axis (e0z = 27.7 mm)

1000 1000
900 900
800 800
700 700
600 600
Load (kN)

Load (kN)

500 500
400 400
300 300
200 Test 200 Test
FE FE
100 100
FE without strain hardening FE without strain hardening
0 0
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 2.5 3.0
End rotation θ (deg) End rotation θ (deg)

(k) A4 Major axis (e0y = 75.1 mm) (l) A4 Minor axis (e0z = 29.8 mm)

Fig. 16 (continued).

developed FE models. The small differences in initial stiffness between The accuracy of the FE models has been further verified by comparing
the numerical and experimental results are attributed to contributions the numerically obtained ultimate load Nu,FE with the respective test
from non-explicitly modelled sources, such as small levels of friction values from compression tests of 22 semi-compact I-sections [3]. The
in the hinges, out-of-flatness of the endplates and unexpected eccen- comparisons between the test and FE results are presented in Table 6
tricities, while the generally slightly conservative FE predictions of ulti- for both stub columns and combined loading tests, where the definitions
mate strength (more notably for the specimens in compression and of the loading eccentricity e and angle α are shown in Fig. 17. The FE
compression plus major axis bending) may be due to variation in mate- models show excellent ability to predict the ultimate resistances of
rial stress-strain properties around the cross-sections, which is not fully hot-rolled steel semi-compact I-sections, with the mean numerical-to-
captured by the results of the tensile coupon tests that are performed on test ratio Nu,FE/Nu,test very close to unity and with small scatter. The com-
material extracted at discrete locations from within the cross-sections parisons between the test results and numerical predictions have led to
[32,33]. In particular, stress-strain curves obtained from tensile tests the conclusion that the developed FE models are accurate and reliable
on full structural sections were found to have consistently shorter for predicting the ultimate resistances of hot-rolled steel I-sections sub-
yield plateaus (and hence earlier initiation of strain hardening) than jected to combined loading and, thus, suitable for performing the com-
those obtained from coupon tests [32,33]. prehensive parametric studies presented in the companion paper.
 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 419

1200 1200

1000 1000

800 800

Load (kN)

Load (kN)
600 600

400 400

Test Test
200 200
FE FE
FE without strain hardening FE without strain hardening
0 0
0.0 0.5 1.0 1.5 2.0 2.5 0.0 1.0 2.0 3.0 4.0 5.0
End rotation θ (deg) End rotation θ (deg)

(m) B4 Major axis (e0y = 75.5 mm) (n) B4 Minor axis (e0z = 31.2 mm)

1000 1200
900
800 1000

700
800
600

Load (kN)
Load (kN)

500 600
400
300 400

200 Test Test

200 FE
FE
100
FE without strain hardening FE without strain hardening
0 0
0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 0.0 1.0 2.0 3.0 4.0 5.0
End rotation θ (deg) End rotation θ (deg)

(o) A5 Minor axis (e0z = 44.8 mm) (p) B5 Minor axis (e0z = 44.6 mm)

900 900

800 800

700 700

600 600
Load (kN)
Load (kN)

500 500

400 400

300 300

200 Test 200 Test

FE FE
100 100
FE without strain hardening FE without strain hardening
0 0
0.0 0.5 1.0 1.5 0.0 1.0 2.0 3.0 4.0 5.0
End rotation θ (deg) End rotation θ (deg)

(q) A6 Major axis (e0y = 46.0 mm) (r) A6 Minor axis (e0z = 50.6 mm)
1000 1000
900 900
800 800
700 700
Load (kN)

600 600
Load (kN)

500 500
400 400
300 300
200 Test 200 Test
FE FE
100 100
FE without strain hardening FE without strain hardening
0 0
0.0 0.3 0.6 0.9 1.2 1.5 0.0 1.0 2.0 3.0 4.0 5.0
End rotation θ (deg) End rotation θ (deg)

(s) B6 Major axis (e0y = 45.4 mm) (t) B6 Minor axis (e0z = 50.6 mm)

Fig. 16 (continued).
420 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421

Table 6
Comparison between numerical and experimental results for the Class 3 (semi-compact) I-sections.

Section and Specimen L Loading conditions Eccentricity e Angle α Nu,test Nu,FE Nu,FE/Nu,test
material
(mm) (mm) (deg) (kN) (kN)

HE 260 AA S235 sc_A1-2 900 Major axis bending plus compression 303.6 −0.3 585.2 544.1 0.93
sc_A1-3 301.7 0.3 812.4 798.4 0.98
sc_A2-1 Biaxial bending plus compression 301.8 9.5 556.9 545.1 0.98
sc_A2-2 298.1 11.0 554.3 543.3 0.98
sc_A3-1 299.9 40.9 404.5 398.7 0.99
sc_A3-2 298.2 39.6 396.6 405.8 1.02
sc_A4-1 Minor axis bending plus compression 98.9 90.1 826.8 857.0 1.04
sc_A4-2 97.8 90.4 824.8 865.2 1.05
sc_A10-2 95.8 89.6 853.4 879.9 1.03
sc_A5-1 Axial compression 0.0 0.0 2135.0 2068.3 0.97
sc_A5-2 0.0 0.0 2134.0 2068.3 0.97
HE 260 AA S355 sc_A7-1 900 Major axis bending plus compression 299.2 −0.2 809.6 836.0 1.03
sc_A7-2 298.5 −0.3 772.3 837.5 1.08
sc_A1-1 300.2 −0.1 768.5 833.9 1.08
sc_A8-1 Biaxial bending plus compression 298.8 10.5 790.7 800.7 1.01
sc_A8-2 298.6 11.4 769.9 794.7 1.03
sc_A9-1 299.4 39.7 559.1 608.4 1.09
sc_A9-2 299.1 39.8 602.4 608.4 1.01
sc_A10-1 Minor axis bending plus compression 99.3 89.7 1299.7 1288.8 0.99
sc_A10-3 99.4 90.1 1408.6 1361.4 0.97
sc_A11-1 Axial compression 0.0 0.0 3138.4 3164.2 1.00
sc_A11-2 0.0 0.0 3252.3 3164.2 0.97
Mean 1.01
COV 0.04

4. Conclusions design provisions for hot-rolled steel I-sections under combined loading
and to extend the scope of application of a more efficient deformation
A total of 14 experiments on hot-rolled steel I-sections, consisting of based design approach, the continuous strength method (CSM), to
two different material grades, subjected to either compression or com- such cases.
bined loading has been carried out and presented in this paper. The test
specimens were stocky (compacted) cross-sections and all were Class 1 Notation
according to EN 1993-1-1 [1]. The stockiness of the tested cross-sections
was such that material strain hardening had an influence on their cross- The following symbols are used in this paper:
sectional capacity and this influence is assessed in the companion paper.
Parallel numerical analysis of hot-rolled steel I-sections under different A cross-section area
loading conditions was performed using the finite element (FE) model- a flat width of the most slender constituent plate element in
ling program ABAQUS. The FE models were validated against the test re- the cross-section under compression
sults obtained in the present paper and from the literature [3]. It was B width of section
found that the FE models were capable of replicating accurately the d distance between the extreme fibres
structural behaviour of the test specimens, and are therefore suitable E Young's modulus
for performing parametric studies, as presented in the companion e distance between loading point and centroid of cross-section
paper. The combined experimental and numerical results, which cover em measured loading eccentricity
a range of cross-section slenderness, material grades and combinations e0 calculated loading eccentricity
of loading, are then used to assess the accuracy of the current codified fy yield stress
fy,aw weighted average material yield stress
fu ultimate tensile stress
H depth of section
Minor axis z-z I second moment of area
L length of specimen
Le effective length of specimen
Mel elastic moment capacity
Eccentric
Mpl plastic moment capacity
loading point
MT total bending moment
MT,u total bending moment at ultimate load
M1 Ne0 is the first order bending moment
M2 NΔ is the second order bending moment
N axial load
Major axis y-y Nu ultimate axial load
Ny Afy,aw is the yield load
ri web-flange internal corner radius
tf flange thickness
tw web thickness
α angle to define the position of eccentric loading
εpl
ln logarithmic plastic strain
Fig. 17. Definition of the loading eccentricity e and angle α in [12]. εm flexural strain due to bending moment
 X. Yun et al. / Journal of Constructional Steel Research 147 (2018) 408–421 421

εmax measured strain at the maximum compressive fibre [8] A.S. Hasham, K.J.R. Rasmussen, Section capacity of thin-walled I-section beam-
columns, J. Struct. Eng. ASCE 124 (4) (1998) 351–359.
εmin measured maximum tensile or minimum compressive strain [9] A.S. Hasham, K.J.R. Rasmussen, Interaction curves for locally buckled I-section beam-
at the other extreme fibre columns, J. Constr. Steel Res. 58 (2) (2002) 213–241.
εnom engineering strain [10] B. Kato, Rotation capacity of H-section members as determined by local buckling,
J. Constr. Steel Res. 13 (2) (1989) 95–109.
εsh strain hardening strain [11] B. Kato, Deformation capacity of steel structures, J. Constr. Steel Res. 17 (1) (1990)
εu strain at ultimate tensile stress 33–94.
εy fy/E is the yield strain [12] D. Beg, L. Hladnik, Slenderness limit of class 3 I cross-sections made of high strength
steel, J. Constr. Steel Res. 38 (3) (1996) 201–217.
σnom engineering stress [13] M. Seif, B.W. Schafer, Local buckling of structural steel shapes, J. Constr. Steel Res. 66
σtrue true stress (10) (2010) 1232–1247.
δu end shortening of stub columns at ultimate load [14] A. Foster, Stability and design of steel beams in the strain-hardening range(Ph.D. thesis)
Department of Civil and Environmental Engineering, Imperial College London, London,
κ curvature under bending moment
U.K, 2014.
θ end rotation [15] M.P. Byfield, D.A. Nethercot, An analysis of the true bending strength of steel beams,
θu end rotation at ultimate load Proc. Inst. Civ. Eng. Struct. Build. 128 (2) (1998) 188–197.
ωf measured maximum local geometric imperfection amplitudes [16] L. Gardner, D.A. Nethercot, Stainless steel structural design: a new approach, Struct.
Eng. 82 (21) (2004) 21–28.
from the flange faces of test specimens [17] S. Afshan, L. Gardner, Experimental study of cold-formed ferritic stainless steel hol-
ωw measured maximum local geometric imperfection amplitudes low sections, J. Struct. Eng. ASCE 139 (5) (2013) 717–728.
from the web face of test specimens [18] M. Ashraf, L. Gardner, D.A. Nethercot, Structural stainless steel design: resistance
based on deformation capacity, J. Struct. Eng. ASCE 134 (3) (2008) 402–411.
Δ lateral deflection at mid-height of specimens [19] X. Yun, L. Gardner, Stress-strain curves for hot-rolled steels, J. Constr. Steel Res. 133
Δu mid-height lateral deflection at ultimate load (2017) 36–46.
[20] X. Yun, L. Gardner, N. Boissonnade, The continuous strength method for the design
of hot-rolled steel cross-sections, Eng. Struct. 157 (2018) 179–191.
Suffixes ‘y’ and ‘z’ denote bending about the major and the minor [21] L. Gardner, X. Yun, L. Macorini, M. Kucukler, Hot-rolled steel and steel-concrete
axis, respectively, and Suffixes ‘test’ and ‘FE’ denote results obtained composite design incorporating strain hardening, Structure 9 (2017) 21–28.
from experiments and FE models, respectively. [22] I.S.O. 6892-1, Metallic materials, Tensile Testing Part 1: Method of Test at Ambient
Temperature, European Committee for Standardization (CEN), Brussels, 2009.
[23] R.D. Ziemian, Guide to Stability Design Criteria for Metal Structures, 6th ed. Wiley,
Acknowledgements Hoboken, NJ, 2010.
[24] Compression tests of stainless steel tubular columns, Tech Rep S770, Centre for Ad-
vanced Structural Engineering, University of Sydney, Australia, 1990.
The authors would like to thank Mr. Yanis Schaller, Mr. Jean-Paul
[25] X. Yun, L. Gardner, N. Boissonnade, Ultimate capacity of I-sections under combined
Andrey and Mr. Dominique Delaquis for their assistance in the loading – part 2: parametric studies and CSM design, J. Constr. Steel Res. (2018)
experimental programme. The financial support provided by the China (submitted).
Scholarship Council (CSC) for the first author's PhD study at Imperial [26] ABAQUS Version 6.13. Hibbit, Karlsson and Sorensen, Pawtucket, RI.
[27] M. Ashraf, L. Gardner, D.A. Nethercot, Finite element modelling of structural stain-
College London is gratefully acknowledged. less steel cross-sections, Thin-Walled Struct. 44 (10) (2006) 1048–1062.
[28] J. Becque, K.J.R. Rasmussen, Numerical investigation of the interaction of local and
References overall buckling of stainless steel I-columns, J. Struct. Eng. ASCE 135 (11) (2009)
1349–1356.
[1] EN 1993-1-1, Eurocode 3: Design of Steel Structures – Part 1–1: General Rules and [29] X. Yun, J. Nseir, L. Gardner, N. Boissonnade, Experimental and numerical investiga-
Rules for Buildings, European Committee for Standardization (CEN), Brussels, 2005. tion into the local imperfection sensitivity of hot-rolled steel I-sections, Proc., 7th
[2] ANSI/AISC 360-16, Specification for Structural Steel Buildings, American Institute of Int. Conf. on Coupled Instabilities in Metal Structures, Baltimore, Maryland, U.S,
Steel Construction (AISC), Chicago, Illinois, 2016. 2016.
[3] M. Kettler, Elastic-plastic Cross-sectional Resistance of Semi-compact H- and Hollow [30] E.N. 1993-1-5, Eurocode 3: Design of Steel Structures – Part 1–5: Plated Structural
Sections(PhD thesis) Faculty of Civil Engineering, Graz University of Technology, Elements, European Committee for Standardization (CEN), Brussels, 2006.
Graz, Austria, 2008. [31] Ultimate limit state calculation of sway framed with rigid joints, Technical Commit-
[4] N. Boissonnade, J.P. Jaspart, R. Oerder, K. Weynand, A new design model for the re- tee 8 of European Convention for Constructional Steelwork (ECCS), Tech. Rep., No.
sistance of steel semi-compact cross-sections, Proc., 5th European Conf. on Steel 33, 1984.
Structures, Eurosteel, Graz, Austria, 2008. [32] J. Wang, S. Afshan, N. Schillo, M. Theofanous, M. Feldmann, L. Gardner, Material
[5] A.J. Davids, G.J. Hancock, Compression tests of short welded I-sections, J. Struct. Eng. properties and compressive local buckling response of high strength steel square
ASCE 112 (5) (1986) 960–976. and rectangular hollow sections, Eng. Struct. 130 (2017) 297–315.
[6] W.X. Ren, Q.Y. Zeng, Interactive buckling behaviour and ultimate load of I-section [33] A.S.J. Foster, L. Gardner, Y. Wang, Practical strain-hardening material properties for
steel columns, J. Struct. Eng. ASCE 123 (9) (1997) 1210–1217. use in deformation-based structural steel design, Thin-Walled Struct. 92 (2015)
[7] M.A. Bradford, Inelastic local buckling of fabricated I-beams, J. Constr. Steel Res. 7 (5) 115–129.
(1987) 317–334.