Behavior and Design of Normal- and High-Strength

Steel SHS and RHS Columns

Xin Meng 1 and Leroy Gardner 2

Abstract: The behavior and design of hot-rolled and cold-formed steel square and rectangular hollow section (SHS and RHS) columns,
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

made of both normal- and high-strength material, are addressed in this paper. A series of experiments on hot-rolled high-strength steel SHS
columns was first conducted—six tests on S690 SHS 100 × 100 × 4 columns and six tests on S770 SHS 120 × 120 × 6.3 columns were
performed. Finite-element (FE) models were developed to replicate the experimental results and to carry out parametric studies to expand the
column buckling data pool. The accuracy of the European and North American buckling design rules for normal- and high-strength steel SHS
and RHS columns was evaluated through comparisons with the freshly generated test and FE results, as well as with existing test data
collected from the literature. Finally, a modified approach was proposed and statistically verified in accordance with existing standards;
the new approach features an imperfection factor that is a continuous function of yield strength, reflecting the reducing relative influence
of residual stresses and global imperfections with increasing steel grades. Improved consistency in resistance predictions over the existing
design provisions is demonstrated across a wide range of steel grades and relative slenderness values. DOI: 10.1061/(ASCE)ST.1943-
541X.0002728. © 2020 American Society of Civil Engineers.
Author keywords: Column buckling tests; Experiments; High-strength steels; Member stability; Numerical simulation; Rectangular
hollow sections (RHS); Square hollow sections (SHS); Testing.

Introduction and Education on Construction of Tubular structures (CIDECT)

projects have also been undertaken to explore the column buckling
Structural steel square and rectangular hollow section (SHS and behavior of SHS and RHS members, including those by Guiaux
RHS) elements are widely used in the construction industry as struc- (1972), Salvarinas et al. (1978), Braham et al. (1979), and Sedlacek
tural components due to their aesthetic appearance and excellent et al. (1996, 1999), in which a large number of buckling tests on
structural qualities, particularly under predominantly compressive both hot-rolled and cold-formed columns were reported. Extensive
loading. SHS and RHS products are commonly manufactured by experimental research has also been carried out on welded box sec-
means of hot-rolling (CEN 2006c), cold-forming (CEN 2006a), tion columns by, for example, Rasmussen and Hancock (1995),
or welding four plates into a box shape. With the increasing use Ban et al. (2013), and Schillo and Feldmann (2018), but these sec-
of high-strength steel SHS and RHS in construction, the develop- tions are outside the scope of the present study.
ment of suitable rules to ensure their safe and efficient design is Codified design rules for normal-strength steel columns have
imperative. been developed and used extensively over many decades, whereas
The flexural buckling behavior and design of hot-rolled and corresponding rules for high-strength steel columns are still rather
cold-formed SHS and RHS columns in normal- and high-strength limited. The current version of Eurocode 3 (EC3) includes provi-
steel grades are studied in the present paper. A number of previous sions for the design of columns in steel grades up to S700 (CEN
experimental investigations into the behavior of structural steel 2005, 2007), and the current version of AISC 360 (AISC 2016)
SHS and RHS members under axial compression have been carried is applicable to steel plates with nominal yield strengths up to
out. Laboratory testing on hot-rolled SHS and RHS columns has 690 N=mm2 , but to hollow sections only up to yield strengths of
been conducted by Rondal (1984) and Wang and Gardner (2017), 485 N=mm2 (AISC 2016; ASTM 2015, 2018). In general, the cur-
and buckling tests on cold-formed SHS and RHS compression rent codified design provisions for high-strength steel columns are
members have been performed and reported by Bjorhovde (1977), a simple extension to those for normal-strength steels, with scope
Rondal (1984), Key et al. (1988), Sully and Hancock (1996), for improvements in terms of accuracy and consistency.
Pavlovcic et al. (2012), SSAB (2014), Ma (2016), and Somodi and This study is therefore aimed at expanding the pool of column
Kövesdi (2017). Several Committee for International Development buckling test data on SHS members in the high-strength steel range
and developing improved design rules for both normal- and high-
1
Ph.D. Candidate, Dept. of Civil and Environmental Engineering, strength steel SHS and RHS members under axial compression. A
Imperial College London, London SW7 2AZ, UK (corresponding author). series of column buckling tests on hot-finished high-strength steel
Email: [email protected] SHS is firstly presented. Finite-element (FE) models are then estab-
2
Professor of Structural Engineering, Dept. of Civil and Environmental lished, validated, and used to generate supplementary numerical data.
Engineering, Imperial College London, London SW7 2AZ, UK. ORCID: The column design rules in EC3 and the US specification (AISC
https://orcid.org/0000-0003-0126-6807. Email: leroy.gardner@imperial
2016) are subsequently evaluated using the test and FE data and
.ac.uk
Note. This manuscript was submitted on November 22, 2019; approved found to exhibit some shortcomings. A modified EC3 design ap-
on March 2, 2020; published online on August 18, 2020. Discussion period proach in which the influence of yield strength is included in the im-
open until January 18, 2021; separate discussions must be submitted for perfection factor is then proposed to improve the accuracy and
individual papers. This paper is part of the Journal of Structural Engineer- consistency of the resistance predictions; recommendations are also
ing, © ASCE, ISSN 0733-9445. made to improve the AISC column design method. Finally, reliability

© ASCE 04020227-1 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 analysis of the current EC3 design approach and the proposed modi- are listed in Table 1. These material properties, which were deter-
fied EC3 approach is carried out. mined from flat tensile coupon tests, are adopted throughout this
study to represent the material behavior over the whole cross
section because the uniformity of properties for hot-rolled hollow
Experimental Investigation sections has been previously confirmed by Gardner et al. (2010)
and Wang et al. (2017).
General The key results from the stub column tests are summarized in
Table 2, where H, B, R, r, and t are the height, breadth, outer corner
An experimental program was conducted to investigate the flex- radius, inner corner radius, and thickness of the cross section, re-
ural buckling behavior of hot-finished high strength steel SHS spectively, L is the specimen length, c=tε is the EC3 local slender-
members under axial compression. Two quenched and tempered ness parameter, in which c is the width of the flat region of the
high strength steel SHS profiles in two different grades, namely cross-section faces, ε2 ¼ 235=f y , N u is the ultimate test load,
S690 SHS 100 × 100 × 4 (cross-section ID: 1) and S770 SHS and δu is the end shortening at N u .
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

120 × 120 × 6.3 (cross-section ID: 2), were examined. According

to the slenderness limits set out in the latest revision to Eurocode 3,
i.e., EN 1993-1-1:2018 (CEN 2018), the studied S690 SHS 100 × Column Buckling Tests
100 × 4 and S770 SHS 120 × 120 × 6.3 are Class 3=4 and Class 2 A total of 12 column specimens, six with S690 SHS 100 × 100 × 4
cross sections, respectively. In total, 12 SHS column specimens profiles (SC1 series) and six with S770 SHS 120 × 120 × 6.3 pro-
with varying lengths were tested. files (SC2 series), were tested to investigate their flexural buckling
responses under axial compression. The specimen lengths were var-
Material and Cross-Section Tests ied to cover a wide spectrum of EC3 relative slenderness values
λ̄EC3 [as defined by Eqs. (1) and (2) for Classes 1–3 and Class
Material tests and stub column tests on the two examined cross
4 cross sections, respectively, where N c;EC3 is the EC3 cross-
sections have been previously conducted and reported by Meng
sectional resistance to compression, A and Aeff are the gross
and Gardner (2020b); a brief summary of the results is provided
and effective cross-sectional areas, respectively, and N cr is the
herein. The engineering stress-strain curves from the tensile coupon
Euler buckling load] from 0.47 to 1.47. The measured geometric
tests are plotted in Fig. 1, and the obtained mechanical properties,
properties and the EC3 relative slenderness for each specimen are
including the Young’s modulus E, yield strength fy , ultimate
summarized in Table 3
strength fu , strain at the ultimate strength εu , and plastic strain at
sffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffi
fracture measured over the standard gauge length εf (CEN 2016),
N c;EC3 Af y
λ̄EC3 ¼ ¼ for Class 1–3 sections ð1Þ
N cr N cr
1000
S770 SHS 120×120×6.3 sffiffiffiffiffiffiffiffiffiffiffiffiffi sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
900 N c;EC3 Aeff fy
800 λ̄EC3 ¼ ¼ for Class 4 sections ð2Þ
N cr N cr
700
Stress (N/mm2)

600 All column buckling tests were conducted using an Instron

S690 SHS 100×100×4
500 2,000-kN hydraulic testing machine (Instron, Norwood, Massachu-
400 setts), as shown in Fig. 2. The column specimens were placed be-
300 tween a pair of wedge plates, with pin-ended boundary conditions
200 achieved through a pair of knife edges at the top and bottom ends.
100 The critical (effective) length Lcr of each specimen is equal to the
0 member length L plus an additional length for the knife edge and
0 2 4 6 8 10 12 wedge plate equal to 75 mm at each end, as summarized in Table 3.
Strain (%) Adjustable end clamps were employed at both ends to prevent any
movement of the specimens during loading. Stainless-steel shims
Fig. 1. Engineering stress-strain curves from tensile coupon tests re-
were inserted between the specimen ends and the wedge plates
ported by Meng and Gardner (2020b).
when necessary to eliminate any small gaps arising from the

Table 1. Average measured material properties from tensile coupon tests

Cross section E (N=mm2 ) fy (N=mm2 ) f u (N=mm2 ) εu (%) εf (%) f u =f y εu =εy
S690 SHS 100 × 100 × 4 211,000 787.3 825.8 6.2 17.1 1.049 16.6
S770 SHS 120 × 120 × 6.3 206,300 859.0 888.9 5.7 15.4 1.035 13.8
Source: Data from Meng and Gardner (2020b).

Table 2. Key results from stub column tests

Cross section H (mm) B (mm) R (mm) r (mm) t (mm) L (mm) c=tε N u (kN) δ u (mm)
S690 SHS 100 × 100 × 4 100.76 100.05 7.9 4.4 3.99 300.03 38.5 1,184.3 1.19
S770 SHS 120 × 120 × 6.3 119.00 118.62 9.5 3.8 6.29 360.20 30.0 2,492.8 3.38
Source: Data from Meng and Gardner (2020b).

© ASCE 04020227-2 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 Table 3. Measured geometric properties of test specimens and key results from column buckling tests
Cross section ID H (mm) B (mm) R (mm) r (mm) t (mm) L (mm) Lcr (mm) λ̄EC3 ωg (mm) N u (kN)
S690 SHS 100 × 100 × 4 SC1-1 100.49 100.48 8.4 4.3 4.04 802.0 952.0 0.47 0.98 1,148.1
SC1-2 100.13 100.94 7.9 4.3 4.01 1,199.0 1,349.0 0.67 1.37 1,053.6
SC1-3 100.17 100.70 8.6 4.5 4.02 1,599.8 1,749.8 0.87 1.77 1,071.1
SC1-4 99.53 101.34 7.8 4.6 4.02 2,000.3 2,150.3 1.08 2.06 807.0
SC1-5 100.98 100.24 8.4 4.6 4.03 2,397.4 2,547.4 1.26 2.51 668.4
SC1-6 100.40 100.47 8.3 4.7 4.02 2,800.0 2,950.0 1.47 2.95 498.1
S770 SHS 120 × 120 × 6.3 SC2-1 120.12 118.66 10.5 4.6 6.37 2,003.0 2,153.0 0.96 2.22 1,873.5
SC2-2 118.71 120.35 11.3 4.6 6.37 2,201.3 2,351.3 1.06 2.29 1,610.6
SC2-3 120.13 119.18 11.3 5.3 6.34 2,399.8 2,549.8 1.14 2.46 1,549.2
SC2-4 119.30 119.13 10.0 4.2 6.30 2,601.8 2,751.8 1.23 2.74 1,343.3
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

SC2-5 119.07 119.17 9.5 3.6 6.33 2,798.9 2,948.9 1.32 2.72 1,223.8
SC2-6 119.32 119.23 9.6 3.7 6.31 2,923.0 3,073.0 1.38 2.96 1,148.4

Lcr =1,000 was achieved for each test; this value is the typically
assumed global imperfection amplitude in the development of col-
Load cell umn buckling design rules (Beer and Schulz 1970; Ziemian 2010).
Prior to testing, each specimen was preloaded to about 10% of the
Inclinometer predicted failure load, enabling the global imperfection amplitude
ωg to be back-calculated using Eq. (3). The initial loading eccen-
tricity was then carefully adjusted, and the procedure repeated
Adjustable end clamps
until the resulting ωg was sufficiently close to the target value of
Lcr =1,000. The final values of ωg for all column tests are reported
Adjustable screw ram in Table 3
EIðεmax − εmin Þ
LVDT ωg ¼ −Δ ð3Þ
HN
where I = second moment of area about the axis of buckling; εmax
Strain gauges and εmin = average strains (from the pair of strain gauges) on the
concave and convex sides of each column, respectively; N = ap-
plied axial load; and Δ = measured midheight lateral deflection.
Column specimen The load-midheight lateral deflection curves obtained from the
tests are presented in Figs. 3 and 4 for the S690 SHS 100 × 100 × 4
Wedge plate
and S770 SHS 120 × 120 × 6.3 columns, respectively. The ulti-
mate loads N u are reported in Table 3, and all deformed test spec-
Knife edge
imens are displayed in Figs. 5 and 6. All tested specimens buckled
in the direction dictated by the global imperfections. For the shorter
three specimens in the SC1 series, i.e., SC1-1, SC1-2, and SC1-3,
failure was dominated by local buckling, whereas for the longer
specimens, i.e., SC1-4, SC1-5, and SC1-6, only global buckling
was evident at the ultimate load, with local buckling occurring
after the peak load under the combination of a lower level of axial
compression but an increasing second-order bending moment. The
Fig. 2. Column buckling test setup.
local buckling occurred in an abrupt manner for the SC1 series,
resulting in a sudden drop in applied load and a jump in the re-
corded deformations, as indicated by the dashed lines in Fig. 3.
out-of-flatness of the end sections. Four electrical resistance strain For the SC2 series, all specimens failed by global buckling, with
gauges were affixed at the midheight of the two opposite faces of local buckling forming gradually beyond the ultimate load, except
the test specimens parallel to the axis of buckling at a distance of 3t for Specimens SC2-5 and SC2-6, where the tests were terminated
from the edges for the measurement of the maximum and minimum before the onset of local buckling.
outer fiber strains. The lateral deflection at midheight and rotations
at the top and bottom ends of the columns were measured using a
LVDT and a pair of inclinometers, respectively. The test outputs, Numerical Simulations
including the load, machine displacement, and readings from the
strain gauges, LVDT, and inclinometers, were recorded and logged
at 0.5-s intervals using a proprietary in-house developed data logger General
and software. In addition to the laboratory testing, a numerical modeling program
The global imperfections ωg of the columns consisted of the out- was carried out to simulate the buckling behavior of SHS and RHS
of-straightness of the member and the initial loading eccentricity, members under axial compression. FE models were developed and
the latter of which was adjusted such that a total imperfection (out- validated against available column test results on the hot-rolled
of-straightness + initial loading eccentricity) as close as possible to and cold-formed SHS and RHS. Additional numerical results for

© ASCE 04020227-3 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 1400

1200

Load N (kN) 1000

800

600
SC1-1 SC1-2
400

200

0
0 2 4 6 8 10 12 14 16 18 20
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

Mid-height lateral deflection (mm)

1400

1200
SC1-3
1000
SC1-4
Load N (kN)

800
SC1-5
600
SC1-6
400

200

0
0 10 20 30 40 50 60 70 80 90 100
Mid-height lateral deflection (mm)

Fig. 3. Experimental load-midheight lateral deflection curves for S690

SHS 100 × 100 × 4 (SC1 series). Fig. 5. Failure modes of S690 SHS 100 × 100 × 4 (SC1 series, scale
in millimeters).

2000
SC2-1
1800 SC2-2
1600 SC2-3
1400 SC2-4
Load N (kN)

1200
1000
SC2-5
800
SC2-6
600
400
200
0
0 10 20 30 40 50 60 70 80 90 100
Mid-height lateral deflection (mm)

Fig. 4. Experimental load-midheight lateral deflection curves for S770

SHS 120 × 120 × 6.3 (SC2 series).

a range of steel grades and geometries were then generated from

parametric studies as an expansion of the data pool for SHS and
RHS column buckling.

Description of FE Models
The flexural buckling behavior of SHS and RHS members was si-
mulated by means of geometrically and materially nonlinear analy-
ses with imperfections (GMNIA) using the finite-element analysis
Fig. 6. Failure modes of S770 SHS 120 × 120 × 6.3 (SC2 series, scale
package Abaqus version 2016. The basic assumptions of the
in millimeters).
numerical modeling approach are descried herein. The four-noded

© ASCE 04020227-4 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 shell element with reduced integration S4R (Abaqus) was employed Validation
throughout the investigation, as commonly used in previous numeri-
To confirm the accuracy of the developed FE models, the numerical
cal studies on structural hollow sections, such as those reported by predictions were compared with the key results from experiments,
Yun and Gardner (2018), Meng et al. (2020), and Zhao et al. (2019). including the ultimate loads, load-deformation histories, and failure
In the flat regions of the modeled cross sections, a finer mesh of modes. The 12 column buckling tests on hot-finished SHS from
ðH þ BÞ=30 was applied within a length of 4H of the midheight this study and 11 tests on cold-formed SHS and RHS from a pre-
of the columns to capture the potential local buckling, and a coarser vious study by Ma (2016) were adopted for the validation study. A
mesh of ðH þ BÞ=15 was adopted for the remainder of the member local imperfection amplitude of ωl ¼ c= 200 was assumed, as rec-
length to reduce computational time; for each corner region, six el- ommended in EN 1993-1-5:2006 (CEN 2006b) and successfully
ements were employed. employed in several previous numerical studies, e.g., Yun et al.
Symmetry was also exploited by establishing quarter-models (2018) and Walport et al. (2019). The measured local imperfection
with suitable symmetry boundary conditions applied (Fig. 7) to fur- amplitudes (Ma et al. 2016; Meng and Gardner 2020b) for the con-
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

ther enhance the computational efficiency. The end section of the sidered SHS and RHS profiles were, on average, slightly smaller
models was coupled to a reference point through kinematic cou- than the value of c=200 recommended in EN 1993-1-5:2006 (CEN
pling, with boundary conditions applied to the reference point to 2006b), and thus, slightly conservative resistance predictions from
reflect the corresponding test end conditions. the FE models were anticipated.
Both local and global geometric imperfections were introduced A series of global imperfection amplitudes ωg , the measured val-
into the FE models in the form of the lowest local and global elastic ues, Lcr =500, Lcr =1,000, and Lcr =2,000 were incorporated into the
buckling shapes from a prior linear bifurcation analysis (LBA) per- FE models to assess their sensitivity to variation in global imper-
formed for each member. Residual stresses were not modeled fection levels. For the hot-rolled sections, the measured material
explicitly for either the hot-rolled or the cold-formed SHS and RHS properties were applied uniformly to the full modeled cross sec-
members. This is because for hot-rolled hollow sections, residual tions, whereas for the cold-formed sections, corner strength
stress magnitudes have been shown to be negligible compared with enhancements that arise due to plastic deformation during the roll-
the yield strength (Law and Gardner 2012; Wang et al. 2016), ing process were incorporated by assigning the measured corner
whereas for cold-formed hollow sections, the effect of the dominant material properties to the corner regions of the cross sections plus
bending residual stresses is already implicitly incorporated into the to an extension of 2t into the flat regions (Cruise and Gardner 2008;
stress-strain curves obtained from tensile tests performed on cou- Afshan et al. 2013). Deviation of the actual distribution of mechani-
pons cut from the examined profiles (Rasmussen and Hancock cal properties from this idealized pattern is a potential source of
1993; Jandera et al. 2008). It has also been confirmed in a number modeling error, although accurate results have been consistently
of previous numerical studies, such as those by Wang and Gardner obtained in previous numerical models of cold-formed SHS and
(2017), Huang and Young (2018), Meng and Gardner (2020a), and RHS developed on this basis (Yun and Gardner 2018; Ma et al.
Toffolon and Taras (2019), that FE models without the explicit in- 2019).
corporation of residual stresses are capable of accurately replicating Comparisons of the ultimate loads N u;FE derived from the FE
the experimentally observed physical responses of both hot-rolled models with those obtained from the experiments N u;test are given
and cold-formed steel tubular members. in Table 4. The FE models can be seen to be relatively sensitive to

2000
1800 Test data
1600 FE data
Finer mesh
Axial load N (kN)

1400
Coarser mesh 1200 SC2-3
1000
Planes of symmetry 800
600
400
Reference 200 SC1-5
Quarter-model
point 0
0 10 20 30 40 50 60 70 80
Mid-height lateral deflection (mm)
Kinematic coupling
Fig. 8. Comparisons of typical load-deformation curves from column
Fig. 7. Quarter-model for SHS and RHS columns. buckling tests and FE models (SC1-5 and SC2-3).

Table 4. Comparisons between column buckling test results and FE predictions with varying levels of global imperfections
Production Evaluation Measured ωg Lcr =500 Lcr =1,000 Lcr =2,000
Source of data route parameter (N u;FE =N u;test ) (N u;FE =N u;test ) (N u;FE =N u;test ) ðN u;FE =N u;test Þ
This study (12 tests) Hot-rolled Mean 0.976 0.909 0.975 1.020
COV 0.033 0.037 0.033 0.037
Ma (2016) (11 tests) Cold-formed Mean 0.981 0.928 0.955 0.960
COV 0.040 0.071 0.061 0.050

© ASCE 04020227-5 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 values of f y and fu from EN 1993-1-1:2018 (CEN 2018) were
used for steel grades up to S700, whereas for S900, f u was taken
as 1.05fy according to the ductility requirements given in EN 1993-
1-12:2007 (CEN 2007).
For the cold-formed SHS and RHS, the predictive models for
corner strength enhancements by Rossi et al. (2013) were used in
conjunction with the stress-strain models of Gardner and Yun
(2018), and the resulting corner stress-strain relationships were as-
signed to the corner regions of the cross section plus to an extension
of 2t into the adjacent flat regions. Two sets of cross-section outer
dimensions, 100 × 100 and 100 × 50 [H × B (mm)], were exam-
ined, with the thickness t varied to achieve a spectrum of local slen-
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

derness within the EC3 Class 1–3 domain. The modeled member
lengths were varied from 160 to 6000 mm, leading to a range of
global slenderness values λ̄ from 0.16 to 2.62.
A total of 2,300 numerical results on hot-rolled and cold-formed
SHS and RHS columns, covering normal- to high-strength steel
members, were generated. These numerical data are combined with
existing experimental data and used in the next section for the as-
sessment of existing and development of new design provisions for
hot-rolled and cold-formed steel SHS and RHS columns.

Stability Design of SHS and RHS columns

Fig. 9. Comparisons of typical failure modes from column buckling
tests and FE models (SC1-5). General
The numerical results generated in the previous section have been
supplemented with experimental results collected from the litera-
ture. In total, 588 existing test data have been collected, with 314
the global imperfection amplitude; when the measured values of
ωg were utilized, excellent accuracy and consistency in the FE pre- falling into the Class 1–3 range in accordance with EN 1993-1-
dictions were achieved, whereas with ωg taken as Lcr =1,000, the 1:2018 (CEN 2018). The freshly generated numerical column
predictions were accurate but slightly conservative. Excellent buckling data, combined with the collected test results for members
agreement between the test and numerical load-deformation histories with nonslender cross sections, are plotted against the codified col-
and failure modes was also observed, as shown in Figs. 8 and 9. umn buckling curves in Fig. 10. A relatively high level of scatter is
Overall, it can be concluded that the developed FE models are observed in the test data points, which is attributed largely to the
capable of accurately replicating the buckling behavior of hot-rolled sensitivity of the column buckling resistance to the inherently scat-
and cold-formed SHS and RHS members under axial compression. tered global imperfections and initial loading eccentricities, as well
as variations in the test conditions.
An assessment of the accuracy of the codified column buckling
Parametric Studies design provisions in EN 1993-1-1:2018 (CEN 2018) and ANSI/
Following the validation study, the developed FE models were AISC 360-16 (AISC 2016) for hot-rolled and cold-formed SHS
used in parametric studies for the generation of additional column and RHS with varying steel grades is firstly conducted using
buckling data on hot-rolled and cold-formed SHS and RHS over a the established test and FE database, where ANSI is the acronym
wider range of steel grades, cross-section geometries, and member for the American National Standards Institute. To address the
lengths. Five steel grades from normal to high strengths, namely shortcomings revealed in both design codes, a modified EC3 de-
S355, S460, S500, S690=S700, and S900, were considered for both sign approach is developed, in which the imperfection factor is
hot-rolled and cold-formed sections. The material models devel- defined as a continuous function of yield strength, leading to im-
oped by Yun and Gardner (2017) and Gardner and Yun (2018) were proved accuracy and consistency; recommendations for improve-
employed for the description of the respective stress-strain curves; ments to the AISC column buckling design approach are also
the adopted input parameters are listed in Table 5. The nominal made. Reliability analysis of the current EC3 and proposed

Table 5. Key material properties used in parametric studies

Hot-rolled SHS and RHS Cold-formed SHS and RHS
Grade E (N=mm2 ) fy (N=mm2 ) fu (N=mm2 ) E (N=mm2 ) f y (N=mm2 ) f u (N=mm2 )
S355 210,000 355 490 210,000 355 430
S460 210,000 460 540 210,000 460 520
S550 210,000 550 600 210,000 550 600
S690=S700 210,000 690 770 210,000 700 750
S900 210,000 900 945 210,000 900 945

© ASCE 04020227-6 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 1.2
FE data
This study
1 Wang & Gardner (2017)
Sedlacek et al. (1999)
0.8 EC3 (curve a)
EC3 (curve a 0 )

Nu / Npl
0.6 AISC

0.4

0.2

0
0 0.5 1 1.5 2 2.5
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

(a)
1.2
FE data
Ma (2016)
1 Sully & Hancock (1996)
Key et al. (1986)
0.8 Salvarinas et al. (1978)
Bjorhovde (1977)
Nu / Npl

0.6 Guiaux (1972)

EC3 (curve c)
0.4 AISC

0.2

0
0 0.5 1 1.5 2 2.5
(b)

Fig. 10. Comparisons of column buckling test and FE data with codified column buckling curves for (a) hot-rolled SHS and RHS; and (b) cold-
formed SHS and RHS.

modified EC3 column design approaches is then conducted to de- into Eq. (6), Eq. (9) is determined (Boissonnade et al. 2006) as
termine a suitable partial safety factor. given by Eq. (10)
sffiffiffiffiffiffiffiffi
Af y
Background of Ayrton-Perry Formulation λ̄ ¼ ð7Þ
N cr
The background to the Ayrton-Perry formulation (Ayrton and Perry
1886) to describe the buckling behavior of imperfect columns is N b;R
χ¼ ð8Þ
first presented. The derivation is made based on the following as- Af y
sumptions: (1) a simply-supported prismatic member under axial
compression; (2) an initial geometric imperfection in the form A
ð1 − χÞð1 − χλ̄2 Þ ¼ e0 χ ¼ ηχ ð9Þ
of a half sine wave with a maximum value e0 at midheight; W el
(3) a nonslender cross section; and (4) a first-yield failure criterion.
For a member under an axial force N, the maximum total lateral A
η ¼ e0 ð10Þ
deflection at midheight vmax (equal to the initial geometric imper- W el
fection amplitude e0 plus the additional midheight lateral deflection
where N b;R = column buckling resistance based on the assumption
Δ) is given by Eq. (4), where N cr is the Euler buckling load, as
of failure at the point of first yield; and η = generalized initial im-
given by Eq. (5). The load corresponding to first-yield of the
perfection factor.
member N b;R can be expressed by Eq. (6)
Finally, the column buckling reduction factor χ can be derived
1 from Eq. (9), as given by Eqs. (11) and (12), with the theoretical
vmax ¼ e ð4Þ column buckling resistance being N b;R ¼ χAf y
1 − N=N cr 0
1
π2 EI χ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð11Þ
N cr ¼ 2 ð5Þ Φ þ Φ2 − λ̄2
Lcr
Φ ¼ 0.5ð1 þ η þ λ̄2 Þ ð12Þ
N b;R 1 N b;R e0
þ ¼ fy ð6Þ
A 1 − N b;R =N cr W el
Eurocode 3—EN 1993-1-1:2018
where A = cross-sectional area; and W el = elastic section modulus.
By substituting the relative slenderness λ̄ [as defined by Eq. (7)] The EC3 column buckling design curves originate from the Ayrton-
and the column buckling reduction factor χ [as defined by Eq. (8)] Perry formulation and have been calibrated against experimental

© ASCE 04020227-7 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 Table 6. Values of imperfection factor α for hollow sections in EN Φ ¼ 0.5ð1 þ ηEC3 þ λ̄2EC3 Þ ¼ 0.5ð1 þ αðλ̄EC3 − 0.2Þ þ λ̄2EC3 Þ
1993-1-1:2018
ð15Þ
Cross-section type α (S235-S420) α (S460-S700)
Hot-rolled hollow sections 0.21 (curve a) 0.13 (curve a0 ) Finally, the design column buckling resistance N b;Rd , as ex-
Cold-formed hollow sections 0.49 (curve c) 0.49 (curve c) pressed by Eq. (16), is calculated as the product of the EC3 column
buckling reduction factor χEC3 and the EC3 cross-section resistance
N c;EC3 , to give N b;EC3 ¼ χEC3 N c;EC3 , divided by the partial safety
and numerical data for different types of cross section (Beer and factor γ M1, which has a recommended value of 1.0 in EN 1993-1-
Schulz 1970; Starting and Vos 1973; Sfintesco and Carpena 1977; 1:2018 (CEN 2018)
Chan et al. 2010; Taras and Greiner 2010). The EC3 relative slen-
derness λ̄EC3 is defined by Eqs. (1) and (2) for Classes 1–3 and N b;EC3 χEC3 N c;EC3
Class 4 sections, respectively, where Aeff is the effective area of N b;Rd ¼ ¼ ð16Þ
γ M1 γ M1
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

the cross section, allowing for local buckling and determined in

accordance with EN 1993-1-5:2006 (CEN 2006b). To assess the accuracy of the EC3 column buckling design pro-
The EC3 column buckling reduction factor χEC3 is deter- visions for SHS and RHS members, the ultimate loads from the
mined from the modified Ayrton-Perry formula, as given by column buckling tests and FE simulations N u are normalized by
Eqs. (13)–(15), where the EC3 generalized initial imperfection the unfactored resistance predictions from EC3 N b;EC3 and plotted
factor ηEC3 given by Eq. (14) is adopted to account for the com- against the relative slenderness λ̄ in Fig. 11. Table 7 summarizes the
bined effects of initial out-of-straightness and residual stresses mean values and coefficients of variation (COV) of the ratio
and α is the imperfection factor, as tabulated in Table 6 for dif- N u= N b;EC3 . The yield strength f y for the cold-formed SHS and
ferent production routes and steel grades RHS is taken as the weighted average yield strength incorporating
the strength enhancements in the corner region, as given by
1 Eq. (17), where fy;f and fy;c are the yield strengths of the flat and
χEC3 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi but χEC3 ≤ 1.0 ð13Þ
Φ þ Φ2 − λ̄2EC3 corner regions respectively, and Af an Ac are the total areas of the
flat and corner regions, respectively

f y;f Af þ fy;c Ac
ηEC3 ¼ αðλ̄EC3 − 0.2Þ ð14Þ fy ¼ for cold-formed SHS and RHS ð17Þ
A

1.6
FE data (S355)
1.5
FE (S460–S900)
1.4 Test data ( fy 460 N/mm2)
1.3
Nu / Nb,EC3

1.2
1.1
1
0.9
0.8
0 0.5 1 1.5 2 2.5
(a)

1.6
FE data (S355)
1.5 FE (S460–S900)
1.4 Test data ( fy 460 N/mm2 )
Test data ( fy 460 N/mm2)
1.3
Nu / Nb,EC3

1.2
1.1
1
0.9
0.8
0 0.5 1 1.5 2 2.5
(b)

Fig. 11. Comparisons of column buckling test and FE resistances with unfactored resistance predictions from EN 1993-1-1:2018 for (a) hot-rolled
SHS and RHS; and (b) cold-formed SHS and RHS.

© ASCE 04020227-8 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 Table 7. Comparisons of column buckling test and FE results with predicted resistances for hot-rolled and cold-formed SHS and RHS
Evaluation EN1993-1-1:2018 Modified EC3 ANSI/AISC 360-16
Cross-section type parameter (N u =N b;EC3 ) (N u =N b;mod−EC3 ) (N u =N b;AISC )
Hot-rolled SHS and RHS Mean 1.029 1.053 1.074
(96 tests þ1,150 simulations) COV 0.037 0.040 0.052
Cold-formed SHS and RHS Mean 1.116 1.084 0.951
(108 tests þ1,150 simulations) COV 0.064 0.051 0.122

For the hot-rolled SHS and RHS, the EC3 design approach gen- αmod ¼ 0.24ε for hot-rolled SHS and RHS ð19Þ
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

erally yields accurate column buckling resistance predictions,

although a relatively wide scatter band is observed because the in- αmod ¼ 0.56ε for cold-formed SHS and RHS ð20Þ
fluence of the yield strength on the column buckling behavior is not
fully captured. For the cold-formed SHS and RHS, the EC3 pre- On the other hand, to address the optimistic EC3 buckling
dictions generally fall on the safe side but exhibit an increasing strength predictions in the lower slenderness range, a shorter pla-
level of conservatism with increasing yield strength, owing again teau length of λ̄mod-EC3;0 ¼ 0.1 is proposed to supersede the current
to the influence of yield strength on the buckling behavior not being EC3 plateau length of λ̄EC3;0 ¼ 0.2, resulting in slightly lower re-
fully captured. Furthermore, a number of data points appear on the sistance predictions for stockier columns. The modified EC3 gen-
unsafe side for λ̄ < 0.7 for both hot-rolled and cold-formed SHS eralized initial imperfection factor ηmod−EC3, incorporating both the
and RHS; this stems, to a large extent, from the fixed plateau length proposed modified imperfection factor and shortened plateau is ex-
of λ̄EC3;0 ¼ 0.2 that features in the current EC3 column buckling pressed by Eq. (21)
curves.
ηmod-EC3 ¼ αmod ðλ̄EC3 − 0.1Þ ð21Þ
Modified EC3 Approach
Finally, the modified EC3 column buckling reduction factor
Residual stresses and global imperfections play an important role in χmod-EC3 is given by Eqs. (22) and (23)
the buckling behavior of compressed steel members. Residual
stresses have been shown to be approximately constant regardless 1
χmod-EC3 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi but χmod-EC3 ≤ 1.0 ð22Þ
of the steel grade (ECCS 1984), and hence reduce as a proportion of
Φmod þ Φ2mod − λ̄2EC3
the yield strength for higher grades. In terms of global imperfec-
tions, the original Ayrton-Perry formulation can be utilized to ex-
amine their influence on the buckling strength of columns in Φmod ¼ 0.5ð1 þ ηmod-EC3 þ λ̄2EC3 Þ
different grades. Considering a pin-ended column with an initial
bow imperfection of amplitude e0 ¼ L=1,000 and no residual stress, ¼ 0.5ð1 þ αmod ðλ̄EC3 − 0.1Þ þ λ̄2EC3 Þ ð23Þ
the generalized initial imperfection factor η can be rearranged into
Eq. (18), indicating that η is proportional to the product of the rel- The accuracy of the proposed modified EC3 design approach
ative slenderness λ̄ and f−1=2 (or, equivalently to the product of λ̄ for hot-rolled and cold-formed SHS and RHS columns is now as-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi y
and ε ¼ 235=f y ) for a given cross-section profile. This means that sessed using the obtained experimental and numerical data. Com-
as the yield strength increases, the generalized initial imperfection parisons of the test and numerical results N u with the predictions
factor becomes smaller for a given level of relative slenderness, cor- from the proposed approach N b;mod−EC3 are presented in a normal-
responding to a less detrimental effect from global imperfections on ized manner in Fig. 12, with the statistical results summarized in
the column buckling curve. Overall, therefore, a higher normalized Table 7.
column buckling resistance is anticipated with increasing yield The data points associated with the various steel grades now
strength follow a significantly tighter trend for both hot-rolled and cold-
formed SHS and RHS columns, showing that the influence of yield
strength on the column buckling behavior is successfully captured
 pffiffiffiffiffiffiffiffiffi
L A π EIA λ̄ by the modified imperfection factor αmod. Meanwhile, owing to the
η¼ ¼ pffiffiffiffiffi adoption of a shorter plateau length, an improved level of accuracy
1,000 W el 1,000 W el fy
 pffiffiffiffiffiffiffiffiffi over the EC3 approach is achieved in the lower slenderness range.
π EIA The improvements in accuracy and consistency over the current
¼ pffiffiffiffiffiffiffiffi λ̄ε ð18Þ EC3 approach are further revealed quantitatively by the statistical
1,000 235 W el
results presented in Table 7, particularly for the cold-formed SHS
and RHS columns.
However, in the current EC3 design approach, these beneficial
effects are only partially captured for hot-rolled SHS and RHS in a
US Specification ANSI/AISC 360-16
stepwise manner by applying different imperfection factors α for
steel grades above and below S460, and they are ignored for cold- The design rules for SHS and RHS compression members in
formed SHS and RHS. In light of this, a modified imperfection factor the US specification ANSI/AISC 360-16 (AISC 2016) are con-
αmod , incorporating ε, is proposed to take due account of the influ- sidered in this section, with the notation adjusted for consistency
ence of yield strength on the column buckling behavior in a con- with EC3. Unlike the EC3 relative slenderness that features the
sistent and continuous manner, as given by Eqs. (19) and (20) for effective cross-sectional area for Class 4 cross sections, the AISC
hot-rolled and cold-formed SHS and RHS, respectively relative slenderness λ̄AISC is defined based on the gross cross-

© ASCE 04020227-9 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 1.6
FE data (S355)
1.5
FE (S460–S900)
1.4 Test data ( fy 460 N/mm2)

Nu / Nb,mod-EC3
1.3
1.2
1.1
1
0.9
0.8
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

0 0.5 1 1.5 2 2.5

(a)
1.6
FE data (S355)
1.5 FE (S460–S900)
1.4 Test data ( fy 460 N/mm2)
Test data ( fy 460 N/mm2)
Nu / Nb,mod-EC3

1.3
1.2
1.1
1
0.9
0.8
0 0.5 1 1.5 2 2.5
(b)

Fig. 12. Comparisons of column buckling test and FE resistances with unfactored resistance predictions from proposed modified EC3 approach for
(a) hot-rolled SHS and RHS; and (b) cold-formed SHS and RHS.

section properties regardless of the local slenderness, as given by N b;Rd ¼ ϕc N b;AISC ¼ ϕc χAISC fy A for nonslender SHS and RHS
Eq. (24) ð27Þ

sffiffiffiffiffiffiffiffi rffiffiffiffiffi N b;Rd ¼ ϕc N b;AISC ¼ ϕc χAISC fy Aeff for slender SHS and RHS
Af y Lcr 1 fy
λ̄AISC ¼ ¼ ð24Þ ð28Þ
N cr i π E
The accuracy of the AISC column buckling design curve is
now evaluated through comparisons of the test and FE data
The AISC column buckling reduction factor χAISC is deter- N u with the unfactored AISC resistance predictions N b;AISC , as
mined from a single two-stage column buckling curve for all given in Fig. 13 and Table 7. The AISC buckling curve generally
production routes and steel grades, as described by Eqs. (25) provides a lower bound to the test and FE data points for the hot-
and (26) rolled SHS and RHS columns, whereas for the cold-formed com-
pression members, a large number of data points appear on the
unsafe side. In general, the shape of the AISC column buckling
2
χAISC ¼ 0.658λ̄AISC for λ̄AISC ≤ 1.5 ð25Þ curve matches less well with the trend of the column buckling
data compared with the Ayrton-Perry formulation used in
EC3, leading to relatively high values of COV in Table 7 for both
hot-rolled and cold-formed SHS and RHS columns. Furthermore,
the influence of yield strength on the column buckling curve is
0.877 not considered in the AISC approach, leading to reduced accu-
χAISC ¼ for λ̄AISC > 1.5 ð26Þ
λ̄2AISC racy and increased scatter when high-strength steel grades are
considered.
The following recommendations are therefore made herein for
The design buckling resistance under axial compression N b;Rd improvements to the AISC column design approach: (1) different
can then be derived from Eqs. (27) and (28) for nonslender and buckling curves should be used for hot-rolled and cold-formed col-
slender SHS and RHS, respectively, where ϕc ¼ 0.9 is the resis- umns to account for their distinct buckling responses arising from
tance factor for compression and Aeff is the effective cross- the different stress-strain characteristics and residual stress distri-
sectional area calculated in accordance with ANSI/AISC 360-16 butions, and (2) the influence of yield strength on the column buck-
(AISC 2016) ling curve should be duly incorporated.

© ASCE 04020227-10 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 1.6
FE data (S355)
1.5 FE (S460–S900)
1.4 Test data ( fy 460 N/mm 2)

Nu / Nb,AISC
1.3
1.2
1.1
1
0.9
0.8
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

0 0.5 1 1.5 2 2.5

(a)
1.5
FE data (S355)
1.4 FE (S460–S900)
1.3 Test data ( fy 460 N/mm2)
Test data ( fy 460 N/mm2)
Nu / Nb,AISC

1.2
1.1
1
0.9
0.8
0.7
0 0.5 1 1.5 2 2.5
(b)

Fig. 13. Comparisons of column buckling test and FE resistances with unfactored resistance predictions from ANSI/AISC 360-16 for (a) hot-rolled
SHS and RHS; and (b) cold-formed SHS and RHS.

Reliability Analysis influence from the variability of f y , A, and E, can then be deter-
mined from Eq. (30) for each test or simulation
The reliability of the buckling design rules in prEN 1993-1-1:2018
(CEN 2018) and the proposed modified EC3 approach for normal-
and high-strength steel SHS and RHS columns is evaluated in this N b;R ¼ kfcy Ad Ee ð29Þ
section following the standard procedure set out in EN 1990:2002
(CEN 2002). As provided in Annex E of prEN 1993-1-1:2018 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
(CEN 2018), the material overstrength fy;m =fy;n (i.e., the ratio be- V rt ¼ ðcV fy Þ2 þ ðdV A Þ2 þ ðeV E Þ2 ð30Þ
tween the mean and nominal yield strengths) is taken as 1.2, 1.15,
and 1.1 for steel grades below S460, equal to S460, and above
S460, respectively, with the corresponding COV of yield strength The design fractile factor kd;n, the COV of the test and FE re-
V fy equal to 0.05, 0.04, and 0.035, respectively. The COV of the sistances relative to the predictions from the resistance model V δ ,
Young’s modulus V E is taken equal to 0.03 (CEN 2018), and the and the combined COV incorporating the variability of the resis-
COV of the cross-sectional area V A is taken as 0.026 for SHS and tance model and the basic variables V r were calculated in accor-
RHS, as calculated based on the statistical parameters of the geo- dance with EN 1990:2002 (CEN 2002). The correction factor b
metric dimensions provided in Annex E of prEN 1993-1-1:2018 was calculated using Eq. (31) instead of the least-squares method
(CEN 2018) and following the procedure described by Afshan et al. recommended in EN 1990:2002 (CEN 2002). This approach pre-
(2015). The mean values of the Young’s modulus E and all the geo- vents the resulting value of b from being biased toward the test
metric properties are taken equal to the nominal values except for and FE specimens with higher resistance values (Meng et al.
the thickness t, which is taken equal to 0.99 times the nominal 2020)
thickness of the SHS and RHS according to Annex E of prEN
1993-1-1:2018 (CEN 2018).
The column buckling design function can be expressed in terms
Table 8. Reliability analysis results for current EC3 (prEN 1993-1-1:2018)
of the Young’s modulus E, yield strength f y , and the cross- approach
sectional area A, as given by Eq. (29), where the dependency of
the resistance function on the variables (i.e., the values of the ex- Cross-section type n kd;n b Vδ Vr γ M1
ponents c, d, and e) can be derived for each individual test or Hot-rolled SHS 1,123 3.098 1.030 0.038 0.064 1.086
numerical simulation following the method set out by Afshan et al. and RHS
(2015). The combined coefficient of variation of the material and Cold-formed SHS 1,128 3.098 1.115 0.068 0.084 1.054
and RHS
geometric basic variables V rt , incorporating the varying degrees of

© ASCE 04020227-11 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 Table 9. Reliability analysis results for modified EC3 approach confirming that the EC3 recommended partial safety factor of unity
Cross-section type n kd;n b Vδ Vr γ M1 is suitable for use with the proposed approach.
Hot-rolled SHS and RHS 1,123 3.098 1.056 0.041 0.065 1.063
Cold-formed SHS and RHS 1,128 3.098 1.085 0.054 0.072 1.041
Acknowledgments

The authors gratefully acknowledge the Skempton scholarship

from Imperial College London for the financial support. The
1X n
re;i authors would also like to thank Trevor Stickland and Soh Harn
b¼ ð31Þ
n i¼1 rt;i Chong for their assistance during the experiments.

where n = number of test and FE data; re = experimental resistance;

Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

and rt = theoretical resistance determined from the resistance References

model.
Afshan, S., P. Francis, N. R. Baddoo, and L. Gardner. 2015. “Reliability
The test and FE data were divided by relative slenderness into
analysis of structural stainless steel design provisions.” J. Constr. Steel
three groups: λ̄ ≤ 0.75, 0.75 < λ̄ ≤ 1.25, and 1.25 < λ̄ ≤ 2; those
Res. 114 (Nov): 293–304. https://doi.org/10.1016/j.jcsr.2015.08.012.
with λ̄ > 2 are deemed beyond the practical range and were there- Afshan, S., B. Rossi, and L. Gardner. 2013. “Strength enhancements in
fore not considered in the reliability analysis. The key reliability cold-formed structural sections—Part I: Material testing.” J. Constr.
analysis results are reported in Tables 8 and 9, where γ M1 is the Steel Res. 83 (Apr): 177–188. https://doi.org/10.1016/j.jcsr.2012.12
required value of partial safety factor for member buckling design. .008.
Values of b, V δ , and V r in Tables 8 and 9 were obtained based on AISC (American Institute of Steel Construction). 2016. Specification for
the entire set of data for illustration purposes. The required partial structural steel buildings. ANSI/AISC 360-16. Chicago: AISC.
safety factors for the proposed approach are equal to 1.063 and ASTM. 2015. Standard specification for cold-formed welded carbon
1.041 for hot-rolled and cold-formed SHS and RHS, respectively, steel hollow structural sections (HSS). ASTM A1085/A1085M. West
which are slightly higher than the target value of 1.0 but lower than Conshohocken, PA: ASTM.
the corresponding values of 1.086 and 1.054 for the current EC3 ASTM. 2018. Standard specification for high-yield-strength, quenched
approach. Therefore, the EC3 recommended partial safety factor and tempered alloy steel plate, suitable for welding. ASTM A514/
γ M1 for member buckling design is deemed applicable to the pro- A514M–18e1. West Conshohocken, PA: ASTM.
Ayrton, W. E., and J. Perry. 1886. “On struts.” The Engineer 62: 464–465,
posed modified EC3 approach, particularly in light of the reliability
513–515.
recommendations given by Simões da Silva et al. (2017).
Ban, H., G. Shi, Y. Shi, and M. A. Bradford. 2013. “Experimental inves-
tigation of the overall buckling behaviour of 960 MPa high strength
steel columns.” J. Constr. Steel Res. 88: 256–266. https://doi.org/10
Conclusions .1016/j.jcsr.2013.05.015.
Beer, H., and G. Schulz. 1970. “Bases théoriques des courbes européennes
A comprehensive study into the buckling behavior of structural de flambement.” [In French.] Construction Métallique, 3: 5–12.
steel SHS and RHS columns has been presented in this paper. Bjorhovde, R. 1977. Strength and behavior of cold-formed HSS columns.
A total of 12 hot-finished S690 and S770 SHS columns covering Rep. No. 65. Edmonton, AB, Canada: Univ. of Alberta.
a wide spectrum of relative slenderness values from 0.47 to 1.47 Boissonnade, N., R. Geriner, J. P. Jaspart, and J. Lindner. 2006. Rules for
were tested with pin-ended boundary conditions. Finite-element member stability in EN 1993-1-1: Background documentation and de-
models were then established and validated against the test results sign guidelines, Guide to stability design criteria for metal structures.
generated herein on hot-finished SHS columns and existing test re- Brussels, Belgium: European Convention for Constructional Steelwork.
sults on cold-formed SHS and RHS columns; excellent agreements Braham, M., J. P. Grimault, and J. Rondal. 1979. Flambement des profils
creux á parois minces, cas des profils rectangulaires chargés axiale-
between the test and numerical responses were achieved. Sub-
ment. CIDECT 2H 79/19. Altendorf, Switzerland: Committee for
sequent parametric studies provided over 2,000 column buckling
International Development and Education on Construction of Tubular
resistance data over a wider range of steel grades and cross-section Structures.
and member geometries. The generated experimental and numeri- CEN (European Committee for Standardization). 2002. Eurocode—Basis
cal data, combined with further existing test results collected from of structural design. EN 1990:2002. Brussels, Belgium: CEN.
the literature, were utilized to assess current codified design rules for CEN (European Committee for Standardization). 2005. Eurocode 3:
hot-rolled and cold-formed SHS and RHS columns. The Ayrton- Design of steel structures—Part 1-1: General rules and rules for build-
Perry-based formula specified in prEN 1993-1-1:2018 was shown ings. EN 1993-1-1:2005. Brussels, Belgium: CEN.
to provide accurate buckling strength predictions for both hot-rolled CEN (European Committee for Standardization). 2006a. Cold formed
and cold-formed SHS and RHS columns, although with some scope welded structural hollow sections of non-alloy and fine grain steels—
for improvement. A modified EC3 approach, featuring a modified Part 2: Tolerances, dimensions and sectional properties. EN 10219-
imperfection factor linked to the yield strength of the material 2:2006. Brussels, Belgium: CEN.
and a shortened plateau length, has been developed and shown to CEN (European Committee for Standardization). 2006b. Eurocode 3—
provide more consistent resistance predictions than the current EC3 Design of steel structures—Part 1-5: Plated structural elements. EN
1993-1-5. Brussels, Belgium: CEN.
provisions.
CEN (European Committee for Standardization). 2006c. Hot finished struc-
The column design rules in the US specification AISC 360-16
tural hollow sections of non-alloy and fine grain steels—Part 2:
were also examined; the use of different column buckling curves Tolerances, dimensions and sectional properties. EN 10210-2.
for hot-rolled and cold-formed sections and for account to be taken Brussels, Belgium: CEN.
of the improved normalized response of columns with increasing CEN (European Committee for Standardization). 2007. Eurocode 3—
yield strength was proposed. Finally, reliability analysis was con- Design of steel structures—Part 1-12: Additional rules for the extension
ducted for the current EC3 and proposed modified EC3 column of EN 1993 up to steel grades S 700. EN 1993-1-12. Brussels, Belgium:
buckling design approaches in accordance with EN 1990:2002, CEN.

© ASCE 04020227-12 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 CEN (European Committee for Standardization). 2016. Metallic materials— Rasmussen, K. J. R., and G. J. Hancock. 1995. “Tests of high strength steel
Tensile testing Part 1: Method of test at room temperature. EN ISO columns.” J. Constr. Steel Res. 34 (1): 27–52. https://doi.org/10.1016
6892-1. Brussels, Belgium: CEN. /0143-974X(95)97296-A.
CEN (European Committee for Standardization). 2018. Eurocode 3— Rondal, J. 1984. “Contribution à l'étude de la stabilité des profils creux à
Design of steel structures—Part 1-1: General rules and rules for build- parois minces.” Ph.D. thesis, Faculté des sciences appliquées, Univ. of
ings. EN 1993-1-1:2018. Brussels, Belgium: CEN. Liège.
Chan, T. M., L. Gardner, and K. H. Law. 2010. “Structural design of ellip- Rossi, B., S. Afshan, and L. Gardner. 2013. “Strength enhancements in cold-
tical hollow sections: A review.” Proc. Inst. Civ. Eng. Struct. Build. formed structural sections—Part II: Predictive models.” J. Constr. Steel
163 (6): 391–402. https://doi.org/10.1680/stbu.2010.163.6.391. Res. 83 (Apr): 189–196. https://doi.org/10.1016/j.jcsr.2012.12.007.
Cruise, R. B., and L. Gardner. 2008. “Strength enhancements induced dur- Salvarinas, J. J., J. D. Barber, and P. C. Birkemoe. 1978. An experimental
ing cold forming of stainless steel sections.” J. Constr. Steel Res. investigation of the column behaviour of cold-formed stress-relieved hol-
64 (11): 1310–1316. https://doi.org/10.1016/j.jcsr.2008.04.014. low structural steel sections. CIDECT 2F-78/10. Altendorf, Switzerland:
ECCS (European Convention for Constructional Steelwork). 1984. Ulti- Committee for International Development and Education on Construc-
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

mate limit state calculation of sway framed with rigid joints. Technical tion of Tubular Structures.
Rep. No. 33. Brussels, Belgium: ECCS. Schillo, N., and M. Feldmann. 2018. “Interaction of local and global buck-
Gardner, L., N. Saari, and F. Wang. 2010. “Comparative experimental ling of box sections made of high strength steel.” Thin-Walled Struct.
study of hot-rolled and cold-formed rectangular hollow sections.” 128 (Jul): 126–140. https://doi.org/10.1016/j.tws.2017.07.009.
Thin-Walled Struct. 48 (7): 495–507. https://doi.org/10.1016/j.tws Sedlacek, G., B. Kuhn, J. Rondal, and P. Boeraeve. 1999. Buckling behav-
.2010.02.003. iour of hot-formed SHS in high strength steel grade E-460. CIDECT
Gardner, L., and X. Yun. 2018. “Description of stress-strain curves for cold- 2T-2/99. Altendorf, Switzerland: Committee for International Develop-
formed steels.” Constr. Build. Mater. 189 (Nov): 527–538. https://doi ment and Education on Construction of Tubular Structures.
.org/10.1016/j.conbuildmat.2018.08.195. Sedlacek, G., J. Rondal, P. Boeraeve, N. Stranghöner, R. Schneider, and D.
Guiaux, P. 1972. Essais de flambement sur profils creux formes a froid, Grotmann. 1996. Buckling behaviour of a new generation of cold-
carres et circulaires. CIDECT 72/28/F. Altendorf, Switzerland: Com- formed hollow sections. CIDECT No. 2R-2-96. Altendorf, Switzerland:
mittee for International Development and Education on Construction of Committee for International Development and Education on Construc-
Tubular Structures. tion of Tubular Structures.
Huang, Y., and B. Young. 2018. “Design of cold-formed stainless steel cir- Sfintesco, D., and A. Carpena. 1977. “Experimental bases of the ECCS
cular hollow section columns using direct strength method.” Eng. column curves.” In Proc., 2nd Int. Colloquium on Stability, 68–75.
Struct. 163 (May): 177–183. https://doi.org/10.1016/j.engstruct.2018 Bethlehem, PA: Lehigh Univ., Secretary, Structural Stability Research
.02.012. Council.
Jandera, M., L. Gardner, and J. Machacek. 2008. “Residual stresses in cold- Simões da Silva, L., T. Tankova, L. Marques, U. Kuhlmann, A. Kleiner, J.
rolled stainless steel hollow sections.” J. Constr. Steel Res. 64 (11): Spiegler, H. H. Snijder, R. Dekker, A. Taras, and N. Popa. 2017. “Safety
1255–1263. https://doi.org/10.1016/j.jcsr.2008.07.022. assessment across modes driven by plasticity, stability and fracture.”
Key, P. W., S. W. Hasan, and G. J. Hancock. 1988. “Column behavior of In Proc., Eurosteel 2017, 3689–3698. Berlin: Ernst & Sohn.
cold-formed hollow sections.” J. Struct. Eng. 114 (2): 390–407. https:// Somodi, B., and B. Kövesdi. 2017. “Flexural buckling resistance of cold-
doi.org/10.1061/(ASCE)0733-9445(1988)114:2(390). formed HSS hollow section members.” J. Constr. Steel Res. 128 (Jan):
Law, K. H., and L. Gardner. 2012. “Lateral instability of elliptical hollow 179–192. https://doi.org/10.1016/j.jcsr.2016.08.014.
section beams.” Eng. Struct. 37 (Apr): 152–166. https://doi.org/10.1016 SSAB. 2014. Axial resistance of double grade (S355, S420) hollow
/j.engstruct.2011.12.008. sections manufactured by SSAB, statistical evaluation based on tests.
Ma, J. L. 2016. “Behaviour and design of cold-formed high strength steel Helsinki, Finland: SSAB.
tubular members.” Ph.D. thesis, Dept. of Civil Engineering, Univ. of Starting, S., and H. Vos. 1973. “Computer simulation of the E.C.C.S. buck-
Hong Kong. ling curve using a Monte-Carlo method.” HERON 19 (2): 3–38.
Ma, J. L., T. M. Chan, and B. Young. 2016. “Experimental investigation on Sully, R. M., and G. J. Hancock. 1996. “Behavior of cold-formed SHS
stub-column behavior of cold-formed high-strength steel tubular sec- beam-columns.” J. Struct. Eng. 122 (3): 326–336. https://doi.org/10
tions.” J. Struct. Eng. 142 (5): 04015174. https://doi.org/10.1061 .1061/(ASCE)0733-9445(1996)122:3(326).
/(ASCE)ST.1943-541X.0001456. Taras, A., and R. Greiner. 2010. “New design curves for lateral-torsional
Ma, J. L., T. M. Chan, and B. Young. 2019. “Cold-formed high-strength buckling—Proposal based on a consistent derivation.” J. Constr. Steel
steel rectangular and square hollow sections under combined compres- Res. 66 (5): 648–663. https://doi.org/10.1016/j.jcsr.2010.01.011.
sion and bending.” J. Struct. Eng. 145 (12): 04019154. https://doi.org Toffolon, A., and A. Taras. 2019. “Development of an OIC-Type local
/10.1061/(ASCE)ST.1943-541X.0002446. buckling design approach for cold-formed unstiffened and groove-
Meng, X., and L. Gardner. 2020a. “Simulation and design of semi-compact stiffened hollow sections.” Thin-Walled Struct. 144 (Nov): 106266.
elliptical hollow sections.” Eng. Struct. 202 (Jan): 109807. https://doi https://doi.org/10.1016/j.tws.2019.106266.
.org/10.1016/j.engstruct.2019.109807. Walport, F., L. Gardner, E. Real, I. Arrayago, and D. A. Nethercot. 2019.
Meng, X., and L. Gardner. 2020b. “Testing of hot-finished high strength “Effects of material nonlinearity on the global analysis and stability of
steel SHS and RHS under combined compression and bending.” stainless steel frames.” J. Constr. Steel Res. 152 (Jan): 173–182. https://
Thin-Walled Struct. 148 (Mar): 106262. https://doi.org/10.1016/j.tws doi.org/10.1016/j.jcsr.2018.04.019.
.2019.106262. Wang, J., S. Afshan, M. Gkantou, M. Theofanous, C. Baniotopoulos, and
Meng, X., L. Gardner, A. J. Sadowski, and J. M. Rotter. 2020. “Elasto- L. Gardner. 2016. “Flexural behaviour of hot-finished high strength
plastic behaviour and design of semi-compact circular hollow sections.” steel square and rectangular hollow sections.” J. Constr. Steel Res.
Thin-Walled Struct. 148 (Mar): 106486. https://doi.org/10.1016/j.tws 121 (Jun): 97–109. https://doi.org/10.1016/j.jcsr.2016.01.017.
.2019.106486. Wang, J., S. Afshan, N. Schillo, M. Theofanous, M. Feldmann, and L.
Pavlovcic, L., B. Froschmeier, U. Kuhlmann, and D. Beg. 2012. “Finite Gardner. 2017. “Material properties and compressive local buckling re-
element simulation of slender thin-walled box columns by implement- sponse of high strength steel square and rectangular hollow sections.”
ing real initial conditions.” Adv. Eng. Software 44 (1): 63–74. https://doi Eng. Struct. 130 (Jan): 297–315. https://doi.org/10.1016/j.engstruct
.org/10.1016/j.advengsoft.2011.05.036. .2016.10.023.
Rasmussen, K. J. R., and G. J. Hancock. 1993. “Design of cold-formed stain- Wang, J., and L. Gardner. 2017. “Flexural buckling of hot-finished high-
less steel tubular members. I: Columns.” J. Struct. Eng. 119 (8): 2349– strength steel SHS and RHS columns.” J. Struct. Eng. 143 (6):
2367. https://doi.org/10.1061/(ASCE)0733-9445(1993)119:8(2349). 04017028. https://doi.org/10.1061/(ASCE)ST.1943-541X.0001763.

© ASCE 04020227-13 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227

 Yun, X., and L. Gardner. 2017. “Stress-strain curves for hot-rolled steels.” Struct. 157 (Feb): 179–191. https://doi.org/10.1016/j.engstruct.2017
J. Constr. Steel Res. 133 (Jun): 36–46. https://doi.org/10.1016/j.jcsr .12.009.
.2017.01.024. Zhao, O., L. Gardner, and B. Young. 2019. “Finite element modelling and
Yun, X., and L. Gardner. 2018. “The continuous strength method for the de- design of stainless steel SHS and RHS beam-columns under moment
sign of cold-formed steel nonslender tubular cross-sections.” Eng. Struct. gradients.” Thin Walled Struct. 134 (Jan): 220–232. https://doi.org/10
175 (Nov): 549–564. https://doi.org/10.1016/j.engstruct.2018.08.070. .1016/j.tws.2018.10.004.
Yun, X., L. Gardner, and N. Boissonnade. 2018. “The continuous strength Ziemian, R. D. 2010. Guide to stability design criteria for metal structures.
method for the design of hot-rolled steel cross-sections.” Eng. 6th ed. Hoboken, NJ: Wiley.
Downloaded from ascelibrary.org by CORNELL UNIV LIBRARIES on 08/18/20. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 04020227-14 J. Struct. Eng.

J. Struct. Eng., 2020, 146(11): 04020227