Thin-Walled Structures 128 (2018) 92–107

Contents lists available at ScienceDirect

Thin-Walled Structures
journal homepage: www.elsevier.com/locate/tws

Full length article

Experimental characterization of steel tubular beam-columns resistance by T

means of the Overall Interaction Concept
⁎
Marielle Hayecka, Joanna Nseirb, Elsy Saloumia, Nicolas Boissonnadec,
a
University of Applied Sciences of Western Switzerland, Fribourg, Switzerland
b
Saint Joseph University, Beirut, Lebanon
c
Laval University, Québec City, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: This paper presents a series of twelve buckling tests on rectangular and circular hollow section beam-column of
Steel section nominal steel grade S355. The columns were fabricated by either the hot-rolling or the cold-forming process, and
Steel member were subjected to different load cases through the application of eccentric compression: mono-axial bending
Design (My) or bi-axial bending (My + Mz) combined with axial compression (N). Preliminary measurements of cross-
Overall Interaction Concept
section geometry, material properties, geometrical imperfections, residual stresses as well as stub column tests
Resistance
are also reported in this paper. The (imperfect) initial geometry was measured along the whole column by means
Stability
Buckling tests of two different procedures; the first method relied on the use of a set of equally spaced Linear Variable
Displacement Transducers (LVDTs) displaced on each specimen's plates. The second method consisted in
scanning the specimen's plates by means of a laser Tracker AT401. Residual stresses were also determined
experimentally, and the sectioning technique was classically used to record the deformations of the released
material; these results have been compared to measurements taken with electrical strain gauges. Finally, beam-
column buckling strengths were plotted in an O.I.C.-type format, and complemented with an extensive
experimental data collected from the literature that comprised various load cases, fabrication processes, yield
strengths, cross-sections shapes, and elements’ lengths. The O.I.C. approach was shown to adequately capture the
behavior of hollow section beam-columns in a straightforward and simple manner, and its potential for
efficiently and rationally predicting carrying capacities was evidenced.

1. Introduction influence of initial imperfections (out-of-straightness, residual stresses,

The experimental results presented in the present paper take place
in a wider research effort that consists in improving the way the
performance and the carrying capacity of tubular members are actually
characterized, through the development of an original “Overall
Interaction Concept” (O.I.C., cf. [1,2]). They follow investigations at
the cross-section level where a series of tests on RHS, SHS and CHS
hollow sections has also been performed [3].
The O.I.C. approach primarily relies on the fact that the behavior of
a steel member is influenced by the interaction between resistance and
instability. The classical concept of ‘resistance’ mainly characterizes the
cross-sectional capacity of the member, assuming that no instability
phenomena can occur, whereas the concept of ‘stability’ describes the
tendency of an ideal member to buckle under the applied loading. The
Overall Interaction Concept makes use of both aspects and includes the
non-homogenous material, etc.) to calculate the resistance of a member
through the generalization of the relative slenderness parameter
defined as λ = Rpl / Rcr , where Rpl represents the ratio by which the
initial loading has to be multiplied to reach the pure resistance limit –
i.e. the plastic load – and Rcr is the ratio leading to the buckling load
(stability) of the ideal member – i.e. the critical load.
Resorting to load ratios makes it capable of directly dealing with
complex loading situations, as well as with various problems (e.g. member
or cross-section resistance-instability interaction, non-linear material
behavior). The relative slenderness value “λ” then leads to the determina-
tion of a “χ” value (Fig. 1) called “reduction factor” that represents the
penalty due to instability effects on the pure resistant behavior. The
interested reader may usefully refer to [1] for details relative to the
historical background of the O.I.C., its connections with existing ap-
proaches and its various features and mechanical interpretations.

Abbreviation:O.I.C., Overall Interaction Concept

⁎
Correspondence to: Civil and Water Engineering Department, Laval University, Adrien Pouliot Bldg., Office 2928G, 1065, Avenue de la Médecine, Quebec, Canada G1V 0A6.
E-mail addresses: marielle_hayeck@hotmail.com (M. Hayeck), Joanna.nseir@usj.edu.lb (J. Nseir), elsysaloumi@gmail.com (E. Saloumi),
nicolas.boissonnade@gci.ulaval.ca (N. Boissonnade).

http://dx.doi.org/10.1016/j.tws.2017.05.019
Received 15 February 2017; Received in revised form 8 May 2017; Accepted 16 May 2017
Available online 25 May 2017
0263-8231/ © 2017 Elsevier Ltd. All rights reserved.
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Nomenclature χL Generalised cross-section local buckling factor

χL + G Generalised member local and global buckling factor
ψ End moment ratio χTest Generalised buckling factor calculated from test results
λL Generalised cross-section relative slenderness (includes Rpl Load ratio to reach to “resistance” limit (plastic capacity)
influence of local buckling behavior) Rcr,L Load ratio to reach to cross-sectional (local) “stability”
λL + G Generalised member relative slenderness (includes influ- limit
ences of local and global buckling behavior) Rcr,G Load ratio to reach to member (global) “stability” limit

Cross-sectional local behaviour (L) Member global behaviour (G)

Rcr,L Rpl Rcr,G

3 key factors:calculated by advanced tools (or by formulae)

R pl χ L R pl
λL = λ L+ G =
Rcr , L Rcr ,G

χL χ L +G Member buckling curve

χL+G Rpl
Cross-section buckling curve Rcr,G
χL Rpl Member global
buckling curve
Rcr,L
Cross-section local
buckling curve
λL+G

λL Design check:
Rb,L+G = ≥ 1.0 (Eurocode 3 format)

Rb , L+G = 1.0 (AISC format)

Fig. 1. Principles and application steps of the Overall Interaction Concept.

As Fig. 1 summarizes, design by means of the O.I.C. consists in
applying the approach first at the cross-section level, then at the
member level, leading to a final, single design check that accounts for
the complex interactions between plasticity, local and global instabil-
ities and imperfections; a particular attention shall be paid to the
definition of the member relative slenderness λL + G which incorporates
the potential influence of cross-sectional local instability on member
behavior through factor χL – in essence, this treatment of local/global
coupled instabilities is similar to what is proposed in many design
codes, such as Eurocode 3 for example [4].
In order to assess the ability of the O.I.C. to characterize accurately
the behavior and response of hollow section shapes, an extensive data
set was collected from many sources and assembled into a convenient
database. A total of 802 test results for members were collected from
Grimault [5], Rondal [6], Greiner [7], Yeomans [8], Guiaux [9],
Salvarinas [10], Braham [11], Sedlacek [12], Kuhn [13] and Pavlovcic
[14] 1; the experimental beam-column tests conducted in the present
work were also considered. This experimental database comprises
different element lengths tested under simple and combined loading
with different shapes of bending moment distribution along the
member, and includes hollow sections well distributed along section
classes (i.e. from stocky to slender).
Although relatively extensive, this database may not allow to
adequately quantify the influence of key design parameters nor to
clearly isolate their effects. In this respect, typical non-linear F.E.
simulations are desirable and shall bear considerable benefits, provided
that (i) the numerical models have shown appropriate and suitable (i.e.
validated against test results) and that (ii) experimental evidence of real
mechanical behavior served as a basis for the developments of F.E.
models.
Accordingly, the need for fully-detailed and documented tests
appeared inevitable. The following Section 2 describes a series of 12
buckling tests performed on rectangular and circular hollow sections
with a nominal steel grade S355. The columns were fabricated by either
the hot-rolling or the cold-forming process, and loaded under combined
compression and bending by varying axial compression load eccentri-
cities ey and ez. Two column lengths were chosen so as inelastic
buckling to be governing (4000–4900 mm).
Besides the main beam-column test series, preliminary measure-
ments were performed that consisted in:

i) Cross-section dimensions’ measurements using a digital caliper;

1 ii) Material testing by means of classical tensile tests extracted from
It shall perhaps be mentioned here that the data extracted from Braham consisted in
cold-formed rectangular hollow sections for which an annealing process was performed the flat faces and from the corners of the corresponding specimens;
for some of these sections. iii) Geometrical imperfections measurements by means of two different

93
M. Hayeck et al.
4900
Test specimen
6000
Fig. 2. Partition of a 4900 mm, CHS 159 × 5 column (dimensions in mm).
procedures. The first method consisted in a set of equally spaced
linear variable displacement transducers (LVDTs), fixed on an
aluminum bar that was displaced laterally on each specimen's
plates. The second method consisted in scanning the specimen's
plates by means of a laser Tracker AT401;
iv) Residual stresses determination through the sectioning method to
measure the deformations experienced by the released material.
Measurements taken with electrical strain gauges were compared to
the ones obtained by the mechanical procedure;
v) Stub column tests for the determination of cross-sectional load
carrying capacities under pure compression.
Further to the results of the tests themselves, the main goal was here
to collect sufficient information for the validation of F.E. numerical
tools, in order to launch consecutive extensive parametric studies on
hot-rolled and cold-formed sections, including a wide scope of para-
meters (such as cross-section shape, steel grade, load cases, etc.). In this
respect, particular care and attention was paid to carefully measure all
preliminary data that shall be accounted for in the numerical models, as
detailed in the various paragraphs of Section 2.
Section 3 then makes use of the experimental database to analyze
experimentally the influence of the various parameters on the member's
response and resistance. These data are then used to (i) assess the
suitability of the O.I.C. approach and to (ii) identify the remaining key
parameters that shall be accounted for in the derivation of so-called
“buckling” or “resistance” curves.
2. Experimental investigations
2.1. Test specimens and experimental setup
Twelve specimens involving 4 different cross-section shapes of
nominal steel grade S355 were tested at the Structural Engineering
laboratory of the University of Applied Sciences of Western Switzerland
– Fribourg: two hot-rolled Circular Hollow Section shapes (CHS 159 ×
5 and CHS 159 × 6.3) as well as two cold-formed Rectangular Hollow
Section shapes (RHS 200 × 100 × 4 and RHS 220 × 120 × 6). Two
column lengths (L = 4000 mm and L = 4900 mm) were considered for
the main beam-column tests; a portion – of approximately L =
1000 mm – was kept for the preliminary measurements of residual
stresses, stub column and tensile tests. Fig. 2 represents the typical
cutting plan of a 4900 mm profile.
The ends of the tested columns were milled flat and welded to end
plates of 20 mm thickness, with potentially different eccentricities,
according to the desired load case (see Fig. 21). Mono-axial and bi-
axial-bending with axial compression load cases were obtained by
applying eccentric compression; the bending moment distributions
applied on the members were therefore linear, either constant (equal
eccentricities at both end of the specimen) or triangular (eccentricities
at one end of the specimen only). Accordingly, end moments ratios2 ψ
= 1 and ψ = 0 were adopted to consider constant and triangular
bending moment distributions, respectively. Table 1 summarizes the
test program of the twelve buckling tests. The test specimens have been
2
ψy and ψz indicate the ratios between end moments about y-y and z-z axes,
respectively.
94
Thin-Walled Structures 128 (2018) 92–107
Res. Stresses Tensile coupons
150 475 475
Stub column Test
labeled so that the section type (RHS or CHS), the fabrication process3
(cold-formed CF or hot-rolled HR) and the cross-section geometry can
be identified from the label. A subsequent letter (T) is used, followed by
an identification number ranging from 1 to 12, denoting the test
number. The full test program can be found in [15].
The general test setup is presented in Fig. 3. A purposely-designed
rigid frame was built to ensure sufficient bracing of the tested column.
The loading rig consisted in a hydraulic jack fixed at the bottom of the
column and used to generate upwards compressive force. Four load
cells were located under the jacks to record the force applied. The end
plates of the specimen were centered at their bottom and top to two
spherical supports (i.e. hinges) designed to provide nearly-pinned end
conditions. Since the hydraulic jack was placed at the bottom extremity,
this extremity was free to move vertically. In order to resist to the shear
forces arising when the column is tested under compression with
triangular bending moment distributions, an additional supporting
system aimed at resisting horizontal forces was specially designed, as
illustrated in Fig. 4. In one direction, these supports were formed by
two triangularly-shaped pieces bolted to a base plate of dimensions 700
× 500 × 50 at each side of the specimen. In the other direction, the
supports were made with welded channels 300 × 220 × 20 connected
at each side of the specimen.
2.2. Tensile tests
The material properties of all rectangular and circular hollow
sections were characterized by means of classical tensile tests. Series
of coupons were extracted from the middle of two opposite flat faces –
not containing the weld – of the six cold-formed rectangular hollow
sections, as well as from the two opposite corners, so as to measure the
expected increase in strength associated with cold-forming effects. Two
tensile coupons were also extracted from each section of the six CHS.
The RHS flat coupons were 270 mm long with nominal gauge width of
10 t, where t represents the thickness of the corresponding plate. The
corner prismatic coupons as well as the CHS coupons were 150 mm
long with coupon dimensions of 3 mm × 3 mm cut within the cross-
section thickness in order to avoid creating eccentric loads while testing
(see Fig. 5 and Fig. 6).
A 100 kN testing machine with hydraulic grips was used to test the
necked coupons. The corner and CHS coupons were tested in a smaller
10 kN testing machine due to their smaller size and cross-section. The
coupons were placed in the testing rig and a calibrated extensometer of
20 mm was mounted at the middle of each coupon. Tensile load was
applied by a constant rate of strain (0.045%/s) until fracture. In total,
36 coupon tests have been performed; Fig. 7 depicts representative
examples of stress-strain curves obtained for hot-rolled section CHS 159
× 5 and cold-formed section RHS 220 × 120 × 6. The finally-kept
yield stress value fy was taken as the average of the yield plateau stress
for hot-rolled sections; 0.2% proof stress was used as a convenient
equivalent yield stress for cold-formed sections where the material
response showed a pronounced non-linear behavior.
Experimental results are presented in Table 2 where Em represents
the mean value of the measured section Young's modulus while (fym,
fum, ftm) represent the mean values of yield, ultimate and fracture
3
CF: cold-formed, HR: hot-rolled.
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Table 1
Test program for buckling tests.

Specimen Fabrication process Load case ψy ψz Measured dimensions

Length H B t D
[–] [–] [mm] [mm] [mm] [mm] [mm]

RHS CF 200 × 100 × 4 T1 Cold-formed N + My 1 – 4001.22 200.12 100.28 4.00 –

RHS CF 200 × 100 × 4 T2 Cold-formed N + My + Mz 1 1 3999.61 200.42 100.08 4.05 –
RHS CF 200 × 100 × 4 T3 Cold-formed N + My 0 – 4000.30 198.66 99.93 3.85 –
RHS CF 220 × 120 × 6 T4 Cold-formed N + My 1 – 4000.09 220.09 119.77 5.90 –
RHS CF 220 × 120 × 6 T5 Cold-formed N + My + Mz 1 1 4000.33 219.91 120.27 5.86 –
RHS CF 220 × 120 × 6 T6 Cold-formed N + My 0 – 3998.58 219.90 119.78 5.83 –
CHS HR 159 × 5 T7 Hot-rolled N + My 1 – 3999.70 – – 5.42 159.80
CHS HR 159 × 5 T8 Hot-rolled N + My 1 – 4895.00 – – 5.30 159.30
CHS HR 159 × 5 T9 Hot-rolled N + My 0 – 4000.05 – – 5.30 159.40
CHS HR 159 × 6.3 T10 Hot-rolled N + My 1 – 4900.05 – – 6.51 159.20
CHS HR 159 × 6.3 T11 Hot-rolled N + My + Mz 1 1 4000.10 – – 6.53 159.40
CHS HR 159 × 6.3 T12 Hot-rolled N + My 0 – 4000.00 – – 6.40 159.20

Fig. 3. Overall view of test setup.

95
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Welded channel Welded channel

300x220x20 300x220x20

Triangularly- Triangularly-
shaped piece shaped piece

Anchor bolts

Fig. 4. Support designed to resist shear forces.

Fig. 5. Tensile coupons extracted from – a) flat faces of cold-formed RHS sections – b) corners of cold-formed RHS sections – c) hot-rolled CHS sections.

stresses along with the corresponding average values of elongation (εym,

εum and εtm), respectively.

2.3. Residual stresses measurements

The sectioning technique was first used to measure the distribution

Fig. 6. Typical dimensions of tensile coupons (dimensions in mm).
and intensity of residual stresses within the sections (see Fig. 8).
Material relaxation was such that each strip, after cutting, exhibited
an axial displacement due to the membrane residual stresses relaxation
as well as a curvature from initial bending residual stresses. Besides,
additional measurements were made with electrical strain gauges and
compared to the ones obtained by the mechanical procedure, in an
effort to improve confidence in the results through redundant informa-

600
600

500 500
Stress [N/mm ]

Stress [N/mm ]
2

400
2

400

300 300
Face_1
200 200 Face_2
Face_1 Corner_1
Face_2 Corner_2
100 100
Ultimate strength fu Ultimate strength fu

0 0
0 2 4 6 8 10 12 14 16 18 20 22 24 0 5 10 15 20 25
Strain [%] Strain [%]
Fig. 7. Typical example of engineering stress-strain curves for – a) hot-rolled CHS 159 × 5 – b) cold-formed RHS 220 × 120 × 6.

96
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Table 2
Measured material properties.

Section Position of coupon Em fym εym fum εum ftm εtm

[kN/mm2] [N/mm2] [%] [N/mm2] [%] [N/mm2] [%]

RHS 200 × 100 × 4 Flat 233.1 475 0.203 583 15.7 434 27.7
Corner 207.9 593 0.285 625 1.5 385 3.53
RHS 200 × 100 × 4 Flat 193.8 450 0.335 566 13.3 410 23.2
Corner 217.1 573 0.266 611 1.3 361 2.67
RHS 200 × 100 × 4 Flat 206.4 480 0.234 589 12 421 21.1
Corner 180.3 588 0.329 618 2.4 406 4.21
RHS 220 × 120 × 6 Flat 184.0 450 0.245 542 14.6 389 25.9
Corner 205.1 590 0.288 641 1.4 438 2.81
RHS 220 × 120 × 6 Flat 182.6 462 0.253 554 13 387 23.8
Corner 188.8 553 0.298 574 0.8 394 1.17
RHS 220 × 120 × 6 Flat 196.1 448 0.229 553 14.6 395 26.1
Corner 183.2 585 0.32 624 1.5 355 3.33
CHS 159 × 5 Flat 205.0 399 0.195 543 16.3 316 23.6
CHS 159 × 5 Flat 198.5 393 0.198 529 16.7 312 23.2
CHS 159 × 5 Flat 202.2 405 0.201 537 16.2 320 23.4
CHS 159 × 6.3 Flat 198.9 396 0.199 539 16.4 324 22.5
CHS 159 × 6.3 Flat 202.7 389 0.192 522 16.1 321 21.5
CHS 159 × 6.3 Flat 206.4 394 0.191 529 14.9 325 19.9

Fig. 8. Principles of the sectioning method (figure from [16]).

tion, since experimental measurements of residual stresses are known to
be quite delicate.
2.3.1. Mechanical measurements
Released residual stresses were first measured by mechanical
means. The specimens were divided into strips of 150 mm length and
30 mm width and labeled. Fig. 9 illustrates the location of the strips in
the RHS 200 × 100 × 4 and RHS 220 × 120 × 6 sections along with
the adopted numbering convention.
Two little 100 mm-spaced circular imprints were punched on each
strip of the cross-section; these marks were used as a precise reference
to measure the length and curvature variations mechanically (see
Fig. 9). Prior to cutting, the length was measured by an extensometer
with an accuracy of ± 5 μm while the curvature reference was mea-
sured by means of a curvature dial. Cutting of the strips was then
performed on an automated milling machine followed by additional
measurements of lengths and curvatures of the released strips. The
differences between initial and final strips length led to membrane
stresses patterns, while the curvature variations provided information
on flexural – through thickness – stresses (see Fig. 10).
2.3.2. Strain gauges’ measurements
After recording all initial lengths and curvatures, a series of strain
gauges were glued to the external surface of each strip in the long-
itudinal direction, followed by initial electrical readings (Fig. 11). The
strips were then cut and final readings could be taken. The total residual
stresses measured from the outer surfaces combined both bending and
membrane stresses; consequently, the membrane residual stresses were

A22 A3 A4 B
B19 B20 B21
B18 B22
A1 A5
B17 B1
A22 A6
B2
32

A21 A7 B16

A20 A8 B15 B3

A19 A9 B14 B4

A18 A10
B13 B5

A17 A11
B12 B6
A16 A12
A115 A14 A13 B11 B
B10 B9 B8 B7

RHS_CF_200
0x100x4_Strips_A RHS_CF_
_220x120x6_Strips_B

Fig. 9. Location of the strips in the hollow sections with the adopted labeling system – circular imprint.

97
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

obtained by subtracting the bending stresses determined from the

curvature dial, and compared to values obtained mechanically. Patterns
of through-thickness residual stresses variations from both the electrical
Rm Inw
ward and the mechanical readings are presented in Fig. 12 for the RHS CF
Re striip 220 × 120 × 6 specimen. One may note that the bending stresses were
v currvature dominant while the membrane stresses were relatively low, as typically
l0 recorded for cold-formed tubes.
a The measured magnitudes were inevitably affected by several
a experimental uncertainties, explaining the differences between the
l0 Outtward membrane residual stresses obtained mechanically and their counter-
v
striip parts obtained through electrical strain gauges. The main reasons
Ri currvature behind these differences are laboratory measurement errors, especially
Rm in the way of holding the extensometer, the variation of the inclination
of the extensometer before and after cutting, etc. The extensometer had
also an accuracy of ± 5 μm. This precision value did not affect
significantly the residual stresses in the corners as much as those found
in the flat faces, because of the small magnitudes measured in the flat
Fig. 10. Geometrical deformation due to (flexural) residual stresses.
faces, especially with respect to membrane residual stresses measure-
ments. It is to be noted that mechanical measurements have a better
level of reliability when the strips were subjected to high values of
residual stresses and therefore had a large variation in length, while
Marks for mechanical electrical measurements give more accurate results for low values of
strain readings released stresses. Accordingly, electrical and mechanical measure-
ments’ pairs complete each other and improve confidence in the
measurements.
Strain gauge for electrical
strain readings
2.4. Geometrical imperfections

The measurement of initial imperfections was performed by means

of two different procedures. The first method was based on the
measurements of a set of equally spaced linear variable displacement
transducers (LVDTs) fixed on an aluminum bar that was displaced
Fig. 11. Electrical strain measurements.
laterally on each specimen's plate in order to record out-of-plane

Stress [MPa] Stress [MPa]

400 400
200 200
64.65 59.3
0 0
-37.5-21.2-41.6
-200 -12.9-36.9-51.9 -200
-400 -400 -255
-393 -398
-600 -600 -487 -430-450
-248
-500
Stress [MPa]

Stress [MPa]
Stress [MPa]

Stress [MPa]
-600
-400
-200

-600
-400
-200
0
400

64.6 200

-200
-400
-600

200
400

400
200
0
-200
-400
-377 -375 -424 -523 -600

200
400
B19 B20 B21 B19 B20 B21
0

B18
-558 -248

B22 B18 B22

59.3

-255
-19.8-38.8-32.8
-31.8 -23.4 -30

-401 -387-360-349-390-395
-381 -349-326-318-347-346

B17 B1 B17 B1
B2 B2
-438

B16
-48.9 -16.4-52-29.7-42-50.7

B16
-30.8-16-30.4-10.8-14-34.4

B15 B3 B15 B3
-347-364

CF CF
-31.2 -42.6-48.1
-40 -65.6-29.4

B14 B4 B14 B4
-431
-447

B13 B5 B5
B13

B12 B6
-529
-560

B12 B6
64.4

-150
-164

B11 B10 B9 B8 B7
41.5

B11 B10 B9 B8 B7

-164
-600 -600 -460 -376 -423
-445 -365 -413
-400 -400
-200 -14.7 -10.7 -10.6 -150
-21.3 -17.6 -29.68 -200
0 0
200 41.5 64.4
200
400 400
Stress [MPa] Stress [MPa]

Membrane Stress Gauge [MPa]

Membrane Stress Stress Total Gauge [MPa]
Mecanichal Method Flexural Stress
[MPa] Mecanichal Method
[MPa]
Fig. 12. Mechanical and electrical measured stresses – a) membrane (right column) – b) flexural/total (left column) stresses of RHS CF 220 × 120 × 6.

98
M. Hayeck et al.
10 displacements
LVDTs measuring
out-of-plane defaults
Perforated
aluminum bar
Fig. 13. Initial geometrical imperfections measuring procedure. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.).
imperfections on each plate's surface. The second method consisted in
scanning the specimen's plates by means of a laser Tracker AT401
capable of measuring positions of scattered points very accurately.
These points were then post-treated with a specific software to get the
out-of-plane defaults of each specimens' plate.
2.4.1. LVDT measurements
Imperfections were measured using an aluminum frame comprising
a sideways-movable aluminum bar drilled at 10 equally-spaced loca-
tions, containing 10 vertical displacements LVDTs spaced at 10 cm
intervals (yellow arrow in Fig. 13). In order to record data over the
complete specimen, the frame was moved along the profile's length (see
green arrow in Fig. 13).
Fig. 14 gives an example of the (magnified) imperfect shape of a
portion selected at mid span of the specimen RHS CF 200 × 100 × 4
T2, as implemented in finite elements models (cf. [15]).
Fig. 15 represents the corresponding out-of-flatness contour plots of
imperfection magnitudes obtained for some of the considered plates
separately.
2.4.2. Topometric measurements
Before each test, the initial geometrical imperfections were also
measured for the six cold-formed RHS by means of a laser Tracker
AT401 with a general accuracy of 0.5 mm and an improved accuracy of
0.2 mm in the expected failure zone. The columns were placed
horizontally and marked with a series of targets; marks were more
densely distributed at the expected locations of local buckling. This
technique consisted in moving a light source along predefined long-
itudinal lines for each plate, and very accurately recording consecutive
positions. The "tracker" technology allowed the instrument to detect a
reflector as illustrated in Fig. 16. The remote measurements were then
post-treated by a specific software.
Fig. 17 illustrates the measured initial imperfections along the
length of specimen RHS CF 200 × 100 × 4 T2 at the mid-width of
the bottom flange and left web plates, by using the LVDT and the laser
tracker procedures – one may notice that there was some extra initial
“deformation” recorded near the end plates, due to the presence of the
welds.
Disparities in measurements are mostly credited to a relative
insufficient accuracy of the LVDT method; the LVDTs procedure indeed
typically slightly overestimated the initial defaults. A higher level of
Upper flange
Fig. 14. Imperfect shape of specimen RHS CF 200 × 100 × 4 T2 (magnified).
99
Thin-Walled Structures 128 (2018) 92–107
LVDTs
displacements
along member
length
LVDTs
Transversal
displacements
confidence and reliability was provided by the laser AT401 measure-
ments. Both sets of measured initial imperfections were introduced in
suitably-built shell F.E. models, along with all measured data, and were
shown to lead to nearly identical numerically-predicted failure loads,
indicating that the observed differences in imperfection patterns shall
be deemed acceptable and sufficient [15].
2.5. Stub column tests
12 stub columns compression tests were performed on all different
cross-section shapes. The length of the tested specimens was chosen
equal to three times the largest cross-sectional dimension, based on the
principle that the length is sufficiently small to prevent member
buckling while long enough to avoid an important influence of the
boundary conditions and leave the development of buckling waves free.
The specimens were set in a 5000 kN hydraulic machine to apply
compressive axial force to the stub column specimens, following a
displacement-controlled procedure. Four transducers were positioned
on the stub ends to measure the axial shortening of the specimens, and
two strain gauges were attached at the mid-length of the specimen's
adjacent plates. The strain gauges provided the load displacement
behavior of the specimen in the elastic range in order to assess the
(indirect) corresponding Young's modulus. Fig. 18 shows an example of
a stub column during testing.
Table 3 details the obtained stub column failure loads and also
allows to compare the maximum capacity obtained during the tests
with the actual load Npl, actual and with the nominal expected load Npl,
nominal. Fig. 19 and Fig. 20 plot examples of the evolution of compression
load N versus end shortening δ and of the normalized axial load N/Npl,
actual versus relative strain ε/ εy – εy being the strain level at first yield.
Together with Table 3, they show that all cold-formed sections (RHS)
have not been able to reach their plastic capacity, sometimes just short
of it (98% of Npl,actual for specimen RHS CF 220 × 120 × 6), sometimes
by far (73% of Npl,actual for specimen RHS CF 200 × 100 × 4). In
contrast, all hot-rolled (CHS) stub columns showed a certain effect of
strain hardening on the carrying capacity, up to 20%.
2.6. Beam-column tests
As previously mentioned, the test program comprised twelve beam-
columns tested under mono-axial and bi-axial-bending with axial
Right web
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Right web:
80 0.8

0.7

0.6

0.5

Magnitude
Width [mm]

[mm]
0 0.4

0.3

0.2

0.1

-80 0
0 500 1000 1500 2000 2500 3000 3500 4000
Length [mm]
Upper flange:
30 1
Width [mm]

Magnitude
[mm]
0 0.5

-30 0
0 500 1000 1500 2000 2500 3000 3500 4000
Length [mm]

Fig. 15. Measured out-of-flatness contour plots of imperfection magnitudes obtained for plates from RHS CF 200 × 100 × 4 T2 specimen.

compression load cases obtained by applying eccentric compression.
Fig. 21 illustrates typical specimens’ arrangements for the case of a
constant and a triangular bending moment distributions.
Various transducers were used to monitor the member's response:
i) 4 load cells were located under the jacks to record the applied force;
ii) Inclinometers were fixed at both ends of the column to measure the
column end rotations in both principal bending planes;
iii) 4 linear variable displacement transducers were positioned on the
mid-span cross-section, to measure lateral and transversal displace-
ments by means of two independent systems (Fig. 22);
iv) 4 linear variable displacement transducers were positioned on the
bottom end plate of the specimen to record the axial shortening and
rotations during testing (see Fig. 23).
obtained with the inclinometers for the same test. The excellent
agreement between both sources indicates a high level of reliability
in the experimental recordings – note also that upper and lower
readings from inclinometers were expected to be identical since the
loading applied on specimen CHS HR 159 × 6.3 T11is symmetric, and
Figs. 25a and 25b indeed show excellent agreements.
Fig. 26 presents an example of load vs. mid-span cross-section
lateral and transversal obtained displacements (specimen RHS CF 220
× 120 × 6 T6).
All twelve columns were tested up to and beyond failure with all
readings being taken using an electronic data acquisition system
recording at 2 Hz pace. The measured eccentricities and maximum
forces of all tested specimens are listed in Table 4.

Fig. 24 illustrates the displacements measured by the four bottom
LVDTs as well as the corresponding average displacement for specimen
CHS HR 159 × 6.3 T11 – averaging the axial shortening measurements
of LVDTs 1–4 was obviously necessary to isolated the member's overall
shortening, since some tests involved biaxial bending whose corre-
sponding effects on the recordings of the LVDTs had to be accounted for
properly.
Fig. 25 shows the rotations measured from LVDTs recordings along
major and minor axes, respectively, compared with the rotations
3. Characterization by means of the O.I.C
All collected test results previously mentioned in the introduction as
well as those from the previously-detailed 12 buckling tests are
presented in an O.I.C. format in Fig. 27 to Fig. 33, where the horizontal
axis usually relates to the generalised member slenderness λL + G while
the vertical axis depicts the member reduction factor χL + G as
presented in Eq. (1) and Eq. (2). In these figures, Eurocode 3 buckling
curves a0 and d are also reported, for sake of comparison.

Reflector moving
along the column
length
Laser Tracker
AT401 detecting
the movement of
the reflector

Column placed
horizontally and
marked

Fig. 16. Measurement of geometrical imperfections – Topometric procedure.

100
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Bottom flange mid-width measurements Left web mid-width measurements

0.4 1.0
LVDT Measurements LVDT Measurements
0.3 Laser tracker AT401 0.264 Laser tracker AT401
0.8
Imperfections [mm]

Maximum magnitude reached

Imperfections [mm]
Maximum magnitude reached
0.2 0.6
0.098
0.37 0.343
0.1 0.4

0.0 0.2

-0.1 0.0

-0.2 -0.2
0 1000 2000 3000 4000 0 1000 2000 3000 4000
Specimen length [mm] Specimen length [mm]
Fig. 17. Initial deformations along the length of the RHS CF 200 × 100 × 4 T2 specimen at the mid-width of each plate obtained by using the LVDT and the laser tracker procedures.

subsequent results, χL values have been evaluated numerically by

means of suitable shell non-linear F.E. simulations so as to provide a
fair estimation of the influence of local buckling on the members’
Milled flat plate overall responses. These shell models were previously validated against
more than 50 cross-sectional tests on multiple hollow section shapes [3].
Stub column Fig. 27 displays comprehensive plot with all gathered tests – all data
reported in O.I.C.-type plots in what follows accounts for actual,
Local buckle measured data. The relatively large scatter noticed is obviously
associated with multiple load combinations, different sections’ shapes
Strain gauge
and slenderness, various steel grades, production routes, member
slenderness, etc. Also, the scatter can be seen to cover the complete
range between Eurocode 3 buckling curves a0 and d and even well
outside, reflecting the complex behavior and response of beam-
LVDTs measuring columns.
axial shortening In the following, the influence of various key parameters is
Milled flat plate investigated, namely the (i) production process (either hot-finished or
cold-formed hollow sections), (ii) cross-section slenderness, (iii) load
Fig. 18. General stub-column test setup. case, and (iv) bending moment distribution. The intention is here to
identify in which extent these factors shall be accounted for in deriving
χL ⋅Rpl suitable χ = f°(λ) resistance curves – note that some of them are
λL + G =
Rcr , G (1) already accounted for in the O.I.C. approach through the R-ratios in the
λL and λL + G definitions, however with respect to ideal behaviors (i.e.
RTest pure resistance or pure instability).
χL + G = The influence of the cross-section resistance on the member
χL ⋅Rpl (2)
response is highlighted in Fig. 28, where the effects of cross-sectional
Rpl represents the load ratio to reach the plastic resistance limit, Rcr,G local buckling were “omitted” and not deduced from the overall
represents the load ratio to reach the instability limit of the member, behavior, unlike through the general flowchart of the O.I.C. The
RTest is the ultimate load multiplier calculated from experimental peak horizontal axis represents the generalised slenderness λG while the
loads and χL represents the cross-section reduction factor. In all vertical axis depicts the reduction factor χG, defined in Eq. (3) and Eq.

Table 3
Stub column test results.

Specimen L Calculated areaa Npl, actual Npl, nominal Nexp Nexp / Npl, actual Nexp / Npl, nominal
[mm] [mm2] [kN] [kN] [kN] [–] [–]

RHS CF 200 × 100 × 4 603 2197 1044 780 770 0.74 0.99
RHS CF 200 × 100 × 4 601 2204 992 783 765 0.77 0.98
RHS CF 200 × 100 × 4 598 2205 1058 783 775 0.73 0.99
RHS CF 220 × 120 × 6 660 3696 1663 1312 1608 0.97 1.23
RHS CF 220 × 120 × 6 657 3694 1705 1311 1622 0.95 1.24
RHS CF 220 × 120 × 6 659 3692 1652 1311 1611 0.98 1.23
CHS HR 159 × 5 475 2682 1070 952 1233 1.15 1.29
CHS HR 159 × 5 478 2665 1047 946 1220 1.16 1.29
CHS HR 159 × 5 475 2668 1082 947 1162 1.07 1.23
CHS HR 159 × 6.3 474 3131 1240 1111 1481 1.19 1.33
CHS HR 159 × 6.3 474 3066 1191 1089 1437 1.21 1.32
CHS HR 159 × 6.3 476 3078 1211 1093 1470 1.21 1.35

a
Calculated areas were determined by dividing the weight of the specimens by their measured lengths and assumed density of steel (G = 78.5 kN/m3).

101
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

1622 kN Npl,fy =1705 kN

0.952

Npl,355 =1311 kN

N / Npl,fy [-]
Load [kN]

RHS_220*120*6_Flange
RHS_220*120*6
RHS_220*120*6_Web
Peak load
Peak load

Axial shortening δ [mm] Strain ratio ε / ε [-]

Fig. 19. Stub column test results – RHS CF 220 × 120 × 6 T5 – a) load displacement curve – b) strain gauges measurements.

1437 kN 1.21

Npl,fy =1192 kN

N / Npl,fy [-]
Npl,355 =1089 kN
Load [kN]

CHS_159*6.3_Flange
CHS_159*6.3 CHS_159*6.3_Web
Peak load Peak load

Axial shortening δ [mm] Strain ratio ε / ε [-]

Fig. 20. Stub column test results – CHS HR 159 × 6.3 T11 – a) load displacement curve – b) strain gauges measurements.

Hinge connection plate Hinge connection plate

60 60
5 Specimen RHS_S355CF_200x100x4_L4000 5
100

100
Endplates 500/300/20 S355 Endplates 500/300/20 S355
Hinge Hinge
4000

Bolts (M20 10.9)

Hinge connection plate Hinge connection plate

60 5 Specimen RHS_S355CF_200x100x4_L4000 60
140

Endplates 500/300/20 S355 Endplates 500/300/20 S355

Hinge Hinge

Bolts (M20 10.9)

4000

Fig. 21. Typical example of column positioning for the cases of – a) constant bending moment distribution – b) triangular bending moment distribution.

(4), respectively. finished vs. cold-formed), and by “plastic” (class 1) or “slender” (class
4) cross-sectional local slenderness. Obviously, class 1 stocky sections
Rpl
λG = exhibit a higher resistance than slender ones exhibiting an important
Rcr , G (3) influence of local buckling. It is clearly seen that the influence of the
section sensitivity to local buckling is important for low values of
RTest
χG = member relative slenderness λG where different values of resistance are
Rpl (4) reached: plastic, elastic or effective depending on cross-section classi-
The experimental results are first sorted by production process (hot- fication. The vertical scatter observed can be shown to be mostly due to

102
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

2 LVDTs 2 LVDTs recording

recording lateral
transversal displacement
displacement

Fig. 22. Measurement of displacements in mid-span cross-section.

2 LVDTs recording 2 LVDTs

axial shortening recording axial
shortening

Fig. 23. Measurement of axial shortening.

305 kN For higher values of relative slenderness (0.8 < λG < 1.6), the elements
fail in inelastic to elastic flexural buckling. For higher values of relative
slenderness, global buckling becomes solely determinant, regardless of
the section slenderness and local buckling that may occur. The vertical
Load [kN]

scatter is seen to be more reduced in this case.

Fig. 29 presents results obtained for SHS and RHS sections by
deducing the cross-section instabilities from the overall behavior, i.e.
applying the complete O.I.C. approach as summarized in Fig. 1 and
LVDT_1
LVDT_2
with Eqs. (1) and (2) definitions. As expected, the resistance of slender
LVDT_3 sections, exhibiting an important influence of local buckling, is
LVDT_4
LVDT_Average
enhanced when considering the effect of the cross-section instabilities,
especially for low values of the relative slenderness. Although higher χL
+ G resistance values are usually observed, a wide scatter is still

Displacement δ [mm]
Fig. 24. Typical axial shortening curves (CHS HR 159 × 6.3 T11).
different section classes (cf. [2] and see Fig. 29). For low values of the
relative slenderness (λG < 0.8), failure of the elements typically is in
large parts driven by cross-sectional resistance, so that the omission of
χL < 1.0 factors in Eq. (3) explains χG values below Eurocode 3 curve d.
noticeable, especially for cold-formed members, indicating that other
key influences shall be identified.
Fig. 30 and Fig. 31 first filter results for pure compression and for
compression with mono-axial bending load cases, respectively. For each
load case, cold-formed test results were again separated from hot-
finished test results. One may notice that hot-finished sections nearly
systematically exhibit higher resistance than cold-formed ones, no
matter the load case considered; this can be attributed to the material

305 kN 305 kN
Load [kN]
Load [kN]

Upper inclinometer
Upper inclinometer
Lower inclinometer
Lower inclinometer
Rotation LVDT
Rotation LVDT

End rotation θy [degrees] End rotation θz [degrees]

Fig. 25. Typical load-rotation curves along major and minor axes (CHS HR 159 × 6.3 T11).

103
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

691 kN 691 kN
Load [kN]

Load [kN]
LVDT_1 LVDT_1
LVDT_2 LVDT_2
LVDT_Average LVDT_Average

Lateral displacement δ [mm] Transversal displacement δ [mm]

Fig. 26. Mid-span cross-section lateral and transversal displacements (RHS CF 220 × 120 × 6 T6).

Table 4
Measured dimensions and obtained ultimate loads for all tested specimens.

Cross-section shape ψ Measured dimensions FTEST

Length H B t D ez ey
[–] [mm] [mm] [mm] [mm] [mm] [mm] [mm] [kN]

RHS CF 200 × 100 × 4 T1 1 4001.22 200.12 100.28 4.00 – 84.6 0 351.92

RHS CF 200 × 100 × 4 T2 1 3999.61 200.42 100.08 4.05 – 88.9 41.6 213.54
RHS CF 200 × 100 × 4 T3 0 4000.30 198.66 99.93 3.85 – 141.0 0 365.14
RHS CF 220 × 120 × 6 T4 1 4000.09 220.09 119.77 5.90 – 72.2 0 700.00
RHS CF 220 × 120 × 6 T5 1 4000.33 219.91 120.27 5.86 – 85.4 43.8 478.20
RHS CF 220 × 120 × 6 T6 0 3998.58 219.90 119.78 5.83 – 120.3 0 691.38
CHS HR 159 × 5 T7 1 3999.70 – – 5.42 159.80 68.9 0 345.43
CHS HR 159 × 5 T8 1 4895.00 – – 5.30 159.30 77.5 0 288.53
CHS HR 159 × 5 T9 0 4000.05 – – 5.30 159.40 130.0 0 317.81
CHS HR 159 × 6.3 T10 1 4900.05 – – 6.51 159.20 78.5 0 319.59
CHS HR 159 × 6.3 T11 1 4000.10 – – 6.53 159.40 88.4 58.7 304.71
CHS HR 159 × 6.3 T12 0 4000.00 – – 6.40 159.20 130.0 0 363.00

EC3 - Curve a0 should be expected since – predominant – weak axis flexural buckling
EC3 - Curve d cumulates with weak axis bending.
1.4
Resistance Line Fig. 32 highlights the influence of the axial force level through
Instability Curve
1.2 Hayeck [15]
relative axial force parameter n = NEd/Nb,Rd = (NEd /χ Npl,Rd or NEd /χ
Grimault [5] Neff,Rd); n therefore can serve as an indicator of the part of member
1.0 Rondal [6] resistance that is necessary to resist flexural buckling alone: small n
Greiner [7]
0.8 values indicate a relative low level of compression compared to acting
Yeomans [8]
Guiaux [9] bending moments, while high n means that axial compression is
0.6 Salvarinas [10] predominant. A careful analysis of the results indicates that the factor
χ

0.4
0.2
0.0
0.0 0.4 0.8 1.2 1.6 2.0
λ element owing to instability and not because of a lack of cross-sectional
Fig. 27. Collected test results and comparison with Eurocode buckling curves.
law relative to each production route and their associated material
imperfections (higher influence of residual stresses on buckling for
cold-formed sections). Consequently, higher interaction curves should
be derived when considering buckling curves for hot-finished hollow
sections – this is typically accounted for in Eurocode 3 through higher
column buckling curves for hot-finished tubes, however solely for
compression cases. Moreover, detailed, deeper analysis of the test
data [15] for hollow sections under compression with constant mono-
axial bending moment distribution shows that these members are less
affected by instability under compression and major-axis moment (N +
My) than under compression and minor-axis moment (N + Mz), as
Braham [11]
n influences as well the member resistance and should be considered
Sedlacek [12]
Kuhn [13] too as a leading parameter for the derivation of appropriate interaction
Pavlovcic [15] curves. It is clearly shown that for a high level of compression (n > 0.3),
global buckling becomes determinant, leading to more scattered results
(λL + G > 0.6). Global buckling from higher levels of compression
2.4 occurs before cross-section full yielding, resulting in the failure of the
resistance. However, for a lower level of compression (n ≤ 0.3), the
results are closer and reach lower values of the generalised relative
slenderness λL + G; in the latter cases, bending is predominant.
mz (Mz / Mplz )
Ω y / z = arctan( ) = arctan
my (My / Mply ) (5)
The degree of biaxiality, characterized through factor Ωy/z (see Eq.
(5) where Mply, Mplz are the plastic cross-section resistances and My, Mz
are the ultimate bending moments about strong and weak axis
respectively), plays a significant role as well and further studies are
under way to sort the results accordingly.
Fig. 33 finally illustrates the application of the O.I.C. under different
(linear) bending moment distributions, represented by the ψ factor

104
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

1.2 EC3 - Curve a0

1.2
EC3 - Curve a0 EC3 - Curve d
1.0 EC3 - Curve d
1.0 Class 4 sections
Class 1-2 sections
Class 4 sections
0.8 Class 1 sections 0.8

0.6 0.6

χ
χ

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.4 0.8 1.2 1.6 2.0 2.4 0.0 0.4 0.8 1.2 1.6
λ λ
Fig. 28. Experimental results relative to section classes obtained by omitting χL for – a) hot-finished SHS and RHS – b) cold-formed SHS and RHS.

1.4 1.4
EC3 - Curve a0
EC3 - Curve a0
1.2 EC3 - Curve d
1.2 EC3 - Curve d Class 4 sections
Class 4 sections Class 1-2 sections
1.0 Class 1 sections
1.0

0.8 0.8

0.6 0.6
χ
χ

0.4 0.4

0.2 0.2
0.0 0.0
0.0 0.4 0.8 1.2 1.6 2.0 2.4 0.0 0.4 0.8 1.2 1.6
λ λ
Fig. 29. Experimental results relative to section classes obtained by considering χL for – a) hot-finished SHS and RHS – b) cold-formed SHS and RHS.

1.4
Kuhn [13] 1.4 Yeomans [8]
1.2 Rondal [6] Guiaux [9]
1.2 Salvarinas [10]
1.0 Braham [11]
1.0 Sedlacek [12]
0.8 Pavlovcic [13]
0.8 Grimault [5]
Rondal [6]
0.6 0.6
χ

Braham_annealed [11]
χ

0.4 0.4
0.2 0.2

0.0 0.0
0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 0.0 0.4 0.8 1.2 1.6 2.0
λ λ
Fig. 30. Experimental results relative to pure compression load cases – a) hot-finished SHS and RHS – b) cold-formed SHS and RHS.

which stands as the ratio between applied end moments: ψ = 1
indicates a constant bending moment distribution, ψ = 0 a triangular
distribution, and ψ = −1 to ψ = −0.5 a bi-triangular pattern. All
results are relative to beam-column members tested under combined
loading situations (either compression with mono-axial bending or
compression with bi-axial bending). Again, it is obvious that the
bending moment distribution is a leading parameter influencing the
resistance; no matter the load cases, one may obviously notice that
members subjected to a triangular bending distribution exhibit a higher
resistance than member subjected to a constant moment distribution
(see also [17] for a deeper analysis of beam-columns behavior and
response). Accordingly, the bending moment distribution shall be
accounted for in the derivation of design resistance curves.
All these results evidence the ability of the O.I.C. direct approach to
capture the behavior of beam-columns in a straightforward manner, i.e.
without resorting to section or member interaction formulae actually
used in current design codes, nor to the calculation of effective
properties. All types of experimental results presented with the various
load cases also clearly show the potential for accurate and safe χ = f (λ)
equations to be derived, and such developments are presently under
way.

105
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

Hayeck [15] - N+My - ψ=1

1.4 1.4
Greiner [7] - N+My - ψ=1 Grimault [5] - N+My - ψ=1
1.2 Rondal [6] - N+My - ψ=1 1.2 Rondal [6] - N+My - ψ=1
Greiner [7] - N+Mz - ψ=1 Grimault [5] - N+Mz - ψ=1
1.0 1.0
Rondal [6] - N+Mz - ψ=1 Rondal [6] - N+Mz - ψ=1
0.8 Pavlovcic [14] - N+Mz - ψ=1
0.8
Braham [11] - N+Mz - ψ=1
0.6 0.6
χ

χ
0.4 0.4
0.2 0.2
0.0 0.0
0.0 0.4 0.8 1.2 1.6 0.0 0.4 0.8 1.2 1.6
λ λ
Fig. 31. Experimental results relative to compression with mono-axial bending load cases for – a) hot-finished SHS and RHS – b) cold-formed SHS and RHS.

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8

0.6 0.6
χ
χ

0.4 0.4
Combined loading - ψ=1 - 0 < n < 0.3 Combined loading - ψ=1 - 0 < n < 0.3
0.2 0.2
Combined loading - ψ=1 - n >0.3 Combined loading - ψ=1 - n > 0.3
0.0 0.0
0.0 0.4 0.8 1.2 1.6 2.0 0.0 0.4 0.8 1.2 1.6 2.0
λ λ
Fig. 32. Experimental results under combined loading for – a) hot-finished SHS and RHS – b) cold-formed SHS and RHS.

1.4 1.4

1.2 1.2

1.0 1.0

0.8 0.8
0.6 0.6
χ
χ

Combined loading - ψ=1

0.4 0.4 Combined loading - ψ=1
Combined loading - ψ=0
Combined loading - ψ=0
0.2 Combined loading - ψ=−0.5 0.2 Combined loading - ψ=−1
0.0 0.0
0.0 0.4 0.8 1.2 1.6 0.0 0.4 0.8 1.2 1.6 2.0
λ λ
Fig. 33. Experimental results under combined loading for – a) hot-finished SHS and RHS – b) cold-formed SHS and RHS.

4. Conclusions iv) Measurements of the residual stresses;

Experimental investigations on tubular rectangular and circular
hollow sections have been presented in the present paper. The test
program included 12 beam-columns tests, where compression was
applied through various eccentricities. Four different cross-section
shapes have been considered, and profiles had a nominal yield stress
fy = 355 N/mm2 and were manufactured through either a hot-rolled or
cold-formed process. Preliminary measurements prior to testing were
also described in detail; they consisted in:
i) Measurements of cross-section dimensions;
ii) Measurements of geometrical imperfections;
iii) Determination of material properties;
v) Testing of stub columns.
In addition to the present experimental series, an experimental
database comprising more than 800 results was assembled from
literature. It comprises various load cases, fabrication processes and
yield strengths, cross-sections shapes, elements lengths, etc. Results
showed that the O.I.C. approach, although being quite simple and
straightforward, was suitable to characterize the behavior and response
of steel hollow section beam-columns; further, the influence of key
parameters such as the fabrication processes, bending moment dis-
tribution, axial force ratio, degree of biaxial bending Ωy/z, and steel
grade was investigated. These data, together with an extensive numer-
ical parametric study, will subsequently serve as a set of reference

106
M. Hayeck et al. Thin-Walled Structures 128 (2018) 92–107

results for the derivation of adequate interaction curves, needed for the [7] R. Greiner, et al., SEMI-COMP: Plastic member capacity of semi-compact steel
sections – a more economic design, Research Fund for Coal and Steel, European
practical prediction of beam-column resistance by means of the O.I.C. Commission, ISBN 978-92-79-11113-6, 2009.
[8] N. Yeomans, Commentary on buckling tests on square and circular structural hollow
Acknowledgements sections, Comité International pour le Développement et l'Étude de la Construction
Tubulaire, programme 2B.
[9] P. Guiaux, Essais de flambement sur profils creux formes à froid, carres et
CIDECT, by means of financial and technical support within circulaires, Comité international pour le développement et l'étude de la construc-
research project “HOLLOPOC”, is gratefully acknowledged. tion tubulaire, 2002.
[10] J.J. Salvarinas, An Experimental Investigation of the Column Behaviour of Cold-
formed Stress-relieved Hollow Structural Steel Sections, University of Toronto,
References Toronto, Canada, 1978.
[11] M. Braham, Flambement des profils creux a parois minces cas des profils
rectangulaires charges axialement, Université de Liege, Liège, Belgium, 1979.
[1] N. Boissonnade, et al., An Overall Interaction Concept for an alternative approach
[12] G. Sedlacek, Buckling behaviour of a new generation of cold-formed hollow
to steel members design, J. Constr. Steel Res 135 (2017) 199–212, http://dx.doi.
sections, Research project No 2R, CIDECT, 1996.
org/10.1016/j.jcsr.2017.02.030.
[13] G. Sedlacek, Buckling behaviour of hot-formed SHS in high strength steel grade E-
[2] N. Boissonnade, J. Nseir, E. Saloumi, The overall interaction concept: an alternative
460. Comité international pour le développement et l'étude de la construction
approach to the stability and resistance of steel sections and members, in:
tubulaire, CIDECT report 2T-2/99, 1999.
Proceedings of the Annual Stability Conference Structural Stability Research
[14] L. Pavlovcic, Finite Element Simulation of Slender Thin-walled Box Columns by
Council, St. Louis, Missouri, April 16–20, 2013.
Implementing Real Initial Conditions, Faculty of Civil and Geodetic Engineering,
[3] J. Nseir, Development of a New Design Method for the Cross-section Capacity of
University of Ljubljana, Slovenia, 2011.
Steel Hollow Sections (Ph.D. thesis), University of Liege, Belgium, 2015.
[15] M. Hayeck, Development of a New Design Method for Steel Hollow Section
[4] CEN (Comité Européen de Normalisation), Eurocode 3: Design of Steel Structures –
Members Resistance (Ph.D. thesis), University of Liege, Belgium, 2016.
Part 1.1: General Rules and Rules for Buildings (EN 1993-1-1), Brussels, 2005.
[16] R. Thiébaud, Résistance au déversement des poutres métalliques de pont (Ph.D.
[5] J. Rondal, J.P. Grimault, A. Plumier, Flambement des profils creux a parois minces,
thesis), EPFL, Lausanne, Switzerland, 2014.
cas des profils rectangulaires charges excentriquement, Comm. Des. communautés
[17] N. Boissonnade, R. Greiner, J.-.P. Jaspart, J. Lindner, Rules for Member Stability in
Eur. (2002).
EN 1993–1–1, Background documentation and design guidelines, Eds. Mem
[6] J. Rondal, Contribution à l'étude de la stabilité des profils creux à parois minces,
Martins, Portugal, E.C.C.S. – ISBN 92-9147-000-84, vol. 119, 2006, p. 259.
Université de Liège, Liège, Belgium, 1984.

107