You are on page 1of 13

Engineering Structures 270 (2022) 114871

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Local–flexural interactive buckling behaviour and resistance of


press-braked stainless steel slender channel section columns
Shuai Li , Ou Zhao *
School of Civil and Environmental Engineering, Nanyang Technological University, Singapore

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

Keywords: The present paper reports experimental and numerical studies of the local–flexural interactive buckling
Design analyses behaviour and resistance of press-braked stainless steel slender channel section columns. A testing programme
Local–flexural interactive buckling was firstly performed, including tensile coupon tests, initial geometric imperfection measurements and ten pin-
New proposal
ended column tests. The pin-ended column test setup, procedures and results were fully reported and analysed.
Numerical modelling
Pin-ended column tests
The progression of local buckling and flexural buckling was discussed in detail. Following the testing programme,
Press-braked slender channel section a numerical modelling programme was conducted, where finite element models were developed to replicate the
Stainless steel test responses and then employed to perform parametric studies to generate further numerical data over a wide
range of cross-section dimensions and member effective lengths. The obtained test and numerical data were
employed to assess the accuracy of the relevant design rules for press-braked stainless steel slender channel
section columns, as set out in the European code, American specification and Australian/New Zealand standard.
The assessment results revealed that the European code leads to conservative and scattered interactive buckling
resistance predictions, while the American specification and Australian/New Zealand standard result in many
unsafe and scattered interactive buckling resistance predictions. A new design approach was then developed,
based on the Eurocode column buckling curve and the continuous strength method, and shown to lead to ac­
curate, consistent and safe interactive buckling resistance predictions for press-braked stainless steel slender
channel section columns. The reliability of the new design approach was confirmed by means of statical analyses.
Besides, the relevant design rules for press-braked carbon steel slender channel section columns, as given in the
North American specification, were assessed, and shown to result in unsafe though consistent interactive
buckling resistance predictions, when used for their stainless steel counterparts.

1. Introduction results, the accuracy of the relevant codified design rules was assessed,
indicating conservatism, and new improved design methods were pro­
Stainless steel has been gaining increasing attention in civil and posed [3–5]. Liang et al. [6,7] conducted comprehensive experimental
offshore engineering, owing to its superior corrosion resistance and and numerical studies on laser-welded stainless steel channel section
durability, high strength and exceptional ductility [1,2]. Channel sec­ stub columns subjected to combined compression and bending moment,
tion with simple geometry is widely used in structural applications, investigated their cross-section resistances, assessed the codified design
including roof purlins, façade columns, built-up column members, lat­ provisions and proposed new accurate design rules. Fang et al. [8,9]
tice structures and framing chords. A brief summary of previous studies numerically investigated the web crippling buckling behaviour and re­
on stainless steel channel section members is firstly provided herein. sistances of press-braked stainless steel channel sections and proposed a
Specifically, Fan et al. [3] and Zhang et al. [4] investigated the local deep-learning framework. Rossi et al. [10] conducted fixed-ended
buckling behaviour and cross-section compression resistances of press- compression tests on press-braked stainless steel lipped channel sec­
braked stainless steel channel section stub columns, while Zhang et al. tion columns to study their flexural-torsional–distortional interactive
[4] and Theofanous et al. [5] studied the flexural behaviour and cross- buckling behaviour, while Becque and Rasmussen [11] performed pin-
section bending resistances of press-braked and laser-welded stainless ended compression tests on press-braked stainless steel lipped channel
steel channel section beams. On the basis of the test and numerical section columns to investigate their local–overall interactive buckling

* Corresponding author.
E-mail address: ou.zhao@ntu.edu.sg (O. Zhao).

https://doi.org/10.1016/j.engstruct.2022.114871
Received 9 November 2021; Received in revised form 18 July 2022; Accepted 19 August 2022
Available online 5 September 2022
0141-0296/© 2022 Elsevier Ltd. All rights reserved.
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

behaviour. Zhang et al. [12] and Dobrić et al. [13] studied the behaviour Table 3 reports the key measured material properties, including the
and resistances of press-braked stainless steel channel section columns Young’s modulus E, the 0.2% proof stress σ0.2, the ultimate stress σ u, the
failing by flexural buckling, highlighted the inaccuracy of the codified strain at the ultimate stress εu, the strain at fracture εf, and the param­
buckling curves and proposed new design methods. eters n and m used in the Ramberg–Osgood material model [19]. Fig. 3
To supplement the aforementioned previous studies, this paper fo­ shows the measured flat and corner material stress–strain curves.
cuses on the local–flexural interactive buckling behaviour and resistance
of press-braked stainless steel slender channel section columns. An 2.3. Initial geometric imperfection measurements
experimental programme was conducted, which included tensile coupon
tests, initial geometric imperfection measurements and ten pin-ended Initial geometric imperfections are an inherent feature of thin-walled
column tests. The experimental programme was accompanied by a nu­ steel structural members and known to influence their buckling behav­
merical modelling programme, where finite element models were firstly iour and resistances [4–6,10–12,20–22]. Therefore, the initial global
developed and validated against the experimental results and then and local geometric imperfections of each press-braked stainless steel
employed to perform parametric studies for generating additional nu­ slender channel section column specimen were measured. Fig. 4(a)
merical data over a wide range of cross-section dimensions and member displays the setup for initial global geometric imperfection measure­
effective lengths. On the basis of the obtained experimental and nu­ ments, where a column specimen is mounted onto the flat work bench of
merical data, the accuracy of the design rules for press-braked stainless a computer numerical control (CNC) router, while a linear variable
steel slender channel section columns, as given in EN 1993-1-4 [14], displacement transducer (LVDT), pointing to the centreline of the web,
SEI/ASCE 8-02 [15] and AS/NZS 4673 [16], was assessed. A new is moved along the member longitudinal direction. The initial global
improved design approach was then proposed. Moreover, the relevant geometric imperfection amplitude at each position about the minor
design rules for press-braked carbon steel slender channel section col­ principal axis was defined as the deviation from the corresponding
umns, as set out in AISI S100 [17], were assessed for the applicability to measured data point to a liner reference line (defined as a straight line
their stainless steel counterparts. connecting the data points measured at the two ends). Fig. 4(b) displays
the measured initial global geometric imperfection distribution for a
2. Experimental programme typical column specimen C1-L2. Table 2 reports the initial mid-height
global geometric imperfection amplitude ωg measured for each col­
2.1. Test specimens umn specimen. The sign convention for ωg was defined as follows:
positive if the column specimen bowed towards the flange tips – see
Given that there have been no tests on press-braked stainless steel Fig. 5(a), but negative if the column specimen bowed towards the web –
slender channel section columns, an experimental programme was see Fig. 5(b). The setup for initial local geometric imperfection mea­
firstly conducted. Two sizes of plain channel sections C 80 × 40 × 2 and surements is similar to that adopted for initial global geometric imper­
C 100 × 50 × 2, press-braked from grade EN 1.4301 austenitic stainless fection measurements, but now with three LVDTs moving transversely
steel sheets, were adopted in the experimental programme and are both along the web and flanges of the mid-height cross-section of each col­
defined as Class 4 according to the slenderness limits in EN 1993-1-4 umn specimen. The initial local geometric imperfection magnitudes for
[14]. The chemical compositions of the used grade EN 1.4301 austen­ each constituent plate element were defined as the deviations from a
itic stainless steel are reported in Table 1. For each channel section, five linear reference line (defined as a linear regression line fitted to the
column specimens with different member lengths were prepared. The whole set of measured data), while the maximum deviation obtained
label of each column specimen consists of the cross-section identifier from the three constituent plate elements was taken as the initial local
(with ‘C1’ representing C 80 × 40 × 2 and ‘C2’ representing C 100 × 50 geometric imperfection magnitude of the column specimen ω0. Fig. 6
× 2), a letter ‘L’ (indicating length) and a number (from ‘1’ to ‘5’), which displays the measured initial local geometric imperfection distributions
is used to differentiate the column specimens with the same cross- for the three constituent plate elements of a typical column specimen
section size but different member lengths. Table 2 reports the C1-L2 at mid-height. Table 2 reports the initial local geometric imper­
measured geometrical dimensions for each press-braked stainless steel fection magnitude ω0 measured for each column specimen.
slender channel section column specimen, where Bw is the outer web
width, Bf is the outer flange width, t is the wall thickness and r is the 2.4. Pin-ended column tests
corner inner radius (see Fig. 1) and L is the member length.
Pin-ended compression tests were conducted on the ten press-braked
2.2. Material tests stainless steel slender channel section column specimens. A
displacement-controlled INSTRON 5000 kN hydraulic testing machine
Tensile coupon tests were conducted to measure the material was adopted to conduct all the column tests at a displacement rate of 0.2
stress–strain curves of the adopted press-braked stainless steel slender mm/min. Both the top and bottom ends of the testing machine were
channel sections. One flat coupon and one corner coupon were cut from equipped with knife-edge devices – see Fig. 7, which provided pin-ended
each channel section and tested in a Schenck 250 kN hydraulic testing boundary conditions to the column specimen ends about the minor
machine under displacement control. The initial displacement rate was principal axis. Moreover, anchor devices (see Fig. 7) were used to clamp
taken as 0.05 mm/min until the nominal 0.2% proof stress, after which a the ends of the column specimens, aimed at preventing the end cross-
faster displacement rate of 0.8 mm/min was adopted. The resulting sections from any deformation. Prior to testing, the relative position
strain rates were in accordance with the requirements of EN ISO 6892–1 between the column specimen and the knife-edge devices was carefully
[18]. Fig. 2 depicts the tensile coupon test setup, including a pair of adjusted to achieve that (i) the cross-section minor principal axis of the
strain gauges attached to the mid-height of the coupon to record the column specimen was parallel to the knife edges and (ii) the member
longitudinal strains and an extensometer mounted onto the necked part longitudinal axis was perpendicular to and intersected with the knife
of the coupon to record the elongations over the gauge length of 50 mm. edges. Upon completion of the member alignment and position

Table 1
Chemical compositions of used grade EN 1.4301 austenitic stainless steel.
Grade C (%) S (%) P (%) Mn (%) Si (%) Cr (%) Ni (%) N (%)

EN 1.4301 0.026 0.004 0.024 1.38 0.48 18.3 8.0 0.06

2
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Table 2
Measured geometric properties and initial geometric imperfections of press-braked stainless steel slender channel section column specimens.
Cross-section Specimen ID L Bw Bf t r σcr,f σcr,w σcr,g ω0 ωg
(mm) (mm) (mm) (mm) (mm) (MPa) (MPa) (MPa) (mm) (mm)

C 80 × 40 × 2 C1-L1 298.0 80.29 40.99 2.05 4.01 267.5 653.2 2036.5 0.03 0.05
C1-L2 498.4 80.72 40.13 2.03 4.05 276.0 633.3 873.0 0.04 0.21
C1-L3 699.1 80.48 39.42 2.06 4.05 297.0 657.9 474.1 0.02 0.16
C1-L4 899.4 79.85 40.40 1.99 4.03 260.2 622.1 322.9 0.04 0.26
C1-L5 1099.8 80.40 40.72 2.08 3.98 279.6 670.3 228.0 0.05 0.18
C 100 × 50 × 2 C2-L1 298.3 101.58 50.86 2.04 4.01 160.9 375.4 3147.8 0.04 − 0.03
C2-L2 499.2 101.05 50.82 2.02 4.06 158.3 373.0 1413.6 0.03 0.12
C2-L3 698.4 101.42 50.89 2.04 3.99 160.6 376.4 804.3 0.05 0.16
C2-L4 899.5 100.73 50.92 2.01 4.05 155.9 371.7 517.5 0.04 − 0.37
C2-L5 1099.1 101.21 50.93 2.03 4.06 159.1 375.5 360.4 0.03 0.31

Table 3
Summary of key measured material properties.
Cross- Coupon E σ0.2 σu εu εf R–O
section ID parameters

(GPa) (Mpa) (Mpa) (%) (%) n m

C 80 × C1-F 201.8 281.8 754.7 48.5 60.3 8.8 1.9


40 × 2
C1-C 196.8 394.3 795.2 39.6 46.0 4.0 3.2
C 100 × C2-F 200.4 244.3 750.5 52.6 56.8 9.2 1.9
50 × 2
C2-C 201.3 403.1 810.4 45.7 51.3 3.5 3.4

Note: ‘F’ denotes flat coupon, while ‘C’ denotes corner coupon.

adjustment, the column specimen was anchored by bolting the anchor


devices to the knife-edge devices. It is worth noting that the distance
from the column specimen end to the rotation centre of the knife-edge
device is equal to 55 mm – see Fig. 7, leading to the member effective
length equal to Le = L + 110 mm.
Fig. 7 depicts the instrumentation adopted for the pin-ended column
tests, including two strain gauges adhered onto the outer faces of the
web and flange tip of the column specimen at mid-height to measure the
Fig. 1. Notations of geometric parameters of plain channel section and loca­
tions of flat and corner coupons.
member longitudinal strains and an LVDT perpendicularly pointing to
the web at mid-height to measure the lateral deflections about the minor
principal axis. On the basis of the readings of the LVDT and strain
gauges, the actual overall global geometric imperfection amplitude em
for each column specimen is calculated by Eq. (1) [12],
EI(εmax − εmin )
em = e0 + ωg = − Δ (1)
NBf

where e0 is the initial loading eccentricity at the specimen end, I is the


second moment of area about the minor principal axis, εmax-εmin is the
difference of the longitudinal strains measured by the two strain gauges,
N is the applied compression load and Δ is the mid-height lateral
deflection measured by the LVDT. It is worth noting that Eq. (1) was
derived based on the assumption that the structural behaviour was close
to linear elastic and it was suggested that no more than 15% of the ex­
pected failure loads be used for determining em [12]. The sign conven­
tion for em was defined as follows: (i) positive if the relative position
between the knife edge and the elastic neutral axis of the mid-height
cross-section was similar to that illustrated in Fig. 8(a) and (ii) nega­
tive if the relative position between the knife edge and the elastic neutral
axis of the mid-height cross-section was similar to that illustrated in
Fig. 8(b). The position of each column specimen was iteratively adjusted
until the achievement of |em|<Le/1000 [12], where |em| denotes the
absolute value of em. Table 4 reports the final value of em and the cor­
responding ratio of |em|/Le for each press-braked stainless steel slender
Fig. 2. Tensile coupon test setup. channel section column specimen.

3
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Fig. 3. Measured material stress–strain curves.

Fig. 4. Measurement of initial global geometric imperfections.


2.5. Column test results and discussions

Upon testing, all the press-braked stainless steel slender channel


section column specimens failed by local–flexural interactive buckling.
Specifically, the column specimens C1-L4 and C1-L5 exhibited local
buckling coupled with minor-axis flexural buckling in the ‘reverse C’
orientation, with typical failure mode shown in Fig. 9, while all the other
column specimens exhibited local buckling coupled with minor-axis
flexural buckling in the ‘C’ orientation, with typical failure mode
shown in Fig. 10. The measured failure load Nu,test and the associated
mid-height lateral deflection Δu for each column specimen are reported
in Table 4, while Fig. 11(a) and 11(b) display the load–mid-height
lateral deflection curves measured for the C 80 × 40 × 2 and C 100 × 50
× 2 column specimens, respectively. Note that the mid-height lateral
deflections were taken as positive if the failure specimen bowed towards
the flange tips but negative if the failure specimen bowed towards the Fig. 5. Sign convention of initial global geometric imperfection ωg.
web, in accordance with the sign convention for ωg.
The progression of local and flexural buckling was monitored using flexural buckling in the ‘C’ orientation at the failure load. Fig. 12(a)
the measured load–mid-height strain curves and according to the shows the measured load–mid-height strain curves for flange and web of
maximum mid-plane strain method [23–25]. For each C 100 × 50 × 2 a typical column specimen C2-L1. It is worth noting that the load–mid-
column specimen, noticeable localised deformation (and local buckling) height strain curve measured from the flange had a reversal point, which
firstly occurred in the compressive flanges, which was followed by indicated the occurrence of sudden local deformation (and local

4
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

different modes. As can be seen from Table 2, the elastic critical local
buckling stress for flange σcr,f was lower than the elastic critical local
buckling stress for web σcr,w and the elastic critical flexural buckling
stress σ cr,g, indicating that the local–flexural interactive buckling was
induced by local buckling of flange [26–28].
The same progression of local and flexural buckling was observed for
the C 80 × 40 × 2 column specimens with relatively short member
lengths (C1-L1, C1-L2 and C1-L3), which was also consistent with the
fact that the measured load–mid-height strain curves for flanges had
reversal points and the values of σ cr,f were lower than those of σcr,w and
σ cr,g – see Table 2. However, for the 80 × 40 × 2 column specimens with
relatively long member lengths (C1-L4 and C1-L5), flexural buckling was
firstly observed, followed by local buckling. Fig. 12(b) shows the
measured load–mid-height strain curves for flange and web of a typical
column specimen C1-L5; both the two measured load–mid-height strain
curves had no reversal points, indicating that flexural buckling firstly
occurred [23–25].

3. Numerical modelling programme

3.1. Overview

Following the experimental programme, a numerical modelling


programme was conducted by means of the finite element (FE) software
ABAQUS [29]. FE models were firstly developed to replicate the test
structural responses and then employed to perform parametric studies
for expanding the test data pool over a wider range of cross-section di­
mensions and member effective lengths.

Fig. 6. Measured distribution of initial local geometric imperfections for col­


umn specimen C1-L2 at mid-height. 3.2. Development of FE models

buckling) of the flange, before the failure load was attained and flexural Each press-braked stainless steel slender channel section column
buckling occurred [23–25]. The progression of local and flexural specimen was modelled based on the measured cross-section dimensions
buckling, as observed from the C 100 × 50 × 2 column tests, was also and member length and using the four-node shell element S4R
consistent with the relative values of elastic critical buckling stresses for [4,6–9,12,13,30–32]. A prior mesh sensitive study examining element

Fig. 7. Setup for pin-ended column tests.

5
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Fig. 8. Sign convention of actual overall global geometric imperfection em.

Table 4
Summary of pin-ended column test results.
Cross-section Specimen ID Le em |em|/Le Failure Failure Nu,test Δu
mode orientation
(mm) (mm) (kN) (mm)

C 80 × 40 × 2 C1-L1 408.0 0.09 1/4533 Driven by local buckling C 68.1 − 0.36


C1-L2 608.4 − 0.18 1/3380 Driven by local buckling C 57.7 − 1.17
C1-L3 809.1 − 0.20 1/4046 Driven by local buckling C 55.6 − 1.01
C1-L4 1009.4 0.30 1/3365 Driven by flexural buckling reverse C 51.6 1.73
C1-L5 1209.8 0.51 1/2372 Driven by flexural buckling reverse C 48.1 2.45
C 100 × 50 × 2 C2-L1 408.3 − 0.17 1/2402 Driven by local buckling C 72.8 − 0.18
C2-L2 609.2 0.25 1/2437 Driven by local buckling C 70.8 − 0.22
C2-L3 808.4 − 0.19 1/4255 Driven by local buckling C 66.0 − 0.18
C2-L4 1009.5 − 0.92 1/1097 Driven by local buckling C 60.9 − 0.83
C2-L5 1209.1 − 1.12 1/1080 Driven by local buckling C 58.1 − 1.60

sizes ranging from 0.2 t × 0.2 t to 2 t × 2 t was conducted to seek suitable Five sets of global and local imperfection magnitude combinations,
element sizes which could result in a good balance between computa­ including |em|+ω0, Le/1000 + t/10, Le/1500 + t/10, Le/1000 + t/100
tional efficiency and accuracy. The final selected element size was taken and Le/1500 + t/100, were used to scale the obtained initial geometric
equal to t × t for the flat parts of each column FE model and a finer mesh imperfection distribution patterns, in order to study the sensitivity of the
with four elements was assigned to the corner parts. Regarding the developed column FE models to imperfection magnitudes. Finally, the
material modelling of stainless steel, the plastic material model with modified ‘Riks’ method [29], considering material and geometric non­
isotropic hardening adopting von Mises yield criterion was used. It linearities, was conducted on the developed column FE models, to
required the inputted material properties to be specified in the form of obtain the numerical failure loads, load–mid-height lateral deflection
true stress and true plastic strain. Therefore, the measured (engineering) curves and failure modes.
flat and corner stress–strain curves were converted into the true
stress–plastic strain curves and then assigned to the corresponding flat 3.3. Validation of FE models
and corner regions of each column FE model.
Boundary conditions were carefully modelled to replicate the pin- Validation of the developed column FE models was conducted
ended boundary conditions used in the column tests. Firstly, each end through comparisons of the numerical results with the experimental
section of the column FE model was coupled to a concentric reference observations. Table 5 reports the numerical to test failure load ratios Nu,
point, which was offset longitudinally from the end section by 55 mm, i. FE/Nu,test for all the press-braked stainless steel slender channel section
e. the distance from the rotation centre of the knife-edge device to the column specimens. The comparison results revealed that (i) both the
end section of the column specimen. Then, the boundary conditions global and local imperfection magnitudes can affect the failure loads of
were set on the reference points; specifically, the top reference point was press-braked stainless steel slender channel section columns failing by
allowed to translate in the longitudinal direction and rotate about the local–flexural interactive buckling, (ii) the best agreement between the
minor principal axis, while the bottom reference point was only allowed test and numerical failure loads was obtained when the measured
to rotate about the same axis. imperfection magnitude combination (|em|+ω0) was used and (iii) the
Regarding the incorporation of initial geometric imperfections, a generalised imperfection magnitude combination (Le/1000 + t/100)
prior elastic eigenvalue buckling analysis [29] was conducted on each also led to accurate predictions of the test failure loads. Comparisons
column FE model to derive the lowest elastic global and local buckling between the test and numerical load–mid-height lateral deflection
mode shapes, which were taken as the initial global and local geometric curves for typical column specimens C1-L1 failing in the ‘C’ orientation
imperfection distribution patterns, respectively. It is worth noting that and C1-L5 failing in the ‘reverse C’ orientation are displayed in Fig. 13
modification of the nodal coordinates of the input file of the column FE (a) and 13(b), respectively, where the test load–deformation histories
model is another efficient way to obtain the buckling mode shapes [33]. are generally well replicated by their numerical counterparts. The test

6
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Fig. 9. Test and FE failure modes for column specimen C1-L5 (with failure in the ‘reverse C’ orientation).

Fig. 10. Test and FE failure modes for column specimen C1-L1 (with failure in the ‘C’ orientation).

7
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Fig. 12. Load–mid-height strain curves for typical column specimens.

Fig. 11. Test load–mid-height lateral deflection curves. Table 5


Comparisons of FE failure loads with test failure loads for varying initial geo­
and numerical failure modes for typical column specimens C1-L5 with metric imperfection magnitude combinations.
‘reverse C’-orientation failure and C1-L1 with ‘C’-orientation failure are Cross- Specimen Nu,FE/Nu,test
shown in Figs. 9 and 10, respectively, indicating good agreement. section ID
| Le/ Le/ Le/ Le/
Overall, it can be concluded that the developed FE models can accu­ em|+ω0 1000 + 1500 + 1000 + 1500 +
rately simulate the experimental responses of the press-braked stainless t/10 t/10 t/100 t/100
steel slender channel section column specimens and were thus deemed C 80 × C1-L1 1.01 0.93 0.93 0.99 1.00
to be validated. 40 × 2
C1-L2 1.03 0.94 0.97 0.96 1.00
C1-L3 0.97 0.87 0.90 0.88 0.91
3.4. Parametric studies C1-L4 1.03 0.98 1.00 0.98 1.00
C1-L5 1.01 0.95 0.98 0.95 0.98
C 100 × C2-L1 0.98 0.88 0.88 0.99 1.00
The validated column FE models were adopted to perform para­ 50 × 2
metric studies, aiming at expanding the test data pool over a wider range C2-L2 1.00 0.87 0.89 1.00 1.00
of cross-section dimensions and member effective lengths. The model­ C2-L3 1.03 0.91 0.92 1.01 1.03
C2-L4 1.00 0.92 0.94 0.99 1.03
ling assumptions, procedures and techniques were the same as those
C2-L5 0.99 0.92 0.95 0.99 1.03
described in Section 3.2, but with the generalised imperfection magni­ Mean 1.01 0.92 0.94 0.98 1.00
tude combination (Le/1000 + t/100) and the material properties of C COV 0.02 0.04 0.04 0.04 0.03
100 × 50 × 2 used for all the FE models in the present parametric
studies. With regard to the geometric dimensions of the modelled

8
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

stainless steel slender channel section columns failing by local–flexural


interactive buckling, as set out in the European code EN 1993-1-4 [14],
American specification SEI/ASCE 8-02 [15] and Australian/New Zea­
land standard AS/NZS 4673 [16], was assessed through comparing the
test and numerical failure loads Nu against the unfactored resistance
predictions Nu,pred. The shortcomings of the codified design rules were
highlighted, followed by development of a new improved design
approach. Moreover, the relevant design rules for press-braked carbon
steel slender channel section columns, as given in the North American
specification AISI S100 [17], were also assessed for the applicability to
their stainless steel counterparts.

4.2. EN 1993-1-4 (EC3)

The European code EN 1993-1-4 [14] provides design rules for


different types of stainless steel components. Specifically, for the design
of non-slender channel section columns failing by flexural buckling,
buckling curves, as developed based on the Perry-Robertson buckling
formula, are used, while for the design of slender channel sections prone
to local buckling, effective width method is adopted. For slender channel
section columns susceptible to local–flexural interactive buckling, EN
1993-1-4 [14] firstly combines the use of buckling curves and effective
width method to account for the interaction between local and flexural
buckling in the design. Moreover, since the effective and elastic neutral
axes of slender channel section do not coincide, an additional bending
moment is induced due to the shift in neutral axis and shall be consid­
ered. In summary, a slender channel section column under pure
compression is required to be designed as a beam-column under com­
bined compression and bending moment, based on the design interac­
tion curve expressed by Eq. (2), where NEC3,Rd is the EC3 column flexural
buckling resistance, eN,EC3 is the shift in neutral axis and calculated
based on the effective width method specified in EN 1993-1-5 [34],
MEC3,Rd is the EC3 cross-section bending resistance and kEC3 is the
interaction factor given by Eq. (3), where λ̄ is the member non-
dimensional slenderness about the axis of flexural buckling and deter­
mined by Eq. (4), where Aeff,EC3 is the effective cross-section area and
calculated based on the effective width method specified in EN 1993-1-5
[34] and NEu = π2 EI/L2e is the Euler buckling load.
Nu,pred Nu,pred eN,EC3
+ kEC3 ⩽1 (2)
NEC3,Rd MEC3,Rd

Nu,pred Nu,pred
Fig. 13. Test and FE load–mid-height lateral deflection curves for typical col­ 1.2⩽kEC3 = 1 + 2(λ̄ − 0.5) ⩽1.2 + 2 (3)
NEC3,Rd NEC3,Rd
umn specimens.
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Aeff ,EC3 σ 0.2
channel sections, the outer web widths were fixed at 240 mm and the λ̄ = (4)
NEu
outer flange widths were varied between 80 mm and 240 mm, leading to
a range of cross-section aspect ratios from 1.0 to 3.0 being considered; The EC3 column flexural buckling resistance NEC3,Rd is determined
moreover, the wall thicknesses ranged from 4 mm to 18 mm, with the from Eq. (5) as the product of the effective cross-section compression
corner inner radii equal to the wall thicknesses. Note that the modelled resistance Aeff,EC3σ 0.2 and the reduction factor for flexural buckling χ ,
channel sections were categorised as Class 4 in accordance with the which is dependent on the EC3 design column buckling curve defined by
slenderness limits set out in EN 1993-1-4 [14]. The member effective Eq. (6), where ϕ is a buckling parameter and given by (7), where α is the
lengths were varied between 450 mm and 15300 mm, resulting in a wide imperfection factor and λ̄0 is the limiting slenderness, reflecting the
range of member non-dimensional slendernesses. For each channel degree of influence of initial geometric imperfection and residual stress
section column FE model, two orientations of initial global geometric on the column buckling resistance; for press-braked stainless steel
imperfection shape were modelled, leading to both ‘C’- and ‘reverse C’- channel section columns with minor-axis flexural buckling, α is equal to
orientation failure being considered. A total of 120 numerical data on 0.76 and λ̄0 is equal to 0.20. The EC3 cross-section minor-axis bending
press-braked stainless steel slender channel section columns have been resistance MEC3,Rd is defined as the 0.2% proof stress σ0.2 multiplied by
generated. the minor-axis effective section modulus Weff,EC3, which is calculated
based on the effective width method specified in EN 1993-1-5 [34].
4. Design analyses
NEC3,Rd = χ Aeff ,EC3 σ0.2 (5)

4.1. Overview 1
χ= √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅⩽1 (6)
2
In this section, the accuracy of the design rules for press-braked ϕ2 + ϕ2 − λ̄

9
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

[ 2]
ϕ = 0.5 1 + α(λ̄ − λ̄0 ) + λ̄ (7)

The accuracy of the EC3 design rules for press-braked stainless steel
slender channel section columns failing by local–flexural interactive
buckling was assessed based on the obtained test and numerical data.
Table 6 reports the mean test and numerical to predicted failure load
ratios Nu/Nu,pred and the corresponding COVs, arranged by failure
orientation. The assessment results revealed that the EC3 design rules
result in conservative and scattered interactive buckling resistance
predictions. This is also evident in Fig. 14, where the ratios Nu/Nu,pred are
plotted against the member non-dimensional slendernessesλ̄. The high
level of conservatism and scatter of the EC3 design rules can be princi­
pally attributed to the rather conservative treatment of local buckling
based on the 0.2% proof stress, i.e. the effective cross-section geometric
properties, including eN,EC3 and Aeff,EC3, are determined at the 0.2%
proof stress rather than the actual column flexural buckling stress.

4.3. SEI/ASCE 8-02 (ASCE) and AS/NZS 4673 (AS/NZS)

The American specification SEI/ASCE 8-02 [15] and Australian/New


Zealand standard AS/NZS 4673 [16] use the same design rules for cold-
formed stainless steel structures. Regarding slender channel section
columns susceptible to local–flexural interactive buckling, they employ
a similar design concept – the combined use of effective width method
(for treatment of local buckling) and buckling curve (for treatment of
flexural buckling) as well as interaction curve (for consideration of shift
in effective neutral axis) – as that adopted in EN 1993-1-4 [14]. The
design interaction curve is defined by Eq. (8), where NA,Rd is the ASCE
(or AS/NZS) column flexural buckling resistance, eN,A is the shift in
neutral axis, MA,Rd is the ASCE (or AS/NZS) cross-section bending
resistance and 1/αN = 1/(1-Nu,pred/NEu) is the magnification factor.
Nu,pred Nu,pred eN,A
+ ⩽1 (8)
NA,Rd αN MA,Rd
The ASCE (or AS/NZS) column flexural buckling resistance NA,Rd is
taken as the product of the effective cross-section area Aeff,A and the
design flexural buckling stress Fn; note that Fn is determined by means of
the tangent modulus approach expressed by Eq. (9), where re is the
radius of gyration about the minor principal axis and Et is the tangent
modulus of the Ramberg–Osgood material model at Fn, as given by Eq.
(10), where n is a parameter used in the Ramberg–Osgood material Fig. 14. Comparisons of test and FE failure loads with interactive buckling
model. The ASCE (or AS/NZS) cross-section minor-axis bending resis­ resistance predictions from EN 1993-1-4.
tance MA,Rd is equal to the 0.2% proof stress σ 0.2 multiplied by the minor-
axis effective section modulus Weff,A. It is worth noting that the ASCE (or The quantitative and graphical assessment results are respectively pre­
AS/NZS) effective width method [15,16] specifies that Weff,A is calcu­ sented in Table 6 and Fig. 15, indicating that the ASCE (or AS/NZS)
lated at σ 0.2, while eN,A and Aeff,A are calculated at Fn (which is no greater design rules lead to unsafe and scattered interactive buckling resistance
than σ 0.2). predictions, which are mainly owing to the overprediction of the design
π 2 Et flexural buckling stress by the tangent modulus approach [15,16].
Fn = ( )2 ⩽σ0.2 (9)
Le /re
4.4. New proposal
Eσ0.2
Et = ( )n− (10)
The codified design rules were found to result in inaccurate and
1
Fn
σ 0.2 + 0.002nE σ0.2
scattered interactive buckling resistance predictions for press-braked
The ASCE (or AS/NZS) design failure loads for press-braked stainless stainless steel slender channel section columns. Moreover, the concept
steel slender channel section columns failing by local–flexural interac­ of designing slender section column as beam-column based on interac­
tive buckling were assessed against the test and numerical failure loads. tion curve leads to cumbersome calculations. To address the

Table 6
Comparisons of test and FE failure loads with predicted interactive buckling resistances from different design methods.
Failure orientation No. of data EN 1993-1-4 SEI/ASCE 8-02 (or AS/NZS 4673) New proposal AISI S100

Test FE Mean COV Mean COV Mean COV Mean COV

C 8 60 1.61 0.40 1.00 0.30 1.10 0.08 0.84 0.10


Reverse C 2 60 1.69 0.38 0.98 0.18 1.23 0.06 0.91 0.10
Total 10 120 1.65 0.39 0.99 0.25 1.16 0.10 0.88 0.10

10
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

results are reported in Table 6, with the mean Nu/Nu,pred ratio equal to
1.16 and the COV of 0.10, revealing much more consistent as well as
accurate and safe interactive buckling resistance predictions for press-
braked stainless steel slender channel section columns than the codi­
fied design rules. This is also evident from the graphical evaluation re­
sults shown in Fig. 16. It is worth highlighting that the proposed design
method is specific for stainless steel slender channel section columns,
while its applicability to other metallic materials may need further
investigations.
The reliability of the proposed design method for press-braked
stainless steel slender channel section columns was assessed in accor­
dance with the requirements and procedures set out in EN 1990 [38]. In
the present reliability analyses, the material over-strength ratio for
austenitic stainless steel was equal to 1.3, with the COV of 0.06, and the
COV of the geometric dimensions of stainless steel members was equal to
0.05 [39]. Table 7 reports the key calculated statistical parameters,
where kd,n is the design (ultimate limit state) fractile factor, b is the mean
ratio of test and numerical to design model resistances, Vδ is the COV of
the test and numerical resistances relative to the resistance model, Vr is
the combined COV incorporating all variable uncertainties and γ M1 is the
(required) partial safety factor. The calculated (required) partial safety

Fig. 15. Comparisons of test and FE failure loads with interactive buckling
resistance predictions from SEI/ASCE 8-02 (or AS/NZS 4673).

shortcomings of the codified design rules, a new design method is pro­


posed in this section. The local–flexural interactive buckling resistance is
now calculated in two steps:
Step 1: Determine the flexural buckling resistance NEC3,Rd based on
the EC3 design column buckling curve, but with the use of the gross
cross-section area A;
Step 2: Reduce the flexural buckling resistance NEC3,Rd to account for
local buckling, based on the continuous strength method, as given by Eq.
(11) [35], where λ̄l,p is the local buckling slenderness and determined
from Eq. (12), where σcr,c is the elastic critical local buckling stress of
channel section in compression [36,37].


⎪ NEC3,Rd for λ̄l,p ⩽0.68
⎨( )
Nu,pred = 0.222 1 (11)

⎩ 1 − 1.050
⎪ N
1.050 EC3,Rd
for λ̄l,p > 0.68
λ̄l,p λ̄l,p
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅
NEC3,Rd
λ̄l,p = (12)
Aσ cr,c

The proposed design method for press-braked stainless steel slender


channel section columns failing by local–flexural interactive buckling
was evaluated through comparisons of the test and numerical failure
Fig. 16. Comparisons of test and FE failure loads with interactive buckling
loads against the predicted resistances. The quantitative evaluation resistance predictions from new proposal.

11
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Table 7
Reliability analysis results calculated according to EN 1990.
Design method No. of test and FE data kd,n b Vδ Vr γM1

New proposal 130 3.17 1.09 0.09 0.12 1.02

factor for the proposed design method, as reported in Table 7, is less than
the current limit value of 1.1 used in EN 1993–1-4 [14], therefore
demonstrating the reliability of the proposed design method for press-
braked stainless steel slender channel section columns.

4.5. AISI S100 (AISI)

The North American specification AISI S100 [17] provides design


rules for cold-formed carbon steel structural members and connections.
Regarding slender channel section columns failing by local–flexural
interactive buckling, AISI S100 [17] specifies two steps in calculating
the load-carrying capacity:
Step 1: Determine the flexural buckling resistance NAISI,Rd = AFcr,
where Fcr is the critical stress and calculated from the AISI strength curve
defined by Eq. (13), where λ̄ is calculated based on Eq. (4), but with the
use of the gross cross-section area A;
⎧ ( 2
)
⎪ λ̄
⎨ 0.658 σ 0.2 for λ̄⩽1.5

Fcr = ( ) (13)

⎪ 0.877
⎩ 2
σ 0.2 for λ̄ > 1.5
λ̄
Step 2: Reduce the flexural buckling resistance NAISI,Rd to account for
local buckling [40], as given by Eq. (14), where λ̄l,AISI is the AISI local
buckling slenderness and determined by Eq. (15).


⎪ NAISI,Rd for λ̄l,AISI ⩽0.776
⎨( )
Nu,pred = 0.15 1 (14)

⎩ 1 − 0.8
⎪ 0.8
NAISI,Rd for λ̄l,AISI > 0.776
λ̄l,AISI λ̄l,AISI
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅
NAISI,Rd
λ̄l,AISI = (15)
Aσcr,c

The applicability of the AISI design rules to press-braked stainless


steel slender channel section columns was assessed by comparing the
predicted interactive buckling resistances against the test and numerical Fig. 17. Comparisons of test and FE failure loads with interactive buckling
failure loads. The quantitative and graphical assessment results are re­ resistance predictions from AISI S100.
ported in Table 6 and Fig. 17, respectively, revealing that the AISI design
rules result in consistent interactive buckling resistance predictions 4673 [16] result in many unsafe and scattered interactive buckling
when used for press-braked stainless steel slender channel section col­ resistance predictions. A new design approach was then proposed and
umns, but the majority of the predictions are unsafe. shown to yield a much higher level of design accuracy and consistency
than the existing design codes. The reliability of the new design
5. Conclusions approach was also confirmed by means of statical analyses. Moreover,
the relevant design rules for press-braked carbon steel slender channel
An experimental and numerical modelling programme has been section columns, as specified in AISI S100 [17], were assessed for the
conducted to investigate the local–flexural interactive buckling behav­ applicability to their stainless steel counterparts, with the assessment
iour and resistance of press-braked stainless steel slender channel sec­ results indicating unsafe though consistent design.
tion columns. The experimental programme included tensile coupon
tests, initial geometric imperfection measurements and ten pin-ended CRediT authorship contribution statement
column tests. The experimental results were systematically analysed
and then employed in the numerical modelling programme to validate Shuai Li: Conceptualization, Formal analysis, Investigation, Meth­
FE models, which were then used to perform parametric studies to odology, Visualization, Writing – original draft. Ou Zhao: Conceptual­
generate further numerical data over a wide range of cross-section di­ ization, Formal analysis, Funding acquisition, Investigation,
mensions and member effective lengths. On the basis of the obtained Methodology, Resources, Supervision, Writing – review & editing.
experimental and numerical data, the accuracy of the design rules for
press-braked stainless steel slender channel section columns, as set out Declaration of Competing Interest
in EN 1993-1-4 [14], SEI/ASCE 8-02 [15] and AS/NZS 4673 [16], was
evaluated. The quantitative and graphical evaluation results revealed The authors declare that they have no known competing financial
that (i) EN 1993-1-4 [14] leads to conservative and scattered interactive interests or personal relationships that could have appeared to influence
buckling resistance predictions and (ii) SEI/ASCE 8-02 [15] and AS/NZS the work reported in this paper.

12
S. Li and O. Zhao Engineering Structures 270 (2022) 114871

Data availability [18] ISO EN. 6892-1. Metallic materials-Tensile testing-Part 1: Method of test at room
temperature. International Organization for Standardization, 2009.
[19] Arrayago I, Real E, Gardner L. Description of stress–strain curves for stainless steel
Data will be made available on request. alloys. Mater Des 2015;87:540–52.
[20] Chen M-T, Young B, Martins AD, Camotim D, Dinis PB. Experimental investigation
Acknowledgements on cold-formed steel stiffened lipped channel columns undergoing
local–distortional interaction. Thin-Walled Struct 2020;150:106682. https://doi.
org/10.1016/j.tws.2020.106682.
The authors thank Bahru Stainless SDN. BHD. for sponsoring press- [21] dos Santos ES, Batista EM, Camotim D. Experimental investigation concerning
braked stainless steel slender channel section column specimens. The lipped channel columns undergoing local–distortional–global buckling mode
interaction. Thin-Walled Structures 2012;54:19–34.
assistances from Mr Chelladurai Subasanran, Mr Lim Yong Cheng and [22] Santos WS, Landesmann A, Camotim D. Distortional strength of end-bolted CFS
Mr Chin Xin Yu Melvin during tests are also acknowledged. lipped channel columns: experimental investigation, numerical simulations and
DSM design. Thin-Walled Struct 2020;148:106469. https://doi.org/10.1016/j.
tws.2019.106469.
References [23] Shi G, Zhang Z, Zhou Le, Yang Lu, Zhou W. Experimental and numerical
investigation on local–overall interactive buckling behavior of welded I-section
[1] Baddoo NR. Stainless steel in construction: a review of research, applications, steel columns. Thin-Walled Struct 2020;151:106763. https://doi.org/10.1016/j.
challenges and opportunities. J Constr Steel Res 2008;64(11):1199–206. tws.2020.106763.
[2] Gardner L. Stability and design of stainless steel structures – review and outlook. [24] Vann WP, Sehested J. Experimental techniques for plate buckling. In: in: 2nd
Thin-Walled Struct 2019;141:208–16. International Specialty Conference on Cold-Formed Steel Structures; 1973.
[3] Shenggang F, Yuelin T, Baofeng Z, Fang L. Capacity of stainless steel lipped C- p. 83–105.
section stub column under axial compression. J Constr Steel Res 2014;103:251–63. [25] Roorda J, Venkataramaiah KR. Analysis of local plate buckling experimental data.
[4] Zhang L, Tan KH, Zhao O. Local stability of press-braked stainless steel angle and In: in: 6th International Specialty Conference on Cold-Formed Steel Structures;
channel sections: testing, numerical modelling and design analysis. Eng Struct 1982. p. 45–74.
2020;203:109869. [26] Wadee MA, Bai Li. Cellular buckling in I-section struts. Thin-Walled Structures
[5] Theofanous M, Liew A, Gardner L. Experimental study of stainless steel angles and 2014;81:89–100.
channels in bending. Structures 2015;4:80–90. [27] Bai Li, Wadee MA. Mode interaction in thin-walled I-section struts with semi-rigid
[6] Liang Y, Zhao O, Long Y-L, Gardner L. Stainless steel channel sections under flange–web joints. Int J Non Linear Mech 2015;69:71–83.
combined compression and minor axis bending–part 1: experimental study and [28] Shen J, Wadee MA, Sadowski AJ. Interactive buckling in long thin-walled
numerical modelling. J Constr Steel Res 2019;152:154–61. rectangular hollow section struts. Int J Non Linear Mech 2017;89:43–58.
[7] Liang Y, Zhao O, Long Y-L, Gardner L. Stainless steel channel sections under [29] ABAQUS. ABAQUS/standard user’s manual. Version 6.17. Dassault Systemes
combined compression and minor axis bending–Part 2: parametric studies and Simulia Corp. USA; 2017.
design. J Constr Steel Res 2019;152:162–72. [30] Roy K, Chen B, Fang Z, Uzzaman A, Chen X, Lim JB. Local and distortional buckling
[8] Fang Z, Roy K, Chi Y, Chen B, Lim JBP. Finite element analysis and proposed design behavior of back-to-back built-up aluminium alloy channel section columns. Thin-
rules for cold-formed stainless steel channels with web holes under end-one-flange Walled Struct 2021;163:107713.
loading. Structures 2021;34:2876–99. [31] Peiris M, Mahendran M. Behavior of cold-formed steel lipped channel sections
[9] Fang Z, Roy K, Ma Q, Uzzaman A, Lim JBP. Application of deep learning method in subject to eccentric axial compression. J Constr Steel Res 2021;184:106808.
web crippling strength prediction of cold-formed stainless steel channel sections [32] Vy ST, Mahendran M, Sivaprakasam T. Built-up back-to-back cold-formed steel
under end-two-flange loading. Structures 2021;33:2903–42. compression members failing by local and distortional buckling. Thin-Walled
[10] Rossi B, Jaspart J-P, Rasmussen KJR. Combined distortional and overall flexural- Struct 2021;159:107224.
torsional buckling of cold-formed stainless steel sections: experimental [33] Shen J, Wadee MA. Sensitivity of elastic thin-walled rectangular hollow section
investigations. J Struct Eng 2010;136(4):354–60. struts to manufacturing tolerance level imperfections. Eng Struct 2018;170:
[11] Becque J, Rasmussen KJR. Experimental investigation of local–overall interaction 146–66.
buckling of stainless steel lipped channel columns. J Constr Steel Res 2009;65(8-9): [34] EN 1993-1-5. Eurocode 3: Design of steel structures – Part 1–5: Plated structural
1677–84. elements. Brussels: European Committee for Standardization (CEN); 2015.
[12] Zhang L, Tan KH, Zhao O. Press-braked stainless steel channel section columns [35] Zhao O, Afshan S, Gardner L. Structural response and continuous strength method
failing by flexural buckling: testing, numerical simulation and design. Thin-Walled design of slender stainless steel cross-sections. Eng Struct 2017;140:14–25.
Struct 2020;157:107066. [36] Schafer BW, Ádány S. Buckling analysis of cold-formed steel members using
[13] Dobrić J, Ivanović J, Rossi B. Behavior of stainless steel plain channel section CUFSM: conventional and constrained finite strip methods. Eighteenth
columns. Thin-Walled Struct 2020;148:106600. International Specialty Conference on Cold-Formed Steel Structures. 2006.
[14] EN 1993-1-4:2006+A2:2020. Eurocode 3: Design of steel structures – Part 1.4: [37] Fieber A, Gardner L, Macorini L. Formulae for determining elastic local buckling
General rules – Supplementary rules for stainless steels, including amendment A2 half-wavelengths of structural steel cross-sections. J Constr Steel Res 2019;159:
(2020). Brussels: European Committee for Standardization (CEN); 2020. 493–506.
[15] SEI/ASCE 8-02. Specification for the Design of Cold-Formed Stainless Steel [38] EN 1990. Eurocode – basis of structural design. Brussels: European Committee for
Structural Members, American Society of Civil Engineers (ASCE), Reston, 2002. Standardization (CEN); 2002.
[16] AS/NZS 4673. Cold-formed stainless steel structures. Vol. 4673, AS/NZS, Sydney, [39] Afshan S, Francis P, Baddoo NR, Gardner L. Reliability analysis of structural
2001. stainless steel design provisions. J Constr Steel Res 2015;114:293–304.
[17] AISI S100. North American specification for the design of cold-formed steel [40] Schafer BW. Advances in the Direct Strength Method of cold-formed steel design.
structural members. American Iron and Steel Institute, 2016. Thin-Walled Struct 2019;140:533–41.

13

You might also like