You are on page 1of 11

Journal of Constructional Steel Research 196 (2022) 107395

Contents lists available at ScienceDirect

Journal of Constructional Steel Research


journal homepage: www.elsevier.com/locate/jcsr

Biaxial bending design of solid steel storage rack uprights in global buckling
Nima Talebian a, *, Benoit P. Gilbert b, Dane Miller a, Hassan Karampour b
a
Faculty of Society and Design, Bond University, Australia
b
School of Engineering and Built Environment, Griffith University, Australia

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

Keywords: This study aims at investigating the biaxial bending capacity of cold-formed steel (CFS) storage rack uprights
Cold-formed steel undergoing global (lateral-torsional) buckling. A previously published and validated biaxial bending numerical
Biaxial bending model for local and distortional buckling of CFS rack uprights is initially validated in this paper for global
Global (lateral-torsional) buckling
buckling and then used to evaluate and analyse the biaxial bending response of nine unperforated upright cross-
Finite element modelling
Direct strength method (DSM)
sections. Nine biaxial bending configurations were considered per cross-section. Results show that a nonlinear
interactive relationship typically governs the biaxial bending of the studied uprights. This relationship is dis­
cussed and analysed for different cross-sectional slenderness. The numerical capacities are compared with both
the classical DSM predictions and the use of inelastic reserve capacity in the DSM. Results show that the classical
DSM equations underestimate the biaxial bending capacity by up to 55% but results in better predictions when
the inelastic reserve capacity is considered. Finally, an extended range of inelastic reserve strength for global
buckling is proposed which results in a 14% improvement of the DSM predictions. The appropriate form of the
DSM to estimate the global biaxial bending capacity of unperforated cold-formed steel storage rack uprights is
discussed.

1. Introduction equations in [12] were evaluated. Bernuzzi and Maxenti [13] and Ber­
nuzzi and Simoncelli [14] evaluated different European design proced­
Clad-racks, also referred to as rack-supported buildings, are a rela­ ures of beam-column storage rack uprights and investigated the
tively efficient type of buildings for which the racking is integrated into geometry of the cross-sections, perforations, the upright slenderness and
the building enclosure, resulting in more economical buildings but also loading condition. Zhao et al. [15] experimentally investigated the
complex structural systems. With their growing applications, it is response of storage rack uprights subjected to axial compression and
necessary that the structural behaviour of these systems and their ele­ studied the effect of perforations on the failure mode of rack sections.
ments is studied in more detail to produce safe and economical design Baldassino et al. [16] studied rack uprights in compression and bending.
guidelines. In these structures, the uprights, i.e., the vertical members The authors investigated the influence of perforations and evaluated key
which are usually perforated monosymmetric cold-formed steel (CFS) design geometric parameters.
open sections, undergo biaxial bending due to the combined actions of Recently, Talebian et al. [1,2] experimentally and numerically
the wind and the vertical loads of the stored goods [1,2]. Unlike the investigated the local and distortional biaxial bending behaviour of
current CFS structures design specifications [3–5] that consider a linear perforated and unperforated CFS storage rack uprights. Parametric
interaction equation to design members in biaxial bending, several nu­ studies were performed on unperforated cross-sections to evaluate the
merical and experimental studies have reported that the linear equation accuracy of the linear biaxial bending design equation, as well as several
is conservative, and a nonlinear relationship governs the biaxial bending forms of the Direct Strength Method (DSM) predictions. Results revealed
response of cold-formed steel members [1,2,6–11]. that the linear interaction equation is conservative and underestimates
For steel storage racks in global buckling, the response of long the biaxial bending capacity by up to 39% and 46% for local and
storage rack uprights subjected to compression and biaxial bending was distortional buckling, respectively. However, global buckling was not
experimentally investigated by Kumar and Jayachandran. The current investigated.
linear interaction equation and the proposed DSM beam-column design The main attempt of this paper is to capture the global (lateral-

* Corresponding author.
E-mail address: ntalebia@bond.edu.au (N. Talebian).

https://doi.org/10.1016/j.jcsr.2022.107395
Received 10 February 2022; Received in revised form 19 June 2022; Accepted 19 June 2022
Available online 25 June 2022
0143-974X/© 2022 Elsevier Ltd. All rights reserved.
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

torsional) biaxial bending behaviour of CFS storage uprights through ends of the upright, to reproduce the simply supported test setup
numerical parametric studies. The biaxial bending Finite Element (FE) and constant bending moment over the length of the uprights.
model for local and distortional buckling of CFS rack uprights in [1] is II. Material modelling: Material nonlinearity in the specimens was
validated herein for global buckling. The model is then used to run the considered using the von Mises yield criterion and isotropic
analyses and quantify the global biaxial bending responses and capac­ hardening. A multilinear stress-strain relationship was used for
ities of unperforated uprights with a wide range of slenderness values. the material properties of the flat parts of the cross sections, with
Nine different cross-sections are considered, and analyses are run for nominal properties used in [1] when running parametric studies
nine biaxial bending configurations per cross-section type. Biaxial (Young's modulus of 200 GPa and a yield stress of 450 MPa).
bending responses of all studied uprights are discussed and presented in Elastic–perfectly plastic behaviour was assumed for the corner
the paper. zones of the upright sections with enhanced yield strength
Ultimately, the accuracy of the DSM in predicting the global biaxial determined by the equations proposed in [27]. These equations
bending capacity of unperforated cold-formed steel storage rack up­ were derived from coupon tests and therefore consider the re­
rights is assessed. The numerical failure moments obtained from the sidual stress. True stress and strain were employed in the nu­
parametric study are compared with DSM-based approaches. These merical model.
approaches consist of (i) the classical DSM equations given in AISI-S100 Geometric imperfections: Pure global geometric imperfections
[3] and AS/NZS 4600 [4], (ii) considering the inelastic reserve capacity were introduced to the perfect mesh by using the first global
for compact cross-sections, as permitted in the AISI-S100 [3] and AS/ elastic buckling mode when the upright was subjected to the
NZS 4600 [4], and (iii) using a new extended range of the cross-sectional biaxial bending moment of interest. The elastic buckling
slenderness for the inelastic reserve capacity in global buckling, similar deformed shapes were scaled based on an amplitude of L/1000
to the one proposed by Pham and Hancock [17] for local and distortional [28], where L is the length of the upright.
buckling. III. Analysis: The Modified Riks method was selected to perform
While storage rack uprights are typically perforated sections, this geometric and material nonlinear analyses [24]. The analyses
study aims to investigate the global biaxial bending response of solid were run statically to determine the peak failure moment.
rack uprights only. Influence of perforations on biaxial bending response
of rack uprights requires further investigation when the pure biaxial Fig. 1 shows the FE model and boundary conditions for a typical
bending behaviour of solid rack uprights is understood. upright. The applied moment was about global Y-axis and the upright
was rotated by an angle α about its centroidal axis to apply biaxial
2. Biaxial bending finite-element model for global buckling bending moment, as shown in Fig. 1.

2.1. Published experimental tests (background) 2.3. Finite element validation of the model for global buckling

In Talebian et al. [2], 78 tests were performed on short (local The experimental tests performed by Put et al. [29] on ten biaxial
buckling mode) and medium length (distortional buckling mode) bending configurations and cold-formed steel Zed beams are used herein
perforated and unperforated storage rack uprights under biaxial as the only available experimental tests in the literature on pure global
bending. Two different types of storage rack upright cross-sections, biaxial bending to provide the extra level of confidence on the accuracy
referred to as Type A (semi-compact cross-section) and Type B of the FE model for global buckling. Fig. 2 shows the FE model and
(compact cross-section), were tested. Steel plates were welded to both boundary conditions for the tests conducted by Put et al. [29]. The
ends of the uprights to connect to the test rig and restrain warping. Seven beams were simply supported, with partially restrained warping at the
different biaxial bending configurations per upright type were tested supports, and loaded at mid-span, 60 mm below the shear centre. To
and the ultimate failure moments were obtained. Cross-sectional de­ simulate the partially restrained warping at the supports, the nodes of
formations were measured at the mid-length of each specimen. The full the web at each end of the beam, over a 50 mm wide area [30], were
experimental setup and results are detailed in [2]. constrained as shown in Fig. 2. The load was applied at the same location
as the experiment. The inclinations of the load to the web of the Z-sec­
2.2. FE model tion varied from − 14 to 45 degree to generate biaxial bending. Element
type and mesh size remained the same as in Section 2.2. In [29], the
Finite element modelling of cold-formed steel members has been maximum global geometric imperfection was measured and included in
well established in the literature and the FE model in this study follows the model by appropriately scaling the first elastic buckling mode when
the recommendations in [18–23]. The non-linear biaxial bending re­ the Z-section was subjected to the biaxial bending moment of interest.
sponses of the uprights tested in [2] for local and distortional buckling Measured tensile test results obtained from coupon tests (for both flat
were numerically reproduced in [1] using the software ABAQUS [24]. parts and corners) were used in the FE model, therefore they included
The FE model developed for local and distortional buckling is adapted the effect of the residual stress. See Put et al. [29] for further information
herein for global buckling and then used to investigate the biaxial on the experimental setup.
bending behaviour of steel storage rack uprights in global buckling. The Table 1 presents the ten biaxial bending test results against predicted
main characteristics of the global biaxial bending model are described as FE failure loads. As shown in Table 1, the developed FE model can
follows: accurately predict the global buckling biaxial bending response with an
average test-to-predicted bending strength of 0.97 and an associated
I. Element type, mesh size and boundary conditions: The uprights and coefficients of variation (COV) of 4%. Fig. 3 presents the FEA and
their end plates were modelled using four-node S4R shell ele­ experimental load versus lateral deflection at mid-span for ZL25A14 and
ments. Convergence studies were performed on all uprights in ZL25A37. The FEA model is also able to capture well the lateral
bending about major axis. An element size of approximately 3 deflection observed in the biaxial bending experimental tests.
mm × 3 mm was found adequate for these cases. This mesh size
was used for all upright sections in different configurations [25]. 3. Parametric studies of biaxial bending failing in global
Similar to the experimental tests, warping was restrained by buckling
using plates rigidly connected to the ends of the uprights
[1,2,26]. A concentrated biaxial bending moment was then Similar to local and distortional buckling in [1], parametric studies
applied at the location of the upright's centroidal axis, at both were performed over a wide range of cross-sectional shapes and

2
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Fig. 1. FE model and boundary conditions (shown for Type M upright in Section 4.1).

Fig. 2. FE model and boundary conditions for biaxial bending experiments of Z beams [29].

3
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Table 1
Biaxial bending test results in [29] and predicted FE failure strength.
Test Section Length (mm) Load to web inclination, α (degrees) Exp. failure moment (kN.m) FE failure moment (kN.m) MEXP / MFEA

ZL19A − 14 Z10019 1900 − 14 2.81 2.83 0.99


ZL19A0 Z10019 1900 0 3.25 3.19 1.02
ZL19A14 Z10019 1900 14 4.06 4.33 0.94
ZL19A37 Z10019 1900 37 3.39 3.66 0.93
ZL19A45 Z10019 1900 45 2.87 3.05 0.94
ZL25A-14 Z10019 2500 -14 2.62 2.55 1.03
ZL25A0 Z10019 2500 0 2.86 2.98 0.96
ZL25A14 Z10019 2500 14 3.17 3.28 0.97
ZL25A37 Z10019 2500 37 2.62 2.85 0.92
ZL25A45 Z10019 2500 45 2.48 2.59 0.96
Average 0.97
COV (%) 4.00

Fig. 3. FE model and experimental load versus lateral deflection at mid-span for Z beams (a) ZL25A14 (b) ZL25A37.

slenderness values to fully capture the biaxial bending behaviour of CFS


storage rack uprights in global buckling. Slender, semi-compact and
compact unperforated upright cross-sections were considered.

3.1. Investigated configurations and upright lengths

Nine cross-sectional shapes were investigated using the same up­


rights as in [1]. Nine different biaxial bending configurations per upright
type were tested. In reference to the local y-axis of symmetry and minor
x-axis, these included pure bending about the local y-axis (Conf 0), My =
2.5Mx (Conf 1), My = Mx (Conf 2), Mx = 2.5My (Conf 3), bending about
the minor x-axis when the lip stiffeners are in compression (Conf 4), My
= − 2.5Mx (Conf 5), My = − Mx (Conf 6), Mx = − 2.5My (Conf 7), and
bending about the minor x-axis when the web are in compression (Conf
8), where My and Mx are the moments applied about the y- and x-axes,
respectively. Mx positive signifies that lip is in compression.
The upright cross-sectional shapes are shown in Fig. 4. Types G and I
were taken from the literature [14]. Types C, D, E, F, M and O cover
cross-sectional shapes encountered in the industry and were considered
herein with two different thicknesses. The main cross-sectional di­
mensions and properties of all upright types are given in Table 2.
The length of the uprights in the numerical model was determined
based on their “signature curves” obtained from elastic buckling ana­
Fig. 4. Upright cross sections: (a) Type C; (b) Type D; (c) Type E; (d) Type F; (e)
lyses performed in ABAQUS [24] on the simply supported and warping
Type G; (f) Type H; (g) Type I (k) Type M; and (m) Type O.
restrained model. Per upright type, the length of the upright was kept
the same for all nine investigated biaxial bending configurations. The
lengths were chosen so that no buckle interaction modes would occur. (McrG) of the selected buckling length, of Type M upright with a thick­
For all sections and configurations, the elastic local buckling moment ness of 1.8 mm in Conf 0 (major axis), Conf 4 and 8 (minor axis). For this
was sufficiently higher than the global buckling moment [31,32]. upright, Conf 8 (bending about the minor x axis when the web is in
Regarding distortional-global interaction mode, the lengths of the up­ compression) governs the length of the upright to be 3500 mm, ensuring
rights were initially chosen so that the global-to-distortional elastic that interactive buckling failure modes would not occur in all investi­
buckling moment ratio (RGD) was <0.85 for all sections and configura­ gated configurations. The lengths of all uprights are given in Table 2.
tions [31]. Fig. 5 shows examples of Mcr (critical elastic buckling
moment) vs. L curves, with associated critical global buckling mode

4
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Table 2 failure modes of 2 mm thick Type D upright with RGD = 0.56, Type E
Nominal cross-sectional dimensions and properties of investigated uprights in upright with RGD = 0.7 and Type G upright with RGD = 0.6, in Conf 1. As
global buckling. shown in Fig. 7, Type G upright fails in pure global buckling with no
Upright Thick. Depth Width Imajor / Upright length evidence of distortional deformations, whereas SDI develops in Type D
type (mm) (mm) (mm) Iminor (mm) and E uprights. It should be noted that running the analyses for longer
Type C 2.0 140 100 2.53 7500 uprights, i.e., Type E upright (L = 12 m, RGD = 0.66 in Conf 1, RGD = 0.56
2.5 140 100 2.53 6800 in Conf 2 and RGD = 0.55 in Conf 3) results in the same SDI failure mode
Type D 1.2 90 72 1.58 5000 for these configurations and the linear biaxial bending design equation
2.0 90 72 1.58 4000
still to be unconservative. Further investigations are needed to evaluate
Type E 1.2 90 72 2.06 5000
2.0 90 72 2.06 4000 influence of cross-sectional shape and dimensions in beams experiencing
Type F 1.5 125 100 1.79 6700 SDI failure mode. It will be shown in Section 6.1 that both the global and
2.0 125 100 1.79 5500 SDI failure modes are safely handled by the DSM global strength curve,
Type G 2.0 100 110 0.94 8000 eliminating the need to define any border between these two failures.
Type H 2.5 100 90 1.41 8000
Type I 2.5 100 80 2.13 8000
Type M 1.8 80 60 2.17 3500 5. Direct strength method equations to predict global biaxial
2.5 80 60 2.17 3200 bending capacity
Type O 2.0 120 60 5.05 5500
2.5 120 60 5.05 5000
The accuracy of available DSM-based approaches in predicting the
biaxial bending capacity of steel storage rack uprights in global buckling
4. Global biaxial bending response of the uprights and is evaluated. These approaches are detailed in the following sections and
interactive behaviour are: (i) using the classical global DSM equations given in the AISI-S100
[3] and AS/NZS 4600 [4], (ii) using global inelastic reserve capacity for
Elastic or inelastic global buckling failure modes were observed for stocky cross sections in the AISI-S100 [3] and AS/NZS 4600 [4], (iii)
all investigated uprights. The normalised ultimate biaxial moment ca­ using a newly proposed extended inelastic reserve strength in global
pacities (Mx/ Mbx and My/ Mby) for all upright types are summarized in buckling similar to the one proposed by Pham and Hancock [17] for
Table 3. Mbx and Mby are taken as the numerical bending moment ca­ local and distortional buckling.
pacities about the x- (minor) and y- (major) axes, respectively. The The classical DSM [Approach (i)] nominal member moment capac­
global slenderness ratio λG in Table 3 is ity, Mbe, for global buckling, ignoring inelastic reserve capacity, is
√̅̅̅̅̅̅̅̅̅̅ defined as [3,4,33].
My
λG = (1) Mbe = My if λG < 0.6 (3)
McrG
( )
where McrG is determined from the FE model running linear buckling 10
Mbe = My 1 −
10My
if 0.6 ≤ λG ≤ 1.336 (4)
analyses (LBA) and My is the yield biaxial moment. The associated 9 36McrG
interactive biaxial moment capacity is obtained from the linear equation
currently provided in the standards [3–5]. Mbe = McrG if λG > 1.336 (5)

Mx My where McrG and My are elastic global buckling moment and yield
+ ≤ 1.0 (2) moment, respectively, and λG is nondimensional slenderness ratio
Mbx Mby
defined by Eq. (1).
Fig. 6 illustrates, in a different manner than Table 3, the linear design The recent AISI-S100 [3] and AS/NZS 4600 [4] considers inelastic
equation versus the normalised biaxial bending results obtained from reserve capacity, with the nominal member moment capacity enabled to
the nine different investigated configurations. Table 3 and Fig. 6 show range between My and the plastic moment Mp for compact cross-
that the governing interaction relationship is not linear. When the lip is sections, i.e., if λG < 0.6. When considering the global inelastic reserve
in compression (Conf 1–3), the linear equation is unconservative for capacity [Approach (ii)], Mbe becomes.
Type C (Conf 1 and 2), Type D (Conf 1 and 2), Type E (Conf 1–3), Type F ⎛√̅̅̅̅̅̅̅ ⎞
(Conf 1 and 2), Type H (Conf 1–3), Type I (Conf 1 and 2), Type M (Conf My
( )⎜ − 0.23⎟
1) and Type O (Conf 1) uprights with interaction ratios ranging from Mbe = Mp − Mp − My ⎜
McrG
⎟ ≤ Mp if λG < 0.6 (6)
⎝ ⎠
0.68 (Type M – Conf 1) to 0.97 (Type D – Conf 1). For configurations 1 to 0.37
3 and for all uprights, on average, the linear equation results in
unconservative estimation of the biaxial bending capacity by 3% with a The global elastic buckling moments McrG for each tested configu­
coefficient of variation (COV) of 19%. ration to be inputted in the DSM expressions were obtained by running
When the web is in compression (Conf 5–7), the linear equation is LBA. For each of the tested configurations, the yield moment (My) and
conservative for all upright types and gives ratios ranging from 1.21 plastic moment (Mp) were calculated about the biaxial bending axis with
(Type G – Conf 5) to 1.45 (Type I – Conf 6). Considering all uprights and the yield stress of 450 MPa used in the parametric studies.
configurations, the linear equation conservatively estimates the biaxial Similar to the extended inelastic reserve capacity for local and
bending capacity by 32% on average with a COV of 5%. distortional buckling proposed by Pham and Hancock [17], a new
Secondary distortional-bifurcation distortional-global interaction extended range of the cross-sectional slenderness for which the inelastic
(SDI) may occur for sections with RGD < 1.0 subject to high yield to non- strength can be applied [Approach (iii)] is proposed herein for global
critical distortional buckling moment ratio (Ry) [31,32]. The uprights buckling. For global buckling, the inelastic reserve capacity can be
and configurations at which the linear equation is found to be uncon­ extended and defined as [3,4,17]:
servative, exhibit 0.45 < RGD < 0.84 and 1.03 < Ry < 1.86. The ( / )( )
assessment of the failure modes for these uprights and configurations MnyG = My + 1 − 1 CyGn 2
Mp − My (7)
indicated that distortional deformations tended to develop near failure,
suggesting the development of the secondary distortional-bifurcation D- where
G interaction (SDI) affecting ultimate strength of the uprights [31]. √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Fig. 7 compares the global and distortional elastic buckling modes and CyGn = 1 ≤ 1.336/λG ≤ 3 (8)

5
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Fig. 5. Mcr vs. L curves and critical global buckling mode (McrG) of the selected length for 1.8 mm thick Type M upright a) Conf 0 (Mx = 0) b) Conf 4 (My = 0 and lip
in compression) c) Conf 8 (My = 0 and web in compression).

MnyG is then used in the classical DSM [Eqs. (3–5)] instead of My, and classical DSM [Approach (i)] and the new proposed extended range
λGn is defined as: [Approach (iii)], for all tested configurations. As shown in Fig. 8 and
√̅̅̅̅̅̅̅̅̅̅ Table 3, the classical DSM (without the inelastic reserve capacity)
MnyG typically results in conservative estimation of the bending capacity of
λGn = (9)
McrG the studied uprights, with the FEA to DSM capacity ratios ranging be­
tween 0.89 (for 1.2 mm thick Type E – Conf 2) and 3.07 (for Type I –
is used instead of λG to obtain the new nominal member capacity with
Conf 0). On average, the DSM without the inelastic reserve capacity
extended range, MbGn.
conservatively estimates the bending capacity by 55%, with a coefficient
of variation (COV) of 26% for all tested uprights and configurations.
5.1. Comparison of direct strength method design approaches with Out of total 135 tested configurations in biaxial bending of nine
parametric results global buckling sections, 45 configurations have slender cross-sections
with λG > 1.336, 73 configurations have cross-sections with 0.6 ≤ λG
Table 3 provides the FEA biaxial failure moment MFEA to the DSM ≤ 1.336, and 17 configurations have compact cross-sections with λG <
predicted moment MDSM ratio for the three different approaches and 0.6. When λG > 1.336, the classical DSM global strength curve un­
global buckling. Fig. 8 graphically compares the classical DSM global derestimates the capacity by about 64%, with a COV of 34%. In other
buckling curve to the normalised FEA predicted capacities for the

6
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Table 3
Comparison of parametric studies results with linear equation and DSM for global buckling.
DSM DSM

Up- Conf. λG My/ Mx/ Eq. (2) MFEA/ MFEA/ MFEA/ Up- Conf. λG My/ Mx/ Eq. (2) MFEA/ MFEA/ MFEA/
right Mby Mbx Linear MDSM MDSM MDSM right Mby Mbx Linear MDSM MDSM MDSM
(no (with (extended (no (with (extended
a b c a b c
reserve) reserve) range) reserve) reserve) range)

Type 0 1.54 1.00 0.00 1.00 1.68 1.68 1.68 Type 0 1.23 1.00 0.00 1.00 1.40 1.40 1.39
C-2 1 1.65 0.45 0.31 0.76 1.12 1.12 1.12 G-2 1 1.27 0.85 0.66 1.51 1.62 1.62 1.61
mm 2 1.38 0.33 0.55 0.88 0.97 0.97 0.97 mm 2 1.39 0.56 1.10 1.66 1.73 1.73 1.73
3 1.15 0.19 0.81 1.00 1.20 1.20 1.15 3 1.43 0.23 1.10 1.33 1.50 1.50 1.50
4 1.15 0.00 1.00 1.00 1.40 1.40 1.34 4 1.52 0.00 1.00 1.00 1.32 1.32 1.32
5 1.25 0.97 0.37 1.34 1.42 1.42 1.41 5 0.99 0.94 0.28 1.22 1.37 1.37 1.27
6 0.92 0.73 0.71 1.44 1.36 1.36 1.23 6 0.88 0.75 0.55 1.30 1.36 1.36 1.21
7 0.63 0.38 0.93 1.31 1.73 1.73 1.24 7 0.82 0.44 0.81 1.25 1.57 1.57 1.32
8 0.56 0.00 1.00 1.00 1.80 1.63 1.22 8 0.84 0.00 1.00 1.00 1.69 1.69 1.46
Type 0 1.37 1.00 0.00 1.00 1.53 1.53 1.53 Type 0 1.62 1.00 0.00 1.00 2.34 2.34 2.34
C- 1 1.35 0.48 0.31 0.79 0.92 0.92 0.92 H-2.5 1 1.61 0.35 0.37 0.72 1.16 1.16 1.16
2.5 2 1.24 0.34 0.54 0.88 0.95 0.95 0.94 mm 2 1.64 0.20 0.55 0.75 1.12 1.12 1.12
mm 3 1.03 0.22 0.86 1.08 1.41 1.41 1.28 3 1.70 0.12 0.83 0.95 1.47 1.47 1.47
4 1.01 0.00 1.00 1.00 1.58 1.58 1.41 4 1.90 0.00 1.00 1.00 1.71 1.71 1.71
5 1.12 0.90 0.38 1.28 1.30 1.30 1.26 5 1.20 0.93 0.36 1.29 1.79 1.79 1.77
6 0.84 0.68 0.72 1.38 1.34 1.34 1.17 6 0.93 0.71 0.69 1.40 1.71 1.71 1.54
7 0.57 0.36 0.93 1.29 1.79 1.65 1.22 7 0.80 0.38 0.93 1.31 1.71 1.71 1.46
8 0.50 0.00 1.00 1.00 1.90 1.52 1.22 8 0.84 0.00 1.00 1.00 1.45 1.45 1.34
Type 0 1.45 1.00 0.00 1.00 1.72 1.72 1.72 Type 0 1.92 1.00 0.00 1.00 3.07 3.07 3.07
D- 1 1.41 0.43 0.39 0.82 1.07 1.07 1.07 I-2.5 1 1.88 0.49 0.31 0.80 1.92 1.92 1.92
1.2 2 1.45 0.26 0.57 0.83 1.04 1.04 1.04 mm 2 1.80 0.38 0.59 0.97 2.24 2.24 2.24
mm 3 1.46 0.15 0.85 1.00 1.31 1.31 1.31 3 1.59 0.22 0.83 1.05 2.59 2.59 2.59
4 1.63 0.00 1.00 1.00 1.51 1.51 1.51 4 1.60 0.00 1.00 1.00 2.94 2.94 2.94
5 1.02 0.90 0.33 1.23 1.45 1.45 1.34 5 1.53 0.93 0.39 1.32 2.39 2.39 2.39
6 0.78 0.71 0.65 1.36 1.55 1.55 1.26 6 1.15 0.71 0.74 1.45 1.83 1.83 1.78
7 0.64 0.40 0.90 1.30 1.71 1.71 1.27 7 0.83 0.36 0.94 1.30 1.89 1.89 1.58
8 0.67 0.00 1.00 1.00 1.49 1.49 1.24 8 0.76 0.00 1.00 1.00 1.81 1.81 1.48
Type 0 1.13 1.00 0.00 1.00 1.42 1.42 1.40 Type 0 1.39 1.00 0.00 1.00 1.59 1.59 1.59
D-2 1 1.13 0.56 0.43 0.99 1.13 1.13 1.09 M-1.8 1 1.35 0.42 0.26 0.68 0.92 0.92 0.92
mm 2 1.12 0.33 0.65 0.98 1.12 1.12 1.07 mm 2 1.29 0.43 0.66 1.09 1.50 1.50 1.49
3 1.12 0.18 0.86 1.04 1.27 1.27 1.21 3 1.24 0.23 0.89 1.12 1.74 1.74 1.72
4 1.24 0.00 1.00 1.00 1.33 1.33 1.32 4 1.35 0.00 1.00 1.00 1.83 1.83 1.83
5 0.83 0.91 0.39 1.30 1.48 1.48 1.28 5 1.02 0.90 0.39 1.28 1.36 1.36 1.27
6 0.62 0.68 0.72 1.40 1.65 1.65 1.27 6 0.73 0.69 0.74 1.43 1.51 1.51 1.22
7 0.51 0.35 0.94 1.29 1.81 1.50 1.20 7 0.54 0.35 0.96 1.31 1.72 1.53 1.20
8 0.53 0.00 1.00 1.00 1.53 1.39 1.16 8 0.54 0.00 1.00 1.00 1.45 1.35 1.14
Type 0 1.53 1.00 0.00 1.00 1.87 1.87 1.87 Type 0 1.21 1.00 0.00 1.00 1.42 1.42 1.41
E- 1 1.56 0.38 0.40 0.78 0.95 0.95 0.95 M-2.5 1 1.18 0.48 0.29 0.77 0.95 0.95 0.93
1.2 2 1.43 0.24 0.62 0.86 0.89 0.89 0.89 mm 2 1.12 0.43 0.61 1.04 1.35 1.35 1.30
mm 3 1.28 0.13 0.84 0.97 1.02 1.02 1.02 3 1.06 0.23 0.85 1.08 1.66 1.66 1.54
4 1.31 0.00 1.00 1.00 1.17 1.17 1.17 4 1.15 0.00 1.00 1.00 1.73 1.73 1.69
5 1.22 0.92 0.37 1.29 1.43 1.43 1.42 5 1.24 0.88 0.42 1.30 1.86 1.86 1.84
6 0.88 0.71 0.70 1.41 1.50 1.50 1.31 6 0.65 0.67 0.80 1.47 1.59 1.59 1.23
7 0.63 0.38 0.93 1.31 1.82 1.82 1.29 7 0.49 0.32 0.96 1.28 1.74 1.40 1.15
8 0.59 0.00 1.00 1.00 1.80 1.74 1.27 8 0.48 0.00 1.00 1.00 1.49 1.28 1.12
Type 0 1.19 1.00 0.00 1.00 1.47 1.47 1.46 Type 0 2.10 1.00 0.00 1.00 2.35 2.35 2.35
E-2 1 1.22 0.48 0.36 0.84 0.91 0.91 0.90 O-2 1 1.96 0.61 0.31 0.92 1.84 1.84 1.84
mm 2 1.12 0.31 0.59 0.90 0.97 0.97 0.93 mm 2 1.76 0.49 0.62 1.11 2.11 2.11 2.11
3 1.00 0.17 0.80 0.97 1.23 1.23 1.09 3 1.41 0.27 0.85 1.12 2.29 2.29 2.29
4 1.01 0.00 1.00 1.00 1.43 1.43 1.29 4 1.39 0.00 1.00 1.00 2.50 2.50 2.50
5 0.97 0.84 0.39 1.23 1.27 1.27 1.19 5 1.56 0.91 0.38 1.29 1.73 1.73 1.73
6 0.69 0.63 0.74 1.37 1.47 1.47 1.18 6 1.07 0.68 0.72 1.40 1.26 1.26 1.18
7 0.50 0.33 0.96 1.29 1.92 1.53 1.21 7 0.62 0.36 0.93 1.29 1.53 1.53 1.15
8 0.47 0.00 1.00 1.00 1.84 1.42 1.18 8 0.49 0.00 1.00 1.00 1.57 1.35 1.15
Type 0 1.42 1.00 0.00 1.00 1.53 1.53 1.53 Type 0 1.86 1.00 0.00 1.00 1.99 1.99 1.99
F- 1 1.42 0.44 0.39 0.83 0.99 0.99 0.99 O-2.5 1 1.79 0.62 0.29 0.91 1.57 1.57 1.57
1.5 2 1.35 0.27 0.60 0.87 0.94 0.94 0.94 mm 2 1.54 0.51 0.58 1.09 1.80 1.80 1.80
mm 3 1.34 0.16 0.86 1.02 1.17 1.17 1.17 3 1.23 0.29 0.84 1.11 2.05 2.05 2.02
4 1.46 0.00 1.00 1.00 1.27 1.27 1.27 4 1.21 0.00 1.00 1.00 2.33 2.33 2.29
5 1.07 0.87 0.33 1.20 1.33 1.33 1.26 5 1.44 0.91 0.41 1.32 1.48 1.48 1.48
6 0.75 0.68 0.65 1.33 1.44 1.44 1.16 6 0.95 0.68 0.76 1.44 1.23 1.23 1.10
7 0.60 0.38 0.90 1.28 1.65 1.65 1.18 7 0.55 0.36 0.98 1.34 1.62 1.47 1.14
8 0.64 0.00 1.00 1.00 1.47 1.47 1.19 8 0.44 0.00 1.00 1.00 1.58 1.25 1.12
Type 0 1.19 1.00 0.00 1.00 1.33 1.33 1.31 Average (all uprights) 1.10 1.55 1.52 1.41
F-2 1 1.17 0.51 0.39 0.90 0.95 0.95 0.93 COV (%) 18 26 26 29
mm 2 1.13 0.34 0.66 1.00 1.05 1.05 1.01
3 1.11 0.18 0.89 1.07 1.23 1.23 1.16
4 1.22 0.00 1.00 1.00 1.23 1.23 1.21
5 0.88 0.89 0.36 1.25 1.31 1.31 1.16
6 0.64 0.68 0.69 1.37 1.53 1.53 1.16
(continued on next page)

7
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Table 3 (continued )
DSM DSM

Up- Conf. λG My/ Mx/ Eq. (2) MFEA/ MFEA/ MFEA/ Up- Conf. λG My/ Mx/ Eq. (2) MFEA/ MFEA/ MFEA/
right Mby Mbx Linear MDSM MDSM MDSM right Mby Mbx Linear MDSM MDSM MDSM
(no (with (extended (no (with (extended
a b c a b c
reserve) reserve) range) reserve) reserve) range)

7 0.51 0.37 0.93 1.30 1.80 1.48 1.18


8 0.53 0.00 1.00 1.00 1.52 1.37 1.14
a
No inelastic reserve capacity – Approach (i).
b
Inelastic reserve capacity – Approach (ii).
c
Newly proposed extended reserve strength in global buckling – Approach (iii).

Fig. 6. Biaxial bending interaction points for global buckling — all uprights.

words, taking elastic buckling moment as failure moment for slender 6. Reliability analysis
sections is quite conservative. When 0.6 ≤ λG ≤ 1.336, the classical DSM
gives a conservative prediction of the bending capacity of the investi­ Based on the parametric studies and the corresponding predicted
gated uprights, with an average FEA to DSM biaxial moment capacity of capacities, the reliability or safety index, β0, is calculated for [Approach
1.45, with a COV of 18%. For stocky sections with λG < 0.6, the classical (i)] and [Approach (iii)] in accordance with the method outlined in
DSM underestimates, on average, the FEA capacity by 70%, with a COV Chapter K of AISI-S100 [3]. β0 is a measure of reliability of a structure or
of 9%. structural member. Larger β0 indicates higher reliability and lower risk
For bending about the major axis (Conf 0), ten out of 15 configura­ of failure. The strength of the tested members should satisfy Eq. K2.1.1-
tions have slender cross-sections (λG > 1.336) and fail in elastic buckling 1a for LRFD from AISI-S100 [3] defined as
at a moment less than My. The other eight configurations have slen­ ∑
derness ratio of 0.6 ≤ λG ≤ 1.336, and also fail in elastic buckling except γi Qi ≤ ϕRn (10)
one configuration (2 mm thick Type D). The average MFEA/MDSM for ∑
bending about the major axis is 1.78. In bending about the minor axis where γ iQiis the required strength (factored loads) based on the most
(Conf 4 and 8), the average MFEA/MDSM is 1.66. For Conf 1–3, DSM is critical load combination determined in accordance with Section B2 for
more accurate in predicting the moment capacity with the average LRFD. Rn is the mean value of all test results. The resistance factor ϕ can
MFEA/MDSM of 1.35. For Conf 5–7, the DSM predictions leads to the be determined by
average MFEA/MDSM of 1.58. √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
̅
(11)
− β V 2m +V 2F +CP V 2P +V 2Q
The use of the DSM with inelastic reserve capacity [Approach (ii)], as ϕ = Cϕ (Mm Fm Pm )e 0
in the AISI-S100 [3] and AS/NZS 4600 [4], results in an overall 3%
where Cϕis the calibration factor defined as 1.52 for LRFD method; Mm is
improvement, on average, of the predictions when compared with the
the average material factor equal to 1.1; Fm is the average fabrication
classical DSM.
factor equal to 1.0; Pm is the average of the test (FEA) to predicted (DSM)
Regarding the DSM predictions using the new extended range of the
ratio; β0 is the target reliability index which is considered as 2.5 for cold-
inelastic reserve capacity for global buckling, Table 3 and Fig. 8 show
formed members; Vm is the coefficient of variation for the material factor
that the proposed method [Approach (iii)] provides better strength
and is taken as 0.1; Vf is the coefficient of variation for the fabrication
predictions (primarily for stocky sections) when compared with the
factor and is taken as 0.05; CP is the correction factor based on the
previous two DSM approaches. On average, for all configurations and
number of the tests; VP is the coefficient of variation of the test to pre­
upright types, this method overestimates the FEA capacity by 41%, with
dicted ratio; and VQ is the coefficient of variation for the load effect
a COV of 29%.

8
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Fig. 7. Global buckling (McrG), distortional buckling (McrD), and failure deformed shapes (scale factor = 3) in Conf 1 a) Type D upright (thk. = 2 mm) b) Type E
upright (thk. = 2 mm) c) Type G upright.

taken as 0.21 for the LRFD method [3]. biaxial bending interaction relationship was found to be nonlinear, and
Table 4 presents the results of the reliability analyses performed the linear biaxial bending design equation to be conservative with fail­
using Eq. (11) on [Approach (i)] and [Approach (iii)]. The average of the ure occurring at ratios given by the design equation ranging from 1.00 to
test to predicted ratio (Pm) and the corresponding coefficient of varia­ 1.47. When the flange is in compression, the linear equation is uncon­
tions (VP) are also given in Table 4. The resulting safety indexes β0 are servative for some investigated uprights, due to observed secondary-
shown for typical resistance factor for beams (ϕ=0.9). Furthermore, the distortional bifurcation D-G interaction. The accuracy of the currently
resulting ϕ values are given for the target reliability index of 2.5. For all codified DSM global strength curve was also assessed to estimate the
three investigated cases in Table 4, the corresponding safety index biaxial bending capacity of CFS storage rack uprights failing in global
β0 and resistance factor ϕ for [Approach (iii)] are lower than those of buckling with or without considering the inelastic reserve capacity as
[Approach (i)] but higher than the target safety index of 2.5. offered in design specifications. It was found that the classical DSM
As shown in Table 4, when 0.6 ≤ λG ≤ 1.336, [Approach (iii)] gives underestimates the biaxial bending capacity for all tested configura­
14% improvements of the predictions when compared with [Approach tions. On average, the capacity to DSM prediction ratio was equal to
(i)]. For this case, the safety index β0 and resistance factor ϕ are 2.90 and 1.55. When the inelastic reserve capacity is considered, this ratio
1.02, respectively. changed to 1.52. When using the newly proposed extended range in the
when λG < 0.6 [Approach (iii)] results in an overall 52% improve­ paper for global inelastic reserve strength, the DSM equations better
ment, on average, of the predictions when compared with the classical predicted the biaxial capacity, with a capacity to prediction equal to
DSM. The corresponding safety index β0 and resistance factor ϕ of this 1.41. Future studies would need to be performed on perforated uprights.
case are 3.26 and 1.08, respectively, which are well above the target
safety index of 2.5. CRediT authorship contribution statement

7. Conclusion Nima Talebian: Conceptualization, Investigation, Methodology,


Software, Validation, Writing – original draft. Benoit P. Gilbert:
A numerical study on global biaxial bending of cold-formed steel Conceptualization, Investigation, Methodology, Software, Validation,
storage rack uprights was reported in this paper. Nine different unper­ Writing – review & editing. Dane Miller: Conceptualization, Investi­
forated upright sections were analysed to assess the accuracy of the gation, Methodology, Software, Validation, Writing – review & editing.
linear biaxial bending design equation in the international design Hassan Karampour: Conceptualization, Investigation, Methodology,
specifications for global buckling. When the web is in compression, the Software, Validation, Writing – review & editing.

9
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Fig. 8. Comparison of the DSM curve to parametric studies data for global buckling [Approach (i) and (iii)] a) all configurations b) Conf 0 (major axis) c) Conf 1 d)
Conf 2 e) Conf 3 f) Conf 4 (lip in compression) g) Conf 5 h) Conf 6 i) Conf 7 and j) Conf 8 (web in compression).

10
N. Talebian et al. Journal of Constructional Steel Research 196 (2022) 107395

Table 4
Reliability analyses for Approach (i) and Approach (iii).
Case Approach (i) Approach (iii)

Pm VP β0 (ϕ=0.9) ϕ (β0 = 2.5) Pm VP β0 (ϕ=0.9) ϕ (β0 = 2.5)

All specimens 1.55 0.26 2.98 1.07 1.41 0.29 2.55 0.92
λG < 0.6 1.70 0.09 4.48 1.50 1.18 0.04 3.26 1.08
0.6 ≤ λG ≤ 1.336 1.45 0.18 3.30 1.14 1.31 0.19 2.90 1.02

Declaration of Competing Interest [14] C. Bernuzzi, M. Simoncelli, European design approaches for isolated cold-formed
thin-walled beam–columns with mono-symmetric cross-section, Eng. Struct. 86
(2015) 225–241.
The authors declare that they have no known competing financial [15] X. Zhao, C. Ren, R. Qin, An experimental investigation into perforated and non-
interests or personal relationships that could have appeared to influence perforated steel storage rack uprights, Thin-Walled Struct. 112 (2017) 159–172.
the work reported in this paper. [16] N. Baldassino, C. Bernuzzi, A. di Gioia, M. Simoncelli, An experimental
investigation on solid and perforated steel storage racks uprights, J. Constr. Steel
Res. 155 (2019) 409–425.
Data availability [17] C.H. Pham, G.J. Hancock, Experimental investigation and direct strength design of
high-strength, complex C-sections in pure bending, J. Struct. Eng. 139 (ASCE) (11)
(2013) 1842–1852.
Data will be made available on request. [18] B.W. Schafer, Z. Lia, C.D. Moen, Computational modelling of cold-formed steel,
Thin-Walled Struct. 48 (2010) 752–762.
References [19] C. Yua, B.W. Schafer, Simulation of cold-formed steel beams in local and
distortional buckling with applications to the direct strength method, J. Constr.
Steel Res. 63 (5) (2007) 581–590.
[1] N. Talebian, B.P. Gilbert, C.H. Pham, R. Chariere, H. Karampour, Parametric
[20] C.H. Pham, A.F. Davis, B.R. Emmett, Numerical investigation of cold-formed
studies and design rules for local and distortional biaxial-bending capacity of cold-
lapped Z purlins under combined bending and shear, J. Constr. Steel Res. 95 (2014)
formed steel storage-rack uprights, J. Struct. Eng. ASCE 146 (3) (2020) 04020009.
116–125.
[2] N. Talebian, B.P. Gilbert, C.H. Pham, R. Chariere, H. Karampour, Local and
[21] M. Chen, B. Young, Experimental and numerical investigation on cold-formed steel
distortional biaxial bending capacities of cold-formed steel storage rack uprights,
semi-oval hollow section compression members, J. Constr. Steel Res. 151 (2018)
J. Struct. Eng. ASCE 144 (6) (2018) 04018062.
174–184.
[3] AISI(American Iron and Steel Institute), North American Specification for the
[22] J. Ye, S.M. Mojtabaei, I. Hajirasouliha, Local-flexural interactive buckling of
Design of Cold-Formed Steel Structural Members. AISI-S100, AISI, Washington, DC,
standard and optimised cold-formed steel columns, J. Constr. Steel Res. 144 (2018)
2016.
106–118.
[4] AS/NZS (Standards Australia), Cold-Formed Steel Structures. AS/NZS 4600, AS/
[23] M. Peiris, M. Mahendran, Behaviour of cold-formed steel lipped channel sections
NZS, Sydney, Australia, 2018.
subject to eccentric axial compression, J. Constr. Steel Res. 184 (2021), 106808.
[5] CEN (European Committee for Standardization), Eurocode 3. Design of steel
[24] ABAQUS, Abaqus ver. 6.14 user manual., ABAQUS, Providence, RI, 2015.
structures, in: General rules. Supplementary Rules for Cold-Formed Members and
[25] N. Talebian, B.P. Gilbert, N. Baldassino, H. Karampour, Factors contributing to the
Sheeting. EN 1993-1-3, CEN, Brussels, Belgium, 2006.
transverse shear stiffness of bolted cold-formed steel storage rack upright frames
[6] M.B. Put, Y.-L. Pi, N.S. Trahair, Lateral buckling tests on cold-formed Z-beams,
with channel bracing members, Thin-Walled Struct. 136 (2019) 50–63.
J. Struct. Eng. ASCE 125 (1999) 1277–1283.
[26] J.S. Rajkannu, S.A. Jayachandran, Flexural-torsional buckling strength of thin-
[7] S. Torabian, D.C. Fratamico, B.W. Schafer, Experimental response of cold-formed
walled channel sections with warping restraint, J. Constr. Steel Res. 169 (2020),
steel zee-section beam-columns, Thin-Walled Struct. 98 (2016) 496–517.
106041.
[8] S. Torabian, B. Zheng, B.W. Schafer, Development of a new beam-column design
[27] K.W. Karren, Corner properties of cold-formed steel shapes, J. Struct. Eng. ASCE 93
method for cold-formed steel lipped channel members, in: Proc., 22nd Int.
(1) (1967) 401–432.
Specialty Conf. on Cold-Formed Steel Design and Construction, Missouri Univ. of
[28] AS (Standards Australia), Steel Storage Racking. AS4084, Sydney, Australia: AS,
Science and Technology, Toronto, 2014, pp. 359–376.
2012.
[9] S. Torabian, B. Zheng, B.W. Schafer, Experimental response of cold-formed steel
[29] M.B. Put, Y.-L. Pi, N.S. Trahair, Biaxial bending of cold-formed Z-beams, J. Struct.
lipped channel beam-columns, Thin-Walled Struct. 89 (2015) 152–168.
Eng. 125 (11, 1999) 1284–1290.
[10] L. Bertoccia, D. Comparinia, G. Lavacchinib, M. Orlandoc, L. Salvatoric,
[30] V.M. Guimarães, B.P. Gilbert, N. Talebian, B. Wang, Shape optimisation of singly-
P. Spinellic, Experimental, numerical, and regulatory P-Mx-My domains for cold-
symmetric cold-formed steel purlins, Thin-Walled Struct. 161 (2021), 107402.
formed perforated steel uprights of pallet-racks, Thin-Walled Struct. 119 (2017)
[31] A.D. Martins, D. Camotim, P.B. Dinis, Distortional-global interaction in lipped
151–165.
channel and zed-section beams: strength, relevance and DSM design, Thin-Walled
[11] S. Torabian, B.W. Schafer, Development and experimental validation of the direct
Struct. 129 (2018) 289–308.
strength method for cold-formed steel beam-columns, J. Struct. Eng. ASCE 144 (10,
[32] A.D. Martins, D. Camotim, P.B. Dinis, On the distortional-global interaction in cold-
2018), 04018175.
formed steel columns: relevance, post-buckling behaviour, strength and DSM
[12] J.V. Kumar, S.A. Jayachandran, Experimental investigation and evaluation of
design, J. Constr. Steel Res. 145 (2018) 449–470.
direct strength method on beam-column behavior of uprights, Thin-Walled Struct.
[33] B.W. Schafer, Review: the direct strength method of cold-formed steel member
102 (2016) 165–179.
design, J. Constr. Steel Res. 64 (7) (2008) 766–778.
[13] C. Bernuzzi, F. Maxenti, European alternatives to design perforated thin-walled
cold-formed beam–columns for steel storage systems, J. Constr. Steel Res. 110
(2015) 121–136.

11

You might also like