Thin-Walled Structures 191 (2023) 111038

Contents lists available at ScienceDirect

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

Full length article

Ultimate behaviour of aluminium alloy I-beams: New experimental tests and

European codification
Rosario Montuori a , Elide Nastri b , Vincenzo Piluso b , Alessandro Pisapia b ,∗
a
University of Salerno, Department of Pharmacy, Italy
b
University of Salerno, Department of Civil Engineering, Italy

ARTICLE INFO ABSTRACT

Keywords: In this work, the ultimate behaviour of I-shaped aluminium beams and a comparison with the current design
Local buckling provisions reported in the Eurocodes are presented. A total number of 15 three-point bending tests are carried
Aluminium alloys out on extruded beams made of 6060-T66, 6082-T6 and 6005A-T6 aluminium alloys at the Strength laboratory
Interactive buckling
of the University of Salerno. The tested cross-sections are characterized by different width-to-thickness ratios
Experimental test
and, three different lengths are considered: 1300 mm, 800 mm and 500 mm. Material properties, maximum
I-sections
Effective thickness method
flexural resistance, rotation capacity and failure mode were fully reported within this manuscript. Moreover,
Bending moment a total number of 86 experimental tests are collected from the scientific literature and, a comparison between
the experimental results and those obtained by current design provision EN 1999 − 1 − 1 is provided. In the
case of bending moment, the experimental results are used to evaluate the accuracy of the new methodologies
currently proposed in the Annex L of the new Eurocode 9 draft, i.e. the extension of effective thickness method.
The EN 1999 proposal showed a higher accuracy in estimating the I-shaped flexural resistances.

1. Introduction good ductility of the members made of aluminium alloys of 6xxx series
allows to employ this material in the seismic framework to realize
In the last decades, the application of aluminium alloys in the struc- seismic links, dissipative joints, shear walls and fuses [4–7].
tural engineering is increasing due to its favourable properties as low As for stainless and high strength steels, the aluminium alloys are
mass weight, high strength-to-weight ratio, good plasticity, ease of fab- characterized by a strain hardening behaviour which can be described
rication, good corrosion resistance and high degree of workability [1]. by means of Ramberg–Osgood constitutive model [8], as reported in
The 25% of the global aluminium production is currently employed in the current Annex E of Eurocode 9 [9].
the structural field [2]. The extrusion process allows to realize complex Nowadays, the ultimate resistance of cross-section aluminium mem-
tailored cross-section shapes, suitable for structures where more com- bers is evaluated according to the design rules reported in Section 6 of
monly used materials, as carbon steel or concrete cannot be applied. EN 1999 − 1 − 1 [9], namely ‘‘Ultimate Limit State for Members’’.
The corrosion resistance allows to adopt aluminium in the marine The aluminium alloy structural codes are strongly influenced by
environments without surface protection and with low maintenance the standard provisions used for steel [10], especially concerning the
costs. The low weight makes the aluminium applied for long span roof evaluation of the maximum resistance and deformation capacity of
systems, movable bridges, and prefabricated frameworks. members. In fact, these provisions are based on an elastic-perfectly
Aluminium is not just one material, but it gives rise to a family plastic analysis performed on an effective cross-section. However, it is
of different groups of alloys whose mechanical properties widely vary recognized that this current approach is too conservative for aluminium
from one group to another and within each group itself. From the point alloys preventing the best exploitation of their member capacities.
of view of the technological use, these alloys, according to the American The main reason is that this approach neglects the real behaviour of
Association classification, can be grouped into nine series. Typically, aluminium alloys, which stress–strain relationship is characterized by
the aluminium alloys adopted in the structural applications belong a continuous strain-hardening curve [11]. Moreover, it is important to
to 6xxx and 7xxx series. The chemical composition of the first one observe that a wide variety of aluminium alloys is present, and this
is Aluminium–Silicon–Magnesium, the second series is characterized makes the determination of general rules valid for all alloys even more
by the presence of the Zinc [1,3]. Comparing the stress–strain curves complex.
from different aluminium alloys, it is evident that 7xxx series shows Consequently, many research works have been devoted to filling
higher yield stress, but lower ductility compared to 6xxx series. The the knowledge gaps on aluminium structures. With reference to the

∗ Corresponding author.
E-mail addresses: [email protected] (R. Montuori), [email protected] (E. Nastri), [email protected] (V. Piluso), [email protected] (A. Pisapia).

https://doi.org/10.1016/j.tws.2023.111038
Received 21 March 2023; Received in revised form 29 June 2023; Accepted 17 July 2023
Available online 1 August 2023
0263-8231/© 2023 Elsevier Ltd. All rights reserved.
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Nomenclature 𝝌 𝑳𝑻 represents the reduction factor for lateral–

torsional buckling
𝑩 flange width
𝝀𝐋𝐓 Relative slenderness related to lateral–
𝑯 Height of section
torsional buckling
𝒕𝒘 Web thickness
𝜱𝐋𝐓 Perry–Robertson relation
𝒕𝒇 𝒊 Bottom flange thickness
𝜶 𝐋𝐓 Imperfection factor
𝒕𝒇 𝒔 Upper flange thickness
𝝀0,𝐋𝐓 Limit of horizontal plateau
𝒃𝒇 Width of the outstand part element
𝑴 𝒄𝒓 Elastic critical bending moment
𝒉𝒘 Width of internal part of web plate
𝜿 𝒘𝒕 Non-dimensional torsional parameter
𝑳𝒕𝒐𝒕 Total length of the tensile coupon
𝜻𝒈 Relative non-dimensional coordinate of the
𝑳𝒄 Parallel length of the tensile coupon load application
𝑳𝑺 Gripping length of the tensile coupon 𝜻𝒋 Relative non-dimensional cross-section
𝑺 Gripping part width of the tensile coupon mono-symmetry parameter
𝑺𝒐 Reduced width of the tensile coupon 𝑰𝒕 Torsion constant
𝒆̇ 𝑳𝒄 Strain rate 𝑰𝒛 Second moment of area about of minor axis
𝒗𝒄 Speed test 𝑰𝒘 Warping constant
𝑬 Young’s modulus 𝒌𝒚 ,𝒌𝒛 , 𝒌𝒘 Coefficients related to restrain conditions at
𝒇 0.1 Stress at a residual strain of 0.1% the end of the beam
𝒇 0.2 Stress at a residual strain of 0.2% 𝑪𝒊 Coefficients related to the loading and end
𝒇𝒖 Ultimate stress restraint conditions
𝝈 Engineering stress 𝑷 𝐖𝐂 Maximum resistance due to transverse loads
𝜺 Engineering strain according to EN1999-1-1
𝜺𝒖 Strain corresponding to 𝑓𝑢 𝑳𝒆𝒇 𝒇 Effective length according to EN1999-1-1
𝜺𝒓 Ultimate strain 𝝌𝑭 Reduction factor for web crippling
𝜺0.𝒖 Strain level equal to ‘‘𝜀𝑢 − 0.002’’ 𝝀𝑭 Relative slenderness for web crippling
𝒏𝒊 Ramberg–Osgood in the inelastic range 𝑭 𝒄𝒓 Elastic critical load due to web crippling
𝒏𝒑 Ramberg–Osgood in the plastic range 𝒔𝒔 Length of stiff bearing
𝑳𝒊 Tested length of beam 𝒌𝑭 Coefficient related to load application
𝑰 Inertia modulus of beam section 𝒍𝒚 Effective loaded length
𝑴 0.2 Bending moment corresponding to 𝑓0.2 𝒎1 , 𝒎2 Non-dimensional parameters related to web
𝜽0.2 Chord rotation corresponding to 𝑀0.2 crippling
𝝌 0.2 Curvature corresponding to 𝑀0.2 𝜷∕𝝐 0 Slenderness parameter of plate according to
𝜹∗ Theoretic limit displacement EN1999-1-1
𝑴𝒖 Maximum bending moment 𝒌 Buckling factor accounting for the edge
𝑭𝒖 Maximum load at the midspan restraining conditions
𝜹𝒖 Maximum displacement at the midspan 𝜻 Correction factor accounting for interactive
buckling
𝜽𝒖 Chord rotation corresponding to 𝑀𝑢
𝑳𝒔 Shear length
𝝌𝒖 Maximum curvature at the midspan
𝒕𝒆𝒇 𝒇 Effective thickness of plate
𝑬 𝒔 ,𝑬 𝒕 Secant and tangent moduli
𝝍 Parameter accounts for the stress distribu-
𝝂 Poisson’s coefficient
tion
𝝃 Plastic coefficient related to non-linearity of
aluminium material
𝝆𝒄 Reduction factor due to local buckling
𝜺𝒊 ,𝜺𝒔 Strains obtained by the strain gauges local buckling and post-buckling behaviour of aluminium sections un-
𝒅𝜺 Distance between the strain gauges der compression, the previous research works carried out by Faella
𝑴 𝐁𝐌 Bending moment at the midspan according et al. [12], Mazzolani et al. [13,14], Su et al. [15], Piluso Pisapia [16]
to EN1999-1-1 and Piluso et al. [17] have improved the understanding of the structural
𝑾 𝒆𝒍 Elastic modulus of section material, covering various cross-shapes: channels, angles, box sections
𝑾𝒑 Plastic modulus of section with or without internal cross stiffeners.
𝑾 𝒆𝒇 𝒇 Effective section modulus In the case of the flexural behaviour, the first experimental studies
have been carried out by Dumont and Hill [18] which investigated the
𝑨𝒗 Shear area of beam section
lateral stability of equal flanged I-beams subjected to pure bending.
𝑽 𝐏𝐒 Shear action according to EN1999-1-1
In 1947, Panlio [19] investigated the behaviour of two-span statically
𝑽 𝑬𝒅 Shear action indetermined beams. Subsequently, Welo [20] performed tests under
𝑴 𝐁𝐒 Bending moment reduced due to shear uniform moment for determining the moment–curvature behaviour.
action according to EN1999-1-1 While Moen et al. [21,22], De Matteis et al. [23,24] and Manganiello
𝑴 𝐋𝐓 Lateral–torsional buckling resistance ac- et al. [25] carried out several experimental and numerical tests into
cording to EN1999-1-1 the strength and rotation capacity of aluminium beams subjected to
non-uniform bending. From the analytical point of view, Eberwein
Valtinant [26] provided a mathematical method to define the moment–
curvature curve of symmetrical aluminium cross-sections, while Schafer
Pek oz
̈ [27] developed the direct strength method (DSM), initially for

2
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

design of cold-formed steel members. Subsequently, it was extended to Finally, the experimental results, including those collected in the
aluminium alloy sections and verified against the beam tests performed state of the art, are compared with those derived by the design rules
by Zhu Young [28]. of European standard checking the failure mode of each tested beam.
As the strain hardening, and ductility of aluminium material have Furthermore, the experimental and numerical results were employed
a strong influence on the ultimate flexural behaviour of aluminium to assess the accuracy of the new effective thickness approach [41]
beams, Kim Pek oz ̈ [29] carried out experimental tests and developed provided in Annex L of new proposed version of Eurocode 9 [42].
numerical models to obtain the inelastic flexural bending capacities
of aluminium alloy stocky sections, where it was found that the ul- 2. State of art review
timate material resistance could be achieved. Another mathematical
approach, including the post-yield material properties of aluminium In this section, a review of the experimental tests performed on the
alloys, is the continuous strength method (CSM) [30,31]. It represents a aluminium alloy I-beams is provided. In particular, the main research
deformation-based design approach, initially developed for non-linear works about this topic and presented in the scientific literature since
metallic structural members. This method estimates a limiting strain 1940, have been analysed and the main results of three-point and pure
for the section which is used in combination with a strain hard- bending tests have been collected.
ening material model to define load–displacements capacities. The Dumont and Hill [18] and Wang et al. [49] investigated the lateral–
previous approach is currently reported in Chinese, American and torsional buckling resistance of I-shaped beams subjected to the pure
Australian/New Zealand specifications [32–34] as an alternative to the bending. The test setup has been the same for both experimental
traditional design procedures. campaign and a schematic representation is shown in Fig. 1. Two
To predict the inelastic response of box-shaped aluminium beams, identical beams were placed parallel and bolted to supports to avoid
taking into account the main slenderness parameters of cross-section lateral–torsional rotation from end supports. The simply supported
and the strain hardening behaviour, Castaldo et al. [35,36] and Nastri boundary conditions were provided by one roller plate. Hence, in-
Piluso [37] provided the empirical relationships obtained by means plane rotation of end supports could bring pure bending moment to
of finite element simulations. Starting from an extensive parametric specimens. Consequently, the in-plane rotation is free while the out-
analysis, the mathematical formulas have been derived as function plane rotation and the rotation around the longitudinal direction are
of the flange slenderness, the flange-to-web slenderness ratios, the prevented. Moreover, the length of the beam laterally bolted to the end
non-dimensional shear length, and the Ramberg–Osgood coefficient.
support is equal to 330 mm for each specimen. A single compressive
Subsequently, the same mathematical regressions have been derived
load by the hydraulic jack was transferred to two end supports by a
for I-shaped aluminium beams [38–40]. Moreover, Nastri et al. [41]
rigid distributive girder with jack located at the midspan.
proposed an extension of Effective thickness method (ETM), currently
Zhang and Wu [50] and Guo et al. [51] performed the three-
adopted by Eurocode 9 [9], to estimate more accurately the ultimate
point bending tests to study the lateral torsional resistance of I-beams
flexural resistance of aluminium beams under uniform compression or
subjected to non-uniform bending. In these experiments, all the beam
in bending. It is important to underline that the mathematical formulas,
specimens were considered as simple beams. Consequently, one of the
previously described, and the new version of ETM are currently pro-
supports is movable while the other is fixed, see Fig. 2. Two rigid
vided within the Annex L belonging to the revised version of Eurocode
clamps, whose distance can be modified by three bolts, are used to fix
9 [42]. Recently, Li et al. [43,44] investigated the ultimate behaviour
the test beams (Fig. 3a). The end supports were intended to provide
of 7A04-T6 high-strength aluminium alloy SHS and RHS beams under
simple supports that prevent from in and out-of-plane deflections and
three and four bending tests to evaluate the applicability of the current
twist rotations, but do not restrain in- and out-of-plane rotations and
design provisions in Chinese, European and American codes.
warping displacements since the clamping length is so small that the
Huynh et al. [45] presents an experimental and numerical investi-
warping restraint can be neglected. In most cases, the vertical load
gation on the behaviour and strength of cold-rolled aluminium beams
has been applied at the top centre of the test beam (Fig. 3b). Only in
subjected to distortional buckling, while Pham et al. [46] performed a
series of four-point bending tests in order to study the global buckling five tests carried out by Guo et al. [51], a concentrated force has been
of aluminium alloy channel beams. Other researchers focused on the applied at the centre of cross-section (Fig. 3c).
ultimate behaviour of lipped channels beams (LCB). In particular, M. As previously mentioned, at the beginning of 2000s, Moen et al. [21]
Rouholamin et al. [47,48] carried out the experimental tests on the studied the ultimate behaviour of aluminium beams under non-uniform
aluminium LCB beams in shear with and without web openings. bending in terms of maximum flexural resistance and rotational capa-
Although in the last three decades the studies regarding the ultimate bility. The test specimens were supported on two steel cylinders that
response of aluminium beams in bending have significantly increased, were placed in steel bearings resting on top of the support beam, as
the gap of knowledge respect to steel members is still very high, reported in Fig. 4. Moreover, in order to avoid the lateral–torsional
especially in the experimental field. In fact, as reported in Section 2, buckling, the angle cross-sections with stiffening plates were located on
few studies on I-shaped aluminium beams under different loading both sides of the I-beams at the midspan. The loading device consisted
condition have been performed. It gives rise to a great restriction on of a compact steel block with a total width of 150 mm with the corners
the application and development of aluminium alloys in the structural rounded by a radius equal to10 mm.
engineering. Regarding the resistance of I-beams subjected to concentrated load-
Consequently, in this work, the attention is focused on the ultimate ing, there are few experimental tests in the scientific literature. 4
behaviour of I-shaped beams under different loading conditions. Firstly, experiments have been carried out by Tryland et al. [52] according
an extensive state of art on the experimental research is performed to to the geometric configuration reported in Fig. 5. In particular, 6
collect the experimental results in terms of the ultimate resistances. experimental tests were performed according to the scheme reported
Three main collapse modes have been considered: (1) Bending moment in Fig. 5a, where the compressive load was applied at the center of
at the midspan; (2) Lateral-tortional buckling; (3) Web crippling due to beam specimens and the distance between the support and the end of
concentrated load. the beam (𝐶) is fixed to 60 mm. The configuration shown in Fig. 5b was
Subsequently, an experimental campaign, performed at the Uni- used for three specimens with 𝐶 = 20 mm, while 𝐶 = 5 mm was adopted
versity of Salerno, is presented. In particular, 15 three-point bending for the latest three tests. Furthermore, the beams were restrained in the
tests have been carried out on five different cross-sections made of lateral direction using two of the machined details shown in Fig. 5c.
alloys 6060-T66, 6082-T6 and 6005/A-T6. The tested sections are Subsequently, Wang et al. [53] performed 6 experiments with different
characterized by different width-to-thickness ratios ranging from 8 to width-to-thickness ratios and by varying the tested length. The loading
24. The failure mode has been determined for each tested I-beam. conditions was the same for all tests and it is shown in Fig. 6a. It is

3
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 1. Geometric scheme of pure bending test (a); Details of the supports (b) [49].

important to underline that the specimens were clamped supports in

order to avoid lateral instability at the beam ends as depicted in Fig. 6b.
A summary of the experimental tests on aluminium alloy I-beams,
presented in the scientific literature, is shown in Table 1. Authors, date
of publication, the number of tests, and tested alloys are indicated in
the first four columns. The fifth column provides the loading conditions:
Pure bending (P.B.), Non uniform bending moment (N.B.) and Concen-
trated loading (C.L). Finally, the last column indicates the investigated
failure mode: Lateral–torsional buckling (LT), Bending moment (BM)
and Web crippling due to transverse load (WC).
All experimental data used in this work are collected in Annex.
In particular, the mechanical properties of aluminium alloys, declared
by the authors, are depicted in Table A.1. It is interesting to observe
that the value of the Young’s modulus is almost always around the
value of 70 GPa. According to Fig. A.1, the geometric properties and
the maximum experimental loads (𝐹𝑢.𝑒𝑥𝑝 ) of tested I-beams, provided
by the authors, are reported in Table A.2. It is worth mentioning
that the aluminium cross-sections employed by Wang et al. [49] are
characterized by different flange plates (𝐵𝑠 ≠ 𝐵𝑖 ). Moreover, regarding Fig. 2. Geometric scheme of non-uniform bending test [50].
the tests subjected to pure bending, the value of 𝐿 indicates the laterally
unbraced length. So, the total length (𝐿𝑡𝑜𝑡 ) is given by the sum of 𝐿
and twice the supported length of beam (330 mm). The cross-section

4
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table 1
Summary of the experimental tests reported in the scientific literature.
Author(s) Date Number of tests Aluminium alloy Loading Failure mode
Dumont & Hill [18] 1940 10 S27-T P.B. LT
Moen et al. [21] 1999 8 6082-T6 N.B. BM
Tryland et al. [52] 1999 4 6082-T6 C.L. WC
Zhang & Wu [50] 2006 10 6061-T6 N.B. LT
Wang et al. [49] 2012 20 6061-T6 P.B. LT
6063-T5
Guo et al. [51] 2014 8 6061-T6 N.B. LT
Wang et al. [53] 2016 6 6061-T6 C.L. WC
6063-T5

Fig. 3. Details of experimental equipments reported in [51].

6060-T66, 6082-T6 and 6005A-T6. These alloys are corrosion resistant

and particularly suitable for extrusion, but also rolled sections as well
as tubes can be produced. They are used either in welded structures
and in bolted or riveted connections. By means heat-treatment the yield
strength of these alloys is increased up to 250 MPa, with a quite good
ductility, being the ultimate strain up to 12% [1]. These alloys are
composed by Aluminium–Silicon–Magnesium.
According to EN 573-3 [54], their chemical compositions are re-
ported in Table 2. It is possible to observe that 6082 alloy contains
more iron (Fe) and Manganese (Mn) than the other two alloys and, con-
sequently, the nominal yield stress is higher. In the case of aluminium
temper, the T designation is used for any cast alloy that has been given
a solution heat treatment followed by a suitable quench and either
natural (i.e., in air) or artificial (i.e., in a furnace) aging. Temper T66
represents a variation of temper T6. It provides mechanical property
Fig. 4. Experimental setup adopted in [21]. level higher than T6 achieved through special control of the process.
Both heat treatments are suitable for structural applications because of
its high resistance [3]. According to EN 1999 − 1 − 1 [9], the nominal
classification is provided according to EN 1999 − 1 − 1 [9]. Moreover, mechanical properties of each aluminium alloy are reported in Table 2,
With reference to the I-beams subjected by pure bending [18,49], the where 𝑡 represents the nominal thickness, 𝑓𝑢 represents the ultimate
values of 𝐹𝑢.𝑒𝑥𝑝 are referred to the parallel pairs of members. So, the tensile strength, 𝑓0.2 is the characteristic value of 0.20% proof strength,
maximum resistance of a single beam could be obtained as half of the 𝐵𝐶 represents the buckling curve and 𝑛 represents the strain hardening
declared values. coefficient according the Ramberg–Osgood’s law [8] (see Table 3).

3. Experimental campaign 3.2. Geometric properties

To improve the knowledge of ultimate resistance of aluminium alloy The experimental tests have been performed on 5 section types:
I-beams, 15 three-point tests have been performed at the University of H90, H100, H120, H140, H160 and each section has been tested for
Salerno. In particular, 5 cross-sections with different width-to-thickness three different lengths (𝐿1 = 1300 mm; 𝐿2 = 800 mm; 𝐿3 = 500 mm). All
ratios have been tested with reference to three lengths: 𝐿1 = 1300 mm, specimens have been labelled by cross-section dimension and the test
𝐿2 = 800 mm and 𝐿3 = 500 mm. Geometric and mechanical properties length. As an example, by considering the label ‘‘H90 - L1’’, it represents
of the aluminium alloys and the experimental results are presented in the specimen with nominal height equal to 90 mm and test length of
the following section. 1300 mm. With reference to the geometric scheme shown in Fig. 7, 𝐵
and 𝐻 are, respectively, the flange width and the total height of section;
3.1. Chemical composition 𝑡𝑤 is the thickness of web plate, while 𝑡𝑓𝑖 and 𝑡𝑓𝑠 are, respectively, the
thicknesses of the lower and upper flange. Moreover, 𝑏𝑓 represents the
Three point bending tests have been carried out on 15 I-shaped width of the outstand part element and ℎ𝑤 is the width of internal
aluminium beams. They are extruded profiles of aluminium alloys part of web plate. The measured geometric values of each specimen

5
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 5. Geometric configuration of test setup adopted in [52].

Fig. 6. Details of experimental equipments reported in [53].

Table 2
Nominal chemical composition of aluminium alloys according to EN 573-3 [54].
Alloy Si Fe Cu Mn Mg Cr Zn Ti Others[%] Al
[%] [%] [%] [%] [%] [%] [%] [%] [%]
6060 Min 0.30 0.10 – – 0.35 – – – Each Total rest
Max 0.60 0.30 0.10 0.10 0.60 0.05 0.15 0.10 0.05 0.15
6082 Min 0.70 – – 0.40 0.60 – – – Each Total rest
Max 1.30 0.50 0.10 1.00 1.20 0.25 0.20 0.10 0.05 0.15
6005A Min 0.50 – – – 0.40 – – – Each Total rest
Max 0.90 0.35 0.30 0.50 0.70 0.30 0.20 0.10 0.05 0.15

are listed in Table 4. The geometric cross-sections have been chosen in while the height-to-thickness ratios for the web 𝑏𝑤 ∕𝑡𝑤 were ranged
order to evaluate the ultimate behaviour of beams with different slen- from 7.88 to 24.01. Moreover, according to EN 1999 − 1 − 1 [9], the
derness ratios of the plates composing the sections and compatibly with cross-section classification is depicted in Table 4. In particular, the
commercially available profiles. The beam lengths have been selected specimens H90 and H160 belong to Class 4, consequently, they should
to evaluate different failure collapse modes under three-point tests. In be affected by elastic local buckling, and they should not have any
fact, as will be highlighted above, the main collapse modes, which flexural overstrength. The secH100 and H120 are Class 3, so they
have occurred during the experimental tests, are: Bending moment should locally buckle before the development of the plastic sources, so
at the midspan, lateral–torsional buckling and web crippling due to their plastic deformation capacity is not achieved. Finally, the sections
transverse load. It is possible to observe that the width-to-thickness H140 are Class 2, consequently, they could develop their whole plastic
ratios of the compressive flange 𝑏𝑓 ∕𝑡𝑓 were ranged from 8.17 to 16.36, resistance with a limited plastic deformation capacity.

6
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table 3 width of the parallel length of a flat test piece; 𝑆 is the width of the
Nominal material properties according to EN 1999 − 1 − 1 [9].
gripped ends; 𝐿𝑐 is the parallel length; 𝐿𝑡𝑜𝑡 is the total length of test
Alloy Temper 𝑡[mm] 𝑓0.2 [MPa] 𝑓𝑢 [MPa] BC n piece; 𝐿𝑆 is the length of the gripped ends. The monotonic tests have
6060 T66 𝑡≤3 160 215 B 16 been performed under displacement control according to Method A2
3 < 𝑡 ≤ 15 150 195 B 18 as reported in UNI EN ISO 6892 − 1 − 1 [55]. For predicting the elastic
6082 T6 𝑡≤5 250 290 A 32 and inelastic properties of aluminium alloys, four intervals of speed
5 < 𝑡 ≤ 15 260 310 A 25
(𝑒̇ 𝐿𝑐 ) have been considered as provided in UNI EN ISO 6892 − 1 −
𝑡≤5 225 270 A 25 1 [55] and reported in Fig. 9. A specific strain rate 𝑒̇ 𝐿𝑐 is defined as a
6005A T6
5 < 𝑡 ≤ 10 215 260 A 24
function of the parallel length 𝐿𝑐 for each interval and, consequently,
10 < 𝑡 ≤ 25 200 250 A 20
the corresponding speed test (𝑣𝑐 ) is equal to:

𝑣𝑐 = 𝐿𝑐 ⋅ 𝑒̇ 𝐿𝑐 (1)

The nominal values of coupons and the displacement speeds are

reported in Table 6, while Fig. 10 shows the setup of tensile tests.
Through the tensile tests, the following mechanical properties have
been defined: the elastic modulus (𝐸𝑒𝑥𝑝 ), the stress at a residual strain of
0.1% (𝑓0.1 ), the stress at a residual strain of 0.2% (𝑓0.2 ), the engineering
maximum stress (𝑓𝑢 ), the engineering strain corresponding to maximum
stress (𝜀𝑢 ), the ultimate strain (𝜀𝑟 ). The coefficients 𝑛𝑖 and 𝑛𝑝 represents
the values of the Ramberg–Osgood coefficient in the inelastic and
plastic region, respectively. So, with reference to the Ramberg–Osgood
law [8]:
( )𝑛
𝜎 𝜎
𝜀= + 0.002 (2)
𝐸 𝑓0.2
According to Annex E of EN 1999 − 1 − 1 [9], 𝑛𝑖 and 𝑛𝑝 are equal to:
𝑙𝑛2 𝑙𝑛(0.002∕𝜀0.𝑢 )
𝑛𝑖 = ( ) 𝑛𝑝 = ( ) (3)
𝑓0.2 𝑓
𝑙𝑛 𝑓0.1
𝑙𝑛 𝑓0.2
𝑢

where 𝜀0.𝑢 is the residual strain corresponding to the maximum stress

Fig. 7. Geometric scheme of I-shaped section. 𝑓𝑢 , and it is equal to 𝜀0.𝑢 = 𝜀𝑢 − 0.002. According to ANNEX E of EN
1999 − 1 − 1 [9] and with reference to Fig. 11, the coefficient 𝑛 is
equal to 𝑛𝑖 , if the analysis concerns the range of elastic deformations,
i.e., between 𝑓0.1 and 𝑓0.2 . While 𝑛 is equal to 𝑛𝑝 in the range between
3.3. Mechanical properties
𝑓0.2 and 𝑓u , i.e., in the plastic region.
These properties are listed in Table 7 for all plates of each H-
The main material properties of each profile have been obtained by section. In particular, 𝑃𝑓𝑖 and 𝑃𝑓𝑠 denote, respectively, the coupon of
means of standard tensile tests according to UNI EN ISO 6892−1−1 [55]. bottom and upper flange, while 𝑃𝑤 indicate the coupon of web plate.
In particular, the tensile tests have been carried out through a Schenck Fig. 12(a) depicts the experimental stress–strain curves. It is possible
Hydropuls S56 testing machine (maximum load 630 kN, piston stroke to observe that the values of strain corresponding to the maximum
± 125 mm), with one strain gauge and one extensometer recording the stress are ranged from 5.25 to 8.77%, while the ultimate strains (𝜀𝑟 )
strain response during testing. The adopted strain gauge is FLAB-3-23 are around 7%–13%. A comparison between the experimental stress
produced by Tokyo Instruments Lab and its main characteristics are values with those reported in EN 1999 − 1 − 1 [9] is reported in Table 8.
reported in the following table (see Table 5): In the case of the yield stress, the experimental values are greater than
According to Fig. 8, The tensile coupons are extracted along the corresponding nominal values with a percentage difference between 2%
longitudinal direction of all plate elements constituting each cross- and 22%. In the case of the ultimate stresses, the difference is lower
section. The geometric dimensions of the tensile coupons are defined than the yield stresses and it is ranged in 0.5÷13%, in fact, the mean
according to UNI EN ISO 6892 − 1 − 1 [55] where 𝑆0 is the original of the ratios 𝑓𝑢.𝑒𝑥𝑝 ∕𝑓𝑢.𝑛𝑜𝑚 is about 1.03 with a coefficient of variation

Table 4
Measured geometric dimensions of specimens.
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐴𝑙𝑙𝑜𝑦 𝐻 [mm] 𝐵 [mm] 𝑡𝑤 [mm] 𝑡𝑓𝑖 [mm] 𝑡𝑓𝑠 [mm] ℎ𝑤 ∕𝑡𝑤 [−] 𝑏𝑓 ∕𝑡𝑓 [−] 𝐶𝑙𝑎𝑠𝑠 𝑆𝑒𝑐𝑡𝑖𝑜𝑛
H90 - L1 6060 – T66 89.94 69.94 4.95 4.05 4.04 16.54 16.07 4
H90 - L2 6060 – T66 90.06 69.94 4.95 3.99 3.97 16.59 16.33 4
H90 - L3 6060 – T66 89.94 69.95 5.00 3.99 3.95 16.40 16.36 4
H100 - L1 6082 – T6 100.20 100.23 10.15 10.08 10.1 7.88 8.93 3
H100 - L2 6082 – T6 100.24 100.22 10.12 10.07 10.09 7.91 8.94 3
H100 - L3 6082 – T6 100.26 100.21 10.10 10.07 10.09 7.93 8.94 3
H120 - L1 6060 – T66 120.06 120.16 8.20 8.04 8.05 12.68 13.92 4
H120 - L2 6060 – T66 120.14 120.16 8.17 8.07 8.04 12.73 13.90 4
H120 - L3 6060 – T66 120.14 120.16 8.20 8.04 8.04 12.69 13.93 4
H140 - L1 6005A – T6 140.08 89.85 7.94 9.98 9.97 15.13 8.21 2
H140 - L2 6005A – T6 140.08 89.75 8.09 9.96 10.02 14.85 8.17 2
H140 - L3 6005A – T6 139.93 89.75 8.00 9.96 9.92 15.01 8.22 2
H160 - L1 6082 – T6 160.00 120.26 6.15 8.15 8.15 23.37 14.00 4
H160 - L2 6082 – T6 160.00 120.22 6.00 8.08 8.06 23.98 14.15 4
H160 - L3 6082 – T6 160.00 120.22 5.99 8.08 8.06 24.02 14.15 4

7
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 8. Tensile coupon test according to UNI EN ISO 6892 − 1 − 1 [55].

Fig. 9. Strain rate for each interval provided by UNI EN ISO 6892 − 1 − 1 [55].

Fig. 10. Setup of tensile tests.

equal to 0.05. The tensile coupons after test are reported in Fig. 12 (b). 3.4. Three-point test setup
The necking phenomenon involves all the tested coupons, consequently
the aluminium alloys are characterized by a good ductility. Finally, The 15 three-point bending tests have been performed on the five
Fig. 13 shows a comparison of stress–strain curves defined according to I-shaped aluminium beams. To evaluate the influence of shear length
Ramberg–Osgood model, as reported in Eq. (2) for each tested section. on the ultimate resistance of aluminium members, three different test
The main parameters, adopted to define the continuous models, have lengths have been considered: 𝐿1 = 1300 mm, 𝐿2 = 800 mm and
been defined as the mean of the experimental results provided in 𝐿3 = 500 mm. The test setup is shown in Fig. 14, where a simply
Table 7 and their values are depicted in Table 9. supported beam is vertically loaded at the midspan by means of the

8
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 11. Stress–strain curve of a typical aluminium alloy.

Table 5 characterized by a longitudinal stress gradient due to shear action.

FLAB-3-23 strain gauge properties.
Indeed, the presented experimental work was performed within the
Type FLAB-3-23
framework of Eurocodes’ revision. Also Annex L of prEN 1999 − 1 −
Test condition 23 ◦ C-50% RH 1 is aimed to the evaluation of the plastic rotation capacity under
Gauge Length 3 mm
non-uniform bending.
Gauge Width 1.70 mm
Backing Length 8.80 mm The three-point bending tests have been carried out under displace-
Backing Width 3.50 mm ment control and the test speeds have been defined as a function of
Gauge Resistance 120 ± 0.3 Ω a theoretical limit displacement (𝛿 ∗ ) computed for each specimen. In
Gauge Factor 2.09 ± 1%
particular, 𝛿 ∗ represents the displacement corresponding to reaching
the conventional elastic limit:
𝑀0.2 𝐿2𝑖
hydraulic actuator imposing displacements on the loading device. The 𝛿∗ = (4)
12𝐸𝐼
loading device consists of a compact steel half cylinder with diameter
where 𝑀0.2 is the theoretical bending moment corresponding to the
equal to 50 mm. The beam is placed on top of half cylindrical supports
conventional yield stress 𝑓0.2 , 𝐿𝑖 represents the test length, i.e., the
which are fixed on the upper flange of a stiff supporting steel beam HEA
distance between the supports of the tested beams (Fig. 14), 𝐼 is
220. In this way, the test setup should be stiffener than the specimens.
the inertia modulus computed according to the geometric properties
Moreover, during the tests, for checking, possible displacements, or
reported in Table 4. Instead, the mechanical properties are computed
rotations out of the plane have been monitored through two linear
according to Table 7.
differential transducers located at the supports, see Fig. 15. Other two
For each specimen, two displacement speeds have been imposed:
displacement transducers have been located at the midspan to mea-
an initial speed (𝑣𝑖 ) up to a limit displacement and a final speed (𝑣𝑓 )
sure the displacements corresponding to the center of the specimens
until the beam collapse. The initial speed 𝑣𝑖 is defined by imposing that
(Fig. 15). For evaluating the values of curvature at the mid span,
two strain gauges have been fixed to the web plates and two others each beam test to achieve the inelastic range in a minimum of 4 min,
have been located to the inside of the flange plates, see detail A of while the final speed (𝑣𝑓 ) is defined about three times the initial speed
Fig. 15. the adopted strain gauges have the same properties as those (𝑣𝑖 ). The values of loading protocol for each specimen are depicted in
used for the tensile tests (Table 5). Concerning the experimental setup, Table 10.
the restrain conditions at the supports are consistent with those adopted
by Moen et al. [21,22]. Conversely, no lateral restraint is provided in 3.5. Experimental results
the midspan section where the load is applied.
Moreover, it is worthwhile the adoption of the three-point bending In this section, the experimental results of three-point tests are
scheme for the experimental setup was preferred because such loading reported. For each tested beam, the values are summarized in Table 11.
condition is typically used when the obtained experimental results are The label 𝐹𝑢.𝑒𝑥𝑝 indicates the maximum load while 𝛿𝑢.𝑒𝑥𝑝 represents
aimed to provide additional information relevant for loading conditions the displacement at the midspan corresponding to 𝐹𝑢.𝑒𝑥𝑝 . According to

Table 6
Geometric properties and displacement speed of tensile coupon tests.
Section 𝑆0 [mm] 𝑆 [mm] 𝐿𝑐 [mm] 𝐿𝑠 [mm] 𝐿𝑡𝑜𝑡 [mm] 𝑣(𝐼)
𝑐 [mm/s] 𝑣(𝐼𝐼)
𝑐
[mm/s] 𝑣(𝐼𝐼𝐼)
𝑐
[mm/s] 𝑣(𝐼𝑉
𝑐
) [mm/s]

H90-H100 15 30 80 90 275 0.0056 0.020 0.16 0.54

H120-H140-H160 20 40 100 100 320 0.0070 0.030 0.20 0.67

9
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 12. Experimental stress–strain curves (left) and the corresponding tensile coupons (right).

the setup scheme described in Fig. 14, the maximum bending moment of the strain gauges. The curvature has been computed as:
𝑀𝑢.𝑒𝑥𝑝 and the corresponding chord rotation 𝜃𝑢.𝑒𝑥𝑝 are defined as: 𝜀1 + 𝜀2
𝜒𝑢.𝑒𝑥𝑝 = (6)
𝐹𝑢.𝑒𝑥𝑝 𝐿𝑖 𝛿𝑢.𝑒𝑥𝑝 𝑑𝜀
𝑀𝑢.𝑒𝑥𝑝 = 𝜃𝑢.𝑒𝑥𝑝 = 2 (5)
4 𝐿𝑖 where 𝜀1 and 𝜀2 are the strain values measured by the strain gauges
where 𝐿𝑖 represents the generic test length. The values of the maximum number 1 and number 2 and 𝑑𝜀 is the distance between the strain
experimental curvature 𝜒𝑢.𝑒𝑥𝑝 have been defined by means of reading gauges. According to Fig. 15, in the case of strain gauges applied

10
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 12. (continued).

Fig. 13. Stress–strain curves according to Ramberg–Osgood model.

Table 7
Mechanical properties of tested specimens.
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐸exp [MPa] 𝑓0.1 [MPa] 𝑓0.2 [MPa] 𝑓𝑢 [MPa] 𝜀𝑢 [%] 𝜀𝑟 [%] 𝑛𝑖 [-] 𝑛𝑝 [-]
𝑃𝑓𝑖 65698.37 157.16 162.84 187.59 5.24 7.91 19.52 22.81
H90
𝑃𝑓𝑠 68005.45 167.99 172.66 202.66 6.29 10.41 25.28 21.32
𝑃𝑤 65434.86 167.64 171.88 200.53 6.39 11.31 27.75 22.26
𝑃𝑓𝑖 76564.13 302.18 308.26 336.05 6.10 8.72 34.80 39.21
H100
𝑃𝑓𝑠 71965.64 314.96 318.33 349.25 7.81 13.11 65.13 39.25
𝑃𝑤 68055.87 302.00 307.17 338.92 7.36 13.08 40.84 36.37
𝑃𝑓𝑖 78896.46 169.59 171.41 199.54 6.03 13.55 64.93 24.37
H120
𝑃𝑓𝑠 65873.24 168.61 171.19 194.06 6.01 12.22 45.64 25.59
𝑃𝑤 68474.20 164.58 168.89 191.32 5.77 11.22 26.81 26.68
𝑃𝑓𝑖 72267.14 253.09 258.28 278.69 5.71 7.50 34.15 40.95
H140
𝑃𝑓𝑠 74986.19 255.38 259.95 279.78 5.66 7.48 39.08 44.98
𝑃𝑤 69290.51 256.74 261.85 283.65 6.14 7.26 35.17 42.41
𝑃𝑓𝑖 69776.81 261.51 267.29 303.49 7.18 12.03 31.71 27.97
H160
𝑃𝑓𝑠 70839.20 262.37 268.87 304.15 8.77 13.83 28.32 30.48
𝑃𝑤 68135.23 258.79 264.29 299.44 7.04 10.55 32.96 28.29

to the flange plates, 𝑑𝜀 coincides with ℎ𝑤 , while in the case of the H160-L3), the strain gauges get unstuck after reaching the maximum
specimen where the strain gauges have been applied to the web plate, load.
the distance 𝑑𝜀 = ℎ𝑤 − 20.
Moreover, the following non-dimensional parameters are defined:
Unfortunately, in two cases (H140-L3; H160-L2), data acquisitions
of all strain gauges were not sufficient to estimate the maximum value 𝑀𝑢.𝑒𝑥𝑝 𝜃𝑢.𝑒𝑥𝑝 𝜒𝑢.𝑒𝑥𝑝
of curvature at the midspan, while in the other two cases (H160-L1; 𝑀𝑢 = 𝜃𝑢 = 𝜒𝑢 = (7)
𝑀0.2 𝜃0.2 𝜒0.2

11
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table 8
Comparison between the nominal and experimental stress values.
𝑓0.2.𝑛𝑜𝑚 𝑓𝑢.𝑛𝑜𝑚
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝑓0.2.𝑛𝑜𝑚 [MPa] 𝑓0.2.𝑒𝑥𝑝 [MPa] 𝑓0.2.𝑒𝑥𝑝
𝑓𝑢.𝑛𝑜𝑚 [MPa] 𝑓𝑢.𝑒𝑥𝑝 [MPa] 𝑓𝑢.𝑒𝑥𝑝

𝑃𝑓𝑖 150 162.84 1.09 195 187.59 0.96

H90
𝑃𝑓𝑠 150 172.66 1.15 195 202.66 1.04
𝑃𝑤 150 171.88 1.15 195 200.53 1.03
𝑃𝑓𝑖 260 308.26 1.19 310 336.05 1.08
H100
𝑃𝑓𝑠 260 318.33 1.22 310 349.25 1.13
𝑃𝑤 260 307.17 1.18 310 338.92 1.09
𝑃𝑓𝑖 150 171.41 1.14 195 199.54 1.02
H120
𝑃𝑓𝑠 150 171.19 1.14 195 194.06 1.00
𝑃𝑤 150 168.89 1.13 195 191.32 0.98
𝑃𝑓𝑖 215 258.28 1.20 260 278.69 1.07
H140
𝑃𝑓𝑠 215 259.95 1.21 260 279.78 1.08
𝑃𝑤 215 261.85 1.22 260 283.65 1.09
𝑃𝑓𝑖 260 267.29 1.03 310 303.49 0.98
H160
𝑃𝑓𝑠 260 268.87 1.03 310 304.15 0.98
𝑃𝑤 260 264.29 1.02 310 299.44 0.97
Mean[𝜇] 1.14 1.03
Dev.St.[𝜎] 0.07 0.05

Table 9 interaction between buckling phenomena has been observed. In partic-

Main parameters of stress-strain curves according to Ramberg–Osgood law.
ular, in some beams characterized by higher values of plate slenderness
𝑆𝑒𝑐𝑡𝑖𝑜𝑛 𝐸 [MPa] 𝑓0.1 [MPa] 𝑓0.2 [MPa] 𝑓𝑢 [MPa] 𝑛𝑖 [-] 𝑛𝑝 [-] ratios, the lateral–torsional buckling occurred simultaneously with local
H90 66379.56 164.26 169.13 196.93 24.18 22.13 instability phenomena regarding the compression elements of cross
H100 72195.21 306.38 311.25 341.41 46.92 38.28
sections.
H120 71081.30 167.59 170.50 194.04 45.80 25.54
H140 72181.28 255.07 260.03 280.70 36.13 42.78 The main failure modes observed before reaching the maximum
H160 69583.75 260.89 266.82 302.36 31.00 28.91 load are: bending moment at the midspan, lateral–torsional buckling
and web crippling due to the concentrated load. More details are
reported in the following Section 4.1.2. Finally, it is interesting to
Table 10
Loading protocol. underline that the cross-sections H90 and H120, classified as fourth
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝛿 ∗ [mm] 𝑣𝑖 [mm/min] 𝑣𝑓 [mm/min] class according to EN 1999-1-1, have reached the post-elastic region
by providing the flexural overstrength.
H90 - L1 7.15 1.79 5.37
H90 - L2 2.71 0.68 2.03
H90 - L3 1.06 0.26 0.79 4. Analysis of experimental results
H100 - L1 5.37 1.34 4.02
H100 - L2 2.03 0.51 1.52 4.1. Eurocode 9 provisions
H100 - L3 0.79 0.20 0.60
H120 - L1 10.46 2.62 7.85 In this section, the main design rules, provided in Section 6 of
H120 - L2 3.96 0.99 2.97
current Eurocode 9 [9], are reported. In particular, five main collapse
H120 - L3 1.55 0.39 1.16
modes have been considered in the prediction of maximum resistance
H140 - L1 6.54 1.63 4.90
of tested beams: (1) Bending moment at the midspan (BM); (2) Pure
H140 - L2 2.48 0.62 1.86
H140 - L3 0.97 0.24 0.73 shear at the supports (PS); (3) Bending and shear at the midspan (BS);
H160 - L1 7.47 1.87 5.60
(4) Lateral–torsional buckling at the midspan (LT) and (5) Failure due
H160 - L2 2.83 0.71 2.12 to concentrated load at the midspan (WC).
H160 - L3 1.11 0.28 0.83
1. In the first case, the maximum flexural resistance at the midspan is
defined according to section 6.2.5 of current Eurocode 9 as:

𝑀BM = 𝛼𝑊𝑒𝑙 𝑓0.2 (9)

where:
𝑀0.2 𝐿 𝑀0.2 where 𝑊𝑒𝑙 is the elastic modulus of section, while 𝛼 is the shape factor
𝑀0.2 = 𝑊𝑒 𝑓0.2 𝜃0.2 = 𝜒0.2 = (8)
6𝐸𝐼 𝐸𝐼 defined as:
Therefore, 𝑀0.2 is the conventional yielding moment, 𝜃0.2 is the ⎧ 𝑊𝑝𝑙 ∕𝑊𝑒𝑙 𝑓 𝑜𝑟 𝑐𝑙𝑎𝑠𝑠 1 𝑎𝑛𝑑 2
corresponding chord rotation and 𝜒0.2 is the corresponding curvature. ⎪
𝛼=⎨ 1 𝑓 𝑜𝑟 𝑡ℎ𝑒 𝑐𝑙𝑎𝑠𝑠 3 (10)
Fig. 16a shows the load–displacement curves of each tested beam. It ⎪
⎩𝑊𝑒𝑓 𝑓 ∕𝑊𝑒𝑙 𝑓 𝑜𝑟 𝑡ℎ𝑒 𝑐𝑙𝑎𝑠𝑠 4
is possible to observe that when the test length decreases or the height
of I-section improves, the maximum load 𝐹𝑢.𝑒𝑥𝑝 increase, conversely, where 𝑊𝑝𝑙 and 𝑊𝑒𝑓 𝑓 represent, respectively, the plastic and effective
the corresponding displacement 𝛿𝑢.𝑒𝑥𝑝 decreases. The final deformed section modulus. In the case of first- and second-class sections, it is
configurations of the specimens after testing (i.e. for displacement necessary to adopt the plastic modulus according to the rigid-plastic
levels largely greater than the one corresponding to the maximum load analysis, while the elastic analysis must be applied for the third-class
bearing capacity) are reported in Fig. 16b where it is possible to observe sections by adopting the elastic modulus 𝑊𝑒𝑙 . The four class sections
local buckling phenomena occurring in the compression elements of are subjected to local buckling effects in the elastic range. Currently,
the Eurocode 9 prescribes the use of a reduced elastic modulus 𝑊𝑒𝑓 𝑓
the cross sections and, in some cases, lateral–torsional post-buckling
deriving by the effective thickness method.
behaviour. By observing the load–displacement curves, it is immedi-
ate to observe that the member stiffness and the ultimate resistance 2. the lateral cross-sections at the supports, where the tested beams
decrease when the test length increases. Moreover, during testing, the have been placed, were subjected to pure shear. This action can be

12
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 14. Setup of Three Point Bending Test.

√
defined according to Section 6.2.6 of current EN 1999 − 1 − 1 as: where 𝜀 = 250∕𝑓0.2 .
𝐴𝑣 𝑓0.2 3. In the three-point bending tests, the middle cross-section is subjected
𝑉PS = √ (11)
3 to both bending moment and shear stresses. Generally, the presence of
shear action reduces the ultimate resistance of beam section. According
where 𝐴𝑣 represents the shear area and, according to the scheme of
to Section 6.2.8, in the case of an equal-flanged I-section classified
Fig. 7, it is equal to:
as first or second class in bending, the resulting value of the reduced
𝐴𝑣 = ℎ𝑤 𝑡𝑤 (12) bending moment is equal to:
(𝑟)
It is important underlining that Eq. (11) can be applied for non- ( ) 𝑡𝑤 ℎ2𝑤 𝑓0.2
𝑀BS = 𝑡𝑓 𝐵 𝐻 − 𝑡𝑓 𝑓0.2 + (14)
slenderness sections. Consequently, before applying Eq. (11), it was 4
verified that: where 𝐵, 𝐻, ℎ𝑤 , 𝑡𝑓 , 𝑡𝑤 represent the geometric properties of cross-
ℎ𝑤 section defined according to Fig. 7, 𝑓0.2 is the conventional yield stress
≤ 39𝜀 (13) of aluminium alloys, while 𝑓0.2 (𝑟)
is the reduced value of yield strength
𝑡𝑤

13
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 15. Instrumentation and data acquisition scheme.

Fig. 16. Experimental curves load–displacement (left) and the corresponding close-up of beams after testing (right).

14
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 16. (continued).

computed as: where 𝜒𝐿𝑇 represents the reduction factor for lateral–torsional buckling
[ ( ) ] and it is equal to:
(𝑟) 2𝑉𝐸𝑑 2
𝑓0.2 = 𝑓0.2 1 − (15) 1
𝑉𝑃 𝑆 𝜒LT = √ (17)
2 − 𝜆2
𝛷LT + 𝛷LT
where 𝑉𝐸𝑑 is the shear action and 𝑉PS is the shear resistance of cross- LT
section computed by means of Eq. (11). In the case of an equal-flanged where:
I-section classified as third class in bending, the resulting value of 𝑀BS √
is given by Eq. (14) but with the denominator 4 in the second term ( ) 2 𝛼𝑊𝑒𝑙 𝑓0.2
𝛷LT = 0.5[1 + 𝛼LT 𝜆LT − 𝜆0,LT + 𝜆LT ] 𝜆LT = (18)
replaced by 6. Finally, for sections classified as belonging from class 4, 𝑀𝑐𝑟
the Eq. (9) must be applied.
Obviously, 𝜆LT is the relative slenderness, while 𝛷LT represents the
4. The lateral–torsional buckling resistance is defined according to
well-known Perry–Robertson expression, function of the imperfection
Section 6.3.2.1 of the EN 1999 − 1 − 1 [9]:
factor 𝛼LT and the limit of horizontal plateau 𝜆0,LT . The values of 𝛼LT
𝑀LT = 𝜒𝐿𝑇 𝛼𝑊𝑒𝑙 𝑓0.2 (16) and 𝜆0,LT are reported in Table 12.

15
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table 11
Experimental results of three-point tests.
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝑀𝑢.𝑒𝑥𝑝 [kNm] 𝑀 𝑢 [-] 𝛿𝑢.𝑒𝑥𝑝 [mm] 𝜃𝑢.𝑒𝑥𝑝 [rad] 𝜃 𝑢 [-] 𝜒𝑢.𝑒𝑥𝑝 [m−1 ] 𝜒 𝑢 [-]
H90 - L1 17.52 5.69 1.19 13.72 0.021 1.71 0.170 3.00
H90 - L2 24.97 4.99 1.06 8.448 0.021 2.78 0.190 3.36
H90 - L3 36.17 4.52 0.96 8.368 0.033 6.99 0.221 3.90
H100 - L1 97.56 31.71 1.12 20.856 0.032 1.72 0.300 3.49
H100 - L2 165.75 33.15 1.17 19.680 0.049 4.27 0.350 4.07
H100 - L3 243.77 30.47 1.08 7.826 0.031 4.33 0.163 1.90
H120 - L1 66.60 21.65 1.11 8.538 0.013 1.50 0.063 1.58
H120 - L2 104.12 22.82 1.17 11.257 0.028 5.26 0.077 1.93
H120 - L3 123.81 15.48 0.80 12.408 0.050 15.03 0.088 2.20
H140 - L1 96.71 31.75 0.98 11.005 0.017 1.53 0.102 1.98
H140 - L2 169.85 35.97 1.10 9.884 0.025 3.65 0.069 1.34
H140 - L3 188.38 23.55 0.73 8.661 0.035 8.16 – –
H160 - L1 128.26 41.68 0.97 10.171 0.016 1.54 0.043 0.90
H160 - L2 153.38 30.68 0.72 5.275 0.013 2.03 – –
H160 - L3 170.50 21.31 0.50 3.935 0.016 4.01 0.027 0.56

Table 12
Values of coefficients 𝛼𝐿𝑇 and 𝜆0,𝐿𝑇 .
𝐶𝑙𝑎𝑠𝑠 𝑠𝑒𝑐𝑡𝑖𝑜𝑛 𝛼LT 𝜆0,LT
1 0.10 0.60
2 0.10 0.60
3 0.20 0.40
4 0.20 0.40

The elastic critical moment for lateral–torsional buckling is indi-

cated as 𝑀𝑐𝑟 and it is defined according to Annex I of EN 1999-1-1.
In the case of a beam of uniform cross-section symmetrical about the
minor axis (Fig. 17), the critical elastic moment is given by:
√
𝜋 𝐸𝐼𝑧 𝐺𝐼𝑡
𝑀𝑐𝑟 = 𝜇𝑐𝑟 𝐿
[√ ] (19)
𝐶1 2 + (𝐶 𝜁 − 𝐶 𝜁 )2 − (𝐶 𝜁 − 𝐶 𝜁 )
𝜇𝑐𝑟 = 𝑘𝑧
1 + 𝜅𝑤𝑡 2 𝑔 3 𝑗 2 𝑔 3 𝑗

where 𝜅𝑤𝑡 is the non-dimensional torsion parameter, 𝜁𝑔 represents the

relative non-dimensional coordinate of the point of load application
related to shear centre and 𝜁𝑗 is the relative non-dimensional cross- Fig. 17. Geometric scheme of I-section symmetrical about the minor axis (z).
section mono-symmetry parameter. Their expressions are given by:

√ √ √
𝜋 𝐸𝐼𝑤 𝜋𝑧𝑔 𝐸𝐼𝑧 𝜋𝑧𝑗 𝐸𝐼𝑧 et al. [57] proposed a method similar to the previous one but applied
𝜅𝑤𝑡 = 𝜁𝑔 = 𝜁𝑗 = (20) to aluminium tubular sections. While Nguyen et al. [58] provided a
𝑘𝑤 𝐺𝐼𝑡 𝑘𝑧 𝐿 𝐺𝐼𝑡 𝑘𝑧 𝐿 𝐺𝐼𝑡
new method for cold-formed steel sections based on the well-known
where 𝐼𝑡 is the torsion constant, 𝐼𝑧 represents the second moment of Direct Strength Method (DSM). However, in this work the attention is
area about of minor axis, 𝐼𝑤 the warping constant and 𝐿 indicates focused on the comparison with the European rules. So, according to
the beam length between points that have the lateral restraint. The
Section 6.7.5 of the EN 1999-1-1, the maximum resistance web plates
coefficient 𝑧𝑔 is the coordinate of the point of load application related
due to transverse loads applied through upper flange, can be computed
to shear centre and it is function of the loading position 𝑧𝑎 , while 𝑧𝑗
as:
is function of the coordinate of the shear centre related to centroid 𝑧𝑠
and it is equal to zero for cross-sections with double symmetry. The 𝑃WC = 𝐿𝑒𝑓 𝑓 𝑡𝑤 𝑓0.2.𝑤 (21)
coefficients 𝑘𝑦 , 𝑘𝑧 and 𝑘𝑤 are referred to the restrain conditions at the
end of the beams. Their values are between 0.5–1.00. In particular, 𝑘𝑦 where 𝑡𝑤 is the web thickness, 𝑓0.2.𝑤 is the characteristic value of
is related to the movement in plane of loading, if 𝑘𝑦 = 1.00, the beam strength of the web plate, while 𝐿𝑒𝑓 𝑓 is the effective length expressed
is free to rotate. The coefficient 𝑘𝑦 represents the restraint to lateral as:
movement, if 𝑘𝑧 = 1.00, the beam is free to rotate on plan. Finally,
𝐿𝑒𝑓 𝑓 = 𝜒𝐹 𝑙𝑦 (22)
𝑘𝑤 restrains against rotation about longitudinal axis, the beam is free
to warp if 𝑘𝑤 = 1.00. Finally, the coefficients 𝐶𝑖 are factors depending where 𝑙𝑦 is the effective loaded length (Section 6.7.5 of EN 1999-1-1)
mainly on the loading and end restraint conditions. and 𝜒𝐹 represents the reduction factor due to local buckling and it is
5. The last possible failure mode is related to the presence of the defined by:
concentrated load at the upper flange of I-beams. Regarding the fail- 0.50
𝜒𝐹 = (23)
ure due to web crippling, in addition to the methods proposed in 𝜆𝐹
the current American and Australian provisions [33,34], alternative where the slenderness 𝜆𝐹 is equal to:
approaches have been, recently, presented in the scientific literature. √
Duarte Silvestre [56] provided a new slenderness-based approach for 𝑙𝑦 𝑡𝑤 𝑓0.2.𝑤
𝜆𝐹 = (24)
the web crippling design applied to plain channel steel beams. Bock 𝐹𝑐𝑟

16
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 18. Load applications and buckling coefficients.

Table 13
Comparison between experimental results with Eurocode 9.
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹 𝑎𝑖𝑙𝑢𝑟𝑒 𝑚𝑜𝑑𝑒 𝐹BM [kN] 𝐹PS [kN] 𝐹BS [kN] 𝐹LT [kN] 𝐹WC [kN]
H90 - L1 BM 13.83a 79.12 13.83 13.83 32.23
H90 - L2 BM 22.17a 79.37 22.17 22.17 31.71
H90 - L3 WC 35.40 80.07 35.40 34.99 31.83a
H100 - L1 BM 87.10 291.91 87.10 76.94a 264.09
H100 - L2 LT 141.47 291.26 141.47 135.97a 263.33
H100 - L3 BM 226.34a 290.76 226.34 226.34 263.00
H120 - L1 LT 58.24 167.84 58.24 55.47a 108.60
H120 - L2 BM 94.79a 167.33 94.79 94.79 108.50
H120 - L3 WC 151.48 167.99 151.48 151.48 108.54a
H140 - L1 LT 139.13 286.39 116.21 109.98a 179.74
H140 - L2 WC 227.49 291.73 188.54 221.73 182.02a
H140 - L3 WC 361.14 288.36 295.55 361.14 179.87a
H160 - L1 LT 124.77 272.28 124.77 108.05a 157.53
H160 - L2 WC 199.82 265.93 199.82 194.66 150.36a
H160 - L3 WC 319.63 265.49 319.63 319.63 151.14a
a
Minimum value calculated among all collapse failure modes.

According to Fig. 7, the elastic critical load 𝐹𝑐𝑟 is obtained as: that: 4 tested beams (H90-L1, H90-L2, H100-L3, H120-L2) have been
governed by the collapse related to simple bending moment (BM);
0.90𝑘𝐹 𝐸𝑡3𝑤
𝐹𝑐𝑟 = (25) the Lateral–torsional buckling (LT) influenced 5 specimens (H100-
ℎ𝑤 L1, H100-L2, H120-L1, H140-L1, H160-L1) and, finally, 6 I-beams
The coefficient 𝑘𝐹 depends on the load application and it is com- (H90-L3, H120-L3, H140-L2, H140-L3, H160-L2, H160-L3) have been
puted as shown in Fig. 18, where 𝑠𝑠 represents the length of stiff governed by the web crippling (WC) due to transverse load (case a).
bearing, and, according to Fig. 14, it is equal to 0. Instead, failures due to the shear resistance at the supports and to
Referred to type (a), the effective loaded length 𝑙𝑦 is obtained as the bending/shear at the midspan are never governing for the tested
follows: specimens.
√ However, it is worthwhile mentioning that the failure modes iden-
𝑙𝑦 = 𝑠𝑠 + 2𝑡𝑓 (1 + 𝑚1 + 𝑚2 ) (26)
tified in Table 13 are based on the numerical values, coming from the
where 𝑚1 and 𝑚2 are the nondimensional parameters defined as: application of code provisions, leading to the smallest capacity. Indeed,
( )2 failure modes observed during the experimental tests, in some cases,
𝑓0.2.𝑓 𝑏𝑓 ℎ𝑤 are not clear and seem to be a combination of different failure modes.
𝑚1 = 𝑚2 = 0.02 𝑖𝑓 𝜆𝐹 > 0.50 otherwise 𝑚2 = 0
𝑓0.2.𝑤 𝑡𝑤 𝑡𝑓 In particular, this occurs when the code provisions corresponding to
different failure modes provide quite close results.
(27)
4.1.2. Comparison with current Eurocode 9
4.1.1. Evaluation of failure mode of experimental tests
All experimental results provided by the scientific literature, includ-
With reference to the three-point tests performed at the University
ing those presented in Section 3.5, have been compared with those
of Salerno, the collapse mechanism has been evaluated as the minimum
obtained by the current design rules reported in Section 4.1.1. So,
value between all possible failure modes previously described. In par-
the experimental results have been collected according to the specific
ticular, according to the test setup presented in Section 3.4, the collapse
collapse mode: Bending moment (BM), Lateral–torsional buckling (LT)
modes are given by:
and Web crippling due to transverse load (WC).
4𝑀BM 4𝑀BS 4𝑀LT With reference to the simple bending moment at the midspan, only
𝐹BM = 𝐹PS = 2𝑉PS 𝐹BS = 𝐹LT = 𝐹WC = 𝑃WC
𝐿 𝐿 𝐿 12 experimental results have been used in this analysis. In particular, 8
(28) experimental tests are provided by Moen et al. [21] and 4 experiments
are presented in this work. So, taking into account the class section, the
The theoretical maximum load 𝐹𝑢.𝑡ℎ can be defined as: theoretic bending moment has been computed by means of Eqs. (9) and
{ } (10). The result of comparison is depicted in Fig. 19a. The mean value
𝐹𝑢.𝑡ℎ = min 𝐹BM , 𝐹PS , 𝐹BS , 𝐹LT , 𝐹WC (29)
of the ratio between the theoretical values (𝐹BM ) and the experimental
The numerical results are depicted in Table 13 where the bold val- values (𝐹𝑢.𝑒𝑥𝑝 ) is equal to 0.82 with the standard deviation of 0.09. The
ues highlight the failure mode of the specimens. It is easy to recognize numerical values are reported in Table A.3 of Annex.

17
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Fig. 19. Comparison between experimental results with those obtained by EC9 according to different failure modes: Bending moment (a), LT Buckling-1 (b), LT-Buckling-2 (c),
LT Buclking-3.1 (d), LT Buckling 3.2 (e), Web crippling due to transverse load (f).

In the case of collapse by lateral–torsional buckling, two different scheme (third scheme of Table 14). The first scheme has been adopted
groups have been identified: (1) LT Buckling due to concentrated force when the total length divided by 4 is lower than the laterally supported
at the midspan; (2) LT Buckling due to pure bending loading. The first length (330 mm). In this case 𝑘𝑤 = 0.50 because, as depicted in Fig. 1,
one includes the experimental results provided by: Zhang Wu [50], the specimens are fixed at the end plates by means of bolted angle
Guo et al. [51] and 5 tests presented in Section 3.5. The values of joints and, consequently, the beam ends are not free to warp. This case
𝑘𝑦 , 𝑘𝑧 , 𝑘𝑤 factors and 𝐶1 , 𝐶2 , 𝐶3 coefficients, adopted in this case, are includes the following specimens: S9, S10, 6061 − 7 − 𝐴 (B), 6061 − 8 − 𝐴
reported in Table 14. As shown in Fig. 19b (LT Buckling 1), the mean (B), 6061 − 9 − 𝐴 (B), 6061 − 10 − 𝐴 (B), 6063 − 8 − 𝐴 (B), 6063 − 9 − 𝐴
value of 𝐹LT ∕𝐹𝑢.𝑒𝑥𝑝 is about 0.93 with the standard deviation equal to (B) and 6063 − 10 − 𝐴 (B). The comparison between the experimental
0.10. The numerical values are reported in Table A.4 of Annex. and theoretical results are reported in Fig. 19c (LT Buckling 2). The
The second group includes the experimental campaigns carried out mean value of the ratio between the theoretical values (𝐹LT ) and the
by Dumont Hill [18] and Wang et al. [49]. As declared by the authors, experimental values (𝐹𝑢.𝑒𝑥𝑝 ) is equal to 0.88 with the standard deviation
𝑘𝑦 is equal to 1.00, while 𝑘𝑧 is assumed equal to 0.50. According to the of 0.14. The numerical values are reported in Table A.5 of Annex.
Annex I of EN 1999 − 1 − 1 [9] and with reference to the setup tests If the total length divided by 4 is greater than the laterally supported
presented in [18,49], two geometric scheme have been considered: (1) length (330 mm), uniform moment scheme has been adopted. In this
Four-point scheme (second scheme of Table 14); (2) Uniform moment case, the analyses have been carried out by assuming the distance

18
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table 14
Values of factors according to the loading and support conditions.
𝐿𝑜𝑎𝑑𝑖𝑛𝑔 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 𝑘𝑦 𝑘𝑧 𝑘𝑤 𝐶1.0 𝐶1.1 𝐶2 𝐶3

1.00 1.00 1.00 1.348 1.363 0.553 0.411

1.00 0.50 0.50 0.922 0.945 0.237 0.772

1.00 0.50 1.00/0.50 1.00 1.127 – 1.019

*𝐶1 = 𝐶1.0 + (𝐶1.1 − 𝐶1.0 )𝜅𝑤𝑡 ≤ 𝐶1.1

between the loading application points as the length of the beams. For being 𝑏 the plate width, 𝑡 the plate thickness and 𝑘 the buckling factor
this reason, it is not easy to recognize that the beams are free to warp or accounting for the edge restraining conditions and the stress distri-
not. So, the analyses have been performed for both 𝑘𝑤 values, i.e., 0.50 bution along the loaded edges. The coefficients 𝐸 and 𝐸𝑠 represent,
(warping fixity) and 1.00 (free to warp). The comparison between the respectively, the elastic and secant modulus, while 𝜈 is the Poisson’s
experimental results with those provided by EN 1999 − 1 − 1 are shown ratio in the elastic–plastic range according to Gerard and Wildhorn
in Fig. 19d (LT Buckling 3.1) and Fig. 19e (LT Buckling 3.2). It is relation [59]:
easy to recognize that the case 𝑘𝑤 = 0.50 provides values closer to the 𝐸𝑠 ( )
experimental ones. In particular, the mean value of the ratio between 𝜈 = 𝜈𝑝 − 𝜈 − 𝜈𝑒 (32)
𝐸 𝑝
the theoretical values (𝐹LT ) and the experimental values (𝐹𝑢.𝑒𝑥𝑝 ) is equal
where 𝜈𝑝 = 0.50 is the Poisson’s ratio for fully plastic behaviour and
to 0.93 with the standard deviation of 0.06. The numerical values are
𝜈𝑒 = 0.30 is the elastic value. The coefficient 𝜉 takes into account
reported in Table A.6 of Annex.
the mechanical non-linearity of aluminium material i.e., local buckling
Regarding the collapse mode due to the concentrate load, 14 ex- in the elastic–plastic range, according to the relationship presented
perimental tests have been considered: 4 experiments by Tryland in [16,41]:
et al. [52], other 4 by Wang et al. [53] and 6 experimental results [ √ ]
presented in Section 3.5. The comparison between the theoretical 𝐸𝑠 𝑛 − 8 8 𝐸𝑡
𝜉= + (33)
results (𝐹WC ) and the experimental values (𝐹𝑢.𝑒𝑥𝑝 ) is shown in Fig. 19f. 𝐸 8 𝑛 𝐸𝑠
In this case, the average value of the ratios 𝐹WC ∕𝐹𝑢.𝑒𝑥𝑝 is equal to 0.92
where 𝐸𝑡 is the tangent modulus and 𝑛 is the Ramberg–Osgood coef-
with a standard deviation of 0.21. The numerical values are reported
ficient. In Eq. (30) 𝜁 is a correction factor accounting for interactive
in Table A.7 of Annex.
buckling and represents the ratio between the buckling factor 𝑘 ac-
counting for interactive buckling and the buckling factor 𝑘0 evaluated
4.2. Annex L of revised Eurocode 9
for the isolated plate element, i.e. 𝜁 = 𝑘∕𝑘0 . In the case of unstiffened
cross-section parts, acting as flange:
In the revised version of prEN 1999 − 1 − 1 [42], the current Annex L
0.35𝑏2 ∕𝑏1 ( )3
provides an alternative approach to predict the maximum bending mo- 𝜁 = 3.00 − − 0.0059 𝑏2 ∕𝑏1 ≥ 1 (34)
ment of aluminium members by considering the local buckling effects 0.60 + 𝑏2 ∕𝑏1
in the elastic–plastic range. This procedure represents an extension of where, according to Fig. 7, 𝑏1 and 𝑏2 are equal to:
effective thickness approach (ETM) accounting for interactive plastic 𝐵
buckling. 𝑏1 = 𝑏2 = 𝐻 − 𝑡 𝑓 (35)
2
The new version of ETM is presented in [16,41] and it can be The Eq. (34) is derived from the expression of k given by BS5950-
applied to the aluminium members in compression or in bending. It 5 [60] considering that, in this case, 𝑘0 = 4. In the case of the plate
represents a refinement of the traditional effective thickness method elements acting as webs 𝜁 = 1.0 is adopted. Finally, 𝛼𝐿𝑆 is the factor
commonly used to consider the elastic local buckling effects referring accounting for the influence of the longitudinal stress gradient, the
to fourth class sections. The revised ETM is based on a strain-dependent following relation can be employed [61]:
evaluation of the effective thickness considering the local buckling in
4
the elastic–plastic region and the interaction between the plate ele- 𝛼𝐿𝑆 = 1 + ( )0.95 (36)
𝐿𝑠
ments constituting the I shaped section. Moreover, the formulation for − 0.60
𝑏
evaluating the effective thickness also accounts for the stress gradient
in the longitudinal direction. According to a generic plate in compres- where 𝐿𝑠 is the shear length and 𝑏 is the plate width in compression.
sion, the slenderness parameter of the plate elements constituting the The shear length is defined as the distance between the point of zero
member section can be evaluated as a function of the strain level [16] bending moment and the section where the maximum bending moment
as follows: occurs (in this case 𝐿𝑠 = 𝐿∕2).
√ By combining Eq. (30) with the buckling curves of EN1999-1-1, it
𝛽 𝑏 𝜀
= 17.54𝜂 (30) is possible to determine the effective thickness in the elastic–plastic
𝜖0 𝑡 𝜇𝜁 𝛼𝐿𝑆
region as function of the strain level 𝑡𝑒𝑓 𝑓 = 𝑡𝑒𝑓 𝑓 (𝜀) = 𝜌𝑐 𝑡. The reduction
where the factors 𝜂 and 𝜇 are given by: factor accounting for local buckling is computed as:
𝜉𝐸 1 ( √ )
2 𝛽 1
𝜂= √ 𝜇= (31) 𝜌𝑐 = 1 𝑖𝑓 ≤ 𝑐1 + 𝑐12 − 𝑐2 (3 + 𝜓) (37)
𝑘 𝐸𝑠 1 − 𝜈 2 𝜖0 2

19
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table 15
Summary of comparison between theoretical and experimental values.
Failure mode Reference 𝐹𝑢.𝑡ℎ ∕𝐹𝑢.exp
(EN1999-1-1/prEn1999-1-1)
𝜇[-] 𝑐𝑜𝑣[-]
Bending moment (BM) Section 6.2.5 0.82 0.11
Bending moment (BM(L) ) Annex L 0.95 0.06
Lateral–torsional buckling (LT-1) Section 6.3.2.1 0.93 0.11
Lateral–torsional buckling (LT-2) Section 6.3.2.1 0.88 0.16
Lateral–torsional buckling Section 6.3.2.1 0.93 0.06
(LT-3, 𝑘𝑤 = 0.50)
Lateral–torsional buckling Section 6.3.2.1 0.76 0.13
(LT-3, 𝑘𝑤 = 1.00)
Web crippling Section 6.7.5 0.92 0.23

and:
( √ )
𝑐1 𝑐 (3 + 𝜓) 𝛽 1
𝜌𝑐 = − 2( )2 𝑖𝑓 > 𝑐1 + 𝑐12 − 𝑐2 (3 + 𝜓) (38)
𝛽∕𝜖0 4 𝛽∕𝜖0 𝜖0 2

The parameter 𝜓 accounts for the stress distribution along the

loaded edge of the plate. It is given by the ratio between the maximum
compression strain at one end of the plate and the strain at the second
end of the plate element. In the case of uniform compression, it results
in 𝜓 = 1 while 𝜓 < 0 when the second end of the plate element is
subject to tension. The coefficients c1 and c2 are reported in Eurocode
9 and they are defined according to the buckling curves.
So, the effective thickness of plate elements as a function of the
strain level can be computed by applying Eqs. (30), (37), (38). The
previous relationships can be easily employed within a numerical pro-
cedure which works for increasing values of the curvature where the
neutral axis is derived by exploiting the translation equilibrium equa-
tion and the bending moment corresponding to the imposed curvature
is obtained by the rotation equilibrium. Therefore, the moment versus
Fig. 20. Comparison between experimental results with those obtained by Annex L.
curvature relation is derived accounting for the occurrence of local
buckling in the elastic–plastic range. Also in this case, the numerical
(𝐿)
values, expressed as maximum load 𝐹BM according to Eq.(28), have
been computed with reference to the average values of geometric and considered: (1) Bending moment at the midspan; (2) Lateral–torsional
mechanical properties reported in Tables 6 and 7. buckling due to pure bending; (3) Lateral–torsional buckling due to
A comparison between the experimental results and those obtained non-uniform bending moment; (4) Web crippling due to concentrated
by ETM method, is depicted in Fig. 20 and the numerical values load.
are reported in Table A.3. It is evident that this procedure provides The comparison between the experimental results with those ob-
theoretical values closer to the experimental ones than the current tained by the current design rules highlighted that in the case of the
(𝐿)
design rules. In fact, the mean value of the ratios 𝐹BM ∕𝐹𝑢.𝑒𝑥𝑝 is equal prediction of the flexural resistance, the theoretical values are very
to 0.95 with a standard deviation of 0.06. conservative compared to the experimental ones. In fact, the mean
A final comparison between the theoretical values (𝐹𝑢.𝑡ℎ ), obtained value of the numerical/experimental ratio (𝐹BM ∕𝐹𝑢.𝑒𝑥𝑝 ) is about 0.82
by EN 1999 − 1 − 1 [9] and prEN 1999 − 1 − 1 [42] and the experimental with a standard deviation equal to 0.09.
results (𝐹𝑢.𝑒𝑥𝑝 ), presented in Section 3.5, is reported in Table 15 in terms In the case of the collapse due to the lateral torsional buckling,
of mean values (𝜇) and coefficients of variation (𝑐𝑜𝑣) with reference to two cases have been considered: (1) concentrated load at the midspan;
𝐹𝑢.𝑡ℎ ∕𝐹𝑢.𝑒𝑥𝑝 ratios. (2) Unform bending moment along the specimen. In the first one,
the accuracy is quite high. In fact, the mean value of the numer-
5. Conclusions ical/experimental ratio (𝐹BM ∕𝐹𝑢.𝑒𝑥𝑝 ) is about 0.93 with a standard
deviation equal to 0.10. Instead, in the second case, according to Annex
In this paper, an extensive overview on the ultimate behaviour I of EN 1999-1-1, two main loading conditions have been considered:
of I-shaped beams under different loading conditions is presented. (1) Four-point scheme; (2) Uniform moment scheme. The theoretical
Preliminarily, the main experimental tests, presented in the scien- results are strongly influenced by the different values of the warping
tific literature, have been analysed highlighting the principal collapse factor 𝑘𝑤 and the coefficients 𝐶𝑖 , as reported in Annex.
modes of beams subjected to pure bending, non-uniform bending, and Regarding the failure mode due to the concentrated load, the design
concentrated loading. All mechanical and the experimental results have prescriptions are not very accurate because, in some cases, the theoreti-
been collected. cal resistance is overestimated, while in others it is underestimated with
To increase the number of the experimental results, an experimental a high standard deviation of 0.21.
campaign on the flexural behaviour of I-shaped aluminium beams Finally, with reference to the cases governed by flexural behaviour
subjected to non-uniform bending moment has been performed at the only, Annex L of revised version of Eurocode 9 is applied. Annex L
University of Salerno considering 15 three-point bending tests of dif- presents an extension of the effective thickness approach, currently
ferent aluminium alloys (6060-T66, 6082-T6, 6005-T6) with different adopted in EN 1999-1-1, which considers the interactions between plate
width-to-thickness ratios and for three different lengths. Then, the elements of section and the strain-hardening behaviour of aluminium
numerical analyses have been carried out according to the current Eu- alloys in the predictions of the ultimate response of beams in bending.
rocode 9 provisions. In particular, the following failure mode have been By comparing the experimental values with those obtained by Annex

20
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

L, it is evident that the prediction of the ultimate resistance is more Methodology, Supervision, Writing – review & editing. Alessandro
accurate than the one reported by the current version of EN 1991–1. Pisapia: Data curation, Formal analysis, Investigation, Resources,
The finite element analyses have been carried out in order to Validation, Writing – original draft.
improve the experimental results derived by the tests. However, for
sake of shortness, the results obtained are not herein delivered as they Declaration of competing interest
will be presented in a forthcoming work where the attention will also
be focused on the evaluation of the main mechanical and geomet-
Authors state that there is no conflict of interest.
ric parameters which influence the ultimate behaviour of aluminium
H-shaped beams subjected to three-point test.
Data availability
CRediT authorship contribution statement
All data are available within the manuscript.
Rosario Montuori: Conceptualization, Supervision, Writing –
review & editing. Elide Nastri: Resources, Supervision, Visualization, Annex A
Writing – review & editing. Vincenzo Piluso: Conceptualization,

Table A.1
Mechanical properties of experimental tests collected by the scientific literature.
𝐴𝑢𝑡ℎ𝑜𝑟𝑠 𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐴𝑙𝑙𝑜𝑦 𝐵𝐶 𝐸 [MPa] 𝑓0.2 [MPa] 𝑓𝑢 [MPa] 𝑛[-] 𝜀𝑢 [%]
S1 27ST – 68948 393.69 427.78 – –
S2 27ST – 68948 393.69 427.78 – –
S3 27ST – 68948 393.69 427.78 – –
S4 27ST – 68948 393.69 427.78 – –
Dumont & Hill [18] S5 27ST – 68948 393.69 427.78 – –
S6 27ST – 68948 393.69 427.78 – –
S7 27ST – 68948 393.69 427.78 – –
S8 27ST – 68948 393.69 427.78 – –
S9 27ST – 68948 393.69 427.78 – –
S10 27ST – 68948 393.69 427.78 – –
I1-2m-1 6082-T6 A 66716 312.20 324.20 74.00 6.90
I2-1m-1 6082-T6 A 66874 279.40 300.70 62.00 6.90
I2-1m-2 6082-T6 A 66874 279.40 300.70 62.00 6.90
Moen et al. [21] I2-1m-3 6082-T6 A 66874 279.40 300.70 62.00 6.90
I2-2m-1 6082-T6 A 66874 279.40 300.70 62.00 6.90
I2-2m-2 6082-T6 A 66874 279.40 300.70 62.00 6.90
I2-2m-3 6082-T6 A 66874 279.40 300.70 62.00 6.90
I2-3m-1 6082-T6 A 66874 279.40 300.70 62.00 6.90
H-480 6082-T6 A 70774 308.90 331.30 – 7.34
Tryland et al. [52] H-960 6082-T6 A 70774 308.90 331.30 – 7.34
𝐻 − 𝐶 = 20 6082-T6 A 70774 308.90 331.30 – 7.34
𝐻 −𝐶 =5 6082-T6 A 70774 308.90 331.30 – 7.34
I2167-1000-1 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1000-2 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1200-1 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1200-2 6061-T6 A 69125 240.80 279.85 27.30 7.00
Zhang & Wu [50] I2167-1400-1 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1400-2 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1600-1 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1600-2 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1800-1 6061-T6 A 69125 240.80 279.85 27.30 7.00
I2167-1800-2 6061-T6 A 69125 240.80 279.85 27.30 7.00
6061 − 1 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 1 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 2 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 2 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 3 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 3 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 4 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 4 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 5 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 5 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 6 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 6 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 7 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 7 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 8 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 8 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 9 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 9 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6061 − 10 − 𝐴 6061-T6 A 54750 283.00 295.00 31.50 8.00
Wang et al. [49] 6061 − 10 − 𝐵 6061-T6 A 54750 283.00 295.00 31.50 8.00
6063 − 1 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 1 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
(continued on next page)

21
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table A.1 (continued).

6063 − 2 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 2 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 3 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 3 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 4 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 4 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 5 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 5 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 6 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 6 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 7 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 7 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 8 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 8 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 9 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 9 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 10 − 𝐴 6063-T5 B 55000 146.00 184.50 16.50 12.00
6063 − 10 − 𝐵 6063-T5 B 55000 146.00 184.50 16.50 12.00
I07 6061-T6 A 65669 247.70 290.30 – –
I08 6061-T6 A 65669 247.70 290.30 – –
I09 6061-T6 A 65669 247.70 290.30 – –
Guo et al. [51] I10 6061-T6 A 65669 247.70 290.30 – –
I11 6061-T6 A 65669 247.70 290.30 – –
I12 6061-T6 A 65669 247.70 290.30 – –
I13 6061-T6 A 65669 247.70 290.30 – –
I14 6061-T6 A 65669 247.70 290.30 – –
T5-4.4-50.5 6063-T5 B 65919 158.93 205.13 15.88 –
T5-12-1-56-1 6063-T5 B 65919 158.93 205.13 15.88 –
Wang et al. [53] T5-9.8-72-3 6063-T5 B 65919 158.93 205.13 15.88 –
T6-7.9-42-9 6061-T6 A 70185 248.55 283.88 24.88 –
T5-8.8-71.6 6063-T5 B 65919 158.93 205.13 15.88 –
T6-5.0-56.1 6061-T6 A 70185 248.55 283.88 24.88 –

Legend
𝐵𝐶 ∶ Buckling Class according to EN 1999 − 1 − 1.
𝐸 ∶ Young’s modulus [MPa].
𝑓0.2 ∶ Stress at the residua strain 0.2% [MPa].
𝑓𝑢 ∶ Ultimate stress [MPa].
𝑛 ∶ Ramberg–Osgood coefficient [-].
𝜀𝑢 ∶ Maximum strain corresponding to the ultimate stress [MPa].

Table A.2
Geometric properties of experimental tests collected by the scientific literature.
𝐴𝑢𝑡ℎ𝑜𝑟𝑠 𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐵𝑠 [mm] 𝐵𝑖 [mm] 𝐻[mm] 𝑡𝑓 [mm] 𝑡𝑤 [mm] 𝐿[mm] 𝐶𝑙𝑎𝑠𝑠𝑆𝑒𝑐𝑡𝑖𝑜𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN]
S1 35.18 35.18 101.55 2.05 2.46 2235 4 8.01
S2 35.18 35.18 101.55 2.05 2.46 1981 4 9.73
S3 35.18 35.18 101.55 2.05 2.46 1727 4 12.17
S4 35.18 35.18 101.55 2.05 2.46 1473 4 15.38
Dumont & Hill [18] S5 35.18 35.18 101.55 2.05 2.46 1219 4 20.41
S6 35.18 35.18 101.55 2.05 2.46 965 4 31.16
S7 35.18 35.18 101.55 2.05 2.46 838 4 38.24
S8 35.18 35.18 101.55 2.05 2.46 711 4 45.76
S9 35.18 35.18 101.55 2.05 2.46 584 4 52.97
S10 35.18 35.18 101.55 2.05 2.46 457 4 57.16
I1-2m-1 119.90 119.90 120.20 7.96 5.09 2000 4 73.76
I2-1m-1 70.00 70.00 80.35 4.94 4.97 1000 4 40.51
I2-1m-2 70.00 70.00 80.35 4.94 4.97 1000 4 39.17
Moen et al. [21] I2-1m-3 70.00 70.00 80.35 4.94 4.97 1000 4 39.17
I2-2m-1 70.00 70.00 80.35 4.94 4.97 2000 4 19.42
I2-2m-2 70.00 70.00 80.35 4.94 4.97 2000 4 19.59
I2-2m-3 70.00 70.00 80.35 4.94 4.97 2000 4 19.92
I2-3m-1 70.00 70.00 80.35 4.94 4.97 3000 4 12.61
𝐻 − 480 − 1 120.00 120.00 120.00 7.91 5.05 480 4 146.60
Tryland et al. [52] 𝐻 − 960 − 1 120.00 120.00 120.00 7.91 5.05 960 4 119.00
𝐻 − 𝐶 = 20 − 1 120.00 120.00 120.00 7.91 5.05 760 4 101.68
𝐻 −𝐶 =5−1 120.00 120.00 120.00 7.91 5.05 715 4 83.68
I2167-1000-1 66.00 66.00 102.00 6.00 6.00 1000 3 34.26
I2167-1000-2 66.00 66.00 102.00 6.00 6.00 1000 3 32.12
I2167-1200-1 66.00 66.00 102.00 6.00 6.00 1200 3 24.88
I2167-1200-2 66.00 66.00 102.00 6.00 6.00 1200 3 25.85
Zhang & Wu [50] I2167-1400-1 66.00 66.00 102.00 6.00 6.00 1400 3 19.60
I2167-1400-2 66.00 66.00 102.00 6.00 6.00 1400 3 19.90
I2167-1600-1 66.00 66.00 102.00 6.00 6.00 1600 3 14.56
(continued on next page)

22
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table A.2 (continued).

I2167-1600-2 66.00 66.00 102.00 6.00 6.00 1600 3 15.53
I2167-1800-1 66.00 66.00 102.00 6.00 6.00 1800 3 12.55
I2167-1800-2 66.00 66.00 102.00 6.00 6.00 1800 3 12.50
6061 − 1 − 𝐴 34.94 17.74 98.00 3.91 2.88 1800 2 14.05
6061 − 1 − 𝐵 34.94 17.40 97.80 3.92 2.84 1800 2 14.05
6061 − 2 − 𝐴 34.98 17.48 97.68 3.94 2.86 1500 2 19.07
6061 − 2 − 𝐵 34.96 17.56 97.96 3.94 2.84 1500 2 19.07
6061 − 3 − 𝐴 17.78 17.78 97.98 3.93 2.84 1250 2 6.27
6061 − 3 − 𝐵 17.36 17.70 97.72 3.92 2.86 1250 2 6.27
6061 − 4 − 𝐴 17.62 17.76 97.92 3.94 2.86 1050 2 7.71
6061 − 4 − 𝐵 18.02 18.00 97.74 3.94 2.84 1050 2 7.71
6061 − 5 − 𝐴 17.36 17.30 97.82 3.94 2.84 850 2 10.21
6061 − 5 − 𝐵 17.40 17.66 97.78 3.94 2.84 850 2 10.21
6061 − 6 − 𝐴 16.88 16.28 97.96 3.90 2.82 700 2 12.79
6061 − 6 − 𝐵 17.70 17.74 97.70 3.91 2.82 700 2 12.79
6061 − 7 − 𝐴 17.92 17.98 97.66 3.93 2.86 580 2 18.44
6061 − 7 − 𝐵 17.78 17.84 98.00 3.95 2.86 580 2 18.44
6061 − 8 − 𝐴 17.70 17.74 97.88 3.94 2.90 480 2 22.70
6061 − 8 − 𝐵 18.20 17.22 97.96 3.92 2.88 480 2 22.70
6061 − 9 − 𝐴 17.96 34.86 97.86 3.93 2.88 420 2 27.79
6061 − 9 − 𝐵 17.86 34.88 98.04 3.93 2.92 420 2 27.79
6061 − 10 − 𝐴 17.68 34.92 97.96 3.92 2.90 350 2 34.68
Wang et al. [49] 6061 − 10 − 𝐵 17.52 34.94 97.90 3.93 2.88 350 2 34.68
6063 − 1 − 𝐴 34.98 17.64 98.02 3.93 2.90 2200 1 10.52
6063 − 1 − 𝐵 35.00 17.60 97.90 3.92 2.90 2200 1 10.52
6063 − 2 − 𝐴 34.96 17.84 97.64 3.92 2.78 1800 1 14.38
6063 − 2 − 𝐵 34.98 17.82 97.96 3.93 2.86 1800 1 14.38
6063 − 3 − 𝐴 17.64 17.74 97.84 3.93 2.86 1500 1 5.01
6063 − 3 − 𝐵 17.84 17.76 97.88 3.93 2.86 1500 1 5.01
6063 − 4 − 𝐴 17.84 17.98 98.02 3.95 2.92 1250 1 5.99
6063 − 4 − 𝐵 17.82 17.92 97.82 3.93 2.86 1250 1 5.99
6063 − 5 − 𝐴 17.98 17.80 97.78 3.94 2.88 1050 1 7.88
6063 − 5 − 𝐵 17.96 17.54 97.96 3.93 2.86 1050 1 7.88
6063 − 6 − 𝐴 17.42 17.82 97.70 3.93 2.86 850 1 9.87
6063 − 6 − 𝐵 17.74 17.54 98.10 3.94 2.92 850 1 9.87
6063 − 7 − 𝐴 17.76 17.98 97.80 3.94 2.86 700 1 13.15
6063 − 7 − 𝐵 18.08 17.88 97.94 3.95 2.90 700 1 13.15
6063 − 8 − 𝐴 17.78 18.08 97.96 3.94 2.90 600 1 12.23
6063 − 8 − 𝐵 17.96 17.84 97.82 3.94 2.86 600 1 12.23
6063 − 9 − 𝐴 18.28 34.92 97.90 3.96 2.88 550 1 16.30
6063 − 9 − 𝐵 17.70 34.94 97.76 3.96 2.86 550 1 16.30
6063 − 10 − 𝐴 17.80 34.96 97.70 3.96 2.82 450 1 19.65
6063 − 10 − 𝐵 18.12 34.96 97.84 3.93 2.84 450 1 19.65
I07 70.00 70.00 121.00 7.00 5.00 1500 3 19.30
I08 70.00 70.00 121.00 7.00 5.00 1500 3 19.16
I09 70.00 70.00 121.00 7.00 5.00 1800 3 28.30
Guo et al. [51] I10 70.00 70.00 121.00 7.00 5.00 1800 3 27.48
I11 70.00 70.00 121.00 7.00 5.00 2100 3 24.19
I12 70.00 70.00 121.00 7.00 5.00 2100 3 22.38
I13 70.00 70.00 121.00 7.00 5.00 2400 3 9.44
I14 70.00 70.00 121.00 7.00 5.00 2400 3 7.83
T5-4.4–50.5 64.70 64.70 239.75 6.88 4.47 609 2 97.62
T5-12-1-56-1 168.80 168.80 239.25 6.83 4.02 609 4 90.27
Wang et al. [53] T5-9.8-72-3 119.70 119.70 270.00 5.91 3.57 600 4 74.27
T6-7.9-42-9 163.50 163.50 348.50 9.85 7.66 799 4 207.12
T5-8.8–71.6 103.70 103.70 270.00 5.68 3.61 1200 4 65.92
T6-5.0–56.1 108.80 108.80 449.80 10.06 7.65 1200 4 277.70

Fig. A.1. Geometric scheme of I-shaped beam.

23
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table A.3
Numerical results in the case of bending moment (BM).
(𝐿)
𝐹BM (𝐿) 𝐹BM
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝐹BM [kN] 𝐹𝑢.𝑒𝑥𝑝
𝐹BM [kN] 𝐹𝑢.𝑒𝑥𝑝

I1-2m-1 73.76 61.12 0.83 72.15 0.98

I2-1m-1 40.51 30.23 0.75 35.07 0.87
I2-1m-2 39.17 30.23 0.77 35.07 0.90
I2-1m-3 39.17 30.23 0.77 35.51 0.91
I2-2m-1 19.42 15.11 0.78 17.69 0.91
I2-2m-2 19.59 15.11 0.77 17.69 0.90
I2-2m-3 19.92 15.11 0.76 17.47 0.88
I2-3m-1 12.61 10.08 0.80 11.63 0.92
H90_L1 17.52 13.83 0.79 16.17 0.92
H90_L2 21.19 22.17 1.05 23.05 1.09
H100_L3 245.66 226.34 0.92 284.35 1.16
H120_L2 104.12 94.79 0.91 103.21 0.99
𝜇 0.82 0.95
𝜎 0.09 0.06

Table A.4
Numerical results in the case of LT Buckling 1 (LT).
𝐹LT
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝐹LT [kN] 𝐹𝑢.𝑒𝑥𝑝

I2167-1000-1 34.26 31.42 0.92

I2167-1000-2 32.12 31.42 0.98
I2167-1200-1 24.88 23.07 0.93
I2167-1200-2 25.85 23.07 0.89
I2167-1400-1 19.60 17.49 0.89
I2167-1400-2 19.90 17.49 0.88
I2167-1600-1 14.56 13.68 0.94
I2167-1600-2 15.53 13.68 0.88
I2167-1800-1 12.55 10.99 0.88
I2167-1800-2 12.50 10.99 0.88
I07 19.30 20.73 1.07
I08 19.16 20.73 1.08
I09 28.30 25.23 0.89
I10 27.48 25.23 0.92
I11 24.19 19.30 0.80
I12 22.38 19.30 0.86
I13 9.44 8.27 0.88
I14 7.83 8.27 1.06
H100_L1 97.56 76.94 0.79
H100_L2 165.75 135.97 0.82
H120_L1 50.93 55.47 1.09
H140_L1 96.70 109.98 1.14
H160_L1 117.66 108.05 0.92
𝜇 0.93
𝜎 0.10

Table A.5
Numerical results in the case of LT Buckling 2 (LT).
𝐹LT
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝐹LT [kN] 𝐹𝑢.𝑒𝑥𝑝

S9 52.97 35.60 0.67

S10 57.16 40.82 0.71
6061 − 7 − 𝐴 23.17 16.13 0.70
6061 − 7 − 𝐵 23.17 15.95 0.69
6061 − 8 − 𝐴 28.53 21.18 0.74
6061 − 8 − 𝐵 28.53 22.30 0.78
6061 − 9 − 𝐴 34.93 31.26 0.90
6061 − 9 − 𝐵 34.93 31.01 0.89
6061 − 10 − 𝐴 43.58 40.67 0.93
6061 − 10 − 𝐵 43.58 39.85 0.91
6063 − 8 − 𝐴 15.37 15.00 0.98
6063 − 8 − 𝐵 15.37 15.19 0.99
6063 − 9 − 𝐴 20.49 21.03 1.03
6063 − 9 − 𝐵 20.49 19.62 0.96
6063 − 10 − 𝐴 24.69 26.53 1.07
6063 − 10 − 𝐵 24.69 27.42 1.11
𝜇 0.88
𝜎 0.14

24
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

Table A.6
Numerical results in the case of LT Buckling 3 (LT).
𝐹LT 𝐹LT 𝐹LT 𝐹LT
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝐹LT [kN] 𝐹𝑢.𝑒𝑥𝑝
𝐹LT [kN] 𝐹𝑢.𝑒𝑥𝑝
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝐹LT [kN] 𝐹𝑢.𝑒𝑥𝑝
𝐹LT [kN] 𝐹𝑢.𝑒𝑥𝑝

𝑘w = 0.50 𝑘w = 1.00 𝑘w = 0.50 𝑘w = 1.00

S1 8.01 7.14 0.89 5.33 0.67 6061 − 6 − 𝐴 12.79 11.82 0.92 8.98 0.70
S2 9.73 8.53 0.88 6.20 0.64 6061 − 6 − 𝐵 12.79 13.36 1.05 9.85 0.77
S3 12.17 10.52 0.86 7.42 0.61 6063 − 1 − 𝐴 10.52 9.74 0.93 9.17 0.87
S4 15.38 13.48 0.88 9.21 0.60 6063 − 1 − 𝐵 10.52 9.71 0.92 9.15 0.87
S5 20.41 18.15 0.89 12.09 0.59 6063 − 2 − 𝐴 14.38 12.90 0.90 11.93 0.83
S6 31.16 25.74 0.83 16.83 0.54 6063 − 2 − 𝐵 14.38 13.07 0.91 12.11 0.84
S7 38.24 30.82 0.81 20.54 0.54 6063 − 3 − 𝐴 5.01 4.59 0.92 4.12 0.82
S8 45.76 36.06 0.79 25.67 0.56 6063 − 3 − 𝐵 5.01 4.68 0.93 4.19 0.84
6061 − 1 − 𝐴 14.05 13.17 0.94 12.18 0.87 6063 − 4 − 𝐴 5.99 6.03 1.01 5.23 0.87
6061 − 1 − 𝐵 14.05 13.06 0.93 12.12 0.86 6063 − 4 − 𝐵 5.99 5.91 0.99 5.11 0.85
6061 − 2 − 𝐴 19.07 17.60 0.92 16.06 0.84 6063 − 5 − 𝐴 7.88 7.61 0.97 6.34 0.81
6061 − 2 − 𝐵 19.07 17.61 0.92 16.06 0.84 6063 − 5 − 𝐵 7.88 7.52 0.95 6.28 0.80
6061 − 3 − 𝐴 6.27 5.89 0.94 5.09 0.81 6063 − 6 − 𝐴 9.87 9.64 0.98 7.57 0.77
6061 − 3 − 𝐵 6.27 5.64 0.90 4.90 0.78 6063 − 6 − 𝐵 9.87 10.07 1.02 7.99 0.81
6061 − 4 − 𝐴 7.71 7.38 0.96 6.14 0.80 6063 − 7 − 𝐴 13.15 13.44 1.02 9.96 0.76
6061 − 4 − 𝐵 7.71 7.69 1.00 6.36 0.82 6063 − 7 − 𝐵 13.15 14.04 1.07 10.47 0.80
6061 − 5 − 𝐴 10.21 9.60 0.94 7.58 0.74 𝜇 0.93 0.76
6061 − 5 − 𝐵 10.21 9.71 0.95 7.61 0.75 𝜎 0.06 0.10

Table A.7 [12] C. Faella, F.M. Mazzolani, V. Piluso, G. Rizzano, Local buckling of aluminium
Numerical results in the case of web crippling (WC). members: testing and classification, J. Struct. Eng. 126 (3) (2000) 353–360, ASCE
𝐹WC 2000.
𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝐹𝑢.𝑒𝑥𝑝 [kN] 𝐹WC [kN] 𝐹𝑢.𝑒𝑥𝑝
[13] F.M. Mazzolani, V. Piluso, G. Rizzano, Experimental analysis of aluminium
𝐻 − 480 − 1 146.60 135.78 0.93 alloy channels subjected to local buckling under uniform compression, in: XVIII
𝐻 − 960 − 1 119.00 134.99 1.13 Congresso C.T.a. Venice 2001.
𝐻 − 𝑐 = 20 − 1 101.68 135.15 1.33 [14] F.M. Mazzolani, V. Piluso, G. Rizzano, Local buckling of aluminum alloy angles
𝐻 −𝐶 =5−1 83.68 101.20 1.21 under uniform compression, J. Struct. Eng. 137 (2) (2011) 173–184.
T5-4.4–50.5 97.62 71.73 0.73 [15] M.N. Su, B. Young, L. Gardner, Testing and design of aluminium alloy cross
T5-12-1-56-1 90.27 61.77 0.68 sections in compression, J. Struct. Eng. 140 (9) (2014) 04014047.
T5-9.8-72-3 74.27 44.74 0.60
[16] V. Piluso, A. Pisapia, Interactive plastic local buckling of box-shaped aluminium
T5-8.8–71.6 65.92 44.23 0.67
members under uniform compression, Thin-Walled Struct. 164 (2021) 107828.
T6-5.0–56.1 277.70 227.17 0.82
[17] V. Piluso, A. Pisapia, G. Rizzano, Local buckling of aluminium channels under
H90_L3 36.17 31.83 0.88
uniform compression: Theoretical analysis and experimental tests, Thin-Walled
H120_L3 123.81 108.54 0.88
Struct. 179 (2022) 109511.
H140_L2 169.85 182.02 1.07
[18] C. Dumont, H.N. Hill, Lateral Stability of Equal Flanged Aluminium Alloy I-Beams
H140_L3 188.38 179.87 0.95
Subjected To Pure Bending, NACA Technical Note n. 770, National Advisory
H160_L2 153.38 150.36 0.98
Committee for Aeronautics, Washington, 1940.
H160_L3 170.50 151.14 0.89
[19] F. Panlio, The theory of limit design applied to magnesium alloy and aluminium
𝜇 0.92 alloy structures, R. Aerinaut. Soc. (1947) 534–571.
𝜎 0.21 [20] T. Welo, Inelastic Deformation Capacity of Flexurally-Loaded Aluminium Alloy
Structures, Norwegian University of Science and Technology, 1990.
[21] L.A. Moen, O.S. Hopperstad, M. Langseth, Rotational capacity of aluminum
beams under moment gradient. I: Experiments, J. Struct. Eng. 125 (8) (1999)
References 910–920.
[22] L.A. Moen, G. De Matteis, O.S. Hopperstad, M. Langseth, R. Landolfo, F.M.
Mazzolani, Rotational capacity of aluminum beams under moment gradient. II:
[1] F.M. Mazzolani, 3D aluminium structures, Thin-Walled Struct. 61 (2012)
numerical simulation, J. Struct. Eng. 125 (8) (1999) 921–929.
258–266.
[23] G. De Matteis, L.A. Moen, M. Langseth, R. Landolfo, O.S. Hopperstad, F.M.
[2] E. Georgantzia, M. Gkantou, G.S. Kamaris, Aluminium alloys as structural
Mazzolani, Cross-sectional classification for aluminium beams: a parametric
material: A review of research, Eng. Struct. 227 (2021) 111372.
study, J. Struct. Eng. 127 (3) (2001) 271–279.
[3] J.G. Kauffman, Introduction To Aluminum Alloys and Tempers, ASM
[24] G. De Matteis, R. Landolfo, M. Manganiello, F.M. Mazzolani, Inelastic be-
International, 2000.
haviour of I-shaped aluminium beams: Numerical analysis and cross-sectional
[4] G. De Matteis, G. Brando, F. Caldoso, F. D’Agostino, Seismic performance of
classification, Comput. Struct. 82 (2004) 2157–2171.
dual steel frames with dissipative metal shear panels, Ingegneria Sismica 35 (2)
[25] M. Manganiello, G. De Matteis, R. Landolfo, Inelastic flexural strength of
(2018) 124–141.
aluminium alloy structures, Eng. Struct. 28 (4) (2006) 593–608.
[5] G. De Matteis, G. Brando, Metal shear panels for seismic protection of buildings: [26] U. Eberwien, G. Valtinat, The fullness method: a direct procedure for calculation
recent findings and perspectives, Ingegneria Sismica 33 (3) (2016) 5–27. of the bending moment of a symmetrical aluminium cross section, in: Proceedings
[6] G. De Matteis, G. Brando, F.M. Mazzolani, Hysteretic behaviour of bracing-type of the Eighth International Conference in Aluminium, INALCO, Germany, 2001,
pure aluminium shear panels by experimental tests, Earthq. Eng. Struct. Dynam. pp. 1–10, pp. 3.3.
40 (10) (2011) 1143–1162. [27] B.W. Schafer, T. Peköz, Direct strength prediction of cold-formed steel members
[7] Y. Wu, L.X. Peng, X.C. He, J.S. Yang, Experimental and numerical analyses on using numerical elastic buckling solutions, in: International Specialty Confer-
mechanical properties of aluminum alloy L-shaped joints under cyclic load, Eng. ence on Cold-Formed Steel Structures: Recent Research and Developments in
Struct. 245 (2021) 112854. Cold-Formed Steel Design and Construction, 1998, pp. 69–76.
[8] W. Ramberg, WR Osgood: Description of Stress–Strain Curves By Three [28] J.H. Zhu, B. Young, Design of aluminum alloy flexural members using direct
Parameters, NACA Technical Note n. 902, Washington, 1943. strength method, J. Struct. Eng. 135 (5) (2009) 558–566.
[9] EN 1999-1-1. Eurocode 9, Design of Aluminium Structures – Part 1.1: General [29] Y. Kim, T. Peköz, Ultimate flexural strength of aluminium sections, Thin-Walled
Structural Rules, European Committee for Standardisation, Brussels, Belgium, Struct. 48 (10–11) (2010) 857–865.
2007. [30] M.N. Su, B. Young, L. Gardner, Deformation-based design of aluminum alloy
[10] EN 1993-1-1. Eurocode 3, Design of Steel Structures - Part 1-1: General Rules beams, Eng. Struct. 80 (2014) 339–349.
and Rules for Buildings, European Committee for Standardisation, 2005. [31] M.N. Su, B. Young, L. Gardner, Flexural response of aluminium alloy SHS and
[11] A. Pisapia, E. Nastri, V. Piluso, A. Formisano, F.M. Mazzolani, Experimental RHS with internal stiffeners, Eng. Struct. 121 (2016) 170–180.
campaign on structural aluminium alloys under monotonic and cyclic loading, [32] GB 50429, Code for Design of Aluminium Structures, Ministry of Construction
Eng. Struct. 282 (2023) 115836. of the People’s Republic of China, 2007.

25
R. Montuori, E. Nastri, V. Piluso et al. Thin-Walled Structures 191 (2023) 111038

[33] North American Specification for the Design of Cold-Formed Steel Structures [48] M. Rouholamin, S. Gunalan, K. Poologanathan, H. Karampour, Shear strength
Members, American Iron and Steel Institute, Washington, DC, 2001. reduction of aluminium lipped channel beams due to web openings, Thin-Walled
[34] Australian/New Zealand standard (AS/NZS): aluminium structures part 1: Limit Struct. 164 (2021) 107697.
stade design, sydney, 1997. [49] Y.Q. Wang, H.X. Yuan, Y.J. Shi, M. Cheng, Lateral–torsional buckling resistance
[35] P. Castaldo, E. Nastri, V. Piluso, FEM simulations and rotation capacity eval- of aluminium I-beams, Thin-Walled Struct. 50 (2012) 24–36.
uation for RHS temper T4 aluminium alloy beams, Composites B 115 (2017) [50] Q.L. Zhang, Y. Wu, Numerical and experimental study on flexural-torsional
124–137. buckling of aluminium beams, Struct. Eng. Int. 4 (2006) 312–318.
[36] P. Castaldo, E. Nastri, V. Piluso, Ultimate behaviour of RHS temper T6 aluminium [51] X. Guo, Z. Xiong, Z. Shen, Flexural–torsional buckling behavior of aluminum
alloy beams subjected to non-uniform bending: Parametric analysis, Thin-Walled alloy beams, Front. Struct. Civil Eng. 9 (2015) 163–175.
Struct. 115 (2017) 129–141. [52] T. Tryland, M. Langseth, O. Hopperstad, Nonperfect aluminium beams subjected
[37] E. Nastri, V. Piluso, The influence of strain-hardening on the ultimate behaviour to concentrated loading, J. Struct. Eng. 125 (8) (1999) 900–909.
of aluminium RHS-beams under moment gradient, Thin-Walled Struct. 157 [53] Y.Q. Wang, Z.X. Wang, F.X. Yin, L. Yang, Y.J. Shi, J. Yin, Experimental study
(2020) 107091. and finite element analysis on the local buckling behavior of aluminium alloy
[38] V. Piluso, A. Pisapia, E. Nastri, R. Montuori, Ultimate resistance and rotation ca- beams under concentrated loads, Thin-Walled Struct. 105 (2016) 44–56.
pacity of low yielding high hardening aluminium alloy beams under non-uniform [54] UNI.-E.N. 573-3, Alluminio e leghe di alluminio - composizione chimica e forma
bending, Thin-Walled Struct. 135 (2019) 123–136. dei prodotti semilavorati - parte 3: Composizione chimica e forma dei prodotti,
[39] R. Montuori, E. Nastri, V. Piluso, A. Pisapia, Ultimate behaviour of high-yielding 2019.
low-hardening aluminium alloy I-beams, Thin-Walled Struct. 146 (2020) 106463. [55] UNI-EN-I.S.O. 6892-1, Metallic Materials – Tensile Testing – Part 1: Method of
[40] R. Montuori, E. Nastri, V. Piluso, A. Pisapia, The influence of the material Test At Room Temperature, European Standard, 2020.
properties on the ultimate behaviour of aluminium H-shaped beams, Open [56] A.P.C. Duarte, N. Silvestre, A new slenderness-based approach for the web
Constru. Build. Technol. J. 15 (2021) 176–188. crippling design of plain channel steel beams, Int. J. Steel Struct. 3 (2013)
[41] E. Nastri, V. Piluso, A. Pisapia, Numerical application of effective thickness 421–434.
approach to box aluminium sections, J. Composit. Sci. 5 (11) (2021) 291. [57] M. Bock, S. Gupta, M.F. Hassanein, Y. Sheng, A slenderness-based method for
[42] CEN/TC250/SC9 N888 - prE.N. 1999-1-1, Eurocode 9: Design of aluminium web crippling design of aluminium tubular sections, J. Struct. Eng. 148 (12)
structures - part 1-1: General structural rules, in: CEN/TC250/SC9 Working Draft (2022) 04022205.
Document, 2018. [58] V.V. Nguyen, G.J. Hancock, C.H. Pham, New developments in the direct strength
[43] B. Li, Y. Wang, Y. Zhang, H. Yuan, X. Zhi, C.C. Baniotopoulos, Flexural behaviour method (DSM) for design of cold-formed steel sections under localised loading,
of 7404-T6 high-strength aluminium alloy SHS and RHS beams under moment J. Struct. Eng. 10 (3) (2017) 227–233.
gradient, Eng. Struct. 259 (2022) 114138. [59] G. Gerard, S. Wildhorn, A Study of Poisson’s Ratio in the Yield Region, NACA
[44] B. Li, Y. Wang, Z. Wang, X. Zhi, Y. Zhang, Y. Ouyang, 7A04-T6 high-strength Technical Note 2561, National Advisory Committee on Aeronautics, Washington,
aluminium alloy SHS and RHS beams under pure bending—Testing, modelling D.C, 1952.
and design recommendations, Thin-Walled Struct. 177 (2022) 109400. [60] B.S. 5950-5, Structural Use of Steelwork in Building - Formed Thin Gauge
[45] L. Huynh, C.H. Pham, K. Rasmussen, Distortional buckling behaviour and Sections Part 5: Code of Practice for Design of Cold Formed Thin Gauge Sections,
strength of cold-rolled aluminium alloy beams, J. Construct. Steel Res. 187 Steel Construction Institute, Silwood Park, Ascot, Berkshire SL5 7QN, 1998.
(2021) 106980. [61] R. Rebiano, N. Silvestre, D. Camotim, GBT formulation to analyze the buckling
[46] N.H. Pham, C.H. Pham, K. Rasmussen, Global buckling capacity of cold-rolled behavior of thin-walled members subjected to non-uniform bending, Int. J. Struct.
aluminium alloy channel section beams, J. Construct. Steel Res. 179 (2021) Stab. Dyn. 7 (1) (2007) 23–54.
106521.
[47] M. Rouholamin, S. Gunalan, K. Poologanathan, H. Karampour, Experimental
study of roll-formed aluminium lipped channel beams in shear, Thin-Walled
Struct. 153 (2020) 106687.

26