Revision of EN 1993-1-1: Design rules for member buckling

Citation for published version (APA):

Bureau, A., Snijder, H. H., Knobloch, M., Kuhlmann, U., Gardner, L., & Simoes da Silva, L. (2022). Revision of
EN 1993-1-1: Design rules for member buckling. In The International Colloquium on Stability and Ductility of
Steel Structures (pp. 385-393). (CE/Papers; Vol. 5, No. 4). Ernst & Sohn. https://doi.org/10.1002/cepa.1769

DOI:
10.1002/cepa.1769

Document status and date:

Published: 01/01/2022

Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)

Please check the document version of this publication:

• A submitted manuscript is the version of the article upon submission and before peer-review. There can be
important differences between the submitted version and the official published version of record. People
interested in the research are advised to contact the author for the final version of the publication, or visit the
DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page
numbers.
Link to publication

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.

If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please
follow below link for the End User Agreement:
www.tue.nl/taverne

Take down policy

If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.

Download date: 27. avr.. 2024

 ORIGINAL ARTICLE

Revision of EN 1993-1-1 – Design Rules for Member

Buckling
Alain Bureau1, Bert Snijder, Markus Knobloch, Ulrike Kuhlmann, Leroy Gardner, Andreas Taras, Luis da
Silva

Correspondence Abstract

Alain Bureau In the framework of the revision of Eurocode 3 Part 1-1, several amendments have been proposed
Centre Technique Industriel de la Construction and accepted in order to improve the rules for the resistance to member buckling. For clarification,
Métallique a flow-chart connecting the global analysis (1st or 2nd order), the imperfections and the type of
Espace Technologique – Immeuble Apollo
verification has been implemented for ease of use. Since the publication of the standard in 2005,
L’orme des merisiers
F-91193 Saint Aubin many research projects have been carried out across Europe on this topic and their results have
Email: abureau@cticm.com contributed to provide appropriate answers to problems identified in practice. Therefore, the
revised code provides new design rules for stability. Important works of calibration have been
performed in these different projects to derive appropriate values of the partial factor on the
resistance side.

For example, a new formulation for lateral torsional buckling has been introduced for the
calculation of the reduction factor. The consequence is a reduction of the discrepancy between
the results obtained by these new methods and those from experimental or numerical tests. In
order to extend the scope of Eurocode 3 Part 1-1, additional methods have been implemented in
an annex to cover the stability of members with mono-symmetric cross-section under
compression axial force, biaxial bending, with or without torsion. The format of the resistance
criteria remains similar to the format of the current interaction formulae so that the designers can
easily identify the evolution of the rules. This paper presents in a systematic way the new
implementations in prEN 1993-1-1.

Keywords

Eurocode, global analysis, classification of cross-sections, cross-section resistance, member

buckling, lateral torsional buckling, code calibration

1 Introduction
The revision of the Eurocodes effectively started in 2015 with the
creation of project teams. EN 1993-1-1 is the main part of
Eurocode 3 since it includes the general rules for steel structures,
and more specifically rules for the strutural analysis and the
verification of its members. With EN 1993-1-8 (design of joints), it
has been revised in the first phase of the project of revision of the
Eurocodes, under the mandate M/515 of the European Commission.
The document is now stabilized and is now at the stage of formal
voting.
This paper gives an overview of the main changes in FprEN 1993-1-
1 [1]. In a first step, it is explained that the different methods for the
structural analysis have been clarified. Then, regarding the
classification and the resistance of the cross-sections, new rules
have been added to enlarge the field of application (elliptical
sections, open sections under torsion, elasto-plastic resistance of
semi-compact sections). Thanks to several research projects in the
last decades, the rules for member buckling (flexural buckling,
lateral torsional buckling, interactions) have also been improved and
cover a wider range of applications. These evolutions have been
adopted with respect to the level of safety required by EN 1990.
2 Global analysis
2.1 General
For better applicability, the new section 7 ‘Structural analysis’ in the
formal vote draft FprEN 1993-1-1 [1] has been restructured
compared with the current section 5 in EN 1993-1-1 [2] on
structural analysis, while largely retaining the same content. This
new section links the calculation of internal forces with the
verification for the ultimate limit state. Essentially, it is about the
conditions under which determination of internal forces considering
second order effects with the assumption of imperfections is
required and whether a stability member check may be performed.
Similar guidance is currently given in EN 1993-1-1 cl. 5.2.2.

© 2022 Ernst & Sohn GmbH. · ce/papers 5 (2022), No. 4

https://doi.org/10.1002/cepa.1769 wileyonlinelibrary.com/journal/cepa 385
386 |
However, these rules often led to questions and misunderstandings of considering second order effects. In the following, the individual
in practice and are actually interpreted very differently based on methods of analysis are dealt with in detail.
traditional approaches in the various countries. Within the revision
these different views and traditions were harmonised and put
together as a result in a flow chart, see Figure 1. This constitutes an
important tool for ease of use by clearly connecting the type of
structural analysis and the choice of imperfections with the method
of verification at the ultimate limit state. Depending on the extent
and type of consideration of structural deformations according to
second order theory, three approaches are distinguished:

− complete calculation of internal forces according to second

order theory considering global and local imperfections
(Method M5),
− partially considering second order effects and imperfections in
the global analysis and partially through individual stability
checks of members (Method M2 to M4),
− for simple systems by individual stability checks of equivalent
members (EM).

Further details of these procedures are also given in [3], [4].

2.2 Criteria for considering second order effects

The criteria for considering second order effects in the global

analysis, the rules for methods of structural analysis as a function of
the verification methods at the ultimate limit state and the rules for
imperfections are key components of the revised section on
structural analysis. The formal vote draft FprEN 1993-1-1
distinguishes between non-sway (ns) and sway (sw) buckling modes
and therefore contains two criteria for the decision whether second
order effects should be considered or not.
The first criterion in Eq. (1) looks at non-sway buckling modes.
Hence, second order effects due to in-plane or out-of-plane member
buckling may be neglected for the global analysis if the criterion is
satisfied. The k0 value is a Nationally Determined Parameter (NDP)
that may be specified by the national standardization bodies of the
different countries. The recommended value for k0 is 25 and results
from the fact that the stability behaviour of the individual member
must be considered from an influence of 1/25 or 4 % corresponding
to a relative slenderness of 0,2.
Figure 1 Methods of structural analysis applicable for the ultimate limit state
design
For the simplest case, where a first order analysis suffices (Method
M0), both criteria for the member (non-sway) buckling mode and for
the global (sway) buckling mode Eqs. (1) and (2), are fulfilled. The
same applies for the case of method M1, where an additional lateral
torsional buckling (LTB) verification is necessary.
Method M2 refers to structural systems, where the sway buckling
mode is prevented by stabilizing elements such as concrete walls or
bracings. Hence, only member (non-sway) buckling modes have to
be considered when verifying the individual members according to
cl. 8.3 of FprEN 1993-1-1. Nevertheless, a sway imperfection should
be considered in the structural analysis, in order to consider
stabilizing forces for the stiffening elements.

𝐹𝑐𝑟,𝑛𝑠
Methods M3 and M4 belong to design situations, where both
𝛼𝑐𝑟,𝑛𝑠 = ≥ 𝑘0 (1) criteria according to Eqs. (1) and (2) fail. Consequently, a ‘mixed’
𝐹𝐸𝑑
procedure according to EN 1993-1-1 cl. 5.2.2(3), case b), is applied.
The second criterion, see Eq. (2), was established to check whether The two methods differ according to the extent of the second order
first order analysis may be used to determine the in-plane sway analysis. The in-plane analysis of method M3 considers only the
bending moments. Compliance with this criterion implicitly assumes global second order effects and imperfections, but may assume the
that the increase in the internal forces and moments due to sway internal forces and moments between the member ends according
second order effects is no more than 10 % of the original internal to first order theory. However, Method M4 covers both – global and
forces and moments according to first order theory. member second order effects and imperfections for the in-plane
analysis. This latter procedure allows the application of the member
𝛼𝑐𝑟,𝑠𝑤 =
𝐹𝑐𝑟,𝑠𝑤
≥ 10 (2) check to be restricted to Eq. (8.89) in FprEN 1993-1-1 representing
𝐹𝐸𝑑
the out-of-plane member buckling verification, usually on the
2.3 Methods of analysis ‘weak’ axis of the cross-section including lateral torsional buckling.

The most complex method M5 considers all in-plane and out-of-

Six different methods of analysis have been defined. The list of
plane second order effects, including torsional effects, as well as
methods of analysis, see Figure 1, starts with the simplest case
(sway) imperfections and in-plane and out-of-plane local bow
where a first order analysis suffices (Method M0) and ends up with
imperfections. As a consequence, only a cross-sectional check
a calculation based on second order theory considering global and
according to cl. 8.2 of FprEN 1993-1-1 considering the second order
local imperfections in- and out-of-plane (Method M5). The different
internal forces and moments is necessary for the verification.
methods of analysis were not newly developed. These methods are
already common practice within the current version of EN 1993-1-
The ‘equivalent member’ (EM) method is an independent
1 [2], but the conditions and way of application of the various
alternative approach compared to methods M0 to M5. This
methods currently are not clearly distinguished. Now the choice of
traditional method also covers the global sway buckling mode by
the methods is clearly built on certain decisions concerning the need

applying an effective buckling length for the stability member check
using only first order internal forces and moments. Imperfections
are not explicitly considered in the structural analysis, but indirectly
by the reduction factors of the member check. This simple approach
suffers from the lack of knowledge on the “real” structural
behaviour, e.g. at the joints or at adjacent members. For that reason,
safe sided simplified assumptions have to be considered.
3 Classification and resistance of cross-sections
The revised FprEN 1993-1-1 has improved and extended provisions
for the classification and design of steel cross-sections, a summary
of which is provided in the following sub-sections.
3.1 Classification limits and effective section properties
Classification (or slenderness) limits are used to assign cross-
sections to discrete behavioural classes (Class 1, 2, 3 or 4) to reflect
their susceptibility to local buckling. Some updates to these limits –
primarily for internal plate elements and circular hollow sections -
have been made in the revised code. For internal plate elements, the
new (slightly stricter) provisions remove an existing inconsistency
between the Class 3 limits in EN 1993-1-1 and the limit at which a
local buckling reduction factor below unity would arise from the
effective width formulae given in EN 1993-1-5. Similar changes to
the Class 1 and 2 limits in compression, and hence the limits under
combined loading, have also been made. For CHS, a substantial
improvement has been made to the Class 3 slenderness limit in
bending, which has increased from D/t2 = 90 (where D and t are the
outer diameter and thickness of the CHS, respectively) to 140. This
change follows a re-evaluation of existing test data [6], reflects the
treatment in other international standards and moves many
commonly used CHS profiles (particularly in the higher strength
steel grades) out of the slender (Class 4) range, facilitating ease of
use. Ease of use has been further facilitated by the introduction of
simple effective area and section modulus formulae for determining
the reduced resistance of Class 4 CHS in compression and bending
respectively due to local buckling.
3.2 Semi-compact (Class 3) cross-sections
The rules for semi-compact (Class 3) cross-sections have been
enhanced to remove the artificial step in resistance predictions that
arise in the transition from fully plastic (Class 1 and 2 cross-sections)
to fully elastic (Class 3 and 4 cross-sections) design treatments. In
the existing design rules, a drop in bending moment resistance from
the plastic moment capacity Mpl to the elastic bending moment
resistance Mel would occur at the Class 2 to 3 boundary. This drop is
around 10-15% for I-sections in major axis bending and about 25%
for hollow sections, and could be larger under combined loading. In
the new provisions [7], [8], illustrated in the context of CHS in
Figure 2, the resistance of Class 3 sections linearly reduces with
increasing slenderness from Mpl at the Class 2 slenderness limit to
Mel at the Class 3 slenderness limit, providing increased design
efficiency and a better reflection of observed physical behaviour.
The new rules are implemented through the introduction of an
elastic-plastic section modulus Wep, set out in Annex B of the revised
code.
| 387
Revised
Figure 2 Illustration of new slenderness limit (D/t2=140) and linearly reducing
resistance of semi-compact (Class 3) CHS in bending [8].
3.3 Elliptical hollow sections
Reflecting the recent addition of elliptical hollow sections (EHS) to
the family of tubular steel products available to structural
engineers, a set of new rules has been introduced in FprEN 1993-1-
1 for their design. The classification of EHS is based on the use of the
slenderness limits for CHS together with the definition of an
equivalent diameter De, which is a function of both the aspect ratio
(h/b) of the cross-section and the loading to which the section is
subjected. Formulae are also provided for the calculation of shear
resistance (including the proportions beyond which shear buckling
should be considered), resistance under combined bending and
shear and resistance under combined axial compression and
bending moments [9], [10]. As for CHS, the design of Class 4 EHS has
also been facilitated by the provision of effective area and effective
section modulus formulae to account for the loss of effectiveness
due to local buckling.
3.4 Other revisions
Other key revisions pertaining to the design of cross-sections are as
follows. Simplified rules for the assessment of the resistance of
sections to concentrated transverse forces have been introduced. In
the net section check for tension members, a new factor k has been
introduced instead of the constant of 0.9. The factor k depends on
how the bolt holes have been cut and whether the member will be
subjected to fatigue loading. New formulae have been introduced to
provide shear areas to calculate shear resistances for a wider range
of cross-sections. Finally, the provisions for torsion and its
interaction with other internal forces, have been improved,
following the research presented in [11], with the bi-moment now
explicitly introduced, simplifying the design of sections where
warping torsion is dominant e.g. I-sections.
4 Buckling of members in compression
Though members in pure compression do not occur much in
practice, this is the basic buckling case and therefore the first
stability case treated in FprEN 1993-1-1 [1]. Compared to the
current code EN 1993-1-1 [2], the structure of the clause on uniform
members in compression (cl. 8.3.1) has been changed and is clearer
now. First the buckling resistance is treated in general terms, then
the slenderness of compression members is defined and
subsequently buckling reduction factors for flexural buckling, and
for torsional and torsional-flexural buckling are provided.
4.1 Buckling resistance
In cl. 8.3.1.1 on the buckling resistance, there are no major changes
compared to the content of EN 1993-1-1: the design value of the
388 |
compressive force shall be smaller than the design buckling
resistance, which is the buckling reduction factor 𝜒 times the
characteristic value 𝑁𝑅𝑘 of the cross-sectional resistance divided by Table 1 Flexural buckling curve selection table (Table 8.3 in FprEN 1993-1-1)
the partial factor 𝛾𝑀1 . Contrary to the current situation in EN 1993-
1-1, the relevant characteristic cross-sectional resistances to
compression are now listed in the clause for the resistance of cross-
sections (i.e. cl. 8.2).

4.2 Slenderness of compression members

Slenderness definitions for all relevant buckling cases are gathered

in cl. 8.3.1.2: flexural buckling about the (strong) y-y axis, flexural
buckling about the (weak) z-z axis, torsional buckling and flexural-
torsional buckling. Content wise, the definitions are the same as
currently in EN 1993-1-1. The relative slenderness 𝜆̅ is firstly given
in its general format as the square root of the ratio between the
characteristic value 𝑁𝑅𝑘 of the cross-sectional resistance and the
elastic critical force 𝑁𝑐𝑟 . The latter is then specified for all buckling
cases mentioned, leading to specific relative slendernesses for these
buckling cases: 𝜆̅𝑦 for flexural buckling about the y-y axis, 𝜆̅𝑧 for
flexural buckling about the z-z axis, 𝜆̅ 𝑇 for torsional buckling and 𝜆̅ 𝑇𝐹
for torsional-flexural buckling. The relevant elastic critical forces for
these buckling cases, as in the current code, are not specified in the
code itself. However, a Technical Report CEN/TR 1993-1-103 is
foreseen that will provide elastic critical buckling solutions.
Secondly, the relative slenderness is, only for flexural buckling, given
based on the buckling length. These rules had a more prominent
position in the current EN 1993-1-1.

4.3 Buckling reduction factor for flexural buckling

The buckling curves for flexural buckling and their representation in

formulas and figures are given in cl. 8.3.1.3 and have not changed:
the ECCS buckling curves a0, a, b, c and d are still in place. However,
the buckling curve selection table has been modified substantially,
on the one hand to include innovations in the steel sector, on the
other hand to achieve a uniform level of safety over all flexural
buckling cases specified in the buckling curve selection table, shown
here as Table 1. The innovations introduced in the buckling curve
selection table concern elliptical hollow sections, heavy sections
with flange thickness greater than 100 mm, very heavy thick-walled
rolled angles and high-strength steel.
Elliptical hollow sections (EHS) are architecturally appealing and
now also available on the market besides circular, square and
rectangular hollow sections (CHS, SHS and RHS). It was shown that
buckling curve c, applicable for CHS, SHS and RHS, also applies to
EHS [5]. Therefore, in the row for hollow sections in Table 1, the
drawing for EHS has been added.
In the current code EN 1993-1-1, the buckling curve selection table
does not give buckling curves for heavy HD- and HL-sections with
depth-to-width ratio ℎ/𝑏 > 1.2 and flange thicknesses 𝑡𝑓 > 100 mm.
However, these sections are available on the market nowadays and
are also widely used for amongst others high-rise buildings. By
carrying out numerical research using conventional imperfections
and residual stress patterns based on measurements [5] [12] [13]
[14], it could be shown that the buckling curves currently used for
ℎ/𝑏 > 1.2 and flange thicknesses 40 < 𝑡𝑓 ≤ 100 mm also apply to
heavy HD- and HL- sections with ℎ/𝑏 > 1.2 and 𝑡𝑓 > 100 mm. Since
buckling about the weak z-z axis is more affected by residual
stresses than buckling about the strong y-y axis, for high-strength
steels with ℎ/𝑏 > 1.2 and 𝑡𝑓 > 40 mm, the buckling curve had to be
changed from a to b, see Table 1.
Heavy thick-walled rolled angles are currently available on the
market, which are in competition with welded angles. However, the
current buckling curve selection table in EN 1993-1-1 does not
distinguish between rolled and welded angles, though the drawing
in the buckling curve selection table suggests that rolled angles are
meant. In the current buckling curve selection table of EN 1993-1-
1, buckling curve b is specified for all (rolled) angles regardless the
steel grade. However, high-strength steels suffer less from residual
stresses than steel grades with moderate strength. Also, welded
sections show higher residual stresses needing a lower buckling
curve than rolled sections. By numerical research using geometrical
imperfections and residual stress patterns based on measurements
[5] [15] [16] [17] [18], the buckling curves as shown in Table 1 for
angles could be justified. The buckling curve selection table now
distinguishes clearly between rolled and welded angles and
between steel grades with moderate strength and high-strength
steels.
High-strength steel grades up to and including S700 currently have
their own code part: EN 1993-1-12 [19]. However, with high-
strength steels being more frequently used, it was decided to
include the design rules of the current EN 1993-1-12 in the
appropriate code part. Therefore, the buckling curve selection table
now has been extended up to and including S700, see table 1. The
current buckling curve selection table in EN 1993-1-1 specifies the
same buckling curves for rolled I- or H-sections for strong y-y axis
buckling and weak z-z axis buckling for the highest steel grade S460.
However, residual stresses are more detrimental for weak z-z axis
buckling than for strong y-y axis buckling. Numerical research [5]

[20] [21] [22] shows that indeed weak z-z axis buckling requires a
lower buckling curve, which now is reflected in Table 1. There is
however one exception for rolled I- or H-sections, namely for ℎ/𝑏 ≤
1.2 and 𝑡𝑓 > 100 mm where buckling curve c is specified for both
strong y-y and weak z-z axis buckling, which still is under debate in
the code committee. Also currently under debate are the buckling
curves for welded high-strength I-sections. This may lead to
amendments to Table 1 in the near future.
The modified buckling curve selection table of FprEN 1993-1-1
(Table 1), compared to the one in EN 1993-1-1, results in more
consistent safety levels over all buckling cases considered. Several
studies, amongst others [44], show that the buckling curves
proposed in Table 1 can be used with a partial factor on the
resistance side being 𝛾𝑀1 = 1.0.
4.4 Buckling reduction factors for torsional and torsional-flexural
buckling
The current design code EN 1993-1-1 only specifies that for
torsional and torsional-flexural buckling, the buckling curve for
flexural buckling about the weak z-z axis may be used. In cl. 8.1.3.4
of FprEN 1993-1-1 (see [23] [24] [25] for background information)
it is first stated that for members with open cross-sections, the
torsional or torsional-flexural buckling resistance may be governing
over the flexural buckling resistance. In a note it is clarified that this
may be the case for: doubly-symmetric cross-sections having
intermediate lateral restraints and cruciform and asymmetric cross-
sections, e.g. channel-, l-, T-, and I- or H-sections with single
symmetry. Then it is stated that the torsional and torsional-flexural
buckling coefficients 𝜒𝑇 and 𝜒𝑇𝐹 may be determined using the
flexural buckling curves specified in Table 1 for weak axis z-z
buckling. The appropriate relative slenderness, 𝜆̅ 𝑇 for torsional
buckling and 𝜆̅ 𝑇𝐹 for torsional-flexural buckling, needs to be used.
Figure 3 Geometrical definitions for torsional-flexural buckling of laterally braced
doubly-symmetric sections
Subsequently, for doubly-symmetric I- and H-sections in
compression with continuous or discrete intermediate lateral
restraints (Figure 3) a specific and more accurate formula is
provided to obtain the torsional-flexural buckling coefficient 𝜒𝑇𝐹 .
This formula can also be used to obtain the torsional buckling
coefficient 𝜒𝑇 but then of course the torsional relative slenderness
𝜆̅ 𝑇 should be used instead of the torsional-flexural relative
slenderness 𝜆̅ 𝑇𝐹 . These relative slendernesses should be determined
| 389
using the weak z-z axis buckling length 𝐿𝑐𝑟 of the unrestrained
flange. A weak axis flexural buckling check between intermediate
discrete supports should be carried out as well.
5 Lateral torsional buckling of members in bending
5.1 General
EN 1993-1-1 currently provides two different sets of LT-buckling
curves: one set covers “general sections” without pronounced
torsional properties, while the other set covers “rolled or welded
sections” and accounts for their distinct torsional rigidity. While the
first set has a plateau-length corresponding to merely 𝜆̅𝐿𝑇 = 0.2 –
which is the value also applicable in the case of flexural buckling –,
the second set defines the plateau length at 𝜆̅𝐿𝑇 = 0,4 [2]. This second
set of LT-buckling curves was based on numerical simulations [26]
as well as on beam buckling tests from different sources (e.g. [27]
[28]). In recent years and during the development of the second
generation of Eurocode 3 [1], the accuracy and degree of
conservatism of this set of design rules has been questioned by
various researchers. It was generally found that the so-called
“general case” rules of EN 1993-1-1, which cover doubly- as well as
mono-symmetric sections, lead to quite conservative results for
many, particularly, stockier sections. The “special case”, which is
supposed to cover doubly-symmetric rolled sections in particular, is
also inaccurate for many sections and somewhat unconservative for
others [29] [30]. The degree of unconservatism is rather
pronounced for certain load cases, e.g. with transverse loads. For
these reasons, it was decided to implement newer, more precise and
mechanically consistent lateral buckling design rules in section 8 of
FprEN 1993-1-1, which were first described and published in [31].
5.2 New buckling reduction factors for doubly-symmetric sections
In the past two decades, the knowledge about the specific lateral-
torsional (LT)-buckling behaviour of imperfect members has
significantly increased as a result of numerous numerical (GMNIA)
simulations using the FE-Method. In addition to this numerical work,
progress has also been achieved on the analytical side, which led to
amended and more consistent formulations of “Ayrton-Perry” type
equations for LT-buckling. In [29], formulations for lateral torsional
buckling of I- and H-sections were developed that were based on the
concepts originally applied by Ayrton and Perry to the design
against flexural buckling, i.e. a calibration of a design formulation
based on elastic second-order theory and a linear cross-sectional
failure criterion through a generalized imperfection definition. The
calibration and validation were finally conducted against a very
large series of numerical results and for various load cases. Thereby,
a formulation for LTB was analytically obtained that resembles in
most parts the familiar column buckling formula (which is itself an
“Ayrton-Perry” formulation), but includes an additional term
(𝜆̅𝐿𝑇 /𝜆̅𝑧 )2 which accounts for and is heavily dependent on the
torsional stiffness of the considered cross-section in relation to its
weak-axis bending stiffness (equations (3) to (5)). In addition, the
derived formulation contains an imperfection term  = 𝛼𝐿𝑇 ·
(𝜆̅𝑧 − 0.2), which was used for calibration in an analogous way as for
the familiar column buckling curves.
1
𝜒𝐿𝑇 = ≤ 1.0 (3)
̅𝐿𝑇 2
𝛷+√𝛷2 −𝜆
1 2
with: 𝛷 = · (1 + 𝜂∗ + 𝜆̅𝐿𝑇 ) (4)
2
̅𝐿𝑇 2 ̅𝐿𝑇 2
𝜂∗ = 𝜂 ·
𝜆
2 = 𝛼𝐿𝑇 · (𝜆̅𝑍 − 0.2) ·
𝜆
2 (5)
̅
𝜆𝑍 ̅
𝜆𝑍
390 |
𝛼𝐿𝑇 = 𝑐𝑎𝑙𝑖𝑏𝑟𝑎𝑡𝑖𝑜𝑛 𝑓𝑎𝑐𝑡𝑜𝑟
In the calibration of the formula, the different behaviour of I- and H-
sections was captured both by the consideration of the mentioned
factor (𝜆̅𝐿𝑇 /𝜆̅𝑧 )2 and by an additional factor (Wz/Wy)0.5, which allowed
for a much more accurate calibration of the LT factor by means of
least-square analyses to match the GMNIA buckling curves. It
should also be noted that, through the introduction of this factor,
the grouping of the corresponding section shapes no longer follows
the previous criterion h/b > or < 2.0, but that the section types are
now classified in the same way as for flexural buckling, with a
classification limit at h/b = 1.2. This underlines the consistency of the
approach: all sections usually considered to feature the same
amplitude of residual stresses are classified into the same group
independently of the given buckling mode.
The accuracy of the new concept in terms of buckling reduction
factors  is illustrated in Figure 4 . Up to a slenderness of 𝜆̅𝐿𝑇 = 1.2,
the correspondence of the design curves and the GMNIA-curves is
very high. The rather modest deviations on the conservative side –
all occurring only at higher slenderness - can be tolerated in light of
the fact that I-section beams of lengths that result in values of 𝜆̅𝐿𝑇 >
1.2 when loaded in uniform bending are rarely found in practice.
Therefore, the new approach has proven that a consistent
derivation allows to establish very accurate design formulae for LT-
buckling on the basis of the Ayrton-Perry format. However, the pre-
condition for this is that the imperfection coefficients are properly
calibrated.
1.0
GMNIA
E ul e
prEN.
r
0.8
0.6
 =M /M [-]
pl
HEM320
HEB400
R
0.4 IPE240
LT
HEAA900
0.2
M M
0.0
0.0 0.5 1.0 1.5 2.0
 LT
Figure 4 Comparison of GMNIA resistances against lateral-torsional buckling
(LTB) with the new design rules for LTB of doubly-symmetric I- and H-section in
FprEN1993-1-1
It should finally be noted that the new design formulation leads to a
unique LTB design curve for each section; this needs to be kept in
mind primarily if it is desired to use tabulated values for the buckling
reduction factors LT.
The mentioned formulation was further expanded to account for
the effect of various bending moment diagrams. This is done
through the use of a factor “fM”, which is given for a large set of
practically relevant bending moment diagrams in FprEN 1993-1-1
and may be seen as an “over-strength” factor for moment diagrams
that diverge from the uniform one and mainly accounts for the
following two mechanical effects:
i. Firstly, the factor accounts for the fact that the point with
the maximum in-plane bending moment may significantly
differ from the point in the beam that is most affected by
lateral-torsional deformations caused by buckling. Since
the design equations only make direct reference to the
point with the maximum bending moment, this causes an
increase of the applicable value of LT.
ii. In addition, bending moment diagrams with steep
gradients and thus sharp decreases of the moment around
the point most affected by LT-buckling provide a degree of
support (through a smaller spread of plasticity) that again
manifests itself through a higher value of the applicable
LT.
Further details on the application of the new design rules for LTB
and their expansion to various load cases may be found in [4] and [5].
6 Interaction formulae for member buckling
The application of the simplified rules for stability verification of cl.
6.6.3 of EN 1993-1-1:2005 [2] is limited to structural steel members
with doubly-symmetric cross-sections under biaxial bending and
axial compression. FprEN 1993-1-1 [1] extends the application
limits and introduces a normative Annex C for this purpose. Annex
C is based on rules of the Austrian National Annex [32] to EN 1993-
1-1:2005 and EN 1993-6 [33] for the design of crane runway
girders. It provides supplementary rules to the member stability
verification rules, which are part of cl. 8.3.3 of FprEN 1993-1-1 [1].
Annex C.1 contains additional rules for members with mono-
symmetric I-, H- and welded box sections, which have flanges of
different size and are symmetric about the z-z axis. Annex C.2 deals
with members under torsional loading.
6.1 Members with mono-symmetric cross-section – background
and application of Annex C.1
Annex C.1 gives specific rules for the stability member verification
of mono-symmetric cross-sections. These rules extend the design
formulae of cl. 8.3.3 of FprEN 1993-1-1 [1], which essentially
represent the rules from Annex B of the current EN 1993-1-1:2005
[2]. The rules in Annex C.1 are based on studies by Kaim [34], which
included numerical simulation studies and a comparative study with
experimental results. These studies considered mono-symmetric I-
shaped cross-sections, in which either one flange was reduced in
width, or one flange was doubled in thickness. Moreover, mono-
symmetric RHS-sections and T-sections were included in the
numerical simulation study, which considered common approaches
for imperfections, i.e., a bow imperfection of L/1000 and typical
residual stress patterns.
The study revealed characteristic behaviours that occur in mono-
symmetric cross-sections in addition to those of doubly-symmetric
sections. In the case of buckling failure due to deformations
predominately in the x-z-plane (buckling about the y-y axis), the
interaction coefficients kyy for doubly-symmetric cross sections (see
cl. 8.3.3 of FprEN 1993-1-1) may lead to unconservative design
results if positive bending moments cause compressive stresses in
the smaller flange of the cross-section. Modified interaction factors
kyy are therefore introduced in Annex C.1 to cover this effect and to
obtain conservative design results. However, for negative bending
moments causing compressive stresses in the larger flange, the kyy
values of the doubly-symmetric cross sections are sufficiently
conservative; however, a distinction between positive and negative
bending moments was omitted in favour of unambiguous, easy-to-
apply rules.
For members with open cross-sections that fail in buckling mainly
due to out-of-plane deformations and torsion, the interaction
formulae had to be modified to account for the fact that the mode of

failure in compression is torsional-flexural buckling and not flexural
buckling about the z-z axis. Thus, the buckling resistance considers
the torsional-flexural buckling reduction factor χTF. When the axial
compressive force acts in conjunction with a positive bending
moment (causing compressive stresses in the smaller flange), the
interaction coefficients kzy of cl 8.3.3 of FprEN 1993-1-1 describe
the buckling strength of the beam-column quite accurately.
However, if the axial compressive force acts in conjunction with a
negative bending moment, the structural member behaviour of the
mono-symmetric cross-section changes significantly because the
bending moment relieves the smaller flange. In this case, the
interaction formula for LTB is still valid, but the term related to the
axial force in this case must use the weak-axis flexural buckling
coefficient χz in the interaction formula of FprEN 1993-1-1, 8.3.3. In
addition, for these cases with combined axial compression and
negative bending moment, it must be verified separately that the
axial compressive resistance to torsional-flexural buckling is
sufficient to withstand the acting compressive force.
Finally, the rules in Annex C.1 take into account that the equivalent
uniform moment factors Cm used in the interaction formulae of cl,
8.3.3 of FprEN 1993-1-1 to account for the beneficial effect of non-
uniform bending moment distributions, are no longer directly
applicable for beam-columns with mono-symmetric cross-sections.
The difficulty stems primarily from the fact that the direction of
bending (causing compression in either the larger or smaller flange)
is no longer irrelevant to the application of Cm factors. Hence, the
standard values of Cm can no longer be used with certainty,
particularly for bending moment diagrams that change sign. More
accurate Cm coefficients for such cases could be derived for mono-
symmetric sections, but they would heavily depend on the degree of
asymmetry of the section and would thus be cumbersome and prone
to errors in application. For this reason, it was decided to include
simplified rules in FprEN 1993-1-1 where the Cm factors do not take
into account a change in the sign of the bending moment and have to
be calculated assuming an equivalent bending moment diagram with
the same sign over the entire member length.
6.2 Members in bending, axial compression and torsion – the new
Annex C.2
The rules of Annex C.2 for the stability verification of members in
bending, axial compression and torsion of FprEN 1993-1-1 [1] are
based on the alternative rules for the lateral-torsional buckling of
crane runway girders in biaxial bending and torsion of Annex A of
EN 1993-6 [33] as well as novel findings of the German national
research project AiF/IGF 19044 N [35]. The former is based on
theoretical and experimental investigations [36]. Supplementary
theoretical investigations were carried out in [37] and [38]. Specific
attention was paid to consistency with both the buckling checks of
cl. 8.3.3 of [1] for members without torsional loading and rules for
mono-symmetric cross-sections of Annex C.1.
Annex C.2 extends the member stability rules by the term of the
bimoment according to [36]. The interaction factors kij according to
Tables 8.7 and 8.8 for doubly-symmetric and Table C.1 for mono-
symmetric cross-sections should be used in the same way as the
equivalent uniform moment factors of Table 8.9 for non-constant
moment distributions. For class 1 and class 2 cross-sections, the
bending moment resistances Mpl,y and Mpl,z as well as the bimoment
resistance Bpl according to plastic theory may be used for design. For
I-shaped cross-sections with unequal flanges, the bending moment
resistance Mpl,z and the bimoment resistance Bpl do not arise under
fully plastic stress distribution. These two resistances according to
plastic theory result from a stress state with partial yielding [37]. For
| 391
class 3 cross-sections, the corresponding resistances according to
the elastic theory should be used. For the computation of the
resistances Mel,z and Bel, the mono-symmetric cross-section can be
reduced to a doubly-symmetric cross-section with the width of the
narrower flange. Alternatively, the rules for semi-compact cross-
sections of Annex B (section 3 of this paper) may be applied and
partially plastic stress states may be used for the moment
resistances. Members with Class 4 cross-sections are not covered
by Annex C.2.
The application of Annex C.2 is subject to further limits. The rules
should be applied to simply supported, I-section members, with
equal flanges or with unequal flanges provided that the ratio of the
moments of inertia about the z-z axis of the flanges is greater than
or equal to 0.2 and smaller than or equal to 5.0. These limits cover
the scope of the theoretical investigations carried out and prevent
the application of the rules to T-shaped cross-sections with only one
flange or mono-symmetric I-sections with one very narrow and one
significantly wider flange. The load-carrying behaviour of such
cross-sections can differ significantly from the behaviour of typical
I-shaped sections. The design value of the bimoment BEd is limited to
30 % of the bimoment resistance to ensure consistency with the
member buckling verification rules of cl. 8.3.3 of FprEN 1993-1-1
for doubly-symmetric cross-sections in biaxial bending and axial
compression and to cover sections with low torsional resistance, i.e.,
IPE- and equivalent mono- and doubly-symmetric cross-sections by
considering plastic capacities.
The influence of the bimoment on the structural stability behaviour
of a simple member can be neglected if the product of the amplifier
according to second order theory kα and the bimoment ratio
BEd/(BRk/γM1) does not exceed 0.07. In this case the assessment
method can lead to unconservative design results up to 5 % at the
level of the load amplifiers. Alternatively, a conservative rule can be
applied for small bi-moment values. If the bi-moment ratio
BEd/(BRk/γM1) is smaller than 0.035, its effect may also be neglected.
Sufficient resistance of the cross-sections must be verified in
addition to the structural stability check. For members with class 1
and 2 cross-sections, non-linear plastic interactions can be used, e.g.,
the partial internal forces method [39].
7 Calibration of partial factors
In the field of construction, it is essential that a design code leads to
a satisfactory level of safety. EN 1990 ([40] and [41]) defines the
general scheme for the safety management for all the Eurocodes
using partial factors on the action side (F) and on the resistance side
(M).
Many works have been carried out in the recent past to assess the
values of the partial factors M in EN 1993. The SAFEBRICTILE
project [42] aimed at a calibration of these factors for steel
structures covering modes driven by ductility, stability and fracture.
Firstly, in the frame of this project, detailed guidance was developed
on the application of the Annex D methodology for the safety
assessment of stability design rules and the corresponding partial
factors M [43]. Secondly, a large number of measured data for
material and geometrical properties were collected and gathered in
a European database. From these data, recommendations for the
statistical distributions were specified and used for the assessment
of the values of the partial factors. The statistical parameters (mean
value, coefficient of variation) have been included in an informative
annex (Annex E) of FprEN 1993-1-1.
The reliability assessment of the FprEN 1993-1-1 rules for prismatic
392 |
members in compression [44] has shown that the specification of
buckling curves in Table 1 for rolled I- or H-sections leads to
consistent levels of safety across the different buckling modes and
cross-section geometries, in line with a M1 = 1.0.
Concerning the reliability assessment of the lateral-torsional
buckling resistance of prismatic beams with hot-rolled I-shaped
cross sections, the conservative nature of the set of LT-buckling
curves for general sections without pronounced torsional
properties was confirmed, and the accuracy of the new buckling
reduction factors for doubly-symmetric sections was confirmed,
leading to M1 = 1.0 [45].
The reliability assessment of the FprEN 1993-1-1 rules for prismatic
members in bending and compression with hot-rolled I-shaped cross
sections [46] led to required partial factors that are in general lower
than the ones obtained for columns and beams. Since the
verification is performed as an interaction between the resistances
obtained for columns and beams, it somehow shows that the
interaction factors provide additional safety in relation to the
column and beam case. The results for the different buckling modes,
with and without, lateral-torsional buckling were found very similar,
showing that the interaction coefficients are calibrated with
sufficient accuracy for both cases. The results highlight lower M1
values than the ones calculated for columns and beams separately
and thus indicate that the interaction factors provide appropriate
safety.
Finally, the assessment of a global value of the partial factor
encompassing columns, beams and beam-columns obtained as an
average M1 from all sub- sets and for all buckling modes led to the
recommended value of M1 = 1.0. It is finally noted that this
conclusion is valid as long as the yield stress of steel is taken from
the relevant product standard.
8 Conclusion
The revision of EN 1993-1-1 is primarily based on the feedback from
the designers and also on the most recent knowledge derived from
many research projects carried out across Europe in the last
decades. FprEN 1993-1-1 will be a modern code based on the same
principles as the current code but adapted to the practice of the
steel construction industry and its evolution like the use of high-
strength steel grades. It provides a harmonized level of reliability in
accordance with the basis of design defined in EN 1990. In addition,
the code was drafted with the aim to improve the ease-of-use and to
harmonize the rules within the different parts of Eurocode 3 and
related standards.
References
[1] FprEN 1993-1-1 (2022) Eurocode 3: Design of steel structures -
Part 1-1: General rules and rules for buildings. Formal vote draft,
Document TC250/SC3/N 3504.
[2] EN 1993-1-1 (2005) Eurocode 3: Design of steel structures - Part
1-1: General rules and rules for buildings.
[3] Kuhlmann, Ulrike; Schmidt-Rasche, Christina; Jörg, Fabian;
Pourostad, Vahid; Spiegler, Jennifer; Euler, Mathias (2021):
Update on the revision of Eurocode 3. In: Steel Construction 14
(1), pp. 2–13. DOI: 10.1002/stco.202000048.
[4] Knobloch, M.; Bureau, A.; Kuhlmann, U.; da Silva, L. S.; Snijder,
H. H.; Taras, A., Bours, A.-L., Jorg, F. (2020): Structural member
stability verification in the new Part 1-1 of the second
generation of Eurocode 3 - Part 1: Evolution of Eurocodes,
background to partial factors, cross-section classification and
structural analysis. In: Steel Construction 13 (2), pp. 98–113.
[5] Knobloch, M., Bureau, A., Kuhlmann, U., da Silva, L.S., Snijder,
H.H., Taras, A., Bours, A.-L., Jorg, F. (2020): Structural member
stability verification in the new Part 1-1 of the second
generation of Eurocode 3 - Part 2: Member buckling design
rules and further innovations. Steel Construction, Vol. 13, No. 3,
pp. 208–222.
[6] Chan, T. M.; Gardner, L. (2008) Bending strength of hot-rolled
elliptical hollow sections. Journal of Constructional Steel
Research 64(9), 971-986.
[7] SEMI-COMP. (2009) Plastic Member Capacity of Semi-compact
Steel Sections — a More Economic Design, European Commission.
[8] Meng, X.; Gardner, L.; Sadowski, A. J.; Rotter, J. M. (2020)
Elasto-plastic behaviour and design of semi-compact circular
hollow sections. Thin-Walled Structures 148, 106486.
[9] Chan, T. M.; Gardner, L.; Law, K. H. (2010) Structural design of
elliptical hollow sections: a review. Proceedings of the Institution
of Civil Engineers - Structures and Buildings 163(6), 391-402.
[10] Meng, X.; Gardner, L. (2020) Simulation and design of semi-
compact elliptical hollow sections. Engineering Structures 202,
109807.
[11] Mirambell, E.; Bordallo, J.; Real, E. (2016) Torsion and its
interaction with other internal forces in EN 1993-1-1 – a new
approach. Steel Construction 9, 240–248.
[12] Snijder, H.H., Cajot, L.-G., Popa, N., Spoorenberg, R.C. (2014)
Buckling curves for heavy wide flange steel columns, Romanian
Journal of Technical Sciences – Applied Mechanics, 59, Nos. 1–2,
pp. 178–204.
[13] Spoorenberg, R.C., Snijder, H.H., Cajot, L.-G., May, M.S. (2013)
Experimental investigation on residual stresses in heavy wide
flange QST steel sections, Journal of Constructional Steel
Research, 89, pp. 63–74.
[14] Spoorenberg, R.C., Snijder, H.H., Cajot, L.-G., Popa, N. (2014)
Buckling curves for heavy wide flange QST columns based on
statistical evaluation, Journal of Constructional Steel Research,
101, pp. 280–289.
[15] Cajot, L.-G. (2014) Buckling curves for L-sections, Doc. CEN-
TC250-SC3_N2017.
[16] Zhang, L., Jaspart, J.-P. (2013) Stability of members in
compression made of large hot-rolled and welded angles,
Université de Liege.
[17] Beg, D., Rejec, K., Sinur, F. (2013) Determination of buckling
curves for large angle profiles considering different steel grades and
different residual stress patterns, University of Ljubljana, Faculty
of Civil & Geodetic Engineering.
[18] Beg, D., Može, P., Rejec, K., Sinur, F. (2013) Report on the residual
stress measurements and numerical determination of buckling
curves for large angle profiles, University of Ljubljana, Faculty of
Civil & Geodetic Engineering.
[19] EN 1993-1-12 (2005) Eurocode 3: Design of steel structures – Part
1-12: Additional rules for the extension of EN 1993 up to steel

grades S700, CEN, Brussels.
[20] Lindner, J., Simoes da Silva, L. (2015) Classification of rolled I-
Profiles fabricated in steel grade S460 within Table 6.2 of EN 1993-
1-1, Doc. CEN-TC250-SC3_N2164.
[21] Lindner, J. (2015) Classification of rolled I-profiles fabricated in
steel grade S460 within table 6.2 of EN 1993-1-1, Rev. 3, Report
to Working Group 1.
[22] Simoes da Silva, L., Tankova, T., Marques, L., Rebelo, C. 2015)
Safety assessment of EC3 stability design rules for flexural buckling
of columns, v40, Report to Evolution Group 1.
[23] Greiner R., Taras A. (2010) New design rules for LT and TF
buckling with consistent derivation and code-conform
formulation, Steel Construction - Design and Research, 3/2010,
pp. 176-186.
[24] Taras, A. (2010) Contribution to the Development of Consistent
Stability Design Rules for Steel Members, PhD thesis, Graz
University of Technology.
[25] Taras, A., Greiner, R., Unterweger, H. (2013) Proposal for
amended rules for member buckling, semi-compact cross-
section design, and extension of member buckling rules of
Annex B to monosymmetric sections, AM-1-1-2012-01 to 06,
Consolidated version of documents of the same title submitted
to the SC3 Evolution Group 1993-1-1.
[26] Greiner, R.; Kaim, P., Comparison of LT-buckling design curves
with test results, ECCS TC8 – Report, April 23-2001, European
Convention for Constructional Steelwork (2001), Brussels.
[27] Byfield, M.P., Nethercot, D.A., An analysis of the true bending
strength of steel beams. Proceedings of the Institution of Civil
Engineers – Structures and Buildings 128 (1998), pp. 188-197.
[28] CEC, ENV background doc. 5.03 (1988), Commission of the
European Community, Brussels.
[29] Rebelo, C., Lopes, N., Simões da Silva, L., Nethercot, D., Vila
Real, P. (2009). Statistical evaluation of the lateral-torsional
buckling resistance of steel I-beams - Part 1: Variability of the
Eurocode 3 design model, Journal of Constructional Steel
Research, 65(4), pp. 818-831.
[30] Simões da Silva, L., Rebelo, C., Nethercot. D, Marques, L.,
Simões, R., Vila Real, P. (2009). Statistical evaluation of the
lateral-torsional buckling resistance of steel I-beams - Part 2:
Variability of steel properties", Journal of Constructional Steel
Research, 65(4), pp. 832-849.
[31] Taras A., Greiner R., New design curves for lateral torsional
buckling - Proposal based on a consistent derivation, Journal of
Constructional Steel Research, 70, Elsevier London/Amsterdam,
2010.
[32] ÖNORM B 1993-1-1:2006: Eurocode 3; Bemessungs und
Konstruktion von Stahlbauten – Teil 1-1: Allgemeine
Bemessungsregeln
| 393
[33] EN 1993-6 (2007) Eurocode 3: Design of steel structures – Part 6:
Crane supporting structures, CEN, Brussels.
[34] Kaim, P. (2004) Spatial buckling behaviour of steel members under
bending and compression. PhD thesis, TU Graz.
[35] Kuhlmann, U. (2019) Interaktionsbeziehungen für Normalkraft,
Biegemomente und Torsion: Harmonisierung und Ergänzung der
Stabilitätsnachweise für Stäbe mit Standard-Walzprofilen. Project
AiF/IGF: 19044 N, 1 Feb 2016 – 30 Apr 2019.
[36] Lindner, J., Glitsch, T. (2004) Vereinfachter Nachweis für I- und
U-Träger – beansprucht durch doppelte Biegung und Torsion.
Stahlbau 73(9): 704–715.
[37] Bours, A.-L., Winkler, R., Knobloch, M. (2019) Ergänzende
Untersuchungen zum Tragverhalten einfachsymmetrischer I-
Querschnitte unter Biegung, Druck und Torsion. Stahlbau
88(9): 836–850.
[38] Winkler, R., Walter, A., Knobloch, M. (2018) Zum
Stabilitätsnachweis von Stahlbauteilen aus einfach- und
doppeltsymmetrischen I-Querschnitten unter Biegung, Druck
und Torsion. Stahlbau 87(5): 476–490.
[39] Winkler, R., Kindmann, R., Knobloch, M. (2019) Assessment of
the plastic capacity of I-shaped cross-sections according to the
partial internal forces method. Engineering Structures 191: 740–
751.
[40] EN 1990 (2003) Eurocode, Basis of structural design.
[41] prEN 1990 (2021) Eurocode, Basis of structural and geotechnical
design.
[42] SAFEBRICTILE (2017). Standardization of Safety Assessment
Procedures across Brittle to Ductile Failure Modes, RFSR-CT-
2013-00023, Final Report EUR 28906. European Commission.
[43] Tankova, T, Simões da Silva, L., Marques, L., Rebelo, C. and
Taras, A. (2014). Towards a standardized procedure for the
safety assessment of stability design rules, Journal of
Constructional Steel Research, 103, 290-302.
[44] Simões da Silva, L., Tankova, T., Marques, L., Rebelo, C.,(2016).
Safety assessment of EC3 stability design rules for the flexural
buckling of columns, Advanced Steel Construction, 12(3), 328-
358.
[45] Simões da Silva, L., Tankova, T., Marques, L., Rebelo, C. and
Taras, A. (2018). Safety assessment of EC3 stability design
rules for the lateral-torsional buckling of prismatic beams”,
Advanced Steel Construction, 14(4), 668-693.
[46] Simões da Silva, L., Tankova, T., Rebelo, C. (2020). Safety
Assessment of Eurocode 3 Stability Design Rules for Prismatic
Members in Bending and Compression. Int J Steel Struct, 20,
343-354.