You are on page 1of 12

Journal of Constructional Steel Research 157 (2019) 347–358

Contents lists available at ScienceDirect

Journal of Constructional Steel Research

Buckling resistance of steel H-section beam–columns: In-plane


buckling resistance
M.A. Gizejowski ⁎, Z. Stachura, R.B. Szczerba, M.D. Gajewski
Faculty of Civil Engineering, Warsaw University of Technology, Warsaw, Poland

a r t i c l e i n f o a b s t r a c t

Article history: Buckling resistance predictions resulting from flexural and flexural–torsional buckling of double tee section
Received 12 February 2019 members subjected to compression and bending are considered. A novel analytical model is developed for estab-
Accepted 3 March 2019 lishing design criteria based on decomposition of the member buckling behaviour into in-plane and out-of-plane
Available online xxxx
resistance. The former is based on second-order bending relationships of load effects of structural members with
in-plane equivalent imperfections, while the latter is based on the stability theory of thin-walled open sections.
Keywords:
Steel structures
First part of this study presents an analytical formulation of the in-plane buckling resistance of beam–columns. In
Imperfect beam–columns this regard, further decomposition is postulated for the in-plane first-order bending moment diagram that results
In-plane resistance in the loading state to be the superposition of two components. The first component is related to symmetrical
FEM nonlinear analysis loading and the second to antisymmetric loading. To consider the second-order effects, prebuckling displace-
Eurocodes ments generated by the abovementioned loading components are amplified with regard to the inversion of
the residuum of the buckling force utilisation ratio in order to obtain approximate values of second-order
displacements and internal moments. As a result, the in-plane interaction curve, expressed in dimensionless
coordinates, that describes the beam–column in-plane flexural buckling resistance without considering
lateral–torsional buckling effects, is obtained. The results of nonlinear finite element simulations are used for
the verification of the developed analytical formulation. It is concluded that this proposal yields less conservative
predictions than those based on the interaction relationships of clause 6.3.3 of Eurocode 3, Part 1–1.
© 2019 Elsevier Ltd. All rights reserved.

1. Introduction In the former approach, two interpolation formulas of clause 6.3.3


[1] deal with the member buckling resistance as a linear combination
The buckling resistance of columns and beams in EN 1993-1-1 [1] is of the two stability utilisation ratio components:
described by the analytical Ayrton–Perry model [2] of the behaviour of
a) axial compression buckling resistance utilisation ratio, and.,
initially crooked members with profile imperfections described by the
b) lateral–torsional buckling (LTB) utilisation ratio for moment about
lowest eigenmode, the amplitude of which has an equivalent value
the y-y axis multiplied by the interaction coefficients kij for ij = yy
representing globally the influence of all the imperfections of material
or zy that consider the nonlinear character and complexity of the be-
and geometric nature. As the direct application of Ayrton–Perry analyt-
haviour of steel beam–column elements of actual structural systems.
ical formulation for intermediate cases of beam–columns, for which the
element is subjected to nonzero values of both the axial compressive
force and the bending moment about the y-y axis, yields a highly com- The kij coefficients were derived from the general second-order rela-
plex analytical solution, the Eurocode verification criteria are based on tionships and presented in [1] in the form of two methods: Method 1
two approaches. In the case of mono-axial bending about a major axis from Annex A, which is more accurate but also more complicated for
they are as follows: hand calculations, or Method 2 from Annex B, which is more suitable
for quick resistance verification. These coefficients contain the so-
1. A set of interpolation formulas of [1], regarding clauses 6.2.9 dealing called equivalent uniform bending moment factors and are dependent
with the cross-sectional resistance and 6.3.3 dealing with the mem- upon a number of other parameters, which are identified through a sen-
ber buckling resistance, for the moment load effect Mz,Ed = 0. sitivity analysis as the most critical factors influencing the accuracy of
2. The so-called general method of clause 6.3.4 [1]. design interpolation criteria. The verification of the accuracy of these
design method was reported in [3,4].
⁎ Corresponding author. The in-plane behaviour of beam–columns is critical in engineering
E-mail address: m.gajewski@il.pw.edu.pl (M.A. Gizejowski). practice and the design of steel structures, e.g. planar portal and

https://doi.org/10.1016/j.jcsr.2019.03.002
0143-974X/© 2019 Elsevier Ltd. All rights reserved.
348 M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358

multi-storey frames. Chen and Atsuta [5] reported the in-plane behav- In the case of mono-axial bending and in-plane buckling resistance,
iour of beam–columns with perfect geometry. Guidelines for the stabil- as considered in Part 1 of this study, the Eurocode LTB reduction factor
ity and design of imperfect elements are reported in [6,7]. Columns and in [1] becomes equal to one and the formulae for interaction factors kij
beam–columns in the case of in-plane behaviour are sensitive to flex- become simplified. The safety level of the Eurocode approach with the
ural buckling (FB). A review of the development of the column stability use of these simplified interpolation formulas is compared in this part
criteria contained in studies and design codes was reported by of the study with that of the analytical formulation developed herein,
Bjorhovde [8]. An experimental and numerical study of stocky beam– and verified by the results of a wide range of finite element simulations
columns made of hot-rolled steel I-sections under combined compres- conducted specifically for this study. Issues related to the assumptions
sion and bending moment was reported by Yun et al. [9]. Experimental used in the finite element simulations of the in-plane flexural resistance
and numerical studies of laterally restrained steel columns with variable were reported in [34].
web-tapered I cross-section were reported in [10,11]. The buckling The application of the general method approach proposed in [1] re-
behaviour of high-strength compression steel columns was experimen- lates to the out-of-plane buckling resistance. The background of this
tally investigated by Ban et al. [12,13]. method was explained in detail by Bijlaard et al. [35], Simoes da Silva
An investigation considering the application of the so-called “Level 1” et al. [36], and in the ECCS design manual [37]. Papp and Szalai [38] illus-
[14] and “Level 2” [15] beam–column interaction formulae to isolated trated the use of the general method by means of a number of examples
members with arbitrary loading and end support conditions was and compared them with the classical method of [1].
reported by Goncalves and Camotim [16]. They compared the analytical Recently, research efforts for establishing design criteria for beam–
in-plane resistance of members with values yielding from second-order column resistance have been focused on the Ayrton–Perry approach
plastic zone finite element analyses to the use of the initial bow imper- to stability problems. Tankova et al. [39] presented a simplification of
fection and residual stress distributions. In their study, attention was the “exact solution” yielded from the differential equilibrium state
paid to issues related to the correct choice and determination of the of crooked beam–column elements for their use in the codification of
equivalent moment factor that is used in analytical interaction formulae. FTB resistance of beam–columns. The method proposed in [39] is closely
A safety assessment of EN 1993-1-1 [1] rules for FB of columns made based on the theoretical derivation of the generalisation of the Ayrton–
of hot-rolled I-shaped cross sections was reported by Silva et al. [17]. It Perry formula reported by Szalai and Papp [40] and referenced to LTB
was shown that the Eurocode imperfection factors currently prescribed problems of beams and beam–columns. A complete closed-form
in [1] for FB about the minor axis of steel-columns made of S460 steel universal Ayrton–Perry format-type solution for all types of buckling
are inadequate and they proposed that buckling curves were more suit- modes of steel beam–columns was derived by Szalai [41]. Simple con-
able. In the current version of EN 1993-1-1 [1], FB curves for hot-rolled I- cepts of the Ayrton–Perry analytical formulation were developed in
shaped cross sections with height-to-width ratios h/bf N 1.2 and flange [42] for FB of columns and in [43] for FTB resistance of beam–columns.
thicknesses tf N 100 mm are not available. Proposals for these types of These concepts are further investigated in this study and presented in
heavy I-shaped cross sections made from mild and high-strength steel two parts as a unified approach to predicting the resistance of beam–
were reported in [18,19] with the use of existing European buckling columns with regard to their in-plane and out-of-plane failure modes.
curves [1]. Taras and Greiner [20] reported that the European buckling Part 1 of this study considers design criteria for in-plane bending
curves from [1] do not satisfactorily represent torsional and flexural– behaviour. In terms of the Eurocode 3 approach, such criteria are
torsional buckling phenomena of laterally restrained columns. A new based on the second-order bending relationship of imperfect struc-
flexural–torsional buckling (FTB) curve for columns subjected to uni- tural members with in-plane equivalent imperfections discussed
form compression, which is fully consistent with the background and below. The obtained general solution can be simplified to one that
methodology of the European column buckling curves, was presented is also valid for the in-plane buckling resistance of perfect beam–
by Greiner and Taras [21]. columns. The out-of-plane stability criterion in the second part of
According to EN 1993-1-1 [1], the resistance of steel columns and this study [44] utilises:
beam–columns can be calculated by second-order analysis considering
equivalent initial bow imperfections. Verification of FB according to [1] a) the perfect member in-plane inelastic resistance without LTB effects
using bow imperfections was made by Lindner et al. [22]. However, considered, and.,
the use of equivalent imperfections according to [1] could be conserva- b) the elastic FTB limit curve formulation according to the thin-walled
tive [22,23]. In [23], it was suggested that structural analysis could be sections theory,
performed with equivalent column bow imperfections extracted di-
rectly from the Ayrton–Perry formulation. Chladny and Stujberova
as the basis for the development of the general method approach ap-
[24] presented the procedure using the equivalent unique global and
plied for the refined FTB resistance assessment.
local imperfection in the shape of the elastic buckling mode. A design
The analytical model is verified by the results of finite element sim-
example of a planar steel multi-storey frame was considered. It was
ulations. Comparisons of results from the proposed analytical model
noted that it could be necessary to consider the shape of the imperfec-
and the Eurocode design criteria according to Method 1 and Method 2
tion given by buckling modes higher than the lowest mode [25]. A pro-
[1] are also presented.
posal for the generalisation of an overall imperfection method utilising
linear buckling analysis for columns, beams, and beam–columns was re-
ported by Papp [26]. Lechner [27] reported the application of the 2. Buckling resistance criterion based on generalised Ayrton–Perry
Eurocode design methods in [1] to determine the FB resistance of a formulation
steel planar portal frame. Effects of geometric imperfections on FB resis-
tance of laterally braced columns can be found in [28]. The influence of We present a hypothetical situation of the independent treatment
initial imperfections variability on the behaviour of steel columns was of ultimate limit states referring to the in-plane inelastic behaviour
verified by a Sobol sensitivity analysis and reported by Kala [29]. (bending and buckling about the major axis discussed below and
In-plane resistance of structural steel elements can be verified with identified by the subscript ip = y) and the out-of-plane stability be-
the use of both stiffness reduction [30,31] and direct strength [32] haviour of I-section beam–columns (flexural-torsional instability
methods. Recently, Tankova et al. [33] reported a novel general formu- discussed in second part of this study [44] and identified therein by
lation that is based on stress utilisation, with the buckling mode as the the subscript op). These two independent solutions are combined in
shape of the initial imperfection that can be used to check in-plane resis- [44] to form the Ayrton–Perry approach for a single design criterion
tance of beam–columns. of the beam–column FTB resistance verification.
M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358 349

2.1. Background to evaluation of in-plane buckling resistance Fig. 1c shows the decomposition of the deflection state produced
byan arbitrary moment gradient case (general case of bending about
Let us consider a simply supported beam–column under compres- the y-y axis, denoted by GMY) into the linear combination of two bend-
sion and bending about the y-y axis that is subjected to a moment gra- ing moment diagram components, namely a symmetric component
dient resulting from the application of two different end moments: Myd,L (denoted by SMY) and an antisymmetric one (denoted by AMY), each
at the left support, and Myd,R at the right support (see Fig. 1a). The inter- corresponding to the respective components of the first order in-plane
nal bending moments at the left and right ends, being in equilibrium internal moment state.
with the applied moments, are My,Ed,L and My,Ed,R, respectively (see The ultimate strength of the beam–column can be represented by
Fig. 1b). The assumption of the maximum absolute value of the left the resistance utilisation ratio UF equated to unity. The UF ratio is a linear
end moment is adopted and it assigned as being positive, therefore, combination of the elementary resistance utilisation ratios defined in
the distribution of first order moment can be described by a single mo- accordance with the type of analysis introduced in clause 2.2 of EN
ment gradient ratio (ψy), as follows: 1993-1-6 [45] of the following form:
h   i
MIy;Ed ðξÞ ¼ 0:5My;Ed 1 þ ψy þ 1−ψy ð1−ξÞ ð1Þ U F ¼ U M þ U P‐δ þ U imp ð5Þ

where UM corresponds to the first order moment under applied Myd,L


where ξ = 2×/L is the dimensionless member coordinate (L is the beam
and Myd,R, UP–δ to the moment produced by the Px,d −δz,Ed effect, and
length); X(ξ) is the X variable being a function of ξ; and ψy = My,Ed,R/
Uimp to the moment from an equivalent imperfection profile (denoted
My,Ed,L, the major axis support moments ratio (My,Ed,L = My,Ed,max =
Udist in the case of a quasi-perfect element, i.e. the element with an infin-
My,Ed and My,Ed,R = My,Ed,min = ψy My,Ed,max = ψy My,Ed).
itesimally small value of the imperfection amplitude that ensures the
By decomposition, the first order moment can be represented by
convergence of numerical solutions for the perfect model inelastic buck-
symmetric Mys,Ed and antisymmetric Mya,Ed components, as follows:
ling load estimate).
  Table 1 presents the structure of the elementary utilisation factor com-
Mys;Ed ¼ 0:5 1 þ ψy My;Ed ð2Þ ponents to be used in the analytical formulations according to different
types of analysis. Additional symbols to those of clause 2.2 of EN 1993–1-
 
Mya;Ed ¼ 0:5 1−ψy M y;Ed ð3Þ 6 [45] are introduced, namely LA+ and MNA+, and GNA+ and GMNA+.
The first two symbols refer to the inclusion of unamplified Px, d − δIz, Ed mo-
ments, and the latter two refer to the inclusion of the disturbance of the first
so that Eq. (1) can be rewritten as: order moment of an infinitesimally small value in order to obtain numeri-
cally stable solutions for the buckling resistance of perfect members evalu-
MIy;Ed ðξÞ ¼ Mys;Ed þ ð1−ξÞM ya;Ed ð4Þ ated in the case of compression without bending moments.

Fig. 1. Loading case considered: a) loading, b) first order moment diagram, and c) first order displacement profile.
350 M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358

Table 1 moments do not exist:


Resistance utilisation ratio components.
2 3
Analysis type UM UP–δ Uimp(Udist) MIIys;dist 1 þ ψy 1−ψy
mIIys;dist ¼ ¼ my;dis 4 2
þ 2
5 ð11Þ
LA, MNA mIy;Ed – – Mcy;Rk 1−nEd λy 1−0:25nEd λy
mN;cy;Rk
LA+, MNA+ mIy;Ed mIyδ;Ed – where my,dis is the dimensionless disturbing moment (the value 10−6 is
mN;cy;Rk mN;cy;Rk suggested hereafter),
GNA, GMNAa mIy;Ed mIIy;Ed –
nEd – dimensionless axial force:
mN;cy;Rk mN;cy;Rk
GNA+, GMNA+b mIy;Ed mIIy;Ed mIIys;dist N Ed
mN;cy;Rk mN;cy;Rk mN;cy;Rk
nEd ¼ ð12Þ
Nc;Rk
GNIA mIy;Ed mIIy;Ed mIIys;imp
mN;y;el;Rk mN;y;el;Rk mN;y;el;Rk where Nc,Rk is the cross-sectional resistance in compression according
GMNIA mIy;Ed mIIy;Ed mIIys;imp to [1],
mN;y;pl;Rk mN;y;pl;Rk mN;y;el;Rk mN,cy,Rk – section class dependent reduced moment resistance
−2
a
Applied for mIy, Ed N 0 while for mIy, Ed = 0 the solution is nEd ¼ λy ≤1. defined as follows:
b
Applied for mIy, Ed ≥ 0.
M N;cy;Rk
mN;cy;Rk ¼ ð13Þ
M cy;Rk
Regarding the introduced decomposition of the load and load effect
components, the following notation is used in Table 1: that for class 1 and 2 cross-sections and plastic design (analysis types
mIy, Ed– dimensionless first order moment: MNA, MNA+, GMNA, GMNA+, GMNIA) is, according to [1], given by:

MN;y;pl;Rk n −ay
M Iy;Ed ðξmax Þ
Mys;Ed þ ð1  ξmax ÞM ya;Ed mN;cy;Rk ¼ mN;y;pl;Rk ¼ ¼ 1− Ed but mN;cy;Rk ≤1 ð14Þ
mIy;Ed ¼ ¼ ¼ Mcy;pl;Rk 1−ay
M cy;Rk M cy;Rk
¼ mys;Ed þ hð1  ξmax Þm ð6Þ
 ya;Ed ¼ i where ay = 0.5a, where a is the section parameter according to [1],
¼ 0:5my;Ed 1 þ ψy þ 1  ψy ð1  ξmax Þ while for class 3 and 4 cross-sections and elastic design (LA, LA+,
GNA, GNA+, GNIA) or any conservative design situation regardless of
where Mcy,Rk is the cross-sectional resistance in bending according to the section class:
[1], and the dimensionless bending moment is calculated as follows:
MN;y;el;Rk
mN;cy;Rk ¼ mN;y;el;Rk ¼ ¼ 1−nEd ð15Þ
M cy;el;Rk
My;Ed
my;Ed ¼ ð7Þ
Mcy;Rk
MN,y,pl,Rk, MN,y,el,Rk – reduced plastic or elastic cross-section resistance
in bending expressed according to [1], respectively,
mIyδ, Ed– dimensionless first order moment because of Px, d − δIz, Ed Mcy,pl,Rk, Mcy,el,Rk – plastic or elastic section resistance in bending
effect: about the y-y axis according to [1], respectively.
h i The first order deflections produced by applied moments Myd,L and
NEd δIzs;Ed ðξmax Þ þ δIza;Ed ðξmax Þ   Myd,R can be calculated by superposition of symmetric and antisymmet-
mIyδ;Ed ¼ ¼ nEd ηIys;Ed þ ηIya;Ed ð8Þ ric components and represented in dimensionless format according to
M cy;Rk
the following approximate relationships:

where NEd is the axial compressive force; for nEd, ηIys,Ed, and ηIya,Ed see Nc;Rk δIzs;Ed; max
Eqs. (12), (16), and (17), respectively, ηIys;Ed ¼ ηys ws ðξmax Þ ¼ ws ðξmax Þ ð16Þ
M cy;Rk
mIIy, Ed – dimensionless second-order moment because of amplified
Px, d − δIz, Ed effect: Nc;Rk δIza;Ed; max
ηIya;Ed ¼ ηya wa ðξmax Þ ¼ wa ðξmax Þ ð17Þ
h i M cy;Rk
NEd δIIzs;Ed ðξmax Þ þ δIIza;Ed ðξmax Þ  
mIIy;Ed ¼ ¼ nEd ηIIys;Ed þ ηIIya;Ed ð9Þ where ws ðξmax Þ, wa ðξmax Þ, δIzs,Ed,max, and δIza,Ed,max are expressed by
M cy;Rk
Eqs. (23)–(26), respectively.
Their second-order counterparts are the amplified ηIys,Ed and ηIya,Ed
where ηIIys,Ed and ηIIya,Ed are presented in Eqs. (18) and (19),
components, which can be approximated by:
mIIys,imp – dimensionless second-order bending moment from the
equivalent geometric bow imperfections at the lowest eigenmode pro- ηIys;Ed
file: ηIIys;Ed ¼ 2
ð18Þ
1−nEd λy
NEd wimp ðξmax Þ nEd ηys;imp
mIIys;imp ¼ ¼ ð10Þ ηIya;Ed
Mcy;Rk N 1−nEd λy
2
1− Ed ηIIya;Ed ¼ 2
ð19Þ
Ny 1−0:25nEd λy

where Ny is the elastic critical force for the relevant buckling mode; λy is An additional dimensionless parameter ηys,imp is introduced and is
the flexural slenderness ratio about the y-y axis according to [1]; and associated with the equivalent imperfection profile of the lowest buck-
wimp(ξmax) and ηys,imp are expressed by Eqs. (20) and (22), ling mode that, for pure compression, reproduces the buckling curve
mIIys,dist – dimensionless second-order bending moment, being used for design of actual (imperfect) structural members of steel struc-
a disturbing factor for the ideal beam–column when the applied tures in [1]. For the element considered herein, simple (fork type)
M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358 351

boundary conditions are used, for which the in-plane eigenmodes re- of the cross section under compression (the parameter necessary for
lated to in-plane buckling about the y-y axis are defined by a sine- classification of a cross section subjected to bending and compression
function series of the following form: according to [1]) can be estimated according to [49] for I- and
  X   H-sections. A review of cross-sectional classifications based on the sec-
πξ iπξ tion ductility in four major current design specifications (European,
wimp ðξÞ ¼ e0y sin þ e0y;i sin ð20Þ
2 i≥2
2 American, Japanese, and Chinese codes) was made by Chen et al. [50].
A novel concept of section classification based on the member ductility
where, for i = 1, the amplitude of the lowest eigenmode imperfection rather than on the section ductility is reported in [51]. The section con-
profile is given by: sidered here is of class 1 according to the abovementioned classification
criteria.
M cy;Rk The beam–column resistance of the HEB 300 section investigated in
e0y ¼ ηy ð21Þ
Nc;Rk [42] considered two finite element types, beam and shell elements, in
numerical simulations using ABAQUS software. The constitutive steel
The dimensionless Maquoi–Rondal equivalent amplitude is ηy ¼ model was piece-wise-linear elastic-plastic with isotropic hardening.
The imperfection amplitude was calculated according to Eq. (21) using
α y ðλy −0:2Þ, where αy is the imperfection factor associated with the
the plastic properties of the section. It was observed that, for small
buckling curve according to [1]. Therefore, the following relationship
values of the axial force, the beam model (BM) produces a resistance
holds for the parameter ηys, imp:
curve that is below that produced by the shell model (SM). This can
Nc;Rk e0y be attributed to the fact that the SM cannot represent the gradual yield-
ηys;imp ¼ ws ðξmax Þ ¼ ηy sinð0:5πξmax Þ ð22Þ ing of flange fibres through the thickness. In modelling of imperfections
Mcy;Rk
by the equivalent geometric initial profile, yielding of the flange in SM
occurs abruptly and encompasses all the fibres simultaneously, while
As the imperfection-related resistance utilisation ratio component is
yielding in BM is gradual and represented by the through-thickness nu-
represented by the sine function, it is convenient to use trigonometric
merical integrations at Gauss points. However, the differences are small,
shape functions when describing the beam–column deflected shapes
so that the shell modelling technique, which is more general, is consid-
under symmetric and antisymmetric loading components. The two low-
ered hereafter for the calibration of the proposed analytical formulation
est eigenmodes are therefore considered hereafter as shape functions:
of the beam–column in-plane buckling resistance.
The other issue in finite element modelling (FEM) and analytical for-
ws;Ed ðξmax Þ ¼ δIzs;Ed; max ws ðξmax Þ ¼ δIzs;Ed; max sinð0:5πξmax Þ ð23Þ
mulation is related to the adoption of the imperfection profile and its
amplitude. The effect of a single eigenmode and a superposition of mul-
wa;Ed ðξmax Þ ¼ δIza;Ed; max wa ðξmax Þ ¼ δIza;Ed; max sinðπξmax Þ ð24Þ tiple eigenmodes for representing the imperfection profile was reported
in [52], using the equivalent lowest mode amplitude according to
where the first order amplitudes corresponding to the symmetric and Eq. (21), and the superposition of multiple eigenmodes according to
antisymmetric loading components are given by: Eq. (20) for i ≤ 2 (together with the plastic section properties). The re-
sults of finite element simulations indicated that the most important
M y;Ed L2  
imperfection profile is that corresponding to the lowest eigenmode, re-
δIzs;Ed; max ¼ 1 þ ψy ð25Þ
16EIy gardless of the shape of the bending moment diagram. The question is
whether the application of plastic properties in the evaluation of equiv-
My;Ed L2   alent amplitudes will yield similar results to those obtained when using
δIza;Ed; max ¼ 1−ψy ð26Þ
128EIy elastic properties. On one hand, it was shown that the adoption of elastic
section bending resistance yields results for the reduction factor χy =
in which the other notations are according to [1]. Nby,Rk/Nc,Rk (where Nby,Rk is the buckling resistance) that are in good
All the components presented in Table 1 will be calculated for the agreement with the Eurocode buckling curves [1] for slender members,
section at the highest stress the dimensionless coordinate of which is and marginally above those of the buckling curve for stocky members.
ξ = ξmax. For a given section, the value of the ξmax coordinate depends On the other hand, the adoption of plastic section bending resistance
on ψy as well as the ratio of nEd/my,Ed. It is calculated by expressing the yields results for stocky members and members characterised by a me-
buckling resistance utilisation ratio UF in terms of an arbitrary ξ coordi- dium slenderness ratio, and are less than those from the application of
nate, finding the function of its first derivative over ξ, equating (d/dξ)UF the elastic section bending resistance. In addition, it was concluded
to zero, and solving for ξ = ξmax. If the equation (d/dξ)UF = 0 has no that the numerical results for slender members were similar to each
positive solution, the most stressed section is at the beginning of beam other and to the buckling curve of [1]. The results obtained using equiv-
(at My,Ed,max, see Fig. 1), therefore for the coordinate ξ = ξmax = 0. alent amplitude based on the plastic bending moment resistance are
used hereafter for calibration of the model parameters and verification
2.2. Aspects of finite element modelling affecting in-plane buckling of the refined analytical formulation (RAF) developed in the following
resistance section.

Finite element methods are currently the most common methods


used in steelwork verification of analytical models, as laboratory exper- 2.3. Analytical formulation of in-plane buckling resistance
imental investigations are both costly and time consuming. The latest
recommendations and useful guidelines for finite element modelling The in-plane buckling resistance evaluation considers beam–
techniques, simulations, and results verification are reported in columns of the section class 1 or 2, for which option GMNIA in Table 1
[46–48]. This technique is also used for obtaining the numerical results is relevant. The resistance utilisation ratio in GMNIA is normalised
used for verifying the analytical model developed in the preceding with the plastic moment resistance Mcy,Rk = Mcy,pl,Rk. Because the resis-
subsection. tance utilisation ratio component Uimp in compression has to be at the
As an example, we consider the in-plane buckling resistance about elastic level, regardless of the section class, this term in Table 1 has
the y-y axis of an HEB 300 section of S235 grade steel. Typically, in the the same format for GNIA and GMNIA, in order to analytically evaluate
case of bending and axial force, the parameter describing the portion the resistance, as that of clause 6.3.1 in [1].
352 M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358

The analytical model originally developed assumes concentrated so that the interrelationship between dimensionless stress resultants
plasticity for members of both class 1 or 2 sections, i.e. the beam– can be expressed as the functions of αEd:
column ultimate limit state is reached when the plastic hinge is formed
in the most stressed cross section under the interactive action of com- α Ed
my;Ed ¼ n ð31Þ
pression and bending (so-called full-plastic resistance). Therefore, the 1−α Ed Ed
ultimate limit state criterion of the in-plane bending and buckling
about the y-y axis is described by UF = 1 as: 1−α Ed
nEd ¼ my;Ed ð32Þ
α Ed
h   i  
0:5my;Ed 1 þ ψy þ 1  ψy ð1  ξmax Þ þ nEd ηIIys;Ed þ ηIIya;Ed
mN;y;pl;Rk Eq. (27) can now be expressed as a function of αEd for each propor-
ð27Þ
nEd ηys;imp tion of nEd and my,Ed and solved accordingly. Repeating the procedure for
þ ¼1 different values of nEd and αEd, or my,Ed and αEd, the dimensionless values
 2 nEd
mN;y;el;Rk 1  λ
y
of the applied bending moment Myd,L = My,Ed or the applied force Px,d =
NEd can be calculated.
where the reduced plastic moment resistance mN,y,pl,Rk according to The ultimate limit strength curves from the presented analytical for-
Eq. (14) is applied by rearranging the relationship given in the Eurocode mulation are compared in Fig. 2 with the results of the FEM simulations.
[1]: The concentrated plasticity approach referring to the ultimate limit
8 strength curves is based on two versions of the inelastic interaction
<1 for α n ≤ay curves at the plastic hinge formation. The “full-plastic resistance” (de-
mN;y;pl;Rk ¼ nEd −ay ð28Þ
: 1− for α n Nay noted by FPR) curves are based on the bilinear interaction of My,Ed–NEd
1−ay in cross-sectional resistance of [1], while the “partial-plastic resistance”
(denoted by PPR) is based on the conservative linear interaction of the
where: bending moment and the axial force at the plastic hinge formation.
The former gives the analytical upper bound and the latter the analytical
nEd
αn ¼ ð29Þ lower bound. Bounds created by the results of the above two stated ver-
my;Ed
sions of the analytical formulation GMNIA presented in Table 1, and the
results from FEM simulations using the plastic amplitude of the equiva-
Using the factor αEd ranging from zero to unity, the current load lent imperfection profile, correspond to three different values of ψy. The
effects can be expressed as functions of that factor. Let us consider the areas between the upper and lower bounds are shaded. The analytical
following relationship for αEd: formulation using the linear elastic cross-sectional resistance and
GNIA is the most conservative, and not referred to in Fig. 2.
1 In conclusion, using GMNIA and the linear plastic cross-sectional
α Ed ¼ ð30Þ
1 þ αn resistance relationship yields the conservative solution, however, it is

Fig. 2. Comparison of buckling resistance curves using different assumptions of most stressed cross-section ultimate limit state curve: a) ψy = 1.0, b) ψy = 0.0, and c) ψy = −1.0.
M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358 353

significantly closer to that obtained from FEM simulations than that cal- 3. Verification of developed refined formulation using finite element
culated with GNIA and the linear elastic cross-sectional resistance. The simulations
most appropriate approach is that using the bilinear plastic cross-
sectional resistance relationship (according to clause 6.2.9.1 [1]) that 3.1. Verification by finite element modelling of in-plane resistance
constitutes the upper bound solution in reference to FEM results,
allowing for the attainment of plastic hinge moment resistance for The developed FEM technique discussed in Section2.2 is adopted for
low values of the dimensionless axial force. Observations made with re- simulation of the in-plane resistance of beam–columns under compres-
gard to the results from the analytical formulation based on the bilinear sion and mono-axial bending about the y-y axis. Three cases of the mo-
plastic hinge moment resistance and FEM simulations suggest that the ment gradient are investigated. They are characterised by ψy = 1, 0, and
analytical results are on the safe side, particularly for low values of − 1, and the imperfection profile being the lowest eigenmode with the
axial force and members with high slenderness ratios. This can be amplitude in line with that of Maquoi–Rondal of [1]. The proposed RAF
clearly seen for a negative value of ψy, when the member is bent in a considering the distributed plasticity (see Section2.3 above) is verified
double curvature. The observation of plasticity development in the with the use of FEM simulations based on the plastic equivalent imper-
most stressed section and along the member length indicates that, for fection amplitude. The results are presented in Fig. 3.
more slender elements, the member ultimate limit state is associated Using the refined analytical model, the buckling resistance curves
with an inelastic distribution of stresses in the most stressed section can be accurately reproduced taking into consideration the fact that,
being below that corresponding to the full plastic redistribution, and for small values of nEd, the FEM SM simulations are somewhat less con-
with the plastic zones being less extensive over the member length. servative. This is because they are unable to reproduce gradual yielding
A refinement is therefore needed to better represent the effects of dis- through the flange thickness, and the cross-section resistance is repre-
tributed plasticity, both in the most stressed section, and along the lon- sented by the bilinear relationship that overestimates the actual resis-
gitudinal axis on both sides of the most stressed section. Such a tance in the neighbourhood of αn = ay [cf. Eq. (28)]. Therefore, it is
refinement can be carried out by shifting the section limit curve down- concluded that the developed refined model can reproduce FEM results
wards towards the linear one and increasing the value of second-order more accurately than the original plastic hinge approach. In addition, it
displacements for members with greater slenderness ratios. Therefore, provides results on the safe side and, more importantly, for all values of
the following refinements are used: the moment gradient and slenderness ratios, the shape of ultimate limit
1. The resistance utilisation ratio UM remains to be normalised with curves remains similar to that obtained from the FEM simulations.
the use of the unreduced plastic resistance in the bending about the It can be seen that, for my,Ed = ηIIys,Ed = ηIIya,Ed = 0, Eq. (35) is being
y-y axis, My,pl,Rk. reduced to that for the characteristic resistance evaluation of imperfect
2. The resistance utilisation ratio UP-δ is normalised with the reduced columns in accordance with the original Ayrton–Perry approach, the so-
elastic–plastic resistance in bending MN,y,el-pl,Rk, in which: lution of which is used in clause 6.3.1 [1]. On the other hand, the devel-
oped formulation can also be used for the evaluation of the buckling
a) the effect of inelastic section redistribution of stresses is accounted
resistance curves of perfect beam–columns where my,Ed, ηIIys,Ed, and/or
for by the factor:
ηIIya,Ed are greater than zero. The refined formulation presented in
Section2.3 has to be modified in this case in accordance with GMNA+
by applying the disturbing resistance utilisation ratio component Udist.
8
> 1 for λy ≤0:2 Therefore, the following relationship holds:
>
>
W y;pl < W y;pl
¼ 1− ð33Þ h i
W y;el−pl >> W y;pl W y;el 0:5my;Ed 1 þ ψy þ 1  ψy 1  ξmax
>
: þ   for λy N0:2
W y;el 1 þ cγcy λy −0:2 þ
mN;y;pl;Rk
W y;pl γ nEd ηys;Ed þ ηya;Ed
II II
þ þ ð36Þ
W y;elpl 2 mN;y;elpl;Rk 3
b) and the effect of the plastic zones by the factor: my;dis 4 1 þ ψy 1  ψy
5¼1
þ þ
mN;y;pl;Rk 1  n λ  2 1  0:25nEd λ
Ed y
2
y

8
>
< mN;y;pl;Rk for λy ≤0:2
The verification of perfect beam–column buckling resistance is
mN;y;el−pl;Rk ¼ mN;y;pl;Rk −mN;y;el;Rk ð34Þ
>
: mN;y;el;Rk þ   for λy N0:2 shown in Fig. 4, where the FEM curves are reproduced from [34] and
1 þ cby λy −0:2 the analytical results are calculated by Eq. (36).
It can be seen that the refined analytical solution developed in this
where γ, ccy, and cby are constants fitted to the FEM resistance curves. study covers a wide range of specific design situations, not only those
Considering the HEB 300 section, the following model parameters related to buckling of imperfect beam–columns, but also to buckling of
are proposed: ccy = cby = 3.75, and γ = 2. The original formulation perfect beam–columns. It can be seen that, for my,Ed = ηIIys,Ed = ηIIys,Ed
presented above is therefore refined by replacing Eq. (27) with the fol- = 0, Eq. (36) is reduced to that allowing the dimensionless resistance
lowing equation: −2
evaluation of perfect columns to be 0:25λy ≤1 in the case of an anti-
symmetric bending moment diagram (Udist component for ψy = −1.0
h   i −2
0:5my;Ed 1 þ ψy þ 1  ψy ð1  ξmax Þ corresponds to the second lowest eigenmode) and λy ≤1 in other
þ cases of bending moment diagrams (Udist component for ψy N −1.0 cor-
mN;y;pl;Rk 
  responds to the lowest eigenmode).
W y;pl γ nEd ηys;Ed þ ηya;Ed
II II
ð35Þ
þ þ
W y;elpl mN;y;elpl;Rk 3.2. FEM equilibrium paths and development of plasticity in most stressed
nEd ηys;imp
þ ¼1 section
 2 nEd
mN;y;el;Rk 1  λ
y
Fig. 5 shows the comparison of the numerical results of the exem-
plary equilibrium paths from the FEM incremental analysis related to
354 M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358

Fig. 3. Numerical results for imperfect beam–column resistance based on single lowest buckling mode imperfection profile (solid line) and developed analytical model (dashed line):
a) ψy = 1.0, b) ψy = 0.0, and c) ψy = −1.0.

Fig. 4. Numerical results for perfect beam–column resistance [34] (solid line) and developed analytical model (dashed line): a) ψy = 1.0, b) ψy = 0.0, and c) ψy = −1.0.
M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358 355

SM simulations for a chosen ψy and slenderness ratio. The stress distri- For ψy = 1, mya,Ed and ηya,Ed are equal to zero and mys,Ed and ηys,Ed are
bution patterns in the bulk of the stressed cross-sections at the ultimate nonzero. The most stressed section coordinate is at mid-length, regard-
limit state are discussed below. It can be seen that, for the members less of the value of the dimensionless moment, therefore ξmax = 1.0.
with higher slenderness ratios, the buckling resistance is associated It can be seen that the refined analytical model accurately predicts
with an inelastic redistribution of stresses in the most stressed section the most stressed section coordinate, and how the value of this coordi-
and is farther from that corresponding to the plastic hinge. This supports nate travels from mid-length (ξmax = 1 when ψy = 1) to the left end
the way in which refinements have been introduced to the plastic hinge section of the maximum applied moment (ξmax → 0 when ψy → –1).
formulation in order to define the developed analytical modelling
technique. 3.3. Comparison of proposed refined analytical formulation and Eurocode's
Prediction of the most stressed section coordinate ξmax by the FEM results
simulations is presented in Figs. 6 and 7 for two values of ψy, namely
ψy = 0 and ψy = −1. The curves presented in Figs. 6 and 7 are for the The refined analytical model is used in this subsection for the verifi-
slenderness ratio equal to unity. cation of the results from Eurocode interpolation Method 1 and Method
2 [1].

Fig. 5. Comparison of equilibrium path and stress distribution pattern in most stressed section in case of my,Ed = 0.8: a) ψy = 1.0, b) ψy = 0.0, and c) ψy = −1.0.
356 M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358

Fig. 6. Comparison of estimation of most stressed section coordinate ξmax, for ψy = 0: a) loading case, and b) value of coordinate ξmax depending on my,Ed.

Fig. 8 shows the comparison of the RAF developed herein and the 4. Summary and conclusions
results of the Eurocode interpolation relationships according to
Method 1. In this, Part 1 of the study, the authors focussed on issues related
The verification of the analytical model presented in Section3.1 to the stability resistance of single-span steel members which are
indicates that the developed model represents the behaviour of im- sufficiently laterally and torsionally restrained and subjected to
perfect beam–columns with sufficient accuracy for practical appli- axial compression and bending about the y-y axis, which is the stron-
cations for ψ y or member slenderness ratio. The comparison in ger inertia axis. Because of the adopted assumptions, the considered
this section indicates that the developed analytical model is less elements were only sensitive to in-plane interaction of buckling and
conservative than that of Eurocode Method 1, and represents the bending about the y-y axis (no out-of-plane effects occur; thus lat-
behaviour of imperfect beam–columns with a better accuracy for eral, torsional, and lateral–torsional buckling were not taken into
practical applications for any ψy or member slenderness ratio. The consideration).
advantage of such a formulation is primarily that there is no need The authors proposed the novel analytical model based on the RAF,
for the introduction of the so-called equivalent uniform moment which allows for the estimation of in-plane beam–column buckling re-
factor, as used in the Eurocodes model. Application of this factor sistance. To stabilise the various parameters of the formulation, numer-
renders the Eurocode Method 1 results more conservative, with ous FEM analyses were conducted with beam and shell finite elements,
the difference increasing with decreasing ψy (antisymmetric load- considering different load conditions (uniform, triangle, and antisym-
ing component is more pronounced). metric bending moment diagrams were considered). To make the prob-
Fig. 9 shows the comparison of the RAF developed herein and the lem more general, perfect and imperfect (with initial crookedness based
results of the Eurocode interpolation relationships according to on scaled buckling modes) members were analysed. The results of the
Method 2. numerical calculations and their comparison with the proposed analyt-
The results shown in Fig. 9 indicate that the developed model ical model were discussed in detail and presented in the study.
is considerably less conservative than that of Eurocode Method The results obtained from the developed analytical approach were
2. The comparison of results in Figs. 8 and 9 clearly supports the compared with those from EN 1993-1-1 [1] analytical procedures
conclusion that the concept of an equivalent uniform moment re- based on interpolation relationships. The proposed analytical model
sults in significantly conservative results, specifically for the neg- has only one design condition because the concept of equivalent uni-
ative moment gradient ratio and when Eurocodes Method 2 is form moment is no longer maintained. Therefore, the cross-section ca-
used. pacity and the overall stability are not required to be checked

Fig. 7. Comparison of estimation of most stressed section coordinate ξmax for ψy = −1: a) loading case, and b) value of coordinate ξmax depending on my,Ed.
M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358 357

Fig. 8. Comparison of refined analytical formulation and Eurocode Method 1: a) ψy = 1.0, b) ψy = 0.0, and c) ψy = −1.0.

Fig. 9. Comparison of refined analytical formulation and Eurocode Method 2: a) ψy = 1.0, b) ψy = 0.0, and c) ψy = −1.0.
358 M.A. Gizejowski et al. / Journal of Constructional Steel Research 157 (2019) 347–358

separately, which is obligatory when using EN 1993-1-1 [1] procedures. [26] F. Papp, Buckling assessment of steel members through overall imperfection
method, Eng. Struct. 106 (2016) 124–136.
In addition, the proposal of the authors, as opposed to the standard ap- [27] A. Lechner, Flexural buckling of frames according to the new EC3 rules – a compar-
proach, has a simple physical interpretation. Detailed conclusions are ative, parametric study, Proceedings of the International Colloquium on Stability
presented in the relevant subsections. and Ductility of Steel Structures, Lisbon, Portugal, 2006.
[28] C. Dou, Y. Pi, Effects of geometric imperfections on flexural buckling resistance of lat-
This part of the study is the introduction to second part [44], in erally braced columns, J. Struct. Eng. 142 (9) (2016).
which the authors present the issues related to the out-of-plane buck- [29] Z. Kala, Sensitivity assessment of steel members under compression, Eng. Struct. 31
ling resistance criteria of elements sensitive to lateral–torsional (2009) 1344–1348.
[30] M. Kucukler, L. Gardner, L. Macorini, A stiffness reduction method for the in-plane
deformations. design of structural steel elements, Eng. Struct. 73 (2014) 72–84.
[31] M. Kucukler, L. Gardner, L. Macorini, Development and assessment of a practical
References stiffness reduction method for the in-plane design of steel frames, J. Constr. Steel
Res. 126 (2016) 187–200.
[1] Eurocode 3, EN 1993-1-1: Design of Steel Structures, Pat 1-1: General Rules and [32] A. Taras, Derivation of DSM-type resistance functions for in-plane global buckling of
Rules for Buildings, CEN, 2005. steel beam-columns, J. Constr. Steel Res. 125 (2016) 95–113.
[2] W.E. Ayrton, J. Perry, On struts, Engineer 62 (1886) 464–465. [33] T. Tankova, L. Simões da Silva, L. Marques, Buckling resistance of non-uniform steel
[3] A. Boissonnade, J.-P. Jaspart, J.-P. Muzeau, M. Villette, New interaction formulae for members based on stress utilization: general formulation, J. Constr. Steel Res. 149
beam-columns in Eurocode 3: the French-Belgian approach, J. Constr. Steel Res. 60 (2018) 239–256.
(2004) 421–431. [34] M.A. Gizejowski, R.B. Szczerba, M.D. Gajewski, Z. Stachura, Buckling resistance as-
[4] R. Greiner, J. Lindner, Interaction formulae for members subjected to bending and sessment of steel I-section beam-columns not susceptible to LT-buckling, Arch.
axial compression in Eurocode 3 – the method 2 approach, J. Constr. Steel Res. 62 Civ. Mech. Eng. 17 (2017) 205–221.
(2006) 757–770. [35] F. Bijlaard, M. Feldmann, J. Naumes, G. Sedlacek, The “general method” for assessing
[5] W.F. Chen, T. Atsuta, Theory of Beam-Columns, Vol. 1, In-Plane Behavior and Design, the out-of-plane stability of structural members and frames and the comparison
McGraw-Hill, New York, 1976. with alternative rules in EN 1993 – Eurocode 3 – part 1-1, Steel Constr. 3 (2010)
[6] N.S. Trahair, M.A. Bradford, D.A. Nethercot, L. Gardner, The Behaviour and Design of 19–33.
Steel Structures to EC3, Taylor & Francism, London– New York, 2008. [36] L. Simoes da Silva, L. Marques, C. Rebelo, Numerical validation of the general method
[7] R.D. Ziemian, Guide to Stability Design Criteria for Metal Structures, 6th ed. John in EC3-1-1 for prismatic members, J. Constr. Steel Res. 66 (2010) 575–590.
Wiley & Sons Inc, New York, 2010. [37] L. Simoes da Silva, R. Simoes, H. Gervasio, Design of Steel Structures, Eurocode 3: De-
[8] R. Bjorhovde, Evolution and state-of-the-art of column stability criteria, J. Civ. Eng. sign of Steel Structures, Pat 1-1: General Rules and Rules for Buildings, 2nd ed. ECCS,
Manag. 16 (2010) 159–165. Ernst & Sohn, 2016.
[9] X. Yun, L. Gardner, N. Boissonnade, Ultimate capacity of I-sections under combined [38] F. Papp, A. Rubert, J. Szalai, DIN EN 1993–1-1 based integrated stability analysis of
loading – part 1: experiments and FE model validation, J. Constr. Steel Res. 147 2D/3D steel structures, Stahlbau (2014) 325–342 Part 1–83 (1) (2014) 1–15; Part
(2018) 408–421. 2–83 (2) (2014) 122–141; Part 3–83 (5), in German.
[10] I.M. Cristutiu, D.L. Nunes, A. Dogariu, Experimental study on laterally restrained [39] T. Tankova, L. Marques, A. Andrade, L. Simoes da Silva, A consistent methodology for
steel columns with variable I cross sections, Steel Compos. Struct. 13 (2012). the out-of-plane buckling resistance of prismatic steel beam-columns, J. Constr.
[11] T. Tankova, J.P. Martins, L. Simões da Silva, R. Simões, H.D. Craveiro, Experimental Steel Res. 128 (2017) 839–852.
buckling behaviour of web tapered I-section steel columns, J. Constr. Steel Res. [40] J. Szalai, R. Papp, On the theoretical background of the generalization of Ayrton-
147 (2018) 293–312. Perry type resistance formulas, J. Constr. Steel Res. 66 (2010) 670–679.
[12] H. Ban, G. Shi, Y. Shi, Y. Wang, Overall buckling behavior of 460MPa high strength [41] J. Szalai, Complete generalization of the Ayrton-Perry formula for beam-column
steel columns: experimental investigation and design method, J. Constr. Steel Res. buckling problems, Eng. Struct. 153 (2017) 205–223.
74 (2012) 140–150. [42] M.A. Gizejowski, Z. Stachura, M.D. Gajewski, R.B. Szczerba, A new method of buck-
[13] H. Ban, G. Shi, Overall buckling behaviour and design of high-strength steel welded ling resistance evaluation of laterally restrained beam-columns, in: M.A.
section columns, J. Constr. Steel Res. 143 (2018) 180–195. Gizejowski, A. Kozlowski, J. Marcinowski, J. Ziolko (Eds.), Recent Progress in Steel
[14] R. Greiner, Background Information on the Beam-Column Interaction Formulae at and Composite Structures, Taylor & Francis Group, London 2016, pp. 102–103 , (e-
Level 1, Report TC8–2001-002, ECCS Technical Committee 8, 2001. book on CD: 197–205).
[15] A. Boissonnade, J.-P. Jaspart, J.-P. Muzeau, M. Villette, Improvement of the interac- [43] M.A. Gizejowski, Z. Stachura, Generalized Ayrton-Perry approach for the evaluation
tion formulae for beam columns in Eurocode 3, Comput. Struct. 80 (2002) of beam-column resistance, in: A. Zingoni (Ed.), Insights and Innovations in Struc-
2375–2385. tural Engineering, Mechanics and Computation, Taylor & Francis Group, London
[16] R. Gonçalves, D. Camotim, On the application of beam-column interaction formulae 2016, pp. 253–254 , (e-book on CD: 713-719).
to steel members with arbitrary loading and support conditions, J. Constr. Steel Res. [44] M.A. Gizejowski, Z. Stachura, R.B. Szczerba, M.D. Gajewski, Buckling resistance of
60 (2004) 433–450. steel H-section beam–columns: Out-of-plane buckling resistance, J. Constr. Steel
[17] L. Simões da Silva, T. Tankova, L. Marques, C. Rebelo, Safety assessment of Eurocode Res. (submitted).
3 stability design rules for the flexural buckling of columns, Adv. Steel Constr. 12 [45] Eurocode 3, EN 1993–1-6: Design of Steel Structures, Part 1–6: Strength and Stabil-
(2016) 328–358. ity of Shell Structures, CEN, 2007.
[18] H.H. Snijder, L.G. Cajot, N. Poppa, R.C. Spoorenberg, Buckling curves for heavy wide [46] R. Kindmann, M. Kraus, Steel Structures – Design Using FEM, Ernst & Sohn, Berlin,
flange steel columns, Rom. J. Tech. Sci. Appl. Mech. 59 (1/2) (2014) 178–204. 2012.
[19] R.C. Spoorenberg, H.H. Snijder, L.G. Cajot, N. Popa, Buckling curves for heavy wide [47] E. Ellobody, R. Feng, B. Young, Finite Element Analysis and Design of Metal Struc-
flange QST columns based on statistical evaluation, J. Constr. Steel Res. 101 (2014) tures, Butterworth-Heinemann, Waltham, 2014.
280–289. [48] M. Abambres, M.R. Arruda, Finite element analysis – steel structures – a reviewof
[20] A. Taras, R. Greiner, Torsional and flexural torsional buckling – a study on laterally useful guidelines, Int. J. Struct. Integr. 7 (2016) 490–515.
restrained I-sections, J. Constr. Steel Res. 64 (2008) 725–731. [49] L. Gardner, D.A. Nethercot, Designers' Guide to EN 1993-1-1 Eurocode 3: Design of
[21] R. Greiner, A. Taras, New design curves for LT and TF buckling with consistent der- Steel Structures: General Rules and Rules for Buildings, Thomas Telford, SCI,
ivation and code-conform formulation, Steel Constr. 3 (3) (2010) 176–186. London, 2005.
[22] J. Lindner, U. Kuhlmann, A. Just, Verification of flexural buckling according to [50] Y. Chen, X. Cheng, D.A. Nethercot, An overview study on cross-section classification
Eurocode 3 part 1-1 using bow imperfections, Steel Constr. 9 (2016) 349–362. of steel H-sections, J. Constr. Steel Res. 80 (2013) 386–393.
[23] J. Jönsson, T.-C. Stan, European column buckling curves and finite element model- [51] M. Shokouhian, Y. Shi, Classification of I-section flexural members based on member
ling including high strength steels, J. Constr. Steel Res. 128 (2017) 136–151. ductility, J. Constr. Steel Res. 95 (2014) 198–210.
[24] E. Chladny, M. Stujberova, Frames with unique global and local imperfection in the [52] M.A. Gizejowski, R.B. Szczerba, M.D. Gajewski, Z. Stachura, Beam-columnin-plane
shape of the elastic buckling mode, Stahlbau 82 (8) (2013) 609–617. resistance based on the concept of equivalent geometric imperfections, Arch. Civ.
[25] A. Aguero, L. Pallares, F.J. Pallares, Equivalent geometric imperfection definition in Eng. 62 (2016) 35–71.
steel structures sensitive to flexural and/or torsional buckling due to compression,
Eng. Struct. 96 (2015) 160–177.

You might also like