Structures 29 (2021) 1445–1462

Contents lists available at ScienceDirect

Structures
journal homepage: www.elsevier.com/locate/structures

New method for metal beams sensitive to lateral torsional buckling with an
equivalent geometrical UGLI imperfection
A. Agüero a, *, I. Baláž b, Y. Koleková c
a
Dept. of Continuous Medium Mechanics and Theory of Structures, Universitat Politècnica de València, C/Camino de Vera s/n, 46022 Valencia, Spain
b
Slovak University of Technology in Bratislava ⋅ Faculty of Civil Engineering, Department of Metal and Timber Structures, Bratislava, Slovak Republic
c
Slovak University of Technology in Bratislava ⋅ Faculty of Civil Engineering, Department of Structural Mechanics, Bratislava, Slovak Republic

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

Keywords: The Eurocodes EN 1993-1-1:2005, prEN 1993-1-1:2020, and EN 1999-1-1:2007, prEN 1999-1-1:2020 propose
Lateral torsional buckling the same method. They take into account the geometrical equivalent Unique Global and Local Initial (UGLI)
Imperfection shapes and amplitudes imperfections. According to the Eurocodes, the assumed shape of UGLI imperfections may be derived from the
Equivalent UGLI imperfection
elastic buckling mode and provide a method that enables computing the amplitude of a UGLI imperfection in
EN 1993-1-1:2005
prEN 1993-1-1:2020
slender metal structures. The authors method presented enables computing the amplitude of the initial imper­
fection for the lateral torsional buckling of beams with a doubly symmetric section. This is a generalization of
Eurocode rules that is valid only for members in compression. Numerical examples are presented in order to
understand the authors method. The buckling resistance is verified with the test of Wieschollek et al. and
compared with the results of four other authors, the results of a Geometrically and Materially Non-linear Analysis
(GMNIA) with imperfections and the results of other methods used in the current Eurocode 3 and a draft of its
new generation.

(ii) Direct method using a 2nd order analysis with the equivalent
1. Introduction geometric imperfections.

1.1. Background According to clauses 5.3.4(3) [1,2] and 7.3.4(3) [4], “Taking account
of lateral torsional buckling of a member in bending the imperfections
As outlined in EN 1993-1-1 [1] and EN 1999-1-1 [2], the design of may be adopted as k e0 where e0 is the equivalent initial bow imper­
metal structures with compression elements must consider the effects of fection of the weak axis of the profile considered. In general an addi­
imperfections, including any residual stresses and geometrical imper­ tional torsional imperfection need not to be allowed for. The value k =
fections. The following imperfections could be taken into account: a) 0.5 is recommended. The National Annex may choose the value of k.”
global imperfections for frames and bracing systems; b) local imper­ According to 7.3.3.2 [3], “For a 2nd order analysis taking account of
fections for individual members; c) the shape of the elastic critical lateral torsional buckling of a member in bending, the equivalent
buckling model ηcr of the structure, which may be applied according to imperfection may be determined according to (7.11), where e0,LT is the
clauses 5.3.2(11) in [1] and [2] as the geometrical equivalent Unique equivalent bow imperfection about the weak axis of the profile consid­
Global and Local Initial (UGLI) imperfection. ered. In general, an additional torsional imperfection may be neglected”.
The methods which enable obtaining the buckling resistance of In this paper a numerical method is presented that enables obtaining
beams sensitive to lateral torsional buckling according to [1–4] are as the equivalent initial imperfection for beams with a doubly symmetric
follows: section susceptible to lateral torsional buckling.
The comparisons of the relative buckling resistances αb (eq. (12)) are
(i) Indirect method performing a linear analysis. This method in­ given in Fig. 1 for the methods used in [1] and [3]. The results differ in
cludes geometrical, and material non-linearity and imperfections the range of the smallest relative slendernessλLT . In order to ensure that
by means of buckling curves to obtain the reduction factor χ LT they will lead to the same imperfection there should be a relation be­
according to clause 6.3 in [1]. tween the imperfection and the relative slenderness. For a smaller

* Corresponding author.
E-mail address: anagra@mes.upv.es (A. Agüero).

https://doi.org/10.1016/j.istruc.2020.11.047
Received 20 July 2020; Received in revised form 13 October 2020; Accepted 19 November 2020
Available online 30 December 2020
2352-0124/© 2020 Institution of Structural Engineers. Published by Elsevier Ltd. All rights reserved.
A. Agüero et al. Structures 29 (2021) 1445–1462

Nomenclature aluminium alloy [2]: 0.6 for Class 1 and 2 cross-sections

and 0.4 for Class 3 and 4 cross-sections in 6.3.2.2)
α is the imperfection factor related to the flexural buckling β is the correction factor for the lateral torsional buckling
curve (Tables 6.1 and 6.2 of EN 1993-1-1 [1]; Tables 3.2 curves (6.3.2.3 in [1;2] does not know such quantity and
and 6.6 in EN 1999-1-1 [2]) symbol)
αLT is the imperfection factor for the lateral torsional buckling {ηinit} is the UGLI imperfection in the shape of the elastic critical
related to the buckling curve (Tables 6.3, 6.4 and 6.5 in EN buckling mode
1993-1-1 [1]; 6.3.2.2 in EN 1999-1-1 [2]) {ηcr} is the shape of the elastic critical buckling mode
αcr is the minimum load amplifier for the axial force ηcr,w is the displacement of the buckling shape component
configuration in the members to reach the elastic critical perpendicular to the axis y
buckling load (5.2.1(3) in [1]; 5.2.1(3) in [2]) ηcr,v is the displacement of the buckling shape component
αult is the amplifier for the load of the members to reach the perpendicular to the axis z
characteristic resistance of the critical cross-section (6.3.4 ηcr,θx is the torsional rotation of the buckling shape component
(2) in [1], where more convenient symbol αult,k is used; [2] about shear center axis
does not know such quantity and symbol) A is the cross-sectional area
αb is the relative lateral torsional buckling resistance (eq. η0 is the amplitude of the UGLI imperfection.
(12)); the symbol is not used in [1] and [2]) E is the modulus of elasticity (210 000 MPa for steel [1]; 70
γM0 is the partial safety factor for the resistance of cross-section 000 MPa for aluminium alloy [2])
whatever the Class cross section is (6.1 in [1;2] does not Iy, Iz are the second moments of area with respect to y and z axes
know such quantity and symbol) Iw is the warping constant
γM1 is the partial safety factor for the resistance of members to Ncr is the elastic critical force of the relevant buckling mode
instability assessed by member checks (6.1 in [1] and based on the gross cross-section properties
6.1.3; Table 6.1 in [2]) My,Ed is the design value for the bending moment about y axis
χ is the reduction factor for the relevant buckling curve NRk is the characteristic resistance of the normal force in the
(6.3.1.2 in [1]; 6.3.1.2 in [2]) critical section
χLT is the reduction factor for the lateral torsional buckling MRk is the characteristic resistance of the bending moment in
λ is the relative slenderness (6.3.1.2 in [1]; 6.3.1.2 in [2]) the critical section xcr
λ0 is the plateau length of the buckling curves (for steel [1]: UMy+Mz+B is the utilization factor due to My,Ed , Mz,Ed, and B,Ed
0.2; for aluminium alloy [2]: 0.1 for Buckling Class A and Wz is the section modulus about z axis
0.0 for Buckling Class B) Wy is the section modulus about y axis
λLT is the relative slenderness for the lateral torsional buckling WB is the warping section modulus
(6.3.2.2 and 6.3.2.3 in [1]; 6.3.2.2 in [2]) xcr is the critical section where the utilization factor is greater
than factors at all the other sections.
λLT,0 is the plateau length of the lateral torsional buckling curves
(for steel [1]: 0.2 in 6.3.2.2 and 0.2–0.4 in 6.3.2.3; for

Fig. 1. Comparisons of the relative lateral torsional buckling resistances αb as a Fig. 2. Comparisons of the relative lateral torsional buckling resistances αb as a
function of the relative slenderness λLT for two methods: a) 6.3.2.2 [1] (“a” function of slenderness λLT for two methods: a) 6.3.2.2 [1] (“a” buckling curve)
buckling curve) and b) the 2nd order (E-P) analysis with imperfection according and b) the 2nd order (E-P) analysis with imperfection according to 7.3.3.2 [3].
to 7.3.3.2 [3]. Simply supported beam, steel grade S235, IPE 200 and HEB Simply supported beam, steel grade S 235, IPE 200 profile.
300 profiles.

1446
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 4. Lateral torsional buckling of continuous beam.

⎛ ( ) ⎞
Fig. 3. Comparisons of the relative lateral torsional buckling resistances αb for
2
⎜α λ − λ0 1− λ χ
fy ⎟
ηinit,v (x) = η0 ⋅ηcr,v (x) = ⎜ ⎟
γM1
two methods: a) 6.3.2.2 [1] (“a” buckling curve) and b) the 2nd order (E-P) ⎜ 2 2 ( )⎟ ⋅ηcr,v (x)
⎝ 1 − λ χ E⋅ Iz d2 ηcr,v ⎠
analysis with imperfection according to 7.3.3.2 [3]. The cantilever with a point λ
Wz dx2
load F at the free end acting in shear centre, steel grade S 235, IPE 200 profile. Xcr
The 2nd order analysis uses three different shapes and amplitudes of initial (3)
imperfection.
In the case of prismatic elements with uniform axial forces, no iter­
slenderness, the imperfection should be smaller in order to match the ation is needed; the critical section will take place where the curvature
6.3.2.2 [1] results. A second order (E-P) analysis performed elastic has maximum. In most practical cases, the iteration is needed.
calculation of internal forces taking into account a plastic linear inter­ This proposal was further generalized by Chladný et al. [5,6,7], when
action equation (22) between the bending moments and bimoment for a the flexural buckling occurs about both axes as follows:
plastic cross-section resistance. { } { }
ηinit,v (x) η (x)
In Fig. 2, the relative buckling resistances αb calculated according to ηinit,w (x)
= η0 ⋅ cr,v
ηcr,w (x)
the 2nd order analysis are greater than the ones computed according to ⎛ ( ) ⎞
6.3.2.2 [1]. α 2
⎜ λ− λ 0 λ
1− γM1 χ ⎟ { η (x) }
From Fig. 3 it can be concluded that it is very important to take into =⎜
⎜
( ⃒
fy
⃒ ⃒
⎟
⃒ ) ⎟ ⋅ cr,v
account the initial torsional rotation imperfection in order to have ⎝ λ
2 2
1− λ χ E⋅ Iz ⃒⃒d2 ηcr,v ⃒⃒ + Iy ⃒⃒d2 ηcr,w ⃒⃒ ⎠ ηcr,w (x)
Wz ⃒ dx2 ⃒ Wy ⃒ dx2 ⃒
similar relative buckling resistance αb as in 6.3.2.2 [1]. Without the Xcr
initial torsional rotation imperfection, the amplitude in the y direction (4)
has to be multiplied by two (2L/200 = L/100, i.e., ¼ sinus wave) to
The formulae (2), (3) and (4) may be simplified by deleting the
obtain similar results as with the torsional imperfection (L/200 buckling
partial safety factor γM1. This is not recommended because in such a
shape).
case, the basic feature of the method will be destroyed, and the results
The imperfection can be expressed in the form of a single imper­
will not be the same as the results of the equivalent member method for
fection, similar to the buckling mode ηcr(x) of the structure (clauses 5.3.2
NEd = Nb,Rd.
(11) [1,2], 7.3.6(1) [3], 7.3.2(11) [4]), and it is called the geometrical
The new methodology has been developed during recent years in
equivalent Unique Global and Local Initial (UGLI) imperfection, which is
different publications with examples obtaining the flexural buckling
fully described in Chladný et al. [5,6,7].
resistance for members with a non-uniform cross-section, non-uniform
The proposals in [1–4] are based on equation (1), when flexural
axial force, and arch structures (Baláž et al. [8,9,10,11]). In [8], the
buckling occurs about a strong axis due to the compression:
amplitude of this imperfection was derived in another way by
In general the imperfections can be written as: {ηinit (x) } =
comparing it with Chladný‘s method.
η0 {ηcr (x) }
One generalization in the case of flexural torsional buckling due to
⎛ ( ) ⎞
2
compression was done by Agüero et al. [12,13], where the analysis of
⎜α λ − λ0 1− λ χ
fy ⎟ the imperfect structure is described as follows:
ηinit (x) = η0 ⋅ηcr (x) = ⎜ ⎟
(1)
γM1
⎜ 2 2 ( )⎟ ⋅ηcr (x) ⎧ ⎫ ⎧ ⎫
⎝ λ 1 − λ χ E⋅ I d2 ηcr ⎠ ⎨ ηinit,v (x) ⎬ ⎨ ηcr,v (x) ⎬
W dx2
Xcr η (x) = η ⋅ η (x)
⎩ ηinit,w (x) ⎭ 0 ⎩ ηcr,w (x) ⎭
General formula (1) applied for the flexural buckling about a strong init,θx
⎛
cr,θx
( ) ( ⎞
)
axis: 2
⎜ fy 1−
λTF χ TF
α λ TF − λ 0 ⎟ ⎧η ⎫
⎛ ( ) ⎞ ⎜ γM1 ⎟ ⎨ cr,v ⎬
⎜ ⎟ η
2 =⎜ ( ) ⎟
⎜α λ − λ0 1− λ χ
fy ⎟ ⎜ d 2η I d 2η d 2η 2
( 2
) ⎟ ⎩ η cr,w ⎭
ηinit,w (x) = η0 ⋅ηcr,w (x) = ⎜ ⎟ I
⎝E z 2 + I ⎠
⎜
γM1
( )⎟ ⋅ηcr,w (x)
cr,v
Wz dx
y
Wy
cr,w
2
dx
+ w
WBi dx
cr,θx
2 λ TF 1− λTF χ cr,θx
⎝ 2 2
λ 1 − λ χ E⋅ Iy d ηcr,w ⎠
2
Xcr
Wy dx2
Xcr (5)
(2)
According to clause 8.3.1.4 [3] for doubly symmetric I- and H-sec­
General formula (1) applied for the flexural buckling about a weak tions, the following equation can be used:
axis:

1447
A. Agüero et al. Structures 29 (2021) 1445–1462

⎧ ⎫ ⎧ ⎫
⎨ ηinit,v (x) ⎬ ⎨ ηcr,v (x) ⎬ U = 1.0, if the equivalent pinned-ended member has the same properties
η (x) = η η (x) as the cross-section in the critical section xcr. Similarly for the members
⎩ ηinit,w (x) ⎭ 0 ⎩ ηcr,w (x) ⎭
init,θx cr,θx in bending, the method proposed by the authors, which uses the relative
( )
⎛
2
( ) ⎞ slendernessλLT defined at the critical section xcr, has the same lateral
λTF χ TF ⎧ ⎫
⎜ fy 1− γM1
αTF λz − λ0 ⎟ ⎨ ηcr,v ⎬ torsional buckling resistance as the resistance obtained according to
⎜ ⎟
=⎜
⎜ ( ) ( ⎟
) ⎟ ⎩ ηcr,w ⎭ 6.3.2 [1].
⎝E Iz d2 ηcr,v
+
Iy d2 ηcr,w
+
2
Iw d ηcr,θx
λ
2
1− λ
2
χ ⎠ ηcr,θx
Wz dx 2 Wy dx 2 WBi dx 2 z TF TF
Xcr 2. Method to obtain the amplitude of the imperfection
(6)
Non-linear geometric elastic analysis is used in Sections 2 and 3.
Another generalization in the case of lateral torsional buckling was
According to Agüero et al. [14], equation (7) must be used to obtain the
done by Agüero et al. [12,14] (Fig. 4) , where the analysis of the
UGLI imperfection. Equation (8) [3] could also be used, but that is
imperfect structure is described as follows:
beyond the scope of this paper.
{ } { }
ηinit,v (x) η (x) The buckling shape has been scaled to have a maximum value 1.0, i.
= η0 ⋅ cr,v
ηinit,θx (x) ηcr,θx (x) e., max[ηcr,v(x)] = 1.0. The meaning of η0 is the amplitude of the
⎛ ( )
( ) ⎞
2
imperfection at the shear centre.
⎜ αLT λLT − λLT,0 fy 1− λLTγM1χLT ⎟ { } ( ) ( ) ( )
⎜ ⎟ ηcr,v η0 = ηinit,v xmax,v = max ηinit,v (x) = max η0 ηcr,v (x) (9)
=⎜
⎜ 2 ( ) ( ⃒ ⃒ ⃒ ⃒ ) ⎟
⎟
2 ⃒ 2 ⃒ ⃒ 2 ⃒
⎝λLT 1− βλLT χ LT E⋅ Iz ⃒d ηcr,v ⃒ + Iw ⃒d ηcr,θx ⃒ ⎠ ηcr,θx ( ) ( ) ( )
Wz ⃒ dx ⃒ WBi ⃒ dx ⃒
2 2 ηinit,θx xmax,θx = max ηinit,θx (x) = max η0 ηcr,θx (x)
Xcr
(7) Four steps to obtain the imperfection:
The first step: calculation of the buckling load αcr and buckling
For doubly symmetric sections, the curvatures are considered as an
shape ηcr(x). They can be computed using the Finite Element Method
absolute value.
(FEM), e.g., according to Trahair [18].
According to clause 8.3.2.3 [3] for doubly symmetric I- and H-sec­
The second step: calculation of the bending moments about weak z
tions and fork supports at both ends, the following equation can be used,
axis and the bimoments related to the buckling mode. The internal
considering fM = 1 (Table 8.6 in [3]):
forces of bending and torsion are obtained as follows
{ } { }
ηinit,v (x) ηcr,v (x)
ηinit,θx (x) = η0 ηcr,θx (x) Mzη (x) = EIz
d2 ηcr,v
; Bη (x) = EIw
d2 ηcr,θx
(10)
⎛ ( ) ⎞ dx 2 dx2
( ) 2

The relevant stresses are calculated from equation

λLT χ LT
⎜ αLT λz − λ0 fy 1− γM1 ⎟ { }
⎜ ⎟ η
=⎜
⎜ 2 ( ) ( ⃒ ⃒ ⃒ ⃒ ) ⎟
⎟ ⋅ cr,v
2 ⃒ ⃒ ⃒ ⃒ ηcr,θx EIz d2 ηcr,v EIw d2 ηcr,θx
⎝λz 1− λLT χ LT E⋅ Iz ⃒d ηcr,v ⃒ + Iw ⃒d ηcr,θx ⃒ ⎠ (11)
2 2

Wz ⃒ dx ⃒ WBi ⃒ dx ⃒
σ Mz (x) + σB (x) = +
2 2
Wz dx2 WB dx2
Xcr
(8) The third step: the first iteration of the calculation uses the initial
guess
Bijlaard et al. [15] and Wieschollek et al. [16] have also generalized
√̅̅̅̅̅̅̅̅̅
the equation given in Eurocodes [1] and [3] in the case of lateral W ⋅f αult,1 αult,1 ⋅χ LT,1
αult,1 = min( y y )→λLT,1 = →χ LT,1 →αb,1 = (12)
torsional buckling, considering cross-section flanges as members under MyEd αcr γM1
compression sensitive to flexural buckling using Chladny’s method.
Papp [17] solved a case of buckling under bending and compression. Another initial guess can be considered given by the position where
the given eq. (11) has maximum.
It is needed to calculate the imperfection (scale factor Ω1(x)) in each
1.2. Research significance
section in order to reach the cross-section resistance αb,1 in the moment
when the buckling load level is reached:
The innovative aspect of this research is the numerical method,
which enable obtaining the equivalent initial imperfection for metal MyEd ⋅αb,1 Ω1 (x)
( (
I d2 ηcr,v Iw d2 ηcr,θx
))
fy
members susceptible to lateral torsional buckling. The method is +( ) E⋅ z + = (13)
Wy αcr Wz dx2 WB dx2 γM0
consistent with the Eurocodes [1,2,3,4] for members in compression and αb,1 − 1
allows obtaining the critical section and the amplitude of the imper­
( )( )
fection. The lateral torsional buckling resistance is verified by fy MyEd ⋅αb,1
comparing with tests [16], the results of other authors, the results of −γM0 Wy
⋅ ααb,1
cr
− 1
Eurocode procedures, and the GMNIA results. Ω1 (x) = ( ) (14)
d2 ηcr,v d2 η
E⋅ WIzz dx2
+ WIwB dxcr,θx
2

1.3. Basic assumptions

The minimum of these scale factors is the objective of the first iter­
In this work the following assumptions have been made: ation, i.e., η0,1 takes place at the critical section xcr,1.
The first assumption is that clause 6.3.2 [1] allows designers to use (
η0,1 = min(Ω1 (x) ) = Ω1 xcr,1
)
(15)
the χ LT factor, which was originally deduced from flexural buckling
experiments on members in compression as well as for members in The fourth step: the second iteration
bending sensitive to lateral-torsional buckling. ( ) ( ) √̅̅̅̅̅̅̅̅̅
Wy xcr,1 ⋅fy xcr,1 αult,2 αult,2 ⋅χ LT,2
The second assumption is that the basis of the authors‘ proposal is the αult,2 = ( ) →λLT,2 = →χ LT,2 →αb,2 = (16)
same as in the method for the members in compression given in Euro­ MyEd xcr,1 αcr γ M1
code clauses 5.3.2 (11) [1,2], 7.3.6(1) [3], 7.3.2(11) [4]. It uses the Compute the utilization factor U(x) along the beam:
elastic buckling mode ηcr(x) as the shape of the UGLI imperfection
ηinit(x). The method using the UGLI imperfection and the Equivalent
Member Method (EMM) leads to the same value of the utilization factor

1448
A. Agüero et al. Structures 29 (2021) 1445–1462

⃒ ⃒ ⃒ ⃒
⃒d 2 η ⃒ ⃒d 2 η ⃒
EIz ⃒⃒ dxcr,v ⃒ EIw ⃒⃒ dxcr,θx ⃒

Numberof cases[Source]
2 ⃒ 2 ⃒
MyEd ⋅αb,2 η0,1 η
UMy+Mz+B (x) = ( ) + ( ) ( ) + ( 0,1 ) ( )
fy αcr fy αcr fy
Wy γM0 αb,2 − 1 Wz γM0 αb,2 − 1 WB γM0

(17)

12[17]

10[16]

11[19]
For the doubly symmetric sections, the maximum utilization is ob­

9[17]

6[20]
7[21]
tained as the sum of the absolute values of the partial utilizations.
Calculation of the next critical section xcr,2, where the utilization

My,Ed = My,pl (1 – x/L) – My,pl (x/L),My,pl = 439.1 kNm

FEd = 14.75 kN + FEd = 14.75 kN in the shear centre, factor U is the maximum
( ) ( )
max UMy+Mz+B (x) = UMy+Mz+B xcr,2 (18)
location:2.64 m + 1.76 m + 1.76 m + 2.64 m

If the critical section is the same as in the previous iteration, the

critical section xcr has been found and also as a consequence the value of
My,Ed = 50 (1 – x/L) + 37.5 x/L kNm

the initial imperfection amplitude η0. The initial imperfection is

obtained.
My,Ed = 32.68 (1 - x/L) kNm

My,Ed = My,pl = 86.16 kNm

{ηinit } = η0 {ηcr } (19)

Loading [kNm], [kN]

My,Ed = 26.094 kNm

If the location of the critical section differs from the previous one,
another iteration is needed in step four until the position of the critical
section is obtained
⎛( )( )⎞
fy MyEd ⋅αb,2 αcr
⎜ γM0 − Wy
⋅ αb,2 − 1 ⎟
η0,2 = ⎜ ⎜ (
⎟
) ⎟ (20)
⎝ Iz d2 ηcr,v 2
Iw d ηcr,θx
⎠
fy [MPa]

E⋅ Wz dx2 + WB dx2
337.7

Xcr,2
235
235

235
235
235

3. Benchmark beams calculated by the authors‘ method,

welded tapered beam, Table 3[Fig. 8 – Fig. 9]

rolledtapered beam,Table 7[Fig. 20 – Fig. 21]

according to [1,3] and by of other authors

An overview of the benchmark beams is presented in Table 1.

HEB 300, Table 5[Fig. 14 – Fig. 15]
IPE 240, Table 6[Fig. 16 – Fig. 19]

The values of the relative lateral torsional buckling resistance are

checked with the test of Wieschollek et al. [16] and the GMNIA results of
IPE 120, Table 4[Figs. 10-12]
Cross SectionTable, [Figures]

IPE 200, Table 2[Figs. 5–7]

Papp [17], Gizejowski et al. [19], Kucukler et al. [20] and Tankova et al.
[21]. In all the cases γM1 = 1.0.
Overview of beams in investigation. Number of calculated cases: 55 and 5 for example 4.1, 22 figures.

The authors performed the following calculations using a second

order elastic (E) analysis in the calculation of internal forces with a
elastic (E) or plastic (P) verification of a cross-section depending on the
Class of cross-section. In a case of E-P analysis the limitation of the
plastic cross-sectional reserves was taken into account consistent with
Eurocode [1]. The residual stresses for rolled or welded cross-sections
are taken according to [22]:
4.4 + 4.4

23.879
5.879
L [m]

2.72

7.75

1) The GMNIA was performed in Examples Nos.1, 2 and 6 with the

9

shape of the geometrical initial imperfection proportional to the

simply supported, tension flange restrained in the midspan

eigenmode deflection with the amplitude e0 = L/1000 and with a

simply supported, elastic spring at flange in compression

structural initial imperfection in the form of residual stress patterns

for hot rolled or welded sections taken according to [22].
2) The authors‘ method used the calculated buckling mode as the
imperfection and the clauses of the current Eurocode [1]: (i) 6.3.2.2
with parameters λLT,0 = 0.2, βmax = 1.0, (ii) λLT,0 = 0.4, β = 0.9, f =
1, (iii) 6.3.2.3 with parametersλLT,0 = 0.4, βmin = 0.75, f = 1.
3) Eurocode procedures [1] which use the clauses: (i) 6.3.2.2 with pa­
rameters λLT,0 = 0.2, βmax = 1.0, (ii) case 4: λLT,0 = 0.4, β = 0.9, f = 1,
two span continuous beam

(iii) 6.3.2.3 with parametersλLT,0 = 0.4, βmin = 0.75, f = 1.

simply supported
simply supported

simply supported

In the authors‘ method and in Eurocode procedure [1], the relative

lateral torsional buckling resistances αb in the critical sections xcr were
Supports

calculated after computing the relevant χ LT values.

4) The 2nd order theory with the shape of the initial imperfection in the
form of a quadratic parabola or complete sinus wave or half-wave
Exam- ple No.

sinus with amplitude ke0 = 0.5e0 according to 5.3.4(3) in [1]

Table 1

4.2
1
2
3

5
6

1449
 Table 2
Example 1. Papp [17]. Comparisons of the results obtained by 12 ways of calculations. Simply supported beam, L = 5.879 m, IPE 200, S235, γM1 = 1.0, linear bending moment distribution My,Ed = 32.68 (1 - x/L) kNm
(Figs. 5-7).
A. Agüero et al.

EXAMPLE 1Fig. 5 – a) GMNIA Material imperfection: residual Results according to authors‘ method and [results according to EMM* in current Theory of 2nd order with amplitude according to [1] 5.3.4(3) l) Papp [17]
Fig. 7 stress patterns for hot rolled sections [1] clauses 6.3.2.2 or 6.3.2.3] and [draft [3] 7.3.3.2]
b) Case 1, Fig. 6: [c) Case d) Case 3, [e) Case 4];[1] f) Case 5 [g) Case 6];[1] h) Case 7 [1], Fig. 7; [i) Case 8 j) Case 9, [1], [k) Case 10
2];[1] 6.3.2.2 β = 1.0, 6.3.2.3, f = 1, β = 0.9, 6.3.2.3, f = 1, β = 0.75, [3], [3],
λLT,0 = 0.2 λLT,0 = 0.4 λLT,0 = 0.4

Geometrical Buckling mode including torsional Authors‘ method: buckling mode ηinit,v(x) including torsional imperfection ηinit,θ Quadratic parabola, without Half-wave sinus, without Buckling mode including
imperfection shape imperfection x(x) ; torsional imperfection. torsional imperfection torsional imperfection
Imperfection L/1000 = 5.9 mm η0 = 7 mm ≡ ≡ηinit,v η0 = 7 mm ≡ ≡ηinit,v η0 = 4 mm ≡ ≡ηinit, 0.5e0 = 14.7 mm 0.5e0 = 14.7 mm 10.28 mm
amplitude [mm] (2.528 m) (2.528 m) v(2.528 m) [e0,LT = 29.4 mm] [e0,LT = 29.4 mm]
Relative buckling 1.00 0.975 [0.888] 0.975 [0.887] 1.059 [0.941] 1.038 [0.919] 1.050 [0.935] 1.024
resistance αb
Critical section – 1.352 [ – ] 1.293 [ – ] 1.293 [ – ] 1.293 [1.352] 1.352 [1.470] 1.654
xcr [m]
Relative buckling
resistance αb

* Equivalent Member Method – EMM in [1] is based on experimental tests and numerical investigation without torsional rotation imperfection.
Calculations of amplitudes: h) and j) k e0 = 0.5 L/200 = 0.5 × 5879 mm/200 = 14.7 mm, i) and k) e0,LT = βLT L /ε = 5879 mm/200/1.0 = 29.4 mm

1450
Table 3
Example 2. Papp [17]. Comparisons of the results obtained by 9 ways of calculations. Simply supported welded tapered beam, L = 2.72 m, I-section, hmin = 200 mm, hmax = 600 mm b = 100 mm tf = 8.5 mm tw = 5.6 mm ,
S235, γM1 = 1.0, linear bending moment distribution My,Ed = 50 (1 - x/L) + 37.5 x/L kNm (Fig. 8 – Fig. 9).
EXAMPLE 2Fig. 8 – a) GMNIA Material imperfection: residual Results according to authors‘ method and [results according to Theory of 2nd order with amplitude according to [1] 5.3.4(3) and l) Papp [17]
Fig. 9 stress patterns for welded sections EMM* in current [1] clauses 6.3.2.2 or 6.3.2.3] [draft [3] 7.3.3.2]

b) Case 1, Fig. 8:[1] d) Case 2, [1] f) Case 3, [1] h) Case 4 [1], Fig. 9, [i) Case 5 j) Case 6, [1], [k) Case 7 [3],
6.3.2.2 β = 1.0, 6.3.2.3, f = 1, 6.3.2.3, f = 1, [3],
λLT,0 = 0.2 β = 0.9,λLT,0 = 0.4 β = 0.75,λLT,0 = 0.4

Geometrical Buckling mode including torsional Authors‘ method: buckling mode ηinit,v(x) including torsional Quadratic parabola, without Half-wave sinus, without Buckling mode including
imperfection shape imperfection imperfection ηinit,θx(x) ; torsional imperfection torsional imperfection torsional imperfection
Imperfection L/1000 = 2.7 mm η0 = 8 mm ≡ ≡ηinit, η0 = 6 mm ≡ ≡ηinit, η0 = 5 mm ≡ ≡ηinit,v 0.5e0 = 6.8 mm 0.5e0 = 6.8 mm 4.9 mm
amplitude [mm] v(1.378 m) v(1.378 m) (1.378 m) [e0,LT = 27.2 mm] [e0,LT = 27.2 mm]
Relative buckling 1.167 1.044 [-] 1.168 [-] 1.225 [-] 1.32 [1.170] 1.335 [1.182] 1.132
resistance αb
Critical section xcr [m] – 1.741 [ – ] 1.813 [ – ] 1.868 [ – ] 2.122 [1.940] 2.049 [1.868] 1.955
Relative buckling
resistance αb

* Equivalent Member Method – EMM in [1] is based on experimental tests and numerical investigation without torsional rotation imperfection.
Calculations of amplitudes: h) and j) k e0 = 0.5 L/200 = 0.5 × 2720 mm/200 = 6.8 mm, i) and k) e0,LT = βLT L /ε = 2720 mm/100/1.0 = 27.2 mm
Structures 29 (2021) 1445–1462
 Table 4
Example 3. Wieschollek et al. [16]. Comparisons of the results obtained by 10 ways of calculations. Continuous beam with two spans L + L = 4.4 m + 4.4 m, IPE 120, fy = 337.7 MPa, γM1 = 1.0, loaded by two point loads
FEd = 14.75 kN acting in shear center, located LF = 1.76 m from intermediate support (Figs. 10-12).
A. Agüero et al.

EXAMPLE 3Figs. 10- a) GMNIA Material imperfection: residual Results according to authors‘ method and [results according to EMM* in current Theory of 2nd order with amplitude according to [1] 5.3.4 l) Wieschollek et al [16]
12 stress patterns for hot rolled sections [1] clauses 6.3.2.2 or 6.3.2.3] (3) and[draft [3] 7.3.3.2] [test]

b) Case 1, Fig. 11,[c) Case d) Case 3, [e) Case 4];[1] f) Case 5, [g) Case 6];[1] – j) Case 7, [1], [k) Case 8, [3],
2];[1] 6.3.2.2 β = 1.0, 6.3.2.3, f = 1, β = 0.9, 6.3.2.3, f = 1, β = 0.75, Fig. 12]
λLT,0 = 0.2 λLT,0 = 0.4 λLT,0 = 0.4

Geometrical Buckling mode including torsional Authors method: buckling mode ηinit,v(x) including torsional imperfection Quadratic parabola, Half-wave sinus, without Buckling mode including
imperfection shape imperfection ηinit,θx(x) ; without torsional imp. torsional imperfection torsional imperfection
Imperfection – η0 = 6 mm ≡ ≡ηinit,v η0 = 6 mm ≡ ≡ηinit,v η0 = 2 mm ≡ ≡ηinit,v – 0.5e0 = 11 mm –
amplitude [mm] (2.200 m) (2.200 m) (2.200 m) [e0,LT = 26.4 mm]
Relative buckling – 0.975 [0.875] 0.989 [0.876] 1.098 [0.941] – 1.069 [0.955] 0.869[1.0]
resistance αb
Critical section – 6.160 [ – ] 6.160 [ – ] 6.160 [ – ] – 6.160 [6.160] 6.160
xcr [m]
Relative buckling
resistance αb

* Equivalent Member Method – EMM in [1] is based on experimental tests and numerical investigation without torsional rotation imperfection.
√̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
Calculations of amplitudes: j) k e0 = 0.5 L /200 = 0.5⋅4400 mm/250 = 11 mm, k) e0,LT = βLT L /ε = 4400 mm/200/ 235/337.7 = 26.4 mm.

1451
Table 5
Example 4.2. Gizejowski et al. [19]. Comparisons of the results obtained by 11 ways of calculations. Simply supported beam, L = 23.879 m, HEB 300, S235, γM1 = 1.0,linear bending moment distribution My,Ed = My,pl (1 –
x/L)- My,pl (x/L), My,Ed = My,pl = 439.1 kNm (Fig. 14 – Fig. 15).
EXAMPLE 4.2Fig. 14 a) GMNIAMaterial imperfection: residual Results according to authors‘ method and [results according to EMM* in current [1] Theory of 2nd order with amplitude according to [1] 5.3.4 k) Gizejowski [19]
– Fig. 15 stress patterns for hot rolled sections clauses 6.3.2.2 or 6.3.2.3] (3) and [draft [3] 7.3.3.2]
b) Case 1, Fig. 14,[c) Case 2]; d) Case 3, [e) Case 4];[1] f) Case 5, [g) Case 6];[1] h) Case 7 [1], Fig. 15; [i) (j) Case 9 [3],
[1] 6.3.2.2 β = 1.0,λLT,0 = 0.2 6.3.2.3, f = 1, β = 0.9, 6.3.2.3, f = 1, β = 0.75, Case 8 [3],
λLT,0 = 0.4 λLT,0 = 0.4

Geometrical Buckling mode including torsional Authors method: buckling mode ηinit,v(x) including torsional imperfection ηinit,θx(x) ; Sinus, without torsional Half-wavesinus, without Buckling mode including
imperfection imperfection imperfection torsional imperfection torsional imperfection
shape
Imperfection L/1000 = 23.9 mm η0 = 26 mm ≡ ≡ηinit,v η0 = 26 mm ≡ ≡ηinit,v η0 = 20 mm ≡ ≡ηinit,v 0.5e0 = 40 mm [e0,LT = 119.4 mm] 23.9
amplitude [mm] (18.15 m) (18.15 m) (18.15 m) [e0,LT = 59.7 mm]
Relative buckling 0.93 0.857 [0.755] 0.855 [0.755] 0.898 [0.78] 0.962 [0.902] 0.995 0.93
resistance αb
Critical section xcr – 21.252 [ – ] 21.252 [ – ] 21.252 [ – ] 1.672 [21.491] 0.0 –
[m]
Relative buckling
resistance αb

* Equivalent Member Method – EMM in [1] is based on experimental tests and numerical investigation without torsional rotation imperfection.
In this example two imperfection shapes were considered: (h) and i) complete sinus wave with amplitudes: h) k e0 = 0.5 (L/2)/150 = L/600 = 23879 mm/600 = 40 mm, i) e0LT= (L/2)/200 = L/400 = 23879 mm/400 =
59.7 mm, this is the worst shape and, j) half wave e0LT = L/200 = 23879 mm/200 = 119.4.
Structures 29 (2021) 1445–1462
A. Agüero et al. Structures 29 (2021) 1445–1462

5) The 2nd order theory with the shape of the initial imperfection in the

Example 5. Kucukler et al. [20]. Comparisons of the results obtained for different spring stiffnesses. Simply supported beam, L = 7.75 m. Elastic spring at the compression flange in midspan with stiffness (0, 0.2, 0.4, 0.6,

Results according to authors‘ method. Buckling curve “b”, αLT = 0.34, f = 1, β = 0.9, λLT,0 = 0.4 according to clause 6.3.2.3 [1].Elastic spring at the compression flange in midspan with the stiffness kx108 N/mm
form of a quadratic parabola or complete sinus wave or half-wave
sinus with amplitude e0,LT according to 7.3.3.2 in [3].

η0 = 11 mm ≡ ≡ηinit,v(3.875 m) η0 = 14 mm ≡ ≡ηinit,v(3.875 m) η0 = 15 mm ≡ ≡ηinit,v(3.875 m) η0 = 14 mm ≡ ≡ηinit,v(3.875 m) η0 = 14 mm ≡ ≡ηinit,v(3.875 m) η0 = 4 mm ≡ ≡ηinit,v(5.813 m)

More details are given in Table 2 (example 1), Table 3 (example 2),
Table 4 (example 3), where test result [16] also is given, Table 5
(example 4.2), Table 6 (example 5), Table 7 (example 6).

1.0*108 N/mm
Fig. 18. Case 6
The plots of the following parameters were implemented to have a
better understanding of the examples using the authors‘ method and the

5.813 m
second order analysis with an imperfection used according to [1] and

0.628
[3] :

a) Imperfection: the imperfection displacements of the shear centre and

the torsional rotation are plotted along the beam. The values are
divided by the maximum values in order to have the relative
imperfection with the maximum value 1.0.
0.8*108 N/mm

b) Deformations: the displacement and the torsional rotation due to the

imperfection suggested by [1] and [3] are plotted along the beam.
5.657 m

The values are divided by the maximum values in order to have

Case 5

0.592

relative deformations with the maximum value 1.0.

c) The bending moments and bimoments due to the imperfection when
the load level is αb are plotted along the beam. The values are divided
by the maximum value in order to have the relative bending mo­
ments and bimoments with the maximum value 1.0.

d2 η d2 ηcr,θx
0.6*108 N/mm

η0 η
Fig. 17. Case 4

Mz (x) = ( ) EIz cr,v ; B(x) = ( 0 ) EIw (21)

0.8, 1.0)x108 N/mm, IPE 240, γM1 = 1.0, uniform bending moment distribution ψ M = 1.0, My,Ed = My,pl = 86,16 kNm (Figs. 16-19).

dx 2 dx2
αcr αcr
αb − 1 αb − 1
5.503 m
0.546

d) Utilization factors. (i) the partial factor UMy due to the bending
moments about strong y axis; (ii) the partial factor UMz due to the
bending moments about weak z axis, (iii) the partial factor UB due to
the bimoments and (iv) the total factor UMy+Mz+B, show how close
each section is to its maximum strength and also the influence of the
0.4*108 N/mm

partial factors. The total utilization factor UMy+Mz+B = 1.0 is in the

critical section xcr.
5.192 m
Case 3

0.491

UMy+Mz+B (x) = UMy (x) + UMz (x) + UB (x)

⃒ ⃒ ⃒ ⃒
⃒d2 η ⃒ ⃒d 2 η ⃒
EIz ⃒⃒ dxcr,v
2 ⃒
⃒ EIw ⃒⃒ dxcr,θx
2 ⃒
⃒
MyEd ⋅αb η η
= ( )+( 0 ) ( )+( 0 ) ( )
Calculated buckling mode

fy αcr f αcr f
Wy γM0 αb − 1 Wz γM0y αb − 1 WB γM0y

(22)
0.2*108 N/mm

4.728 m

Example No.1. Example 1 (Fig. 5) was solved by Papp [17]. The

Case 2

0.424

description of all the input data and the results of the calculations are
given in Table 2 and accompanying Figs. 6 – 7. The results of the
Eurocode procedure [1] have been verified by the computer program
DRILL [23] using the second order analysis with an imperfection. The
results were practically the same.
In the upper part of Fig. 7, there are deformations (displacement v
Fig. 16. Case 1

and torsional rotation Θ) due to the imperfection plotted along the

beam. Their values are divided by the maximum values in order to have
0.0 N/mm

3.875 m

relative deformations with a maximum value 1.0. In Fig. 7 there are also
Relative buckling resistance αb 0.342

relative bending moments and bimoments due to the imperfection and

the load level responding to αb. The maximum values of all the quantities
Imperfection amplitude [mm]

together with their locations are given in the box.

Example 5.Fig. 16 – Fig . 19

In the lower part of Fig. 7, there are distributions of the partial uti­
Critical section xcr [m]

lization factors U showing the influences of the bending moments My, Mz

Figure No. Case No.

Imperfection shape

and bimoment B together with the total utilization factor U ≈ 1.0. The
Spring stiffness

numerical values of all the utilization factors in the critical section xcr =
[N/mm]

1.293 m are given in the box.

Table 6

Example No.2. Example 2 was solved by Papp [17]. The description of

1452
A. Agüero et al. Structures 29 (2021) 1445–1462

Example 6. Tankova et al. [21]. Comparisons of the results obtained by 7 ways of calculations. Simply supported hot rolled tapered beam, L = 9 m, I-section, hmin = 100 mm, hmax = 200 mm b = 100 mm tf = tw = 10 mm .

torsional imperfection
Theory of 2nd order with amplitude l) Tankova [21]

Buckling mode
including

0.9312
[–]

[–]
according to [1] 5.3.4(3) and [draft

– Complete-wave sinus, without

– j) Case 4, [1], [k) Case 5 [3],

torsional imperfection

η0 = 10 mm ≡ ≡ηinit, – 0.5e0 = 15 mm [e0,LT

– 0.995 [0.953]
= 22.5 mm] Fig. 5. Example 1. Description. MyEd = 32.68 (1 - x/L) kNm.

– 0.0 [1.26]
[3] 7.3.3.2]

Fig. 21;

all the input data and the results of the calculations are given in Table 3
and the accompanying Figs. 8 – 9.
f) Case 3,[1] 6.3.2.3,

Example No.3. Example 3 (Fig. 10) was solved by Wieschollek et al.

EXAMPLE 6Fig. 20 – Fig. 21 a) GMNIA Material imperfection: residual stress patterns for hot Results according to authors‘ method and [results according to EMM* in current

Authors‘ method: buckling mode ηinit,v(x) including torsional imperfection ηinit,θ

[16]. The description of all the input data and the results of the calcu­
f = 1, β = 0.75,

lations are given in Table 4 and the accompanying Figs. 11 – 12.

λLT,0 = 0.4

0.924[ – ]
m)

1.32 [ – ]

Calculations of amplitudes: j) k e0 = 0.5 (L/2)/150 = 0.5 (9000 mm/2)/150 = 15 mm, k) e0,LT = βLT (L/2) /ε = (9000 mm/2)/200/1.0 = 22.5 mm. Example No. 4.1. This example was solved by Gizejowski et al. [19].
v(2.28

* Equivalent Member Method – EMM in [1] is based on experimental tests and numerical investigation without torsional rotation imperfection.

It is similar to example 1 with My = My,pl (1 – x/L). A parametric study

was performed for different relative slenderness λLT using HEB 300 and
Tension flange restrained in the midspan. S235, γM1 = 1.0. Uniform bending moment distribution, My,Ed = 26.094 kNm (Fig. 20–Fig. 21).

steel grade 235. The procedure using (6.3.2.2) was applied to compute
d) Case 2,[1] 6.3.2.3,

χ LT. Fig. 13 shows the values of the buckling strength calculated for the
η0 = 11 mm ≡ ≡ηinit,v(2.28 m) η0 = 12 mm ≡ ≡ηinit,

different values of slenderness λLT according to GMNIA, the authors‘

f = 1, β = 0.9,

method, clause 6.3.2.2 [1], and the second order analysis with an
m)
λLT,0 = 0.4

0.902 [ – ]

1.440 [ – ]

imperfection using [1] and [3].

v(2.280

Example No.4.2. This is similar to example 4.1, but the bending

moment diagram is given by My = My,pl (1 - x/L) - My,pl x/L, L = 23.879
[1] clauses 6.3.2.2 or 6.3.2.3]

b) Case 1, Fig. 20,[1] 6.3.2.2

m. Example 4.2 was solved by Gizejowski et al. [19]. The description of

all the input data and the results of the calculations are given in Table 5
and the accompanying Figs. 14 – 15.
β = 1.0,λLT,0 = 0.2

Example No.5. Example 5 was solved by Kucukler et al. [20]. The

description of all the input data and the results of the calculations are
0.903 [ – ]

1.44 [ – ]

given in Table 6 and the accompanying Figs. 16 – 18. Comparisons of the

x(x) ;

GMNIA relative lateral torsional buckling resistances αb as the function

of different spring stiffnesses k with the results of authors‘ method ob­
tained for (i) β = 0.9, λLT,0 = 0.4 and (ii) β = 0.75, λLT,0 = 0.4 according
to clause 6.3.2.3 [1] are given in Fig. 19.
Dux et al. [24] performed GMNIA calculations in such a way that
incorporated a geometrical lateral bow and initial twist imperfections,
Buckling mode including torsional imperfection

assuming their shapes as a half-sine wave in each laterally unrestrained

span into the finite element models.
According to the plots, the results of Kucukler et al. [20] overlapped
with the GMNIA results (Fig. 19).
Example No.6. Example 6 was solved by Tankova et al. [21]. The
description of all the input data and the results of the calculations are
given in Table 7 and the accompanying Figs. 20 and 21.
L/1000 = 9 mm
rolled sections

4. Discussion of the results

a) The shape of the geometrical initial imperfection. The worst

Relative buckling resistance 0.902

imperfection is defined by the buckling mode, including the torsional

–

imperfection. In some cases higher modes have to be considered

Relative buckling resistance
Geometrical imperfection

[13]. The parabolic shape is worse than the sinus wave shape. In [1]
Imperfection amplitude

Critical section xcr [m]

and [3] guidelines should be given about geometrical initial imper­

fection shapes for boundary conditions different from fork supports,
e.g., for a cantilever (Fig. 3). In simply supported beams with a linear
bending moment ψ = -1 (example 4.2), it can be concluded that a
[mm]
shape
Table 7

complete sinus wave is worse than a half-wave sinus (Table 5).

αb

αb

1453
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 6. Example 1. Case 1. Authors‘ method. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial imperfection ηinit,θx, bending moment Mz, bimoment
B and utilization factors U. Buckling curve “a” and αLT = 0.21 according to clause 6.3.2.2 [1].

Fig. 7. Example 1. Case 7. Critical section xcr, lateral displacement v, torsional rotation θx, bending moment Mz, bimoment B and utilization factors U. Imperfection
with parabolic shape. Amplitude ke0 according to 5.3.4(3) [1]

1454
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 8. Example 2. Case 1. Authors‘ method. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial imperfection ηinit,θx, bending moment Mz, bimoment
B and utilization factors U. Buckling curve “d” and αLT = 0.76 according to clause 6.3.2.2 [1].

Fig. 9. Example 2. Case 4. Critical section xcr, lateral displacement v, torsional rotation θx, bending moment Mz, bimoment B and utilization factors U. Imperfection
with parabolic shape. Amplitude ke0 according to 5.3.4(3) [1]

1455
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 10. Example 3. Description. IPE 120. L = 4.4 m, LF = 1.76 m, FEd = 14.75 kN acting in shear centre.

Fig. 11. Example 3. Case 1. Authors‘ method. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial imperfection ηinit,θx, bending moment Mz,
bimoment B and utilization factors U. Buckling curve “a” and αLT = 0.21 according to clause 6.3.2.2 [1].

b) The amplitude of the geometrical initial imperfection. The value of equivalent welded sections in bending, Eurocode [1] gives the rules
amplitude e0,LT suggested in [3] is greater than the amplitude ke0 in clause 6.3.2.3. It is written in [1]: “The parameters β and λLT,0 and
suggested by [1] and greater than the amplitude of authors‘ method any limitation of validity concerning the beam depth or h/b ratio
which takes into account the torsional imperfection, which is may be given in the National Annex”.
significant. c2) The authors‘ method uses the same parameters as [1] and the
findings are the same as the calculations according to [1]. When the
Further studies should therefore be carried out with different authors‘ method is used, the value αb is the smallest if parameter
boundary conditions to give guidelines for the shape and the amplitude λLT,0 = 0.2 is used; it is greater if λLT,0 = 0.4 and β = 0.9 are used; it is
of the geometrical initial imperfection. There is a great difference be­ the greatest if λLT,0 = 0.4 and β = 0.75 are used. In some cases β =
tween the values of the amplitudes defined in [1] and [3], where the 0.75 leads to an overestimation of αb value, i.e., it is greater than the
amplitude e0,LT may be even 3-times greater than ke0 in [1], especially GMNIA or test value.
for the higher values of the yield strength fy. c3) Eurocode EN 1993-1-1 enables the use of the theory of the 2nd
order with a geometrical initial imperfection with the amplitude ke0
c) The relative buckling resistance αb. defined in [1] or with the amplitude e0,LT defined in the draft [3]. In
c1) For a general case, Eurocode [1] gives the rules in clause 6.3.2.2, example 4.1 (Fig. 13) the relative buckling resistance αb calculated
where λLT,0 = 0.2 and the “hidden parameter” β = 1.0. For rolled or

1456
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 12. Example 3. Case 8. Critical section xcr, lateral displacement v, torsional rotation θx, bending moment Mz, bimoment B and utilization factors U. Imperfection
with half-wave sinus shape in both spans. Amplitude e0,LT according to 7.3.3.2 [3].

In the authors‘ method the parameters from clause 6.3.2.2 or from

clause 6.3.2.3 with β = 0.9 should be used, because β = 0.75 leads in
some cases to values greater than the GMNIA ones.
The relative differences between the αb values and GMNIA values are
given in Fig. 22.

5. Conclusions

The advantages of using the elastic buckling mode and equation (7)
for the design of beams to define the
imperfection shape are:

– The nonlinear analysis with an imperfection given by the elastic

buckling mode is easy to perform.
– The method can be applied to any structural or geometrical type
since a unique global and local imperfection can be obtained from
the buckling mode.
– The imperfection of the members is calibrated to ensure that the
buckling resistance is equal to the one computed according to the
clauses 6.3.2 [1,2], 8.3.2 [3] and 8.3.2 [4].
Fig. 13. Example 4.1. Comparisons of the relative lateral torsional buckling
resistances αb as a function of λLT calculated by 5 different ways.
– At the cross section level buckling is considered as additional dis­
placements and forces that verify the compatibility and equilibrium
according to the authors‘ method using 6.3.2.2 leads to αb values equations. To verify the structure one just has to check it at the cross
smaller than the ones obtained by other methods. The theory of the section level adding the internal forces and moments due to the
2nd order with amplitudes ke0 [1] and e0,LT [3] gives αb values, imperfection to the others, instead of stability checks at the member
which are similar to GMNIA αb values (Fig. 13). This study was level and thereby reducing the strength of the members.
performed for slenderness λLT from 0.6 to 1.2.
– The buckling problem is understood as a global issue and is analyzed
In example 5 (Fig. 19) the values of αb calculated according to the by considering the interaction of all the members of the structure
authors‘ method are closer to the GMNIA values when β = 0.75 is used rather than only by the members under compression or bending,
instead of β = 0.9, which could be used in the National Annex. which means that secondary internal forces will also appear in the
stabilizing beams or tension members.
d) Choice of reduction factor χ LT

1457
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 14. Example 4.2. Case 1. Authors‘ method. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial imperfection ηinit,θx, bending moment Mz,
bimoment B and utilization factors U. Buckling curve “a” and αLT = 0.21 according to clause 6.3.2.2 [1].

Fig. 15. Example 4.2. Case 7. Critical section xcr, lateral displacement v, torsional rotation θx, bending moment Mz, bimoment B and utilization factors U. Imper­
fection with complete-wave sinus shape. Amplitude ke0 according to 5.3.4(3) [1].

1458
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 16. Example 5. Case 1. Authors‘ method. Elastic spring stiffness k = 0.0 N/mm. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial imper­
fection ηinit,θx, bending moment Mz, bimoment B and utilization factors U. Buckling curve “b”, αLT = 0.34, β = 0.9, λLT,0 = 0.4 according to clause 6.3.2.3 [1].

Fig. 17. Example 5. Case 4. Authors‘ method. Elastic spring stiffness k = 0.6x108 N/mm. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial
imperfection ηinit,θx, bending moment Mz, bimoment B and utilization factors U. Buckling curve “b”, αLT = 0.34, β = 0.9, λLT,0 = 0.4 according to clause 6.3.2.3 [1].

1459
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 18. Example 5. Case 6. Authors‘ method. Elastic spring stiffness k = 1.0x108 N/mm. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial
imperfection ηinit,θx, bending moment Mz, bimoment B and utilization factors U. Buckling curve “b”, αLT = 0.34, β = 0.9, λLT,0 = 0.4 according to clause 6.3.2.3 [1].

of metal members in compression only. In this paper the method was

generalised for the members in bending susceptible to lateral torsional
buckling.
The authors‘ method described in section 2 shows how to obtain the
shape and amplitude of UGLI imperfections (displacement and rotation),
the relative buckling resistance αb, and the partial and total utilization
factors U found in critical section xcr.
The authors method can be applied to any type of beams with a
doubly symmetric cross-section since the Unique Global and Local Initial
(UGLI) imperfection can be obtained from the buckling mode. In single
members the imperfection is calibrated to ensure that the buckling
resistance is equal to one computed according to clauses 6.3.2 [1] and
8.3.2 [3].
More than 60 cases have been investigated. The seven examples (1,
2, 3, 4.1, 4.2, 5, 6) have been investigated using 12 different calculation
methods, taking into account thevarious shapes of geometrical imper­
fections and their amplitudes. The GMNIA results, the results of the
calculations according to the current EN 1993-1-1 [1], and its working
draft [3] together with the theory of the 2nd order and the results and
tests of other authors, i.e., (Papp [17], Wieschollek et al. [16], Gize­
Fig. 19. Example 5. Comparisons of GMNIA relative lateral torsional buckling jowski et al. [19], Kucukler et al. [20] and Tankova et al. [21]) have also
resistances αb as function of different values of spring stiffness k with results of been presented.
the authors‘ method obtained for (i) f = 1, β = 0.9, λLT,0 = 0.4 and (ii) f = 1, β = The discussion of the results are summarized in section 4 relating to:
0.75, λLT,0 = 0.4 according to clause 6.3.2.3 [1]. a) the shape of the geometrical imperfections, b) the amplitude of the
imperfections, c) the relative buckling resistance and d) choice of
We would like to stress that under wording “elastic or plastic anal­ reduction factor χ LT.
ysis” in Table 5.1 in [1], Eurocode users should understand a “elastic or
plastic cross-section analysis”. This is now clearly stated in Table 7.1 in
[3]. Declaration of Competing Interest
Eurocodes [1–4] allow also designing members with the method
using the geometrical Unique Global and Local Initial (UGLI) imper­ The authors declare that they have no known competing financial
fection (displacement only). The method is used for the flexural buckling interests or personal relationships that could have appeared to influence
the work reported in this paper.

1460
A. Agüero et al. Structures 29 (2021) 1445–1462

Fig. 20. Example 6. Case 1. Authors‘ method. Critical section xcr, lateral initial imperfection ηinit,v, torsional initial imperfection ηinit,θx, bending moment Mz,
bimoment B and utilization factors U. Buckling curve “d” and αLT = 0.76 according to clause 6.3.2.2 [1].

Fig. 21. Example 6. Case 5. Critical section xcr, lateral displacement v, torsional rotation θx, bending moment Mz, bimoment B and utilization factors U. Imperfection
with half-wave sinus shape. Amplitude e0,LT according to 7.3.3.2 [3].

1461
A. Agüero et al. Structures 29 (2021) 1445–1462

[5] Chladný E, Štujberová M. Frames with unique global and local imperfection in the
shape of the elastic buckling mode (part1). Stahlbau 2013;8:609–17.
[6] Chladný E, Štujberová M. Frames with unique global and local imperfection in the
shape of the elastic buckling mode (Part 2). Stahlbau 2013;9:684–94.
[7] E. Chladný, Štujberová. Errata: frames with unique global and local imperfection in
the shape of the elastic buckling mode. Part 1. Stahlbau 2013;82:H. 8, S. 609–617.
Part 2. Stahlbau 2013;82:H. 9, S. 684–694. Stahlbau 2014;83:H. 1, S.64.
[8] Baláž, I.Determination of the flexural buckling resistance of frames with members
with non-uniform cross-section and non-uniform axial compression forces. Zborník
z XXXIV. Aktívu pracovníkov odboru OK so zahraničnou účast’ou “Teoretické a
konštrukčné problémy ocel’ových a drevených konštruckcií a mostov”. 16.–17. 10;
2008, Pezinok. p. 17–22.
[9] Baláž I, Koleková. Y. Metal frames with non-uniform members and/or nonuniform
normal forces with imperfections in the form of elastic buckling mode. Engineering
research. Anniversary volume honoring Amália and Miklós Iványi. Pollak Milhaly
Faculty of Engineering. Univeristy of Pécs; October 25–26; 2010. p. B:3–B:15.
[10] Baláž, I. Koleková .Y. In plane stability of two hinged arches. In: Proceedings of
European conference on steel and composite structures, Eurosteel 2011. 31. August
– 2. September; 2011. Budapest, vol. C. p. 1869–74.
[11] Baláž, I. Koleková. Y. Structures with UGLI imperfections. In: CD proceedings of
18th international conference engineering mechanics 2012. Svratka, Czech
republic, May 14–17; 2012, paper no. 233. p. 61–86.
[12] Agüero A, Pallarés FJ. Proposal to evaluate the ultimate limit state of slender
structures. Part 1: Technical aspects. Eng Struct 2007;29:483–97.
[13] Agüero A, Pallarés L, Pallares FJ. Equivalent geometric imperfection definition in
steel structures sensitive to flexural and/or torsional buckling due to compression.
Fig. 22. Comparisons of relative lateral torsional buckling resistances αb Eng Struct 2015;96(1):160–77.
[14] Agüero A, Pallarés FJ, Pallares L. Equivalent geometric imperfection definition in
calculated for examples 1, 2, 3, 4.2 and 6 according to 5 different methods. steel structures sensitive to lateral torsional buckling due to bending moment. Eng
Differences [%] from GMNIA results. Struct 2015;96(1):41–55.
[15] Bijlaard F, Feldmann M, Naumes J, Sedlacek G. The “general method” for assessing
the out of plane stability of structural members and frames and the comparison
Acknowledgement with alternative rules in EN1993 – Eurocode3 – Part1-1. Steel Constr 2010;3(1):
19–33.
This research did not receive any specific grant from funding [16] Wieschollek M, Schillo N, Feldmann M, Sedlacek G. Lateral–torsional buckling
checks of steel frames using second-order analysis. Steel Constr 2012;5(2):71–86.
agencies in the public, commercial, or not-for-profit sectors. Authors [17] F. Papp. Buckling assessment of steel members trough overall imperfection
devote the paper to the memory of Prof. Ing. Eugen Chladný, Ph.D. method, Engineering Structures, Vol. 106, 2016, pp. 124− 136.
(01.08.1928–07.07.2020), the author of UGLI imperfection method [18] N.S. Trahair .Flexural–torsional buckling of structures E & FNspon (1993).
[19] Gizejowski, M. Szczerba, R. Gajewski. M. LTB resistance of rolled I-section beams:
used in both metal Eurocodes.
FEM verification of Eurocode’s buckling curve formulation. ce/papers. 2017; 1.
1325-1334. 10.1002/cepa.174.
References [20] Kucukler M, Gardner L, Macorini L. Lateral–torsional buckling assessment of steel
beams through a stiffness reduction method. J Constr Steel Res 2015;109:87–100.
https://doi.org/10.1016/j.jcsr.2015.02.008.
[1] EN 1993-1-1:2005 and Corrigendum AC (2006) and Corrigendum AC (2009) and
[21] Tankova T, Simões da Silva L, Marques L. Buckling resistance of non-uniform steel
Amendment A1 (2014). Eurocode 3: Design of steel structures. Part 1.1: General
members based on stress utilization: General formulation. J Constr Steel Res 2018;
rules and rules for buildings. CEN Brussels.
149:239–56. https://doi.org/10.1016/j.jcsr.2018.07.022.
[2] EN 1999-1-1:2007 and Amendment A1 (2009) and Amendment A2 (2013).
[22] prEN 1993-1-14: 2018 Eurocode 3 - Design of steel structures. Part 1-14: Design
Eurocode 9: Design of aluminium structures.Part 1.1: General structural rules. CEN
assisted by finite element analysis.
Brussels.
[23] Program DRILL, version 2009.150, FIDES DV-Partner GmbH, München, Germany.
[3] prEN 1993-1-1:2020. Eurocode 3: Design of steel structures. Part 1.1: General rules
Author: Prof. Dr.- Ing. Harald Friemann.
and rules for buildings. September 2020, CEN Brussels.
[24] Dux PF, Kitipornchai S. Inelastic beam buckling experiments. J Constr Steel Res
[4] prEN 1999-1-1:2020-07-01. Eurocode 9: Design of aluminium structures. Part 1.1:
1983;3(1):3–9. https://doi.org/10.1016/0143-974X(83)90011-1.
General structural rules.

1462