applied

sciences
Article
Assessment of in-Plane Behavior of Metal
Compressed Members with Equivalent
Geometrical Imperfection
Antonio Agüero 1, * , Ivan Baláž 2 , Yvona Koleková 3 and Pedro Martin 1
1 Department of Continuous Medium Mechanics and Theory of Structures,
Universitat Politècnica de València, c/Camino de Vera s/n, 46022 Valencia, Spain; pmartin@upvnet.upv.es
2 Department of Metal and Timber Structures, Faculty of Civil Engineering,
Slovak University of Technology in Bratislava, 810 05 Bratislava, Slovakia; Ivan.Balaz@stuba.sk
3 Department of Structural Mechanics, Faculty of Civil Engineering,
Slovak University of Technology in Bratislava, 810 05 Bratislava, Slovakia; yvona.kolekova@stuba.sk
* Correspondence: anagra@mes.upv.es




Received: 27 October 2020; Accepted: 15 November 2020; Published: 18 November 2020


Abstract: A new procedure was presented with the objective of proving that it is the generalization
of current attempts in designing compressed members and structures which is able to solve cases
where other authors have problems. It is the further development of the former methods published
by Chladný, Baláž, Agüero et al., which are based on the shape of the elastic critical buckling
mode of the structure. Chladný’s method was accepted by CEN/TC 250 working groups creating
Eurocodes. Both current Eurocodes EN 1993-1-1:2005 and EN 1999-1-1:2007 in their clauses 5.3.2(11)
enable applying the geometrical equivalent unique global and local initial (UGLI) imperfection.
The imperfection has the shape of the elastic critical buckling mode with amplitude defined in
5.3.2(11). UGLI imperfection is an alternative to the global sway and local bow initial imperfections
defined in 5.3.2(3) and to the imperfections described in the clause 5.3.2(6). The determination of
the location and value of UGLI imperfection proved to be onerous by some authors, especially in
cases of members with variable cross-sections or/and axial forces. The paper also provides for special
cases a procedure to detect the critical cross-section along the member which is defined as the one
in which the utilization factor obtains maximum values. The new approach is validated by the
investigation of five complex structures made of steel and one made of aluminum alloy solved by
other authors. Comparisons of the results with those of other authors and with the Geometrically and
Materially Nonlinear Analysis with Imperfections (GMNIA) results showed very good agreements
with negligible differences. The information concerning the differences between current Eurocodes
EN 1993-1-1:2005 and EN 1999-1-1:2007 is provided. Working drafts of Eurocodes of new generation
prEN 1993-1-1:2020 and prEN 1999-1-1:2020 are also commented on.

Keywords: buckling in plane of the structure; UGLI imperfection based on buckling modes; steel;
aluminum alloy; EN 1993-1-1; EN 1999-1-1

Highlights
- A new procedure is presented to obtain the location of the critical section and the value of the
amplitude of unique global and local initial (UGLI) imperfection having the shape of elastic
critical buckling mode. The new procedure is also used for the structures consisting of strongly
irregular members in compression.
- Validation of the procedure by the investigation of five complex structures made of steel and one
made of aluminum alloy solved by other authors.

Appl. Sci. 2020, 10, 8174; doi:10.3390/app10228174 www.mdpi.com/journal/applsci

Appl. Sci. 2020, 10, 8174 2 of 22

- Verification of the procedure by comparisons with results of other authors and Geometrically and
Materially Nonlinear Analysis with Imperfections (GMNIA) results.
- Information about differences in methods given in four Eurocodes: EN 1993-1-1:2005,
EN 1999-1-1:2007, prEN 1993-1-1:2020 and prEN 1999-1-1:2020.

1. Introduction

1.1. Overview and Analysis of Current State

According to Eurocode EN 1993-1-1 [1] and EN 1999-1-1 [2], second order effects and initial
imperfection depending on the type of frame structure may be taken into account in global analysis.
The design of steel structures with compression elements must consider the existence of residual
stresses and geometric imperfections such as initial out-of-plumbness, flatness and fit tolerances.
The negative effects of all these imperfections are covered in the global analysis by equivalent
geometrical imperfection with a relevant shape and value of amplitude. There are the following types
of imperfections in [1]: (i) global initial sway imperfection for the global analysis of frames; (ii) local
initial bow imperfection for global and member analysis; (iii) imperfection for the analysis of bracing
systems; and (iv) unique global and local initial (UGLI) imperfection based on elastic critical buckling
modes. In this paper, the global analysis with UGLI imperfection is utilized.
Section 5.3.2(11) of [1] provides formulae which are the base of a procedure for obtaining UGLI
imperfection. Unfortunately, the procedure in [1] is limited to the structures with a uniform cross-section
under uniform axial force distribution. No iterations are needed in such cases. The critical section will
take place where the curvature has maximum value. The detailed description of the development of
this method is given by Chladný in [3]. The procedure described in [3] is without the above limitations
and may be used for the most practical cases. Generally, the critical section must be found by iterations.
The more general procedure without above limitations is given in EN 1999-1-1 [2]. The new generation
of Eurocodes is in preparation and it should be available to the technical public in 2023. In the
working drafts prEN 1993-1-1 [4] and prEN 1999-1-1 [5] the more general procedure was improved.
The formulation in draft [5] is more useful and clear compared to its former edition [2]. The working
draft prEN 1993-1-1 [4] differs from its former edition [1] and from [2] and [5]. In Eurocode [4],
the partial material safety factor γM1 was removed from the formula for the imperfection amplitude e0
of the equivalent member. The consequence is that the value of imperfection amplitude e0 is in [1,2,5]
greater than in [4].
The working draft Eurocode prEN 1999-1-1:2020 [5] contains the following formulae for:

(i). The characteristic value of the initial imperfection amplitude of the equivalent member related to
the critical section m:

  MRk,m
e0,k,m = α λm − λ0 (1)
NRk,m

(ii). Design value of the initial imperfection amplitude of the equivalent member related to the critical
section m:

2 2
χm λm χ λ
1− γM1   MRk,m 1 − γmM1m
e0,d,m = e0,k,m 2
= α λm − λ0 (2)
1 − χm λm NRk,m 1 − χ λ2
m m

Note that the symbol e0,d,m used in Equation (2) must not be confused with the amplitude of the
initial local bow imperfection eo given in [1] or [2] in Table 5.1.
Appl. Sci. 2020, 10, 8174 3 of 22

(iii). Design value of UGLI imperfection amplitude ηinit,max :

Ncr,m
ηinit,max = e0,d,m   (3)
EIm η00 cr,m 
For the simply supported members with a uniform cross-section and uniform axial force
distribution ηinit,max = e0,d,m . The symbols e0,k,m and ηinit,max are used here for the better understanding
of the procedure. They are not explicitly used in Eurocodes [1,2,4,5].
The differences of Eurocodes [1,2,4] compared with the working draft Eurocode prEN
1999-1-1:2019 [5] are as follows:

(i). Current Eurocode EN 1993-1-1:2005 [1] contains the above Equations (2) and (3) but in
Corrigendum AC (2009) the index “d“ was removed and some print errors were corrected.
(ii). Current Eurocode EN 1999-1-1:2007 [2] contains the above Equations (2) and (3) but in Amendment
A2 (2013) [2] there are several changes and index “d“ was removed.
(iii). Working draft Eurocode prEN 1993-1-1:2020 [4] removed from the above Equations (2) and (3) the
partial material safety factor γM1 and index “d”. Removing γM1 from Equation (2) destroyed the
main assumption of Chladný’s method defined below, but the code committee followed this idea
because it leads to an ease of use in practical applications [6]. The authors intend to publish the
results of the study showing the consequences of this decision leading to an important difference
between the values of UGLI imperfection amplitude calculated according to [4] or [5].

Eurocodes use the symbol ηinit,max for the design value of UGLI imperfection amplitude. In the
paper, the symbol η0 = ηinit,max is also used, which is not in Eurocodes. The symbol η0 is used in the
text, Equations and Tables because it is shorter than ηinit,max . The critical section is indicated in this
paper by symbol xcr and not by index “m“ as it is done in Eurocodes. Equation (4) can be rearranged in
the form of Equation (5):

Ncr,m
ηinit,m (x) = e0,d,m   ηcr (x) = ηinit,max ηcr (x) (4)
EIm η00 cr,m 

where χ is the reduction factor for the relevant buckling curve depending on the relevant cross-section,
see 6.3.1.
Equation (4) is used in Eurocodes in the clauses: 5.3.2(11) [1], 5.3.2(11) [2], 7.3.6(1) [4] and
7.3.2(11) [5]. Equation (5) is just another form of (4). It is more convenient for programming.
When flexural buckling occurs about the strong axis, UGLI imperfection may be written in the form:
  2 
 λ χ
 α λ y − λ0 1 − γy y
 
M1
fy 
ηinit,w (x) =   ηcr,w (x) = η0 ηcr,w (x) (5)
 
2 2

I y d2 ηcr,w 

λy 1 − λ y χ y E· W
 
2
 
y dx xcr

When flexural buckling occurs about the weak axis, UGLI imperfection is as follows:
  2 
λz χz

 α λz − λ0 1 − γM1 fy
 

ηinit,v (x) =  2 2
  η (x) = η η (x)
 cr,v 0 cr,v (6)
Iz d ηcr,v 
2

 λz 1 − λz χz E· W 2

z dx xcr

The method with UGLI imperfection was developed by Chladný and published with necessary
details and applications in [3,7,8]. The first applications of Chladný’s method may be found in [9],
where two examples were solved: a portal frame and a member with a non-uniform cross-section.
In the second edition [10], the same two examples as in [8] may be found together with an example
of the steel structure of three-story building. Baláž, who participated in developing Chladný’s
Appl. Sci. 2020, 10, 8174 4 of 22

method (see pages 693–694 [7]), derived this method in a different way and published the details of
a step-by-step procedure for non-uniform members in [11]. Baláž and Koleková published detailed
illustrative numerical examples in [12,13]. They investigated in [14] the in-plane stability of the large
two-hinged arch bridge (Žd’ákov bridge in Czech Republic). Baláž showed that for the uniform
members, UGLI imperfection amplitude has important geometrical meaning, which enables doing
calculation and results verification by “hand calculation”. The importance of so called “hand calculation”
was illustrated in several examples in [15]. In [3,7,8], Chladný solved in detail a member with a
non-uniform cross-section and three-story building as in [8] and three new examples relating to the
stability of parts of bridge structures: (i) the bottom flange of the continuous composite steel and
concrete road bridge; (ii) the upper chord of the steel railway half-through truss bridge; and (iii)
the basket handle type of the arch bridge. Numerical examples in [3,7,8] provide guidance of the
proper application of Chladný’s method. Chladný showed there the good agreement with other
methods of Eurocode EN 1993-1-1:2005 described in clauses 5.2.2(3) and 5.3.2(3). The guidance for
the design of members of the steel basket handle arch bridges is given in the clause NB.3 of Slovak
National Annex to Eurocode STN EN 1993-2 Design of steel structures—Part 2 Steel bridges. Chladný
pointed out there that it is not enough for a such type of bridges to take into account only the first
buckling mode. The members of such bridges are in biaxial bending. Chladný in [3,7,8] generalized his
method for members when flexural buckling occurs about both axes. For a member in biaxial bending,
UGLI imperfection may be defined as follows:
  2 
λ χ

ηinit,v (x)  α λ − λ0 1 − γM1 fy ηcr,v (x) ηcr,v (x)
( )   ( ) ( )

=   = η0 (7)
ηinit,w (x) ηcr,w (x) ηcr,w (x)

2 2

d2 ηcr,v d2 ηcr,w 
 
Iy
 λ 1 − λ χ E· WIz + Wy

z dx2 dx 2
xcr

The new methodology has been developed in recent years in different works. Generalization for:
(i) flexural torsional buckling of member due to compression was given by Agüero et al. in [16,17]:

ηinit,v (x) ηcr,v (x) ηcr,v (x)

   2
    
λ χTF
 α(λTF −λ0 ) 1− TF
       
fy
     
γM1
η η η η
      
init,w (x) =  cr,w (x) = cr,w (x) (8)
     
0

 λ2TF 2 d2 ηcr,θx 
!
Iz d2 ηcr,v I y d2 ηcr,w
     
1−λTF χTF Iw
    
 η
  E· Wz dx2 + W y dx2 + WB
  η
   η
 
init,θx (x) dx2 cr,θx (x) cr,θx (x)
  
xcr

and for (ii) lateral torsional buckling of member due to bending moment by Agüero et al. [16,18]:
 2 ! 
λLT χLT
ηinit,v (x)  αLT (λLT −λLT,0 ) 1− γM1 ηcr,v (x) ηcr,v (x)
( )   ( ) ( )
fy
= η0

=   (9)
ηinit,θx (x) 2 d2 ηcr,θx  ηcr,θx (x) ηcr,θx (x)
  !
2
λLT 1−βλLT χLT E· Iz d2 ηcr,v Iw
Wz dx2 + WB
 
dx2
xcr

The method with UGLI imperfection was also used by two Baláž’s Ph.D. students: Kováč [19]
and Dallemule [20], where other numerical examples may be found. Dallemule created a general
computer program based on Chladny’s method. Bijlaard et al. [21] and Wieschollek et al. [22] have
also generalized the equations given in [1] for the lateral torsional buckling of a member in bending,
and Papp [23] investigated the buckling of member under bending and compression.
Brodniansky [24], another Ph.D. student from Bratislava, used a modified Dallemule’s computer
program and pointed out some numerical obstacles when calculating the location of the critical section
xcr for the column with step change in cross-sectional parameters and step change in normal force
along the members.

1.2. Research Significance, Contributions and Basic Assumptions of Presented New Method
The innovation of the presented research is the numerical method for obtaining the equivalent UGLI
imperfection amplitude for metal members susceptible to buckling due to axial forces. The method is
Appl. Sci. 2020, 10, 8174 5 of 22

consistent with the Eurocode procedure and allows to obtain the location of critical section xcr and
consequently UGLI imperfection amplitude for any case.
Several assumptions are considered in this work:

(i). The first assumption is that the first buckling mode of the examined structural member is
dominant and therefore the effect of the higher buckling modes can be neglected. However,
in some cases, the higher buckling modes have to be considered in the design. See suggestions by
Agüero et al. [17].
(ii). The second assumption is that Chladný’s method given in [3] remains valid. Quoting Chladný [3],
where experimentally established values are unavailable, the amplitude of the equivalent
imperfection in the shape of the elastic buckling mode may be determined assuming that the
buckling resistance of a structure with axially loaded members shall be equal to the buckling
resistance of the equivalent member. The buckling resistance of axially loaded columns is
defined in clauses 6.3.1.1–6.3.1.4 [1]. The relative slenderness λ relates to the critical section.
The equivalent member has pinned ends, its cross-section and axial force are the same as in the
critical cross-section m of the frame and its length is such that its critical force equals the axial
force in the critical cross-section m at the critical loading of the structure. The position of the
critical cross-section m is determined by the condition that the utilization Um , with allowance for
the effect of the axial force and bending moments due to imperfections in the critical cross-section
m, is greater than the utilization U(x) at all other cross-sections of the verified member or
frame structure.
(iii). The third assumption is that the linear interaction Formula (22) is used to verify the member
according to Eurocodes and [25,26].
(iv). The fourth assumption is that if plastic cross-section resistance is taken into account, a maximum
plastic shape factor of 1.25 may be taken into account [4,26]. This means that if the criterion is
applied with reference to the plastic resistance, the design value of the moment resistance Mc,Rd
should be limited to 1.25 Mel,Rd for both a strong axis and weak axis.
(v). The fifth assumption accepts the common method for finding the equivalent local bow
imperfections given in [1] 5.3.2 (3). For the new generation of Eurocodes, intensive investigations
were done to enhance these values, now in draft [4] see 7.3.3.1 (1) [25,26]. In practice, we often
deal with members under compression and bending. If the external bending moments are present,
the results may be more unfavorable, because a larger part of the cross-section in longitudinal
direction may be plastified, thus leading to a stiffness reduction and therefore higher necessary
bow imperfections for simplified calculations. If this is neglected, the load carrying capacity may
be underestimated. In our cases of centrally loaded members, there is no need to use the fourth
and fifth assumptions.

2. Method for Obtaining UGLI Imperfection Amplitude

According to Eurocodes [1,2], Equation (4) shall be used to obtain the value of the initial
imperfection amplitude of the equivalent member. The following method is presented when buckling
about the y axis takes place. This can be applied in a similar way for buckling about the z axis.
The presented method and Chladny’s method as well may be used in both cases: when buckling about
the y axis and z axis takes place.
The buckling shape has been scaled to have a maximum value of 1.0, i.e., max(ηcr,w ) = 1.0.
The meaning of η0 is the amplitude of UGLI imperfection. The flow chart is shown in Figure 1.
Appl. Sci. 2020, 10, 8174 6 of 22
Appl. Sci. 2020, 10, x FOR PEER REVIEW 6 of 24

Figure 1.
Figure 1. Flow
Flow chart
chart describing
describing the
the procedure
procedure onon how
howto
toobtain
obtainthe
thelocation
locationof ofthe
thecritical
criticalsection
sectionxxcr
cr
and the unique global and local initial (UGLI) imperfection amplitude η 0. Illustrative example.
and the unique global and local initial (UGLI) imperfection amplitude η0 . Illustrative example.

Steps
Steps for
for Obtaining
Obtaining UGLI
UGLI Imperfection
Imperfection:
The
The first
first step
stepisisobtaining
obtainingthe thebuckling loadααcrcr and
bucklingload andbuckling
bucklingshape
shape{η{ηcrcr}.}. They can be
be computed
computed
using
using finite
finite element
element method,
method,for forinstance,
instance,according
accordingto toTrahair
Trahair[27].
[27].
The
The second
second stepstep is
is the
the calculation
calculation of
of bending
bending moments
moments and and stresses
stresses related
related to to the
the buckling
buckling mode
mode
as
as follows:
follows:
d2 ηcr,w
M y,η = −EI y d 2η2 (10)
M = − EI dx cr ,w
y ,η y
(10)
d2 ηcr,wdx 2
EI y dx2
σM = (11)
d yηcr , w
W
2

EI y (11)
The third step is the first iteration. σM = dx 2
Initial guess: Wy
The third step is the first iteration.
A· f y −
r
αult,1 αult,1 ·χ1
Initial guess: αult,1 = min( ) → λ1 = → χ1 → αb,1 = (12)
NEd αcr γM1
A· f
y
− α α ·χ
α ult ,1 may
Another initial guess = min( ) → λ1 in
be considered ,1
→ χ1where
= theultlocation → α b ,1the ult ,1 1
= stress calculated according
(12)
Equation (11) has a maximum value: Ed N α cr γ M1
The imperfection
Another (scale
initial guess factor
may Ω1 (x)) needed
be considered at each
in the section
location in order
where to reach
the stress the cross-section
calculated according
resistance when
Equation (11) the
has buckling load
a maximum level is reached αb,1 is calculated from formula:
value:
The imperfection (scale factor Ω1(x)) needed at each section in order to reach the cross-section
d2 ηcr,w
αb,1 is reached
NEd ·level
resistance when the buckling load Ω1 (x) αEIb,1 yis dx
calculated
2 ffrom
y formula:
+ = (13)
γM0

A αcr Wy
α − 1
b,1
Appl. Sci. 2020, 10, 8174 7 of 22

fy NEd · αb,1 αcr Wy

! !
Ω1 (x) = − −1 (14)
γM0 A αb,1 EI
d2 ηcr,w
y dx2

The minimum of those scale factors is the objective in the first iteration, i.e., η0,1 takes place at the
critical section xcr,1 :
η0,1 = min(Ω1 (x)) = Ω1 (xcr,1 ) (15)

The fourth step is the second iteration:

αult,2 αult,2 ·χ2

r
A(xcr,1 )· f y (xcr,1 ) −
αult,2 = → λ2 = → χ2 → αb,2 = (16)
NEd (xcr,1 ) αcr γM1

The imperfection (scale factor Ω2 (x)) needed at each section in order to reach the cross-section
resistance when the buckling load level is reached αb,2 is calculated as follows:

fy NEd · αb,2 αcr Wy

! !
Ω2 (x) = − −1 (17)
γM0 A αb,2 EI
d2 ηcr,w
y dx2

η0,2 is the amplitude at the second iteration. It takes place at the critical section xcr,2 :

η0,2 = min(Ω2 (x)) = Ω2 (xcr,2 ) (18)

If the critical section is the same as in the previous iteration, the critical section is found and
consequently the value of UGLI imperfection amplitude may be calculated. If the critical section is
different, then the other iteration is needed starting from the fourth step. Finally, the distribution of
UGLI imperfection ηinit (x) is calculated:

ηinit = η0 ηcr
 
(19)

3. Validation of Presented Procedure

Three Chladný’s complex examples [3,7,8] are very interesting in order to validate the presented
method. The fourth example shows that differences are negligible between αb values calculated by
the new procedure and αb values obtained by Marques et al. [28] with the help of Geometrically and
Materially Nonlinear Analysis with Imperfections (GMNIA).
Plots of the following parameters are implemented to have a full understanding of the examples:
- The relative UGLI imperfection is plotted along the beam. The values are divided by the maximum
value of imperfection in order to obtain the relative UGLI imperfection with maximum value 1.0.
- The distribution of relative bending moments due to UGLI imperfection is plotted along the beam,
the values are divided by the maximum bending moment value in order to have a maximum
value 1.0:

η0 d2 ηcr,w
EI y (20)
(αcr − 1) dx2
- The distribution of the relative shear forces due to UGLI imperfection is plotted along the beam.
The values are divided by the maximum shear force in order to have a maximum value 1.0:

η0 d3 ηcr,w
EI y (21)
(αcr − 1) dx3
- The plot of utilization factors U distribution: due to axial forces UN , due to bending moments
UM and the global utilization factor UN+M . This factor shows how close each section is to its
maximum strength and the influence of axial forces and bending moments:
 d ηcr ,w
   EI y 
 N Ed   η0 dx 2

U N +M = (U N ) + [U M ] = + (22)
  f y    (α cr − 1)  fy  
 A     Wy  
Appl. Sci. 2020, 10, 8174
  γ M 0     γ M 0  
8 of 22
- The value of the scale factor distribution along the member is plotted showing the first, the
second and the last iteration. The minimum value of each plot is η0,i. The maximum value shown
is 10 times η0:    
d2 ηcr,w 
  η0 EI y dx2 
  
 N
UN+M = (UN ) + [UM ] =   Ed   +  (22)
 f y NEd ⋅α A f 
b,i  y α
 W  
α

( − 1 ) fy 
Ωi x =  ( ) − ⋅
γM0 −1?
cr  cr y
W
 
y γM0
γ A α  d 2η (23)
 M0   b ,i  EI cr , w
y 2
- dxshowing
The value of the scale factor distribution along the member is plotted the first, the second
and the last iteration. The minimum value of each plot is η0,i . The maximum value shown is
10 times η0 : { ( )}
η0,i = min Ωi x = Ωi xcri ( ) (24)

The values of UGLI imperfection amplitude η0, the location of critical section xcr and other values
fy Ed · αb,i αcr in brackets. W y Very good agreements were
! !
have been compared with Chladný’s values N
which are given
Ωi (x) = − · −1 · (23)
achieved. In the examples No. 3 and γ4,M0the influence A of αthe
b,i localEI d2 ηcr,w is neglected. It is beyond
buckling
y dx2
the scope of this paper. The authors have solutions of these examples which also take into account
local buckling. They will be published η0,i with
= minallthe
Ωi (xnecessary
) = Ωi (details

xcri ) in another paper. (24)
Example 1:
The
The values
example of UGLI imperfection
1 is described in Figure 2. It isη0the
amplitude , the location
scheme of of critical
the bottom section
flange xcrofand
theother values
continuous
have been compared
composite with Chladný’s
steel and concrete road bridge.valuesMore
which are given
details aboutinthe brackets.
bridge Very may be good agreements
found were
in Section 5.4
achieved. In the Examples 3 and 4, the influence
[7]. Steel grade S355 and γM1 = 1.1 are taken into account. of the local buckling is neglected. It is beyond the
scopeThe
of this paper.
bottom The of
flange authors have
the steel solutions of in
cross-section these examples is
compression which also take
subjected into account
to lateral local
buckling. It
buckling. They will be published with all the necessary details in another
may be analyzed by modeling the elements as a column subjected to the compression force NEd and paper.
supported by continuous and discrete elastic restraints modeled as springs (Figure 2). The springs
Example TheC1
stiffnesses1.are Example 1 is described
= 10 kN·mm −1, C2 =in C3Figure
= C4 2.= 4It kN·mm
is the scheme
−1. The of elastic
the bottom flangedue
restraint of thetocontinuous
the cross-
composite
section web is c = 5.863 kN·m . The axial force distribution is given by formula NEd (x) = −90005.4
steel and concrete road
−2 bridge. More details about the bridge may be found in Section [7].
1062 x
Steel grade
– 17.49 S355 and γM1 = 1.1 are taken into account.
x2 (kN).

Figure 2. Example
Figure 2. Example 1: description taken
1: description taken from
from [7],
[7], Section
Section 5.4.
5.4.

The bottom flange of the steel cross-section in compression is subjected to lateral buckling. It may be analyzed
by modeling the elements as a column subjected to the compression force NEd and supported by continuous
and discrete elastic restraints modeled as springs (Figure 2). The springs stiffnesses are C1 = 10 kN·mm−1 ,
C2 = C3 = C4 = 4 kN·mm−1 . The elastic restraint due to the cross-section web is c = 5.863 kN·m−2 . The axial
force distribution is given by formula NEd (x) = −9000 1062 x − 17.49 x2 (kN).
The distributions of UGLI imperfection, bending moment M(x) and shear force V(x) due to UGLI
imperfection and utilization factors U(x) are plotted in Figure 3. The critical section takes place after several
iterations at xcr = 1.44 m. The significant step in shear force distribution V is located where the column is
supported by the spring with the greatest stiffness. The scale factor is plotted in Figure 4. The minimum value at
the last iteration gives the amplitude of the imperfection η0 = 39 mm. Chladný’s values in the brackets confirm
the very good agreements of the results (Figure 3). According to the direct method [23] the critical section takes
place where the second order effects have a maximum. This method gives a value of amplitude η0 = 29 mm which
differs from 39 mm a lot. In all the other examples, the direct method gave similar results as Eurocode’s method.
 several iterations at xcr = 1.44 m. The significant step in shear force distribution V is located where the
column is supported by the spring with the greatest stiffness. The scale factor is plotted in Figure 4.
column is supported by the spring with the greatest stiffness. The scale factor is plotted in Figure 4.
The minimum value at the last iteration gives the amplitude of the imperfection η0 = 39 mm.
The minimum value at the last iteration gives the amplitude of the imperfection η0 = 39 mm.
Chladný’s values in the brackets confirm the very good agreements of the results (Figure 3).
Chladný’s values in the brackets confirm the very good agreements of the results (Figure 3).
According to the direct method [23] the critical section takes place where the second order effects
According to the direct method [23] the critical section takes place where the second order effects
have a maximum. This method gives a value of amplitude η0 = 29 mm which differs from 39 mm a
have a maximum. This method gives a value of amplitude η0 = 29 mm which differs from 39 mm a
Appl. Sci. 2020,lot.
10,In
8174
all the other examples, the direct method gave similar results as Eurocode’s method. 9 of 22
lot. In all the other examples, the direct method gave similar results as Eurocode’s method.

Figure 3. Example 1: UGLI imperfection ηinit, bending moment M, shear force V, utilization factor U.
Figure 3. Example
Figure 3. Example
Values in1: UGLI1:
brackets imperfection
are from [7]. ηinitη,initbending
UGLI imperfection , bending moment
moment M,M,
shear forceforce
shear V, utilization factor U. factor U.
V, utilization
Values in brackets are from [7].
Values in brackets are from [7].

Figure 4. Example
Figure 4. Example
Figure 4. 1: the scale
Example 1: thefactor Ω(x)Ω(x)
1: the scale factor
scale factor
Ω(x) for utilizationfactor
for for utilization factor UU
utilization N+M
N+M = 1.0=in1.0
factor UN+M = 1.0 in bay
bay
No.1.
inNo.1. In
bay In other bays,
No.1.
other In other bays,
bays,
the scale factor is greater than 10 times the minimum value.
the scale factor is greater
the scale factor is than
greater10than
times the the
10 times minimum
minimumvalue.
value.

The value η0 = 39 mm given in Figure 4 was obtained after seven necessary iterations. Table 1 shows all
the details of the iterations.

Table 1. Example 1: key values for each iteration.

− η0,i
Iteration αult λ χ αb xcr,i η0

1 1.2228 0.4557 0.8106 0.9011 1.9200 1.6435

2 1.5532 0.5136 0.7698 1.0870 1.0200 0.7413
3 1.3762 0.4835 0.7910 0.9896 1.6800 1.2012
4 1.5023 0.5051 0.7757 1.0594 1.3200 0.8779
5 1.4311 0.4930 0.7842 1.0203 1.5000 1.0586
6 1.4660 0.4990 0.7800 1.0396 1.4400 0.9702
7 1.4542 0.4970 0.7814 1.0331 1.4400 1.0000
 4 1.5023 0.5051 0.7757 1.0594 1.3200 0.8779
5 1.4311 0.4930 0.7842 1.0203 1.5000 1.0586
6 1.4660 0.4990 0.7800 1.0396 1.4400 0.9702
7 1.4542 0.4970 0.7814 1.0331 1.4400 1.0000
Appl. Sci. 2020, 10, 8174 10 of 22

Example 2:
The example
Example 2 is described
2. The Example in Figure
2 is described 5. It is a5.schema
in Figure of the upper
It is a schema of the chord of the of
upper chord steel
the railway half-
steel railway
through truss
half-through bridge.
truss More
bridge. Moredetails
detailsabout
about the bridgemay
the bridge maybebe found
found in Section
in Section 5.5 [7].
5.5 [7]. Steel
Steel gradegrade
S235S235
and
and γM1 = 1.1 are taken into account.
γM1 = 1.1 are taken into account.

Figure 5. Example
Figure 5. Example 2:
2: description taken from [7] Section 5.5.
5.5.

The
Thetruss
trusschord
chordunder
under compression
compression that that
is subjected to theto
is subjected lateral buckling
the lateral may be analyzed
buckling by modeling
may be analyzed by
the elementsthe
modeling aselements
a column as subjected
a column to the compression
subjected to theforce NEd and supported
compression force NEd by anddiscrete elasticbyrestraints
supported discrete
modeled as springs.
elastic restraints The springs
modeled stiffnesses
as springs. The are C0 =stiffnesses
springs 11 MN/m are andCC 0 ==115.068
MN/m MN/m.
and CInput
= 5.068 values
MN/m. for
the iterative procedure [7] are given in Figure 5. In the two edge fields
Input values for the iterative procedure [7] are given in Figure 5. In the two edge fields on both sides on both sides of the truss chord:
of=the
A 0.01909 2 , I = 96.4 × 10−6 m42, W = 459 cm3−6 A = of m2 ,
truss mchord: A = 0.01909 m , I = 96.4 × 10 , in mthe
4, W four middle
= 459 cm3,fields
in the of the
fourtruss chord:
middle fields 0.02898
the truss
= 177.525
Ichord: A = 0.02898
−6
× 10 mm,2,W 4
I ==177.525
3
845 cm ×. 10 The
−6 mloading
4, W =is defined
845 cm3. inTheFigure
loading5. is defined in Figure 5.
The
Thedistributions
distributions of UGLI
of UGLI imperfection,
imperfection,bending momentmoment
bending M(x) andM(x) shearandforceshear
V(x) force
due to V(x)
the imperfection
due to the
and the utilization factors U(x) are plotted in Figure 6. The steps in the
imperfection and the utilization factors U(x) are plotted in Figure 6. The steps in the shear shear force distribution are due to steps in
force
the axial forces and in spring supports. The critical section
distribution are due to steps in the axial forces and in spring supports. takes place at x cr = 13.15 and 18.25 m simultaneously
The critical section takes place
due
at xtocr =the symmetry
13.15 and 18.25 of themproblem. The scale due
simultaneously factortoisthe
plotted in Figure
symmetry of 7.theThe scale factor
problem. Theminimum
scale factorled tois
Appl.
UGLISci. 2020, 10, x
imperfectionFOR PEER REVIEW
amplitude η = 9 mm. The significant values are shown
plotted in Figure 7. The scale0 factor minimum led to UGLI imperfection amplitude η0 = 9 mm. The in Table 2. Only one 11 of
iteration 24
was
needed.
significantThe comparison
values arewith shownChladný’s
in Tablevalues 2. given
Onlyinone brackets showswas
iteration veryneeded.
good agreement.
The comparison with
Chladný’s values given in brackets shows very good agreement.

Figure
Figure 6. Example
6. Example 2: UGLI
2: UGLI imperfection ηinitη,init
imperfection , bending
bending moment M, M,
moment shear
shear force
force V, and
V, and utilization
utilization factor
factor
U. Values in brackets are from
U. Values in brackets are from [7].[7].
 Figure 6. Example 2: UGLI imperfection ηinit, bending moment M, shear force V, and utilization factor
U. Values in brackets are from [7].
Appl. Sci. 2020, 10, 8174 11 of 22

Appl. Sci. 2020, 10, x FOR PEER REVIEW 12 of 24

Table 2. Example 2: key values for each iteration.

Figure
Figure 7. Example
7. Example 2: 2:
thethe scale
scale
− factor Ω(x).
factor Ω(x). Utilization
Utilization factor
factor η0,==i 1.0.
UN+M
UN+M 1.0.
Iteration αult λ χ αb xcr,i
Table 2. Example 2: key values for each iteration. η0
1 1.9376 − 0.6777 0.7961 1.4023 13.2273 1 η0,i
Iteration αult λ χ αb xcr,i η0

Chladný used 1 the method

1.9376 also 0.6777 0.7961
for more complicated 1.4023 13.2273
cases, where 1 method is not
the simplified
applicable, e.g., for the Maria Valeria Bridge across the Danube, Štúrovo–Esztergom. See Note 5 in
[7] referring
Chladnýto [29].
used the method also for more complicated cases, where the simplified method is not applicable,
According
e.g., for the MariatoValeria
the direct method,
Bridge theDanube,
across the same results are obtained. See Note 5 in [7] referring to [29].
Štúrovo–Esztergom.
Example 3:
According to the direct method, the same results are obtained.
The example 3 is described in Figure 8, with steel grade S355 and γM1 = 1.0
Example 3. The Example 3 is described in Figure 8, with steel grade S355 and γM1 = 1.0

Figure 8. Example 3: description taken from [9,10] and [7] Section 5.2.
Figure 8. Example 3: description taken from [9,10] and [7] Section 5.2.
The member is fabricated from an IPE 400 section. It was cut in two pieces and consequently welded
together
Theinmember
such a way that the depths
is fabricated from of its
an cross-section at theItends
IPE 400 section. wasarecut
560inand
two240 mm. and consequently
pieces
welded Thetogether
distributions of UGLI
in such a wayimperfection, bendingofmoment
that the depths M(x) and shear
its cross-section at theforce
endsV(x)
aredue
560toand
the imperfection
240 mm.
and the utilization factors U(x) are plotted in Figure 9. The critical section is
The distributions of UGLI imperfection, bending moment M(x) and shear force V(x) duefound at x cr = 12 m. Thetoscale
the
factor minimum led to UGLI imperfection amplitude
imperfection and the utilization factors U(x) are plotted η 0 = 36.31
in Figure 9. The critical section is foundfor
mm (Figure 10). The significant values at all
xcr
=iterations
12 m. The arescale
shown in Table
factor 3. Onlyled
minimum twotoiterations were needed. amplitude η0 = 36.31 mm (Figure 10). The
UGLI imperfection
significant values for all iterations are shown in Table 3. Only two iterations were needed.
 Appl. Sci. 2020, 10, x FOR PEER REVIEW 13 of 24

Appl.
Appl. Sci.
Sci. 2020,
2020, 10,10, x FOR PEER REVIEW
8174 1213
ofof
2224

Figure 9. Example 3: UGLI imperfection ηinit, bending moment M, shear force V, and utilization factor
U. Values in brackets are from [7,10].
Figure 9.9.Example
Figure Example3:3:UGLI
UGLIimperfection ηinit
imperfection ηinit, ,bending momentM,
bendingmoment M,shear forceV,V,and
shearforce andutilization
utilizationfactor
factor
U.U.
Values in brackets are from [7,10].
Values in brackets are from [7,10].

Figure
Figure 10. 10. Example
Example 3: the
3: the scale
scale factor Ω(x).
factor Ω(x). Utilization
Utilization factor
factor UN+M==1.0.
UN+M 1.0.

Figure 10. Example 3: the scale factor Ω(x). Utilization factor UN+M = 1.0.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 14 of 24

Table 3. Example 3: key values for each iteration.

Appl. Sci. 2020, 10, 8174 13 of 22

− η0,i
Iteration αult λ χ αb xcr,i
Table 3. Example 3: key values for each iteration. η0
1 1.3674 −0.7834 0.7346 1.0045 11.95 1.2343 η0,i
Iteration αult λ χ αb xcr,i η0
2 1.4777 0.8144 0.7155 1.0574 12 1
1 1.3674 0.7834 0.7346 1.0045 11.95 1.2343
Chladný’s values
2 are in brackets 0.8144
1.4777 and they confirm
0.7155 the very good agreement
1.0574 12 of the
1 results (Figure
9).
The direct method
Chladný’s values aregives the value
in brackets andof imperfection
they confirm the amplitude η0 = 34.56of
very good agreement mmthewhich
resultsapproximately
(Figure 9).
equalsThe value 36 mm (Figure 10).
direct method gives the value of imperfection amplitude η0 = 34.56 mm which approximately equals
valueExample 4
36 mm (Figure 10).
Example 4.1
The example
Example 4. (a) The4 was published
Example by Marques
4 was published et al. [28],
by Marques with
et al. steel
[28], grade
with steel235,
gradeand γM1
235, γM1It=is1.0.
= 1.0.
and a
tapered column
It is a tapered similar
column to that
similar toin Figure
that 8 but8made
in Figure fromfrom
but made IPE IPE200,200,
withwith
different boundary
different boundary conditions.
conditions.
The
The linearly varyingheight
linearly varying heightisisdefined
definedbybythethe taper
taper ratioratio of h/hmax
of hmax min = 3. It is simply supported at both ends,
/hmin = 3. It is simply supported at both
L = 12.9
ends, L = 12.9 m. The
m. The membermember is restrained
is restrained againstagainst out-of-plane
out-of-plane buckling. buckling. The compression
The compression = 500NkN
force NEd force Ed

=was
500applied.
kN wasThe computed load factor αb = 1.004 with GMNIA. The method proposed by the authors leads to
applied. The computed load factor α b = 1.004 with GMNIA. The method proposed by
the
αb authors
= 0.991. leads to αb = 0.991.
The difference The difference is −1.29%.
is −1.29%.
In
Inexample 4.1, GMNIA
Example 4(a), GMNIA was wasperformed
performed by Marques
by Marques et al.
et al. [28] [28]ABAQUS
using using ABAQUS software
software taking intotaking
account
into
a geometrical imperfection with the shape of the buckling mode and an amplitude of L/1000, also considering of
account a geometrical imperfection with the shape of the buckling mode and an amplitude the
L/1000,
residualalso considering
stresses of a weldedthecross-section.
residual stresses of a welded cross-section.
The
The distributions
distributions ofofUGLI
UGLI imperfection,
imperfection, bending
bending moment moment
M(x) and M(x) and
shear shear
force V(x)force
due toV(x)
the due to the
imperfection
imperfection and the utilization factors U(x) are plotted in Figure 11. The critical section
and the utilization factors U(x) are plotted in Figure 11. The critical section is found at xcr = 10.2 m. The scale is found at
xfactor
cr = 10.2 m. The scale
minimum led to factor
UGLI minimum
imperfectionled to UGLIη0imperfection
amplitude amplitude
= 28.04 mm (Figure 12). ηThe
0 = 28.04 mm (Figure
significant 12).
values for all
The significant values for all iterations are shown
iterations are shown in Table 4. Three iterations were needed. in Table 4. Three iterations were needed.

Figure11.
Figure Example
11.Example 4.1:4(a):
UGLI UGLI imperfection
imperfection ηinit , bending
ηinit, bending moment
moment M, force
M, shear shearV,force V, utilization
utilization factor
factor U. Value (1.004) is obtained by Geometrically and Materially Nonlinear Analysis
U. Value (1.004) is obtained by Geometrically and Materially Nonlinear Analysis with Imperfections with
Imperfections
(GMNIA). (GMNIA).
Appl. Sci. 2020, 10, x FOR PEER REVIEW 15 of 24
Appl. Sci. 2020, 10, 8174 14 of 22

Figure 12.Example
Figure12. Example4(a):
4.1:the
thescale factor
scale Ω(x).
factor Utilization
Ω(x). UN+MU=N+M
factorfactor
Utilization 1.0.= 1.0.

Table 4. Example 4: key values for each iteration.

Table 4. Example 4: key values for each iteration.
− η0,i
Iteration αult λ χ αb xcr,i η0
− η1.4135
1.2828 α
1 Iteration
ult
0.8319
λ
0.7046χ α
0.9038
b
10.0878
xcr,i 0,i
η0
2 1.5123 0.9033 0.6591 0.9967 10.268 0.9775
3 1 1.49751.28280.8989
0.8319 0.6619
0.7046 0.9912
0.9038 10.0878
10.268 1.4135
1
2 1.5123 0.9033 0.6591 0.9967 10.268 0.9775
3 the maximum
In the Example 4(a), 1.4975 moment
0.8989 Mmax
0.6619 0.9912
(x = 7.654 m) can10.268 1 Equation (25),
be obtained from
which serves to verify the value Mmax (x = 7.654 m) = 30.441 kNm given in Figure 11:
In the example 4.1, the maximum moment Mmax(x = 7.654 m) can be obtained from Equation (25),
FEd · η0 500 · 28/1000
which serves to verify the value
Mmax =M max(x = 7.654
= m) = 30.441 ≈kNm
30.441given in Figure 11:
kN · m (25)
1 1
1− 1−
FEd α⋅η
cr
500 ⋅ 28 /1000
1.852
M max = = 0
≈ 30.441kN ⋅ m
According to the direct method,the imperfection
1  amplitude1 is η0 =25.86 mm. It is a similar result. (25)
1 − for
(b) In Figure 13, the αb values calculated 1 − slenderness
 different  at the critical section are compared with
the GMNIA results of Marques et al.[28].αThe  1.852 
cr maximum difference is −8% for relative slenderness λ = 0.773
(buckling curve b). This analysis is performed for a hot-rolled tapered column b = 100 mm, tf = tw =10 mm,
According to the direct method, the imperfection amplitude is η0 =25.86mm. It is a similar result.
hmin = 100 mm and hmax = 400 mm, steel S235 with a linearly varying height, and a simply supported member.
Example 4.2
The uniform axial force NEd = Npl,Rd,min = 658 kN.
In Figure 13, the αb values calculated for different slenderness at the critical section are compared
with the GMNIA results of Marques et al. [28]. The maximum difference is −8% for relative
slenderness λ = 0.773 (buckling curve b). This analysis is performed for a hot-rolled tapered column
b = 100 mm, tf = tw =10 mm, hmin = 100 mm and hmax = 400 mm, steel S235 with a linearly varying height,
and a simply supported member. The uniform axial force NEd = Npl,Rd,min = 658 kN
Appl. Sci. 2020, 10, x FOR PEER REVIEW 16 of 24

Appl.
Appl. Sci.Sci. 2020,
2020, 10,10, x FOR PEER REVIEW
8174 15 16 of 24
of 22

Figure 13. Example 4.2: comparison of αb values calculated for the different slenderness obtained by
GMNIA and by the proposed procedure of the authors.
Figure Example
13.13.
Figure Example 4(b): comparisonofofαbαbvalues
4.2:comparison valuescalculated
calculatedfor
forthe
thedifferent
differentslenderness obtained
slenderness byby
obtained
In example
GMNIA
GMNIA andand4.2,
byby GMNIA
the proposed
the was
proposed performed
procedure
procedure ofby
of the Marques
authors. et al. [28] using ABAQUS software taking
authors.
the
into account a geometrical imperfection with the shape of the buckling mode and amplitude of
L/1000,InIn
Example
example
also 4(b),
4.2,GMNIA
considering GMNIA was
wasperformed
the residual stressesbyofMarques
performed by Marques
a hot-rolledet al.cross-section.
et[28]
al. using ABAQUS
[28] using ABAQUS software takingtaking
software into
account a geometrical imperfection with the shape of the buckling mode and
into account a geometrical imperfection with the shape of the buckling mode and amplitude of amplitude of L/1000, also considering
theL/1000,
4. residualalso
Applicationstresses of a Case
to the
considering hot-rolled
where
the cross-section.
Problems
residual stresses Were
of aFound
hot-rolledto Obtain the Critical Section and to the
cross-section.
Member Made of Aluminum Alloy
4. Application to the Case Where Problems Were Found to Obtain the Critical Section and to the
4. Application
Member Made the oftoAluminum
the Case where Problems Were Found to Obtain the Critical Section and to the
Alloythese
Following other examples, are solved where the step changes in the cross-section and
Member Made of Aluminum Alloy
in theFollowing
axial forces thecreate
other numerical
examples, problems to find where
these are solved the critical
the stepsection xcr. These
changes in theare pure academic
cross-section and
problems Following
found inthethe other examples,
literature, these
however, are
there solved
is the where
need to the
show
in the axial forces create numerical problems to find the critical section xcr . These are pure step changes
that they in
maythebecross-section
solved.
academic and
in the axial
Example forces
5 create numerical problems to find the critical
problems found in the literature, however, there is the need to show that they may be solved. section x cr. These are pure academic

problems
A Ph.D.found
student in wrote
the literature, however,
in [24] about there is the
the obstacles need
of the methodto show that they
because he was may notbeable
solved.
to obtain
the Example
location of 5
the critical section due to the never-ending cycle by
Example 5. A Ph.D. student wrote in [24] about the obstacles of the method because he was not able togoing from finding the “critical
obtain
section” A to
the locationPh.D. thestudent
ofthe other
criticalonewroteandin
section [24]
back.
due about
to theThe the obstacles
fifth
never-ending cycleofby
example the
shows method
going that because
fromproposed
finding thehe was notsection”
method
“critical ableno
has totoobtain
such
the
the
problem.location
A of the
cantilever critical
column section
with adue
step to the
change never-ending
in the cycle
cross-sectional by going from
parameters
other one and back. The Example 5 shows that proposed method has no such problem. A cantilever column with finding
and axialthe “critical
forces is
section”
investigated to the
(Figure other14). one
The and back.
material The
used fifth
in [24] example
is the shows
steel grade that proposed
S355. The
a step change in the cross-sectional parameters and axial forces is investigated (Figure 14). The material used method
safety factor has
γ M1 no
= such
1.0.
in problem.
[24] is the A cantilever
steel grade S355. column with afactor
The safety step γ change
M1 = 1.0.
in the cross-sectional parameters and axial forces is
investigated (Figure 14). The material used in [24] is the steel grade S355. The safety factor γM1 = 1.0.

Figure 14. Example 5: description taken from [24].

Figure 14. Example 5: description taken from [24].

Figure 14. Example 5: description taken from [24].

 Appl. Sci. 2020, 10, x FOR PEER REVIEW 17 of 24
Appl. Sci. 2020, 10, 8174 16 of 22
Appl. Sci. 2020, 10, x FOR PEER REVIEW 17 of 24
The position of the section xU where the greatest utilization factor UN+M,max < 1.0 takes place may
depend on the load level. The position of the section xU may differ from the location of the critical
Theposition
The position of the
of the section
section xU where
xU where the greatest
the greatest utilization
utilization factor Ufactor
N+M,max
UN+M,max < 1.0 place
< 1.0 takes takesmayplace may
depend
section xcr defined in the nomenclature. This is case in Figures 15 and 16. In the rare cases, as it is in
ondepend
the loadonlevel.
the load level. The
The position position
of the section of xUthe
maysection xU may
differ from differ from
the location of thethe location
critical of the
section critical
xcr defined
example 5 (Figure 14), the location of the critical section xcr may be found in accordance with its
section x defined in the nomenclature. This is case in Figures 15 and 16. In
in the nomenclature. This is case in Figures 15 and 16. In the rare cases, as it is in Example 5 (Figure
cr the rare cases, as it is
14),in
definition given in nomenclature only if UN+M,max = 1.0 (Figure 17). The cases in which the almost
example 5 (Figure 14), the location of the critical section x may be found in accordance
the location of the critical section xcr may be found in accordance with its definition given in nomenclature only if
cr with its
identical values of UN+M,max appear in the same time in more than one section are very rare. Example
Udefinition
N+M,max = given in nomenclature
1.0 (Figure only
17). The cases in which if U
the almost
N+M,max = identical
1.0 (Figure 17).
values of The
U casesappear in which
in thethe
samealmost
time
5 is a such rare case. The comparison of the values UN+M,max (xcr = 10 N+M,max m) = 1.0 with UN+M (x = 0 m) =
identical values of U appear in the same time in more than one section are
in more than one section are very rare. Example 5 is a such rare case. The comparison of the values UN+M,max
N+M,max very rare. Example
0.9378 is shown in Figure 17. These values would be closer if steel S235 would be used instead of steel
(x5cris=a10
such
m) =rare
1.0 case.
with U The comparison
N+Mthe(x = 0 m) = of the values
0.9378 is shownUin Figure
N+M,max (xcr17.
= 10 m) values
These = 1.0 with
wouldUbeN+M (x = if
closer 0 steel
m) =
S355: 1.0 and 0.9965, and value of η0 would be decreased from 101 to 77 mm.
0.9378 is shown in Figure 17. These values would be closer if steel S235 would
S235 would be used instead of steel S355: 1.0 and 0.9965, and the value of η0 would be decreased from 101be used instead of steel
to
S355:
77 mm. 1.0 and 0.9965, and the value of η 0 would be decreased from 101 to 77 mm.

Figure 15. Example 5: UGLI imperfection ηinit, bending moment M, shear force V, and utilization
factor.Values in brackets areimperfection
taken from [24]. Value (1.24) was obtained by GMNIA.
Figure 15. Example
Figure15. UGLI imperfection ηηinit
Example 5:5: UGLI , bending moment M, shear force V, and utilization
init, bending moment M, shear force V, and utilization
factor.Values in brackets are taken from [24]. Value (1.24) was obtained by GMNIA.
factor.Values in brackets are taken from [24]. Value (1.24) was obtained by GMNIA.

Figure
Figure 16. 16. Example
Example 5: the
5: the scale
scale factor Ω(x).
factor Ω(x). Utilization
Utilization factor
factor UN+M==1.0.
UN+M 1.0.

Figure 16. Example 5: the scale factor Ω(x). Utilization factor UN+M = 1.0.
 Appl. Sci. 2020, 10, 8174 17 of 22
Appl. Sci. 2020, 10, x FOR PEER REVIEW 18 of 24

Figure
Figure 17.17. Example
Example 5: UGLI
5: UGLI imperfection
imperfection ηinit , bending
ηinit, bending moment
moment M,force
M, shear shearV,force V, and utilization
and utilization factor
factor U
UN+M = 1.0.N+M = 1.0.

Thedistributions
The distributions ofofUGLI
UGLIimperfection,
imperfection, bending moment
bending M(x) and
moment M(x) shear
andforce
shearV(x) due V(x)
force to thedue
imperfection
to the
and the utilization factors U(x) are plotted in Figure 15. The resulting critical
imperfection and the utilization factors U(x) are plotted in Figure 15. The resulting section is x = 10 m. The minimum
cr critical section is
value of the scale factor led in the second and the last iteration to UGLI imperfection
xcr = 10 m. The minimum value of the scale factor led in the second and the last iteration to UGLI amplitude η 0 = 101 mm
(Figure 16). amplitude
imperfection The significant
η0 = values
101 mm for(Figure
all iterations
16). Thearesignificant
shown in Table
values5. With
for allthe procedure
iterations areof shown
the authors
in
Table 5. With the procedure of the authors presented in this article, there is no problem forload
presented in this article, there is no problem for solving this example. The utilization at the buckling level is
solving
shown
this in Figure
example. The17. The buckling
utilization bucklingαload
at theresistance b = 1.189
level was calculated
is shown according
in Figure 17. Theto authors’
bucklingprocedure
resistance and
checked with GMNIA value
αb = 1.189 was calculated according α b = 1.24 (Figure 15).
to authors’ procedure and checked with GMNIA value αb = 1.24
(Figure 15). Table 5. Example 5: key values for each iteration.

Table − η0,i
Iteration αult 5. Example
λ
5: key values
χ for each
αb iteration.xcr,i η0

1 3.0523 1.5139− 0.3372 1.0292 η

0.0 2.0485
2
Iteration α
9.5979 ult 2.6845λ
χ
0.1224
α1.1749
b xcr,i
10
0,i
η0 1

1 3.0523 1.5139 0.3372 1.0292 0.0 2.0485

In Example 5, GMNIA 2 was performed2.6845
9.5979 by the authors
0.1224(Figure
1.174915) using
10 the 1Open System for Earthquake
Engineering Simulation (OPENSEES) software taking into account a geometrical imperfection with the shape
of the
In buckling
examplemode and amplitude
5, GMNIA was performed cr /1000),
of L/500 (Lby also considering
the authors theusing
(Figure 15) residual
thestresses
Open of a hot-rolled
System for
cross-section.
Earthquake Engineering Simulation (OPENSEES) software taking into account a geometrical
In Example
imperfection with5,the
theshape
maximum
of themoment
bucklingMmax (x =and
mode 0 m)amplitude
can be obtained from(L
of L/500 Equation (26), which serves to
cr/1000), also considering
verify the value M
the residual stresses (x = 0 m) = 101.731 kNm
maxof a hot-rolled cross-section.given in Figure 15:
In example 5, the maximum moment M max(x = 0 m)x can be obtained
x = 5 from Equation (26), which
FEd1 ·ηxinit= 15 FEd2 ·ηinit= 10 FEd3 ·ηinit
M(xmax
serves to verify the value Mmax = 0= m) =(101.7311
1− αcr )
+
kNm given
1
(1− αcr )
+ Figure
in =
(1− α1cr )15:
(26)
= 101/1000
1 ( 95 + 0.41 · 225 + 0.095 · 550 ) ≈ 101.731 kN · m
(1− 1.313 )
Appl. Sci. 2020, 10, 8174 18 of 22

If the value of the force of FEd,2 = 225 kN (Figure 14) changes, the values of UGLI imperfection amplitude
and the critical section position xcr are given in Table 6. There is no problem to find the location of the critical
section xcr . From Table 6, it can be concluded that UGLI imperfection amplitude has the step change when the
critical section position changes from xcr = 10 to 5 m. It changes from η0 = 101 to 145 mm for steel grade S355.

Table 6. Example 5: influence of the value of FEd,2 and the steel grade. Values in brackets are from [24].

Steel FEd,2 (kN) 100 125 150 175 200 (225) 250 275 300 325
S 235 η0 (mm) 68.8 71.1 73.4 75.7 78 77.7 113 115.5 115.5 115.3
S 275 η0 (mm) 75 77.4 80 82.4 85 87.4 126.1 126 126 125.8
S 355 η0 (mm) 86 88.8 91.6 94.5 97.4 101 (100.9) 145 145 144.9 144.7
xcr (m) 10 10 10 10 10 10 (10) 5 5 5 5

In Table 7, the buckling resistance αb calculated according to the authors’ procedure is checked with the
GMNIA value for steel grades 235, 275 and 355.

Table 7. Example 5: buckling resistance αb calculated according to the authors’ procedure and with GMNIA.

Steel FEd,2 (kN) 100 125 150 175 200 225 250 275 300 325
αb (Authors) 1.381 1.333 1.287 1.242 1.2 1.159 1.065 1.023 0.985 0.949
S 235
αb (GMNIA) 1.455 1.396 1.346 1.294 1.246 1.193 1.138 1.09 1.07 1.03
αb (Authors) 1.397 1.348 1.301 1.256 1.213 1.171 1.082 1.04 1.001 0.965
S 275
αb (GMNIA) 1.46 1.41 1.37 1.318 1.26 1.22 1.15 1.11 1.07 1.04
αb (Authors) 1.42 1.37 1.322 1.275 1.231 1.189 1.148 1.064 1.025 0.988
S 355
αb (GMNIA) 1.48 1.45 1.39 1.34 1.29 1.24 1.21 1.169 1.121 1.08

In Brodniansky’s work [24], where for UGLI imperfection the incorrect name EUGLI is used, the critical
section xcr could not be found using the utilization factor matrix due to numerical issues. According to the
procedure proposed in this work, the critical section xcr is possible to find. It may be concluded that Chladný’s
method has no obstacles.

Example 6. Höglund solved in [30] by equivalent member method given in [2] the member made of aluminum
alloy EN AW-6005A-T6, buckling class A, with a yield strength fo = 215 MPa, ultimate strength fu = 260 MPa
and γM1 = 1.1. He also took into account welds located in the fixed end and in the place where a member changes
his cross-section.
We took from example 6.11 [30] only the geometry L = 5 m and loading. The purpose was to show that also
Appl. Sci. 2020, 10, x FOR PEER REVIEW 20 of 24
in example 6.11 [30], which is similar to Example 6, the presented procedure may obtain desired results without
problems.
The section Our member (Figure
properties of EP 18) is made of
(extruded the aluminum
profile) are A1 =alloy
38.8EN
cmAW-6061-T6,
2, I1 = 1673 cmbuckling
4, W1 = class
220 cm = 70,000
A, 3Eand A2 =
MPa, f o = 240 MPa, fu = 260 MPa
91 cm , I2 = 8091 cm , W2 = 736 cm .
2 4 3(Table 3.2b [2]), γ M1 = 1.1. The section properties of EP (extruded profile)
are A1 = 38.8 cm2 , I1 = 1673 cm4 , W1 = 220 cm3 and A2 = 91 cm2 , I2 = 8091 cm4 , W2 = 736 cm3 .

Figure 18. Example 6: description of the member made of aluminum alloy taken from [30] example 6.11.
Figure 18. Example 6: description of the member made of aluminum alloy taken from [30] example
6.11.

Nothing may be compared with Höglund’s example 6.11, except the values of buckling lengths,
which are the same.
Appl. Sci. 2020, 10, 8174 19 of 22
Figure 18. Example 6: description of the member made of aluminum alloy taken from [30] example
6.11.
Nothing may be compared with Höglund’s example 6.11, except the values of buckling lengths, which are
the same. Nothing may be compared with Höglund’s example 6.11, except the values of buckling lengths,
which are the same.
The distributions of UGLI imperfection, bending moment M(x) and shear force V(x) due to the imperfection
The distributions of UGLI imperfection, bending moment M(x) and shear force V(x) due to the
and the utilization
imperfectionfactors
and U(x) are plotted
the utilization in Figure
factors 19.plotted
U(x) are The critical section
in Figure takes
19. The at xcr =takes
placesection
critical 0 m.place
The minimum
value of scale factor led in the last iteration to UGLI imperfection amplitude η
at xcr = 0 m. The minimum value of scale factor led in the last iteration to UGLI
0 = 56 mm (Figure 20).
imperfection amplitude The influence
η0 =is56beyond
of the welds mm (Figure 20). The
the scope influence
of this paper. ofThe
theauthors
welds isintend
beyondtothe scope
take of this
it into paper.inThe
account authors
another paper.
intend to take it into account in another paper.

Appl. Sci. 2020, 10, x FOR PEER REVIEW 21 of 24

Figure 19. Example 6: UGLI6: imperfection

Figure 19. Example ηinitηinit
UGLI imperfection , bending moment
, bending moment M,M, shear
shear V, andV,
forceforce and utilization
utilization factor factor U.
U.

Figure
Figure 20. 20. Example
Example 6: the6:scale
the scale
factor Ω(x).
factor Ω(x).Utilization
Utilization factor
factorUU N+M = 1.0.
N+M = 1.0.

5. Conclusions
A new more general procedure is presented based on Chladný’s method, which was accepted
by CEN/TC 250 working groups for Eurocodes [1,2,4,5]. It enables to obtain the geometric equivalent
UGLI imperfection ηinit (x), the critical section xcr and UGLI imperfection amplitude η0 for complex
slender metal structures. This procedure has been validated by the recalculation of several examples
of steel members published by Chladný [7], Papp [23] and Marques et al. [28]. The comparison of the
results showed very good agreement with the results of other authors and results of GMNIA. It is
shown that the presented procedure is also able to solve case [24] in which the critical section xcr could
Appl. Sci. 2020, 10, 8174 20 of 22

5. Conclusions
A new more general procedure is presented based on Chladný’s method, which was accepted by
CEN/TC 250 working groups for Eurocodes [1,2,4,5]. It enables to obtain the geometric equivalent
UGLI imperfection ηinit (x), the critical section xcr and UGLI imperfection amplitude η0 for complex
slender metal structures. This procedure has been validated by the recalculation of several examples of
steel members published by Chladný [7], Papp [23] and Marques et al. [28]. The comparison of the
results showed very good agreement with the results of other authors and results of GMNIA. It is
shown that the presented procedure is also able to solve case [24] in which the critical section xcr could
not be found. The example of the member made of aluminum alloy investigated by Höglund [30] is
also solved without problem.
The paper provides information about: (i) the history of method development used in Eurocodes
and (ii) differences between the current Eurocodes [1,2] and their newest working drafts [4,5].
The method is described in detail in such a way that the users of Eurocodes will also be able to use it
for complex frame structures made of steel or aluminum alloys.
The topics which are beyond scope of this paper are namely the influence of: (i) partial safety
factor γM1 ; (ii) local buckling at steel structures; and (iii) transverse or longitudinal welds at structures
made of aluminum alloys, will be investigated in the next papers. The results of direct method are very
close to the ones obtained by the authors’ method, except the results in the Example 1. Further studies
should be performed to determine if the direct method could be an alternative to the current method
proposed in the Eurocodes.

Author Contributions: A.A.: Conceptualization, methodology, software, validation, writing—original

draft, writing—review and editing, supervision; I.B.: conceptualization, validation, writing—original draft,
writing—review and editing, supervision; Y.K.: writing—original draft, writing—review and editing, supervision;
P.M.: writing—original draft, writing—review and editing, supervision. All authors have read and agreed to the
published version of the manuscript.
Funding: This research did not receive any specific grant from funding agencies in the public, commercial,
or non-profit sectors.
Acknowledgments: Authors devote the paper to the memory of Prof. Ing. Eugen Chladný, Ph.D.
(01.08.1928–07.07.2020), the author of UGLI imperfection method used in both metal Eurocodes.
Conflicts of Interest: The authors declare no conflict of interest.

Nomenclature
α is the imperfection factor related to the flexural buckling curve (Tables 6.1 and 6.2 of EN 1993-1-1 [1]; Tables
3.2 and 6.6 in EN 1999-1-1 [2]). αLT is the imperfection factor for lateral torsional buckling related to buckling
curve (Tables 6.3, 6.4 and 6.5 in EN 1993-1-1 [1]; 6.3.2.2 in EN 1999-1-1 [2]). αcr is the minimum load amplifier
for the axial force configuration in members to reach the elastic critical buckling load (5.2.1(3) in [1]; 5.2.1(3) in
[2]). αult is the amplifier for the load of members to reach the characteristic resistance of the critical cross-section
(6.3.4(2) in [1], where more the convenient symbol αult,k is used; [2] does not know such quantity and symbol). αb
is the load amplifier of members to reach the ultimate buckling load (see Equations (12) and (16); the symbol is
not used in [1,2]). γM0 is the partial safety factor for the resistance of cross-section whatever the class is (6.1 in
[1]; [2] does not know such quantity and symbol). γM1 is the partial safety factor for the resistance of members
to instability assessed by member checks (6.1 in [1] and 6.1.3; Table 6.1 in [2]). χ is the reduction factor for the
relevant buckling curve (6.3.1.2 in [1]; 6.3.1.2 in [2]). λ is the relative slenderness (6.3.1.2 in [1]; 6.3.1.2 in [2]). λ0 is
the plateau length of buckling curves (for steel [1]: 0.2; for the aluminum alloy [2]: 0.1 for buckling class A and 0.0
for buckling class B). λLT is the relative slenderness for lateral torsional buckling (6.3.2.2 and 6.3.2.3 in [1]; 6.3.2.2
in [2]). λTF is the relative slenderness for torsional flexural buckling. λLT,0 is the plateau length of lateral torsional
buckling curves. β is the correction factor for lateral torsional buckling curves (6.3.2.3 in [1]; [2] does not know
such quantity and symbol). η0 is the amplitude of UGLI imperfection. {ηinit } is UGLI imperfection in the shape
of elastic critical buckling mode. {ηcr } is the shape of elastic critical buckling mode. ηcr,w is the displacement of
the buckling shape component perpendicular to axis y along the x axis. ηcr,v is the displacement of the buckling
shape component perpendicular to axis z along the x axis. ηcr,θx is the torsional rotation of the buckling shape
component about shear center axis along the x axis. A is the cross-sectional area. E is the modulus of elasticity
(210,000 MPa for steel [1]; 70,000 MPa for aluminum alloy [2]). e0,d (e0,k ) is the design (characteristic) value of the
initial imperfection amplitude of the equivalent member used in the calculation of UGLI imperfection amplitude.
It is given in 5.3.2(11) [1,2] or in 7.3.6(1) [4] or in 7.3.2(11) in [5]. The indices d and k are sometimes omitted.
Appl. Sci. 2020, 10, 8174 21 of 22

e0 is the local bow imperfection given in Table 5.1 [1,2] or in 7.3.3.1(1) [4] or in 7.3.2(3)b) [5]. It is used when
performing second order analysis including member imperfections related to flexural buckling. It is not used in
UGLI imperfection method and must not be commuted with the above amplitude. Iy , Iz are the second moments
of area with respect to the y, z axes. Iw is the warping constant. Ncr is the elastic critical force of the relevant
buckling mode based on the gross cross-section properties. NEd is the design value of normal force. NRk is the
characteristic resistance of normal force in the critical section (depending on Class 1, 2 and 3 cross-section). MRk is
the characteristic resistance of bending moment in the critical section (depending on Class 1, 2 and 3 cross-section).
Wz is the section modulus about z axis (depending on Class 1, 2 and 3 cross-section). Wy is the section modulus
about y axis (depending on Class 1, 2 and 3 cross-section). WB is the warping section modulus. xcr is the critical
section where the utilization factor is greater than at all the other sections.

References
1. EN 1993-1-1:2005 and Corrigendum AC (2006) and Corrigendum AC (2009) and Amendment A1 (2014). Eurocode 3:
Design of Steel Structures. In Part 1.1: General Rules and Rules for Buildings; CEN: Brussels, Belgium, 2005.
2. EN 1999-1-1:2007 and Amendment A1 (2009) and Amendment A2 (2013). Eurocode 9: Design of Aluminium
Structures. In Part 1.1: General Structural Rules; CEN: Brussels, Belgium, 2007.
3. Chladný, E.; Štujberová, M. Frames with unique global and local imperfection in the shape of the elastic
buckling mode (part1). Stahlbau 2013, 8, 609–617. [CrossRef]
4. prEN 1993-1-1:2020. Eurocode 3: Design of Steel Structures. In Part 1.1: General Rules and Rules for Buildings;
CEN: Brussels, Belgium, 2020.
5. prEN 1999-1-1:2020. Eurocode 9: Design of aluminium structures. In Part 1.1: General Structural Rules;
CEN/TC 250/SC 9 N 888; Secretariat BSI: London, UK, 2020.
6. Sedlacek, G. Consistency of the Equivalent Geometric Imperfection Used in Design and in the Tolerances for Geometric
Imperfection Used in Execution; CEN/TC 250/SC 3 document N1721; CEN: Brussels, Belgium, 2010; pp. 1–144.
7. Chladný, E.; Štujberová, M. Frames with unique global and local imperfection in the shape of the elastic
buckling mode (Part 2). Stahlbau 2013, 9, 684–694. [CrossRef]
8. Chladný, E.; Štujberová, M. Errata: Frames with unique global and local imperfection in the shape of the
elastic buckling mode. Stahlbau 2013, 82, 684–694. [CrossRef]
9. Chladný, E. Imperfections for global analysis in STN EN 1993-1-1. In Implementation of Eurocodes in Practice.
Design of Steel Structures according to STN EN 1993-1-1: November 2006 and STN EN 1993-1-8 April 2007, 1st ed.;
SKSI Bratislava: Bratislava, Slovakia, 2007; Chapter 5. (In Slovak)
10. Chladný, E. Imperfections for Global Analysis in STN EN 1993-1-1. In Implementation of Eurocodes in Practice.
Design of Steel Structures According to STN EN 1993-1-1: November 2006 and STN EN 1993-1-8 April 2007 and
Their Corrigenda and National Annexes, 2nd ed.; Baláž, I., Ed.; SKSI Bratislava: Bratislava, Slovakia, 2010;
Chapter 5. (In Slovak)
11. 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. In Aktívu Pracovníkov Odboru
OK so Zahraničnou Účast’ou “Teoretické a Konštrukčné Problémy Ocel’ových a Drevených Konštruckcií a Mostov”;
STU Bratislava: Pezinok, Slovakia, 2008; pp. 17–22.
12. 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. In Anniversary Volume Honoring
Amália and Miklós Iványi; Univeristy of Pécs: Pécs, Hungary, 2010; pp. 3–15.
13. Baláž, I.; Koleková, Y. Structures with UGLI imperfections. In Proceedings of the 18th International
Conference Engineering Mechanics 2012, Svratka, Czech Republic, 14–17 May 2012; pp. 61–86.
14. Baláž, I.; Koleková, Y. In plane stability of two hinged arches. In Proceedings of the European Conference on
Steel and Composite Structures, Eurosteel, Budapest, Hungary, 31 August–2 September 2011; pp. 1869–1874.
15. Baláž, I.; Koleková, Y.; Kováč, M.; Živner, T. Generalization of Unique Global and Local Initial Imperfection
Used in EN 1993-1-1 and EN 1999-1-1. In Proceedings of the 8th International Conference on Steel and
Aluminium Structures, Hong Kong, China, 7–9 December 2016; pp. 1398–1411.
16. Agüero, A.; Pallarés, F.J. Proposal to evaluate the ultimate limit state of slender structures. Part 1: Technical
aspects. Eng. Struct. 2007, 29, 483–497. [CrossRef]
17. Agüero, A.; Pallarés, L.; Pallares, F.J. Equivalent geometric imperfection definition in steel structures sensitive
to flexural and/or torsional buckling due to compression. Eng. Struct. 2015, 91, 160–177. [CrossRef]
Appl. Sci. 2020, 10, 8174 22 of 22

18. Agüero, A.; Pallarés, F.J.; Pallares, L. Equivalent geometric imperfection definition in steel structures sensitive
to lateral torsional buckling due to bending moment. Eng. Struct. 2015, 96, 41–55. [CrossRef]
19. Kováč, M. Buckling Resistance of Metal Members and Frame Structures, Application of New Methods from
Eurocodes. Ph.D. Thesis, STU in Bratislava, Bratislava, Slovakia, 2012.
20. Dallemule, M. In Plane Buckling of Arch Structures. Ph.D. Thesis, STU in Bratislava, Bratislava, Slovakia,
2013. (In Slovak).
21. 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 with alternative rules in
EN1993—Eurocode3—Part1-1. Steel Constr. 2010, 3, 33. [CrossRef]
22. Wieschollek, M.; Schillo, N.; Feldmann, M.; Sedlacek, G. Lateral–torsional buckling checks of steel frames
using second-order analysis. Steel Constr. 2012, 5, 71–86. [CrossRef]
23. Papp, F. Buckling assessment of steel members trough overall imperfection method. Eng. Struct. 2016, 106,
124–136. [CrossRef]
24. Brodniansky, J. Equivalent unique global and local initial imperfection - Imperfection in EN 1993-1-1 and EN
1999-1-1 Clause 5.3.2 (11)—Calculation procedure and discovered obstacles. Pollack Period. 2017, 12, 33–42.
[CrossRef]
25. Lindner, J.; Kuhlmann, U.; Just, A. Verification of Flexural Buckling According to Eurocode 3 Part 1-1 using
bow imperfections. Steel Constr. 2016, 9, 349–362. [CrossRef]
26. Lindner, J.; Kuhlmann, U.; Jörg, F. Initial bow imperfection e0 for the verification of Flexural Buckling
According to Eurocode 3 Part 1-1—Additional considerations. Steel Constr. 2018, 11, 30–41. [CrossRef]
27. Trahair, N.S. Flexural–Torsional Buckling of Structures; E & FNspon: London, UK, 1993.
28. Marques, L.; Taras, A.; Silva, L.S.; Greiner, R.; Rebelo, C. Development of a consistent. buckling design
procedure for tapered columns. J. Constr. Steel Res. 2012, 72, 61–74. [CrossRef]
29. Chladný, E.; Chladná, M. Some remarks to the application of equivalent bow imperfection in analyses of
compressed steel members and structures. In Proceedings of the 10th International Conference on Metal Structures;
Ziółko, J., Supernak, E., Eds.; PAN: Gdansk, Poland, 2001; Volume 2, pp. 351–356.
30. Höglund, T.; Tindall, P. Designer’s Guide to Eurocode 9: Design of Aluminium Structures; EN 1999-1-1 and -1-4;
ICE Publishing: London, UK, 2012; ISBN 978-0-7277-5737-1.

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional
affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).