Professional Documents
Culture Documents
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
a r t i c l e i n f o a b s t r a c t
Article history: The purpose of this paper is to present a proposal for the design of steel structures sensitive to buckling
Received 4 July 2014 due to compression in order to fill in the gaps in the current Standard EN 1993-1-1 guidelines for
Revised 1 December 2014 obtaining the magnitude of the geometric equivalent imperfection. The proposal generalizes the
Accepted 30 March 2015
approach provided in Clause 5.3.2(11) of EN 1993-1-1 (EC-3) for cases in which a torsional or flexural–
Available online 17 April 2015
torsional buckling mode may occur. The extension of the procedure also allows designers to obtain the
magnitude and shape of the imperfection as well as the worst direction of the imperfection due to
Keywords:
the external loads applied. It also identifies the cases in which it is necessary to consider the shape of
Buckling
Torsional–flexural
the imperfection given by higher buckling modes.
Steel design Ó 2015 Elsevier Ltd. All rights reserved.
Geometric imperfection
Nonlinear analysis
http://dx.doi.org/10.1016/j.engstruct.2015.03.065
0141-0296/Ó 2015 Elsevier Ltd. All rights reserved.
A. Agüero et al. / Engineering Structures 96 (2015) 160–177 161
Nomenclature
a is the imperfection factor (Tables 6.1 and 6.2 of EC-3) re- It is the torsional constant
lated to each buckling curve Iw is the warping constant
acr is the minimum force amplifier for the axial force con- KB is the generic element of the structure
figuration in members to reach the elastic critical buck- ½K L is the linear stiffness matrix
ling load ½K G is the geometric stiffness matrix taking into account ax-
ault;k is the minimum force amplifier for the axial force con- ial forces
figuration in members to reach the characteristic resis- L is the total length of the structure
tance N Rk of the most axially stressed cross section M Rk is the characteristic moment resistance of the critical
without taking buckling into account cross section(Mel;Rk in class 3, Mpl;Rk in class 2)
cM1 is the partial safety factor for the resistance of members M pl;y;Rd ; M pl;z;Rd are the plastic moment resistance about y and z
to instability assessed by member checks axes
cM0 is the partial safety factor for the resistance of cross sec- Ncr;T is the elastic torsional buckling force
tions whatever the class is Ncr;TF is the elastic torsional–flexural buckling force
v is the reduction factor for the relevant buckling curve NB is the number of elements
k is the nondimensional slenderness N0 is the reference axial design force (NEd;0 according to
ginit is the vector of imperfections [1]), related to one section of the member
gcr is the shape of the elastic critical buckling mode t is the thickness of the flange/web
EIg00cr;max is the bending moment due to the buckling mode gcr T is the torsional moment
at the critical cross section Tt is the internal St. Venant torsional moment
xðy; zÞ is the warping function Tw is the internal warping torsional moment
hx is the torsional rotation. u is the displacement in direction x of the centroid
A is the cross sectional area Vy, Vz are the shear forces in y and z directions
Bpl;Rd is the plastic bimoment resistance v; w are the displacements in the principal directions y and z
fdNL g is the vector of nonlinear displacements of the shear center
ec is the magnitude of the imperfection Wz is the bending section modulus about z axis
E is the modulus of elasticity for steel Wy is the bending section modulus about y axis
fF ext g is the vector of external forces Wx is the bimoment section modulus
G is the shear modulus for steel x is the axis of the member
is is the radius of gyration related to centroid ysc; zsc are the coordinates of the shear center
Iy; Iz are the second moment of area with respect to y, z axes y, z are the coordinates of the check point
The disadvantages of Method 2.1 are: sensitive to flexural and/or torsional buckling; the worst direction
of the imperfection to be considered; the number of buckling
– A large number of combinations of local and global imperfec- modes to be considered in order to define the imperfection and
tions are possible, and the designer must find the worst [7,8] the way to deal with structures formed with tapered columns, arch
from all the likely combinations. (See example of a pitched roof geometry, or members under non uniform axial internal force.
frame Fig. 2.) Several authors have tried to improve the methods proposed in
– The buckling resistance obtained with Method 2.1 (Clause 5.3.2) the EC-3. Chladný and štujberová [10] proposed modifying Clause
is not consistent with the buckling resistance obtained using 5.3.2(11) to deal with structures sensitive to flexural buckling with
Method 1 (Clause 6.3.1) and therefore with the numerical and tapered columns and/or non-uniform distribution of the compres-
experimental studies on which this standard is based. sion force along their length and/or arch geometry. Marques et al.
– The method is easily applied to building frames but might be [11] proposed a design consistent with current rules for uniform
difficult to apply to other structural geometries. Additional columns provided in EC-3, i.e. Clause 6.3.1. for the relevant case
recommendations for arch imperfections are provided in of in-plane flexural buckling of linearly tapered columns subject
Clause D.3.5 (Annex D of EN 1993-2 steel bridges EN-1993-2 to constant axial force. Agüero and Pallarés [8] proposed a simpli-
[9]). fied method of obtaining the geometric equivalent imperfection in
structures sensitive to flexural–torsional buckling due to either
The advantages of Method 2.2 are: bending or compression where shear stress due to Saint Venant
torsional moment could be neglected, scaling the first buckling
– The nonlinear geometric analysis with an imperfection given by mode. The general method for flexural and lateral torsional buck-
the elastic buckling mode is easy to perform (see Section 2.2 of ling checks in accordance with clause 5.3.2(11) and 6.3.4 was
the article). developed by Bijlaard et al. [12], Wieschollek et al. [13] and
– The method can easily be applied to any structural type or Naumes [14] considering the flanges as members under compres-
geometry (see Figs. 3 and 4) since a unique global and local sion sensitive to flexural buckling.
imperfection can be obtained from the buckling mode.
– In single members the imperfection is calibrated to ensure that 1.1. Research significance
the buckling resistance with method 2.2 is equal to the buckling
resistance computed by method 1 (Clause 6.3.1). The main innovations in this research are (Fig. 1):
However, Method 2.2, provided by the EC-3 for flexural buck- – A procedure to assess the imperfection for structures sensitive
ling, presents some uncertainties from a practical point of view, to flexural and/or torsional buckling due to compression, con-
such as: the definition of the imperfection (ginit ) in structures sistent with the buckling load obtained from EC-3. Like the
162 A. Agüero et al. / Engineering Structures 96 (2015) 160–177
Fig. 1. Summary of the methods to obtain the buckling resistance and the scope of this research.
method proposed by Chladný and štujberová [10], the approach experiments, for members under a torsional and flexural–torsional
described in this paper can be used for tapered columns and/or buckling load. The approach proposed in this paper is valid if dif-
arch geometry and/or nonuniform normal forces distribution ferent v factors were taken into account, which are better adjusted
(Fig. 4). to torsional and flexural–torsional buckling [15–19].
– An energy method to estimate the worst direction of the imper- The second assumption is that the basis of the proposal is the
fection with regard to the external loads. same as in the method given in Clause 5.3.2(11), using the elastic
– An approach to establish the shape of the imperfection that buckling mode as the shape of the imperfection and assuming
should be considered if the first buckling mode is not the most the results of the equivalent member method, which extracts an
adverse shape for the design. equivalent member with the same buckling length but with pin-
ended boundary conditions, and computes its buckling strength
Several examples of the application of these proposals are given on the basis of column buckling curves. Quoting Chladný and štu-
in Section 7 of this paper. jberová [4], where experimentally established values are unavail-
able, the amplitude of the equivalent imperfection in the shape
1.2. Basic assumptions of the elastic buckling mode may be determined assuming that
buckling resistance of a structure with axially loaded members is
In this work two assumptions are considered: equal to the buckling resistance of the equivalent member. The
The first is that Clause 6.3.1.4 allows designers to use the v fac- buckling resistance of axially loaded columns for EC-3 is defined
tor, originally deduced from compression–flexural buckling in 6.3.1.1–6.3.1.4. and the non dimensional slenderness
k is defined
A. Agüero et al. / Engineering Structures 96 (2015) 160–177 163
Fig. 3. Out of plane imperfections for a pitched roof frame according to the present proposal.
164 A. Agüero et al. / Engineering Structures 96 (2015) 160–177
Fig. 4. In plane and out of plane imperfection according to the present proposal.
2 !2 !2
at the critical section [4,5]. This proposal is also included in EN Z 2 2 2
1999-1-1:2007 [20]. U1 ¼
1 4EA du þ E Iy d w þ E Iz d v
2 dx 2 2
Chladný and štujberová [4,5] concluded that the differences in L dx dx
imperfection effects according to Method 2.1 and Method 2.2 2 !2 3
2
dhx d hx 5
(5.3.2(3) and 5.3.2(11) in EC-3) are small for continuous frames þG It þ E Iw 2
dx ð2Þ
and, if Method 1 (5.2.2(3) c) is used, a good agreement with the dx dx
aforementioned methods is obtained, leading to conservative
"
2 2
results in most cases. 1
Z
dwL
dv dh 2
2 x
U2 ¼ þ N þ is þ y2sc þ z2sc
2 0 dx dx dx
2. Buckling study and analysis of the structure with a geometric dhx dv dhx dw T
þ2zsc 2ysc dx fdNL g fF ext g ð3Þ
imperfection ginit dx dx dx dx
The analysis presented in this paper is limited to Class 2 and 3 The value of the critical buckling load N cr cancels the determi-
sections in terms of EC-3 nomenclature. Class 1 and 4 are excluded nant of the stiffness matrix of the structural system,
for simplicity. j½K L þ N cr ½K G j ¼ 0. The critical buckling coefficient (acr ) can be
On the other hand, the equilibrium of the structure can be defined as the ratio between the critical buckling load and the
addressed using the finite element method and imposing the first reference load acr ¼ N cr =N 0 ; the first buckling mode fgcr g is asso-
variation of the total potential is zero, i.e. the sum of the first varia- ciated with the first critical buckling load, so can be written:
tion of the potential of all the elements is zero. According to the
ð½K L þ acr N0 ½K G Þfgcr g ¼ 0 ð7Þ
energy approach, the global balance of a structure sensitive to
buckling due to compression can be written in weak form The following sections describe the procedures to obtain the magni-
[21,22] adding the strain energy and the potential of external tude of the imperfection according to the type of buckling mode,
forces as follows: specifying whether the expressions of the warping constant (Iw )
A. Agüero et al. / Engineering Structures 96 (2015) 160–177 165
and/or if the shear stress due to Saint Venant torsional moment can Bimoment:
be neglected. Since the field of steel structures distinguishes 2
between open and closed thin-walled sections, flexural or torsional d gcr;hx
Bg ¼ EIw 2
ð16Þ
buckling may take place. Closed sections are only sensitive to flexu- dx
ral buckling (due to their high torsional stiffness) and open sections Torque:
are sensitive to flexural and/or torsional buckling. For isolated
3
members, three buckling cases are usually studied: dgcr;hx d gcr;hx
T g ¼ T t;g þ T w;g ¼ GIt EIw 3
ð17Þ
dx dx
1. Flexural buckling about y, z axis and torsional buckling are
uncoupled. The member can buckle in the xy plane, xz plane Saint Venant Torsional moment:
or can develop a torsional buckling mode. The buckling load dgcr;hx
and related modes can be obtained. The minimum buckling T t;g ¼ GIt ð18Þ
dx
load will define the buckling mode of the column.
Shear force in y and z direction:
2. Flexural torsional buckling. Bending about z axis is coupled
with torsions. 3 3
d gcr;v d gcr;w
3. Flexural torsional buckling. Bending about y and z axis are cou- V y;g ¼ EIz 3
and V z;g ¼ EIy 3
ð19Þ
dx dx
pled with torsions.
Due to an imperfection given by the first buckling mode the stress
As an example, Fig. 1 shows the buckling mode of different cross resultants will be obtained by amplifying the previous equations,
sections for an isolated member under compression with fork sup- according to the load level NðN Ed according to [1]), by the factor
1
port at both ends. To calculate the buckling load and mode in acr 1, where acr ¼ N cr =N.
tapered beams the recommendations given by Trahair [23] have
to be taken into account.
2.2.2. General imperfection
2.2. Analysis of the imperfect structure Generally, any imperfection fginit g can be expressed as a
n o
combination of the buckling modes gj taking into consideration
The equilibrium equation of the imperfect structure is given by:
the orthogonal properties of the buckling modes [22] (i.e.
ð½K L þ N 0 ½K G ÞfdNL g ¼ fF ext g N0 ½K G fginit g ð8Þ n o
T
if i – j ! fgi g ½K L gj ¼ 0), so:
n o
2.2.1. Imperfection defined by the first buckling mode X
n n o fginit gT ½K L gj
To obtain the displacements in the imperfect structure one fginit g ¼ bj gj ; where bj ¼ n oT n o : ð20Þ
must solve ð½K L þ N 0 ½K G ÞfdNL g ¼ N 0 ½K G fginit g. If the imperfection j¼1 gj ½K L gj
is defined by the first buckling mode fginit g ¼ fgcr g, the solution to
the aforementioned equation is fdNL g ¼ qfgcr g. The displacements associated with the imperfection are given by
Pn bj n o 1
This yields: fdg ¼ j¼1 aj 1 gj . For larger critical loads (aj "), the term aj 1
ð½K L þ N 0 ½K G Þqfgcr g ¼ N0 ½K G fgcr g; ð9Þ becomes very small and related imperfections should not be consid-
ered in the analysis.
and taking into account that:
ð½K L þ N cr ½K G Þfgcr g ¼ 0 ! ½K L fgcr g ¼ Ncr ½K G fgcr g ð10Þ 3. Procedure to assess the imperfection fginit g
then
The expressions obtained in Section 2 are defined in shape but
q Ncr ½K G fgcr g þ q N0 ½K G fgcr g ¼ N0 ½K G fgcr g ð11Þ not in magnitude. The scale factor ec , by which the first buckling
mode should be multiplied to completely define the imperfection
from which one can obtain
fginit g ¼ ec fgcr g ¼ ec fgg, is obtained by imposing the restriction
1 that the buckling resistance of the structure, obtained according to
q¼ ; ð12Þ
acr 1 Clause 6.3 EC-3 derived in Section 3.1 ab ¼ ault;k v=cM1 , is equal to
the buckling resistance obtained by applying the criterion of collapse
and therefore:
to the imperfect structure:
1
fdNL g ¼ fgcr g ð13Þ
acr 1 – In Class 2 sections the plastic resistance is reached.
Likewise, associated with the imperfection or buckling mode, the – In Class 3 sections the maximum von Mises stress reaches the
stress resultants are: bending moments about y, z axis (My;g ; Mz;g ) yield strength (6.2.1 EC-3).
bimoment ðBg Þ, torque (T g ) and shear (V y;g ; V z;g ) where g is asso-
3.1. Buckling resistance of the structure
ciated to the buckling mode:
Considering axial forces (ab ), the load level that makes the
bending moment about y axis:
structure collapse can be obtained following the next three steps
2
d gcr;w consistent with Chladný and štujberová [4], Bijlaard et al. [12],
My;g ¼ EIy 2
ð14Þ Wieschollek et al. [13] and Naumes [14]:
dx
bending moment about z axis: 1. From a linear analysis, ault;k ¼ min AðxÞ f y ðxÞ=NðxÞ is the mini-
2
mum force amplifier for the axial force configuration in mem-
d gcr;v bers to reach the characteristic resistance of the most axially
Mz;g ¼ EIz 2
ð15Þ
dx stressed cross section without taking buckling into account.
166 A. Agüero et al. / Engineering Structures 96 (2015) 160–177
According to Chladný and štujberová [4], Bijlaard et al. [12], On the other hand, in open thin-walled sections due to Saint
Wieschollek et al. [13] and Naumes [14] ault;k must be defined Venant torsional moment, shear stress generated in the imperfect
by ault;k ðxcr Þ ¼ Aðxcr Þ f y ðxcr Þ=Nðxcr Þ, where xcr is the critical structure can be obtained as:
section carrying out the nonlinear analysis of the imperfect
structure. Further details about the iterative procedure to find n ec T t;g t
s¼ ð22Þ
the critical section are described by Baláž [24–27], there is also ðw 1Þ It
graphical interpretation of the imperfection amplitude and the
location of the critical section [25,27]. n value between [0, 1] and depends on the check point.
2. Then, one can obtain the reduced slenderness of the structure For thin-walled sections, it can be assumed that shear stresses
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
from the expression k ¼ ault;k =acr , and related to shear forces and warping torsion due to the imperfection
3. Finally, the last step is to obtain the coefficient v associated can be neglected.
with the EC-3 buckling curves, which will depend on the type To obtain Cðx; y; zÞ which defines the magnitude of the
of section, whether it is rolled or welded and the buckling mode imperfection to reach the yield stress at point (x, y, z), the
2
of the member in which xcr takes place. Hence, the collapse load restriction r2 þ 3 s2 ¼ f y =cM0 must be imposed substituting r
level of the structure will be ab ¼ ault;k v=cM1 . and s:
2
N Cðx; y; zÞ Mz;g My;g Bg
3.2. Procedure to assess the magnitude ec for Class 3 sections ab þ þ þ
A w1 Wz Wy Wx
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2
First of all, checkpoints where the von Mises stress r2 þ 3 s2 n Cðx; y; zÞ T t;g t fy
þ3 ¼ ð23Þ
is expected to be highest should be chosen in the structure (x, y, z), ð w 1Þ It cM0
with engineering criterion.
For Class 3 sections, the magnitude of ec is obtained restricting
to the yield strength the largest von Mises stress in the structure
that takes place at one of the aforementioned checkpoints (x, y, z); This equation can be written as:
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
this condition can be expressed as max r2 þ 3 s2 ¼ f y =cM0 .
a Cðx; y; zÞ2 þ b Cðx; y; zÞ þ c ¼ 0 ð24Þ
To obtain the value of ec , it is possible to define a function
Cðx; y; zÞ that provides the magnitude of the imperfection to reach The solution to this equation is:
the yield stress at a generic point (x, y, z). The value of ec will be pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
given by the minimum of Cðx; y; zÞ, hence ec ¼ min jCðx; y; zÞj ¼ b
2
b 4a c
Cðxcr ; ycr ; zcr Þ. Cðx; y; zÞ ¼ ð25Þ
2a
To obtain von Mises stress in the imperfect structure, the
normal and shear stresses must be previously computed. The where
internal forces are then obtained at the buckling load consistent !2
with the EC-3 primary axial forces (ab N) and stress resultants cM0
related to the imperfection, which can be obtained scaling a¼ x21 þ 3x23 ;
f y ðw 1Þ
ec
{M y;g ; M z;g ; Bg ; T t;g } by the factor w1 , where w ¼ aacr .
b cM0
On one hand, normal stresses generated (Fig. 5) in the imperfect b¼2 x1 x2 ; c ¼ x22 1 ð26Þ
f y ðw 1Þ
structure can be obtained as:
where
N ec Mz;g M y;g Bg
r ¼ ab þ yþ z þ xðy; zÞ
A w 1 Iz Iy Iw Mz;g My;g Bg ab N cM0
x1 ¼ þ þ ; x2 ¼ ;
N ec Mz;g My;g Bg Wz Wy Wx A fy
¼ ab þ þ þ ð21Þ
A w 1 Wz Wy Wx n T t;g t
x3 ¼ ð27Þ
It
Iz Iy Iw
where W z ¼ ; W y ¼ ; W x ¼
y z xðy;zÞ
Table 1
Definition of x1 as a function of the stress resultants and the displacements of the buckling mode.
Flexural torsional buckling. Bending about z axis coupled with torsions M z;g B 2
d g d2 g
þ Wgx ¼ E WIz z 2v þ WIwx 2hx
M z;g
¼ EI z
2
d gv
Wz Wz Wz 2
dx dx dx
a
Torsional buckling Bg
¼ EI
2
w d ghx 0
Wx Wx 2
dx
Flexural buckling M z;g
¼
2
EIz d gv
Wz W z dx2
a
In case of torsional buckling if the warping constant is negligible Iw 0, the shear stresses cannot be neglected; otherwise it is not possible to obtain the magnitude of the
imperfection ec . This particular case is developed in Section 3.2.2.
Considering a linear interaction formula and neglecting torque 4.3. Several buckling loads close to the first buckling load
effects, a simplified expression can be used to obtain the plastic
resistance in the most critical section: On the other hand, when several independent buckling modes
n o
gj have their buckling loads (aj ) very close to the first buckling
N ab eC M z;g M y;g Bg
max þ þ þ ¼1 ð37Þ load (a1 ), it is also possible to estimate the worst direction apply-
Npl;Rd w 1 M pl;z;Rd M pl;y;Rd Bpl;Rd n o
T P
ing the energy method: U ¼ fkgggk fF EXT g, where fgg ¼ nj¼1 cj gj
To obtain the value of ec , a function XðxÞ may be defined that gives 2 n oT n o
the magnitude of the imperfection that makes the section located in
and gj ¼ gj ½K L gj , where (cj ) are defined as the magni-
position (x) reach the plastic strength. The value of ec will be given tudes that lead to maximum work.
by the minimum of XðxÞ, hence ec ¼ min jXðxÞj ¼ Xðxcr Þ.
N ab 1
XðxÞ ¼ 1 ð w 1Þ ð38Þ 5. Number of buckling modes to define the imperfection shape
Npl;Rd M z;g
þ
My;g
þB
Bg
M pl;z;Rd M pl;y;Rd pl;Rd
5.1. Checking if the imperfection is the most adverse for all the
and therefore implemented in the EC-3.
members under compression
4. Procedure to estimate the most adverse direction In some structural systems, considering a single imperfection
defined by the first buckling mode can lead to incorrect designs.
The direction that defines the most adverse imperfection must To illustrate this, a simple example is shown in Fig. 1, where it
be used in order to obtain the worst effects of the external loads can be seen that we need to define the imperfection using several
on the imperfect structure. Following a proposal to provide a buckling modes. For example, this may occur in cases in which two
suggestion of the worst direction of the imperfection is explained. or more substructures buckle independently, perhaps due to the
existence of moment releases. If when the first buckling mode is
obtained, one finds compression members ðKBÞ for which the
4.1. The first buckling load is related to one buckling mode
strain energy related to the buckling mode is close to zero
Once the magnitude and shape of the buckling mode is fg1 gT ½K L fg1 g 0, this means that the imperfection considered
KB
obtained, it is necessary to choose the direction among the two is not producing any adverse effect on this compression member,
likely buckling shape modes gþ1 ; g1 (see Fig. 1). but there may be another imperfection that could have an adverse
To estimate the most adverse direction an energy method is effect on this column. If both imperfections are not taken into
proposed, obtaining the work ðU Þ of the external loads when the account the structure could collapse because of poor design.
structure undergoes displacements defined by the imperfection. Another example is if the imperfection related to the first buckling
U ¼ fginit gT fF ext g. If this value is positive the chosen direction mode is affecting a part of the structure, and the most severely
provides the most adverse case, but if it is negative, the opposite loaded bar bearing external loadings is somewhere else. It should
direction should be considered. The criterion gives good results be ensured that there is an imperfection that affects the most
in the cases tested (see Section 7). severely loaded bar.
If the strength of the structure without considering the imper-
fection is smaller than considering the imperfection, designers
must consider changing the direction of the imperfection [7], 5.2. Criterion to use higher buckling modes to define the imperfection
so that it is always recommended to check the strength of the
n o
structure without imperfections. The lowest buckling mode j gj for which the strain energy
related to the buckling mode in member ðKBÞ is greater than zero
4.2. The first buckling load is related to several independent buckling n o n o
T
modes gj ½K L gj > 0 must therefore be found. If the critical load
KB
level associated aj to this buckling mode is less than 25, (aj < 25)
It may be possible that the first buckling load (a1 ) has asso- (the value of 25 was taken from the EC-3 recommendation in
n o
ciated j independent buckling modes gj , as can happen in sym- Clause 6.3.1.2 (4) NNcr < 0:04 ! acr ¼ NNcr > 25), the imperfection
metrical structures like towers, masts, chimneys (EN 1993-3-1) or should be taken into consideration as calculated by the method
silos (EN 1993-4-1). In this case the analysis of the structure described in Sections 2 and 3 of this paper, using the corresponding
considering infinite imperfections with the shape associated to critical load and buckling mode.
different buckling modes has to be performed and the strength of An extended criterion of the EC-3 applicable to all structures
the structure will be the minimum. This option is not acceptable could be to consider all the imperfections associated with a buck-
from a practical point of view, as only a few imperfections could ling mode whose critical load is lower than 25 (aj < 25). It must be
be checked. It is suggested that one of the shapes to estimate the remembered that the computation of imperfections should be
worst imperfection is given by the buckling mode that maximizes performed for each significant load combination and cannot be
the aforementioned work. considered simultaneously.
A. Agüero et al. / Engineering Structures 96 (2015) 160–177 169
k v 2
a k 0:2 1 cM1 fy
ec ¼ ð46Þ
k2 1 k v ðNcr;TF ySC Þ= W y Ncr;TF =N cr;y 1 þ ðNcr;TF zSC Þ=ðW z ðNcr;TF =Ncr;z 1ÞÞ þ p22 EIw
2
L Wx
– Estimate the most adverse direction of the imperfection Example 2. Unequal angle section (LD) column under compres-
(Section 7.2) Examples 10 and 11. sion (Fig. 7) fysc – 0; zsc – 0g.
– Prove the need to consider higher buckling modes to define the The buckling mode and load are given by the same expressions
shape of the imperfection (Section 7.3) Example 12. as in Example 1.
– Apply the simplified nonlinear analysis of the imperfect struc- The maximum von Mises stress is reached at midspan,
ture (Section 7.4) Example 13. where s ¼ 0 (Fig. 7), so that the same expression can be obtained
– Obtain the amplification of linear effects in a structure sensitive for the scale factor ec as in Example 1, but neglecting the term
to flexural–torsional buckling (Section 7.5) Example 14. p2 EIw 0.
2
L Wx
7.1. Examples of obtaining the geometric equivalent imperfection Example 3. I-section column under compression with the sup-
ports and shape of the imperfection shown in Figs. 8 and 9.
Nine examples illustrate the definition of the geometric imper- To see how the equivalent imperfection is obtained, the normal,
fection. Table 3 classifies the examples according to the type of shear and von Mises stress in the imperfect beam can be seen in
buckling mode and whether or not Iw can be neglected. Fig. 9.
170 A. Agüero et al. / Engineering Structures 96 (2015) 160–177
Fig. 8. Lateral imperfection at the shear center, top and bottom flange (Example 3).
Fig. 10. Lateral imperfection at the shear center, top and bottom flange (Example 4).
Fig. 14. Lateral imperfection at the shear center, top and bottom flange (Example 7).
1. For b between 90° and 270° the strength of the structure could
be increased, leading to unsafe designs, and the maximum
strength is obtained for b = 180°. If the energy method is
applied, U becomes negative.
2. For b between 90° and 90° the strength of the structure is
reduced, the minimum strength is obtained for b = 0°, so the
worst imperfection is that obtained with the energy method
proposed. In this case allowable values of b to define the imper-
fection will be those very close to 0. If the energy method is
applied, the work U is positive.
Example 11. The previous example (Fig. 18) can be analyzed but
For the cantilever there are infinite buckling modes related to considering an elliptical section where the major and minor axis
the first buckling load. The buckling mode i, depending on b is: have very similar lengths. In this case the first and second buckling
px modes have almost the same buckling load. Solving the problem,
ginit v ðxÞ ¼ ec 1 cos cos ðbÞ; the same conclusions are obtained as in Example 10.
2L
px
ginit w ðxÞ ¼ ec 1 cos sin ðbÞ ð59Þ 7.3. Example of proving the need to consider higher buckling modes to
2L
define the imperfection
Applying the energy method suggested in Section 4, the work
(U) is given by:
Example 12. Given a frame with out-of-plane buckling restrained
U ¼ fginit gT fF ext g ¼ M ginit hy þ Q ginit v ð60Þ (Fig. 19a), we check whether the design of the structure can be
done using only the first buckling mode to define the imperfection.
The displacement at the free end is:
The imperfection related to the first buckling mode (Fig. 19b)
ginit ðLÞ ¼ eC ð61Þ does not affect the column on the right; this can be numerically
detected as the strain energy related to this buckling mode at this
At the free end, the imperfection in direction y is given by:
column is zero g1 T ½K L g1 0. Nevertheless, the imper-
ginit v ðLÞ ¼ ec cosðbÞ ð62Þ KB
fection related to the second buckling mode (Fig. 19c) has the
The rotation about y axis is given by: worst effect on the design of this column. So both imperfections
p ec must be considered for a proper design of the structure.
ginit hy ðLÞ ¼ sinðbÞ ð63Þ
2L
The work is provided by: 7.4. Example of applying the simplified nonlinear analysis of the
imperfect structure
M p sin ðbÞ
U ¼ ec þ Q cos ðbÞ ð64Þ
2L
Example 13. Given a frame with out-of-plane buckling restrained,
And the angle that makes the work maximum is: TanðbÞ ¼ p2LQ
M
. the expression for the nonlinear displacements is obtained.
If Q = 0 then b=-90°, and if M = 0 then b=0°. Considering that the first buckling mode is a sway mode gS
A. Agüero et al. / Engineering Structures 96 (2015) 160–177 175
Fig. 19. Frame definition (a), first buckling mode (b) and second buckling mode (c) (Example 12).
and the second buckling mode is a nonsway mode gNS , as shown 7.5. Example of obtaining the amplification of linear effects in a
in Fig. 20, their corresponding buckling loads (aS ) and (aNS ) are structure sensitive to flexural–torsional buckling
lower than 10 and the other buckling modes have a critical load
larger than 10. Example 14. Given a HEB240-section column (length, L = 6 m)
The expression of the nonlinear displacements of the imperfect with the supports shown in Fig. 21, the system is analyzed when
structure is given by: the critical load level is 1.4 (a1 ¼ 1:4), subjected to the four
load cases shown in Fig. 21. The linear out-of-plane displacements
fgS gT fF ext g 1 fgNS gT fF ext g
fdNL g ¼ fdL g þ fgS g þ are obtained and compared with the exact nonlinear and the
fgS g ½K L fgS g ðaS 1Þ
T
fgNS gT ½K L fgNS g approximate nonlinear displacements using one buckling
1 fg g
fg g þ init k ð65Þ mode fdNL g ¼ fdL g þ ða1a1 1
Þ g1 , and two buckling modes
ðaNS 1Þ NS ðak 1Þ
fdNL g ¼ fdL g þ ða1a1
1
Þ g 1 þ a2
ða2 1Þ g 2 .
The second term in the above equation includes the amplifications
In the simplified procedure the displacements are obtained
related to the sway mode known as P–D effect. The third term in the P n o
aj
above equation includes the amplifications related to the non-sway using: fdNL g ¼ fdL g þ nj¼1 a 1 g thus the designer must
ð j Þ j
mode known as P–d effect, usually included as an interaction compute the buckling modes associated with critical loads
formula.
aj 1 6 10. Fig. 22 shows the first five buckling modes (the first
The fourth term is the imperfection, and the buckling modes
g , third g3 and fourth g4 are torsional modes, the second
associated with critical loads lower than 25 (aj < 25) should be 1
considered, as mentioned in the previous example. g2 and fifth g5 are flexural modes), and for each external load
T
If the case in which only the sway mode has a critical load lower fgj g fF ext g
the coefficients aj ¼ T are computed.
than 10 is considered, only P–D effects are considered: fgj g ½K L fgj g
From Fig. 21 it can be concluded that if two buckling modes are
fgS gT fF ext g 1 fg g
fdNL g ¼ fdL g þ fgS g þ init k ð66Þ used, the results can be accepted as exact. With one buckling mode
fgS g ½K L fgS g ðaS 1Þ
T ðak 1Þ
the displacements in the top flange are the same as in the linear
The EC-3 in Clause 5.2.2 proposes calculating the nonlinear dis- case, due to the fact that in the first buckling mode the top flange
placements associated with horizontal loads by scaling the linear does not have any displacements. To make a good guess at the
displacements by 1=ð1 1=aS Þ. This can be deduced from the above nonlinear displacements due to the Bimoment (Load Case 4) only
equation, since linear displacements fdL g are almost proportional to one buckling mode has to be taken into account.
the sway mode displacements fgS g Once the displacements are obtained the stress resultants
torque, bimoment and minor axis moment are computed using
fdL g fdL g
fdNL g ¼ fdL g þ ¼ ð67Þ mechanics of materials equations.
ðaS 1Þ ð1 1=aS Þ
Fig. 20. Nonlinear effects related to sway and nonsway mode (Example 13).
176 A. Agüero et al. / Engineering Structures 96 (2015) 160–177
Fig. 21. Comparison of displacements: linear, nonlinear exact and approximated (Example 14).
Fig. 22. Nonlinear effects related to the buckling modes (Example 14).
[4] Chladný E, štujberová M. Frames with unique global and local imperfection in [18] Taras A, Gonzalez M, Unterweger H. Behaviour and design of members with
the shape of the elastic buckling mode (part1). Stahlbau 2013;8:609–17. monosymmetric cross-section. Proc ICE: Struct Build 2013;166(8):413–23.
[5] Chladný E, štujberová M. Errata: frames with unique global and local [19] Trahair NS. Strength design of cruciform steel columns. Eng Struct
imperfection in the shape of the elastic buckling mode. Part 1. Stahlbau 2012;35:307–13.
2013;82:H. 8, S. 609–617. Part 2. Stahlbau 2013;83:H. 9, S. 684–694. Stahlbau [20] EN 1999-1-1:2007. Eurocode 9: design of aluminium structures. Part 1.1:
2013;82:H. 9 S. 684-694, Stahlbau 2014;1:64-64. General structural rules.
[6] Gonçalves R, Camotim D. On the incorporation of equivalent member [21] Trahair NS. Flexural–torsional buckling of structures. E&Fnspon; 1993.
imperfections in the in-plane design of steel frames. J Constr Steel Res [22] Chen WF, Atsuta T. Theory of beam–columns. McGraw-Hill; 1977.
2005;61:1226–40. [23] Trahair NS. Bending and buckling of tapered steel beam structures. Eng Struct
[7] Clarke MJ, Bridge RQ, Hancock GJ, Trahair NS. Advanced analysis of steel 2014;59:229–37.
building frames. J Constr Steel Res 1992;23:1–29. [24] Baláž I. Determination of the flexural buckling resistance of frames with
[8] Agüero A, Pallarés FJ. Proposal to evaluate the ultimate limit state of slender members with non-uniform cross-section and non-uniform axial compression
structures. Part 1: Technical aspects. Eng Struct 2007;29:483–97. forces. Zborník z XXXIV. Aktívu pracovníkov odboru OK so zahraničnou
[9] EN 1993-2. Eurocode 3: Design of steel structures. Part 2: Steel bridges. účast’ou ‘‘Teoretické a konštrukčné problémy ocel’ových a drevených
[10] Chladný E, štujberová M. Frames with unique global and local imperfection in konštruckcií a mostov’’. 16.–17. 10. 2008, Pezinok. p. 17–22.
the shape of the elastic buckling mode (Part 2). Stahlbau 2013;9:684–94. [25] Baláž I, Koleková Y. Metal frames with non-uniform members and/or
[11] Marques L, Taras A, Simões da Silva L, Greiner R, Rebelo C. Development of a nonuniform normal forces with imperfections in the form of elastic buckling
consistent buckling design procedure for tapered columns. J Constr Steel Res mode. Engineering research. Anniversary volume honoring Amália and Miklós
2012;72:61–74. Iványi. Pollak Milhaly Faculty of Engineering. Univeristy of Pécs; October 25–
[12] Bijlaard F, Feldmann M, Naumes J, Sedlacek G. The ‘‘general method’’ for 26, 2010. p. B:3–B:15.
assessing the out of plane stability of structural members and frames and the [26] Baláž I, Koleková Y. In plane stability of two hinged arches. In: Proceedings of
comparison with alternative rules in EN1993 – Eurocode3 – Part1-1. Steel European conference on steel and composite structures, Eurosteel 2011. 31.
Constr 2010;3(1):19–33. August – 2. September 2011. Budapest, vol. C. p. 1869–74.
[13] Wieschollek M, Schillo N, Feldmann M, Sedlacek G. Lateral–torsional buckling [27] Baláž I, Koleková Y. Structures with UGLI imperfections. In: CD proceedings of
checks of steel frames using second-order analysis. Steel Constr 2012;5(2): 18th international conference engineering mechanics 2012. Svratka, Czech
71–86. republic, May 14–17, 2012, paper no. 233. p. 61–86.
[14] Naumes J. Biegeknicken und Biegedrillknicken von Stäben und Stabsystemen [28] Trahair NS, Pi YL. Torsion, bending and buckling of steel beams. Eng Struct
auf einheitlicher Grundlage. RWTH Aachen University, doctoral thesis, Aachen, 1997;19(5):372–7.
Shaker Verlag, Institution series – steel structures, no. 70; 2009. [29] Rotter M. Shell buckling design and assessment and the LBA-MNA
[15] Taras A, Greiner R. Development of consistent buckling curves for torsional methodology. Stahlbau 2011;80(11):791–803.
and lateral–torsional buckling. Steel Constr 2008;1:42–50. [30] Ayrton WE, Perry J. On struts. Engineer 1886;62:464–5. a. p. 513–5.
[16] Taras A, Greiner R. Torsional and flexural–torsional buckling – a study on [31] Rondal J, Maquoi R. Formulation d’Ayrton–Perry pour leflambement des barres
laterally restrained I-sections. J Constr Steel Res 2008;64:725–31. métaliques. Constr Métall 1979;4:41–53.
[17] Taras A, Greiner R. New design curves for LT and TF buckling with consistent [32] Maquoi R, Rondal J. Mise en équation des nouvellescourbes européennes de
derivation and code-conform formulation. Steel Constr 2010;3(3):176–86. flambement. Constr Métall 1978;1:17–30.