Documente Academic
Documente Profesional
Documente Cultură
In many cases, instability is the most important limit state for structural members
made of steel or aluminium alloys. This phenomenon can affect a part of the
member, the entire member, a part of the structure or the entire structure.
This is the type of loss of stability that affects bars in compression or compressed
parts of bars. At a certain value of the load, the bar in compression or the
compressed part of the bar (involving the un-compressed part too) finds equilibrium
in a deformed shape, in the neighbourhood of the straight one.
This problem is probably the most studied one in the history of stability problems.
Beginning with Euler, 1744 [1], different researchers tried to express the equilibrium
and the failure mode of a perfectly straight member subject to axial compression [2].
When subject to an axial compression force, a straight member may lose its stability
in one of the following forms (Fig. 1.1):
• flexural buckling (v ≠ 0; ϕ = 0) (Fig. 1.1a);
• torsion buckling (v = 0; ϕ ≠ 0) (Fig. 1.1b);
• flexural-torsion buckling (v ≠ 0; ϕ ≠ 0) (Fig. 1.1c);
where v means the lateral displacement in the plane of the cross-section and ϕ is the
rotation of the cross-section in its plane.
1
Fcr Fcr Fcr
ϕ ϕ
v v
The buckling load is the critical force Fcr at which a perfectly straight member in
compression assumes a deflected position (Fig. 1.1). Buckling is a limit state, in the
meaning that once the force Fcr is reached, the deflection increases until the collapse
of the bar is reached. The member should be subjected only to loads inferior to the
critical force (F < Fcr) [2].
In the same way as for any member in compression, the buckling problem appears
for the compressed flange of a beam. Generally, buckling (Fig. 1.2) may not occur in
the plane of the web, as the compressed flange is continuously connected through
the web material to the tensioned part of the cross-section, the tension flange. The
stabilizing effect of the tension zone transforms free transverse buckling into lateral-
torsional buckling, causing lateral bending and twisting of the beam [2].
2
Lateral buckling
of the flange
Torsion
(twisting
of the
beam)
span
Fig. 1.2. Lateral-torsional buckling of a beam [2]
The arch is a very efficient structural member for resisting symmetrical loads acting
in its plane, normally to the line of its supports. Its important strength comes from the
arm lever exiting between the compressed part (the arch) and the tensioned part (the
line of supports) (Fig. 1.3), which is much bigger than the distance between the
compressed flange and the tension one in the case of a beam.
3
The horizontal stiffness of the supports (the abutment stiffness) is vital for the
performance of the arch; the smaller this one is, the closer the behaviour is to a
“curved beam” and strong values of the bending moment can be found along the
arch.
Local buckling of a plate may occur as a result of the action of in-plane normal
compression stresses (σ), of shear ones (ττ), or of their combination (Fig. 1.5).
4
1.1.2.2. Local buckling of shells
Similarly to the case of plates, in-plane compression stresses (σ) can lead to local
buckling of shells (Fig. 1.6).
Because of the big values of the angles among bars (Fig. 1.7), strong compression
forces result in the bars. Because of the strains in these bars, the geometry of the
structure changes and, in some circumstances, equilibrium can be found in tension.
5
1.1.3.2. Instability of structures containing members in compression
A part of a structure or the entire structure can go unstable, as the end supports or
the joints of members in compression or in compression and bending have limited
stiffness. Figure 1.8 shows the case of a plane frame and of the reticulated structure
of a roof.
Tension members do not generate static instability of structures. However, they can
be subject of dynamic instability of structures like vibrations, resonance, “flutter” etc.
A very well-known example is the failure of Tacoma Narrows Bridge on November 7,
1940 (Fig. 1.9).
6
1.2. COMMENTS
There are well known situations when the recorded value of the
snow load, for instance, were much higher than the design ones and
no failure was noticed, as no instability problems were involved.
7
1.3. CLASSES OF CROSS-SECTIONS
Generally, given the strength of steel and aluminium alloys, failure of a metal
member subjected to loads other than tension occurs by buckling or by local
buckling. Depending on the slenderness of the element, this can happen either in the
elastic range (0 – Y in figure 1.10) or in the plastic range (Y – F in figure 1.10). To
manage this, EN 1993-1-1 [6] defines four classes of cross-sections of structural
members. They are best expressed for members in bending. In these definitions, the
behaviour of the material is presumed perfectly elastic up to the yielding limit and
perfectly plastic for elongations superior to the strain corresponding to the yielding
limit (Fig. 1.10). This model is known as the Prandtl model.
σ
real
Y Prandtl F
fy
0 εy εu ε
Depending on the stress state that causes local buckling, cross-sections of structural
members are classified as [6] (Fig. 1.11):
Class 1 – cross-sections that can form a plastic hinge with sufficient rotation
capacity to allow redistribution of bending moments. Only class 1 cross-
sections may be used for plastic design.
Class 2 – cross-sections that can reach their plastic moment resistance but local
buckling may prevent development of a plastic hinge with sufficient
rotation capacity to permit plastic design (redistribution of bending
moments).
8
Class 3 – cross-sections in which the calculated stress in the extreme
compression fibre can reach the yield strength but local buckling may
prevent development of the full plastic bending moment.
Class 4 – cross-sections in which it is necessary to take into account the effects of
local buckling when determining their bending moment resistance or
compression resistance.
For practical reasons, the limits between these classes are expressed in terms of
slenderness. Tables 1.1, 1.2, 1.3 show the requirements for different cross-sectional
classes. The class of a cross-section is the maximum among the classes of its
components.
(–)
y y x x
(+)
y y x x
z
Fig. 1.11. Possible stress distribution, depending on the cross-section class
The plastic hinge is a concept. It is a model of a cross-section where all the fibres
reached the yielding limit in tension or compression (Fig. 1.11) generated by a
bending moment, presuming a bi-linear (Prandtl) behaviour diagram for the material,
while in the neighbour cross-sections the stress state is elastic. In reality, the stress
and strain state is more complex (Fig. 1.12): the material behaviour is not ideally
elasto-plastic and the plastic deformations extend on a certain length.
9
Fig. 1.12. The stresses in the region of a plastic hinge
Tab. 1.1. Limitations for the slenderness of internal walls (EN 1993-1-1 [6] Tab. 5.2)
c c c c
Bending axis
t t t t
t t t
t
c c c
c
Bending axis
10
fy fy
Stress distribution fy
c c c
fy c/2
ψfy
c 42ε
when ψ > –1: ≤
3 c c t 0,67 + 0,33ψ
≤ 124ε ≤ 42ε
t t
≤ 62ε (1 − ψ) (− ψ)
c
when ψ ≤ –1:
t
Tab. 1.2. Limitations for the slenderness of flanges (EN 1993-1-1 [6] Tab. 5.2)
t t t t
c c c
c
Stress distribution αc αc
c c c
1 c c 9ε c 9ε
≤ 9ε ≤ ≤
t t α t α α
2 c c 10ε c 10ε
≤ 10ε ≤ ≤
t t α t α α
11
Stress distribution
c c c
3 c c
≤ 14ε ≤ 21ε k σ
t t
Tab. 1.3. Limitations for the slenderness of the walls of round tubes (EN 1993-1-1 [6]
Tab. 5.2)
In a similar manner, the American code ANSI/AISC 360-16 [7] distinguishes between
compact, noncompact and slender cross-sections. The Japanese code JSCE [8]
makes reference to EN 1993-1-1 [6] and associates compact to class 1 and class 2
cross-sections, noncompact to class 3 and slender to class 4 ones, using its own
limits.
12
state of the cross-section in compression and bending which is defined using the
following relations:
• for class 1 and class 2 cross-sections, the check for bending moment is done
with the following relation:
M Ed ≤ M N , Rd (EN 1993-1-1 [6] rel. (6.31)) (1.1)
where MN,Rd is the design plastic moment resistance reduced due to the axial
force NEd; therefore, to determine the class of the cross-section, one must
establish the stress distribution associated to MN,Rd (Fig. 1.13).
For doubly symmetrical I- and H- sections or other flanges sections, no
reduction of the plastic resistance moment MN,y,Rd about the y-y axis needs to
be done when both the following criteria are satisfied:
N Ed ≤ 0,25 ⋅ N pl, Rd (EN 1993-1-1 [6] rel. (6.33)) and (1.2)
0,5 ⋅ h w ⋅ t w ⋅ f y
N Ed ≤ (EN 1993-1-1 [6] rel. (6.34)) (1.3)
γ M0
Otherwise, the following reduction is done:
1− n
M N , y , Rd = M pl, y , Rd ⋅ but M N , y, Rd ≤ M pl, y, Rd (EN 1993-1-1 [6] rel. (6.36)) (1.4)
1 − 0,5 ⋅ a
where:
N Ed
n=
N pl, Rd
A − 2 ⋅ b ⋅ tf
a= but a ≤ 0,5
A
A ⋅ fy
N pl,Rd = (EN 1993-1-1 [6] rel. (6.6)) (1.5)
γ M0
tw – web thickness;
hw – web height;
b – flange width;
tf – flange thickness;
Wpl,y ⋅ f y
M pl, y,Rd = (EN 1993-1-1 [6] rel. (6.13)) (1.6)
γ M0
Wpl– plastic section modulus;
γM0 – partial factor for resistance of cross-sections whatever the class is; the
recommended value in EN 1993-1-1 [6] is γM0 = 1,0;
13
• for class 3 cross-sections, the check for bending moment is done with the
following relation:
fy
σ x ,Ed ≤ (EN 1993-1-1 [6] rel. (6.42), (6.43)) (1.7)
γ M0
therefore, to determine the class of the cross-section, one must establish
the stress distribution corresponding to reaching the yielding limit fy in
the farthest fibre from the centre of gravity of the cross-section (Fig.
1.13).
Depending on the class of cross-section and on the amount of axial force NEd,
relations (1.1) and (1.7) define an interaction curve as the one given in figure 1.13.
In practical situations, the stress state on the cross-section is inferior to the limit one.
The question that arises is how to reach the limit, as several options are possible
(Fig. 1.13):
• increasing the axial force (Fig. 1.13(a));
• increasing the axial force and the bending moment according to a variation
law (Fig. 1.13(b));
• increasing the bending moment (Fig. 1.13(c));
(c)
(b)
(a)
14
The approach in figure 1.13(c), by increasing the bending moment only, is consistent
to the present day version of EN 1993-1-1 [6]. In this idea, using the notations and
the stress distributions given in tables 1.1 and 1.2, the values of α, for the distinction
between class 1 and class 2, and ψ, for the distinction between class 3 and class 4,
should be determined. In these tables, (+) means compression.
In the case of cross-sections subjected to axial force NEd and bending moment about
the y – y axis, My,Ed, (Mz,Ed = 0), there is no normal stress (σ) variation with the width
of the flange (σ is constant on the entire width of the flange). Therefore, the class of
the flange is determined by simply comparing the slenderness of the wall to the
corresponding limits given in table 1.2:
• “non-internal” wall – for flanges of I, H and channel cross-sections and for side
parts of the flanges of box cross-sections;
• “internal” wall in constant compression – for central parts of the flanges of box
cross-sections.
In the case of cross-sections subjected to axial force NEd and bending moment about
the y – y axis, My,Ed, the limits for the class of the web depend on the value of the
axial force NEd and they are determined in two main step:
• determine α – the fraction of the web in compression – for classes 1 and 2;
• determine ψ – the ratio between the stress in the extreme fibre opposite to the
one (possibly) yielding in compression and the yielding limit – for class 3 (and 4).
15
N Ed
hN = (1.8)
nw ⋅ tw ⋅fy
3. If hN < c, the height of the tensioned part (for plastic stress distribution) of the
web is
c − hN
hT = (1.9)
2
4. α – the fraction of the web that is in compression (Tab. 1.1)
hT + hN
α= (1.10)
c
5. The obtained value for α is used for determining the limits (for class 1 and 2)
in table 1.1.
6. If the slenderness of the web is bigger than the limit for class 2, the limit
between class 3 and class 4 must be checked.
7. Determine the stress generated by the compression force on the area of the
cross-section (A) (Tab. 1.1)
N Ed
σN = (1.11)
A
8. Determine the stress range “available” for bending (Tab. 1.1)
σM = f y − σ N (1.12)
9. ψ – the ratio between the stress in the extreme fibre opposite to the one
(possibly) yielding in compression and the yielding limit (Tab. 1.1)
− σM + σ N
ψ= (1.13)
fy
16
1.3.2.3. Establish the class of the cross-section
For each component of the cross-section, the check is done successively for class 1,
2 and 3, till the class is established (if the slenderness is bigger than the limit for
class 3, then it is class 4). The class of the cross-section is the maximum of the
classes of its components.
17
Chapter 2
INSTABILITY OF BARS
2.1. TORSION
It occurs when all the following assumptions are accomplished (Fig. 2.1):
• the torsion moment is constant along the bar;
• the area of the cross-section is constant along the bar;
• there are no connections at the ends or along the bar that could prevent
warping.
18
2.1.1.1. Stress and strain state
T Ed = τ × r ⋅ dA (2.2)
A
• each cross-section rotates like a rigid disk (it goes out of plane but the shape
does not change);
• the rotation between neighbour cross-section is the same along the bar.
dϕ
θ= = const. (2.3)
dx
It occurs anytime when at least one of the St. Venant assumptions is not fulfilled
(Fig. 2.3).
19
Fig. 2.3. Warping torsion
20
2.1.2.2. Equilibrium equations
X Ed ,i = 0 N Ed = 0 σ w dA = 0 (2.6)
A
M y , Ed ,i = 0 M y , Ed = 0 σ w ⋅ zdA = 0 (2.7)
A
M z , Ed ,i = 0 M z , Ed = 0 σ w ⋅ ydA = 0 (2.8)
A
• in each cross-section, the torsion moment is the sum of the St. Venant
component and the warping component (Fig. 2.5):
TEd = τ ⋅ r ⋅ dA + Vw ⋅ h e (2.9)
A
TEd = Tt , Ed + Tw , Ed (2.10)
where:
Tt,Ed – the internal St. Venant torsion;
Tw,Ed – the internal warping torsion.
21
For a I- or H- cross-section, the force F, acting in the plane xOz, generates only
bending moment about the y – y axis (and shear force) and no torsion moment, as
the resultant forces Ff on the flanges are balanced (Fig. 2.6).
Fig. 2.7. Shear stresses for force acting in the centre of gravity
TEd = Ff ⋅ h e + Fw ⋅ e (2.11)
22
Fig. 2.8. Shear stresses for force acting in the shear centre
VEd ⋅ c = Ff ⋅ h e + Fw ⋅ e (2.13)
Ff ⋅ h e + Fw ⋅ e
c= (2.14)
VEd
Ff
Notations: α= ; Fw = VEd (2.15)
VEd
α ⋅ VEd ⋅ h e + VEd ⋅ e
c= (2.16)
VEd
c = α ⋅ he + e (2.17)
23
2.1.4. Torsion – calculation
1
b IT = ⋅ b ⋅ t3 (2.19)
3
dϕ T
θ= = ϕ′ = Ed = const. (2.20)
dx G ⋅ IT
TEd = G ⋅ I T ⋅ ϕ′ (2.21)
TEd ,1 TEd,n T Ed ,i
TEd
θ= = ... = = i
= (2.22)
G ⋅ I T ,1 G ⋅ I T ,n n
G ⋅ IT
G ⋅ I T ,i
1
1 n
IT = ⋅ b i ⋅ t 3i (2.23)
3 1
Remark: For hot-rolled shapes,
α n
IT = ⋅ b i ⋅ t 3i α = 1,1 … 1,3 (2.24)
3 1
TEd ⋅ t max
τ max = tmax = maximum thickness (2.25)
IT
TEd = G ⋅ I T ⋅ ϕ′ (2.26)
The case of hollow sections (Fig. 2.10)
24
TEd = Va ⋅ b + Vb ⋅ a (2.27)
Vb TEd
τb = = (2.31)
b ⋅ tb 2 ⋅a ⋅ b ⋅ tb
TEd
τ max = (2.32)
2 ⋅ A ⋅ t min
An exact calculation would consider the bar as a sum of shells (Fig. 2.11).
25
In daily practice a simplified approach is used, based on the Vlasov theory. The
simplifying assumptions are the following ones:
1. rigid disk behaviour:
• each cross-section rotates one about the other;
• the rotation varies from one cross-section to the other but it is constant
for all the points on the same cross-section;
• the rotation occurs around an axis parallel to the axis of the bar (Fig.
2.12);
2. the shear deformations are zero in the mid-line of the cross-section (Fig.
2.13);
mid-line
26
Based on these assumptions, the cross-section of the bar is reduced to its mid-line
(Fig. 2.14) and the following relations can be written between in-plane strains and
longitudinal ones (Fig. 2.15), considering rotation around point C:
nn ' = dv (2.33)
du dv
= (2.34)
ds dx
nn ' = nn ′′ ⋅ cos α (2.35)
dv = nn ' = nn ′′ ⋅ cos α (2.36)
mid-line
nn ′′ = Cn ⋅ dϕ (2.37)
dv = nn ' = Cn ⋅ dϕ ⋅ cos α (2.38)
r = Cn ⋅ cos α (2.39)
dv = r ⋅ dϕ (2.40)
du r ⋅ dϕ dϕ
= du = r ⋅ ds ⋅ (2.41)
ds dx dx
du
ε= ε = ω ⋅ ϕ′′ (2.45)
dx
du
27
By definition (Fig. 2.15),
r ⋅ ds
r ⋅ ds = dω = 2 ⋅ (2 × area of the triangle ) (2.42)
2
Notation (Fig. 2.15):
[ ]
s s
ω = r ⋅ ds = dω L2 normalised warping function (coordonată sectorială) (2.43)
0 0
σ w ⋅ ω ⋅ dA = E ⋅ ϕ′′ ⋅ ω2 ⋅ dA (2.47)
(bimoment de încovoiere-răsucire)
Vz , Ed ⋅ S y M w , Ed ⋅ S w
τz = τw = (2.51)
t ⋅ Iy t ⋅ Iw
ω ⋅ z ⋅ dA
yC = A
(2.53)
Iy
ω ⋅ y ⋅ dA
zC = A
(2.54)
Iz
28
2.1.5. Cross-section characteristics associated to torsion
• the position of the shear centre S from the bottom fibre of the cross-section:
t2 b3 ⋅ t
z SC = + hs ⋅ 3 1 13 (2.55)
2 b 2 ⋅ t 2 + b1 ⋅ t1
The first known theoretical approach for solving a bar in compression belongs to
Euler (1744) [1]. He started by writing the following equilibrium equation (Fig. 2.17)
for a pin connected bar axially loaded in compression:
29
d2v
dx 2 M 1
=− = (2.58)
dv 2 32
EI ρ
1 +
dx
where:
M = F⋅ v (2.59)
L
F x
e0
Fig. 2.17. The equilibrium of a pin connected bar in compression
for the critical force that generates buckling of the bar and:
π⋅x
z = e 0 ⋅ sin (2.61)
L
for the deformed shape of the bar.
This relation was then extended to other types of restraints at the ends, by inscribing
the bar on an equivalent pin-connected bar (Fig. 2.18). To allow this, the buckling
length was defined as a concept. All these theoretical approaches are based on the
theory of bifurcation of equilibrium.
The system length (EN 1993-1-1 [6] def. 1.5.5) is the distance in a
given plane between two adjacent points at which a member is
braced against lateral displacement in this plane, or between one
Definition such point and the end of the member.
The buckling length (Lcr) (EN 1993-1-1 [6] def. 1.5.6) is the system
length of an otherwise similar member with pinned ends, which has
the same buckling resistance as a given member or segment of
Definition member.
30
It is also defined as the distance between two consecutive inflection
points along the deformed shape of a bar. Sometimes, in practice, it
is replaced by the system length.
In everyday situations, bars are part of a structure, they are connected to other bars
and so the joints are not purely fixed or purely pinned. As a result, the buckling
length of an element depends on its loading state and on the stiffness of the
neighbour bars. Relations for calculating it are given in different books and
publications like Annexe E (informative) of the previous version of Eurocode 3 – ENV
1993-1-1 [12], or SN 008a-EN-EU [13]. For defining the buckling length of a column,
(parts of) structures are separated in “sway” and “non-sway”, depending whether the
(lateral) displacements of the joints at the end of the bar are permitted or not. This
separation is done by means of stiffness criteria; such a criterion is formulated in
ENV 1993-1-1 [12], where it is stated that a steel structure can be considered as
31
braced if the brace system reduces its horizontal displacements by at least 80%.
Usually, the “non-sway” behaviour is guaranteed by means of bracings. The
distribution factors ηi used in figure 2.19 – 2.22 are calculated using the following
relations:
KC
η1 = (ENV 1993-1-1 [12], rel. (E.1)) (2.63)
K C + K11 + K12
KC
η2 = (ENV 1993-1-1 [12], rel. (E.2)) (2.64)
K C + K 21 + K 22
where:
KC – stiffness of the column (I/L);
Kij – stiffness of the beam ij.
Remark: A more precise formulation for Kij would be “stiffness of the connection”
between beam ij and column, as semi-rigid connections could be used. In this case
a more careful analysis should be carried out.
The buckling length for non-sway buckling mode is presented in figure 2.19 [12] and
it is determined using factors obtained according to figure 2.21 [12].
Fig. 2.19. Non-sway buckling mode (ENV 1993-1-1 [12] Fig. E.2.3)
32
The buckling length for sway buckling mode is presented in figure 2.20 [12] and it is
determined using factors obtained according to figure 2.22 [12].
Fig. 2.20. Sway buckling mode (ENV 1993-1-1 [12] Fig. E.2.3)
Fig. 2.21. End fixity condition, k, for non-sway buckling (ENV 1993-1-1 [12] Fig.
E.2.1)
33
Fig. 2.22. End fixity condition, k, for sway buckling (ENV 1993-1-1 [12] Fig. E.2.2)
This model (for sway and for non-sway buckling) can be expanded to continuous
columns, presuming the loading factor N/Ncr is constant on their entire length. If this
does not happen (which is the actual case) the procedure is conservative for the
most critical part of the column [12]. For continuous columns, the distribution factors
are calculated using the following relations:
K C + K1
η1 = (ENV 1993-1-1 [12], rel. (E.3)) (2.65)
K C + K1 + K11 + K12
KC + K2
η2 = (ENV 1993-1-1 [12], rel. (E.4)) (2.66)
K C + K 2 + K 21 + K 22
where K1 and K2 are the values of the stiffness of the neighbour columns (Fig. 2.23).
34
Fig. 2.23. Distribution factors for continuous columns (ENV 1993-1-1 [12] Fig. E.2.4)
Presuming the beams are not subject to axial forces, their stiffness can be taken
from table 2.1, as long as they remain in the elastic range (ENV 1993-1-1 [12]).
Tab. 2.1. Stiffness of a beam in the elastic range (ENV 1993-1-1 [12] Tab. E.1)
35
For regular buildings with rectangular frames and reinforced concrete floors, subject
to uniform loads, it is accepted to consider the stiffness of the beams given in table
2.2.
Tab. 2.2. Stiffness K of beams – structures with reinforced concrete floors (ENV
1993-1-1 [12] Tab. E.2)
When the beams are subject to axial forces, stability functions must be used for
expressing their stiffness. A simplified conservative approach is proposed in ENV
1993-1-1 [12], neglecting the increase of stiffness generated by tension and
considering only compression in the beams. Based on these assumptions, the
values in table 2.3 can be considered.
Tab. 2.3. Stiffness of beams in compression (ENV 1993-1-1 [12] Tab. E.3)
where:
π 2 ⋅ EI
NE = (2.67)
L2
36
2.2.3. Empirical relations for the buckling length of columns
ENV 1993-1-1 [12] provides empirical expressions as safe approximations that can
be used as an alternative to the values from figures 2.21 and 2.22. The k coefficient
for the buckling length can be calculated by the following relations:
a. for non-sway buckling mode (Fig. 2.21)
k = 0,5 + 0,14 ⋅ (η1 + η2 ) + 0,055 ⋅ (η1 + η2 )
2
([12], rel. (E.5)) (2.68)
or, alternatively,
1,0 + 0,145 ⋅ (η1 + η2 ) − 0,265 ⋅ η1 ⋅ η2
k= ([12], rel. (E.6)) (2.69)
2,0 − 0,364 ⋅ (η1 + η2 ) − 0,247 ⋅ η1 ⋅ η2
b. for sway buckling mode (Fig. 2.22)
If the buckling length is generally easy to identify for members subject to axial
compression forces, the effective lateral buckling length is a more delicate subject,
given the complexity of the deformed shape (combination of flexural buckling and
torsion). This leads to a temptation to simplified approaches, like considering the
effective lateral buckling length as equal to the distance between points of zero (Fig.
2.24) in the bending moment diagram, or between inflection points of the deformed
shape of the beam about the strong axis of its cross-section [14].
37
Fig. 2.24. Zero bending moment points along a beam [14]
The previous approaches are valid only in the elastic range, using the theory of
bifurcation of equilibrium. Several researchers tried to express buckling when at least
one of the following Euler’s requirements for the bifurcation of equilibrium is not
fulfilled:
• the axis of the member is rigorously straight;
• the compression load acts strictly in the centre of gravity of the cross-section;
• the cross-section is bi-symmetrical;
• the moment of inertia of the cross-section is constant all along the bar;
• the deflected shape is a sinusoid;
• the material is homogenous and has a perfectly elastic behaviour (E = constant).
38
Britain, Italy, Netherlands and Yugoslavia) during the decade 1960 – 1970. Following
this, at the beginning one buckling curve and then several ones were drawn,
depending on:
• the shape of the cross-section;
• the plane in which buckling occurs – the axis of the cross-section;
• the yield limit of the steel grade.
There are situations in practice when a bar in compression does not have the same
cross-section on its entire length. This could be the case of a column with a change
in cross-section (Fig. 2.25) or of a pin-connected member, like a bar connected on
gusset plates at its ends (Fig. 2.26). In such cases, an equivalent buckling length can
be used, based on the theoretical approaches of Timoshenko [15].
I2 I1
P
L2 L1
Fig. 2.25. Column with a change in cross-section
Considering the notations in figure 2.25, Timoshenko writes the following equilibrium
equations:
39
d 2 y1
E ⋅ I1 ⋅ = P ⋅ (δ − y1 )
dx 2
(2.71)
d 2 y2
E ⋅ I2 ⋅ = P ⋅ (δ − y 2 )
dx 2
where:
δ – the displacement at the free end of the column;
y1 – the deformed shape of part 1;
y2 – the deformed shape of part 2.
L = L1 + L2
Using the notations,
P P
k 12 = k 22 = (2.72)
E ⋅ I1 E ⋅ I2
the following transcendental equation is obtained for determining the critical load:
tan (k 1 ⋅ L1 ) × tan (k 2 ⋅ L 2 ) =
k1
(2.73)
k2
Replacing the following (Fig. 2.26):
a L−a
L2 → and L1 → (2.74)
2 2
the critical force can be expressed as (in this case, δ is the displacement in the
middle of the length of the bar):
m ⋅ E ⋅ I2
Pcr = (2.75)
L2
where the values of the factor m are given in table 2.4.
I1 I2 I1
P P
40
where:
π
β= (2.77)
m
Tab. 2.4. Values for the m factor [15] for figure 2.26
a/L
0,2 0,4 0,6 0,8
I1/I2
0,01 0,15 0,27 0,60 2,25
0,1 1,47 2,40 4,50 8,59
0,2 2,80 4,22 6,69 9,33
0,4 5,09 6,68 8,51 9,67
0,6 6,98 8,19 9,24 9,78
0,8 8,55 9,18 9,63 9,84
The previous equations are written using the elastic approach, by writing elastic
equilibrium equations. An alternative solution can be obtained using the energy
method. In this idea, the deformed shape in figure 2.25 is considered [15]:
π⋅x
y = δ ⋅ 1 − cos (2.78)
2⋅L
As a result, for the problem in figure 2.25, the critical force is obtained [15]:
π2 ⋅ E ⋅ I2 1
Pcr = ⋅ (2.79)
4⋅L 2
L 2 L1 I 2 1 I 2 π ⋅ L2
+ ⋅ − ⋅ − 1 ⋅ sin
L L I1 π I1 L
L 2 L1 I 2 1 I 2 π ⋅ L2
β = 2⋅ + ⋅ − ⋅ − 1 ⋅ sin (2.80)
L L I1 π I1 L
Similarly, for the problem in figure 2.26, the critical force is [15]:
π2 ⋅ E ⋅ I 2 1
Pcr = ⋅ (2.81)
a L − a I2 1 I2 π⋅a
2
L
+ ⋅ − ⋅ − 1 ⋅ sin
L L I1 π I1 L
a L − a I2 1 I2 π⋅a
β= + ⋅ − ⋅ − 1 ⋅ sin (2.82)
L L I1 π I1 L
41
It is known that the energy approaches generally overestimate the value of the
critical force, leading to values superior to the actual ones.
L a
x
Fig. 2.27. Column with a continuous variation of the cross-section
Using the notations in figure 2.27, the variation of the cross-section is described by:
n
x
I(x ) = I1 ⋅ (2.83)
a
where I1 is the second moment of the area (moment of inertia) of the top cross-
section of the column (where x = a).
Timoshenko [15] writes the following equilibrium equation:
n
x d y
2
E ⋅ I1 ⋅ ⋅ 2 = −P ⋅ y (2.84)
a dx
In the general case, the equation can be solved using Bessel functions for any value
of n. For the particular case of n = 2, that would correspond to an I-shape with
constant flanges and continuous variation of the height of the web, the critical force
can be expressed [15] as:
m ⋅ E ⋅ I2
Pcr = (2.85)
L2
where I2 is the second moment of the area (moment of inertia) of the bottom cross-
section of the column (where x = a + L). The buckling length can be expressed using
relations (2.76) and (2.77) and values of the factor m given in table 2.5.
42
Tab. 2.5. Values for the m factor [15] for figure 2.27 and n = 2
I1/I2 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0
m 0,250 1,350 1,593 1,763 1,904 2,023 2,128 2,223 2,311 2,392 π2/4
Bracings are essential components for the structural stability control. They can
provide the lateral supports that are needed for preventing buckling.
Various criteria are used for classifying bracings, as: nodal or relative; punctual or
continuous; against translation, against rotation or against both, specific for buckling
or for lateral buckling etc. Usually, the main supports have also stability functions. A
standard case is that of the classical fork support (Fig. 2.28). Some usual bracing
cases are presented in figure 2.28.
43
Fig. 2.28. Usual bracing systems [14]
44
2.3.1. The bow imperfection (imperfecțiunea inițială în arc)
To take into account the fact that the bar is not perfectly straight, EN 1993-1-1 [6]
uses an equivalent bow imperfection (Fig. 2.29).
P P N P
H H
q α
q
e0 e0
L L
45
q ⋅ L2
M max = N ⋅ e 0 = (2.90)
8
The values recommended for e0 are given in table 2.6.
Tab. 2.6. Recommended values for bow imperfection (EN 1993-1-1 [6] Tab. 5.1)
For the buckling curves given in EN 1993-1-1 [6], Hmax results as (0.9...2.7)% of the
compression force.
To take into account the fact that the force does not act exactly on the centroid line
of the bar, EN 1993-1-1 [6] uses an equivalent sway imperfection (Fig. 2.30).
P P
H
46
H = φ⋅P (2.91)
where:
φ = φ0 ⋅ α h ⋅ α m (EN 1993-1-1 [6], rel. (5.5)) (2.92)
ϕ0 – basic value, ϕ0 = 1/200;
αh – the reduction factor for height h (h = L in Fig. 2.30) applicable to columns:
2 2
αh = but ≤ αh ≤ 1 (2.93)
h 3
h – “the height of the structure in meters”; in this case, h = L;
αm – “the reduction factor for the number of columns in a row”;
1
α m = 0,5 ⋅ 1 + (2.94)
m
m – “the number of columns in a row including only those columns which carry a
vertical load NEd not less than 50% of the average value of the column in the
vertical plane considered” (Fig. 2.31).
Fig. 2.31. Equivalent sway imperfections (EN 1993-1-1 [6] Fig. 5.2)
where:
HEd – “the design value of the horizontal reaction at the bottom of the storey to the
horizontal loads and fictitious horizontal loads”;
VEd – “the total design vertical load on the structure on the bottom of the storey”.
47
2.4. STRUCTURAL STABILITY ANALYSES
The present day computer programs for structural analysis and computing devices
used to run these applications allow analysing a structure by writing the equilibrium
equations on its deformed shape. However, this type of analysis is not always
necessary. In general, two types of analyses can be carried out on a structure:
• first order analysis – equilibrium is expressed on the initial shape of the
structure (efforts do not increase because of the displacements);
• second order analysis – equilibrium is expressed on the deformed shape of
the structure (efforts increase, significantly, because of the displacements).
Provided that the compression forces in the beams are not important, αcr for portal
frames with shallow roof slopes can be calculated with the following approximative
formula (Fig. 2.32):
H h
α cr = Ed ⋅
(EN 1993-1-1 [6], rel. (5.2)) (2.97)
VEd δ H ,Ed
48
Fig. 2.32. Notations for determining αcr (EN 1993-1-1 [6] Fig. 5.1)
where:
HEd – the design value of the horizontal reaction at the bottom of the storey to the
horizontal loads and fictitious horizontal loads;
VEd – the total design vertical load on the structure on the bottom of the storey;
h – the storey height;
δH,Ed – the horizontal displacement at the top of the storey, relative to the bottom of
the storey, when the frame is loaded with horizontal loads and fictitious
horizontal loads which are applied at each floor level.
Remarks:
1. In the absence of more detailed information, in the previous formulation, a roof
slope may be considered to be shallow if it is not steeper than 1:2 (26°).
2. In the absence of more detailed information, in the previous formulation, the axial
compression in the beams may be assumed to be significant if:
A ⋅ fy
λ ≥ 0,3 ⋅ (EN 1993-1-1 [6], rel. (5.3)) (2.98)
N Ed
where:
λ – the in-plane non dimensional slenderness calculated for the beam considered
as hinged at its ends of the system length measured along the beam;
NEd – the design value of the compression force.
49
For single storey frames, in the elastic range, if αcr ≥ 3,0, second order sway effects
due to vertical loads may be calculated by increasing the horizontal loads HEd and
equivalent loads φ × VEd due to imperfections by the factor:
1
(EN 1993-1-1 [6], rel. (5.4)) (2.99)
1
1−
α cr
“When performing the global analysis for determining end forces and end moments
to be used in member checks, local bow imperfections may be neglected. However
for frames sensitive to second order effects local bow imperfections of members
additionally to global sway imperfections should be introduced in the structural
analysis of the frame for each compressed member where the following conditions
are met:
• at least one moment resistant joint at one member end;
A fy
• λ > 0,5 (EN 1993-1-1 [6], rel. (5.8)) (2.100)
N Ed
where:
λ – the in-plane non-dimensional slenderness calculated for the member
considered as hinged at its ends;
NEd – the design value of the compression force”.
where:
χ ⋅ λ2
1−
γ M1
e 0 = α ⋅ (λ − 0,2 ) ⋅
M Rk
⋅ for λ > 0,2 (EN 1993-1-1 [6], rel. (5.10)) (2.102)
N Rk 1 − χ ⋅ λ2
50
α ult ,k
λ= is the relative slenderness (EN 1993-1-1 [6], rel. (5.11)) (2.103)
α cr
Lateral-torsional buckling may occur in the case of beams or lattice girders. It can be
prevented either by performing checks using suitable relations, or by introducing
lateral bracings whose purpose is to reduce the distance on which this phenomenon
may occur.
The relation given in EN 1993-1-1 [6] for the lateral-torsional buckling resistance is:
fy
M b,Rd = χ LT ⋅ Wy ⋅ (EN 1993-1-1 [6], rel. (6.55)) (2.104)
γ M1
where:
χLT – the reduction factor for lateral-torsional buckling;
Wy – the appropriate section modulus:
Wy = W pl,y for Class 1 or 2 cross-sections
Wy = W el,y for Class 3 cross-sections
Wy = W eff,y for Class 4 cross-sections
fy – the yielding limit;
51
γM1 – the partial safety factor for resistance of members to instability assessed by
member checks (γM1 = 1,0 in the National Annex of EN 1993-1-1 [6]);
χLT is taken form the appropriate buckling curve, based on the non-dimensional
slenderness:
Wy ⋅ f y
λ LT = (2.105)
M cr
where Mcr is the elastic critical moment for lateral-torsional buckling, whose
expression is not given in EN 1993-1-1 [6]. Under these circumstances, it can be
taken from recognised sources like publications, or computer programs. Some
examples of such sources are the following ones:
• the previous version of Eurocode 3 - ENV 1993-1-1 [12];
• published books (ex. Timoshenko, Gere [15] etc.);
• „Non-Contradictory Complementary Information” documents (NCCI), ex.
SN003b-EN-EU [16];
• computer programs like LTBEAM [17], developed at CTICM (Centre
Technique Industriel de la Construction Métallique).
The relation recommended by SN003b-EN-EU [16] is:
π 2 ⋅ E ⋅ I z k z I w (k z ⋅ L ) ⋅ G ⋅ I T
2
( )
2
M cr = C1 ⋅ ⋅ + + C ⋅ z
2
− C ⋅ z (2.106)
(k z ⋅ L)2 k w I z π2 ⋅ E ⋅ I z
2 g 2 g
where:
E – modulus of elasticity (Young’s modulus E =210000N/mm2);
G – shear modulus (G =81000N/mm2);
Iz – the second moment of the area about the weak axis (z – z);
IT – the torsion constant;
Iw – the warping constant;
L – the beam length between points which have lateral restraints;
kz – the effective length factor that refers to the end rotation about the z – z axis;
kw – the effective length factor that refers to the end warping;
zg – the distance between the point of load application and the shear centre;
C1, C2 – coefficients depending on the loading and end restraint conditions.
It is to note that the value of the critical moment is influenced by the position of the
loading point. The load can have a stabilizing or a destabilizing effect (Fig. 2.33).
52
destabilizing stabilizing
compression flange
effect effect
tension flange
The relation (2.104), given in EN 1993-1-1 [6], can be used by means of three
methods given in the code.
For all cross-sections, unless otherwise specified, the following relation can be used:
1
χ LT = but χ LT ≤ 1,0 (EN 1993-1-1 [6], rel. (6.56)) (2.107)
2
Φ LT + Φ 2LT − λ LT
where:
[ ( ) 2
Φ LT = 0,5 ⋅ 1 + α LT ⋅ λ LT − 0,2 + λ LT ] (2.108)
αLT – imperfection factor, given in the (Romanian) National Annex of EN 1993-1-1
[6]; the recommended values can be taken from table 2.7.
Tab. 2.7. Values for lateral torsional buckling (EN 1993-1-1 [6] Tab. 6.3)
Buckling curve a b c d
Imperfection factor αLT 0,21 0,34 0,49 0,76
53
The proper buckling curve, depending on the type of cross-section, is chosen based
on the recommendations given in table 2.8.
Tab. 2.8. Recommended values for lateral torsional buckling curves for cross-
sections using relation (2.107) (EN 1993-1-1 [6] Tab. 6.4)
M Ed 2
For values of the slenderness λ LT ≤ λ LT, 0 or for ≤ λ LT , 0 , lateral torsional buckling
M cr
effects may be ignored and only cross sectional checks apply. The maximum
recommended value for λ LT, 0 , which was adopted in the Romanian National Annex
of EN 1993-1-1 [6] is 0,4.
2.6.2. The specific method for rolled sections or equivalent welded sections
This method is a particular case for rolled sections or equivalent welded section. In
this case,
χ LT ≤ 1,0
1
χ LT = but 1 (EN 1993-1-1 [6], rel. (6.57)) (2.109)
χ LT ≤ 2
2
Φ LT + Φ 2LT − β ⋅ λ LT λ LT
where:
[ ( ) 2
Φ LT = 0,5 ⋅ 1 + α LT ⋅ λ LT − λ LT 0 + β ⋅ λ LT ] (2.110)
αLT – imperfection factor.
λ LT 0 ≤ 0,4 the value in the Romanian National Annex of EN 1993-1-1 [6] λ LT 0 = 0,4
β ≥ 0,75 the value in the Romanian National Annex of EN 1993-1-1 [6] β = 0,75
The values of αLT are taken from table 2.7, depending on the proper buckling curve,
chosen based on the recommendations given in table 2.9.
54
Tab. 2.9. Recommended values for lateral torsional buckling curves for cross-
sections using relation (2.109) (EN 1993-1-1 [6] Tab. 6.5)
This method is also specific for rolled sections or equivalent welded section. A
correction is introduced, to take into account the bending moment diagram along the
bar. In this idea, a modified reduction factor is calculated, as follows:
χ LT
χ LT ,mod = but χ LT , mod ≤ 1,0 (EN 1993-1-1 [6], rel. (6.57)) (2.111)
f
where:
χ LT – the reduction factor obtained at 2.6.2, using relation (2.109);
[ (
f = 1 − 0,5 ⋅ (1 − k c ) ⋅ 1 − 2,0 ⋅ λ LT − 0,8 )]
2
but f ≤ 1,0 (2.112)
Moment distribution kc
1,0
ψ=1
1
1,33 − 0,33ψ
-1 ≤ ψ ≤ 1
0,94
0,90
0,91
0,86
0,77
55
0,82
In the case where the compressed flange is provided with discrete lateral restraints
(the recommended situation) at a distance Lc between two consecutive ones, the
beam is not susceptible to lateral-torsional buckling if:
k cLc M c ,Rd
λf = ≤ λ c0 (EN 1993-1-1 [6], rel. (6.59)) (2.113)
i f ,z λ 1 M y,Ed
where:
My,Ed– the maximum design value of the bending moment within the restraint
spacing;
fy
M c, Rd = Wy ⋅ ;
γ M1
kc – a slenderness correction factor for moment distribution between restraints,
given in table 2.10;
if,z – the radius of gyration of the equivalent compression flange composed of the
compression flange plus 1/3 of the compressed part of the web area, about
the minor axis of the section;
I eff ,f
i f ,z = ;
1
A eff ,f + ⋅ A eff , w ,c
3
where:
Ieff,f – the effective second moment of area of the compression flange about
the minor axis of the section;
Aeff,f – the effective area of the compression flange;
Aeff,w,c – the effective areas of the compressed part of the web.
λ c 0 – a slenderness limit of the equivalent compression flange defined as follows; it
is given in the National Annex of EN 1993-1-1 [6]; the recommended value is
λ c 0 = λ LT ,0 + 0,1 ;
56
E
λ1 = π = 93,9ε ;
fy
235
ε= (fy in N/mm2).
fy
If the requirements from relation (2.113) are not fulfilled, the design buckling
resistance moment is expressed as:
M b,Rd = k fl ⋅ χ ⋅ M c ,Rd but M b.Rd ≤ M c.Rd (EN 1993-1-1 [6], rel. (6.60)) (2.114)
where:
χ – the reduction factor of the equivalent compression flange determined with λ f ;
kfℓ – the modification factor accounting for the conservatism of the equivalent
compression flange method.
The recommended value is kfℓ = 1,10 and it was adopted in the Romanian National
Annex of EN 1993-1-1 [6]. χ is determined based on λ f , using curve d for welded
h
sections having ≤ 44ε and curve c for all other situations, where:
tf
h – the overall depth of the cross-section;
tf – the thickness of the compression flange.
Buckling and lateral-torsional buckling checks strongly depend on the position of the
bracing points along the structural member in discussion. The requirements for a
bracing system are not only in terms of strength but stiffness too. In fact, if a bracing
system needs important displacements to resist against forces generated by the
buckling trend of an element, buckling of the braced element could occur.
57
by EN 1993-1-1 [6], as in the American code requirements are expressed not only as
strength but as stiffness too. Winter, Yura and others [7] showed that the minimum
theoretical stiffness of bracings, evaluated for the ideal member, is not enough for
the actual imperfect member. Figure 2.34 and table 2.11 show examples [14] of
strength and stiffness requirements for bracing systems in ANSI/AISC 360-16 [7].
P P
P
Lb Pr,b
βbr
Lb
Pr,b
βbr
P P Lb
Pr,b Pr,b
βbr βbr Lb Pr,b
Lb Lb Lb Lb
Lb Pr,b βbr
βbr Pr,b
βbr
(a) (b) (c) (d) (e)
Fig. 2.34. Strength and stiffness requirements in ANSI/AISC 360-16 [7]
58
(b) Pr ,b = 0,004 ⋅ Pr ([7], rel. (A-6-1)) 1 2 ⋅ Pr
β br = ⋅ ([7], rel. (A-6-2))
φ L b
c) Pr ,b = 0,01 ⋅ Pr ([7], rel. (A-6-3)) 1 8 ⋅ Pr
β br = ⋅ ([7], rel. (A-6-4))
φ L b
(d) M r ⋅ Cd 1 4 ⋅ M r ⋅ Cd
Pr ,b = 0,008 ⋅ ([7], rel. (A-6-5)) β br = ⋅ ([7], rel. (A-6-6))
h0 φ L b ⋅ h 0
(e) M r ⋅ Cd 1 10 ⋅ M r ⋅ C d
Pr ,b = 0,02 ⋅ ([7], rel. (A-6-7)) β br = ⋅ ([7], rel. (A-6-8))
h0 φ L b ⋅ h 0
In figure 2.34 and table 2.11, case (d) refers to relative brace while case (e) is for
nodal brace. Relative brace – brace that controls the relative movement of two
adjacent brace points along the length of a beam or column or the relative lateral
displacement of two stories in a frame (ANSI/AISC 360-16 [7]). Nodal brace – brace
that prevents lateral movement or twist independently of other braces at adjacent
brace points (ANSI/AISC 360-16 [7]). Bracing – members or system that provides
stiffness and strength to limit the out-of-plane movement of another member at a
brace point (ANSI/AISC 360-16 [7]).
59
2.8.1. Resistance of cross-sections in compression and bending
where MN,Rd is the design plastic moment resistance reduced due to the axial
force NEd;
• for class 3 and class 4 cross-sections, in the absence of shear force, the general
relation is:
fy
σ x ,Ed ≤ (EN 1993-1-1 [6], rel. (6.42), (6.43)) (2.116)
γ M0
where σx,Ed is the design value of the local longitudinal stress due to moment
and axial force taking account of fastener holes where relevant.
For bisymmetrical I- and H- sections or other flanges sections, no reduction of the
plastic resistance moment MN,y,Rd about the y-y axis needs to be done when both the
following criteria are satisfied:
N Ed ≤ 0,25 ⋅ N pl, Rd and (EN 1993-1-1 [6], rel. (6.33)) (2.117)
0,5 ⋅ h w ⋅ t w ⋅ f y
N Ed ≤ (EN 1993-1-1 [6], rel. (6.34)) (2.118)
γ M0
For bisymmetrical I- and H- sections or other flanges sections, no reduction of the
plastic resistance moment MN,z,Rd about the z-z axis needs to be done if:
hw ⋅ tw ⋅ fy
N Ed ≤ (EN 1993-1-1 [6], rel. (6.35)) (2.119)
γ M0
Otherwise, if the requirements in relations (2.117), (2.118), or (2.119) are not fulfilled,
the following reductions must be done:
1− n
M N , y , Rd = M pl, y , Rd ⋅ but M N , y, Rd ≤ M pl, y, Rd (EN 1993-1-1 [6], rel. (6.36))(2.120)
1 − 0,5 ⋅ a
for n ≤ a: M N ,z , Rd = M pl,z , Rd (EN 1993-1-1 [6], rel. (6.37)) (2.121)
n − a 2
for n > a: M N ,z , Rd = M pl,z , Rd ⋅ 1 − (EN 1993-1-1 [6], rel. (6.38)) (2.122)
1 − a
where:
60
N Ed
n=
N pl, Rd
A − 2 ⋅ b ⋅ tf
a= but a ≤ 0,5
A
For bi-axial bending:
α β
M y , Ed M z , Ed
+ ≤1 (EN 1993-1-1 [6], rel. (6.41)) (2.123)
M N , y, Rd M N ,z , Rd
in which α and β are constants, which may conservatively be taken as unity,
otherwise as follows:
• I and H sections:
α = 2 ; β = 5n but β ≥ 1
α = 2 ;β = 2
1,66
α=β= but α = β ≤ 6
1 − 1,13 n 2
EN 1993-1-1 [6] uses a pair of two relations for buckling check of a member in
compression and bending. These relations can lead to different results, as some of
the factors that intervene can be calculated according to two different procedures,
given in Annexes A and B. The checking relations are the following ones:
N Ed M y , Ed + ∆M y , Ed M + ∆M z , Ed
+ k yy + k yz z , Ed ≤1
χ y N Rk M M z , Rk
χ LT y , Rk
γ M1 γ M1 γ M1
61
where NEd, My,Ed and Mz,Ed are the design values of the compression force and the
maximum moments about the y-y and z-z axis along the
member, respectively
∆My,Ed, ∆Mz,Ed are the moments due to the shift of the centroidal axis for
class 4 sections
kyy, kyz, kzy, kzz are the interaction factors that depend on the chosen
method (from Annex A or from Annex B)
In the Romanian National Annex, the method in Annex A is recommended.
Depending on the class of the cross-section, characteristics given in table 2.12
should be used.
Tab. 2.12. Values for NRk = fy Ai, Mi,Rk = fy W i and ∆Mi,Ed (EN 1993-1-1 [6], Tab. 6.7)
Class 1 2 3 4
Ai A A A Aeff
Wy Wpl,y Wpl,y Wel,y W eff,y
Wz Wpl,z Wpl,z Wel,z W eff,z
∆My,Ed 0 0 0 eN,y NEd
∆Mz,Ed 0 0 0 eN,z NEd
Example diagram is triangular, with the maximum value at one end and 0
(zero) at the other end. The cross-section of the bar has a 10 × 400
mm web and the flanges are 20 × 300 mm. The length of the bar
considered in this example is 5m.
62
• two buckling relations using Annex A factors;
• two buckling relations using Annex B factors.
The resistance of the bar is obtained as the minimum given by the
three relations, depending on the Annex that is used.
ANSI/AISC 360-16 [7] uses the following relations for buckling check of members in
compression and bending:
a) when Pr/Pc ≥ 0,2:
Pr 8 M rx M ry
+ ⋅ + ≤ 1,0 (ANSI/AISC 360-16 [7], rel. (H1-1a)) (2.126)
Pc 9 M cx M cy
Pr M M ry
+ rx + ≤ 1,0 (ANSI/AISC 360-16 [7], rel. (H1-1b)) (2.127)
2 ⋅ Pc M cx M cy
where:
Pr – required axial strength;
63
Pc – available axial strength;
Mr – required flexural strength;
Mc – available flexural strength;
x – subscript relating symbol to major axis bending;
y – subscript relating symbol to minor axis bending.
2.8.4. Commentary
There are situations when simplified relations are useful for the design of arches or
single storey frames. The following relations can be used [18].
• Arches – for two- and three-hinged arches (Fig. 2.36) with the ratio h/L
between 0,15 and 0,5 and essentially uniform cross-section, the in-plane
buckling length may be taken as:
L cr = 1,25 ⋅ s (2.128)
where s is half of the arch length;
64
Fig. 2.36. Buckling length for a two-hinged arch [18]
• Two- and three-hinged frames – if the inclination of the columns is less than
15°, the following relation can be used for the column buckling length (Fig.
2.37):
I⋅s E⋅I
L cr = h ⋅ 4 + 3,2 ⋅ + 10 ⋅ (2.129)
I0 ⋅ h h ⋅ Kr
For the buckling length of the rafter, the following relation can be used:
I⋅s E⋅I I ⋅N
L cr = h ⋅ 4 + 3,2 ⋅ + 10 ⋅ ⋅ 0 (2.130)
I0 ⋅ h h ⋅ Kr I ⋅ N0
where N and N0 are the axial forces in the column and in the rafter; in the
case of tapered cross-sections, the cross-sections at 0,65h or 0,65s (Fig.
2.37) respectively should be used;
65
• Columns or rafters with knee bracing – the in-plane buckling length for
columns (Fig. 2.38(a)) and for rafters (Fig. 2.38(b)) can be estimated as:
L cr = 2 ⋅ s l + 0,7 ⋅ s 0 (2.131)
(a) (b)
Fig. 2.38. Portal frame (a) and frame with V-shaped columns (b) [18]
a 3 ⋅ π2 ⋅ a ⋅ E ⋅ I
k = 1 + 2 ⋅ + 10 ⋅ (2.132)
s 4 ⋅ s 2 ⋅ (1 + a s ) ⋅ K r
where EI is the bending stiffness of the rafter for bending about the vertical
axis and Kr is the rotational stiffness of the connection between the rafter and
the compression ring for bending about the vertical axis.
66
Chapter 3
INSTABILITY OF PLATES
EN 1993-1-5 [19] uses two methods for considering plate buckling effects [20]:
• the effective cross-section approach (metoda secțiunii transversale eficace);
• the reduced strength method (metoda tensiunilor reduse).
The relations are developed presuming the resistance of the cross-section is
governed by the most compressed fibre. The following assumptions are considered
to be fulfilled [20]:
• the individual plate elements are quasi-rectangular (the angle between the
longitudinal edges is less than 10°);
• stiffeners, if any, are placed on orthogonal directions about the direct stresses;
• openings and cut outs, if any, are small;
• the cross-section is uniform;
• flange induced web buckling is prevented by constructional recommendations.
For panels having the edges at angles higher than 10°, it is conservative to consider
a panel having the dimensions equal to the maximum of the actual ones. The
openings are considered to be less than 5% of the element width.
67
Both methods are equivalent for single plate elements; differences can be noticed in
the case of cross-sections composed of several plates. The effective cross-section
approach allows the use of cross-sections with higher slenderness than the reduced
strength method and, as a result, the importance of serviceability limit state
requirements increases in this case [20].
In this method, a separate check is made for each individual loading situation. Then,
interaction formulae are used to check for the combined effects. The design is
governed by the yield strength in the most compressed fibre of the reduced cross-
section of the member [20].
The main steps of using the effective cross-section method are as follows [20]:
1. Determine the stress distribution based on the full cross-section;
2. Based on this distribution, determine the reduced cross-section of each individual
plate that is part of the cross-section;
3. Determine the stress distribution on the obtained effective cross-section,
consisting of reduced cross-sections of each component plate;
4. Establish the effective cross-section, based on the stress distribution from step 3.
5. Repeat steps 3 and 4 till the stress distribution is compatible with the cross-
section.
The post-buckling strength of a plate depends on the aspect ratio and on the
possible presence of stiffeners. Therefore, three models are used for the behaviour
of a plate element:
• the plate type behaviour;
• the column type behaviour;
• an interpolation zone.
68
3.1.1.1. Plate behaviour
A correct analysis of the elastic critical plate buckling stress can be done by means
of appropriate computer programs. Alternatively, EN 1993-1-5 [19] provides two
simple approaches, depending on the number of longitudinal stiffeners that are
present in the compression zone:
• multiple stiffeners (three or more);
• one or two stiffeners.
In the case of multiple stiffeners, the element is treated like an orthotropic plate with
smeared stiffeners, meaning that the total rigidity of the stiffeners is distributed on
the plate, transforming it into a fictitious one.
The effective cross-section is obtained by using a reduced width for the compression
parts.
b c ,eff = ρ ⋅ b c (3.1)
where:
ρ – the reduction factor for plate buckling;
bc – the actual width of the compressed part;
ρ depends on the stress distribution ψ across the width of the plate and on the
support condition along the longitudinal edges. It is calculated as follows:
• for internal compression elements (two longitudinal edges supported):
ρ = 1,0 for λ p ≤ 0,673 (3.2)
λ p − 0,055 ⋅ (3 + ψ )
ρ= ≤ 1,0 for λ p > 0,673 , where (3 + ψ ) ≥ 0
λ2p
λ p − 0,188
ρ= ≤ 1,0 for λ p > 0,748 ([19], rel. (4.3)) (3.5)
λ2p
as a column slenderness [20], as square root of the ratio between the characteristic
69
resistance of the cross-section and the critical force, considering Ac,eff the effective
area of the compression zone:
A c,eff ⋅ f y fy
λp = = (3.6)
A c,eff ⋅ σ cr σ cr
π2 ⋅ E
2
t
σ cr = k σ ⋅ σ E = k σ ⋅ ⋅
12 ⋅ (1 − ν ) b
2
(3.7)
fy b t
λp = = (3.8)
σ cr 28,4 ⋅ ε ⋅ k σ
where:
235
ε=
fy
α n
2
kσ = + (3.9)
n α
where:
a
α=
b
a – length of the considered panel in the direction of the direct stress;
b – width of the considered panel in the direction of the direct stress;
n – number of half sine waves in the direction of compression corresponding to α.
70
Tab. 3.1. Internal compression elements (EN 1993-1-5 [19], Tab. 4.1)
Tab. 3.2. Outstand compression elements (EN 1993-1-5 [19], Tab. 4.2)
71
3.1.1.2. Column behaviour
π2 ⋅ E
2
t
σ cr ,c ⋅
12 ⋅ (1 − ν ) a
= 2
(EN 1993-1-5 [19], rel. (4.8)) (3.10)
fy
λc = (EN 1993-1-5 [19], rel. (4.10)) (3.11)
σ cr ,c
and the reduction factor χc is obtained using the relations that describe the buckling
curves in EN 1993-1-1 [6], where the factor α is replaced by a factor αe, defined in
EN 1993-1-5 [19].
where:
σ cr ,p
ξ= − 1 but 0 ≤ ξ ≤ 1
σ cr ,c
72
Fig. 3.1. Interpolation between column behaviour χc and plate behaviour ρ [20]
3.1.1.4. Checks for instability produced by bending moment (and axial force)
73
3.1.1.5. Checks for instability produced by shear force
Slender web panels in shear can develop an important post-buckling resistance due
to the tension field that is formed after local buckling. The present day models used
in EN 1993-1-5 [19] are based on Höglund’s model [21].
f yw 3 f yw
λw = = 0,76 (EN 1993-1-5 [19], rel. (5.3)) (3.15)
τ cr τ cr
and the elastic shear buckling stress of a perfect shear panel is:
2
π2 ⋅ E t
τ cr = k τ ⋅ σ E = k τ ⋅ ⋅
( )
12 ⋅ 1 − ν h w
2
(3.16)
74
h w 31
o ≥ ⋅ ε ⋅ kτ for a stiffened web (3.19)
tw η
the resistance to shear buckling should be checked and transverse stiffeners should
be provided at the supports. The recommended values for η are:
o η = 1,20 for S235, S275, S355, S460;
o η = 1,00 for steel grades higher than S460.
and they were adopted in the Romanian National Annex of EN 1993-1-5 [22].
The relation given in EN 1993-1-5 [19] for calculation of the shear buckling
resistance is:
η ⋅ f yw ⋅ h w ⋅ t w
Vb,Rd = Vbw ,Rd + Vbf ,Rd ≤
3 ⋅ γ M1
(EN 1993-1-5 [19], rel. (5.1)) (3.20)
where Vbw,Rd is the contribution of the web:
χ w ⋅ f yw ⋅ h w ⋅ t w
Vbw ,Rd = (EN 1993-1-5 [19], rel. (5.2)) (3.21)
3 ⋅ γ M1
and Vbf,Rd is the contribution of the flanges, considered only when the resistance of
the flanges is not completely used for resisting the bending moment (MEd < Mf,Rd),
where:
χw – the reduction factor for the shear resistance, depending on the web
slenderness λ w ; the values are given in table 3.3;
γM1 – the partial factor for the resistance to instability;
η – coefficient that includes the increase of shear resistance at smaller web
slenderness [20];
λ w ≥ 1,08 (
1,37 0,7 + λ w ) 0,83 λ w
75
The presence of η bigger than 1,0 is explained by test results [20] that showed an
increase of the shear strength up to 0,7 … 0,8 times the yielding limit for webs with
reduced slenderness; there is also a contribution of the flanges but, as these two
influences could not be separated, η is not allowed for single plates, without flanges.
b f ⋅ t f2 ⋅ f yf M Ed
2
(EN 1993-1-5 [19], rel. (5.8))
Vbf ,Rd = ⋅ 1− (3.22)
c ⋅ γ M1 M f ,Rd
where:
bf – width of the compressed flange, not larger than 15εtf on each side of the web;
Mf,Rd – moment resistance of the cross-section consisting of the effective area of the
flanges only (Fig. 3.3);
c – distance between the plastic hinges on the flange at the limits of the tension
field (Fig. 3.4).
According to Höglund [21], for steel plate girders, the distance c can be estimated
with the following approximate formula [20]:
M pl,f 1,6 b f t f2 f yf
c = a 0,25 + 1,6 ⋅ = a 0,25 + ([20], rel. (5.25)) (3.23)
M pl,w t w h 2w f yw
76
Fig. 3.4. Tension field carried by bending resistance of flanges ([20] Fig. 5.10)
where:
b f t f2 f yf
M pl,f = is the plastic moment resistance of the flanges (3.24)
4
t w h 2w f yw
M pl,w = is the plastic moment resistance of the web (3.25)
4
In the presence of an axial force NEd, the plastic moment resistance Mf,Rd must be
reduced by multiplying with the following factor:
1 − N Ed (EN 1993-1-5 [19], rel. (5.9)) (3.26)
(A f 1 + A f 2 ) ⋅ f yf
γ M0
Usually, the contribution of the flanges is small and can be neglected. It may be
important at the end supports, where the flanges are less loaded. The values of c
observed in tests are usually larger [20].
Based on the previous approaches, the contribution of the flanges, Vbf,Rd, can be
calculated as (Fig. 3.4):
4 ⋅ M pl,f ,Rd b f ⋅ t f2 ⋅ f yf
Vbf ,Rd = = (3.27)
c ⋅ γ M1 c ⋅ γ M1
Presuming the bending moment is resisted only by the flanges, leading to axial
forces in the flanges that reduce their plastic moment resistance, a correction factor
is used:
77
M
2
1 − Ed (3.28)
M f ,Rd
and so, relation (3.22) is obtained.
78
13. SN008a-EN-EU – NCCI: Buckling lengths of columns: rigorous
approach, Access Steel
14. Diacu I., Ștefănescu B. – Stability bracing in practice, Steel – A
New and Traditional Material for Building, Proceedings of the
International Conference in Metal Structures – Poiana Brașov,
România, September 20-22, 2006, D. Dubină, V. Ungureanu
Editors, Taylor & Francis Group, 2006, pag. 109-117
15. Timoshenko, S.P., Gere, J.M. – Theory of elastic stability, 2nd
Edition. McGraw-Hill, 1961
16. SN003b-EN-EU – NCCI: Elastic critical moment for lateral
torsional buckling, Access Steel
17. LTBEAM – http://www.cticm.com/content/ltbeam-version-1011
18. Blass H.J. – Timber Engineering STEP 1: Basis of design,
material properties, structural components and joints, Almere:
Centrum Hout, 1995
19. EN 1993-1-5 – Eurocode 3: Design of Steel Structures – Part 1-5:
Plated structural elements
20. Johansson B., Maquoi R. , Sedlacek G. , Müller C. , Beg D. –
Commentary and Worked Examples to EN 1993-1-5 “Plated
Structural Elements”, JRC – ECCS Joint Report, First Edition,
October 2007, EUR 22898 EN – 2007
21. Höglund T. – “Shear buckling resistance of steel and aluminium
plate girders” – Thin-Walled Structures, Vol. 29, Nos. 1-4,
pag.13-30, 1997
22. SR EN 1993-1-5/NA – Eurocod 3: Proiectarea structurilor din oțel
– Partea 1-5: Elemente structurale din plăci plane solicitate în
planul lor – Anexa Națională
79